1. Introduction
Skin wound healing starts immediately after injury and evolves in three phases. The first one is an inflammatory phase during which platelets tend to aggregate, while inflammatory cells are recruited to the wound site. The second proliferative phase is characterized by the formation of granulation tissue and re-epithelialization due to the migration and proliferation of keratinocytes, fibroblasts, and ECs, and by ECM deposition. The last one is the so-called remodeling phase during which the regenerative process comes to an end and the wound becomes avascular and acellular, thereby allowing the reorganization of the connective tissue to promote scar formation [
1,
2].
ICOS (CD278) is a T cell co-stimulatory receptor, member of the CD28 family [
3], mainly expressed on activated T-cells. ICOS binds ICOSL (CD275, also called B7h, GL50, B7H2), a member of the B7 family. ICOS triggering in T cells promotes not only the activation of effector T cells in peripheral tissues but also the development of regulatory T cells [
4]. ICOSL is expressed on multiple cell types, including antigen presenting cells (APCs), activated ECs, epithelial cells, fibroblasts, and keratinocytes [
5,
6]. ICOSL triggering mediated by ICOS drives a “reverse signal” that inhibits the migration of endothelial, dendritic, and tumor cells, modulates cytokine secretion while promoting antigen cross-presentation in dendritic cells, and inhibits osteoclast differentiation and functions [
7,
8,
9,
10,
11].
We have recently shown that ICOSL also binds osteopontin (OPN) at a different site from that used to bind ICOS [
12], which suggests that the ICOSL/OPN axis may play a role in wound healing besides tumorigenesis. This hypothesis is also supported by the observation that OPN can act as both an ECM component and a soluble cytokine involved in inflammation and angiogenesis [
13,
14]. Indeed, ICOSL triggering by OPN induces tumor cell migration and promotes tumor angiogenesis, both of which are counteracted by ICOS-mediated activation of ICOSL [
12].
The formal demonstration of a functional role of the ICOS/ICOSL pathway in wound healing comes from the observation that ICOS
−/−, ICOSL
−/−, and ICOS/ICOSL
−/− mice show delayed wound healing [
15] likely due to decreased production of IL-6 [
16]. In good agreement with a role of the ICOS/ICOSL dyad in normal tissue repair, we have recently shown that CCl
4-induced liver damage, which is dependent on massive recruitment of blood-derived monocytes/macrophages, is dramatically worsened in both ICOS
−/− and ICOSL
−/− mice [
17]. Interestingly, we were able to rescue this impairment by treating mice with ICOS-Fc, a recombinant soluble protein composed of the ICOS extracellular portion fused to the IgG1 Fc portion, which has been previously shown to trigger ICOSL, thereby inhibiting the development of experimental tumor metastases in vitro and tumor angiogenesis in vivo [
9,
11,
17,
18].
As the aforementioned findings support a functional role of ICOS/ICOSL in tissue repair, in the present study, we sought to determine the effect of ICOS-Fc in both in vitro and in vivo models of skin wound healing. Our in vivo results show that ICOS-Fc improves would healing likely by increasing angiogenesis and recruitment of reparative macrophages.
3. Discussion
The present study shows that ICOS and ICOSL cooperate in skin wound healing and that triggering of ICOSL by instillation of ICOS-Fc into the wound bed favors tissue repair in vivo. These results extend those obtained by Maeda et al. [
15] showing that wound healing is delayed in ICOS
−/−, ICOSL
−/−, or ICOS/ICOSL
−/− mice, possibly due to defective production of IL-4, IL-10, and, especially, IL-6 at the wound site. Since this defective repair was overcome by adoptive transfer of wild-type T cells (expressing ICOS) in ICOS
−/− but not ICOSL
−/− mice, the authors concluded that the healing defect in KO mice could be ascribed to the impaired development of T helper type 2 cells due to the lack of ICOS-mediated co-stimulation of T cells.
Even though our findings confirm that wound healing is defective in mice lacking ICOS or ICOSL, the observation that ICOSL stimulation by ICOS-Fc is sufficient to accelerate the early phases of the healing process underscores the importance of ICOSL in ICOS/ICOSL-mediated tissue repair. Indeed, enhanced wound healing in response to ICOS-Fc treatment is readily apparent in both wild-type and ICOS−/− mice, but not in mice lacking ICOSL, which indicates that this effect is not due to the inhibition of ICOS activity in T cells, but it is instead caused by ICOSL-mediated “reverse signaling” in other cell types. The fact that ICOS-Fc treatment is effective also in immunodeficient NSG mice confirms that T cells are not involved in ICOS-Fc-induced wound healing. Moreover, the lack of effect in ICOSL−/− mice rules out possible confounding effects due to the potential interaction of ICOS-Fc with Fcγ receptors.
A key effect of ICOS-Fc is represented by increased angiogenesis and recruitment of fibroblasts at day 3 and 4, as judged by histologic analysis, both of which are preceded by upregulation of CD31 and VEGF-α—two markers of angiogenesis—and αSMA—a marker of reparative myofibroblasts—mRNA expression at day 1 and 2, respectively. Enhanced angiogenesis in response to ICOSL triggering was unexpected since previous works had shown that in vivo treatment with ICOS-Fc curbed neoplastic angiogenesis in several mouse tumor types, and in vitro experiments showed that ICOS-Fc had no effect on angiogenesis induced by VEGF whereas it inhibited that induced by OPN [
8,
18].
Another interesting observation from our histological analysis is that ICOS-Fc treatment can also modulate the infiltration of inflammatory cells by decreasing neutrophils and increasing T cells and macrophages. The decrease in neutrophils is in line with previous data showing that ICOS-Fc inhibits neutrophil adhesion to ECs, which may affect their recruitment into inflamed tissues [
8]. The increase in T cells might be ascribable to the enhanced vascularization of the wound or to the functional antagonism between ICOS-Fc and ICOS expressed on T cells given that, at least in tumors, ICOS-Fc treatment is known to increase effector T cells and decrease regulatory T cells [
18,
20]. The increase in macrophages is quite intriguing as it is accompanied by a five-fold increase in TREM2/TREM1 expression ratio, which suggests that ICOS-Fc favors recruitment of TREM2
+ M2-like reparative macrophages, as compared to TREM1
+ M1-like inflammatory macrophages. This possibility is also supported by our cell migration experiments in vitro, showing that ICOS-Fc enhances the migration of M2 macrophages, whereas it inhibits that of M1 macrophages. The increased migration of M2 macrophages was unexpected, since ICOS-Fc had always inhibited the migration response of all cell types analyzed until then [
8,
9,
10,
12,
20,
21]. The different response of mouse M1 and M2 macrophages may be due to differences in their migration and adhesive properties likely caused by higher expression levels of β2 integrins in M1 vs. M2 cells [
22]. Intriguingly, ICOS-Fc treatment also led to increased migration of keratinocytes, as judged by our scratch assay analysis, which could be the result of changes in size, shape, adhesiveness, and organization of keratin intermediate filaments of these cells as shown previously [
23].
Overall, the effects of ICOS-Fc on wound healing are in line with our previous work showing that CCl
4-treated ICOS
−/− or ICOSL
−/− mice develop a more severely acute inflammatory liver damage, along with a reduction of reparative macrophages, compared to their wild-type counterparts. Moreover, treatment with ICOS-Fc protected ICOS-deficient mice from this increased damage, simultaneously restoring the number of reparative macrophages, whereas it had no effects in ICOSL
−/− mice [
17]. These findings are also in line with the aforementioned study by Maeda et al. [
15], showing that mice lacking ICOS and/or ICOSL display decreased angiogenesis and a reduction of T cells and macrophages at the wound site. Intriguingly, the authors observed decreased IL-6 in the wounds of these mice, and local application of exogenous IL-6 in the initial phase of healing (day 1) led to a substantial improvement of tissue repair. A potential role of ICOSL-induced IL-6 production in wound healing is also supported by our observation that treatment with ICOS-Fc of wounded wild-type mice increases the expression of IL-6 at day 2 [
15].
4. Materials and Methods
4.1. Scratch Assay
HaCat cells (human keratinocytes) were purchased from ATCC (Manassas, VN, USA) and grown in DMEM (Life Technologies, Carlsbad, CA, USA) medium plus 10% fetal bovine serum (FBS; Life Technologies). HaCat cells were plated in six-well plates at a concentration of 106 cells/well and grown to confluence. To prevent cell proliferation, cells were incubated for 12 h in FBS-free medium. Cell monolayers were wounded by scratching with a sterile plastic pipette tip along the diameter of the well. Cells were then incubated in culture medium in the absence or presence of 2 µg/mL human ICOS-Fc or F119SICOS-Fc, an ICOS-Fc mutant unable to bind ICOSL. To monitor cell migration in the wound, five fields of each wound were analyzed and photographed immediately after scratching (0 h) and 24 h later. The wound closure was calculated with the following formula: (1 − (scratch width of the treated group/scratch width of the control group)) × 100%.
ICOS and ICOSL expression was assessed by immunofluorescence and flow cytometry (Attune NxT, Thermo-Fisher, Waltham, MA, USA) using PE-conjugated mAb to ICOS or ICOSL (R&D System, Minneapolis, MN, USA). The mean fluorescence intensity ratio (MFI-R) was calculated according to the following formula: MFI of the stained sample histogram (arbitrary units)/MFI of the control histogram (arbitrary units).
4.2. Mice
C57BL6/J (WT), NOD-SCID-IL2R γ-null mice (NSG) and knockout B6.129P2-Icostm1Mak/J (ICOS−/−) and B6.129P2-Icosltm1Mak/J (ICOSL−/−) mice (The Jackson Laboratory, Bar Harbor, ME, USA) were bred under pathogen-free conditions in the animal facility at Università del Piemonte Orientale, Department of Health Sciences (Authorization No. 217/2020-PR) and treated in accordance with the Ethical Committee and European guidelines.
4.3. In Vivo Wounds
The day before wound induction (day-1), WT, NSG, ICOS−/−, and ICOSL−/− mice were anesthetized with 2% isoflurane and their back was shaved. At day 0, mice were anesthetized as above, and wounds were made on their back using a 4 mm puncher (Kai Medical, Solingen, Germany). The wound area was photographed and measured using the following formula: (a/2) × (b/2) × 3.14, where “a” and “b” are the two perpendicular diameters. In the following days, wound closure was calculated using the following formula: (wound areaT0-wound areaTX)/wound areaT0 × 100. Mice were treated daily with 10 µg/mouse ICOS-Fc in PBS instilled directly into the wound site; controls were treated with an equal volume of PBS. Mice were monitored daily for 12 days, at which point in time the wound was closed. In some experiments, mice were sacrificed at day 1, 2, 3, and 4 to harvest and analyze the healing tissue. Each experiment involved 4–7 mice for each condition tested; each condition was tested in 2–3 independent experiments. Sample size was calculated using G*Power (RRID:SCR_013726) software (Power: 80%; Significance: 95%).
4.4. Real-Time PCR Analysis
Total RNA was isolated from skin samples collected at day 1, 2, and 3 post-injury, or from in vitro-differentiated macrophages using TRIzol reagent (Sigma-Aldrich, St. Louis, MO, USA). RNA (1 µg) was retro-transcribed using QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany). Expression of the IL-6, TNF-α, TGF-β, IL-33, IL-10, IL-4, IFN-γ, OPN, TREM1, TREM2, VEGF-α, α-SMA, ICOS, NOS2, ARG1, and ICOSL mRNA were evaluated by real-time PCR (Assay-on Demand; Applied Biosystems, Foster City, CA, USA). The β-actin gene was used to normalize the cDNA amounts. Real-time PCR was performed using the CFX96 System (Bio-Rad Laboratories, Hercules, CA, USA) in duplicate for each sample in a 10 µL final volume containing 1 µL of diluted cDNA, 5 µL of TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA), and 0.5 µL of Assay-on-Demand mix. The results were analyzed with a ΔΔ threshold cycle method.
4.5. Histological Analysis
Skin samples were collected at day 3 and 4 post-injury and processed for paraffin embedding. Samples were cut at 4-µm thickness and stained with hematoxylin and eosin (H&E) (Sigma-Aldrich) for tissue morphology and fibroblast evaluation, or with picrosirius red (Abcam, Cambridge, UK) to evaluate the extent of fibrosis.
Immunohistochemical staining of CD31, MPO, CD3, and F4/80 was performed to detect neo vessel formation and infiltration of immune cells (i.e., neutrophils, T cells, and macrophages). Samples were treated with citrate buffer (Vector Laboratories, Burlingame, CA, USA) for antigen retrieval, and endogenous peroxidases were blocked with 3% H2O2 (Sigma-Aldrich). To avoid secondary antibody unspecific binding, samples were pre-incubated with 5% normal goat serum (NGS) (Sigma-Aldrich) for 1 h at room temperature (RT). Samples were stained with rabbit antibodies against CD31 (Abcam, 1:50), MPO (Invitrogen, 1:100), CD3 (Invitrogen, 1:150), or F4/80 (Invitrogen, 1:100) overnight at 4 °C and, then, with a goat anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibody (Sigma-Aldrich), followed by 3,3′-diaminobenzidine (DAB) (Agilent Dako, Santa Clara, CA, USA). Successively, samples were counterstained with hematoxylin (Sigma-Aldrich), dehydrated, and mounted on cover slips. Slides were acquired using Pannoramic MIDI (3D Histech, Budapest, Hungary) at 200× magnification. The positive areas for CD31, fibroblasts, and collagen were calculated using the following formula: (positive area/total area) × 100%. MPO, CD3, and F4/80 positive cells were expressed as cell number/field counted in 15 fields for each sample.
4.6. Macrophage Migration Assay
Spleen cells were separated by density gradient centrifugation using the Ficoll-Hypaque reagent (Lympholyte-M, Cedarlane Laboratories, Burlington, ON, Canada) and incubated in tissue culture dishes for 2 h with DMEM supplemented with 10% FBS. Subsequently, supernatants and non-adherent cells were discarded, and the adherent cells were rinsed three times and cultured in DMEM (Life Technologies) medium supplemented with 10% FBS, 1% glutamine, and 1% penicillin/streptomycin plus 20 ng/mL M-CSF (Immunotools, Friesoythe, Germany) for 14 days (normal DMEM medium). At day 14, adherent cells were cultured for additional 48 h with interferon-γ (IFN-γ; 100 U/mL Immunotools) to obtain M1 macrophages, and with interleukin-4 (IL-4; 20 ng/mL Immunotools) to obtain M2 macrophages; each culture condition was performed in the presence or absence of LPS (LPS; 100 ng/mL Sigma).
Macrophage migration was assessed by the Boyden chamber migration assay (BD Biosciences, San Jose, CA, USA). Cells were plated (10
4 cell/well) onto the apical side of 50 µg/mL Matrigel-coated filters in serum-free medium in the presence or absence of msICOS-huFc (2 µg/mL), composed by the extracellular portion of murine ICOS fused to the Fc of human IgG1, or human
F119SICOS-Fc (2 µg/mL). Mouse CCL2 (30 nM, Immunotools) or OPN (10 µg/mL) were used as chemoattractants in the bottom chamber. After 6 h, the cells on the apical side were wiped off with Q-tips. Cells on the bottom of the filter were stained with crystal violet and all counted (quadruplicate filter) with an inverted microscope. Data are shown as number of migrating cells [
12].
4.7. Statistical Analyses
Statistical analyses were performed using Mann–Whitney U test, Wilcoxon test, Dunnett’s test, or Student’s t-test using GraphPad Instat Software (GraphPad Software, San Diego, CA, USA), as indicated. Data are expressed as mean and standard error of the mean (SEM) and statistical significance was set at p < 0.05.