Identification of Rigosertib for the Treatment of Recessive Dystrophic Epidermolysis Bullosa–Associated Squamous Cell Carcinoma

Patients with the devastating inherited skin disease recessive dystrophic epidermolysis bullosa (RDEB) are at significantly increased risk of developing aggressive cutaneous squamous cell carcinoma (SCC), which is the cause of death by age 45 years in 70% of individuals with the severe generalized form of the disease (1, 2). Furthermore, 5-year survival in all RDEB subtypes diagnosed with SCC is close to 0% (2), making RDEB SCC one of the most aggressive forms of this tumor type. Current clinical guidelines offer limited treatment options for RDEB SCC and consist of wide local excision, radiotherapy, and in late stages, limb amputation (3). New approaches to therapy for RDEB SCC are in urgent need.

RDEB is caused by mutations in COL7A1, the gene encoding type VII collagen (4). However, COL7A1 does not behave as a classical tumor suppressor as heterozygous carriers are not at increased risk of skin cancer and COL7A1 has not been identified as a significantly mutated gene in genetic profiling of any tumor type to date. RDEB is characterized by skin fragility, trauma-induced skin blistering, and chronic nonhealing wounds (5), and work by us and others has implicated the tumor microenvironment as a driving mechanism of cancer development (6, 7). Our recent comprehensive genetic characterization of RDEB SCC demonstrates that these tumors are driven by somatic mutation in driver genes that are identical to UV-induced skin SCC and head and neck SCC (HNSCC; ref. 8), and we have consistently been unable to identify genetic mechanisms which explain the aggressive nature of RDEB SCC (9, 10). Our early microarray gene expression profiling failed to significantly differentiate RDEB from skin SCC in cultured keratinocytes (10). However, this work did identify polo-like kinase 1 (PLK1) as a therapeutic target in RDEB SCC. Here, we follow up this finding with a screen of compounds originally designated as PLK1 inhibitors and identify ON-01910, or rigosertib, as a lead candidate for therapy in RDEB SCC.

Cells were isolated from biopsies taken as part of routine surgical or diagnostic procedures. Informed written consent was obtained from each patient or in the case of underaged children from their parents or guardian. Ethical approval for this investigation was obtained from all local ethics committees, and this study was performed in accordance with the Helsinki declaration. All cells were isolated as described (11) and cultured at 37°C in 5% CO₂. Keratinocytes were grown in DMEM (Corning cellgro, Mediatech Inc.)/Ham's F12 medium (3:1), supplemented with 10% FBS (PEAK Serum, Cat PS-FB1), 10 ng/mL of EGF, 10⁻¹⁰ mol/L cholera toxin, 0.4 μg/mL of hydrocortisone, 5 μg/mL of transferrin, 5 μg/mL of insulin, and 13 ng/mL liothyronine. Supplementary Table S1 details all cells used in this study. All cells are routinely tested for Mycoplasma using either a PCR approach or a colorimetric detection assay (MycoAlert, Lonza). SCCRDEB2, SCCRDEB3, SCCRDEB70, RDEB84, and RDEB86 were Mycoplasma positive upon isolation and were treated with Plasmocure (InvivoGen). Only cells which yielded a negative Mycoplasma test were used in the assays described.

Total lysates were quantified using Pierce bicinchoninic assay Protein Assay kit (Fisher Scientific, Cat. # 23225), and 5 to 50 μg of protein were loaded onto SDS-PAGE gel. The signal from Western blot analysis was quantified with Image J. P-CRAF and P-AKT were quantified relative to GAPDH.

Rigosertib (ON 01910.Na) was either purchased from Selleckchem.com or provided by Onconova Therapeutics, Inc. All other drugs were purchased from Selleckchem.com. Note that 2 to 4 × 10⁵ keratinocytes were plated in a 6-well dish. On the following day, the medium was changed with the drug at 1 μmol/L or vehicle control. Medium was left for 48 hours, and cells were lysed with radioimmunoprecipitation assay buffer. Lysate was placed in a centrifuge for 5 minutes at 4°C, and the supernatant was mixed with a 6x Laemmli loading buffer. Samples were boiled for 5 minutes at 95°C before being loaded onto SDS-PAGE gels. Primary antibodies used were PLK1 (Cell Signaling Technology, 208G4), CRAF polyclonal (Cell Signaling Technology, 9422S), P-CRAF (Cell Signaling Technology, 9427S), P-AKT (Cell Signaling Technology, 9271), AKT (Cell Signaling Technology, 9272), and GAPDH (Santa Cruz Biotechnology, 6C5). Resolved proteins were transferred onto nitrocellulose membrane with a BioRad Trans-Blot-Turbo (Bio-Rad), blocked in TBS-0.1% Tween-20 with 5% milk or 5% BSA according to requirements of the primary antibody, and incubated overnight with the primary antibody. After incubation with IgG-HRP–conjugated secondary antibody (Santa Cruz Biotechnology), membrane was incubated with Pierce ECL Western blotting substrate (Fisher Scientific) and exposed to CL-XPosure X-ray film (Fisher Scientific).

RDEB SCC and human control skin biopsies were frozen in optimal cutting temperature (Sakura Finetek USA) and cut at 6 μm on a cryostat. Sections were fixed with 50:50 methanol: acetone mixture for 5 minutes. Slides were rinsed with 1x PBS and permeabilized with PBS/0.1% Tween-20 (Sigma Aldrich) for 5 minutes followed by blocking for 1 hour with PBST/3% BSA (Sigma Aldrich). Primary antibodies were incubated for 1.5 hours at room temperature. Secondary antibody Alexa Fluor 594 goat anti-rabbit (1:800, Molecular Probes) was applied for 1 hour at room temperature. Slides were cover-slipped with hard set 4,6-diamidino-2-phenylindole (DAPI; Vector Labs) and examined by fluorescence microscopy (EVOS FL cell imaging system, ThermoFisher).

Colorimetric assays of mitochondrial dehydrogenase activity were performed using the MTS CellTitre 96 AQueous One Solution Cell Proliferation Assay (Promega) according to the manufacturer's instructions. In each case, 1 to 4 × 10⁴ cells were seeded into replica 96-well plates, and readings were taken at intervals up to a maximum of 96 hours. We have previously shown that an increase in mitochondrial dehydrogenase activity using this assay directly correlates with primary non-SCC and SCC cell number determined using a CASY Model TT cell counter (Roche Diagnostics Ltd.; ref. 10).

For crystal violet assay of cell growth, 1 to 4 × 10⁴ cells were seeded into replica 96-well plates and fixed with 70% ETOH at intervals up to a maximum of 96 hours. After fixation, wells were incubated with a solution of 0.2% crystal violet in 2% ETOH for 1 hour at room temperature followed by three washes with tap water. Plates were allowed to air dry and were then photographed, and 100 μL of 70% ETOH was added to each well for 20 minutes, the plate was agitated to homogenize the solution, and the absorbance at 595 nm was measured using a plate reader (Flexstation 3, Molecular Devices). For photographic quantification, black and white images were produced using the threshold function in Adobe Photoshop C6 (Adobe) and quantified using the measure tool.

5-Bromodeoxyuridine (BrdU; Sigma-Aldrich) was added to cells at 30 mmol/L final concentration for 2 hours. Cells were collected and washed in PBS and then fixed overnight in cold 70% ETOH. RNase A (Sigma) was added at 100 μg/mL, in 5% Tween-20 and PBS for 30 minutes at 37°C. Pepsin (Sigma) was added at 1 mg/mL in 30 mmol/L HCL for 30 minutes at 37°C, and DNA was denatured with 2N HCl containing 0.5% Tween-20 in PBS for 30 minutes at 37°C. An anti-BrdU antibody (Invitrogen) diluted in PBS/0.1% Tween/1% BSA was added for 2 hours followed by a wash with 1% BSA. After passing the cells through a 70 μm filter, propidium iodide (Sigma) was added in the final wash step at a concentration of 25 mg/mL. Samples were analyzed using a FACScan flow cytometer and CellQuest software (Becton Dickinson).

Apoptosis was measured using the Cell Death Detection ELISA^PLUS (ROCHE, Cat. # 11774425001). Briefly, non-SCC keratinocytes were plated at 5 × 10³ cells/well and SCC keratinocytes at 2.5 × 10⁴ cells/well in 96-well plates. On the next day, the cells were treated with the drug (1 μmol/L) for 16 to 24 hours. Cells were lysed for 30 minutes, and 20 μL of the lysate was incubated with 80 μL immunoreagent. After incubation, the reaction was stopped with ABTS solution, and absorbance was read using a plate reader at 405 nm and 490 nm (Flexstation 3, Molecular Devices).

siRNA oligonucleotides targeting PLK1 (FlexiTube siRNA, Qiagen, Cat. #: 1027416), a negative scrambled control (AllStars Negative Control siRNA, Qiagen, Cat. #: SI03650318), and a positive cell death control (AllStars Hs Cell Death siRNA, Qiagen, Cat. #: SI04381048) were used. Cells were seeded in 6-well plates at 2.5 × 10⁵ cells/well and 24 hours later transfected with siRNA (40 nmol/L final concentration) using Lipofectamine 2000 (Invitrogen) diluted in Opti-MEM (Invitrogen) according to the manufacturer's instructions. Cells were left for 16 to 24 hours before being lysed for Western blotting or trypsinized and seeded in 96-well plates at 5 × 10³ cells/well in 100 μL of media. For comparison of siRNA with vinblastine, cells were seeded directly in 96-well plates at 1 × 10⁴ cells/well in 100 μL of media and transfected 24 hours later as described.

All animal experiments were conducted in accordance with UK Home Office (Project License # 60/4054) and Thomas Jefferson IACUC (Protocol # 01719) regulations. For tumorigenicity assays, a suspension of 1 to 4 × 10⁶ tumor cells was mixed with high-concentration Matrigel (Becton Dickinson) and injected subcutaneously into the flanks of adult female SCID Balb/c mice (between 6 and 12 weeks of age and 18–27 g in weight) purchased from The Jackson Laboratories (stock # 001803) or from Taconic (model # CB17SC). Tumors were measured by a caliper, and treatment began when volume reached 100 to 200 mm³. Animals were treated with 20 mg/kg rigosertib dissolved in PBS, and animals were distributed into treatment or vehicle control groups by alternating between treatment (first animal to reach minimum 100 mm³) and vehicle control (second animal). For the experiment shown in Supplementary Fig. S1, 25 mice were injected with 4 × 10⁶ SCCRDEB16 cells, and BI-2536 treatment was included as a positive control for 2 animals where every third animal was allocated to this group. For the experiment shown in Fig. 5, 20 mice were injected with 1 × 10⁶ SCCRDEB106 cells. In all cases, animals where tumors were attached to the rib-cage or growing superficially were sacrificed and not included in the experiment.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) was performed using the TACS2 TdT_Fluor In Situ Apoptosis Detection kit (Trevigen) according to the manufacturer's instructions. Slides were fixed with mounting medium containing DAPI and were viewed under a fluorescence microscope using a 495 nm filter.

Paired two-tailed t test was used for statistical analysis using Prism 8 (GraphPad Software). EC₅₀ values were calculated with nonlinear regression using Prism 8.

We previously identified PLK1 mRNA was increase in 4 RDEB SCC keratinocyte cultures compared with normal primary keratinocyte controls (10). Here, we extend this observation to a total of 10 separate SCC populations and identify PLK1 increased in 10 of 10 fresh-frozen RDEB SCC tissue samples using IHC (Fig. 1).

After confirmation of increased PLK1 in all RDEB SCC assessed, we set out to compare the efficacy of commercially available PLK1 inhibitors for targeting RDEB SCC. We initially assessed the effects of 6 PLK1 inhibitors on cell metabolism using an MTS assay after 48-hour exposure to normal primary keratinocytes (n = 1), RDEB primary fibroblasts (n = 4), and RDEB SCC keratinocytes (n = 4; Supplementary Table S2). These data identified the compound ON-01910, also known as rigosertib, as having the greatest specificity for RDEB SCC keratinocytes when compared with normal primary keratinocytes and RDEB primary fibroblasts (Supplementary Table S2).

Next, these data were confirmed in all 10 separate patient-derived RDEB SCC cell lines using crystal violet uptake by live cells after 48- 72-hour exposure to varying drug concentrations (Fig. 2A and B; Supplementary Fig. S1). These data identified clear specificity for RDEB SCC keratinocytes when compared with primary normal or primary non-SCC RDEB keratinocytes. We also observed that whereas higher concentrations of rigosertib could inhibit the growth of non-SCC primary keratinocytes in culture (both RDEB and non-RDEB, Fig. 2A and B), no obvious signs of cell death were evident while in RDEB SCC keratinocyte populations cell death was widespread (Fig. 2C).

In agreement with our previous data using PLK1 siRNA and the PLK1 inhibitors BI-2536 and GW-843682 (10), as well as published studies of rigosertib (12–14), exposure of RDEB SCC keratinocytes to rigosertib induced cell-cycle arrest and apoptosis (Fig. 3A and B).

Since its identification as an inhibitor of PLK1, rigosertib has been shown to inhibit a variety of kinases, including AKT (12), and recently rigosertib has been shown to compete with activated RAS for binding to CRAF (15). More recently rigosertib was shown to bind to and destabilize microtubules (16), and studies in other malignancies have shown synthetic lethality of PLK1 and microtubule-destabilizing agents (17).

To investigate potential mechanisms of action of rigosertib, we first exposed RDEB SCC keratinocytes to inhibitors of MEK and AKT as well as the microtubule-destabilizing agent vinblastine. These data showed some efficacy of MEK inhibition at lower concentrations but no specificity over normal cells at higher doses, no apparent effect of AKT inhibition and differential selectivity of vinblastine (Fig. 4A). Next, we performed siRNA knockdown of PLK1 and showed inhibition of cell growth in all RDEB SCC populations tested (Fig. 4B). Four of the populations showed only moderate growth inhibition after PLK1 siRNA (45%–75%), and in these cells, we performed siRNA knockdown in the presence of vinblastine but failed to see any overt synergy (Fig. 4C) as has previously been reported in rhabdomyosarcoma (17). Treatment of RDEB SCC keratinocytes with rigosertib inhibited phosphorylation of both AKT and CRAF which was also evident after PLK1 siRNA knockdown (Supplementary Fig. S2).

Finally, we sought to establish the ability of rigosertib to target RDEB SCC keratinocytes in vivo. Tumor xenograft models showed that either local tumor injection (Supplementary Fig. S3) or systemic delivery (Fig. 5) effectively inhibited the growth of two separate RDEB SCC keratinocyte cell lines in vivo. Ki-67 and TUNEL IHC showed that tumors harvested from those animals treated with rigosertib had significantly fewer proliferative tumor cells and significantly more apoptosis than vehicle control-treated animals (Fig. 5D and E). Unlike our experiments with BI-2536 (Supplementary Fig. S3 and ref. 10), we did not see any erythema or ulceration of tissue associated with the injection site when using rigosertib.

SCC is a lethal complication for the majority of patients with the already devastating genetic disease RDEB, and currently, there are no approved or effective therapies for SCC in this patient population (3). Here, we present evidence that the multikinase inhibitor rigosertib effectively induces apoptosis in 10 of 10 separate RDEB SCC populations and targets in vivo tumor growth in xenograft studies, data which support a clinical trial of rigosertib in RDEB SCC. In particular, the lack of significant targeting of nontumor cells by rigosertib (Fig. 2), the lack of any observed toxicity in our animal studies coupled with the overall favorable side-effect profile for this drug as determined by trials in patients with myelodysplastic syndrome (18) supports the notion that even in a patient group such as RDEB, where numerous clinical complications abound as a result of lack of mechanical integrity of the epidermis, rigosertib may be well tolerated.

Previous studies of rigosertib in the context of HNSCC have demonstrated similar efficacy in a subset of tumor cell lines/populations (19). The data presented in our study show remarkable sensitivity of RDEB SCC to rigosertib and suggest that this subtype of SCC may respond to rigosertib treatment more favorably. We have recently increased our current understanding of RDEB SCC through comprehensive genomic characterization of 27 tumors, including 9 of 10 cell lines described here (8), which shows that other than mutation burden and mechanism, the somatic mutation profile in RDEB SCC is no different to HNSCC or UV-induced skin SCC, and that RDEB SCC are most similar to the basal and mesenchymal subtypes of HNSCC at the level of transcriptomics (8). This similarity to subtypes of HNSCC at the level of mRNA expression may begin to offer opportunity for identifying biomarkers of drug sensitivity, and we are actively pursuing this line of inquiry. Another feature identified in RDEB SCC is homogeneity at the level of somatic tumor driver-gene mutation (8), and this may be advantageous in the context of targeted therapies where resistance that often develops is associated with genetic heterogeneity (20).

With regard to mechanism of action of rigosertib in RDEB SCC, we have investigated pathways reported to be targeted by rigosertib and showed that only siRNA knockdown of PLK1 has a similar profile in targeting all RDEB SCC cell lines with little effect to normal cells (Fig. 4). However, the level of PLK1 expression did not seem to show any clear correlation with the sensitivity of individual RDEB SCC cell lines to either PLK1 siRNA or rigosertib, which does not rule out contribution of another pathway when considering RDEB SCC sensitivity to rigosertib over normal cells. Further understanding of this sensitivity may offer precision medicine opportunities for use of rigosertib in subsets of sporadic SCC arising at different anatomical locations, and future work will concentrate on understanding the relationship between apoptosis induction and either PLK1 siRNA or rigosertib exposure in RDEB SCC.

Finally, the data described here have supported the approval of a clinical trial comparing the tolerability and efficacy of rigosertib in RDEB SCC (ClinicalTrials.gov Identifier: NCT03786237), and we aim to analyze patient response in the context of biomarker expression, tumor and patient genotype, as well as sensitivity of patient material in the laboratory.

J.W. Bauer reports receiving speakers bureau honoraria from and is a consultant/advisory board member for Castle Creek Pharma, and holds ownership interest (including patents) in Diaderm. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.P. South

Development of methodology: V.S. Atanasova, C. Pourreyron, M. Farshchian, C.A. Brown IV, J.C. Salas-Alanis, J.W. Bauer, A.P. South

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Pourreyron, M. Farshchian, M. Lawler, C.A. Brown IV, S. Wright, M. Warkala, I. Fuentes, M. Prisco, E. Rashidghamat, C. Has, J.C. Salas-Alanis, F. Palisson, A. Hovnanian, J.A. McGrath, J.E. Mellerio, J.W. Bauer, A.P. South

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V.S. Atanasova, M. Farshchian, M. Lawler, C.A. Brown IV, S. Wright, M. Warkala, I. Fuentes, A.P. South

Writing, review, and/or revision of the manuscript: C. Pourreyron, M. Farshchian, C.A. Brown IV, S. Wright, I. Fuentes, M. Prisco, E. Rashidghamat, C. Has, J.C. Salas-Alanis, A. Hovnanian, J.A. McGrath, J.E. Mellerio, J.W. Bauer, A.P. South

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V.S. Atanasova, S.A. Watt, C. Guttmann-Gruber, J. Piñón Hofbauer, A.P. South

Study supervision: A.P. South

Other (performed experiments): V.S. Atanasova

This work was supported by DEBRA International—funded by DEBRA UK (grants to A.P. South, and to A.P. South, J.W. Bauer, and J.E. Mellerio) and the Office of the Assistant Secretary of Defense for Health Affairs and the Defense Health Agency J9, Research and Development Directorate, through the Congressionally Directed Medical Research Program under Award No. W81XWH-18-1-0382 (to A.P. South). Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the Department of Defense. M. Farshchian is supported by the Sigrid Juse′lius Foundation, Orion Research Foundation, and the Finnish Society of Dermatology. We would like to thank all the patients who contributed to this study, Evan Greenawalt for technical assistance and the Flow Cytometry Shared Resource at the Sidney Kimmel Cancer Center, supported by the NCI, grant 5P30CA056036-17.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.