Growth modulatory effects of fenretinide encompass keratinocyte terminal differentiation: a favorable outcome for oral squamous cell carcinoma chemoprevention

Abstract

Oral squamous cell carcinoma (OSCC) is worldwide health problem associated with high morbidity and mortality. From both the patient and socioeconomic perspectives, prevention of progression of premalignant oral intraepithelial neoplasia (OIN) to OSCC is clearly the preferable outcome. Optimal OSCC chemopreventives possess a variety of attributes including high tolerability, bioavailability, efficacy and preservation of an intact surface epithelium. Terminal differentiation, which directs oral keratinocytes leave the proliferative pool to form protective cornified envelopes, preserves the protective epithelial barrier while concurrently eliminating growth-aberrant keratinocytes. This study employed human premalignant oral keratinocytes and an OSCC cell line to evaluate the differentiation-inducing capacity of the synthetic retinoid, fenretinide (4HPR). Full-thickness oral mucosal explants were evaluated for proof of concept differentiation studies. Results of this study characterize the ability of 4HPR to fulfill all requisite components for keratinocyte differentiation, i.e. nuclear import via binding to cellular RA binding protein-II (molecular modeling), binding to and subsequent activation of retinoic acid nuclear receptors (receptor activation assays), increased expression and translation of genes associated with keratinocyte differentiation [Reverse transcription polymerase chain reaction (RT-PCR), immunoblotting] upregulation of a transglutaminase enzyme essential for cornified envelope formation (transglutaminase 3, functional assay) and augmentation of terminal differentiation in human oral epithelial explants (image-analyses quantified corneocyte desquamation). These data build upon the chemoprevention repertoire of 4HPR that includes function as a small molecule kinase inhibitor and inhibition of essential mechanisms necessary for basement membrane invasion. An upcoming clinical trial, which will assess whether a 4HPR-releasing mucoadhesive patch induces histologic, clinical and molecular regression in OIN lesions, will provide essential clinical insights.

This study employed human premalignant and malignant oral keratinocytes to elucidate mechanisms by which fenretinide engages all of the requisite steps, i.e. nuclear transport, modulation of gene expression and protein translation, corneocyte formation and desquamation necessary for keratinocyte terminal differentiation.

Graphical Abstract

Graphical Abstract.

Introduction

Oral squamous cell carcinoma (OSCC), which arises from the surface epithelium that lines the oral cavity, is a worldwide health problem (1). The primary treatment modality for OSCC is wide surgical excision, which results in loss of structures that are vital for esthetics and function such as speech and eating (2). In addition to its debilitating effects, clinical care costs for persons with OSCC are among the highest for solid cancers (3). Collectively, the loss of persons from the workforce combined with their health care expenses that include both treatment and rehabilitation convey a high negative socioeconomic impact (3). Despite aggressive surgery, frequently combined with intraoperative radiation therapy, many patients develop massive, fatal locoregional recurrences and/or second primary tumors (2). Even those individuals who survive OSCC encounter significant treatment-related morbidities (2). Early intervention at the considerably more manageable premalignant stages (oral intraepithelial neoplasia, OIN) via effective chemoprevention prior to OSCC development is clearly a preferable strategy to avert OSCC’s extensive morbidity and mortality.

Secondary chemoprevention entails the use of natural or synthetic compounds including chemicals to induce regression or prevent progression of a premalignant process (4). At the time Sporn developed the chemoprevention concept, chemicals were regarded as the primary human carcinogens (4). Subsequent studies by Sporn et al., which provided mechanistic data that demonstrated select chemicals convey chemopreventive effects, alleviated these concerns and initiated acceptance of chemicals including drugs in cancer prevention efforts (4).

The vast majority of previous OSCC chemoprevention trials have employed systemic agent delivery and generated disappointing results (5). Additional OIN chemoprevention clinical trials have attempted to repurpose systemically delivered chemotherapeutic drugs through varied systemically administered routes including highly inconvenient intravenous infusions, which are highly inconvenient for the non-cancer patient (6–8). Commensurate with their cytotoxic mechanisms of action, these drugs induced significant side effects including interstitial lung disease, bradycardia, pleural effusions, colitis, hepatitis and renal failure (6–8). Systemic delivery limitations, which include drug-related systemic toxicities and the inability to achieve therapeutic levels at the target site, prompted our laboratory to develop formulations amenable for targeted delivery directly to the lesional site. Notably, local delivery formulations provide a pharmacologic advantage via delivery of chemopreventive-relevant levels to the treatment site without deleterious drug-related systemic effects (9). Furthermore, as the oral cavity is directly visible, OIN lesions are amenable to direct agent application by the patient and monitoring of treatment effects by the health care team (10).

The apoptosis-inducing ability of the synthetic vitamin A derivative, 4HPR, is well established (11). Our laboratory has shown 4HPR possesses a wide extent of additional chemopreventive effects including induction of anoikis, high-affinity binding/inactivation of signaling kinases often hijacked during malignant transformation, i.e. FAK, Pyk2, STAT3, c-Src and c-Abl, and Wnt, perturbation of cytoskeletal components necessary for invasion and migration, and inhibition of activation and function of matrix metalloproteinases that are essential for basement membrane invasion (12–15). Based on its kinase inhibitory effects (12–15), 4HPR can be regarded as a small molecule protein inhibitor.

As the oral cavity is the gateway to the respiratory and gastrointestinal tracts from the external environment, an intact surface oral epithelial barrier and preservation of its associated mucosal immunity are integral (16). While apoptosis or anoikis eliminate keratinocytes from the proliferative pool, extensive keratinocyte loss disrupts epithelial integrity via pronounced atrophy or overt ulceration (17). In contrast, terminal differentiation, whereby chemopreventive-treated keratinocytes leave the proliferative pool to form protective cornified envelopes, preserves the protective epithelial biologic barrier while concurrently eliminating growth-aberrant keratinocytes (18).

Similar to apoptosis, keratinocyte terminal differentiation is a complex event that requires an orderly cessation of pro-proliferative processes (18). Vitamin A mediated differentiation is initiated following retinoic acid (RA) binding with several cognate receptors including retinoid acid receptors (RARs) and retinoid acid X receptors (RXRs) within the nucleus (19). Modulation of gene expression ensues, resulting in a coordinated series of events that entail cessation of growth and proliferation, synthesis of cornified envelope substrates, intraepithelial activation of isopeptide protein–protein cross-linking enzymes (transglutaminases), followed by superficial disruption of keratinocyte desmosomes and corneocyte desquamation (20). This current study employed human premalignant oral keratinocyte lines and an OSCC cell line to evaluate the differentiation-inducing capacity of 4HPR. Full-thickness oral mucosal explants were evaluated for proof of concept local delivery studies. Collectively, these studies tested the hypothesis that 4HPR has the capacity to function as a terminal differentiating agent in human oral keratinocytes.

Materials and methods

Cell culture, cell line validation and characterization by immunoblotting

A continuum of human oral epithelial cell lines comprised of immortalized human oral keratinocytes (EPI), immortalized human oral keratinocytes that have undergone the epithelial–mesenchymal transition (EPI-EMT) and a low population doubling cell line from a stage III OSCC (JSCC23) were used for these experiments. The EPI and EPI-EMT are premalignant oral keratinocyte cell lines that were immortalized via transduction with the E6/E7 oncogenes of human papillomavirus (HPV) type 16 (generous gifts from Dr Walee Chamulitrat, Deutsches Krebsforschungszentrum, Department of Applied Tumorvirology, Heidelberg/Germany) (21). Following transduction, the EPI-EMT cell line was additionally treated with chronic ethanol exposure (1). The JSCC23 cells were isolated from a stage III OSCC of tongue. The original tumor exhibited perineural invasion and positive ipsilateral nodes (2 of 58 evaluated). The JSCC23 cells used in this experiment are low-passage, population doubling level ~11. EPI, EPI-EMT and JSCC23 cells were cultured in Advanced DMEM supplemented with 1× Glutamax and 5% heat-inactivated FBS (GIBCO; Life Technologies; ‘complete’ medium). The EPI, EPI-EMT and JSCC23 cell lines were authenticated via short tandem repeats profiling analyses at the Genetic Resources Core Facility (Johns Hopkins University, Baltimore, MD). Western immunoblotting was conducted to evaluate for transduction stability and key intermediate filaments (keratins and vimentin). Fifty micrograms of EPI, EPI-EMT and JSCC23 crude cell lysates were resolved on an SDS-PAGE gel, transferred to a nitrocellulose membrane and probed using antibodies to HPV16 E6 (SC-460, 1/200, Santa Cruz Biotechnology, Dallas, TX), E7 (SC-6981, 1/200), Pancytokeratin proteins (SC-8018, 1/200) and vimentin (ab92547, 1/1000, Abcam, Cambridge, MA).

Comparison of EPI and EPI-EMT directed invasion at baseline and following treatment with 4HPR and all-trans-retinoic acid

EPI and EPI-EMT cells were cultured in advanced DMEM supplemented with 1× Glutamax and 5% heat-inactivated FBS (GIBCO; Life Technologies; ‘complete’ medium). Log growth cells received the following treatments: (i) control 0.01% DMSO; (ii) 1 µM 4HPR; (iii) 1 µM all-trans-retinoic acid (ATRA) for 6 days, with fresh treatment and medium every 48 h (EPI n = 4; EPI-EMT n = 5). To optimize cell responsiveness to chemoattractants, 24 h prior to the invasion assay, cells were cultured in BASE medium (same as above without serum) with the same treatments. Cells were then seeded onto ECMatrixTM-coated microporous polycarbonate membrane (Millipore QCM™ 96-well Invasion Assay, Massachusetts). The lower chamber contained conditioned medium JSCC3 (previously determined to be highly chemoattractant to OSCC cells (2) was used as the chemoattractant). After 18 h of invasion (37°C, 5% CO₂), cell detachment buffer was used to dislodge successfully invaded cells. Cells were subsequently lysed and detected by CyQuant GR dye. Invaded cell number was determined relative to the fluorescently labeled, cell-line-specific standard curve by running a fluorescent cell standard curve. Fluorescence was assessed using a FLUOstar Omega (BMG LABTECH) fluorescence plate reader using 480/520 nm filter set (12).

Molecular modeling to assess feasibility of 4HPR nuclear translocation and binding to all isoforms of RAR, RXRs and peroxisome proliferator-activated receptors (22–29)

The protein structures for the proteins mentioned above were obtained from the Protein Databank. The crystal structures for CRABP-II associated with ATRA and a synthetic retinoid (1CBS), were obtained from National Center for Biotechnology Information (NCBI). In the event that a protein structure was incomplete, SwissModel was used to homology model the protein to fill in the missing structure. The protein structures were cleaned and optimized with Yasara using the default clean and minimization algorithms. All ligands were constructed in Spartan18 and minimized using MMFF. All relevant protein structures were used to provide alternate protein conformations at the binding site. The optimized protein structure and ligands were docked using AutoDock Vina¹¹ using an exhaustiveness of 500. Each calculation was repeated three times to ensure a thorough exploration of the binding site. Calculated binding free energies were used to determine a binding affinity (K_a) and dissociation constant (K_d) to compare to experimental data: $\mathrm{\Delta }G=\mathrm{R}\mathrm{T}\mathrm{l}\mathrm{n}\left(K_{\mathrm{a}}\right)$ or $K_{\mathrm{a}}={\mathrm{e}}^{-\left(\frac{\mathrm{\Delta }G}{\mathrm{R}\mathrm{T}}\right)}$ and $K_{\mathrm{d}}=\frac{1}{K_{\mathrm{a}}}$.

Determination if 4HPR binding activates isoforms of RARs, RXRs and PPARs

These functional activation assays were conducted using Indigo Biosciences (State College, PA) receptor activation kits. This system employs Chinese hamster ovary cells (CHO) that express high levels of human RAR (IB02301-32P), RXR (IB00831-32P) and peroxisome proliferator-activated receptors (PPAR) (IB00101-32) receptors, respectively, with distinct isoforms. As these cells possess a luciferase reporter gene, quantification of luciferase expression provides a surrogate, albeit intensified and measure of receptor activity in treated cells. Following a 24 h incubation the reporter cells were then treated for 24 h with 1 µM of 4HPR, 4-oxo-HPR, ATRA (n = 8) or 9-cis-RA (n = 8 for every group). A positive control receptor-activating agonist was included with every assay kit.

Impact of vitamin A and derivatives on cell growth

EPI, EPI-EMT and JSCC23 cells were seeded into plastic 6-well plates (Corning) at 5 × 10⁴ cells/well and allowed to adhere for 24 h (n = 3 for every group). Cells were treated with 4HPR, 4-oxo-HPR, ATRA, 9-cis-RA at doses of 1 and 2.5 µM, with fresh medium and RA derivatives provided every other day. The effects of retinoid acid derivatives on viable cell number were determined over a 6-day time course. 0.05% trypsin has been used to achieve a single-cell suspension, the cell numbers were counted using BioRad TC20 Automated Cell Counter (BioRad, Hercules, CA).

Effects of 4HPR on keratinocyte differentiation gene expression and translation

Following 48 h treatment with 1 µM 4HPR, RNA was isolated using PureLink RNA kit (Thermo Fisher Scientific, Waltham, MA), converted to cDNA (Superscript IV First-Strand Synthesis System, Thermo Fisher Scientific) and RT-PCR conducted (Power SYBR Green PCR Master Mix, 40 cycles, QuantStudio3 Real-Time PCR Systems, Thermo Fisher Scientific) using primer pairs designed. The RT-PCR primer sequences are contained in Supplementary Table 1, available at Carcinogenesis online. Target gene expression levels were calculated as Cts after subtracting housekeeping gene β-actin Ct values. Every experiment was repeated three times (n = 3). Subsequent protein assessment consisted of time course studies (harvests at 24, 48, 96 and 144 h) with EPI, EPI-EMT and JSCC23 cells treated with 1 μM of fenritinide (fresh medium + drug q 48 h) were conducted. The cells were solubilized in RIPA buffer (Thermo Fisher Scientific) with proteinase inhibitors (Thermo Fisher Scientific). Twenty to 50 ng crude cell lysates were resolved on an SDS-PAGE gel and transferred to a nitrocellulose membrane. The targeted proteins were individually probed by a validated specific first antibody and then the signal intensities were detected by fluorescence-conjugated secondary antibody in the immune complex with a computer-aided reader (Biorad, Hercules, CA). Target protein antibodies manufacturer: Santa Cruz Biotechnology (SC, Dallas, TX): CREB1 (SC-240, 1/200) and GAPDH (SC-47724, 1/4000). Cell Signaling Biotechnology (Danvers, MA): RARa (62294, 1/1000), RARg (8965, 1/1000), RXRa (3085, 1/1000), RXRb (8715, 1/1000), RXRg (5629, 1/1000) and FABP5 (39926, 1/1000). Abcam (Waltham, MA): Loricrin (ab137533, 1/1000), Involucrin (ab53112, 1/1000), RARb (ab53161, 1/700) and Vimentin (ab92547, 1/1000). Proteintech (Rosemont, IL): TGM1 (67004, 1/2000). Thermo Fisher Scientific-Invitrogen: CYP26A1 (PA5-115080, 1/1000).

Evaluation of the effects of 4HPR on functional activity of transglutaminase 1 and transglutaminase 3 enzymes

EPI, EPI-EMT and JSCC23 cells were seeded onto 6-well plates in complete medium, treatment with 1 μM 4HPR for 4 days, fresh medium + 4HPR every other day. Cells were then lysed (Biovision, Milpitas, CA) and immunoprecipitation using transglutaminase 1 (TGM1) and transglutaminase 3 (TGM3) antibodies (Proteintech Rosemont, IL, Invitrogen, Rockford, IL, respectively) and protein A/G bead separation (Pierce, Thermo Fisher Scientific) was conducted. The functional activities of the TGM-enriched samples were then analyzed using the Zedira transglutaminase fluorogenic assay (Darmstadt, Germany, #T036) in accordance with manufacturer’s instructions using a fluorescence spectrophotometer (BMG Labtech, Germany). The kinetic assays for the EPI and EPI-EMT cell line matched controls (no 4HPR treatment) and 4HPR treated samples were conducted concurrently (n = 3). The JSCC23 concurrently analyzed control and treated samples were run separately from the EPI-EPI-EMT assays.

Effect of 4HPR on extent of keratinization in 3-D in full-thickness human oral mucosal explants

Studies to assess the effects of 4HPR on surface epithelial differentiation, were conducted using full-thickness human oral tissue explants (MatTek EpiOral, Ashland, MA). Tissue explants were treated with 1 µM 4HPR for 48 h incubation at 37°C, 5% CO₂. Since the method of tissue processing can impact the ability to conduct subsequent analyses, we opted to use two methods of tissue fixation. Following incubation, half of the explant tissues were removed, placed in cryomolds, O.C.T. compound (Thermo Fisher Scientific, Waltham, MA) added, then snap frozen with liquid nitrogen-cooled isopentene. The frozen OCT samples were sectioned via a cryostat. The remaining half of the tissue explants were placed in neutral buffered formalin for 24 h followed by routine histologic processing and tissue sectioning. The differentiation proteins assessed by immunohistochemistry were: involucrin, loricrin and TGM1, using the antibodies described in the immunoblotting methods section. Ki-67 immunohistochemical staining (Abcam, Waltham, MA) was used to assess ongoing cell proliferation. Photomicrographs were captured using the Leica DMi8 microscope and Leica ICC50W camera (Leica, Wetzlar, Germany). The extent of explant differentiation was determined by conduction of a complete scan and image capture of the H&E stained slides [Leica Dmi8 microscope with Leica Application Suite X (Buffalo Grove, IL)] followed by quantitative image analysis of the desquamated keratinocytes [ImagePro software (Media Cybernetics, Rockville, MD)].

Statistical analyses

The capacity of the EPI and EPI-EMT cells to invade a synthetic basement membrane was evaluated using a two-tailed Mann–Whitney U-test. A Kruskal–Wallis follow by a Dunn’s multiple comparison post hoc test was employed to evaluate the capacity of RA and derivatives to affect synthetic basement membrane invasion in premalignant oral keratinocytes, activate RA nuclear receptors (isoforms of RARs, RXRs and PPARs), modulate cell proliferation and affect terminal differentiation in oral epithelial explants. The impact of 4HPR on TGM1 and TGM3 functional activity was analyzed using a two-tailed, non-paired Wilcoxon rank-sum test. For all statistical analyses, a P value of <0.05 was considered statistically significant.

Ethics statement

No human subjects were included in this study.

Results

Cell characterization studies confirm the presence of cytokeratin and E6 and/or E7 in all cell lines

Phase contrast images of log growth cultures 100 X image (Nikon DS-Ri1 color digital microscope camera) revealed phenotypic differences (Figure 1A). The EPI cell line exhibits a uniform nuclear appearance and a relatively higher amount of cytoplasm. In contrast, the EPI-EMT cultures show a more spindly, angular phenotype. Finally, the JSCC23 cell line, which was derived from a primary OSCC with lymph node metastases, demonstrates higher nuclear-to-cytoplasmic ratios, greater nuclear and cellular pleomorphism and loss of contact inhibition. All three cell lines contained cytokeratin and variable levels of the oncogenic HPV16 proteins E6 or E7 (Figure 1B). The EPI-EMT cells appeared more spindled and contained appreciably higher levels of the intermediate filament, vimentin; findings suggestive of cells undergoing the epithelial–mesenchymal transition (EMT). This microscopic appearance was supported by the EPI-EMT cells’ increased capacity to invade a synthetic basement membrane (Figure 1C). Similar to previous findings on human OSCC cell lines (16), 4HPR also significantly inhibited invasion in the EPI-EMT cells, as did ATRA. Neither 4HPR nor ATRA affected EPI invasion.

Figure 1.

Characterization of short tandem repeat-authenticated EPI, EPI-EMT and JSCC23 cell lines. (A) Phase contrast images of log growth cultures 100× image (Nikon DS-Ri1 color digital microscope camera). While all cell lines exhibited high proliferation indices in culture, qualitative differences were noted. The EPI cell line exhibits a uniform nuclear appearance and a relatively higher amount of cytoplasm. In contrast, the EPI-EMT cultures show a more spindly, angular phenotype. Finally, the JSCC23 cell line, which was derived from a primary OSCC with lymph node metastases, demonstrates higher nuclear-to-cytoplasmic ratios, greater nuclear and cellular pleomorphism and loss of contact inhibition. (B) Immunoblotting was used to characterize selective intermediate filament proteins and presence or absence of oncogenic HPV E6 and E7 proteins. Consistent with their epithelial origin, all cell lines expressed pancytokeratin and also showed concurrent vimentin expression. Vimentin levels, however, were highest in the EPI-EMT cells which also possessed the most spindled ‘fibroblast like’ appearance (30). All cell lines expressed at least one oncogenic HPV E6 or E7 protein. The EPI-EMT cells were the exclusive line that expressed both E6 and E7. (C) Collagen type IV simulated basement membrane invasion assays were conducted to assess whether or not the EPI-EMT phenotypic appearance correlated with enhanced invasive capacity. Log growth cells received the following treatments: (i) Control 0.01% DMSO; (ii) 1 µM 4HPR; (iii) 1 µM ATRA for 6 days. To optimize cell responsiveness to chemoattractants, 24 h prior to the invasion assay, cells were cultured in BASE medium (same as above without serum) with the same treatments. Cells were then seeded onto ECMatrixTM-coated microporous polycarbonate membrane (Millipore QCM™ 96-well Invasion Assay, Massachusetts). The lower chamber contained conditioned medium JSCC3. After 18 h of invasion (37°C, 5% CO₂), cell detachment buffer was used to dislodge successfully invaded cells. Cells were subsequently lysed and detected by CyQuant GR dye. Invaded cell number was determined relative to the fluorescently labeled, cell-line-specific standard curve by running a fluorescent cell standard curve. Fluorescence was assessed using a FLUOstar Omega (BMG LABTECH) fluorescence plate reader using 480/520 nm filter set. Assay results confirmed a markedly enhanced invasiveness in the EPI-EMT cell lines and also confirmed the capacity of both 4HPR and ATRA to significantly inhibit invasion (mean + SEM, EPI n = 4, EPI-EMT n = 5, * P < 0.05, Kruskal–Wallis with Dunn’s Multiple Comparisons post hoc test). (D) Molecular modeling studies confirm the capacity for 4HPR to undergo nuclear translocation. In order for retinoids to impact gene expression, transportation to the nucleus is essential. RA and its derivatives employ binding to the cellular RA binding protein, CRABP-II, for successful nuclear importation. Our molecular modeling data confirm that oxidized 4HPR can bind to and function as competitive inhibitors of RA binding the CRABP-II. Superimposition of 4HPR (background, yellow) and 1CBS (foreground, green). The 4HPR structure extends deeper into the protein's binding pocket due to the additional length from the 4-hydroxyphenyl group where if OH is placed in the same location of the CO₂ of the RA.

Molecular modeling studies confirm the capacity for 4HPR nuclear transport and binding to intranuclear RARs

Cellular RA binding protein (CRABP-II) molecular modeling studies confirm the capacity for 4HPR and its oxidized metabolite, 4-oxo-HPR, to bind to the nuclear translocation protein CRABP-II, albeit at lower levels than the endogenous ligand, RA (Figure 1D). Additional molecular modeling assessments provided similar findings, i.e. capacity for both 4HPR and 4-oxo-HPR binding to RARa, RARb, RARg isoforms (Figure 2A and B) and/or the PPARa, PPARb/d and PPARg isoforms (Figure 3A).

Figure 2.

4HPR activates the nuclear transcription regulatory RARs. The capacity for 4HPR to bind (molecular modeling assessment) and activate the RARs was assessed using RA reporter assay system. This system employs human cells that express high levels of RARs and possess a luciferase reporter gene. Quantification of luciferase expression, therefore, provides a surrogate measure of RAR activity in treated cells. (A.i) Molecular modeling studies showed that in all cases, the docking with the endogenous ligand matched the pose from the crystal structure. 4HPR and 4-oxo-4HPR, bind with similar affinities as RA or 9-cis-RA indicating that these compounds will be a competitive inhibitor for RAR, particularly the RARa isoform. Model compounds labeled as 9-cis-RA (green), RA (red), 4-oxo-4HPR (blue) and 4HPR (cyan). (A.ii) Following a 24 h incubation the reporter cells were treated with 1 µM of 4HPR, 4-oxo-HPR, ATRA and 9-cis-RA. Adapalene served as the positive assay control. All agents except 4-oxo-HPR significantly increased RAR receptor activity in all three isoforms, α, β and γ. Mean + SEM, n = 8, *P < 0.05, **P < 0.010. Kruskal–Wallis, Dunns multiple comparison post hoc test. (B.i) Molecular modeling studies to assess 4HPR and it oxidized metabolite 4-oxo-HPR revealed that 4HPR (green) does not bind in the same pocket as the others and has a significantly smaller K_d. Model compounds labeled as 4HPR (green) in the binding pocket with the endogenous ligand 9-cis-RA (pink) and 9-cis-4HPR (Yellow). (B.ii) Cells were handled in an identical fashion as that described for the RAR studies. The ATRA and 9-cis-RA provided strong, positive induction of all three isoforms (RARa, RARb and RARg) of the RXR receptors. In contrast, neither 4HPR nor 4-oxo-HPR demonstrated any inductive effect. Mean + SEM, n = 8, **P < 0.01 Kruskal–Wallis, Dunns multiple comparison post hoc test.

Figure 3.

4HPR does not activate the PPARs nor induce cell proliferation. The capacity of 4HPR to bind (molecular modeling assessment) and activate PPARs was assessed using the PPAR reporter assay system. (A.i) Molecular modeling compounds are labeled as 4HPR (blue) and 4-oxo-HPR (red) in the binding pocket with the endogenous ligand 9-cis-RA (yellow). 4HPR does not bind in the same pocket as the others and has a significantly smaller K_d. Modeling studies indicated that 4HPR interacted with the PPAR binding site on all isoforms, comparable to moderate to high potency PPAR targeted inhibitors, e.g. GW2331. (A.ii) CHO cells were treated with 1 µm 4HPR, 4-oxo-HPR, ATRA and 9-cis-RA for 24 h, followed by assessment of PPAR activation using Indigo reporter assay kit. Despite comparable calculated molecular binding affinities, 9-cis-RA and 4HPR exhibited different effects on PPAR isoform signaling. 9-cis-RA significantly increased receptor activation (P < 0.01) in all three isoforms PPARa, PPARb/d and PPARg whereas 4HPR significantly repressed signaling for the PPARa and PPARg isoforms, (#P < 0.05, respectively). Addition of ATRA also significantly increased the g isoform signaling (P < 0.01). Mean + SEM, n = 8, Kruskal–Wallis with a Dunns multiple comparison post hoc test. (B) The effects of RA derivatives on cell proliferation were determined over a 6-day time course. Cells were treated with 4HPR, 4-oxo-HPR, ATRA and 9-cis-RA at doses of 1 and 2.5 µM, with fresh medium and RA derivatives provided every other day. Cell numbers were determined using BioRad TC20 Automated Cell Counter (BioRad, Hercules, CA). The cell growth curves demonstrated that the EPI-EMT cells displayed the highest growth rate. In none of the cultures, at any retinoid dose, did the addition of retinoids augment cells proliferation. The established apoptosis inducer, 4-oxo-HPR significantly decreased cell numbers in the EPI and EPI-EMT cells at 1 µM. All three cell lines showed significant growth inhibition at day 6 with the addition of 1.0 µM 4-oxo-HPR (P < 0.05) and 2.5 µM 4-oxo-HPR (**P < 0.01) n = 3, Kruskal–Wallis with a Dunns multiple comparison post hoc test.

4HPR activates RARs in reporter cells but does not impact RXRs or PPARs

As depicted in Figure 2, 4HPR binding to RARa, RARb or RARg in reporter cells increased luciferase reporter activity relative to control cells. RAR isoform-specific responsiveness, was noted with greatest 4HPR activation seen in the gamma isoform. ATRA, 9-cis-RA and the positive control adapalene all induced reporter activity. The oxidized 4HPR metabolite, 4-oxo-HPR, did not significantly increase reporter activity. While challenge with 4HPR and 4-oxo-HPR did not affect luciferase activity for any of the RXR isoforms, all three responded following ATRA and 9-cis-RA challenge. 9-cis-RA increased luciferase reporter activity in PPARa, PPARb/d and PPARg, with more modest effects only in PPARg following ATRA challenge. While PPARa, PPARb/d and PPARg exhibited significant increases in reporter activity following 9-cis-RA challenge, introduction of 4HPR induced either no response (PPARb/d) or modest inhibition (PPARa and PPARg) isoforms. Similarly, the oxidized 4HPR metabolite, 4-oxo-HPR, did not affect PPAR relative receptor responsiveness.

4HPR and its oxidized metabolite 4-oxo-HPR do not promote oral keratinocyte proliferation

Regardless of dose, addition of ATRA or 9-cis-RA did not affect proliferation in any of the three cell lines. 4HPR and 4-oxo-HPR, however, modulated proliferation in a cell line and dose-dependent fashion. Regardless of dose (1 or 2.5 µM), 4HPR exhibited negligible effect on either the EPI or JSCC23 cells with modest anti-proliferative effects on the EPI-EMT cells. In contrast, 2.5 µM 4-oxo-HPR appreciably reduced proliferation in all three cell lines (Figure 3B). Agents’ effects on cell proliferation are expressed as fold changes relative to the intragroup starting cell number (5 × 10⁴).

Addition of 4HPR modulates expression and translation of retinoid metabolism and terminal differentiation proteins

An expression analysis focused on genes associated with keratinocyte terminal differentiation, revealed a 1 µM 4HPR 48 h treatment of premalignant oral keratinocytes induced a statistically significant increase in genes associated with terminal differentiation, retinoic metabolism, retinoid signaling and retinoid transport (Figure 4A). Some of the 4HPR upregulated genes are regulated by RA (30). Corresponding immunoblot analyses revealed similar findings, i.e. low dose (1 µM 4HPR, 96 h treatment) also increased translation of proteins essential for keratinocyte terminal differentiation and retinoid metabolism (see Figure 4B, Supplementary Table 2, available at Carcinogenesis online).

Figure 4.

4HPR modulates gene expression and protein translation toward keratinocyte differentiation. (A) Addition of low dose 4HPR (1 µM, 48 h treatment) induced a statistically significant increase in genes associated with terminal differentiation, retinoic metabolism, retinoid signaling and retinoid transport. One of the 4HPR upregulated genes (CYP26A1) is known to be regulated by RA. The fold change in expression was determined in accordance the delta CT calculation (n = 3) (31). (B) Time course studies to assess the impact of low dose 4HPR (1 µM) on cell production of proteins associated with retinoid-associated nuclear translocation, gene expression, metabolism and terminal differentiation were conducted. 4HPR-administrated periods were specified to 1, 2, 4 and 6 days, respectively, to each cell lines for expression assessment; same concentration (volume ratio) of DMSO treatment were served as control (see Supplementary Table 2, available at Carcinogenesis online).

4HPR treatment induces functional activity of an enzyme associated with keratinocyte cornified envelope formation (TGM3)

Functional activity of baseline (no 4HPR control) and 1 µM 4HPR (96 h treatment, fresh medium + 4HPR every 48 h) revealed both the EPI and EPI-EMT control cell lines exhibited functional transglutaminase 1 (TGM1) and transglutaminase 3 (TGM3) enzyme activity, with relatively higher activities noted in the EPI cells for both enzymes. While treatment with 4HPR did not impact TGM1 function in either cell line, statistically significant increases (P < 0.05) were noted following 4HPR treatment in both the EPI and EPI-EMT cells (see Figure 5). While the OSCC JSCC23 cells retained TGM1 and TGM3 function, 4HPR treatment had no impact on functional activity (data not shown).

Figure 5.

4HPR augments functional activity of a key differentiation enzyme, transglutaminase 3. Studies were conducted to assess the effects of low dose 4HPR (1 µM, 96 h, fresh medium and 4HPR every other day) on the enzymatic function of two transglutaminases [TGM1 (A) and TGM3 (B)] that are essential for cornified envelope formation and keratinocyte terminal differentiation. All enzymatic functional assays (Zedira transglutaminase fluorogenic assay, Darmstadt, Germany) were conducted concurrently, using the same 96-well black polystyrene microplate. Our data confirm that both the EPI and EPI-EMT cell lines exhibited basal levels of both TGM1 and TGM3 activity, with greater activity demonstrated by the EPI cells. Treatment with 4HPR resulted in a significant increase in TGM3 function in both the EPI and EPI-EMT cultures. *P < 0.05, n = 3, two-tailed, non-paired Wilcoxon rank-sum test. While the JSCC23 cells also exhibited basal level TGM1 and TGM3 activity, no upregulation was noted with 4HPR treatment (data not shown).

4HPR increases corneocyte formation and desquamation in oral epithelial explants

Explant treatment with 4HPR increased keratinocyte differentiation as evidenced by formation of terminally differentiated keratinocytes undergoing superficial desquamation (see Figure 6A). Although control explants displayed some keratinocyte desquamation, the extent in the control tissues was negligible relative to 4HPR-treated explants (see Figure 6A). Notably, keratin envelope formation and corneocyte desquamation was observed in the 4HPR-treated explants regardless of fixation method (OCT or formalin–paraffin embedded) employed. Extensive, cytosolic staining for TGM1 (see Figure 6B) and involucrin were noted while loricrin staining was less intense. Ki-67 nuclear staining was noted in basal layer keratinocytes of all of the control and 4HPR-treated explants (see Figure 6C).

Figure 6.

4HPR promotes corneocyte formation and desquamation in oral mucosal explants. (A) MatTek EpiOral explant treatment with 1 µM 4HPR increased keratinocyte differentiation as evidenced by formation of terminally differentiated keratinocytes undergoing superficial desquamation. Although control explants displayed some keratinocyte desquamation, the extent in the control tissues was negligible relative to 4HPR-treated explants. Notably, keratin envelope formation and corneocyte desquamation was observed in the 4HPR-treated explants regardless of fixation method (OCT or formalin–paraffin embedded) employed. In addition, stratified squamous intact epithelium remained underlying the desquamated keratinocytes. Microscopic image analyses to quantify desquamated keratinocytes confirmed 4HPR significantly increased keratinocyte terminal differentiation and desquamation explants. (mean ± SD; **P< 0.01; 4HPR bolus delivery, n = 8; no 4HPR control, n = 3; Kruskal–Wallis, Dunns multiple comparison post hoc test). (B) Diffuse cytosolic staining for TGM1 is present in both the 4HPR treated and control explants. Immunohistochemical stains for the other two cytosolic proteins were comparable regardless of 4HPR treatment. (C) Ki-67 nuclear staining (highlighted by arrows), indicative of cell proliferation/explant viability, is observed in some of the non-desquamated basal layer keratinocytes in both the control and 4HPR treated explants.

Discussion

Capacity of RA to induce cellular differentiation was documented over 30 years previously when Huang et al. employed ATRA to induce clinical remission in patients with acute promyelocytic leukemia (32). Despite the time lapse since this discovery, the underlying molecular mechanisms of RA-mediated differentiation have not yet been fully elucidated (33). More recently, investigations expanded to include differentiation properties of RA derivatives and their associated impact on non-hematologic cell types (33–35). With regard to OSCC chemoprevention, the benefits of dysplastic keratinocyte differentiation, i.e. preservation of a epithelial protective barrier with suppression of inappropriate epithelial proliferation are readily apparent. Consequently, identification of agents that induce OIN terminal differentiation combined with a delivery strategy that provides chemopreventive-relevant levels at the treatment site have the potential to transform the OIN chemoprevention paradigm. Results of this current study depict the ability of 4HPR to fulfill all requisite components for keratinocyte differentiation, i.e. nuclear import via binding to CRABP-II, binding to and subsequent activation of RA nuclear receptors, increased expression and translation of genes associated with retinoid transport, function and keratinocyte differentiation, upregulation of a transglutaminase enzyme (TGM3) essential for cornified envelope formation and augmentation of terminal differentiation in human oral epithelial explants. To our knowledge, this study is the first to comprehensively characterize the induction of 4HPR in terminal differentiation in human premalignant oral epithelial cells.

Similar to other hydrophobic retinoids, 4HPR is unstable in an aqueous environment and therefore requires retinoid binding proteins for stability (36). Depending on the location, the 4HPR-retinoid protein binding requirement can occasionally be problematic. During systemic 4HPR administration, the higher binding affinity of 4HPR displaces vitamin A from sera retinol binding protein resulting in nyctalopia (37, 38) Based on these recognized 4HPR-retinol binding protein interactions, the molecular modeling data that demonstrated 4HPR binding to the nuclear import protein CRABP-II was not surprising.

Subsequent to intranuclear transport, in order to impact gene expression, 4HPR must bind to and activate RA gene expression-enabling RARs. Molecular modeling studies indicated 4HPR and its 9-cis derivative have comparable binding affinities as RA or 9-cis-RA to all three retinoic acid receptors (RARa, RARb and RARg). The subsequent nuclear receptor functional assays, which demonstrated 4HPR has ability to activate all RAR isoforms in reporter cells, corroborated the molecular modeling studies.

The inability of 4HPR oxidized metabolite, 4-oxo-HPR, to activate RARs appears consistent with its predilection for redox chemistry and the ability of 4-oxo-HPR to induce apoptosis-as opposed to differentiation in 4HPR-resistant cells (39). While 4HPR activated all three retinoic acid receptors (RARa, RARb and RARg), no 4HPR-mediated activation of any RXR or PPAR isoforms was noted.

The role of PPARs in carcinogenesis, which reflects numerous components including the specific PPAR agonist, PPAR isoform(s) activated and the contributions of the tumor microenvironment, is extremely complex (40, 41). Research discrepancies have arisen over PPAR responsiveness, or lack thereof, to retinoid challenge (41–43). Studies by Muhkerjee et al., which demonstrated that RXR-selective ligands, labeled rexinoids, concurrently activated PPARa in vivo, stimulated interest in PPAR agonists and their physiological impact (30). Subsequent work by Schug et al. reported that RA concurrently activated both RARa and PPARb/d in immortalized skin keratinocytes (HaCaT keratinoyctes) and the ensuing PPARb/d activation also enhanced cell survival (43). It was hypothesized that the relative differences in retinoic nuclear receptor activation reflected variations in the nuclear transport proteins CRABP-II and FABP5 relative ratios (43). In our study, 9-cis-RA, but no other ligand, activated the PPARa, PPARb/d and PPARg receptors. These results may reflect the relative greater potency of 9-cis-RA relative to all-trans-RA with regard to nuclear RA receptor activation (44) and the chemical structural differences between 4HPR relative to RA. Our accompanying cell growth studies did not show a 9-cis-RA or 4HPR mediated increase in cell proliferation. Studies by Borland et al. evaluated RA’s ability, applied at comparable levels relative to Schug et al., to function as a PPAR agonist via concurrent evaluation of proliferation, apoptosis and gene expression in HaCaT cells. Similar to our findings, these investigators did not identify a prosurvival or a PPARb/d activation function for RA (41). Rieck et al.’s studies, which evaluated the abilities of selective ligands and ATRA to activate PPARs, also failed to show any ATRA-mediated PPAR activation (42). Consistent with its variable functions described previously, PPAR activation can either promote or inhibit tumorigenesis (40). Consequently, PPAR agonists and antagonists have been evaluated for their therapeutic benefits for multiple human diseases including cancer prevention (45). The capacity of 4HPR to reduce PPARa or PPARg basal level signaling is enigmatic and may reflect 4HPR-mediated signal quenching. The accompanying cell proliferation studies, which showed 4HPR did not enhance growth in any of the cell lines, in conjunction with the absence of 4HPR-initiated PPAR and RXR receptor activation, imply the growth regulatory effects of 4HPR are primary mediated through RAR signaling in premalignant oral keratinocytes.

Confirmation of the ability of 4HPR to modulate gene expression and protein translation was essential to establish a growth regulatory, differentiation-inducing capacity. Results of the expression analyses, which focused on genes established to be regulated by RA (46) and/or those essential for RA signaling or differentiation confirmed 4HPR induces RA-like gene modulation. Not only were genes central for terminal differentiation, e.g. involucrin, loricrin, transglutaminases significantly upregulated, but additional genes associated with retinoid signaling, intranuclear transport and a key cytochrome P450 (CYP26A1) for which retinoids serve as both substrates and potent inducers (47) also showed increased expression. While increases in gene expression frequently correlate with increased protein production, this is not always the situation (48). The accompanying immunoblot analyses, which showed all three cell lines treated with 4HPR exhibited increases in retinoid metabolism and differentiation-associated proteins, were commensurate with the expression analyses.

All of the cell lines evaluated exhibited basal level activity of both TGM1 and TGM3. As calcium is an established inducer of keratinocyte differentiation, retention of basal transglutaminase functions may reflect cell culture in calcium-containing bovine serum. Consistent with their epithelial predominant phenotype, the EPI cells exhibited higher basal and induced TGM activities relative to the vimentin-expressing more ‘epithelial myoepithelial’-like EPI-EMT cells (49). Unlike the premalignant keratinocytes, the JSCC23 cells did not demonstrate a 4HPR-mediated TGM induction. Due to the progressed phase of molecular disruptions in overt carcinoma, the lack of TGM induction in the JSCC23 cells was not surprising.

Terminal differentiation, which entails cessation of DNA synthesis in conjunction with transglutaminase-mediated cross-linkage of specific proteins, degradation of the keratinocyte desmosomes in the outermost keratinocytes and outermost corneocyte shedding via desquamation, occurs in healthy stratified squamous epithelium (21, 50). Results of the human oral explant studies, which showed 4HPR effectively induced keratinocyte terminal differentiation and corneocyte desquamation, recapitulate features seen in healthy, stratified squamous oral epithelium (50). Furthermore, the retention of Ki-67 nuclear labeling confirmed ongoing cell proliferation and that the desquamated keratinocytes did not reflect the loss of explant viability. These data align well with previous in vivo studies from our laboratory which evaluated 4HPR local delivery from a mucoadhesive patch to rabbit oral mucosa (10 days, 30 min application qd) (51). Oral mucosal samples with final 4HPR levels of <5 µM demonstrated evidence of differentiation as noted in significant increases in keratinocyte-specific transglutaminase (TGM1) intraepithelial protein and reduced proliferation (Ki-67 labeling) while 4HPR levels >5 µM significantly increased apoptosis. All 4HPR patch-treated oral mucosal samples, regardless of final 4HPR levels achieved, exhibited modestly increased surface keratin thickness (51). Similar to our current findings, an in vitro study by Chen et al., that evaluated low-level 4HPR (1 µM) treatment, demonstrated induced neuronal differentiation as confirmed by select protein expression and cell phenotypic changes in ARPE-19 human retinal epithelial cells (34).

Elucidation of an agent’s mechanisms of action is central for rational drug selection and optimization of its clinical impact. Topical application of retinoids has been and continues to be used widely used in the management of dermatologic conditions ranging from acne to basal cell carcinoma [reviewed in Ref. (42). In contrast, retinoid use for intraoral clinical applications was discontinued following unsuccessful attempts at OSCC chemoprevention by systemically administered 4HPR (52). Due to appreciable inactivation during first-pass metabolism and the resulting inability to even achieve chemopreventive-relevant 4HPR sera levels (52), it is apparent that systemic administration of 4HPR is not a viable strategy for intraoral applications. The greater level of success with retinoid-based dermatologic formulations probably reflects the site of application, i.e. a dry cutaneous surface. In contrast, the saliva-rich oral cavity provides challenges for the local delivery of hydrophobic compounds like 4HPR. Innovative pharmaceutical chemistry, however, enables development of formulations that incorporate 4HPR stabilization to provide a bioavailable local delivery strategy for OIN chemoprevention (16, 51, 53). Our team is finalizing preparations for an upcoming clinical trial to assess the capacity of a modified 4HPR-releasing mucoadhesive patch to induce OIN regression. The primary parameters to be assessed include histologic grade, loss of heterozygosity at putative tumor suppressor gene loci and clinical size and appearance. Additional studies to determine proliferation and apoptotic indices, as determined by Ki-67 and caspase-3 labeling, will also be conducted. Provided the established pharmacologic advantage conveyed by local delivery and the multiple chemopreventive mechanisms of action of 4HPR (13–16, 51), we anticipate treatment-associated increases in keratinocyte differentiation as reflected by reductions in both keratinocyte proliferation and apoptosis.

Supplementary material

Supplementary data are available at Carcinogenesis online.

Glossary

Abbreviations

EMT

epithelial–mesenchymal transition

OSCC

oral squamous cell carcinoma

OIN

oral intraepithelial neoplasia

RA

retinoic acid

RARs

retinoid acid receptors

RXRs

retinoid acid X receptors

Funding

National Institute of Health-National Cancer Institute (R01CA227273) to S.R.M. and (R01CA258757) to S.R.M., J.L.

Conflict of Interest Statement: None declared.

Authors’ contributions

D.W.: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology. P.P.: Data curation, formal analysis, validation, investigation, visualization, methodology. F.S.: Data curation, formal analysis, validation, investigation, visualization, methodology. C.R.S.: software, investigation, visualization, methodology. A.C.: investigation, visualization. J.L.: project administration, supervision, funding acquisition, validation, visualization. S.R.M.: Conceptualization, resources, project administration, supervision, formal analysis, funding acquisition, validation, visualization, writing–original draft, project administration, and writing–review and final editing.

Data Availability

All of the data underlying this manuscript are available in this articles and in its online supplementary material.