Expression of a Tumor Necrosis Factor-a Transgene in Murine Lung Causes
Lymphocytic and Fibrosing Alveolitis
A Mouse Model of Progressive Pulmonary Fibrosis
Yoshitaka Miyazaki, Kimi Araki, Christian Vesin, Irene Garcia, Yusuf Kapanci, Jeffrey A. Whitsett,*Pierre-FrangoisPiguet,
and Pierre Vassalli
Department ofPathology, University of Geneva, CH 1211 Geneva 4, Switzerland; and *Division ofPulmonary Biology, Children's
Hospital Medical Center, Department ofPediatrics, College ofMedicine, University of Cincinnati, Cincinnati, Ohio 45229-2899
The murine TNF-a gene was expressed under the control of
the human surfactant protein SP-C promoter in transgenic
mice. A number of the SP-C TNF-a mice died at birth or
after a few weeks with very severe lung lesions. Surviving
mice transmitted a pulmonary disease to their offspring, the
severity and evolution of which was related to the level of
TNF-a mRNA in the lung; TNF-a RNA was detected in
alveolar epithelium, presumably in type II epithelial cells.
In a longitudinal study of two independent mouse lines,
pulmonary pathology, at 1-2 mo ofage, consisted ofa leuko-
cytic alveolitis with a predominance of T lymphocytes. Leu-
kocyte infiltration was associated with endothelial changes
and increased levels of mRNA for the endothelial adhesion
molecule VCAM-1. In the following months, alveolar spaces
enlarged in association with thickening of the alveolar walls
due to an accumulation of desmin-containing fibroblasts,
collagen fibers, and lymphocytes. Alveolar surfaces were
lined by regenerating type H epithelial cells, and alveolar
spaces contained desquamating epithelial cells in places.
Platelet trapping in the damaged alveolar capillaries was
observed. Pulmonary pathology in the SP-C TNF-a mice
bears a striking resemblance to human idiopathic pulmo-
nary fibrosis, in which increased expression of TNF-a in
type II epithelial cells has also been noted. These mice pro-
vide a valuable animal model for understanding the patho-
genesis of pulmonary fibrosis and exploring possible thera-
peutic approaches. (J. Clin. Invest. 1995. 96:250-259.) Key
words: surfactant * promoter * pneumocytes * VCAM-1.
Idiopathic pulmonary fibrosis is a group of usually fatal disor-
ders of the lung characterized by an alveolitis which progresses
to interstitial fibrosis (1). It is generally assumed that the infil-
tration of inflammatory cells and the proliferation of interstitial
Address correspondence to Pierre Vassalli, Department of Pathology,
University of Geneva, 1, rue Michel-Servet, CH 1211 Geneva 4, Swit-
zerland. Phone: 22-70-25-745; FAX: 22-70-25-746.
Receivedfor publication 30 January 1995 and accepted in revised
form 22 March 1995.
cells associated with fibrosis are caused by an overproduction of
cytokines. Platelet-derived growth factor (PDGF), transforming
growth factor /3 (TGF-/3), and tumor necrosis factor a (TNF-
a) have been incriminated in the pathogenesis of idiopathic
pulmonary fibrosis (2-5). TNF-a is a cytokine with both in-
flammatory and fibrogenic activities (6). TNF-a mRNA and
protein have been detected in lungs from patients with idio-
pathic pulmonary fibrosis (7, 8) and in lungs from mice with
pulmonary fibrosis elicited by exposure to bleomycin (9) or
silica (10). In those experimental models, pulmonary inflamma-
tion and fibrosis were prevented by injection of anti-TNF anti-
bodies (9, 10) or of TNF-a antagonists (11).
Administration of TNF-a, either intravenously or intratra-
cheally, elicits a severe alveolitis resembling the adult respira-
tory distress syndrome but with little fibrosis (12); this, how-
ever, is an acute type of exposure to TNF-a. To assess the
potential role ofTNF-a in the pathogenesis of the lesions associ-
ated with pulmonary fibrosis, it may be necessary to expose the
lung to small amounts of this cytokine for a prolonged period
of time. We have therefore produced transgenic mice in which
the TNF-a gene was placed under the control of the transcrip-
tional promoter of the surfactant protein C (SP-C)1 gene. SP-C
is synthesized and secreted by alveolar type II epithelial cells
and serves to enhance the spreading of surfactant phospholipids
at the alveolar surface to reduce surface tension of the air-
liquid interface (13). TNF-a was expressed in the lungs of these
transgenic mice, producing severe alveolitis, alveolar disrup-
tion, and progressive pulmonary fibrosis.
Mice. C57BLJ6xDBA2 F1 (B6D2Fl), C57BL/6, Balb/c mice of either
sex were purchased from IFFA Credo (L'Abresles, France). Transgenic
founder mice (B6D2FI) were backcrossed with C57BL16 or Balb/c mice
in our animal facilities to generate F1 hybrid transgenic mice.
Construction of the SP-C/TNF-a. A 3.7-kb fragment from the 5'-
flanking region of the human SP-C gene (14, 15) was cloned into the
Jolla, CA). Two kinds of chimeric constructs were prepared: (a) SPC/
TNF-a-UT, which contained the entire 3' untranslated region (UTR) of
the mouse TNF-a gene (16), obtained by introducing the NarI/SalI
fragment containing all the coding sequence and 3'UTR of the TNF-a
gene into the EcoRV site of pBSH containing the SP-C promoter; and
(b) SPCfTNF-a-pA, which lacked the 3'UTR sequence of the TNF-a
gene, obtained by introducing the NarI/EcoRI fragment of the TNF-a
site of a pBluescript II SK- plasmid (pBSII) (Stratagene, La
1. Abbreviations used in this paper: ICAM-1, intercellular adhesion
molecule-i; SP-C, surfactant protein C; UTR, untranslated region;
VCAM-1, vascular cell adhesion molecule-i.
Miyazaki et al.
J. Clin. Invest.
X The American Society for Clinical Investigation, Inc.
Volume 96, July 1995, 250-259
Figure 1. Restriction map of the two con-
structs used as transgenes. In the upper con-
struct, the 3'UT region of the TNF-a gene
has been replaced by a globin intron and a
polyadenylation sequence (pA) while the
lower construct contains the intact 3'UT re-
gion (3'UTR) of the TNF-a gene.
gene into the EcoRV site of pBSII containing the SP-C promoter. A
rabbit globin intron and a polyadenylation signal sequence obtained
from the pCAGGS plasmid (17) were placed 3' of the TNF-a gene.
The structure of these two constructs is shown in Fig. 1.
Generation and identification of transgenic mice. The SPC/TNF-a-
UT and SPCUTNF-a-pA DNAs were excised from the parental plasmids
by digestion with BamHI and XhoI or XhoI and purified by agarose
gel electrophoresis and passage through NACS PREPAC (GIBCO BRL,
Gaithersburg, MD). Recovered DNA was resuspended in 10 mM Tris-
HCl (pH 7.4)/I mM EDTA and the concentration was adjusted to 50
ng/,l.Purified DNA, kept at -20°C, was diluted 10 times with H20
before microinjection into the male pronucleus of B6D2F1 fertilized
eggs. After egg transplantation into pseudopregnant foster mothers,
transgenic offspring were identified by PCR analysis of genomic DNA
isolated from tail biopsies, using a 5' primer specific to the SPC promoter
(5' AGATATGTGGGAGGAGGCAA 3') and a 3' primer specific to
the TNF-a gene (5'GAGAAGAGGGTGAGACATAG 3'). Transgene
copy numbers were estimated by Southern blots of genomic DNA.
Northern blot analysis. RNAs, prepared from a variety of mouse
tissues by a guanidine isothiocyanate extraction procedure, were ana-
lyzed on Northern blots as described previously (18) by hybridization
with murine cRNAs probes. Equality of sample loading and efficiency
of the transfer were controlled by staining the filters with methylene-
blue in 0.5 M sodium acetate, pH 4.8. Filters were hybridized with 32P-
labeled cRNA probes for: TNF-a (18), interleukin-l (18), interleukin-
6, granulocyte macrophage colony stimulating factor (GM-CSF) (19),
c-sis PDGF-B (20), TGF-a (21) and
cell adhesion molecule-l (VCAM-1) (24), and intercellular adhesion
molecule-l (ICAM-1) (25).
In situ hybridization. 33P-labeled mouse TNF-a cRNA probes and
control sense probes were used for in situ hybridization, which was
performed as described elsewhere with detection of the radioactive
probes on emulsion coated slides (26) or by exposure to x-ray film (27).
Hydroxyproline assay. Hydroxyproline assay was performed as de-
scribed elsewhere (28).
Histological and morphometric analysis. Mice were exsanguinated
by opening the aorta under ether anesthesia. Tissues were taken and fixed
immediately by immersion in 4% buffered paraformaldehyde solution.
Lungs were fixed by intratracheal instillation of 2% glutaraldehyde or
4% paraformaldehyde after opening of the thorax. Tissues sections were
stained by hematoxylin and eosin, Prussian-blue, Grocott, and Gram or
by the Gomori's silver impregnation method (29). For morphometric
analysis, horizontally oriented histologic sections taken from the left
lung were examined with a Vidas image analysis system (C. Zeiss,
Oberkochen, Germany) using the software package VIDAS release 2.1
(Kontron Electronik, GmbH, Eching, Germany). The sections were
-(3 (22), E-selectin (23), vascular
scanned with a black and white Burle CCD camera, type FMC4005
(AVT-Horn, Aalen, Germany) through an Axiophot photomicroscope.
The CCD camera was connected via a fram grabber board to a 386
Intel PC. Three to five random selected fields were examined for each
section. Images were stored on an optical disk and displayed on a VGA/
RGGB monitor via a TV video input/output. The background of the
image was balanced within a rectangular 512 x 512 pixel frame. The
air spaces were reconstructed by a grain function and the size and
number of different areas were recorded. Areas in pixels were converted
to square microns using a calibrated reference surface.
Immunohistochemistry. Stainings were performed either on deparaf-
finized sections obtained from formol-fixed and paraffin-embedded tis-
sue or on cryostat sections from lungs instilled with OCT 4583 (Miles
Inc., Elkhart, IN). The following antibodies were used to identify the
cells or their content: for fibroblasts, biotinylated mouse monoclonal
anti-human a-smooth muscle actin (30), rabbit anti-human desmin
(31) (kind gifts from G. Gabbiani, University of Geneva, Geneva, Swit-
zerland), and goat polyclonal anti-human vimentin (Sigma lInmuno-
chemicals, St. Louis, MO); for epithelial cells, rat monoclonal antipan-
cytokeratin (Sigma Immunochemicals); for macrophages and T and B
lymphocytes, rat monoclonal antibodies against macrophages (F4/80)
and against mouse CD4 (GK1.5), anti-CD8 (H-35) T lymphocytes
(American Type Culture Collection, Rockville, MD), and biotinylated
anti-mouse Ig (Dako, Copenhagen, Denmark). Rabbit antibody anti-
TGF-,B synthetic peptides were a kind gift of Kathleen Flanders (Na-
tional Cancer Institute, National Institutes of Health, Bethesda, MD)
(32). Antibodies were detected by FITC-labeled goat anti-rat immuno-
globulin or by biotinylated anti-rat IgG (Dako) followed by avidin
Platelet trapping assay. Platelets were prepared in citrate-glucose
solution and labeled with 5'Cr as described (33). About 105 platelets
3 h later and radioactivity was measured in various organs.
Bacterial quantification. Lungs were isolated and homogenized in
a sterile culture medium. Dilutions of the homogenates were added to
Data analysis. All statistical data are expressed as mean values
+SD. One way ANOVA was performed using MicroCal ORIGINE 2.8
(MicroCal Software, Northampton, MA) for Microsoft Windows 3.1.
-105 cpm) were injected intravenously; mice were killed
Generation of transgenic mice. From 360 eggs injected with
the SPCfINF-a-pA DNA construct, which contains the mouse
TNF-a gene without its 3' untranslated region, no transgenic
Transgenic Model ofPulmonary Fibrosis
Figure 2. Northern blots and in situ hybridization of lung mRNAs. For
Northern blotting, equal amounts of lung RNA were placed in each
track, and the evenness of loading was judged by methylene blue stain-
ing of the filters showing the ribosomal RNA bands. TNF: lane 1, normal
mouse; lanes 2-5, transgenic mice of the same age (see text); mice 2
and 5 had more severe lesions. TGF,/: lanes I and 4, normal littermates;
lanes 2 and 3, transgenic mice. VCAM, E-selectin (E-SEL), and ICAM:
lane 1, normal littermates; lane 2, normal mice 2 h after LPS injection;
lane 3, transgenic mice. In situ hybridization of sections of the whole
lung are shown at top right: TNF, hybridization with a 33P-labeled
antisense TNF-a probe; C, TNF-a sense probe, with prolonged exposure
in order to see lung outline.
mice were obtained. This suggested that expression of this con-
struct results in lethality in utero. To further explore this possi-
bility, an additional 90 eggs were injected and the foster mothers
were killed on the 19th day of gestation. 5 dead fetuses were
recovered together with 12 living fetuses, which were killed for
DNA analysis; none of the living fetuses were transgenic, while
the 5 dead fetuses bore the transgene.
From the 500 eggs injected with the SPC/TNF-a-UT DNA,
which contains the entire 3'UT region of the TNF-a gene, 13
transgenic mice were obtained. One was dead at birth and two
died within 24 h, with cyanosis and signs of respiratory distress.
From the 10 transgenic mice which lived and had no early
signs of disease, 2 failed to gain weight normally and displayed
progressive respiratory symptoms, which led to their death at
-1 mo old. Eight mice developed normally and, after mating
with normal mice, passed the transgene to offspring following
Mendelian rules. Transgenic offspring of each of these lines
were killed between 1 and 2 mo of age and their lungs were
examined histologically. In each case, varying degrees of
lymphocytic alveolitis were observed. Only 2 lines (Tg-2 and
Tg-10), which by Southern blot analysis had
of the transgene (not shown), were kept by further breeding for
more detailed examination of the progression of the disease.
There were no consistent differences in the nature and severity
of the lesions observed in these two lines (except for some
variations within line Tg-2, as mentioned below), and the pa-
thology of the transgenic mice will thus be considered together.
Expression ofTNF-a mRNA in the lungs oftransgenic mice.
TNF-a mRNA was detected on Northern blots of lung RNAs
in all transgenic mice examined (Fig. 2); using the same condi-
tions of detection, TNF-a mRNA was not detected in other
tissues tested (liver, spleen, heart, kidney, uterus, brain, large
bowel, and lymph nodes, not shown). By in situ hybridization
of slices of whole lung, TNF-a mRNA was found to be evenly
10 copies each
expressed (Fig. 2), in contrast to the nodular distribution ob-
served previously in lungs of experimental silicosis using the
same technique (10). On histologic sections, the main source
of TNF-a mRNA appeared to be in alveolar cells, whose local-
ization suggested that they were mainly type II epithelial cells,
which is consistent with the pattern of expression of SP-C in
the mouse. In advanced lesions, however, the possibility that
some labeled cells were macrophages could not be ruled out
(Fig. 3 d); this would presumably correspond to TNF-a mRNA
of endogenous rather than transgenic origin. Since the severity
of the lesions observed on histologic sections from line Tg-2
showed some variability, nine transgenic mice and their lit-
termates were killed at 5 mo of age; the right lung was used
for RNA extraction and the left lung for histologic analysis.
Two of these mice had a level of TNF-a mRNA clearly higher
than the others (Fig. 2), and they also had the most severe lung
lesions. To assess whether the severity of the lung disease might
be influenced by infection, bacterial cultures of lung extracts
were performed, and yielded number of colonies per lung simi-
lar to those of control lungs; no bacteria were detected by Gram
stains of lung sections and by electron microscopy (see below).
It thus appeared that the variation observed in the Tg-2 line
reflected variation in the level of transgene expression perhaps
resulting from variable degree of methylation (34). Severity of
disease in offspring from Tg-10 line was more consistent and
correlated with the levels of TNF-a mRNA in the lung. At no
stage of the disease was TNF-a detected in the blood by a
biological assay. In general, mice were killed after 7 mo to
avoid death from progressive respiratory failure.
Pulmonary lesions analyzed by histology, electron micros-
copy, and immunohistochemistry. The two mice that died within
24 h after birth showed very severe pulmonary hypoplasia with
immature, dense alveolar tissue. The other two mice that died
at 1 mo had very severe interstitial pneumonia with large areas
of pulmonary consolidation, similar in severity to those de-
scribed below in older mice. From the two transgenic lines
chosen for extensive analysis, Tg-2 and Tg-10, 64 F1 mice were
killed at various intervals between 2 and 6 mo of age. For the
sake of clarity, average lesions at 2 and 6 mo will be described.
In Tg-2 line offspring, variation in the severity of the disease
resulted in lesions occasionally as severe at time of killing at 2
mo as those observed for other mice, killed at 6 mo, with a few
mice dying from respiratory failure before this age.
At 2 mo of age, the lungs were usually macroscopically
normal. Leukocytic alveolitis, usually more patchy than diffuse,
was a constant finding, with lymphocytic infiltrates more promi-
nent within the interlobular septa, around the extraalveolar small
vessels, and under the pleura. Immunohistochemistry showed a
majority of CD4+ lymphocytes, but CD8+ and B lymphocytes
were also present. Macrophages (identified by staining with the
F4/80 antibody) were not especially numerous; neutrophils were
also observed, in variable numbers. The air spaces were occa-
sionally lined by cuboid cells, which were identified by electron
microscopy as type II pneumocytes. In cases fixed by tracheal
instillation rather than by immersion, ruptures of some alveolar
septa, resulting in an enlargement of the alveolar spaces, were
observed (Fig. 4, a and b); this lesion was quantified by morpho-
metric image analysis (Fig. 5). By electron microscopy, an addi-
tional observation was that of frequent endothelial swelling in
the alveolar capillaries. The extent of fibrosis, as judged by
252Miyazaki et al.
123 4 5
Figure 3. (a) Histologic section stained with Gomori's silver impregnation method, showing the extent of collagen deposition (black) in the thickened
alveolar walls, x 100. (b) Immunohistochemical staining for desmin, showing numerous desmin-positive fibroblasts in the thickened alveolar walls,
X80. (c) Prussian blue staining showing, in black, cells laden with iron, either lining the alveolar space or in the air space. These cells appear to
be in part alveolar macrophages, in part alveolar epithelial cells. Iron is also seen in some interstitial cells of the thickened alveolar walls, x80.
(d) In situ hybridization with a 33P-labeled TNF-a cRNA probe. Labeled cells are numerous, and their localization is compatible with that of
pneumocytes. Hybridization with a sense probe showed very little background and no grain-bearing cell, x250.
silver staining of histologic sections, was usually minimal at
At 6 mo of age, the lungs did not collapse at the opening
of the thorax and had an irregular, yellow surface. On histologic
sections (Figs. 3 and 4), marked thickening of the alveolar septa
with remodeling of the air space was widespread. Lymphocytic
accumulation, sometimes associated with the presence of poly-
morphonuclear leukocytes and plasma cells, was in places still
conspicuous (Fig. 4), but most commonly decreased or even
absent. The respective contribution, in the distortion of lung
histologic structures, of epithelial cells, fibroblasts, and macro-
phages could be best assessed by immunocytochemistry or elec-
tron microscopy. Thickening of the septa resulted in large part
from an increased number of fibroblasts (Fig. 4 c), as judged
by the extent of immunohistochemical staining for desmin (Fig.
3 b). These last cells, however, were not stained by anti-smooth
muscle a-actin antibody, in contrast to what has been observed
with the myofibroblasts of bleomycin-induced pneumopathy in
rats (35). A marked fibrosis of the alveolar spaces was observed
on silver-stained sections (Fig. 3 a), and, by electron micros-
copy, collagen fibrils accumulation was extensive (Fig. 6 c).
To quantify the extent of fibrosis, hydroxyproline content was
determined in the lungs of 6-mo-old transgenic mice and of
their normal littermates, demonstrating an approximate three-
fold increase compared with controls (Fig. 7 a). The thickened
alveolar septa were lined by type II epithelial cells (Fig. 4 e),
which by electron microscopy (Fig. 6 a) usually appeared poor
in lamellated bodies and with empty vacuoles, a change proba-
bly indicative of rapid regeneration. In places, these cells were
stratified or desquamated in the alveolar lumen where they ap-
peared as clusters of foam cells, sometimes with cholesterol
crystals (Fig. 4 e). By histology, identification of the cells in
the alveolar spaces as macrophages or type H epithelial cells
was often difficult; immunofluorescent staining suggested that
they were more frequently epithelial cells than macrophages,
and on electron micrographs they were most often clearly identi-
fied as type II epithelial cells (Fig. 6 e). Altogether, these epithe-
lial alterations suggested a rather extensive process of cell dam-
age and regeneration among type II pneumocytes. Prussian blue,
detecting iron, conspicuously stained some alveolar and intersti-
tial macrophages and apparently some epithelial cells as well
(Fig. 3 c), probably reflecting previous microhemorrhages
which may have resulted from alveolar wall destruction. In
general, the various pulmonary lesions observed in these
Transgenic Model ofPulmonary Fibrosis
Miyazaki et al.
Alveolar surfaces (
transgenic mice bore a marked resemblance to the variety of
lesions observed in cases of idiopathic pulmonary fibrosis in
humans (see for comparison Fig. 4f).
Finally, a feature which could be observed only on ultrathin
sections was that of vascular changes, consisting mostly of a
thickening of the endothelial cytoplasm in the alveolar capillar-
ies (Fig. 6 d). Platelets lining the capillary endothelium and
platelet clumps were observed with unusual frequency. To as-
sess potential platelet trapping in these lungs, 51Cr-labeled plate-
lets were injected into groups of transgenic mice and control
littermates, demonstrating increased platelet trapping in the
lungs of transgenic mice (Fig. 7 b).
Expression ofmRNAsfor cytokines and adhesion molecules
in the diseased lungs. RNA from lungs of transgenic mice and
littermates killed at 5-6 mo were studied by Northern blot
analysis to determine the levels of mRNAs for interleukin-l,
TGF-a,TGF-fl1,GM-CSF, and c-sis PDGF-B. No significant
changes were observed in these cytokine mRNAs; lack of accu-
mulation ofTGF-f3mRNA (which showed slight variations
between controls as well as between transgenic mice, see Fig.
2) was in agreement with the failure to detect significant depos-
its or accumulation ofTGF-/3 by immunohistochemistry, except
/Figure 5. Morphometric analysis
of the size distribution of air
spaces seen on histologic sections
oflungsof normal (black bars) or
transgenic (hatched bars) mice.
Results are mean of the value ob-
served with five mice. Lungs of
transgenic mice have fewer air
spaces in the 22-144 um2 range,
while the number of larger spaces
in occasional macrophages (Fig. 8). In RNAs from the same
lungs, levels of mRNAs for E-selectin, ICAM-1, and VCAM-
1 were assessed, since exposure of venular endothelial cells to
TNF-a is known to induce the expression of these adhesion
molecules (36); RNAs from lungs of control littermates and
from the lungs of normal mice 2 h after LPS injection (a condi-
tion which markedly increases the level of these mRNAs) were
used for comparison. Only VCAM mRNA was found to be
increased in the transgenic mice (Fig. 2).
The mouse SP-C-TNF-a transgenic lines show that the perma-
nent production of TNF-a by type II pneumocytes leads to a
range of pulmonary lesions of varying severity. The prenatal
lethality observed with a transgene not including an intact TNF-
a 3' untranslated region (37) is likely to reflect a requirement
for this sequence to control the level ofTNF-a synthesis through
mechanisms of transcriptional and probably also translational
regulation (38). When the transgene included this regulatory
sequence, some of the transgenic mice, probably those express-
ing the highest amount of lung TNF-a, died at birth, with com-
Figure 4. Histologic sections of lungs at various stages of pulmonary disease. Hematoxylin and eosin stain. (a) Irregular and large alveolar spaces
probably resulting from ruptures of alveolar septa (morphometric analysis of this type of lesion is shown in Fig. 5). Lymphocytic infiltration is
most conspicuous in the sub-pleural region (left border of the field). Fixation by tracheal instillation, x40. (b) The thickened alveolar septa are
hypercellular and infiltrated mostly by lymphocytes (in majority CD4 and CD8 T lymphocytes, as detected by immunofluorescent staining). Fixation
by tracheal instillation, x80. (c) Higher magnification (x200) of the rather heterogeneous cell infiltrate of the interalveolar walls, which is made
of lymphocytes and of some polymorphonuclear leukocytes, detected by their convoluted nuclei. Fixation by immersion, as in the following plates.
(d) A nodular perivascular lymphocytic infiltration (with some degree of apoptosis) is seen at the center, with lymphocytes extending into thickened
adjacent septa, x160. (e) The severely thickened alveolar walls contain various cells, among which fibroblasts are a majority (best identified by
immunohistochemistry detecting intracytoplasmic desmin, see Fig. 3 b). The grossly altered alveolar space is lined by hyperplastic alveolar type H
cells (best identified by electron microscopy, see Fig. 6 a). The alveolar lumen on the right of the picture contains a cellular aggregate composed
of macrophages or type II epithelial cells (see ultrastructural appearance on Fig. 6 e), with cholesterol crystals (*), x250. (f) Histology of a human
case of pulmonary fibrosis, classified as "Usual Interstitial Pneumonitis," shown for comparison, x250; the severe thickening of the alveolar walls
also results mostly from fibroblast accumulation, and the alveoli are lined with cuboidal epithelial cells.
Transgenic Model ofPulmonary Fibrosis
Miyazaki et al.
Figure 7. Determination of hydroxyproline content(left)
trapping after injection of 51Cr-labeled platelets (right),
matched normal and transgenic mice.
plex pulmonary lesions indicative of abnormal lung develop-
ment. In all other mice bearing the transgene, pulmonary disease
was observed, whose difference in rapidity ofevolution, ranging
from death within afew weeks from birth to progression through
over 6 mo, appeared to correlate well with the transgene expres-
sion, as judged by lung TNF-a mRNA levels.
The first conspicuous lesion was leukocytic alveolitis; the
dominant cells were lymphocytes, mostly of the T lineage, al-
though neutrophils and monocytes were also present. This lesion
is likely to result from endothelial changes known to be induced
by TNF-a, in particular from the expression by endothelial
cells of cell adhesion molecules instrumental in leukocyte local
infiltration (36). A progressive local lymphocytic infiltration has
also been observed in mice bearing a TNF-a transgene placed
under the control ofan insulin promoter; this resulted in massive
insulitis, apparently secondary to venular endothelial changes
in the pancreatic islets (39). In the SP-C-TNF-a transgenic mice,
the increased lung expression of VCAM-1 mRNA, in the ab-
sence of variations in E-selectin or ICAM-1 mRNAs, suggests
that VCAM-1, which acts as a ligand for the integrin VLA-4
borne by T lymphocytes, may play a role in alveolitis. It is
unlikely that the lymphocytic infiltration, which did not show
signs of lymphocyte activation as judged by electron micros-
copy, played a significant role in subsequent alterations of the
alveolar walls, in particular fibrosis. In the TNF-a transgenic
mice with pancreatic insulitis, some degree of islet sclerosis was
also observed, and it was clearly independent from lymphocytic
infiltration, since it was ofthe same intensity when the transgene
was expressed in scid mice, which have neitherT norB lympho-
cytes and displayed little leukocytic insulitis (39). It is neverthe-
less clear that, in the presence of stimulating local antigens,
lymphocyte activation can play an important role in the evolu-
tion of lesions; for instance, in transgenic mice with TNF-a-
induced insulitis, local expression of the T lymphocyte costimu-
lating molecule B7 (directed by a second transgene) led to
lymphocyte activation, which resulted in islet beta cell destruc-
tion and lethal diabetes (40). It is reasonable to think that
lymphocytic alveolitis accompanied by local immune lympho-
cytic stimulation, as appears to be present in a number ofhuman
interstitial alveolitis, may, similarly, lead to more rapidly pro-
gressive pulmonary damage. Neutrophils may also be involved
in the progressive alveolar damage observed in the SP-C TNF-
a transgenic mice; while these cells were less conspicuous than
lymphocytes in the leukocytic infiltrates, their shorter life span
mightaccount for their lack of accumulation.Septal ruptures
compatible with polymorphonuclear elastase release were ob-
served in early lesions (Fig. 4, a and b); the iron deposits found
in such lesions (Fig. 3 c) probably resulted from focal alveolar
Two features indicated progressive epithelial damage. One
was the hyperplasia of type II alveolar epithelial cells, often
conspicuous in the lining of the distorted alveolar or residual
air spaces, a finding very suggestive of epithelial regeneration.
Another conspicuous lesion was intraalveolar epithelial desqua-
mation, taking usually the appearance of intraalveolar aggre-
gates of foamy cells, sometimes containing cholesterol crystals
(Fig. 4 e). While it is possible that alveolar macrophages also
contributed to this lesion, its ultrastructural appearance, with
lamellar bodies and focal intercellular adherence areas, indi-
cated that it was most often made of epithelial cells.
Progressive accumulation of fibroblasts in the alveolar wall
could be best evaluated by immunohistochemical staining for
desmin, since pulmonary fibroblasts are characterized by their
high desmin content (41). Fibroblasts were obviously responsi-
ble for the deposition of collagen and progressive destruction
of alveolar structures which culminated in extensive fibrosis.
While TNF-a has been observed to induce fibroblast prolifera-
tion in vitro (42) and in vivo (6), other cytokines may be indi-
rectly involved. TGF-/3 or PDGF-B has been incriminated as
playing an important role in human or experimental pulmonary
and of platelet
using lungs of
Figure 6. Electron micrographs of pulmonary alveolar wall lesions observed in the lungs of 4-6-mo-old mice. (a) The alveolar space on the left
is lined by regenerating type II alveolar cells. In the thickened septum, two lymphocytes (L), large interstitial cells, probably fibroblastic (F), and
a distorted capillary (C) with swollen endothelium are seen. (b) A thickened alveolar wall contains lymphocytes (L) and interstitial cells, not clearly
identifiable, an accumulation of collagen fibrils (CF), swollen endothelial cell probably of a neo-capillary (E) not yet surrounded by a well-formed
basement membrane, and several distorted capillaries containing red blood cells. (c) A severely damaged junction of alveolar spaces with folded
basal lamina suggestive of prior alveolar collapses. A dense accumulation of collagen fibrils (CF) is seen toward the center of the junction. (d) An
interalveolar capillary with a folded basement membrane contains a small platelet clump together with red blood cells. (e) The alveolar space on
the right of the picture, lined on its left border by alveolar type II cells, contains two desquamated large alveolar cells, which correspond to the
type of cells seen on Fig. 4 c.
Transgenic Model of Pulmonary Fibrosis
Figure 8. Immunofluorescent staining of diseased
lung forTGF-fi. On the left, three labeled cells, prob-
ably macrophages, are seen in a very thickened alveo-
lar wall, x80; such a concentration of TGF-fi-con-
taining cells is not representative. On the right, an
alveolar macrophage contains TGF-13, x250.
diseases (2-5, 43). In the SP-C-TNF-a transgenic mice, no
significant increase in lung levels of TGF-l3 and PDGF-B
mRNAs was observed at various stages of the lesions, and the
only accumulation ofTGF-,/ detected by immunochemistry was
in isolated macrophages (Fig. 8); activated alveolar macro-
phages are known to synthesize TGF-a (44). It cannot be con-
cluded, however, that these cytokines were not involved in pul-
monary fibrosis. Clumps of blood platelets, a major source of
both of these cytokines, were commonly detected by electron
microscopy in altered alveolar capillaries, and enhanced platelet
trapping in the lung was observed after transfer of labeled plate-
lets. This may have resulted from a combination of endothelial
alterations and enhanced expression of platelet adhesion mole-
cules induced by the local presence of TNF-a (45, 46). Experi-
mental evidence suggests that increased platelet trapping in the
lung is correlated with the development of other forms of lung
fibrosis in which TNF-a is involved (46). It must be emphasized
that prolonged release of "latent" TGF-/3 in the lung is not
necessarily harmful; we have generated several mouse SP-C-
transgenic lines in which the latent uncleaved form of
TGF-,3 is released in the lung without generating lesions (Miya-
zaki, Y., J. Whitsett, P.-F. Piguet, and P. Vassalli, unpublished
observations), suggesting that local additional production of
proteolytic enzymes able to process this cytokine in its active
form is essential for the induction of lesions.
The most striking conclusion resulting from the study of
the SP-C-TNF-a transgenic mice is that of the resemblance of
all steps of this progressive lung disease with those of idiopathic
pulmonary fibrosis or "fibrosing alveolitis" in humans, a dis-
ease of unknown etiology and pathogenesis. It is thus of special
interest to note that the presence of TNF-a mRNA and protein
has been detected recently in alveolar epithelial cells, mainly
type II pneumocytes, in this last disease (7, 8). While the patho-
genic mechanisms responsible for TNF-a overexpression by
alveolar epithelial cells in human pulmonary fibrosis are not
understood, the present findings with mice similarly overex-
pressing TNF-a in this localization suggest that this functional
disturbance may represent an initial event from which all subse-
quent lung pathology, including leukocytic infiltration and pro-
gressive fibrosis, might result. In this respect, the SP-C-TNF-a
transgenic mice represent an excellent experimental model of
pulmonary fibrosis, in particular to explore a variety of thera-
peutic approaches, directed at neutralizing the effects of TNF-
a or at decreasing its synthesis (47), or at preventing collagen
We want to thank Ms. J. Ntah for secretarial work, Mr. J.-C. Rumbeli
and Mr. E. Denkinger for photographic work, and Pedro Herrera for
performing the immunofluorescent staining forTGF-fl.
This work was supported by grants from the Swiss National Founda-
tion (31-37516.93) and by the Cystic Fibrosis Foundation (HL 51838),
Center for Gene Therapy.
1. Colby, T. V., and A. C. Churg. 1986. Patterns ofpulmonary fibrosis. Pathol.
2. Martinet, Y., W. N. Rom, G. R. Grotendorst, G. R. Martin, and R. G.
Crystal. 1987. Exaggerated spontaneous release of platelet-derived growth factor
by alveolar macrophages from patients with idiopathic pulmonary fibrosis. N.
Engl. J. Med. 317:202-209.
3. Antoniades, H. N., M. A. Bravo, R. E. Avila, T. Galanopoulos, J. Neville-
Golden, M. Maxwell, and M. Selman. 1990. Platelet-derived growth factor in
idiopathic pulmonary fibrosis. J. Clin. Invest. 86:1055-1064.
4. Vignaud, J. M., M. Allam, N. Martinet, M. Pech, F. Plenat, and Y. Martinet.
1991. Presence of platelet-derived growth factor in normal and fibrotic lung is
specifically associated with interstitial macrophages, while both interstitial macro-
phages and alveolar epithelial cells express the c-sis proto-oncogene. Am. J.
Respir. Cell Mol. Biol. 5:531-538.
5. Khalil, N., R. N. O'Connor, H. W. Unruh, P. W. Warren, K. C. Flanders,
A. Kemp, 0. H. Bereznay, and A. H. Greenberg. 1991. Increased production and
immunohistochemical localization of transforming growth factor-fB in idiopathic
pulmonary fibrosis. Am. J. Respir. Cell Mol. Biol. 5:155-162.
6. Piguet, P. F., G. E. Grau, and P. Vassalli. 1990. Subcutaneous perfusion
of tumor necrosis factor induces local proliferation of fibroblasts, capillaries, and
epidermal cells, or massive tissue necrosis. Am. J. Pathol. 136:103-110.
7. Piguet, P. F., C. Ribaux, V. Karpuz, G. E. Grau, and Y. Kapanci. 1993.
Expression and localization of tumor necrosis factor-a and its mRNA in idiopathic
pulmonary fibrosis. Am. J. Pathol. 143:651-655.
8. Nash, J. R. G., P. J. McLaughlin, D. Butcher, and B. Corrin. 1993. Expres-
sion of tumour necrosis factor-a in cryptogenic fibrosing alveolitis. Histopathol-
9. Piguet, P. F., M. A. Collart, G. E. Grau, Y. Kapanci, and P. Vassalli.
1989. Tumor necrosis factor cachectin plays a key role in bleomycin-induced
pneumopathy and fibrosis. J. Exp. Med. 170:655-663.
10. Piguet, P. F., M. A. Collart, G. E. Grau, A. P. Sappino, and P. Vassalli.
1990. Requirement of tumor necrosis factor for development of silica-induced
pulmonary fibrosis. Nature (Lond.). 344:245-247.
11. Piguet, P. F., and C. Vesin. 1994. Treatment by human recombinant
soluble TNF receptor of pulmonary fibrosis induced by bleomycin or silica in
mice. Eur. Respir. J. 7:515-518.
12. Nelson, S., G. J. Bagby, B. G. Bainton, L. A. Wilson, J. J. Thompson,
and W. R. Summer. 1989. Compartmentalization of intraalveolar and systemic
Miyazaki et al.
lipopolysaccharide-induced tumor necrosis factor and the pulmonary inflammatory
response. J. Infect. Dis. 159:189-194.
13. Johansson, J., T. Curstedt, and B. Robertson. 1994. The proteins of the
surfactant system. Eur. Respir. J. 7:372-391.
14. Korfhagen, T. R., S. W. Glasser, S. E. Wert, M. D. Bruno, C. C. Daugherty,
J. D. McNeish, J. L. Stock, S. S. Potter, and J. A. Whitsett. 1990. Cis-acting
sequences from a human surfactant protein gene confer pulmonary-specific gene
expression in transgenic mice. Proc. Natd. Acad. Sci. USA. 87:6122-6126.
15. Glasser, S. W., T. R. Korfhagen, C. M. Perme, T. J. Pilot-Matias, S. E.
Kister, and J. A. Whitsett. 1988. Two SP-C genes encoding human pulmonary
surfactant proteolipid. J. Biol. Chem. 263:10326-10331.
16. Semon, D., E. Kawashima, C. V. Jongeneel, A. N. Shakhov, and S. A.
Nedospasov. 1987. Nucleotide sequence of the murine TNF locus, including the
TNF-a (tumor necrosis factor) and TNF-/3 (lymphotoxin) genes. Nucleic Acids
17. Niwa, H., K. I. Yamamura, and J. I. Miyazaki. 1991. Efficient selection
for high-expression transfectants with a novel eukaryotic vector. Gene (Amst.).
18. Collart, M. A., D. Belin, J. D. Vassalli, S. de Kossodo, and P. Vassalli.
1986. Gamma interferon enhances macrophage transcription of the tumor necrosis
factor/cachectin, interleukin-1, and urokinase genes which are controlled by short
lived repressor. J. Exp. Med. 163:2113-2118.
19. Thorens, B., J. J. Mermod, and P. Vassalli. 1987. Phagocytosis and in-
flammatory stimuli induce GM-CSF mRNA in macrophages through posttran-
scriptional regulation. Cell. 48:671-679.
20. Wong-Staal, F., R. Dalla Favera, E. P. Gelmann, V. Manzari, S. Szala,
S. F. Josephs, and R. C. Gallo. 1994. The v-sis transforming gene of simian
sarcoma virus is a new oncogene of primate origin. Nature (Lond.). 294:273-
21. Derynck, R. 1986. Transforming growth factor-a: structure and biological
activities. J. Cell. Biochem. 32:293-304.
22. Derynck, R., J. A. Jarrett, E. Y. Chen, and D. V. Goeddel. 1986. The
murine transforming growth factor-f3 precursor. J. Biol. Chem. 261:4377-4379.
23. Becker-Andre, M., R. H. Van Huijsduijnen, C. Losberger, J. Whelan, and
J. F. DeLamarter. 1992. Murine endothelial leukocyte-adhesion molecule 1 is a
close structural and functional homologue of the human protein. Eur. J. Biochem.
24. Janssen, Y. M. W., B. Van Houten, P. J. A. Borm, and B. T. Mossman.
1993. Biology of disease. Cell and tissue responses to oxidative damage. Lab.
25. Horley, K. J., C. Carpenito, B. Baker, and F. Takei. 1989. Molecular
cloning of murine intercellular adhesion molecule (ICAM-1). EMBO (Eur. Mol.
Biol. Organ.) J. 8:2889-2896.
26. Pardue, M. L. 1994. In situ hybridization. In Nucleic Acid Hybridization:
A Practical Approach. B. D. Hames and S. J. Higgins, editors. IRL Press, Oxford.
27. Sappino, A. P., J. Huarte, D. Belin, and J. D. Vassalli. 1989. Plasminogen
activators in tissue remodeling and invasion. Messenger RNA localization in
mouse ovaries and implanting embryos. J. Cell Biol. 109:2471-2479.
28. Thrall, R. S., J. R. McCormick, R. M. Jack, R. A. McReynolds, and P. A.
Ward. 1979. Bleomycin-induced pulmonary fibrosis in the rat. Am. J. Pathol.
29. Ganter, P., and G. Jolles. 1969. Histochimie Normale et Pathologique.
Gauthier-Villars, Paris. 1632-1633.
30. Skalli, O., P. Ropraz, A. Trzeciak, G. Benzonana, D. Gillessen, and G.
Gabbiani. 1986. A monoclonal antibody against a-smooth muscle actin. A new
probe for smooth muscle differentiation. J. Cell Biol. 103:2787-2796.
31. Kapanci, Y., C. Ribaux, C. Chaponnier, and G. Gabbiani. 1992. Cytoskele-
tal features of alveolar myofibroblasts and pericytes in normal human and rat
lungs. J. Histochemn CytochemL 40:1955-1963.
32. Flanders, K. C., N. L. Thompson, D. S. Cissel, E. Van Obberghen-Schil-
ling, C. C. Baker, M. E. Kass, L. R. Ellingsworth, A. B. Roberts, and M. B.
Sporn. 1989. Transforming growth factor-,l6: histochemical localization with anti-
bodies to different epitopes. J. Cell Biol. 108:653-660.
33. ICHS Panel. 1988. Recommended method for Indium-111 platelet sur-
vival. J. Nucl. Me& 29:564-566.
34. Araki, K., K. Akagi, J. Miyazaki, K. Matsubara, and K. Yamamura. 1990.
Correlation of tissue-specific methylation with gene inactivity in hepatitis B virus
transgenic mice. Jpn. J. Cancer Res. 81:1265-1271.
35. Vyalov, S. L., G. Gabbiani, and Y. Kapanci. 1993. Rat alveolar myofi-
broblasts acquire a-smooth muscle actin expression during bleomycin-induced
pulmonary fibrosis. Am. J. Pathol. 143:1754-1765.
36. Bevilacqua, M. P. 1993. Endothelial-leukocyte adhesion molecules. Annu.
Rev. Inmmunol. 11:767-804.
37. Keffer, J., L. Probert, H. Cazlaris, S. Georgopoulos, E. Kaslaris, D. Ki-
oussis, and G. Kollias. 1991. Transgenic mice expressing human tumour necrosis
factor: a predictive genetic model of arthritis. EMBO (Eur. Mol. Biol. Organ.) J.
38. Kruys, V., K. Kemmer, A. Shakhov, V. Jongeneel, and B. Beutler. 1992.
Constitutive activity of the tumor necrosis factor promoter is canceled by the 3'
untranslated region in nonmacrophage cell lines; a trans-dominant factor over-
comes this suppressive effect. Proc. Natl. Acad. Sci. USA. 89:673-677.
39. Higuchi, Y., P. Herrera, P. Muniesa, J. Huarte, D. Belin, P. Ohashi, P.
Aichele, L. Orci, J. D. Vassalli, and P. Vassalli. 1992. Expression of a tumor
necrosis factor a transgene in murine pancreatic P cells results in severe and
permanent insulitis without evolution towards diabetes. J. Exp. Med. 176:1719-
40. Herrera, P. L., D. M. Harlan, L. Fossati, S. Izui, J. Huarte, L. Orci, J. D.
Vassalli, and P. Vassalli. 1994. A CD8+ T-lymphocyte-mediated and CD4+ T-
lymphocyte-independent autoimmune diabetes of early onset in transgenic mice.
41. Zhang, K., M. D. Rekhter, D. Gordon, and S. H. Phan. 1994. Myofi-
broblasts and their role in lung collagen gene expression during pulmonary fibro-
sis. Am. J. Pathol. 145:114-125.
42. Sugarman, B. J., B. B. Aggarwal, P. E. Hass, I. S. Figari, M. A. Palladino,
and M. H. Shepard. 1985. Recombinant human tumor necrosis factor-alpha: ef-
fects on proliferation of normal and transformed cells in vitro. Science (Wash.
43. Broekelmann, T., A. H. Limper, T. V. Colby, and J. A. McDonald. 1991.
Transforming growth factor /31 is present at sites of extracellular matrix gene
expression in human pulmonary fibrosis. Proc. Natl. Acad. Sci. USA. 88:6642-
44. Madtes, D. K., E. W. Raines, K. S. Sakariassen, R. K. Assoian, M. B.
Spern, G. I. Bell, and R. Ross. 1988. Induction of transforming growth factor-a
in activated human alveolar macrophages. Cell. 53:285-293.
45. Piguet, P. F., C. Vesin, J. E. Ryser, G. Senaldi, G. E. Grau, and F.
Tacchini-Cottier. 1993. An effector role for platelets in systemic and local lipo-
polysaccharide-induced toxicity in mice, mediated by a CD1 ia- and CD54-depen-
dent interaction with endothelium. Infect. Immun. 61:4182-4187.
46. Piguet, P. F., and C. Vesin. 1994. Pulmonary platelet trapping induced by
bleomycin: correlation with fibrosis and involvement of the /32 integrins. Int. J.
Exp. Pathol. 75:321-328.
47. Lee, J. C., J. T. Laydon, P. C. McDonnell, T. F. Gallagher, S. Kumar, D.
Green, D. McNulty, M. J. Blumenthal, J. R. Heys, S. W. Landvatter, et al. 1994.
A protein kinase involved in the regulation ofinflammatory cytokine biosynthesis.
Nature (Lond.). 372:739-746.
Transgenic Model ofPulmonary Fibrosis