Abstract
The VAV proteins VAV1, VAV2 and VAV3 have been identified as important molecules in tumorigenesis, tumor growth and cell migration. In addition to the full-length isoforms, a much shorter family member, VAV3.1, also known as VAV3 isoform 2, is known to be differentially expressed in a broad variety of tissues. Furthermore, VAV3.1 was shown to be down-regulated in cultured keratinocytes by the growth factors epidermal growth factor (EGF) EGF and transforming growth factor beta (TGFβ) TGFβ which in turn play important roles in the pathogenesis of oral squamous cell carcinoma (OSCC). Herein we showed that VAV3.1 is underexpressed in OSCC tissue samples compared to corresponding normal mucosa. We further demonstrated a trend of distinctive down-regulation of mRNA for VAV3.1 in tissues of locally advanced OSCC that have already metastasized to regional lymph nodes, indicating an increased malignant potential of tumors with low VAV3.1 mRNA expression. Moreover, in other studies a correlation between increased VAV3 expression and cancer progression was shown. In the present study, the analyzed OSCC tissue samples showed no significant change of VAV3 mRNA expression. Taken together, our data support the hypothesis that molecular interactions and signaling cascades of VAV3 can be regulated or directed by the competing molecule VAV3.1. Additionally, discrete and different functions of VAV3.1 in metastasis and tumorigenesis are conceivable.
VAV1, VAV2 and VAV3 share a typical arrangement of eight functional domains, which allow interactions with multiple other proteins and therefore can orchestrate various signaling cascades, including tumorigenesis, tumor growth and cell migration (1-4). In addition to the full-length VAV3, a much shorter family member, VAV3.1, is differentially expressed in a broad variety of tissues (5). This transcript variant encodes for the three carboxy-terminal functional domains (Src-homology domains SH3, SH2 and SH3) of VAV3 (Figure 1). Additionally, these SH2 and SH3 domains have a conserved sequence homology of 46% to 70% within the corresponding domains of VAV1 and VAV2. The postulated that VAV3.1 protein comprises the 287 amino acids, including a short stretch of eight unique amino-terminal amino acids that differ from VAV3 and which do not have sequence homology to that of VAV1 nor VAV2 genes.
The few data published about VAV3.1 show a different mRNA expression pattern compared to VAV3 in a wide variety of tissues, indicating a possible regulatory aspect of VAV3.1 for VAV3 function. Interestingly, the expression of the VAV3.1 transcript can be down-regulated by up to 10-fold in normal human epidermal keratinocytes and in the immortalized keratinocyte cell line HaCaT by stimulation with transforming growth factor β (TGFβ) or epidermal growth factor (EGF) (5-7), whereas VAV3 mRNA expression was not altered in these experiments. However, the enzymatic function of the VAV3 protein can be regulated through phosphorylation after EGF stimulation (8). In a different study, down-regulation of VAV3.1 expression was seen after astragaloside treatment of the hepatocellular carcinoma cell line HepG2, which resulted in reduced colony formation (9).
Experiments using a truncated 287-amino acid VAV3 construct differing from the postulated VAV3.1 protein by only the unique eight amino-terminal amino acids showed biological interactions such as binding to the ROS proto-oncogene receptor (ROS) or phosphorylation after receptor stimulation (8).
It has been postulated that VAV3.1 can regulate VAV3 or even VAV1 and VAV2 functions by substitution or competition, and can act as an adaptor molecule in certain signal transduction pathways (10). Furthermore, a dominant-negative effect of VAV3.1 was shown in a cell proliferation assay using the human ALVA-31 prostate cancer cell line. VAV3-transfected ALVA-31 cells showed increased proliferation compared to VAV3- and VAV3.1-co-transfected cells under androgen (R1881) stimulation. Additionally, cell proliferation was reduced by transfection of VAV3.1 into androgen receptor (AR)-transfected ALVA-31 cells, indicating a modulation of cell proliferation via a VAV3-VAV3.1-AR pathway (11). Genomic analysis revealed the possibility that VAV3.1 is transcribed from its own promoter, rather than being a splice variant of the VAV3 transcript. This is especially interesting in terms of different regulatory pathways for the expression of VAV3.1 and VAV3 mRNAs.
Primer and probe design for VAV3, VAV3.1 and Porphobilinogen deaminase (PBGD) amplicons in 5’-3’ orientation. For primer and probe design, the primer design program Primer3 (33) and the Roche Universal Probe Library probe Assay Design program (http://lifescience.roche.com/webapp/wcs/stores/servlet/CategoryDisplay?catalogId=10001&tab=Assay+Design+Center&identifier=Universal+Probe+Library&langId=-1) were used. The designed amplicons were intron spanning to prevent amplification of genomic DNA. The primers were synthesized and purified by high-performance liquid chromatography (TIB Molbiol, Berlin, Germany). The sequence-specific fluorescent probes were chosen from the Roche Universal Probe Library. The VAV3 amplicon is located at the VAV3-specific 5’-end of VAV3 cDNA whereas the VAV3.1-specific amplicon is partially located within the 110 bp specific VAV3.1 sequence and partially in the shared VAV3-VAV3.1 sequence.
Since the first descriptions of VAV3 (12, 5) many different VAV3 signaling cascades and functions have been identified. VAV3 is phosphorylated after ligand stimulation of different receptor tyrosine kinases such as EGF receptor (EGFR), plateled-derived growth factor receptor (PDGFR), ROS receptor, insulin receptor (IR) and insulin-like growth factor I receptor (IGFR). Moreover, stimulation of the B-cell and the T-cell receptors (BCR and TCR) induces phosphorylation of VAV3 (13, 14, 8).
VAV3 has guanine nucleotide exchange factor activity for the activation of Rho and Rac GTPases, by which diverse cellular processes, including oncogenic transformation, cell growth and survival, actin cytoskeleton organization, differentiation, gene transcription and migration, can be regulated (15).
The oncogenic potential of VAV3 has been shown in several studies. VAV3 or VAV3 mutants can induce focus formation, colony formation, morphological changes, and cell motility in cell culture assays as well as angiogenesis in in vivo assays through downstream pathways such as phosphatidyinositol-3-kinase, mitogen activated protein kinase and Rho family GTPases (16, 17).
Furthermore, VAV3 activates the estrogen receptor and is postulated to be involved in breast cancer (18). Additionally VAV3 activates the AR and is up-regulated in prostate cancer and an involvement in development, progression and recurrence of prostate cancer is postulated (19, 20, 21, 11). An involvement of the EGF-VAV3 signaling axis was also shown in prostate cancer cells (22). Expression of VAV3 was associated with poor patient survival in glioblastoma (23).
The potential ability of VAV3 to facilitate metastasis is of particular interest. Correlations of VAV3 expression and invasive and migratory capabilities in glioblastoma and neuroblasts were recently shown (24, 23). Furthermore, VAV3 also seems to play an important role within the tumor microenvironment. Impaired tumor growth, invasiveness and survival was seen in VAV3-deficient host cells, suggesting that the malignant potential of VAV3 is at least partially localized in the tumor stroma (25). Consistent with this, it was shown that VAV3 and VAV2 are cooperatively involved in autocrine and paracrine programs that regulate keratinocyte proliferation and favor the initiation and promotion of skin tumors (26)
Oral squamous cell carcinoma (OSCC) accounts for 2-3% of all cancer manifestations (27). The individual prognosis is particularly dependent on tumor localization and tumor stage. The tumor stage depends on tumor size, involvement of regional lymph nodes, distant metastasis and differentiation of tumor tissue. Surgical treatment includes total resection of the tumor and neck dissection. Radiotherapy with/without chemotherapy, as well as anti-EGFR therapy are options for adjuvant therapies of locoregionally advanced tumors or for inoperable tumors. The overall five-year survival rate for patients diagnosed with OSCC is about 50%, ranging from 15% to 85% (28).
Structural domains of VAV3 and VAV3.1. VAV3 consists of eight functional domains: calponin homology, CH; acidic, AC; Dbl homology, DH; pleckstrin homology, PH; zinc finger, ZF; Src homology 3, SH3; and Src homology 2, SH2. VAV3.1 is a short variant of VAV3 and mainly consists of the three carboxy-terminal domains of VAV3 (SH3, SH2 and SH3). The amino-terminal end of VAV3.1 contains a stretch of a unique 5’ sequence. Hence both variants can be distinguished. VAV3 has a length of 847 amino acids, VAV3.1 codes for 287 amino acids.
Relative mRNA expression of VAV3 (A) and VAV3.1 (B) in normal and tumor tissue. The quantitative real-time polymerase chain reaction (qRT-PCR) data from 20 tumor samples and neighboring normal mucosa were normalized using the internal reference gene porphobilinogen deaminase (PBGD). The relative mRNA expressions are shown in box plot diagrams. The boxes represent the 25th and 75th percentiles, the whiskers represent the 9th and 91st percentiles. Outliers are plotted as dots. The solid line in the box represents the median, and the dashed line represents the mean.
In the present study, we investigated the expression of VAV3 and VAV3.1 in OSCC tissue samples, which seemed particularly applicable for studying VAV3 and VAV3.1 expression, since growth factor assays in normal human keratinocytes and in the immortalized keratinocyte cell line HaCaT, had shown that EGF and TGFß induced down-regulation of VAV3.1 (5). Expression of these growth factors and their receptors is known to be associated with proliferation, migration and invasion, resulting in poor prognosis in OSCC (29, 30, 31).
Materials and Methods
Sample collection and study design. In a prospective study design, 20 specimens from patients with OSCC and their adjacent non-neoplastic native matched tissue were obtained. Evaluation of clinical data and experimental procedures were approved by the local authorities and by the Ethics Committee of the University of Lübeck (approval number 06-205).
The patients' ages ranged from 26 to 87 years (median=60 years), 14 of the patients were male and 6 were female. Tumor localizations were the floor of the mouth in 9, alveolar ridge in 6, tongue in 4 and buccal mucosa in 1 patient. According to the TNM classification (32), the tumor size was as follows: pT1 in 4 patients, pT2 in 7 patients, pT3 in 6 patients and pT4a in 3 patients. In 12 patients, no lymph node metastasis was evident (pN0), whereas in 8 patients histopathologically verified lymph node metastasis (pN1-pN2b) was seen. None of the patients had distant metastases (cM0). The histopathological grading showed moderately differentiated carcinoma (G2) in 12 patients and poorly differentiated carcinoma (G3) in the remaining 8 patients.
Representative OSCC tissue specimens and adjacent normal mucosa were sampled from the macroscopic center of the tumor directly after its surgical resection. The samples were frozen immediately in liquid nitrogen until RNA preparation. A fraction of the frozen tumor tissue was additionally analyzed for the presence and amount of tumor cells using histomorphometric analysis. The main part of the resected tissue underwent standardized analysis for pathological staging and grading.
Histomorphometrical analysis. A representative fraction of the frozen tumor tissue was used for histomorphometrical analysis. Each specimen was taken from a site adjacent to the fraction used for mRNA analysis and was fixed by neutral phosphate-buffered formalin (4% buffered formalin; BÜFA, Hude, Germany) and embedded in paraffin (Leica Microsystems, Wetzlar, Germany). Sections of 2 μm-thick were cut by microtome, placed on Super Frost® glass slides (Menzel, Braunschweig, Germany) and underwent conventional hematoxylin-eosin staining. For evaluation of the percentage of tumor tissue in each sample, areas of tumor cells were outlined and the percentage of tumor cells in relation to the total field of view was calculated in three consecutive fields using an Axioplan® microscope (Zeiss, Jena, Germany) and the professional Soft Imaging System analySIS® (Soft Imaging System GmbH, Münster, Germany).
RNA preparation. For RNA preparation, a total of 24 mm3 tissue of each frozen sample was used. After mortar and pestle tissue homogenization, total RNA was extracted using the RNeasy Mini kit and an additional DNAse step on the silica column according to the manufacturer's instructions (Qiagen, Hilden, Germany). In order to assess the RNA quality, RNA samples were electropheretically separated and analyzed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Böblingen, Germany). Only RNA samples of high quality with high RNA integrity numbers (RIN >6.3) were used for further analyses.
Primer and probe design. For primer and probe design, the primer design program Primer3 (33) and the Roche Universal Probe Library probe Assay Design program (Roche Applied Science; http://lifescience.roche.com/webapp/wcs/stores/servlet/CategoryDisplay?catalogId=10001&tab=Assay+Design+Center&identifier=Universal+Probe+Library&langId=-1) were used.
The designed amplicons were intron spanning to prevent amplification of genomic DNA. The primers were synthesized and purified by high performance liquid chromatography (TIB Molbiol, Berlin, Germany). The sequence-specific fluorescent probes were chosen from the Roche Universal Probe Library. The VAV3 amplicon is located at the VAV3-specific 5’-end of VAV3 cDNA, whereas the VAV3.1-specific amplicon is partially located within the 110 bp-specific VAV3.1 sequence and partially in the shared VAV3/VAV3.1 sequence (Table I).
Quantitative real-time reverse transcription PCR (qRT-PCR). For first-strand synthesis, each reaction contained 3.6 μg total RNA, 1 μl 100 μM oligo(dT), 1 μl 100 μM random hexamer primer, 1 μl (200 U) SuperScript II reverse Transkriptase (Invitrogen, Life Technologies, Darmstadt, Germany), 1 μl 100 μM oligo(dT), 1 μl 100 μM random hexamer primer, 5 μl 10 mM dNTPs, 5 μl 0,1 M dithiothreitol (DTT), 10 μl 5× buffer (Invitrogen), and 1 μl (40 U) RNasin Ribonuclease Inhibitor (Promega, Mannheim, Germany). After incubation of the RNA at 65°C for 10 min, the reaction mix was added to it to a final volume of 50 μl and incubated for 10 min at 37°C, followed by incubation at 42°C for 1 h and a denaturation step for 5 min at 95°C. Additional samples without reverse transcriptase were incubated to control for genomic DNA contamination in the qRT-PCR steps.
For the enzymatic quantitative real-time PCR (qPCR) reactions, 5 μl cDNA and 15 μl master-mix from the LightCycler TaqMan Master Kit were mixed in glass capillaries and were temperature cycled in a Roche LightCycler 2.0 according to the manufacturer's instructions (Roche, Mannheim, Germany).
The temperature profile for the qPCR started with an initial denaturation step at 95°C for 10 min and was followed by 45 cycles at 95°C for 10 s, 59°C for 30 s and 72°C for 24 s. With qPCR, the generated PCR products were measured in real time during the cycling process. The relative amount of mRNA of the analyzed genes was determined from the amount of generated PCR products. All PCR reactions were performed in duplicates and the mean value was used for further analysis. qRT-PCR expression data of VAV3.1 and VAV3 mRNAs were normalized to that of an internal reference mRNA (porphobilinogen deaminase, PBGD, housekeeping gene) which was amplified in duplicate parallel reactions and the relative mRNA expression was calculated (relative expression=2− [ΔCPsample−ΔCPcontrol]). In order to analyze the differential gene expression, the qRT-PCR data from the tumor samples were compared pairwise to those of the corresponding normal mucosal samples. Differences were statistically evaluated using a Wilcoxon signed-rank test and a randomization test (pairwise fixed reallocation randomization test) using REST-384-software version 1 (34).
Additionally the VAV3.1/VAV3 ratio was calculated. The qRT-PCR products were furthermore analyzed by agarose gel electrophoresis to confirm correct product size.
Results
Gene expression of VAV3 in OSCC. VAV3 mRNA expression pattern in OSCC did not significantly differ when compared to that of normal mucosa (Figure 2A). Furthermore, no significant differences in VAV3 mRNA expression were detected among subgroups regarding lymph node status, tumor size or histopathological grading.
It is noteworthy that expression of VAV3 was very low in the oral mucosa and the OSCC samples analyzed. The crossing point (CP) of the employed amplicon was an average of 32 PCR cycles. At such low expression levels, reliable data are difficult to obtain because at high cycle numbers systematic errors can arise. In the present study, all PCR experiments were carried-out in duplicates and the average error was less than 5%. We, therefore, conclude that despite the low expression profile and the resulting high cycle numbers, the analysis is still reliable. To additionally exclude a merely inefficient amplicon, the VAV3 amplicon was tested in other tissues which showed higher expression rates, with CPs at considerably lower cycle numbers. These included Merkel cell carcinoma (CP=25 cycles), pleomorphic adenoma and cystadenolymphoma of the parotid gland (CP=26 cycles) (data not shown), confirming the generally low VAV3 mRNA expression in OSCC and normal mucosa.
VAV3.1 is down-regulated in OSCC. VAV3.1 mRNA expression analysis revealed a significant two-fold down-regulation of the mRNA expression level of VAV3.1 in OSCC tissue compared to normal oral mucosa (n=20; p=0.006, Wilcoxon signed-rank test) (Figure 2B).
VAV3.1/VAV3 mRNA ratio in normal and tumor tissues. The mRNAs from eight tumor samples and neighboring normal mucosa from patients with proven lymph node metastasis were analyzed by quantitative real-time polymerase chain reaction (qRT-PCR). The 3.5-fold difference of the VAV3.1/VAV3 mRNA ratio in the carcinoma samples compared to the neighboring normal mucosa is clear (n=8; p=0.009, Wilcoxon signed-rank test).
The CP of the VAV3.1 amplicon employed was an average of 26.5 PCR cycles, whereas the expression of the housekeeping gene PBGD was higher, with an average of 23 PCR cycles at the CP. The difference in expression levels between VAV3.1 and the housekeeping gene seemed small enough to sufficiently control for a substantial error due to differences in cycle numbers.
To further analyze the down-regulation of VAV3.1 in the tumor samples, the expression data were tested for correlation with clinical and histopathological data. In regard to histopathological grading and tumor size, no correlation was observed. However, particularly with regard to the correlation of lymph node metastasis and VAV3.1 mRNA expression, we observed a trend for a profound down-regulation of VAV3.1 mRNA in the samples of tumor associated with lymph node metastasis [6-fold down-regulation; n=8, p=0.05, pairwise fixed reallocation randomization test (34)]. Tumors without lymph node metastasis (n=12) showed no significant regulation.
Differential gene expression of VAV3.1 and VAV3. Assuming that the VAV3 signaling cascade can be regulated by coordinated expression of both variants VAV3 and VAV3.1, the expression ratio of the two genes was analyzed. This simple calculation allows for direct analysis of the expression data without the need for further normalization by an internal reference or housekeeping gene.
Subdividing the 20 tissue pairs according to the histopathological grade or the tumor size and extent revealed no significant differences in the expression ratios.
Again we observed a significant correlation in patients with evident lymph node metastasis. In these more aggressive carcinomas with poor prognosis, the ratio of VAV3.1 to VAV3 expression was lowest and was 3.5-fold different compared to the neighboring normal mucosa (n=8; p=0.009, Wilcoxon signed-rank test) (Figure 3).
Histomorphometrical analysis. Representative fractions of the frozen tumor tissues were analyzed histomorphometrically. The amount of tumor tissues compared to the adjacent non-cancerous tissue in the specimens ranged from 4% to 47%, with a mean of 26% and a standard deviation of 14%. This indicates the variability of OSCC tissue specimens regarding the mere number of tumor and stroma cells. Furthermore, no correlation was seen between the proportion of tumor cells and the mRNA expression of VAV3 or VAV3.1, indicating that the changes in mRNA expression rather depend on the individual genetic characteristics of the tumor or stroma than on a general change found in all OSCC tissues.
Discussion
To our knowledge, this is the first study to show a differential gene expression of VAV3.1 mRNA in cancer tissue. OSCC samples exhibited significant down-regulation of VAV3.1 mRNA expression compared to normal adjacent oral mucosa. On the other hand, no significant change was observed for VAV3 mRNA expression in these tissues. Looking at the VAV3.1/VAV3 ratio, we found a significant difference in the tissue pairs of patients with evident lymph node metastasis. These data substantiate a hypothetical model where VAV3.1 can modulate biological functions of VAV3 by competition in VAV3 pathways. Down-regulation of VAV3.1 could, therefore, have similar effects to up-regulation of VAV3. This hypothesis is supported by data from other studies where up-regulation of VAV3 has been associated with increasing malignant potential in different tumor types as glioblastoma, and breast, prostate and gastric cancer (23, 19, 18, 35, 20). Furthermore, it can be assumed that in some tumor entities, VAV3.1 mRNA is not regulated or is even not expressed and therefore VAV3 functions would have to be primarily regulated by the level of VAV3 expression or activation. Looking at more than 50 different cell lines and tissues, a great diversity in the expression ratio of VAV3 and VAV3.1 mRNA was seen, not only including samples with very low or no VAV3.1 mRNA expression, but also samples with a moderate-to-high expression rate in about half of the analyzed tissues and cell lines (5, 8). Taken together, these data substantiate the hypothesis of a tissue-specific expression ratio of VAV3.1/VAV3 and an interacting modulatory function of these proteins.
Other VAV3-independent functions of VAV3.1 are conceivable, but will have to be analyzed in further studies.
It is noteworthy that several previous VAV3 expression studies did not distinguish between VAV3 and VAV3.1 transcripts because PCR amplicons are often located within the common 3’ sequence. In the present study, we used such an amplicon as a control and did see the expression rate consequently being higher than that of each transcript separately, most likely due to additional amplification from more than one transcript variant (data not shown). Future experiments should generally aim to distinguish between both variants.
In contrast to VAV3, the presence of the VAV3.1 protein has not yet been verified. The available VAV3 antibodies are directed against amino-terminal VAV3 structures beyond the VAV3.1 sequence, or would not differentiate between VAV3 and VAV3.1 in the common carboxy-terminal sequence. In further studies, it would be beneficial to establish a VAV3.1-specific antibody to be able to provide further evidence for the presence of the VAV3.1 protein and to characterize its functions.
VAV3.1 might also be involved in the regulation of functions of the other family members VAV1 and VAV2. Due to the conserved amino acid sequence of VAV3.1 found in VAV1 and VAV2, ranging from 46 to 70% identity within the SH2 and SH3 domains, interaction with the same proteins is conceivable. This hypothesis is substantiated by the fact that interaction of VAV1, VAV2 and VAV3 with the same proteins was already shown (13) and redundant roles were described (36).
No comparable transcript variants of VAV1 or VAV2 have yet been identified, although an even shorter transcript variant of VAV1, VAV-T, containing the carboxy-terminal SH3 domain, has been described. Its expression was detected only in murine testicular germ cells (37), whereas VAV3.1 is expressed in a wide variety of human tissues and cell lines, as well as in murine tissues (5, 8).
Roles for VAV2 in OSCC have already been described. Activated VAV2 appears to modulate cellular invasion by activation of the GTPases RAC1 and cell division control protein 42 homolog (CDC42) in OSCC (38). Additionally, a connection to the EGFR pathway via EGFR-VAV2-RAC1 has been shown (39). Furthermore VAV2 can regulate EGFR signaling by slowing receptor internalization and degradation through its interaction with endosome-associated proteins (40). As EGF also plays an important role in VAV3 and VAV3.1 regulation, a tight connection of VAV2, VAV3 and VAV3.1 signaling seems plausible.
As we did not observe any correlation between the proportion of cancer or stroma cells and mRNA pattern of VAV3.1 in the analyzed tissue samples, it would be beneficial to further subdivide the tissue specimens in order to identify the cell type that undergoes VAV3.1 regulation. It was recently shown that stromal changes such as TGFβ1 activation or alpha smooth muscle actin (α-SMA) expression promote invasion and metastasis and correlate with OSCC mortality (41). Therefore it would be of great interest to evaluate the effect of VAV3.1 and VAV3 on extracellular matrix protein production which is at least partly responsible for metastasis of OSCC (42).
The aggressive behavior of malignant tumors is amongst other factors caused by a dysregulation of the actin system. Cell functions such as adhesion, movement, exocytosis, endocytosis and cell division are therefore influenced (43). The already known functions of VAV3 include regulation of the actin system via RHO GTPases (44, 1, 17), substantiating the relevance of VAV3 and possibly VAV3.1 in invasive growth and metastasis in OSCC and other cancer entities.
Therefore therapeutic approaches altering the expression of VAV3 or VAV3.1 to treat patients with OSCC might be considered.
The possible role of VAV3.1 as a diagnostic marker is especially interesting. Patients with potentially metastasizing tumors usually undergo cervical lymph node dissection, not only in cases of evident lymph node metastasis, but also as a preventative treatment in cases of probable micrometastasis. Additionally, radiotherapy can be applied as a treatment for regional lymph nodes. The possible substantial morbidity resulting from these kinds of treatment could be avoided in selected cases if a biological marker for the probability of metastasis could be assessed. For patients with tumors with low risk of metastasis, additional treatment and the resulting morbidity could consequently be avoided; for patients with high risk of metastasis, the treatment options could, in contrast, be utilized to the fullest extent.
Acknowledgements
The Authors would like to thank Dr. Merle Hanke for her excellent scientific advice and her support in data analysis, and the staff of the Department of Pathology, University of Lübeck for their sincere support and for providing technical equipment. This work was supported by a Research Grant of the University of Lübeck.
Footnotes
-
This article is freely accessible online.
-
Conflicts of Interest
The Authors declare there exist no conflicts of interest with regard to this study.
- Received January 19, 2015.
- Revision received February 4, 2015.
- Accepted February 9, 2015.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
