Cancer Cell Biology
Escape from microenvironmental control and progression of intraepithelial neoplasia
Article first published online: 26 APR 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 116, Issue 6, pages 885–893, 10 October 2005
How to Cite
Zhang, W., Vaccariello, M. A., Wang, Y., Alt-Holland, A., Fusenig, N. E. and Garlick, J. A. (2005), Escape from microenvironmental control and progression of intraepithelial neoplasia. Int. J. Cancer, 116: 885–893. doi: 10.1002/ijc.21103
- Issue published online: 28 JUL 2005
- Article first published online: 26 APR 2005
- Manuscript Accepted: 28 JAN 2005
- Manuscript Received: 4 JUL 2004
- National Institutes of Dental and Craniofacial Research. Grant Number: DE-11250-06
- cell–cell interactions;
- intraepithelial neoplasia;
- tumor microenvironment;
- organotypic culture;
We previously reported that normal human keratinocytes controlled neoplastic progression of tumor cells at an early stage of transformation in stratified squamous epithelium. We now studied if cells at a more advanced stage of transformation were also subject to such microenvironmental control. To accomplish this, 3D human tissues that mimic intraepithelial neoplasia were fabricated by mixing genetically marked (β-gal), early-stage (II-4 cells) or advanced-stage (SCC13) transformed keratinocytes with normal keratinocytes, and tumor cell fate and phenotype were monitored in organotypic culture and after surface transplantation to nude mice. In vivo, SCC13 cells evaded local growth suppression to undergo connective tissue invasion at significantly lower tumor cell volumes (12:1, 50:1 normal:tumor cells) than II-4 cells. This behavior was explained by the growth suppression of II-4 cells, while advanced-stage tumor cells escaped this control and continued to undergo clonal expansion in mixed cultures to form large, intraepithelial tumor clusters. These communities of tumor cells underwent autonomous growth that was associated with altered expression of markers of differentiation (keratin 1) and cell–cell communication (connexin-43). Furthermore, significantly greater numbers of SCC13 cells expanded into a basal position after low-calcium stripping of suprabasal cells of mixed cultures compared to II-4 cells, suggesting that expansion of these cells enabled tumor cell invasion after transplantation. These findings demonstrated that early tumor development in human stratified squamous epithelium required escape from microenvironmental growth control that was dependent on the transformation stage of intraepithelial tumor cells during the premalignant stage of cancer progression. © 2005 Wiley-Liss, Inc.
The multistage model of skin carcinogenesis has revealed predictable and progressive stages during the development of SCC in rodent and human epithelium.1, 2 Much progress has been made in the identification of genetic changes that are characteristic of each of these stages in the evolution of skin cancer. The sequential activation of cellular oncogenes3 and the inactivation of tumor-suppressor genes4 have been shown to determine the properties of individual cells at varying stages of neoplastic progression. Cells at a premalignant stage of progression have been associated with non-random, sequential chromosomal alterations that result in a cell phenotype imparting a selective growth advantage permissive for clonal outgrowth of the altered cell.5, 6, 7, 8 However, in addition to these well-characterized mutations, studies have shown that signaling events inherent in the maintenance of normal tissue architecture control malignant progression and repress the tumor phenotype in a number of tissue types.9, 10, 11, 12
An emerging view in tumor biology is that the tumor cell microenvironment plays an important role in the control of neoplastic progression through dynamic reciprocity between tumor cells and their surrounding tissues.13 In this way, the phenotype of the tissue microenvironment may alter the manifestation of those genetic defects that are associated with distinct stages in carcinogenesis.11, 14, 15 In our previous studies, we determined that interactions between tumor cells at an early stage of transformation and their adjacent normal neighbors could also modulate the early stages of neoplastic progression in stratified squamous epithelium. Using 3D tissue models that mimic the onset of malignant disease in stratified epithelium, we demonstrated that early neoplastic progression of low-grade malignant keratinocytes could be suppressed by normal cell context and tissue architecture.11 This growth suppression limited clonal expansion of early-stage transformed cells and prevented expression of their neoplastic phenotype. We subsequently demonstrated that this suppression could be overcome through modification of the immediate cellular microenvironment by altering interactions between tumor cells and cells adjacent to them14, 15 or by the presence of basement membrane in 3D cultures.16 The fate and neoplastic potential of keratinocytes at an early stage of malignant transformation in stratified epithelium was thus dependent on their interactions with neighboring cells and with underlying stroma during the intraepithelial stages of neoplastic progression. While murine keratinocytes at a more advanced stage of malignant transformation are not subject to this tissue-induced suppression of neoplastic phenotype,17 this has yet to be studied in human tissues.
The purpose of this investigation was to determine if neoplastic development of intraepithelial tumor cells at an advanced stage of transformation that manifest more aggressive biologic behavior than cells at an early stage of transformation, could also be controlled by normal cells in the microenvironment. To accomplish this, we constructed 3D tissue by mixing genetically marked (β-gal), early-stage (II-4) or advanced-stage (SCC13) transformed keratinocytes with NHKs at varying ratios to generate dysplastic tissues. The fate of these β-gal-expressing cell lines was followed in organotypic culture in vitro and after transplanting cultures to nude mice. In this way, we directly determined how interactions between NHKs and intraepithelial tumor cells could influence the fate and phenotype of tumor cells in premalignant tissue during the early phases of neoplastic progression. We found that only tumor cells at an advanced stage of transformation had acquired the properties needed to circumvent microenvironmental growth control and that escape from this regulation was dependent on the stage of tumor cell transformation.
Material and methods
Epidermal NHKs were cultured from newborn foreskin by the method of Rheinwald and Green18 in keratinocyte medium described by Wu et al.19 Cultures were established through trypsinization of foreskin fragments and grown on irradiated 3T3 fibroblasts. 3T3 cells were maintained in DMEM containing 10% bovine calf serum. Tumor cells at a late stage of transformation were SCC13 cells, which were originally derived from a cutaneous squamous cell carcinoma20 and shown to be highly tumorigenic in vivo21 and to have very limited potential to undergo terminal differentiation in vitro.22 Tumor cells representing an early stage of the transformation process were HaCaT-II-4 cells (II-4 cells), which were derived by transfection of the spontaneously immortalized human keratinocyte line (HaCaT)23 with activated c-Harvey-ras oncogene.24 These cells display severe dysplasia in organotypic culture and low-grade malignant behavior after in vivo transplantation.11 For 2D monolayer cultures, the HaCaT-derived cell line (II-4) was grown in DMEM containing 5% FCS while SCC13 keratinocytes were grown in keratinocyte medium. 3D organotypic cultures were prepared as previously described.11 Briefly, early-passage human dermal fibroblasts were added to neutralized Type I collagen (Organogenesis, Canton, MA) to a final concentration of 2.5 × 104 cells/ml. Three milliliters of this mixture were added to each 35 mm well of a 6-well plate and incubated for 4–6 days, until the collagen matrix showed no further shrinkage. At this time, 5 × 105 NHKs, II-4 cells, SCC13 cells or mixtures of NHKs with either II-4 or SCC13 cells were plated on the contracted collagen gel. Cultures were submerged in low-calcium EGM for 2 days, submerged for 2 days in normal-calcium EGM and raised to the air–liquid interface by feeding from below with normal calcium cornification medium for 3 days.25 For proliferation assays, BrdU (Sigma, St. Louis, MO) was added to organotypic cultures 8 hr prior to harvesting at a final concentration of 10 μM. 3D tissues were analyzed in triplicate for 3 independent experiments.
Retroviral vectors and transduction of keratinocytes
The MFG-gal vector is a Moloney murine leukemia virus–based vector which contains the gene for bacterial β-gal.26 Transduction of II-4 and SCC13 cells with this vector was performed as previously described.27 Briefly, cells were transduced 24 hr after plating 1 × 106 cells in a 100 mm dish using fresh, filtered (0.45 μ; Gelman, Ann Arbor, MI) supernatant from confluent amphotropic cells producing the MFG-gal vector. Transduced cells were passaged at clonal density, and clones were screened for persistence of transgene expression after 3 passages and after transplantation to nude mice. Only clones maintaining transgene expression in 100% of cells were expanded and used for organotypic culture and grafting experiments. No deleterious effect on cell growth or phenotype was seen after transduction. Producer lines were maintained in 10% bovine calf serum.
Transplantation of organotypic cultures to nude mice
Organotypic cultures were trimmed using a surgical punch 1.4 cm in diameter. Six-week-old male Swiss nude mice (N:NIHS-nuf DF; Taconic Farms, Germantown, NY) were anesthetized using xylazine:ketamine (2:3), and 1.3 cm of dorsal skin was removed. Organotypic cultures were placed onto this area, covered with petrolatum gauze (Sherwood Pharmaceuticals, St. Louis, MO) and secured with bandages (Baxter Scientific, McGraw Park, IL). These dressings were changed after 7 days and removed completely after 14 days. Three independent experiments were performed, and 3 animals were killed at both 4–6 weeks and 8–9 weeks after transplantation for each experiment. All animal experiments were performed with necessary approval from the SUNY Animal Welfare Committee.
Stripping of suprabasal cells in low-calcium medium and regrowth of basal cells in organotypic culture
Low-calcium medium (LCEM) was used to strip suprabasal cell layers from organotypic cultures by adapting the protocol described for submerged cultures.28 LCEM was prepared by mixing minimal essential medium without calcium or magnesium (GIBCO BRL, Gaithersburg, MD) with nonessential amino acids (GIBCO BRL) and 10% FCS that had been depleted of divalent cations by treatment with Chelex 100 (Bio-Rad, Richmond, CA). Organotypic cultures were submerged for 4 days and then incubated in LCEM for 60 hr. The remaining organotypic construct consisting of the basal cells and collagen matrix was bisected, and one-half of the culture was processed for frozen and paraffin-embedded tissue analysis. The other half was regrown in organotypic culture in low-calcium EGM for 2 days, submerged for 2 days in normal-calcium EGM and raised to the air–liquid interface for an additional 3 days in normal-calcium cornification medium. After recovery and stratification, cultures were processed for immunohistochemistry and hematoxylin and eosin staining as described below.
Specimens were frozen in embedding medium (Triangle Biomedical, Durham, NC) in liquid nitrogen vapor after being placed in 2 M sucrose for 2 hr at 4°C. Tissues were serial sectioned at 6 μm and mounted onto gelatin/chrome alum–coated slides. Tissue sections were washed with PBS and blocked with 10 μg/ml goat IgG, 5% goat serum and 0.2% BSA vol/vol in PBS without fixation. Sections were incubated with rabbit polyclonal antiserum to bacterial β-gal (Cortex Pharmaceuticals, San Leandro, CA) and detected with Alexa 488–conjugated goat antirabbit IgG (Molecular Probes, Eugene, OR). Double-stain immunofluorescence was performed by staining for β-gal expression using a MAb to either K1 (Enzo Diagnostics, Farmingdale, NY), BrdU (Boehringer-Mannheim, Indianapolis, IN) or CX43 (Chemicon, Temecula, CA), detected with Alexa 594–conjugated goat antimouse IgG (Molecular Probes). Slides were coverslipped with Vectashield containing 1 μg/ml DAPI (Vector, Burlingame, CA). Fluorescence was visualized using a Nikon (Tokyo, Japan) OptiPhot microscope, and double-exposure photomicroscopy was performed using FITC and Texas red filters. For routine light microscopy, tissues were fixed in 10% neutral buffered formalin and embedded in paraffin and 4 μm sections were stained with hematoxylin and eosin.
Indices of proliferation, clonal expansion and persistence in a basal position
To further quantify immunohistochemical data, tumor cell expansion, proliferation and the percentage of II-4 and SCC13 cells in the basal layer of the epithelium were measured in 5 tissue sections which were at least 100 μm apart after serial section. The degree of expansion was measured using the NIH Image program (version 1.61; NIH, Bethesda, MD) and determined as the area of β-gal-positive cells divided by the total area of cells in the tissue section. PI was calculated as the percentage of β-gal-positive cells that incorporated BrdU during S phase by determining the fraction of cells that coexpressed β-gal and BrdU for mixed cultures and the percentage of BrdU-positive cells in pure cultures. Basal persistence of II-4 or SCC13 cells was quantified as the percentage of basal cells which were β-gal-positive. These sections were counterstained with DAPI to allow identification and counting of all nuclei.
II-4 cells were growth-suppressed while SCC13 cells continued to proliferate and undergo clonal expansion in mixed organotypic cultures
To gain insight into the behavior of early- and advanced-stage transformed keratinocytes after mixing with NHKs in 3D organotypic cultures, the distribution of genetically marked cells in normal cell context was mapped by immunohistochemical detection of β-gal. To follow the fate of marked, transformed cells, it was first determined that β-gal expression was maintained in all cells by detection of β-gal in pure II-4 and SCC13 cultures (Fig. 1a,b). These cultures generated a well-stratified epithelium that demonstrated proliferating cells in both the basal and suprabasal layers, as seen by BrdU-positive nuclei in these tissues. No β-gal positivity was observed in control cultures of NHKs where BrdU-positive cells were limited to the basal layer (data not shown). Mixtures generated with equal numbers of NHKs and tumor cells (1:1) showed expansion of tumor cell populations which dominated most of the tissue for both II-4 and SCC13 cells (Fig. 1c,d, Table I). In contrast, in both 4:1 and 50:1 mixtures (NHK:tumor cell), II-4 cells appeared in the upper layers of the epithelium as individual cells that did not expand (Fig. 1e,g, arrows). In contrast, SCC13 cells were present as clusters that continued to undergo expansion and proliferate at 4:1 and 50:1 mixing ratios, as seen by the BrdU-positive nuclei present in expanded β-gal-positive clusters (Fig. 1f,h, arrows). Interestingly, the β-gal-positive SCC13 clusters were smaller in 50:1 mixtures than in 4:1 mixtures. However, the presence of BrdU-positive nuclei in these cells (Fig. 1h, arrow) showed that they were not growth-suppressed. This demonstrated the ability of advanced-stage tumor cells to circumvent the NHK-induced growth suppression seen for II-4.
|Tumor cell type and (initial ratio NHK:tumor cell)||NHK:tumor cell ratio at day 7||PI of tumor cells||% of basal cells that are β-gal-positive|
The lack of SCC13 growth suppression was also evident when the percentage of β-gal-positive area was calculated in multiple tissue sections from organotypic cultures (Table I). While the fraction of the epithelium that was occupied by β-gal-positive II-4 cells was essentially unchanged after 7 days in organotypic culture, NHK:SCC13 mixtures cultured at these ratios demonstrated a 4- to 6-fold increase in the percentage of SCC13 cells present (Table I). When PI was determined, 12:1, 4:1 and 1:1 mixtures (NHK:SCC13) showed 6%, 22% and 23% BrdU-positive SCC13 cells, respectively. In contrast, there were no BrdU-positive nuclei detected in any II-4 cells in either 12:1 or 4:1 (NHK:II-4) mixtures, while II-4 cells in 1:1 mixtures demonstrated LI of 13%. Nonmixed, pure control cultures demonstrated PIs of 24.5% and 45% for II-4 and SCC13, respectively. These results further demonstrated that II-4 cells were growth-suppressed at tumor cell volumes (4:1 and 12:1 ratio) at which SCC13 could continue to proliferate and expand.
SCC13 cells persisted in a basal position in mixed organotypic cultures and repopulated the epithelium, while II-4 cells were growth-suppressed upon epithelial regrowth
The distribution of intraepithelial tumor cells was further analyzed by determining the behavior of the SCC13 and II-4 cells that were present in a basal position in mixed organotypic cultures. To accomplish this, suprabasal layers from organotypic cultures of 1:1 mixtures of NHK:II-4 and NHK:SCC13 were stripped in low-calcium medium and the remaining basal cells were regrown for 1 week in 3D cultures. Cultures generated at these ratios demonstrated altered tissue architecture and foci of aberrant cell clusters for II-4 (Fig. 2i) and SCC13 (Fig. 2j) cells that were similar in appearance to the mixed cultures shown in Figure 3e,i. Hematoxylin and eosin staining of stripped cultures demonstrated a cell monolayer at the basement membrane interface for both NHK:II-4 (Fig. 2a) and NHK:SCC13 (Fig. 2e) 1:1 mixtures. Upon β-gal staining, these stripped cultures revealed few persisting II-4 cells (Fig. 2c, arrow) while large numbers of SCC13 cells were retained in this monolayer (Fig. 2g). The presence of these cells in a basal position was further illustrated upon regrowth of these cultures at the air–liquid interface to regenerate 3D cultures. Regrown NHK:II-4 cultures demonstrated normal epithelium (Fig. 2b), which contained scattered II-4 cells in a suprabasal position upon β-gal staining (Fig. 2d, arrows). In contrast, basal SCC13 cells reconstituted a dysplastic epithelium (Fig. 2f) comprised almost entirely of β-gal-positive SCC13 cells (Fig. 2h). These findings demonstrated that most II-4 cells were sorted to a suprabasal position in original 1:1 mixtures and lost upon stripping, while SCC13 cells expanded to occupy a position in the basal layer. To further analyze the position of labeled cells, the percentage of β-gal tumor cells that were present in the basal layer of organotypic mixtures before stripping was calculated in mixtures generated at higher ratios (12:1, 4:1) as well. No II-4 cells persisted in a basal position in 12:1 or 4:1 mixtures of NHK:II-4 (Table I), further demonstrating that the displacement of II-4 keratinocytes from contact with the basement membrane zone was associated with the growth suppression of these early-stage transformed cells. In contrast, most tumor cells in NHK:SCC13 mixtures were initially present in a suprabasal position (Fig. 1f, Table I) and able to proliferate and expand to occupy the basal layer (Fig. 1d). This showed that SCC13 cells present at low intraepithelial tumor cell volumes could preferentially expand to occupy a basal position.
Advanced-stage (SCC13) tumor cells manifest more aggressive behavior and invade at lower tumor volume than early-stage (II-4) cells after in vivo transplantation
To determine the behavior of early- and advanced-stage transformed cells after in vivo transplantation, II-4 and SCC13 cells were grown as pure cultures or as mixed cultures with NHK at 1:1, 4:1, 12:1 and 50:1 (NHK:tumor cell) ratios and grafted to nude mice. The appearance of these tissues immediately prior to grafting and 8 weeks after transplantation is shown in Figure 3. II-4 keratinocytes formed a dysplastic epithelium in organotypic culture (Fig. 3a) and demonstrated elongated rete pegs harboring dysplastic epithelial cells (Fig. 3b, arrows) 8 weeks after grafting. In contrast, while SCC13 cells formed a dysplastic epithelium in organotypic culture (Fig. 3c), they generated a highly infiltrative and poorly differentiated carcinoma after grafting (Fig. 3d, arrows). II-4 cells mixed with NHK at a 1:1 ratio showed dysplastic tumor cell foci in organotypic culture (Fig. 3e, arrows) and focal areas of aberrant, dysplastic intraepithelial tumor cells associated with elongated rete pegs (Fig. 3f) 8 weeks after grafting. In 4:1 mixtures (NHK:II-4), small dysplastic foci of II-4 cells were seen in a suprabasal position within a well-differentiated epithelium in organotypic culture (Fig. 3g, arrow). However, this mixture was grafted for 8 weeks, and the resulting epithelium showed normal tissue architecture (Fig. 3h). In contrast, 4:1 mixtures containing SCC13 cells (NHK:SCC13) showed large intraepithelial clusters of dysplastic SCC13 cells in organotypic culture (Fig. 3i, arrows) and invading islands of poorly differentiated epithelium in the connective tissue 8 weeks after grafting (Fig. 3j). Even at 50:1 mixtures, SCC13 cells demonstrated intraepithelial expansion in vitro (Fig. 3k, arrow) and strands of epithelium protruding into the connective tissue 8 weeks after transplantation (Fig. 3l). These results indicated that II-4 cells generated a low-grade carcinoma while SCC13 cells formed a higher-grade carcinoma in vivo. This difference in histologic grade was manifested in dysplastic cell mixtures, demonstrating that II-4 cells progressed to malignancy only at high intraepithelial tumor cell volumes when equal numbers of II-4 cells and NHKs were present in premalignant tissues. In contrast, SCC13 cells were able to persist, expand and invade at much lower tumor cell volumes.
Tumor cell expansion and invasion in vivo determined by stage of transformation of intraepithelial tumor cells
To further understand the intraepithelial dynamics between NHKs and each of the two tumor cell populations, mixed 7-day organotypic cultures of NHK:II-4 and NHK:SCC13 were transplanted to nude mice and harvested at either early (4–6 weeks) or late (8–9 weeks) time points and the fate of β-gal-marked tumor cells was mapped in vivo by immunofluorescent detection of β-gal expression. When pure cultures were transplanted for 4 weeks, β-gal-positive II-4 cells had expanded throughout the epithelium (Fig. 4a) while SCC13 cells invaded into the connective tissue as small clusters and islands (Fig. 4b). NHK:II-4 mixtures grafted at a 1:1 ratio showed persisting II-4 cells that expanded into the basal position after 4 weeks (Fig. 4c, arrow) and underwent further expansion and invasion 8 weeks after grafting (Fig. 4d, arrow). NHK:II-4 mixtures grafted at a 4:1 ratio showed β-gal-positive cells limited to the most superficial stratum corneum at 4 weeks (Fig. 4e), while no II-4 cells were retained in the epithelium after 8 weeks (Fig. 4f). This indicated that intraepithelial, suprabasal cells seen in vitro had been desquamated from the epithelium after transplantation. In contrast, SCC13 cells grafted at a 4:1 ratio (NHK:SCC13) expanded to dominate the entire epithelium by 6 weeks (Fig. 4g) and were fully invasive after 9 weeks (Fig. 4h, arrow). SCC13 cells grafted at a 50:1 ratio (NHK:SCC13) showed β-gal-positive cells that had expanded laterally in a basal position 6 weeks after grafting (Fig. 4i) and showed contiguous areas of dysplastic cells by 9 weeks (Fig. 4j). Small numbers of SCC13 cells were present in a basal position 6 weeks after grafting 500:1 mixtures (NHK:SCC13) (Fig. 4k), and these cells continued to expand to occupy most of the epithelium 9 weeks after grafting (Fig. 4l). These results indicated that advanced-stage transformed keratinocytes were retained in the tissue to undergo further intraepithelial expansion and invasion even when surrounded by considerably larger numbers of NHKs. This demonstrated that NHKs could not suppress the neoplastic potential of these tumor cells in vivo. In contrast, neoplastic progression of tumor cells at an earlier stage of transformation (II-4) was abrogated when NHKs surrounded these cells.
Neoplastic progression associated with lack of expression of markers of differentiation and cell–cell communication in vitro but not in vivo
Since only advanced-stage, transformed SCC13 cells underwent clonal expansion and invasion in vivo at mixing ratios that were not permissive for expansion of II-4 cells, it was important to further define the phenotype of tumor cells that enabled this behavior. To accomplish this, mixtures were stained by double immunofluorescence for β-gal and either K1 or CX43 after growth in organotypic culture for 7 days and after grafting for 9 weeks. These markers were chosen since the normalizing effect of adjacent normal keratinocytes on intraepithelial tumor cell expansion has been shown to induce differentiation of growth-suppressed tumor cells11 and appears to be regulated through direct cell–cell contact and communication.14, 29, 30 While coexpression of β-gal and K1 was seen in some expanding suprabasal II-4 cells in 1:1 mixtures (Fig. 5a, arrow, yellow stain), expanding clusters of SCC13 cells showed no K1 expression and adjacent NHKs correctly expressed this protein in a strata-specific manner (Fig. 5c). CX43 was expressed in nearly all NHKs surrounding either II-4 (Fig. 5e) or SCC13 (Fig. 5g) clusters yet was not expressed by these tumor cells in organotypic culture. This showed that a high fraction of tumor cells did not express these cytoskeletal or cell–cell communication markers in organotypic mixtures during intraepithelial expansion. In contrast, both SCC and II-4 mixtures upregulated expression of these markers in vivo, as seen by their tissue distribution after grafting. Expression of K1 (Fig. 5b) was completely normalized in NHK:II-4 mixtures at a 1:1 ratio as seen by staining limited to the immediate suprabasal cells in both the invading epithelial tongue and the surface epithelium. In contrast, SCC13 cells in grafted mixtures demonstrated K1 expression that was characterized by an expansion of basal-type cells, as evidenced by the delayed onset of K1 expression to the mid-spinous layer in vivo (Fig. 5d). The lower degree of differentiation of SCC13 cells compared to II-4 cells was reflected by the absence of K1 expression in invading SCC13 tumor islands (Fig. 5d) and by the presence of this protein in invading II-4 cells (Fig. 5b). CX43 expression was upregulated after grafting of SCC13 and II-4 mixtures as II-4 cells expressed CX43 in nearly all cells as seen by colocalization of CX43 and β-gal (Fig. 5f, yellow) while this protein was expressed in a heterogeneous pattern in SCC13 cells (Fig. 5l). In summary, a higher degree of differentiation in vivo was seen in tumors that arose from 1:1 II-4 mixtures than from SCC13 mixtures. SCC13 did not express either K1 or CX43 in organotypic mixtures and showed an altered pattern of differentiation characterized by disorganized and delayed maturation of the epithelium. In contrast, while a heterogeneous pattern of K1 expression was seen in vitro, II-4 cells in 1:1 mixtures demonstrated significant normalization of differentiation in vivo, as seen by the appropriate compartmentalization of K1 and CX43. This induction of more complete differentiation of both tumor cell types after grafting demonstrates that the in vitro tissue model lacks microenvironmental mesenchymal factors that may play a central role in regulating tumor cell phenotype.
We have developed 3D tissue models that simulate premalignant disease and allow cellular and molecular investigation of IEN in stratified epithelium in vitro and in vivo.11, 14, 15, 16 In the current study, we show that intraepithelial tumor cells at an advanced stage of transformation have the capacity to evade the growth-suppressive effects of normal keratinocytes in the cellular microenvironment and to undergo clonal expansion in vitro and invasion in vivo. In contrast, tumor cells at an early stage of transformation (II-4 cells) were growth-suppressed by adjacent normal keratinocytes and failed to form tumors after transplantation to nude mice. This suggested that loss of normal keratinocyte control over early neoplastic progression was related to the high-grade behavior manifested by these cells in vivo and in vitro. In this light, microenvironmental control of tumor cell behavior is limited to cells that retain sensitivity to this regulation, such as intraepithelial tumor cells at an early stage of malignant transformation.
The study of the early stages of malignant progression in human stratified squamous epithelium has been hampered by difficulty in the establishment of cell lines from precancerous, dysplastic lesions. For this reason, the II-4 cell line used in our studies is particularly useful in studying early cancer development as it has been characterized as an early-stage malignant cell line on the basis of its phenotypic and genetic alterations.8 This cell line harbors many of the important genetic hallmarks of intraepithelial and early, invasive squamous cell carcinoma,1 such as mutations in p53 and ras, and exhibits low-grade malignant behavior in vivo. In contrast, the SCC13 cell line represents a significantly more advanced stage of transformation than II-4, as seen by its higher histologic grade, less differentiated phenotype and more aggressive growth behavior. Our comparison of these early- and advanced-stage cell lines demonstrated phenotypic differences in expression of markers associated with differentiation and cell–cell interactions that may partially explain their divergent behaviors when grown in the context of normal cells in 3D tissues. The higher level of morphologic differentiation of II-4 cells was reflected in the expression of K1 in a large number of these tumor cells in mixed organotypic cultures and in the normalized distribution of this protein after transplantation. However, the lower differentiation capacity of SCC13 cells was seen by the absence of K1 expression in mixed organotypic cultures and by the delayed and disorganized expression pattern of this protein after grafting. This was likely due to the significant impairment in the control of proliferation and terminal differentiation of SCC13 cells that has been described for these cells in 3D cultures.31, 32 The lack of microenvironmental control exerted by neighboring normal keratinocytes on intraepithelial tumor cells therefore appears due to the greater degree of autonomous growth that is an essential feature of cells at a more advanced stage of transformation.
We have previously shown that suppression of early neoplastic progression in stratified squamous epithelium requires the maintenance of normal tissue architecture.11, 14, 15 For example, intraepithelial expansion of II-4 cells was not suppressed when these cells were grown in the context of the immortalized but nontumorigenic parental HaCaT cell line.14 In the current report, we demonstrate that small numbers of advanced-stage tumor cells are also in an environment permissive for intraepithelial expansion. This expansion may then lead to significant disruption of tissue architecture, which may further alter the growth-regulatory network imposed by normal cell context. In intestinal epithelium, altered cell–cell adhesion mediated by β-catenin and E-cadherin leads to the development of aberrant tissue architecture and of adenomas10 and supports the view that altered 3D tissue structure can result in loss of tumor cell growth control during the early stages of colorectal tumor development, as well. We previously found that abrogation of E-cadherin-mediated adhesion in II-4 cells resulted in loss of tissue architecture and induced tumor cell invasion in 3D culture, suggesting that maintenance of intercellular adhesion is required to limit intraepithelial tumor cell progression.29, 30 The complete absence of CX43 expression in SCC13 cells undergoing intraepithelial expansion in vitro suggested that this may be related to the capacity of transformed cells at an advanced stage of transformation to be more refractory to the influence of neighboring normal cells. The increased expression of proteins that modulate cell-growth regulation and cell–cell interactions after in vivo transplantation of organotypic mixtures demonstrates that the in vitro environment lacks the presence of factors that can modulate the differentiation potential and proliferative activity of these tumor cells.
It is known that mesenchymal factors determining the degree to which tumor cells are sensitive to microenvironmental control are a function of the stage of neoplastic progression of the tumor cell.9, 13, 16, 32, 33 In addition, normal murine epidermal cells were able to suppress the growth of cells derived from papillomas but not the growth of a malignant variant of these cells after in vivo transplantation.17 However, to date, the phenotypic control of human epithelial tumors cells of different transformation stages resulting from their interactions with adjacent normal human epithelial cells has not been studied. Models to study early tumor progression have been limited by the nature of monolayer culture and heterotypic transplantation, both of which do not provide the proper tissue architecture in which tumorigenesis is initiated in vivo. The bioengineered human tissue models described here provide an environment in which tumor cell behavior and fate during early carcinogenesis can be followed in a stepwise fashion in a cellular context that maintains 3D tissue architecture before and after transplantation. In this way, the predictable progression from small communities of intraepithelial tumor cells in vitro to invasive tumors in vivo provides a biologically meaningful context to study the cellular and molecular pathways that regulate early developmental stages of squamous cell carcinomas.
In summary, it appears that the ability of intraepithelial tumor cells to undergo expansion in the context of a 3D network of normal cells is a function of the accumulation of genetic and phenotypic alterations required to override the inhibitory effect of the immediate cellular microenvironment and initiate tumorigenesis. This suggests that microenvironmental control of early tumor development is dominant only when tumor cells are responsive to the microenvironmental signals inherent in cell–cell interactions. Thus, initiation of tumorigenesis in stratified epithelium involves the acquisition of adequate autonomous growth capacity, loss of differentiation potential and alterations in cell–cell interactions that permit intraepithelial tumor cell expansion and lead to sufficient disruption of tissue architecture to allow escape from intraepithelial, microenvironmental control mechanisms.
We thank Drs. L. Taichman and S. Ghazizadeh for critical discussions; R. Harrington, T. Kolodka, L. Pfeiffer and L. Ning for technical assistance; J. Rheinwald for SCC13 cells; and R. Mulligan for the MFG vector. In addition, we thank Ms. C. Dowling, Ms. K. Henrickson, Mr. S. Suchit, Ms. Jennifer Landmann, and Ms. J. Neville for assistance in the preparation of illustrations.
- 25Human skin equivalent models of epidermal wound healing: tissue fabrication and biological implications. In: MaibachH, RoveeD, eds. Epidermal wound healing. Boca Raton: CRC Press, 2004. 3–23..
- 27Retroviral vectors. In: LeighIM, LaneEB, WattFM, eds. Keratinocyte methods. Cambridge: Cambridge University Press, 1994. 181–3..