Cytokeratin 8 (K8) is a structural protein, which forms intermediate filaments within the cytoplasm of simple epithelial cells 1 as a dimer with CK18 2. Along with other keratins, K8/CK18 generate a stabilizing framework, which is cell shape determining and allows cells to cope with mechanical stress. Cytokeratin filaments further on represent a mesh of "paths" on which signalling molecules, metabolites, and pathogens can travel the cell in an orientated fashion. The regulation of the localization of K8 within cells and polymerization into intermediate filaments is dependent upon its phosphorylation. Two main kinase families are instrumental in this context: the MAP kinase family member p38 3 and PKC-ε related kinase 4. Phosphorylation of K8 at serine in position 73 (Ser⁷³) is mediated by p38 under stress such as orthovanadate treatment, and regulates keratin organization 5. High p38 kinase activity correlated with the formation of keratin granules, while low p38 activity, ergo low K8 Ser⁷³ phosphorylation, was associated with a prevented disassembly of the filament network 5. As a potential counter-regulator and eventually in order to balance the phosphorylation status of K8, the catalytic subunit of protein phosphatase 2A (PP2A) associates with and dephosphorylates K8 after hyposmotic stress 6. However, dephosphorylation was site-specific and concerned Ser⁴³¹, not Ser⁷³. Additionally, K8 and CK18 hyperphosphorylation is a valuable marker for the progression of liver diseases such as non-cirrhotic hepatitis C infection or cirrhosis 7. Disease associated mutations of K8 were reported for the case of cryptogenic liver diseases with single point mutations leading to the exchange of glycine at position 61 to a cysteine residue and of tyrosine⁵³ to a histidine 89. Gly⁶¹ → Cys mutation was of major importance as it diminished the capacity of cells to reorganize keratin filament. Recently, Ku and colleagues reported on an animal model for the Gly⁶¹ → Cys mutation. In transgenic mice, this point mutant of K8 predisposed animals to liver injury along with a decreased Ser73 phosphorylation 10. When ectopically expressed at the plasma membrane of carcinoma cells 11, K8 serves as a tissue-type plasminogen activator (tPA) 12131415 and might help tumour cells to remodel or invade surrounding tissue 16.

Generally speaking, K8 is believed to be involved in the process of carcinogenesis 1718192021 and silencing of it resulted in sensitization for cisplatin 22. We have isolated K8 as a tumour-associated antigen, which elicits a humoral response in vivo in patients suffering from carcinomas of the head and neck area 23. A continuative study on the profile of K8-specific autoantibodies in healthy donors and patients revealed that autoantibody titers allowed to differentiate normal and diseased persons, but not to discriminate between cases of benign and malignant disease 24. Normal squamous epithelium, which represents the great majority of epithelia of the head and neck and of malignancies thereof, is devoid of K8. De novo expression of K8 was observed for head and neck carcinomas, however in a small patients cohort 25. Studies including larger numbers of patients with head and neck malignancies are to the best of our knowledge missing so far and therefore the topic of the present investigation.

Here, we present a comprehensive survey of K8 expression in normal mucosa, leukoplakia, head and neck squamous cell carcinomas (HNSCC), and lymph node metastases of the head and neck area. We have used immunohistochemistry on cryosections for this purpose as it allows thorough detection of K8 and, importantly, the assignment of staining to particular cell types within samples as opposed to RT-PCR or immunoblotting. K8 positivity was a hallmark of transformed epithelia where it correlated with early stages of carcinogenesis in dysplastic leukoplakia. K8 was also strongly over-expressed in the majority of HNSCC tested. Importantly, K8^high was a characteristic of disseminated tumour cells and in overt lymph node metastases. For this reason, K8 represents a valuable marker for head and neck malignancies in early stages of the disease and of invasive growth of tumour cells.

Reliable markers for pre-malignant lesions retained their paramount importance, as they are believed to bring about significant improvements of patients' care and overall survival 27. This notion was best exemplified by the use of prostate-specific antigen PSA for the early diagnosis of prostate carcinoma 28. Clearly, the earlier a malignancy of the head and neck area is diagnosed, the better the prognosis for the patient 29. Hence, detection of pre-malignant lesions is highly desirable as is the visualization of disseminated tumour cells, which are to be seen as founders of metastases 303132. With these prerequisites in mind it is interesting to retrieve from the present stuy that K8 (i) is absent in normal mucosa composed of squamous epithelium and in contrast to adenomatous epithelium, that (ii) K8 expression differentiates dysplastic lesions, carcinomas in situ, and small established carcinomas (i.e. T1-2) from normal tissue and hyperplastic lesions within oral leukoplakia, and finally that (iii) K8 thoroughly marks single detached tumour cells. These features and the de novo expression of qualify K8 as a worth candidate for the early detection of pre-malignant lesions, which might progress to overt malignancies with significantly enhanced probability 33, and for disseminated tumour cells in head and neck carcinomas. Additionally, K8 was released as a circulating marker from apoptotic non-small cell lung carcinoma cells as a full-length and proteolytic cleavage-resistant protein and might hence represent a valuable biomarker for head and neck malignancies 34.

From a mechanistic and molecular point of view, and with respect to the over-expression of K8 in cancer cells, two eventualities are to be envisaged. Firstly, disseminating and invading cells might require a reorganization of their cytoskeleton to improve motility and epithelial-to-mesenchymal transition (EMT), which could be warranted by keratins of simple epithelia as is K8. Data from Chu and colleagues disclosed an ability of K8 and K18 to foster the invasive potential when ectopically expressed in murine L fibroblasts 16. Along this line, assessment of p38 kinase activity in disseminated cells appears expedient. K8 Ser⁷³ serves as a substrate for p38 kinase and phosphorylation at this position is crucial for the destabilization of intermediate filaments 3535. Reorganization, especially destabilization of intermediate filaments occurs under various physiological conditions such as mechanic stress, during epithelial cell homeostasis, exposure to chemicals (vanadate, ocadaic acid), and during mitosis. Common to all these processes is the recruitment of active p38 kinase to depolimerized keratin granules and phosphorylation of K8 at Ser⁷³ 511. Hence, one may also envisage the simultaneous assessement of K8 expression and of Ser⁷³ phosphorylation as an additional surrogate marker for the mitotic index. A second scenario must be considered, in which K8 serves as a receptor for plasminogen and tPA at the plasma membrane of tumour cells 121314. By doing so, K8 will enhance the proteolytic activity at the plasma membrane and facilitate tissue remodelling and invasion. Such an eventuality is supported by data presented herein. Single tumour cells that were already detached from the main tumour at the time of sample asservation displayed highest K8 expression, which may foster proteolytic activity and invasion. Regulatory processes involved in the de novo expression of K8 in these tumour cells are poorly understood. Recent work by Lacina et al. disclosed a potential for non-malignant stroma cells, i.e. tumour-associated fibroblasts, to induce K8 expression on normal keratinocytes in vitro 3637. Thus, the tumour microenvironment needs to be seen as a strong modulator of protein cell surface expression of carcinoma cells. Clearly, both eventualities, namely remodelling of cell structures and improvement of the proteolytic appartus of carcinoma cells, are not mutually exclusive but may even be both instrumental in parallel.

The findings presented in this study of K8 expression are in accordance with and complementing previous data that demonstrated de novo synthesis of K8 in dysplastic lesions as well as in head neck carcinomas. However, these former studies were conducted in substantially smaller cohorts and without differentiation of tumour sub-localizations 2538. Subdividing head and neck tissues according to their precise localization revealed minor differences in the K8 expression profile. Hypopharynx, oropharynx, larynx, valecula, and vocal cords were characterized by > 85% of samples expressing intermediate to strong levels of K8. In contrast, carcinomas of the tongue were different since they displayed a more heterogeneous repartition with 37.5% of samples expressing no or weak levels of K8. Also, normal mucosa from the larynx was always positive for K8, pinpointing differences amongst healthy epithelia, too. Therefore, tongue and laryngeal carcinomas appear less amenable to a K8-based diagnosis.

Taken together our data qualify K8 as an excellent marker for head and neck carcinomas, lymph node metastases, and for tumour cells that have already detached from the primary tumour.

The intermediate filament protein K8 is known as a tumour-associated antigen. We present a survey of K8 expression in head and neck epithelia that demonstrates the specific staining of K8 in pre-malignant and malignant tissue versus normal cells. Dysplastic and tumour cells expressed K8 to strong levels, as did disseminated tumour cells. Hence, K8 is an excellent marker for the visualisation and diagnosis of pre-malignancies and DTCs in the head and neck area.

K8: cytokeratin 8; K18: cytokeratin 18; DTC: disseminated tumour cell; HNSCC: head and neck squamous cell carcinoma; MAP kinase: mitogen-activated protein kinase; PBS: phosphate balance salt solution; PKC: protein kinase C; RT: room temperature.

The authors declare that they have no competing interests.

CM and AB were in charge of patients' recruitment, sample assessments, and study design. BM performed all stainings shown in the manuscript. OG wrote the manuscript, analysed the acquired data, and was responsible for the study design together with CM. All authors read and approved the final manuscript.