As IPF is a prevalent disease in males (Raghu et al., 2011; Cottin et al., 2013; Caminati et al., 2015), all experiments were conducted in C57Bl/6j male mice (n = 12, Janvier Labs, Le Genest-Saint-Isle, France). Animals were aged about 8 weeks and weighed 23.5 ± 1.0 g at the beginning of experiments. The animals were housed in standard cages with a 12 h/12 h lighting conditions and received a standard diet with ad libitum access to drinking water in our animal facility (agreement number C9300801). After 1 week of acclimatization, experimental procedures were initiated.

As recently published (Yegen et al., 2022), we have developed an original model of pulmonary fibrosis in C57Bl/6j mice, obtained by repeated intra-tracheal instillations of BLM (Sigma-Merck, Saint-Quentin-Fallavier, France) under general anesthesia (3% isoflurane). Briefly, 6 intra-tracheal instillations of low doses of BLM (0.8 UI. g^-1, n = 6) or PBS (n = 6) were performed with an interval of 2 weeks between two administrations (Figure 1A) (Boncoeur et al., 2022; Yegen et al., 2022). Survival and body weight were monitored during the entire experimental protocol (See Yegen et al., 2022 (Yegen et al., 2022) for more details). All the following analyses were carried out at day 90 (D90), 2 weeks after the last instillation (Figure 1A).

As previously described (Voituron et al., 2009; Niane et al., 2011; Samillan et al., 2013; Jeton et al., 2022), breathing variables were measured non-invasively in unanesthetized and unrestrained animals using whole-body flow barometric plethysmograph (Emka technologies, Paris, France). This method consists of measuring the pressure variations in a recording chamber during spontaneous ventilation. Briefly, mice were placed in a recording chamber ventilated with air at room temperature (21°C–22°C) and the ventilatory variables were analysed through a differential pressure transducer that measured the pressure difference between the recording chamber and a reference chamber. Thus, whole body plethysmography provides a measurement of respiratory frequency (f _R in cycle per min, c. min^-1), as well as tidal volume (V_T in µl) and minute ventilation ( V ˙ e in ml. min^-1), which were normalized by the mice body weight (V_T, μl g^-1 and V ˙ e, ml.g^-1. min^-1). The Ti/Ttot and V_T/Ti ratios were also determined to provide an indication of breathing time and an index of inspiratory drive (Milic-Emili and Grunstein, 1976).

After short sedation with isoflurane 3% during 2 min, animal neck was shaved and then a collar sensor was put in place. After waking, mice were habituated for 10 min and the peripheral blood saturation (SpO₂, %) was recorded during 10 min in normoxic condition through a non-invasive infrared pulse oximetry (mouseOx Plus, Starr Life Science) (Lax et al., 2014; Gille et al., 2018).

Pulmonary compliance was evaluated in mice using plethysmography (Emka technologies). Briefly, mice were sedated (ketamine/xylazine I.P. injection; 100 mg/kg and 20 mg/kg respectively), tracheotomised and ventilated (RoVent Jr., Kent Scientific Corporation; respiratory frequency = 150 bpm, Tidal Volume = 0.27 mL; Insp/Exp ratio = 0.40). A differential pressure transducer was used to obtain a flow signal, which reflects the expansion and contraction of the thorax during each ventilation cycle. Compliance and resistance were calculated following the acquisition of flow and pressure signals. Volume signal was obtained by integration of the flow signal measured by the differential pressure transducer. As the pneumotachograph was the only way for air to flow into and out of the chamber, the difference of pressure between inside and outside was proportional to this flow.

After measuring lung compliance, a thoracotomy was performed. Then, right pulmonary lobes were isolated through a ligation to the hilum to prevent the passage of the perfusion solution. Mice were then transcardially perfused with 0.9% saline-buffered solution followed by 4% paraformaldehyde (Chem Cruz, SC-253236B) in 0.1 M Phosphate-buffered saline solution (pH 7.4). Thus, the left lobes as well as the brain were fixed. After fixation, the brain and the heart/lung unit were removed. The right lung lobes were separated, frozen in liquid nitrogen and stored at −80°C for other studies. The left lobes were placed in 4% paraformaldehyde for 24 h and paraffin embedded. Sections were cut at 5 µm thickness for Sirius Red staining. Finally, the brain was post-fixed for 48 h at 4°C and then cryoprotected in 30% sucrose solution for 48 h at 4°C for immunohistochemistry.

Lung sections of 5 µm thickness were dewaxed in two xylene (Carlo Erba, 392602) baths and then rehydrated by successive ethanol baths at decreasing concentration (absolute ethanol: VWR, 20820362). The collagen fibers 1 and 3 were stained with 0.1% Sirius Red (Sigma, Cat#365548) and washed with acidified water. Then, preparations were dehydrated with absolute alcohol, cleared with xylene and mounted with a hydrophobic mounting solution.

Additional 5 µm paraffin sections of the left lung were used. After paraffin removal with xylene (Carlo Erba, 392602), the tissue was re-hydrated by successive baths of decreasing concentration of ethanol (absolute ethanol: VWR, 20820362), the alveolar epithelium was counterstained with 0.1% Fast Green (Merck, F7252) and type 1 and 3 collagen fibers were stained with 0.1% Sirius Red (Sigma, Cat#365548). Subsequently, the sections were dehydrated, mounted and observed under the optic microscope. To quantify the severity of fibrosis, we developed a new score derived from the Aschroft score with a panel of five images ranging from an image of a healthy lung with a normal lung without architectural remodeling (score 1) to a completely remodeled lung (score 5) which describes the major remodeling observed in pulmonary fibrosis (Ashcroft et al., 1988). The score 1 corresponds to a lung with no fibrotic burden at the most flimsy small fibers in some alveolar walls and with normal lung. Score 2 correspond to clearly fibrotic changes with alveoli partly enlarged and rarefied, but no fibrotic masses. Score 3 correspond to presence of single fibrotic masses (>10% of microscopic field). Score 4 correspond to lung with confluent fibrotic masses (>10%) with a lung structure severely damaged but still preserved and score 5 correspond to large continuous fibrotic masses (>50%) and lung architecture mostly not preserved.

Graph and statistical analyses were performed with GraphPad Prism (GraphPad Software, version 9). Data were presented as the mean ± SD. D’Agostino-Pearson Omnibus normality test was performed to assess the distribution of the data. Comparisons among groups were assessed using Mann-Whitney test. An α-level of 0.05 was used for all tests.

Fifteen days after the first instillation, body weight was significantly lower in BLM group as compared to control (data not shown) and this difference persists until the end (D90) (Table 1). At the lung level, the extent of fibrosis was assessed on whole lung sections stained with Sirius red showing collagen deposition (Figures 1B, C). As expected, a large, compact and extensive collagen deposition with air space reduction was observed in the lung of BLM group as compared to PBS group (Figures 1B, C, 2). Lung fibrosis was then quantified with a score of lesion severity (Figures 2A, B) indicating fibrotic injury. A mean score of 1.6 ± 0.6 (out of 5) was observed in PBS group while a score of 4.5 ± 0.7 (out of 5) was observed in the BLM group (Figures 2A, B).

At the beginning of the protocol (day 0, D0), respiratory variables were not different between PBS and BLM groups (Table 1; Figures 1D, E). By contrast, after 90 days of BLM treatment, minute ventilation ( V ˙ e) was significantly increased in BLM group as compared to PBS group (Table 1; Figures 1F, G). Indeed, mice with BLM-induced pulmonary fibrosis displayed an important increase in V_T and f _R (Table 1; Figures 1E, G). Moreover, at D90, the BLM group displayed a significant increase in the Ti/Ttot and V_T/Ti ratio as compared to the PBS group (Table 1). Finally, the BLM group showed a significant decrease in lung compliance (Figure 1H) and SpO₂ (Figure 1I) as compared to the PBS group. Our results showed also that the greater the fibrosis the lower is the SpO₂ (Figure 2C). Conversely, when the histological fibrosis score increased, V ˙ e increased (Figure 2D).

The pulmonary fibrosis-induced neuronal plasticity at the respiratory network level was evaluated through immunodetection of the long-term neuronal markers FOSB/ΔFOSB. As compared to the PBS group, the BLM group displayed a significantly higher number of FOSB/ΔFOSB-positive cells at the level of the first central relay of the peripheral afferents (SolC, +200% and SolM, +180%; Table 2; Figures 3A–D). Some other areas of the central pattern generator increased their long-term activities following induction of pulmonary fibrosis (Table 2), such as SolVL (Figures 3A–D), RTN/pFRG (Figures 3E, F) and RMg (Figures 3C, D), while other areas did not show any change (Table 2). About the RMg, our work shows that 67% of the neurons having modified their long-term activity are serotoninergic (5-HT, Figures 4B, D). NK1-R staining was used to localise more precisely the inspiratory rhythm generator [the Pre-Bötzinger complex (Gray et al., 2001)]. The localisation of the NK1-R distribution associated with that of FOSB/ΔFOSB-positive cells (Figure 5) suggested that neurons of the Pre-Bötzinger complex, the inspiratory rhythm generator, displayed a long-term modification of their activity following an experimental induction of pulmonary fibrosis (Figures 5B, D). Indeed, among the FOSB/ΔFOSB-positive cells in Pre-Bötzinger complex, 100% expressed NK1R in PBS group and 77% expressed it in the BLM group.

We observed alteration of ventilatory variables in our mouse model of chronic lung fibrosis, which reminds us, at least in part, what is observed in IPF patients (Olukogbon et al., 2016). These patients display an increase in minute ventilation associated with increased respiratory rate and decreased tidal volume (Javaheri and Sicilian, 1992; Fumeaux et al., 2003). Indeed, in patients, an impairment of ventilatory variables has been observed, characterized by a low value of forced vital capacity associated with an increase in f _R, and a decrease in V_T (Javaheri and Sicilian, 1992). In contrast, mice with BLM-induced pulmonary fibrosis displayed a large increase in V_T and f _R. However, the absolute value of V_T and the minute ventilation derived from Drorbaugh and Fenn equation (Drorbaugh and Fenn, 1955) are indicative while temporal measurements such as f _R are reliable. Indeed, the whole body plethysmography is a barometric plethymography method. For this, mice were placed in subject chamber whereas reference chamber remains empty. The pressure in the subject chamber varies with the subject’s breathing. Each of the subject’s breaths creates two distinct but related signal: a pressure modification generated by flow of air into and out of the nose and pressure modification generated by compression/expansion of air when the thorax rises and falls. Thus, results given by plethysmography approach reflect the functioning of the respiratory system as a whole (thoraco-pulmonary system, airways.) and did not only reflect V_T unlike that observed in patients where V_T was measured directly via a mouthpiece and reflects the volumes entering and leaving the airways. Thus, the observed increase in V_T may reflect an increase in respiratory effort in the face of decreased compliance. Indeed, the pressure oscillation was affected not only by V_T of the mechanical chest but also by f _R and by airway resistance (Enhorning et al., 1998). Moreover, inflammation is recognized as a mediators affecting control of breathing (Andrade et al., 2018). Furthermore, inflammation undermines neuroplasticity, including serotonin-dependent phrenic long-term facilitation (pLTF) following moderate acute intermittent hypoxia (Marciante and Mitchell, 2023). However, our model has the advantage of inducing a persistent lung fibrosis without major inflammation, reproducing histology features observed in UIP (Yegen et al., 2022) (Yegen et al., 2022). Although this remains to be verified, we assume that there is no inflammation in the CNS either, and therefore the observed increase in V ˙ e is probably not related to this. Similarly, the breathing time as well as an index of the inspiratory drive (given respectively by The Ti/Ttot and V_T/Ti ratios; (Milic-Emili and Grunstein, 1976)) was also increased following the induction of pulmonary fibrosis in our model. The increase in the V_T/Ti ratio was an indication of increased ventilatory drive. In mice, the functional decline has been positively correlated with the degree of pulmonary fibrosis (Manali et al., 2011) and the work of breathing was shown to be significantly correlated with lung fibrosis histopathology score (Phillips et al., 2012), in line with our results. Our murine model of pulmonary fibrosis obtained after repeated BLM instillations presents a diffuse collagen deposition in the whole lung (Yegen et al., 2022). This could explain, at least in part, the decrease in pulmonary compliance. This decrease is likely the consequence of increased stiffness of the pulmonary parenchyma and is consistent with that observed in the literature for murine models of pulmonary fibrosis obtained by repeated instillations (Redente et al., 2021). As there are mechanical afferents from mechanoreceptors located in the lung parenchyma, bronchi and respiratory muscles (Raux et al., 2007), the modification of lung stiffness could affect these thoraco-pulmonary mechanoreceptors, which in turn can affect the control of ventilation (Figure 6). Indeed, a disturbance of ventilatory mechanics can have a significant effect on the respiratory control (Maszczyk et al., 1990). Furthermore, we observed that mice with pulmonary fibrosis, 90 days after the first intratracheal instillation of BLM, showed a decrease in SpO₂, as observed in patients with severe IPF (Stephan et al., 2007; Takei et al., 2020). The observed desaturation could activate peripheral chemoreceptors (Khoo et al., 1982; Lahiri et al., 1983) and participate in the augmentation of V ˙ e (Figure 6). Thus, modifications of ventilatory variables in our model could be the consequence of either altered pulmonary mechanics and/or altered gas exchanges at the level of the alveolar-capillary membrane (Figure 6). The pulmonary fibrosis-induced increase in ventilation could reflect the need for the animals to make a ventilatory “effort” to compensate the increase in pulmonary stiffness and/or could be related to the presence hypoxemia resulting from the alteration of gas exchange through the thickening of the alveolar-capillary membrane (Launois et al., 1991).

The mechanism of breathing results from the rhythmic contraction of the ventilatory muscles, which depends on the central respiratory drive that determines f _R and V_T. A neuronal network located at the ponto-medullary level generates this central command and receives different kinds of afferents including metabolic afferents from central and peripheral chemoreceptors (Raux et al., 2007; Dean and Putnam, 2010). The variations in ventilatory variables observed in our mouse model of pulmonary fibrosis could also reflect a change in the functioning of this respiratory neuronal network. Our hypothesis was that the modification of the pulmonary mechanics and/or hypoxemia due to fibrosis could be at the origin of a neuroplasticity. Neuroplasticity can be divided into two major mechanisms: neuronal regeneration/collateral sprouting and functional reorganization. More precisely, the respiratory neuroplasticity describes “the persistence of morphological and/or functional neuronal changes following previous experiences” (Mitchell and Johnson, 2003). Neuroplasticity could be induced by several mechanisms such as the reactivation axon terminals, changes in the synthesis, release or re-uptake of neurotransmitters/neuromodulators, a modification in the excitability of neurons by changes in the membrane potential or the formation of new nerve connections. As mentioned above, these mechanisms generate STP and LTP through a cascade of reactions which can lead to a modification of the neuronal network, the stability of a synaptic connection, and/or the number of axon terminals that stimulate a dendrite (Fuller and Mitchell, 2017). Thus, we hypothesize that modification of lung architecture due to pulmonary fibrosis can lead to chronic stimuli through peripheral chemoreceptors and/or mechanoreceptors, which are likely to induce long-lasting functional changes in the central respiratory drive.

In IPF patients, the increase in ventilatory effort induced by exercise could be attributed to increased afferents from the lungs and/or chest wall (Van Meerhaeghe et al., 1981). Furthermore, the disturbance of the ventilatory mechanics can have a significant effect on the respiratory control (Maszczyk et al., 1990). In our mouse model of pulmonary fibrosis, analysis of the various structures within the respiratory neural network revealed an increase in the number of FOSB/∆FOSB positive cells suggesting that some of them modify their activities in response to lung damage. Indeed, we assessed the neuroplasticity by immunodetection of the long-term neuronal markers FOSB/∆FOSB (Baum et al., 2018). In this latter study, as in our study, it was shown that chronic intermittent hypoxia induced modification of FOSB/∆FOSB at the level of SolC and rVLM (Baum et al., 2018). Modification of Sol activity contributes to both the development and maintenance of plasticity (Ostrowski et al., 2023). Indeed, SolC neurons received primary afferent from the lungs, carotid body, etc. and are able to modulate respiratory response through connections with different brain region. Among these regions there are respiratory-related areas of the medulla, such as the rVLM, the RTN/pFRG (Otake et al., 1992; Accorsi-Mendonça et al., 2011) via excitatory transmission (Accorsi-Mendonça et al., 2011; Lima-Silveira et al., 2019). Thus, our results suggested that the thoraco-pulmonary mechanoreceptors and/or the chemoreceptors, via their afferents, could be at the origin of a neuroplasticity phenomenon affecting respiratory control. It was previously suggested that the ventilatory alterations observed in patients with stable diffuse interstitial lung disease would be the result of neural mechanisms (Gaultier, 1983). Our results tried to elucidate this point. It seems that lung damage could be associated with long-term changes in the sensitivity of pulmonary sensors (Undem, 2005) and/or altered synaptic plasticity in second-order neurons in the nucleus of the solitary tract as well as in higher-order synapses (Getsy et al., 2019). Indeed, SolC and SolM are the first central relays of information from the peripheral chemosensory and mechanical afferents (Torrealba and Claps, 1988). Neurons in the SolC and SolM integrate and relay information to the respiratory neuronal network including PreBötC, RTN/pFRG or Raphe Nuclei (Takakura et al., 2006; Smith et al., 2013). Therefore, changes in chemo and/or mechanoreceptors activity can affect the entire respiratory neuronal network and explain the observed increase at the level of the PreBötC and CPG structures leading to an increase of the respiratory motor output, which could modify the respiratory pattern (Figure 6). Thus, it seems that chest wall mechanoreceptors sensitive to rib cage expansion, as well as chemoreceptors, could contribute to these responses. Other experiences would be useful to differentiate the involvement of each mechanism. Furthermore, pulmonary fibrosis induced an increase in the number of FOSB/∆FOSB positive cells at the level of the RTN/pFRG and the 5-HT neurons of RMg, two areas involved in central chemosensitivity (Dias et al., 2007; Guyenet and Bayliss, 2015). These results suggested that the decrease in the ventilatory response to hypercapnia reported in patients with pulmonary fibrosis (Launois et al., 1991) could have a central origin. This point requires further experiments. Moreover, serotoninergic systems are known to be involved in many processes of ventilatory neuroplasticity, making 5-HT one of the chemical messengers of neuroplasticity. Thus our results suggest a form of neuroplasticity in the respiratory neuronal network.

Patient with IPF are disabled by dyspnea, which impairs their quality of life (Martinez et al., 2017; Plantier et al., 2018). Decrease in lung compliance, hypoxemia, increase in dead space ventilation or increase of work of breathing have all been widely described as being involved in dyspnea. However the role of these factors as a primary determinant of dyspnea is unclear. The dyspnea could also result from a mismatch between the respiratory motor output developed by the neuronal respiratory network and the accomplished ventilatory motor activity (Fukushi and Okada, 2019; Fukushi et al., 2021). Indeed, higher brain centers compare the respiratory motor command corollary discharge to the information coming from lung mechanoreceptors, and dissociation between both signals may be at the origin of dyspnea. Thus an alteration of the functioning of the respiratory neural network as presented in this paper could participate and reinforce this feeling of respiratory discomfort. Improving ventilation by a pharmacological impact on the respiratory neural network could break the vicious circle of dyspnea and could be a way to improve the quality of life of patients. Further studies will be necessary to explore this possibility.

Another important point would be to know what is the contribution of the mechanoreceptors versus chemoreceptors in the observed responses. Furthermore, it would be necessary to be able to distinguish the sites of projections of the mechanoreceptors from those of the chemoreceptors. Indeed, we suggested that the thoraco-pulmonary mechanoreceptors and/or the chemoreceptors, via their afferents, could be at the origin of a neuroplasticity phenomenon affecting respiratory control. Thus in not-hypoxemic patient, respiratory activity could be affected only by mechanical aspect of architectural modification. Indeed, we could hypothesize that inputs from the mechanical receptors are predominant signals for respiratory alterations in mild fibrosis whereas those from chemical receptors further augments respiration in severe fibrosis. Further studies will be necessary to confirm this hypothesis.