3D in vitro M2 macrophage model to mimic modulation of tissue repair

Establish 3D in vitro model and macrophage morphology

Current wound-healing models are based on 2D tissue culture plastic, lacking many biophysical and biochemical features found in native wound tissue. In this work, we aimed to establish a biomimetic 3D cell culture models to replicate modulation of the initiation and resolution phases of the tissue repair. Our cell culture model facilitates the development of well-defined wound-healing models for high-throughput screening in drug development, studying of cell–cell and cell–matrix interactions, as well as other biomedical studies. We utilized 3D collagen matrices as a 3D cell culture model. Our 3D collagen matrices reconstituted at 2 mg/ml collagen concentration, exhibit a pore size of 8.1 ± 0.8 µm, a fibril diameter of 652.0 ± 32.0 nm, and an elastic modulus of 85.5 ± 7.3 Pa.

As high-throughput studies require large number of cells and to overcome donor-to-donor variation of primary cells, we utilized THP-1-derived macrophages as a cell model. To generate M2 macrophage subtypes, we first differentiated embedded THP-1 cells into uncommitted macrophages using PMA and further activated them using interleukin-4 (IL-4)/IL-13 and IL-10 into M_IL-4/IL-13 and M_IL-10, respectively (Fig. 1a). After macrophage activation, we quantified cell morphology by means of measuring cell area (Fig. 1c) and cell aspect ratio (Fig. 1d) using an automated image analysis toolbox³⁸. Both M_IL-4/IL-13 and M_IL-10 cells exhibited a round morphological appearance, as visualized in Fig. 1b and quantified by cell aspect ratio (Fig. 1d), no significant difference could be found between both macrophage subtypes. Macrophages are distributed homogeneously throughout 3D collagen matrices (Supplementary Fig. 1). For further characterization, we comprehensively studied cell surface markers and cytokine secretion profile. Later, both activated macrophages subtypes were co-cultured with fibroblasts and myofibroblasts to demonstrate differential cellular functions. Specifically, we aim to demonstrate the advantage of our 3D cell culture models for in-depth study of collagen reorganization during the initiation and resolution phases of the tissue repair.

Establish and characterization of M_IL-4/IL-13 and M_IL-10 macrophage

To distinguish between the macrophage subtypes used in this study, we first characterized both M_IL-4/IL-13 and M_IL-10 cells using their surface marker profile. The expression of CD14, CD68, CD80, CD86, CD105, CD163, CD206, and HLA-DR was measured via immunocytostaining using flow cytometry and quantified by geometric mean fluorescence intensity (gMFI) (Fig. 2). Histogram plots of quantified surface markers are depicted in Supplementary Fig. 2.

As shown in Fig. 2, both M2 subtypes shared similar expression levels of CD68 and CD86. Since CD68 is a pan-macrophage marker³⁹, it is not surprising that a recent report suggested expression level of CD68 increased upon differentiation in THP-1-derived macrophages, as compared to undifferentiated cell⁴⁰. We further showed that THP-1-derived macrophages activated towards pro-inflammatory subtype and treated using IL-4/IL-13 showed no difference in CD68 expression⁴⁰. CD86, on the other hand, is reported to be highly expressed in pro-inflammatory macrophages³⁹ and hence used to discriminate M2a and M2c from pro-inflammatory macrophages⁴¹.

M_IL-4/IL-13 exhibited higher expression levels of CD105, CD206, and HLA-DR when compared to M_IL-10. HLA-DR is generally expressed approximately 5- to 10-fold higher in pro-inflammatory macrophages when compared to anti-inflammatory macrophages for both PBMC- and THP-1-derived macrophages^27,42. In this work, we found a higher expression of HLA-DR in M_IL-4/IL-13 when compared to M_IL-10 macrophages, suggesting a higher pro-inflammatory activity of M_IL-4/IL-13. As for the expression of CD105 and CD206, it is reported that both these surface markers can be induced by IL-4/IL-13 in both PBMC- and THP-1-derived macrophages^24,25, explaining the high expression of CD105 and CD206 measured in our M_IL-4/IL-13. We previously showed that THP-1-derived pro-inflammatory macrophages and treated with IL-4/IL-13 showed similar levels of CD206.

M_IL-10 expressed high CD14, CD80, and CD163. CD14 is known used as an M2c-specific marker³¹. It has been showed that IL-10 enhanced CD14 expression⁴³. In general, CD80 surface markers have been reported to be highly expressed in pro-inflammatory macrophages^44,45. However, a recent report demonstrated that the presence of CD80 in anti-inflammatory macrophages is involved in the differentiation of regulatory T cells which play an important role in anti-rejection of transplants and maintaining tolerance^46,47. CD163 is identified as an anti-inflammatory receptor, which expression can be induced by IL-10 (ref. ⁴⁸). The high expression of CD14, CD80, and CD163 suggest an advanced suppressive role of pro-inflammatory response^49,50.

To further characterize both M2 phenotypes, we studied the cytokine secretion profile using a multiplex bead-based immunoassay (Fig. 3). To ensure that IL-4, IL-10, and IL-13 were autologously secreted, and not absorbed and released by the collagen through our exogenous application during macrophage activation, collagen matrices were incubated with those cytokines and the release profile in the media was determined over 3 days. Additionally, matrices were digested using collagenase and cytokine levels were quantified. Both soluble or matrix-bound IL-4, IL-10, and IL-13 were undetectable using the bead-based immunoassay, and therefore quantified cytokines were secreted by cells.

As shown in Fig. 3, we found significantly higher secretion levels of interferon-γ (IFN-γ), IL-2, IL-4, IL-8, IL-12p70, MCP-1, and TGF-β1 in M_IL-4/IL-13 when compared to M_IL-10. Both cells secreted similar levels of IL-1β, IL-6, tumor necrosis factor-ɑ (TNF-ɑ), and IP-10. In general, IFN-γ, IL-1β, IL-6, IL-12p70, IP-10, and TNF-ɑ are pro-inflammatory cytokines secreted by macrophages during the early wound-healing process⁵¹; however, they are generally expressed by M1 macrophages^18,40,52. Especially, IFN-γ, IL-2, IL-4, and IL-12 have been associated with accelerated tissue repair^53,54 and reported to contribute to effective wound healing. Any dysregulation in the secretion of those cytokines can lead to tissue fibrosis^55,56,57,58. In addition to pro-inflammatory cytokines, M_IL-4/IL-13 secreted a high amount of TGF-β1, a well-known multifunctional cytokine involved in the process of fibroblast differentiation. It has been demonstrated that TGF-β1 is secreted by macrophages activated with IL-4/IL-13 in both PBMC and THP-1-derived macrophages^18,30. The secretion of TGF-β1 also supports the maintenance of M2a phenotype⁵⁹.

On the other hand, M_IL-10 produced high amounts of IL-17 and IL-10 which were undetectable from M_IL-4/IL-13. While most analyzed pro-inflammatory cytokines have been demonstrated to enhance tissue repair, IL-17 is a pro-inflammatory cytokine that is involved in the activation of M2 macrophages via NFκB signaling⁶⁰ and the administration of recombinant IL-17A leads to delayed wound healing⁶¹, which might relate to stimulation of differentiation of human anti-inflammatory macrophages M2c in response to IL-10 (ref. ⁶²). IL-10 is the anti-inflammatory cytokine reported to be highly secreted in IL-10 activated PBMC-derived macrophages (M2c) in particular⁵². It is coupled with the regulation of the different phases during wound healing^{35,63,64,65,66}.

The cytokine secretion profile of our M_IL-4/IL-13 and M_IL-10 matches that of PBMC-derived M2a and M2c, respectively⁵². However, it will be incorrect to compare actual levels of secreted cytokines from both PBMC- and THP-1-derived macrophages due to experimental differences in cell number, activation protocols, cell culture dimensionality, biophysical and biochemical properties of cell culture substrate, and cell culture conditions. It should be reiterated that our cytokine analysis is based on activated macrophages in 3D cell culture models. We previously have reported that many cytokines, including IL-1, IL-6, IL-8, IL-10, IL-12, MCP-1, and TGF-β1, are secreted in collagen density-dependent manner by macrophages¹⁸. In that latter study, pro-inflammatory cytokines, IL-1β, IL-6, IL-8, and TNFα, were elevated in macrophages treated with LPS/IFNγ, namely pro-inflammatory macrophages as compared to M_IL-4/IL-13.

In sum, surface markers and cytokine secretion profiles suggest that M_IL-4/IL-13 and M_IL-10 activated macrophages demonstrate distinct M2 phenotypes, which are in line with previous findings on M2a and M2c macrophage subtypes in primary human macrophages^22,28,52, respectively.

Mimicking the early phase of tissue repair

To further differentiate between both M2 phenotypes and demonstrate their potential in regulating fibroblast differentiation during the early phase of tissue repair, human primary dermal fibroblasts were co-cultured with M2 phenotypes for 3 days and myofibroblast-related genes were analyzed using qPCR analysis (Fig. 4). Respectively, as negative and positive control of myofibroblast differentiation, fibroblasts were monocultured in the absence and presence of 10 ng/ml TGF-β1 for 3 days. TGF-β1 is known to trigger fibroblast differentiation in vitro^21,67. Fibroblasts cultured in the presence of TGF-β1 and co-cultured with our M_IL-4/IL-13 exhibited more pronounced actin stress fibers with αSMA incorporation as well as a more flattened cell morphology, suggesting a differentiation of fibroblast into myofibroblast (Fig. 4a). This could not be observed in untreated fibroblasts (control) and those co-cultured with M_IL-10.

To confirm the fibroblast differentiation, we quantified αSMA expression, a myofibroblast marker^8,68, using immunocytostaining (Fig. 4b) and gene expression (Fig. 4c) analysis. αSMA-positive cells and gene expression increased in fibroblast culture treated with TGF-β1 and when co-cultured with M_IL-4/IL-13, when compared to control and co-cultured with M_IL-10. Both αSMA-positive cells and gene expression correlated nicely. This result also corroborates with previous reports which demonstrate the effect of TGF-β1 (refs. ^8,66,69) and M_IL-4/IL-13 mediated fibroblast differentiation¹⁸. Besides αSMA expression, we analyzed matrix protein production potential, through analyzing alpha 1 chain of the type 1 collagen (Coll1a1) and fibronectin containing Extra Domain-A (EDA-FN). The expression of both Coll1a1 (Fig. 4d) and EDA-FN (Fig. 4e) follow the same trend of αSMA expression. Again, Coll1a1 and EDA-FN were highly expressed in fibroblast co-cultured with M_IL-4/IL-13 rather than TGF-β1-treated fibroblasts. Fibroblasts co-cultured with M_IL-10 did not show an enhanced expression and the expression level of both matrix proteins was in the same range as untreated fibroblasts.

In sum, by co-culturing fibroblasts with both macrophage subtypes, we found that only M_IL-4/IL-13 is able to differentiate fibroblast into myofibroblast, as demonstrated by αSMA and matrix protein expression. This could be explained by the higher TGF-β1 expression in M_IL-4/IL-13 when compared to M_IL-10 (Fig. 3). Interestingly, the expression levels of αSMA and matrix protein in fibroblast co-cultured with M_IL-4/IL-13 are significantly higher than fibroblast directly treated with TGF-β1, which is to our previous finding¹⁸. This variation could be explained by the sustained release of TGF-β1 by M_IL-4/IL-13. As demonstrated using controlled release microbeads, pulsed paracrine delivery of TGF-β1 in a picogram range (100 pg/ml) from polyethylene glycol (PEG) microbeads embedded in 3D collagen matrix could fully differentiate fibroblasts into myofibroblast similar to a continuous 10 ng/ml systematic delivery⁶⁹.

Besides production of matrix proteins, another major feature reserved for myofibroblasts is their ability to remodel the matrix. Collagen matrices were decellularized using osmotic pressure by incubating in bi-distilled water and stained using TAMRA-SE for visualization by confocal laser microscopy (Fig. 5a). Changes in matrix characteristics could be visually observed. The quantitative analysis of matrix porosity and fibril diameter using a custom-built image analysis toolbox demonstrated that matrices of TGF-β1 treated fibroblasts exhibit a higher collagen density than the untreated counterpart and matrices of fibroblast co-cultured with both M2 macrophage (Fig. 5b). Matrices of fibroblast co-cultured with M_IL-4/IL-13 showed significantly smaller pore size than when co-cultured with M_IL-10. Interestingly, an increase in collagen fibril diameter was only found in matrices of fibroblast co-cultured with M_IL-4/IL-13. This observation is consistent with the initial development of fibrosis, where an increase of collagen fibril can be found in pro-inflammatory milieu⁷⁰, which at low levels, our M_IL-4/IL-13 are able to secrete (Fig. 3). In addition, it has been reported that collagen fibril diameter can mechanically actuate myofibroblast differentiation⁷¹, which might also explain why fibroblast differentiation in co-culture with M_IL-4/IL-13 exhibited higher expression of αSMA (Fig. 4b, c) and matrix proteins (Fig. 4d, e).

Furthermore, we analyzed the elastic modulus of the remodeled matrices using a contactless rheometer (Fig. 5d). Matrices of untreated fibroblast and fibroblast co-culture with M_IL-10 showed a similar range of elastic modulus. Interestingly, matrices of fibroblast co-cultured with M_IL-4/IL-13 showed higher elastic modulus than that from TGF-β1-treated fibroblasts. As demonstrated in various works^71,72,73,74, the contribution of collagen fibril diameter on bulk matrix modulus is not negligible. For example, collagen matrices with bigger pore size and thicker fibrils demonstrated higher elastic modulus than matrices with smaller pore size and thinner fibrils^72,74. These previous findings provide an explanation for our obtained results of mechanical analysis.

Mimicking the resolution phase of tissue repair

While M_IL-4/IL-13 has shown their potential in initiating tissue repair, M_IL-10 has been hypothesized to terminate the tissue repair phase due to their low pro-inflammatory cytokine secretion and high secretion of IL-10 (Fig. 3). It has been reported that exogenously applied IL-10 is able to convert myofibroblasts into fibroblast^35,66. However, few studies have been performed in co-culture conditions to investigate this phenomenon³⁵. To address this, we first derived myofibroblasts by treating fibroblasts with 10 ng/ml TGF-β1 on 2D cell culture plastic for 3 days. Afterwards, myofibroblasts were detached and cultured on top of the collagen matrices pre-embedded with and without M2 subtypes. It has to be noted that the cell culture media of the myofibroblast in the control condition is TGF-β1 supplemented (10 ng/ml) to prevent myofibroblast apoptosis, as previously reported⁶⁶. It is known that TGF-β1 can block the intrinsic apoptotic pathway by inhibiting the pro-apoptotic protein BCL-2-associated death promoter (BAD) via the FAK–PI3K–AKT signaling pathway in fibroblast⁷⁵.

While myofibroblasts in monoculture exhibit pronounced actin stress fibers with αSMA incorporation (Fig. 6a), this observation is attenuated when IL-10 is present systemically. Quantitative analysis of αSMA-positive cells (Fig. 6b) and αSMA expression (Fig. 6c) confirmed the observation, along with reduction matrix proteins expression, namely Coll1a1 (Fig. 6d) and EDA-FN (Fig. 6e). In co-culture conditions, myofibroblastic phenotypes are maintained in the presence of M_IL-4/IL-13, while myofibroblasts co-cultured with M_IL-10 manifests a reduction in αSMA-positive cells and expression. The expression of matrix proteins was also reduced and correlated fittingly with the reduction of αSMA expression. Both results suggested the dedifferentiation of myofibroblast into fibroblast phenotypes when co-cultured with M_IL-10, which is in line with dedifferentiation of myofibroblasts by PBMC-derived anti-inflammatory macrophages using macrophage colony-stimulating factor (M-CSF)³⁵. To elucidate that myofibroblast dedifferentiation is related to the secretion of IL-10, we subsequently supplemented cell culture media with anti-IL-10 antibody (aIL-10). Results suggested that in the presence of aIL-10, the number of αSMA-positive cells and expression, as well as the expression of matrix proteins, remained at a similar level to control and co-culture with M_IL-4/IL-13. Our results are in line with the presence of M2c, as well as a detected IL-10 secretion in the late state of the tissue repair phase^35,66,76, which emphasized the importance of IL-10 in the myofibroblast dedifferentiation process, and hence, termination of the tissue repair phase.

The question arises whether de-differentiated myofibroblasts are able to remodel collagen matrices. To address this underlying question, we decellularized collagen matrices and analyzed their matrix porosity, fibril diameter, and matrix elasticity. Changes in matrix characteristics could be visually observed (Fig. 7a) and further quantitative analysis of matrix porosity showed that matrices of myofibroblasts treated with IL-10 and in co-culture with M_IL-10 were significantly higher than those found in the co-culture with M_IL-4/IL-13, and M_IL-10 in the presence of aIL-10 (Fig. 7b). By analyzing collagen fibril diameter, we found that fibril diameter increased only in co-cultured conditions with M_IL-4/IL-13, but not in monoculture and co-culture with M_IL-4/IL-13 and M_IL-10 with aIL-10 (Fig. 7b). This observation is similar to the one found in co-culture of fibroblast and M_IL-4/IL-13 (Fig. 5), which might be caused by the enhanced pro-inflammatory milieu, tissue stiffness, microRNA^77,78. Significantly lower matrix elasticity could be observed in myofibroblast treated with IL-10 and in co-culture with M_IL-10 (Fig. 7d). Together with αSMA expression and matrix production, our matrix remodeling data suggest the dedifferentiation of myofibroblasts into fibroblast phenotypes in the presence of IL-10 and M_IL-10. By blocking IL-10 using neutralizing antibody (aIL-10), we found that the expression of αSMA, matrix proteins, and the matrix remodeling remains at similar levels as myofibroblasts. It suggests that the dedifferentiation of myofibroblast is in response to the presence of supplemented or M_IL-10 secreted IL-10. Our data emphasize the importance of IL-10 on the reduction of matrix remodeling by initiation of myofibroblast dedifferentiation.

Fibroblast differentiation state is correlated with migratory behavior

Besides matrix remodeling, another advantage of the 3D cell culture models is they allow to study cell migration. Migration of fibroblast cells during different stages of wound healing is important for tissue repair, but not much experimental evidence exists on the migratory behavior of fibroblasts at different differentiation stages and in co-culture conditions. We performed a migration analysis of both fibroblasts and myofibroblasts in mono- and co-culture conditions to address this underlying problem. Representative side-view images of fibroblast (Fig. 8a) and myofibroblast (Fig. 8d) show that fibroblast migration into 3D collagen matrices is significantly higher than that of myofibroblasts (TGF-β1-treated conditions), as required during early tissue repair phase. This data corroborate with our previous findings^8,66. In addition, fibroblast co-culture with M_IL-10 exhibited a similar migration capacity in both percentage of migrating cells (Fig. 8b) and maximal migration depth (Fig. 8c) when compared to untreated fibroblasts, but exhibited significantly higher migration when compared to myofibroblast and in co-culture with M_IL-4/IL-13. In another experiment, modeling the late tissue repair phase, where pre-differentiated myofibroblasts were seeded onto 3D collagen matrices, we found that myofibroblasts showed limited migration into collagen matrices (Fig. 8d). However, an enhanced migration of myofibroblast in the presence of IL-10 and in co-culture with M_IL-10 could be observed both in the migrating cell number (Fig. 8e) and maximal migration depth (Fig. 8f). Myofibroblasts co-culture with M_IL-4/IL-13 and M_IL-10 in the presence of aIL-10 showed similar migration patterns as those found in myofibroblast alone.

Overall, our data suggest that myofibroblast exhibits limited migration capacity when compared to fibroblast, which is well correlated with the high expression of αSMA and the contractile phenotypes, as previously published^8,17,79. The dedifferentiation of myofibroblast into fibroblast phenotypes in the presence of IL-10 and in co-culture with M_IL-10 could be confirmed by migratory behavior. In addition, by neutralizing IL-10 using antibody in the co-culture with M_IL-10, the migratory behavior remains at the level of myofibroblasts, confirming the effect of IL-10 in myofibroblast dedifferentiation.

General discussion and conclusion

To establish wound-healing models for high-throughput assessment of pathobiological changes and testing of anti-fibrotic drugs, well-defined biosystems, including cells and their cell culture substrate, are required. Although primary human cells have become a major interest in the field of regeneration, they cannot be utilized to establish a well-defined wound model due to patient-to-patient variability as well as possible lack of data regarding their underlying activation stage. To overcome these issues, we used THP-1, a human monocytic cell line widely used as an established macrophages model in many biomedical studies such as wound repair, infection models, and immuno-oncology⁸⁰. Although THP-1 based M1-like macrophage models have been extensively established, few studies attempted to study M2a- and M2c-like macrophages and address their cellular functions in the context of wound repair. In this work, we utilized collagen matrices as a biomimetic wound model and comprehensively analyzed and demonstrated that surface markers and cytokines secretion profile of M_IL-4/IL-13 and M_IL-10 is akin to M2a and M2c macrophages derived from human PBMC, respectively^22,28,52. We showed that M_IL-4/IL-13 can modulate tissue repair by controlled secretion of TGF-β1 to induce fibroblast differentiation, while M_IL-10 macrophages secrete high amounts of IL-10 to resolve inflammation and tissue repair processes. Besides, we demonstrate that IL-10 can reverse myofibroblast into fibroblast phenotypes, as published elsewhere^35,66. By neutralizing IL-10 with antibody in co-culture with M_IL-10, no dedifferentiation of myofibroblast could be observed, emphasizing the role of IL-10 in resolution of the tissue repair phase. An illustration summary of our results in different tissue repair phases is depicted in Fig. 9. Interestingly, we found an increase in collagen fibril diameter only in the co-culture of myofibroblasts and M_IL-4/IL-13 (M2a). This suggests that the contribution of M_IL-4/IL-13 to the tissue repair process is through remodeling provisional wound tissue²². Our recent report demonstrates that thicker collagen fibrils could activate fibroblast differentiation via cell contractility⁷¹, advancing the development of tissue scarring and fibrosis. Alongside with that, a prolonged presence of M2a might also contribute to fibrosis^81,82. Overall, our results pinpoint the importance of the co-culture model of fibroblast and macrophages for biomimetic wound healing, instead of fibroblast monoculture. In addition, our established biomimetic model can guide the development of well-defined high-throughput platforms for improving tissue healing and anti-fibrotic drugs testing, as well as other biomedical studies.