Mesenchymal stem cells inhibit cutaneous radiation-induced fibrosis by suppressing chronic inflammation


  • Author contributions: J.H.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; K.H. and B.S.: collection and/or assembly of data and manuscript writing; E.J.C.: collection and/or assembly of data and data analysis and interpretation; A.W.: collection and/or assembly of data; J.F.B.: data analysis and interpretation and manuscript writing; D.C.: conception and design, financial support, data analysis and interpretation, manuscript writing, and final approval of manuscript.

Correspondence: Deborah E. Citrin, M.D., Section of Translational Radiation Oncology, Radiation Oncology Branch, Center for Cancer Research, The National Cancer Institute, 10-CRC—Hatfield Clinical Research Center, B2–3500, 10 Center Drive, Bethesda, Maryland 20892-1682, USA. Telephone: 301–496-5457; Fax: 301–480-5439; e-mail:


Exposure to ionizing radiation (IR) can result in the development of cutaneous fibrosis, for which few therapeutic options exist. We tested the hypothesis that bone marrow-derived mesenchymal stem cells (BMSC) would favorably alter the progression of IR-induced fibrosis. We found that a systemic infusion of BMSC from syngeneic or allogeneic donors reduced skin contracture, thickening, and collagen deposition in a murine model. Transcriptional profiling with a fibrosis-targeted assay demonstrated increased expression of interleukin-10 (IL-10) and decreased expression of IL-1β in the irradiated skin of mice 14 days after receiving BMSC. Similarly, immunoassay studies demonstrated durable alteration of these and several additional inflammatory mediators. Immunohistochemical studies revealed a reduction in infiltration of proinflammatory classically activated CD80+ macrophages and increased numbers of anti-inflammatory regulatory CD163+ macrophages in irradiated skin of BMSC-treated mice. In vitro coculture experiments confirmed that BMSC induce expression of IL-10 by activated macrophages, suggesting polarization toward a regulatory phenotype. Furthermore, we demonstrated that tumor necrosis factor-receptor 2 (TNF-R2) mediates IL-10 production and transition toward a regulatory phenotype during coculture with BMSC. Taken together, these data demonstrate that systemic infusion of BMSC can durably alter the progression of radiation-induced fibrosis by altering macrophage phenotype and suppressing local inflammation in a TNF-R2-dependent fashion. Stem Cells 2013;31:2231–2241


Radiation is a commonly used therapy for solid tumors, but exposure of adjacent normal tissues may lead to injury. Acute skin injury from radiation is characterized by erythema, alopecia, and dermatitis, which are typically self-limiting. In contrast, late radiation skin injury results in irreversible epidermal hyperplasia, capillary disorganization, chronic inflammation, and dermal thickening due to collagen accumulation [1, 2]. The resulting limitations in range of motion and cosmetic deficits can negatively impact quality of life for cancer survivors.

The etiology of cutaneous radiation-induced fibrosis remains incompletely understood. It is proposed that fibrosis may result from an unrestrained repair program, suggesting that local dysregulation of immune cell activity may underlie this pathology [3]. Despite trials of numerous immunomodulatory agents as mitigants of fibrosis, very few durable successes have been realized to date [1].

Bone marrow-derived mesenchymal stem cells (BMSC), also referred to as bone marrow stromal cells, are a multipotential lineage characterized by a capacity for extracorporeal expansion and the ability to differentiate into bone, cartilage, and adipose tissues [4]. Mesenchymal stem cells appear to function as potent immunomodulators [5]. Furthermore, allogeneic or xenogeneic BMSC do not elicit an immune rejection response presumably due to low surface expression of major histocompatibility complex class II (MHC-II) molecules [6]. These features have led to great interest in their use as a cell-based immunomodulatory therapy. Early results from preclinical models have shown that BMSCs are potent effectors of immunosuppression via T-cell suppression [7], and/or inhibiting inflammatory cytokine release by macrophages (MΦ), and promoting their transition toward an interleukin-10 (IL-10) secreting regulatory phenotype [8]. Several clinical trials assessing the efficacy of BMSC in immune-mediated diseases are currently underway [9-11].

Given the complex inflammatory milieu associated with radiation injury, it is possible that anti-inflammatory BMSC therapy would be beneficial. We hypothesized that systemic infusion of BMSC would ameliorate the progression of radiation-induced fibrosis. Using a well-characterized murine model, we demonstrated that systemic delivery of BMSC after resolution of acute dermatitis was sufficient to mitigate fibrosis. Through transcriptional profiling and protein expression assays, we found that BMSC induce IL-10 production by MΦ, and reduce the levels of proinflammatory cytokines. Finally, we used cell-based assays to elucidate the mechanisms by which BMSC modulate the inflammatory phenotype of MΦ. We demonstrated that BMSC modulation of MΦ IL-10 secretion requires tumor necrosis factor-receptor 2 (TNF-R2) signaling. Taken together, these results suggest that BMSC may represent a new therapeutic avenue for treatment of radiation-induced skin fibrosis.

Materials and Methods

Mouse Strains

All procedures using animal subjects were performed under an institutionally approved protocol deemed in accordance with the guidelines of the Institute of Laboratory Animal Resources, National Research Council. C3H/HeN mice were obtained from Frederick National Laboratory (Frederick, MD; C57BL/6-Tg(UBC-GFP)30Scha/J, B6.129S2-Tnfrsf1btm1Mwm/J, and c57BL6/J mice were purchased from Jackson Laboratories (Bar Harbor, ME; All animals were housed in a specific-pathogen-free environment with 12-hour photoperiods and ad libitum access to standard chow and water.

Murine Model of Cutaneous Radiation-Induced Fibrosis

We used a well-described murine model of cutaneous radiation-induced fibrosis for our experiments [12, 13]. Briefly, 10-week-old female C3H/HeN mice were immobilized in a custom lucite jig, and the right hind leg was irradiated to 35 Gy at a dose rate of 2.6 Gy/minute using an X-RAD 320 x-ray irradiator (Precision X-Ray, North Branford, CT, operating at 320 kVp with 2.0 mm aluminum filtration. Beam collimation and lead shielding were used to protect the remainder of the body from irradiation, including the contralateral limb, which was used as an internal control. All radiation experiments were completed in groups of at least 10 mice. Acute skin injury was assessed using a modification of a previously described phenotypic scale, where 1 = normal, 2 = hair loss, 3 = erythema, 4 = dry desquamation, 5 = <30% area moist desquamation, and 6 = >30% area moist desquamation [14]. Skin contracture was serially assessed at 90, 120, and 150 days after irradiation by measuring passive extension of the irradiated and contralateral hind limbs as described previously [12].

Additional groups of mice were treated in an identical manner to provide tissue for histologic analysis and biochemical assay at select time points following irradiation. For histologic studies, skin was spread on filter paper, fixed in formalin, and embedded in paraffin. For biochemical assays, tissues were cleaned of adherent fascia and adipose tissue, snap frozen, and stored at −80°C.

Isolation, Expansion, and Infusion of Primary BMSCs

All cell culture media and reagents were obtained from Life Technologies (Carlsbad, CA, unless otherwise noted. Bone marrow was collected by flushing bilateral tibiae and femora of 4–6-week-old mice with 3 ml/bone ice-cold Hank's buffered salt solution supplemented with 2% fetal calf serum (FCS), and 1× antibiotic-antimycotic solution (AAS). BMSC were expanded for two passages in αMEM media supplemented with 4 mM l-glutamine, 1× AAS, 100 μM 2-mercaptoethanol, 10 nM dexamethasone (Sigma, St. Louis, MO;, 100 μM ascorbate (Sigma), and lot-selected FCS (20% [v/v], Gemini Biosciences, West Sacramento, CA; After the second passage, hematopoietic contaminants were depleted with magnetic cell sorting (Lineage Depletion Kit, Miltenyi Biotec Inc. Auburn, CA; Flow cytometry was performed to assess cell surface expression of Sca-1, CD11b, CD29, CD44, CD45, and CD106 (Mouse Mesenchymal Stem Cell Marker Kit, R&D Systems, Minneapolis, MN; Multipotentiality was demonstrated by inducing adipogenic, chondrogenic, and osteogenic differentiation according to protocols described previously [15].

Twenty-four hours after cell sorting, BMSC were harvested by trypsinization and resuspended in ice-cold phosphate buffered saline (PBS) with 100 U/ml heparin. Six weeks post-ionizing radiation (IR), mice were randomized to receive infusion of 5 × 105 BMSC or the heparinized saline vehicle (200 μl) via the lateral tail vein. Infused cells were derived from male donors of C3H/HeN (syngeneic/sBMSC) or C57BL/6-Tg(UBC-GFP)30Scha/J (allogeneic/aBMSC) origin.

Histology and Immunohistochemistry

Masson trichrome staining (Sigma) was performed to assess skin thickness and the extent of collagen deposition. Slides were examined by light microscopy and digitally captured at 20× magnification. Image analysis was performed with ImageJ software v1.46e (National Institutes of Health, Bethesda, MD; Ten measurements of skin thickness were taken at uniform intervals across three randomly selected 20× fields for each tissue sample, with thickness measurements extending from the external surface of the epidermis to the dermal-subcutaneous interface.

Immunohistochemistry was performed to assess inflammatory cell infiltration with the Impress System (Vector Labs, Burlingame, CA; Following antigen retrieval, sections were incubated in primary antibodies against CD45 (R&D Systems), green fluorescent protein (GFP), CD68, CD80/B7.1, and CD163 (AbCam, Cambridge, MA;, followed by incubation with a horseradish peroxidase- or alkaline phosphatase-conjugated secondary antibodies. Immunoreactivity was visualized by diaminobenzidine (ImmPact, Vector Labs) or nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (Thermo Scientific, Rockford, IL; histochemistry. The number of infiltrating MΦ (CD68+), proinflammatory/classically activated MΦ (CD80/B7.1+), and anti-inflammatory/regulatory MΦ (CD163+) [16] was counted in at least three randomly selected 20× fields per mouse [17, 18].

Detection of BMSC in Irradiated Tissue

A real-time PCR-based assay was used to detect the presence of donor cells in tissues of mice receiving BMSC infusion. Genomic DNA was extracted from skin or blood samples by the DNEasy method (Qiagen, Valencia, CA,, and 0.5 μg aliquots of total DNA was used in PCR reactions. PCR reactions were run using Quantitect SybrGreen PCR reagents (Qiagen) and run on an Applied Biosystems 7500 real-time PCR system (Life Technologies). Genomic DNA from the male donor cells was amplified with primers directed against a Y-chromosomal sequence (F:5′-CCTGCATCCCAGCTGCTTG-3′;R:5′-CAGAATCCCAGCATGCAAAATACA-3′). A genomic DNA sequence located inthe noncoding region of Apolipoprotein-B locus was amplified using a primer set (F:5′-CGTGGGCTCCAGCATTCTA3′;R:5′-CGTGGGCTCCAGCATTCTA-3′) as an internal control. All PCR experiments were performed with appropriate controls, including no-template, Y-chromosomal negative, female DNA positive (Y-chromosomal negative control), gDNA from male donor mice (Y-chromosome positive control), and all samples were assayed in duplicate reactions. Reaction specificity was verified by melting-curve dissociation analysis and agarose gel electrophoresis. Samples were considered to be “Y-positive” when the following criteria were met: sample Ct value at least one cycle less than the Y-negative control; when a single melting peak was observed that was within ±0.5°C of the positive control; and PCR products migrated at the expected band size by agarose gel electrophoresis.

Gene Expression Analysis by Pathway-Focused RT-PCR Array

The expression levels of 84 genes associated with fibrosis were determined using the RT2-PCR array system (PAMM-120, SA Biosciences, Valencia, CA, Skin tissue was homogenized in Trizol reagent (Life Technologies) followed by purification with the RNEasy system (Qiagen). RNA was reverse-transcribed and amplified by PCR using the reagents recommended by the manufacturer. Normalized gene expression data were derived by the 2−ΔΔCT method using the RT2 Profiler PCR Array Data Analysis Suite (SA Biosciences). Gene expression values were normalized across samples and treatment groups by the geometric mean expression of the reference genes β-actin, HPRT, and GusB. Differences in gene expression were considered significant when p ≤ .05 by paired t test.

Assessment of Tissue Collagen Content and Inflammatory Cytokine Expression

To determine changes in collagen content, samples of skin tissue were weighed and digested in 0.5 M acetic acid/0.5% porcine gastric pepsin (Sigma) to release soluble collagen. Collagen content was assessed using a previously described colorimetric Sirius-red binding assay [19]. Sample collagen content was calculated by interpolation to a standard curve of known concentrations of rat tail type I collagen (Sigma), and expressed as μg collagen per mg wet tissue weight.

Levels of IL-1α, IL-1β, TNF-α, interferon-γ (IFN-γ), IL-1 receptor antagonist (IL-1RN), IL-10, TNF-R2, and total transforming growth factor β1 (TGF-β1) were determined by enzyme-linked immunosorbent assay (ELISA). Briefly, frozen skin tissue was weighed and homogenized in T-PER lysis buffer (10 mg tissue/ml buffer) supplemented with HALT protease-phosphatase inhibitor cocktail (Thermo Scientific). Following centrifugation, supernatants were analyzed with ELISA according to the manufacturer's recommended protocol (Duoset ELISA kits, R&D systems).

Coculture Model of BMSC Interaction with MΦ

Primary BMSC and MΦ were isolated from 4- to 6-week-old C3H/HeN, wild-type c57BL6/J, or mice deficient in TNF-R2. BMSC were prepared as described above and used at the second through fourth passages. Primary MΦ were isolated from flushed marrow and expanded in DMEM:F12 media supplemented with 10% heat-inactivated FCS (Gemini), 10% L-929 conditioned media, and 1% AAS as described by Zhang et al. [20]. Primary MΦ were harvested by scraping and plated at 1 × 105/cm2. For coculture experiments, primary BMSC at 1 × 105/cm2 were added 1 hour after MΦ plating. Six hours following addition of BMSC, cultures were stimulated for 24 hours with 1 μg/ml lipopolysaccharides (LPS) from Escherichia coli strain 0111:B4 (Sigma). Sterile, endotoxin-free PBS was used as a control. A TNF-R2 neutralizing antibody was used to block this receptor in cocultures of wild-type C3H/HeN origin cells (1 μg/ml, AF726, R&D Systems) for 6 hours prior to addition of LPS. Conditioned media were collected and stored at −80°C until ELISA.

Statistical Analysis

Unless stated otherwise, all data presented are mean ± SD. Data were analyzed by Student's t test or ANOVA, and differences were accepted as statistically significant when p ≤ .05 (StatView v5.01, SAS Institute, Cary, NC).


BMSC Characterization

BMSC are characterized by expression of several surface markers and display multipotential differentiation along mesenchymal lineages [4]. Phenotypic characterization by flow cytometry verified that the magnetically sorted BMSC product was >95% positive for Sca-1, CD29, CD106, and CD44 and negative for CD45 and CD11b (Fig. 1A). Chemically defined media were used to induce in vitro differentiation of the magnetically sorted BMSC product toward adipogenic, osteogenic, and chondrogenic phenotypes, verifying their mesenchymal multipotentiality (Fig. 1B, 1C).

Figure 1.

Phenotype and multipotency of bone marrow-derived mesenchymal stem cells (BMSC). (A): Cell surface expression of Sca1, CD29, CD44, CD106, CD45, and CD11b (red) or isotype (blue). (B): BMSC induced toward osteogenic (Alizarin red; ×20 magnification bar = 50 μm), adipogenic (oil red O/hematoxylin; ×40 magnification, bar = 50 μm), or (C) chondrogenic phenotypes (alginate bead section, alcian blue/nuclear fast red; ×10 magnification, bar = 100 μm).

Cutaneous Radiation-Induced Fibrosis Is Significantly Reduced in Mice Receiving BMSC Infusion

Skin contracture, collagen deposition, and dermal thickening are characteristic of cutaneous radiation-induced fibrosis. In this experiment, acute skin damage was evaluated regularly at early time points, while skin contracture was assessed at later time points by measuring hind limb extension. Acute injury progressed to moist desquamation by 28 days post-irradiation, and rapidly resolved to erythema prior to BMSC infusion. No significant difference in acute injury was observed between vehicle or BMSC-treated mice at any time point through 66 days post-irradiation (Supporting Information Fig. S1). Thereafter, mice irradiated with 35 Gy showed progressive contracture of the irradiated limb (Fig. 2A), with limb extension in the vehicle-treated group reduced to 79.9% at 150 days after IR. In contrast, contracture in mice receiving either sBMSC or aBMSC 6 weeks after IR was significantly improved at both 120 (86.4% and 85.7%, respectively) and 150 (88.5% and 87.4%, respectively) days in comparison to vehicle-treated control mice (p ≤ .0079). Relative to vehicle-treated mice, mice infused with sBMSC or aBMSC showed a 37% and 34% reduction in leg contracture.

Figure 2.

BMSC suppress radiation-induced skin contracture, thickening, and collagen deposition. (A): Hind limb extension, (B) skin thickness, and (C) collagen content were improved in BMSC-infused mice. Brackets show differences, p < .05. (D): Partial improvement of skin architecture in BMSC-treated mice. Masson trichrome stain, ×20 magnification, bar shows 150 μm. Abbreviation: BMSC, bone marrow-derived mesenchymal stem cells.

Dermal thickening and collagen accumulation characterize fibrosis, which contribute to loss of skin elasticity and skin contracture. The irradiated skin showed significant thickening in all treatment groups when compared with skin from the unirradiated contralateral limb (Fig. 2B, p ≤ .0071). Skin thickness was significantly reduced in mice treated with BMSC of either syngeneic or allogeneic origin (p ≤ .0197) when compared with vehicle-treated mice. Collagen content was assayed from skin samples obtained 150 days after IR (Fig. 2C). Collagen content was significantly greater in the irradiated skin of all treatment groups relative to the unirradiated contralateral control skin (p < .0001). However, in comparison to the irradiated skin of vehicle-treated mice, significant reductions in collagen content were observed in the irradiated skin of mice treated with either syngeneic (p = .0032) or allogeneic BMSC (p = .0109). There was no significant difference in collagen content of the unirradiated skin when comparing vehicle control or BMSC treatment groups (p > .4196).

Histologically (Fig. 2D), irradiated skin from vehicle-treated mice showed profound thickening of the dermis with dense deposition of collagen, numerous infiltrative inflammatory cells, and absence of dermal appendages. Epidermal hyperplasia and extensive hyperkeratosis were also observed. Although irradiated skin from BMSC-treated mice exhibited dermal thickening and epithelial hyperplasia, the extent of these changes was markedly reduced compared to irradiated skin from vehicle-treated mice. Interestingly, cells derived from syngeneic or allogeneic donors showed similar efficacy in reducing contracture, skin thickness, and collagen accumulation. Taken together, these results demonstrate BMSC are capable of altering the progression of radiation-induced cutaneous fibrosis.

Persistence of BMSC in Irradiated Tissues

We used a PCR assay to detect genomic DNA unique to the Y-chromosome of donor-derived BMSC from irradiated skin samples collected at 14, 42, and 108 days following systemic infusion. Donor-derived cells were detectable by PCR in 75% of skin samples at 14 days, 25% at 42 days, and 10% at 108 days after injection. Donor-derived cells were not detectable by PCR in peripheral blood samples at any time point, suggesting that the cells detected in the host tissue were not transitory. By immunohistochemistry (Supporting Information Fig. S2), we observed rare GFP-positive cells in the dermis and subcutaneous fascia, adjacent to or surrounding small blood vessels, but never in the vascular lumen or in the epidermis, in agreement with the findings of Zhang et al. [21]. Furthermore, we did not observe colocalization of GFP immunoreactivity with that of the common leukocyte antigen CD45, demonstrating that the donor cells detected were unlikely to be derived from expansion of a hematopoietic stem cell contaminating the infused cell product.

BMSC Infusion Alters the Expression of Several Fibrosis-Associated Genes in Irradiated Skin

A number of gene products have been identified as mediators of both fibrosis and chronic inflammation. In order to better understand what effect BMSC have in the context of radiation-injured skin, we assayed differential expression of 84 fibrosis-associated genes at the peak of their retention (14 days after infusion). A total of thirty-six genes were differentially expressed in comparisons between unirradiated, irradiated/vehicle, and irradiated/BMSC groups. Unsupervised hierarchical clustering (Fig. 3) was used to organize all evaluated genes on the basis of common levels of gene expression between sample groups, although no functional or coregulatory relationship is implied. Ten genes showed significant differences (p < .05) in expression in the irradiated skin of BMSC or vehicle infused mice. Of these, the greatest differences in expression were observed for IL-1β and Serpine1, which were downregulated in BMSC-infused mice (−2.54-fold, p = .0088 and −5.61-fold, p = 0.287, respectively). In contrast, irradiated skin from BMSC-treated mice showed significant upregulation of IL-10 (2.32-fold, p = .0037) and platelet-derived growth factor-α (PDGFα; 2.50-fold, p = .0009).

Figure 3.

BMSC alter fibrosis-associated gene expression in irradiated skin. Unsupervised hierarchical clustering of 84 fibrosis-associated transcripts, comparing unirradiated skin to the irradiated skin of mice receiving either vehicle or sBMSC. IL-10, IL-1β, PDGF-A, and Serpine1 were significantly greater than twofold differentially expressed between the vehicle and BMSC-infused mice. Abbreviation: BMSC, bone marrow-derived mesenchymal stem cells.

BMSC-Infused Mice Show an Altered Profile of Inflammatory Cytokine Expression and MΦ Infiltration

Based on the results of the RT-PCR array suggesting that BMSC may alter the levels of immunoregulatory cytokines, we assayed the level of IL-1α, IL-1β, TNF-α, IL-1RN, IL-10, and total TGF-β1 (Fig. 4A–4F, respectively) in skin tissue collected 45, 60, and 150 days post-irradiation (corresponding to 3, 18, and 108 days after BMSC infusion). At early time points, skin from BMSC-infused mice showed significantly decreased levels of IL-1α (p ≤ .0412) and IL-1β (p ≤ .0236) relative to irradiated skin from vehicle-treated mice, while levels of IL-1RN (p ≤ .0086) and IL-10 (p ≤ .0001) were increased. At later time points, BMSC-treated mice showed decreased levels of TNF-α (p ≤ .0001) and TGF-β1 (p ≤ .0032).

Figure 4.

Temporal alteration of immunoregulatory cytokine levels by BMSC infusion. Tissue homogenates prepared for enzyme-linked immunosorbent assay to determine levels (A) IL-1α, (B) IL-1β, (C) TNF-α, (D) IL-1RN, (E) IL-10, and (F) total TGF-β1. Abbreviations: BMSC, bone marrow-derived mesenchymal stem cells; IL-1α, interleukin-1α; IR, ionizing radiation; TGF-β1, transforming growth factor-β1; TNF-α, tumor necrosis factor-α.

Infiltrative cells derived from the monocyte-macrophage lineage participate in the successive phases of inflammation, wound repair, and tissue regeneration in response to a variety of injuries. Furthermore, IL-1β and IL-10 are known products of classically activated MΦ and regulatory inflammatory MΦ, respectively [17]. Based on the differences in expression of IL-1β and IL-10 that we observed in irradiated skin from mice receiving BMSC infusion compared to vehicle, we hypothesized that differential polarization of MΦ could be responsible for these effects. Therefore, we sought to compare the phenotype of MΦ in irradiated skin by immunohistochemical staining for MΦ markers CD68, CD80/B7.1, and CD163 (Fig. 5). Compared to their unirradiated skin, irradiated skin from vehicle-treated mice exhibited significantly increased numbers of cells staining for the pan-MΦ marker CD68 (p < .0001) or CD80/B7.1 a marker of classically activated MΦ (p = .0091). Relative to the irradiated skin of vehicle-treated mice, BMSC treatment reduced the number CD68 and CD80/B7.1-positive cells, and increased the number of CD163-positive regulatory MΦ (p = .0004).

Figure 5.

BMSC suppress macrophage infiltration in irradiated skin. (Left) Representative immunohistochemistry and (right) positive cell counts for (A) CD68, (B) CD80/B7.1, and (C) CD163. DAB histochemistry with Gill's hematoxylin counterstain; ×20 magnification, bar shows 150 μm, and brackets p ≤ .05. Abbreviation: BMSC, bone marrow-derived mesenchymal stem cells.

BMSC Enhance Production of Anti-Inflammatory Molecules by LPS-Activated MΦ In Vitro

Our in vivo experiments showed that BMSC treatment resulted in a persistent reduction of chronic inflammation, and warranted a refined study of the interaction between BMSC and activated MΦ. To this end, we examined the profile of cytokine secretion by LPS-activated primary MΦ when cocultured with BMSC. MΦ activated with LPS released massive quantities of TNF-α (p < .0001, Fig. 6A). Similarly, LPS activation induced production of IL-10 and IL-1RN, an effect that was significantly amplified by coculture with BMSC (p ≤ .0001 Fig. 6B, 6C). We did not however observe a significant effect of coculture on the level of TGF-β elaborated by MΦ (Fig. 6D, p > 0.0500). These results demonstrate that the direct interaction of BMSC and MΦ promotes transition toward an anti-inflammatory phenotype in both in vivo and in vitro model systems.

Figure 6.

Cytokine secretion by LPS-stimulated bone marrow-derived mesenchymal stem cells (BMSC)-MΦ cultures. Mono- and cocultures of BMSC or MΦ were exposed to either PBS (control) or LPS for 24 hours, and enzyme-linked immunosorbent assay interrogated media supernatants for levels of secreted (A) TNFα, (B) IL-10, (C) IL-1RN, and (D) total TGF-β1. Abbreviations: IL-10, interleukin-10; IL-1RN, interleukin-1 receptor antagonist; LPS, lipopolysaccharides; TGF-β1, transforming growth factor-β1; TNF-α, tumor necrosis factor-α.

Signaling through the TGF-β [22, 23] and TNF-α [3] pathways is known to be critical promoters of fibrosis. In contrast, localized exposure of antigen-presenting cells to TGF-β and TNF-α results in TNF-R2-mediated immune tolerance [24]. In irradiated skin tissue, we observed progressive decreases in both TGFβ and TNFα following BMSC infusion. In contrast, we measured an immediate and massive increase in TNF-R2 levels following BMSC infusion. In our coculture experiments, we observed increased expression of TNF-α but no significant change in TGF-β secretion when activated MΦ were cocultured with BMSC compared to activated MΦ alone. This difference may reflect the vastly simplified nature of the coculture experiments, which were designed solely to examine the interaction of BMSC and MΦ, and cannot reasonably recapitulate all aspects of the in vivo microenvironment. While macrophages are likely the predominant source of TNFα in both models, there are multiple cell lineages present in the skin (e.g., keratinocytes, (myofibroblasts, vascular cells, and other leukocyte lineages) that are known to secrete significant amounts of TGFβ in response to radiation injury [14, 23, 25, 26]. Nonetheless, these findings, coupled with the enhancement of IL-10 in coculture led us to hypothesize that TNF-R2 signaling may mediate the BMSC-induced transition of MΦ toward a regulatory phenotype. We tested the influence of TNF-R2 on IL-10 secretion by control or LPS-stimulated cocultures of BMSC and MΦ in the presence of a TNF-R2 neutralizing antibody. Consistent with our hypothesis, antibody-mediated neutralization of TNF-R2 suppressed IL-10 release in coculture experiments to a level comparable to that of unstimulated cultures (Fig. 7A). To clarify whether TNF-R2 deficiency in MΦ or BMSC mediated this effect, coculture experiments with cells derived from TNF-R2 deficient mice were performed. Coculturing MSC and MΦ from various combinations of wild-type and TNF-R2 deficient mice revealed that IL-10 production is impaired only when the MΦ component was TNF-R2 deficient (Fig. 7B). Supporting our in vitro findings, we observed increased expression of TNF-R2 in the irradiated skin of BMSC-treated mice compared to vehicle controls (Fig. 7C, p ≤ .0085). Taken together, these findings demonstrate that BMSC suppress chronic inflammation by inducing IL-10 production by activated macrophages in a TNF-R2-regulated mechanism.

Figure 7.

TNF-R2 expression is increased by BMSC, and mediates IL-10 production. IL-10 release by (A) cocultures exposed to a TNF-R2 neutralizing antibody or (B), mixed cocultures of MΦ and BMSC from wild-type and TNF-R2 mice. (C): ELISA of TNF-R2 levels in skin tissue. Abbreviations: BMSC, bone marrow-derived mesenchymal stem cells; IL-10, interleukin-10; IR, ionizing radiation; LPS, lipopolysaccharides.


Cutaneous fibrosis has long been recognized as a significant adverse event following exposure to IR, yet few successful therapies are currently available. Recently, BMSC have been demonstrated to accelerate healing of excisional wounds and thermal burns [27-30]. Similarly, Francois et al. reported that systemic infusion of human BMSC 24 hours after radiation exposure reduced the severity of acute radiation dermatitis in immunodeficient mice. Several groups have recently reported that locally injected mesenchymal stem cells may be effective in healing of ulcerative radiation wounds [31-33]; in contrast, their ability to prevent or treat radiation fibrosis is unexplored.

Using a murine model, we tested the hypothesis that cutaneous radiation-induced fibrosis could be attenuated by late systemic administration of BMSC. We sought to clearly distinguish the ability of BMSC to mitigate late skin injury from acute injury, and thus chose to deliver the therapeutic intervention after resolution of acute dermatitis. Furthermore, we anticipated that clinical translation would be most realistic in the mitigation or treatment setting versus a prophylactic approach.

Mice infused with BMSC from allogeneic or syngeneic donors displayed significantly reduced fibrotic skin contracture, dermal thickening, and collagen accumulation relative to vehicle-infused mice. Previous studies have shown systemic BMSC administration inhibited bleomycin-induced pulmonary fibrosis [34, 35] and that local application improved cutaneous wound healing [21, 28, 36]. To our knowledge, this is the first study documenting a reduction of radiation-induced fibrosis with BMSC-based therapy.

Although it is generally accepted that nonsyngeneic BMSC are capable of evading the host immune response [37], it is possible that conditions under which BMSC are produced or administered may alter the immune recognition of the BMSC [4, 38-40]. It has been reported that BMSC evade rejection by the host immune system due to low levels of MHC-II molecules on the cell surface. Others have shown that MHC-II expression on BMSC may be inducible, particularly by IFN-γ, resulting in elimination of the mismatched donor cells [41-43]. Radiation has been shown to induce IFN-γ in some settings [44], raising the question of whether allogeneic BMSC will be as effective in radiation injury models. Based on this controversy, we evaluated the persistence and efficacy of syngeneic and allogeneic BMSC in our model. We found no differences in the efficacy or persistence of BMSC from syngeneic or allogeneic sources, suggesting that radiation-injured skin may be an environment that is permissive of MSC retention.

To identify the potential mechanisms by which BMSC may elicit their antifibrotic effect, we profiled the differential expression of 84 fibrosis-associated genes at the peak of BMSC retention in the irradiated skin. By this approach, we found significant differential regulation of several genes that are associated with fibrosis including PDGF-A, Serpine1, IL-10, and IL-1β.

Our decision to focus on immune interactions was derived from the finding that mice treated with MSC showed differential expression of two mutually antagonistic genes (IL-1β and IL-10) that have been previously associated with the immunomodulatory activity of BMSC. The skin of BMSC-treated mice displayed significant downregulation of IL-1β, which clustered with several other downregulated genes that have been shown to promote fibrosis. IL-1β is a major proinflammatory cytokine induced by a number of proinflammatory stimuli and secreted by activated MΦ upon inflammasome assembly [45]. Conversely, IL-10 expression was significantly upregulated in the skin of mice infused with BMSC after irradiation. IL-10 functions as an anti-inflammatory cytokine, suppressing release of proinflammatory cytokines by activated MΦ and promoting their polarization toward a regulatory phenotype [46].

These data suggested that BMSC might be altering the local inflammatory environment through interactions with tissue MΦ, a source of both proinflammatory IL-1 α and IL-1β as well as anti-inflammatory IL-10. It was recently discovered that BMSC promote IL-10 secretion by LPS-activated MΦ in a contact-dependent, prostaglandin E2-mediated mechanism [8]. BMSC have also been shown to suppress classically activated MΦ [47], and promote their transition toward a regulatory MΦ phenotype [48]. Our in vitro studies confirmed that BMSC promote the production of IL-10, and IL-1RN by LPS-activated MΦ in a paracrine manner. We also provide new evidence that that the anti-inflammatory effect of BMSC characterized by IL-10 release is mediated at least in part by signal transduction through TNF-R2 by the activated macrophages.

Despite significant phenotypic overlap between various populations of MΦ, it is clear that specific subsets are integral to the inflammatory, repair, and resolution phases of radiation injury [49]. Classically activated MΦ, which secrete the proinflammatory cytokines IL-1β and TNFα, optimize the early immune response to injury [50]. The precise role of regulatory MΦ is not completely understood, but they are thought to suppress chronic inflammation, which if unrestrained may lead to tissue fibrosis [51, 52].

Our data showed a significant increase in the level of IL-10 transcript and protein in the irradiated skin of mice that received BMSC infusion. These findings suggested that redirection of classically activated MΦ toward the regulatory phenotype MΦ was occurring after BMSC infusion. Indeed, MΦ infiltration was reduced in irradiated skin after BMSC infusion and a greater proportion of the remaining MΦ were of the CD163+ regulatory phenotype. These findings were further supported by our experiments in vitro, demonstrating that the production of IL-10 by LPS-activated MΦ is greatly enhanced by coculture with BMSC. These results are consistent with reports by others which have demonstrated a similar ability of BMSC to stimulate MΦ production of IL-10 [8, 48].

Prior studies have largely focused on paracrine signaling mechanisms used by BMSC to affect macrophage activity. Nemeth et al. [8] demonstrated that BMSC-induced IL-10 production by activated MΦ is TNF-α dependent, requiring TNF-R1, but not TNF-R2, activation on BMSC for this effect. The signaling pathways in MΦ responsible for BMSC-mediated stimulation in MΦ IL-10 production are largely unexplored. We determined that MΦ TNF-R2 signaling is necessary for BMSC-mediated transition of activated MΦ toward a IL-10 secreting regulatory phenotype.

It is also possible other anti-inflammatory mechanisms potentiated by BMSC may contribute to the antifibrotic effect of BMSC infusion. In response to bleomycin challenge [34], BMSC were previously found to suppress macrophage infiltration, proinflammatory cytokine release, and the development of pulmonary fibrosis by upregulated expression of IL-1RN. Similar to this finding, we observed increased levels of IL-1RN in the irradiated skin of BMSC-treated mice when compared with irradiated skin from vehicle-treated mice. We also observed increased expression of IL1-RN expression by LPS-activated MΦ over untreated cells, and this was further increased, albeit modestly, by coculture with BMSC.

Collectively, these data suggest that BMSC result in a durable alteration of the local inflammatory environment via manipulation of the secretome and polarization of MΦ. The in vitro effects we observed, together with our temporal evaluation of cytokine expression by ELISA, support the notion that BMSC may act in an immunomodulatory capacity to repress fibrosis. This conclusion was further supported by our observation of a durable reduction of the number of infiltrating MΦ in irradiated skin. In contrast to the effects observed with small molecule inhibitors targeting profibrotic pathways, a single infusion of BMSC appears capable of providing durable alterations in inflammatory cytokine balance and MΦ infiltration in irradiated skin.


In conclusion, this study has demonstrated a persistent antifibrotic effect of a single late systemic infusion of BMSC. The durable suppression of macrophage infiltration and cytokine expression in the irradiated skin of BMSC-treated mice suggests that late infusion of BMSC fundamentally alters the biology of irradiated skin tissue. While we did not achieve complete resolution of the fibrosis in this study, it is possible that further refinement of the delivery strategy by early or repeated earlier delivery of BMSC may provide an even more effective antifibrotic effect BMSC. Experiments to test this possibility are currently underway.


This research was supported by the National Institutes of Health Intramural Research Program (National Cancer Institute) and the Radiation/Nuclear Medical Countermeasures Program (National Institute of Allergy and Infectious Diseases).

Disclosure of Potential Conflicts of Interest

The authors have no potential conflicts of interest to disclose.