Cell culture

Human tongue squamous cell carcinoma cells HSC-3 (JCRB 0623; Osaka National Institute of Health Sciences, Osaka, Japan), SAS (JCRB 0260; Osaka National Institute of Health Sciences, Osaka, Japan) and human dysplastic oral keratinocytes DOK (European Collection of Cell Cultures 94122104, Salisbury, Wilts., UK) were cultured in 1:1 DMEM/F-12 (Invitrogen) supplemented with 100 U/ml penicillin, 100 µg/ml streptomycin, 50 µg/ml ascorbic acid, 250 ng/ml fungizone, 5 µg/ml insulin (bovine pancreas), 0.4 µg/ml hydrocortisone (all from Sigma-Aldrich), and 10% heat-inactivated fetal bovine serum (FBS). For zymography, fetal bovine serum was replaced by 0.5% lactalbumin (Sigma-Aldrich). Human bone marrow-derived BMMSCs were originally obtained from patients operated for hip fracture or osteoarthritis. The ethical committee of Oulu University Hospital had approved the study protocol (Statements 4/2000,58/2009 and 21/2011; Research Diary 180/2001 and 12/2004) and the patients had given their informed written consent for participation in the study. The BMMSCs used in this study were harvested and cultured as described previously [25][26]. All experiments were carried out with cells with low passage numbers 3 - 4. Human gingival fibroblasts (GF) used in this study were obtained from biopsies of healthy gingiva as described earlier [27]. Carcinoma-associated fibroblasts (CaDEC12) [11]derived from a specimen of tongue SCC as well as normal oral fibroblasts (NOFs) [28] were cultured in the same media as GFs [27]. All cells were cultured in a humidified atmosphere of 5% CO₂ at 37°C. In BMMSC or GF co-cultures with HSC-3, SAS or DOK cells BMMSC or GF culture media was used.

Organotypic invasion assay

The organotypic invasion assay and the quantification of invasion were performed as described previously [29]. In monoculture assays 2.0 x 10⁵ - 4 x 10⁵ HSC-3 or SAS cells or 2 x 10⁵ DOK cells were cultured on the top of myoma disk for 10 - 14 days. In co-culture assays 0.5 - 1.5 x 10⁵ BMMSCs were added together with cancer cells on the top of the myoma disks (OTSCC:BMMSC ratio varied from 2:1 to 5:1). The histological sections of myoma disks were stained with monoclonal pancytokeratin antibody (DAKO, clone AE1/AE3). The average of invasion area or depth of HSC-3, SAS, DOK or any other control cells grown in monoculture was set as 100 %. For the inhibition of invasion, 50 µg/ml of monoclonal antibody against human CCL5 (R&D Systems, MAB678), CXCL1 (R&D Systems, MAB275) or isotype matched mouse normal IgG (Jackson Laboratories) were added to cell culture media.

Proliferation assays

Quantification of proliferation in the organotypic invasion assay was performed from triplicate myoma disks per cell line used. Histological sections cut from the disks representing inner parts of the disk were immunostained with polyclonal antibody against Ki67 (Abcam, #15580). Cell proliferation rate was determined as the percentage of Ki67-expressing cells among all cells per microscopic field in the cell layer on the top of the myoma disk. Cancer cells were first labeled with Vybrant® CM-DiI cell-tracking solution (Life Technologies) for the discrimination of cancer cells from BMMSCs. Altogether four microscopic fields were counted per section (12 fields per sample). For the cell culture proliferation assay 2 x 10⁴ HSC-3 cells were cultured as a monoculture or as a co-culture with 1 x 10⁴ BMMSCs or GFs on four chamber slides (Lab-Tek) per test for 24 h. The slides were stained with polyclonal antibody against Ki67 and Alexa Fluor® 488 secondary antibody (Molecular Probes) and counterstained with DAPI. 10 microscopic fields per chamber slide were examined and cell proliferation rate was determined as a percentage of Ki67-expressing cells among all cells per microscopic field.

Scratch assay

A total of 1 x 10⁵ HSC-3 cells and 1 x 10⁵ BMMSCs or GFs were co-cultured in 10% FCS complete culture media in 24-well plates (Costar) in triplicate wells until confluence after which a wound was made by using a pipet tip. The wells were washed with cell culture media and the wounded areas were photographed immediately after wounding (0 h) and again in the end of the study (20 h) when cells were stained with crystal violet. The size of the wound area and the closure of wound were analyzed with ImageJ (version 1.45s, http://imagej.nih.gov/ij/). The completely healed wound area was set to a value 100.

Zymography

Zymography performed as previously described [29] was used to detect the expression of metalloproteinases 2 and 9 (MMP-2 and -9) in HSC-3, SAS and DOK cell culture media after o/n incubation with 100 ng/ml of recombinant CCL5 (rCCL5) (R&D Systems, #278-RN).

Microarray

HSC-3 cells were cultured as a monoculture or in a co-culture with BMMSC cells (cell ratio 1:1) on 6-well plates in two separate experiments for 24 h. Next HSC-3 cells were sorted out from co-cultures with FACS (FACScan, Becton Dickinson). RNA was extracted and purified from the cells with the Qiagen RNeasy kit according to the manufacturer's instructions and pooled for microarray.

In another set of co-culture assay, 80% confluent cultures of HSC-3 were grown in HSC-3 media with 2% FBS for 24 h. The media were collected, centrifuged, transferred to BMMSC cultures and incubated for 24 h, after which the cells were collected. RNA was extracted as described above from three separate co-culture sets.

Affymetrix Human GeneChip Arrays were used for Microarray analysis. Experimental procedures for GeneChip were performed according to the Affymetrix GeneChip Expression Analysis Technical Manual. The expression data was analyzed using the Affymetrix GeneChip Operating System (Affymetrix) and dChip software [30]. The array data were also deposited in the GEO (accession number GSE44458).

Determination of chemokine levels

For the measurement of CCL5 expression 2 x 10⁴ HSC-3, SAS and DOK cells were cultured as a monoculture or in co-culture with 2 x 10⁴ BMMSCs or GFs (cell ratio 1:1 to 2:1) on 24-well plates (Costar). Cell culture media was collected 40 h after cell plating and filtered. The expression of chemokine CCL5 was determined from cell culture media by Quantikine CCL5/RANTES Immunoassay (R&D Systems).

Patient samples and clinicopathological data

Archival specimens of 105 OTSCC and ten lymph node metastases (pN1) from patients, surgically treated at the Oulu University Hospital between the years 1981-2009, were retrieved from the Oulu University Hospital, Department of Pathology, Oulu, Finland. Additional nine patients signed out as tumor-free lymph nodes (pN0) were retrieved from The Chaim Sheba Medical Center, Tel Hashomer, Ramat Gan, Israel. The median age of 114 patients was 64 years (range 27-87). The median follow-up time was 101 months (range 18-298 months) in the surviving patients (n=53). The median follow-up time for the study patients was 47 months (range 1-298 months). Patient survival data was acquired from Statistics Finland and other relevant data from patient records (Table 1). We could not retrieve treatment data from three of the patients and survival data from two of the patients. The retrieval of the patient data was approved by the local ethics committees and Finnish National Supervisory Authority for Welfare and Health.

Immunohistochemistry

Immunostaining was performed on all of our OTSCC samples, which were previously selected to be representative of the tumour mass in the resected specimens. After a high temperature antigen retrieval method with citrate buffer for CCL5 (REAL Target Retrieval Solution, pH 6; Dako, Glostrup, Denmark) or Tris/EDTA for PINP (10 mmol/L Tris, 1 mmol/L EDTA, pH 9), sections were blocked by incubation with normal goat serum for 30 min. Slides were incubated overnight at 4°C with the polyclonal goat anti-human CCL5 antibody (#AF-278-NA, R&D Systems) at a 1:100 dilution. For the polyclonal rabbit anti-human antibody PINP [31] slides were incubated for one hour at room temperature with the PINP primary antibody at a 1:5000 dilution. Samples were incubated with secondary antibodies specific for each species. For detection we used Dako EnVision kit (Dako) with Diaminobenzidine (Dako basic DAB-kit) as a chromogen. Counterstaining was done in the Dako Autostainer (Dako, Copenhagen, Denmark). In negative controls IgG of each respective species was used instead of primary antibody.

Immunohistochemical assessment of PINP and CCL5 expression

Expression of PINP and CCL5 was evaluated through immunohistochemistry. Samples were analyzed by three independent researchers (K.M., J.H.K., and T.S. or S.S., J.H.K., and T.S.; lymph nodes free of metastatic disease were examined by D.D. and M.V.). Different patterns of classifying expression were chosen. At first samples were classified based on the overall difference of staining between the superficial and invasive areas of the tumors were categorized (staining appears more intensive in superficial areas of the tumor or staining appears more intensive in invasive areas of the tumor). The percentage of positive cells in the stromal and tumor cells in both the superficial and invasive areas of the tumor was scored on a four point scale (0 = 0%, 1 = 0-25%, 2 = 25-50% and 3 = >50%, titled no, low, moderate and high expression, accordingly). Additionally, the presence of PINP staining in blood and lymphatic vessels was assessed. The subepithelial stroma of morphologically normal epithelium was also stained. At dysplastic mucosa sites both epithelial cells and the subepithelial stromal cells were analyzed.

RNA Interference

Three commercial CCR5 short hairpin RNA (shRNA) oligonucleotides (Thermo Scientific #VGH5518-98903425, #VGH5518-99159618, #VGH5518-99294601) and non-silencing control (Thermo Scientific #RHS4348) were obtained from Thermo Scientific Open Biosystems GIPZ Lentiviral shRNAmir Library (Thermo Scientific). The transductions of HSC-3 cells with viral and control vectors were performed according to manufacturer’s instructions with puromycin (Sigma-Aldrich) selection. The efficiency of knockdown by CCR5-shRNA was determined by flow cytometric analysis using monoclonal anti-CCR5 antibody (BD Pharmingen, #556903) and isotype matched IgG (BD Pharmingen, #555576) as a control. About 70 % knockdown of CCR5 was obtained compared to non-silencing control cells with one of the commercial oligonucleotides (Thermo Scientific #VGH5518-99159618).

Statistical analysis

All assays were repeated 2 - 4 times, except the 3D invasion assay with SAS and DOK cells that was performed only once with triplicate myoma disks per mono- or co-culture. Differences in cell proliferation and wound healing, invasion area and depth were evaluated by using Student’s t-test. In all experiments, a p-value of less than 0.05 was considered statistically significant. For statistical analyses of the patient material we used PASW Statistics 18 (IBM corp.) software. A chi-square-test was used to calculate statistically significant differences between prognostic and clinicopathological variables. Survival tables were calculated according to the Kaplan-Meier method, and were compared with the log-rank test. Multivariate survival and recurrence analyses were done with the Cox proportional hazards model using the following covariates: gender (male or female), age at the time of diagnosis (<55 yrs, 55-70 yrs and >70 yrs), tumour stage (1–4) and tumour histologic grade (well =1, moderately = 2 and poorly = 3 differentiated carcinomas according to World Health Organization head and neck carcinoma classification (2005) together with PINP expression pattern variables as indicated. Cox regression was done using backward stepwise selection of variables, and a p value of 0.05 was adopted as the limit for inclusion of a covariate.

BMMSCs induce tongue cancer cell invasion in the 3D in vitro organotypic model

The effect of BMMSCs on tongue cancer cell invasion was examined in the myoma organotypic model [29] that is more faithful to the human TME than previous animal tissue derived models. Human tongue carcinoma cells, HSC-3, SAS or dysplastic DOK cells were cultured as a monoculture or in co-culture with BMMSCs on the top of myoma disks. After 10 - 14 days invasion area and depth were quantitated by analyzing images taken from pancytokeratin stained histological sections from monoculture and co-culture disks. As shown in Figure 1, BMMSCs induced an increase in invasion area, and even a clearer effect was seen on the invasion depth in HSC-3-BMMSCs co-cultures (Figure 1A-1B). A slight increase in the invasion area was seen also with SAS cells in the presence of BMMSCs, although not statistically significant. However, there seemed to be no effect on invasion depth (Figure 1C-1D). Interestingly, BMMSCs inhibited the invasion of dysplastic DOK cells (Figure 1D-1E) suggesting a different effect of BMMSCs on carcinoma cells compared to non-malignant cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. BMMSCs enhance tongue cancer cell invasion in 3D organotypic model.

Bone marrow-derived BMMSCs enhance the invasion of HSC-3 cells (A, B), but not dysplastic cells (E, F) in 3D myoma organotypic model [29]. Minor increase, although not statistically significant, was seen in the invasion area of SAS (C,D). Histological sections obtained from myoma disks were stained with monoclonal pancytokeratin antibody (DAKO, clone AE1/AE3) and photographed at 100 x magnifications with a DMRB photo microscope connected to DFC 480 camera using QWin V3 software (Leica Microsystems). Scale bar 100 µm.

https://doi.org/10.1371/journal.pone.0077692.g001

To study if the increased invasion of HSC-3 cells was due to increased cell proliferation, HSC-3 and SAS cells were cultured on top of myoma disk as a monoculture or co-culture with BMMSCs or GFs. Histological sections obtained from these disks were immunostained with the proliferation marker Ki67 and positive cells were calculated as described in Materials and Methods. Cell proliferation levels were remarkably lower in HSC-3 and SAS co-cultures with BMMSCs as compared to co-cultures with GFs (Figure 2A-2B). To confirm that the effect seen in histological sections came from the reduced proliferation of cancer cells, HSC-3 cells were cultured as a mono- or co-culture on cell culture chamber slides with BMMSCs labeled with a fluorescent dye. After co-culture overnight, cells were stained for Ki67. As shown in Figure 2C, BMMSCs inhibited HSC-3 cell proliferation significantly also in the cell co-culture assay. The viability of HSC cells remained still high and no sign of apoptosis was detected (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. BMMSCs inhibit cell proliferation in tongue cancer cells.

Histological sections obtained from myoma disks were stained with polyclonal anti-Ki67 antibody. BMMSCs inhibit proliferation of HSC-3 and SAS cells in organotypic invasion model (A, B) as well as in cell culture assays (C), and shift HSC-3, but not dysplastic DOK cells towards a migratory phenotype with less cell-cell contacts (D, arrowheads). Normal gingival fibroblasts, GFs, promote wound closure more than BMMSCs potentially partly due to the increase in cell proliferation of cancer cells (E). Scale bar 50 µm.

https://doi.org/10.1371/journal.pone.0077692.g002

As proliferation was not increased, to evaluate if the increased invasion was a consequence of a higher cancer cell migration a scratch assay was performed with HSC-3 and DOK cells seeded alone or together with BMMSCs or GFs to 24-well plates. BMMSCs shifted HSC-3 cells, but not dysplastic DOK cells, towards a migratory phenotype with lamellipodia structures and less cell-cell contacts (Figure 2D). However, GFs promoted faster wound closure than BMMSCs, which may be partly due to effects of serum on the proliferation capacity of cancer cells (Figure 2D-2E)

The expression of chemokine CCL5 is up-regulated in BMMSCs after interaction with cancer cells

BMMSCs were then cultured with conditioned media obtained from HSC-3. Although several genes were up-regulated, the highest increase in gene expression in BMMSCs was seen in chemokine genes (Table 2). When the chemokine expression was examined at protein level the expression of CCL5 was found to be primarily a result of the interaction between HSC-3 and BMMSC cells rather than from the co-cultures of HSC-3 cells and GFs (Figure 3A). Similar findings were obtained from co-cultures of SAS and BMMSCs, but not from DOK cells interacting with BMMSCs (Figure 3A). At protein level, the up-regulation of most of the chemokines listed on the Table 2 was found to originate from OTSCC cell lines interactions with both BMMSCs and GFs. Some chemokines, such as CCL2, CCL20 and CXCL8, were also up-regulated as a result of interactions between BMMSCs and DOK cells, as studied in immunoassays (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. The expression of chemokine CCL5 is up-regulated in co-cultures of BMMSCs cells and OTSCC.

Co-culture of OTSCC cells with BMMSCs results in the high expression of chemokine CCL5 in HSC-3 and SAS cells, but not in dysplastic DOK cells (A). Function-blocking monoclonal antibody against CCL5 (Mab CCL5) slightly inhibits BMMSC enhanced invasion area, but has no effect on the invasion depth, whereas the antibody against CXCL1 does not increase the inhibitory effect in OTSCC invasion (B, C).

https://doi.org/10.1371/journal.pone.0077692.g003

Function-blocking antibody against CCL5 inhibits BMMSC increased invasion area

We next explored the importance of CCL5 in the spread of tongue cancer, since CCL5 has been shown to promote the motility of oral cancer cells [32]. CCL5 was also shown to be important in tumor progression of several cancers, such as breast and colorectal carcinoma [17,33]. Therefore, we tested the effect of CCL5 function-blocking antibody on OTSCC and BMMSC co-cultures in the 3D invasion model. We also tested the potential effect of function blocking antibody against CXCL1 on invasion, since CXCL1 was also expressed from BMMSC-HSC-3 interaction and it has previously been demonstrated to be present in cancers including breast, lung, colorectal and prostate cancers either supporting or suppressing tumor progression [34–36]. In oral cancer CXCL1 has been suggested to have a role in tumor progression since in patient samples the expression of CXCL1 has been shown be associated with leukocyte infiltration, lymph node metastasis, and angiogenesis [37]. As shown in Figure 3, the CCL5 function–blocking monoclonal antibody was able to significantly attenuate BMMSC-promoted HSC-3 cell invasion area, but had no effect on invasion depth (Figure 3B-3C). CXCL1 did not have an effect on invasion as it did not increase the inhibitory effect of CCL5 in the 3D invasion model (Figure 3B-3C).

In patient samples CCL5 is expressed mainly by inflammatory cells and some cancer cells, but only sparse CAFs are CCL5 positive

Next, to evaluate the importance of CCL5 expression in patient tumor samples for diagnostics of tongue cancer, histological sections representing different stages of dysplastic lesions and tongue carcinomas from human patients were stained with anti-CCL5 antibody. The expression of CCL5 was mostly detected in inflammatory cells and in some cancer cells (Figure 4A-4B), but only sparse CAFs were CCL5 positive, as assessed by immunohistochemical co-localization of CCL5 and CAF marker αSMA (data not shown). However, the expression of CCL5 was not associated with the clinical outcome of the OTSCC patients (not shown). The expression of CCL5 was further studied in normal primary oral fibroblasts (NOF, [28]) and CaDEC12 cells [11]. Immunoassay detecting secreted CCL5 in cell culture media showed very low or no CCL5 expression in normal fibroblasts, but considerable higher expression in CaDEC12 cells (Figure 4C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. CCL5 expression in OTSCC patient samples.

CCL5 is mainly expressed by inflammatory cells in TME and some cancer cells in patient samples as shown by immunostaining with Pab CCL5, but only sparse CAFs are positive for CCL5 (A, B). CaDEC12 cells, the primary cultures of CAFs obtained from tongue cancer patients, express relative amount of CCL5 compared to NOFs, normal oral fibroblasts (C). Scale bar 100 µm.

https://doi.org/10.1371/journal.pone.0077692.g004

The role of CCL5/CCR5 axis is not critical in BMMSC increased tongue cancer invasion

The CCL5 signaling in cancer cells has been proposed to be mediated mainly, but not necessary exclusively, via CCR5 receptor [38]. Both HSC-3 and SAS cells were shown to express this receptor (Figure 5A). In co-culture roughly all cancer cells expressed CCR5, however, in monocultures approximately only 25 % of cancer cells were positive for CCR5 suggesting that a contact with stromal cells or CCL5 induction is needed for higher CCR5 expression (flow cytometry analysis, data not shown). The association of CCL5 and CCR5 expression has been proposed recently also by others [39,40]. As the CCL5-CCR5 axis has been shown to promote the motility of human oral cancer cells in vitro [32] and to induce the expression of MMP9 [32], we also examined the effect of rCCL5 on MMP9 expression levels in HSC-3 and SAS as well as in dysplastic DOK cells. Similarly to previous observation by Chung and co-workers [32] rCCL5 clearly increased the expression of MMP9 in HSC-3 cells and SAS cells, but not in dysplastic DOK cells (Figure 5B). However, rCCL5 did not induce epithelial-mesenchymal transition (EMT) as analyzed by Western blotting with anti-vimentin antibody, or proliferation of cancer cells (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Signaling via CCL5/CCR5 is not critical in OTSCC invasion.

The CCL5 receptor CCR5 is expressed in HSC-3 and SAS cells, but not in BMMSCs (A). rCCL5 can induce the expression of MMP9 in HSC-3 and SAS, but not in DOK cells (B). The silencing of the expression of CCR5 receptor in HSC-3 cells downregulates the gene expression about 70 % compared to control cells and blocks the signaling increasing the expression of MMP9 in cancer cells (C, D). CCR5 knockdown have minor or no effect on BMMSC enhanced invasion of HSC-3 cells in 3D organotypic model (E, F).

https://doi.org/10.1371/journal.pone.0077692.g005

To explore the importance of CCL5/CCR5 axis in the spread of tongue cancer, we performed a 3D invasion study with SAS and HSC-3 OTSCC cells with reduced CCR5 expression and with function-blocking antibody against CCL5. First, we inhibited the expression of CCR5 by transducing HSC-3 cells with CCR5-shRNA and produced a stable cell line with ~70 % inhibition of CCR5 expression and reduced response to rCCL5 induction (Figure 5C, 5D). In the 3D co-culture invasion assay the CCR5 knockdown cells invaded equally or than the wild-type or control HSC-3 cells transduced with non-silencing shRNA (Figure 5E, 5F). Similar results were also obtained with a monoclonal antibody against CCR5 (data not shown).

Expression of collagen and epithelial plasticity components are induced in cancer cells after BMMSC interaction

Potential mechanisms of the invasion promoting effect of BMMSCs on cancer cells were further studied by microarray analysis. HSC-3 cells were used for the experiments, since they responded more clearly to BMMSCs than SAS cells, especially in the 3D invasion model. HSC-3 cells were cultured as a monoculture or co-culture with BMMSC cells on 6-well plates for 24 h after which HSC-3 cells were sorted out from co-cultures with FACS (not shown) and RNA was extracted for microarray analysis. The overall gene profile of HSC-3 cells after interaction with BMMSCs showed up-regulated gene expression of ECM components, with the highest induction occurring for type I collagen mRNA expression. In addition to various collagens, BMMSC interaction with HSC-3 cells has also up-regulated modulators of ECM, such as lysyl oxidase (LOX) and other markers of epithelial plasticity, such as cadherin-11. Molecules associated with cell motility and cancer invasion, such as α-actinin-4 mRNAs, were also induced in HSC-3 cells by the interaction with BMMSCs (Table 3).

Type I procollagen is highly expressed in the invasion front in tumor samples obtained from oral tongue cancer patients and correlated with worse prognosis

To further evaluate the relevance of type collagen I expression originating from the cancer cell-stromal interactions to the aggressiveness of OTSCC, we performed immunohistochemistry for type I procollagen fragment (PINP). PINP has previously been shown to be associated with clinical outcome in breast and ovarian cancers, but so far has never been shown in OTSCCs [22-24,31]. In our 114 OTSCC cases, PINP was not found in epithelial cells in histologically normal nor dysplastic peritumoral tissue (Figure 6A-6B). In some tumors PINP was almost absent (Figure 6F), however, most of the cases showed high expression by both carcinoma and TME cells (Figure 6E, 6G, 6H). In general overview, PINP expression was more intense either in more invasive or superficial areas of the tumor (Figure 6C-D). Similar to the pattern of expression found in the primary tumors, PINP was also expressed in OTSCC lymph node metastases (Figure 6I). Furthermore, in a couple of lymph nodes originally signed out as free of metastases (pN0), PINP was expressed in a cluster of cells revealing previously unrecognized occult carcinoma metastases which also stained positive for pancytokeratin (not shown). We then compared the pattern of PINP expression and patient clinical data. In our survival analyses, there was a correlation between worse prognosis and expression of PINP by OTSCC cells at invasive areas (p=0.018) (Figure 7A). Increased PINP expression by both carcinoma and stromal cells at the invasive areas was also correlated to worse prognosis (p=0.004) (Figure 7B). Both the high expression of PINP in SCC cells (HR 1.728 95% CI [1.072 - 2.910]) as well as increased PINP expression by both carcinoma and stromal cells (HR 1.924, 95% CI [1.127 - 3.285]) at invasive areas were identified as independent indicators of overall survival along with patient age and high tumor stage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. OTSCC samples show staining for type I collagen N-terminal propeptide (PINP).

The expression of PINP was not found in any histologically normal epithelial cells of the carcinoma sections, except in a few of the mesenchymal cells within wound bed (arrow heads). The wound epithelial border is marked by dotted line (A). PINP expression was not found in dysplastic epithelium but was present in a few spindle cells in the submucosa (B). In various cases PINP expression was more intense in either superficial (C) or invasive (D) areas (arrow heads). Some cancers were mainly negative for PINP staining with a few sporatic positive areas (F), whereas other cases showed intense staining for both cancer and mesenchymal cells (E). In a couple of cases PINP was expressed in the periferal cells of invading islands (G) or mainly in the spindle carcinoma associated mesenchymal cells (H). Lymph node metastases were occasionally positive for PINP (I). Scale bar 200µm .

https://doi.org/10.1371/journal.pone.0077692.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Survival analyses for PINP expression.

In our set of samples, high PINP expression (PINP >50 %; n=41, 27 deaths in the group) compared to low expression (PINP 0-50%, n=71, 32 deaths) in OTSCC cells at the invasive front (>50%) correlated with increased overall deaths (A). In general overview, the increased expression of PINP in both OTSCC and stromal cells in the invasive areas of the tumor was observed in 76 patients (35 deaths in the group), whereas increased expression of PINP in superficial areas was observed in 36 patients (24 deaths). The higher expression of PINP in invasive areas associated with worse prognosis, p=0.004 (B).

https://doi.org/10.1371/journal.pone.0077692.g007