Porcupine expression promotes the progression of oral carcinogenesis

Highlights

• •-PORCN is upregulated in oral dysplasia and OSCC.
• •-Oral dysplasia and OSCC display higher Wnt/β-catenin activity.
• •-Progression from oral dysplasia to OSCC is prevented by the PORCN inhibitor LGK974.
• •-The use of LGK974 reduces Wnt/β-catenin activity in oral dysplasia and OSCC.

Abstract

Oral squamous cell carcinoma (OSCC) is the most common type of oral cancer, which is usually preceded by a potentially malignant disorder histologically diagnosed as dysplasia. We and others have provided evidence for the pro-carcinogenic role of the Wnt/β-catenin pathway in this context, in which Wnt ligands stabilize and allow relocalization of β-catenin to the nucleus for transcription of pro-survival and pro-proliferation genes. However, the contribution of Porcupine (PORCN), an O-acyltransferase that catalyzes the palmitoylation of Wnt ligands, to OSCC carcinogenesis is not known. Moreover, the effectiveness of LGK974, a novel PORCN inhibitor remains to be elucidated. By using different ex vivo, in vivo and in vitro OSCC carcinogenesis models, we show that PORCN expression is significantly increased in high-grade dysplasia as well as moderately/poorly- differentiated OSCC. Consistent with these observations, expression of key proteins involved in the Wnt/β-catenin pathway are elevated as well. Importantly, the treatment with LGK974, a chemical PORCN inhibitor, reduced the number and size of oral lesions in mice treated with 4-Nitroquinoline 1-oxide (4NQO), a tobacco smoke surrogate. These results highlight the role of PORCN during OSCC carcinogenesis.

Introduction

Oral cancer is a global burden affecting nearly 380,000 persons per year [1]. Among oral cancers, the Oral Squamous Cell Carcinoma (OSCC) represents 90 % of the cases [2]. During OSCC carcinogenesis, continuous exposure to risk factors, including tobacco smoking or alcohol consumption, provoke adaptation mechanisms in the oral mucosa [2]. Some of these alterations are macroscopically observed as oral lesions, collectively called Oral Potentially Malignant Disorders (OPMDs) [3]. OPMDs include leukoplakia, erythroplakia and lichen planus, among others [3]. Histologically, nearly 80 % of OPMDs are diagnosed as oral dysplasia, characterized by epithelial cell atypia and increased patchy proliferation [4]. The probability that an OPMD with oral dysplasia evolves to OSCC is around 12 % [3].

We have previously found that the Wnt/β-catenin is important for OSCC carcinogenesis. Activation of this pathway leads to cytoplasmic stabilization followed by nuclear translocation of β-catenin, for transcription of proliferation and death resistance genes, i.e. cyclin D1 (CCND1), baculoviral IAP repeat containing 5 (BIRC5; Survivin), among others [5]. During OPMDs with oral dysplasia and OSCC, higher levels of Wnt3a, an extracellular ligand that promotes the sequestration of the β-catenin destruction complex are observed [6]. Sequestration of the destruction complex, which includes AXIN1, APC regulator of WNT signaling pathway (APC), dishevelled segment polarity protein 1 (DVL1) and the Ser/Thr kinases glycogen synthase kinase 3 beta (GSK3B) and casein kinase 1 (CK1), phosphorylates β-catenin at Ser33/Ser37/Thr41, promotes the stabilization of total and transcriptionally active (non-phosphorylated) β-catenin [5]. β-catenin expression is significantly correlated with OSCC clinicopathological features, including T-status, N-status, clinical stage and histological grade [7]. Loss of β-catenin membrane localization is associated with T-status, N-status and clinical stage, while nuclear β-catenin expression is associated with clinical stage and histological grade [7].

Despite all this, it is not clear how Wnt3a levels are upregulated. We have recently demonstrated that chemical inhibition of Porcupine (PORCN) with a compound known as C-59 effectively attenuates OSCC carcinogenesis by reducing Wnt/β-catenin activity [8]. PORCN is an O-acyltransferase that catalyzes the palmitoylation of Wnt ligands required for their maturation and extracellular transport [9]. Whether PORCN levels in OPMDs with oral dysplasia and OSCC are associated with dysregulation of the Wnt/β-catenin pathway is not known yet.

Here we used multiple approaches to address this question, including retrospective biopsies from patients with OPMDs with low- and high- grade oral dysplasia, as well as samples from patients with well- and moderately/poorly- differentiated OSCCs (ex vivo OSCC carcinogenesis model); tumors grown in mice from dysplastic oral cell lines (DOK) transformed with Benzo[a]pyrene and dimethylbenz[a]anthracene (B(a)P/DMBA; DOK-TC), both tobacco smoke-related carcinogens (in vivo OSCC tumorigenesis); cultured DOK and DOK-TC cells (in vitro OSCC carcinogenesis model); and finally oral lesions from mice challenged with 4NQO, a carcinogen that resembles the effects of tobacco smoking (in vivo OSCC carcinogenesis model). In all these models, we evaluated the levels of PORCN and the activation status of the Wnt/β-catenin pathway. We found that PORCN expression is upregulated in high-grade dysplasia as well as well- and moderately/poorly- differentiated OSCC. Similar upregulation of PORCN was observed in DOK-TC tumors, DOK-TC cell cultures as well as dysplastic and OSCC lesions in mice treated with 4NQO. Consistently, higher activation of the Wnt/β-catenin pathway was associated with elevated levels of PORCN. Finally, using the in vivo carcinogenesis model, we tested the effects of LGK974, a novel PORCN inhibitor developed by Novartis and currently under evaluation in a clinical trial phase 1 for Wnt-dependent malignancies [10]. Importantly, LGK974 reduced the size and number of murine oral lesions induced by 4NQO, as well as prevented the histological evolution of the lesions towards OSCC.

Methods

Materials

Anti-Wnt3a antibody (ab219412) was from Abcam (Cambridge, UK). Anti-PORCN antibody (PA5-43423) was from Invitrogen (Waltham, Massachusetts, USA). Anti-total-β-catenin antibody (M3539) was from Dako (Glostrup, Glostrup, Denmark). Anti-non-phosphorylated β-catenin Ser33/37/Thr41 antibody (8814) was from Cell Signaling Technology (Danvers, Massachusetts, USA). Anti-GAPDH (sc-365062) was from Santa Cruz Biotechnology (Dallas, Texas, USA). Paraformaldehyde (PFA), 4-Nitroquinoline 1-oxide (4NQO), Benzo[a]pyrene (B(a)P) and dimethylbenz[a]anthracene (DMBA) were from Sigma-Aldrich (San Luis, Misuri, USA). EZ-ECL chemiluminescent substrate was from Pierce Chemical (Dallas, Texas, USA). Goat anti-rabbit and goat anti-mouse antibodies coupled to horseradish peroxidase (HRP) were from Bio-Rad Laboratories (Hercules, California, USA). DMEM High glucose, penicillin, streptomycin, and Fetal Bovine Serum (FBS) were from Corning (Corning, New York, USA). Protein lysis buffer T-PER, and protease and phosphatase inhibitors were from Thermo Fisher Scientific (Waltham, Massachusetts, USA). Hydrocortisone was from Tocris (Bristol, UK). Nitrocellulose membranes were from Cytiva (Marlborough, Massachusetts, USA). Horseradish peroxidase VECTASTAIN ABC system was from Vector Laboratories (Newark, California, USA).

Retrospective OSCC carcinogenesis samples from patients

A total of 105 retrospective paraffin-embedded biopsies from the Anatomical Pathology service of the Faculty of Odontology, Universidad de Chile were used. The use of the retrospective samples was approved by the Ethics Committee of the Faculty of Medicine, Universidad de Chile (code 234-2021). These samples included 5 Normal oral mucosa, 25 Low-grade oral dysplasias, 25 High-grade oral dysplasias, 25 Well-differentiated OSCC and 25 Moderately/Poorly-differentiated OSCC. Histological classification was assessed by an oral pathologist in sections previously stained with hematoxylin-eosin, according to WHO 2022 Diagnostic criteria for epithelial dysplasia [11]. Samples were characterized according to sex (female or male), age (more or less than 60 years), clinical diagnosis (leukoplakia, erythroplasia, leukoerythroplasia or oral cancer), localization of the lesion (tongue, palate, mouth floor, gum or cheek) and smoking habit (smoker, no smoker or ex-smoker). Retrospective biopsies were used for immunohistochemistry.

Tumorigenicity assay

Human oral dysplasia cell lines (DOK) were maintained in DMEM high glucose, supplemented with 20 % FBS, 5 μg/mL hydrocortisone, penicillin (10,000 U/mL) and streptomycin (10 μg/mL). Malignant transformation of DOK cells (DOK-TC) was promoted by intermittent treatment with 70 μmol/L B(a)P/DMBA mixture for 3 months. 1 × 10⁶ DOK or DOK-TC cells in 100 μl of saline solution were subcutaneously inoculated into the left flack of 5 NOD/SCID mice per condition. Tumor growth was allowed for 33 days. After euthanasia, tumors were excised, measured and processed for immunohistochemistry. Experiments were conducted in accordance with approved protocols of the Institutional Committee for the Care and Use of Animals (CICUA), code 22554-MED-UCH.

Experimental OSCC carcinogenesis model in vivo

30 C57Bl/6 female mice 4–6 weeks old were randomly divided into 3 groups: (A) a control group maintained with purified water ad libitum for 22 weeks (n=10); (B) a 4NQO group treated for 16 weeks with 50 μg/mL 4-Nitroquinoline 1-oxide (4NQO) in the drinking water, and then 6 weeks with purified water. At week 20, citrate buffer, LGK974 vehicle, was administered by oral gavage for 14 days (n=10); (C) and a 4NQO+LGK974 group treated with 50 μg/mL 4NQO in the drinking water for 16 weeks and purified water until week 22. At week 20, 3 mg/kg/day LGK974 were administered by oral gavage for 14 days (n=10). All mice were maintained under a 12/12 h light/dark cycle, 22 °C, and food and water ad libitum. Mice 4 to 6 weeks old were preferred because early onset of oral lesions after carcinogen challenge has been described for younger animals [12]. Prior to euthanasia, mice were anesthetized, and macroscopic analysis of tumor size and number was performed. At week 22 all mice were euthanized by cervical dislocation and tongues were processed for immunohistochemistry or western blotting. Experiments were conducted in accordance with approved protocols of the Institutional Committee for the Care and Use of Animals (CICUA), code 22554-MED-UCH.

Immunohistochemistry (IHC)

Mouse tongues obtained from the experimental OSCC carcinogenesis model in vivo were fixed in 4 % paraformaldehyde and included in paraffin blocks. Retrospective biopsies from patients with Normal oral mucosa, Low grade dysplasia, High grade dysplasia, Well-differentiated OSCC and Moderately/Poorly-differentiated OSCC, were previously included in paraffin blocks. 3 μm sections were subsequently incubated with xylene, descending ethanol concentrations and distilled water, for deparaffination and rehydration. For antigen retrieval and endogenous peroxidase inactivation, samples were incubated with citrate buffer pH 6.0 and 3 % hydrogen peroxide in methanol, respectively. Then samples were blocked with horse serum and incubated with primary antibodies (overnight, 4 °C). Primary antibodies used were anti-PORCN (PA5-43423, Invitrogen; 1:100), anti-Wnt3a (ab219412, Abcam; 1:100), anti-β-catenin (M3539, Dako; 1:100), anti-non-phosphorylated β-catenin (8814, Cell Signaling Technology; 1:100). For detection of the primary antibodies, biotinylated secondary antibodies and HRP-conjugated streptavidin were used. The reaction was visualized with diaminobenzidine and samples were stained with Harris hematoxylin. Images were obtained with an optical microscope (Olympus CX41, 400 ×) and analyzed using ImageJ software. For PORCN and Wnt3a expression, staining intensity was scored on a scale from 0 to 3, being 0 the lack of staining, 1 mild staining, 2 moderate staining, and 3 intense staining of the cells. For β-catenin and non-phosphorylated β-catenin expression, the percentage of cells with positive nuclear staining was determined. All analyzes were done by randomly choosing five fields, observing a total of 1000 cells per sample.

Western blotting

Mouse tongues obtained from the experimental OSCC carcinogenesis model in vivo were lysed in T-PER buffer (Thermo Fisher Scientific), supplemented with protease and phosphatase inhibitor cocktails (Thermo Fisher Scientific). 20 μg of total protein per sample were mixed with Laemmli buffer and SDS-PAGE was performed. Proteins were then transferred to nitrocellulose membranes in a wet tank (30 V, overnight). Membranes were blocked with 5 % non-fat milk in 0.1 % Tween-TBS and incubated with primary antibodies: anti-PORCN (PA5-43423, Invitrogen; 1:1000), anti-β-catenin (M3539, Dako; 1:1000), anti-non-phosphorylated β-catenin (8814, Cell Signaling Technology, 1:1000), anti-GAPDH (Santa Cruz Biotechnology, sc-365062; 1:5000). Secondary HRP-conjugated antibodies and ECL substrate were used for the chemiluminescence reaction. Luminescence was detected with a C-DiGit Blot scanner (LI-COR).

Statistical analysis

t-test and ANOVA, Bonferroni post-test, were used. Statistical significance was considered as p < 0.05, using GraphPad Prism 6 software.

Results

PORCN levels in OSCC carcinogenesis ex vivo

As an ex vivo model of the different histological grades of oral dysplasia and OSCC, retrospective biopsies from 105 patients were used (Table 1). Compared to normal oral mucosa, histological staining with Hematoxylin & Eosin (H&E) revealed augmented proliferation with cytological and architectonical changes in the basal layer for low grade oral dysplasia, while increased patchy proliferation with cytological and architectonical changes in more than 2/3 of the epithelial layers was observed for high grade oral dysplasia (Fig. 1A). Elevated proliferation and slight stromal invasion were observed in well-differentiated OSCC, while complete stromal invasion characterized moderately/poorly-differentiated OSCC (Fig. 1A).

Table 1.

Clinical demographic table of patients with oral dysplasia and OSCC. 25 patients with low-grade oral dysplasia, 25 patients with high-grade oral dysplasia, 25 patients with well-differentiated OSCC and 25 patients with moderately/poorly-differentiated OSCC were classified according to sex (male or female), age (equal/more or less than 60 years old), clinical diagnosis (leukoplakia, erythroplasia, leukoerythroplasia or oral cancer), lesion localization (tongue, palate, mouth floor, gum or other) and smoking habit (smoker, no smoker, ex-smoker or data not recorded).

	Low-grade dysplasia	High-grade dysplasia	Well differentiated OSCC	Moderately/poorly differentiated OSCC
Total cases	25 (100 %)	25 (100 %)	25 (100 %)	25 (100 %)
SEX Female Male	11 (44 %) 14 (56 %)	18 (72 %) 7 (28 %)	12 (48 %) 13 (52 %)	9 (36 %) 16 (64 %)
AGE < 60 years ≥ 60 years	10 (40 %) 15 (60 %)	13 (52 %) 12 (48 %)	9 (36 %) 16 (64 %)	11 (44 %) 14 (56 %)
SMOKING Yes No Ex-smoker Data not recorded	15 (60 %) 2 (8 %) 3 (12 %) 5 (20 %)	17 (68 %) 1 (4 %) 0 (0 %) 7 (28 %)	18 (72 %) 0 (0 %) 3 (12 %) 4 (16 %)	14 (56 %) 3 (12 %) 3 (12 %) 5 (20 %)
CLINICAL DIAGNOSIS Leukoplakia Erythroplakia Leukoerythroplakia Oral cancer	18 (72 %) 5 (20 %) 2 (8 %) 0 (0 %)	14 (56 %) 7 (28 %) 1 (4 %) 3 (12 %)	4 (16 %) 0 (0 %) 2 (8 %) 19 (76 %)	0 (0 %) 0 (0 %) 9 (36 %) 16 (64 %)
LOCALIZATION Tongue Palate Mouth floor Gum Cheek	16 (64 %) 4 (16 %) 0 (0 %) 3 (12 %) 2 (8 %)	11 (44 %) 1 (4 %) 8 (32 %) 3 (12 %) 2 (8 %)	19 (76 %) 0 (0 %) 6 (24 %) 0 (0 %) 0 (0 %)	12 (48 %) 4 (16 %) 5 (20 %) 3 (12 %) 1 (4 %)

Fig. 1.

PORCN expression in OSCC carcinogenesis ex vivo. Retrospective samples from 105 patients diagnosed as normal oral mucosa (n=5), low grade dysplasia (n=25), high grade dysplasia (n=25), well-differentiated OSCC (n=25) and moderately/poorly-differentiated OSCC (n=25) were used. For histologic analysis, Hematoxylin & Eosin (H&E) staining was performed. Representative H&E images (400 × amplification) are shown in (A). To determine the levels of PORCN, immunohistochemistry was performed. Representative images (400 × amplification) are displayed in (B), while quantification of the staining intensity of the cells (0 no staining, 1 mild staining, 2 moderate staining, and 3 intense staining) is graphically depicted in (C). (low grade dysplasia vs normal oral mucosa p=0.5001; high grade dysplasia and well- and moderately/poorly-differentiated OSCC vs normal oral mucosa p < 0.0001; moderately/poorly-differentiated OSCC vs well-differentiated OSCC p = 0.0196; n=105).

To evaluate the levels of PORCN in these retrospective biopsies, immunohistochemistry was performed. All 5 normal oral mucosa tissues displayed mild PORCN staining intensity. In low-grade oral dysplasia, 17 of 25 samples showed mild PORCN staining intensity and 8 of 25 samples presented moderate staining intensity (p = 0.5001). PORCN staining intensity in high-grade oral dysplasia was predominantly moderate (17 of 25) or high (8 of 25) (p < 0.0001) (Fig. 1B and C). In well-differentiated OSCC, 8 of 25 tissues showed moderate PORCN staining intensity while 17 of 25 tissues displayed high staining intensity (p < 0.0001). Finally, high PORCN staining intensity was observed in all of the 25 moderately/poorly-differentiated OSCC tissues (p < 0.0001) (Fig. 1B, C). This indicates that, compared to normal oral mucosa, PORCN expression is significantly higher in high grade dysplasia, well-differentiated OSCC as well as moderately/poorly-differentiated OSCC.

Wnt/β-catenin levels in OSCC carcinogenesis ex vivo

Given that PORCN is an O-acyltransferase important for Wnt ligand maturation and export from the cell, we evaluated the levels of the Wnt ligand Wnt3a, and its downstream target β-catenin -as well as its non-phosphorylated (active) form- in retrospective biopsies from patients with oral dysplasia and OSCC. Immunohistochemical staining for Wnt3a in normal oral mucosa was of mild intensity in 3 of 5 samples, while for the other 2 no staining was detected. In low-grade oral dysplasia, for all 25 samples analyzed Wnt3astaining intensity was mild. In high-grade oral dysplasia, Wnt3a staining intensity was moderate in 17 of 25 and high in 8 of 25 tissue samples (Fig.2A and B). For high-grade oral dysplasia, expression of Wnt3a was statistically higher than for normal oral mucosa (p < 0.0001). Of the 25 samples of well-differentiated OSCC, 14 and 11 samples showed moderate or high Wnt3a staining intensity, respectively. Moreover, Wnt3a staining intensity was high in all 25 samples of moderately/poorly-differentiated OSCC (Fig. 2A and B). Also significant differences were detected for well- and moderately/poorly- differentiated OSCC, compared to normal oral mucosa (p < 0.0001).

Fig. 2.

Wnt/β-catenin pathway activity in OSCC carcinogenesis ex vivo. Retrospective samples from 105 patients diagnosed as normal oral mucosa (n=5), low grade dysplasia (n=25), high grade dysplasia (n=25), well-differentiated OSCC (n=25) and moderately/poorly-differentiated OSCC (n=25) were used. To determine the expression of Wnt3a, total nuclear β-catenin and non-phosphorylated (active) nuclear β-catenin, immunohistochemistry was performed. Representative images (400 × amplification) of Wnt3a immunohistochemistry is shown in (A), and quantification of Wnt3a staining intensity of the cells (0 no staining, 1 mild staining, 2 moderate staining, and 3 intense staining) is presented in (B) (low grade dysplasia vs normal oral mucosa p=0.2205; high grade dysplasia, well-differentiated OSCC and moderately/poorly-differentiated OSCC vs normal oral mucosa p < 0.0001; n=105). Representative images (400 × amplification) of total β-catenin immunohistochemistry is shown in (C), quantification of the percentage of cells with nuclear localization of β-catenin is presented in (D) (low grade dysplasia, high grade dysplasia, well-differentiated OSCC and moderately/poorly-differentiated OSCC vs normal oral mucosa p < 0.0001; n=105). Representative images (400 × amplification) of non-phosphorylated β-catenin immunohistochemistry is shown in (E), and quantification of the percentage of cells with nuclear localization of non-phosphorylated β-catenin is presented in (F) (low grade dysplasia, high grade dysplasia, well-differentiated OSCC and moderately/poorly-differentiated OSCC vs normal oral mucosa p < 0.0001; n=105).

Since stimulation by Wnt ligands results in stabilization and nuclear translocation of β-catenin, we also evaluated the expression of total β-catenin in oral dysplasia and OSCC. In normal oral mucosa, β-catenin was mainly found at the cell membrane of keratinocytes (therefore, 0 % of cells were considered positive for nuclear localization of β-catenin), while in low-grade oral dysplasia, nuclear, membranous and cytoplasmic localizations were more frequent (55 % of cells with positive nuclei) (p < 0.0001) (Fig. 2C and D). An increase in nuclear β-catenin was observed in high grade oral dysplasia, as well as in all histological differentiation types of OSCC (p < 0.0001) (Fig. 2C and D). Indeed, for high-grade oral dysplasia, well- and moderately/poorly- differentiated OSCC, 90 %, 75 % and 95 % of cells with positive nuclear β-catenin staining, respectively, were detected (Fig. 2C and D). This suggests that nuclear translocation of β-catenin is significantly augmented in high- and low- grade dysplasia, as well as in well- and moderately/poorly- differentiated OSCC.

The expression of the transcriptionally active, non-phosphorylated form of β-catenin was determined by immunohistochemistry. Non-phosphorylated β-catenin was found exclusively in the plasma membrane of oral keratinocytes in normal oral mucosa samples, equivalent to 0 % of positive nuclear β-catenin staining in cells (Fig. 2E and F). For low- and high- grade dysplasia, progressive increases in the nuclear localization of non-phosphorylated β-catenin were detected, whereby in 61 % and 92 % of the cells, respectively, nuclei stained positive (p < 0.0001) (Fig. 2E and F). Similarly, for well- and moderately/poorly- differentiated OSCC 77 % and 98, respectively, of the cell nuclei stained positive for non-phosphorylated β-catenin (p < 0.0001) (Fig. 2E and F). These observations indicate that transcriptionally active β-catenin translocation to the nucleus occurs in high- and low- grade dysplasia, as well as in well- and moderately/poorly- differentiated OSCC.

PORCN and Wnt/β-catenin levels in an in vivo OSCC tumorigenesis model

We have previously characterized an in vivo OSCC tumorigenesis model, where dysplastic oral keratinocytes (DOK) were treated with a mixture of the carcinogens Benzo(a)pyrene (B(a)P) and 7,12-dimethylbenz(a)anthracene (DMB) to promote malignant transformation of the cells (DOK-TC) [8]. DOK and DOK-TC cells were then subcutaneously injected into NOD/SCID mice to allow tumor growth (Fig. 3A). At different time points, tumors were then extracted and fixed (Fig. 3B). To evaluate the levels of PORCN and the Wnt/β-catenin activation, tumor samples were analyzed using immunohistochemistry. High staining intensity for PORCN was observed in all 5 DOK-TC tumors, while DOK tumors displayed mild (2/3) and moderate staining (1/3) intensity (p=0.0005) (Fig. 3C, D). Consistently, Wnt3a staining intensity was mostly moderate in DOK-TC tumors (4/5), compared to the mild staining intensity observed in all 3 DOK tumors (p=0.0240) (Fig. 3C, E). In DOK tumors, nuclear localization of total and non-phosphorylated β-catenin, were observed in 10 % and 21 % of the cases, respectively (Fig. 3C, F, G). Significant increases in total and non-phosphorylated β-catenin were observed in DOK-TC tumors, where values reached 74 % (p < 0.0001) and 51 % (p=0.0249), respectively (Fig. 3C, F, G). To confirm that transformed malignant cells display higher PORCN levels, extracts from DOK and DOK-TC cells were analyzed (Fig. 3H). Compared to DOK cells, PORCN levels were significantly increased in DOK-TC cells (p = 0.0318) (Fig.3I). Original western blot images are shown in Supplementary Fig. 1. Altogether, these results suggest that tumors formed by malignant oral cell lines express higher PORCN levels coincident with increased activation of the Wnt/β-catenin pathway.

Fig. 3.

PORCN expression and Wnt/β-catenin pathway activity in in vivo tumorigenesis. Oral dysplastic cell lines (DOK) were treated with 70 μmol/L B(a)P/DMBA mixture for 3 months for malignant transformation of the cells (DOK-TC). For tumor growth, subcutaneous inoculation of 1 × 10⁶ DOK or DOK-TC cells into the left flack of 5 NOD/SCID mice per cell type was performed. After 33 days, tumors were excised and processed for immunohistochemistry. Representative images of subcutaneous tumor growth from DOK or DOK-TC cells after 33 days are shown in (A), while excised tumors are presented in (B). Representative immunohistochemistry images (400 × amplification) for PORCN, Wnt3a, Total β-catenin, Non-phosphorylated (active) β-catenin, as well as H&E staining, of tumors from DOK and DOK-TC cells are displayed in (C). Quantification of PORCN and Wnt3a staining intensity of the cells (0 no staining, 1 mild staining, 2 moderate staining, and 3 intense staining) in tumors from DOK and DOK-TC cells is presented in (D) (DOK-TC tumor vs DOK tumors p=0.0005; n=5) and (E) (DOK-TC tumor vs DOK tumors p=0.0240; n=5), respectively. Quantification of the percentage of cells with nuclear localization of total and non-phosphorylated β-catenin is displayed in (F) (DOK-TC tumor vs DOK tumors p < 0.0001; n=5) and (G) (DOK-TC tumor vs DOK tumors p=0.0249; n=5), respectively. DOK and DOK-TC cells were cultured and lysed for western blotting. Representative western blot images of PORCN and GAPDH (loading control) are shown in (H), while densitometric analysis and quantification of PORCN levels is graphically displayed in (I) (DOK-TC cells vs DOK cells p = 0.0318; n = 4).

Macroscopic and histological evaluation of OSCC lesions in vivo after treatment with LGK974

Following a similar protocol as described previously [8], 30 C57BL/6 female mice were equally distributed into 3 groups: a control group was left for 22 weeks with regular water ad libitum; an oral carcinogenesis group was treated for 16 weeks with 4NQO in the drinking water followed by 6 weeks with regular water (4NQO group). Between the weeks 20 and 22, citrate buffer, the vehicle for LGK974, was administered daily by oral gavage. A LGK974 treatment group was left for 16 weeks with 4NQO in the drinking water and 6 weeks with regular water (4NQO+LGK974 group). Between the weeks 20 and 22, 3 mg/kg/day LGK974 were administered by oral gavage. The schematic representation of times and stimulations for each group are depicted in Fig. 4A. After 22 weeks, mice were euthanized and tongues were removed for macroscopic evaluation of the lesions. In the 4NQO group, lesions were white in appearance with exophytic growth and warty and irregular surfaces, while lesions in the 4NQO+LGK974 group were white with flat surfaces (Fig. 4B). On average the size of the lesions in the 4NQO group was three-times greater than the ones in the 4NQO+LGK974 group (p < 0.0001) (Fig. 4C). The mean number of lesions found in the tongues of the 4NQO group was 3.3, while in the 4NQO+LGK974 group only 1.3 lesions per tongue were found (p < 0.0001) (Fig. 4D). Histological analysis revealed that, in the 4NQO group, all 10 mice presented lesions consistent with OSCC and oral dysplasia. In the 4NQO+LGK974 group, on the other hand, all 10 mice had lesions consistent with oral dysplasia (Fig. 4E). These results indicate that treatment with LGK974 significantly reduces the number and size of the oral lesions induced by 4NQO, and impedes the progression of oral dysplasia towards OSCC.

Fig. 4.

Effect of LGK974 in oral carcinogenesis in vivo. Three groups of C57Bl/6 mice (4 to 6 weeks old; 10 mice per group) were used. Timelines and treatments are described in (A). The control group was maintained with purified water ad libitum for 22 weeks. The 4NQO group was challenged with 50 μg/mL 4-Nitroquinoline 1-oxide (4NQO; 6 weeks) in the drinking water, and then maintained with purified water (6 weeks). The last two weeks, citrate buffer (LGK974 vehicle) was administered by oral gavage. Finally, the 4NQO+LGK974 group was challenged with 50 μg/mL 4NQO in the drinking water (16 weeks) and purified water (6 weeks). The last two weeks, 3 mg/kg/day LGK974 were administered by oral gavage. All mice were euthanized at week 22 and tongues were extracted. Representative images of the tongues of control, 4NQO and 4NQO+LGK974 groups are shown in (B). Size and number of lesions the tongues in 4NQO and 4NQO+LGK974 groups (control group did not present any lesions) were quantified and graphically depicted in (C) and (D), respectively (4NQO+LGK974 group vs 4NQO group p < 0.0001; n=10). Tongues were fixated in PFA and processed for histological analysis. Representative images of H&E staining (a single 20-fold magnification and a 400-fold magnification from two tissue areas) in control, 4NQO and 4NQO+LGK974 tongues are presented in (E). Lesions in 4NQO group were consistent with OSCC and dysplasia, while lesions in 4NQO+LGK974 group were consistent with dysplasia.

PORCN levels in OSCC carcinogenesis in vivo

To evaluate the expression of PORCN, immunohistochemistry was performed on tongue samples obtained from 30 mice treated with either regular water for 22 weeks (control group) or 4NQO for 16 weeks + 4 weeks with regular water (4NQO group). The 4NQO group tissues were histologically classified as oral dysplasia or OSCC, as shown in Fig. 4E. PORCN staining in the control group was predominantly observed in the basal epithelial stratum (10 out of 10 mice with mild staining intensity) (Fig. 5A and B). Expression of PORCN was detected in all epithelial layers in oral dysplasia and OSCC tissues. In oral dysplasia tissues, for 7 lesions staining intensity was moderate, while in 3 lesions staining intensity was high (p < 0.0001) (Fig. 5A and B). In OSCC tissues, all 10 lesions presented high staining intensity (p < 0.0001) (Fig. 5A, B). This suggests that significant upregulation of PORCN occurs in dysplasia and OSCC lesions induced by 4NQO. To confirm this, we analyzed tongue tissue extracts from control and 4NQO mouse groups (3 mice per group) to determine by western blotting the levels of PORCN. Of note, since tissues were solubilized for analysis, lesions could not be pre-classified as dysplasia or OSCC. In oral lesions from mice treated with 4NQO, the levels of PORCN were higher than the levels observed in oral tissues from control mice (p=0.0494) (Fig. 5C and D).

Fig. 5.

PORCN levels in oral carcinogenesis in vivo. Tongues from control and 4NQO mouse groups were fixated and immunohistochemistry for PORCN was performed. For 4NQO group, lesions consistent with dysplasia and OSCC are shown. Representive immunohistochemistry images (400 × magnification) are presented in (A), while quantification of PORCN staining intensity of the cells (0 no staining, 1 mild staining, 2 moderate staining, and 3 intense staining) for each group is graphically depicted in (B) (dysplasia and OSCC in 4NQO group vs control group p < 0.0001; n=10). Tongues from control and 4NQO groups were lysed and immunowestern blotting for PORCN was performed. Representative western blot images of PORCN and GAPDH (loading control) are shown in (C), while densitometric analysis and quantification of PORCN levels is graphically displayed in (D) (4NQO group vs control group p = 0.0494; n=3).

Wnt/β-catenin levels in OSCC carcinogenesis in vivo

Given that, consistent with the results observed in ex vivo human samples of oral carcinogenesis, PORCN levels were increased in oral dysplasia and OSCC lesions, we sought to determine whether the application of the PORCN inhibitor LGK974 precluded activation of the Wnt/β-catenin pathway in vivo. To this end, we evaluated the expression of Wnt3a, non-phosphorylated β-catenin and total β-catenin by immunohistochemical analysis in control mice and mice treated with 4NQO alone or 4NQO+LGK974, as previously described (Fig. 6A). Importantly, as shown in Fig. 4E, lesions induced by 4NQO could be classified into dysplasia or OSCC, but in lesions from mice treated with LGK974 (and challenged with 4NQO) only dysplasia was observed.

Fig. 6.

Wnt/β-catenin pathway activity in oral carcinogenesis in vivo. Tongues from control, 4NQO and 4NQO+LGK974 mouse groups were fixated and immunohistochemistry for Wnt3a, Total β-catenin and Non-phosphorylated (active) β-catenin. Representative immunohistochemistry images (400 × amplification) for Wnt3a, Total β-catenin and Non-phosphorylated β-catenin of each mouse group are displayed in (A). For 4NQO group, lesions consistent with dysplasia and OSCC are shown, while for 4NQO+LGK974, since no OSCC lesions were detected, only lesions consistent with dysplasia are shown. Quantification of PORCN and Wnt3a staining intensity of the cells (0 no staining, 1 mild staining, 2 moderate staining, and 3 intense staining) is presented in (B) (dysplasia and OSCC in 4NQO group vs control group; dysplasia and OSCC in 4NQO+LGK974 group vs 4NQO group; p < 0.0001; n=10). Quantification of the percentage of cells with nuclear localization of total and non-phosphorylated β-catenin is displayed in (C) (dysplasia and OSCC in 4NQO group vs control group; dysplasia and OSCC in 4NQO+LGK974 group vs 4NQO group; p < 0.0001; n=10) and (D) (dysplasia and OSCC in 4NQO group vs control group; dysplasia and OSCC in 4NQO+LGK974 group vs 4NQO group; p < 0.0001; n=10), respectively. Tongues from control, 4NQO and 4NQO+LGK974 groups were lysed and immunowestern blotting for Total β-catenin and Non-phosphorylated β-catenin was performed. Representative western blot images of Total β-catenin, Non-phosphorylated β-catenin and GAPDH (loading control) are shown in (E), while densitometric analysis and quantification of Total β-catenin and Non-phosphorylated β-catenin levels is graphically displayed in (F) (4NQO group vs control group p=0.0061; 4NQO+LGK974 group vs control group p=0.0292; n=3) and (G) (4NQO group vs control group p=0.0115; 4NQO+LGK974 group vs control group p=0.1138; n=3), respectively.

In control conditions, Wnt3a was found in both the epithelium and stroma, whereby staining intensity was mild in 2 of 10 mice, and moderate in 8 of 10 mice (Fig. 6B). After treatment with 4NQO, Wnt3a was evenly distributed in the epithelium of dysplasia and OSCC lesions (Fig. 6A). Of the 10 oral dysplasia lesions induced by 4NQO, 3 showed moderate staining intensity for Wnt3a, while for 7 staining intensity was high. For all 10 OSCC lesions with 4NQO staining intensity was high (Fig. 6B). Notably, treatment with LGK974 blunted stromal localization and significantly decreased Wnt3a expression, in which cases staining intensity was weak (p < 0.0001) (Fig. 6A and B).

Total β-catenin, on the other hand, displayed an epithelial membranous localization in oral tissues from control mice, with no nuclear expression (0 % positive cells with nuclear β-catenin) (Fig. 6A, C). Exposure to 4NQO resulted in increased nuclear β-catenin expression in oral dysplasia and OSCC lesions (63.5 % and 73.0 %, respectively). In oral dysplasia lesions, total β-catenin was mainly found in nuclear and cytoplasmic localizations. Alternatively, in OSCC lesions, β-catenin was found in nuclei, as well as the plasma membrane and the cytoplasm. Treatment with LGK974 dramatically diminished the presence of β-catenin in nuclei of oral lesions diagnosed as dysplasia (4,4 %), while increasing localization of β-catenin to the cytoplasm (p<0.0001) (Fig. 6A, C).

In control tissues, non-phosphorylated (active) β-catenin was observed in the epithelial cell plasma membrane (0 % nuclear positivity) (Fig. 6A, D). Nuclear expression of non-phosphorylated β-catenin increased significantly after treatment with 4NQO, in oral dysplasia and OSCC lesions (82.5 % and 85.0 %, respectively). In both cases, non-phosphorylated β-catenin was detected, besides the nucleus, at the plasma membrane and in the cytoplasm of epithelial cells. Like total β-catenin expression, nuclear non-phosphorylated β-catenin expression was drastically reduced in oral dysplastic lesions after treatment with LGK974, with a positivity percentage of 6.3 % (p < 0.0001). Interestingly, membrane localization of non-phosphorylated β-catenin was predominantly observed after treatment with LGK974 (Fig. 6A, D). This suggests that treatment with LGK974 significantly reduced the activation of the Wnt-β-catenin signaling pathway in vivo.

To confirm this, we evaluated the levels of total β-catenin and non-phosphorylated β-catenin by western blotting in tongue samples from control mice and mice treated with 4NQO alone or 4NQO+LGK974. Exposure to 4NQO resulted in a five-fold increase in the levels of non-phosphorylated β-catenin compared to those observed in samples from control mice (p=0.0061) (Fig. 6E and F). Similarly, total β-catenin levels were significantly higher in samples from 4NQO-induced lesions than in those from control tissues (p=0.0115) (Fig. 6E, G). A tendency towards reduced total β-catenin levels was observed after treatment with LGK974 (p=0.1138) (Fig. 6E, G). Treatment with LGK974 reduced non-phosphorylated β-catenin levels (p=0.0292) (Fig. 6E and F), suggesting that LGK974 prevents activation of β-catenin in oral carcinogenesis in vivo.

Taken together, these observations indicate that inhibition of PORCN with LGK974 effectively reduces Wnt/β-catenin activity in dysplasia and OSCC in vivo.

Discussion

The use of PORCN inhibitors for cancer treatment has been reviewed elsewhere [9]. LGK974 is a PORCN inhibitor that blocks the palmitoleoyl-CoA binding site, interfering with the post-translational modifications required for Wnt ligand maturation [13]. Despite that LGK974 is still under evaluation in clinical trials, its use as an anti-cancer drug looks promising. Cancer cells that are highly dependent on Wnt-ligand driven proliferation are known as “Wnt-ligand addicted” [14]. These cells are especially susceptible to LGK974 treatment, not only by suppressing the expression of β-catenin controlled genes associated with proliferation, but also by disrupting the cancer cell metabolic status as is the case in pancreatic adenocarcinoma [15,16], by increasing Nuclear factor (erythroid-derived 2)-like 2 (NRF2) related radiosensitivity [17], or by switching Tumor‑associated macrophages into an M1 state in non‑small cell lung cancer [18]. Here we observed that during OSCC carcinogenesis PORCN and Wnt3a levels were consistently upregulated ex vivo (Figs. 1B, and C, 2A, and B), in vivo (Figs. 3C–E, 5A, 5B, 6A, and B) and in vitro (Fig. 3H and I), suggesting that in high-grade dysplasia and OSCC, oral cells behave as if they were “Wnt-ligand addicted”. This likely explains why LGK974 was useful in our oral carcinogenesis model, where application of the inhibitor reduced the size and number of the oral lesions induced by 4NQO (Fig. 4B–D). The results obtained analyzing Wnt3a expression in OSCC carcinogenesis ex vivo (Fig. 2A, and B) are consistent with the expression reported for other Wnt ligands. For instance, Wnt5a expression is significantly elevated in high-grade dysplasia and well- and poorly- differentiated OSCC, but not in low-grade dysplasia [19]. Information of overall survival of the patients from which the biopsies were obtained was not available, representing a limitation of our study.

Some studies have found increased PORCN expression in cancers other than OSCC, but with different outcomes. In tissues from patients with hepatocellular carcinoma, PORCN upregulation is detected, compared to the adjacent normal tissues [20]. This upregulation is transcriptionally driven, as higher levels of PORCN coding mRNAs are also detected in these tumors. High expression of PORCN in hepatocellular carcinoma is associated with reduced survival, highlighting its value as a prognosis marker. Indeed, overexpression of PORCN in the human hepatocarcinoma cell line SMCC-7721 promotes the formation of larger subcutaneous tumors after injection into nude mice [20]. This is consistent with the results obtained in our in vivo OSCC tumorigenesis model, where DOK-TC tumors displayed higher PORCN levels (Fig. 3). A limitation of our study, however, is that we did not evaluate whether PORCN levels were upregulated due to increased synthesis or reduced degradation.

In non-small cell lung carcinoma, high expression of PORCN is associated with the expression S100 calcium-binding protein P (S100P), a protein involved in cell cycle control and differentiation regulated by nuclear factor kappa B (NF-κB) and cyclin D1 [21]. Unlike for hepatocellular carcinoma, in non-small cell lung carcinoma no changes in PORCN coding mRNAs is observed [21]. Notably, tissues of nasopharyngeal carcinoma -a type of Head and Neck Squamous Cell Carcinoma- present augmented levels of S100P [22]. In patients, higher S100P expression levels correlate with decreased overall survival, suggesting that downstream PORCN targets might be useful as prognosis markers in some types of Head and Neck cancers [22]. Indeed, chemical inhibition of PORCN reduces migration and induces apoptosis in Head and Neck Cell Carcinoma cells [23]. For Cutaneous Squamous Cell Carcinoma, established after malignant transformation of skin keratinocytes by HPV infection, higher PORCN expression is observed in cancer stem cells [24]. Also, in vivo tumorigenesis of malignant skin keratinocytes following treatment with LGK974 results in reduced tumor weight [24], consistent with the results obtained in our in vivo model of OSCC carcinogenesis following treatment with LGK974 (Fig. 4B, and C). Similarly, intestinal stem cells require Wnt secretion in order to initiate colorectal cancer tumorigenesis, as LGK974 reduces stem cell replacement and increases the stem cell size [25].

Natural compounds such as 1,25-dihydroxycholecalciferol (Calcitriol), the active form of Vitamin D, can also downregulate the Wnt/β-catenin pathway. In skin SCC cell lines, treatment with Calcitriol and other Vitamin D derivatives reduces proliferation and decreases transcription of CTNNB1, the gene coding for β-catenin [26]. Consistently, Vitamin D receptor (VDR) levels are significantly downregulated in oral dysplasia, and treatment with Calcitriol blunts the migration of oral dysplasia cell lines by reducing β-catenin activation [27]. Whether Calcitriol prevents upregulation of PORCN in oral dysplasia and OSCC is not known yet.

During the past few years has become relevant the role of tumors on systemic responses of the organism, ultimately facilitating carcinogenesis. Neuropeptides, classical hypothalamic hormones, biogenic amines and cathecolamines, among others, are released into circulation to directly enhance the carcinogenic features of the tumor, or to modify the tumor microenvironment (i.e. immune suppression) [28]. Dysruption of neuroendocrine communication also results in anxiety and altered cognitive state of the patients [28]. In OSCC, cathecolamines in the plasma are significantly higher than control conditions, suggesting a neuroendocrine modulation of the tumor [29]. Given that OSCC cells express adrenergic β-receptors, is feasible that cathecolamines act directly on OSCC cells to promote malignant features [29]. Indeed, chronic stress in vivo and cathecolamine stimulation in vitro results in increased OSCC tumor growth and OSCC cell lines aggressiveness, respectively [30]. These effects are mediated by C-X-C motif chemokine ligand 3 (CXCL3)-dependent activation of Wnt/β-catenin [30]. Future studies should address whether PORCN-mediated Wnt release in high-grade oral dysplasia and OSCC could modify neuroendocrine responses to drive oral carcinogenesis.

Together, this study not only highlights the role of PORCN as a progression marker during OSCC carcinogenesis, but also suggests the effectiveness of LGK974 for OSCC and/or potentially malignant lesions.

Conclusion

In summary, we determined that PORCN levels are significantly elevated in high-grade dysplasia and well- and moderately/poorly- differentiated OSCC in retrospective samples from patients, in tumors grown from malignant-transformed oral cell lines, in cultured malignant-transformed oral cell lines, as well as in oral murine lesions induced by the tobacco smoke-mimetic 4NQO. Consistently, we also observed increased Wnt3a expression, as well as augmented nuclear localization of total and active β-catenin in all experimental models employed. Oral administration of 3 mg/kg/day LGK974 in OSCC carcinogenesis in vivo -after 4NQO challenge- reduced the number and size of the oral lesions and prevented the activation of the Wnt/β-catenin pathway. Given that LGK974 is currently under evaluation in clinical trials assessing effectivity for the treatment of several cancers, additional studies should be included to determine its pharmacological effectiveness in the treatment of OSCC carcinogenesis.

Translational relevance

Oral cancer represents a huge economic and emotional burden to patients. The patient's prognosis is established by a tumor, node and metastasis system (TNM), in combination with a histopathological grading. However, in many cases these methods fail to provide an accurate prognosis, so novel biomarkers are required. We and others have studied the Wnt/β-catenin signaling pathway in Oral Squamous Cell Carcinoma. For Wnt ligand export, these molecules must be post-transcriptionally modified with a cysteine palmitoylation by an enzyme known as Porcupine. Here we show that the levels of Porcupine are modified during the different OSCC stages, highlighting its usefulness as a prognosis biomarker. Additionally, chemical inhibition of Porcupine is a feasible complementary pharmacological treatment for Oral Squamous Cell Carcinoma. We expect that the results of this research will shed light on patients prognosis and pharmacological opportunities.

CRediT authorship contribution statement

Daniel Peña-Oyarzún: Writing – review & editing, Writing – original draft, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Andrew F.G. Quest: Writing – review & editing, Writing – original draft, Resources, Methodology, Funding acquisition, Formal analysis. Lorena Lobos-González: Writing – review & editing, Writing – original draft, Resources, Methodology, Funding acquisition, Formal analysis. Andrea Maturana-Ramírez: Writing – review & editing, Writing – original draft, Resources, Methodology, Formal analysis. Montserrat Reyes: Writing – review & editing, Writing – original draft, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.