|Home | About | Journals | Submit | Contact Us | Français|
Darier’s Disease (DD), caused by mutations in the endoplasmic reticulum (ER) Ca2+ ATPase ATP2A2 (SERCA2b), is a skin disease that exhibits impaired epidermal cell-to-cell adhesion and altered differentiation. Although previous studies have shown that keratinocyte Ca2+ sequestration and fluxes are controlled by sphingolipid signaling, the role of this signaling pathway in DD previously has not been investigated. We show here that sphingosine levels increase and sphingosine kinase (SPHK1) expression decreases after inactivating SERCA2b with the specific SERCA2 inhibitors thapsigargin (TG) or siRNA to SERCA2b. Conversely, inhibiting sphingosine lyase rescues the defects in keratinocyte differentiation, E-cadherin localization, Desmoplakin (DP) translocation, and ER Ca2+ sequestration seen in TG-treated keratinocytes. To our knowledge, it was previously unreported that the keratinocyte sphingolipid and Ca2+ signaling pathways intersect in ATP2A2- controlled ER Ca2+ sequestration, E-cadherin and desmoplakin localization and Ca2+ - controlled differentiation, and thus may be important mediators in DD.
Darier’s disease (DD) (Darier, 2000,White, 1889) is an inherited skin disease characterized by keratotic papules and plaques found mostly in skin flexures and on the chest, upper back, face and scalp. Histopathology of DD skin reveals impaired differentiation, acantholysis and pathologic apoptosis. Patients with DD also have been reported to suffer from neuropsychiatric disease (Gordon-Smith, et al., 2010). Mutations in the endoplasmic reticulum (ER) Ca2+ ATPase (gene ATP2A2; protein SERCA2b), which sequesters Ca2+ in the ER, cause DD (Sakuntabhai, et al., 1999a, Sakuntabhai, et al., 1999b). The disease is thought to be caused by haploinsufficiency (Foggia, et al., 2006). Treatment for DD largely has been empiric, and is only partially effective in some patients, although a variety of pharmacologic and physical agents have been proposed (Kragballe, 1995; Reuter, et al., 2007; Stewart and Yell, 2008; Cooper and Burge, 2003).
Ca2+ signaling is implicated in keratinocyte and epidermal differentiation, cell-to-cell adhesion and apoptosis. Normal Ca2+ signaling requires both IP3-mediated ER and Golgi Ca2+ release and Ca2+ influx though Ca2+ -permeable ion channels, leading to transcription of differentiation-specific genes (Bikle, et al., 2001) and stimulating formation of adherens junctions containing E-cadherin (Green, et al., 2010). Well known as a structural protein, E-cadherin also recently has been shown to have important signaling functions, as two kinases contained within the E-cadherin complex (Xie and Bikle, 2007; Xie, et al., 2009) maintain the prolonged Ca2+ increase essential for calcium induced differentiation.
Sphingolipid metabolism (fig 1A) constitutes a complementary Ca2+ signaling system in keratinocytes. Heightened levels of intracellular Sphingosine-1-phosphate (S1P) increase cytosolic Ca2+ levels by direct mobilization of Ca2+ from Thapsigargin (TG) sensitive stores (Meyer zu Heringdorf, et al., 2003), while increases in extracellular S1P concentrations activate Store Operated Calcium Entry (SOCE), through activation of G-protein coupled receptors (Mehta, et al., 2005). In keratinocytes, increased intracellular S1P enhances keratinocyte growth and differentiation, and blocks apoptosis (Hong, et al., 2008; Lichte, et al., 2008). Activating sphingosine kinase (SPHK1) increases intracellular S1P; thus, agents that activate this enzyme enhance expression of the differentiation markers filaggrin and involucrin in HaCAT keratinocytes (Hong, et al., 2008). In contrast, blocking SPHK1 increases apoptosis in UVB-treated keratinocytes (Uchida, et al., 2010). S1P levels also can be increased by preventing S1P breakdown (Figure 1A) via Sphingosine phosphate lyase (SGPL1) (Reviewed by Kumar and Saba, 2009).
We show here that the sphingolipid and Ca2+ signaling pathways intersect in controlling SERCA2- ER Ca2+ sequestration. SERCA2b inhibition decreases SPHK1 expression and reproduces the defects in cell growth, adhesion and differentiation seen in DD. In contrast, inhibiting SGPL1 compensates for decreased SPHK1 and rescues or ameliorates defects in differentiation, E-cadherin processing, desmoplakin translocation, and ER Ca2+ sequestration. Currently, SGPL1 is being targeted as possible therapeutic target for rheumatologic disease and multiple sclerosis. These therapeutic agents could also prove valuable as therapy for DD.
Increased sphingosine can enhance cell death/cell cycle arrest, previously reported in a canine model of DD (Muller, et al., 2006), while increased levels of S1P promote cell growth and viability. The relative amounts of these lipids are determined by the activity of SPHK1, which metabolizes sphingosine to S1P; and SGPL1, which degrades S1P to phosphoethanolamine and hexadecanol (Fig. 1A). Because abnormal sphingolipid signaling and DD share impaired cell differentiation, cell survival and calcium signaling, we first examined whether pharmacologically inactivating the ER Ca2+ ATPase SERCA2b could alter sphingolipid metabolism. We found that low concentrations of TG (100nM) increased sphingosine concentrations (Fig. 1B). We were able to detect decreased sphingosine kinase mRNA expression after treatment with even lower TG concentrations (10-100nM) (Fig. 1C). A similar decrease in SPHK1 expression was seen after inactivating SERCA2b with siRNA (Fig. 1C). Both siRNA and TG were used to impair SERCA2 function in subsequent experiments. The dose of SERCA2b siRNA used for these experiments was chosen to reproduce the average 50% reduction in SERCA2b expression seen in DD patients (Foggia, et al., 2006) (figure S1). In the same paper, the authors report an average basal level of ER calcium for DD cells that is 60% lower than control as observed by transient cytosolic calcium increase after TG exposure. We observed that a similar decrease in ER calcium was obtained by treating cells with 10nM TG (Celli, et al., 2010, and figure 7c) and we therefore used this concentration for the functional studies presented in this report.
The ER Ca2+ store is important for normal keratinocyte signaling and differentiation (Callewaert, et al., 2003; Foggia, et al., 2006). Previous experiments demonstrated that early differentiation protein expression is inhibited by disabling the ER Ca2+ ATPase (Li, et al., 1995b; Li, et al., 1995a). Early differentiation proteins also are decreased in SERCA2 knockout mouse keratinocytes (Hong, et al., 2010), and defects in E-cadherin localization have been reported in human DD (Tada and Hashimoto, 1998). Decreased E-cadherin levels and/or trafficking may directly lead to decreased cell-to-cell adhesion via impaired formation of adherens junctions and may additionally impair IP3 mediated keratinocyte Ca2+ signaling because the E-cadherin complex formation in the membrane is essential for PLCγ1 activity (Xie and Bikle, 2007; Xie, et al., 2009). We therefore assessed whether modulating the sphingolipid pathway could reproduce or rescue defects in keratinocyte differentiation and E-cadherin expression or localization. We observed abnormal keratinocyte morphology and growth 24 hours (data not shown) and 48 hours after TG treatment (Fig. 2D). Inhibition of SGPL1 with siRNA rescued these defects (Fig 2F). Western blotting showed that SERCA2 inhibition with TG also led to defects in involucrin protein synthesis and E-cadherin protein synthesis and processing. These defects could be normalized or ameliorated by inhibiting SGPL1 synthesis (Fig. 3). We found that SGPL1 inhibition was more effective in normalizing the defects in involucrin and E-cadherin expression caused by lower doses of TG (10 nM) than after higher (100nM) TG (data not shown), likely because the effects of SERCA2 inhibition became irreversible at higher doses. Treatment with exogenous S1P had no effect (Fig. S3). Vitamin D and cannabinoid analogues also were ineffective (Fig. S3).
The additional, higher molecular weight band seen on immunoblots after TG treatment suggested that E-cadherin intracellular processing and localization might also be disrupted when ER Ca2+ stores were depleted, similar to what has previously been reported for DD (Hakuno, et al., 2000). To test this, we localized E-cadherin first using an antibody that selectively binds to the extracellular epitope (Fig. 4A-D). E-cadherin normally responds to raised extracellular Ca2+ by forming continuous, fine intercellular strands (Fig. 4A). TG treatment lead to a dose-dependent disruption in these cell-to-cell contacts, with E-cadherin localization becoming coarse and irregular at very low (10nM) TG concentrations (Fig. 4B), and discontinuous at 100 nM TG concentrations (data not shown). SGPL1 inhibition normalized E-cadherin localization at 10 nM TG (Fig. 2D), but only partially rescued TG-induced E-cadherin disruption at the 100 nM dose (not shown). Treatment with scrambled siRNA had no significant effect (Fig. 4C). These data suggest that E-cadherin is not properly localized to the plasma membrane after SERCA2 inhibition.
To test whether SERCA2 inhibition lead to defective E-cadherin, we next localized E-cadherin using an antibody that binds to an intracellular E-cadherin epitope. After raising extracellular Ca2+, the vast majority of E-cadherin was found at cell-to-cell junctions in normal keratinocytes (Fig. 4E) with only punctuate foci of intracellular E-cadherin seen. After treatment with 10 nM TG, extracellular E-cadherin again appeared irregular and beaded (Fig. 4F). In addition, perinuclear aggregates of E-cadherin also were seen (Fig. 4F), suggesting that ER Ca2+ depletion leads to intracellular E-cadherin accumulation. This process also was dose-dependent, and the vast majority of E-cadherin was localized intracellularly after treatment with 100 nM TG (not shown). As above, inhibition of SGPL1 was able to normalize E-cadherin localization in keratinocytes treated with 10 nM TG (Fig.4H), but only partially rescued keratinocytes treated with the higher 100 nM TG (not shown). Treatment with scrambled siRNA did not rescue the defects caused by TG treatment (Fig. 4G).
Downregulation of SERCA2b expression by siRNA resulted in a similar pattern of intracellular and extracellular E-cadherin staining (Fig. 5D and J respectively) that could be normalized by SGPL1 siRNA treatment (Fig. 5E and K). Treatment with a scrambled siRNA sequence had no effect (Fig. 5F and L).
These studies demonstrate that E-cadherin requires normal ER Ca2+ stores to properly localize to the plasma membrane, and that the defects seen in cell-to-cell adhesion in DD may be due to defective E-cadherin localization.
To further investigate the effects of SERCA2 inactivation on cell to cell junctions, we assessed the effect of 10 nM TG on desmoplakin (DP) localization, which also is disrupted following SERCA2 inactivation (Hobbs, et al., 2011). In agreement with this previous report, we also observed DP localization at the cell to cell borders within 3 hours after raising extracellular Ca2+ in control cells (Fig. 6A-C). 48 hours after raising extracellular Ca2+, DP was almost entirely localized to cell-to-cell borders (Fig. 6G-I). Like E-cadherin, TG disrupted DP localization in a dose-dependent manner. Cells treated with 10 nM TG are shown in figure 6D and J, and cells treated with 100 nM TG are shown for comparison in Fig. S5. DP staining was mostly intracellular in TG-treated cells. Defects in DP staining and gaps between cells were noted at both the 3 and 48 hours time points (Fig. 6D and J respectively). Treatment with SGPL1 siRNA completely rescued DP localization 48 hours after calcium switch, but only partially at the earlier 3 hour time point (Fig. 6K and E respectively). Treatment with scrambled siRNA did not rescue the abnormal DP localization (figure 6C, F, I, L). SGPL1 siRNA also partially rescued DP localization in cells treated with 100nM TG (Fig. S5).
We used the ER-targeted Ca2+ sensor D1ER to monitor ER Ca2+ depletion after SERCA2 inhibition, and to assess whether SGPL1 inhibition could ameliorate this loss. Figure 7A and figure 7B present the data as average FRET ratios and average calcium concentrations respectively. A decrease in FRET ratio indicates lower Ca2+ concentrations. FRET ratios were converted into calcium concentration following the protocol in (Rudolf, et al., 2006). siRNA transfection alone, with either a scrambled sequence or SGPL1 siRNA, leads to a small decrease in ER calcium levels. In agreement with a previous report (Celli, et al., 2010), treatment with 10nM thapsigargin caused a dramatic drop in ER calcium concentration in both untreated and scrambled siRNA transfected samples. However, treatment with 10nM TG had no significant effect on ER Ca2+ in cells pretreated with SGPL1 siRNA (Fig. 7A and B). TG treatment produced more than a 50% drop in ER Ca2+ concentration in both untreated keratinocytes and scrambled siRNA-transfected keratinocytes (Fig. 7C). In contrast, SGPL1 inhibition prevented this TG- induced Ca2+ loss (Fig. 7C). These data suggest that modulating sphingosine metabolism may normalize TG-induced abnormalities in keratinocyte differentiation and cell-to-cell adhesion by mitigating ER Ca2+ loss after SERCA2 inactivation. 100 nM TG almost completely emptied ER Ca2+ stores and could not be rescued by SGPL1 inhibition (data not shown). SGPL1 inhibition did not change SERCA2 protein levels or SPHK1 mRNA expression, suggesting that Ca2+ was normalized by a mechanism independent of SERCA2 or SPHK1 synthesis (Fig. S1 and S2).
This report demonstrates that inhibiting SGPL1 rescues ER Ca2+ depletion-induced defects in keratinocyte differentiation and adhesion, and restores ER Ca2+ stores depleted by SERCA2 inhibition. How sphingosine metabolism and SERCA2 Ca2+ signaling interact in directing keratinocyte adhesion and differentiation is not yet well defined. Certainly, the actions of exogenous S1P in controlling intracellular Ca2+ have been studied extensively. However, we found that exogenous S1P was ineffective in rescuing adhesion and differentiation defects caused by SERCA2b inactivation (Fig. S3). These data suggest that generation of intracellular S1P is essential for rescuing the DD phenotype. Alternatively, extracellular S1P-stimulated Ca2+ release from the ER may be blunted when these stores already are depleted due to SERCA2 dysfunction. Exposure to increased intracellular S1P levels, on the other hand, has been shown to increase calcium levels in TG sensitive stores and to mobilize calcium from these stores into the cytosol, independently of the plasma membrane G-protein coupled S1P receptor (Claas, et al., 2010; Meyer zu Heringdorf, et al., 2003).
In a recent paper, Calloway et al (Calloway, et al., 2009) found that treatment of RBL mast cells with sphingosine and its positively charged analogues inhibited the Ca2+ influx via SOCE usually seen after TG treatment, by flipping to the inner leaflet of the plasma membrane and preventing Ora1/CRACM1 from interacting with STIM1. Thus, SERCA2b inactivation might cause not only ER Ca2+ depletion, but also lead to increased intracellular sphingosine, via SPHK1 inhibition. The increased sphingosine then might interfere with Ora/CRACM1/STIM interaction, exacerbating the low ER Ca2+ caused by SERCA2b loss.
We chose to study the effects of ER calcium depletion and SGPL1 down-regulation on E-cadherin because this protein has both a structural function and a role in calcium signaling (Xie and Bikle, 2007; Xie, et al., 2009). Moreover, calcium dependent E-cadherin aggregation at cell-to-cell adhesion points is known to be perturbed in DD (Tada and Hashimoto, 1998). In contrast to a previous study (Hobbs, et al., 2011) our experiments demonstrated that E-cadherin localization was disrupted after SERCA2 inhibition. This discrepancy may be explained by the longer timecourse (24–48 hours vs. 3 hours) used in our studies, based on previous findings that E-cadherin continues to aggregate and complex at cell-to-cell junctions for at least 7–20 hours (Vasioukhin, et al., 2000). Immunofluorescence micrographs of SERCA2 siRNA treated cells harvested 3 hours after calcium switch do not show significant defects in E-cadherin localization (Fig. S4) as compared to the longer time points, thus lending support to this hypothesis.
SGPL1 inhibition also normalized desmoplakin (DP) trafficking, which has been implicated in DD (Dhitavat, et al., 2003; Hobbs, et al., 2011). These studies suggest that modulating the sphingolipid pathway may ameliorate the defective cell-to-cell adhesion that is a hallmark of DD.
We studied cells treated with thapsigargin or siRNA against SERCA2 in lieu of DD keratinocytes, as previous studies conducted on DD keratinocytes revealed a wide heterogeneity in calcium mobilization in these cells (Foggia, et al., 2006). While further investigation on the effects of SGPL1 inhibition in DD cells is needed, we show here that reversing SERCA2 inhibition caused by thapsigargin or SERCA2 siRNA ameliorates the junctional and differentiation defects that are known to underlie DD pathology.
Normal human keratinocytes, passage 2-3, were plated in 0.03 mM Ca+2 154 CF media (Cascade, Life Sciences, NY). For experiments where TG was used, cells were treated with either scrambled siRNA (as a negative control) or siRNA against SGPL1 (Hs_SGPL1_2 Flexitube, Quiagen, Valencia, CA) when 80% confluent. 8 hours after siRNA treatment, thapsigargin at a final concentration of 10 or 100 nM, was added to some of the cells for 2 hours. The media was then replaced with 1.2 mM Ca+2 154 CF (Cascade, Life Sciences, NY), to induce differentiation. For experiments where siRNA against SERCA2 (silencing sequence: AAGCAGGACATCAATGAGCAA) was used, cells were transfected with siRNA against SERCA2, or a scrambled siRNA sequence for control, when cells were 60% confluent and treated with SGPL1 siRNA, or with a scrambled siRNA for control, 24 hours after. Hyperfect transfection reagent (Quiagen, Valencia, CA) was used for all siRNA transfections. Cells were harvested at time points noted in each figure after thapsigargin treatment. Vehicle (for TG) or scrambled siRNA were used as controls.
60% confluent cells were transfected with D1ER (gift of Prof. Tsien’s laboratory) using Mirrus Keratinocyte Transit lipid reagent, according to the manufacturer’s protocol. 24 hours after transfection, cells were treated with SGPL1 siRNA or scrambled siRNA and TG as described above. Dual channel fluorescence measurements were performed 24 hours later using a Zeiss LSM Meta confocal system (Carl Zeiss Inc., NY). Calibration of the Calcium sensitivity range of the probe and conversion of the FRET ratio into Calcium concentrations was performed following the protocol in (Palmer and Tsien, 2006; Rudolf, et al., 2006).
Total lipids were extracted from keratinocytes and then sphingoid bases were converted to o-phthalaldehyde derivatives after dephosphorylation with alkaline phosphatase as previously reported (Min, et al., 2002). Sphingoid base o-phthalaldehyde derivatives were separated on C18 reverse phase HPLC column (Luna C18(2), 250 x 4.6 mm, 5μm, Phenomenex, Torrance, CA) using methanol:water (9:1) (v/v) with a 1 ml/min flow. Fluorescence intensity was monitored at Ex 360 nm and Em 430 nm. The sphingoid base content was reported as pmol per mg of protein.
Normal human keratinocytes were treated as described above (see Cell Culture). Samples were formaldehyde-fixed. Cells were stained with an antibody that stained both intracellular and extracellular E-cadherin (36/E-Cadherin BD Transduction Laboratories, Sparks, MD), or only the extracellular epitope of E-cadherin (HECD-1, abcam, Cambridge, MA). For desmoplakin immunofluorescence we used a mouse monoclonal antibody (1G4 gift of Dr. Kathy Green).
Photomicrographs shown in figure 4 were taken on a Zeiss 510 Meta confocal microscope (Carl Zeiss Inc. NY). Identical laser power and detector gain levels were used for all images. Photomicrographs in figures 5 and and66 were taken using a Zeiss observer/spinning disk confocal system (Carl Zeiss Inc. NY). Identical laser power and detector gain levels were used for all images.
Cells were homogenized, and proteins were isolated using RIPA buffer (Sigma-Aldrich, Saint Louis, MO) containing protease inhibitors (Roche Applied Science, Indianapolis, IN). Protein quantification for equal loading was made with Thermo Scientific Pierce’s BCA assay kit. SDS-PAGE and transfer was performed using NuPAGE Novex 4-12% sodium dodecyl sulfate-polyacrylamide gels and nitrocellulose membranes according to Invitrogen’s NuPAGE protocol. Membranes were blocked in PBS with 5% non-fat dry milk and 0.05% Tween-20 and incubated with the following primary antibodies and dilutions: anti-Involucrin (Sigma Aldrich, Saint Louis, MO) at 1:100, anti-Ecadherin (Invitrogen, Life Science, NY) at 1:500, and anti-β-Actin-HRP (Sigma-Aldrich, Saint Louis, MO) at 1:30,000 anti-SERCA2. Chemiluminescent detection was performed with ECL Plus detection reagent (Thermo Scientific Pierce, Waltham, MA) using the Fujifilm LAS-3000 imaging system.
All cDNA was generated using Roche Applied Science’s Transcriptor first strand cDNA synthesis kit. Quantitative PCR (QPCR) was performed using TaqMan fast universal PCR master mix and the ABI 7900HT real-time PCR machine (Applied Biosystems, Life sciences, NY) and target genes were normalized to GAPDH (primers and dual-labeled probes from Applied Biosystems). Human SPHK1 primers and probes used: CTGTCTGCTCCGAGGACTG forward, CGTGGTTCTTACACCACTGC reverse, TGACCTGCTACCTCGGCCGC probe.
Data were analyzed using the paired, two-tailed Students T test and one-way ANOVA.
This work was supported by National Institutes of Health grants AR051930 and the San Francisco Veterans Administration Medical Service.
Conflict of interest: The authors have no conflict of interest.