ORIGINAL ARTICLE

Panx1 Regulates Cellular Properties of Keratinocytes and Dermal Fibroblasts in Skin Development and Wound Healing Silvia Penuela1, John J. Kelly1, Jared M. Churko1,3, Kevin J. Barr2, Amy C. Berger2 and Dale W. Laird1,2 Pannexin1 (Panx1), a channel-forming glycoprotein is expressed in neonatal but not in aged mouse skin. Histological staining of Panx1 knockout (KO) mouse skin revealed a reduction in epidermal and dermal thickness and an increase in hypodermal adipose tissue. Following dorsal skin punch biopsies, mutant mice exhibited a significant delay in wound healing. Scratch wound and proliferation assays revealed that cultured keratinocytes from KO mice were more migratory, whereas dermal fibroblasts were more proliferative compared with controls. In addition, collagen gels populated with fibroblasts from KO mice exhibited significantly reduced contraction, comparable to WT fibroblasts treated with the Panx1 blocker, probenecid. KO fibroblasts did not increase asmooth muscle actin expression in response to TGF-b, as is the case for differentiating WT myofibroblasts during wound contraction. We conclude that Panx1 controls cellular properties of keratinocytes and dermal fibroblasts during early stages of skin development and modulates wound repair upon injury. Journal of Investigative Dermatology (2014) 134, 2026–2035; doi:10.1038/jid.2014.86; published online 27 March 2014

INTRODUCTION The channel-forming family of glycoproteins, known as pannexins, consists of three members: Panx1, Panx2, and Panx3 (Panchin et al., 2000). Although all three pannexins form functional single-membrane channels (Penuela et al., 2009), Panx1, which is the most ubiquitously expressed of the three (Baranova et al., 2004), has been more extensively studied (Penuela et al., 2013). The role of Panx1 as a mechanosensitive ATP-release channel (Bao et al., 2004) has been linked to paracrine signaling involving Ca2 þ wave propagation (Locovei et al., 2006), vasoconstriction (Billaud et al., 2011), airway defense (Seminario-Vidal et al., 2011), and even facilitating HIV viral infection (Seror et al., 2011). Panx1 has also been shown to mediate cellular responses in the immune system, acting as the ATP-release channel of ‘find-me’ signals from apoptotic cells to monocytes during their recruitment for cell clearance (Chekeni et al., 2010). Under pathological conditions, Panx1 has been reported to increase the severity of seizures in status epilepticus (Santiago 1

Department of Anatomy and Cell Biology, University of Western Ontario, London, Ontario, Canada and 2Physiology and Pharmacology, University of Western Ontario, London, Ontario, Canada

3

Current address: Department of Medicine (Cardiology), Stanford School of Medicine, Stanford, California 94305, USA. Correspondence: Silvia Penuela, Department of Anatomy and Cell Biology, University of Western Ontario, Dental Science Building DSB 00076, London, N6A5C1, Ontario, Canada. E-mail: [email protected]

Abbreviations: KO, knockout; REK, rat epidermal keratinocyte; a-SMA, a-smooth muscle actin; WT, wild type Received 16 October 2013; revised 8 January 2014; accepted 2 February 2014; accepted article preview online 12 February 2014; published online 27 March 2014

2026 Journal of Investigative Dermatology (2014), Volume 134

et al., 2011). Panx1 channel opening increases neuronal cell death under ischemic conditions (Thompson et al., 2006, 2008), and promotes death of enteric neurons in colitis (Gulbransen et al., 2012). In normal development, all three pannexins have been reported to have an important role in different cell types. Panx1 is important in differentiation of keratinocytes (Celetti et al., 2010), Panx3 in osteoblasts (Bond et al., 2011; Ishikawa et al., 2011) and chondrocytes (Iwamoto et al., 2010; Bond et al., 2011), and Panx2 in development of neuronal cells (Swayne et al., 2010). Panx1 has been implicated in the differentiation of cardiac fibroblasts into myofibroblasts in response to ATP released from cardiomyocytes (Dolmatova et al., 2012). Recently, it was also reported that mast cell– derived histamine can activate ATP release from human subcutaneous fibroblasts via Panx1 channels through H1 receptor activation (Pinheiro et al., 2013). Interestingly, Panx1 is not only functional at the cell surface but has also been shown to be expressed in intracellular compartments, such as the endoplasmic reticulum, where it is proposed to function as an intracellular Ca2 þ leak channel (Vanden Abeele et al., 2006; D’Hondt et al., 2011). In adult human skin, we have previously reported the expression of Panx1 and Panx3 in the different layers and appendages of the skin with various localization profiles (Penuela et al., 2007; Cowan et al., 2012). In mouse skin, we detected Panx1 and Panx3 expression as early as day E13.5 in the developing epidermis and dermis (Celetti et al., 2010). At the cellular level, we demonstrated that an overexpression of Panx1 in rat epidermal keratinocytes (REKs) reduced their proliferation rate, and disrupted their normal differentiation in a model of organotypic epidermis & 2014 The Society for Investigative Dermatology

S Penuela et al. Panx1 in Skin and Wound Healing

grown in three-dimensional cultures from these engineered REKs (Celetti et al., 2010). In addition, Panx1 expression in basal and squamous skin cell carcinomas, as well as in the more deadly melanomas, has been shown to change with malignant transformation of the skin. In human keratinocyte tumors, both Panx1 and Panx3 are significantly reduced (Cowan et al., 2012), whereas mouse melanomas express high levels of Panx1 that can be downregulated to revert them into a more normal melanocytic phenotype (Penuela et al., 2012). Using the Panx1 knockout (KO) mouse model reported by Qu et al. (2011), we set out to evaluate the role of Panx1 in skin development and response to injury in vivo. We discovered that Panx1 is necessary for early skin development, has decreasing expression in aged skin, and is upregulated again upon wounding. Panx1 has a role in the migration of keratinocytes and regulates differentiation of dermal fibroblasts into myofibroblasts for proper wound contraction. When Panx1 is missing, fibroblasts proliferate more and do not properly differentiate in response to TGF-b, leading to the formation of fibrotic scars. RESULTS Panx1 expression decreases in aging dorsal skin

Levels of endogenous Panx1 protein expression in the dorsal skin of neonatal mice and in mice up to 4 weeks of age were higher than those in adult mice (12 weeks old) and in aged mice of up to 18 months (Figure 1a). This same trend was observed in studies using Panx1 immunofluorescence labeling of skin sections of different ages (Figure 1b and c). Panx1 is localized at the cell surface of basal keratinocytes and hair follicles of neonatal skin (Figure 1b, white arrowheads), whereas dermal fibroblasts and differentiated keratinocytes of the upper layers of the epidermis present an intracellular distribution (Figure 1b, stars and open arrowheads). As the skin ages, Panx1 not only decreases in the different layers of the skin, but its localization pattern is less prevalent at the cell surface of the different skin cell types (Figure 1c). Characterization of the Panx1 KO mouse model

As previously described and characterized by Qu et al. (2011), a gene-targeting strategy was used to generate Panx1-null mice that were confirmed to lack Panx1 in tissues in which expression is normally abundant, such as the brain, thymus, and spleen (Figure 1d and e). Neonatal dorsal skin of the Panx1 KO was also devoid of Panx1. To verify the absence of any Panx1 transcript in the KO mouse, brain and thymus tissue from WT and KO mice were used for qPCR of Panx1 (Figure 1e). Panx1 is absent in KO mice dorsal skin (Figure 1f), whereas Panx3 was found to be upregulated in young and adult Panx1 KO skin compared with controls (Figure 1g). Decreased dermal thickness and increased hypodermal fat in Panx1 KO mice

Panx1 KO mice are viable, fertile, and do not show an overt skin phenotype. However, comparing histological sections of dorsal skin from 1-day to 7-week-old WT mice with those from KO mice, we detected a significant decrease in total dermal area (dermis and epidermis) of KO dorsal skin in every

age group. At the same time, the hypodermal fat area was significantly increased in the KO mice as early as 4 days after birth and this trend prevailed into adulthood (Figure 2). Panx1 is upregulated in dorsal skin wounds and its depletion delays wound healing, increasing fibrosis

Healing of dorsal wounds on 12-week-old males was significantly reduced in KO mice compared with that in WT controls at every stage of healing, including inflammation, proliferation, and tissue remodeling (Figure 3a). Although wounds eventually closed, there was a prominent scar on the KO mouse skin, which when examined in histological sections revealed an area of increased fibrosis (Figure 3b) that was quantified by measuring the maximum wound depth (N ¼ 4, n ¼ 12, Po0.01). As mentioned before, mouse dorsal skin at 12 weeks of age has low basal levels of Panx1 expression (Figure 1c). However, upon wounding, Panx1 expression detected by immunohistochemistry is increased near the site of the wound as early as 3 days after surgery in WT dorsal skin (Figure 3c). Panx1 is apparent in both dermis and epidermis, reaching maximum expression at day 6 after surgery. By day 13, Panx1 protein expression returns to basal levels around the wound site, but it is still highly expressed in the new re-epithelialized tissue (above dotted wound line, Figure 3c). Primary basal keratinocytes express high levels of Panx1 in early stages of differentiation

To elucidate the mechanism responsible for delayed wound healing in KO mice skin, we isolated primary keratinocytes from neonatal dorsal skin. WT basal keratinocytes are enriched in Panx1 (Figure 4a). A growth curve assay showed no significant differences in the growth rate of KO keratinocytes compared with that in WT controls (Figure 4b). However, an in vitro scratch wound assay revealed a significantly dysregulated migration/proliferation of KO keratinocytes after 12 hours of migration in the presence of serum (Figure 4c). To test for the ability of keratinocytes to differentiate in culture, cells were exposed for 2 days to medium containing CaCl2 solution. Panx1 protein levels were significantly reduced 2 days after calcium treatment (Figure 4d). However, both KO and WT keratinocytes could start to differentiate in vitro, as the differentiation marker involucrin was significantly increased in both groups of keratinocytes (Figure 4e). There were no apparent differences between WT and KO cells in the levels of CK14, a marker of basal keratinocytes (Figure 4e). To test whether the Panx1 present in these primary keratinocytes was forming active channels at the cell surface, we performed a sulforhodamine B dye uptake assay as described earlier by our group (Celetti et al., 2010). WT keratinocytes could readily uptake dye, whereas KO cells exhibited significantly reduced dye uptake. Dextran-rhodamine, which is too large to pass through pannexin channels, did not enter the cells, indicating that the cells were healthy and the plasma membrane was intact (Figure 4f). All dye uptake assays were performed in the presence of physiological levels of calcium and magnesium to minimize any potential uptake due to calcium-gated connexin hemichannels. www.jidonline.org 2027

S Penuela et al. Panx1 in Skin and Wound Healing

ks ee

*

W

ee W

1 Day-old dorsal skin

18

12

4

4

D

ay s W ee ks

ks

Panx1

Epidermis -50 Gly2-

Panx1 Gly1-

-37

Gly0-

Dermis

GAPDH

1 Day

12 Weeks

4 Weeks

Thymus WT

KO

Panx1 gene expression (normalized to β-2 micoglobulin)

Panx1

4 Days

Spleen WT

KO -50

Panx1

-37

GAPDH

4 3 2 1 0 WT brain

*** *** KO KO WT brain thymus thymus

WT

WT KO WT KO

KO 63-

50-

Panx3 Panx1

37-

37GAPDH 4 Days

GAPDH 4 Weeks 12 Weeks

Panx3 protein expression (~43kDa band normalized to GAPDH)

Panx3 in dorsal skin Panx1 in dorsal skin

*** 0.06 0.04 * 0.02 0.00 WT

KO

4 Weeks

WT

KO

12 Weeks

Figure 1. Panx1 decreases in aging skin and is absent in Panx1 knockout (KO)–derived tissues. (a) Western blots and (b, c) immunohistochemistry revealed that Panx1 is elevated in neonatal skin and it decreases with aging. Cell surface localization of Panx1 in the basal epidermal keratinocytes and hair follicle cells (solid arrows) is prevalent in 1-day-old skin (b) but is reduced 4 days after birth (c) and in differentiated keratinocytes (*). (b) Panx1 is also expressed in dermal fibroblasts (open arrows) of neonatal skin. (d) Western blots of protein lysates from Panx1 KO tissues, revealed no expression of Panx1 compared with wild type (WT). (e) qPCR of transcripts from brain and thymus confirmed the absence of Panx1 mRNA in the KO mice. (f) Panx1 is absent in dorsal skin of KO mice. (g) The B43-kDa isoform of Panx3 is highly upregulated in 4-week-old Panx1 KO mouse skin and, to a lesser extent, in adult skin, whereas the B70-kDa band of Panx3 does not change. Student’s t-test, N ¼ 4, Po0.0001 or Po0.05. Bars ¼ 20 mm. Blue, nuclei. Protein sizes in kDa. Gly0, 1, 2 represent Panx1 bands with different levels of glycosylation. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as loading control.

Panx1 KO–derived primary dermal fibroblasts proliferate less than WT fibroblasts and have reduced dye uptake capabilities

Primary dermal fibroblasts from WT neonatal mice abundantly express Panx1 (Figure 5a). KO-derived fibroblasts grew faster than WT control fibroblasts as revealed by growth curve analysis (Figure 5b). This difference in growth was due to proliferation and not due to decreased apoptosis, as shown by differential labeling of nuclei with the proliferation marker phospho-histone 3 (Figure 5c) and by apoptotic nuclei labeled with Apoptag (TUNEL assay, Figure 5d). There was no 2028 Journal of Investigative Dermatology (2014), Volume 134

significant difference between WT and KO fibroblasts in the total migrated distance after a scratch wound assay in vitro (Figure 5e). Similar to primary keratinocytes, WT primary dermal fibroblasts were also capable of sulforhodamine B dye uptake (Figure 5f), indicating that there were functional channels present at the cell surface. Panx1 KO fibroblasts showed significantly reduced dye uptake, and dextran rhodamine controls were again negative (not shown). The Panx1 blocker probenecid (PBN) had a dose-dependent response in the

S Penuela et al. Panx1 in Skin and Wound Healing

WT

KO

6,000 5,000 4,000 3,000 2,000 1,000 0

**

1,000 500

Hypodermal area (μm2)

2,500 2,000 1,500 1,000 500 0

1,500

Hypodermal area

0

*

WT

KO

NS

WT

KO

*

1,750 1,500 1,250 1,000 750 500 250 0 WT

KO

*

WT

800 600 400 200 0

KO

Hypodermal area (μm2)

WT

Hypodermal area (μm2)

7 Weeks

Dermal area (μm2)

4 Days

Dermal area (μm2)

1 Day

Dermal area (μm2)

Dermal area 2,000

3,000 2,500 2,000 1,500 1,000 500 0

KO

*

WT

KO

Figure 2. Skin thickness is decreased in Panx1 knockout (KO) mice, whereas hypodermal fat increases compared with wild-type (WT) controls. Histological analysis of dorsal skin sections from KO mice showed a reduction in dermal area (dermis and epidermis) when compared with WT controls. N ¼ 4, Po0.05. As early as 4 days after birth, Panx1 KO mouse skin shows an accumulation of hypodermal adipose tissue that persists into adulthood, N ¼ 4, Po0.05 (NS, not significant).

inhibition of dye uptake by WT fibroblasts but had no added blocking effect on dye uptake from KO cells (Figure 5f and g). Contractile properties of Panx1 KO–derived dermal fibroblasts are reduced, and they do not increase smooth muscle actin in response to TGF-b stimulation

To test whether Panx1 KO–derived dermal fibroblasts were able to differentiate into myofibroblasts and efficiently contract a free-floating collagen gel, we seeded equal numbers of WT and KO fibroblasts into collagen lattices. The gel surface area was measured for 96 hours. A significant decrease in contraction was observed with KO-derived fibroblasts at all time points (Figure 6a). This decrease was comparable to that observed with WT-derived fibroblasts and subjected to the Panx1 blocker PBN (Figure 6b). A dose-dependent response was evident with 0.5 or 1 mM PBN, indicating that the channel blocker had an effect on the contractile properties of the Panx1-rich WT fibroblasts. Control gels without cells or without serum showed no significant contraction as expected, suggesting that the contraction is a mechanistic effect of the myofibroblasts (Figure 6b). Mouse dermal fibroblasts tend to spontaneously differentiate when plated on hard surfaces or on collagen-coated plastic dishware (Elliott et al., 2012). Both WT and KO dermal fibroblasts had some level of differentiation and expressed basal levels of smooth muscle actin (a-SMA) as seen on both western blots and immunofluorescence (Figure 6c and d). However, WT fibroblasts responded to the addition of TGF-b by increasing the levels of a-SMA, a necessary step in normal

differentiation into contractile myofibroblasts. On the other hand, Panx1 KO fibroblasts failed to further respond to TGF-b maintaining the same levels of a-SMA, consistent with their inability to fully contract (Figure 6a). Morphologically, KO fibroblasts did not respond to TGF-b stimulation, whereas WT fibroblasts acquired a contractile myofibroblast morphology after stimulation (Figure 6d). DISCUSSION Normal skin development requires a carefully orchestrated set of events involving different cell types derived from the ectoderm and the mesenchyme. We have previously demonstrated that Panx1 is expressed as early as day E13.5 in the developing epidermis and dermis (Celetti et al., 2010). In neonatal skin, Panx1 is highly expressed in keratinocytes, fibroblasts, and hair follicle cells. In basal keratinocytes and hair follicles, Panx1 is expressed at the cell surface in a profile reminiscent of previously reported ectopic expression in reference cells (Boassa et al., 2007; Penuela et al., 2007). As keratinocytes differentiate and move upwards in the epidermal layer, their Panx1 expression decreases and localizes to intracellular compartments. On the basis of our previous work (Penuela et al., 2009) and other reports (Boassa et al., 2007), we would predict that the most complex glycosylated form of Panx1 (Gly2) would be more prevalent at the cell surface, whereas Gly1 and Gly0 would be more abundant intracellularly. The change in Panx1 expression was recapitulated in vitro when Panx1-rich primary basal keratinocytes were induced to differentiate in the presence www.jidonline.org 2029

S Penuela et al. Panx1 in Skin and Wound Healing

WT

KO

35

Day 1

Day 6 Day 9

30 Wound size (mm2)

Day 3

WT

***

KO

***

**

25 20 15 10

***

***

5 Day 13 0 0

6

3

12

9

15

Days

WT

KO

Maximum scar depth (mm)

** 5 4 3 2 1 0 WT

Day 1

Day 3

Day 6

Day 13

KO

Figure 3. Panx1 is transiently upregulated at wound sites and its depletion increases fibrosis and delays wound healing. (a) Wound-healing assays performed with dermal punch biopsies on the dorsal area of 12-week-old male mice showed significant delays in wound closure of Panx1-null skin compared with wild-type controls during all phases of wound repair. Duplicate wounds, n ¼ 28, N ¼ 14, Po0.01 or 0.001. Bars indicate SEM. (b) Histology of healed wounds at 13 days after surgery show significant fibrosis and increased scar depth in knockout (KO) wounds, whereas controls healed normally. (c) Immunolabeling of Panx1 (red) at different time points during healing of wild-type (WT) skin wounds revealed an upregulation of Panx1 at the wound site (basal levels are low in 12-week-old skin, day 1) as early as 3 days after surgery, with a peak intensity at day 6. Panx1 expression returns to basal levels near healed wounds (day 13) but is still expressed in re-epithelialized wound tissue. Dashed line indicates wound edge. Nuclei, blue. Bars ¼ 20 mm.

of calcium, and began to decrease their Panx1 expression while increasing expression of involucrin, a differentiation marker. However, Panx1 KO–derived primary keratinocytes 2030 Journal of Investigative Dermatology (2014), Volume 134

were still able to differentiate, so Panx1 may not be required for initiating this process. Panx1 is important for proper skin architecture, as Panx1 KO skin tends to have a thinner dermal area than WT mice. This may be due to the effect of Panx1 deletion not only on the keratinocytes but also on the dermal fibroblasts that appear to show dysregulated proliferation and differentiation. As skin ages, Panx1 expression decreases in all layers of the skin, indicating that this channel protein is needed early on in development but is downregulated as skin matures. This is similar to earlier reports of Panx1 expressed at high levels in the embryonic and young neonatal rat brain that significantly decline in adult rat brain (Vogt et al., 2005). Only when the skin is challenged with an injury is Panx1 again needed at the wound site, with expression peaking around day 6 after surgery but returning to basal expression levels typical of adult mouse skin once the wound is healed. The thinner dermal area seen on KO mouse skin appears to be offset by an increase in the hypodermal fat area, which may simply occur as a response to the need to maintain a thermal balance of the thinner skin. Alternatively, the increase in hypodermal fat could be due to potential changes in hormone levels as has been reported for male mice with low testosterone (Azzi et al., 2005), or other possible changes in adipocyte regulation that move beyond the scope of this study. Similarly, without challenge, KO mice of different ages seem to have no overt skin phenotypes other than the changes in thickness seen under histological observations. This may be explained in part by the upregulation and potential compensation of Panx3 in KO skin of Panx1-null mice, particularly at early stages when Panx1 would be needed for proper development. However, upon injury of the dorsal skin at 12 weeks of age, the lack of Panx1, even in the presence of Panx3 expression in the skin, results in significant delays in wound healing, and after wound closure the scars that remain on the KO skin are highly fibrotic. We observed a significant difference between WT and KO skin in all phases of wound healing. Early stages of wound healing could involve pannexins due to a compromised inflammation response at days 1–3. Panx1 has been reported to have a role in the inflammasome (Silverman et al., 2009), is known to be expressed in macrophages (Pelegrin and Surprenant, 2006), and could very likely be involved in the signaling process of immune cells for phagocytic clearance of apoptotic cells (Chekeni et al., 2010). The granulation or proliferation phase is also significantly delayed in which both Panx1-rich keratinocytes and dermal fibroblasts have a major role (Gurtner et al., 2008). Here, the deletion of Panx1 dysregulates migration of keratinocytes and increases proliferation of dermal fibroblasts, and these effects may be due in part to a lack of ATP release via Panx1 channels present in these two cell types. ATP released from Panx1 channels could potentially act on purinergic receptors to activate downstream signaling that affects proliferation, differentiation, and apoptosis of different skin cell types, as reviewed by Burnstock et al. (2012). ATP released in response to mechanical stimulation and cell damage is involved in proper wound healing (Wang et al., 1990), hence we could speculate that the complex of Panx1

S Penuela et al. Panx1 in Skin and Wound Healing

WT KO

1.3×106

KO Total cells per ml

WT 50 -

Panx1 37 -

1.0×106 7.5×105 5.0×105 2.5×105 0

β-Tubulin

0

2

4

6

10

8

Days

Migration distance after 12 hours (μm)

250

**

200 Day: 0

150 100

2

0

2

1

Involucrin -

50

CK14 -

0

WT

KO Day:

0

1

2

0.8 Panx1 0.6

** 0.4 0.2

Involucrin expression (normalized to GAPDH)

WT

Panx1 expression (normalized to GAPDH)

1

KO

0.5 0.4 0.3 0.2 0.1 0.0 Day0 Day1 Day2 Day0 Day1 Day2

0.0

Sulforhodamine B

Day1

Phase

Day2

WT

KO

Dextran-rhodamine

WT

KO

Dye uptake incidence (dye area/cell area)

Day0

0.3 0.2

*

0.1 0.0 WT

KO

Figure 4. Primary keratinocytes derived from knockout (KO) neonatal skin are more migratory than those from wild-type (WT) controls, and their dye uptake ability is impaired. (a) Panx1 is highly expressed in WT basal keratinocytes, as seen on western blots of total protein lysates, but its depletion does not significantly alter their growth (b). P40.05, N ¼ 6. (c) Scratch wound experiments revealed that Panx1-null primary keratinocytes were more migratory than controls. N ¼ 5, Po0.01. (d) Western blots of differentiating primary keratinocytes from wild-type mice in the presence of 1.4 mM CaCl2 revealed a progressive decrease (from day 0 to day 2) in the levels of Panx1. N ¼ 6, Po0.01. (e) A marker of keratinocyte differentiation (involucrin) increased, whereas basal marker cytokeratin 14 (CK14) remained unchanged with no significant differences between WT and KO genotypes. (f) Sulforhodamine B dye uptake in Panx1 null– derived primary cells (KO) was reduced under conditions of physiological calcium and mechanical stimulation when compared with wild-type controls (WT). Po0.05, N ¼ 4. Negative control experiments using dextran-rhodamine revealed no dye uptake. Error bars indicate SEM. GAPDH and b-tubulin were used as loading controls.

and purinergic receptors orchestrates the proper signaling cascade needed for the healing response. In addition, KO fibroblasts proliferate more in vitro, and can migrate as well as the WT fibroblasts, but we hypothesize that once at the site of injury, they cannot fully respond to TGF-b for proper differentiation into a-SMA-enriched myofibroblasts that serve to contract the wound (Desmouliere et al., 1993; Tomasek et al., 2002). This lack of differentiation could be due to the absence of ATP release needed for the proper differentiation and

contraction of fibroblasts (Ehrlich et al., 1986; Dolmatova et al., 2012) or due to other molecules that may pass through the channel. Some Panx1 channel-independent functions are also possible, such as direct binding of the C-tail of Panx1 to actin (Bhalla-Gehi et al., 2010). However, as PBN, although an imperfect channel blocker (Silverman et al., 2008), seems to block the dye uptake function in WT fibroblasts and mimics the contraction impairment exhibited by KO fibroblasts, the channel function of Panx1 is likely involved. The last phase of wound www.jidonline.org 2031

S Penuela et al. Panx1 in Skin and Wound Healing

Total cells per ml

KO

50Panx1 37β-Tubulin

*

WT KO

2.0×105 WT

1.5×105

**

*

1.0×105 5.0×104 0

0.4 0.3 0.2 0.1 0.0 WT

2

4

6

0.20 Migration distance after 8 hours (μm)

***

0.5

Apoptotic/total nuclei

Proliferating/total nuclei

0

NS

0.15 0.10 0.05 0.00

KO

WT

8

NS

200 150 100 50 0

KO

WT

KO

0.4 Dye uptake incidence (dye area/cell area)

0.2 mM PBN

Control

Sulforhodamine uptake WT KO

a

0.3

a

0.2 0.1 b

b

b

b

KO

WT

KO

0.0 1.0 mM PBN

WT

KO

Control

WT

0.2 mM PBN

1.0 mM PBN

Figure 5. Dermal fibroblasts from knockout (KO) mice dorsal skin proliferate more than those from wild-type (WT) controls and have impaired dye uptake function. (a) Dermal fibroblasts from WT neonatal skin are rich in Panx1 as shown by western blot. b-Tubulin was used as loading control. (b) Dermal fibroblasts from KO neonatal mice subjected to growth curve analysis had a significant increase in proliferation. N ¼ 6, Po0.05 or 0.01. (c) Increased cell proliferation was confirmed by quantification of nuclei, stained with phospho-histone 3 (pHis3), N ¼ 3, Po0.001. (d) Apoptag labeling (TdT terminal deoxynucleotidyl transferase) of nuclei undergoing apoptosis showed no difference between WT and KO primary fibroblasts. (e) Scratch wound experiments revealed that Panx1-null primary fibroblasts had no significant difference in migration compared with WT controls. N ¼ 5, P40.05. (f) Sulforhodamine-B dye uptake in Panx1 KO–derived primary cells was reduced under conditions of physiological calcium and mechanical stimulation when compared with WT controls. WT fibroblasts responded in a dose-dependent manner to probenecid (PBN) inhibition of dye uptake, whereas the Panx1 blocker had no further inhibitory effect on dye uptake in KO-derived cells. Po0.05, N ¼ 4. Error bars indicate SEM (NS, not significant).

healing, known as tissue remodeling (Singer and Clark, 1999) may also be compromised in the absence of Panx1 as KO scars show increased fibrosis. Fibrotic and hypertrophic scars are characterized by the excessive production of collagen and other extracellular matrix components by the proliferating fibroblasts that are not correctly remodeled, as is the case of keloid scars (Tredget et al., 1997). Fibrotic scars could be explained by increased proliferation of the KO fibroblasts at the wound site, but also may be due to defects in extracellular matrix remodeling and collagen reabsorption (Gurtner et al., 2008). 2032 Journal of Investigative Dermatology (2014), Volume 134

Collectively, we have shown that Panx1 has a key role in proper skin development and its expression is not needed in aging skin, unless there is an injury that requires its upregulation. In keratinocytes and dermal fibroblasts, Panx1 is required for proper migration and differentiation as KO fibroblasts do not respond to TGF-b by increasing a-SMA to form fully differentiated myofibroblasts that can contract properly and aid wound closure. Panx1 may be a new target for wound-healing treatments and re-epithelialization.

S Penuela et al. Panx1 in Skin and Wound Healing

90

% Contraction

80

WT

70

***

***

***

60

WT 1

WT 2

KO 1

KO 2

KO No cells

*** ***

50 40 0 0

24

48 Time (hours)

72

96

90 80 % Contraction

70

* ***

60 50

***

Control 0.5 mM PROB 1 mM PROB No FBS No cells

*** ***

** ***

40

***

30

No FBS

Control

0.5 mM

1 mM

20 10 0 0

24

48 72 Time (hours)

96

– TGFβ

WT WT KO KO + + – TGFβ: –

WT

α-SMA

KO

α-SMA

α-Smooth muscle actin

48-

β-Tubulin

α-Smooth muscle actin/β-tubulin (normalized to controls)

+ TGFβ

*

1.75

NS

1.50 1.25 1.00 0.75 0.50 0.25 0.00 TGFβ:

WT –

WT +

KO –

KO +

Figure 6. Contractile properties of dermal fibroblasts are compromised upon Panx1 deletion. (a) In a collagen lattice contractility assay, knockout (KO) dermal fibroblasts embedded in free-floating collagen gels contracted significantly less than those embedded with wild-type (WT) fibroblasts (two-way ANOVA, N ¼ 6, n ¼ 36, Po0.001) over 4 days. Gels without cells did not contract. (b) Images in a and b are representative of gels at the 96 hours time point. WT fibroblasts in a similar time course contractility assay exposed to 0.5 or 1.0 mM of the Panx1 blocker probenecid (PROB) had reduced contractility in a dose-dependent manner (two-way ANOVA, N ¼ 6, n ¼ 36, Po0.001 ). Gels without cells or with cells but no fetal bovine serum (FBS) had no significant gel contraction. Bars: SEM. (c) Fibroblasts derived from KO mice neonatal skin do not respond to the addition of TGF-b1 (NS, not significant), whereas WT fibroblasts respond by increasing the expression of a-smooth muscle actin (SMA; two-way ANOVA, N ¼ 3, Po0.05). In two-dimensional culture, WT fibroblasts in the presence of TGF-b1 can change morphology by differentiating into contractile myofibroblasts, whereas KO fibroblasts do not change cell morphology (d). SMA, red. Nuclei, blue. Bars ¼ 20 mm. b-Tubulin was used as a loading control.

MATERIALS AND METHODS Detailed methods in Supplementary Section online.

Panx3 antibodies used (specific to the C terminus) were described in the study by Penuela et al. (2009).

Ethic statement and animals

Immunolabeling

Animal experiments were approved by the Animal Use Subcommittee of the University Council on Animal Care at Western University. Panx1 KO mice were a kind gift from Dr Vishva Dixit at Genentech (San Francisco, CA) and were described by Qu et al. (2011). WT mice of the C57BL/6N strain were from Charles River Canada (SaintConstant, PQ).

Coverslips with primary cells were fixed in either methanol acetone or 10% neutral buffered formalin and immunolabeled with primary antibodies as described previously (Penuela et al., 2007). For proliferation assays, cells were labeled with the M-phase-specific marker phospho-histone 3 (Millipore, Billerica, MA). For apoptosis assays, cells were labeled following the Apoptag Fluorescein kit instructions (Millipore) and nuclei were stained with Hoechst 33342. Immunolabeling of dorsal skin samples was performed as described in the study by Celetti et al. (2010). For histology assays, parallel tissue sections were stained with hematoxylin/eosin. Images

Western blots Western blots of tissues and primary cell lysates were performed as described in the study by Penuela et al. (2007, 2009). Panx1 and

www.jidonline.org 2033

S Penuela et al. Panx1 in Skin and Wound Healing

were collected using a Leica DM IRE2 inverted epifluorescence microscope as described in the study by Churko et al. (2011a). ImageJ was used to determine the dermal area (epidermis and dermis) as well as the hypodermal area (area between dermis and panniculus carnosus). All statistical analyses were performed using GraphPad v.4.1.

Primary keratinocyte and dermal fibroblast cultures Neonatal dorsal skin was dissected from KO and WT pups. After disinfection, the skin was processed either for primary keratinocyte extraction and differentiation following the protocol described in the study by Churko et al. (2012), or for extraction of dermal fibroblasts and TGF-b differentiation as reported by Churko et al. (2011b).

Quantitative PCR Total RNA was extracted with Qiagen RNeasy kits (Qiagen, Toronto, ON, Canada) from brain and thymus dissected from 7-week-old females from WT and Panx1 KO genotypes. Using the RevertAid H minus, first-strand cDNA synthesis kit from Fermentas (Burlington, ON, Canada), total RNA was used to generate the cDNA template. Panx1 transcript levels were determined using mouse Panx1-specific primers (forward: 50 -ACAGGCTGCCTTTGTGGATTCA-30 and reverse: 50 -GGGCAGGTACAGGAGTATG-30 ) for qPCR amplification with iQ SYBR green Supermix (Bio-Rad, Mississauga, ON, Canada) in a Bio-Rad CFX96 real-time system. Results were normalized to b-2 microglobulin expression.

Cell growth and migration assays Primary cells were evaluated in a growth curve assay as described before (Churko et al., 2011b, 2012). Scrape wounding migration assays were performed as described by Churko et al. (2011b).

Dye uptake assays Primary cells were subjected to dye uptake assays under physiological calcium as reported by Celetti et al., (2010) and Penuela et al. (2012). PBN (Sigma, Oakville, ON, Canada) inhibition of dye uptake was done at 0.2 mM or 1.0 mM PBN, with pre-incubation at 37 1C in growth media for 1 hour. All dye uptake solutions and washes contained the same dose of PBN. Controls were performed using the vehicle solvent (D-PBS with calcium and magnesium, pH 7.4, Invitrogen, Burlington, ON, Canada).

Wound healing Twelve-week-old male mice (WT and KO) were anesthetized with isoflurane gas and dorsal skin incisions were performed and evaluated as described by Churko et al. (2011b).

Collagen gel contractility assay Free-floating collagen gels were prepared from primary mouse fibroblasts isolated from six wild-type mice and six Panx1 KO mice. This technique has been described elsewhere (Grinnell et al., 1999; Ngo et al., 2006; Kelly et al., 2012). For drug-treated gels, the pannexin-1 inhibitor probenecid (Silverman et al., 2008) was added to the cell suspensions and culture media to give a final concentration of 0.5–1 mM.

CONFLICT OF INTEREST The authors state no conflict of interest.

2034 Journal of Investigative Dermatology (2014), Volume 134

ACKNOWLEDGMENTS We thank Vishva Dixit (Genentech) for generously providing the Panx1 KO mice. We also thank Daniella Ostojic and Cindy Sawyez for their valuable contributions to this work. This work was supported by the Canadian Institutes of Health Research (CIHR). SUPPLEMENTARY MATERIAL Supplementary material is linked to the online version of the paper at http:// www.nature.com/jid

REFERENCES Azzi L, El-Alfy M, Martel C et al. (2005) Gender differences in mouse skin morphology and specific effects of sex steroids and dehydroepiandrosterone. J Invest Dermatol 124:22–7 Bao L, Locovei S, Dahl G (2004) Pannexin membrane channels are mechanosensitive conduits for ATP. FEBS Lett 572:65–8 Baranova A, Ivanov D, Petrash N et al. (2004) The mammalian pannexin family is homologous to the invertebrate innexin gap junction proteins. Genomics 83:706–16 Bhalla-Gehi R, Penuela S, Churko JM et al. (2010) Pannexin1 and pannexin3 delivery, cell surface dynamics, and cytoskeletal interactions. J Biol Chem 285:9147–60 Billaud M, Lohman AW, Straub AC et al. (2011) Pannexin1 regulates alpha1adrenergic receptor- mediated vasoconstriction. Circ Res 109:80–5 Boassa D, Ambrosi C, Qiu F et al. (2007) Pannexin1 channels contain a glycosylation site that targets the hexamer to the plasma membrane. J Biol Chem 282:31733–43 Bond SR, Lau A, Penuela S et al. (2011) Pannexin 3 is a novel target for Runx2, expressed by osteoblasts and mature growth plate chondrocytes. J Bone Miner Res 26:2911–22 Burnstock G, Knight GE, Greig AV (2012) Purinergic signaling in healthy and diseased skin. J Invest Dermatol 132:526–46 Celetti SJ, Cowan KN, Penuela S et al. (2010) Implications of pannexin 1 and pannexin 3 for keratinocyte differentiation. J Cell Sci 123:1363–72 Chekeni FB, Elliott MR, Sandilos JK et al. (2010) Pannexin 1 channels mediate ’find-me’ signal release and membrane permeability during apoptosis. Nature 467:863–7 Churko JM, Chan J, Shao Q et al. (2011a) The G60S connexin43 mutant regulates hair growth and hair fiber morphology in a mouse model of human oculodentodigital dysplasia. J Invest Dermatol 131:2197–204 Churko JM, Kelly JJ, Macdonald A et al. (2012) The G60S Cx43 mutant enhances keratinocyte proliferation and differentiation. Exp Dermatol 21:612–8 Churko JM, Shao Q, Gong XQ et al. (2011b) Human dermal fibroblasts derived from oculodentodigital dysplasia patients suggest that patients may have wound-healing defects. Hum Mutat 32:456–66 Cowan KN, Langlois S, Penuela S et al. (2012) Pannexin1 and Pannexin3 exhibit distinct localization patterns in human skin appendages and are regulated during keratinocyte differentiation and carcinogenesis. Cell Commun Adhes 19:45–53 D’Hondt C, Ponsaerts R, De Smedt H et al. (2011) Pannexin channels in ATP release and beyond: an unexpected rendezvous at the endoplasmic reticulum. Cell Signal 23:305–16 Desmouliere A, Geinoz A, Gabbiani F et al. (1993) Transforming growth factor-beta 1 induces alpha-smooth muscle actin expression in granulation tissue myofibroblasts and in quiescent and growing cultured fibroblasts. J Cell Biol 122:103–11 Dolmatova E, Spagnol G, Boassa D et al. (2012) Cardiomyocyte ATP release through pannexin 1 aids in early fibroblast activation. Am J Physiol Heart Circ Physiol 303:H1208–18 Ehrlich HP, Rajaratnam JB, Griswold TR (1986) ATP-induced cell contraction in dermal fibroblasts: effects of cAMP and myosin light-chain kinase. J Cell Physiol 128:223–30 Elliott CG, Wang J, Guo X et al. (2012) Periostin modulates myofibroblast differentiation during full-thickness cutaneous wound repair. J Cell Sci 125:121–32

S Penuela et al. Panx1 in Skin and Wound Healing

Grinnell F, Ho CH, Lin YC et al. (1999) Differences in the regulation of fibroblast contraction of floating versus stressed collagen matrices. J Biol Chem 274:918–23

Qu Y, Misaghi S, Newton K et al. (2011) Pannexin-1 is required for ATP release during apoptosis but not for inflammasome activation. J Immunol 186: 6553–61

Gulbransen BD, Bashashati M, Hirota SA et al. (2012) Activation of neuronal P2  7 receptor-pannexin-1 mediates death of enteric neurons during colitis. Nat Med 18:600–4

Santiago MF, Veliskova J, Patel NK et al. (2011) Targeting pannexin1 improves seizure outcome. PLoS One 6:e25178

Gurtner GC, Werner S, Barrandon Y et al. (2008) Wound repair and regeneration. Wound Repair Regen 453:314–21 Ishikawa M, Iwamoto T, Nakamura T et al. (2011) Pannexin 3 functions as an ER Ca(2 þ ) channel, hemichannel, and gap junction to promote osteoblast differentiation. J Cell Biol 193:1257–74 Iwamoto T, Nakamura T, Doyle A et al. (2010) Pannexin 3 regulates intracellular ATP/cAMP levels and promotes chondrocyte differentiation. J Biol Chem 285:18948–58 Kelly JJ, Forge A, Jagger DJ (2012) Contractility in type III cochlear fibrocytes is dependent on non-muscle myosin II and intercellular gap junctional coupling. J Assoc Res Otolaryngol 13:473–84 Locovei S, Wang J, Dahl G (2006) Activation of pannexin 1 channels by ATP through P2Y receptors and by cytoplasmic calcium. FEBS Lett 580:239–44 Ngo P, Ramalingam P, Phillips JA et al. (2006) Collagen gel contraction assay. Methods Mol Biol 341:103–9 Panchin Y, Kelmanson I, Matz M et al. (2000) A ubiquitous family of putative gap junction molecules. Curr Biol 10:R473–4 Pelegrin P, Surprenant A (2006) Pannexin-1 mediates large pore formation and interleukin-1beta release by the ATP-gated P2  7 receptor. EMBO J 25:5071–82 Penuela S, Bhalla R, Gong XQ et al. (2007) Pannexin 1 and pannexin 3 are glycoproteins that exhibit many distinct characteristics from the connexin family of gap junction proteins. J Cell Sci 120:3772–83 Penuela S, Bhalla R, Nag K et al. (2009) Glycosylation regulates pannexin intermixing and cellular localization. Mol Biol Cell 20:4313–23 Penuela S, Gehi R, Laird DW (2013) The biochemistry and function of pannexin channels. Biochim Biophys Acta 1828:15–22 Penuela S, Gyenis L, Ablack A et al. (2012) Loss of pannexin 1 attenuates melanoma progression by reversion to a melanocytic phenotype. J Biol Chem 287:29184–93 Pinheiro AR, Paramos-de-Carvalho D, Certal M et al. (2013) Histamine induces ATP release from human subcutaneous fibroblasts via pannexin-1 hemichannels leading to Ca2 þ mobilization and cell proliferation. J Biol Chem 288:27571–83

Seminario-Vidal L, Okada SF, Sesma JI et al. (2011) Rho signaling regulates pannexin 1-mediated ATP release from airway epithelia. J Biol Chem 286:26277–86 Seror C, Melki MT, Subra F et al. (2011) Extracellular ATP acts on P2Y2 purinergic receptors to facilitate HIV-1 infection. J Exp Med 208:1823–34 Silverman W, Locovei S, Dahl G (2008) Probenecid, a gout remedy, inhibits pannexin 1 channels. Am J Physiol Cell Physiol 295:C761–7 Silverman WR, De Rivero Vaccari JP, Locovei S et al. (2009) The pannexin 1 channel activates the inflammasome in neurons and astrocytes. J Biol Chem 284:18143–51 Singer AJ, Clark RAF (1999) Cutaneous wound healing. N Engl J Med 341: 738–46 Swayne LA, Sorbara CD, Bennett SA (2010) Pannexin 2 is expressed by postnatal hippocampal neural progenitors and modulates neuronal commitment. J Biol Chem 285:24977–86 Thompson RJ, Jackson MF, Olah ME et al. (2008) Activation of pannexin-1 hemichannels augments aberrant bursting in the hippocampus. Science 322:1555–9 Thompson RJ, Zhou N, MacVicar BA (2006) Ischemia opens neuronal gap junction hemichannels. Science 312:924–7 Tomasek JJ, Gabbiani G, Hinz B et al. (2002) Myofibroblasts and mechanoregulation of connective tissue remodelling. Nat Rev Mol Cell Biol 3: 349–63 Tredget EE, Nedelec B, Scott PG et al. (1997) Hypertrophic scars, keloids and contractures: the cellular and molecular basis for therapy. Surg Clin North Am 77:701–30 Vanden Abeele F, Bidaux G, Gordienko D et al. (2006) Functional implications of calcium permeability of the channel formed by pannexin 1. J Cell Biol 174:535–46 Vogt A, Hormuzdi SG, Monyer H (2005) Pannexin1 and Pannexin2 expression in the developing and mature rat brain. Mol Brain Res 141:113–20 Wang DJ, Huang NN, Heppel LA (1990) Extracellular ATP shows synergistic enhancement of DNA synthesis when combined with agents that are active in wound healing or as neurotransmitters. Biochem Biophys Res Commun 166:251–8

www.jidonline.org 2035