PDGFRα-positive cells in bone marrow are mobilized by high mobility group box 1 (HMGB1) to regenerate injured epithelia

Abstract

The role of bone marrow cells in repairing ectodermal tissue, such as skin epidermis, is not clear. To explore this process further, this study examined a particular form of cutaneous repair, skin grafting. Grafting of full thickness wild-type mouse skin onto mice that had received a green fluorescent protein-bone marrow transplant after whole body irradiation led to an abundance of bone marrow-derived epithelial cells in follicular and interfollicular epidermis that persisted for at least 5 mo. The source of the epithelial progenitors was the nonhematopoietic, platelet-derived growth factor receptor α-positive (Lin⁻/PDGFRα⁺) bone marrow cell population. Skin grafts release high mobility group box 1 (HMGB1) in vitro and in vivo, which can mobilize the Lin⁻/PDGFRα⁺ cells from bone marrow to target the engrafted skin. These data provide unique insight into how skin grafts facilitate tissue repair and identify strategies germane to regenerative medicine for skin and, perhaps, other ectodermal defects or diseases.

Bone marrow (BM) cells contribute a substantial proportion of cells, both inflammatory and noninflammatory, that have roles in tissue homeostasis, repair, and regeneration. Such cells may be derived from either hematopoietic or mesenchymal stem cell populations, and subpopulations thereof can differentiate into both hematopoietic and mesenchymal lineage cells (for review, see refs. 1 and 2).

In skin, studies have shown that BM provides fibroblast-like cells in the dermis (of hematopoietic and mesenchymal lineages) and that the number of these cells increases after skin wounding (3, 4). BM can also generate epithelial cells, i.e., keratinocytes, in the epithelia, although the precise derivations and mechanisms to raise BM-derived keratinocytes are not fully known (3, 5–16). Human/mouse studies involving transplantation of sex-mismatched or genetically tagged BM cells have shown that keratin-positive bone marrow-derived cells can be found in skin epidermis, hair follicles, and sebaceous glands (6, 8, 9, 11, 12, 14, 16), sites that harbor skin stem cell niches (17). Moreover, in humans who have undergone BM transplantation (BMT), donor cells that have differentiated into keratinocytes can be detected in the epidermis for at least 3 y (8).

BM cells also contribute to skin development: Infusion of green fluorescent protein (GFP) BM cells in utero in mice leads to accumulation of a subpopulation of GFP-positive cells in nonwounded skin dermis, particularly in association with developing hair follicles (18). With regard to skin injury, both embryonic and postnatal transplantation of BM cells into mice lacking the skin protein, type VII collagen (Col 7) as well as postnatal studies in mice lacking type XVII collagen, basement membrane components that normally help secure adhesion between the epidermal and dermal skin layers, have demonstrated the capacity of BM to promote skin wound healing and to correct the intrinsic basement membrane defect (18–20). Most recently, a clinical trial of allogeneic whole BMT in humans lacking Col 7 (who have the inherited blistering skin disorder, recessive dystrophic epidermolysis bullosa, RDEB; OMIM226600) (21) has demonstrated that BM cells can repair fragile skin and restore Col 7 expression in skin basement membrane (22).

Collectively, however, these animal and human studies have shown that BM-derived keratinocytes are an extremely rare finding in the epidermis. Indeed, analyses in two different murine models have shown that BM-derived keratinocytes comprise only ≈0.0001–0.0003% of all keratinocytes in the new epidermis (13). The relative scarcity of such cells therefore raises questions about their biological significance. For example, it is not known whether these cells have a physiological role in epithelial regeneration and, if they do, under what circumstances? Moreover, it is not clear which particular cells in BM contribute to the epithelial repair. Furthermore, there is little awareness of the mechanism(s) through which the damaged epithelium signals to invoke mobilization and recruitment of the key BM cells. In this study, we have begun to address these questions, and here report identification of a specific subset of BM cells with epithelial differentiation potential as well as a unique in vivo mechanism through which these cells contribute to epithelial regeneration and maintenance.

Results

We first examined the contribution of BM-derived cells to epithelial regeneration in murine skin wounds. The wounds were created by using a scalpel and involved excision of a piece of full-thickness skin (i.e., including epidermis and dermis) thus creating an ulcer. The mice selected for wounding had received lethal dose irradiation followed by GFP-BMT, thereby allowing us to evaluate the contribution of GFP-BM cells to skin regeneration after injury (Fig. 1A). At 4 wk after the injury, GFP-positive keratinocytes were not obvious (Fig. 1B), indicating a minimal, if any, contribution of BM cells to epithelial regeneration, findings consistent with previous data (13). We then used the same mouse model to examine a different form of skin repair, skin grafting, to further explore the potential contribution of BM cells (Fig. 1C). Surprisingly, we noted that significant numbers of GFP-positive cells expressing skin-specific keratin 5 formed epidermal proliferative unit (EPU)-like clusters in the epidermis of the skin graft in biopsies taken 4 wk after the engraftment (Fig. 1D and Fig. S1A). Furthermore, the BM-derived GFP-positive keratinocytes were maintained in the epidermis and hair follicles 5 mo after the engraftment (Fig. 1E and Fig. S1B). Given that mouse epidermis is renewed every 2–3 wk by epithelial stem cell-derived keratinocytes, those long-residing GFP BM-derived epithelial cells in the 5-mo-old graft are likely to contain epithelial progenitor/stem cells.

Fig. 1.

BM-derived cells contribute to epidermal and follicular renewal in skin grafts, but not in skin wounds. (A) Schematic outlining details of the skin wound model on GFP-BMT mice. (B) Confocal-laser microscopy of skin sections from the reepithelialized wounds. Green fluorescence in the dermis represents BM-derived cells. Note that there is no GFP fluorescence within the regenerated epithelia. (C) Schematic outlining details of the skin engraftment of GFP-BMT mice. (D) Confocal-laser microscopy of skin sections from the 4-wk skin graft. Green fluorescence in the epidermis and the dermis represent BM-derived cells. Arrows indicate EPU-like structures that are composed of BM-derived epithelial cells emanating from the basal layer (white arrows) within the epidermis to the horny layer (red arrowheads). (E) Confocal-laser microscopy pictures of skin sections from the 5-mo skin graft. Green fluorescence in the epidermis, the hair follicles, and the dermis represent BM-derived cells. Arrows in Upper indicate sustained BM-derived epithelial cells emanating from the basal layer (white arrows) to the horny layer (red arrows). White arrows in Lower indicate sustained BM-derived follicular cells. Note that hair shafts show nonspecific auto-fluorescence (asterisks). White dotted lines indicate the dermal-epidermal and dermal-follicular junctions. (Scale bars: 50 μm.)

To assess whether the BM-derived epidermal cells might have functional significance, we then searched for BM-derived cells in the skin of Col 7-null mice engrafted onto the GFP-BMT mice (Fig. 2A). Because these mice have complete detachment of the epidermis after birth due to extensive skin and mucous membrane fragility (Fig. 2A) (23), we anticipated a potentially greater contribution of BM-derived cells in the regenerating epithelia of the engrafted Col 7-null mouse skin. The engrafted Col 7-null mouse skin initially showed extensive subepidermal detachment, similar to the skin pathology seen in human patients with RDEB (21). At 4 wk after the engraftment, we noted an even greater contribution of GFP-positive cells expressing keratinocyte-specific keratins in the regenerating epidermis and hair follicles of the engrafted Col 7-null mouse skin (Fig. 2 B–D and Fig. S2). Moreover, new Col 7 protein was present at the cutaneous basement membrane zone in the engrafted Col 7-null mouse skin (Fig. 2E). Notably, Col 7 labeling was maximal in the basement membrane adjacent to GFP-BM–derived epithelial cells. Evidence for BM-derived epithelial cells was further confirmed by demonstration of GFP-positive individual keratinocytes from the grafted skin by means of both flow cytometric analysis and cell culture (Fig. 2 F and G).

Fig. 2.

BM cells substantially contribute to the repair of grafted Col 7-null epidermis. (A) Schematic picture of the Col 7-null skin engraftment on GFP-BMT mice (Left); hematoxylin-eosin staining of Col 7-null mouse skin (Left Center). Asterisks indicate separation of the epidermis (E) from the dermis (D); Col 7 immunostaining in the Col 7-null skin (Right Center). (Right) Arrows indicate dermal-epidermal junction, Col 7 staining of wild-type mouse skin. Green color indicates Col 7 staining. (B) Accumulation of GFP fluorescence (Lower) within the region of the Col 7-null skin graft (Upper). The white dotted line indicates the margins of the skin graft. (C and D) Confocal-laser microscopy pictures of sections from the Col 7-null mouse skin that had been surgically grafted. Blue color indicates DAPI staining. Green fluorescence in the epidermis and the dermis represents BM-derived cells. Red labeling indicates keratin 5 (K5; C) or keratin 10 (K10; D) immunofluorescence. Yellow color image represents BM-derived cells that express both GFP (green) and K5 (red; C) or K10 (red; D). (E) Col 7 expression in grafted Col 7-null mouse skin. White dotted lines indicate dermal-epidermal and dermal-follicular junctions. White arrows indicate immunofluorescence for Col 7 (red) at the basement membrane zone. (Scale bars: 50 μm.) (F) Flow cytometric analysis for K5 and GFP in the epidermal cell suspension of Col 7-null skin engrafted onto a GFP-BMT mouse (Upper Left), wild-type mouse skin (Lower Left), and the GFP-transgenic mouse skin (Lower Right). Upper Right shows isotype control for K5 in the epidermis of the grafted Col 7-null skin. (G) Confocal-laser micrographs of cultured BM-derived epithelial cells isolated from the epidermis of the Col 7-null skin graft labeled with GFP or K5.

Collectively, these data showed that a subpopulation of BM cells contribute to epithelial regeneration and maintenance in these murine skin graft models. Analyses of sex chromatin numbers and fusion-dependent enhanced GFP expression in BM-derived epithelial cells in the grafted skin did not show any evidence for cell fusion, suggesting differentiation of the BM-derived cells as the likely mechanism for raising BM-derived keratinocytes in the skin graft (Figs. S3 and S4). Reverse wild-type BMT in a GFP mouse, followed by engrafting Col 7-null mouse skin onto the back of this mouse, excluded other possible extrinsic sources of keratinocytes in the skin graft besides the BM (Fig. S5).

We then investigated the subpopulation of BM cells that has the potential for epithelial differentiation. Recent studies have shown that the PDGFRα-positive nonhematopoietic BM cell population contains ectoderm-derived mesenchymal stem cells (MSCs) (24–26), indicating that perhaps PDGFRα⁺ BM cells might be a putative source of BM-derived keratinocytes in the skin graft. To test this hypothesis, we examined BM from a heterozygous knock-in mouse in which a histone H2B-GFP fusion gene was inserted into the locus of the PDGFRα gene (PDGFRα-H2BGFP mouse); this knock-in results in accumulation of GFP fluorescence within the nuclei of PDGFRα-expressing cells. We found that the PDGFRα⁺ BM cells were exclusively enriched in the Lineage-negative (Lin⁻) cell population from the knock-in mouse BM (Fig. S6) and that the Lin⁻ population provided proliferating PDGFRα⁺ fibroblastic cells in culture (Fig. 3A). Lin⁻ indicates negativity for the cell surface antigens CD5, B220, CD11b, Gr-1, 7–4, and Ter-119 and excludes mature hematopoietic cells, such as T cells, B cells, monocytes/macrophages, granulocytes, and erythrocytes/their committed precursors from BM. Flow cytometric analysis indicated that the Lin⁻ and PDGFRα⁺ (Lin⁻/PDGFRα⁺) cells were independent from the Lin⁻/c-kit⁺ cell population, which includes a hematopoietic stem cell pool (Lin⁻/c-kit⁺/Sca-1⁺) in BM (Fig. 3B and Fig. S7). We found that Lin⁻ cell populations collectively accounted for ≈5.6% of the total number of BM cells (Fig. 3B). The Lin⁻/PDGFRα⁻/c-kit⁺ and Lin⁻/PDGFRα⁻/c-kit⁻ BM cells comprised ≈2.28% and ≈3.11% of total BM cells, respectively. We observed that the Lin⁻/PDGFRα⁺ BM cells (≈0.22% of the total BM cells) could exclusively generate BM-derived epithelial cells expressing keratin 5 in culture after supplementation with skin soaked buffer (SSB), in which excised newborn mouse skin had been soaked in phosphate buffer saline (PBS) for 24 h (Fig. 3C and Fig. S8). The Lin⁻/PDGFRα⁻ cell population also contained adherent and proliferative cells in culture, but none of these cells showed differentiation into keratin 5-positive keratinocytes with SSB supplementation (Fig. S8). These data suggest that the BM-derived keratinocytes are not of hematopoietic origin, but instead are derived from a specific subpopulation of Lin⁻/PDGFRα⁺ BM cells. In this context, additional flow cytometry analysis of the Lin⁻/PDGFRα⁺ BM cells did not show expression of CD146 or CD271 (Fig. S9), both of which are established markers of human BM MSCs (27), indicating different cell surface molecule profiles for human BM MSCs and mouse Lin⁻/PDGFRα⁺ BM cells.

Fig. 3.

Characterization of BM cells of the PDGFRα knock-in mouse demonstrates that PDGFRα⁺ subpopulation give rise to epithelial progenitors. (A) Phase-contrast (Left) and GFP-fluorescent (Right) micrographs of the Lin⁻ BM cells in culture. All adherent and proliferating Lin⁻ cells were positive with PDGFRα-GFP labeling in their nuclei. (B) Flow cytometry analysis for PDGFRα-GFP and c-kit expression on the Lin⁻ cells in total BM cells of the PDGFRα knock-in mice. The number (%) in the chart represents the population of each fraction in the total number of BM cells. Note that there is no population in the PDGFRα-positive and c-kit-positive fraction (upper right corner) in BM. (C) Confocal-laser micrographs of cultured bone marrow cells (BMCs) from a heterozygous knock-in mouse with a histone H2B-GFP fusion gene inserted into the PDGFRα gene locus. PDGFRα promoter-dependent H2B-GFP expression was noted by accumulation of GFP fluorescence in the nuclei (PDGFRα/H2B-GFP, indicated by white arrows; Upper Right) of cells expressing K5 (Lower Left).

We then investigated the mechanism through which the transplanted skin graft is able to recruit Lin⁻/PDGFRα⁺ cells from the BM. First, we established a Boyden chamber migration assay to demonstrate that Lin⁻/PDGFRα⁺ BM cells migrate toward one or more chemoattractants in SSB (Fig. 4A). We then assessed which molecules in the SSB have the capacity to induce migration of Lin⁻/PDGFRα⁺ BM cells. We noted that a heparin-binding fraction of SSB was able to induce robust migration of the Lin⁻/PDGFRα⁺ BM cells (Fig. 4B). This finding supports the notion that the excised skin graft can release heparin-binding molecules capable of attracting these particular BM cells. To find the precise molecules involved, we fractionated SSB by heparin-affinity chromatography and obtained several fractions with strong activity for inducing Lin⁻/PDGFRα⁺ BM cell migration (Fig. S10). Some fractions had strong cell-migrating activity but comparatively less protein expression: These fractions were then subjected to SDS/PAGE analysis (Fig. 4C). Three prominent silver-stained proteins were observed in the gel, which were then further analyzed by liquid chromatography/tandem mass spectrometry. They were identified as nucleolin, anti-thrombin III (AT-III), and high mobility group box 1 (HMGB1) (Fig. 4C). Nucleolin is a eukaryotic nucleolar phosphoprotein that is involved in the synthesis and maturation of ribosomes in nucleoli (28). AT-III is a well-characterized anticoagulant molecule generated in the liver, and which exists in blood plasma (29). HMGB1, also known as amphoterin, is a nuclear protein that can regulate chromatin structure and gene expression (30). It is also released from necrotic cells and some apoptotic cells and acts as an inflammatory regulator. Other studies, however, have indicated that HMGB1 may also act as a local chemo-attractant for various hematopoietic and nonhematopoietic cells that can regulate tissue remodeling (31).

Fig. 4.

Elevation of HMGB1 in serum is induced by skin grafting and mobilizes Lin⁻/PDGFRα⁺ cells from BM. (A) Schematic illustration of the Boyden chamber approach to assess migration of the Lin⁻/PDGFRα⁺ cells to SSB or PBS. The actual and relative migration of these cells to both buffers is shown. (B) Assessment of Lin⁻/PDGFRα⁺ cells migration to heparin-bound (H/B) or unbound (H/U) fractions of the SSB. (C) SDS/PAGE of H/B SSB fractions with Lin⁻/PDGFRα⁺ cell migration activity. Results of liquid chromatography-tandem mass spectrometry for three major proteins in the gel are noted on the right. (D) Western blot against nucleolin, AT-III, and HMGB1 in the SSB obtained at sequential time periods. The lowest blot shows Lin⁻/PDGFRα⁺ cell migration activities in each SSB. (E) Lin⁻/PDGFRα⁺ cell migration assay in a Boyden chamber with recombinant HMGB1, AT-III, and nucleolin. (F) HMGB1 levels in mouse sera after full-thickness skin grafting with newborn wild-type mouse skin. Asterisks (*) indicate statistical significance vs. Day 0 (P < 0.05, n = 4). (G) Flow cytometric analyses of Lin⁻/PDGFRα⁺/CD44⁺ cells of peripheral blood mononuclear cells of mice 12 h after systemic administration of HMGB1 (10 μg in 400 μL of PBS) or PBS (400 μL). Asterisk (*) represents statistical significance (P < 0.01, n = 4). (H) Intravital two-photon imaging of the calvaria BM in PDGFRα-H2BGFP mice 12 h after systemic administration of PBS (400 μL; Left) or HMGB1 (10 μg in 400 μL of PBS; Right) via the tail vein. Green color, GFP expressed under the promoter of PDGFRα; red color, BM microvasculature visualized by i.v. injection of 70-kDa dextran-conjugated Texas Red. (Scale bar: 50 μm.)

The next objective was to determine the time course for release of the proteins from excised skin graft into the buffer (Fig. 4D). Of note, both HMGB1 and AT-III were rapidly released within a few minutes into the SSB fraction that demonstrated strong chemoattractant activity for Lin⁻/PDGFRα⁺ BM cells. AT-III secretion continued at similar levels for at least 48 h, whereas HMGB1 release gradually declined after ≈8 h—a time course that paralleled the chemoattraction findings. With regard to nucleolin, its presence in the SSB only started 2 h after incubation, possibly reflecting the consequences of necrosis in the excised skin. To evaluate the chemoattractant properties, we expressed mouse HMGB1 in HEK293 cells and compared the in vitro activity of the purified recombinant protein to induce Lin⁻/PDGFRα⁺ BM cell migration. Recombinant AT-III and nucleolin were also assessed, but cell migration assays demonstrated that only HMGB1 could induce migration of these particular BM cells (Fig. 4E). We also explored the nature of the receptor on Lin⁻/PDGFRα⁺ BM cells relevant to the cell migration and excluded a role for two known receptors that can mediate the extracellular cytokine effects of HMGB1, the receptor for advanced glycation endproduct (RAGE) and toll-like receptor (TLR) 4 (32–34) (see SI Results and Discussion and Fig. S11 for details).

Next, we explored the source of HMGB1 in the grafted skin. Immunofluorescent microscopy analysis of HMGB1 protein in the skin graft showed abundant staining in the epidermis and much less in the dermis, reflecting the higher cellularity in the epidermis (Fig. S12). We then analyzed Col 7-null mouse skin for HMGB1 release and noted that the detached epithelia (blister roofs) released significant amounts of HMGB1 after soaking in PBS (Fig. S13 A and B). These observations suggest that the epithelial tissue in the skin graft could be a significant source of HMGB1 in vivo. Further support for damaged epithelium as a source of HMGB1 was demonstrated by finding elevated HMGB1 in freshly generated subepidermal blister fluid of human subjects with RDEB (n = 3) (Fig. S13C). We then investigated the systemic effects of HMGB1 after skin injury. We first measured HMGB1 levels in the sera of mice that had received a skin graft of wild-type newborn mouse skin (Fig. 4F). We observed a marked increase in HMGB1 serum levels 3 d after grafting. Of note, however, no increase in serum HMGB1 was noted in mice with full thickness wounds but no skin graft (Fig. S13D), suggesting that the transplanted epithelial tissue is likely to be the source of the elevated HMGB1 in the sera. We also detected ≈60-fold higher levels of HMGB1 in the sera of individuals with RDEB (n = 3) compared with similarly aged normal control subjects (n = 3) (Fig. S13E). These observations led us to hypothesize that systemic elevation of HMGB1 in the blood might positively induce recruitment of Lin⁻/PDGFRα⁺ cells from BM to raise BM-derived keratinocytes (as well as fibroblasts) in the regenerating injured skin, and that this might be one mechanism through which the practice of skin grafting achieves its clinical goals.

To confirm this hypothesis, we systemically administered recombinant HMGB1 at levels similar to that seen in the sera of skin grafted mice to wild-type mice. We observed that this action could mobilize Lin⁻/PDGFRα⁺ BM cells into the blood circulation (Fig. 4G). Lower doses of HMGB1 failed to mobilize these cells (Fig. S14). We noted no local or systemic inflammation or other potentially adverse effects in the mice, despite the high doses of systemic HMGB1 administered (Fig. S15). To further investigate the mechanics of this mobilization by HMGB1 in vivo, we performed intravital two-photon imaging of calvaria BM in living PDGFRα-H2BGFP mice. This experiment showed that HMGB1 could mobilize PDGFRα-positive cells, allowing them to congregate around blood vessels and, thereby, allow egress into the circulation in vivo (Fig. 4H).

To confirm that the mobilized BM-derived PDGFRα⁺ circulating cells provide the epithelial cells in vivo, we combined FACS-sorted PDGFRα⁺/GFP⁺ BM cells with wild-type PDGFRα⁻ BM cells and transplanted these cells to lethally irradiated mice, which then received skin grafts of Col 7-null mouse skin (Fig. 5A). Very few cells were GFP-positive in the peripheral blood mononuclear cell populations of the PDGFRα⁺/GFP⁺ BM transplanted mice (Fig. 5B). However, those GFP-positive circulating cells that originated from the transplanted PDGFRα⁺ BM cells had adherent and proliferative capacities in culture (Fig. 5B). Four weeks after the Col 7-null skin engraftment, multiple foci of GFP-positive cells expressing keratin 5 were observed in the epithelia of the engrafted skin (Fig. 5C), suggesting that the BM-derived PDGFRα⁺ circulating cells contain a population that can differentiate into epithelial cells in the skin graft.

Fig. 5.

Mobilized Lin⁻/PDGFRα⁺ BM-derived cells in circulation contribute to epithelial regeneration of the skin graft in vivo. (A) Schematic illustration showing Col 7-null skin graft on a mouse transplanted with PDGFRα-positive/GFP-BM cells (Left); bright field (Center) and dark field (Right) stereomicroscopic pictures of femoral BM in a PDGFRα⁺/GFP-BMT mouse. Green fluorescence indicates PDGFRα⁺/GFP-BM cells. (B) Bright field (Upper) and dark field (Lower) fluorescent microscopic pictures of cultured peripheral blood mononuclear cells that were obtained from the PDGFRα⁺/GFP-BMT mouse engrafted with Col 7-null skin. (Left) A single PDGFRα⁺/GFP-BM-derived cell in culture (Day 1). (Center) Proliferation of the single PDGFRα⁺/GFP-BM-derived cell in culture (Day 5). (Right) Dividing adherent cells indicated in the white lined box in Lower Center. (C) Confocal-laser microscopic pictures of a section of grafted Col7-null skin onto a PDGFRα⁺/GFP-BMT mouse. GFP fluorescence was merged with the red immunofluorescence of K5 to provide the yellow color. DAPI staining (Upper Left), GFP fluorescence (Upper Right), K5 staining (Lower Left), and a merged image (Lower Right). (Scale bar: 50 μm.)

Discussion

This work clearly demonstrates that Lin⁻/PDGFRα⁺ cells from BM significantly contribute to the regeneration of the epidermis after skin grafting in vivo, and that one biological repair mechanism involves the key cells being mobilized in response to elevated HMGB1 levels in serum, the source of which is the skin graft. The observation that tissue damage can recruit BM stromal cells for tissue repair is well-established (35), but here we have defined a subpopulation of cells that have the capacity to repair skin, including epidermis. The Lin⁻/PDGFRα⁺ BM cell marker profile is not unique to one particular cell population, and our data suggest that it is shared by ≈1 in 450 BM cells. PDGFRα is not expressed by hematopoietic stem cells but by MSCs in bone marrow that can give rise to mesenchymal lineage cells as well as neuroepithelial and neural crest lineage cells (24–26), suggesting that the Lin⁻/PDGFRα⁺ BM cells contain an MSC fraction (discussed further in SI Results and Discussion).

Our study has shown that in situations in which there is significant damage to the epidermis, such as Col 7 deficiency leading to subepidermal blistering, at least some of the Lin⁻/PDGFRα⁺ cells have the plasticity to become BM-derived epithelial progenitors, to generate and sustain new keratinocytes, and to correct the intrinsic lack of Col 7. Moreover, we have identified HMGB1 as a specific factor involved in Lin⁻/PDGFRα⁺ BM cell responses. Our data indicate that HMGB1, which is rapidly released from the detached or blistered Col 7-deficient epithelia, can mobilize Lin⁻/PDGFRα⁺ BM cells into the circulation and accelerate regeneration of the skin by recruiting these cells to raise BM-derived epithelial cells and BM-derived mesenchymal cells in the epidermis and dermis of Col 7-null skin, respectively.

HMGB1 is a highly conserved, abundant, and ubiquitously expressed 30-kDa nonhistone protein with diverse biologic functions (36, 37). With regards to human and murine HMGB1, only 2 of 215 amino acids show species differences (>99% identical) and both reside in the C-terminal region outside of the known receptor- and DNA-binding motifs (38, 39). HMGB1 can act as a mobile and dynamic nucleo-cytoplasmic protein influencing multiple processes in chromatin such as transcription, replication, recombination, and DNA repair (39, 40), but HMGB1 can also be secreted into the extracellular milieu as a signaling molecule when cells are stressed (41). HMGB1 can bind exogenous and endogenous agents such as endotoxin, microbial DNA, and nucleosomes, and induces adaptive and innate immune responses via TLR2/4/9 followed by NF-κB activation (42, 43), contributing to inflammation, autoimmune dysregulation, and carcinogenesis. However, purified recombinant HMGB1 has little, if any, proinflammatory activity (44). Indeed, free HMGB1, that is unbound to endogenous/exogenous inflammatory factors, is able to suppress inflammatory reactions in noninjured tissues by inhibiting TLR-mediated NF-κB signaling (33). Biologically, HMGB1 can be regarded as a critical factor for maintaining tissue homeostasis in cleaning damaged/infected tissues (i.e., promoting intralesional inflammation), but also in protecting surrounding noninjured tissues (i.e., suppressing inflammation), and accelerating regeneration of damaged tissues by mobilizing and recruiting specific BM cells, that include epithelial progenitors when there is extensive epithelial injury, such as in Col 7-deficient RDEB skin.

The concept that a particular threshold concentration of HMGB1 in serum is relevant to mobilizing Lin⁻/PDGFRα⁺ BM cells and targeting them to damaged tissue also offers unique possibilities to augment a variety of other tissue repair mechanisms. It is likely that, in several other situations, systemic administration of HMGB1 to achieve serum levels similar to those observed in RDEB could be used as a therapeutic strategy to recruit stem/progenitor cells to accelerate regeneration of damaged tissue. Precisely which tissues might benefit from HMGB1-induced mobilization of Lin⁻/PDGFRα⁺ BM cells remains to be determined. What has also not been determined thus far are the dynamics of both the release of the Lin⁻/PDGFRα⁺ cells from the BM and the events that promote recruitment and migration within the target tissue and the other potential local microenvironment-induced and biochemical processes that contribute to improved wound healing (35) (discussed further in SI Materials and Methods). Nevertheless, we believe our data represent a significant advance in identifying a direction for potentially bringing stem/progenitor cell regenerative medicine to a broader clinical arena.

Materials and Methods

BMT.

BM cells were isolated under sterile conditions from 8- to 10-wk-old male C57BL/6 transgenic mice that ubiquitously expressed enhanced green fluorescent protein (GFP). Recipients were 8- to 10-wk-old female C57BL/6 mice that were lethally irradiated with 10 Gy of X-rays, and each irradiated recipient received 5 × 10⁶ BM cells from GFP transgenic mice kindly provided by Masaru Okabe (Osaka University). Experiments were performed on the BMT mice at least 6 wk after the BMT.

Mouse Skin Transplantation.

Full-thickness skin from wild-type and Col 7-null newborn mice (graft size ≈2 × 2 cm) was carefully isolated by excision after the mice had been euthanized under systemic anesthesia, and engrafted onto the backs of the GFP-BMT mice, wild-type BMT mice, and K5-Cre-GFP-BMT mice, with grafting just above the muscular fascia.

Immunofluorescent Microscopy Analysis.

The engrafted skins were removed, fixed with 2% paraformaldehyde, and subjected to immunofluorescent analysis. The excised skins were embedded in Tissue-Tec OCT Compound (Sakura Finetek), frozen on dry ice, and stored at −20 °C. For immunofluorescence staining, 6-μm-thick sections were labeled with rabbit polyclonal anti-mouse antibodies. Subsequently, sections were stained with goat anti-rabbit IgG secondary antibody.

Cell-Migration Assay.

Chemokinetic migration of Lin⁻/PDGFRα⁺ cells was assayed by using a modified Boyden chamber. In brief, 1.0 μg of HMGB1 in 27 μL of DMEM was added in the lower chambers, and 10⁶ cells/mL of Lin⁻/PDGFRα⁺ cells suspended in 50 μL of DMEM containing 10% FBS were added to the upper chamber. The cells on the lower side of the membrane were stained with Diff-Quick (Sysmex).