Keratinocyte integrin α3β1 induces expression of the macrophage stimulating factor, CSF-1, through a YAP/TEAD-dependent mechanism

Abstract

The development of wound therapy targeting integrins is hampered by inadequate understanding of integrin function in cutaneous wound healing and the wound microenvironment. Following cutaneous injury, keratinocytes migrate to restore the skin barrier, and macrophages aid in debris clearance. Thus, both keratinocytes and macrophages are critical to the coordination of tissue repair. Keratinocyte integrins have been shown to participate in this coordinated effort by regulating secreted factors, some of which crosstalk to distinct cells in the wound microenvironment. Epidermal integrin α3β1 is a receptor for laminin-332 in the cutaneous basement membrane. Here we show that wounds deficient in epidermal α3β1 express less epidermal-derived macrophage colony-stimulating factor 1 (CSF-1), the primary macrophage-stimulating growth factor. α3β1-deficient wounds also have fewer wound-proximal macrophages, suggesting that keratinocyte α3β1 may stimulate wound macrophages through the regulation of CSF-1. Indeed, using a set of immortalized keratinocytes, we demonstrate that keratinocyte-derived CSF-1 supports macrophage growth, and that α3β1 regulates Csf1 expression through Src-dependent stimulation of Yes-associated protein (YAP)-Transcriptional enhanced associate domain (TEAD)-mediated transcription. Consistently, α3β1-deficient wounds in vivo display a substantially reduced number of keratinocytes with YAP-positive nuclei. Overall, our current findings identify a novel role for epidermal integrin α3β1 in regulating the cutaneous wound microenvironment by mediating paracrine crosstalk from keratinocytes to wound macrophages, implicating α3β1 as a potential target of wound therapy.

INTRODUCTION

Following cutaneous injury, epidermal repair is rapidly triggered to restore the skin barrier. During this process of repair, wounded skin is vulnerable to pathogens. Macrophages are critical to the wound healing process because they clear debris, modulate inflammation, and coordinate tissue repair [1]. Apart from their phagocytic activity, macrophages secrete and respond to various cytokines/chemokines and growth factors into the wound milieu that promote healing. CSF-1, also known as macrophage colony stimulating factor, or M-CSF, is a hemopoietic growth factor and the primary regulator of macrophage growth, survival, polarization, and recruitment [2]. CSF-1 can be contributed by macrophages, but it is also expressed by other cell types including epithelial cells, fibroblasts, osteoblasts, and many types of tumor cells [3].

Integrins are heterodimeric transmembrane receptors that bind to extracellular matrix (ECM) and facilitate bidirectional cellular-extracellular signaling [4]. Integrins expressed in the epidermis have been shown to play critical roles in wound healing. Beyond mediating keratinocyte adhesion and migration, integrins have also been shown to regulate the expression of genes that encode proteins crucial for proper wound healing [5]. Keratinocyte integrin α3β1 binds to laminin-332 in the basement membrane of the skin where it regulates the keratinocyte secretome, thereby promoting crosstalk to distinct cells in extra-epidermal compartments of the skin [6]. Indeed, we have shown that keratinocyte α3β1 promotes angiogenesis and regulates fibroblast differentiation in cutaneous wounds via secreted factors MRP-3 and IL-1α, respectively [7, 8]. Furthermore, mass spectrometry analysis revealed an α3β1-regulated keratinocyte secretome consisting of several cytokines and immune cell modulators, including CSF-1 [6]. α3β1-deficient epidermal tumors had less tumor cell-derived CSF-1 and fewer tumor-associated macrophages in a murine model of skin tumorigenesis [6], although how epidermal α3β1 regulates CSF-1 had been unclear.

Yes-associated protein (YAP) and transcriptional co-activator with a PDZ-binding domain (TAZ), are paralogous transcriptional co-activators that regulate gene expression through the TEAD family of transcription factors and have critical roles in epithelial homeostasis, organ development, tissue regeneration, immune modulation, and repair [9, 10]. YAP and TAZ have been shown to be elevated after wounding [11, 12]. YAP and TAZ are negatively regulated by the Hippo Pathway, a kinase cascade that, when active, leads to the LATS-mediated phosphorylation of YAP and TAZ resulting in their cytoplasmic sequestration or degradation [13]. Thus, the absence of Hippo signaling allows for unphosphorylated YAP and TAZ to enter the nucleus where they can interact with TEAD family transcription factors to regulate target gene expression, controlling cellular processes including motility, proliferation, apoptosis, and differentiation [14–16]. While numerous cellular pathways can influence the Hippo pathway to regulate YAP and TAZ [10, 17], several Hippo Pathway-independent mechanisms of regulation also exist [18]. The Hippo-YAP/TAZ pathway has established roles in both the dermal and epidermal compartments of the skin during development, homeostasis, and repair, and dysregulation of the Hippo Pathway is linked to several skin diseases [19]. However, it remains unclear which YAP/TAZ-dependent gene expression changes drive skin homeostasis and wound repair, and mechanisms of YAP and TAZ activation during wound healing are poorly understood.

The dynamic processes of tumorigenesis and wound healing share many similarities, and tumors have long been described as wounds that do not heal [20]. The idea that tumors co-opt the wound healing response to create a stroma that is hospitable to their maintenance and growth continues to gain support [21]. This notion, combined with our recent findings in skin tumors mentioned above [6], led us to hypothesize that integrin α3β1 on keratinocytes of healing wounds stimulates wound macrophages through regulation of CSF-1. α3β1 has been shown to regulate YAP/TAZ activity through focal adhesion kinase (FAK) to maintain tissue renewal in a mouse incisor model [22], and YAP induced the expression and secretion of CSF-1 in a model of pancreatic ductal adenocarcinoma [23], leading us to further hypothesize that α3β1-dependent activation of YAP promotes the expression of CSF-1 in keratinocytes. In the current study we used a genetic model of inducible α3 knockout to show that wounds deficient in epidermal α3β1 do indeed have reduced CSF-1 and fewer wound-proximal macrophages. We also used a set of immortalized keratinocytes to show that α3β1 promotes Csf1 mRNA expression through a Src-YAP/TAZ-TEAD-dependent mechanism, and that keratinocyte-derived CSF-1 supports macrophage growth. Our findings demonstrate a novel role for α3β1 as a critical regulator of keratinocyte-to-macrophage crosstalk within the cutaneous wound microenvironment, and they identify α3β1-dependent, YAP-mediated induction of Csf1 as a critical component of this crosstalk.

RESULTS

Absence of integrin α3β1 in wound epidermis is associated with reduced epidermal expression of CSF-1 and fewer wound-proximal macrophages

To test the hypothesis that α3β1 on keratinocytes promotes CSF-1 production by the epidermis during wound healing, and thus impacts the wound macrophage population, we utilized our recently developed keratin-14 (K14) CreERT:α3^flx/flx mice, wherein the Itga3 gene that encodes the integrin α3 subunit is knocked out specifically in epidermal keratinocytes by topical application of (Z)-4-hydroxytamoxifen (4OHT) [6]. The backs of mice were pre-treated five and three days prior to wounding with 4OHT, or vehicle as control, then using a biopsy punch, full thickness wounds were made. Immunostaining of biopsy core skin showed that α3 expression in 4OHT-treated mice was undetectable at the time of wounding (Fig. S1). Expression of α3 remained undetectable 3 days post-wounding (Fig. 1a), which is the wound timepoint on which we focus in this study. Hereafter, wounds of 4OHT-treated mice will be referred to as α3 epidermal knockout (α3eKO), and wounds of vehicle-treated mice as control. In control wounds, CSF-1 was upregulated in wound edge epidermis compared to epidermis of distal unwounded skin (Fig. S2), coincident with the upregulation of integrin α3β1 in control wound epidermis (Fig. 1a). Epidermis of α3eKO wounds expressed significantly less CSF-1 protein, as determined by immunofluorescence (IF) (Fig. 1b, Fig. S2). Since several stromal cell types may contribute CSF-1 to the wound microenvironment, we also assessed Csf1 mRNA expression within the wound epidermis using RNAScope in situ hybridization (ISH), which allows for single-molecule mRNA visualization within the spatial context of a tissue [24]. Positive and negative control probes were routinely used for technical quality control (Fig. S3). Analysis of wound cryosections with a probe set specific for murine Csf1 mRNA showed that this transcript was reduced substantially in epidermis of α3eKO wounds, compared with control wounds (Fig. 1c). Importantly, α3eKO wounds had fewer F4-80-positive macrophages in the wound bed beneath re-epithelialized wound epidermis compared to control wounds (Fig. 1d). Taken together, these results suggest that integrin α3β1 on wound keratinocytes induces CSF-1 expression that may stimulate wound macrophages.

Figure 1.

α3eKO wounds express less epidermal CSF-1 and have fewer wound-proximal macrophages. Cryosections of 3-day wounds were prepared from control or α3eKO mice as in Fig. S1 and stained by IF with anti-α3 (green, upper and lower panels) and DAPI (blue, lower panels only); arrowheads indicate edge of wound epidermis; scale bars, 25 μm (upper panel), 100 μm (lower panel). (b) IF with anti-CSF-1 (red) and DAPI (blue); dashed line, outlines wound epidermis; scale bar, 500 μm. Graph shows relative CSF-1 mean fluorescence intensity. (c) ISH to detect epidermal-derived Csf1 mRNA was performed on cryosections; Csf1 mRNA (purple) and DAPI (blue); images were taken fully within wound epidermis; scale bar, 25 μm. Graph shows quantification of relative mRNA puncta. (d) IF with anti-F4/80 (red), anti-K14 (green), and DAPI (blue); box indicates area of inset that is magnified on right with (lower) and without (upper) DAPI; arrows, wound-proximal macrophages; scale bar, 50 μm. Graph shows percentage of cells in wound bed that are F4/80-positive. (b–d) Data are mean +/− s.e.m.; n ≥ 5 wounds per genotype; * p<0.05; *** p<0.001. (a,b,d) e, epidermis; d, dermis; s, scab; wb, wound bed.

Keratinocyte integrin α3β1 promotes CSF-1 expression and supports macrophage growth

To study the mechanism through which epidermal integrin α3β1 regulates CSF-1, we used a set of mouse keratinocyte (MK) cell lines that we derived previously from a wildtype mouse (MKα3+/+) or an α3-null mouse (MKα3−/−) [25], as well as a variant of the MKα3−/− cell line in which α3β1 was restored through stable transfection with human α3 (MKα3res) [26] (Fig. 2a). Consistent with our in vivo results, MKα3−/− cells express significantly reduced Csf1 mRNA levels compared to MKα3+/+ cells (Fig. 2b). Furthermore, immunoblot analysis of conditioned medium (CM) showed that secretion of CSF-1 is reduced in MKα3−/− cells compared to MKα3+/+ cells (Fig. 2c,d). Ponceau staining of CM from the MK set shows comparable protein loading (Fig. S4). Reconstitution of α3β1 in MKα3res cells restored CSF-1 mRNA expression and increased the level of secreted CSF-1 in the CM from these cells (Fig. 2b–d). Additionally, in vitro macrophage growth was substantially higher when cultured in CM from either of the α3-expressing MK lines (MKα3+/+ or MKα3res) compared to CM from MKα3−/− cells (Fig. 2e), suggesting that keratinocyte integrin α3β1 supports macrophage growth via a secreted factor(s).

Figure 2.

Keratinocyte integrin α3β1 promotes CSF-1 expression and supports macrophage growth. (a) Immunoblot of whole cell lysates to validate α3 expression levels in MK cells that lack (α3−/−) or express (α3+/+, α3res) α3β1; ERK, loading control. (b) qPCR quantification of relative Csf1 mRNA levels in MK cells. (c) Representative immunoblot of secreted CSF-1 in CM from MK cells. (d) Quantification of secreted CSF-1 shown in panel c. (e) Macrophage density across 6 consecutive days, represented as fold-change from day 1, when grown in medium conditioned by MK cells; statistical analysis performed on day 6 data (b, d, e). Data are mean +/− s.e.m.; n ≥ 3; * p<0.05; ** p<0.01.

To determine whether keratinocyte-derived CSF-1 is required for macrophage growth, we utilized siRNA to knockdown CSF-1 in α3β1-expressing MKα3+/+ cells. Two distinct Csf1-targeting siRNAs each significantly reduced Csf1 mRNA (Fig. 3a) and secreted CSF-1 protein (Fig. 3b) compared to a control siRNA. Importantly, macrophages cultured in CM from CSF-1-knockdown cells had reduced growth compared to CM from control siRNA-treated cells (Fig. 3c). Overall, these results suggest that CSF-1 secreted by keratinocytes is required to promote macrophage growth in a paracrine manner.

Figure 3.

Macrophage growth is reduced in CM from CSF-1-knockdown keratinocytes. MKα3+/+ cells were treated with one of two distinct siRNAs (#1 or #2) that target Csf1 gene transcripts or a non-targeting siRNA as control. (a) qPCR quantification of Csf1 mRNA levels. (b) Quantification of relative secreted CSF-1, determined by immunoblot of CM from MK cells. (c) Macrophage density across 6 consecutive days, represented as fold-change from day 1, when grown in medium conditioned by MKα3+/+ cells treated with control siRNA (circles), or with CSF-1-targeting siRNA #1 (triangles) or #2 (squares). Statistical analysis was performed on day 6 data. (a–c). Data are mean +/− s.e.m.; n = 3; * p<0.05; ** p<0.01.

Integrin α3β1 promotes YAP/TEAD-dependent expression of Csf1

It has been shown previously that integrin α3β1 promotes YAP activation in a mouse incisor model [22], and that YAP activity promotes expression of CSF-1 in a model of pancreatic ductal carcinoma [23], as well as in fibroblasts [27]. Since YAP and TAZ regulate many genes by partnering with TEADs, we hypothesized that keratinocyte α3β1 promotes Csf1 expression in a YAP/TAZ-TEAD dependent manner. Utilizing a TEAD transcriptional reporter construct in which a TEAD-dependent promoter drives the expression of firefly luciferase [28, 29], we found that MKα3−/− cells had dramatically reduced TEAD transcriptional activity compared to MKα3+/+ cells (Fig. 4a), indicating that α3β1 is required to maintain YAP/TAZ-TEAD activity in keratinocytes. Next, to test if inhibition of YAP/TAZ-TEAD function leads to reduced CSF-1 expression, we treated MKα3+/+ cells with a pharmacological TEAD inhibitor, MGH-CP1 [30]. As expected, MGH-CP1 treatment reduced the activity of the TEAD reporter compared to vehicle-treated cells (Fig. 4b), as well as the expression of the canonical YAP/TAZ-TEAD-dependent genes, Ctgf and Cyr61 (Fig. 4c). Consistent with our hypothesis, MGH-CP1-treated MKα3+/+ cells also showed significantly reduced expression of Csf1 (Fig. 4c). However, expression of Bmp1, another α3β1-dependent gene in MK cells [31] was not significantly altered by TEAD inhibition (Fig. 4c), indicating that not all α3β1-regulated genes are TEAD-dependent. These data place YAP/TAZ-TEAD-dependent transcription downstream of integrin α3β1 and upstream of Csf1 expression.

Figure 4.

Inhibition of integrin α3β1-dependent YAP/TAZ-TEAD activity reduces Csf1-expression in MK cells. TEAD-dependent transcriptional activity was measured using TEAD-reporter assays on (a) MKα3+/+ versus MKα3−/− cells, or (b) MKα3+/+ cells treated with the TEAD-inhibitor, MGH-CP1, or control (DMSO). For each n, the average normalized luciferase levels (Firefly luciferase/Renilla Luciferase) from duplicate wells is plotted. (c) MKα3+/+ cells were treated with 10uM MGH-CP1 or DMSO (dotted line); gene expression of Csf1, Ctgf, Cyr61, and Bmp1 was determined by qPCR. (a–c) Data are mean +/− s.e.m.; n = 4; * p<0.05; ** p<0.01; *** p<0.001; ns, not significant.

Next, we used a series of mutant constructs to determine whether restoring YAP or TAZ activity in α3β1-deficient MKα3−/− cells is sufficient to rescue Csf1 expression. For these experiments, we used mutant forms of YAP (YAP-S127A) or TAZ (TAZ-S89A) that have increased transcriptional activity due to serine to alanine mutations of the LATS phosphorylation site that mediates cytoplasmic sequestration [32, 33]. Expression of YAP-S127A or TAZ-S89A in MKα3−/− cells caused a significant induction of TEAD-dependent transcriptional activity (Fig. 5a) and increased the expression of the canonical YAP/TAZ-TEAD target genes Ctgf (Fig. 5b) and Cyr61 (Fig. 5c). Expression of either YAP-S127A or TAZ-S89A was sufficient to increase the expression of Csf1 (Fig. 5d), while the expression of the α3β1-regulated Bmp1 gene was unchanged (Fig. 5e). To confirm the role of TEADs in the regulation of CSF-1 by α3β1, we included a mutant form of YAP-S127A that cannot bind TEADs (YAP-S127A/S94A), and therefore is unable to promote TEAD-dependent transcription [28, 34]. As expected, YAP S127A/S94A expression did not increase TEAD-dependent transcriptional activity in MKα3−/− cells (Fig. 5a), nor did it increase expression of the canonical YAP/TAZ-TEAD target genes Ctgf (Fig. 5b) and Cyr61 (Fig. 5c). In contrast to YAP-S127A (which can bind TEADs), YAP-S127A/S94A was unable to restore Csf1 expression to MKα3−/− cells (Fig. 5d), indicating that the ability of YAP to promote Csf1 expression in MK cells is dependent on TEAD binding. Collectively, these results show that keratinocyte α3β1 promotes YAP/TAZ-TEAD activity that is required for CSF-1 expression.

Figure 5.

YAP activation in MKα3−/− cells rescues the expression of Csf1 and other YAP/TAZ target genes in a TEAD-dependent manner. (a–e) MKα3−/− cells were transfected with empty expression vector (control), or vector that expresses YAP S127A, TAZ S89A, or YAP-S127A/S94A (see Materials and Methods for details). (a) TEAD-dependent transcriptional activity was measured using TEAD-reporter assays. For each n, the average normalized luciferase levels (Firefly luciferase/Renilla Luciferase) from duplicate wells is plotted. (b–e) qPCR was performed to assay gene expression of (b) Ctgf, (c) Cyr61, (d) Csf1, or (e) Bmp1. Data are mean +/− s.e.m.; n = 4; ** p<0.01; *** p<0.001; **** p<0.0001; ns, not significant.

Integrin α3β1-dependent Src activation promotes YAP activity and expression of Csf1

We showed previously that binding of integrin α3β1 to laminin-332 activates FAK-Src signaling in keratinocytes [35]. It has also been shown that α3β1 signals through FAK to promote YAP/TAZ activity that maintains tissue renewal in a mouse incisor model [22]. To test a requirement for Src in α3β1-dependent YAP activity and downstream Csf1 expression, we utilized two pharmacological inhibitors of Src, PP2 and dasatinib. Treatment of MKα3+/+ cells with either PP2 or dasatinib efficiently inhibited Src activity as well as the Src-dependent phosphorylation of FAK at tyrosine 925 (Fig. 6a,b), which we showed previously was α3β1-dependent [35, 36]. Notably, treatment of MKα3+/+ cells with either PP2 or dasatinib dramatically reduced the activity of the TEAD reporter (Fig. 6c,d). Src inhibition also reduced the expression of Csf1 and of the canonical YAP/TAZ-TEAD target genes, Ctgf and Cyr61 (Fig. 6e,f). Furthermore, expression of a constitutively-activate Src mutant, Src-Y527F, was sufficient to restore TEAD-dependent transcriptional activity in MKα3−/− cells (Fig. S5). Taken together, these results show that Src is both necessary and sufficient to promote YAP activation in MK cells, placing α3β1-dependent FAK-Src signaling [35] upstream of YAP activation.

Figure 6.

Inhibition of integrin α3β1-dependent Src activity in MK cells reduces the expression of Csf1 and other YAP/TAZ target genes. (a,b) Whole cell lysates were prepared from MKα3+/+ cells treated with (a) PP2 or (b) dasatinib, or equal volume of DMSO as control, then immunoblotted for active Src (pSrc Y416), total Src (tSrc), phospho-FAK (pFAK Y925) or GAPDH. Quantitation relative to DMSO and normalized to GAPDH is reported below blots as an average of three experiments. (c-f) MKα3+/+ cells were treated with PP2 (c,e) or dasatinib (d,f) versus DMSO as control. (c,d) TEAD-dependent transcriptional activity was measured using TEAD-reporter assays. For each n, the average normalized luciferase levels (Firefly luciferase/Renilla Luciferase) from duplicate wells is plotted. (e,f) Relative gene expression of Csf1, Ctgf, and Cyr61, was determined by qPCR. Data are mean +/− s.e.m.; n = 3; * p<0.05; *** p<0.001; **** p<0.0001; ns, not significant.

Integrin α3β1-deficient wounds have a reduced number of keratinocytes with YAP-positive nuclei

Finally, we investigated whether α3β1 modulates YAP within the wound epidermis in vivo. YAP was detected predominately in the nuclei of basal keratinocytes of wound epidermis (Fig. 7a), as determined by IF of wound cryosections. This finding is consistent with an earlier report that YAP localizes to the nucleus in basal keratinocytes of skin and is elevated upon wounding [12]. We detected fewer YAP-positive nuclei in the basal keratinocytes of α3eKO wounds compared to control wounds (Fig. 7a,b), revealing an important role for α3β1 in maintaining maximal YAP activity in the epidermis of healing wounds. Taken together, our findings indicate that epidermal keratinocytes can stimulate wound macrophages through integrin α3β1-FAK-Src-YAP/TAZ-TEAD-dependent induction of the Csf1 gene, promoting secretion of CSF-1 that mediates crosstalk to support macrophages in the wound stroma (Fig. 7c).

Figure 7.

Integrin α3β1-deficient wound epidermis has a reduced number of keratinocytes with YAP-positive nuclei. Cryosections of 3-day wounds were prepared as in Figure 1. (a) Representative cryosections stained with anti-YAP (red) and DAPI (blue); e, epidermis; wb, wound bed; dashed line, underlines wound epidermis; scale bar, 50 μm. (b) Graph shows percentage of wound keratinocytes with YAP-positive nuclei. Data are mean +/− s.e.m.; n ≥ 7 wounds per genotype; ** p<0.01. (c) Model: α3β1 on wound keratinocytes activates FAK-Src signaling that promotes YAP-TEAD-dependent transcription of Csf1. CSF-1 protein is secreted from the wound epidermis to support macrophages of the wound bed.

DISCUSSION

This study identifies a new role for the epidermal integrin α3β1 in regulating the cutaneous wound microenvironment by mediating crosstalk to macrophages. It is well known that CSF-1 is a critical regulator of macrophages in diverse tissue remodeling processes [2]. Our study shows that integrin α3β1 promotes the keratinocyte contribution of CSF-1 to the wound microenvironment, which supports the macrophage population in wounds in a paracrine manner. CSF-1 is known to regulate macrophage survival, proliferation, differentiation, and recruitment [2]. Since CSF-1 has many cellular sources [3], we cannot rule out important roles for other cell types in contributing CSF-1 to the wound. However, a significant amount of CSF-1 is in the epidermal compartment of the wound, and CSF-1 levels were dramatically reduced by keratinocyte-specific loss of α3β1 (Fig 1a), suggesting that keratinocytes are an important source of CSF-1 within the wound microenvironment. Furthermore, our study shows that epidermal α3β1 and epidermal-derived CSF-1 each promote macrophage growth (Figs. 2, 3), which could be influenced by effects on proliferation and/or survival. The impact of epidermal-derived CSF-1 in vivo may also extend to recruitment and differentiation of macrophages, which could be elucidated in future studies and may be relevant in the contexts of both wounds (this study) and skin tumors [6].

Additionally, our findings demonstrate that integrin α3β1 regulates YAP/TAZ-TEAD-dependent transcription during wound healing and that Csf1 is one of the genes regulated by α3β1 through YAP/TAZ-TEAD (Figs. 4, 5). YAP/TAZ-TEAD can regulate diverse sets of genes that play critical roles in numerous cellular processes, so it is likely that other genes with important roles in wound healing will also be regulated through the α3β1-YAP/TAZ-TEAD signaling axis. Previous studies have placed YAP and/or TAZ downstream of integrin signaling, as reviewed [37]. One of these studies showed that integrin signaling regulates YAP and TAZ to control skin homeostasis, positing that YAP and TAZ function as sensors of apical and basal signals [12]. Consistent with our current findings, the above study found that YAP and TAZ are elevated upon wound healing and nuclear localized within the basal layer of skin. In this same study, integrin β1 blocking experiments in HaCaT cells showed that ligand engagement is required for integrin-dependent activation of YAP [12]. This work concluded that integrin engagement by the basal ECM was required for YAP activation (nuclear localization), consistent with vastly more abundant nuclear YAP in basal keratinocytes compared to suprabasal keratinocytes, as we similarly observed in wound epidermis (Fig. 7). Although we did not directly test a requirement for ligand binding in the current study, we speculate that binding of integrin α3β1 to laminin-332 in the epidermal basement membrane promotes YAP nuclear localization and expression of its target genes in wound epidermis.

Other studies have also established critical roles for YAP/TAZ-TEAD in skin development, hair follicle morphogenesis, homeostasis, and wound healing [12, 38–40]. Importantly, in a collaborative study from the Klein group, integrin α3β1 was specifically linked to FAK-YAP-mTOR signaling that regulates stem cell expansion in a mouse incisor model [22]. This work revealed that α3β1-FAK signaling promotes YAP activation through PP1A-dependent dephosphorylation of YAP independent of LATS [22]. Similarly, another group reported that YAP activation in pancreatic endoderm cells, differentiated from embryonic stem cells, promotes transcription of pancreatic endoderm-associated genes through α3β1-FAK-CDC41-PP1A signaling rather than through the canonical Hippo signaling pathway [41]. Earlier studies from our group demonstrated α3β1-dependent activation of FAK in keratinocytes [35, 36], which together with our current findings supports an α3β1-FAK-YAP signaling axis in keratinocytes.

YAP/TAZ-TEAD activation is also known to be regulated by Src, another α3β1 signaling effector in keratinocytes [35]. Indeed, integrin-induced autophosphorylation of FAK is followed by Src-dependent phosphorylation of FAK on different tyrosine residues (including Y925), and this FAK-Src complex drives signaling downstream of integrin-mediated FAK activation [reviewed in [42]]. Furthermore, we have previously shown that FAK-Src signaling is α3β1-dependent in keratinocytes, as α3β1 expression is required for Src-dependent phosphorylation of FAK, including of Y925 [35, 36]. Studies from us [43] and others [reviewed in [37]] have revealed several distinct Src-dependent pathways that promote YAP/TAZ-TEAD activity. Many of these studies implicated integrins, including in the skin where increased integrin-Src signaling drives YAP/TAZ activation to promote hair follicle regeneration and wound repair [12, 44]. Consistently, we showed that Src activity is both necessary and sufficient to promote YAP/TAZ-TEAD activity in keratinocytes (Fig. 6, Fig. S5), indicating an α3β1-FAK-Src-YAP/TAZ-TEAD signaling axis that promotes expression of CSF-1. Future studies may determine whether this α3β1-FAK-Src-dependent regulation of YAP occurs in a Hippo-dependent or -independent manner.

YAP activity was shown to induce the expression and secretion of several immune factors, including CSF-1, in a model of pancreatic ductal adenocarcinoma [23]. Interestingly, several other proteins of the α3β1-dependent secretome [6] are encoded by known YAP/TAZ target genes, including SPARC [45], SERPINE1/PAI-1 [46], and LAMC2 [47]. Our current work adds CSF-1 to this list. Conversely, we observed that another α3β1-dependent gene, Bmp1 [31], which encodes the secreted protease BMP-1, was not regulated by YAP/TAZ, indicating that YAP/TAZ signaling controls only a subset of the α3β1-regulated secretome [6].

Importantly, a recent study demonstrated a YAP-CSF-1 signaling axis in fibroblasts whereby microenvironmental sensing by fibroblasts alters YAP activity, and YAP activation elevates fibroblast expression of CSF-1 that leads to an increase in the number of macrophages at steady state [27]. This indicates that mechanosensing by cells capable of contact inhibition (e.g., fibroblasts, keratinocytes) may coordinate tissue composition by regulating lineage-specific growth factors for hematopoietic cells, which are not confined by space [27]. Our findings extend such YAP-dependent, CSF-1-mediated paracrine stimulation of macrophages to wound keratinocytes. Moreover, we show that the laminin binding integrin α3β1, which is positionally poised to participate in mechanosensing, is required for this regulation.

Aberrant integrin function or expression is a likely contributor to wound pathologies, providing rationale for utilizing keratinocyte integrins as targets of therapy [5, 48]. The ability of α3β1 to stimulate crosstalk to other cellular compartments, thereby regulating wound angiogenesis [7], fibroblast differentiation [8], and the wound macrophage population (this study), implicates this integrin as an intriguing candidate for therapeutic targeting. Presumably, stimulating α3β1 function in chronic or hard-to-heal wounds may provide benefit while, on the other hand, inhibiting α3β1 function may be favorable in the case of fibrotic or over-exuberant wound healing (e.g., hypertrophic scars). Remarkably, in full-thickness skin wounds, siRNA-mediated knockdown of YAP/TAZ delayed the rate of wound closure [11] as did application of a CSF-1-neutralizing antibody [49], while topical application of exogenous CSF-1 accelerated wound healing [49]. These findings support the concept that increasing CSF-1 through stimulation of α3β1 may hasten wound healing and elicit therapeutic benefit in certain scenarios, perhaps in diabetic or elderly patients in whom wound healing responses are often delayed or insufficient.

METHODS

Murine studies

Information on K14CreERT:α3^flx/flx mice, and of wound healing experiments conducted in this model, were previously described [6, 36]. Briefly, Itga3^flx/flx mice were intercrossed with K14 promoter-driven, tamoxifen-inducible Cre mice to generate K14CreERT:α3^flx/flx mice on a mixed strain background. Either 1 mg of 4-hydroxytamoxifen (4OHT; Caymen Chemical Company) in 200 μL acetone (promoting deletion of floxed Itga3 in the epidermis; α3eKO) or acetone alone (control) was topically applied to the backs of shaved 12-week old K14CreERT:α3^flx/flx mice, at five and three days prior to wounding. On the day of wounding, mice were anaesthetized, backs were re-shaved, and with a sterile 4 mm biopsy punch, four full-thickness wounds were created on each mouse as described [7]. Three days later, CO₂ narcosis was used to euthanize mice. Surgically excised wounds were frozen in OCT (Electron Microscopy Sciences). Female and male mice were randomized and used in equal ratios. Approval for all murine experiments was granted by Albany Medical College’s Institutional Animal Care and Use Committee.

Histology

Preparation of frozen wound sections and IF was performed as before [6, 36]. Briefly, 10 μm wound sections were rehydrated (0.02% Tween-20/PBS) for 10 minutes, fixed (4% paraformaldehyde/PBS) for 5–30 minutes, permeabilized (0.4% TritonX-100/PBS) for 5–30 minutes, blocked (0.5% BSA, 10% goat serum, 0.1% Tween-20) for 30 minutes, then stained with anti-α3 integrin subunit (1:750) [50], anti-K14 (1:1000; Biolegend; Cat. #905301), anti-CSF1 (1:100; anti-M-CSF; Abcam; Cat. #ab233387), anti-F4/80 (1:100; Bio-Rad; Cat. #MCA497), or anti-YAP (1:100; Cell Signaling Technology; Cat. #14074). Secondary antibodies (1:250; Molecular Probes) were Alexa Fluor 594 goat anti-rat IgG (Cat. #A-11007), Alexa Fluor 594 goat anti-rabbit IgG (Cat. #A-11012), or Alexa Fluor 488 goat anti-rabbit IgG (Cat. #A-11008). Sections were co-stained with anti-K14 to mark the wound epidermis, where possible, and with DAPI. Sections were mounted and imaged as before [6]. NIS Elements AR 3.2 software was used for quantitative analyses, determining percent positive cell counts or mean fluorescence intensity, as indicated. To determine percent F4/80+ or YAP+ cells, the number of labeled cells was divided by the total number of DAPI-positive nuclei within a region of interest (ROI), then multiplied by 100. For F4/80 IF, the ROI was the wound bed beneath the re-epithelializing wound epidermis. For YAP IF, the ROI was the re-epithelializing wound epidermis.

In situ RNA detection

ISH was performed as before [6] using the RNAScope Fluorescent Multiplex V1 kit (Advanced Cell Diagnostics; Newark, CA) on fresh frozen tissue sections (see above). Briefly, sections were fixed in 4% paraformaldehyde for 1 hour, dehydrated, and digested with protease IV for 20 minutes, then incubated for 2 hours, 40°C with probes to detect Csf1 mRNA (Advanced Cell Diagnostics; Cat. #315621-C2 or C3), Krt14 mRNA (Advanced Cell Diagnostics; Cat. #422521; encodes K14) and co-stained with DAPI, then mounted and imaged as described above. Quantitative analysis of the Csf1 puncta in the Krt14-positive area marking the wound epidermis was performed using FIJI image J software and normalized to cell number (DAPI). Technical controls included in the kit were run routinely, including negative control probes that detect bacterial RNA, as well as positive control probes that detect RNA from murine housekeeping genes.

Cell Culture

Derivation of MK cells that express α3β1 (MKα3+/+) or lack α3β1 (MKα3−/−), and of MKα3−/− cells with restored α3 expression (MKα3res), was previously described [25, 26]. MK cell variants were maintained at 33°C, 8% CO₂ on collagen-coated culture dishes in supplemented low calcium EMEM as described [51–53] and have been Mycoplasma-tested using a PCR-based protocol [54]. RAW 264.7 murine macrophages (ATCC) were cultured in DMEM (Corning Life Science) containing 10% heat-inactivated FBS (GeminiBio), 100 μg/ml streptomycin and 100 units/ml penicillin. Macrophages were maintained at 37°C, 5% CO₂.

siRNA transfection

MKs were seeded in full medium 24 hours prior to transfection. Lipofectamine RNAiMAX (Life Technologies) reagent was used for transfection of 40 pmol of siRNA. Two Csf1-targeting siRNA’s (Life Technologies) were as follows: siRNA #1– 5’-GGACUAUCUCUUUAUGGAAtt-3’, siRNA #2–5’-GCAGGAGUAUUGCCAAGGAtt-3’. As a control, a non-targeting siRNA was utilized. After three days, cells were (1) isolated for either immunoblot or qPCR analyses, or (2) washed and serum-starved in preparation for CM collection, as described below.

RNA isolation and qPCR

RNA was isolated from MK cells with RNeasy Plus Mini Kit (Qiagen) or TRIzol (Invitrogen), and cDNA was made with iScript Reverse Transcription Supermix (Bio-Rad) or qScript cDNA SuperMix (QuantaBio). SsoAdvanced Universal SYBR green Supermix (Bio-Rad) or PerfeCTa SYBR green Supermix (QuantaBio) were used for qPCR. Primer sequences are as follows: Csf1: 5’-GCCTCCTGTTCTACAAGTGGAAG-3’, 3’-ACTGGCAGTTCCACCTGTCTGT-5’; Ctgf: 5’-CTCCACCCGAGTTACCAATG-3’, 3’- TGGCGATTTTAGGTGTCCG-5’; Cyr61: 5’-ACAAAGCGTCCACCATACAT-3’, 3’-GAGGTA AGAGGCTTGTGGTTT-5’; Bmp1: 5’-GACAACTCGGTACAGAGGAAAG-3’, 3’-CGAACTGGGCATGGGAATAA-5’; Gapdh: 5’-CTTTGTCAAGCTCATTTCCTGG-3’, 3’-TCTTGCTCAGTGTCCTTGC-5’. Gapdh was used for normalization within same MK line, while normalization for comparison of Csf1 across MK lines was achieved using the geometric mean of three reference genes (Tbp, Ppia, Nono; Bio-Rad) as done previously [55]. Relative mRNA levels were calculated as before [6].

Immunoblot

Non-reducing buffer (Cell Signaling Technology) was used for the preparation of whole-cell lysates. For preparation of CM, MKs were grown to ~80% confluence, then washed with PBS and cultured in serum-free medium for 24 hours. CM was centrifuged at 2×10³ g for removal of cell debris, then CM was concentrated using 30,000 MWCO centrifugal filters (Millipore). The BCA Protein Assay kit (Pierce) was used to elucidate protein concentrations. Reducing 10% SDS-PAGE resolved samples of equal protein concentration. Immunoblot was performed using anti-α3 integrin subunit (1:1,000) [50], anti-ERK (1:1,000; Santa Cruz Biotechnology; Cat. #sc-154, anti-CSF1 (anti-M-CSF; 1:1,000; Abcam; Cat. #ab233387), anti-pSrc Y416 (1:1,000; Cell Signaling; Cat. #2101), anti-Src (1:1,000; Cell Signaling; Cat. #2123), anti-pFAK 925 (1:1,000; Cell Signaling; Cat. #3283), or anti-GAPDH (1:5,000; Cell Signaling; Cat. #2118), followed by HRP‒conjugated goat anti-rabbit IgG (1:2,000; Cell Signaling Technology; Cat. #7074). Chemiluminescence and imaging was done as before [53].

Macrophage growth assays

1×10³ macrophages per well were seeded in triplicate on a 96-well plate (Day 0) and grown in a mixture of 95% MK CM (see ‘Preparation of CM’ description within ‘Immunoblot’ subsection above) and 5% heat-inactivated FBS (GeminiBio). Acquisition of phase images occurred on specified days using a SpectraMax plate reader (Molecular Devices). Softmax Pro Software (Molecular Devices) quantified cell count and covered area.

TEAD Transcriptional Reporter Assay

The TEAD transcriptional reporter assays utilize a YAP/TAZ-TEAD reporter construct (pGL3-5xMCAT(SV)-49) with a minimal SV40 promoter that contains 5 repeats of a TEAD-binding motif driving expression of the Firefly Luciferase gene [43]. A control Renilla luciferase construct (PRL-TK (Promega, Cat. #E2231)) was co-transfected and used for normalization. Cells plated on 12-well dishes in duplicate were co-transfected with 400 ng of a 20:1 mixture of pGL3-5xMCAT(SV)-49 and PRL-TK using Lipofectamine 3000 (4 μL/well) and 2 μL/well of the P3000 reagent (Invitrogen, Cat. #L3000001) in 100 μL of Opti-MEM. A Dual-Luciferase Reporter Assay System (Promega, Cat. #E1910) was used to quantify Firefly and Renilla luciferase activity as described previously [43]. For quantification, normalized luciferase levels (Firefly Luciferase value/Renilla Luciferase value) were calculated and averaged for duplicate wells, then plotted for each experiment. For some experiments, cells were treated with inhibitors or co-transfected with the indicated YAP,TAZ, or Src vectors prior to measuring luciferase activity as described below.

Inhibitor Treatments

Twenty-four hours following luciferase transfection (above), MKs were treated with either MGH-CP1 (Selleck Chemicals) [30] at the indicated concentrations, 7.5μM PP2 (Calbiochem), 0.75μM Dasatinib (Medchem Express), or DMSO as vehicle control. Working stocks were 10mM in DMSO. The volume of DMSO used in MGH-CP1 experiments was equivalent to that needed for the highest MGH-CP1 dose. Twenty-four hours post-treatment, luciferase activity was read, or qPCR or immunoblotting was performed, as described above.

Transient Transfection of YAP, TAZ and Src vectors

For qPCR analysis, MK cells were plated on 6 wells and transfected with 1000 ng DNA of a control vector, MSCV-IRES-Hygro [28], or MSCV-flag-hYAP^S127A-IRES-Hygro [28], MSCV-flag-hYAP^S127A/S94A-IRES-Hygro [28], or MSCV-2HA-TAZ^S89A-IRES-Hygro [43] using Lipofectamine3000 (ThermoFisher). RNA was extracted 48 hours later for qPCR analysis (see above). For transcriptional reporter assays, MKs were seeded on 12-wells in duplicate and co-transfected with 500 ng of the vectors above, or with either MSCV-IRES-Puro [43] or MSCV-c-Src^Y527F-IRES-Puro [43], and 500 ng of a 20:1 mass ratio of the luciferase reporter constructs (see above).

Statistical Analyses

For data analysis from two experimental groups, a 2-tailed student’s t-test was used, while single sample t-tests were performed for analysis of an experimental group relative to control set at y=1. For analyses involving more than two experimental groups, one-way ANOVA with Dunnett multiple comparisons post-test was used. p<0.05 is considered significant; * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001; ns, not significant.

Abbreviations:

CSF1

colony stimulating factor 1

YAP

Yes-associated protein

TEAD

Transcriptional enhanced associate domain

ECM

extracellular matrix

TAZ

transcriptional co-activator with a PDZ-binding domain

FAK

focal adhesion kinase

α3eKO

α3 epidermal knockout

IF

immunofluorescence

MK

mouse keratinocyte

CM

conditioned medium

K14

keratin-14

4OHT

(Z)-4-hydroxytamoxifen

ISH

in situ hybridization