J Oral Pathol Med (2015) 44: 674–679 © 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
doi: 10.1111/jop.12321
wileyonlinelibrary.com/journal/jop
Single cell migration in oral squamous cell carcinoma – possible evidence of epithelial–mesenchymal transition in vivo David H. Jensen1, Jesper Reibel2, Ian C. Mackenzie3, Erik Dabelsteen4 1
Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet and University of Copenhagen, Copenhagen, Denmark; 2Oral Pathology and Oral Medicine, Department of Odontology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark; 3Blizard Institute of Cell and Molecular Science, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London, UK; 4Department of Odontology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
BACKGROUND: The invasion of cancer cells into the surrounding normal tissue is one of the defining features of cancer. While the phenomena of tumour budding, epithelial–mesenchymal transition and the presence of myofibroblasts have independently been shown to be related to a poor prognosis of oral carcinomas, their relationship has not been examined in detail. METHODS: Paraffin-embedded tissues from 28 patients with oral squamous cell carcinomas were stained with antibodies to cytokeratin, a-SMA, vimentin, E-cadherin, N-cadherin and Twist and evaluated for their expression in relation to invasive cancer cells and the surrounding tumour stroma. RESULTS AND CONCLUSIONS: A direct, histological relationship between invading, budding tumour cells and myofibroblasts was occasionally seen but was not a general feature. Most of the budding tumour cells at the invasive front had a decreased expression of E-cadherin, but we did not find that this was associated with a consistent or clear increase in either N-cadherin or vimentin. We therefore suggest that the budding of tumour cells is not dependent upon either myofibroblasts or a complete epithelial–mesenchymal transition and that these phenomena most likely represent separate processes in tumour progression. J Oral Pathol Med (2015) 44: 674–679 Keywords: invasion; migration; myofibroblast; oral cancer
Correspondence: David H. Jensen, Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet and University of Copenhagen, 2100 Copenhagen, Denmark. Tel: +4535459776, Fax: +4535452690, E-mail: [email protected] Accepted for publication March 2, 2015
Introduction Most squamous cell carcinomas of the oral mucosa (OSCC) are high to moderately differentiated with major multicellular units having a differentiation pattern very similar to normal tissue and with regions of dedifferentiation with widespread cellular dissociation being found at the invasive front (1). These small groups of cells are often referred to as tumour buds, and increased numbers of these appear to be a strong predictor of a severe prognosis (1, 2). It is known that invading and disseminating carcinoma cells often show great plasticity and undergo epithelial– mesenchymal transition (EMT) (3–5). It is thus recognized that oral carcinoma cells in vitro may migrate as single cells in conjunction with an EMT. It is also well established that activated tumour stromal fibroblasts of the myofibroblast phenotype establish a permissive environment supporting tumour growth associated with secreted cytokines and matrix modifications (6–10). In vivo budding of oral carcinoma cells may show a mesenchymal-like phenotype, as demonstrated by decreased membranous staining of E-cadherin (11, 12), but the relationship at the invasive front between activated fibroblasts and the tumour buds is less clear. In this work, we show that tumour buds are often present without a direct histological relationship to myofibroblasts (13, 14). We also observe that invading tumour cells in vivo may demonstrate reduced E-cadherin expression without a concomitant change to the elongated spindle-shaped morphology as seen in in vitro studies. We demonstrate that these invading cells retain keratin expression and show little or no vimentin expression. Finally, we observe that although single invading cells do occur, the vast majority of tumour buds are connected to larger tumour strands.
Materials and methods Specimens from 28 patients with oral squamous cell carcinomas were chosen at random from the archive at the
Single cell migration in oral cancer Jensen et al.
Department of Odontology, University of Copenhagen. The specimens were archival biopsies routinely fixed in neutral buffered formalin and embedded in paraffin. Spearman’s rank correlation was used to assess the significance of relationships observed. Histology and immunohistochemical procedure Immunohistochemical staining with a pankeratin antibody was used to identify epithelial cells. We demonstrated myofibroblasts by staining with antibodies directed towards a-SMA. Double staining with antibodies against vimentin and keratin was used to identify cells with both epithelial and mesenchymal filaments indicating EMT. Triple staining with keratin, vimentin and E-cadherin antibodies was made to further designate an EMT. Staining was performed on 4-lm sections of paraffin-embedded tissue using a microwave antigen retrieval technique as previously described (15) and working dilutions of the antibodies were established by titration. The antibodies and the retrieval technique used are described in Table 1. When rabbit antibodies were used as the primary antibody, Alexa Fluor-594 goat anti-rabbit IgG (H+L)conjugated antibodies (Life Technologies, Naerum, Denmark) were used as the secondary antibody, whereas staining with mouse monoclonal antibodies used Alexa Fluor 488 conjugated goat anti-mouse antibody (H+L) (Life Technologies, Naerum, Denmark). Triple staining for keratin, vimentin and E-cadherin was performed by first staining for E-cadherin using an immunoperoxidase system on the Ventana Benchmark Ultra autostainer (Roche A/S, Hvidovre, Denmark), with the OptiView detection system (Roche A/S, Hvidovre, Denmark). The plastic coverslip was then dissolved in methanol and the residual adhesive dissolved by placing the slide in xylene. A second round of antigen retrieval was performed in a Dako Link pH 9.0 retrieval solution (DAKO, Glostrup, Denmark), and the tissue section subsequently stained for keratin and vimentin using fluorescence markers as previously described (16). Incubation with irrelevant primary antibodies of the same isotype as the specific antibodies was used as control for the monoclonal antibodies and deletion of the primary antibody served as control for the conjugate. At the invasive front, tumour buds were defined as single cells, or groups of up to four cells, and were selected on the basis of positive cell staining for keratin. The invasive front
was defined as the band of tissue between the tumour front and adjacent normal stromal tissue. The a-SMA staining was semi-quantitatively assessed using a previously described score system (7, 9). In brief, this is described as follows: 0: no staining, 1: moderate staining in a single focus, 2: staining in most parts of the tumour and 3: massive stromal staining. We modified the system and combined 0 and 1, and 2 and 3. The slides were evaluated by two pathologists who were asked to give their a-SMA scores according to these defined criteria, and were not informed about the purpose of the study.
675
3D confocal imaging To investigate to what degree tumour buds represent truly isolated cells or tissue sectioning artefacts, we used confocal microscopy. The tissues used for the 3D reconstruction were cut as thick (25–30 lm) sections. Confocal pictures were acquired using Zeiss LSM 510 (Carl Zeiss, Jena, Germany) or Zeiss LSM 780 (Carl Zeiss) confocal microscopes and Zen software (Carl Zeiss). Images were median-filtered (3 to 3), and the 3D representations were reconstructed from orthogonal projections of 20–65 optical sections in 0.3– 1 lm increments using the Zen software. For each tumour sample, a field containing a high degree of cellular dissociation at the invasive front was chosen for the 3D reconstruction. A total of 10 samples were examined and, to observe the relationship between tumour cells and fibroblasts, the samples were stained for pancytokeratin, a-SMA and DAPI. The analysis used only cells for which the entire cell could be visualized. Cell lines To test whether the tumour buds seen in vivo expressed the same structural proteins as migrating EMT cancer cells observed in vitro, cells of several OSCC cell lines (17) were plated at low density in six well plates. All cell lines were grown in the highly supplemented epithelial growth medium (termed FAD) with 10% FBS (18). Cells with holoclone, meroclone and paraclone morphologies were identified in addition to the subpopulation of migrating individual cancer stem cells. Before staining, cells were fixed for 10 min in ice-cold acetone/methanol, permeabilized in Tween (2% in PBS) and rinsed in PBS. The primary antibody was applied and left overnight at 4°C. They were then stained with the secondary antibodies and mounted with DAPI Prolong Gold Antifade (Life Technologies, Naerum, Denmark).
Table 1 List of antibodies and their dilution used in all experiments Antibody Anti-E-Cadherin Anti-N-Cadherin Anti-Pancytokeratin Anti-a-SMA Anti-b-Catenin Vimentin Twist
Species Mouse Mouse Rabbit Mouse Mouse Mouse Rabbit
Clonality Monoclonal Monoclonal Polyclonal Monoclonal Monoclonal Monoclonal Polyclonal
IgG1 IgG1 IgG2a IgG1 IgG1
Clone NCH-38 6G11 Z0622 1A4 Clone 14 V9 Ab49254
Manufacturer Dako Dako Dako Sigma-Aldrich Ventana Medical Systems Dako Abcam
Retrieval technique HiAR, HiAR, HiAR, HiAR, HiAR, HiAR, HiAR,
pH pH pH pH pH pH pH
a
9.0 9.0a 9.0 9.0 9.0 9.0 9.0
Dilution 1:50 1:200 1:4000 1:400 Pre-diluted 1:800 Titrated until 1:10
HiAR, heat-induced antigen retrieval. a Staining was performed on a Ventana Benchmark Ultra autostainer. J Oral Pathol Med
Single cell migration in oral cancer Jensen et al.
676
A
B
C
Figure 1 Cell line triple stained for (A) keratin (red), (B) vimentin (green) and DAPI (blue). Image (C) is overlay of image (A) and (B). All cells stain for keratin. On the right, a holoclone colony with a single, elongated vimentin-positive, spindle-shaped cell at its periphery (arrow). Vimentin-positive cells with a squamous appearance can be seen in the holoclone colony as well.
Results Staining of cell lines All cells stained positively with antibodies to keratin. Single spindle-shaped cells could be found at the periphery of the holoclones (Fig. 1). Most such cells also stained positively for vimentin, indicating that they had undergone an EMT (Fig. 1B,C). Vimentin expression could also be observed in the cells of holoclone colonies (Fig. 1B). These findings indicate that although oral squamous carcinoma cells may express mesenchymal markers indicative of EMT, they retain keratin microfilaments. We therefore assumed that staining with antibodies to keratin could be used to trace tumour buds in histological sections even if a complete morphological EMT occurs. Histological sections Four cases of the randomly chosen 28 cases were excluded due to lack of tumour buds. The remaining 24 cases showed, based on staining with antibodies to keratin, a great variability in the number of tumour buds at the invasive front. Alpha smooth muscle actin and keratin double staining Eight of 24 biopsies showed tumour buds but no associated myofibroblasts. In five further biopsies, myofibroblasts were present in tumour stroma, mainly between major central tumour islands, but not at the invasive front and without a close relationship of myofibroblasts to the tumour buds (Fig. 2A). The remaining samples frequently showed a histologically close association between myofibroblasts and tumour cells (Fig. 2C). In some tumours, less than half of the tumour buds were associated with myofibroblasts, but in others, all of the tumour buds were closely surrounded by myofibroblasts (Fig. 2B). In these latter samples, myofibroblasts dominated the entire tumour stroma and were thus not specifically localized to areas surrounding the tumour buds (Fig. 2C). Vimentin, keratin and E-cadherin triple staining Vimentin-positive tumour buds were not found in 10 of the 24 specimens, but other specimens showed single cells, or J Oral Pathol Med
small groups of cells that were positive for both vimentin and keratin (Fig. 3D). We found no correlation between the presence of myofibroblasts and number of vimentin-positive tumour buds (P = 0.36). The staining for vimentin was generally weak and located around the nucleus. In only one case was vimentin strongly expressed in keratin-positive tumour cells (Fig. 3D). Reduced expression of E-cadherin was found in cells at the invasive front; however, tumours often showed buds in which some expression of E-cadherin was retained (Fig. 4B,D). The reduced expression of Ecadherin was not seen to be accompanied by an increased expression of vimentin (Fig. 4C,E). 3D confocal imaging Single cells without a clear relationship to other keratinpositive cancer cells were observed in five of 10 samples (Fig. 5). However, although individual optical slices of the invasive tumour front often showed apparently single cells, examination of adjacent slices typically showed them to be connected to other tumour cells.
Discussion Cancer-associated fibroblasts are most often a-smooth muscle actin-positive (19), and other molecules have also been described as markers of activated fibroblasts (20). Staining for such markers is usually only partially overlapping, suggesting the presence of discrete subsets of tumour stromal fibroblasts of which myofibroblasts constitute a major part. We have investigated the distribution of myofibroblasts because clinical and experimental evidence suggests a correlation between stromal myofibroblasts and progression of oral squamous cell carcinomas (21) and because in vitro studies have shown a direct relationship between myofibroblasts, invasion and EMT. We therefore set out to look for in vivo evidence of EMT occurring in tumour buds at the invasive front of OSCC and to examine the relationship between EMT cells and myofibroblasts (13). We assumed that the tumour buds would contain the most aggressively invading cells and that this phenotype would be in part due to stimulation by adjacent myofibroblasts
Single cell migration in oral cancer Jensen et al.
677
A
B
C
Figure 2 Representative images from the invasive front of OSCC specimens that are double and triple stained. The relationship between single cells and myofibroblasts is illustrated. The myofibroblasts are a-SMA-positive (green) with cancer cells keratin positive (red), and nuclei have been stained with DAPI (blue). (A) Invasive single cells (arrow) not surrounded by myofibroblasts. (B) Arrows indicate carcinoma cells that are ‘wrapped’ by a-SMA-positive cells. (C) Carcinoma cells (red) and myofibroblasts (green) are randomly distributed.
A
B
C
D
Figure 3 The invasive front of an OSCC specimen. Sections are stained with (A) vimentin (green), (B) keratin (red). Image (C) is an overlay of A and B, while (D) is a double exposure in which cells that express both proteins appear orange. Cells expressing both vimentin and keratin are marked with arrows.
that are capable of driving a mesenchymal transition of the epithelial tumour cells. Our findings, however, indicate that the presence of invading tumour buds is not dependent on the existence or close proximity of myofibroblasts as has been suggested by in vitro studies (13). Even when myofibroblasts are present, a close relationship between
myofibroblasts and budding tumour cells is not consistently found. We cannot exclude that tumour-activated and growth-stimulating stromal fibroblasts that are not a-SMApositive are present around the tumour buds; leukaemia inhibitory factor has in this way been established as a driver of fibroblast-mediated extracellular matrix remodelling that J Oral Pathol Med
Single cell migration in oral cancer Jensen et al.
678
A
B
C
D
E
F
Figure 4 The invasive front of an OSCC specimen. Representative examples of triple staining for (A) E-cadherin (brown), (B) keratin (red) and (C) vimentin (green). (D) is an overlay of E-cadherin (A) and keratin (B), where keratin-positive cells that are negative for E-cad are marked by arrows. These single cells are also weakly positive for vimentin and negative for E-cadherin, as can be seen from the overlay of E-cadherin and vimentin in (E) and keratin and vimentin in (F).
Figure 5 3D reconstruction of an approximately 30-lm slide of an OSCC specimen. Notice the one isolated cell on the right, which is not related to the tumour island on the right. Also notice that this cell does not have an elongated, mesenchymal-like morphology.
is independent of a-SMA expression (22). A recent study demonstrate that a-SMA-positive fibroblasts have a protective rather than promoting role in tumour progression further illustrates the complexity of the crosstalk between tumour cells and the microenvironment (23). However, it remains obscure how these findings should be interpreted considering the studies showing that tumours with a high proportion J Oral Pathol Med
of myofibroblasts have a poor prognosis (8, 21, 24). Recently, the absence or presence of infiltrating immune cells at the invasive front has been demonstrated to be of prognostic importance and could thus help to explain some of the discordant findings (25). Although absent or reduced E-cadherin staining was frequently found at the invasive front, this was not accompanied by an upregulation of N-cadherin. Also, we did not observe any clear relationship between the loss of E-cadherin and gain of vimentin. This is in agreement with studies that demonstrate that E-cadherin loss is insufficient to induce an epithelial–mesenchymal transition (26, 27). We have also stained both cell cultures and histological sections for Twist, a well-known EMT regulator, but have only found a weak signal in few cultured cells, and none in cells of tumour buds in histological sections (data not shown). We performed 3D confocal microscopy on 30-lm-thick slices to assess whether cells appearing as isolated single cells in thin histological sections were likely to be isolated cells rather than single cells connected to a larger tumour island by finger-like projections not in the plane of section. This technique made it possible to identify single cells that were not continuous with any larger tumour island. Few of the apparently single cells were found actually to be isolated, but most were part of finger-like projections, as previously noted (28). These findings confirm that OSCC in vivo tend to migrate collectively, by so-called chain migration and not predominantly as single cells as seen in vitro (29), and may help to explain the difference between our in vivo findings and previously reported in vitro results of the expression of EMT markers, such as vimentin (17). It cannot, however, be excluded that the invasive cells only
Single cell migration in oral cancer Jensen et al.
variably express vimentin for a short period during active migration, resulting in a sparse expression in histological sections. In support of this, Wong et al. demonstrated that a malignant cell line (MCF-10A) showed a phenotypic plasticity in the expression of biomarkers as cells interconvert between individual and collective migration and vice versa (30). Alternatively, the levels of expression of vimentin in the tumour cells could be so low compared to the vimentin expression in the surrounding stromal cells that it would not be readily apparent with the methods we have used. Indeed, the few carcinoma cells we identified in the histological sections that were positive for vimentin expressed it to a much lesser degree than cells in the surrounding tumour stroma. These cells were also not overtly mesenchymal in morphology, that is elongated, and this could support the idea of a so-called partial EMT, where cells migrate as multicellular groups, but the tip of the migrating cells undergoes a partial EMT (31).
References 1. Bryne M. Is the invasive front of an oral carcinoma the most important area for prognostication? Oral Dis 1998; 4: 70–7. 2. Wang C, Huang H, Huang Z, et al. Tumor budding correlates with poor prognosis and epithelial-mesenchymal transition in tongue squamous cell carcinoma. J Oral Pathol Med 2011; 40: 545–51. 3. Brabletz T. To differentiate or not — routes towards metastasis. Nat Rev Cancer 2012; 12: 425–36. 4. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin Investig 2009; 119: 1420–8. 5. Zlobec I, Lugli A. Epithelial mesenchymal transition and tumor budding in aggressive colorectal cancer: tumor budding as oncotarget. Oncotarget 2010; 1: 651–61. 6. Bello IO, Vered M, Dayan D, et al. Cancer-associated fibroblasts, a parameter of the tumor microenvironment, overcomes carcinoma-associated parameters in the prognosis of patients with mobile tongue cancer. Oral Oncol 2011; 47: 33–8. 7. Leemans CR, Braakhuis BJM, Brakenhoff RH. The molecular biology of head and neck cancer. Nat Rev Cancer 2011; 11: 9– 22. 8. Marsh D, Suchak K, Moutasim KA, et al. Stromal features are predictive of disease mortality in oral cancer patients. J Pathol 2011; 223: 470–81. 9. Sobral LM, Bufalino A, Lopes MA, et al. Myofibroblasts in the stroma of oral cancer promote tumorigenesis via secretion of activin A. Oral Oncol 2011; 47: 840–6. 10. Vered M, Dobriyan A, Dayan D, et al. Tumor-host histopathologic variables, stromal myofibroblasts and risk score, are significantly associated with recurrent disease in tongue cancer. Cancer Sci 2010; 101: 274–80. 11. Wang X, Zhang J, Fan M, et al. The expression of E-cadherin at the invasive tumor front of oral squamous cell carcinoma: immunohistochemical and RT-PCR analysis with clinicopathological correlation. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009; 107: 547–54. 12. Bagutti C, Speight PM, Watt FM. Comparison of integrin, cadherin, and catenin expression in squamous cell carcinomas of the oral cavity. J Pathol 1998; 186: 8–16.
13. Brentnall TA. Arousal of cancer-associated stromal fibroblasts: palladin-activated fibroblasts promote tumor invasion. Cell Adh Migr 2012; 6: 488–94. 14. Gaggioli C, Hooper S, Hidalgo-Carcedo C, et al. Fibroblastled collective invasion of carcinoma cells with differing roles for RhoGTPases in leading and following cells. Nat Cell Biol 2007; 9: 1392–400. 15. Bilde A, von Buchwald C, Dabelsteen E, et al. Molecular markers in the surgical margin of oral carcinomas. J Oral Pathol Med 2009; 38: 72–8. 16. Osman TA, Øijordsbakken G, Costea DE, et al. Successful triple immuno - enzymatic method employing primary antibodies from same species and same immunoglobulin subclass. Eur J Histochem 2013; 57: e22. 17. Biddle A, Liang X, Gammon L, et al. Cancer stem cells in squamous cell carcinoma switch between two distinct phenotypes that are preferentially migratory or proliferative. Cancer Res 2011; 71: 5317–26. 18. Locke M, Heywood M, Fawell S, et al. Retention of intrinsic stem cell hierarchies in carcinoma-derived cell lines. Cancer Res 2005; 65: 8944–50. 19. Thode C, Jørgensen TG, Dabelsteen E, et al. Significance of myofibroblasts in oral squamous cell carcinoma. J Oral Pathol Med 2011; 40: 201–7. 20. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer 2006; 6: 392–401. 21. Kellermann MG, Sobral LM, Silva SDD, et al. Myofibroblasts in the stroma of oral squamous cell carcinoma are associated with poor prognosis. Histopathology 2007; 51: 849–53. 22. Albrengues J, Bourget I, Pons C, et al. LIF mediates proinvasive activation of stromal fibroblasts in cancer. Cell Rep 2014; 7: 1664–78. € 23. Ozdemir Berna C, Pentcheva-Hoang T, Carstens Julienne L, et al. Depletion of carcinoma-associated fibroblasts and fibrosis induces immunosuppression and accelerates pancreas cancer with reduced survival. Cancer Cell 2014; 25: 719–34. 24. Yamashita M, Ogawa T, Zhang X, et al. Role of stromal myofibroblasts in invasive breast cancer: stromal expression of alpha-smooth muscle actin correlates with worse clinical outcome. Breast Cancer 2012; 19: 170–6. 25. Galon J, Mlecnik B, Bindea G, et al. Towards the introduction of the ‘Immunoscore’ in the classification of malignant tumours. J Pathol 2014; 232: 199–209. 26. Chen A, Beetham H, Black MA, et al. E-cadherin loss alters cytoskeletal organization and adhesion in non-malignant breast cells but is insufficient to induce an epithelial-mesenchymal transition. BMC Cancer 2014; 14: 552. 27. Nilsson GM, Akhtar N, Kannius-Janson M, et al. Loss of Ecadherin expression is not a prerequisite for c-erbB2-induced epithelial-mesenchymal transition. Int J Oncol 2014; 45: 82– 94. 28. Bronsert P, Enderle-Ammour K, Bader M, et al. Cancer cell invasion and EMT marker expression: a three-dimensional study of the human cancer–host interface. J Pathol 2014; 234: 410–22. 29. Wolf K, Friedl P. Mapping proteolytic cancer cell-extracellular matrix interfaces. Clin Exp Metastasis 2009; 26: 289–98. 30. Wong IY, Javaid S, Wong EA, et al. Collective and individual migration following the epithelial–mesenchymal transition. Nat Mater 2014; 13: 1063–71. 31. Revenu C, Gilmour D. EMT 2.0: shaping epithelia through collective migration. Curr Opin Genet Dev 2009; 19: 338–42.
679
J Oral Pathol Med
doi: 10.1111/jop.12321
wileyonlinelibrary.com/journal/jop
Single cell migration in oral squamous cell carcinoma – possible evidence of epithelial–mesenchymal transition in vivo David H. Jensen1, Jesper Reibel2, Ian C. Mackenzie3, Erik Dabelsteen4 1
Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet and University of Copenhagen, Copenhagen, Denmark; 2Oral Pathology and Oral Medicine, Department of Odontology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark; 3Blizard Institute of Cell and Molecular Science, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London, UK; 4Department of Odontology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
BACKGROUND: The invasion of cancer cells into the surrounding normal tissue is one of the defining features of cancer. While the phenomena of tumour budding, epithelial–mesenchymal transition and the presence of myofibroblasts have independently been shown to be related to a poor prognosis of oral carcinomas, their relationship has not been examined in detail. METHODS: Paraffin-embedded tissues from 28 patients with oral squamous cell carcinomas were stained with antibodies to cytokeratin, a-SMA, vimentin, E-cadherin, N-cadherin and Twist and evaluated for their expression in relation to invasive cancer cells and the surrounding tumour stroma. RESULTS AND CONCLUSIONS: A direct, histological relationship between invading, budding tumour cells and myofibroblasts was occasionally seen but was not a general feature. Most of the budding tumour cells at the invasive front had a decreased expression of E-cadherin, but we did not find that this was associated with a consistent or clear increase in either N-cadherin or vimentin. We therefore suggest that the budding of tumour cells is not dependent upon either myofibroblasts or a complete epithelial–mesenchymal transition and that these phenomena most likely represent separate processes in tumour progression. J Oral Pathol Med (2015) 44: 674–679 Keywords: invasion; migration; myofibroblast; oral cancer
Correspondence: David H. Jensen, Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet and University of Copenhagen, 2100 Copenhagen, Denmark. Tel: +4535459776, Fax: +4535452690, E-mail: [email protected] Accepted for publication March 2, 2015
Introduction Most squamous cell carcinomas of the oral mucosa (OSCC) are high to moderately differentiated with major multicellular units having a differentiation pattern very similar to normal tissue and with regions of dedifferentiation with widespread cellular dissociation being found at the invasive front (1). These small groups of cells are often referred to as tumour buds, and increased numbers of these appear to be a strong predictor of a severe prognosis (1, 2). It is known that invading and disseminating carcinoma cells often show great plasticity and undergo epithelial– mesenchymal transition (EMT) (3–5). It is thus recognized that oral carcinoma cells in vitro may migrate as single cells in conjunction with an EMT. It is also well established that activated tumour stromal fibroblasts of the myofibroblast phenotype establish a permissive environment supporting tumour growth associated with secreted cytokines and matrix modifications (6–10). In vivo budding of oral carcinoma cells may show a mesenchymal-like phenotype, as demonstrated by decreased membranous staining of E-cadherin (11, 12), but the relationship at the invasive front between activated fibroblasts and the tumour buds is less clear. In this work, we show that tumour buds are often present without a direct histological relationship to myofibroblasts (13, 14). We also observe that invading tumour cells in vivo may demonstrate reduced E-cadherin expression without a concomitant change to the elongated spindle-shaped morphology as seen in in vitro studies. We demonstrate that these invading cells retain keratin expression and show little or no vimentin expression. Finally, we observe that although single invading cells do occur, the vast majority of tumour buds are connected to larger tumour strands.
Materials and methods Specimens from 28 patients with oral squamous cell carcinomas were chosen at random from the archive at the
Single cell migration in oral cancer Jensen et al.
Department of Odontology, University of Copenhagen. The specimens were archival biopsies routinely fixed in neutral buffered formalin and embedded in paraffin. Spearman’s rank correlation was used to assess the significance of relationships observed. Histology and immunohistochemical procedure Immunohistochemical staining with a pankeratin antibody was used to identify epithelial cells. We demonstrated myofibroblasts by staining with antibodies directed towards a-SMA. Double staining with antibodies against vimentin and keratin was used to identify cells with both epithelial and mesenchymal filaments indicating EMT. Triple staining with keratin, vimentin and E-cadherin antibodies was made to further designate an EMT. Staining was performed on 4-lm sections of paraffin-embedded tissue using a microwave antigen retrieval technique as previously described (15) and working dilutions of the antibodies were established by titration. The antibodies and the retrieval technique used are described in Table 1. When rabbit antibodies were used as the primary antibody, Alexa Fluor-594 goat anti-rabbit IgG (H+L)conjugated antibodies (Life Technologies, Naerum, Denmark) were used as the secondary antibody, whereas staining with mouse monoclonal antibodies used Alexa Fluor 488 conjugated goat anti-mouse antibody (H+L) (Life Technologies, Naerum, Denmark). Triple staining for keratin, vimentin and E-cadherin was performed by first staining for E-cadherin using an immunoperoxidase system on the Ventana Benchmark Ultra autostainer (Roche A/S, Hvidovre, Denmark), with the OptiView detection system (Roche A/S, Hvidovre, Denmark). The plastic coverslip was then dissolved in methanol and the residual adhesive dissolved by placing the slide in xylene. A second round of antigen retrieval was performed in a Dako Link pH 9.0 retrieval solution (DAKO, Glostrup, Denmark), and the tissue section subsequently stained for keratin and vimentin using fluorescence markers as previously described (16). Incubation with irrelevant primary antibodies of the same isotype as the specific antibodies was used as control for the monoclonal antibodies and deletion of the primary antibody served as control for the conjugate. At the invasive front, tumour buds were defined as single cells, or groups of up to four cells, and were selected on the basis of positive cell staining for keratin. The invasive front
was defined as the band of tissue between the tumour front and adjacent normal stromal tissue. The a-SMA staining was semi-quantitatively assessed using a previously described score system (7, 9). In brief, this is described as follows: 0: no staining, 1: moderate staining in a single focus, 2: staining in most parts of the tumour and 3: massive stromal staining. We modified the system and combined 0 and 1, and 2 and 3. The slides were evaluated by two pathologists who were asked to give their a-SMA scores according to these defined criteria, and were not informed about the purpose of the study.
675
3D confocal imaging To investigate to what degree tumour buds represent truly isolated cells or tissue sectioning artefacts, we used confocal microscopy. The tissues used for the 3D reconstruction were cut as thick (25–30 lm) sections. Confocal pictures were acquired using Zeiss LSM 510 (Carl Zeiss, Jena, Germany) or Zeiss LSM 780 (Carl Zeiss) confocal microscopes and Zen software (Carl Zeiss). Images were median-filtered (3 to 3), and the 3D representations were reconstructed from orthogonal projections of 20–65 optical sections in 0.3– 1 lm increments using the Zen software. For each tumour sample, a field containing a high degree of cellular dissociation at the invasive front was chosen for the 3D reconstruction. A total of 10 samples were examined and, to observe the relationship between tumour cells and fibroblasts, the samples were stained for pancytokeratin, a-SMA and DAPI. The analysis used only cells for which the entire cell could be visualized. Cell lines To test whether the tumour buds seen in vivo expressed the same structural proteins as migrating EMT cancer cells observed in vitro, cells of several OSCC cell lines (17) were plated at low density in six well plates. All cell lines were grown in the highly supplemented epithelial growth medium (termed FAD) with 10% FBS (18). Cells with holoclone, meroclone and paraclone morphologies were identified in addition to the subpopulation of migrating individual cancer stem cells. Before staining, cells were fixed for 10 min in ice-cold acetone/methanol, permeabilized in Tween (2% in PBS) and rinsed in PBS. The primary antibody was applied and left overnight at 4°C. They were then stained with the secondary antibodies and mounted with DAPI Prolong Gold Antifade (Life Technologies, Naerum, Denmark).
Table 1 List of antibodies and their dilution used in all experiments Antibody Anti-E-Cadherin Anti-N-Cadherin Anti-Pancytokeratin Anti-a-SMA Anti-b-Catenin Vimentin Twist
Species Mouse Mouse Rabbit Mouse Mouse Mouse Rabbit
Clonality Monoclonal Monoclonal Polyclonal Monoclonal Monoclonal Monoclonal Polyclonal
IgG1 IgG1 IgG2a IgG1 IgG1
Clone NCH-38 6G11 Z0622 1A4 Clone 14 V9 Ab49254
Manufacturer Dako Dako Dako Sigma-Aldrich Ventana Medical Systems Dako Abcam
Retrieval technique HiAR, HiAR, HiAR, HiAR, HiAR, HiAR, HiAR,
pH pH pH pH pH pH pH
a
9.0 9.0a 9.0 9.0 9.0 9.0 9.0
Dilution 1:50 1:200 1:4000 1:400 Pre-diluted 1:800 Titrated until 1:10
HiAR, heat-induced antigen retrieval. a Staining was performed on a Ventana Benchmark Ultra autostainer. J Oral Pathol Med
Single cell migration in oral cancer Jensen et al.
676
A
B
C
Figure 1 Cell line triple stained for (A) keratin (red), (B) vimentin (green) and DAPI (blue). Image (C) is overlay of image (A) and (B). All cells stain for keratin. On the right, a holoclone colony with a single, elongated vimentin-positive, spindle-shaped cell at its periphery (arrow). Vimentin-positive cells with a squamous appearance can be seen in the holoclone colony as well.
Results Staining of cell lines All cells stained positively with antibodies to keratin. Single spindle-shaped cells could be found at the periphery of the holoclones (Fig. 1). Most such cells also stained positively for vimentin, indicating that they had undergone an EMT (Fig. 1B,C). Vimentin expression could also be observed in the cells of holoclone colonies (Fig. 1B). These findings indicate that although oral squamous carcinoma cells may express mesenchymal markers indicative of EMT, they retain keratin microfilaments. We therefore assumed that staining with antibodies to keratin could be used to trace tumour buds in histological sections even if a complete morphological EMT occurs. Histological sections Four cases of the randomly chosen 28 cases were excluded due to lack of tumour buds. The remaining 24 cases showed, based on staining with antibodies to keratin, a great variability in the number of tumour buds at the invasive front. Alpha smooth muscle actin and keratin double staining Eight of 24 biopsies showed tumour buds but no associated myofibroblasts. In five further biopsies, myofibroblasts were present in tumour stroma, mainly between major central tumour islands, but not at the invasive front and without a close relationship of myofibroblasts to the tumour buds (Fig. 2A). The remaining samples frequently showed a histologically close association between myofibroblasts and tumour cells (Fig. 2C). In some tumours, less than half of the tumour buds were associated with myofibroblasts, but in others, all of the tumour buds were closely surrounded by myofibroblasts (Fig. 2B). In these latter samples, myofibroblasts dominated the entire tumour stroma and were thus not specifically localized to areas surrounding the tumour buds (Fig. 2C). Vimentin, keratin and E-cadherin triple staining Vimentin-positive tumour buds were not found in 10 of the 24 specimens, but other specimens showed single cells, or J Oral Pathol Med
small groups of cells that were positive for both vimentin and keratin (Fig. 3D). We found no correlation between the presence of myofibroblasts and number of vimentin-positive tumour buds (P = 0.36). The staining for vimentin was generally weak and located around the nucleus. In only one case was vimentin strongly expressed in keratin-positive tumour cells (Fig. 3D). Reduced expression of E-cadherin was found in cells at the invasive front; however, tumours often showed buds in which some expression of E-cadherin was retained (Fig. 4B,D). The reduced expression of Ecadherin was not seen to be accompanied by an increased expression of vimentin (Fig. 4C,E). 3D confocal imaging Single cells without a clear relationship to other keratinpositive cancer cells were observed in five of 10 samples (Fig. 5). However, although individual optical slices of the invasive tumour front often showed apparently single cells, examination of adjacent slices typically showed them to be connected to other tumour cells.
Discussion Cancer-associated fibroblasts are most often a-smooth muscle actin-positive (19), and other molecules have also been described as markers of activated fibroblasts (20). Staining for such markers is usually only partially overlapping, suggesting the presence of discrete subsets of tumour stromal fibroblasts of which myofibroblasts constitute a major part. We have investigated the distribution of myofibroblasts because clinical and experimental evidence suggests a correlation between stromal myofibroblasts and progression of oral squamous cell carcinomas (21) and because in vitro studies have shown a direct relationship between myofibroblasts, invasion and EMT. We therefore set out to look for in vivo evidence of EMT occurring in tumour buds at the invasive front of OSCC and to examine the relationship between EMT cells and myofibroblasts (13). We assumed that the tumour buds would contain the most aggressively invading cells and that this phenotype would be in part due to stimulation by adjacent myofibroblasts
Single cell migration in oral cancer Jensen et al.
677
A
B
C
Figure 2 Representative images from the invasive front of OSCC specimens that are double and triple stained. The relationship between single cells and myofibroblasts is illustrated. The myofibroblasts are a-SMA-positive (green) with cancer cells keratin positive (red), and nuclei have been stained with DAPI (blue). (A) Invasive single cells (arrow) not surrounded by myofibroblasts. (B) Arrows indicate carcinoma cells that are ‘wrapped’ by a-SMA-positive cells. (C) Carcinoma cells (red) and myofibroblasts (green) are randomly distributed.
A
B
C
D
Figure 3 The invasive front of an OSCC specimen. Sections are stained with (A) vimentin (green), (B) keratin (red). Image (C) is an overlay of A and B, while (D) is a double exposure in which cells that express both proteins appear orange. Cells expressing both vimentin and keratin are marked with arrows.
that are capable of driving a mesenchymal transition of the epithelial tumour cells. Our findings, however, indicate that the presence of invading tumour buds is not dependent on the existence or close proximity of myofibroblasts as has been suggested by in vitro studies (13). Even when myofibroblasts are present, a close relationship between
myofibroblasts and budding tumour cells is not consistently found. We cannot exclude that tumour-activated and growth-stimulating stromal fibroblasts that are not a-SMApositive are present around the tumour buds; leukaemia inhibitory factor has in this way been established as a driver of fibroblast-mediated extracellular matrix remodelling that J Oral Pathol Med
Single cell migration in oral cancer Jensen et al.
678
A
B
C
D
E
F
Figure 4 The invasive front of an OSCC specimen. Representative examples of triple staining for (A) E-cadherin (brown), (B) keratin (red) and (C) vimentin (green). (D) is an overlay of E-cadherin (A) and keratin (B), where keratin-positive cells that are negative for E-cad are marked by arrows. These single cells are also weakly positive for vimentin and negative for E-cadherin, as can be seen from the overlay of E-cadherin and vimentin in (E) and keratin and vimentin in (F).
Figure 5 3D reconstruction of an approximately 30-lm slide of an OSCC specimen. Notice the one isolated cell on the right, which is not related to the tumour island on the right. Also notice that this cell does not have an elongated, mesenchymal-like morphology.
is independent of a-SMA expression (22). A recent study demonstrate that a-SMA-positive fibroblasts have a protective rather than promoting role in tumour progression further illustrates the complexity of the crosstalk between tumour cells and the microenvironment (23). However, it remains obscure how these findings should be interpreted considering the studies showing that tumours with a high proportion J Oral Pathol Med
of myofibroblasts have a poor prognosis (8, 21, 24). Recently, the absence or presence of infiltrating immune cells at the invasive front has been demonstrated to be of prognostic importance and could thus help to explain some of the discordant findings (25). Although absent or reduced E-cadherin staining was frequently found at the invasive front, this was not accompanied by an upregulation of N-cadherin. Also, we did not observe any clear relationship between the loss of E-cadherin and gain of vimentin. This is in agreement with studies that demonstrate that E-cadherin loss is insufficient to induce an epithelial–mesenchymal transition (26, 27). We have also stained both cell cultures and histological sections for Twist, a well-known EMT regulator, but have only found a weak signal in few cultured cells, and none in cells of tumour buds in histological sections (data not shown). We performed 3D confocal microscopy on 30-lm-thick slices to assess whether cells appearing as isolated single cells in thin histological sections were likely to be isolated cells rather than single cells connected to a larger tumour island by finger-like projections not in the plane of section. This technique made it possible to identify single cells that were not continuous with any larger tumour island. Few of the apparently single cells were found actually to be isolated, but most were part of finger-like projections, as previously noted (28). These findings confirm that OSCC in vivo tend to migrate collectively, by so-called chain migration and not predominantly as single cells as seen in vitro (29), and may help to explain the difference between our in vivo findings and previously reported in vitro results of the expression of EMT markers, such as vimentin (17). It cannot, however, be excluded that the invasive cells only
Single cell migration in oral cancer Jensen et al.
variably express vimentin for a short period during active migration, resulting in a sparse expression in histological sections. In support of this, Wong et al. demonstrated that a malignant cell line (MCF-10A) showed a phenotypic plasticity in the expression of biomarkers as cells interconvert between individual and collective migration and vice versa (30). Alternatively, the levels of expression of vimentin in the tumour cells could be so low compared to the vimentin expression in the surrounding stromal cells that it would not be readily apparent with the methods we have used. Indeed, the few carcinoma cells we identified in the histological sections that were positive for vimentin expressed it to a much lesser degree than cells in the surrounding tumour stroma. These cells were also not overtly mesenchymal in morphology, that is elongated, and this could support the idea of a so-called partial EMT, where cells migrate as multicellular groups, but the tip of the migrating cells undergoes a partial EMT (31).
References 1. Bryne M. Is the invasive front of an oral carcinoma the most important area for prognostication? Oral Dis 1998; 4: 70–7. 2. Wang C, Huang H, Huang Z, et al. Tumor budding correlates with poor prognosis and epithelial-mesenchymal transition in tongue squamous cell carcinoma. J Oral Pathol Med 2011; 40: 545–51. 3. Brabletz T. To differentiate or not — routes towards metastasis. Nat Rev Cancer 2012; 12: 425–36. 4. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin Investig 2009; 119: 1420–8. 5. Zlobec I, Lugli A. Epithelial mesenchymal transition and tumor budding in aggressive colorectal cancer: tumor budding as oncotarget. Oncotarget 2010; 1: 651–61. 6. Bello IO, Vered M, Dayan D, et al. Cancer-associated fibroblasts, a parameter of the tumor microenvironment, overcomes carcinoma-associated parameters in the prognosis of patients with mobile tongue cancer. Oral Oncol 2011; 47: 33–8. 7. Leemans CR, Braakhuis BJM, Brakenhoff RH. The molecular biology of head and neck cancer. Nat Rev Cancer 2011; 11: 9– 22. 8. Marsh D, Suchak K, Moutasim KA, et al. Stromal features are predictive of disease mortality in oral cancer patients. J Pathol 2011; 223: 470–81. 9. Sobral LM, Bufalino A, Lopes MA, et al. Myofibroblasts in the stroma of oral cancer promote tumorigenesis via secretion of activin A. Oral Oncol 2011; 47: 840–6. 10. Vered M, Dobriyan A, Dayan D, et al. Tumor-host histopathologic variables, stromal myofibroblasts and risk score, are significantly associated with recurrent disease in tongue cancer. Cancer Sci 2010; 101: 274–80. 11. Wang X, Zhang J, Fan M, et al. The expression of E-cadherin at the invasive tumor front of oral squamous cell carcinoma: immunohistochemical and RT-PCR analysis with clinicopathological correlation. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009; 107: 547–54. 12. Bagutti C, Speight PM, Watt FM. Comparison of integrin, cadherin, and catenin expression in squamous cell carcinomas of the oral cavity. J Pathol 1998; 186: 8–16.
13. Brentnall TA. Arousal of cancer-associated stromal fibroblasts: palladin-activated fibroblasts promote tumor invasion. Cell Adh Migr 2012; 6: 488–94. 14. Gaggioli C, Hooper S, Hidalgo-Carcedo C, et al. Fibroblastled collective invasion of carcinoma cells with differing roles for RhoGTPases in leading and following cells. Nat Cell Biol 2007; 9: 1392–400. 15. Bilde A, von Buchwald C, Dabelsteen E, et al. Molecular markers in the surgical margin of oral carcinomas. J Oral Pathol Med 2009; 38: 72–8. 16. Osman TA, Øijordsbakken G, Costea DE, et al. Successful triple immuno - enzymatic method employing primary antibodies from same species and same immunoglobulin subclass. Eur J Histochem 2013; 57: e22. 17. Biddle A, Liang X, Gammon L, et al. Cancer stem cells in squamous cell carcinoma switch between two distinct phenotypes that are preferentially migratory or proliferative. Cancer Res 2011; 71: 5317–26. 18. Locke M, Heywood M, Fawell S, et al. Retention of intrinsic stem cell hierarchies in carcinoma-derived cell lines. Cancer Res 2005; 65: 8944–50. 19. Thode C, Jørgensen TG, Dabelsteen E, et al. Significance of myofibroblasts in oral squamous cell carcinoma. J Oral Pathol Med 2011; 40: 201–7. 20. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer 2006; 6: 392–401. 21. Kellermann MG, Sobral LM, Silva SDD, et al. Myofibroblasts in the stroma of oral squamous cell carcinoma are associated with poor prognosis. Histopathology 2007; 51: 849–53. 22. Albrengues J, Bourget I, Pons C, et al. LIF mediates proinvasive activation of stromal fibroblasts in cancer. Cell Rep 2014; 7: 1664–78. € 23. Ozdemir Berna C, Pentcheva-Hoang T, Carstens Julienne L, et al. Depletion of carcinoma-associated fibroblasts and fibrosis induces immunosuppression and accelerates pancreas cancer with reduced survival. Cancer Cell 2014; 25: 719–34. 24. Yamashita M, Ogawa T, Zhang X, et al. Role of stromal myofibroblasts in invasive breast cancer: stromal expression of alpha-smooth muscle actin correlates with worse clinical outcome. Breast Cancer 2012; 19: 170–6. 25. Galon J, Mlecnik B, Bindea G, et al. Towards the introduction of the ‘Immunoscore’ in the classification of malignant tumours. J Pathol 2014; 232: 199–209. 26. Chen A, Beetham H, Black MA, et al. E-cadherin loss alters cytoskeletal organization and adhesion in non-malignant breast cells but is insufficient to induce an epithelial-mesenchymal transition. BMC Cancer 2014; 14: 552. 27. Nilsson GM, Akhtar N, Kannius-Janson M, et al. Loss of Ecadherin expression is not a prerequisite for c-erbB2-induced epithelial-mesenchymal transition. Int J Oncol 2014; 45: 82– 94. 28. Bronsert P, Enderle-Ammour K, Bader M, et al. Cancer cell invasion and EMT marker expression: a three-dimensional study of the human cancer–host interface. J Pathol 2014; 234: 410–22. 29. Wolf K, Friedl P. Mapping proteolytic cancer cell-extracellular matrix interfaces. Clin Exp Metastasis 2009; 26: 289–98. 30. Wong IY, Javaid S, Wong EA, et al. Collective and individual migration following the epithelial–mesenchymal transition. Nat Mater 2014; 13: 1063–71. 31. Revenu C, Gilmour D. EMT 2.0: shaping epithelia through collective migration. Curr Opin Genet Dev 2009; 19: 338–42.
679
J Oral Pathol Med
