Dominant effect of gap junction communication in wound‐induced calcium‐wave, NFAT activation and wound closure in keratinocytes

Wounding induces a calcium wave and disrupts the calcium gradient across the epidermis but mechanisms mediating calcium and downstream signalling, and longer‐term wound healing responses are incompletely understood. As expected, live‐cell confocal imaging of Fluo‐4‐loaded normal human keratinocytes showed an immediate increase in [Ca2+]i at the wound edge that spread as a calcium wave (8.3 µm/s) away from the wound edge with gradually diminishing rate of rise and amplitude. The amplitude and area under the curve of [Ca2+]i flux was increased in high (1.2 mM) [Ca2+]o media. 18α‐glycyrrhetinic acid (18αGA), a gap‐junction inhibitor or hexokinase, an ATP scavenger, blocked the wound‐induced calcium wave, dependent in part on [Ca2+]o. Wounding in a high [Ca2+]o increased nuclear factor of activated T‐cells (NFAT) but not NFkB activation, assessed by dual‐luciferase receptor assays compared to unwounded cells. Treatment with 18αGA or the store‐operated channel blocker GSK‐7975A inhibited wound‐induced NFAT activation, whereas treatment with hexokinase did not. Real‐time cell migration analysis, measuring wound closure rates over 24 h, revealed that 18αGA essentially blocked wound closure whereas hexokinase and GSK‐7975A showed relatively minimal effects. Together these data indicate that while both gap‐junction communication and ATP release from damaged cells are important in regulating the wound‐induced calcium wave, long‐term transcriptional and functional responses are dominantly regulated by gap‐junction communication.


| INTRODUCTION
Tight spatial and temporal regulation of intracellular calcium ([Ca 2+ ] i ) signalling allows this ion to control a wide variety of physiological responses including cell growth and differentiation. A steep vertical calcium gradient exists within the epidermis with calcium concentration being low in the proliferative basal layer, progressively increasing through the spinous layer reaching a maximum in the granular layer; levels then fall at the outermost stratum corneum (Elias et al., 2002). Experiments in a wide variety of organisms and cell types have shown that wounding leads to an immediate rise in cytosolic calcium at the wound edge which is followed by a calcium wave radiating away from the site of injury and passing from cell to cell (Tran et al., 1999;Tu et al., 2019). In contrast to excitable cells, wound-induced calcium waves in keratinocytes decay and gradually diminish over relatively short distances (Kobayashi et al., 2014).
The wound-induced rapid elevation in [Ca 2+ ] i and subsequent calcium wave have been shown to be crucial for the downstream calciumdependent signalling pathways that promote wound closure and organism survival (Tu et al., 2019;Xu & Chisholm, 2011). During wounding of the skin, the epidermal barrier is breached, the dermis is damaged and the wounded tissue is exposed to serum which contains a high concentration of calcium (Bandyopadhyay et al., 2006). However, measurement of wound fluid immediately after cutaneous excisional wounding of pigs showed a reduction of calcium concentration compared to plasma, possibly due to Ca 2+ sequestration (Grzesiak & Pierschbacher, 1995). A progressive increase in calcium concentration was noted at the incisional skin wound site up to Day 5 in rats (Lansdown et al., 1999). Together, these studies underscore the importance of extracellular calcium to cutaneous wound healing.
Cytosolic increases of calcium can arise essentially from extracellular influx through plasma membrane ion channels, entry of calcium from adjacent cells through gap junctions or [Ca 2+ ] i release from intracellular stores. The latter occurs primarily from the endoplasmic reticulum (ER), although the Golgi apparatus (Xue et al., 1994) and the mitochondria (De Stefani et al., 2011) have also been reported to be major calcium stores.
Calcium release from the ER is primarily triggered by local increases of the soluble mediator inositol 1,4,5-trisphosphate (IP 3 ) binding to its receptor on the ER and activating Ca 2+ release through the IP 3 receptor's intrinsic Ca 2+ channel. Although both calcium and IP 3 may pass through gap junctions (Boitano et al., 1992;Saez et al., 1989), the diffusibility of calcium is relatively limited, due in part to its propensity to interact with a variety of proteins. IP 3 transfer with subsequent triggering of Ca 2+ release from the ER in adjacent cells is the more likely scenario (Leybaert & Sanderson, 2012), including in keratinocytes (Kobayashi et al., 2014).
Depletion of calcium from the ER is sensed by stromal interaction molecule 1 (STIM1) resulting in the translocation of STIM1 to the plasma membrane, activation of calcium release-activate calcium channel protein 1 (Orai1) and subsequent entry of external calcium through Orai1 channels (Stathopulos et al., 2013). This process, termed store-operated calcium entry (SOCE), inextricably links external calcium concentrations to intracellular calcium fluxes and regulation of cellular processes. Consequently, intra and extracellular calcium sensing proteins play key roles tightly regulating in these signalling pathways by activation of positive and negative feedback loops (Bird & Putney, 2018;Kobayashi et al., 2014).
Although wound healing is an evolutionary conserved process, the mechanisms contributing to the wound-induced calcium wave vary among organisms and cell types. In keratinocytes, gap junction signalling and release, followed by diffusion of ATP to adjacent cells, where it binds to G protein coupled receptors activating phospholipase C and generating IP 3 , have both been shown to play a role in mediating the woundinduced [Ca 2+ ] i wave (Karvonen et al., 2000;Kobayashi et al., 2014).
Recent studies have also identified a role for the calcium-sensing receptor in mediating the wound-induced calcium flux and the migratory response of keratinocytes (Tu et al., 2019). However, the interplay and relative contribution of ATP signalling, gap junctions and extracellular calcium influx to the generation and propagation of the calcium wave in human keratinocytes is incompletely understood. Moreover, the mechanisms linking calcium signalling occurring in the seconds and minutes after injury to activation of physiological processes required for wound healing remain to be fully defined. IP 3 -mediated SOCE signalling cascade is fundamental to nuclear factor of activated T-cells (NFAT) 1-4 activation and all four calcium-dependent family members are expressed in keratinocytes. In keratinocytes, calcineurin accompanies NFAT1 to the nucleus where it retains NFAT in a dephosphorylated state and therefore NFAT remains nuclear (Al-Daraji et al., 2002). We and other investigators have previously reported that inhibition of calcineurin/NFAT signalling inhibits keratinocyte migration and scratch wound closure (Brun et al., 2014;Jans et al., 2013). In this study, we aimed to analyse specific parameters of the wound-induced calcium wave in human keratinocytes and to delineate pharmacologically the contribution of extracellular ATP signalling, gap-junctional communication, and extracellular calcium influx to further understand the mechanisms of wave transmission. Finally, we designed experiments to establish the relative effects of gap-junctional communication, extracellular ATP and SOCE on both wound-induced NFAT transcriptional activation and keratinocyte migration to close the wound.

| Reagents
All chemicals were purchased from Sigma-Aldrich unless otherwise specified.

| Cell culture and treatments
Normal human epidermal keratinocytes were isolated from normally discarded healthy adult skin samples from patients undergoing a surgical procedure (Jans et al., 2013;Todd & Reynolds, 1998) and expanded in human keratinocyte growth supplement medium (Invitrogen). The study was approved by the Newcastle and North Tyneside local ethics committee, and written informed patient consent was obtained. Keratinocytes were used between Passage 1 and 3.

| [Ca 2+ ] i imaging
Keratinocytes seeded in Willco glass-bottomed dishes (Intracel) were loaded with 3 µM Fluo4-AM, 200 mM sulphinpyrazone and 0.1% pluronic acid F-127 and subjected to Ca 2+ i imaging using Fluo-4-AM (Invitrogen) as described (Ross et al., 2007). Post-de-esterification, media was replaced with fresh media with a calcium concentration of either 0.06 or 1.2 mM for 5 min before wounding. Single scratch wounds were made approximately 30 s after imaging commenced using a 200 µl pipette tip.
After 20 min, 3 µM thapsigargin (Tg) was added as a positive control.
Imaging was carried out using a Nikon spinning disc TIRF system (Nikon UK Limited) with 20×1.2 lens and excitation wavelength 488 nm. Live images were captured at 3.7 frames per second (fps). Quantification was performed using Volocity (Improvision).

| ATP release postwounding
ATP release from keratinocytes postwounding was detected using the luminescence-based Cell-titre glo assay (Promega UK).
Transfection efficiency and cell viability were controlled by cotransfecting a Renilla luciferase control vector (pRLTK; Promega). Keratinocytes were seeded in 12-well plates and transfected using 0.5 μg of firefly reporter DNA plus 0.05 μg of Renilla luciferase DNA using Fugene 6 (Roche Applied Sciences) at a 6:1 Fugene:DNA ratio. After 24 h, the cells were stimulated as indicated, lysed, and assayed for dual luciferase activity. Luciferase assays were performed using the dual luciferase assay system (Promega). Reporter firefly luciferase values were normalized to the Renilla values.

| Wound closure rates
A 200 µl pipette tip was used to create a "cross hatch" wound, on confluent keratinocytes seeded in 24-well plates. Cells were imaged using the Nikon Biostation CT (Nikon UK Limited). ImageJ was used to manually measure the initial wound area and then the subsequent wound area at specified time points.

| Statistical analysis
Statistical analysis was conducted using Prism (GraphPad 5 or 6 software). Data presented represent mean ± standard error of the mean (SEM). N depicts the number of independent donors and n indicates the total number of independent experiments (culture dishes) or number of cells analysed (as specified).   Figure 1d shows that that the rate of rise of the [Ca 2+ ] i flux decreased exponentially the further back from the wound the cells were located. Cells at the wound edge had a rate of rise of 0.67 ± 0.23 (ΔF t /F 0 /s) and those six cells back from the wound had a rate of rise of 0.10 ± 0.03 (ΔF t /F 0 /s) (p < .05). Overall, our analysis indicates that after wounding, the calcium wave reduces in amplitude, flux and rate of rise as it travels across cells away from the wound edge. The mechanisms for this dampening remain to be formally determined experimentally, but the pattern of response is most consistent with [Ca 2+ ] i release gradually diminishing as the wave moves away from the wound.
This model is compatible with both gap-junction communication and paracrine ATP signalling (Fujii et al., 2017;Kobayashi et al., 2014) and the diminishing ability of IP 3 and/or extracellular ATP to regenerate IP 3 (Sneyd et al., 1995). (i) Wound size was measured at each time point using Image J and percent of wound closure calculated (mean ± SEM; 18 experiments from six independent donors; n = 18, N = 6). Two-way ANOVA p < .0001; **p < .01 Bonferroni post hoc test between the two external calcium concentrations. ANOVA, analysis of variance; AUC, area under the curve; CI, confidence interval; KGM, keratinocyte growth medium in cells located two to six rows from the wound, in 1.2 mM [Ca 2+ ] o compared to 0.06 mM [Ca 2+ ] o (Figure S1b-f). Additionally, as shown in Figure 1g, the time taken to reach maximum F t /F 0 was reduced when wounding was conducted in 1.2 mM [Ca 2+ ] o . Intriguingly, rate of rise of [Ca 2+ ] i was the only parameter examined that was not affected by external calcium at cell locations either close to and further back from the wound edge. If rate of [Ca 2+ ] i rise is taken as an approximation to the rate at which calcium is released from intracellular stores, such as the ER, this supports a model in which IP 3 transfer across cells induces calcium release from the ER which leads to increased calcium entry in 1.2 mM (Boitano et al., 1992). 3.3 | GSK-7975A inhibits SOCE in keratinocytes GSK-7975A, a pyrazole derivative, is an effective and highly selective SOCE inhibitor that interferes with ion permeation through the Orai1 selectivity pore (Chaudhari et al., 2020;Derler et al., 2013). However, its efficacy in primary keratinocytes has to date, not been reported. To address this, primary human keratinocytes in 1.2 mM [Ca 2+ ] o KGM were treated with 3 µM thapsigargin (Tg) to deplete ER calcium stores and initiate SOCE ( Figure S2). Real time calcium flux imaging showed that 10 µM GSK-7975A or 10 µM diethylstilbestrol (DES) (Jans et al., 2013) blocked SOCE in primary keratinocytes ( Figure S2). However, sulforhodamine B (SRB) cell viability assays demonstrated that DES was toxic to primary human keratinocytes at 24 h whereas GSK-7975A did not appear to reduce cell viability. As we were interested in correlating early calcium flux changes with later wound healing responses, we subsequently used GSK-7975A to inhibit SOCE and study its role in wound-induced calcium wave.  (Figure 2c). These data suggest that blocking SOCE channels at the wound edge may release a reciprocal inhibitory mechanism enhancing calcium entry through store independent calcium entry (SICE) channels (Zhang et al., 2018), for example, as previously described (Mignen et al., 2001(Mignen et al., , 2003.

| Effect of increased extracellular calcium concentration on wound-induced calcium wave characteristics
Maximum

| Role of gap junction communication in mediating the wound-induced calcium wave
In mammalian cells, intercellular calcium wave propagation is thought to occur predominantly through a combination of gap-junction communication and paracrine ATP signalling (Fujii et al., 2017;Kobayashi et al., 2014). Passage of IP 3 through gap-junctions, binding to IP3R on the ER and further initiation of SOCE has been shown in a variety of cell types. ATP-mediated calcium waves are slower compared to direct gap junctional signalling but can spread between cells either side of a cell-free zone (Frame & de Feijter, 1997) and even across different cell types (Koizumi et al., 2004). ATP released via connexin and pannexins (Bao et al., 2004;Barr et al., 2013) is able to bind to P2YRs on neighbouring cells, activating phospholipase C (PLC) and inducing an IP 3 -mediated release of calcium from the ER and subsequently SOCE. However, there is a lack of consensus regarding the relative contribution of each pathway in mediating calcium signal transmission in keratinocytes (Korkiamaki et al., 2005;Tsutsumi et al., 2009). We therefore sought to investigate pharmacologically the role of gap junctions and extracellular signalling on wound induced calcium flux in keratinocytes.
18α glycyrrhetinic acid (18αGA) (Davidson et al., 1986), a stereoisoform metabolite of glycyrrhizin, blocks gap junction communication through modification of phosphorylation of residues in the cytoplasmic tail of Connexin 43 (Cx43) leading to changes in the assembly of connexins required for a functional gap-junction (Guan et al., 1996). As Cx43 is the most abundant connexin in mammals and is expressed throughout the epidermis (Martin et al., 2014), and 18αGA blocks dye transfer across gap-junction in keratinocytes (Easton et al., 2019), we utilised 18αGA to investigate the relative contribution of gap-junction communication to wound-induced calcium wave in human keratinocytes.  To investigate the contribution of ATP release to wound induced [Ca 2+ ] i flux, we utilised the ATP scavenger hexokinase (Liu et al., 2008) (Cheek et al., 1989) (Figures 4b and 4d) (Figures 4d and 4f). Kinetic analysis showed a reduction in time taken to reach maximum F t /F 0 in cells 5 and 6 rows back from the wound with hexokinase treatment (Figure 4g Calcineurin dephosphorylates the calcium dependent factor NFAT (isoforms 1-4), exposing its nuclear localisation domain resulting in NFAT translocation to the nucleus and transcriptional activation (Hogan et al., 2003). As keratinocytes express all four calciumdependent NFAT isoforms (Al-Daraji et al., 2002) which are activated by a variety of stimuli (Flockhart et al., 2008) and play a functional role in regulating growth, differentiation and migration (Jans et al., 2013;Wilson et al.), we hypothesised that wounding would activate NFAT in primary human keratinocytes. Moreover, by using F I G U R E 5 Differential effects of SOCE, ATP and gap junction signalling on wound-induced NFAT activation and wound closure in human keratinocytes. (a-d) Primary human keratinocytes cultured in 0.06 mM [Ca 2+ ] o KGM were transfected with NFAT firefly and renilla luciferase constructs and cultured until 100% confluent. Luciferase activity (mean firefly/renilla ratio ± SEM normalised to unwounded 0.06 mM [Ca 2+ ] o ) was measured 24 h after wounding. (a) Medium was replaced with either 0.06 or 1.2 mM [Ca 2+ ] o KGM before cross-hatch wounding; 47 experiments from 17 independent donors (n = 47, N = 17). Two-way ANOVA p = .0002; ***p < .001 Bonferroni post-hoc. (b) Medium was replaced with either 0.06 or 1.2 mM [Ca 2+ ] o KGM with 10 µM GSK-7975A or 1:1000 vehicle (DMSO) for 1 h before cross-hatch wounding (n = 19, N = 7). Two-way ANOVA p < .0001; ***p < .001 Bonferroni post-hoc. (c) Medium was replaced with either 0.06 or 1.2 mM [Ca 2+ ] o KGM with 20 µM 18αGA or 1:1000 vehicle (DMSO) for 1 h before cross-hatch wounding (n = 15, N = 5). Two-way ANOVA p < .0001; ***p < .001, **p < .01 Bonferroni post-hoc. (d) Medium was replaced with either 0.06 or 1.2 mM [Ca 2+ ] o KGM with 50 U/ml hexokinase or control for 5 min before cross-hatch wounding (n = 18; N = 6). Two-way ANOVA, p = .8802. (e-h) Primary human keratinocytes cultured in 0.06 mM [Ca 2+ ] o KGM were grown to confluency and treated with 20 µM 18αGA (a, g), 10 µM GSK-7975A (e) and 50 U/ml hexokinase (h) in 1.2 mM [Ca 2+ ] o for 1 h before wounding. Images were taken at predetermined co-ordinates every hour for 24 h. (e) Representative images of wound closure. Black lines depict wound edge (Scale bar = 250 µm). (f-g) Wound area was measured using Image J and calculated as a percentage of the original wound size. Graphs show mean ± SEM. (f) Two-way ANOVA p = .01 comparing wound closure rates between untreated and GSK-7975A treated cells; 10 experiments from three independent donors (n = 10, N = 3). (g) Two-way ANOVA p < .0001; ***p < .001 Bonferroni post hoc test comparing wound closure in untreated and 18αGA treated cells (n = 9, N = 3). (h) Two-way ANOVA p = .01, *p < .05 Bonferroni post-hoc test comparing wound closure rates between untreated and hexokinase treated cells (n = 12, N = 4). 18αGA, 18α-Glycyrrhetinic acid; ANOVA, analysis of variance; DMSO, dimethyl sulfoxide; KGM, keratinocyte growth medium; NFAT, Nuclear factor of activated T-cell; SOCE, store-operated calcium entry NFAT as a functional readout of wound-induced [Ca 2+ ] i flux, we aimed to investigate the relative contribution of SOCE, gap junctions and extracellular ATP signalling to wound-induced NFAT activation.
We measured wound-induced NFAT activation, in primary keratinocytes using a NFAT transactivation luciferase assay. Preliminary experiments established that wounding induced maximal increase in NFAT activity at 24 h. Unwounded cells exposed to 1.2 mM [Ca 2+ ] o showed a small 1.55 ± 0.14-fold increase in NFAT activation com-  Figure S8) consistent with the role of NfκB in regulating keratinocyte differentiation (Hinata et al., 2003).
These data indicate that a high external calcium environment, is required for NFAT activation postwounding.   (Faniku et al., 2018) and Kawano et al. showed in human bone marrow-derived mesenchymal stem cells that 18αGA blocked ATP-induced calcium oscillations and NFAT activation (Kawano et al., 2006). Notably though pretreatment of keratinocytes with 18αGA did not significantly reduce woundinduced ATP release ( Figure S11). Nevertheless, it is likely that ATP release and gap-junction signalling interact synergistically to trigger an intercellular calcium wave, NFAT activation and cell migration. For example, Lacobus et al. demonstrated this by analysing the characteristics of the intercellular calcium wave. These authors showed that wound-induced IP 3 -mediated waves only travelled a short distance, about five cells but combining ATP and IP 3 signalling mechanisms resulted in a longer propagation distance (Iacobas et al., 2006). In our study, it appears though that signalling through gap junctions was apparently unable to overcome blockade of ATP signalling in 1.2 mM [Ca 2+ ] o , reinforcing a model in which complex feedback loops integrate signals emanating from ATP signalling, calcium entry, and gap junctions.

| Relative contribution of SOCE, gap junction communication and extracellular ATP signalling in mediating wound-induced NFAT activation
As SOCE regulates NFATc1 translocation in keratinocytes (Jans et al., 2013) and ciclosporin and RNAi-mediated knockdown of NFATc1 inhibits lysophosphatidic acid-induced keratinocyte migration (Jans et al., 2013), we investigated whether scratch wounding resulted in transcriptional activation of NFAT. Our findings that wounding only activated NFAT in 1.2 mM [Ca 2+ ] o which was markedly attenuated by GSK-7975A underscores the role of SOCE in wound-induced NFAT activation.
Wounding in the presence of hexokinase did not attenuate transcriptional activity and the addition of ATP at a physiologically relevant concentration to unwounded keratinocytes did not induce NFAT transcriptional activation. Therefore, extracellular ATP released by keratinocytes following wounding does not appear to play a significant role in NFAT activation. This is in contrast to human bone marrow-derived mesenchymal stem cells (hMSCs) in which extracellular ATP released through hemi-gap-junction channels induced spontaneous calcium oscillations and downstream nuclear translocation of NAFTc3 (Kawano et al., 2006). Pharmacological inhibition of ATP release prevented induction of the calcium oscillations and blocked nuclear translocation of NFAT (Kawano et al., 2006). However, when these stem cells differentiated into adipocytes both spontaneous oscillations and NFAT activation ceased (Kawano et al., 2006). Similar results were seen in microglial cells in which Ferrari et al. observed that ATP was able to activate NFAT (Ferrari et al., 1999).
The propagation of the calcium wave through the population of cells appears to be important for downstream transcriptional activation, as determined by a reduction in wound-induced NFAT activation when wounding was performed in the presence of 18αGA. By studying Cx45 −/− mice embryos, Kumai et al. showed that signalling through gapjunctions was required to activate NFATc1 (Kumai et al., 2000). In knockout mice compared to wild type embryos, the authors showed that Cx45 regulated nuclear translocation of NFATc1. Consistent with these findings, we found that 18αGA, significantly inhibited wound in-