Multimodality Approach to Characterizing the Distinctive Hallmarks of Lung Fibrosis: A Mouse Model with Bleomycin and Indocyanine Green

Background. Idiopathic Pulmonary Fibrosis is a progressive disease with short life expectancy and no disease-modifying pharmacological therapy. The continuous renement of animal models and the integration of in-vivo imaging techniques is fundamental for the selection of new antibrotic drugs. Indocyanine Green (ICG), a uorescent dye, was administered by oropharyngeal aspiration (OA) to mice with Bleomycin (BLM) to map the lung exposure. Methods.


Background
Idiopathic Pulmonary Fibrosis (IPF) is a heterogenous, chronic disease characterized by progressive lung scarring and impaired pulmonary function, ultimately leading to respiratory failure and death. A resolutive pharmacological therapy has yet to be identi ed, with lung transplantation often remaining the only treatment option [1,2]. Even though animal models of IPF are crucial to satisfy this urgent unmet medical need, none of the currently available models can fully replicate the human disease [3,4]. Among the animal models of IPF, administration of the chemotherapeutic agent bleomycin (BLM) in mice is the most widely used [5]. However, this model presents some drawbacks, which are challenging to predict and standardize, such as spontaneous resolution of brosis, variability and mortality. Nonetheless, preclinical data are needed for selection of the best drug candidate for clinical trial and then clinical use, thus the continuous re nement of animal models is fundamental, as is the integration of in-vivo imaging techniques in preclinical studies [6].
Imaging technologies play an important role in decreasing intra-experimental variability, and, as a result, signi cantly reduce the number of animals used per experiment fully in compliance with the 3R's principles [13]. Notably, thoracic imaging represents a surrogate outcome in clinical trials and clinical practice; therefore, the development of preclinical imaging model is expected to integrate well with optimized drug development.
In the present work, Indocyanine Green (ICG), a water-soluble uorescent dye, was mixed with BLM and the solution was administered by oropharyngeal aspiration (OA) to mice.
ICG has been used as tracer to elucidate the distribution and exposure of BLM in the lungs, in light of its common use for clinical and preclinical applications [7][8][9].
We sought to analyze a new model of lung brosis in mice based on a single OA and with multimodality approach for characterization of the disease progression.
Prior to use, animals were acclimatized for at least 7 days to the local vivarium conditions (room temperature: 20-24°C; relative humidity: 40-70%; 12-h light-dark cycle), having free access to regular rodent chow and softened tap water. Sterile sun ower seeds and hydrogel were supplied as diet integration to prevent excessive body weight loss. All animal experiments described below were approved by the intramural animal-welfare committee for animal experimentation of Chiesi Farmaceutici under protocol number: 809/2020-PR and comply with the European Directive 2010/63 UE, Italian D.Lgs 26/2014 and with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines [10]. A Visual Analogue scale (0-10) for pain assessment was assessed daily by a designated veterinarian or trained technicians. VAS ≥7 and/or body weight loss ≥ 20% were considered as humane endpoints, as well as signs of dyspnoea or apathy evaluated by a designated veterinarian.
Administration of either solution (day 0) was performed after mice were slightly anesthetized with 2.5% iso urane (IsoFlo, Zoetis Inc., New Jersey, USA) delivered in a box. The animals were placed on an intubation platform, positioning their incisor on the wire, the tongue was then pulled out with forceps using a small laryngoscope and 50 µL of solution was administered via oropharyngeal aspiration (OA) using a micropipette [11].
The operator kept the laryngoscope in place for 5-10 seconds to ascertain the correct and complete aspiration of the instilled liquid. Mice were held upright for 10-15 seconds before being placed back into the cage.
Three independent experiments were carried out. The whole experimental procedure is summarized in supplementary gure 1.

Fluorescence imaging
Animals were slightly anesthetized with 2% iso urane, shaved and imaged using IVIS Lumina II (PerkinElmer Inc., Waltham, MA) [12]. Mice were imaged in prone position at 0, 7, 14 and 21 days after OA treatment; an average of total uorescence signal emitted from the chest region was calculated for each mouse.

Micro-Computed Tomography:
Acquisition protocol and image post-processing analysis. Micro-Computed tomography (Micro-CT) was performed at day 7, 14 and 21 using a Quantum GX Micro-CT (PerkinElmer, Inc. Waltham, MA). Mice were slightly anesthetized with 2% iso urane and images were acquired with the following parameters: 90KV, 88µA, total scan time of 4 minutes (over a total angle of 360°). The 'high speed' acquisition protocol was used, and a respiratory gated technique was applied. The entire set of projection radiographs was reconstructed using a ltered back-projection algorithm with a Ram-Lak lter and resulted in two stacks of 512 slices with a nominal resolution of 50 µm.
The reconstructed datasets were analysed with Analyze software (Analyze 12.0; Copyright 1986-2017, Biomedical Imaging Resource, Mayo Clinic, Rochester, MN). Following the analysis protocol, CT scans were ltered and then converted from grey levels to CT numbers (Houns eld Units-HU).
For the quantitative assessment of the aeration degrees, the total lung volume was extracted from the reconstructed image [11,13] through manual segmentation. the lung aeration compartments were then determined applying 'HU preclinical ranges' [13]  -860] HU), expressed as percentage of total lung volume.
The hypo-and non-aerated lung tissues areas refer to those with a low gas/tissue ratio, which was previously developed to quantify lung brosis progression and evaluate anti brotic drug e cacy [13].
Micro CT Visual categorisation.
Micro-CT scans were visually evaluated by three independent radiologists: four patterns were de ned to categorize the lung parenchyma features (see supplementary gure 2). The nal decisions reached by consensus.
Healthy lung was de ned by homogenous density of lung parenchyma with regular-shaped blood vessels and airways. The reconstructed datasets revealed an entirely detectable lung volume, with normally aerated areas covering around 80% of total lung volume.
Typical BLM brosis (TBF) was de ned by a patchy pattern with high attenuation areas randomly distributed within the lung parenchyma without visible low attenuation areas [11,13,16].
Fibrotic lung associated with moderate airspace enlargement (FAE) was de ned by a patchy pattern with heterogeneous density from hazy brotic pattern with moderate presence of low attenuation areas, which are di cult to isolate from the surrounding tissue.
Fibrotic lung with severe airspace enlargement (FAE + ) was de ned by a patchy pattern with heterogeneous density into an overt mosaic pattern characterized by multiple and markedly detectable low-density structures, with well-de ned margins appearing in a bubble-like shape in lung parenchyma.
Each Micro-CT scan was attributed to one of these categories at days 7, 14 and 21 and the number of animals assigned to each category was calculated at every time point.
Histological assessment of brosis At 7, 14 and 21 days, after the in-vivo imaging, subsets of 5 (Sal+ICG) and 12 (BLM+ICG) mice were euthanized with an overdose of anaesthetic followed by bleeding from the abdominal aorta. The lungs were removed and in ated with a cannula through the trachea by gentle infusion of 0.6 mL of 10% neutral-buffered formalin and xed for 24h. For histological assessment, the samples were dehydrated in a graded ethanol series, clari ed in xylene, and para n embedded. Sections of 5 μm thickness were cut with a rotary microtome (Slee Cut 6062; Slee Medical, Mainz, Germany) and then stained with hematoxylin and eosin (H&E) and Masson's trichrome (TM), according to the manufacturer's speci cations (Histo-Line Laboratories). The whole slide images (WSI) were acquired by the NanoZoomer S-60 Digital slide scanner (Hamamatsu, Japan) for analysis. Three sections for each lung sample were stained with TM and scored on a scale of 0 to 8 by two independent investigators who were blinded to the treatments. Fibrotic modi cations were assessed morphologically and semi-quantitatively graded according to the scale de ned by Ashcroft et al. [14] and modi ed by Hübner et al [15]. The nal score was expressed as a mean of individual scores observed across all microscopic elds. To quantify the distribution of pulmonary brosis, the Ashcroft scores were graded in three classes of increasing values ranging from 0 to 3 (mild), 4 (moderate), and ≥5 (severe).
Histo-morphometric assessment of airspace enlargements

Fluorescence imaging and prolonged ICG retention in lungs
In-vivo uorescence of Sal+ICG and BLM+ICG mice in prone position and ex-vivo lung signals at different time points are shown in Figure 1A and B. The images showed a uorescent signal well localized in the chest region in all mice. The quanti cation of uorescent signal did not show any statistically signi cant difference between Sal+ICG and BLM+ICG treated mice at any time ( Figure 1C and D).
The signal was present at the time of OA and through the pre-de ned time points, in particular: uorescence peaked at day 7 and gradually decreased at later time points, either in-vivo or ex-vivo. At day 21, a conspicuous ICG signal was still detected (7 and 9 Log radiant e ciency, respectively) ( Figure 1C, D).
These results clearly indicate a remarkable persistence of ICG in the lungs when the dye is administered alone or in association with bleomycin. The in-vivo imaging consolidated two hypotheses: rst, ICG can be used as a tracer to map the distribution and exposure of BLM in lungs, and second, OA is con rmed as a suitable route of administration to obtain uniform lung distribution of the instilled solution [11].

Micro-CT analysis
The representative Micro-CT imaging of 3D lung segmentation at different time points (Figure 2A) showed hypo-and non-aerated areas, associated with appearance of hyper-in ated lung tissue in BLM+ICG mice, whilst Sal+ICG treated animals had normal lung parenchyma throughout the time course.
Hypo-and non-aerated tissues were used to classify the lung compartments with different gas/tissue ratio corresponding to moderate or severe brosis.
The time course quanti cation of lung aeration degrees ( Figure 2B) showed a signi cant increase in hypo-and non-aerated tissues already at day 7 with peak for both compartments at day 14.
At day 21, a signi cant increase (p<0.05) of normo-and a reduction of hypo-aerated tissues were observed compared to 14 day, whilst no signi cant variations occurred regarding non-aerated areas.
Notably, a signi cant amount of hyper-in ated tissue was detected along the time course of the experiment, highest at day 21.

Micro-CT scans visual categorization
Sal+ICG treated mice showed a healthy lung parenchyma at each time point of observation.
BLM+ICG mice were classi ed as TBF, FAE and FAE + and lung aeration degrees were quanti ed ( Figure  4C), to compare different categories regardless of the time-point.
The FAE, with respect to TBF class, showed a markedly higher (p<0.01) hypo-and non-aerated tissue, as well as reduced normo-aerated areas, whilst the hyper-in ated tissue was augmented although not statistically signi cantly. FAE + category, as compared with TBF, exhibited a signi cantly lower normoaerated tissue (p<0.01), whereas hypo-and non-aerated areas were not signi cantly different, although slightly higher than TBF. Finally, a signi cant increase in hyper-in ated tissue (p<0.05) was observed in FAE + with respect to TBF.

Histological assessment of lung brosis
Histological analysis was performed on subsets of Sal+ICG and BLM+ICG mice sacri ced at 7, 14 or 21 days. BLM+ICG histology exhibited a patchy pattern of severe lung brosis, characterized by an increasing deposition and compactness of the substitutive extracellular matrix within the brotic foci. This pattern was already well-established at day 7, worsened at day 14 and remained stable up until day 21 ( Figure 3A). Moreover, the destruction and collapse of alveolar walls with alveolar airspaces enlargement, already observed at day 7, tended to evolve at 14 and 21 days, along with widening of the balloon-like airspace resembling emphysema-like morphology. (Figure 3A). The Sal+ICG group instead, showed normal lung architecture throughout the study ( Figure 3A). Ashcroft score revealed a severe lung brosis and was signi cantly increased (p<0.001) at all time points, although slightly higher at day 14, in BLM+ICG compared to Sal+ICG treated mice ( Figure 3B). Ashcroft score Frequency distribution grouped as mild (0-3), moderate (4), and severe (≥5) signi cantly changed (p<0.05) as compared to Sal+ICG, revealing a prevalence of severe brosis at each time point, with a peak at 14 days, while the moderate values remained stable for BLM+ICG mice during time ( Figure 3C).
Representative micrographic images of Sal+ICG and BLM+ICG groups at different time are shown in gure 5.
The BLM+ICG group showed signi cantly more large airspaces and less medium airspaces at day 14 (p<0.05), compared to Sal+ICG group. This pattern of airspace abnormalities was even more pronounced at day 21 (p<0.05).
These data were con rmed by MLI, which showed signi cant structural changes at day 21 in the BLM+ICG group ( Figure 5C).
Changes in AAA and MLI described disease progression resembling an emphysema-like pattern, in agreement with the outcomes arising from Micro-CT analysis, which were characterized by multiple and markedly detectable low-density structures, appearing in a bubble-like shape in the lung parenchyma.

Discussion
One of the novelties provided by this study is the introduction of a simple in-vivo method to follow lung distribution of BLM. This approach can help reduce variability within and between experimental groups in preclinical studies on lung brosis, leading to the reduction of sample size, thus improving compliance with the 3R's principles.
This study showed that the mouse model with OA of BLM + ICG grants homogeneous distribution of compounds to lung parenchyma, as shown by in-vivo and ex-vivo uorescence signals. The solution of BLM and ICG re ected one of a kind pulmonary damage with long-term evolution of hypoattenuating lung areas on micro-CT and histological evidence of severely enlarged airspaces in association with severe brosis, different from TBF.
IPF represents an urgent unmet medical need. Thus, it is imperative to develop disease-modifying therapies, for which purpose re ned animal models are crucial. The rst objective of the present study was to set up a method to monitor the distribution of BLM after OA through uorescence imaging. Indeed, although performing OA requires lower technical skills with respect to intratracheal instillation, variability in terms of lung exposure remains a common issue, leading to biased outcomes and increasing the number of animals required to produce statistically signi cant results.
Administration via OA generated homogeneous lung distribution as revealed by the quanti cation of invivo and ex-vivo uorescence signals (Fig. 1), con rming that ICG can be used as a tracer to early identify mice that for technical reasons would exhibit a lower or non-uniform lung exposure.
However, Micro-CT and histological analyses revealed a more severe lung brosis compared to our classical BLM model, even though a single rather than a double administration via OA was applied [11,13,16].
A severe lung brosis appeared as soon as day 7, peaked at 14 days and remained stable up to day 21. In fact, a manual segmentation of micro-CT images was necessary for the BLM + ICG mice to detect nonaerated areas ( Fig. 2A, B), in line with the Ashcroft frequency distribution (Fig. 3B) revealing the prevalence of severe lesions at every time point. The coupling of the destruction and collapse of the alveolar walls with alveolar airspaces enlargement evolved in a severe brotic lesion interconnected with emphysema-like morphology, characterized by balloon-like appearance of parenchyma and producing a resemblance to honeycombing (Fig. 3A).
Other distinctive features highlighted by Micro-CT, compared to typical BLM brosis, were low density areas with a bubbles-like shape of hyper-in ated tissue (Fig. 4A, supplementary Fig. 2) which appeared at days 7 and 14 and increased at day 21. These lung tissue structures, resembling emphysema-like features, were categorized as TBF, FAE and FAE + through a visual Micro-CT scans classi cation at the different time-points (Fig. 4B). These features have not been reported in our previous studies on BLMinduced brosis [11,16], representing distinctive hallmarks of a BLM + ICG induced lung disease.
The categories identi ed by visual score were also signi cantly different in terms of lung aeration degrees (Fig. 4C), as mice classi ed as FAE + showed a 3-fold increase in hyper-in ated areas compared to other categories.
This evolution was consistent with substantial prevalence of long-term FAE + , under the newly developed BLM + ICG condition. In line with these ndings, histo-morphometric evaluation of airspace size revealed severe abnormalities at day 21 (Fig. 5A), as con rmed by the signi cant increase in AAA and MLI (Fig. 5B, C).
The progressive lung disease with severe brosis and emphysema-like features reported by both Micro-CT and histo-morphometric analyses cannot be attributed to the tracer itself, as animals receiving saline + ICG had healthy lungs. ICG is considered a safe and biologically inactive probe, as in over 40 years of clinical practice adverse reactions have been anecdotal [24]. Nonetheless, in our study the combination of BLM and ICG triggered a disease that was clearly different from the typical BLM-induced lung brosis with a single administration corresponding to half of the dose we routinely use [16.] ICG is known to have a very short half-life and to be rapidly eliminated by the liver [25] upon intravenous administration, which seems to be in contrast with our ndings, as we detected a remarkable uorescence signal in lungs up to 21 days (Fig. 1). However, there is robust evidence indicating that ICG strongly binds to proteins and phospholipids (PL), which deeply affect the uorescence yield and half-life [26,27] of the tracer. For this reason, in clinical practice, different ICG-based formulas in which the dye is combined with proteins have been used to improve the retention in target organs [28,29]. Lung surfactant being composed mainly of PL and proteins 30 we could hypothesize that interaction between ICG and surfactant's components led to the prolonged retention of the probe within the lungs, despite surfactant deprivation being postulated to occur in models of lung brosis [31].
We could hypothesize that the prolonged retention of ICG in the lungs exacerbated the pulmonary damage triggered by bleomycin. However, a thorough mechanistic explanation has yet to be clari ed and represents one of the key steps for future studies.

Conclusions
In the present work, we have described a model of pulmonary brosis with peculiar characteristics that may be added to the list of experimental tools to be exploited in drug discovery, to foster the development of more effective therapeutics.
In addition, a common issue in pharmacological studies based on the BLM model is that brosis starts to spontaneously resolve after 14 days, thus hindering the achievement of a proper time window to reveal the e cacy of potential new drugs. Notably, in this model the brosis develops early and does not seem to be resolving at 21 days, as well as being associated with progressive emphysema -like features, resembling human IPF [32,33].
Although, signi cant anatomical alterations were reported, namely emphysema, further investigations will be required to elucidate the mechanisms underlying the development of the disease.
We suppose that airspace size and hyper-in ated areas increase could be driven by a severe brosis exerting traction on lung parenchyma, together with lung functional alteration.
Functional measurements, such as functional Micro-CT analysis, could provide a more extensive comprehension of the disease progression.
However, further investigations will be required to elucidate the mechanisms underlying the development of brosis with enlarged airspaces elicited by BLM + ICG.
To sum up, we have highlighted the pivotal role of Micro-CT technology in animal models of lung disease, unveiling the hallmarks of disease progression by both visual and quantitative methods. We strongly believe that a continuous re nement of animal models together with an integrative approach with Micro-CT and other imaging technologies represents a potential complement for clinical translatability. Declarations Availability of data and materials.
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.     asterisks within the bars indicate signi cant differences from saline, while asterisks above the horizontal lines indicate signi cant differences between 7-14-21 days (**p < 0.01; *p < 0.05) (B). Alveolar air spaces evaluated by mean linear intercept (MLI), asterisks within the bars indicate signi cant differences from saline, while asterisks above the horizontal lines indicate signi cant differences between 7-14-21 days (**p < 0.01; *p < 0.05) (C).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.