Abstract
Recently, oral mucosal epithelial cells were proposed as a cell source of the autologous cell transplant therapy for corneal trauma or disease. The question addressed is to know if the biological conditions of grafting could induce certain cellular, molecular, and genetic alterations that might increase the risk of mutations and possibly of cellular transformation. Recent progress in cancer research enables us to depict the generation mechanisms and basic characteristics of human cancer cells from molecular, cytological, and biological aspects. The aim of this study is to evaluate the risk of tumorigenicity of the oral mucosal epithelial culture process in order to mitigate that risk, if any, before clinical application. Oral mucosal epithelial cells from three different human donors were investigated by combinational examinations to detect possible tumorigenic transformation. We investigated (i) clonogenic and karyology types, (ii) the validation of proliferation rate, (iii) the epithelial–mesenchymal transition, (iv) anchorage-independent growth potential, and (v) tumorigenicity on nude mice. Results show that the culture process used in this study presents no risk of tumorigenicity.
Introduction
Regenerative medicine using tissue-engineering technology is expected to open a new era in medical science (11). However, there remain several technical issues to be solved before practical use. One of the important issues is the estimation and reduction of the risk of tumorigenesis, which may be developed after transplantation with tissue-engineered products. However, there has not been established any evaluation design that has consensus to rightly assess the risk of the tumorigenesis. Recent progress in cancer research has enabled depicting the generation mechanisms and basic characteristics of human cancer cells from molecular, cytological, and biological aspects.
The transformation of primary human cells into cancer cells is a multistep process that requires sequential alterations of different well-described pathways. According to previous studies (3,7,9,19,21), the transformation of primary cultures of epithelial cells requires the “targeting” of at least three mechanisms: the combined inactivation of the p53 and Rb pathways, constitutive activation of telomerase, and oncogenic activation of cell proliferation. However, these events have been rarely observed in primary human epithelial cell cultures to date. The epithelial cells preserve their capacity to enter in replicating senescence that is associated with growth arrest, and do not reactivate telomerase activity (6,13). Besides, inactivation of the p53 and the Rb pathways or spontaneous activation of the Ha-Ras pathway has been rarely observed in these primary cultures. Thus, it is quite difficult to evaluate the transformation risk of the human epithelial cells with only molecular biological analysis.
Findings on the cytological characteristics of human cancer cells have also been accumulated. Chromosomal alteration, such as translocations and deletions of specific chromosomes, and aneuploidy, is a hallmark of human cancer cells (17). For example, human small cell lung carcinomas often reveal losses of 3p, 5q, 13q, and 17p (18). The presence of aneuploidy or tetraploid populations is seen in 90–95% of esophageal adenocarcinomas. Alteration of growth characteristics in vitro also occurs accompanied to tumorigenic transformation. Loss of anchorage dependency of growth has been well documented to occur during cell transformation. In the case of transformation, the epithelial cells undergo morphological changes and increase their growth rate (16). In some carcinomas, epithelial–mesenchymal transition, which is defined by a drastic E-cadherin disappearance with vimentin expression by epithelial cells, is often observed during malignant transformation and progression. Although these cellular markers are useful to suspect transformation, they are still not enough to demonstrate neoplastic transformation. Tumor formation assay using immunodeficient mice is still the most relevant and direct bioassay to prove the tumorigenic transformation of the cells to date.
We report here the results of combinational examinations to detect tumorigenic transformation against cultured human oral mucosal cell sheets, a novel tissue-engineered product for treatment of human corneal limbal stem cell deficiency. From this experience, we will propose a set of assays to assure the safety of tissue-engineered product from the risk of tumorigenesis.
Materials and Methods
This study was reviewed and approved by Institutional Animal Care and Use Ethics Committee of Centre Léon Bérard and was conducted in accordance with guideline on animal welfare.
Tissue Procurement
Human oral mucosal epithelial cells were isolated from maxilofacial surgical residues of three volunteer donors, 28, 34, and 39 years old, treated for orthognathic surgery (CCE07083, CB08002, and CCS07097) after informed consent was obtained. Active infection by HIV and hepatitis virus was ruled out in all donors using serological analysis. For inclusion criteria, patients had to be more than 18 years old. History of neoplasic disease was an exclusion criteria.
Cell Isolation and Culture
Three oral mucosal cell cultures from three independent donors (CCE07083, CB08002, and CCS07097) were established and analyzed. Dissected human oral mucosal tissues were treated with dispase II (Roche, 3 mg/ml). Following that step, the epithelial layers were separated from the connective tissue compartments under a dissection microscope. Oral mucosal epithelial layers were then dissociated into single cells by trypsin-EDTA 0.05% treatment at 37°C for 10 min. Cells obtained were seeded on culture dishes (CellSeed Inc; UpCell®) at the densities of 100 to 16,000 cells/cm2 in Green medium, a 3:1 mixture of DMEM and Ham's F12 (Invitrogen) supplemented with 10% fetal calf serum (HyClone fetal II), epidermal growth factor (EGF) (Austral Biologic, 10 ng/ml), insulin (Lilly, 0.12 IU/ml), hydrocortisone (Upjohn, 0.4 μg/ml), triiodo-l-thyronine (Sigma, 5 μg/ml), selenium (Aguettant, 0.033 μg/ml), Isuprel (Abbott, 0.4 mg/ml), penicillin 100 UI/ml, streptomycin 100 μg/ml, and fungizone 1 μg/ml. All the experiments performed for tumorigenic test use the same cocktail of medium and supplements, including material as those used clinically.
Colony Forming Efficiency (CFE) and Growth Rate (GR)
For clonogenic potential determination, cells were seeded on three dishes containing feeder layer for each strain. After mitomycin C treatment (8 mg/ml) during 2 h, feeder layer cells of murin fibroblast NIH3T3 were seeded in Green medium at a density of 10×16 × 103 cells/cm2 24 h before keratinocyte seeding. The epithelial cells were inoculated to the dishes at the densities of 10, 20, or 40 cells/cm2. After cultivation for 12–14 days, the cells were fixed and stained with rhodamine B to calculate colony forming unit (CFU). Then, the number of colonies was counted under a dissecting microscope. The clonogenic potential was estimated by the CFE (%), which represents the percentage of cells giving colonies: CFE = CFU/inoculum x 100.
For growth rate (GR) determination, cells were seeded on three other flasks and cultured for 14 days. Then, the cells were dissociated by a trypsin treatment and counted flask by flask. Population doubling (PD) was calculated as: PD = ln(cell number/CFF)/ln2. Growth rate was determined as: GR = PD/culture day number.
Chromosomal Analysis
For this experiment, cells were cultured in Keratinocytes-SFM (Gibco) supplemented with 0.25% (v/v) of bovine pituitary extract and 2.5 μg/L of EGF and antibiotics to stimulate cell growth without feeder layer. Standard cytogenetic techniques using GTG (Giemsa-Trypsin-Giemsa) and RHG (Reverse banding using Heat and Giemsa) banding were applied for the chromosome examination. The analyses were performed on all the observable mitoses per strain (approximately 8) on 46 chromosomes.
Detection of Epithelial-Mesenchymal Transition by Western Blotting
Epithelial–mesenchymal transition, which is drastic disappearance of E-cadherin with important vimentin expression by keratinocytes, was analyzed by measuring E-cadherin and vimentin expression by Western blotting. Cultured oral mucosal cells at early (P1) and high passages (P5) were trypsinized and collected by centrifugation. The resulting cell pellets were frozen at −80°C until analyses. Cells were lysed by adding a lysis buffer [50 mM Tris-HCl, pH 7.4, 250 mM NaCl, 0.1% SDS, 0.5% Nonidet P-40 (NP40) containing protease inhibitors (2 μg/ml aprotinin, 500 μM phenylmethylsulfonyl fluoride, 0.5 μg/ml leupeptin, 1 pg/ml pepstatin, and 2 mM dithiothreitol)] and placed on ice for 30 min. Extracts were subsequently centrifuged at 13,000 rpm (Centrifuge 5804, Eppendorf) for 15 min at 4°C. Supernatants were collected. Protein concentration was determined using Bradford reagent (BioRad, CA, USA). Protein extracts (30 μg) were separated on 7.5% SDS-PAGE and transferred to polyvinylidine difluoride (PVDF, Amersham, England) membranes by electroblotting. Membranes were blocked with 5% dry milk in PBS-NP40 (0.05%) for 1 h at room temperature and incubated overnight with primary antibody diluted in 1% PBS-NP40 0.001% containing 1% dry milk. The following antibodies were used: anti-E-cadherin clone 36 (Becton-Dickinson) at the concentration of 0.25 μg/ml, and anti-vimentin clone V9 (Dako) at dilution of 1:500. The membranes were washed and incubated with peroxidase-conjugated secondary anti-mouse IgG antibody (Dako) at dilution of 1:1000. The signals were detected by chemiluminescence kit (Amersham).
TP53 Status
For mutation detection, DNA was extracted from the three cells strain at passage 5. TP53 mutation analysis of exons 5–8 was performed by polymerase chain reaction (PCR)-based direct sequencing. The following forward (F) and reverse (R) primers were used: 5F (5′-TGTT CACTTGTGCCCTGACT-3′) and 5R (5′-CAGCCCTG TCGTCTCTCCAG-3′); 6F (5′-GCCTCTGATTCCTCA CTGAT-3′) and 6R (5′-TTAACCCCTCCTCCCAGAG A-3′); 7F (5′-CTTGCCACAGGTCTCCCCAA-3′) and 7R (5′-AGGGGTCAGCGGCAAGCAGA-3′); 8F (5′-TTC CTTACTGCCTCTTGCTT-3′) and 8R (5′-AGGCATA ACTGCACCCTTGG-3′).
PCR conditions were as follows. For exons 5, 6, and 8, denaturation at 94°C for 2 min followed by annealing for 45 s at temperatures from 63°C to 60°C, decreasing 0.5°C every three cycles and extension at 72°C for 1 min. For exon 7, the selected condition was denaturation temperature of 95°C for 15 min followed by 94°C for 30 s, annealing at 60°C for 30 s, and extension at 72°C for 1 min. Aliquots of PCR products were examined by electrophoresis on 2% agarose gel containing ethidium bromide. PCR products were treated with 2 μl ExoSAP-IT (GE Healthcare) at 37°C for 15 min followed by inactivation at 80°C for 15 min and directly sequenced using Applied Big Dye Terminator v1.1 cycle sequencing method on ABI Prism 3100 Genetic Analyzer (Applied Biosystems). The sequences obtained were compared with the reference sequence, X54156, from Genbank (http://p53.iarc.fr/TP53sequenceX54156.html).
Soft Agar Assay
To determine the anchorage-independent growth potential, colony formation was measured in soft agar. Cells of each strain were seeded into T75 culture flasks and grown to 60–70% confluency. Agar media (0.75% and 0.45%) were prepared by mixing 1.5% or 0.9% agar, respectively (Seaplaque agarose, Cambrex Tebu Bio), and DMEM-HAM's F12 2X (Sigma-Aldrich) with 20% fetal bovine serum in proportion 1:1. For the under-layer, 1.5 ml of 0.75% agar medium was added to 60-mm wells. 293T cells were used as positive control. They were generated by transformation of cultures of normal human embryonic kidney cells with sheared adenovirus and transfected by SV40 large T-antigen. These cells are able to grow in soft agar and induced tumors in nude mice (22). Cells at P5 were harvested by incubation in trypsin-EDTA, centrifuged at 1000 rpm, and 5000 cells were suspended in 1.5 ml of 0.45% agar medium. Cell suspensions were then plated on the under-layer of 0.75% agar. Agar plates were incubated at 37°C for 3 weeks. Cell colonies were scored using an inverted microscope. By convention, cell cluster measuring more than 80 μm in greatest diameter were considered a colony. Each soft agar assay was performed in triplicate.
Tumorigenicity Assay on Nude Mice
The animal system chosen was female athymic mice aged of 6 weeks (nu/nu genotype), from Charles River (Lyon, France). After 3 weeks of quarantine and acclimation during at least 3 days, mice are put in animal facilities compliant with regulation. HPKIA cell line at passage 345, a transformed cell line, was used as a positive control, and HPKIA at passage 28, nontumorigenic cell line, was used as a negative control (15). The oral mucosal cells and the HPKIA tumor cells as reference strains were harvested, washed with serum-free medium, and counted. The inoculum of each donor was subcutaneously injected in three nude mice near the scapula. Each animal received 5 × 106 cells suspended in 0.2 ml of PBS. The different cell strains were injected into separate groups. All the animals including the reference group were observed and palpated once a week for the formation of nodules at the sites of injection. The period of observation was 3 months, and at the end all animals including the reference group were killed and autopsied. Injection site area, kidney, lung, liver, and lymph node were analyzed.
Results
Clonogenicity and Karyology
Clonogenicity was measured by colony forming efficiency (CFE). As shown in Table 1, CFE and GR have a tendency to decrease with the passage. Moreover, the cells gradually altered their morphology as a whole into a senescent one along with the progression of culture periods (Fig. 1). Indeed, they became a little bigger with a more differentiated aspect, and the cells died after 10 passages. Cells that were suggested to have transformed phenotypes, such as spindle-shaped or rapidly growing in small dense colonies, were never found during cultivations, and died after 10 passages.
Evolution of the morphological aspect of the three cell strains in culture with the passages. The microscopic pictures show that the three cell strains present different morphology with the passages. At P2, the cells form large colonies with smooth perimeter and contain small cells. At P10, the cells form clones with wrinkled perimeter and the cells are large and flattened. Scale bars: 10 μm for early passage and 20 μm for late passage. Clonogenicity of the Three Cell Strains The colony forming efficiency (CFE), population doubling (PD), and growth rate were calculated for each cell strain at P1 to P5. The percentage of CFE decreases with the passages. The results of karyotype is indicated where performed. Values are mean ± SD. NE, not evaluated.
Continuous serial passaging of any cell types will eventually lead to chromosomal rearrangements, sometimes resulting in a transformed phenotype. To assess the chromosomal stability of cultured oral mucosal epithelial cells, the karyotypes were evaluated at P4 and P5 (Fig. 2). Any major chromosomal abnormalities could not been observed in any mitoses of the oral mucosal epithelial lines. The data indicate that the manufacturing process of cultured human oral mucosal cell sheets is likely to maintain a normal karyotype for at least four passages over this use with culture process.
Karyotype of CB 08002. Example of karyotype obtained by standard cytogenetic techniques using GTG (Giemsa-Trypsin-Giemsa) and RHG (Reverse banding using Heat and Giemsa) banding.
E-cadherin and Vimentin Expression in the Cells at P1 and P5
Recent evidences in cancer research have revealed the important roles of the epithelial–mesenchymal transition (defined by a complete disappearance of E-cadherin with vimentin expression) in the premalignant stage such as oral submucous fibrosis, as well as in the progression stage for the development of human malignant tumors (10,20). To evaluate EMT, E-cadherin expression, a specific marker of epithelial cells, and vimentin expression, a specific marker of mesenchymal cells, were investigated by Western blotting on the cells cultured at P1 and P5. Figure 3 shows the positive control EMT cells, expressing high level of vimentin, and drastic decrease of E-cadherin expression as expected. Whatever the passage, P1 or P5, oral mucosal epithelial cells always well expressed E-cadherin and no or low level of vimentin, as shown in Figure 2. A band can be seen at vimentin size in CCE0783 at P5 and CB08002 at P1 probably due to the fibroblasts feeder layer or human fibroblast contamination. So, we can conclude that no epithelial–mesenchymal transition occurs during the culture.
E-cadherin and vimentin expression at P1 and P5. All three cell strains express E-cadherin but not vimentin.
TP53 Status
Mutations in the tumor suppressor gene TP53 are frequent in most human cancers. The TP53 database (www.iarc.fr/P53/) compiles all mutations (somatic and inherited), as well as polymorphisms, that have been reported in the published literature since 1989. Exon 5 to 9 encodes the DNA binding domain containing more than 90% of all mutations in human cancers and only few SNPs (with a low frequency) occur in this region (8,12). No TP53 mutation neither polymorphism was detected in region at exon 5 to 9 in the three cells strain CCE07083, CB08002, and CCS07097 at P5.
Soft Agar Test
The soft agar test was performed to examine the anchorage-independent growth potential of cells, which is a well-known characteristic of transformed cells. This test was performed using a transformed cells line 293T (positive control) and oral mucosal epithelial cells, CCE07083, CB08002, and CCS07097 at P5. After 21 days of culture, we observed that the positive control (293T cells) grew in colonies in the agar and that the negative control did not. The positive control gave rise to 76 colonies of at least 80 μm diameter per well after 5 weeks of culture. This confirmed that the test was functional. No oral mucosal cells, CCE07083, CB08002, or CCS07097, in P5 grew in soft agar (Fig. 4).
Soft agar assay of the three cell strains. The three cell strains are not able to grow after 3 weeks in soft agar, contrary to 293T cells. Scale bars: 70 μm.
Tumorigenicity Assay Using Nude Mice
Cultured oral mucosal cells derived from the three donors were injected into nude mice. The observations were continued at least three months. We used an HPV16-immortalized keratinocytes cell line (HPKIA) as a control, which has acquired the ability to form squamous cell carcinomas in nude mice after gamma-irradiation and long-term culturing in vitro. This cell line expresses the oncoproteins E6 and E7 of human papillomavirus type 16 (HPV16) previously generated by an independent group from an independent institute (5). The cell line becomes immortal after a few passages, but is not able to induce tumors when injected in nude mice at early passages (negative control). In contrast, after cultured over 300 doubling populations, it had acquired a transformed phenotype and their injection in nude mice results in tumor formation (positive control).
Three months after the injection, all the mice receiving positive tumor control gave rise to rachitis and developed tumors at the injection points. On the contrary, three mice receiving the negative control did not show any tumors at the injection site and were in good health. Eight mice of nine that received 5 million proliferative oral mucosal cells at passage P4 were in good health without rachitis and did not present any tumors and nodules by palpation during the 4 months of the experiment, and no tumors were found after autopsy. One mouse died after 1 month but the death was not related to the cell injection, but most likely to the well-known fragility of the nude mice.
Discussion
In the present study, we report the results of combinational examinations to detect tumorigenic transformation in cultured human oral mucosal cell sheets, a novel tissue-engineered product cultured with the same process and ingredients as investigational products used for a clinical trial for treatment of human corneal limbal stem cell deficiency. We have evaluated the tumorigenic risk of cultured oral mucosal epithelial cells from the various aspects, and could not find any evidences that suggest the risk of tumorigenesis. All the tests performed on the oral mucosal epithelial cells from three oral mucosa different donors did not identify any modifications inducing or facilitating malignant transformation.
Due to the recent progress of cancer research, a lot of common characteristics of human cancer cells have been revealed. Alteration of TP53 gene is one of the possible mechanisms for accelerated genetic aberrations. Despite the normal karyotype, some DNA damages might have been present in cells of the original tissue. Altered growth and highly abnormal karyotype are also generally believed to be indicators for tumorigenic conversion of human primary cells. For validation, transplantation assay in nude mice is the reference and the most powerful technique. However, a unified characteristic of cancer cells has not yet been found, and none of these assay systems is enough to indicate the tumorigenic risk of tissue-engineered products precisely.
We therefore attempted to perform combinational assays to investigate any signals of tumorigenesis from the molecular, cytological, and biological aspects. These combinational examinations should reduce the possibility to pass over low signal of tumorigenesis. Our results strongly suggest that the tumorigenic risk of cultured human oral mucosal cells is quite low.
Four reports have demonstrated the spontaneous immortalization of human keratinocytes (1,2,4,14) exhibiting sometimes a transformed phenotype (e.g., HaCaT). However, this phenomenon requires long-term in vitro culture conditions and is normally associated with a lag phase. During this phase, cells stop proliferating, enlarge, and remain growth arrested for a relatively long period (3–6 weeks). Thereafter, islands of small proliferating cells may appear and continue to grow without further interruption. A cellular transformation is quite visible because it goes through a phase of senescence associated with a decrease of the growth rate.
In our case, we obtained the favorable advice of our competent authorities to transplant Cultured Autologus Oral Mucosal Epithelial Cell Sheets (CAOMECS) on patients suffering from total, bilateral limbal epithelial stem cell deficiency (AFSSAPS TC 220) but without other pathology. Twenty-seven patients received this product without serious adverse event with preliminary good success. Presently, 17 patients have more than 1 year follow-up. The trial will be closed in June 2010 (all patients followed up 1 year). The CAOMECS is grafted on human cornea after culturing only at passage P0-P1. The xenogenic feeder layer is separated from the culture epithelial cells thanks to UpCell®-Inserts technology. In those conditions, they only act for releasing soluble growth factors. The oral mucosa epithelial cells of the three donors did not increase in the growth rates after five passages and their morphological aspect tends to be as senescent cells. The P5 cultivated oral mucosal cells did not show any signs of epithelial-mesenchymal transition, including morphological change and marker expression. And the transplantation experiments using cells at P4 were always negative. From these results, it is extremely unlikely that the epithelial cells accumulate any mutations that could lead to the settlement of a neoplastic phenotype during this very short time of culture passage (P0 or P1).
Furthermore, since the beginning of the clinical application of epidermal sheets, no case of cancer resulting from this kind of grafts has been published in the world or been observed at our laboratory since 1988. We have cultured about epithelial cell strains from 200 donors but 4 years ago only one has shown some phenotypes and growth abnormalities at high passages (submitted). Even such a cell line was neither immortalized nor transformed until passage P25. Moreover, it has been shown that biological conditions, such as inflammatory and immunity, improbably induce cellular, molecular, or genetic lesions that could increase the risk of mutations on cultured allogenic epithelial skin cells graft.
Considering our results, our experiences in epidermal cell sheet transplantation, and literatures reported, it is concluded that the tumorigenic risk of human epithelial cells cultured within a few passages, such as cultured human oral mucosal cells in P0 or P1, should be quite low.
Footnotes
Acknowledgments
We thank CellSeed for financial support and Professor Pierre Breton and Doctor Pascal Pierrillas from “service de chirurgie maxilla-faciale de l'hôpital Lyon Sud” for oral mucosa samples. Amélie Thepot is supported by a fellowship of the Comité de Saône et Loire de la Ligue contre le cancer.