Local apoptosis promotes collagen production by monocyte-derived cells in transforming growth factor β1-induced lung fibrosis

The CD14⁺ fraction of peripheral blood contains heterogeneous monocyte progenitors with important roles in tissue injury and repair [1]. A subpopulation of monocytes differentiates into fibrocytes by acquiring a fibroblast-like morphology, gaining expression of collagen I and CD34 while losing CD14 expression [2]. Fibrocytes accumulate in transforming growth factor (TGF)-β1-exposed tissues [3] and are associated with an array of fibrosing disorders including asthma, pulmonary fibrosis, and scleroderma [4–6]. Due to the considerable variability in methods used to identify these cells, controversy exists as to their true phenotype [7, 8] though the presence of fibrocytes in several forms of fibrosis is now well established. The mechanism(s) through which fibrocytes and related CD45⁺ collagen (Col)-I⁺ cells contribute to fibrosis remain unclear, but may be related to immunological regulation of effector cell phenotypes [9] as well as direct production of extracellular matrix proteins or α-smooth muscle actin (SMA) production [10, 11]. This phenotype is specialized for the characteristics that might be required for repair. However, while the administration of human fibrocytes to severe combined immunodeficiency (SCID) mice requires coadministration of bleomycin to result in pathology [12], requirement for injury in the accumulation of CD45⁺Col-I⁺ in the TGF-β1-exposed murine lung has not been shown.

Pulmonary fibrosis is a progressive and often fatal disease for which there are no effective therapies. The current paradigm of pulmonary fibrosis pathogenesis includes recurrent epithelial cell death responses with subsequent recruitment of a monocyte-derived inflammatory infiltrate and the eventual development of myofibroblast activation [13]. These events are believed to be heavily influenced by TGF-β1 [14–17]. While the precise type of injury initiating these events remains unknown, substantial evidence supports a role for apoptosis as a contributing factor [18–20]. Elevations in circulating and/or tissue CD45⁺Col-I⁺ cells have are seen in a broad array of fibrosing lung diseases including idiopathic pulmonary fibrosis (IPF) [4], asthma [5], post-transplant bronchiolitis obliterans syndrome [21], and scleroderma [6]. Many of these diseases are associated with abnormalities in apoptosis [19, 22, 23]; however, a relationship between CD45⁺Col-I⁺ cells, specifically fibrocytes, and apoptosis has not been previously assessed.

We have recently shown that transgenic overexpression of TGF-β1 results in the accumulation of cells that coexpress CD45 and Col-Iα1 [24]. However, the cell surface phenotype of these cells remains unexplored and the local events initiating TGF-β1-induced accumulation of CD45⁺Col-Iα1⁺ cells remain obscure. Because the TGF-β1 phenotype requires apoptosis for the development of fibrosis and remodeling [18] we thought it likely that the increase in CD45⁺Col-Iα1⁺ cells seen in this model were caused by increases in this form of cell death. To test this hypothesis we explored the identity of CD45⁺Col-Iα1⁺ cells in a mouse model of pulmonary fibrosis caused by transgenic overexpression of the bioactive human TGF-β1 gene and examined whether caspase-mediated apoptotic responses are required for the appearance of these cells. The human relevance of these findings was explored in studies of cultured cells obtained from patients with multiple forms of pulmonary fibrosis. Our results indicate that CD45⁺Col-Iα1⁺ cells recruited to the lung by TGF-β1 are enriched for the expression of CD14 and that their appearance in the lung requires an increase in apoptotic cell death responses. Our data also demonstrate that CD14⁺ monocytes derived from the circulation of patients with multiple forms of lung fibrosis show robust CD34 expression and display a propensity for collagen production that is reduced when apoptosis is blocked.

These studies provide new insight into the relationship of collagen-producing leucocytes and fibrotic lung disease. Specifically, they demonstrate that lung-targeted overexpression of TGF-β1 induces the intrapulmonary accumulation of a heterogeneous population of collagen-containing leucocytes, many of which express a cell surface phenotype characteristic of monocytes but appear to be distinct from alternatively activated macrophages. Furthermore, inhibition of cellular apoptosis results in a significant reduction in all of these populations and restores the CD45⁺Col-Iα⁺ cell surface phenotype seen in wild-type mice. The human relevance of these findings is demonstrated by recapitulation of these results in the lungs and circulation of patients with two separate forms of fibrotic lung disease. Taken together, these data suggest that in the setting of apoptotic injury, monocytes adopt a reparative program characterized by enhanced production of collagen I.

The identity of the collagen producing leucocytes in our study is not entirely clear at this time but based on the robust expression of CD34 seen the cultured human cells, these cells are likely to be fibrocytes in intermediate state of differentiation. Fibrocytes were first described as blood-borne, fibroblast-like cells that appeared in exudative fluid at the earliest phases of wound repair [2]. They are considered to originate from CD14⁺ myeloid cells and coexpress collagen I, CD45, and the progenitor marker CD34 though this latter marker is downregulated as these cells mature in situ[28]. CD34 is also lost on human fibrocytes during in vitro culture in the setting of TGF-β1 [12] suggesting that CD34 may be an early fibrocyte marker which is lost as the cell matures or is activated or that, as is seen in other settings [29, 30], TGF-β1 exposure preferentially impedes the proliferation and survival of CD34⁺ cells. Our findings support this paradigm and are consistent with prior reports in murine models of both asthma [28] and bleomycin induced lung fibrosis [12] showing that CD34⁺ is rapidly lost upon entry into diseased tissue. Fibrocytes demonstrate a remarkably plastic phenotype, adopting the functional characteristics of both macrophages and fibroblasts in response to local cues. Thus, if the CD14⁺ collagen expressing cells seen in our model are indeed fibrocytes, it is possible that their contributions to disease in this setting would include the fibrocyte functions typically attributed to macrophages such as cytokine and chemokine production [9], antigen presentation [31], inflammatory cell trafficking and activation [9] and promotion of angiogenesis [32], as well as the extracellular matrix (ECM)-producing and wound contractile properties typically attributed to activated myofibroblasts. Further studies investigating the role of collagen-producing leukocytes in lung fibrosis could be pursued in studies which live collagen-producing cells are isolated for in vitro studies and functional analysis. In addition, while it is assumed that these cells are producing collagen, it remains possible that the CD14⁺ cells have engulfed collagen. This question highlights the need for the development of high affinity antibodies that detect the immature form of murine collagen.

It is also noteworthy that while our human findings recapitulated much of what was seen in the mouse, there were a number of differences. For example, CD45⁺Col-Iα1⁺ cells in the TGF-β1-exposed lung were enriched for CD14 expression in the absence of CD34. In contrast, while the CD45⁺Pro-Col-Iα1⁺ cells obtained from obtained from patients with lung disease also expressed high levels of CD14, many of these cells also expressed CD34. These data are could be related to the differences in tissue compartments studied (blood versus lung), the use of a transgenic TGF-β1 modeling system in the mouse, and/or unaccounted for clinical factors in the patients such as age and comorbidities. However, because we did not further phenotype the intrapulmonary CD45⁺Pro-Col-Iα1⁺ cells from humans, comparison with the animal findings are limited at best. The mechanistic significance of CD34 on collagen-producing human leukocytes could perhaps best be examined in studies in which these cells are subdivided based on CD34 expression and then adoptively transferred into murine models of experimentally induced lung fibrosis.

In our studies, the in vivo inhibition of apoptosis in two different cell types, murine lung epithelia and human peripheral blood monocytes, significantly reduced the appearance CD45⁺Pro-Col-Iα1⁺ cells, suggesting that this phenotypic change is a non-specific response to local cell death. Additional investigation will be required to determine why apoptosis is required for maximal accumulation of CD45⁺Col-Iα1⁺ cells. Given the well documented effects of apoptotic bodies on monocyte biology [33, 34], it is possible that the increased production of collagen by monocytes is a direct response to exposure to dead/dying cells as has been previously shown in elegant studies of cultured murine monocytes [33]. This hypothesis is further supported by data from studies demonstrating a reduction in CD45⁺Col-Iα1⁺ cells upon exposure to the short pentraxin serum amyloid P, which modulates monocyte phenotypes in response to engulfment of apoptotic cells [35, 36]. It is also possible that other cell populations such as lymphocytes or other monocyte derived cells respond to apoptosis by increasing the secretion of soluble mediators, such as semaphorin 7a [25], Stromal Derived Factor (SDF-1) [4] and Monocyte Chemotactic Protein-1 (MCP1) [37] that could promote the appearance of fibrocytes. In addition, caspase activation itself could induce monocyte abnormalities that lead to enhanced production of collagen as a form of immunosenesence. The documentation of elevated levels of CD45⁺Col-Iα1⁺ cells in the senescence associated mouse [38], as well as in CD14⁺ cells derived from aged but otherwise healthy humans [6], supports this latter hypothesis. Further work is needed to define the precise relationship between caspase activation, apoptosis, and the accumulation of CD45⁺Pro-Col-Iα⁺ cells in the TGF-β1-exposed lung and in patients with pulmonary fibrosis.

Our studies also provide novel insight into the relationship between CD45⁺Col-Iα1⁺ cells and CD206+ macrophages. We have previously shown that TGF-β1 induced lung fibrosis is dependent upon M2 macrophage accumulation [3]. In the current study we find that apoptosis is required for the appearance of CD45+Col-Iα cells but has a lesser effect on macrophages. Because CD206 is a robust marker of alternative activation [39], these data suggest that accumulation of M2 macrophages alone is insufficient for the development of TGF-β1-induced fibrosis and remodeling. When viewed in combination, these studies support a paradigm in which the profibrotic effects of TGF-β1 require both alternatively activated macrophages and collagen-producing leucocytes for maximal effect. The functional contributions of these populations will require further investigation.

In summary, our studies demonstrate that local apoptotic responses potently stimulate the recruitment of collagen containing leucocytes in the TGF-β1-exposed murine lung. These CD45⁺Col-Iα1⁺ cells exhibit significant phenotypic heterogeneity and appear in response to apoptotic cell death. These effects are seen in monocytes derived from patients with two separate forms of fibrotic lung disease, as well as in monocytes obtained from normal controls. These findings suggest that targeting apoptotic responses in an effort to attenuate collagen production by monocytes and the accumulation of fibrocytes may be beneficial in diseases of lung remodeling and aberrant repair.