Human Dermal CD14+ Cells Are a Transient Population of Monocyte-Derived Macrophages

1Dendritic cells (DCs) and macrophages are a heterogeneous population of leukocytes that are critical in orchestrating immune responses. Human tissues are populated by at least three DC subsets; CD141hi DCs, CD1c+ DCs, and CD14+ DCs. Gene-expression studies suggest that human blood and tissue CD141hi DCs are homologous to murine tissue CD103+ and splenic CD8+ DCs and CD1c+ DCs are homologous to CD11b+CD4+ DCs in the spleen and CD11b+CD24+CD64− DCs in nonlymphoid tissues. However, the precise relationship of human CD14+ DCs to murine tissue populations remains unclear. Excluding Langerhans cells of the epidermis, the apparent paradox of three DC subsets in human interstitial tissues but only two in murine tissues remains unreconciled.

2Human CD14+ DCs were first identified as a spontaneously migrating population from dermal explants cultured ex vivo. These cells were classified as DCs based on major histocompatibility complex (MHC) class II glycoprotein expression and their ex vivo migratory behavior. In vitro generated CD14+ DCs from CD34+ hematopoietic stem cells (HSCs) have been used alongside primary cells to dissect their immunological properties. CD14+ DCs secrete interleukin-10 (IL-10) and IL-6 and have been shown to induce regulatory T cells (Tregs) and helper follicular T cells (Tfh). A notable feature of CD14+ DCs is their poor ability to stimulate allogeneic T cell proliferation.

3CD14+ DCs also express CD141, which is further upregulated during spontaneous migration from skin explant culture and initially presumed to be related to blood CD141+ DCs. More recently, the true counterpart of blood CD141+ DCs has been shown to be tissue CD14−CD141hi DCs. CD14+ cells are related to human and mouse blood monocytes by gene expression and are rapidly reconstituted by donor-derived cells following hematopoietic stem cell transplantation (HSCT), unlike dermal macrophages, which turn over at a much slower rate.

4In mice, steady-state DCs are derived from a lineage dependent on FLT3, in contrast to monocytes and macrophages, which are dependent on colony-stimulating factor-1 receptor (CSF-1R). Circulating murine Ly6Chi monocytes have been shown to extravasate into tissues existing as tissue monocytes and also differentiate into DC-like and macrophage populations in the intestine and dermis. Monocytes as a source of tissue inflammatory DCs are also well-documented. Human blood monocyte differentiation into DCs has been proposed in inflammation as the potential equivalent of in vitro cultured GM-CSF and IL-4 monocyte-derived DCs. However, the precise contribution of circulating monocytes to human tissue DCs and macrophages in steady state is unclear.

5Altogether, these findings led us to question whether CD14+ cells were bona fide DCs and which murine population was their homolog. In this study, we investigated the relationships between circulating blood CD14+ monocytes and tissue macrophages with tissue MHC classII+CD14+ cells, currently defined as “DCs.” We defined the transcriptomic profile of the human monocyte-macrophage lineage distinct from the DC lineage and demonstrated the conserved gene transcripts defining these two lineages in humans and mice. Our findings revealed that CD14+ cells more closely resemble tissue resident, monocyte-derived macrophages than bone fide DCs. In addition, we showed that the murine dermal monocyte-derived macrophages are the homolog of human dermal CD14+ cells.

6We previously showed that CD14+ cells were distinct from dermal macrophages, which possessed dense cytoplasmic melanin granules by morphology, were autofluorescent by flow cytometry analysis, were adherent and nonmigratory, although both populations express CD14. As the existence of tissue monocytes derived from Ly6Chi classical monocytes was recently demonstrated in mice, we therefore first set out to determine the phenotypic distinctions between dermal MHC classIIhiautofluorescent (AF)−CD14+ cells (hereafter referred to as CD14+ cells), AF+CD14+ macrophages (hereafter referred to as dermal macrophages), and blood CD14+ monocytes, the homologs of murine Ly6Chi monocytes. To control for the preparation conditions of freshly isolated CD14+ cells by enzymatic digestion, we cultured CFSE-labeled purified blood CD14+ monocytes with enzymatically-digested skin overnight. Gating on CFSE-labeled cells allowed direct comparison of tissue CD14+ with CD14+ monocytes. Overlay dot plot of CFSE-labeled CD14+ monocytes on digested skin cells showed that skin CD14+ cells were phenotypically distinct from blood monocytes with higher side scatter (SSC) properties, expressing higher amounts of HLA-DR and CD1c (Figure 1A). Both skin CD14+ cells and CFSE-labeled CD14+ monocytes spiked into the skin preparation had variable expression of CD141 (Figure 1A). We noted very few CD14+ cells with an identical profile to the CFSE-labeled monocytes cultured with digesting skin (Figure 1A), which might represent extravasated tissue monocytes in healthy human skin.

7We next compared the expression of selected antigens characterizing monocyte-macrophage cells in human skin and blood-antigen-presenting cell populations (Figure 1B). Unlike CD1c+ DCs from blood and skin, CD14+ cells had variable expression of CD163 similar to blood monocytes and macrophages (Figure 1B). We next compared the expression of skin antigen-presenting cell subsets and blood CD14+ monocytes for the following transcripts; DC transcription factor (TF) ZBTB46, CD209, lymphatic vessel endothelial hyaluronan receptor (LYVE1), and factor XIIIa (F13A1). The latter three antigens were shown to identify dermal macrophages in situ (Figure 1C). Skin CD1c+ and CD141hi DCs expressed higher amounts of ZBTB46 transcripts and low amounts of CD209 and LYVE1. In contrast, dermal CD14+ cells expressed 40% and macrophages 15% of ZBTB46 transcript amounts compared to CD1c+ DCs, consistent with previous reports on human inflammatory DCs and murine dermal macrophage populations. CD14+ cells and macrophages expressed higher amounts of CD209 transcript compared to all other subsets, but LYVE1 transcript expression was highest in macrophages (Figure 1C) similar to F13A1 expression. In addition, dermal CD14+ cells expressed high amounts of IL1Α and gamma-glutamyl transferase 5 (GGT5) transcripts (Figure 1C), which were identified from our previous microarray analysis. Immunostaining of whole-mount dermal sheet for LYVE-1, CD209 (DC-SIGN), and FXIIIa identified LYVE-1−FXIIIaloDC-SIGN+ cells corresponding to the CD14+ cells in situ (Figure 1D).

8Human tissue DCs are depleted in patients lacking circulating DCs and monocytes as shown by patients with genetic deficiency of monocytes and DCs due to GATA2 or IRF8 mutation, but the exact precursor-progeny relationships are difficult to demonstrate conclusively in humans. We previously showed that skin CD14+ cells are derived from donor bone marrow within 40 days of HSCT, but the kinetics of this relationship have not been resolved in detail. In this study, we followed the course of blood and skin monocytes, macrophages, and DCs during preparative cytotoxic therapy for transplantation and for up to 2 weeks after HSCT (clinical data in Table S1). Cytotoxic therapy induced bone-marrow suppression and absolute monocytopenia by the day of transplantation (day 0) (Figure 2A). This was mirrored by a rapid loss of CD14+ cells from the skin within 6 days. After a delay of 6 days, macrophage numbers also declined to approximately 50%, where they remained stable. In the early recovery phase after HSCT, there was a rapid rise in CD14+ blood monocytes, which coincided with a rapid reconstitution of skin CD14+ cells (Figure 2B). The kinetics of DC recovery were slower in the blood and skin and did not attain the same frequency as healthy controls (Figures 2A and 2B). The temporal relationship between blood CD14+ monocytes and skin CD14+ cells is consistent with a precursor-progeny relationship in which CD14+ monocytes give rise to CD14+ cells within a short timeframe. In support of the rapid differentiation step of CD14+ monocytes into skin CD14+ cells suggested by our in vivo findings, we showed that purified CD14+ monocytes upregulated antigens and adopted morphological changes characteristic of CD14+ tissue cells upon culture with primary endothelial cells over 3 days (Figure 2C).

9In order to evaluate the lineage identity of human skin CD14+ cells, we performed microarray transcription profiling of human skin and blood dendritic cells, macrophages, and monocyte subsets. Principal component analysis (PCA) of all subsets analyzed showed separation of DCs from monocyte-macrophages in component one and further definition between monocyte-macrophage subsets and plasmacytoid DCs (pDCs) from myeloid DCs in component 2 (Figure 3A). We also performed a supervised connectivity map (CMAP) gene set enrichment analysis and showed that the CD14+ cell gene set was enriched in dermal macrophages and also weakly in CD14+ blood monocytes but exhibited an inverse expression profile to blood and skin DCs (Figure 3B). This suggests that CD14+ cells are more closely aligned to tissue macrophages and CD14+ blood monocytes than to blood or tissue DCs.

10Further examination of the relationship between CD14+ cells, monocytes, macrophages, and DCs revealed a number of coregulated genes distinguishing monocytes, CD14+ cells, and macrophages from blood and skin DCs (Figure 3C). Functional pathways identified by the monocyte, CD14+ cell, and macrophage gene sets include retinoid X receptor signaling, TREM1 signaling, complement system, and communication between innate and adaptive cell regulation (see Figure S1 available online). In contrast, the human DC gene signature was enriched for cell-cycle control and amino acid, nucleic acid and cholesterol metabolism pathways (Figure S1).

11We next performed cross-species analysis comparing human monocyte, macrophage, and CD14+ cells with murine monocytes, macrophages, and DC subsets obtained from ImmGen and GSE49358 microarray data sets. This analysis identified a set of genes that are differentially expressed in a conserved manner, which include SLC11A1, MAFB, CD14, and FCGR2A (Figure 3D; Table S2). Similarly, human and mouse DC lineage also shared close homology of transcripts across species including FLT3, BTLA, HLA-DOA, and CIITA (Figure 3D; Table S2).

12A defining property of tissue-resident DCs is their migratory capacity to lymph node (LN) directed by CCL19 and CCL21 signaling through CCR7. In vitro culture of explanted tissue mimics this process and resident DCs might be observed entering the lymphatic channels prior to emigrating from the tissue. In addition, this experiment showed that resident macrophages remain fixed in the tissue. The ability of CD14+ cells to leave tissue explants has been invoked as a DC credential, but their route of migration in the ex vivo assay has not yet been established. If CD14+ cells are not fixed in skin explants, then simple redistribution in ex vivo culture would result in apparent emigration from the tissue. Like dermal macrophages, migrated and digested CD14+ cells did not express CCR7 even upon stimulation (Figure 4A). Time-course analysis of skin explants showed the presence of CD14+ cells in the skin explant medium as early as 12 hr after culture (Figure 4B). However, at no stage were CD14+ cells observed within lymphatics as assessed by whole-mount immunostaining of skin explants (Figure 4C). DC-SIGN expression is retained by spontaneously migrating CD14+ cells (Figure S2) and would have permitted their localization within lymphatic channels if this had been the route of migration. The presence of spontaneously migrated CD14+ cells, despite their absence within skin lymphatic lumen, suggests that CD14+ cells exited from the skin without entering the lymphatic vasculature. Dermal macrophages did not migrate spontaneously (Figure 4C), in keeping with previous observations.

13Migratory CD14+ cells have been much studied, but their relationship to CD14+ cells isolated by enzymatic digestion has not been extensively evaluated. We compared cells isolated by digestion and migration for the expression of a custom selection of 96 genes reported in the literature to define DCs and macrophages by Taqman Low Density Array PCR. Unsupervised clustering showed that migrated and digested CD1c+ DCs and CD14+ cells clustered by subset rather than isolation method (Figure 4D). CCR7, CYP27B1, FLT3, FSCN1, INDO, LAMP3, and LY75 were expressed at higher amounts by CD1c+ DCs compared to CD14+ cells, which expressed higher amounts of CCL18, CCL3, CCR1, CD163, CD36, CLEC10A, FCGR1A and 3A, IL10, MARCO, MMP12, MSR1, SIGLEC1, and TREM2 (Figure 4D), transcripts characteristic of monocyte-macrophage cells.

14A cardinal property of DCs as opposed to monocytes and macrophages is their superior ability to activate naive T cell proliferation. It has been previously shown that CD14+ cells from the skin are inferior to CD1c+ DCs in inducing allogeneic naive T cell proliferation, but a direct comparison with dermal macrophages has never been performed. After 5 days culture with CFSE-labeled allogeneic naive CD4+ T cells, CD14+ cells induced 80% lower proliferation of naive CD4+ T cell compared to CD1c+ DCs with negligible T cells proliferation observed with macrophages even at the highest APC: T cell ratio tested of 1:10 (Figure 5A). However, CD1c+ DCs, CD14+ cells, and dermal macrophages were potent inducers of memory CD4+ T cell proliferation and cytokine production upon stimulation with Candida albicans (Figure 5B). Both CD14+ cells and dermal macrophages were comparable to CD1c in their ability to induce IL-17, IL-22, interferon-γ (IFN-γ), and IL-4 production by memory CD4+ T cells (Figure 5B).

15Recent studies have demonstrated heterogeneity within murine nonlymphoid tissues CD11b+ cells, which comprise DCs, monocytes, monocyte-derived DCs, and resident macrophages. In order to precisely identify the murine equivalent of human CD14+ cells, we performed comparative transcriptomics analysis by using microarray data of the recently described CD11b+ monocyte, DC, and macrophage populations in murine dermis. CMAP analysis revealed highest enrichment of human CD1c+ DCs to their murine CD11b+ DC counterpart in both steady state and upon inflammation induced by the contact hypersensitivity allergen DNFB (Figure 6A). The reciprocal relationship was also observed that human dermal macrophages had low but positive CMAP enrichment scores with murine macrophages and monocyte-derived cells (Figure 6A). Blood CD14+ monocytes had the highest enrichment score with mouse Ly6Chi blood monocytes in the steady state (Figure 6A), in keeping with previous analysis. Human CD14+ cells were most enriched with murine macrophages (P4 and P5) followed by Ly6CloMHC class II+ monocyte-derived DC-like cells (P2 and P3) (Figure 6A). These results suggest that human CD14+ cells are related to monocytes but are not the equivalent of murine tissue monocytes (P1). The existence of murine monocyte-derived dermal macrophages was recently reported by Tamatounour et al. as observed by a reduction in P4 and P5 dermal macrophages in Ccr2−/− mice. We therefore hypothesized that human CD14+ cells were the homolog of murine dermal monocyte-derived macrophages. In order to confirm the monocyte origin of some murine dermal macrophages, we used the S100a4-cre × R26YFPmice in which >99% of hematopoietic stem cells (HSCs) and resultant blood monocytes express yellow fluorescent protein (YFP). This fate-mapping model previously demonstrated the independence of resident tissue macrophages from circulating monocytes and HSC progenitors. Our analysis of mouse dermal DC and macrophage fractions showed that >90% of CD11b+ P1-P3 and 80%–90% of P4 and P5 macrophages were indeed monocyte-derived (Figure 6B). Collectively, these results provide further evidence of functional equivalence between monocyte-derived macrophages in the mouse dermis as the homologs of human CD14+ cells.

16The results presented here suggest that CD14+ “DCs” are not related to the human DC lineage but are monocyte-derived macrophages that are resident in healthy skin. Their gene-expression program strongly overlaps with that of blood monocytes and resident tissue macrophages, but they are distinguishable from both of these by phenotypic and functional properties. Although it is difficult to prove unequivocally that they have a monocyte origin, their absence in monocyte deficiency states, kinetics of renewal after HSCT, and similarity to monocytes in short-term culture with endothelial cells are all consistent with a precursor-progeny relationship. Their evident lack of ability to stimulate naive T cell proliferation is consistent with their status as a monocyte-derived macrophage.

17Tissue monocytes coexpressing Ly6C and MHC class II were recently demonstrated as a distinct population in murine tissues. These cells, derived from Ly6ChiMHC class II− circulating monocytes, upregulate MHC class II upon contact with endothelium. In human, all blood and skin DCs, monocytes and macrophages express MHC class II (reviewed in ). The Lin−MHC class II− compartment in human blood does not contain any CD14-expressing cells and primarily comprises basophils. “Spiking” CFSE-labeled blood CD14+ monocytes into digesting skin allowed us to perform a direct comparison between dermal CD14+ and blood monocytes.

18The identity of the circulating precursors of human tissue DCs and macrophages has been a subject of intense debate. Despite the widespread use of in vitro culture protocols to generate monocyte-derived DCs and macrophages, in vivo evidence to support the contribution of such monocyte-derived cells in healthy tissue is limited. Here, we show the rapid decline and reconstitution of dermal CD14+ cells that mirrors the kinetics of blood CD14+ monocytes in patients undergoing HSCT. In contrast, blood and skin CD1c+ DCs remained suppressed up to 14 days after HSCT and are thus unlikely to be the precursors of dermal CD14+ cells. We also show that blood CD14+ monocytes acquire the morphology and phenotypic characteristics of dermal CD14+ cells upon culture with endothelial cells in line with previous reports. The reduction of dermal macrophages after HSCT by approximately 50% is in keeping with the reduction in dermal macrophages in patients with GATA2 and IRF8 mutation who are deficient in peripheral blood DCs and monocytes. Longer-term follow up after HSCT is required to evaluate macrophage recovery and whether proliferation of residual dermal macrophages or differentiation from CD14+ cells is responsible for dermal macrophage reconstitution.

19The utility of transcriptomics analysis to map homologous subsets between humans and mice and more recently to define DCs and macrophages in mouse tissues is evident. Extending our previous CMAP analysis to human microarray data set incorporating dermal macrophages and epidermal LCs, we were able to precisely identify the monocyte-macrophage lineage of dermal CD14+ cells. This identification enabled us to define the human DC and monocyte-macrophage transcriptomic signatures and to perform cross-species analysis with the murine DC, monocyte, and macrophage ImmGen data set. The extension of human DC, monocyte, and macrophage mapping further enhances the utility of the ImmGen data repository. Our analysis confirms a high-degree of conservation for DC and macrophage transcription networks between humans and mice identifying shared genes such as MERTK, CD14, and SLC11A1 (NRAMP) to define monocytes and macrophage and FLT3, BTLA, and KIT to define DCs. Comparative biology analysis of DC subsets has supported the relevance of murine models for both developmental and functional studies and the extension of a similar analysis to the macrophage lineage could present an additional perspective for future studies on these cells.

20The phenotypic and transcriptional assignment of dermal CD14+ cells as monocyte-derived macrophages led us to reevaluate their ability to migrate spontaneously from skin explant cultures ex vivo which has been presumed to simulate lymphatic migration despite absent or very low expression of CCR7 even upon cytokine and LPS stimulation. We observed that although 1 in 5 HLA-DR+ cells after 12 hr and 48 hr culture of skin explant were CD14+ cells, these cells were not observed within skin lymphatic lumen. The dichotomy between spontaneous migration and lymphatic migration was previously observed with murine LCs and DCs from Ccr7−/− mice, which were capable of spontaneously migrating from skin explants but failed to form dermal cords indicative of lymphatic migration or enter skin draining lymph node.

21It is well documented that CD14+ cells are poor stimulators of naive T cells, a property that is expected of tissue-resident cells. However, CD14+ cells and macrophages are on par with DCs in regulating memory CD4+ T cell responses as shown here and in previous reports. Dermal CD14+ cells express high amounts of IL-1α and GGT5, a property not shared by any blood or skin DCs, monocytes, and macrophages, suggesting a role in maintaining epithelial integrity and regulation of skin inflammation including neutrophil migration.

22Murine tissue macrophages and epidermal Langerhans cells (LCs) were recently shown to arise from embryonic yolk sac and fetal liver precursors challenging the traditional dogma of monocytes as precursors of all tissue macrophages. However, recent evidence including this report suggests a contribution by circulating monocytes to the murine tissue macrophage pool. The functional differences between cells derived from these two origins have not been defined. The contemporary view of the mononuclear phagocyte system encompasses several precursor origins including circulating monocytes as precursors of tissue macrophages. Here, we report in humans, two populations of tissue resident cells, CD14+ monocyte-derived macrophages and fixed macrophages that map to the CD11b+ pool of mouse macrophages. Further work is required to evaluate the long-term contribution of monocytes to the resident macrophage pools.

23The functional classification of human tissue CD14+ cells as monocyte-derived macrophages will redirect attention to functional regulation in the skin as opposed to lymph node. A greater understanding of the contribution of monocyte-derived cells in steady state, inflammation, wound healing, and pathology characterized by histiocytosis and granuloma formation will provide further insights into exploiting their origin and functional properties for clinical therapy.

24Human samples were obtained in accordance with a favorable ethical opinion from Newcastle, Singapore SingHealth, and National Health Care Group Research Ethics Committees. Normal skin was obtained from mammoplasty and breast reconstruction surgery and digested whole (1 × 1 cm2) as previously described to obtain single-cell suspension. Migrating cells were collected from whole skin pieces (1 × 1 cm2) cultured in RPMI with 10% FCS, and analyzed at serial time points between 0 and 72 hr. Viability was >90% by DAPI exclusion (Sigma). Where stated, 100,000 CFSE-labeled CD14+ blood monocytes were cultured together with 1 × 1cm2 skin piece during collagenase digestion. Shave biopsies were performed on HSC transplant patients with a DermaBlade® (Shuco). Whole single-cell suspensions were then immunostained and analyzed by flow cytometry. Peripheral blood mononuclear cells were isolated by density centrifugation (Lymphoprep; GE Healthcare). Blood and dermal DC subsets, naive and bulk CD4+ T cells were isolated to >91% purity by fluorescence-activated cell sorting (FACS) using a FACSAriaII and FACSFusion (Becton Dickinson [BD]).

25Flow cytometry was performed on a BDLSRII, BDFortessa, and FACSCanto, and data were analyzed with FlowJo (Treestar). Antibodies used are listed in Supplemental Experimental Procedures.

26Whole-mount immunofluorescence staining was performed with a previously described protocol. We fixed a 200 μm skin sheet in PBS containing 2% paraformaldehyde and 30% sucrose overnight at 4°C. Skin was incubated overnight in PBS containing 0.5% BSA and 0.3% Triton X-100 before staining with primary and secondary antibodies at 4°C overnight at each stage. Antibodies used are listed in Supplemental Experimental Procedures. After staining, tissue samples were immersed in VECTASHIELD mounting medium with DAPI (Vector Laboratories). Specimens were viewed using Axio Imager.Z2 fluorescence microscope with Axiovision software v4.8 and Axiocam MR3 camera (Carl Zeiss, Inc.) or Leica Leica TCS SP2 UV confocal microscope and LCS V 2.51 imaging software (Leica). Three-dimensional reconstruction was performed with Imaris7.6.2 soft- ware (www.bitplane.com).

27We cultured 5,000 FACS-sorted dermal DC subsets with 100,000 CFSE-labeled allogeneic naive (CD4+CD25−CCR7+CD45RO−) T cells in U-bottomed 96-well plate. Proliferation was measured by CFSE dilution on day 6. For memory T cell stimulation, autologous blood CD4+ T cells were used to measure recall memory response to Candida albicans. We pulsed 5,000 FACS-sorted dermal DC subsets with Candida albicans overnight, and cultured them with 100,000 CFSE-labeled (Invitrogen) autologous blood CD4+ T cells on the following day. Media was replenished as necessary throughout the duration of the culture. Cytokine production was measured on day 10, upon stimulation with 20 ng/ml PMA (Sigma-Aldrich; Sigma) and 500 ng/ml Ionomycin (Sigma) for 5 hr in the presence of 10 μg/ml Brefeldin A (Sigma-Aldrich) for the last 2 hr. Cells were fixed and permeabilized (eBioscience Fix/Perm Buffer Set) to allow intracellular cytokine staining. Viaprobe staining (Becton Dickinson; BD) was performed prior to cell fixation to distinguish viable cells.

28HUVECs were cultured in Endothelial Cell Basal Medium 2 (Promo Cell) with the following supplements (FCS, endothelial cell growth supplement, epidermal growth factor, Insulin-like growth factor, vascular endothelial growth factor 165, ascorbic acid, heparin, hydrocortisone [Promo cell]). Dermal fibroblasts were cultured in RPMI with 20% FCS, 1000 u/ml Penicillin and Streptomycin and 2mM L-Glutamine. HUVECs or fibroblasts (30,000/well) were seeded in a 24-well plate and cultured in their respective media (500 μl) for 18 hr at 37°C, 5% CO2. FACS-purified blood monocyte subsets were added to HUVECs or fibroblast cell cultures (50,000 monocytes/well to a final volume 1 ml) and analyzed on days 1 and 3 after coculture.

29Absolute whole-blood leukocyte analysis was performed as described previously. Briefly, 100 μl of whole blood was transferred directly to Trucount tubes (BD). Antibodies were added directly and staining was performed at RT for 20 min, followed by red blood cell lysis by adding 900 μl of BD PharmLyse reagent for 10 min at RT. Samples were then analyzed directly by flow-cytometry. Absolute number of cells per microliters of blood was calculated according to the manufacturer’s protocol.

30S100a4-cre mice were purchased from Jackson Laboratory and crossed to R26-stop-EYFP mice in house. Animals positive for the R26-stop-EYFP-flox construct were used for the experiment. Only sex (female) and aged (6–8 weeks) matched mice were used. Mouse skin cells were isolated as described previously. Briefly, mouse ears were split into dorsal and ventral halves and floated in RPMI-1640 medium (Sigma) containing 1 mg/ml dispase (Invitrogen) for 60 min to allow separation of epidermal and dermal sheets. Dermal sheets were then cut into small pieces and incubated in RPMI containing 10% serum and 0.8 mg/ml collagenase type IV (Worthington-Biochem; 250 U/mg) for 2 hr. Cells were then passed through 19 G syringe and filtered through 100 uM cell strainer (BD Falcon) to obtain a homogenous cell suspension.

31Total RNA was extracted using the RNeasy Micro Kit (QIAGEN) and treated with DNase I according to manufacturers’ instructions (QIAGEN). RNA was used as a template for complementary DNA (cDNA) synthesis using the RevertAid H Minus First Strand cDNA Synthesis Kit with the manufacturers’ protocol (Thermo Scientific, Fisher Scientific UK). Real-time PCR was performed with TaqMan Gene-Expression Master Mix and recommended TaqMan Gene-Expression Assays according to manufacturers’ instructions (Life Technologies). Reactions were performed using an ABI 7900HT Fast Real-Time PCR System with the instrument’s default settings for a standard run. Relative quantification of the messenger RNA (mRNA) amounts was performed using the ΔCT method and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the reference gene. Taqman assay ID and sequence details are available in Supplemental Experimental Procedures. Details of transcriptomics analysis are available in Supplemental Experimental Procedures.

32All statistical analyses were performed using Prism 5.0 (GraphPad Software). All p values are two-tailed using Mann-Whitney U test.

33M.H., N.M., A. Schlitzer, M.G., M.C., and F.G., conceived the study and designed the experiments. M.H., N.M., M.G., A. Schlitzer, L.J., A. Shin, S.P., E.P., X.-n.W, R.D., K.B., S.C., V.B., and P.M. performed the experiments and analyzed the data. M.H., M.G., N.M., F.G., and M.C. wrote the manuscript. A.L. provided intellectual input. K.D., P.S.W., K.G., and M.P. contributed to the bioinformatics analysis and manuscript preparation. D.M.S commented on the manuscript.