The EP₁ subtype of Prostaglandin E₂ Receptor: Role in Keratinocyte Differentiation and Expression in Non-Melanoma Skin Cancer

SUMMARY

We have previously demonstrated that the EP₁ subtype of PGE₂ receptor is expressed in the differentiated compartment of normal human epidermis and is coupled to intracellular calcium mobilization. We therefore hypothesized that the EP₁ receptor is coupled to keratinocyte differentiation. In in vitro studies, radioligand binding, RT-PCR, immunoblot and receptor agonist-induced second messenger studies demonstrate that the EP₁ receptor is up-regulated by high cell density in human keratinocytes and this up-regulation precedes corneocyte formation. Moreover, two different EP₁ receptor antagonists, SC51322 and AH6809, both inhibited corneocyte formation. SC51322 also inhibited the induction of differentiation-specific proteins, cytokeratin K10 and epidermal transglutaminase. We next examined the immunolocalization of the EP₁ receptor in non-melanoma skin cancer in humans. Well differentiated SCCs exhibited significantly greater membrane staining, while spindle cell carcinomas and BCCs had significantly decreased membrane staining compared with normal epidermis. This data supports a role for the EP₁ receptor in regulating keratinocyte differentiation.

INTRODUCTION

The ability of keratinocytes to undergo differentiation and form detergent-insoluble squamous cells or corneocytes is critical to maintaining the skin's permeability barrier. This permeability barrier is essential for the normal functioning of the body's largest organ, as disruption of this barrier results in fluid loss and increased susceptibility to environmental and microbial insults. Moreover, escape from differentiation-induced growth arrest is a hallmark of non-melanoma skin cancer (NMSC). However, NMSC exhibits striking differences in the “squamous” cell phenotype, with basal cell carcinomas (BCC) recapitulating the phenotype of the undifferentiated basal cell compartment and well differentiated squamous cell carcinoma (SCC) exhibiting the full spectrum of differentiation-associated cellular changes. Moreover, the degree to which SCCs retain the differentiated phenotype often inversely correlates with the aggressiveness of the tumor, with poorly differentiated SCCs and spindle cell carcinomas exhibiting a more aggressive course. However, the mechanisms that regulate epidermal differentiation are poorly understood.

Prostaglandins (PG) are formed sequentially by cleavage of arachidonic acid (AA) from cellular phospholipids, conversion of AA to PGH₂ by one of two cyclooxygenases (COX-1 and COX-2), and finally metabolism of PGH₂ to the major prostaglandin species by specific PG synthases [1]. The major prostaglandin species released by epidermal keratinocytes is PGE₂ [2]. PGE₂ acts by binding to one of four heterotrimeric G-protein coupled receptors, termed E-series prostaglandin receptors (EP₁-EP₄)(reviewed in [3, 4]). These receptors differ in their G protein alpha subunit binding specificity and the second messenger pathways that are activated upon ligand binding. In addition, the four receptor subtypes also exhibit differences in PGE₂ binding affinities. The EP₃ and EP₄ exhibit binding affinities for PGE₂ in the subnanomolar range, while the EP₁ and EP₂ receptors are lower affinity PGE₂ receptors, with binding affinities of 9.1 and 4.9 nM, respectively [5].

Several studies indicate that cyclooxygenase products are involved in regulating keratinocyte differentiation. Alterations in epidermal differentiation have been described in COX-1 and COX-2 knockout mice, as well as with transgenic mice overexpressing COX-2 in the epidermis [6, 7]. Moreover, in primary human keratinocytes in vitro, inhibition of cyclooxygenase activity has also been shown to inhibit calcium-dependent cornified envelope production [8]. Significantly, exogenous PGE₂, but not prostacyclin (PGI₂), was able to restore normal calcium-induced keratinocyte differentiation [8]. While the specific receptor subtype(s) mediating PGE₂-induced differentiation are unknown, the concentration of PGE₂ which restored calcium-induced differentiation was 100 nM, suggesting activation of low-affinity PGE₂ receptors [8]. The idea that high levels of PGE₂ are necessary for keratinocyte differentiation is also supported by the observation that COX-2 expression is up-regulated in the differentiated, suprabasalar compartment of normal human epidermis by immunohistochemistry [9]. In addition, COX-2 expression, but not COX-1 expression, is markedly up-regulated by high extracellular calcium concentrations in cultured human keratinocytes [9].

We have previously shown that the EP₂ receptor, and possibly the EP₄ receptor, act to stimulate growth of primary human keratinocytes via production of cyclic AMP [10]. Moreover, loss of EP₂ receptor expression may play a role in the invasive behavior in SCC [11, 12]. In contrast, EP₃ receptor activation stimulates diacylglycerol and ceramide production and results in keratinocyte growth inhibition [13]. However, the role of the EP₁ receptor in keratinocyte biology is unknown. Several clues to the function of the EP₁ receptor in keratinocytes can be inferred by its localization within the epidermis as well as the intracellular signaling pathway that is stimulated by EP₁ receptor activation [14]. In this previous study, immunohistochemistry demonstrated that EP₁ receptor expression was seen throughout the epidermis, but was markedly increased in the granular layer. In another study, it was noted that the EP₁ receptor was highly expressed in human squamous cell carcinoma and actinic keratoses, but was weakly expressed or absent in BCCs [15]. Moreover, we have previously demonstrated that normal human keratinocytes express the EP₁ receptor and that this receptor is coupled to intracellular calcium mobilization [14]. Since calcium is a potent inducer of keratinocyte differentiation, this suggests a potential role for the EP₁ receptor in regulating keratinocyte differentiation. In this current study, we examine how the EP₁ receptor is regulated during calcium and density-dependent induction of human keratinocyte differentiation in vitro. We then examine the role that the EP₁ receptor plays in regulating the differentiated phenotype in vitro. Finally, we examine EP₁ receptor expression by immunohistochemistry in pre-malignant human actinic keratoses (AKs), well differentiated SCCs, poorly differentiated SCCs (PD-SCCs), and spindle cell carcinomas and BCCs.

MATERIALS & METHODS

Materials

PGE₂ receptor agonists, PGE₂, and rabbit polyclonal antibodies against the human EP₁ receptor were purchased from Cayman Chemical (Ann Arbor, MI). Peroxoblock, CAS block, and Picture Plus broad spectrum immunohistochemical staining kits were purchased from Zymed Laboratories (South San Francisco, CA). High pH antigen retrieval solution was purchased from Dako (Carpinteria, CA). Isolated membrane preparations from human embryonic kidney cells (HEK-293) stably expressing the human EP₁ and EP_3A1 receptors were the generous gift of Dr. Kathleen Metters, (Merck-Frosst Centre for Therapeutic Research; Quenbec, Canada) [16]. The stable platelet activating factor (PAF) ligand, carbamyl-PAF was kindly supplied by Dr. Jeffrey Travers (Indiana University, Indianapolis, IN).

Isolation and culture of primary human keratinocytes

Adult primary human keratinocytes were prepared from discarded epidermis obtained from reductive mammoplasties and panniculectomies as previously described [10]. Cells were cultured on tissue culture plasticware precoated with collagen (Vitrogen 100, Collagen Corp., Palo Alto, CA). Cells were grown in either Dulbecco minimal Eagle's medium (DMEM) containing 10% fetal bovine serum and 15 mM HEPES buffer (FBS-DMEM), or in low calcium, serum-free medium (Keratinocyte-SFM (K-SFM), Invitrogen Life Technologies) containing 0.06 mM calcium chloride. Media were supplemented with 40 IU per ml penicillin, 40 μg per ml streptomycin, and 0.1 μg per ml amphotericin B. The cells were cultured in 95% air, 5% CO₂ at 37°C. All studies with human skin have been approved by the Indiana University-Purdue University at Indianapolis and the University of Rochester Institutional Review Boards.

[³H]-PGE₂ competitive binding

Freshly isolated primary human keratinocytes were cultured in FBS-DMEM on 12-well plates. To assess specific [³H]-PGE₂ binding activity, PHK's were plated to reach 80% confluence after an overnight incubation. Both specific PGE₂ binding and cornified envelope quantitation (see below) was performed daily once cells had reached > 90% confluence, or one day prior to reaching 100% confluence (Day-1). On each day, starting at day -1 confluence and continuing until 2 days post-confluence (Day +2), duplicate wells were washed twice with ice-cold HEPES buffered saline (HBS): 15 mM HEPES, pH 7.4 containing 140 mM NaCl, 4.7 mM KCl, 2.2 mM CaCl₂, 1.2 mM MgCl₂, 1.2 mM KH₂PO₄, and 11 mM glucose. After washing, 200 μl of cold HBS containing 3 nM of [³H]-PGE₂ (154 Ci/mmol, Perkin Elmer Life Sciences, Boston, MA) was added in the presence and absence of 1000-fold molar excess of PGE₂ or specific receptor agonists. After incubating with rocking at 4°C for 1 hour, the wells were washed four times with 2 ml cold HBS and the cells trypsinized and radioactive incorporation quantitated by liquid scintillation counting. On the same day, separate wells were processed for cornified envelope formation as detailed below.

Preparation of total cell lysates from primary human keratinocyte cultures

PHK's were grown to the desired cell densities in serum-free media containing 0.06 mM calcium, bovine pituitary extract, and epidermal growth factor (K-SFM, Invitrogen, Carlsbad, California). Pre-confluent cultures were established under low power microscopic exam as monolayer cultures occupying less than 60 % of the available surface area at the time of total cell lysate preparation. Confluent cultures were defined as cultures in which the cell monolayer covered 100% of the tissue culture surface. Total cell lysates were then prepared as previously described [17]. Total protein was determined prior to addition of reducing agent and tracking dye using the D_C Protein Assay (Bio-Rad Laboratories, Hercules, CA).

Western blot analysis

Western blot analysis of EP₁ and EP₂ receptor expression was done using rabbit polyclonal anti-EP₁ and anti-EP₂ receptor antibodies (Cayman Chemical) on heat denatured (EP₂) preparations or non-heat denatured (EP₁) preparations as previously described [14]. EP₁ immunoblots were performed using total cell lysates prepared as described above. In contrast, EP₂ immunoblots were performed using cellular membranes prepared as previously described [11].

Calcium mobilization studies

PHKs were grown to low cell density, <60% confluence (pre-confluent), or high cell density, 100% confluent (post-confluent). Prior to Fura PE3/AM loading, confluent monolayers were pre-treated for 48 hours with indomethacin (3 μg per mL) to block endogenous PGE₂ production. Cells were then loaded with 5 μM Fura PE3/AM-ester (Fisher Scientific, Hanover Park, Illinois) in growth supplement-free K-SFM containing 3 μg per mL indomethacin, 2.5mM probenecid (Sigma, St. Louis, Missouri), and 1 mg/ml Bovine Serum Albumin (BSA) for 1 hour at 37°C. After loading, the monolayer was washed 2 times with Hank's balanced salt solution (HBSS) containing 1.26 mM calcium (Sigma, St. Louis, Missouri), 2.5 mM Probenecid, and 1 mg/ml BSA balanced to pH 7.6. The cells were then detached by trypsinization, centrifuged, and resuspended in the HBSS solution, pH 7.6. Resuspended cells were incubated at room temperature for 30 min. For intracellular Ca²⁺ measurements, 2.5 mL of cells in the HBSS solution were loaded into a standard cuvette containing a magnetic stirring device. The cuvette was placed into a thermostatically controlled magnetic stirring chamber at 37°C in a Hitachi F-2000 fluorescence spectrophotometer. After establishing a baseline calcium level, agonists or vehicle (ethanol) was injected into the cuvette using an injection port. For a positive control, carbamyl-PAF (16 μg per mL) was added. Excitation wavelengths were 340 nm and 380 nm, and emitted light measured at 510 nm. Ratiometric recordings were produced using F-2000 software (Hitachi Instruments, Naperville, Illinois).

cAMP production

For pre-confluent versus post-confluent PGE₂-induced cAMP, freshly isolated PHKs were grown in 6 or 12-well plates in FBS-DMEM. Pre-confluent cells were 50-70% confluent while post-confluent cultures were 1-2 days post-confluent. The cells were pretreated with 3 μg/ml indomethacin overnight to block endogenous PGE₂ formation. The cells were then stimulated with 100 nM PGE₂ for 1 minute in media pre-equilibrated to 37°C in the absence of phosphodiesterase inhibitors to measure the peak cAMP response as previously described [10]. To examine the ability of PGE₂ receptor antagonists to block EP₂ agonist-induced cAMP, PHK's were plated in 24-well plates at a density of 25,000 cells/well in low calcium, serum-free K-SFM media. Indomethacin was added as above to block endogenous PGE₂ production. The cells were then stimulated with 10 nM of the specific EP₂ receptor agonist (CAY10933, Cayman Chemical, Ann Arbor, MI) in the absence or presence of AH6809 (12.5 μM) and SC51322 (500 nM) for 15 minutes. In this case, isobutylmethylxanthine (IBMX) was added at a concentration of 4 mM to inhibit cAMP phosphodiesterase activity. Cyclic AMP levels were quantitated as above. In all cases, cAMP responses were normalized to total monolayer protein following hydrolysis in 1 N NaOH using a BCA assay kit (Pierce Biotechnology, Rockford, IL).

Cornified envelope quantitation and envelope competency assay

For cornified envelope quantitation (Fig 1A), PHKs growing in FBS-DMEM as described in the figure legend were trypsinized, pelleted and resuspended in 1 ml of FBS-DMEM. For total cell counts, 0.1 ml of the cell suspension was removed and counted. For cornified envelope quantitation, 100 μl of 20 % sodium dodecyl sulfate (SDS) with 20 mg/ml dithiothreitol was added to the remainder of the cell suspension. The cell suspensions were then placed in boiling water for 5 min and then allowed to cool to room temperature. DNAse I (5 U) was then added and the tubes incubated at 37°C for 15 minutes. Cornified envelopes were then counted using a hemocytometer and expressed as a percent of total cell counts. For the enumeration of cells competent to form cornified envelopes (Fig 4A), PHKs growing in K-SFM media were trypsinized and plated at a cell density of 80,000 cells/well into 6-well plates. At approximately 50-60% confluent growth, fresh media was added containing either dimethylsulfoxide (vehicle) or EP receptor antagonist (SC51322 or AH6809) as described in the figure legend. Media was changed every other day until 3-4 days after reaching confluence. The cells were then trypsinized, and an aliquot counted for total cell counts. The remaining cells were centrifuged, and the pellet resuspended in K-SFM containing 1.2 mM calcium and 5 μg/ml of the ionophore, A23187 to induce envelope formation as previously described [18]. Cornified envelopes were then quantitated by counting SDS-insoluble corneocytes using a hemocytometer at least 10 times per sample.

Figure 1. Up-regulation of the EP₁ receptor occurs with high cell density and precedes the appearance of corneocytes in primary human keratinocytes (PHKs).

(A). Specific PGE₂ binding activity is induced and precedes cornified envelope formation with increasing cell density. Freshly isolated primary human keratinocytes were plated onto collagen-coated 12-well plates in DMEM with 10% fetal bovine serum and antibiotics. One day prior to 100% confluence (day -1), and daily thereafter until the cells were 2 days post-confluent (Day +2), duplicate wells were incubated with [³H]-PGE₂ in the presence or absence of a 1000-fold molar excess of unlabeled PGE₂. Specific radiolabeled PGE₂ binding was then determined and normalized to total cellular protein. On the same days, duplicate wells were trypsinized and terminally differentiated SDS-insoluble cornified cells were counted. Corneocytes are shown as a total cells present per well. (B). EP₁ receptor mRNA is induced as PHKs acquire a confluent monolayer. PHKs were plated onto collagen-coated 6-well plates as described in panel A above. Starting one day prior to confluence, and daily thereafter through 2 days post-confluent, the cells were lysed and total RNA prepared for real-time PCR analysis of EP₁ receptor expression. Results were normalized to 18S ribosomal RNA. The results represent the mean ± SEM of two experiments done in duplicate. (C). EP₁ specific agonists compete for radiolabeled PGE₂ binding in confluent PHKs. Two day post-confluent keratinocytes were treated with radiolabeled PGE₂ as in panel A. Competition for radioligand binding is shown using 100-fold excess of unlabeled PGE₂ and receptor agonists. Results shown are the mean and SEM of two experiments done in duplicate.

Figure 4. EP₁ receptor antagonists inhibit keratinocyte differentiation.

(A). Primary human keratinocytes were grown in K-SFM (0.06 mM Ca²⁺). Addition of vehicle (0.1% DMSO), 10 μM of the non-specific EP₁, EP₂, EP₃ receptor antagonist, AH6809, or 300 nM of the EP₁ selective antagonist, SC51322 was begun when the cells were 50-60% confluent and every other day thereafter. At 3-4 days after reaching confluence, the capacity of the cells to form SDS-insoluble cornified cell envelopes was determined after trypsinizing the cells, pelleting the cells, and treating the cells for 3 hours with a calcium ionophore and high calcium media to stimulate envelope formation (envelope competence). The results were normalized to total cell counts and expressed as a percent of vehicle control cells. The results represent the mean and SEM of three experiments done in single or duplicate wells, with corneocytes from each well counted at least 10 times using a hemocytometer. * Results significantly different from control cells (P < 0.05; one sample t-test). (B). The non-specific EP₁, EP₂, EP₃ antagonist (AH6809), but not the EP₁ specific antagonist (SC51322), inhibits EP₂ receptor stimulated cAMP production. Primary human keratinocytes were first incubated with 3 μg/ml indomethacin to block endogenous PGE₂ formation. The cells were then stimulated with a highly selective EP₂ receptor agonist (CAY10933, 10 nM) in the absence or presence of AH6809 (12.5 μM) and SC51322 (500 nM) for 15 minutes. Cyclic AMP was measured using a commercial EIA kit. (C). The EP₁ specific agonist SC51322 inhibits calcium-dependent up-regulation of the differentiation specific markers, cytokeratin K10 and epidermal transglutaminase (TGMI). Duplicate wells of primary human keratinocytes were grown in K-SFM (0.06 mM Ca²⁺) until they reached near confluence (1 day prior to attaining confluence). At this time, the cells were treated with vehicle or 300 nM SC51322. One hour later, additional Ca²⁺ (1.2 mM), was added to the wells as indicated. After an additional 48 hours (1 day post-confluent), the cells were processed for isolation of total RNA. Semi-quantitative RT-PCR was then performed on each of the duplicate samples using specific primers for K10, TGM1, or β-actin (loading control). All PCR reactions were stopped during the exponential phase of PCR amplification and visualized by agarose gel electrophoresis and autoradiography. The mean normalized band intensity for both K10 and TGM1 (normalized to β-Actin) is shown under each duplicate radiographic image. In each case, the band intensity is shown as a ratio compared with the low calcium (0.06 mM) vehicle control cells (assigned a value of 1.00). Band intensity was determined by area integration using NIH Image J software.

Quantitative real-time PCR

Total RNA was prepared from cellular monolayers using a commercial kit per the manufacturer's instructions (RNeasy ®, Qiagen, Valencia, CA). Reverse transcription was performed using the Superscript III kit (Invitrogen, Carlsbad, CA) and was followed by real-time quantitative PCR using a Cepheid Smart-Cycler real-time PCR instrument (Fisher Scientific, Pittsburgh, PA). Primers and probes for the human EP₁ receptor were as follows: EP₁ forward primer: 5’-AGCTCGCGCCCACGA-3’; EP₁ reverse primer: 5’-ATGCACGACACCACCATGAT-3’; EP₁ probe: 5’-[DTET]-TGGAGATGGTGGGCCAGCTTGTC-3’. Primers and probe for the EP₂ receptor were as follows: EP₂ forward primer: 5’-GGAGAGGAGGGCGCATCT-3’; EP₂ reverse primer: 5’-GGGAGTCATTGGAGGCATTG-3’; EP₂ probe: 5’-[DTET]-TTTTCCAGGCACCCCACCATGG-3’]. 18S forward primer: 5’-ACATCCAAGGAAGGCAGCAG-3’; 18S reverse primer: 5’-TCGTCACTACCTCCCCGG-3’; 18S probe: 5’-[DFAM]-CGCGCAAATTACCCACTCCCGA-3’]. All primers and probes were synthesized by Sigma-Genosys (St. Louis, Mo). Following real-time PCR, the results were converted to numeric values utilizing a standard curve generated from serial dilutions of a gel-purified PCR fragment of each target. Results for EP₁ and EP₂ receptor are normalized to 18S rRNA expression for each sample.

Semi-quantitative PCR for epidermal transglutaminase (TGMI) and keratin K10

First to third passage neonatal human keratinocytes were plated in 6-well dishes (35 mm dishes) in K-SFM (0.06 mM Ca²⁺). After allowing the cells to attach for 16-20 hours, cells grown to a nearly confluent monolayer (day -1) were treated with 300 nM SC51322 or DMSO (0.05% final concentration). In selected wells, 1.2 mM calcium chloride was added 1 hour after SC51322 addition. Total RNA was prepared from duplicate wells at 48 hours after treatment. cDNA was prepared by heat denaturing 0.75 ug RNA at 68°C for 5 minutes, followed by reverse transcription (Superscript II, Invitrogen) for 60 minutes at 42°C, using 2 μM oligo-dT(12-18) primer (Invitrogen), 1.25 mM each dNTP, 0.5 U RNase inhibitor (Promega), and 10 mM DTT. Relative message levels of transglutaminase 1 (TGM1), keratin 10 (K10), and β-actin were determined by PCR amplification of cDNA, diluted 1:10 with H₂O and added at 1/10th volume to a 25 μl reaction comprising 2.5 U Taq DNA polymerase (Promega), PCR buffer (Sigma), 1.8 mM MgCl₂₂, 400 nM primers, 400 μM each dNTP (Roche Molecular Biochemicals), and 5 μCi α-³²P-dCTP. Samples were removed after 22 cycles (TGM1 and K10) or 18 cycles (β-actin) of PCR amplification which corresponded to the midpoint of exponential amplification as determined in previous experiments. The thermal profile was 94°C for 20 seconds, 57°C for 20 seconds, 72°C for 30 seconds. PCR products were then electrophoresed in 6% acrylamide, 0.5X TBE gels and autoradiographed. Primers of the following sequence were used: TGM1 forward: 5’-TCTGTGGGTCCTGTCCCATCCATCCTGACC-3’, and reverse: 5’-CCCCAACGGCCCACATCGGAACGTGGCCCA-3’; human β-actin forward: 5’-CAGGCTGTGCTATCCCTGTAC-3’ and reverse: 5’-CACGCACGATTTCCCGCTCGG-3’; human cytokeratin K10 forward: 5’-GGCTCTGGAAGAATCAAACTATGAGC-3’ and reverse: 5’-GGATGTTGGCATTATCAGTTGTTAGG-3’.

Tumor sections

Formalin-fixed, paraffin-embedded non-melanoma skin cancer sections were obtained from Department of Pathology archival tissue storage at the Indiana University School of Medicine. Procurement was done following approval by the Indiana University at Indianapolis Institutional Review Board.

Immunohistochemistry

Immunohistochemistry was done on human epidermis acquired from reductive mammoplasties and panniculectomiess. Human tumors were acquired from archival formalin-fixed, paraffin-embedded tumors. Immunolabeling was done using a monoclonal anti-EP₁ receptor antibody, clone 5F12 and the SuperPicture broad spectrum IHC staining kit (Zymed Laboratories, South San Francisco) [14]. Briefly, after deparaffinization and rehydration, heat-induced antigen retrieval was done in 75% glycerol: 25% Dako high pH antigen retrieval buffer as described previously [14], incubating the sections with the primary antibody at a concentration of 10 μg/ml.

Scoring and statistical analysis of human tumors

Following IHC staining, each tumor was scored for both membrane and nuclear staining intensity by three different pathologists (SB, NCP, TMK). For both nuclear and membrane staining, a score of 0 indicated no staining; 1+ = weak staining; 2+ = moderate staining; and 3+ = strong staining. The absence of cytoplasmic granular layer staining with loss of keratohyaline granules was also noted. The distributions of nuclear and membrane scores were compared between each tumor type and normal tissue using Cochran-Armitage exact trend tests.

EP₁ receptor immunolocalization in hyperplastic mouse epidermis and in EP₁ knockout mice

EP₁ knockout mice and C57Bl/6 syngeneic wild-type control mice were housed as previously described [19]. Dorsal skin from wild-type and EP₁ knockout mice was excised Hairless, albino, female SKH1 mice were purchased from Charles Rivers Laboratories (SKH1-Hr^hr; Wilmington, MA). The mice were housed in standard microisolator cages under a simulated 12 hour day/night cycle and were fed and watered ad libitum. UVB-induced hyperplasia in an SKH1 mouse was elicited by 1500 J/m² of UVB irradiation using two Westinghouse FS40 sunlamps (National Biological Corp., Twinsberg, OH). A control mouse received no UVB irradiation. At 72 hours post UVB irradiation, the dorsal skin was excised from the euthanized irradiated and non-irradiated control mice. This time was chosen as it corresponds to the peak time point for UVB-induced hyperplasia[20]. The epidermis was then formalin-fixed and paraffin-embedded. To verify the specificity of the immunohistochemical response, formalin-fixed tissue from wild-type and EP₁ knockout mice was also utilized. The fixed skin was then cut into sections, deparaffinized, and heat-induced antigen retrieval was performed as described for the human tissue. The EP₁ monoclonal antibody was applied at the same concentration and for the same period as that done for the human studies. However, due to the fact that we were utilizing a mouse monoclonal antibody on mouse tissue, this required the use of a specialized kit that blocks non-specific detection of endogenous mouse immunoglobulin (Mouse on Mouse Kit, Vector Laboratories, Burlingame, CA). The addition of blocking reagent, primary and secondary antibodies, and detection reagents were performed in kit buffers and done as detailed by the manufacturer except that CAS block (Zymed Laboratories) was added at a 10% final concentration to primary and secondary antibody buffers. All mouse studies were approved by the Indiana University-Purdue University at Indianapolis Animal Care and Use Committee or the University of Ottawa Animal Care Committee.

RESULTS

Increased EP₁-receptor expression precedes the onset of density-dependent corneocyte formation

Given our hypothesis that the EP₁ receptor regulates keratinocyte differentiation, we first examined whether EP₁ receptor expression is up-regulated before or during keratinocyte differentiation in vitro. Previous studies by others have demonstrated that high cell density and the attainment of cell-cell adhesion is key to the initiation of keratinocyte differentiation in vitro [2, 21, 22]. We therefore examined how EP₁ receptor expression was altered during the period in which primary human keratinocytes (PHKs) attain a confluent monolayer and begin the process of corneocyte formation. In figure 1, we processed PHKs at daily intervals beginning at 1 day prior to reaching confluence to 2 days post-confluence for PGE₂ binding activity and SDS-insoluble corneocyte formation (Fig 1A), as well as EP₁ receptor mRNA expression (Fig 1B). In figures 1A & 1B, we demonstrate that both specific [³H]-PGE₂ binding activity and EP₁ receptor mRNA expression increase at the time PHKs attained a confluent monolayer (day 0). Moreover, this increased PGE₂ binding activity and increased EP₁ receptor mRNA expression precede a parallel increase in the formation of terminally differentiated corneocytes beginning 1 day post-confluence. To further characterize the binding activity seen in figure 1A, we next turned to competitive binding studies. In figure 1C, competition studies demonstrate that the EP₁, EP₂ and EP₃ receptor agonist, 17-phenyl-ω-trinor-PGE₂ (17-pt-PGE₂), and PGE₂ showed equivalent potencies for blocking [³H]-PGE₂ binding. In contrast, the EP₂, EP₃, EP₄ agonist, 11-deoxy-PGE₁ (11d-PGE₁), was approximately half as potent as PGE₂ or 17-pt-PGE₂, while the EP₃ receptor agonist misoprostol was unable to compete for radioligand binding. Thus, this data is in agreement with the RT-PCR data in figure 1B and demonstrates that the EP₁ receptor appears to be at least partially responsible for the increased PGE₂ receptor binding activity that preceded corneocyte formation. However, the competitive binding data indicates that EP₂ receptors likely represent a significant portion of the PGE₂ receptors expressed in post-confluent keratinocytes. The potential involvement of EP₄ receptors is less likely, as we have previously shown that only trace amounts of EP₄ receptor mRNA is present in PHKs in culture [10].

EP₁ receptor expression and EP₁-mediated calcium signaling are up-regulated by high cell density

The data in figures 1A & B indicate that EP₁ receptor expression is up-regulated with increasing cell density and this receptor up-regulation precedes density-induced differentiation. However, to rule out the possibility that the EP₁ receptor is regulated by days in culture rather than high cell density, we next performed a series of experiments to verify that the EP₁ receptor expression and function are dependent on cell density. In this case, we plated cells at high and low cell densities so that we could assess EP₁ receptor expression and function in pre-confluent and post-confluent cultures after the same time in culture. We then examined EP₁ receptor-mediated calcium signaling, mRNA expression, and protein expression.

In a previous study, we demonstrated that the EP₁ receptor is coupled to intracellular calcium mobilization by the EP₁ receptor agonists iloprost, 17-pt-PGE₂ and sulprostone [14]. Given the importance of calcium in regulating differentiation-specific gene expression in keratinocytes, we first examined whether EP₁-mediated intracellular calcium mobilization is cell density dependent. In figures 2A & B, we demonstrate that the ability of iloprost to induce calcium mobilization is restricted to post-confluent keratinocytes. While iloprost is also a potent agonist for the prostacyclin (IP) receptor [5], the possibility that iloprost stimulated calcium mobilization through activation of the IP receptor is remote, as this receptor is known to be coupled to Gs alpha subunit signaling and adenylate cyclase activation [23]. Moreover, two other EP₁ receptor agonists, sulprostone and 17-pt-PGE₂ [5], also induced calcium mobilization in post-confluent, but not pre-confluent PHKs (data not shown). In this case, the rank potency for calcium mobilization in confluent PHKs was: iloprost ≥ 17-pt-PGE₂ > sulprostone. As a positive control, keratinocytes are known to express the platelet activating factor receptor (PAF-R), which is coupled to intracellular calcium signaling [24]. Thus, it is noteworthy that a PAF-R agonist, carbamyl-PAF (CPAF), was able to induce a calcium transient regardless of cell density.

Figure 2. EP₁ receptor signaling and expression are up-regulated in PHKs following the attainment of a confluent monolayer.

(A & B). Agonist induced calcium mobilization is restricted to PHKs grown to a post-confluent monolayer. Cells at low cell density (pre-confluent) or high cell density (post-confluent) were loaded with the fluorescent calcium indicator, Fura PE3/AM. After trypsinization, the cells were stimulated with ethanol (EtOH), 100 μM iloprost in ethanol, or 40 ng/ml of the positive control platelet activating factor receptor (PAF-R) agonist, carbamyl-PAF (CPAF). (A). The EP₁ receptor agonist, iloprost, is unable to induce measurable calcium mobilization in pre-confluent PHKs. In contrast, the positive control PAF-R agonist, CPAF, is shown to induce a calcium mobilization response. (B). The EP₁ receptor agonist, iloprost, induces a robust calcium mobilization response in post-confluent keratinocytes. (C). EP₁ receptor mRNA expression is up-regulated in post-confluent PHKs compared with pre-confluent PHKs under differing culture conditions. RNA was prepared from both pre-confluent and post-confluent cells and EP₁ expression was assessed by quantitative real-time RT-PCR and normalized to 18S rRNA. Results represent the mean ± SEM for n=4-5 experiments done in duplicate; * p < 0.05, one-sample t-test relative to 100 % for pre-confluent controls. (D). EP₁ receptor protein expression is up-regulated in post-confluent PHKs compared with pre-confluent PHKs. An immunoblot was performed on total cell lysates from pre-confluent PHKs and post-confluent PHKs grown in serum free media (K-SFM; 0.06 mM Ca²⁺). EP₁ immunoreactive bands are seen at approximately 35 kDa and 70 kDa. The blot was stripped and reprobed with anti-α-tubulin antibody as a loading control (bottom panel).

We next sought to verify that EP₁ receptor mRNA and protein expression are up-regulated in post-confluent compared to pre-confluent PHKs. We first examined EP₁ receptor mRNA expression by qRT-PCR. In this case, we also examined whether differing culture conditions could also alter EP₁ receptor expression. In particular, high extracellular calcium is a potent stimulus for the induction of differentiation-specific proteins. We therefore utilized serum-free defined media (K-SFM) with both low and high calcium concentrations, as well as the culture conditions utilized in the studies described in figure 1 (FBS-DMEM). In figure 2C, we show that EP₁ receptor mRNA expression is up-regulated in post-confluent PHKs compared with pre-confluent PHKs under all culture conditions. Interestingly, varying the extracellular calcium concentration (K-SFM vs K-SFM + Ca²⁺) did not significantly alter EP₁ receptor mRNA expression. In contrast, density-dependent up-regulation of the EP₁ receptor was less robust in the serum-containing culture conditions utilized in figure 1 (for comparison, see figure 1B for cells at 1-2 days post-confluence; in figure 2C, the confluent cells were 2-3 days post-confluent). Finally, in figure 2D, we show by immunoblot that the EP₁ receptor is up-regulated in post-confluent PHKs relative to pre-confluent PHKs. As expected, two different sized bands were detected [14] (see also supplemental figure 2). Under non-confluent conditions, only the lower molecular weight band was seen. Upon reaching confluence, the lower molecular weight band was increased and the high molecular weight band became apparent. It should be noted that G-protein coupled receptors (GPCR) are frequently observed to exhibit multiple bands by western blot [11, 25-27]. In fact, the large number of GPCR receptors exhibiting multiple banding by immunoblot preparations, along with other lines of evidence, have led to the general belief that these multiple bands represent GPCR homo- and hetero-dimers or higher order oligomers and that this oligomerization is involved in GPCR activity (reviewed in, [28, 29]). Thus, it is possible that these two immunoreactive bands represent monomeric and dimeric forms of the EP₁ receptor. This idea is supported by the apparent molecular weights of the two immunoreactive bands, with the smaller band seen at approximately 35 kDa and the larger band seen at approximately 70 kDa. This is consistent with our previous study, in which the major immunoreactive bands were seen at 35-40 and 61-70 kDa [14]. This minor variability in protein size determinations could be due to a number of factors, including differences in protein MW standards or sample to sample variability in solubilization and/or SDS-dependent denaturation of this highly hydrophobic protein. These results are also consistent with our previous study showing the immunolocalization of the EP₁ receptor within human epidermis [14]; in this case, the EP₁ receptor exhibits greater staining in the more differentiated spinous and granular layers, with a marked increase in staining observed in the granular layer.

EP₂ receptor expression and functional coupling to cAMP are not induced by high cell density

In figure 1C, we show that 17-pt-PGE₂ was as effective as PGE₂ in blocking radiolabeled PGE₂ binding to post-confluent keratinocytes. Moreover, 11d-PGE₁ also competed for binding to membranes from post-confluent keratinocytes. Given that 17-pt-PGE₂ is also an EP₂ receptor agonist and 11d-PGE₁ exhibits weak binding affinity with the EP₁ receptor [30], this suggests that EP₂ receptors are present at relatively high levels in post-confluent keratinocytes. We therefore sought to determine whether EP₂ receptor expression is also altered by high cell density under the serum-containing growth conditions utilized in figure 1. In figure 3A, we show that PGE₂ receptors coupled to cyclic AMP (cAMP) production (EP₂ and/or EP₄) exhibit decreased functional activity in PHKs at high cell density (Post-Conf) versus low cell density (Pre-Conf) growth conditions. In figure 3B, EP₂ mRNA expression was unaltered by cellular confluence in defined media containing both low and high calcium concentrations. In contrast, EP₂ mRNA expression was significantly decreased under the serum containing culture conditions utilized in figures 1C and 3A. Similarly, EP₂ protein expression was not increased by high cell density when assessed by immunoblot (insert in Fig 3B). These studies indicate that, depending on the culture conditions, EP₂ receptor expression and activity is either unaffected or suppressed as keratinocytes achieve a confluent monolayer and begin the process of differentiation. However, while EP₂ receptors are not up-regulated by high-cell density, it is clear that EP₂ receptor function and expression at the RNA and protein level is still relatively high in post-confluent keratinocytes. This is consistent with our previous report utilizing immunohistochemistry [14]. In this case, while the EP₂ receptor is expressed in the superficial spinous and granular layers of the epidermis, it exhibits greater staining in the less differentiated keratinocytes at or near the basal layer.

Figure 3. Depending on the culture media, high cell density either has no effect on EP₂ receptor expression or suppresses EP₂ receptor expression in PHKs.

(A). PGE₂-induced cyclic AMP production is reduced in PHKs at high cell density. Pre-confluent PHKs or post-confluent PHKs cultured in DMEM + 10% FBS were treated with 3 μg/ml indomethacin overnight to block endogenous PGE₂ formation. The cells were then stimulated with 100 nM PGE₂ for 1 minute. Cyclic AMP was measured using a commercial EIA kit as described in the methods section. The results represent the mean and SEM of three experiments. (B). Quantitative real-time PCR was performed for EP₂ receptor mRNA expression as described for the EP₁ receptor in figure 2C above. Inset: Immunoblot for EP₂ receptor expression in pre-confluent (Pre) and post-confluent (Post) PHKs grown in DMEM with 10% fetal bovine serum. Membrane preparations were produced from pre-confluent and 2 days post-confluent PHKs. In each case, 40 μg of the membrane preparation was separated by SDS-PAGE electrophoresis and immunoblot performed using a polyclonal anti-EP₂ receptor antibody as described in the methods section.

Receptor antagonists demonstrate a role for the EP₁ receptor in keratinocyte differentiation

The above data indicates that the EP₁ receptor is up-regulated in PHKs during density-dependent differentiation and that this up-regulation precedes the formation of terminally differentiated SDS-insoluble corneocytes. Moreover, this up-regulation is associated with the ability of EP₁ receptor agonists to stimulate intracellular calcium signaling, which is a well-known inducer of PHK differentiation. Thus, we hypothesized that the EP₁ receptor is involved in density-dependent PHK differentiation. We therefore examined whether EP₁ receptor antagonists would block density-dependent corneocyte formation. In figure 4A, we demonstrate that the capacity to form corneocytes is inhibited by 67% using the EP₁ receptor antagonist AH6809 and by 43% using the more selective EP₁ receptor antagonist SC51322 [5]. Given that EP₂ receptors are expressed in post-confluent keratinocytes, we next show that SC51322 has no effect on EP₂ receptor-mediated signaling through cAMP (figure 4B). It should also be noted that the inability of SC51322 to alter EP₂ receptor activation was seen at a concentration (500 nM) that is nearly two-fold greater than that utilized to assess the role of the EP₁ receptor in keratinocyte differentiation (figures 4A & C). This result confirms a previous report indicating that SC51322 has essentially no binding affinity for the EP₂ receptor [5]. As expected, the non-specific EP_1-3 receptor antagonist, AH6809, was highly effective in blocking EP₂ agonist-stimulated cAMP production (*, p<0.05 compared with EP₂ agonist-stimulated cAMP).

We next examined the ability of SC51322 to inhibit the expression of the differentiation markers cytokeratin K10 (K10) and epidermal transglutaminase (TGM1) [31, 32]. In this case, SC51322 was applied to PHKs over a 48 hour period that corresponded to 1 day pre-confluent to 1 day post-confluent. The ability of SC51322 treatment to inhibit extracellular calcium-induced up-regulation of these differentiation markers was then assessed. In figure 4C, we show that K10 expression is increased approximately 2.5-fold in PHKs by treatment with high concentrations of extracellular calcium. Interestingly, SC51322 inhibited calcium-induced expression of K10 by approximately 60%. SC51322 also inhibited the expression of epidermal transglutaminase (TGM1), the calcium-dependent enzyme that catalyzes the final cross-linking of envelope precursors to form the detergent-insoluble cell envelope [32]. In this case, the EP₁ receptor antagonist inhibited TGM1 expression in both low and high calcium conditions.

The intensity of EP₁ receptor plasma membrane localization correlates with a differentiated phenotype in human non-melanoma skin cancers

In studies by other investigators, it has been noted that the EP₁ receptor is highly expressed in human squamous cell carcinomas (SCCs) and actinic keratoses (AKs), but is weakly expressed or absent in basal cell carcinomas (BCCs) [15]. We therefore extended these studies by examining the immunohistochemical localization of the EP₁ receptor in these lesions as well as poorly differentiated SCCs (PD-SCCs) and spindle cell carcinomas. We have previously demonstrated by IHC that normal human epidermis exhibits a weak cell membrane-associated staining pattern for the EP₁ receptor [14]. In our new studies shown in table 1 and figure 5, we demonstrate that the membrane staining pattern was markedly and significantly increased in areas of well-differentiated SCC compared with normal epidermis (see table 1 and figure 5). In contrast, spindle cell carcinomas and basal cell carcinomas lacked membrane staining. In poorly differentiated SCCs, membrane staining was also absent, except for some faint focal staining. In tumors with mixed areas of poorly differentiated and well-differentiated tumor, membrane staining was present primarily in the areas of well-differentiated keratinized tumor.

Table 1.

Membrane scoring for tumors

Membrane Score	Normal	AK	SCC****	PD-SCC	Spindle****	BCC****
3+	0	0	4 *	0	0	0
2+	4	0	2	1 **	0	0
1+	6	2	3	3 **	0	1
0	7	2	0	8	5 ***	7

^*Three of the four well-differentiated tumors with 3+ membrane staining were keratoacanthoma type SCC.

^**1 or 2+ staining was seen in four tumors with mixed poorly differentiated and well-differentiated tumors. Staining was seen only focally or in areas of well differentiated SCC.

^***2 Spindle cell carcinomas showed 1+ membrane staining in adjacent areas of differentiated SCC

^****distribution significantly different from Normal (P < 0.05; two-tailed exact trend test)

Figure 5. EP₁ receptor immunolocalization in non-melanoma skin cancer.

Immunohisochemical (IHC) analyis of EP₁ receptor expression was performed on formalin-fixed, paraffin-embedded archival tissue samples. In each case, 5 μm sections were deparaffinized and heat-induced epitope retrieval was done as outlined in the methods section. IHC staining was done using a monoclonal anti-human EP₁ receptor antibody (clone 5F12) (A, C-F) or an isotype control (B). (A). Keratoacanthoma (400x magnification). (B). A serial section of the same keratoacanthoma stained with isotype (IgG2bκ) negative control antibody (400x). (C). Hyperplastic skin overlying the basal cell carcinoma seen in panel E. (D). Well-differentiated squamous cell carcinoma (SCC) (200x). (E). Basal cell carcinoma (400x). (F). Hematoxylin & eosin stained section corresponding to the section seen in panel C. (G). Poorly-differentiated SCC (400x). (H). Spindle cell carcinoma (400x).

In addition to the membrane staining pattern, we have previously noted in normal epidermis that the EP₁ receptor is heavily localized to the granular layer, exhibiting a grainy cytoplasmic immunolocalization pattern suggestive of localization to intracellular granules [14]. In AKs and well-differentiated SCCs, this pattern was recapitulated, with intense granular-appearing cytoplasmic staining occurring only in areas of tumor exhibiting keratohyaline granules (figures 1C & F). This pattern was not observed in BCCs, poorly differentiated SCCs (PD-SCCs), or spindle cell carcinomas that lack granular layer differentiation. However, it was seen in areas of normal or hyperplastic epidermis overlying these poorly differentiated tumors (figure 1C). It should be noted that this lack of granular cytoplasmic staining with loss of keratohyaline granules was also noted in two archival samples of psoriasis, an inflammatory skin disease also known for loss of normal granular layer development (data not shown).

Finally, in addition to the cell membrane and cytoplasmic staining patterns described above, we have previously demonstrated that normal human epidermis exhibits a perinuclear/nuclear pattern of EP₁ immunolocalization [14]. In figure 5 and table 2, we demonstrate that both normal human epidermis and human tumors exhibit nuclear EP₁ receptor localization by IHC. Significantly, while all normal epidermal samples, AKs, and well-differentiated SCCs exhibit 1+ or greater nuclear scores, a subset of BCCs, spindle cell carcinomas and poorly differentiated SCCs exhibit a lack of nuclear staining, with a statistically significant decrease in nuclear staining noted in BCCs and PD-SCCs (see table 2). It should be noted that PGE₂ receptors have been demonstrated to be localized to intracellular membranes, particularly the nuclear membrane, by multiple methodologies including electron microscopy, immunofluorescence, radioligand binding studies of cellular and intracellular membrane preparations, and immunolocalization of epitope-tagged receptors [14, 33-38]. Moreover, studies using isolated nuclei demonstrated that the receptor is coupled to nuclear calcium signaling [33].

Table 2.

Nuclear scoring for tumors

Nuclear Score	Normal	AK	SCC	PD-SCC*	Spindle	BCC*
3+	12	3	7	4	3	2
2+	4	1	1	4	1	1
1+	1	0	1	4	0	2
0	0	0	0	1	2	2

^*distribution significantly different from Normal (P < 0.05; two-tailed exact trend test)

To further validate the IHC results observed in figure 5, we next utilized mice with germ-line deletion of the EP₁ receptor. In supplemental figure 1, we demonstrate that the EP₁ receptor is localized in the superficial epidermis that immediately underlies the stratum corneum in normal dorsal epidermis from SKH1 mouse. Faint staining within some nuclei was also noted. However, given that mouse epidermis is much thinner than human epidermis and lacks a clearly defined granular layer, it was difficult to assess whether this staining pattern reproduced that observed in normal human epidermis. We therefore examined EP₁ receptor expression in mouse skin that had been induced to undergo hyperplasia by ultraviolet B (UVB) treatment. This allowed us an opportunity to examine whether the cytoplasmic/perinuclear staining pattern seen in the granular layer of human epidermis (see figure 5C) is also observed in hyperplastic mouse epidermis. It should be noted that UVB-induced epidermal hyperplasia also results in a marked expansion of the granular layer in mice. We therefore treated SKH1 mice with 1500 J/m² UVB and then examined EP₁ receptor at 72 hours post-irradiation. This time point was chosen as it corresponds with the peak period of UVB-induced hyperplasia [20]. In supplemental figure 1B, we show that the UVB-induced hyperplastic epidermis exhibited a prominent grainy cytoplasmic staining pattern that closely mimicked the pattern seen in human epidermis. In addition, the EP₁ receptor was seen to be up-regulated after 72 hours within the nuclei of epidermal keratinocytes. Finally, to verify the specificity of the monoclonal antibody, we demonstrate that mice with germ-line deletion of the EP₁ receptor lacked staining for the EP₁ receptor (Supplemental figure 1D). In contrast, wild-type syngeneic C57Bl/6 mice exhibit a pattern of staining in the superficial epidermis that was very similar to that observed in SKH1 mice (compare supplemental figures 1A and C).

DISCUSSION

In this study, we provide evidence that the EP₁ subtype of PGE₂ receptor is important in regulating normal keratinocyte differentiation in vitro. First, we show that EP₁ receptor expression and the ability of EP₁ agonists to induce intracellular calcium mobilization is up-regulated in a density dependent manner and that this up-regulation precedes the appearance of terminally differentiated keratinocytes. In addition, EP₁ receptor antagonism inhibits density-dependent and calcium-dependent keratinocyte differentiation. Finally, we verify previous reports that EP₁ receptor immunolocalization correlates with a squamous or well-differentiated phenotype in human non-melanoma skin cancer.

Our finding that the EP₁ receptor is coupled to calcium mobilization is consistent with previous studies that demonstrate that the EP₁ receptor is coupled to calcium mobilization in PHKs and other cell types [3, 14, 16, 33]. Inasmuch as calcium mobilization is well known to stimulate keratinocyte differentiation [8, 39], our new findings that EP₁ receptor signaling is up-regulated upon reaching cellular confluence suggests a role for the EP₁ receptor in regulating density-dependent keratinocyte differentiation. Thus, the next studies examined whether the EP₁ receptor is involved in keratinocyte differentiation in vitro. We show that the EP₁ receptor antagonist, SC51322, significantly inhibits cornified envelope formation, as well as keratin K10 and epidermal transglutaminase expression. Importantly, SC51322 is also reported to have essentially no binding activity with the EP₂ receptor (Ki of >100,000 nM) and very poor binding activity with the EP₄ receptor (Ki of 14,032 nM) [5]. Our studies demonstrating that SC51322 fails to block EP₂ receptor-mediated cAMP production confirms the inability of SC51322 to act as an EP₂ receptor antagonist. Thus, the ability of SC51322 to suppress corneocyte formation and differentiation marker expression is consistent with antagonism of the EP₁ receptor, but not the EP₂ receptor. However, it is possible that SC51322 may be exhibiting non-specific activity unrelated to any of the prostaglandin receptors. This appears to be unlikely, as we show that the structurally unrelated EP receptor antagonist, AH6809, also inhibits corneocyte formation. Thus, our results provide pharmacologic support for the idea that the EP₁ receptor, but not the EP₂ receptor, is an important regulator of keratinocyte differentiation.

While SC51322 is reported to be one of the most potent and selective antagonists of the EP₁ receptor (Ki of 13.8 nM), SC51322 also exhibits considerably weaker affinity for the EP₃ receptor (reported Ki of 698 nM) [5]. However, we feel that the EP₁, rather than the EP₃ receptor, is likely the major PGE₂ receptor subtype involved in regulating keratinocyte differentiation. This conclusion is based on the following observations: First, the concentration of SC51322 used in our studies (300 nM) is well above the reported binding affinity of this antagonist for the EP₁ receptor. In contrast, this concentration is less than 50% of the reported Ki of this antagonist for the EP₃ receptor. Second, while SC51322 has modest ability to compete for PGE₂ binding to the EP₃ receptor, it is unclear whether this binding activity is associated with significant EP₃ receptor antagonist activity [40]. In contrast, studies in vivo using EP₁ receptor knockout and wildtype mice demonstrate the effectiveness of SC51322 in inhibiting EP₁-dependent physiologic activity while studies in vitro demonstrate that SC51322 is a potent inhibitor of EP₁ receptor-mediated calcium mobilization in cells over-expressing the EP₁ receptor [41, 42]. Finally, our previous study has demonstrated that the EP₃ receptor is not up-regulated with confluence in PHKs and is primarily expressed in the basal and lower spinous layers of intact human epidermis where it likely acts as a negative regulator of proliferation [13]. This is also consistent with our data in figure 1C, in which the EP₃ receptor agonist, misoprostol, was ineffective in blocking PGE₂ binding to post-confluent keratinocyte membrane preparations. However, we cannot rule out the possibility that the EP₃ receptor may act cooperatively with the EP₁ receptor to stimulate differentiation. This idea would be consistent with the greater effectiveness of AH6809 over SC51322 in blocking corneocyte formation. However, these differences could also be attributed to the different concentrations utilized and differences in the relative potencies of these antagonists for the EP₁ receptor.

In normal epidermis, we have previously demonstrated that the EP₁ receptor is localized throughout the epidermis, with perinuclear/nuclear, membrane and cytoplasmic localization [14]. However, the most prominent staining was observed in the granular layer (stratum granulosum), exhibiting a cytoplasmic granular appearance. In human tumors, we demonstrate that EP₁ receptor expression exhibits increased cell membrane localization in well-differentiated keratinized tumors (well-differentiated SCCs), while membrane staining was noted to be reduced in non-keratinizing tumors (BCCs, poorly-differentiated SCCs, and spindle cell carcinomas). Interestingly, the granular cytoplasmic staining was noted only in areas where keratohyaline granules were observed histologically, suggesting that the EP₁ receptor colocalizes with these keratohyalin and fillagrin-rich intracellular structures [43]. Importantly, faint nuclear staining and marked cytoplasmic/perinuclear granular layer staining is noted in normal mouse epidermis, but not in mice lacking the EP₁ receptor. Moreover, in hyperplastic mouse epidermis that more closely resembles the human epidermis, the cytoplasmic/perinuclear granular layer staining is markedly enhanced, as is the nuclear staining pattern.

The above data demonstrating that the EP₁ receptor is expressed in the differentiated compartment of normal and neoplastic human epidermis is in general agreement with a previous report indicating that the EP₁ receptor is expressed in murine and human SCC, but not in BCC [15]. Similarly, enhanced expression of the EP₁ receptor within the differentiated suprabasal compartment of mouse epidermis was also noted by a second group [44]. However, a third group demonstrated EP₁ immunolocalization within the basal compartment of normal mouse epidermis [45]. The authors attribute the differences in their results to differences in methodologies or strain-specific differences. Since this study used a commercial polyclonal antibody preparation, while our current study utilized a monoclonal anti-EP₁ receptor antibody developed in our laboratory, the different staining patterns could be due to the use of different reagents. However, we have previously utilized the same commercial rabbit polyclonal antibody on normal human epidermis and noted a similar staining pattern to that seen using our anti-EP₁ monoclonal antibody [14]. Thus, the differences in the staining pattern are more likely due to species differences, differences in antigen retrieval methodologies or different commercial lot numbers for the polyclonal antibody. It should be noted that we have noted considerable lot-dependent differences in performance of this commercial antibody source as assessed by immunoblot analysis of HEK cell lysates over-expressing the human EP₁ receptor (see supplemental figure 2). Thus, it should also be noted that the immunoblot data shown in figure 2D was performed using the same commercial lot of polyclonal anti-EP₁ receptor antibody that was used in our previous study [14], and the specificity of this lot of antibody is shown in supplemental figure 2A.

In contrast to our findings with the EP₁ receptor, our data also demonstrate that EP₂ receptor expression and functional activity is either unchanged or suppressed in PHKs upon reaching confluence. This is consistent with previous reports demonstrating that the EP₂ receptor is necessary for PGE₂-induced keratinocyte proliferation and that keratinocyte proliferation is markedly suppressed upon reaching a confluent monolayer [2, 10, 12].

One of the more unusual findings of this study relate to the presence of perinuclear or nuclear EP₁ receptor immunolocalization. The significance of nuclear or perinuclear EP₁ receptor immunolocalization is as yet unclear, although there was a weak association with the keratinizing phenotype: While all of the normal human epidermis and well-differentiated non-melanoma skin cancers demonstrated some degree of nuclear/perinuclear staining for the EP₁ receptor, a subset of basal cell carcinomas, poorly-differentiated squamous cell carcinomas and spindle cell carcinomas lacked this staining pattern. It is also possible that the perinuclear/nuclear staining pattern represents non-specific staining. However, we feel that this is unlikely. First, COX-1 and COX-2 are known to be localized to the endoplasmic reticulum and nuclear membrane [46, 47]. Moreover, nuclear membrane localization has been demonstrated for the EP₁, EP₃, and EP₄ receptors in other cell types or tissues [33, 48, 49]. Finally, nuclear EP₁ staining was increased in UVB-irradiated mouse epidermis (supplemental figure 1), while nuclear EP₁ staining was not noted in EP₁ knockout mice. Thus, we speculate that the EP₁ receptor is responsive to PGE₂ production at the nuclear envelope or endoplasmic reticulum of keratinocytes through an autocrine signaling pathway.

In summary, these studies supply evidence that the EP₁ receptor is up-regulated under conditions that stimulate keratinocyte differentiation, the EP₁ receptor is coupled to intracellular calcium mobilization, and EP₁ receptor antagonism inhibits keratinocyte cornification and markers of keratinocyte differentiation. Taken together, these observations supply strong evidence to support a functional role for the EP₁ receptor in mediating keratinocyte differentiation. Moreover, EP₁ receptor expression in non-melanoma skin cancer correlates with histologic evidence of keratinization. Thus, this data suggests that the role of the EP₁ receptor in regulating differentiation appears to be intact in non-melanoma skin cancer. Interestingly, the function and intracellular signaling pathway elicited by the EP₁ receptor in human keratinocytes appears to differ from that observed for the other PGE₂ receptor subtypes. Earlier studies demonstrated that EP₂ receptor activation elicits a proliferative response in PHKs via a cAMP coupled response [10], while the EP₃ receptor acts to suppress growth and induces intracellular signaling through the lipid mediators diacylglycerol and ceramide [13]. Thus, this new data indicates that the multiple PGE₂ receptors, each coupled to different functional roles and intracellular signaling pathways, act to increase the diversity of cellular responses that can be elicited by PGE₂.