Structural and functional correlations in a large animal model of bleomycin-induced pulmonary fibrosis

Despite recent advances in therapeutics available for IPF, there is a continued need
to find more effective therapies to directly prevent or ideally reverse the fibrosis
in the lungs. The development of novel therapies relies on how well experimental animal
models are able to translate findings to clinically related outcomes but there is
still much debate around the reliability of current animal models for pulmonary fibrosis.
While functional assessments are key end-points in clinical trials and disease staging
in IPF 24], only a few small animal models of pulmonary fibrosis incorporate long term lung
functional measurements into their protocols. Here, we present a model that offers
the ability to measure not only histological and biochemical changes in the lung,
but also incorporates long-term lung function monitoring throughout the experiment.
The segmental infusion approach used in the current model offers several key advantages
over conventional small animal models, which typically induce fibrosis to the whole
lung. Firstly, the model has the ability to temporally deliver infusions of bleomycin
to discrete lung segments. This not only reduces the number of animals used for an
experiment, but also improves statistical power by allowing for direct comparisons
in the same animal between an internal normal lung control segment and a lung segment
receiving an active agent. In addition, all stages of the disease, up to end-stage,
can be examined in a small specific lung segment, while the untreated regions of the
lung are healthy and can undertake normal respiratory functions of a healthy animal.
This is an important consideration, as high mortality rates approaching 20-50 %, and
significant initial weight loss are common issues experienced in murine bleomycin
models which can confound injury responses in the lungs 11], 12], 25].

It is important to note that, as with any animal model, there are limitations to the
sheep model that need to be considered. Foremost, in attempting to replicate a complex
and poorly understood disease such as IPF, it is nearly impossible to fully recapitulate
the exact the nature of the disease. Like other animal models of pulmonary fibrosis,
the sheep model induces fibrosis within a relatively short time over a number of weeks,
compared to the natural history of IPF where human patients take decades to develop
the disease 1]. Animal models also largely fail to create the progressive nature of fibrosis seen
in IPF where the pathology and lung function of IPF patients deteriorate relatively
rapidly after initial diagnosis 4], 26]. Over time, the pulmonary fibrotic responses in animal models typically either plateau,
or resolve, rather than progressively decline. Notwithstanding this, studies in animal
models are still an important tool to help further our improved understanding of the
disease and find suitable treatment and should be used synergistically to exploit
the individual advantages each has to offer. Therefore, the distinguishing features
described the sheep model should be used to complement other experimental systems
in contributing to the understanding of pulmonary fibrosis.

The inclusion of physiological changes provides an additional, more clinically relevant
end-point in our model to assess fibrotic changes throughout a study. Importantly,
this is performed in a species with a respiratory system that more closely resembles
the structure and function of the human lung 27], 28]. Lung function assessment in the current study shows that there was a significant
drop in segmental lung compliance in the bleomycin treated segments, which persisted
for seven weeks following bleomycin injury. Importantly, the lung function decline
was significantly correlated with pathology scores at 7 weeks post bleomycin. Moreover,
sheep that showed progressive declines in lung function over the study period were
found to have the most severe pathology, as indicated by the SMI scoring data. This
data is consistent with a number of early clinical studies, which demonstrate a correlation
between structural and functional changes in IPF lungs 29], 30]. In these studies, lung function measurements of FVC, gas exchange (dLCO) and compliance
were correlated to the extent of both fibrosis and cellularity in IPF patients 29], 30]. The association between deteriorating lung function and concomitant pathology has
been ultimately linked with the survivability of patients 26], 31]. The overall significance of our findings was that sequential lung function assessment
in this large animal model could serve as a reliable indication of the fibrotic changes
occurring in the lung.

To our knowledge, there are no other large animal models for pulmonary fibrosis that
examine histological and physiological changes in the lung. In small animal models,
there are only a limited number of bleomycin models that investigate changes to lung
function beyond the acute stages of injury. In the majority of previously published
animal studies which utilize bleomycin as the fibrotic agent and include lung function
measurements, bleomycin administration appears to have only had a minor impact on
pulmonary mechanics and is generally only observable in the acute stage of injury,
and not the later fibrotic stages 8], 9]. Indeed, very few bleomycin studies report physiological data as an end-point in
pre-clinical investigations, which is an obvious limitation for these models. An exception
is the recent study conducted in mice using a relatively high single dose of bleomycin
which was able to detect measurable differences in lung function for up to 6 months
after bleomycin administration 12]. In that study, while there were significant differences between control mice and
bleomycin-treated mice in lung function indices of diffusion factor of carbon monoxide
(DF
_co
), total lung capacity, and compliance at six months after bleomycin administration,
these parameters were all recovering towards normal at this time-point 12].

Interestingly, our lung function data is consistent with the TGF-? transgene model
of pulmonary fibrosis in rats, which has been able to demonstrate a similar reduction
in lung compliance for up to 7 weeks after the induction of fibrosis 32]. The degree of lung stiffness was also negatively correlated to pathology scores
in the TGF-? rat model as we show for the sheep model 32]. This model utilizes adenovirus carrying active TGF-?1, which causes a transient
overexpression of the protein to induce fibrosis 33]. This stimulates downstream pro-fibrotic responses, rather than continual TGF-?1
production, which result in the subsequent lung function decline. This also appears
to be the situation in the current study, whereby transient TGF-?1 expression appeared
to be sufficient to result in self-propagation of the fibrotic response that led to
a loss of lung function. Whilst both models showed similar physiological changes,
an asset of the sheep model we present here was that physiological data could be obtained
without anesthesia or sedation and assessed throughout the entire experimental period,
including the period just before cull. This allowed us to correlate compliance of
individual lung segments with the fibrosis pathology in these segments.

TGF-? is an important pro-fibrotic cytokine which plays a central role in fibrosis
34]. The release of TGF-? from injured epithelial cells and other sources results in
a fibrotic process with the accumulation of fibroblasts and myofibroblasts in the
injured area. These cell types also express TGF-? to further stimulate this wound
response 34], 35]. In the current study, we found that in TGF-? levels in BAL fluid were significantly
increased in all the bleomycin treated lobes 2 weeks after bleomycin treatment. While
at 6 weeks post treatment, the levels of TGF-? were much more variable, with some
sheep showing increased expression, some unchanged, and others decreased expression.
These levels did not correlate to other parameters, such as lung function, pathology
or ?SMA, expression (data not shown). As mentioned above, transient TGF-?1 expression
appears to act as an important stimulus for the profibrotic response and can lead
to the activation fibrogenic cytokines, including TNF-?, platelet-derived growth factor,
and basic fibroblast growth factor 33]. We have previously shown by immunohistochemistry that TGF-? labelling is particularly
strong in alveolar epithelial cells and hyperplastic Type II AEC cells, especially
in regions of more severe fibrosis 13]. These cell types, as well as others in the lung, may contribute to the release of
TGB- ? into lung luminal fluids. Interestingly, there are differences in the literature
regarding the duration of TGF-? expression in the BALF in bleomycin murine models.
Early studies have shown peak expression at 14 days, returning to normal by 21 days
after bleomycin 36], whilst others are found increasing levels up to this time point 37]. A more recent study suggests that peak levels are detectable early, i.e. 3 days
post bleomycin, in the BAL and gradually recede 38]. This variability in responses may be indicative that active TGF-? eventually becomes
bound to components of matrix such as the small proteoglycan biglycan 39], or fibronectin 40] and is active in these regions, and therefore is no longer detectable in the BAL
fluid.

In this study, we found that overall injury and fibrotic pathology, as assessed by
a modified Ashcroft’s scoring system, was similar in the bleomycin treated lobes at
7 weeks post injury to what was observed in our previous study at 2 weeks post bleomycin
13]. However, the inflammation scores detected at 7 weeks post-injury were significantly
lower than that reported at 2 weeks post-injury 13]. The persistence of fibrosis at the 7 week time point was supported by data from
both the hydroxyproline assay and Masson trichrome staining, which indicate an increase
in collagen deposition at this time point compared with controls. The timing of the
fibrotic and inflammation responses in the sheep bleomycin model is in accordance
with the respective timeframes typically exhibited in bleomycin rodent models, where
inflammation is observed up to 10–14 days post-bleomycin and then subsides to give
rise to the fibrotic response, generally between day 21–28 after bleomycin 6], 10].

The finding that fibrosis persists and the overall cellularity of the lesions decreases
over extended periods has been reported for bleomycin models in rodents that use both
multi-hit 7], 11] and single dose methods 10], 38], but not by others using rodent models where the fibrosis resolved over 6 weeks post-injury
8], 9]. Whilst our findings are comparable with the studies showing persistent fibrosis
7], 10], 11], 38], one defining point of difference between the rodent and sheep models is the pattern
of fibrosis observed in the sheep model. In general, intratracheal administration
of bleomycin in rodent models typically produces a peribronchiolar pattern of fibrosis
8], 41], whilst IV administration of bleomycin to mice induces perivascular fibrosis 10], 41], of which both patterns are more commonly observed other lung diseases, such as asthma
and COPD 23], 42]. In contrast, the patchy fibrosis observed in our study is found predominantly in
the lung parenchyma, which is consistent with the pattern of fibrosis reported for
IPF patients 1].

The sustained fibrotic response may reflect a prevailing profibrogenic environment
within the bleomycin treated lung segments of sheep, as suggested by the increased
positive ?SMA immuno-staining found within the injured region of the segments. ?SMA
is a recognized indicator for the presence of myofibroblasts 43], a cell type which is known to be extensively localized in fibrotic lesions of IPF
lungs 44]. It has been shown that the persistent presence of myofibroblasts appears to be a
primary driver of the progression of fibrosis in IPF 45]. In bleomycin-induced fibrosis, the increased presence of the myofibroblast was found
to occur in parallel to increased fibrotic injury 38]. The myofibroblast has been associated with the expression of several matrix proteins
which contribute to matrix deposition 46].

Radiological examination of the lung is the primary method used to determine the presence
of UIP and a diagnosis of IPF and is also used to help monitor the progression of
fibrosis and stage of severity of disease 1]. Software programs, such as the one used in the current study allow for 3D reconstruction
of the scanned area, which can then subsequently be measured and the specific density
range can be quantified. In the current study, we chose to assess the radiological
changes in the lungs of the sheep that had showed the worst lung function throughout
the study to confirm that injury in the lung occurred in the segment treated with
bleomycin. Additionally, this also enabled us to determine if this tool could be incorporated
into our assessment of fibrosis, as has been successfully done in murine models 10], 32] The scan was performed ex-vivo for practical reasons, as it eliminated the need to
anesthetize and transport the sheep for the procedure, similar to that previously
published in a pre-clinical mice model 10]. In CT images, fibrotic lung tissue appeared denser compared to normal lungs, which
was very clearly visualised on the images obtained from our sheep lung, showing a
clear demarcation of the region locally treated with bleomycin compared to the remainder
of the lung. This result suggested that ex-vivo CT scans could serve as a non-invasive
tool for the assessment of fibrotic changes and intervention strategies in future
studies, similar to that done by others 10], 32].