Extracellular nucleotides are used as signaling molecules by several cell types. In epidermis, their release is triggered by insults such as ultraviolet radiation, barrier disruption, and tissue wounding, and by specific nerve terminals firing. Increased synthesis of hyaluronan, a ubiquitous extracellular matrix glycosaminoglycan, also occurs in response to stress, leading to the attractive hypothesis that nucleotide signaling and hyaluronan synthesis could also be linked. In HaCaT keratinocytes, ATP caused a rapid and strong but transient activation of hyaluronan synthase 2 (HAS2) expression via protein kinase C-, Ca2+/calmodulin-dependent protein kinase II-, mitogen-activated protein kinase-, and calcium response element-binding protein-dependent pathways by activating the purinergic P2Y2 receptor. Smaller but more persistent up-regulation of HAS3 and CD44, and delayed up-regulation of HAS1 were also observed. Accumulation of peri- and extracellular hyaluronan followed 4–6 h after stimulation, an effect further enhanced by the hyaluronan precursor glucosamine. AMP and adenosine, the degradation products of ATP, markedly inhibited HAS2 expression and, despite concomitant up-regulation of HAS1 and HAS3, inhibited hyaluronan synthesis. Functionally, ATP moderately increased cell migration, whereas AMP and adenosine had no effect. Our data highlight the strong influence of adenosinergic signaling on hyaluronan metabolism in human keratinocytes. Epidermal insults are associated with extracellular ATP release, as well as rapid up-regulation of HAS2/3, CD44, and hyaluronan synthesis, and we show here that the two phenomena are linked. Furthermore, as ATP is rapidly degraded, the opposite effects of its less phosphorylated derivatives facilitate a rapid shut-off of the hyaluronan response, providing a feedback mechanism to prevent excessive reactions when more persistent signals are absent.
ATP is the classical energy currency of live cells. In addition, it serves as an important signal transmitter in various tissues, including the epidermis [1,2], and between tissues, particularly the nervous system and target organs . It is released in response to mechanical stimulation , temperature stress , and chemical receptor stimulation  by direct cell lysis, exocytosis, or secretion via, e.g. connexin hemichannels and pannexin channels . ATP is also released after insults such as ultraviolet (UV) radiation [7,8] and tissue wounding or skin barrier disruption [9,10]. Additionally, extracellular ATP mediates inflammatory signals [11,12] and pain sensation (reviewed in ref. ).
ATP is released into the extracellular space at relatively high concentrations and degraded rapidly to ADP, AMP, and further to adenosine , which, except for AMP, are known to activate specific receptors even in a self-sustaining manner . ATP signals through both the ionotropic P2X (cation channels) and the metabotropic P2Y receptors, which activate intracellular second messengers via specific G protein activation. The initial discovery of the role of purinergic signaling in rapid responses regulating neurotransmission and secretion was later followed by observations that extracellular nucleotides also modulate long-term or trophic effects including cell proliferation, differentiation, and migration (reviewed in ref. ). There are seven functional P2X receptors (P2X1–7) and eight P2Y receptors (P2Y1–2, P2Y4, P2Y6, and P2Y11–14; ). In addition, another family of metabotropic P1 receptors with four distinct members (A1, A2A, A2B, and A3) relays the effects of extracellular adenosine . Besides ATP, the versatile P2Y family also utilizes other nucleotides and their derivatives: the main ligands also include UTP (P2Y2 and P2Y4), UDP (P2Y6), and UDP-glucose (P2Y14; see refs [3,17] for more complete listings of the known agonists).
ADP and adenosine utilize the metabotropic P2Y and P1 receptors, respectively. The ADP receptor P2Y1 and the ATP receptors P2Y2 and P2Y11 are coupled to Gq/11 proteins. They activate PLC, resulting in the formation of diacylglycerol and IP3 leading directly and indirectly, via Ca2+ release from the endoplasmic reticulum through IP3 receptor activation, to activation of protein kinase C (PKC). The ADP receptors P2Y12 and P2Y13 couple with Gi/o to inhibit adenylyl cyclase and cAMP production [18,19]. Keratinocytes express several purinergic receptors from the P2Y and P2X families [7,11,20]. Of the P1 receptors, A2B is the main subtype [21,22].
Interestingly, it appears that the expression of P2 receptors may depend on the differentiation status of the keratinocytes, as P2Y1 and P2Y2 are found mainly in the basal cell compartment, while P2X5 is located in the basal and spinous layers, and P2X7 is found in the stratum corneum . Functionally, P2Y2 is then associated with proliferation, P2X5 with differentiation, and P2X7 with apoptosis.
The potency of ATP on the different purinergic receptors varies with measured EC50 values ranging from 0.1 µM for P2X1 to as high as 2–4 mM for P2X7 . For the P2Y family, EC50 values are in the micromolar range for ATP, whereas nanomolar ADP may be effective, depending on the receptor . However, determining concentration dependency is difficult due to the quick metabolism of the agonists as well as the differential expression levels of the receptors, which may also change rapidly in response to physiological stimuli. The potency data also depend on the endpoint measured: e.g. it has been noted that adenosine is least effective in modulating cAMP levels through the P1 receptor A2B, whereas all the family members produce similar responses with respect to MAP kinase activation (reviewed in ref. ).
Hyaluronan is composed of repeating disaccharides of d-glucuronic acid and N-acetyl-d-glucosamine. This linear polysaccharide is synthesized at the inner leaflet of the plasma membrane and translocated into the extracellular matrix by the synthesizing enzymes . The three highly homologous mammalian hyaluronan synthases (HAS1–3; [27,28]) differ in tissue distribution, expressional regulation, and synthetic activity [29,30]. Hyaluronan can remain associated pericellularly via the synthases or specific receptors such as CD44, or be released to associate with other matrix molecules .
Hyaluronan is abundant in the dermal connective tissues and the epidermis . Epidermal hyaluronan synthesis is rapidly up-regulated after tissue wounding , disruption of the permeability barrier , and UVB (ultraviolet B radiation) exposure . As extracellular nucleotides are similarly released after tissue injury, it is relevant to ask whether these processes could be linked. Previous data on the effects of nucleosides and nucleotides on hyaluronan metabolism are scarce, but adenosine is known to up-regulate HAS1 expression and hyaluronan secretion in gingival fibroblasts  and smooth muscle cells , while UDP-Glc, UTP, and UDP enhance HAS2 expression and hyaluronan synthesis in keratinocytes [38,39].
In this work, we wanted to explore the role of ATP and its metabolites in hyaluronan metabolism and signaling in human keratinocytes. Our data indicate that extracellular ATP and its degradation products AMP and adenosine are potent regulators of HAS2, enabling a rapid launch and shut-off of hyaluronan synthesis, depending on nucleotide release and degradation. Furthermore, the up-regulation of HAS2 seemed to depend heavily on the activation of the purinergic P2Y2 receptor and downstream signaling cascades involving the calcium-activated PKC, CaMKII (Ca2+/calmodulin-dependent protein kinase II), and CREB (calcium response element-binding protein) as well as the MAP kinases p38 and pERK. Overall, our data shed light on the complex regulation of HA synthesis in keratinocytes under simulated stress conditions.
Materials and methods
HaCaT cells, a spontaneously immortalized human epidermal keratinocyte cell line developed by Fusenig and colleagues , were cultured in DMEM (Sigma–Aldrich, Inc., St. Louis, MO) containing 10% FBS (GE Healthcare Life Sciences/HyClone, Logan, UT), 2 mM l-glutamine (EuroClone, Milan, Italy), 50 units/ml penicillin, and 50 µg/ml streptomycin (EuroClone). The same base medium was used when treating the cells with the different nucleotides, siRNAs, and chemical inhibitors, unless otherwise specified.
ATP, ATPγS, βγ-methylene ATP, ADP, AMP, and adenosine (Sigma–Aldrich) were dissolved in H2O and stored as stock solutions at −20°C. The nucleotides were applied at 100 µM unless otherwise specified for 15 min–24 h.
A 2-h pretreatment was used with the CaMKII inhibitor KN93 (25 µM; Calbiochem/Merck Millipore, Darmstadt, Germany), the JAK2/EGFR inhibitor AG490 (30 μM; Sigma–Aldrich), and the STAT3 (signal transducer and activator of transcription) inhibitor IX/Cpd188 (50 μM; Calbiochem/Merck Millipore). A 30-min pretreatment before adding ATP was applied with the MEK inhibitor PD98059 (0.5 µM; Calbiochem/Merck Millipore), the PKC inhibitor bisindolylmaleimide I (BIM, 10 µM; Calbiochem/Merck Millipore), the CREB inhibitors KG501 (naphthol AS-E phosphate, 25 µM; Sigma–Aldrich) and naphthol AS-BI phosphate (25 µM; Santa Cruz Biotechnology, Inc., Dallas, TX), the P2Y1 inhibitor MRS2179 (5–30 µM; Tocris Bioscience, Bristol, U.K.), the P2Y11 inhibitor NF340 (10 µM; Tocris Bioscience), and the p38 inhibitor BIRB796 (2 µM; Axon Medchem BV, Groningen, The Netherlands). With pertussis toxin (PTX, 100 ng/ml; Sigma–Aldrich), a 17-h preincubation was used. BIM, naphthol AS-BI phosphate, NF340, MRS2179, KN93, and PTX were dissolved in water, and the STAT3 inhibitor IX, AG490, PD98059, KG501, and BIRB796 in DMSO. Equal amounts of these solvents were added to the control cultures, where necessary.
P2Y2-targeted siRNAs were obtained from Thermo Fisher Scientific (Invitrogen/Thermo Fisher Scientific, Waltham, MA) and the scrambled control siRNA from Eurogentec (Eurogentec/Kaneka, Osaka, Japan). Subconfluent cultures were transfected with 40 nM siRNA (a mixture of two specific sequences targeting P2Y2) using Lipofectamine RNAiMAX (Invitrogen/Thermo Fisher Scientific) according to the manufacturer's instructions. The transfection medium was removed after 4 h and replaced with regular culture medium. Total RNA was collected 2 days after the transfection and a 2-h incubation with ATP. Efficacy of the knockdown was confirmed by qRT-PCR.
RNA extraction and qRT-PCR
Total RNA was extracted with Eurozol (EuroClone). cDNA synthesis with 1 µg of total RNA as a template was performed using a Verso™ cDNA kit (Invitrogen/Thermo Fisher Scientific). Quantitative RT-PCR was run on a Stratagene Mx3000P thermal cycler (Agilent, La Jolla, CA), using the FastStart Universal SYBR Green Master with ROX (Roche, Basel, Switzerland). The cycling conditions were as follows: preincubation for 10 min at 95°C followed by 40 cycles of 15 s denaturation at 95°C, 60 s annealing at a primer specific temperature, and 60 s elongation at 72°C. Gene-specific amplification was confirmed by a melt curve analysis. The primers for the transcripts analyzed and the respective annealing temperatures are presented in Supplementary Table S1. Fold inductions were calculated using the formula , where ΔΔCt is ΔCt(sample replicate) − ΔCt(non-treated replicate), ΔCt is Ct(gene of interest) − Ct(ARP0), and Ct is the cycle at which the detection threshold is crossed.
Proteins were extracted by incubating the cultures on ice for 30 min with RIPA lysis buffer (150 mM NaCl, 50 mM Tris, 1% Nonidet P-40, 0.5% sodium deoxycholate, and 0.1% SDS) containing 10 mM NaF, 1% phosphatase inhibitor cocktail 2, and 0.5% protease inhibitor cocktail (Sigma–Aldrich). Protein content was measured using the Bradford assay . Samples containing 6.5–7.5 µg of total protein were resolved by 10% SDS–PAGE followed by transfer onto the nitrocellulose membrane (Amersham™ Protran™ 0.45 µm NC; GE Healthcare Life Sciences, Little Chalfont, U.K.) by a 2–3 mA/cm2 constant current with a Fastblot B43 semidry blotter (Biometra GmbH, Göttingen, Germany). The membrane was rinsed briefly with Tris-buffered saline (TBS; 10 mM Tris and 150 mM NaCl, pH 7.4), blocked with 1–5% BSA–TBS for 30 min, and incubated overnight at 4°C on a rolling mixer with the following primary antibodies: phospho-STAT3 (Tyr705) or phospho-STAT3 (Ser727) diluted 1 : 1000 in 1% BSA–TBS, phospho-CREB (Ser133) diluted 1 : 1000 in 5% BSA–TBS, and phospho-p38 diluted 1 : 1000 in 5% BSA–TBS (all from Cell Signaling, Danvers, MA), phospho-ERK1/2 diluted 1 : 200 in 2% BSA–TBS (Santa Cruz Biotechnology, Inc.), and either mouse or rabbit β-actin diluted 1 : 4000 in 2% BSA–TBS (Sigma–Aldrich). After washes with 0.1% Tween 20–TBS, the membranes were incubated with the secondary antibodies DyLight anti-rabbit 680 or 800 diluted 1 : 3000 in 2% nonfat milk–TBS (for pSTAT3) or 1 : 4000 in 2% BSA–TBS (for pCREB, phospho-p38, and rabbit β-actin) and antimouse 680 or 800 diluted 1 : 2000 in 2% nonfat milk–TBS (for pERK) or 1 : 4000–1 : 5000 in 2% BSA–TBS (for mouse β-actin; all secondaries from Pierce, Rockford, IL). β-Actin was probed separately after a 10-min stripping with 0.2 M NaOH. Protein bands were visualized and quantified with an Odyssey® infrared imaging system (LI-COR Biosciences, Lincoln, NE), normalizing all protein intensities to β-actin.
Hyaluronan enzyme-linked sorbent assay
For hyaluronan assays, HaCaT cells were cultured on 12- or 24-well plates and treated with the respective nucleotides (100 µM). Here, culture medium supplemented with 1% FBS was used for the entire duration of the nucleotide treatments (4–24 h). Additionally, some wells were supplemented with 1 mM glucosamine to provide an excess of the hyaluronan precursor UDP-GlcNAc (uridine diphosphate N-acetyl glucosamine). The media and trypsinates (representing extra- and pericellular hyaluronan, respectively) were collected for hyaluronan quantification after 4, 6, or 24 h. Cells were released with trypsin and counted for data normalization. A sandwich-type enzyme-linked sorbent assay for hyaluronan measurement (HA-ELSA, hyaluronan enzyme-linked sorbent assay) was performed as described previously .
Quantification of hyaluronan nucleotide sugar precursors by HPLC
HaCaT cells on 10-cm dishes were treated with 100 µM ATP for 1.5, 3, or 4.5 h. Cells were extracted and nucleotide sugars were measured as described previously . Briefly, the cells were washed twice with and scraped into ice-cold PBS, sonicated, and centrifuged. The samples were purified with Superclean Envi-Carb SPE-tubes (Sigma–Aldrich) followed by vacuum centrifugation. The dried samples were dissolved in water and analyzed by anion-exchange HPLC (CarpoPac™ PA1 column; Dionex/Thermo Fisher Scientific) as described previously . Total protein was measured with a Pierce BCA kit (Thermo Fisher Scientific) for normalization of the data.
For visualization of cell-associated hyaluronan, the cells were fixed for 20 min in 2% paraformaldehyde, permeabilized, and blocked for 10 min using 0.1% Triton X-100 in 0.1 M phosphate buffer pH 7.0 (PB), containing 1% BSA. The cultures were incubated overnight at 4°C with a biotinylated hyaluronan-binding complex (bHABC, 3 µg/ml; prepared in-house as described previously ) and subsequently for 1 h at room temperature with the ABC reagent (1 : 200, VECTASTAIN Elite ABC Kit; Vector Laboratories, Burlingame, CA). For visualization of the bound probe, the cultures were incubated for 5 min in 0.05% diaminobenzidine (DAB; Sigma–Aldrich) and 0.03% hydrogen peroxide in PB. The nuclei were counterstained with Mayer's hematoxylin. The stained cultures were viewed and imaged with a Zeiss Axio Imager M2 light microscope (Carl Zeiss Microimaging GmbH, Jena, Germany).
For dual stainings and confocal imaging, the anti-CD44 antibody Hermes 3 (a kind gift from Professor Sirpa Jalkanen, University of Turku) was added to the bHABC solution at a 1 : 100 dilution. FITC-labeled antimouse IgG (1 : 200) and TR-streptavidin (1 : 1000; both from Vector Laboratories) were the secondary reagents, respectively. The nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). The stainings were photographed using a 40× NA 1.3 oil objective on a Zeiss Axio Observer inverted microscope equipped with a confocal module (Zeiss LSM 700; Carl Zeiss Microimaging GmbH). Image processing was performed using the ZEN 2009 (Carl Zeiss Microimaging GmbH) and Adobe Photoshop Elements 9 (Adobe Systems, Inc., San Jose, CA) software.
Scratch wound and cell proliferation assays
To analyze the effects of the nucleotides on keratinocyte proliferation, HaCaT cells were seeded on 12-well plates at 80 000 cells/well and grown for 2 days. After this, the cells were treated with 10 or 100 µM ATP, AMP, or adenosine in medium containing 1% FBS. The cells were trypsinized and counted 24 h later to assess the impact of the treatments when compared with the controls (no added nucleotides).
To analyze differences in migration rates, HaCaT cells were seeded on 24-well plates at 80 000–160 000 cells/well. After 2 days of culture, a cross-shaped area was scraped in each well with a sterile 1 ml pipette tip to get a wound devoid of cells, and fresh medium containing 1% serum with or without 10 µM ATP, AMP, or adenosine was changed. The control and the nucleotide-treated cells (6–8 parallel wells/treatment) were photographed immediately and 6 and 24 h after wounding using an Olympus CK2 inverted phase contrast microscope (4× objective) and a Nikon Digital Sight DS-L1 camera system. The distance the cells migrated was analyzed using the ImageJ software (National Institutes of Health, U.S.A.; http://imagej.nih.gov/ij).
The results were analyzed with PASW Statistics 18 (SPSS, Inc., Chicago, IL). Log- or ln-transformation was used if the data were not normally distributed or the variances were unequal, or the non-parametric Friedman test was chosen. Mixed model ANOVA was applied, and pairwise comparisons between the treatments were performed using the estimated marginal means (LSD) or Dunnett's test. The controls (set as 1) and treatments were compared using the cumulative distribution function p-norm in R for Mac OS X, version 3.2.3 . The results were corrected for multiple comparisons. The one-sample t-test was used for comparisons of a single treatment normalized to control (set as 1). Statistical significances are indicated as *,#,§P < 0.05, **,##,§§P < 0.01, and ***,###,§§§P < 0.001. The data represent means and SEM, unless otherwise indicated.
Extracellular ATP strongly up-regulates
HAS2 expression in HaCaT keratinocytes
Treating HaCaT with ATP for 2 h strongly up-regulated HAS2 expression (Figure 1A) in a dose-dependent fashion (Figure 1B). As the response with 100 µM was robust and slightly larger than with 10 µM, the higher concentration was used in most subsequent experiments with ATP and its metabolites. Maximum induction occurred at 1.5 h and subsided to a 50% inhibition at 6 h (Figure 1C). Another stimulation was seen at 24 h (Figure 1C).
Extracellular ATP up-regulates
HAS expression in HaCaT keratinocytes.
There were no significant changes in the expression of HAS1, HYAL1-2, or CD44 at 2 h (Figure 1A), but HAS1 increased later, at 6 h (Figure 1D). ATP also caused a modest increase in the expression of HAS3, starting at 2 h (Figure 1A), and the transcript remained elevated (Figure 1E). CD44 showed a pattern similar to that of HAS3, being significantly up-regulated at 4–6 h (Figure 1F).
Metabolites of ATP differentially regulate
Like ATP, ADP (100 µM) strongly up-regulated HAS2 mRNA expression (Figure 2A). In contrast, the further dephosphorylated AMP down-regulated HAS2 at 2 h (range: 0.27–0.85; n = 12, P = 0.0001). This inhibitory effect persisted for at least 6 h, but after 24 h HAS2 expression in the AMP-treated cells exceeded the controls (Figure 2B). AMP induced a strong expression of HAS1, which peaked at 4–6 h (Figure 2C), while the effects on HAS3 were more modest, with the transcript levels rising steadily toward 24 h (Figure 2D).
The degradation products of ATP differentially regulate
Resembling the effects of AMP, adenosine inhibited HAS2 at 2–6 h, whereas the transcripts were restored to control levels by 24 h (Figure 2B). The effect of adenosine on HAS1 expression was similar to AMP (Figure 2C), while it did not significantly affect HAS3 (Figure 2D). CD44 did not exhibit significant responses to either AMP or adenosine (Figure 2E).
HAS2 expression predominantly via P2Y2, Ca2+, and mitogen-activated protein kinase signaling
To test the involvement of the P2Y receptors in the HAS2 response, siRNAs and chemical inhibitors were applied. The efficiency of the P2Y2 siRNA was 60% (Figure 3A), which was associated with a 68% reduction in the ATP-induced HAS2 up-regulation (Figure 3A). This suggests that P2Y2 was the dominant P2Y receptor in the HAS2 response, an idea strengthened by the negligible effect of the P2Y11 inhibitor NF340 (Supplementary Figure S1A).
HAS2 expression by ATP involves calcium and MAPK signaling via P2Y2.
Next, ADP receptors were blocked to study whether ATP signals through them. MRS2179, which inhibits P2Y1, slightly suppressed the HAS2 response (Figure 3B). Inhibition of P2Y12 and P2Y13 with PTX, which releases the Gi protein from the receptors  and effectively blocks Gi signaling in HaCaT without exhibiting toxicity , moderately inhibited the ATP-induced HAS2 up-regulation (Figure 3C). Additionally, up-regulation of HAS2 at 2 h by the unhydrolyzable ATP-analog ATPγS was not significantly different from the ATP-induced response (Figure 3D). This suggests that the metabolites of ATP such as ADP are not required for the up-regulation of HAS2 expression and further supports the conclusion that P2Y2 is the main receptor mediating the effect.
Adenosine, most probably acting via the P1 receptors , showed a tendency to inhibit the ATP response (Figure 3D). Additionally, β,γ-methylene ATP (AMP-PCP), an agonist of certain P2X receptors and an efficient local stimulator of cAMP formation when converted to AMP and adenosine , had a minor suppressive influence on HAS2 transcription (Figure 3D).
Signal transduction from the Gq/11 protein-coupled P2Y2 involves the secondary messengers diacylglycerol and Ca2+, and subsequent activation of downstream kinases including CaMKII , MEK/ERK , CREB , and PKC . Inhibiting CREB activity with two structurally similar chemicals preventing its binding to CBP  attenuated the induction in HAS2 transcription (Figure 3E). Upstream inhibitors of CaMKII (KN93) and PKC (BIM) almost totally abolished the ATP response (Figure 3F,G). The mitogen-activated protein kinase (MAPK) pathway (MEK/ERK and p38) inhibitors PD98059 and BIRB796 also significantly suppressed the ATP-induced HAS2 up-regulation (Figure 3H), while the JAK2/EGFR inhibitor AG490 was ineffective (Supplementary Figure S1B).
Western blotting was used to further characterize the effects of ATP: phosphorylations of p38, ERK, CREB, and ATF-1 (another cAMP-dependent transcription factor) were clear and fast with maximum induction after 15–30 min (Figure 4A–C), while that of pSTAT3-Ser727 was weaker, and pSTAT3-Tyr705 appeared later, after 60–120 min (Supplementary Figure S2A). However, this activation of STAT3 was apparently not involved in the regulation of HAS2 expression, as its specific inhibitor STAT IX failed to counteract the ATP-induced increase in HAS2 mRNA (Supplementary Figure S2B).
ATP induces the phosphorylation of p38, ERK, and CREB.
ATP and AMP/adenosine have opposing effects on the hyaluronan content of human keratinocyte cultures
The influence of adenosine nucleotides on the hyaluronan content of HaCaT cultures was analyzed with histochemical and biochemical assays. The histochemical stainings showed a clear increase in hyaluronan in the cell layer already 2 h after the application of 100 µM ATP (Figure 5A,B). Hyaluronan was mainly seen at the plasma membranes, colocalized (yellow color) with its receptor CD44, although an increase in the intracellular pool of HA was also evident (Figure 5D,E,D′,E′; 4 h after adding ATP, white arrowheads in Figure 5E).
ATP induces pericellular accumulation of hyaluronan.
For the biochemical analyses, cells were cultured using medium containing 1% FBS, since medium supplemented with 10% FBS alone stimulated hyaluronan synthesis submaximally and almost overwhelmed the increase induced by ATP (data not shown). Four hours after adding 100 µM ATP, the pericellular hyaluronan was augmented by 32%. After 6 h, the peri- and extracellular hyaluronan pools were increased by 67% and 86%, respectively (Figure 6A,B). At 24 h, there were no differences in either compartment. When the cultures were supplemented with 1 mM glucosamine to increase UDP-GlcNAc, a key substrate and activator of HAS2, hyaluronan content was increased even further both in the trypsinate and the medium (total increases with ATP and glucosamine: 216% and 173% after 6 h, respectively; Figure 6C). Glucosamine alone increased hyaluronan accumulation in the medium only by 37% (P < 0.05) and in the trypsinate by 43% (NS), indicating that both treatments are needed for a maximal effect.
Extracellular adenosine nucleotides differentially regulate hyaluronan metabolism.
Marked changes in hyaluronan stainings were not recognized after 2–4 h treatments with AMP and adenosine (Figure 5C,F,F′). The inhibitory effect of AMP on HAS2 expression was not reflected in hyaluronan accumulation during a 6-h follow-up, while after 24 h a clear 41% reduction was observed in the medium (Figure 6D). Adenosine caused a significant 38% reduction in pericellular hyaluronan already at 6 h and also suppressed hyaluronan content in the medium at 24 h by 36% (Figure 6E).
ATP influences keratinocyte migration
Since increased hyaluronan synthesis and HAS2 expression [55,56] have been connected to enhanced cell migration and proliferation, we tested the effects of ATP on keratinocyte numbers and migratory activity in scratch wound assays. Specifically, 10 and 100 µM ATP did not significantly affect cell numbers after a 24-h treatment (Figure 6G), although the higher dose has been considered potentially growth-inhibiting . In contrast, adenosine significantly lowered, and AMP had a tendency to suppress cell proliferation after a 24-h follow-up period with the 100 µM dose (Figure 6G). We then used for the migration assays the lower (10 µM) concentrations, which had no effect on proliferation. The treatment of the cells with this dose of ATP increased their motility by ∼23% after a 6-h treatment (Figure 6H). However, by 24 h, this effect had subsided (Figure 6H). Although adenosine showed a tendency to slow the early phase of scratch wound closure in three out of four experiments, the effect was not statistically significant (Figure 6H; after 6 h). After 24 h, no difference, when compared with the untreated cultures, were seen with either adenosine or AMP (Figure 6H, after 24 h).
Overall, our data indicate that extracellular ATP and its degradation products are potent regulators of keratinocyte HAS expression and hyaluronan synthesis as well as cell behavior, although with partially opposing effects. The signaling pathways most likely involved in the changed HAS expression are summarized in Figure 7.
Summary of ATP-induced signaling regulating hyaluronan synthesis in HaCaT keratinocytes.
The present work shows for the first time that extracellular adenosine nucleotides are potent regulators of HAS expression. Importantly, ATP and the less phosphorylated AMP and adenosine often showed opposite influences on HAS expression, hyaluronan synthesis, and keratinocyte functions, suggesting the existence of a regulatory loop based on ATP catabolism.
HAS expression and hyaluronan synthesis
ATP caused a rapid and very strong HAS2 up-regulation, which was followed by a decline 4–6 h after the ATP exposure, concomitantly with increased HAS1 and HAS3 expressions. AMP and adenosine also up-regulated HAS1 and HAS3 transcript levels; however, their influence on HAS2 was inhibitory. As hydrolysis of extracellular ATP is rapid in HaCaT , the down-regulation of HAS2 and the rise of HAS1 and HAS3 4–6 h after the ATP exposure may be caused by these degradation products.
ATP exposure was associated with a rapid increase in pericellular hyaluronan content and in the amount of hyaluronan released to the culture medium, in line with the observed changes in HAS mRNA expressions. However, the fold change in HAS2 mRNA with ATP was very high, actually higher than previously seen with potent stimuli such as growth factors or hormones [57–60], and compared with the HAS2 response the change in the hyaluronan content was relatively modest, suggesting that ATP may have caused other, perhaps posttranslational changes in HAS2 activity. It is also possible that the changes in the mRNA and protein levels do not correspond to each other under our experimental conditions, or that the residence and activity of the HAS2 protein at the plasma membrane does not reflect the true extent of the changes in its expression .
Indeed, we observed that ATP reduced the level of UDP-GlcNAc after ∼4 h. This could decrease hyaluronan synthesis through a lower precursor supply and by inhibiting the enzymatic activity due to reduced O-GlcNAcylation of HAS2 . Extracellular ATP and adenosine can also signal for AMPK activation , which can inhibit both the activity of HAS2  and of GFAT1, the key enzyme in the hexosamine pathway toward UDP-GlcNAc [53,64]. Indeed, supplementation of the cultures with glucosamine to enlarge the cellular pool of UDP-GlcNAc [52,65] significantly augmented hyaluronan synthesis in the ATP-treated cultures, suggesting that reduced precursor supply dampened the hyaluronan response with the ATP treatment.
Subjecting keratinocytes to either AMP or adenosine reduced the content of hyaluronan, consistently with their inhibitory effect on HAS2 expression, despite the simultaneous up-regulation of the HAS1 isoform. However, in this case, the fold changes may be deceptive, as the basal expression level of HAS1 is very low in keratinocytes, and the synthetic activity of this isoenzyme is modest [29,65].
Receptors and signals of extracellular adenine nucleotides
The concentrations of extracellular nucleotides are generally considered to be nano- or (sub)micromolar in quiescent cells and normal tissues [25,66,67], being thus considerably lower than those used here. However, extracellular ATP concentrations close to what we used in the present experiments (10 and 100 µM) are found in stimulated polymorphonuclear leukocytes , irritated keratinocytes , and wounded epithelial cells . Even higher concentrations of ATP occur under physio-pathological stress such as contact dermatitis and tumor microenvironments . Hypoxia increases extracellular adenosine levels: in the extracellular fluid of hypoxic brain  and in solid tumors , adenosine can reach micromolar concentrations.
The release of ATP after various insults often seems to be an immediate response, but may also depend on the trigger. Thus, after UVB  and UVA irradiation  as well as stretching , the release occurs within seconds/minutes. After chemical irritation, the release has been reported to occur within 1 h, while after nonmetal hapten exposure a longer period is required .
The degradation of ATP in HaCaT cell cultures has been reported to occur rapidly reaching a stable level of ∼50% during the first 40 min after the change of fresh medium (endogenous ATP), while a decline to ∼60% has been found 1 h after exogenously added ATP . Deduced from this more than half of the ATP is probably degraded within the 4-h time frame, when the inhibitory effect on HAS2 and the stimulatory effect on HAS1 after ATP was observed, allowing time for signaling of the less phosphorylated derivatives.
Considering the signaling pathways, keratinocytes are known to express multiple purinergic receptors both from the P2Y and P2X families [7,11,20]. Experiments with siRNAs and chemical inhibitors indicated that P2Y2 is the principal player in the ATP-induced HAS2 response, while P2Y1 and P2Y12/13 play minor roles. However, we cannot totally exclude contributions from the P2X family, also expressed in epidermal keratinocytes.
AMP does not have unique receptors, but it could use A1, one of the receptors for adenosine . However, inhibition of A1 failed to prevent the AMP-induced HAS2 down-regulation (unpublished observations), in line with its low expression in keratinocytes . Therefore, it is likely that AMP influences HAS expression via conversion to adenosine.
At least under normal conditions , the main adenosine receptor in keratinocytes is A2B [21,22], the activation of which inhibits keratinocyte proliferation and causes cell cycle arrest through increased intracellular calcium . A2A, another adenosine receptor, little expressed in resting keratinocytes, is up-regulated in psoriatic epidermis , and by cytokines such as IL-1β and TNF-α, indicating that adenosine signaling shifts to this receptor under inflammatory conditions. This leads to cell proliferation via p38 signaling . Since adenosine reduced cell numbers, the expression of the adenosine receptor A2B rather than A2A prevailed in the present experiments. Intriguingly, if the expression of A2A dominates, which can occur in stressed tissues  or malignancies , activation of p38 signaling could increase the expression of HAS2, as observed in keratinocytes [35,39], and actually lead to increased hyaluronan production.
However, adenosine can also signal via receptor-independent mechanisms such as through adenosine uptake, followed by conversion to AMP with its intracellular signaling functions , or by influencing membrane phosphatase activity . The delayed responses, including the up-regulation of HAS1 after 4–6 h and the secondary waves of HAS2 and HAS3 activation after 24 h, are likely processes secondary to the initial signals induced by the nucleotides. Interestingly, the induction seen in HAS1 with adenosine parallels the observations of Grandoch et al.  in human coronary artery smooth muscle cells, where HAS1 was most strongly up-regulated after a 6-h treatment with the adenosine analog 5′-N-ethylcarboxamidoadenosine (NECA).
Extracellular ATP inherently induces Ca2+ release and PKC signaling via the Gq/11-coupled P2Y receptors , activating MAP kinases, CaMKII, and CREB. The up-regulation of HAS2 by CREB is anticipated since it has functional binding sites at the proximal HAS2 promoter . Additionally, CaMKII, p38, and ERK regulate HAS expression in response to radiation stress and growth factors [35,55,81,82], and p38 and ERK respond to extracellular ATP in human keratinocytes [83,84].
The time frame of the phosphorylations in p38, ERK, and CREB, with maxima 15–30 min after starting the treatments, coincides well with the maximum induction seen in HAS2 ∼90–120 min. Despite the functional STAT3 response element on the HAS2 promoter , and distinct STAT3 phosphorylation at two sites after the ATP exposure, experiments with the STAT3 inhibitor indicated that HAS2 is not directly induced by STAT3. The modest increase in the phosphorylation level of Ser727 may not have been enough to activate HAS2 transcription. The more delayed Tyr705 phosphorylation suggests a separate signaling route. In line, our preliminary experiments suggested that AMP also induced STAT3 phosphorylation at Tyr705 but not at Ser727 (data not shown). Such a difference in STAT3 phosphorylations may significantly influence the final outcome .
Extracellular UTP, also released from stressed cells, causes an up-regulation of HAS2 similar to that induced by ATP . This is not surprising since they share P2Y2, the main P2 receptor involved in the HAS2 response. UTP and ATP also have many common signaling steps involved in the up-regulation of HAS2, including MAP kinases, CaMKII, and CREB . However, ATP and UTP have also divergent effects on signaling, most obviously in STAT3 activation and its involvement in HAS2 up-regulation . This is probably explained by the contribution of receptors specific for each nucleotide, including P2Y4 for UTP, P2Y1 and P2Y12/13 for ADP, and P2X for ATP. The difference in the signaling between ATP and UTP is signified by their divergent effects on HAS1 and HAS3 transcription, and the ability of AMP, but not UMP, to reduce HAS2 expression .
Keratinocyte proliferation and migration
In addition to acting in the maintenance of extracellular matrices, HAS expression and hyaluronan content have been reported to regulate cell functions including proliferation, migration, differentiation, and apoptosis . The inhibition of hyaluronan synthesis induced by adenosine in HaCaT cells was associated with a reduction in cell numbers, in line with previous reports on human keratinocytes [21,77]. Here, ATP did not clearly affect cell proliferation, which appears to depend on the experimental set-up [20,23,88].
However, ATP stimulated keratinocyte migration, and this coincided with the accumulation of hyaluronan after a 6-h ATP treatment. Such stimulation might be enough to keep keratinocytes activated and even repair minor epidermal damages; however, other factors are needed to mount a full wound healing response . Although exogenous ATP has been shown to stimulate cell migration in many cell types such as in corneal epithelial cells , its influence on keratinocyte motility has been controversial , perhaps due to differences in the experimental conditions used. Endogenously released ATP is more consistently stimulatory, regulating stretch-induced and basal migration , as well as enhancing epithelial cell motility and wound closure .
There is no previous information about the influence of adenosine on keratinocyte migration. However, it stimulates motility in inflammatory cells and in breast cancer cells via autocrine activation of the A3 receptor, but inhibits cell migration when added exogenously (, and references therein). The expression of this receptor is low in keratinocytes , perhaps explaining why no change in migration was found. Actually, there was a tendency toward reduced migration at the early phase, coincidentally with the inhibition of HAS2 expression. However, adenosine may still play a positive role in skin wound healing since NECA, its nonspecific analog, has been shown to counteract corticosteroid-suppressed wound healing in a mouse model .
Extracellular nucleotides and hyaluronan in epidermal biology
Extracellular nucleotides are potent local danger signals that affect skin homeostasis through altered gene expression and matrix remodeling, as previously demonstrated in dermal fibroblasts during wound healing . Especially, signaling via the P2Y2 receptor is implicated. The present data show that extracellular nucleotides regulate the synthesis of hyaluronan, a major component of the ECM in the epidermis , through the same receptor.
During normal tissue homeostasis, hyaluronan synthesis and degradation in the epidermis are active, but balanced . However, whenever this homeostasis is disturbed with insults such as wounding, barrier disruption, or UVB radiation, epidermal hyaluronan content is rapidly increased [33–35], even within the first hour after the trauma , together with other immediate defense reactions in the tissue. The strong but transient surge of HAS2 expression is obviously intended for a rapid response that can be augmented by other activators such as growth factors, hormones, or cytokines, if the noxious condition prevails [98–100]. However, extracellular nucleotides may also play a role in the prolonged hyaluronan response because of the more widespread and/or increased expression of the P2Y2 receptor induced by, e.g. wounding and UVB [101,102].
As changes in HAS expression and protein kinase activation elicited by UVB resemble those caused by the ATP exposure [35,103], they may indeed be related to nucleotide signaling . However, while we have observed that the initial influences of UVB on rat keratinocyte Has2 and Has1  resemble those of adenosine on HaCaT, UVB exposure causes Has2 to peak bimodally at 8 and 36 h in a p38-dependent manner, perhaps due to a secondary release of ATP induced by the initial surge of extracellular nucleotides  or altered expression of the P2Y2 receptor [102,105,106].
Functionally, the hyaluronan accumulation seen after ATP treatment may protect cells from apoptosis, as shown in UV-stressed dermal fibroblasts  and corneal epithelial cells . Hyaluronan acting together with its receptor CD44 supports cell survival via AKT signaling, and CD44 may prevent apoptosis via counteracting ROS activity (reviewed in ref. ). Interestingly, the increased expression of CD44 seen after epidermal insults closely correlates with hyaluronan accumulation , and this pattern was also seen here after the ATP exposure.
Our data showing the transient nature of the ATP-induced HAS up-regulation is physiologically important, constituting a ‘state of alert’, which can be either amplified by other factors, or shut down by the ATP degradation products AMP and adenosine in the absence of additional signals for tissue damage. On the other hand, adenosine under basal conditions tends to subdue hyaluronan production to maintain epidermal tissue homeostasis.
biotinylated hyaluronan-binding complex
Ca2+/calmodulin-dependent protein kinase II
calcium response element-binding protein
hyaluronan enzyme-linked sorbent assay
mitogen-activated protein kinase
- P1, P2X, P2Y
purinergic (nucleotide and nucleoside) receptors
protein kinase C
signal transducer and activator of transcription
uridine diphosphate N-acetyl glucosamine
ultraviolet B radiation
L.R. planned and performed part of the experiments, analyzed data, contributed to paper writing, and prepared the figures. T.J. conceived the study, analyzed data, and contributed to paper writing. R.H.T. co-ordinated the study, analyzed results, and participated in the writing of the paper. R.K. provided technical expertize for most of the experiments. G.B. helped conceptualize the study, presented the data, and revised the text. S.O., P.T., and S.P.-S. performed and analyzed part of the experiments and critically revised the manuscript. M.I.T. took part in the study conception and critically revised the manuscript. All authors approved the manuscript.
The present study was financially supported by grants from Sigrid Juselius Foundation (R.H.T., M.I.T., and S.P.-S.), the Special Government Funding of Kuopio University Hospital (M.I.T.), The Spearhead Funds of the University of Eastern Finland/Cancer Center of Eastern Finland (M.I.T. and R.H.T.), and The Cancer Foundation of Northern Savo (L.R.).
We are grateful to Silja Pyysalo, BSc, for helping with the laboratory analyses.
The Authors declare that there are no competing interests associated with the manuscript.
An earlier version of this manuscript has been included in the doctoral thesis work of Leena Rauhala, PhD, and published (print only) in the Publications of the University of Eastern Finland – Dissertations in Health Sciences (2017).