The skin cancer chemotherapeutic agent ingenol-3-angelate (PEP005) is a substrate for the epidermal multidrug transporter (ABCB1) and targets tumor vasculature

Abstract

Ingenol-3-angelate (Ing3A), extracted from Euphorbia peplus, is currently in clinical trials for eradicating basal cell carcinoma (BCC), actinic keratosis and squamous cell carcinoma (SCC) in situ by topical application. Although structurally related to phorbol esters and a PKC activator, topical Ing3A, but not phorbol 12-myristate 13-acetate (PMA), inhibited the growth of subcutaneous tumors derived from PAM212 (mouse SCC) and B16 (mouse melanoma). Ing3A and PMA both induced acute neutrophilic inflammation on mouse skin, but only Ing3A caused subcutaneous hemorrhage and vascular damage. Both Ing3A and PMA activated Erk1/2 in epidermis, but Ing3A also activated Erk1/2 in skin dermal fibroblasts and endothelial cells. Pretreatment with topical cyclosporin A (CsA), verapamil or XR9576, modulators of P-glycoprotein (P-gp), prevented Ing3A-induced hemorrhage but not neutrophil infiltration. CsA also impaired Ing3A’s anti-cancer activity while the anti-inflammatory dexamethasone did not. Ing3A, but not PMA, blocked photoaffinity labeling of human P-gp with [¹²⁵I]-Iodoaryazidoprazosin and inhibited P-gp mediated drug resistance to HCT-15 cells. The intracellular levels of Ing3A were significantly lower in P-gp expressing cells and treatment with XR9576 increased the levels to those of cells that do not express P-gp, demonstrating that Ing3A binds to and is transported by P-gp. Taken together, our results suggest that P-gp mediated absorptive transport, dermal penetration and vascular damage contribute to the anti-cancer activity of Ing3A in vivo.

Introduction

Nonmelanoma skin cancer (NMSC), primarily basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) is the most prevalent cancer in the Caucasian population. Although rarely lethal in the northern hemisphere, in warmer climates or in immunosuppressed patients it can metastasize and cause death (1). Under all circumstances its treatment constitutes an important component of health care costs. Traditionally, treatment for NMSC involves surgical intervention and consequent esthetic deformity due to the common location on exposed surfaces on which these sun induced lesions are most frequently found. Treatments for the much more prevalent precancerous actinic keratosis skin neoplasms (AK) are often scarifying as well, adding to the personal and financial expense associated with skin cancer diagnosis and therapy. Thus, the discovery of non-deforming but curative medical therapies is a high priority for these conditions.

With advances in cutaneous biology research, understanding the molecular pathology of skin cancer development has expanded. The cause of the great majority of BCC involves alterations in components of the sonic hedgehog pathway (2), and targeted drugs are in clinical trials (3). The mechanism for the less common but more aggressive squamous cell carcinoma has been more difficult to delineate although considerable insight has been achieved (4).

The discovery that PKCδ was a proximal regulator of keratinocyte viability (5) and was catalytically inactivated or downregulated in cutaneous squamous cell tumors presented a drugable target for tumor therapy (6–9). Ingenol-3-angelate (Ing3A), an active component extracted from Euphorbia peplus, is a PKC agonist with selectivity for PKCδ (10–12). This molecule had been reported to be a strong irritant but have weak tumor promoting activity for mouse skin (13). Ing3A has antitumor activity when applied topically on mouse skin overlying subcutaneous mouse and human tumor xenografts (14). A mechanism whereby topical skin treatment could inhibit subcutaneous tumor growth was unclear, but the compound has been pursued as a potential skin cancer treatment by Peplin, Inc (drug name PEP005) and is now in clinical trials with promising results (15). Topical Ing3A elicits a strong inflammatory response subcutaneously, and the subsequent tumor killing was considered a necrotic response to a marked microenvironmental inflammation (16).

Ing3A is structurally similar to phorbol 12-myristate 13-acetate (PMA), the classical mouse skin tumor promoter, and elicits similar responses as PMA in cultured human cancer cells and isolated cell membranes or lipid vesicles (12). Since the two compounds have not been tested side by side in vivo, and PMA is equally pro-inflammatory, we focused on defining a unique activity of Ing3A that may contribute to its anti-tumor properties. We now show that Ing3A, like PMA, engages the epidermis, but then traverses through the epidermis by absorptive transport as a P-glycoprotein (P-gp) substrate and reaches the subepidermal and subcutaneous environment to engage and damage the tumor vasculature. This property is essential to its anti-tumor activity and provides a novel paradigm for cutaneous drug therapy.

Materials and Methods

Chemicals and antibodies

Ing3A was kindly provided by Peplin, Inc (Brisbane, Queensland, Australia). PMA and cyclosporin A (CsA) were purchased from LC Laboratories (Woburn, MA). Dexamethasone, indomethacin and verapamil were purchased from Sigma (St. Louise, MO). XR9576 (tariquidar) was a gift of Xenova Research(Slough, Berkshire, United Kingdom). Anti-PKCα, δ, ε, PKC-pan, and MEK1/2 antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-phospho-PKC-pan (P-PKC-pan), phospho-PKCδ (P-PKCδ), phospho-Erk1/2 (P-Erk1/2), phospho-MEK1/2 (P-MEK1/2), antibodies were purchased from Cell Signaling Inc (Beverly, MA). Anti-P-gp (clone C219) monoclonal antibody was from Fujirebio Diagnostics (Malvern, PA).

Cell culture

Mouse squamous carcinoma cell line PAM212 (17) was cultured as described previously (18). Mouse melanoma B16 cell line was kindly provided by Dr. Glenn Merlino and cultured in DMEM (Lonza, Frederick, MD). Hifive cells overexpressing human P-gp (19), KB-31 and KB-V1 cells (20), HCT-15 and HCT-15 cells overexpressing human P-gp (21) were cultured as previously described. Immortalized mouse endothelial cell line 3B11 was purchased from ATCC (Manassas, VA) and cultured according to the instruction. The endothelial tube formation assay was carried out by plating and culturing 3B11 cells on BD Matrigel^™ (BD Bioscience). The preformed endothelial tubes were then treated with Ing3A, PMA, or CsA and photographed using an inverted phase contrast microscope (Nikon). EpiDerm Full Thickness 400 human skin equivalent was purchased from MatTek Corporation (Ashland, MA) and cultured on rafts. Cultures were incubated in serum-free DMEM for 24 hours, and Ing3A or PMA was applied directly on the air surface.

Tumor grafts and topical treatment of mice

Athymic NCr-nu/nu and BALB/cAnNCr-nu/nu mice were obtained from NCI-Frederick (Frederick, MD), and SKH1 mice were purchased from Charles River Laboratory (Wilmington, MA). PAM212 tumors were established by subcutaneous injection of 5 million cells on the flank of the BALB/cAnNCr-nu/nu mice. B16 tumors were obtained by subcutaneous injection of 250,000 cells on the flank of the Athymic NCr-nu/nu mice. For topical treatment of the mice, Ing3A, PMA, or other reagents were dissolved in acetone. All mouse experiments were performed under a protocol approved by the National Cancer Institute Animal Care and Use Committee.

Immunoblotting

Cell lysates were collected in lysis buffer from Cell Signaling and then processed for immunoblotting. For detection of mdr1 in mouse skin and liver and human skin equivalents, the frozen tissues were first mechanically disrupted, placed into a buffer containing 25 mM Tris, pH 7.5, 150 mM NaCl, 250 mM sucrose, 1 mM EDTA, 1 mM EGTA, 100 μM PMSF, and then sonicated. The particulate fractions of the tissue extracts were isolated by ultracentrifugation at 100,000× g for one hour then subjected to immunoblotting using anti-P-gp antibody. HRP-conjugated goat anti-rabbit or anti-mouse IgG antibodies were used as secondary antibodies (Cell Signaling). The immunoblots were visualized by Enhanced Chemiluminescence (Pierce, Rockford, IL).

Multiplex bead array assay for cytokines/chemokines

Skin samples were collected from mice after treatment and then homogenized in PBS. The supernatants were collected after centrifugation at 14,000 rpm for 30 min. The level of multiple cytokines/chemokines in the skin extracts was detected using a multiplex kit (mouse cytokines/chemokines 22 plex kit, Linco) according to manufacture=s instruction and carried out using BioPlex (BioRad, St. Louise, MO).

RT-PCR

The expression of mouse TNFα RNA by PAM212 cells was detected by reverse transcription followed by PCR using primers described previously (22). The PCR products were resolved by 2% agarose gel. The expression of the mouse mdr1 transcript Abcba1a (primer set from SuperArray, Frederick, MD) from the skin and liver of nude mice and human ABCB1 from skin equivalents was detected by reverse transcription followed by QC RT-PCR (BioRad), and normalized to the level of GAPDH in each sample.

Histology and immunohistochemistry

Tissue samples were fixed in 4% formalin (Fisher Inc.). Tissue sections were stained with H&E. Immunohistochemistry for myeloperoxidase (MPO, Dako), CD31 (PECAM-1, Santa Cruz), CD3 (Dako), F4/80 (CALTAG), and CD45R/B220 (BD) was performed to detect neutrophils, endothelial cells, T lymphocytes, macrophages, and B lymphocytes, respectively.

Labeling and detection of vasculature in mouse skin

To reveal the blood vessels in mouse skin, control or treated mice were anesthetized and then injected with FITC-lectin (Vector Laboratories) via tail vein. After two minutes, the mice were perfused with 4% paraformaldehyde via the left ventricle (23). 100 μm skin sections were mounted with mounting medium containing DAPI (Vector Laboratories) for visualization by confocal microscopy (Zeiss NLO510). A series of images were taken and projected in three dimensions using ZeissLSM Image Browser.

Photolabeling of P-gp with [¹²⁵I]-Iodoaryazidoprazosin (IAAP)

The interaction of Ing3A with P-gp was assessed by an in vitro photo labeling assay as previously described (24). Briefly, crude membranes from Hifive cells overexpressing human P-gp were incubated with Ing3A, PMA or CsA for 5 min after which 3–5 nM [¹²⁵I]-IAAP (2200 Ci/mmole, PerkinElmer) in 50 mM Tris-HCl (pH 7.5) was added. After exposure to UV light for 10 min at room temperature, the samples were separated by electrophoresis, followed by drying and exposing to Bio-Max MR film (Eastman-Kodak). Alternatively, the dried gel was scanned by STORM 860 phosphor- imager system (Molecular Dynamics) and then quantified by the ImageQuaNT software.

Efflux assay for P-gp mediated transport

Inhibition of P-gp-mediated transport by Ing3A or PMA was determined by flow cytometry using P-gp-expressing KB-V1 cells and the P-gp specific substrate, calcein AM, as described previously (24). To evaluate P-gp mediated transport of Ing3A, cell associated Ing3A was determined using cultured KB-V1 and KB-31 cells incubated with Ing3A or Ing3A plus XR9576. The ethanol extracts of cells were collected and subjected to mass spectrum analysis for quantification of Ing3A.

Results

Ing3A activates PKC in mouse keratinocytes and mouse skin

Ing3A directly binds and activates PKC in vitro (12). A single dose of 25 nmol of Ing3A or PMA applied topically onto the back skin of nude mice induced phosphorylation of PKC isoforms and MEK1/2 (Figure 1A and B), suggesting that Ing3A, like PMA, activates PKC in mouse skin. Figure 1C shows the induction of TNFα by both Ing3A and PMA was similar in mouse skin. SCC in situ is one target for Ing3A therapy, and Ing3A, similar to PMA, induced a rapid and dose dependent phosphorylation of Erk1/2, PKC-pan and PKCδ, downregulated PKCα, δ, and ε, and elevated TNFα transcripts in cultures of the mouse squamous cancer cell line PAM212 (Supplementary Figure 1A–C). Although Ing3A and PMA activate PKC similarly in PAM212 cells, Ing3A is more potent in eliciting those responses. Nevertheless, Ing3A and PMA inhibited the growth of PAM212 cells at equivalent doses (Supplemental Figure 1D). These results indicate that Ing3A, similar to PMA, activates a common PKC pathway in both mouse skin and cultured mouse squamous cancer cells, and both agents have similar direct effects on the cancer cells.

Figure 1.

Ing3A and PMA activate PKC in skin. A. PKC and MEK1/2 phosphorylation in vivo. Protein extracts from skin samples treated with acetone (Control), Ing3A (25 nmol), or PMA (25 nmol) for 30 min were subjected to immunoblotting for P-MEK1/2, P-PKC-pan, and β-actin. Each lane is one sample of skin lysate from a single treated mouse. B. Immunoblots from A were scanned and then quantified by ImageJ. The levels of P-MEK1/2 and P-PKC-pan in each sample were normalized by the amount of β-actin. Data presented are mean±SEM (n=6 for control, n=8 for Ing3A and PMA). C. TNFα levels from protein extracts of skin treated with acetone (Control), Ing3A, or PMA were detected using multiplex analysis as described in Materials and Methods. Mean values from each group were indicated. All the experiments were repeated at least twice (in vivo) and similar results were obtained. ** indicates p≤0.05.

Topical application of Ing3A induces cutaneous inflammation and hemorrhage

The effects of Ing3A and PMA on the skin of nude mice, the strain that has been the recipient of tumor xenografts in preclinical studies, were compared. Three hours after initial treatment, topical administration of Ing3A in either the clinical gel formulation or dissolved in acetone caused skin hemorrhage without actual breakage of the skin, but PMA did not cause visible change in the treated area (Figure 2A). H&E staining showed that Ing3A and PMA induced some common inflammatory changes including edema and inflammatory cell infiltration by 3–6 hours, but only Ing3A induced red cell extravasation in the dermis and subcutis of the skin (Figure 2A). MPO staining revealed the predominant infiltrating cells are neutrophils in both treatment groups (Figure 2B) as shown previously for Ing3A (16). Ing3A also induced subepidermal hemorrhage when applied to immunocompetent mice of the hairless SKH1 strain, indicating that immunodeficiency was not responsible for the unique response. Ing3A did not increase B lymphocyte, T lymphocyte, or macrophage infiltration in the skin of SKH1 mice when examined at 5 hours after treatment (Supplementary Figure 2); however, both Ing3A and PMA activated mast cells similarly (Supplementary Figure 3). To investigate the pathways that mediate Ing3A induced inflammatory responses including the hemorrhage, various pathway specific inhibitors were tested by applying the inhibitors 15 to 30 minutes before the application of Ing3A and examining the nude mice 5 hours later. As shown in Figure 2C, dexamethasone and indomethacin successfully blocked Ing3A-induced neutrophil infiltration but not hemorrhage. Both dexamethasone and indomethacin are known to inhibit PMA induced inflammation and inflammatory mediators (25,26). In contrast, CsA blocked hemorrhage but not neutrophil infiltration in Ing3A treated mice. These results suggest that Ing3A-induced hemorrhage and inflammatory cell infiltration are two independent events that are mediated by distinct pathways. In contrast, Ing3A and PMA-induced inflammatory cell infiltration are similar. A multiplex assay was used to determine the level of 22 cytokines/chemokines in skin extracts collected from mice treated with either Ing3A or PMA. Eighteen out of 21 detected cytokines/chemokines were induced by both Ing3A and PMA treatment, and the induction of each cytokine/chemokine was similar by either inducer (Supplementary Table 1). These results indicate that a unique pattern of Ing3A induced cytokine/chemokine secretion in skin is not observed and could not be responsible for Ing3A induced hemorrhage.

Figure 2.

Topical application of Ing3A induces hemorrhage in nude mice. A. Photos and H&E sections from skin of nude mice topically treated with acetone (Control), Ing3A or PMA (25 nmol) for three hours. Bar: 100 μm or 25 μm (inset). B. H&E and MPO staining of skin samples from Ing3A or PMA treated mice. Bars: 500 μm (row 1) or 50 μm (row 2&3) C. Dexamethasone (Dex, 559 nmol/mouse), indomethacin (Indo, 559 nmol/mouse) or CsA (160 nmol/mouse) were applied on the back of nude mice 30 minutes before the application of Ing3A (25 nmol). The H&E skin sections were collected 5 hrs later. Bars: 500 μm or 25 μm.

Ing3A penetrates beyond the epidermis in skin and damages blood vessels in vivo

We hypothesized that Ing3A-induced hemorrhage may be caused by Ing3A directly targeting the blood vessels in the skin. It has been reported previously that activation of PKC in monolayer cultured endothelial cells induces endothelial cell activation but not cell death (27). To test Ing3A in 3 dimensional culture conditions, mouse 3B11endothelial cells were plated on Matrigel where they formed a network of capillary like structures composed of tubes and branches (Figure 3A). Addition of Ing3A or PMA caused disaggregation of the endothelial tube and outward migration of single endothelial cells. Pretreatment with CsA did not block the effect of Ing3A or PMA in vitro, suggesting that CsA does not directly protect endothelial cells when Ing3A is applied after CsA in vivo. Figure 3B shows FITC-lectin labeled blood vessels from control and PMA treated mice displayed continuous green fluorescence. However, loss of continuous FITC-lectin staining in Ing3A treated mice is detected 40 minutes after initial treatment, well before the appearance of hemorrhage on the treated skin. To better visualize the 3D images of FITC-lectin labeled blood vessels, a surface rendering technique (Imaris 6.1 software) was used. The reconstructed FITC-lectin labeled vessels are markedly damaged by Ing3A treatment (Supplementary Figure 4) leading to discontinuous patterns appearing as “holes”. Damage was detected well into the deep dermis in the large or medium sized blood vessels in the skin. Immunohistochemistry for CD31 expression and electron microscopy (EM) also revealed the loss of the structural integrity of the cutaneous blood vessels after topical Ing3A treatment (Figure 3C). Disruption of basement membrane of venules, endothelial cell activation in arterioles, and loss of junctions between endothelial cells were documented by EM (Figure 3C and data not shown). Disrupted extracellular matrix, red blood cell extravasation and hemolysis in the blood vessels were also observed in skin from Ing3A treated mice. Since Ing3A and PMA had similar effects on endothelial cell tubes in vitro, we asked if PMA would mimic Ing3A if it were delivered more directly to endothelial cells in skin. Hemorrhage was observed in mice injected subcutaneously with either Ing3A or PMA as early as 1–3 hrs (Supplementary Figure 5). These results indicate that skin hemorrhage could also be induced by PMA if it is placed closer to the cutaneous blood vessels. To test if topical Ing3A penetrates more deeply in skin than PMA to reach the blood vessels in the dermis and subcutis, P-Erk1/2 immunohistochemistry was performed on skin sections obtained from nude mice treated topically with either Ing3A or PMA. Figure 4A shows that phosphorylation of Erk1/2 induced by Ing3A and PMA in epidermis occurs as early as 30 minutes. In the dermis and subcutis of skin, Ing3A induced P-Erk1/2 in fibroblasts and endothelial cells to a much greater extent than PMA. Pretreatment with CsA reduced Ing3A-induced Erk1/2 phosphorylation in the fibroblasts and endothelial cells in the dermis and subcutis of skin, but not in epidermis. These results suggest that, via a CsA dependent mechanism, Ing3A penetrates into the subepidermal compartment of skin rapidly and activates PKC in multiple compartments while PMA elicits most responses in the epidermis. Since mouse epidermis is relatively thin, we repeated these experiments by topically treating human skin organotypic raft cultures that reproduce the morphology of human skin including a stratum corneum (Figure 4B). As seen for mouse skin in vivo, P-Erk1/2 was stimulated extensively in the reconstructed epidermis by both PMA and Ing3A, but widespread and deep dermal fibroblast stimulation was seen only with Ing3A.

Figure 3.

Ing3A damages blood vessels in vitro and in vivo. A. The preformed capillary network from 3B11 cells with or without CsA (20 μM) for 30 min were treated with Ing3A or PMA (50 nM). Photos were taken one hour later. Bar: 50 μm. B. FITC-lectin labeled blood vessels in mouse skin from acetone (Control), Ing3A, or PMA treated mice. Blood vessels in 100 μm thick skin sections were visualized using confocal microscopy. The pictures (top or side view) shown here are 3D projections from a series of z-sections. Bar: 50 μm. Damaged areas are outlined in Ing3A treated skin samples. Blue=nuclei C. CD31 immunohistochemistry detects blood vessels in skin from acetone (control) or Ing3A treated mice. Bar: 25 μm. Vessel ultrastructure images are displayed in the lower panels. Bar: 2 μm.

Figure 4.

Induction of P-Erk1/2 by Ing3A and PMA. Nude mice were treated with CsA (200 μg/mouse) for 30 min followed by either Ing3A or PMA (25 nmol each) for additional 30 minutes. Skin sections were subjected to immunohistochemistry for phospho-Erk1/2. Bars: 200 μm (column 1) or 10 μm (column 2&3). B. Human skin equivalents were treated with Ing3A or PMA (25 nmol each) on the air surface for 30 min. Tissue sections were subjected to immunohistochemistry for phospho-Erk1/2. Bars: 200 μm (column 1) or 25 μm (column 2&3).

Ing3A, but not PMA, exhibits anticancer activity for squamous carcinoma and melanoma

PAM212 or B16 cells were subcutaneously injected on the dorsum of nude mice at four sites per mouse following the protocol used for the preclinical studies (14). When tumors were visible, the skin over the tumor sites was treated with vehicle (acetone), Ing3A, or PMA for three consecutive days (the clinical protocol) and the tumor size was measured before and after treatment. As shown in Figure 5A for PAM212 treated with Ing3A, five tumors were either eliminated or reduced while two tumors were initially reduced and then had resurgent growth between the two measurements. In contrast, the majority of acetone control and PMA treated PAM212 tumors maintained their size or showed limited growth during the two measurement intervals on day 13 and 23 while several showed enhanced growth without treatment. Mice grafted with B16 cells were treated three days after injection of tumor cells and tumor measurements taken ten days later due to the rapid growth of B16 tumors. Ing3A treatment eliminated the B16 tumor at five of the eight treated sites (Figure 5B). In contrast, PMA treated tumors grew even larger than the controls. Blocking neutrophil infiltration by dexamethasone had no effect on Ing3A-induced tumor reduction. However, blocking hemorrhage by CsA partially reversed Ing3A-mediated anti-cancer activity (Supplementary Table 2). B16 tumors were also collected one or five days after initial Ing3A treatment. As shown in Figure 5C, one day after Ing3A treatment, severe hemorrhage is observed at the tumor periphery. Tumor cells in the upper part of the tumor are necrotic and blood vessels are engorged with cells (Figure 5C_b, inset B and C) but cells in the deeper tumor mass (Figure 5C, inset D) appear viable. Five days after initial treatment, the entire tumor mass was necrotic (Figure 5C_c; inset E). These results suggest topical Ing3A treatment penetrates through the epidermis into the dermis and subcutaneous region and initially targets blood vessel in the tumor periphery eventually damaging the blood supply to the tumor and causing tumor necrosis.

Figure 5.

Topical treatment with Ing3A for 3 days reduces tumor size. A. Skin overlying subcutaneous tumors derived from PAM212 cells was treated with Ing3A or PMA (25 nmol each). Tumor volume was measured on day 13 and 23 after the initial treatment. Each tumor was normalized to its size at day 0 and expressed as percentage of the initial size. Eight tumors from each group are shown. The dashed line in each panel indicates 100%. B. Skin overlying tumors derived from B16 cells was treated with Ing3A or PMA. Tumor volume was measured 10 days after the initial treatment. The mean value in each group is shown. (** p≤0.05). C. H&E staining of the tissue sections from acetone or Ing3A treated B16 tumors. Tumors were collected 1 or 5 days after the initial treatment. Magnified areas are indicated by boxes. Bars: 500 μm (a, b, c), or 25 μm (inset A, B, C, D, E).

Ing3A, but not PMA, binds to and inhibits P-gp function

The inhibitory effect of CsA on Ing3A mediated anticancer activity is not likely to result from the involvement of the calcineurin/NFAt pathway, since both Ing3A and PMA induced similar levels of IL-2 (Supplementary Table 1). In addition, other phosphatase inhibitors did not block Ing3A-induced hemorrhage (not shown). CsA is known to target and inhibit the drug transport activities of P-gp which is linked to the development of multidrug resistance in cancer cells (28). The expression of Mdr1/P-gp has been reported in keratinocytes of neonatal and adult mouse skin (29,30). We also detected the expression of mouse Mdr1/P-gp protein and transcript (Abcb1a) in skin extracts from adult nude mice and from extracts of human skin organotypic raft cultures (Figure 6A). To confirm the involvement of P-gp in Ing3A-induced hemorrhage, other modulators of P-gp, verapamil (31) and XR9576 (32), were applied to the back of nude mice before the application of Ing3A. As shown in Figure 6B, all three compounds successfully blocked Ing3A-induced hemorrhage. Since CsA, verapamil, and XR9576, are specific but structurally unrelated modulators for P-gp, our results suggested that P-gp contributed to Ing3A-induced hemorrhage in skin. In addition, Ing3A, but not PMA, successfully competed with radiolabeled photoaffinity substrate IAAP for direct binding to P-gp (Figure 6D, top panel). At non-toxic concentration, Ing3A, but not PMA, also reversed the resistance to doxorubicin in human P-gp expressing HCT cells (Figure 6C). Furthermore, Ing3A was more potent than PMA in blocking calcein-AM efflux in human P-gp expressing KB-V1 cells (Figure 6D, middle panel). These results demonstrate that Ing3A binds to and inhibits the function of human P-gp. The cell associated Ing3A was significantly less in P-gp overexpressing KB-V1 cells than its parental line KB-31, and this was reversed by P-gp inhibitor XR9576 (Figure 6D, bottom panel), demonstrating that Ing3A is a transport substrate for P-gp. Based on these findings and the other data presented, we suggest that, as a substrate for P-gp, topically applied Ing3A is transported from the epidermis to the subepidermal cutaneous tissue and causes direct damage on the blood vessels in the dermis and subcutis; in contrast, topically applied PMA is more restricted to the epidermis of skin and does not directly damage the vasculature in the subepidermal compartments.

Figure 6.

Ing3A binds to and is transported by P-gp. A. Detection of P-gp protein (immunoblotting) and mRNA in mouse skin and 2 independent samples of human skin reconstructions. B. P-gp inhibitors prevent Ing3A induced hemorrhage in nude mice. CsA (160 nmol/mouse), verapamil (90 nmol/mouse), and XR9576 (250 nmol/mouse) were applied topically 30 minutes before Ing3A (25 nmol) and photographed 5 hours later. C. Ing3A reverses resistance to doxorubicin (Dox). HCT-15 and P-gp-overexpressing HCT-15 cells were treated with Dox in the presence of either Ing3A or PMA (5 μM). Cell growth was determined 2 days later using the MTT assay. D. Ing3A directly binds to P-gp, blocks P-gp mediated efflux, and is a substrate for P-gp. Top panel: Plasma membranes from Hifive cells overexpressing human P-gp were incubated with various amount of Ing3A, PMA, or CsA and then [¹²⁵I]-IAAP. The incorporation of [¹²⁵I]-IAAP into P-gp was shown by autoradiograph and quantified. Middle panel: Human P-gp overexpressing KB-V1 cells were loaded with calcein-AM, incubated with 5 μM of Ing3A or PMA, and analyzed by flow cytometry. Bottom panel: Ing3A (500 nM) alone or in combination with XR9576 (2.5 μM) was added to KB-V1 and KB-31 cells. After 15 min, ethanol extracts were subjected to mass spectrum analysis for quantification of intracellular Ing3A. Amount of Ing3A in each sample was normalized by total protein. Data presented are mean±SEM (n=3).

Discussion

Inhibition of both tumor blood flow and neoangiogenesis has been considered a promising approach to cancer therapy. Bevacizumab, a humanized monoclonal antibody to VEGF, is currently approved as monotherapy or in conjunction with a traditional chemotherapy regime for colorectal and lung cancers (33). This approach has not been considered for common skin cancers as the cost is prohibitive and surgical therapies are available. The sensitivity of endothelial cells and endothelial tubes to disruption by PKC activation has been documented in the past (27), and Ing3A may provide a topical approach to penetrate into invading tumor tissue with an agent that targets intrinsic tumor blood vessels through the PKC pathway.

A previous study of topical Ing3A and placebo on mouse skin clearly showed the importance of the neutrophilic infiltrate to mediate the necrotic response of the tumor and prevent tumor recurrence (16). PKC-mediated neutrophilic cutaneous infiltrates are the response to the upregulation of specific chemokines from activated keratinocytes (34, 35), and both topical Ing3A and PMA induced similar chemokine/cytokine profiles in treated skin and the inflammatory response was inhibited by dexamethasone and indomethacin, the same anti-inflammatory agents shown previously to block PKC mediated inflammation (26,36,37). Inhibitor studies indicated that blocking the Ing3A-induced inflammation was not by itself sufficient to alter the anti-tumor response whereas blocking the associated hemorrhage reduced anti-tumor activity. Irritant activity of Ing3A likely accounts for the weak tumor promoting response on mouse skin (13), but experiments show that irritant activity predominates in eradicating tumors (13). By restricting topical exposure to lesional areas in the therapeutic setting, the irritant activity should not impede efficacy.

The unique finding that CsA inhibited Ing3A-mediated hemorrhage and reduced the anti-tumor activity without altering the inflammatory response suggested that disruption of tumor vasculature and induction of skin hemorrhage are principal actions for Ing3A to be an effective anti-cancer drug on skin. Both Ing3A and PMA exerted the identical disruptive effect on cultured endothelial tubes that would be consistent with in vivo vascular leakage, but in vivo, activated endothelial cells were only detected in the dermis and subcutis after Ing3A treatment of intact skin. Previous studies on ex vivo endothelial cells confirmed that Ing3A treatment activated endothelial cells in a PKC dependent response, upregulating E-selectin, ICAM1 and IL-8 and leading to neutrophil adhesion and migration (38). In our in vivo study, EM, light microscopy, immunohistochemistry and in vivo lectin labeling indicated that only Ing3A disrupted the vasculature below the epidermis and in grafted tumors. However, PMA could also induce hemorrhage if it was delivered to bypass the epidermis. Together these data indicate a unique property of Ing3A is its ability to penetrate beyond the epidermis and into the dermis and subcutis where the vasculature resides. Our results indicate that Ing3A, but not PMA, binds to and is a transport substrate for P-gp, that P-gp is expressed in the skin, that multiple disparate modulators of P-gp prevent Ing3A induced hemorrhage, and that CsA reduces the antitumor activity of Ing3A. Together we conclude that active transport of Ing3A through the ABC transporter, ABCB1, provides the unique activity of Ing3A, once it has penetrated into the epidermal layer, to be actively transported to the subepidermal compartments. The antitumor benefit of Ing3A thus is provided through an action on deeply embedded tumor cells where it may be directly cytotoxic or growth inhibitory (Supplemental Fig. 1D) (39), inflammation provoking and disruptive of the tumor vasculature through PKC activation of endothelial cells.

P-gp has traditionally been associated with resistance to a variety of natural product anti-cancer drugs due to its function as the drug efflux pump and decrease drug accumulation in cancer cells. However, P-gp is present in the membranes of many normal cell types including epithelial cells where it may function to transport environmental toxins and toxic metabolites out of cells to protect the host. Skin is largely protected from environmental toxins through the formation of the lipid barrier encompassed in the stratum corneum (40). Drug penetration through this barrier and the consequent complex metabolic enzymes of the epidermis complicate this site as a therapeutic portal (41). Mouse and human epidermis and keratinocytes express the mdr1a and mdr1b/MDR1 (ABCB1) gene, and the functional activity of human cutaneous P-gp has been demonstrated by studies of rhodamine 123 transport and inhibition by verapamil and other P-gp inhibitors (30, 42). Furthermore, the expression and activity of P-gp in human skin keratinocytes is increased by inflammatory mediators that would be found in a tumor microenvironment (43). The functioning of cutaneous P-gp in “absorptive” transport, carrying drugs from the skin surface to the dermis and subcutis, was aptly demonstrated by the elegant studies of Ito et al (29). Using genetic approaches, these authors demonstrated that P-gp is essential for the efficient delivery of drugs across the epidermis and into the dermis and hypodermis of mouse skin. Further, they showed by pharmacological experiments that inhibition of P-gp prevented absorptive drug transport. Our studies now indicate that the biological effectiveness of Ing3A, an anticancer agent that can pass the stratum corneum barrier and traffics to and functions in the dermis and hypodermis is through P-gp-mediated absorptive drug transport. While our biological conclusions are drawn from mouse preclinical models and known P-gp inhibitors, our studies shown in Figure 4 indicate that Ing3A penetrates into the dermis of human skin organotypic reconstructions and in Figure 6 indicate that Ing3A is a substrate for human P-gp. Given this mechanism of action for Ing3A and potentially other topical therapeutics, the documented polymorphisms in the human MDR1 (ABCB1) gene (28, 44) could be another determinant of therapeutic efficacy for individual patients. The skin is increasingly considered an advantageous site for drug delivery (45). The prospect of individual variation in absorptive drug transport may need to be investigated to predict therapeutic potential through this portal.