Expression of the Epidermal Growth Factor Seven-Transmembrane Member CD97 Correlates with Grading and Staging in Human Oral Squamous Cell Carcinomas Free

The oral squamous cell carcinoma (OSCC) is the sixth most common malignant tumor worldwide and increases in incidence (1). According to the histopathologic grading, OSCCs display different metastatic potentials. Highly differentiated OSCCs (G1) display slow growth and destruction to adjacent muscle, bone, and cartilage. By contrast, high invasiveness, early spread to cervical lymph nodes, and bad prognosis are hallmarks of poorly differentiated OSCCs. OSCCs frequently show local recurrence after the initial surgical or radiological treatment at the primary site. Despite modern research and therapy methods, no significant better progress has been achieved by combining surgery and radiotherapy as “traditional” therapy forms with chemotherapy. Recent studies showed that the overall survival rate in OSCC is still <50% (2). Being aware of this dilemma, clinicians and pathologists are showing a steadily growing interest in the identification of relevant markers involved in oral carcinogenesis. Several models have been proposed to elucidate risk factors, such as smoking and alcoholic beverages as well as markers that are involved in the generation and progress of oral cancer (3, 4).

CD97 is a 75- to 90-kDa protein member of the epidermal growth factor seven-transmembrane (EGF-TM7) family of class II TM7 receptors and structurally related to the secretin receptor family. Within the secretin receptor family, the EGF-TM7 molecules classify as members of the B2 subgroup of class B G-protein-coupled receptors (5). The larger of the two noncovalently associated subunits, the extracellular CD97α, contains up to five EGF-like repeats that are spliced in three alternative isoforms (6). The subunit CD97β provides the TM7 protein part.

The ubiquitous expressed CD55, also known as the decay accelerating factor, is a ligand for CD97 and protects host cells from complement attack by binding to C3b and C4b. Binding to C3b and C4b prevents the formation of the membrane attack complex. CD55 functions as a member of the regulators of the complement activation family and consists of an extracellular portion (NH₂ terminus) with four short consensus repeat domains linked via a heavily O-glycolysated spacer to a COOH-terminal glycosyl-phosphatidylinositol membrane anchor (7). The interaction between the EGF-like domains of CD97 and the NH₂-terminal domains of CD55 is Ca²⁺ dependent, and the smallest isoform of CD97 binds CD55 with the highest affinity (8-10). The interaction with CD55 may serve cellular adhesion potentially resulting in the activation of a signal cascade mediated by the TM7 spanning CD97β subunit (11). To date, we know that CD55 plays a role as a complement deactivator in several human malignancies (12). The physiologic function of CD97 is not yet known.

Previous immunohistochemical studies showed a tissue distribution of CD97 in cells of hematopoietic origin, resting T and B cells, fetal liver, spleen, smooth muscle, dentritic cells, and peripheral macrophages (13). CD97 may be regarded as a dedifferentiation marker in nonmedullary and medullary thyroid carcinomas where we correlated aggressiveness and bad prognosis with high CD97 protein expression levels (14-16). CD97 expression was also studied in carcinoma cells of pancreatic, gastric, esophageal, and bowel cancer (17-19).

The aim of our study was to determine the expression of CD97 and CD55 in normal oral tissues and laser microdissected OSCC clusters. In a retrospective approach, we examined the CD97/CD55 expression in OSCCs of defined staging (pTNM) and grading (G1-G4). In addition, we studied the regulation of CD97 and CD55 expression of three human OSCC cell lines on incubation with the redifferentiation agents retinoic acid (RA) and sodium butyrate to address whether CD97 and CD55 may serve as dedifferentiation markers in OSCC.

To specify our analyses, we employed computer-aided UV-laser microdissection and compared the mRNA expression of microdissected carcinoma cell clusters with nonmalignant cells from the same tissue section (20). UV-laser assisted microdissection allows the dissection and catapultation of cells under direct microscopic/computerized visualization without heat transformation into the investigated tissue (21). The sterile transfer of cells into a RNA isolation reagent makes this technique comfortable for performing expression analyses of different cells from one specimen (22, 23).

This study was approved by the local university ethical committee, and all patients had given their consent. A total amount of 78 patients with primary OSCC were surgically treated between November 2000 and May 2003 at the University Department of Oral, Maxillo, and Plastic Facial Surgery, Martin-Luther-University Halle-Wittenberg. Forty-seven patients displayed lymph node metastases and distant metastases to the lung were diagnosed in four patients. Nonmalignant oral mucosa (n = 10) were obtained from patients undergoing wisdom teeth operations and served as control specimen for the normal CD97/CD55 protein distribution. All tissues had been snap frozen in liquid nitrogen immediately after removal and stored at −80°C until used. The histopathologic diagnosis was routinely determined by the Institute of Pathology and included the pTNM and the tumor grading (Table 1).

We employed the Palm Laser Pressure Catapult System and the PalmRobo 1.0 software (Palm Microlaser, Bernried, Germany) for microdissection of carcinoma cell clusters and normal cells. Frozen sections were prepared at 6 μm under RNase-free conditions using a Microm cryostat (Microm International GmbH, Walldorf, Germany). The sections were positioned on RNase-free membrane slides, which are covered with an all-side sticked polyethylene-naphtalate membrane (Palm). After fixation with 97% ethanol, all sections were stained with H&E (Sigma, Seelze, Germany). For total RNA extraction, carcinoma tissue flakes and normal cell clusters from at least 15 consecutive tissue sections were dissected and catapulted into common microfuge caps. Tissue flakes of each specimen were transferred into tubes containing 10 μL diethylpyrocarbonate-treated water and were spun down for 10 minutes at 12,000 rpm.

Total RNA was extracted from tissue flakes and cell lines using the Trizol reagent (WKS, Frankfurt, Germany). The cDNA synthesis was carried out with the SuperScript II kit according to the manufacturer's instructions (Invitrogen, Groningen, the Netherlands) using 1 μg total RNA. For multiplex reverse transcription-PCR (RT-PCR) analysis, specific oligonucleotide primers for CD55 (CD55 amplicon of 550 bp; sense 5′-TTCAGGCAGCTCTGTCCAGTG-3′; antisense 5′-GAGGCTGAAGTGGAAGGATCG-3′) and CD97 (CD97 amplicon of 442 bp; sense 5′-AGCTATCAGTGTCGCTGCCG-3′; antisense 5′-CTATGAGGTGCCGGACAGGT-3′) were employed. Multiplex RT-PCR was done with the housekeeping gene β-actin and specific oligonucleotide primers were designed. A specific β-actin amplicon served as an internal control for CD97 (β-actin amplicon of 608 bp; sense 5′-GCTGGAAGTGGACAGCGA-3′, antisense 5′-GGCATCGTGATGGACTCCG-3′) and for CD55 (β-actin amplicon of 810 bp; sense 5′-ATCTGGCACCACACCTTCTACAATGAGCTGCG-3′; 5′-CGTCATACTCCTGCTTGCTGATCCACATCTGC-3′).

All PCR products were separated on a 1% low melting point agarose gel, purified by magic column extraction, and cloned into the pGEM-T vector (Promega, Heidelberg, Germany). Sequence analysis of the PCR-amplified cDNA clones was done employing the Thermo Sequenase dye terminator cycle sequencing kit (Amersham Biosystem, Freiburg, Germany) employing T7 and SP6 sequencing primers for bidirectional sequencing of the cloned amplicons.

For multiplex RT-PCR, 2 μL cDNA and 10 pmol/L specific sense and antisense primers, respectively, were incubated with TaqGold (Amersham Biosystem). After an initial denaturation for 10 minutes at 95°C, 35 cycles followed with 1-minute denaturation at 95°C, 1-minute annealing at 67°C (CD97) or 57°C (CD55), and 2-minute elongation at 72°C. After a final elongation step for 10 minutes at 72°C, 20 μL of the PCR products were loaded on a 1% agarose gel containing ethidium bromide for visualization of amplicons. All gels containing PCR amplicons were scanned with the Kodak Image System 440cf and electronically evaluated with the Kodak Digital Science 1D software (Eastman Kodak, New Heaven, CT). The analysis of CD55 and CD97 gene expression was done using the comparative and semiquantitative options of the Kodak Digital Science 1D software, which allows the user to compare the gray scales of the scanned amplicons in relation to the positive control (100% reference point). Human pharyngeal tonsil tissue was used as positive control for CD55, whereas the colon carcinoma cell line CaCo was employed as a positive control for CD97. The expression levels of all investigated specimens were determined in comparison with the band intensity of the positive control (0-25%, no expression; 25-50%, low expression; 50-75%, moderate expression; ≥75%, high expression).

For quantification,1.5 μL of the reverse transcriptase reaction mixture was added to 25 μL reaction mixture consisting of 1× AmpliTaq Gold reaction buffer, 1.5 units AmpliTaq Gold, 0.2× SYBR Green (Biozym, Hess. Oldendorf, Germany), 200 μmol/L of each deoxynucleotide triphosphate, and 0.5 μmol/L each of specific primers. A negative control without template was included. Target cDNAs were run in triplicates in a RG2000 cycler (LTF, Wasserburg, Germany). Initial denaturation at 95°C for 10 minutes was followed by 40 cycles with denaturation at 95°C for 15 seconds, annealing at 67°C for 30 seconds, and elongation at 72°C for 20 seconds. The fluorescence intensity of the double-strand–specific SYBR Green, reflecting the amount of formed amplicon, was read after each elongation step at 82°C using the Rotor-Gene 4.6 software. For verification of the PCR products, melt curves were generated. Relative quantification of CD97 and CD55 gene expression was done using the comparative quantification technique of the Rotor-Gene 4.6 software. This technique allows the user to compare differently treated cells with an unstimulated control. The second derivative of the raw data is taken to calculate the takeoff point. Based on the takeoff point and the reaction efficiency, it calculates the relative concentration of each sample compared with the control sample.

The three OSCC cell lines Cal27, Cal33, and BHY were purchased from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany; refs. (24, 25). The BHY cells were derived from a G1 OSCC. Cal33 cells were obtained from a G2 OSCC, and the Cal27 cell line was established from a G3 carcinoma. All three cell lines were cultured in Ham's F12 medium (Invitrogen) containing 10% FCS.

After the cells reached subconfluent mononlayers, stimulation experiments were carried out with or without 10μmol/L RA or 3 mmol/L sodium butyrate for 72 hours (Fa. Roth, Germany). Untreated cells served as controls (K0). We determined the effects of RA and sodium butyrate on CD97/CD55 transcription by real-time RT-PCR. For immunocytochemistry, 1 × 10⁵ cells were directly cultured on UV-sterilized Superfrost slides (Menzel-Gläser, Braunschweig, Germany) and stimulated with RA or sodium butyrate.

Before this study, several experiments with different sodium butyrate and RA doses were done. Sodium butyrate was used in concentration ranging from 1 to 10 mmol/L. A concentration <2 mmol/L was too weak to evoke any significant effect on the transcription and translation level. Concentrations >5 mmol/L were toxic and strongly affected the viability of the cells. This effect resulted in a rapid cell death and a loss of >80% of the population after 72 hours. Doses <1 μmol/L RA were proven to be an ineffective redifferentiation substance for the oral malignant cells used in this study, and we started investigating the effect of higher doses: 10 μmol/L was found to be the most effective RA concentration. Low-dose RA, such as 0.1 μmol/L, enhanced the cell growth.

Cryostat sections were mounted on Superfrost slides and endogenous peroxidase was blocked in a mixture of 97% ice-cold methanol and 3%H₂O₂ for 20 minutes. All samples were incubated for 15 minutes with normal swine serum diluted 1:4 in PBS containing 1% bovine serum albumin to suppress nonspecific bindings (DAKO, Hamburg, Germany).

Tissue sections were separately incubated for 1 hour at 37°C with monoclonal antibodies against CD55 (clone 143-30, BD Biosciences, Heidelberg, Germany) and CD97 (MEM180, Biermann, Germany) at dilutions of 1:100, respectively. After washing in PBS, the samples were incubated for 30 minutes with a 1:1,000 dilution of biotinylated goat anti-mouse secondary antibody (DAKO). Detection of immunoreactive CD55 and CD97 was accomplished by using the avidin-biotin complex method. A 15% solution of 3,3′-diaminobenzidine (DAKO) was used as chromogen. For a negative control and to determine nonspecific staining, the primary antibody was replaced with a nonimmune. Tissue sections from an anaplastic thyroid carcinoma, known for strong CD97/CD55 expression, were used as positive controls. All sections were examined by three independent investigators who were blinded to the histologic diagnosis (A.E., M.K., and J.S.).

We employed the planimetric measurement features of the PalmRobo software to determine the immunostaining intensity. This software allows the user to encircle cell areas for calculation (μm²). The sum of all immunopositive cell squares (μm²) was calculated and compared with the total section area (100%). The level of staining intensity was classified into four groups (0-25%, no expression; 25-50%, low expression; 50-75%, moderate expression; 75-100%, high expression).

Stimulated and unstimulated cells were detached from culture flasks using Accutase (PAA,Linz, Austria). After two cold washes, standard surface membrane immunofluorescence techniques were used. Cells were stained with monoclonal antibodies against CD97 (MEM180) and CD55 (clone 143-30) or an isotype control at 4°C for 40 minutes. After two cold washes containing sodium azide, cells were labeled with FITC (CD97)–conjugated and phycoerythrin (CD55)–conjugated secondary antibody (goat anti-mouse IgG, Dianova, Hamburg, Germany) at 4°C for 30 minutes and washed. Paraformaldehyde (1%) was used for fixation. The fluorescence was analyzed in a FACScan flow cytometer (Becton Dickinson, Heidelberg, Germany). Ten thousand cells were counted and the results were analyzed with the Lysis II software (Becton Dickinson). Unstimulated cells served as control.

All PCR reactions were carried out thrice for calculation of SD. The t test features of the SPSS 10.0 software were employed to calculate for statistical significance. Ps < 0.05 were considered to be statistically significant.

OSCC carcinoma cell clusters were successfully dissected and catapulted into diethylpyrocarbonate-treated common microfuge caps (Fig. 1A-C).

Nonmalignant oral epithelial cells revealed specific amplicons for CD97 at 442 bp and expression levels were <50% of the positive control as determined by multiplex RT-PCR (Fig. 2A). By contrast, significantly stronger CD97 transcripts were detected in all 78 OSCC tissue specimens (Figs. 2B and 3A). CD97 gene activity correlated inversely with the histopathologic stage of the tumors. CD97 transcriptional activity was increased in advanced stages of OSCC with pathologic evidence of lymph node metastasis but weak in pT₁/T₂ tumors with exclusive intraoral growth (Fig. 3B and C). Highest CD97 transcription levels were detected in tumor specimens obtained from patients with N2 lymph node status. G3/G4 carcinomas revealed significantly stronger CD97 transcripts than G1/G2 tumors (Fig. 3D).

Specific CD55 amplicons at 562 bp were detected in normal microdissected oral mucosa and in all OSCCs (Fig. 2C and D). CD55 mRNA expression was significantly lower (P < 0.05) in nonmalignant cells (<75% of the positive control) than in OSCC (Fig. 3E). As shown in Fig. 3F, for example, we could not correlate any of the pTNM variables with the CD55 mRNA expression in OSCC. All investigated carcinoma specimens revealed strong CD55 transcripts irrespective of the stage and grade.

The nonmalignant oral mucosal specimens showed a CD97 immunostaining, which was clearly restricted to the basal epithelial layer (Fig. 4). All OSCC specimens produced CD97 but in varying amounts. CD97 staining was weak in 19, moderate in 27, and strong in 32 OSCCs. CD97 immunostaining was in the cytoplasm and on the cell membrane of OSCC cells. As shown in Fig. 5A, carcinoma cell areas were significantly stronger stained than the surrounding tissue. A strong marginal brown 3,3′-diaminobenzidine staining was found around carcinoma cell clusters (Fig. 5B). We observed significant differences (P < 0.05) in the CD97 immunostaining between pT₁/T₂ and pT₃/T₄ carcinomas (Fig. 6A). OSCC classified as pT₁/T₂ lesions and highly differentiated tumors (G1) were weakly stained (Fig. 5C). Specimens obtained from G2 and G3/G4 OSCC or from tumors with lymph node infiltration or distant metastases displayed significantly stronger CD97 immunostaining (Figs. 5D and 6B and C). We could not find any significant difference in the CD97 staining intensity between G3 and G4 lesions (Fig. 6C).

OSCC were strongly stained for CD55 protein (Fig.7A). OSCC cell clusters were intensively stained. Exclusive strong brown CD55 immunostaining was detected on malignant cells (Fig. 7B) and adjacent cells were CD55 weakly positive or negative. We did not find any significant differences between the expression of CD55 mRNA/CD55 protein and the staging/grading classification in OSCC.

The three OSCC cell lines Cal27, Cal33, and BHY expressed CD97 mRNA under basal cell culture conditions. As determined by real-time RT-PCR, RA inhibited the CD97 mRNA expression in Cal27 (−37%) and Cal33 (−31%) significantly (Fig. 8A). By contrast, the CD97 transcription rate increased in the highly differentiated cell line BHY (+28%) under the influence of RA (Fig. 8B).

Sodium butyrate inhibited CD97 mRNA in all three cell lines significantly (Fig. 7C). The strongest inhibition of CD97 mRNA occurred in Cal27 (−38%).

Under basal cell culture conditions, all three cell lines revealed strong CD97 immunostaining (Fig. 9A). Significant differences were found in the CD97 immunostaining and CD97 flow cytometry analysis after RA and sodium butyrate treatment. Morphologic changes were not observed neither <10 μmol/L RA nor <3 mmol/L sodium butyrate treatment.

Interestingly, under RA treatment, a weaker CD97 immunostaining was observed in the cytoplasm of Cal27 and Cal33 cells and no 3,3′-diaminobenzidine staining was detectable on the plasma membrane (Fig. 9B). By contrast, BHY cells revealed stronger CD97 immunostaining. Flow cytometry revealed significant CD97 down-regulation in Cal27 and Cal33 after 72 hours (Fig. 10A). We determined a down-regulation of >50%for both cell lines. By contrast, BHY cells showed an immense increase of >100% on stimulation with RA (Fig. 10B).

Sodium butyrate decreased the CD97 protein expression in all three cell lines, and changes occurred in the cellular distribution of CD97 as determined by immunocytochemistry. We did only observe a thin brown 3,3′-diaminobenzidine staining in the cytoplasmic area around the nucleus of sodium butyrate–treated cells, but no membrane bound CD97 immunostaining was detectable (Fig. 9C). Flow cytometry confirmed the immunocytochemistry, and all cell lines revealed significant CD97 down-regulation. Flow cytometry showed that sodium butyrate had the strongest inhibitory effect and the fluorescence intensity was significantly weaker than in the unstimulated control (Fig. 10C).

CD55 mRNA was expressed in all three unstimulated cell lines. RA and sodium butyrate evoked an up-regulation of CD55 mRNA in all cell lines and a strong brown 3,3′-diaminobenzidine staining was detected in the cytoplasm and on the plasma membrane of the carcinoma cells (Fig. 9D). As determined by flow cytometry, the fluorescence intensity increased on RA and sodium butyrate treatment, but we could not find any significant differences among 24, 48, and 72 hours.

We identified CD97 as a specific product in all OSCC. By contrast, CD97 was restricted to the basal cell layer of the oral mucosa, implicating this TM7 protein to be a novel specific marker in dedifferentiated human OSCC. CD97 belongs to an expression profile of several human malignancies. We have shown previously that CD97 serves as a dedifferentiation marker in thyroid carcinomas (14-16). Anaplastic thyroid carcinomas and dedifferentiated C-cell carcinomas displayed significantly stronger CD97 immunostaining compared with the more differentiated tumors of the thyroid. Redifferentiation therapy of the human thyroid carcinoma cell line (FTC-133) with RA caused an inhibition of CD97 (15, 26). Furthermore, a close correlation between the stage of differentiation and the CD97 expression was also shown in several colon carcinoma cell lines (19).

In this present study, we observed a strong expression of CD97 in OSCC obtained from patients with lymph node and/or distant metastases. Scattered cell clusters are features of unsolid tumor growth and dedifferentiation. We could show that a significant higher expression of CD97 is detectable in those OSCC with a dedifferentiated histopathologic character (G3/G4). The presence of an very aggressive phenotype and lymph node infiltration coincided with strong CD97 immunostaining. By contrast, patients with OSCC and weak CD97 protein expression are devoid of lymph node metastases and should have better prognosis. We also observed a strong CD97 immunostaining around carcinoma cell clusters. We could not confirm the results of previous research reports where it was postulated that CD97 is significantly stronger expressed at invasion fronts of squamous cell carcinomas (17). Only a limited number of esophageal squamous cell carcinomas (n = 13) with no information regarding the pTNM, histotype, and tumor location were investigated, and this leads to misinterpretation of data obtained from a usually very common type of carcinoma (17). By contrast, our immunostainings show that CD97 is expressed in OSCC cells irrespective of their location.

All investigated OSCC tissues were found to be CD55 immunopositive. Previously, the role of CD55 as a target for cancer vaccines in gastric and colon carcinomas was discussed (27). The absence of a differential production of CD55, irrespective of the stage, grade, and level of immunoreactive CD97 by OSCC cells, indicates a role for CD97 as the regulatory element. Thus, we speculate that strong interactions of CD97 and CD55 may facilitate adhesion of carcinoma cells to surrounding surfaces or blood vessel walls and will result in tumor cell spread.

In the present study, we also show the down-regulation of CD97 by RA and sodium butyrate in three OSSC cell lines. Our data show different responses to RA treatment and a significant CD97 inhibition occurred in two of three cell lines. We conclude that the redifferentiation effect of RA inhibits or stimulates the expression of CD97 and that carcinoma cells respond to RA differently. The unexpected effect of proliferation instead of antiproliferation was recently shown with the help of cell line BHY, which became more aggressive under the influence of low-dose RA (28). The article reports the effects of 1, 0.1, and 0.01 μmol/L RA doses on BHY cells. Cell growth, Matrigel invasion assay, and expression of RA receptors were investigated. All-trans RA and 13-cis RA inhibited the growth of the cells. Interestingly, low-dose RAs enhanced the activity of tissue-type plasminogen activator and activated pro-matrix metallo proteinases. Furthermore, BHY cells expressed all RA receptors and became more aggressive and invasive under the influence of 0.1 and 0.01 μmol/L RA. The results of Uchida et al. indicate that RA, depending on the dose, enhances or inhibits the in vitro invasiveness of OSCC cells via an activation of pro-MMP2 and pro-MMP9. The authors speculated that the RA receptor expression could be one reason why low-dose RA treatment induces proliferation and not antiproliferation (28). Similar to this, we know that physiologic concentrations of RA enhance migration, invasion, and differentiation of normal embryonal cells. The changes in the CD97 expression in all three cell lines suggest that RA might not only be suitable for the redifferentiation of OSCC cells but also be the reason for therapy failures (29).

CD97 was significantly inhibited in all three cell lines by sodium butyrate. This effect of sodium butyrate on the expression of CD97 in the colon carcinoma cell line CaCo is well known (19). Sodium butyrate was tested as a redifferentiation substance in human colon cancer (30). Our data suggest that sodium butyrate may regulate the CD97 transcriptional activity. We know that the induction and activation of expression profiles by sodium butyrate is accompanied by phosphorylation of proteins between 37 and 97 kDa, which is exactly the molecular weight range where CD97 and its isoforms (75-90 kDa) are detectable (31). It may be possible that neoplastic cells treated with sodium butyrate are induced to restore a proapoptotic/apoptotic expression profile, which on one hand results in an inhibition of CD97.

CD97 and CD55 receptor-ligand interactions and adhesive potentials have been described for lymphocytes. Latest data show a possible involvement of CD97-CD55 interaction in leukocyte passaging through the blood-brain border in multiple sclerosis (32).

Published research reports discussed the adhesive potentials of the CD97-CD55 interaction and presented controversial results (33). The idea that this protein-protein interaction is involved in tumor cell spread can be supported and explained by the increased incidence of metastasis in OSCC or thyroid cancer with strong CD97 immunostaining. Whereas these cellular responses are possibly mediated via the TM7 receptor domain of CD97, alternative signaling pathways are likely functional to counteract increased cell death but instead promote clonal expansion and tumor cell invasion. More recently, it was postulated that chondroitin sulfate glycosaminoglycans were identified as ligands for the largest CD97 isoform (34). There is yet no evidence that this interaction of chondroitin sulfate glycosaminoglycans with CD97 could play a role in vivo. Furthermore, it is highly speculative whether the binding to glycosaminoglycans may facilitate adhesion due to lytic activities of tumor cells. No data on the possible involvement of this interaction in pathologic processes are available.

The structure of CD97 is suitable for hormone recognition, hormone binding, and signal transduction. CD97 shares structural similarities with other G-protein-coupled receptors, such as the thyrotropin receptor or the calcitonin receptor (35). Up-regulation or expression of proteins with similar systematic and structural characteristics attracting considerable interest regarding shifted signal transduction features in cancer.

CD97 is expressed in normal basal epithelial cells, which can be seen as progenitor cells. Thus, CD97 must be important for both normal epithelial stem cells and OSCC cells with unchecked growth and aggressive character. Shifting the molecular balance toward an overrepresentation of the EGF-TM7 protein CD97 in OSCC may be regarded as an important contribution to dislodgment and migration of carcinoma cells during the process of invasion and tumor cell spread.

The still unknown function of CD97 might be cleared in future experiments eventually through small interfering RNA gene silencing and inhibition of CD97 transcriptional activities via the modulation of the AU-rich motive of CD97. Methods employing monoclonal antibodies to target CD97 EGF-like domains and to modulate or modify the binding affinities of CD97 are difficult to interpret. Antibodies induce changes in the proteome, which may evoke shifts toward a more aggressive cellular character.

Taken our data together, we complete previous results on the involvement, distribution, and regulation of CD55 and CD97 in human cancer. For the first time, we have identified CD97 as a marker, which is strongly and exclusively up-regulated in dedifferentiated human OSCC.

We thank Kathrin Hammje, Anja Winkler, and Ilka Mannich for excellent technical support; Wenke Stanarius (College of St. Marks and St. John, Plymouth, United Kingdom) for critical help with the article; and Dr. Edgar Spens (Zentrum für ZMK, Martin-Luther-University, Halle, Germany) for the superb and interesting discussion.