SEARCH

SEARCH BY CITATION

Keywords:

  • oral squamous cell carcinoma;
  • epithelial-mesenchymal transition;
  • Twist;
  • metastasis;
  • clinical outcome

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

Locoregional recurrence and distant metastases are ominous events in patients with advanced oral squamous cell carcinoma (OSCC). The objective of this study was to identify functional biomarkers that are predictive of OSCC progression to metastasis.

METHODS

The expression profile of a network of epithelial-mesenchymal transition (EMT) genes was investigated in a large cohort of patients with progressive OSCC using a complimentary DNA microarray platform coupled to quantitative reverse transcriptase-polymerase chain reaction and immunohistochemical analyses. Therapeutic potential was investigated in vitro and in vivo using an orthotopic mouse model of metastatic OSCC growing in the tongue microenvironment.

RESULTS

Among deregulated EMT genes, the Twist-related protein 1 (TWIST1) transcription factor and several of its regulated genes were significantly overexpressed across advanced stages of OSCC. This result was corroborated by the clinical observation that Twist1 up-regulation predicted the occurrence of lymph node and lung metastases as well as poor patient survival. In support of Twist1 as a driver of OSCC progression, the up-regulation of Twist1 was observed in cells isolated from patients with metastatic OSCC. The inhibition of Twist1 in these metastatic cells induced a potent inhibition of cell invasiveness in vitro as well as progression in vivo.

CONCLUSIONS

The current results provide evidence for the prognostic value and therapeutic potential of a network of Twist genes in patients with advanced OSCC. Cancer 2014;120:352–362. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

The development of oral squamous cell carcinoma (OSCC) metastasis poses clinical challenges because of the limited therapeutic options available.[1, 2] The time course of relapse manifestation and metastasis are unpredictable; and metastases, not primary tumors, account for approximately 90% of all cancer deaths.[3]

Metastasis is a dynamic process in which the interplay of tumor cell, host, and tissue microenvironment factors play a critical role, particularly in the signaling pathways that regulate homotypic and heterotypic cell-cell and cell-matrix interactions.[4] In advanced cancers in general, the final stages of tumor differentiation and progression are characterized by the invasion of tumor cells into the surrounding tissue and their dissemination to form metastases in distant organs.[5] Conceptually, the metastatic “cascade” is multifactorial and comprises the loss of cancer cell adherence at the primary site, cell motility, proteolysis of the extracellular matrix (ECM) and basement membrane, invasion of local stroma, entry of vascular and lymphatic vessels (intravasion), extravasion, recolonization, and expansion into distant sites.[4-6]

During this process, epithelial-mesenchymal transition (EMT) is established as a critical mechanism by which epithelial cells acquire sufficient plasticity and motile properties to become invasive. In addition, increasing evidence supports a link between EMT in the selection of cancer cell variants with stem cell properties and chemotherapy resistance.[6, 7] EMT events require the coordinated expression of several sets of genes and signaling pathways, many of which have demonstrated the ability to regulate specific aspects of malignant transformation and cancer progression in preclinical models.[7-9] In this context, early OSCC progression to invasive stages is often associated with widespread changes in cancer cell morphologic and differentiation features[10] believed to promote cancer cell dissemination to distant organs, particularly the lungs, through either the lymphatic system or blood vessels.[11, 12] In the current study, we gained new insights into the role of a network of Twist genes in OSCC progression as well as the correlation of Twist1 with invasiveness and clinicopathologic outcome. In addition, we examined the therapeutic potential of targeting Twist1 in both in vitro and in vivo OSCC models.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Study Population

Ten fresh OSCC samples classified as poorly differentiated tumors (grade 3)[13] from the tongue were matched with morphologically normal, surgically removed tissues from patients at the AC Camargo Cancer Center (São Paulo, Brazil), 70% of whom developed locoregional or distant metastasis, and were submitted to laser-capture microdissection. These samples were used for complimentary DNA (cDNA) microarrays and technical validation experiments.

For the independent sample set, 74 paraffin-embedded OSCC tissue specimens from 14 patients who had lung metastasis (metastatic cases) and 60 patients who had negative lymph node status without recurrence or metastatic disease (good outcomes; nonmetastatic cases) and were followed for at least 157 months were evaluated using immunohistochemistry assays in a tissue microarray. The OSCC metastatic cases presented with advanced T classification (P = .004), lymph nodes capsular rupture (P = .003), and lower survival probability (P = .001) compared with the nonmetastatic cases (Table 1).

Table 1. Distribution of Patients With Oral Squamous Cell Carcinoma
 No. of Patients (%)a 
 Fresh SamplesParaffin-Embedded Samples 
VariableGrade 3MetastaticNonmetastaticP
  1. Abbreviations: ND: not determined.

  2. a

    Percentages are based on the patients who had complete information.

  3. b

    These P values indicate a statistically significant difference.

Sex    
Men10 (100)11 (78.6)42 (68.9).471
Women0 (0)3 (21.4)18 (31.1) 
Race    
White10 (100)10 (71.4)52 (85.3).218
Nonwhite04 (28.6)8 (14.7) 
Smoking habit    
No1 (16.7)1 (7.7)6 (10.7).745
Yes5 (83.3)12 (92.3)50 (89.3) 
Alcohol consumption   
No1 (16.7)1 (7.7)14 (25).173
Yes5 (83.3)12 (92.3)42 (75) 
Tumor classification   
T1 + T24 (40)1 (7.7)31 (51.7)0.004b
T3 + T46 (60)12 (92.3)29 (48.3) 
Lymph node status   
N03 (30)3 (21.4)60 (100)< .001b
N+7 (70)11 (78.6)0 
Histologic grade   
10 (0)8 (61.5)6 (12.8)ND
20 (0)5 (38.5)12 (25.5) 
310 (100)0 (0)29 (61.7) 
Vascular embolization   
No8 (100)10 (83.3)32 (65.3).227
Yes0 (0)2 (16.7)17 (34.7) 
Perineural infiltration   
No3 (37.5)6 (50.0)29 (93.5).371
Yes5 (62.5)6 (50.0)2 (6.5) 
Capsular rupture   
No4 (50)8 (57.1)29 (93.5).003b
Yes4 (50)6 (42.9)2 (6.5) 
Status    
Alive6 (60)5 (35.7)47 (78.7).001b
Dead4 (40)9 (64.3)13 (21.3) 

The eligibility criteria included previously untreated patients without a second primary tumor who received treatment in the same institution. All patients were advised of the procedures and provided written informed consent. The Brazilian Ethics Committee (CONEP) approved this study (no. 875/07).

Tumor staging was reclassified according to the 2002 version of the International Union Against Cancer (TNM). All patients were followed after treatment, and disease recurrence was confirmed histologically. Histologic grade was determined based on the World Health Organization classification.[13] The medical records of all patients were examined to obtain detailed demographic, clinicopathologic, treatment, and follow-up data (Table 1).

Laser-Capture Microdissection and Total RNA Extraction and Amplification

Non-neoplastic samples and tumor samples were laser-capture microdissected using the PixCell II System (Arcturus Engineering, Inc., Mountain View, Calif). The PicoPure RNA Isolation Kit (Arcturus Engineering, Inc.) was used to obtain total RNA according to the manufacturer's recommendations, and the RNA was submitted to 2-round amplification using the RiboAmp kit (Arcturus Engineering, Inc.). The quantification and quality were evaluated using the NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific Inc., Waltham, Mass) and an Agilent Bioanalyzer (Agilent Technologies, Inc., Santa Clara, Calif), respectively. Only RNA samples that had a 28/18S ribosomal ratio >1 were processed.

Complimentary DNA Microarrays and Probes

cDNA clones from open reading frame-expressed sequence tags (ORESTES) representing human genes were successfully amplified by polymerase chain reaction (PCR), purified using G50 resin (Amersham, Little Chalfont, United Kingdom), and spotted onto glass slides using a Flexsys robot (Genomic Solutions Ltd., Huntingdon, United Kingdom).[14] The cDNA platform representing 2352 genes included 496 negative controls, and 48 positive controls (Q gene fragments from phage lambda). Positive hybridization signals from all spots were considered for evaluation of hybridization quality, normalization, and statistical analysis.

Labeled cDNA was generated using 4 μg of RNA, random hexamer primer (Invitrogen Corporation, Carlsbad, Calif), cyanine 3/cyanine 5 (Cy3/Cy5)-labeled deoxycytidine triphosphate (Amersham), and SuperScript II (Invitrogen Corporation). Equal amounts of Cy3/Cy5-labeled cDNA derived from tumor or morphologically normal tissue were mixed and competitively hybridized. The hybridization experiments were performed with tumor versus normal tissues and included 5 cDNA samples in each group in 4 replicates. Dye swap also was performed in duplicate and was used as a replicate and as a control for dye bias. Data acquisition and normalization were performed as described previously.[14] The genes were classified based on their biologic and signaling processes (Gene Ontology and Kyoto Encyclopedia of Genes and Genomes).[15] The University of California-Santa Cruz Cancer Genomics Browser was used to analyze cancer expression and clinical data.[16]

Primers and Quantitative Reverse Transcriptase-Polymerase Chain Reaction Analysis

cDNAs were synthesized from 2 μg of antisense RNA using Superscript II reverse transcriptase (Invitrogen Corporation) and oligo-dT primers (Invitrogen Corporation). Quantitative reverse transcriptase-PCR was performed in the ABI Prism 7900 (Applied Biosystems, Inc., Foster City, Calif) using SYBR Green (Applied Biosystems, Inc.) in a 20 µL total volume and quality controls as proposed by MIQE (Minimum Information for Publication of Quantitative Real-Time PCR Experiments) guidelines.[17] The reactions were carried out in triplicate (Fig. 1A). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was the most stable control gene from 4 endogenous genes tested (GAPDH, ACTB, HPRT1, and BCR) using the geNorm algorithm.[18] Fold differences in the relative gene expression were calculated using a Pfaffl model.[19]

image

Figure 1. (A,B) Primer characteristics and relative messenger RNA (mRNA) expression levels of Twist-related protein 1 (TWIST1); cadherin 1 type 1 (CDH1); catenin (cadherin-associated protein), β1 (CTNNB1); and vimentin (VIM) were analyzed using quantitative reverse transcriptase-polymerase chain reaction (PCR) in samples of oral cell squamous cell carcinoma (OSCC) and normal epithelium after glyceraldehyde-3-phosphate dehydrogenase (GAPDH) normalization (an asterisk indicates P < .05; t test). (C) Box plots illustrate the relation between mRNA levels and immunohistochemical labeling in the OSCC samples.

Download figure to PowerPoint

Tissue Microarray

Core biopsies were extracted from previously defined areas using a Tissue Microarrayer (Beecher Instruments, Inc., Sun Prairie, Wis). Tissue cores that measured 1.0 mm from each specimen were punched and arrayed in duplicate on a recipient paraffin block. Normal oral tissue samples from surgical margins were included as controls.

Immunohistochemistry

Immunohistochemical reactions were performed as previously described.[20] Incubation with the primary antibodies diluted in phosphate-buffered saline (PBS) were conducted overnight at 4°C for anti-Twist1 (1:100 dilution; Abnova, Jhongli City, Taiwan), anti-E-cadherin (1:400 dilution; Dako, Carpenteria, Calif), anti-β-catenin (1:1500 dilution; Thermo Fisher Scientific Inc.), antivimentin (1:50 dilution; Dako), antivascular endothelial growth factor A (anti-VEGFA) (1:200 dilution; Dako), and anti-Ki-67 antigen (1:300 dilution; Dako). The sections were washed and incubated with secondary antibodies (Dako) for 30 minutes, then the polymer detection system (Dako) was used for 30 minutes at room temperature. Positive controls were included in all reactions in accordance with the manufacturer's protocols. Negative controls consisted of omitting the primary antibody, incubating the slides with PBS, and replacing the primary antibody with normal serum.

The slides were analyzed by 2 certified pathologists who were blinded to clinical information. The reactions were analyzed according to staining intensity, with 0 indicating no visible reaction; 1, weak expression; and 2, strong positivity. Each core was scanned in a low-power field to choose the most stained area predominant in at least 10% of tumor cells.[20] The immunohistochemical analyses were performed in duplicate on different tissue microarray levels, representing 4-fold redundancy for each case. The second slides were 25 sections deeper than the first, resulting in a distance of at least 250 μm between the 2 sections, with different cell samples from each tumor. For statistical analysis, samples were categorized into 2 groups: negative and weakly positive cases versus strongly positive cases.[21]

Statistical Analysis

For frequency analysis in contingency tables, statistical analyses of associations between variables were performed using the Fisher exact test (with significance set at P < .05) and, for continuous variables, the nonparametric Mann-Whitney U test. Overall survival was calculated from the beginning of treatment (surgery) to the date of either death or the last information (for censored observations). Survival probabilities were estimated using the Kaplan-Meier method, and the log-rank test was applied to assess the significance of differences among actuarial survival curves with 95% confidence intervals.

Cell Culture

Normal oral epithelial (NOE) cell (kindly provided by Dr Ala-Eddin Al Moustafa) were isolated from normal human tongue tissue and maintained in culture in keratinocyte serum-free medium supplemented with 5 mg/100 mL bovine pituitary extract, as described previously.[22] The OSCC1.2 cell line was established by our group from a metastatic OSCC and was maintained in culture as previously described.[23, 24]

Small Interfering RNA Expression

We used 2 target sequences for knockdown of TWIST1 (available from GenBank under accession no. NM_000474): 5′-UUGAGGGUCUGAAUCUUGCUCAGCU-3′ and 5′-AGCUGAGCAAGAUUCAGACCCUCAA-3′. Transfections were carried out using 100 nM of small interfering RNA (siRNA) oligonucleotides incubated with DharmaFECT1 (Thermo Fisher Scientific Inc.) in Opti-MEM I reduced serum medium (Invitrogen Corporation) according to the manufacturer's instructions. For in vivo studies, the TWIST1-target siRNA sequence was cloned as inverted repeats into pSR puromycin vector according to the manufacturer's instructions (Oligoengine, Seattle, Wash). PSR was used as a control. These vectors were expressed in target cells using a retroviral system as previously described.[25]

Western Blot Analysis

Total cell extracts were used for Western blot analysis as previously described.[25] Blots were detected using the antibodies for anti-E-cadherin (1:1000 dilution; Cell Signaling Technology, Inc., Beverly, Mass), anti-Twist1 (1:500 dilution; Abnova), anti-β-catenin (1:1000 dilution; Abcam plc., Cambridge, United Kingdom), and anti-GAPDH (1:10,000 dilution; Cedarlane, Burlington, Ontario, Canada), and signals were detected with peroxidase-conjugated secondary antibodies and an enhanced chemiluminescence detection system.

Invasion and Migration Assays

Cell invasion assays were performed using 8-μm porous chambers coated with Matrigel (BD Biosciences, East Rutherford, NJ) according to the manufacturer's recommendations. Cell migration was assayed using a qualitative wound-healing assay. Each experiment was performed at least 3 times, and the results are expressed as the average ± standard deviation. Statistical significance was determined using the Student t test.

In Vivo Preclinical Studies

In vivo studies were conducted in compliance with institutional and federal Canadian guidelines and were approved by McGill University. Exponentially growing cells were suspended in PBS (106 cells per 0.1 mL) and injected orthotopically into the tongues of male severe combined-immunodeficiency (SCID) mice; or, alternatively, cells were implanted into the mice subcutaneously to address distinct tumor microenvironment, as previously described.[23, 24] Tumor length (L) and width (W) were monitored every second day using a caliper that was accurate to 0.5 mm. Tumor volume (V) was calculated using the following formula: V = L × W × 1/2 W. Once the tumor reached the maximal size allowed by institutional guidelines, the mice were killed, and the tumors were dissected and weighted using a precision balance. The invasive phenotype of these tumors was investigated macroscopically and by histologic examination. The results are expressed as the average ± standard deviation (n = 8 mice per condition), and statistical analysis was done using the Student t test.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Twist-Regulated Genes Are Molecular Signatures for Invasive Oral Squamous Cell Carcinoma

We conducted a microarray expression analysis using RNA extracted from matched normal tissue and invasive OSCC samples (Table 1). The results revealed differential expression between normal tissues and advanced OSCC samples for several genes that significantly affected primary functions in cell-cell and cell-matrix interactions and in the signal transduction pathways involved in motility and invasion processes. Among them, Twist1 was strongly correlated with invasive OSCC and was predictive of a poor clinical outcome (Tables 2, 3). In addition, the expression of several Twist-regulated genes (Fig. 1A) was altered, including down-regulation of E-cadherin and up-regulation of vimentin and VEGFA. In agreement with the cDNA microarray results, quantitative reverse transcriptase-PCR analysis confirmed TWIST1 overexpression (P = .0001) and down-expression of cadherin 1 type 1 (CDH1) (P = .050) in progressive OSCC compared with normal epithelium (Fig. 1B). A significant correlation was observed between transcription levels and protein expression in the same set of OSCC samples, as represented in Figure 1C by box-and-whisker plots. In support of our results, analysis of a publically available genomic database revealed an association of TWIST1 overexpression with OSCC and TWIST1 down-regulation in normal tissues (Fig. 2).

Table 2. Protein Expression in Patients with Metastatic and Nonmetastatic Oral Squamous Cell Carcinoma
 No. of Patients (%) 
VariableMetastatic OSCCNonmetastatic OSCCPa
  1. Abbreviations: OSCC, oral squamous cell carcinoma; Twist1, Twist-related protein 1; VEGFA, vascular endothelial growth factor A.

  2. a

    All P values indicate a statistically significant difference.

Twist1   
Negative4 (30.8)31 (68.9).013
Positive9 (69.2)14 (31.1) 
E-cadherin   
Negative7 (70)10 (19.6).001
Positive3 (30)41 (80.4) 
Beta-cadherin   
Negative10 (83.3)10 (19.61)< .001
Positive2 (16.7)41 (80.4) 
Vimentin   
Negative5 (38.5)41 (85.4).001
Positive8 (61.5)7 (14.6) 
VEGFA   
Negative6 (50)43 (86).006
Positive6 (50)7 (14) 
Ki-67   
≤25%3 (27.3)35 (63.6).026
>25%8 (72.7)20 (36.4) 
Table 3. Correlations Among Protein Expression and Clinical Stage, Lymph Node Status, and Histologic Grade in Patients With Oral Squamous Cell Carcinoma
 Tumor Classification No. (%)aLymph Node Status No. (%)aHistologic Grade No. (%)a
VariableT1 + T2T3 + T4PN0N+P123P
  1. Abbreviations: Twist1, Twist-related protein 1; VEGFA, vascular endothelial growth factor A.

  2. a

    Percentages are based on the patients who had complete information.

  3. b

    These P values indicate a statistically significant difference.

Twist1          
Negative15 (44.1)19 (55.9).96231 (91.2)3 (8.8).035b7 (23.3)7 (23.3)16 (53.4).356
Positive10 (43.5)13 (56.5) 16 (69.6)7 (30.4) 5 (22.7)9 (40.9)8 (36.4) 
E-cadherin          
Negative2 (12.5)14 (87.5).005b10 (58.8)7 (41.2)< .001b4 (26.7)5 (33.3)6 (40).651
Positive23 (53.5)20 (46.5) 40 (93.0)3 (6.7) 7 (20)9 (25.7)19 (54.3) 
Beta-catenin          
Negative5 (26.3)14 (73.7).047b10 (50.0)10 (50)< .001b6 (33.3)7 (38.9)5 (27.8).074
Positive23 (53.5)20 (46.5) 42 (97.7)1 (1.3) 5 (15.2)8 (24.2)20 (60.6) 
Vimentin          
Negative24 (54.5)20 (45.5).005b41 (91.1)4 (8.9).001b8 (21.6)11 (29.7)18 (48.7).704
Positive2 (13.3)13 (86.7) 8 (53.3)7 (46.7) 4 (28.6)5 (35.7)5 (35.7) 
VEGFA          
Negative21 (43.8)27 (56.2).89644 (91.7)4 (8.3).001b7 (17.1)13 (31.7)21 (51.2).372
Positive5 (41.7)7 (58.3) 7 (53.9)6 (46.1) 4 (36.4)3 (27.2)4 (36.4) 
Ki-67          
≤25%17 (44.7)21 (55.3).41836 (94.7)2 (5.3).007b6 (18.2)10 (30.3)17 (51.5).726
>25%9 (34.6)17 (65.4) 19 (70.4)8 (29.6) 6 (27.3)6 (27.3)10 (45.4) 
image

Figure 2. Signatures of Twist-related protein 1 (TWIST1) and related genes from the University of California-Santa Cruz Cancer Genomics Browser are shown. Each column corresponds to a single sample, and each row corresponds to a biomolecular entity related to the current study. The genomic heat map was organized according to normal samples versus tumor samples. The TWIST1 gene is overexpressed in tumor samples but is mostly absent in normal cases. Chr indicates chromosome; CDH1, cadherin 1 type 1; CTNNB1, catenin (cadherin-associated protein), β1; VIM, vimentin; VEGFA, vascular endothelial growth factor A; Ki67, Ki-67 antigen.

Download figure to PowerPoint

Proteins Encoded by Twist-Regulated Genes Are Differentially Expressed in Invasive Oral Squamous Cell Carcinoma

To confirm the genomic results, protein expression levels were determined for Twist1 and selected targets, including E-cadherin, β-catenin, vimentin, and VEGFA (Fig. 3A-F). A significant association was observed between lymph node metastasis and increased protein levels of Twist1 (P = .035) and Ki-67 (P = .007) (Table 3). OSCC that presented with vascular embolization had an increased cell proliferation index based on positive Ki-67 status (P = .029). The tumors with increased Ki-67 had up-regulation of both Twist1 (P = .051) and VEGFA (P = .031) protein expression (results not shown). Positivity for E-cadherin was more intense, but not statistically significant, in well differentiated tumors compared with moderately or poorly differentiated tumors. However, the absence or reduced expression of E-cadherin was associated with advanced clinical stage (P = .005) and lymph node involvement (P < .001) (Table 3).

image

Figure 3. Immunohistochemical detection of (A) Twist-related protein 1 (Twist1), (B) E-cadherin, (C) β-catenin, (D) vimentin, (E) vascular endothelial growth factor A (VEGFA), and (F) Ki-67 antigen is illustrated. Nuclear immunoreactivity for (A) the transcriptional repressor Twist1 and (F) Ki-67 antigen was easily identified. (B) E-cadherin had sharply demarcated membrane staining observed in well differentiated tumor areas. (C) Beta-catenin was affected by the loss of expression in most samples. (D) Intracytoplasmatic labeling was observed for vimentin, especially in metastatic samples (original magnification ×400 in A and C-F, ×200 in B).

Download figure to PowerPoint

Of significant relevance to OSCC progression, we observed higher expression levels of Twist1 (P = .013) in distant metastasis compared with nonmetastasized tumors (Table 2). In Twist1-overexpessing tumors, negative expression of E-cadherin was associated with high expression levels of vimentin (P < .001); and the highest expression was observed in tumors in advanced stages (P = .005) (Table 3).

Elevated Twist Expression Correlates With a Poor Prognosis

To determine whether Twist1 could have a prognostic value, survival probability was compared between patients who had lung metastasis and patients who had good outcomes (mean follow-up, 157 months). The median overall survival for the metastatic group was 11.5 months versus 88 months for the nonmetastatic group (long-rank test; P < .0001). The 5-year overall survival rates in patients who had aggressive OSCC and patients who had favorable outcomes were 23.7% and 84.7%, respectively. The mean disease-free survival in patients who presented with distant metastasis was 18.6 months (range, 5-78 months; standard deviation, ±19 months). A significantly lower survival probability was verified in patients who had tumors with Twist1 protein overexpression (log-rank test; P = .0310) (Fig. 4).

image

Figure 4. Overall survival was short for patients with oral squamous cell carcinoma who had Twist-related protein 1 (Twist1) overexpression (log-rank test; P = .0310). The solid green line indicates negative Twist1 expression; dotted red line, Twist1 overexpression (Kaplan-Meier test).

Download figure to PowerPoint

Twist1 Is a Potential Therapeutic Target to Interfere With Oral Squamous Cell Carcinoma Invasiveness

Because Twist1 is strongly expressed in patients with advanced OSCC, and its expression level is correlated with metastasis formation and poor survival, we investigated the potential relevance of this transcription factor as a therapeutic target. We addressed the impact of Twist1 manipulation on the EMT phenotype using an NOE cell line, which was established from normal human tongue tissue, and the OSCC1.2 cell line, which was established from a poorly differentiated, metastatic, human oral cancer (stage T4N2b) that had vascular, lymphatic, and perineural invasion.[23, 24] Exposure of NOE cells to TGFβ induced a clear EMT phenotype in almost 100% of the population (Fig. 5A). In contrast, a majority of OSCC1.2 cells expressed an EMT morphology, and exposure of these cells to TGFβ further induced the morphologic characteristics typical of EMT; this coincided with Twist1 overexpression and down-regulation of E-cadherin (Fig. 5B) as well as increased cell invasion capacity (Fig. 5C). Down-regulation of Twist1 using RNA interference (Fig. 5D) significantly prevented the oncogenic features of these invasive OSCC cells. Compared with control cells infected with empty viral particles, down-regulation of Twist1 expression clearly decreased cell proliferation as well as invasive potential using the Boyden chamber assay (P < .005) (Fig. 5E).

image

Figure 5. (A) Transforming growth factor β (TGFβ)-driven epithelial-mesenchymal transition (EMT) was able to change the morphology to fibroblast-like in normal oral epithelial (NOE) cells. OSCC1.2 indicates an oral squamous cell carcinoma cell line. (B) Western-blot analysis confirmed reduced expression of E-cadherin and increased levels of Twist-related protein 1 (Twist1) after TGFβ-driven EMT. C indicates control. (C) An invasion assay was conducted before and after TGFβ treatment using the Boyden chamber assay. The bar graph represents the mean number (±standard error [SE]) of invaded cells (P < .005). GAPDH indicates glyceraldehyde-3-phosphate dehydrogenase. (D) Western blot analysis revealed the efficient down-regulation of Twist1 obtained by small interfering RNA (siRNA) sequences compared with their matched control cells (C), which expressed the pSR puromycin vector (PSR). (E) The invasion of OSCC1.2-TWIST1-siRNA and the OSCC1.2-PSR control was determined using the Boyden chamber assay. The bar graph represents the mean number (±SE) of invaded cells (P < .001).

Download figure to PowerPoint

In vivo, mice that were implanted with OSCC1.2 cells in the tongue had shorter survival because of fast-growing tumors and respiratory problems caused by lung metastatic lesions, which were observed in >50% of animals through pathologic examination (Fig. 6A-C). E-cadherin immunostaining was observed in normal epithelium but was reduced significantly in the primary tumor and at the metastatic site (Fig. 6D-F). An inverse correlation was observed for Twist1 in the same areas (Fig. 6G-I). Equally important, Twist1 down-regulation inhibited tumor growth when cells were implanted either orthotopically into the tongues of SCID mice or subcutaneously (P < .05) (Fig. 6J-K). No macroscopic lung metastasis was observed after Twist1 down-regulation, whereas the control mice had to be killed because of lung metastatic lesions.

image

Figure 6. An in vivo mouse model reveals normal epithelium from (A) the tongue, (B) the primary tumor, and (C) lung metastasis (hematoxylin and eosin [H&E] stain). (D) E-cadherin immunostaining was observed in normal epithelium but was significantly reduced (E) in the primary tumor and (F) at the metastatic site. OSCC1.2 indicates an oral squamous cell carcinoma cell line. (G-I) An inverse correlation was observed for Twist-related protein 1 (Twist1) in the same areas. (J,K) Tumor weights are illustrated at the time mice were killed (n=8) from animals implanted with OSCC1.2-puromycin vector control (c) cells or with OSCC1.2-TWIST1-small interfering RNA (siRNA)-expressing cells implanted either (J) orthotopically in the tongue of severe combined-immunodeficiency mice or (K) subcutaneously (P < .05).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Transcriptional gene expression profiling revealed the predictive power of gene signatures for cancer progression.[26, 27] In this study, laser-microdissected, invasive tongue carcinomas were evaluated using a cDNA microarray platform that contained genes critical to the biology of cancer metastasis. We identified several genes that were differentially expressed. Not surprisingly, most of the genes associated with OSCC invasiveness have been implicated in the regulation of cell-matrix interaction, loss of intercellular junctions, cell scattering, cytokinesis, angiogenesis, and migration, all of which are important processes of metastasis development.[6, 8, 28] With regard to the signature of OSCC progression, it is significant that we identified Twist1 as a predominant marker (based on high levels of expression). In particular, a significant correlation was observed between high Twist1 expression, tumor cell proliferation, and lymph node and distant metastasis. In addition to the prognostic value of Twist1 for OSCC clinical cases, our preclinical studies established a potential role for Twist1 as a therapeutic target. We demonstrated that the inhibition of Twist1 in an OSCC cell line, which was isolated from metastatic oral cancer, lead to the significant inhibition of cell invasion in vitro and to tumor growth in mice.

Twist1, a highly conserved, basic helix-loop-helix transcription factor mapped at 7q21.2, has a bifunctional role, acting as an activator or a repressor, depending on post-translational modifications and physiologic contexts.[29, 30] Twist1 induces gene transactivation through cis-binding to E-box regulatory regions, which are present in several target genes, and this involves complex homodimerization and heterodimerization mechanisms regulated by protein phosphorylation.[29] In the case of gene repression, Twist1 can repress genes by regulating chromatin remodeling through histone acetyltransferase-dependent/histone deacetylase-dependent mechanisms and through the inhibition of DNA binding activity of transcription factors.[30]

The implication of Twist1 in cell migration is attributed primarily to its ability to contribute to EMT, eg, through the down-regulation of E-cadherin and the up-regulation of mesenchymal markers like vimentin, fibronectin, and N-cadherin, as noted above.[1-5, 7, 26, 27, 31] Previous studies have indicated that Twist1 promotes cell proliferation, migration, and expression of a primitive ECM, thus promoting an undifferentiated state.[31] In addition, Twist1 contributes to the EMT phenotype, which has been associated with resistance to chemotherapy and relapses.[32] We observed that, as noted above (see Results), several of the EMT-associated genes regulated by Twist1 were altered in our advanced OSCC cases, including Twist-activated genes like E-cadherin, β-catenin, and VEGFA.

Other mechanisms by which Twist1 can contribute to OSCC aggressiveness may include effects on other genes known to be associated with cancer metastasis, eg, activation of the cell adhesion protein periostin (POSTN), the inhibition of c-MYC, and p53-dependent apoptosis enhancing cell survival, or the up-regulation of AKT2. Also, Twist1 can regulate cell motility through activation of the promotility Rho GTPase RAC1, because it has been demonstrated that Twist1, in cooperation with BMI1, can suppress let-7i expression, which, in turn, can lead to RAC1 activation by the up-regulation of NEDD9 and DOCK3.[30, 33] Although the mechanism underlying Twist1 overexpression in advanced OSCC remains to be established, it has been demonstrated that Twist1 is up-regulated in response to the activation of several factors, including activation of WNT/β-catenin signaling, SRC, STAT3, HIF-1α, and integrin-linked kinase.[30] It is noteworthy that WNT/β-catenin was altered in our cohort of patients with advanced OSCC. We observed that β-catenin staining was predominately cytoplasmic and/or nuclear in patients with metastatic OSCC. Nuclear translocation of β-catenin can promote the activation of E-cadherin transcriptional repressors, including Twist1, a mechanism implicated in promoting EMT in various systems of embryonic development and tumor progression to metastasis.[34] In this study, we observed that inactivation of Twist1 conferred the morphologic transition of OSCC from a fibroblastic to an epithelial appearance, which was accompanied by a gain of epithelial cell markers like E-cadherin.

These findings highlight the utility of Twist1 and regulated genes as a molecular signature for invasive OSCC. Moreover, our results in OSCC preclinical models support the potential of targeting Twist1 as a therapeutic target. It has been established that several transcription factors play key roles in the cell signaling that drives metastasis; however, translation of these findings into clinical applications has been hampered in part by the difficulty of targeting the transcription factor surface motifs required to achieve high-affinity binding and the difficulty in getting small molecules or peptides to reach the nuclear compartment.[35, 36] However, recent advances in the design of small molecules and peptides and of nanotechnologies for drug-delivery approaches are opening new avenues in drug discovery to target transcription factors like Twist1.[37] In summary, this study provides insights into the relevance of Twist1 and regulated genes for the progression of poorly differentiated OSCC to metastasis and highlights their potential to be exploited as therapeutic targets for advanced OSCC.

FUNDING SUPPORT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

This work was supported by the Fundacao de Amparo a Pesquisa do Estado de Sao Paulo (FAPESP), the Canadian Institutes for Health Research (CIHR), and in part by the Canadian Cancer Society and the Quebec Breast Cancer Foundation.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES