We used immunohistochemistry (IHC) to assess whether fibrocytes and CD8⁺ T cells were in close vicinity in human tissue. Sections of distal lung tissues from 17 COPD and 25 control patients were obtained, from a previously described cohort (Dupin et al., 2019), and labeled to detect CD8⁺ T cells, identified as cells positive for CD8 staining and fibrocytes, identified as cells dual positive for FSP1 and CD45 double staining (Figure 1—figure supplement 1A–D). In agreement with previous studies (Dupin et al., 2019; O’Shaughnessy et al., 1997; Saetta et al., 1998), the density of both CD8⁺ T cells and fibrocytes was increased within the subepithelial area of distal bronchi from COPD patients compared with that of control subjects (Figure 1A–C). Moreover, fibrocytes and CD8⁺ T cells were frequently in close proximity (Figure 1D). To quantify the potential for cell-cell contacts, we determined the density of CD8⁺ T cells in interaction with CD45⁺ FSP1⁺ cells (Figure 1—figure supplement 1A–D). Whatever the magnification used to automatically count interacting cells, the density of CD8⁺ T cells in interaction with fibrocytes was higher in the sub-epithelial region of distal bronchi of COPD patients than in that of control subjects (Figure 1D–F). For subsequent analyses, we chose the dilatation size ‘D8’ (3.6 μm, which represents the radius of a mean ideal round cell in our analysis) to reflect the density of interacting cells. To evaluate the minimal distance between CD45⁺ FSP1⁺ cells and neighboring CD8⁺ T cells, we used a CD8 distance map generated from the CD8 staining image, with the brightness of each pixel reflecting the distance from a CD8⁺ T cell (Figure 1—figure supplement 1E–F). The mean minimal distance between fibrocytes and CD8⁺ cells was significantly smaller in the sub-epithelial region of distal bronchi of COPD patients than in that of control subjects (Figure 1G–H). In contrast, the mean minimal distances between CD8⁺ T cells themselves or between fibrocytes themselves were unchanged (Figure 1—figure supplement 2A–B). The majority of both CD8⁺ T cells and fibrocytes were located beneath the epithelium, with their minimal distance and distribution relative to the basal membrane being similar in control and COPD patients (Figure 1—figure supplement 2C–F). Altogether, no difference in spatial repartition was observed within each cell population between control and COPD patients, but the relative distribution of fibrocytes and CD8⁺ cells was affected in tissues from patients with COPD.

Figure 1 with 4 supplements see all

Download asset Open asset

To further describe the relative spatial organization of both cell types, we used a method based on Delaunay triangulation computed on previously segmented cell barycenters. It is based on a custom-developed plugin to determine congregations of small groups of cells, called ‘clusters’ (Figure 1—figure supplement 3). As expected from our minimal distance analysis, we found a difference neither in the density of single cell-type clusters nor in their size, measured by the mean number of cells by cluster, between control subjects and patients with COPD (Figure 1I–J, left and middle panels). However, the density of clusters containing both cell types (‘mixed cell clusters’) was higher in the distal bronchi of COPD patients than in those of control subjects, with a median number of 5 and 6 cells in these clusters in control and COPD tissues, respectively (Figure 1I–J, right panels). This result indicates that fibrocytes and CD8⁺ T cells are found within close proximity in the peribronchial area of COPD patients, with possible co-organization of CD8⁺ T cells and fibrocytes in mixed cell clusters, indicating that direct and/or indirect fibrocyte-CD8⁺ T cell interactions might occur in vivo.

We determined the univariate correlation coefficients between fibrocyte density, CD8⁺ T cell density, the three variables quantifying the interaction of CD8⁺ T cells with fibrocytes (the interacting cell density, the mean minimal distance between fibrocytes and CD8⁺ T cells, and the density of mixed cell clusters), and various functional and CT parameters (Supplementary files 1-5). In particular, moderate but significant univariate correlations were found between the Forced Expiratory Volume in 1 s/Forced Vital Capacity (FEV₁/FVC) ratio (used to diagnose COPD if below 0.7), and the density of fibrocytes, the density of interacting cells, the mean minimal distance between fibrocytes and CD8⁺ T cells and the density of fibrocytes-CD8⁺ T cells clusters (Figure 1—figure supplement 4A–D). Variables significantly correlated with FEV₁/FVC were entered into stepwise regression analyses in order to find the best model fitting FEV₁/FVC. The best model associated the density of interacting cells and the density of mixed cell clusters. It explained 35% of the FEV₁/FVC variability (Supplementary file 6). The relationships between the FEV₁/FVC ratio, the density of interacting cells, and the density of mixed cell clusters were all statistically significant.

To decipher the molecular mechanisms underpinning the increased cell-cell interaction in COPD bronchi, we investigated cell adhesion and chemotaxis processes in CD8⁺ T cells of patients with COPD compared with those of controls. Using the transcriptomic profile of tissular resident memory and effector memory CD8⁺ T cells of COPD patients compared with that of control subjects in the GSE61397 microarray dataset (http://www.ncbi.nlm.nih.gov/geo/) published elsewhere (Hombrink et al., 2016), we noted significative changes in the abundance of transcripts of genes related to cell adhesion. However, the changes were not consistent with clear increased or decreased adhesive properties in both tissue-resident memory CD8⁺ CD103⁺ T-cells (T_RM) and effector memory CD8⁺ CD103^- T-cells (T_EM) (Figure 2—figure supplement 1). In contrast, transcriptomic data reveal consistent changes in COPD cells versus controls, mostly increases, in chemokines and chemokine receptors (Figure 2A). Most changes in transcripts were compatible with a pro-attractive and a pro-migratory response. In particular, there were increases of CCL2, CCL26, CXCL2, and CXCL8 expression in T_RM from patients with COPD, and CCL3L1 expression in T_EM from patients with COPD (Figure 2A).

Figure 2 with 1 supplement see all

Download asset Open asset

We then investigated whether tissular CD8⁺ T cells secretion from control or COPD patients could affect fibrocytes migration in an in vitro assay (Figure 2B). CD8⁺ T cells were purified from lung resection material sampled either in control subjects or in COPD patients, whose characteristics are reported (Supplementary file 7). Precursors of fibrocytes were purified from blood samples of a separate cohort of COPD patients (i.e. COBRA), whose characteristics are also reported (Supplementary file 8). The migration of fibrocytes was significantly increased by conditioned medium derived from tissular CD8⁺ T cells of COPD patients compared with those from control lungs (Figure 2C).

The secretory profile of these tissular CD8⁺ T cells 36 hr after culture conditions with soluble anti-CD3 and anti-CD28 antibodies were determined. The concentration of CXCL8 was increased in CD8⁺ T cells from COPD patients compared to control cells (Figure 2D) in good agreement with the transcriptomic analysis. By contrast, the concentration of both CCL3 and CCL3L1 was undetectable (data not shown), whereas that of CCL2, CXCL1, CXCL2, CXCL3, CXCL5, and CXCL6 remained unchanged (Figure 2D). Since CXCL8 is a ligand of the chemokine receptors CXCR1 and/or CXCR2, we repeated the migration assay with the addition of the drug reparixin, an antagonist of both CXCR1 and CXCR2 (Bertini et al., 2004). Whereas fibrocyte treatment with reparixin had no significant effect on the control CD8⁺ T cells-mediated migration, it did inhibit the increased migration induced by the secretions of CD8⁺ T cells purified from COPD tissues (Figure 2E). Moreover, an anti-CXCL8 blocking antibody also inhibited the increased migration induced by the secretions of CD8⁺ T cells purified from COPD tissues, in the same extend than the blocking of CXCR1/2 by reparixin (Figure 2F), suggesting that this supplementary chemotaxis is mainly due to CXCL8 and not other CXCR1/2 binding CXCL chemokines. These data indicate that tissular CD8⁺ T cells from patients with COPD promote fibrocyte chemotaxis via CXCL8-CXCR1/2 axis.

As fibrocytes and CD8⁺ T cells reside in close proximity in the subepithelial area, especially that of tissues from COPD patients, we investigated their crosstalk capacity. We developed an autologous in vitro co-culture system allowing precise control over the cell types involved. Fibrocytes and CD8⁺ T cells, both purified from the blood of COPD patients were co-cultured 2 days before image acquisition for the following 12 hr. CD8⁺ T cells were either nonactivated or activated with anti-CD3/CD28 antibodies coated microbeads, whereas fibrocytes were not stimulated. At the beginning of live imaging, nonactivated CD8⁺ T cells were equally allocated in fibrocyte-covered zones (41 ± 8%) and in fibrocyte-free zones (59 ± 8%) (Figure 3A–B). Twelve h later, most (77 ± 9%) of CD8⁺ T cells were present in contact with fibrocytes (Figure 3A–B). Activation of CD8⁺ T cells resulted in similar distribution (Figure 3A–B). These data suggest that both cell types are able to directly interact, and that these interactions progressively increase during co-culture. We tracked individual CD8⁺ T cells during 12 hr time-lapse to capture their spatiotemporal dynamics using multiple variables quantification (Figure 3C and Video 1). For both nonactivated and activated lymphocytes, the mean speed of CD8⁺ T cells decreased upon contact with fibrocytes (Figure 3D). Irrespective of the activation state of CD8⁺ T cells, a majority of intercellular contacts (49 ± 6% and 49 ± 8% for nonactivated and activated CD8⁺ T cells, respectively) were short-lived (<12 min) and dynamic, although some longer interactions (>32 min) could also be detected (30 ± 4% and 27 ± 7% for nonactivated and activated CD8⁺ T cells, respectively) (Figure 3E). The contact coefficient and the mean velocity of CD8⁺ T cells measured in the absence of contact with fibrocytes (‘Mean free speed’) were similar in both conditions of activation (Figure 3F–G). However, we observed a significant decrease in the mean speed for activated CD8⁺ T cells when they were in contact with fibrocytes (‘Mean contact speed’) compared to nonactivated CD8⁺ T cells (Figure 3H), reflecting subtle behavior changes in this condition of activation.

Figure 3 with 5 supplements see all

Download asset Open asset

Video 1

Download asset

Since multiple transient contacts have been shown to be an early trigger of events leading to clonal expansion (Obst, 2015), we wondered whether fibrocytes could promote CD8⁺ T cells proliferation using total cell count and a CellTrace-based co-culture proliferation assay. We designed two different co-culture assays modeling either a direct contact between the two cell types or an indirect contact (transwell assay). The activation of CD8⁺ T cells by anti-CD3/CD28 antibody-coated microbeads slightly increased the basal level of dividing CD8⁺ T cells (comparison of the conditions ‘CD8_NA^’ and ‘CD8_A^’ without fibrocytes in Figure 3I–P). The presence of fibrocytes in the indirect co-culture assay did not affect the proliferation capacity of non-activated CD8⁺ T cells and only moderately increased the number of dividing activated CD8⁺ T cells (Figure 3I–L). The distinction between naïve (CD45RA⁺) and memory (CD45RA^-) CD8⁺ T cells did not reveal any selective effect of fibrocytes on these two CD8⁺ subpopulations (Figure 3—figure supplement 1A, C and Figure 3—figure supplement 1E–H). In the direct co-culture model, the total number of CD8⁺ T cells and the percentage of dividing CD8⁺ T cells were far higher in the presence of fibrocytes irrespective of the activation state of CD8⁺ T cells (Figure 3M–P). This effect seemed to be particularly impressive for naïve CD8⁺ T cells as they demonstrated an average differential of 80 ± 14% and 70 ± 20% of dividing cells between the conditions with and without fibrocytes, respectively, for nonactivated (Figure 3—figure supplement 1I–J, top panels) and activated CD8⁺ T cells (Figure 3—figure supplement 1I–J, bottom panels), vs 67 ± 18% and 52 ± 20% for memory CD8⁺ T cells (Figure 3—figure supplement 1K–L). We also performed co-cultures between fibrocytes and CD4⁺ T cells, with the same settings than for CD8⁺ T cells. The results from these experiments show that fibrocytes did not have any significant effect on CD4⁺ T cells death, irrespective of their activation state (Figure 3—figure supplement 2A–C). Fibrocytes were able to promote CD4⁺ T cells proliferation in the activated condition but not in the non-activated condition (Figure 3—figure supplement 2A–D). Altogether, this implies that a direct rather than indirect interactions between CD8⁺ T cells and fibrocytes increased CD8⁺ T cell proliferation, and that although fibrocyte-mediated effect on proliferation is not specific to CD8⁺ T cells, the extend of the effect is much larger on CD8⁺ T cells than on CD4⁺ T cells.

Taking advantage of the staining of CD8⁺ T cells with the death marker Zombie NIR, we have also quantified CD8⁺ T cell death in our co-culture assay. The presence of fibrocytes in the indirect co-culture assay did not affect CD8⁺ T cell death (Figure 3—figure supplement 3A–B). In direct co-culture, the death of CD8⁺ T cells was significantly increased in the non-activated condition but not in the activated condition (Figure 3—figure supplement 3C–D).

After 6 days of co-culture, a cell population with a low level of CD8 expression (CD8^low) appeared, that was inversely proportional to the level of CD8⁺ T cells strongly expressing a high level of CD8 (CD8^high, Figure 3—figure supplement 4). The CellTrace-based assay showed that those cells highly proliferated during co-culture, especially in the direct co-culture (Figure 3—figure supplement 4E), suggesting that CD8^high cells disappeared in favor of CD8^low cells. As fibrocytes could have contaminated the cell suspension harvested from the direct co-culture, we did check that those CD8^low cells were not CD45⁺ Collagen I⁺ (Figure 3—figure supplement 5). Phenotypic analysis of this CD8^low population indicated that cells were mostly CD45RA^- cells (Figure 3—figure supplement 4A–B, S9D-E), with a low level of cytokine expression (Figure 3—figure supplement 4C, F). Since CD8^low cells may thus represent a population of exhausted T cells, we focused on CD8^high cells in the following, especially regarding the secretion profile characterization. As CD86 and CD54 co-stimulatory molecule and adhesion molecules, respectively, pivotal in immunological synapse formation, are both expressed by fibrocytes (Afroj et al., 2021; Balmelli et al., 2005), we tested the effects of anti-CD54 and anti-CD86 blocking antibodies on fibrocyte-induced proliferation of CD8⁺ T cells. The inhibition of CD86 and CD54 significantly reduced the proliferation of nonactivated CD8⁺ T cells in the direct co-culture with fibrocytes (Figure 4). However, these antibodies failed to alter the stimulatory activity of lymphocyte division by fibrocytes, when CD8⁺ T cells were previously activated (Figure 4). Blocking LFA-1 did not affect the fibrocyte-mediated CD8⁺ T cell division (Figure 4—figure supplement 1A–D), suggesting the existence of compensatory integrins at the surface of the lymphocyte, such as CD11b/CD18, to mediate the interaction with CD54. The inhibition of CD44, a receptor for hyaluronan which has been shown to be produced by fibrocytes (Bianchetti et al., 2012), did not impair the proliferation of CD8⁺ T cells irrespective of their activation state (Figure 4—figure supplement 1E–H).

Figure 4 with 1 supplement see all

Download asset Open asset

In total, these results indicate that direct contacts between fibrocytes and CD8⁺ T cells, such as those mediated by CD54 and CD86, were strong positive signals to trigger CD8⁺ T cell proliferation with the induction of CD8^high and CD8^low phenotypes.

Multiparametric flow cytometry was used to characterize the cytokine expression profile of CD8⁺ T cells in the indirect and direct co-culture with fibrocytes. When nonactivated CD8⁺ T cells were indirectly co-cultured with fibrocytes, the expression of TNF-α, IFN-γ by CD8⁺ T cells was slightly increased (Figure 5A–B). IL-10, IL-17, and Granzyme B were not detected (Figure 5A–B). When CD8⁺ T cells were activated with anti-CD3/CD28, the level of TNF-α and IFN-γ further increased, and the expression of granzyme B and IL-10 was slightly induced (Figure 5A–B). Upon direct co-culture, we observed a massive induction of TNF-α, IFN-γ, granzyme B, IL-10, and IL-17, irrespective of the activation state of CD8⁺ T cells (Figure 5C–D). Altogether, these results show that soluble factors and direct contacts between fibrocytes and CD8⁺ T cells might have an additive effect on CD8⁺ T cell cytokine production. The concentration of TNF-α measured in culture supernatant increased significantly upon co-culture between fibrocytes and non-activated CD8⁺ T cells at day 4, confirming that TNF-α was secreted in the medium upon direct interactions with fibrocytes (Figure 5E). This shows that both soluble factors produced by fibrocytes and direct contacts influence the CD8⁺ T cell secretion profile.

Figure 5 with 1 supplement see all

Download asset Open asset

We then wondered whether glucocorticoid drugs (i.e. budesonide or fluticasone propionate) could reverse the fibrocyte-induced proliferation and differentiation of CD8⁺ T cells. Treatment with glucocorticoid drugs significantly decreased fibrocyte-induced TNF-α secretion by non-activated CD8⁺ T cells, without affecting the proliferation (Figure 5—figure supplement 1). Collectively, these results underline the importance of the interaction with fibrocytes for CD8⁺ T cell activation, possibly by favoring cellular proliferation and local cytokine production.

Having shown that fibrocytes promoted CD8⁺ T cells expression of cytotoxic molecules such as granzyme B, we decided to investigate the cytotoxic capacity of CD8⁺ T cells against primary basal bronchial epithelial cells (see Supplementary file 9 for patient characteristics). Direct co-culture with fibrocytes increased total and membrane expression of the cytotoxic degranulation marker CD107a, which was only significant in non-activated CD8⁺ T cells (Figure 6A–E). A parallel increase of cytotoxicity against primary epithelial cells was observed in the same condition (Figure 6F–H). This demonstrates that following direct interaction with fibrocytes, CD8⁺ T cells have the ability to kill target cells such as bronchial epithelial cells.

Figure 6

Download asset Open asset

To analyze the effect of the interaction on the fibrocyte, we performed proteomic analyses on fibrocytes, alone or in co-culture during 6 days with CD8⁺ T cells either non-activated or activated (Figure 7A). Of the top ten pathways that were most significantly activated in co-cultured versus mono-cultured fibrocytes, the largest upregulated genes were those of the dendritic cell maturation box, the multiple sclerosis signaling pathway, the neuroinflammation signaling pathway, and the macrophage classical signaling pathway, irrespective of the activation state of CD8⁺ T cells (Figure 7B). The changes were globally identical in the two conditions of CD8⁺ T cell activation, with some upregulation more pronounced in the activated condition. They were mostly driven by up-regulation of a core set of Major Histocompatibility Complex class I (HLA-B, C, F) and II (HLA-DMB, DPA1, DPB1, DRA, DRB1, DRB3) molecules, co-simulatory and adhesion molecules (CD40, CD86, and CD54). Another notable proteomic signature was that of increased expression of IFN signaling-mediators IKBE and STAT1, and the IFN-responsive genes GBP2, GBP4, and RNF213. We also observed a strong downregulation of CD14, suggesting fibrocyte differentiation, and an upregulation of the matrix metalloproteinase-9 (MMP9) in the non-activated condition only. These changes suggest that the interaction between CD8⁺ T cells and fibrocytes promotes the development of fibrocyte immune properties, which could subsequently impact the activation of CD4⁺ T cells activation.

Figure 7

Download asset Open asset

All the above mentioned results led us to hypothesize that fibrocyte infiltration into the lung, differential migration of fibrocytes towards CD8⁺ T cells, and subsequent CD8⁺ T cell proliferation, could result in a distinct spatial cellular repartition observed in tissues obtained from patients with COPD, compared to control tissues. To investigate this hypothesis, which could not be experimentally tested, we developed an agent-based (cellular automata) model with local and random cellular interactions. We considered the lamina propria (i.e. the peribronchial zone), located between the bronchial epithelium and the smooth muscle layer, which contains fibrocytes and CD8⁺ T cells. In line with the present analysis, the computational domain (i.e. the lamina propria), corresponds to a zone of 179,000 µm². Fibrocytes and CD8⁺ T cell are considered as individual objects that can move, divide, die, and infiltrate the lamina propria in a stable state and during exacerbation. Their individual behaviors and interactions are supposed to be stochastic and the value of the probabilities has been established from literature (Afroj et al., 2021; Bivas-Benita et al., 2013; Dupin et al., 2016; Dupin et al., 2019; Ely et al., 2006; Freeman et al., 2007; Gribben et al., 1995; Hurst et al., 2010; Ling et al., 2019; McMaster et al., 2015; Mrass et al., 2017; Saetta et al., 1999; Scheipers and Reiser, 1998; Schmidt et al., 2003; Schyns et al., 2019; Siena et al., 2011; Takamura et al., 2016; Zenke et al., 2020) and the present in vitro data, as summarized in the method section and in Supplementary files 9 and 10, and exhaustively described in the Appendix 1 and (Dupin et al., 2023). Initial cell densities were scaled with respect to reference values, corresponding to the mean densities measured in non-smoking subjects. Simulations started with these initial densities and ended 20 years later, to reflect the average time between the beginning of cigarette smoke exposure and COPD onset (Løkke et al., 2006).

All the biological processes are governed by probabilities (Figure 8A). CD8⁺ T cells, but not fibrocytes, are able to proliferate, based on our own unpublished observations and other studies (Ling et al., 2019; Schmidt et al., 2003). The presence of fibrocytes in the local neighborhood of a CD8⁺ T cell can trigger CD8⁺ T cell division with an increased probability, based on the present in vitro experiments showing that the contact between those two cell types greatly enhanced CD8⁺ T cell proliferation. When a CD8⁺ T cell has many other T cells in its neighborhood, it can die with an increased probability, in agreement with (Zenke et al., 2020) and our in vitro results. Fibrocytes and CD8⁺ T cells movements depend on the local neighborhood of cells, reflecting their relative chemo-attractive properties. We then simulated the evolution over 20 years, with two sets of parameters, respectively, for the control and COPD cases (see Appendix 1).

Figure 8 with 2 supplements see all

Download asset Open asset

We first tested the results of simulations against our experimental data from patients’ tissues. First, we compared cell densities, experimentally measured in tissue samples, with theoretical predictions at the final state. Snapshots of the peribronchial area at the end of the simulations show that the densities of cells as well as their relative distribution were different between healthy and COPD situations (Figure 8B). From the simulations (n=160 in each condition), we found a median of 754 CD8⁺ T cells/mm² (95% CI, 748–763) and 106 fibrocytes/mm² (95% CI, 101–108) in the control situation, and 1187 CD8⁺ T cells/mm² (95% CI, 1169–1195) and 212 fibrocytes/mm² (95% CI, 206–216) in the COPD situation. These values are in very good agreement with our experimental findings, and the simulations were also able to reproduce the statistical increase of cell densities in COPD situations compared to that of controls (Figure 8C). Next, we tested if our theory accounted for the experimental relative distribution of CD8⁺ T cells and fibrocytes. The densities of CD8⁺ T cells in interaction with fibrocytes (Figure 8D), the mean minimal distances between fibrocytes and CD8⁺ cells (Figure 8E), the distribution of mean minimal distances (Figure 8—figure supplement 1), and the mean number of mixed cell clusters (Figure 8F) were in good agreement with tissular analyses and mimicked the variations observed between control subjects and patients with COPD. The densities of mixed cell clusters predicted by simulations (control simulations:median = 17 clusters/mm² (95% CI, 18–21), COPD simulations:median = 45 clusters/mm² (95% CI, 46–51), p<0.001) agreed perfectly with experimental measurements (Figure 8G) and were, therefore, chosen as a readout of intercellular interactions in the following analyses. If purely random, the density of mixed clusters was expected to be 28 clusters/µm² (95% CI, 25–29) and 73 clusters/µm² (95% CI, 70–74) in control and COPD situations, respectively (Figure 8—figure supplement 2). These random densities as well as the other parameters quantifying the relative distribution of cells were statistically different from the distributions obtained in both simulations and in situ analyses (Figure 8—figure supplement 2). We conclude that the relative organization of CD8⁺ T cells and fibrocytes in control and COPD bronchi did not result from a pure stochastic mechanism but implicates chemotaxis processes.

One of the strengths of the model is to allow the monitoring of the temporal evolution of the different cellular processes and the numerical detection of a change of regime (Figure 8H–I). CD8⁺ T cells infiltration remained identical in control and COPD situation. Fibrocyte-induced T cell proliferation, that represents the minor part of the total proliferation in control situations, quickly increased in COPD situations over time to reach a plateau after approximately 4 years. As the basal proliferation of CD8⁺ T cells remained similar in healthy and diseased situations, the resulting total proliferation in CD8⁺ T cells over time was higher in the COPD situation compared to the control one. COPD dynamics also affected CD8⁺ T cell death, with a concomitant increase of T cell-induced death. In total, the net balance between the gain and loss of CD8⁺ T cells was around zero for control dynamics and strictly positive for COPD dynamics, explaining the increased CD8⁺ T cell density in COPD simulations. Fibrocytes infiltration remained very similar in control and COPD dynamics (Figure 8I). Fibrocytes death was initially lower in COPD simulations before increasing and reaching a stationary state after approximately 7 years, resulting in a net expansion of fibrocytes population in COPD bronchi after 20 years. Moreover, the simulations allowed us to monitor the interactions between fibrocytes and CD8⁺ T cells. The density of mixed cell clusters gradually increased in the first years of the COPD simulation before reaching a stationary state after approximately 6 years (Figure 8J, Video 2, Video 3, Video 4, Video 5). Altogether, the theory of the influence of local interactions tested by our agent-based (cellular automata) model correctly accounts for the shift of the absolute and relative distribution of CD8⁺ T cells and fibrocytes in peribronchial areas from control subjects to patients with COPD.

Video 2

Download asset

Video 3

Download asset

Video 4

Download asset

Video 5

Download asset

We performed additional simulations to investigate the outcomes of possible therapeutic interventions. First, we applied COPD dynamics for 20 years, to generate the COPD states, that provide the basis for treatment implementation. Then, we applied COPD dynamics for 7 years, that mimics the placebo condition (Figure 9A), and we compared it to a control dynamics (‘Total inhibition’), that mimics an ideal treatment able to restore all cellular processes. As expected the populations of fibrocytes and CD8⁺ T cells, as well as the density of mixed clusters, decreased. These numbers reached levels similar to healthy subjects after approximately 2.5 years, and this time point can, therefore, be considered as the steady state (Figure 9B–E). Monitoring of the different processes revealed that these effects were mainly due to a reduction in fibrocyte-induced CD8⁺ T duplication, and a transient or more prolonged increase in basal fibrocyte and CD8⁺ T death (Figure 9C–D). Then, three possible realistic treatments were considered (Figure 9A). We tested the effect of directly inhibiting the interaction between fibrocytes and CD8⁺ T cells by blocking CD54. This was implemented in the model by altering the increased probability of a CD8⁺ T cell to divide when a fibrocyte is in its neighborhood, as shown by the co-culture results (Figure 4). We also chose to reflect the effect of a dual CXCR1/2 inhibition by setting the displacement function of fibrocyte similar to that of control dynamics, in agreement with the in vitro experiments (Figure 2E). Blocking CD54 only slightly reduced the density of CD8⁺ T cells compared to the placebo condition, and had no effect on fibrocyte and mixed cluster densities (Figure 9B). CXCR1/2 inhibition was a little bit more potent in the reduction of CD8⁺ T cells than CD54 inhibition, and it also significantly decreased the density of mixed clusters (Figure 9B). As expected, this occurred through a reduction of fibrocyte-induced duplication, which was affected more strongly by CXCR1/2 blockage than by CD54 blockage (Figure 9C–E). Combining both therapies (CD54 and CXCR1/2 inhibition) did not strongly major the effects (Figure 9B–E). In all the conditions tested, the size of the fibrocyte population remained unchanged, suggesting that other processes such as fibrocyte death or infiltration should be targeted to expect broader effects.

Figure 9

Download asset Open asset

Lung tissues for the in situ study were obtained from a previously described cohort (Dupin et al., 2019). The study was registered at ClinicalTrials.gov with the identifier NCT01692444 (‘Fibrochir’ study). The study protocol was approved by the research ethics committee (‘CPP’) and the French National Agency for Medicines and Health Products Safety (‘ANSM’). Briefly, subjects more than 40 years of age were eligible for enrolment if they required thoracic lobectomy surgery for cancer (pN0), lung transplantation, or lung volume reduction. A total of 17 COPD patients with a clinical diagnosis of COPD according to the GOLD guidelines (Global Initiative for Chronic Obstructive Lung Disease, 2023) and 25 non-COPD subjects (‘control subjects’) with normal lung function testing (i.e. FEV₁/FVC>0.70) and no chronic symptoms (cough or expectoration) were recruited from the University Hospital of Bordeaux. Due to low quality of some tissue sections, distal bronchial identification or fibrocyte and CD8⁺ T cell quantification was impossible in five control specimens and 5 COPD specimens, which were excluded from the in situ analysis.

Lung tissues for the purification of tissular CD8⁺ T cells and basal bronchial epithelial cells were obtained from a separate cohort of patients requiring thoracic lobectomy surgery for nodule or cancer (pN0) (i.e. TUBE, sponsored by the University Hospital of Bordeaux, which includes its own local ethic committee (CHUBX 2020/54)). According to the French law and the MR004 regulation, patients received an information form, allowing them to refuse the use of their surgical samples for research. For the purification of tissular CD8⁺ T cells, a total of 20 patients with COPD and 26 nonsmokers were prospectively recruited from the University Hospital of Bordeaux, according to the GOLD guidelines (Global Initiative for Chronic Obstructive Lung Disease, s. d.) (Supplementary file 7). For the purification of basal bronchial epithelial cells, a total of two patients were prospectively recruited from the University Hospital of Bordeaux (Supplementary file 9).

To study fibrocyte- CD8⁺ T cells interplay in vitro, blood samples were obtained from a separate cohort of COPD patients, (i.e. COBRA (Bronchial Obstruction and Asthma Cohort; sponsored by the French National Institute of Health and Medical Research, INSERM, Ethics committee number: 2008-A00294-51/1)), as outpatients in the Clinical Investigation Centre of the University Hospital of Bordeaux (Supplementary file 8).

All subjects gave their written informed consent to participate in the studies. The studies received approval from the local or national ethics committees.

Fragments of distal parenchyma were obtained from macroscopically normal lung resection or transplantation material. The samples were embedded in paraffin and sections of 2.5 µm thick were cut, as described previously (Dupin et al., 2019). Sections were deparaffinized through three changes of xylene and through graded alcohols to water. Heat-induced antigen retrieval was performed using citrate buffer, pH 6 (Fisher Scientific, Illkirch, France) at 96 °C in a Pre-Treatment Module (Agilent, Les Ulis, France). Endogenous peroxidases were blocked for 10 min using hydrogen peroxide treatment (Agilent). Nonspecific binding was minimized by incubating the sections with 4% Goat Serum (Agilent) for 30 min, before CD8 staining, and before the double staining for CD45 and FSP1. First, the sections were stained with rabbit anti-CD8 monoclonal antibody (Fisher Scientific) for 45 min, and then incubated with HRP anti-Rabbit (Nichirei Biosciences). Immunoreactivity was detected by using the DAB System (Agilent). Second, the same sections were stained with mouse anti-CD45 monoclonal antibody (BD Biosciences, San Jose, CA) overnight and then with rabbit anti-FSP1 polyclonal antibody (Agilent) for 45 min. They were incubated with Alexa568–conjugated anti-Mouse and with Alexa488–conjugated anti-Rabbit (Fisher Scientific) antibodies. Immunoreactivity was detected by fluorescence for FSP1 and CD45 staining.

The sections were imaged using a slide scanner Nanozoomer 2.0HT with fluorescence imaging module (Hamamatsu Photonics, Massy, France) using objective UPS APO 20 X NA 0.75 combined to an additional lens 1.75 X, leading to a final magnification of 35 X. Virtual slides were acquired with a TDI-3CCD camera. Fluorescent acquisitions were done with a mercury lamp (LX2000 200 W - Hamamatsu Photonics) and the set of filters adapted for DAPI, Alexa 488, and Alexa 568. Brightfield and fluorescence images were acquired with the NDP-scan software (Hamamatsu) and processed with ImageJ.

Quantification of CD8⁺ T cells was performed, as described in Figure 1—figure supplement 1A, C. A color deconvolution plugin was used on brightfield images to isolate the signal corresponding to DAB staining. A binary threshold was applied to this grayscale image, followed by a watershed transformation to the segmented image to separate potential neighboring cells (Figure 1—figure supplement 1C). CD8⁺ T cells were then automatically counted by recording all the positive particles with an area greater than 64 µm². This threshold was empirically determined on our images to select positive cells. Quantification of dual positive cells for FSP1 and CD45 was performed, as described in Figure 1B and D. A binary threshold was applied to fluorescence images corresponding to FSP1 and CD45 stainings. These images were combined using the ‘AND’ function of the Fiji ‘Image Calculator’ to select cells dual positive for FSP1 and CD45 double staining (Figure 1—figure supplement 1D). This was followed by a watershed transformation to separate potential neighboring cells. These CD45⁺ FSP1⁺ cells were then automatically counted by recording all the positive particles with an area greater than 64 µm².

This latter segmented image was then used to quantify CD8⁺ T cells in interaction with CD45⁺ FSP1⁺ cells as described in Figure 1E: each CD8 positive particle with an area greater than 64 µm² was enlarged using the dilatation function (4, 8, 10, and 15 pixels dilatation: used to count the cells, respectively, less than 1.8, 3.6, 4.5, and 6.8 μm apart). This modified image was combined with the segmented image for dual CD45 FSP1 positive staining using the ‘AND’ function of the Fiji ‘Image Calculator’ to select CD8⁺ T cells in interaction with CD45⁺ FSP1⁺ cells. These interacting cells were automatically counted by recording all the positive particles. The lamina propria contour was manually determined on a brightfield image and the area was calculated. For distal bronchi, the lumen area was also determined and only bronchi less than 2 mm in diameter were analyzed as described previously (Hogg et al., 2004). The densities of CD8⁺ T cells, FSP1⁺ CD45⁺ cells, and interacting cells were defined by the ratio between the number of positive cells in the lamina propria divided by the lamina propria area. Tissue area and cell measurements were all performed in a blinded fashion for patients’ characteristics.

The segmented image produced from the DAB staining image was inverted, and a CD8 distance map was built from the latter image (Figure 1—figure supplement 1F). As a result, the brighter the pixel, the closer the distance from a CD8⁺ T cell. Conversely, the darker the pixel, the farther away the distance from a CD8⁺ T cell. On the binary image produced from FSP1 and CD45 staining images, dual positive cells for FSP1 and CD45 were selected in the lamina propria. Each area corresponding to a FSP1⁺ CD45⁺ cell was reported on the CD8 distance map, and the minimal gray value in each area was measured and converted to a distance, allowing to measure the minimal distance between the CD45⁺ FSP1⁺ cell and neighboring CD8⁺ T cells. For each patient, a frequency distribution of all minimal distances (with 7 μm binning) and the mean minimal distance were calculated.

On the segmented image with dual CD45 FSP1 positive staining combined with CD8 positive staining, centroids from positive particles located in the lamina propria were connected by a Delaunay triangulation, using a custom freely available ImageJ plugin (Schneider et al., 2012; Figure 1—figure supplement 3A–C, https://github.com/flevet/Delaunay_clustering_ImageJ; copy archived at Levet, 2022). All triangles sharing one edge with the ROI defining the lamina propria were removed (Figure 1—figure supplement 3C, left panel). On the remaining triangulation a distance threshold, corresponding to the minimal mean distance between fibrocytes and CD8⁺ T cells (40 µm) was applied, allowing to select the connections with a distance lower than the threshold distance (Figure 1—figure supplement 3C, right panel). The number of clusters and their composition were then automatically recorded.

The microarray data of tissular CD8⁺ T cells were downloaded from the Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) using a dataset under the accession code GSE61397. Differential expression analysis between patients with COPD and control subjects was performed using the GEO2R interactive web tool. Heatmaps of the expression profiles for genes related to cell adhesion and chemotaxis were visualized with GraphPad Prism 6 software.

After lung parenchyma resection from control or COPD patients (Supplementary file 7), samples were finely chopped at room temperature using scissors and then enzymatically dissociated with 40 IU/mL of collagenase (ThermoFisher) in DMEM medium for 45 min at 37 °C. The enzymatic reaction was stopped by adding HBSS medium (Hank’s Balanced Salt Solution) without calcium and supplemented with 2 mM EDTA (Invitrogen, Cergy Pontoise, France). The cell suspension was filtered twice using 100 μm gauze and 70 μm cell strainer (Fisher Scientific). Tissular CD8⁺ T cells were purified by positive selection using CD8 microbeads (Miltenyi Biotech, Paris, France). Then, tissular CD8⁺ T cells were resuspended in DMEM supplemented with 8% fetal calf serum, soluble anti-CD3 and anti-CD28 antibodies (respectively, 1 μg and 3 μg for 10⁶ cells) for a final concentration of 0.5 × 10⁶ cells/mL. After 36 hr, supernatants from tissular CD8⁺ T cells were collected and frozen, for migration experiments or for further analyses. Supernatants from different samples obtained either from non-smoking subjects or patients with COPD were pooled for migration experiments. Supernatant concentrations of CXCL1, CXCL3, CXCL5, CXCL6, and CXCL8 were measured using ELISA (BioTechne for CXCL1, 5, 6, 8, Abcam for CXCL3). CCL26, CXCL2, and CCL2 concentrations were measured by using a customized Bio-Plex Assay (BioRad, Hercules, CA), using a special plate reader (Bio-Plex 200 Systems, BioRad) and software (Bio-Plex manager), according to the manufacturer’s instruction.

Fibrocytes precursors were isolated from peripheral blood as described previously (Dupin et al., 2016). Fibrocyte migration was assessed using a modified Boyden chamber assay. The transwell inserts (pore size 8 µm, Dutscher) and the wells were coated for 1 hr at room temperature with poly-lysine-ethylene glycol (PEG-PLL, SuSoS, Dübendorf, Switzerland) to prevent cell adhesion. A total of 0.3 × 10⁶ NANT cells resuspended in 0.2 ml DMEM, containing 4.5 g/l glucose and L‐glutamine, supplemented with penicillin/streptomycin and MEM non-essential amino acid solution were added to the upper compartment of each well. When indicated, NANT cells were pretreated for 30 min at 37 °C with 200 nM reparixin (MedChem Express), an antagonist of CXCR1-2. Supernatants of tissular CD8⁺ T cells from non-smoking control subjects or COPD patients were added to the bottom compartment of each well. When indicated, supernatants were pretreated for 30 min at 37 °C with blocking Ab against CXCL8 (clone 6217, BioTechne, 1 µg/mL) or respective control Ab. After 12 hr, the content of the bottom compartment was removed and DAPI staining was performed to exclude dying cells. Cells were then fixed, permeabilized, and stained with anti-Collagen Type I-FITC (Sigma Aldrich), anti-CD45-APC (BD Pharmingen), anti-CXCR1-PE, and anti-CXCR2-APC-Cy7 (Miltenyi Biotec, Paris, France). Fibrocyte migration was assessed by flow cytometry using double labeling CD45-Collagen I. To obtain absolute values of migratory cells, flow cytometric counts for each condition were obtained during a constant predetermined time period (1 min). The fraction of migratory fibrocytes was defined as the number of CD45⁺ Col1⁺ cells counted in the bottom chamber divided by the number of total added cells. These values were normalized to the fraction of migratory fibrocytes obtained in the control condition.

Peripheral blood mononuclear cells (PBMCs) were first separated from the whole blood by Ficoll-Hypaque (Eurobio Scientific, Les Ulis, France) density gradient centrifugation. Cells were washed twice in cold PBS containing 0.5% bovine serum albumin (BSA, Sigma-Aldrich, Saint Quentin-Fallavier, France) and 2 mM Ethylene Diamine Tetra-acetic Acid (EDTA, Invitrogen, Cergy Pontoise, France). CD8⁺ and CD4⁺ T cells were purified by positive selection using CD8 and CD4 microbeads, respectively (Miltenyi Biotech, Paris, France). CD8⁺/CD4⁺ T cells were washed in a buffered solution (‘CTL-Wash,’ Cellular Technology Limited, Bonn, Germany) and resuspended in a serum-free freezing media (‘CTL-Cryo Medium,’ Cellular Technology Limited, Bonn, Germany) for cryopreservation of freshly-isolated CD8⁺/CD4⁺ T cells during fibrocyte differentiation. The CD8⁺ or CD4⁺ T cells-depleted cell fraction was then depleted from CD3⁺ cells using CD3 microbeads (Miltenyi Biotec). Cell suspension containing fibrocyte precursors was cultured during at least 14 days to induce fibrocyte differentiation: a total of 2 × 10⁶ cells resuspended in 1 ml DMEM (Fisher Scientific, Illkirch, France), containing 4.5 g/l glucose and glutaMAX, supplemented with 20% fetal calf serum (Biowest, Riverside, USA), penicillin/streptomycin and MEM non-essential amino acid solution (Sigma-Aldrich), was added to each well of a 12 well plate. After one week in culture, fibrocyte differentiation was induced by changing the medium to a serum-free medium. Mediums were changed every 2–3 days.

One day before co-culture, CD8⁺ T or CD4⁺ cells were thawed. A buffer solution previously heated to 37 °C (PBS 1 X with 0.5% BSA and 2 mM EDTA) was added to the cell suspension. CD8⁺/CD4⁺ T cells were washed with PBS and resuspended in DMEM supplemented with 8% fetal calf serum for a final concentration of 0.5 × 10⁶ cells/mL. CD8⁺/CD4⁺ T cells were either stimulated with a low dose of CD3 antibody (3 μg/10⁶ cells) to promote cell survival without stimulating cell proliferation (‘non-activated’ condition), or stimulated overnight with anti-CD3/CD28 coated microbeads (Fisher Scientific) with a bead-to-cell ratio of 1:1 (‘activated’ condition). At day 0 (co-culture), these beads were removed, CD8⁺/CD4⁺ T cells were stained with 5 µM CellTrace Violet (Fisher Scientific) in case of proliferation experiments, before being added to fibrocyte cultures (0.5 × 10⁶ CD8⁺/CD4⁺ T cells/well). In blocking experiments, the antibodies (Abs) directed against LFA-1 (clone HI111, BioLegend, 1 µg/mL), CD54 (clone HA58, eBioscience), CD86 (clone IT2.2, eBioscience, 10 µg/mL) or CD44 (clone 82102, BioTechne, 10 µg/mL) were used with their respective control Abs, mIgG1 κ (clone MOPC-21, BioLegend), mIgG2b κ (eBM 2b, eBioscience), mIgG2B (133303, BioTechne). In LFA-1 and CD44 blocking experiments as well as in glucocorticoid drugs experiments, CD8⁺ T cells were preincubated, respectively, with corresponding Abs, budesonide, or fluticasone propionate (10^–8M, MedChemExpress) at 37 °C for 1 hr before being added to fibrocytes. In CD54 and CD86 blocking experiments, fibrocytes were preincubated with corresponding Abs at 37 °C for 1 hr before adding CD8⁺ T cells. For indirect co-culture, CD8⁺ T cells were cultured in 0.4 µm transwell inserts (Sigma-Aldrich) for 12-well plates.

For time-lapse microscopy, cells were imaged after 2 days of co-culture, at 37 °C and with 5% CO₂ on an inverted DMi8 stand microscope (Leica, Microsystems, Wetzlar, Germany) equipped with a Flash 4.0 sCMOS camera (Hamamatsu, Japan). The objective used was an HC PL FL L 20 X dry 0.4 NA PH1. The multi-positions were done with an ASI MS-2000–500 motorized stage (Applied Scientific Instrumentation, Eugene, USA). The 37 °C/5%CO2 atmosphere was created with an incubator box and a gaz heating system (Pecon GmbH, Erbach, Germany). This system was controlled by MetaMorph software (Molecular Devices, Sunnyvale, USA). Phase contrast images were collected every 2 min for 12 hr. Image analysis and measurements were performed with the ImageJ software. Using the plugin ‘Cell counter’ of the Fiji software, the number of CD8⁺ T cells in direct contact with a fibrocyte as well as the number of free CD8⁺ T cells were manually counted at the beginning of the acquisition and after 12 hr of acquisition. Cell tracking was performed using the ‘Manual Tracking’ plugin of the Fiji software to determine the durations of contacts between tracked CD8⁺ T cell with fibrocytes and the frequency of contact. A contact was defined manually by a direct interaction between CD8⁺ T cell and fibrocyte. Five numerical variables were collected to characterize CD8⁺ T cell dynamic over time. The mean speed corresponded to the track length divided by the time of tracking duration. The mean free speed corresponded to the length of the track when the T cell was not interacting with any other cell, divided by the time spent free. The mean contact speed corresponded to the length of the track when the T cell is in contact with a fibrocyte, divided by the time spent in contact. For each T cell and for each contact, a contact time was defined as the time spent in contact until the T cell becomes free again. Then, each T cell can have many contact times with fibrocytes. The contact coefficient was defined by the proportion of time the T cell was in contact with a fibrocyte divided by the time of tracking duration.

Four or 6 days after co-culture, CD8⁺/CD4⁺ T cells were harvested and manually counted before being processed for FACS analysis. Dead cells were excluded by using DAPI or the Zombie NIR fixable viability kit (BioLegend). Intracellular cytokines were assessed following stimulation with PMA (25 ng/ml, Sigma-Aldrich), ionomycin (1 µM, Sigma-Aldrich) for 4 hr, and brefeldin A (5 µg/ml, Sigma-Aldrich) for the last 3 hr. Cells were stained with anti-CD8-PerCP-Vio700 or anti-CD4-PerCP-Vio700, anti-CD45-RA-FITC, and then fixed, permeabilized using the IntraPrep Permeabilization Reagent Kit (Beckman Coulter) and stained with anti-Granzyme-APC, anti-TNF-α-PE, anti-IFN-γ-APC, anti-IL-17- PE-Cy7, anti-IL-10-PE or isotype controls (Miltenyi Biotech, Paris, France). The percentage of cell proliferation was estimated using Cell Trace Violet fluorescence loss. FACS data were acquired using a Canto II 4-Blue 2-Violet 2-Red laser configuration (BD Biosciences). Flow cytometry analyses were performed using Diva 8 (BD Biosciences). Human TNF-α concentration levels were quantified using ELISA following the manufacturer’s recommendations (BioTechne). Values below the detection limit were counted as zero.

Six days after co-culture, CD8⁺ T cells were harvested and stained to assess CD107a expression following stimulation with PMA (25 ng/ml, Sigma-Aldrich) and ionomycin (1 µM, Sigma-Aldrich) for 3 hr. Cells were stained with anti-CD8-PerCP-Vio700, anti-CD45-RA-FITC, or isotype controls (Miltenyi Biotech, Paris, France). For membrane expression analysis, cells were then stained with anti-CD107a-PE or isotype control (BD Biosciences, San Jose, CA). For total expression analysis, cells were fixed, permeabilized using the IntraPrep Permeabilization Reagent Kit (Beckman Coulter), and stained with anti-CD107a-PE or isotype control. FACS data were acquired using a Canto II 4-Blue 2-Violet 2-Red laser configuration (BD Biosciences).

Human basal bronchial epithelial cells were derived from bronchial specimens (Supplementary file 9) as described previously (Trian et al., 2015). Bronchial epithelial tissue was cultured in PneumaCult-Ex medium (Stemcell Technologies, Vancouver, Canada) under a water-saturated 5% CO₂ atmosphere at 37 °C until epithelial cells reached 80–90% confluence. For the cytotoxic assay, basal bronchial epithelial cells were plated on a 24-wells plate in PneumaCult-Ex medium.

After 6 days of co-culture, CD8⁺ T cells were resuspended (2.10⁶ cells/ml), and labeled with 5 µM Cell Trace Violet (Fisher Scientific) following the manufacturer’s recommendations. Cell Trace Violet-labeled CD8⁺ T cells as effector cells and primary basal bronchial epithelial cells as target cells were co-incubated at effector:target ratio of 4:1 in PneumaCult-Ex medium. After 6 hr, CD8⁺ T cells were kept with supernatant, and basal epithelial cells were harvested by using 0.05% trypsin-EDTA (Fisher Scientific). The entire content of each well was stained using Annexin V- Propidium iodide (Fisher Scientific) for the detection of apoptotic and living cells by flow cytometry, according to the manufacturer’s instructions. CD8⁺ T cells exclusion was performed based on the Cell Trace Violet-based fluorescence performed 6 hr before.

Four independent biological replicates on total protein extracts from fibrocytes cultured alone (control condition), or co-cultured with CD8⁺ T cells that have been previously non-activated or activated during 6 days. 10 µg of proteins were loaded on a 10% acrylamide SDS-PAGE gel and proteins were visualized by Colloidal Blue staining. Migration was stopped when samples had just entered the resolving gel and the unresolved region of the gel was cut into only one segment. Each SDS-PAGE band was cut into 1 mm × 1 mm gel pieces. Gel pieces were destained in 25 mM ammonium bicarbonate (NH₄HCO₃), 50% Acetonitrile (ACN), and shrunk in ACN for 10 min. After ACN removal, gel pieces were dried at room temperature. Proteins were first reduced in 10 mM dithiothreitol, 100 mM NH₄HCO₃ for 60 min at 56 °C then alkylated in 100 mM iodoacetamide, 100 mM NH₄HCO₃ for 60 min at room temperature and shrunk in ACN for 10 min. After ACN removal, gel pieces were rehydrated with 50 mM NH₄HCO₃ for 10 min at room temperature. Before protein digestion, gel pieces were shrunk in ACN for 10 min and dried at room temperature. Proteins were digested by incubating each gel slice with 10 ng/µl of trypsin (V5111, Promega) in 40 mM NH₄HCO₃, rehydrated at 4 °C for 10 min, and finally incubated overnight at 37 °C. The resulting peptides were extracted from the gel in three steps: a first incubation in 40 mM NH₄HCO₃ for 15 min at room temperature and two incubations in 47.5% ACN, 5% formic acid for 15 min at room temperature. The three collected extractions were pooled with the initial digestion supernatant, dried in a SpeedVac, and resuspended with 0.1% formic acid for a final concentration of 0.02 µg/µL. NanoLC-MS/MS analysis was performed using an Ultimate 3000 RSLC Nano-UPHLC system (Thermo Scientific, USA) coupled to a nanospray Orbitrap Fusion Lumos Tribrid Mass Spectrometer (Thermo Fisher Scientific, California, USA). Each peptide extract were loaded on a 300 µm ID × 5 mm PepMap C₁₈ precolumn (Thermo Scientific, USA) at a flow rate of 10 µL/min. After a 3 min desalting step, peptides were separated on a 50 cm EasySpray column (75 µm ID, 2 µm C₁₈ beads, 100 Å pore size, ES903, Thermo Fisher Scientific) with a 4–40% linear gradient of solvent B (0.1% formic acid in 80% ACN) in 115 min. The separation flow rate was set at 300 nL/min. The mass spectrometer operated in positive ion mode at a 1.9 kV needle voltage. Data were acquired using Xcalibur 4.4 software in a data-dependent mode. MS scans (m/z 375–1500) were recorded at a resolution of R=120,000 (@ m/z 200), a standard AGC target, and an injection time in automatic mode, followed by a top speed duty cycle of up to 3 s for MS/MS acquisition. Precursor ions (2–7 charge states) were isolated in the quadrupole with a mass window of 1.6 Th and fragmented with HCD@28% normalized collision energy. MS/MS data were acquired in the Orbitrap cell with a resolution of R=30,000 (@m/z 200), a standard AGC target, and a maximum injection time in automatic mode. Selected precursors were excluded for 60 s. Protein identification and Label-Free Quantification (LFQ) were done in Proteome Discoverer 2.5. MS Amanda 2.0, Sequest HT, and Mascot 2.5 algorithms were used for protein identification in batch mode by searching against a Uniprot Homo sapiens database (81,856 entries, released January 20^th, 2023). Two missed enzyme cleavages were allowed for the trypsin. Mass tolerances in MS and MS/MS were set to 10 ppm and 0.02 Da. Oxidation (M) and acetylation (K) were searched as dynamic modifications and carbamidomethylation (C) as static modifications. Peptide validation was performed using the Percolator algorithm and only ‘high confidence’ peptides were retained corresponding to a 1% false discovery rate at the peptide level (Käll et al., 2007). Minora feature detector node (LFQ) was used along with the feature mapper and precursor ions quantifier. The normalization parameters were selected as follows: 1: Unique peptides, 2: Precursor abundance based on intensity, 3: Normalization mode:total peptide amount, 4: Protein abundance calculation:summed abundances, 5: Protein ratio calculation:pairwise ratio based, 6: Imputation mode:Low abundance resampling and 7: Hypothesis test:t-test (background based). Quantitative data were considered for master proteins, quantified by a minimum of two unique peptides, a fold changes above 2, and a statistical p-value adjusted using Benjamini-Hochberg correction for the FDR lower than 0.05. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Deutsch et al., 2020) partner repository with the dataset identifier PXD041402. Proteins were clusterized according to their functions by using the Kyoto Encyclopedia of genes and genome analysis in the search tool for retrieval of interaction between genes and proteins (STRING) database. More global analysis of the data was performed via the use of Ingenuity Pathway Analysis (IPA; Qiagen). We used the ‘Core Analysis’ package to identify relationships, mechanisms, functions, and pathways relevant to a dataset.

Exhaustive description of the mathematical model is provided in the Appendix 1.

To understand the interaction between fibrocyte and CD8⁺ T cells in the spatial cellular organization in the peribronchial area, we constructed a discrete time cellular automata model. Two agent types are introduced - CD8⁺ T cell agents and fibrocytes agents, denoted C and F, respectively. C and F cells evolve on a lattice in two-dimensions. We take as the surface of interest a zone with a crown shape, containing 3 652 lattice sites corresponding to a total area of approximately 179,000 µm², which is in agreement with our in situ measurements. Reflecting (zero-flux) boundary conditions are imposed at the external and internal borders. On each site, there is at most one cell. The lattice is initially randomly seeded with both F and C cells at densities corresponding to the mean distribution of non-smokers subjects, reflecting the ‘healthy’ situation: $n_0\left(C\right)$ = 660 cells/mm², and $n_0\left(F\right)$ = 106 cells/mm². This corresponds to an average value of $N_0\left(C\right)=$ 118 C cells and $N_0\left(F\right)=$ 19 F cells.

We assumed that for a healthy subject as for a patient with COPD, the same model can be applied but with different parameters. These parameters are estimated thanks to experiments and data from the literature (see Appendix 1 and Dupin et al., 2023 for a complete description, Supplementary file 11 for numerical values).

The notations and parameters of the mathematical model are summarized in Supplementary file 10 and their numerical values are given in Supplementary file 11. We now describe the behavior of the cells and their interactions. F and C cells infiltrate into the peribronchial area at the stable state with the respective probabilities $p_{istaF}$ and $p_{istaC}$ , and during exacerbation, a supplementary infiltration can occur, each year, with the probability $p_{iexaF}$ (resp. $p_{iexaC}$). In the model, C cells can proliferate with a very low probability $p_C$ , but the presence of F cells in the local neighborhood of a C cell can induce C cell division with an increased probability $p_{C/F}$ , based on our own results and another study (Afroj et al., 2021). We suppose that fibrocytes do not proliferate, as shown by our own in vitro observations (data not shown) and other studies (Ling et al., 2019; Schmidt et al., 2003). F and C cells can move, with probabilities which are determined by the results from chemotaxis experiments (Figure 2). F and C cells die with a ‘basal’ probability $p_{dC}$ (respectively, $p_{dF}$). C cells also die with an increased probability $p_{dC+}$ when the considered C cell has many other C cells in its neighborhood, in agreement with previous data (Zenke et al., 2020). Some of the probabilities are independent of the local environment ($p_{istaF}$ , $p_{istaC}$ , $p_{iexaF}$ , $p_{iexaC},\mathrm{}{p}_C$), the other ones being dependent of the local environment ($p_{C/F,}$ , $p_{dC+}$, and displacement probabilities) (Figure 8A).

Each simulation represents a total duration of 20 years and is divided into 3,504,000 iterations, of 3 min each. Each type of simulation is performed 160 times. This time period allowed the investigation of COPD development. At the final state (20 years), the total numbers of F and C cells, the densities of C cells in interaction with F cells, the minimal distances between C and F cells, and the number and composition of clusters were quantified in the control and COPD situations. For therapeutic interventions, different dynamics were applied during 7 years, starting from COPD states that have been generated using the application of COPD dynamics during 20 years. At the final state (7 years after the application of therapeutic dynamics), the same outcomes than at 20 years were quantified.

Biological replicates are samples purified from different patients. Technical replicates are repeated measurements in the same biological samples. For in situ analyses, both technical replicates (measurements performed on at least two bronchi) and biological replicates have been performed. For most of in vitro analyses, only biological replicates have been obtained, because our cells of interest have a very limited lifespan in culture and experiments can, therefore, be performed only once. Exceptions include migration assays and ELISA measurements, with both technical duplicates and biological replicates. For simulations, technical replicates have been performed, with each type of simulation being repeated 160 times.

Statistical significance, defined as p<0.05, was analyzed by t-tests and MANOVA for variables with parametric distribution, and by Kruskal-Wallis with multiple comparison z-tests, Mann-Whitney tests, Wilcoxon tests, and Spearman correlation coefficients for variables with non-parametric distribution and by two-way ANOVA for distribution tests, using GraphPad Prism 6 software. RStudio software was used to perform stepwise regression and multivariate regression analyses.