Tissue fibrosis is a universal feature of the response to injury and is associated with high mortality in pulmonary fibrosis, liver cirrhosis, chronic heart failure, and chronic kidney disease. In most cases, the inflammatory response is marked by large numbers of monocyte-derived macrophages, which have been found to be necessary for lung fibrosis (1–3). Our recent work profiled the fibrotic phase of lung injury at the single-cell level, revealing a subset of Cx3cr1+ macrophages that localize to clusters of activated fibroblasts and exert a profibrotic effect (3). With respect to the profibrotic mechanisms of these macrophages, one emerging view is that they activate surrounding fibroblasts by secreting cytokines and growth factors, such as TNFα, TGFβ, and PDGF (2–6). However, the full repertory of signals deriving from macrophages is not known.

Previous in vitro results have shown that extracellular ATP activates purinergic receptors on fibroblasts, leading to calcium influx and collagen expression (7, 8). ATP is a known damage associated molecular pattern that binds to and activates membrane channels for calcium entry in a range of inflammatory contexts (9). Recent reports demonstrate that connexin 43 (Cx43) hemichannels are a conduit for ATP efflux by macrophages, with paracrine effects on calcium transients in bystander cells (10–12). However, this biology has not been explored in reference to macrophage-fibroblast crosstalk.

In this study, we found lung fibroblast calcium is increased by co-culture with lung monocyte-derived macrophages. Furthermore, macrophages were found to secrete ATP in a Cx43-dependent manner after injury, and the fibroblast cytosolic calcium response was decreased in co-culture with Cx43 KO macrophages. In vivo studies indicated that the ATP receptor P2rx4 expressed by fibroblasts was necessary for lung fibrosis, revealing a novel profibrotic interaction of macrophages and fibroblasts in lung injury.

Recent studies have shown that Cx43 hemichannels serve as a conduit for ATP release from macrophages, leading to paracrine effects on other cells in the microenvironment (10–12). To probe the potential significance of macrophage ATP efflux in fibrosis, we employed the bleomycin model of lung fibrosis (21). We first measured ATP in conditioned media of cultured tdTomato+ cells isolated from the lungs of bleomycin-injured Cx3cr1-CreERT2: Rosa26-LSL-tdtomato mice (largely monocytes and monocyte-derived macrophages) and found that Cx43 KO macrophages released less ATP into the extracellular space than wild type ( Figure 1A ).

Extracellular ATP is known to bind to membrane-bound purinergic receptors that mediate extracellular calcium entry into the cytosol. First, to test whether fibroblast cytosolic calcium is elevated in injury, we imaged precision-cut lung slices by two-photon laser scanning microscopy. Slices were prepared from mice with the fibroblast-specific Col1a2-CreERT2 (13) crossed to R26-LSL-Salsa6f (22), which consists of tdTomato fused to the calcium indicator GCaMP; the MacBlue allele (23) was included for visualization of moMacs. These data indicated that fibroblast cytosolic calcium was increased in injured mice compared to uninjured controls ( Figures 1B , Supplementary Videos 1 , 2 ). We then explored the role of macrophage Cx43 by fibroblast calcium imaging in a co-culture system. Co-culture with WT macrophages was associated with higher fibroblast cytosolic calcium than co-culture with Cx43 KO macrophages, or fibroblasts alone ( Figure 1C ). Of note, our previous bulk RNAseq data indicated an increase in Cx43 expression in lung monocyte-derived macrophages compared with tissue-resident alveolar macrophages after bleomycin injury (3). Therefore, to test the functional role of Cx43 in lung fibrosis resulting from injury we used the Cx3cr1-CreERT2, a Cre driver that has been used for functional study of monocyte-derived macrophages in multiple contexts in vivo (3, 24, 25). We measured lung collagen by hydroxyproline assay and found that the increase of lung collagen after bleomycin injury was markedly reduced with Cx43 deletion in moMacs compared to wild type ( Figures 2A, B ). Taken together, these data suggest the hypothesis that macrophages efflux ATP into the extracellular space via Cx43, with a resulting calcium response that is necessary for lung fibrosis after injury.

To pursue this hypothesis further, we first sought to elucidate the relevant ATP receptor in fibroblasts. We had previously identified 12 distinct subsets of collagen-producing cells from control mice and mice treated with bleomycin, as well as 6 distinct populations of collagen-producing cells from normal human lungs and lungs obtained at the time of transplant from patients with Idiopathic Pulmonary Fibrosis (16). Using these datasets, we interrogated expression of all 7 ionotropic ATP receptors in lung fibroblasts by single cell scRNAseq. We found that, among the detected ATP ionotropic receptors, P2rx4 was the most highly expressed in nearly all populations of collagen-producing cells as measured by scRNAseq or qPCR ( Figures 3A–C ); notably, the scRNAseq data indicated that P2rx4 was increased with bleomycin injury and in IPF lung samples relative to control. The increases in P2rx4 were the most pronounced in the cellular subclusters previously found (16) to express the highest levels of pathologic extracellular matrix proteins (clusters 8 in mice and 3 in humans).

Given the high expression of P2rx4 both at steady state and its even higher expression in pathologic fibrosis, we next tested the effect of fibroblast-specific P2rx4 deletion by crossing a P2rx4 floxed allele that we previously generated (15) to the Col1a2-CreERT2 allele (13). In the entire lung fibroblast population isolated from bleomycin-injured mice, there was a modest decrease in P2rx4 ( Figure 4A ); nonetheless, possibly because of focused expression of this collagen promoter-driven Cre in fibroblasts with high collagen production, fibroblasts from Col1a2-CreERT2:P2rx4 fl/fl mice had a pronounced decrease in collagen gene expression after bleomycin injury compared to wild type ( Figure 4A ). Conversely, collagen genes were induced by treatment of primary human and mouse lung fibroblasts with ATPγS (a non-hydrolyzable form of ATP; Figure 4B ). Importantly, fibroblast-specific P2rx4 KO mice using two different Cre drivers, the Col1a2-CreERT2 and the Pdgfrb-Cre, had decreased lung collagen after bleomycin injury compared to wild type ( Figure 4C ). Moreover, we confirmed that P2rx4 signals in fibroblasts via p38 MAP kinase ( Figure 4D ), which has previously found to be necessary for P2rx4 signaling in other systems (26, 27). Taken together, these results suggest a model of ATP signaling via fibroblast P2rx4 as a positive regulator of tissue fibrosis after injury ( Figure 4E ).

How macrophages regulate tissular remodeling in homeostasis and injury is an area of active investigation (28), and much of the focus of the recent literature has been on the ontogeny of macrophages that subserve various functions in pathologic states. A major emerging theme is the role of moMacs, which support a range of pathologic phenotypes in multiple disease models (29, 30). Lung fibrosis is no exception, with monocyte-lineage cells being shown to be associated with poor outcomes in multiple clinical cohorts (31) and in mouse models including our own prior work showing a pro-fibrotic effect of moMacs (1–3, 6). This pathologic effect of macrophages has been proposed to be related to macrophage-derived factors that both support fibroblast growth and lead to fibroblast activation, including TGFβ, PDGF, and other secreted factors (4–6).

Connexin hemichannel-mediated ATP efflux as a paracrine signaling mechanism has been described in multiple contexts (10, 11, 32). Here we found that Cx43 was necessary for lung moMac ATP efflux after bleomycin injury. Furthermore, deletion of Cx43 in moMacs decreased fibrosis, and we noted decreased cytosolic calcium in fibroblasts co-cultured with Cx43 KO moMacs. Both human and mouse lung fibroblasts had increased expression of collagen in response to ATPγS treatment. Thus, we tested the role in lung fibrosis of ATP receptor P2rx4, which we found was the most highly expressed ionotropic ATP receptor in both murine and human lung fibroblasts. Furthermore, fibroblast P2rx4 was upregulated in lung fibrosis and has previously been found both to have a pro-fibrotic role and to be necessary for fibroblast calcium in liver fibrosis (27). Using two different fibroblast Cre drivers, we found that fibroblast-specific deletion of P2rx4 decreased lung collagen in mice, both at the cellular and whole-lung levels. Taken together, these results motivate enthusiasm for therapeutic development of inhibitors to P2rx4, some of which exist but are of relatively low potency (33). Moreover, they support a novel hypothesis of profibrotic macrophage-fibroblast interaction based on macrophage efflux of ATP via connexin hemichannels.

Our study has several limitations. While the finding of elevated fibroblast cytosolic calcium dependent on macrophage co-culture was acquired in vitro, importantly, we also observed injury-induced calcium elevation in fibroblasts in precision-cut lung slices, a closer reflection of the biology of intact tissue. However, future studies should seek to confirm specifically within the context of lung slice imaging the dependence of this fibroblast calcium elevation on macrophages and on extracellular ATP, since other paracrine or cell autonomous factors may be relevant. Furthermore, our in vitro data do not rule out the possibility that macrophage Cx43 could have other effects separate from ATP efflux, including gap junctional communication or secretion of other profibrotic mediators that may regulate cytosolic calcium in adjacent fibroblasts in a paracrine manner. We add that it remains to be determined what initiates ATP efflux by macrophages, and whether the resolution of fibrosis results from downregulation of this signaling. The use of two Cre drivers for our analysis of the effects of fibroblast P2rx4 also deserves comment. We observed modest P2rx4 knockout in fibroblasts in aggregate with the Col1a2-CreERT2, which is likely due to mosaic expression with this driver, although it was reassuring to find consistent effects with the Pdgfrb-Cre, which achieved more marked knockout in fibroblasts in aggregate. Finally, the pathways downstream of P2rx4 that drive expression of collagen genes in fibroblasts, including the role of calcium and p38 MAP kinase signaling, will require further elucidation. Despite these limitations, our results reveal a novel role for macrophages in paracrine regulation of fibroblast calcium, as well as a specific dependency on Cx43, and raise the profile of P2rx4 as a potential therapeutic target for lung fibrosis.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article.

The studies involving human participants were reviewed and approved by UCSF Institutional Review Board. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by UCSF Institutional Animal Care and Use Program.

AB, PT, and PY performed all experiments under the guidance of MB. Imaging experiments were performed by PT and AB under the direction of JJ, X-ZT, and CA. KB and TC prepared breeding and experimental stocks of genetically modified mice under the guidance of PT and MB. RM contributed P2rx4 floxed mice. TT performed analysis of single cell RNA-seq data under the supervision of DS. RS and SN isolated and provided human lung fibroblasts for analysis. MB conceived of the work, supervised experimental planning and execution, and wrote the manuscript with input from AB, PT and RM. All authors contributed to the article and approved the submitted version.

This work was supported by the US Department of Defense (W81XWH2110417 to MB), the National Heart, Lung, And Blood Institute (DP2HL117752 to CA), the National Institute Of Allergy And Infectious Diseases (R21AI130495 to CA), the UCSF Cardiovascular Research Institute, and the UCSF Sandler Asthma Basic Research Center.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.