Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Three-dimensional histoculture represents a useful alternative to monolayer culture because it preserves the native architecture of the cellular complexes while still allowing ease of experimental manipulation. Organ cocultures of tumor cells and normal tissues have been widely exploited in cancer research to study mechanisms of tumor invasion and drug sensitivity (reviewed in 1). We focused on a three-dimensional coculture system of human bronchial epithelium and lung carcinoma that may also provide a useful in vitro test system for diagnostic modalities in the early detection of bronchial neoplasm. This system also appears to be particularly suitable for studying the accumulation and kinetics of fluorescent dyes in tumors and in normal tissues for photodynamic diagnosis and therapy. For this purpose the cocultivation of tumor cells and well-differentiated normal epithelial tissue is a prerequisite. Other tissues, such as tissue from urinary bladder wall or endometrium, have been used to confront malignant cells with their tissue of origin (5, 6). This prompted us to develop and characterize an organotypic coculture system between human lung cancer cell lines and an organ culture from human bronchial tissue.

For creating cocultures, we used three-dimensionally precultured fragments of bronchial epithelium obtained as biopsies by routine fiberoptic bronchoscopy. As has been shown in previous studies, bronchial fragments regenerated a complete covering with epithelium during a short time of culture. The organotypic architecture of the epithelium as well as the ciliary beat were preserved to a remarkable degree (7, 8). Further, we used the human non-small cell lung cancer lines EPLC 32M1, LCLC 103H, and NCI H125 as malignant grafts. EPLC 32M1 was derived from a squamous-cell carcinoma (9). In a recent study, the EPLC 32M1 cells were found to have invasive properties in matrigel and collagen gels, used as models for basement membrane and interstitial space collagen (10). LCLC 103H was started from a pleural effusion of a tumor histologically classified at autopsy as large-cell carcinoma (9). NCI H125 was derived from a subcutaneous nodule obtained from a patient with adenocarcinoma of the lung before therapy.

In this study we discuss different types of confrontation between organ cultures and tumor cells and we describe the histology of cocultures and the kinetics of invasion. To illustrate its use for experiments, particularly those comparing normal epithelium with tumor, we cite an example from our work on 5-aminolevulinic acid (5-ALA)-induced protoporphyrin IX (PPIX) fluorescence in bronchial tissue using the cocultures. 5-ALA-induced PPIX has been successfully tested as fluorescent dye for early detection and therapy of carcinomas in different organs (11).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Organ Culture Preparation

Biopsies from normal bronchial tissue of patients were obtained by routine fiberoptic bronchoscopy. Tissue fragments with 0.5 to 1 mm maximum diameter were placed in agar-coated culture wells. The 0.75% agar was prepared using the following medium formulation: Dulbecco's modified Eagle's medium (DMEM) supplemented with 1,000 µg/ml glucose, four times the prescribed concentration of nonessential amino acids and 10% newborn calf serum, penicillin (10 U/ml), streptomycin (10 µg/ml), and amphotericin B (1 µg/ml) (all ingredients from Life Technologies GmbH, Eggenstein, Germany). Bronchial epithelial cell growth medium (BEGM) (Promocell, Heidelberg, Germany) supplemented with bovine pituitary extract, human epidermal growth factor, insulin, hydrocortisone, epinephrine, T₃, transferrin, retinoic acid, gentamicin, and amphotericin B was added and changed at 48-h intervals. Cultures were incubated at 37°C and 5% CO₂. Biopsies became completely epithelialized during the first 3 d of culture. The organ cultures consisted of an oligocellular stroma of connective tissue and a surrounding multilayered respiratory epithelium.

Preparation of Tumor Cells

All cell lines were growing continuously in RPMI 1640 medium (Seromed, Heidelberg, Germany), supplemented with 10% fetal calf serum, 2 mM L-glutamine, 24 U/ml penicillin, 24 µg/ml streptomycin, and 1.2 µg/ml amphotericin B (all from Life Technologies), in a humidified 5% CO₂ atmosphere.

Confrontation Experiments

To establish the preferential method for the production of cocultures, four types of confrontation between organ cultures and EPLC 32M1 cells were studied. A schematic presentation of experiments is given in Figure 1. In all experiments, organ cultures were used after 2 wk of precultivation: In group 1, intact organ cultures were put on top of the growing EPLC 32M1 monolayer in culture wells for 48 h, resulting in random contact between them. Organ cultures were then washed several times in BEGM to remove single cells, and transferred into new wells for further cultivation. In group 2, organ cultures were divided into two pieces with a sharp scalpel, wounding the epithelial surface and basement membrane and uncovering the connective tissues of the stroma at one side of each piece. Then they were confronted with the EPLC 32M1 monolayer, transferred, and cultured identically with group 1. In group 3, intact organ cultures were confronted on top of wet 0.75% agar side-by-side with highly concentrated EPLC 32M1 cell suspensions. The EPLC 32M1 cell suspensions (1 µl) were seeded next to the organ cultures just in contact with the epithelial surface using a microdispenser pipette (Drumond Scientific Co., Broomall, PA). After 1 h, BEGM and RPMI, mixed in a 4:1 ratio, were added. After 48 h (unless fixed earlier) organ cultures were washed and transferred into new wells. In group 4, after wounding organ cultures as described for group 2, cultures were confronted side-by-side with tumor-cell suspension, transferred, and cultured identically with group 3. Here, the EPLC 32M1 cell suspensions were seeded beside the organ culture in contact with the wounded surface (Figure 2A). During cocultivation a combination of both media was used (BEGM and RPMI mixed in a 4:1 ratio).

View larger version (22K): [in this window] [in a new window]   Figure 1.   Schematic presentation of confrontation experiments. Intact (1) or wounded (2) organ cultures were put on top of the tumor cell monolayer; intact (3) or wounded (4) organ cultures were confronted side-by-side with highly concentrated tumor cell suspensions.

View larger version (149K): [in this window] [in a new window]   Figure 2.   Phase-contrast and scanning electron microscopy of cocultures. (A) Phase-contrast micrograph showing the confrontation of the tumor cell suspension (T) and the wounded organ culture (arrow) on agar. (B) Phase-contrast micrograph of a living coculture after it has been transferred into a new well. The organ culture contains a layer of tumor cells that had remained attached to it at the site of contact (T). N indicates the normal epithelium. (C-E) Scanning electron microscopy of a 1-wk-old coculture. Tumor cells (T) formed pseudopodia (long arrows), extending to the neighboring cells of normal epithelium (N). The thick arrow indicates ciliated islands of normal epithelium. (F) Scanning electron microscopy of a coculture after 4 wk. Despite a long culture period, the normal epithelium remained well-differentiated. The photomicrograph shows cilia (C), goblet cells (G), and tumor cells (T).

Histology

Cocultures were terminated at 1, 2, 4, 8, 21, 30, and 45 d; fixed in 5% formalin; and embedded in paraffin. Each block was completely cut into 5-µm-thick serial tissue sections. Histologic sections were stained with hematoxylin and eosin (H&E).

Scanning Electron Microscopy

Samples were fixed for scanning electron microscopy after dehydration, and subjected to critical-point drying according to standard protocols. Samples were mounted on aluminum stubs and sputter-coated with gold. Tissue was examined at 15 kV in a Yeol scanning electron microscope (Yeol, Tokyo, Japan).

Incubation with 5-ALA

To study the kinetics of PPIX fluorescence in cocultures after exposure to 5-ALA, we used the following methods and materials. The coculture was incubated with a 0.03% 5-ALA solution (Sigma-Aldrich GmbH, Deisenhofen, Germany) at 37°C and 7.2 pH for 15 min. It was washed several times in DMEM. For fluorescence microscopy, the coculture-bearing chamber was positioned on an inverted fluorescence microscope (DMIRB; Leica, Munich, Germany) and monitored with 40-fold magnification. During the experiments, chamber temperature was kept at 37°C. For visualization of PPIX fluorescence, the coculture was illuminated for 2 s at a wavelength of 380 to 470 nm and fluorescence was detected above 470 nm. PPIX fluorescence was determined in defined areas in tumor and in normal epithelium using a CCD-camera (Stemmer Imaging, GmbH, Puchheim, Germany) and a computed image analyzing system at defined time points.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Confrontation and Attachment

Cocultures of all four confrontation types were terminated after 8 d and evaluated by light microscopy of H&E-stained serial tissue sections. In group 1, attachment of tumor cells to the epithelium of organ cultures was observed in 40% (n = 10) of confrontations. Tumor cells adhered to organ cultures in multiple locations. In group 2, organ cultures wounded before confrontation, attachment was observed in all experiments. Tumor cells adhered in multiple locations, preferentially to the wounded side (n = 10). In group 3, when EPLC 32M1 suspensions and unwounded organ cultures were cultivated side-by-side on agar, attachment was rarely observed (19%; n = 16). When medium was added, tumor and organ culture immediately separated. In contrast, when organ cultures were wounded before confrontation (group 4), tumor cells always adhered to the wounded side and attachment was never observed in other localities. During the 48 h of confrontation, tumor cells were found to adhere to the organ cultures next to them as well as to the underlying agar matrix. When organ cultures were transferred into new wells, the major part of the original tumor cell suspension remained on the agar matrix. The transferred organ cultures contained a layer of tumor cells that had remained attached to them at the site of contact. These propagated by further cultivation in a three-dimensional fashion.

Cocultures of groups 1 and 2 proved inappropriate for further studies because the location of grafting and the number of attached tumor-cell aggregates per organ culture were not predictable. Further, the diameter of the attached cell aggregates varied broadly. Cocultures of group 3 were also inappropriate because of the low incidence of attachment. Therefore we routinely used the type of confrontation described in group 4 for further investigations.

Interaction between Tumor Cells and Organ Cultures

As revealed by serial analysis of sections and by scanning electron microscopy, tumor cells adhered to wounded organ cultures at the site of contact but did not spread onto the epithelium. Wounded areas that were not occupied by tumor cells regenerated a new epithelial surface within 48 h. Cancer cells invaded the connective tissue individually and in cell aggregates, but they did not appear to cause damage to the neighboring bronchial epithelium. Infiltration took place parallel and perpendicular to the connective fibers. Invasion through the apical side of the intact epithelium was never observed. EPLC 32M1 cells extended pseudopodia to the neighboring cells of normal epithelium (Figures 2D and 2E). The cells grew progressively inside and outside the organ culture, EPLC 32M1 and NCI H125 without showing central necrosis even after 45 d, but LCLC 103H already with central necrosis after 8 d of culture. Histologic features of EPLC 32M1 cocultures were those of an undifferentiated carcinoma with little evidence of whirl formation and no keratinization (Figure 3). Pleomorphic cells with numerous mitotic figures and occasional giant cells were observed. The frontier with normal bronchial epithelium could be seen clearly. In short-term confrontations, invasion of stromal tissue was not clearly observed before 2 d of cocultivation with EPLC 32M1. The large-cell carcinoma LCLC 103H and the adenocarcinoma NCI H125 did not invade earlier than 8 d after confrontation. In long-term cultures (more than 4 wk), the extent of lateral progression corresponding to depth of invasion into stromal tissues increased together with the number of invading cells. Here, cancer cells massively invaded the connective tissue (Figure 3B). However, invasion of tissue remained limited. Even in cultures maintained over 45 d, infiltration of the whole organ culture was never observed. In most cultures, tumor cells reached no more than a third of the distance. Interestingly, in long-term cultures, evidence of mesenchymal stimulation by tumor cells (desmoplasia) could occasionally be observed. In close proximity to infiltrating cancer cells several fibroblasts and partially endothelial cells were found, whereas tumor-free parts in the same culture were largely acellular.

View larger version (117K): [in this window] [in a new window]   View larger version (126K): [in this window] [in a new window]   Figure 3.   Histologic sections of EPLC 32M1 cocultures of group 4 (tumor cells were confronted side-by-side with wounded organ cultures). (A) One-week-old coculture. Cancer cells invaded the tissue. Invasion through the intact epithelium was absent. The frontier with normal epithelium could be clearly seen. (B) Four-week-old coculture. Cancer cells massively invaded adjacent tissue. H&E staining. Bar = 100 µm.

In side-by-side confrontations of EPLC 32M1 cells with unwounded organ cultures (group 3), if attachment occurred, cancer cells seemed to destroy the underlying epithelium but invasion was absent, most likely because the EPLC 32M1 cells failed to invade the intact basement membrane (Figure 4). In group 1, invasion was observed in few cases although organ cultures were not wounded.

View larger version (107K): [in this window] [in a new window]   Figure 4.   Histologic section of a 2-wk-old coculture of group 1 (completely epithelialized organ cultures were put on top of the growing tumor cell monolayer). In cocultures of this type, tumor cells can be seen at multiple localities of the organ culture (arrowheads). Tumor cells did not invade the unwounded organ culture, most likely because the tumor cells failed to invade the intact basement membrane (long arrow). H&E staining. Bar = 100 µm.

To investigate whether invasion in our cultures could be easily quantified, the lateral infiltration distance was measured in EPLC 32M1 cocultures of group 4 with different incubation times (4, 8, and 30 d). The degree of lateral progression was scored arbitrarily by comparing the maximal distances found for each confrontation, from the upper level of the epithelium-tumor frontier to the most distant tumor cells found in stromal tissues in serial sections (Figure 5). This parameter is thus similar to the Breslow index for melanoma invasiveness (14), and has already been used successfully for quantification of tumor invasion in three-dimensional cocultures (5). The data are shown in Figure 6 and Table 1. After 4, 8, and 30 d of cocultivation, EPLC 32M1 cells were seen at depths of 71, 164, and 272 µm, respectively, implying a migration rate through the tissue of about 18 µm/d between the first and fourth days, 23 µm/d between the fourth and eighth days, and 5 µm/d between the eighth and 30th days. LCLC 103H and NCI H125 cocultures were also performed using the type 4 confrontation and were terminated after 4, 8, and 20 d. Both cell lines formed stable adhesion to organ cultures during confrontation on agar. The adhesion rates were 90 and 69%, respectively. In LCLC 103H cocultures, gigantic and pleomorphic cells with numerous mitotic figures and prominent nucleoli were found (Figure 7A). Desmoplasia could not be observed. NCI H125 cocultures showed poorly cohesive cells with prominent nucleoli (Figure 7B). In contrast to LCLC 103H, desmoplasia was frequently found in NCI H125 cocultures. Both cell lines were invasive after 8 d; however, the pattern of invasion differs obviously from that of EPLC 32M1 cells. LCLC 103H and NCI H125 cells showed occasional protrusions into the stroma but no deep, massive invasion or replacement of stromal tissue, and they did not cause significant damage to the organ culture. The mean lateral invasion distances of LCLC 103H and NCI H125 (100.7 ± 38.2 and 117.2 ± 38.9 µm, respectively) after 8 d of culture were clearly below those observed in EPLC 32M1 cultures (164.2 ± 20.9 µm). The data after 20 d of cocultivation with LCLC 103H and NCI H125 are shown in Figure 8.

View larger version (48K): [in this window] [in a new window]   Figure 5.   Scheme for determination of the lateral invasion distance in cocultures. The arrow measures the distance of interest. One section of a series of the completely cut coculture is shown. The distance was measured for all sections and the maximum was defined as relevant one (T: invading tumor; Ep: epithelium; OC: organ culture).

View larger version (16K): [in this window] [in a new window]   Figure 6.   Lateral infiltration distance in EPLC 32M1 cocultures. Comparison between 4, 8, and 30 d of culture. Tumor cells were seen at depths of 71, 164, and 272 µm, respectively, implying a migration rate through the tissue of about 18 µm/d between the first and fourth days, 23 µm/d between the fourth and eighth days, and 5 µm/d between the eighth and 30th days (t test, P < 0.005).

View larger version (142K): [in this window] [in a new window]   View larger version (138K): [in this window] [in a new window]   Figure 7.   Histologic sections of LCLC 103H and NCI H125 cocultures. (A) LCLC 103H coculture, 20 d old, showing gigantic and pleomorphic cells with numerous mitotic figures and prominent nucleoli invading into the stromal tissue. Bar = 100 µm. (B) NCI H125 coculture, 20 d old, showing poorly cohesive cancer cells with prominent nucleoli (arrowheads). H&E staining. Bar = 50 µm.

View larger version (11K): [in this window] [in a new window]   Figure 8.   Lateral infiltration distances in NCI H125 and LCLC 103H cocultures after 20 d of culture. The mean values ± standard deviation were 131.8 ± 52.9 µm for NCI H125 and 168.6 ± 75.6 µm for LCLC 103H. No significant difference in the invasion depth was found.

5-ALA-Induced PPIX Fluorescence in Cocultures

PPIX fluorescence could be detected in tumor and normal epithelium of investigated cocultures. An example is shown in Figure 9a. The experiments showed marked differences in kinetics and intensity of PPIX fluorescence in normal tissue and tumor, as shown in Figure 9C.

View larger version (68K): [in this window] [in a new window]   View larger version (54K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Figure 9.   5-ALA induced PPIX fluorescence in cocultures. (A) Fluorescence micrograph of the surface of a living coculture after 2 h of incubation with 5-ALA. PPIX red fluorescence could be seen in tumor (thick arrow) and in normal epithelium (long arrow). (B) A scheme of the living 5-ALA incubated coculture shown in A. PPIX red fluorescence is represented by black-colored areas. The arrows indicate different histologic tissue types. Open arrow, tumor; closed arrow, bronchial epithelium. (C) Microscopic fluorescence levels in tumor and in normal epithelium of a coculture as a function of time after incubation with 5-ALA.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The aim of the present study was to establish and characterize a culture system for in vitro investigations of bronchial cancer that is devised to approach the in vivo environment more closely. We used organ cultures of bronchial epithelium as host tissue because they proved to be reliable in many aspects: (1) they mimic the human bronchial epithelium to a remarkable degree and can be maintained in culture for reasonable periods of time without loss of structure or cell viability (7, 8); (2) they provide a three- dimensional tissue structure containing epithelium and stromal tissue (which tumors possess in vivo); and finally, (3) the stroma consists mostly of connective tissue fibers with few cellular components such as fibroblasts and endothelial cells that are histologically identifiable and thus discernible by routine H&E staining from tumor cells.

In our model, wounding the organ cultures before confrontation with cancer cells made attachment more frequent and permitted invasion. The observations correlate well with previous studies of invasion in other tissues, in which connective tissues appeared very sensitive to malignant invasion (14, 15). The absence of invasion of malignant cells into the apical zone of intact epithelia in vitro and in vivo has also been described (15). However, in some coculture assays malignant cell spheroids progressively invaded and replaced the epithelia from the apical side; for instance, the epithelia of the bladder and the endometrium (5, 6). In confrontations with unwounded organ cultures, cancer cells destroyed the epithelium but failed to invade through the basement membrane into the stroma (Figure 4). Therefore, we infer that the lack of invasion into the apical zone of epithelia in our model is most likely due to the intact basement membrane. Other obstacles of invasion could have been the apical cytoskeleton of epithelium or the complex of intercellular junctions that may have remained intact in spite of a partial damaging caused by tumor cells.

The cocultures of group 4, in which tumor cells were confronted side-by-side with wounded organ cultures, proved to be more appropriate for further investigations because of at least three aspects: (1) The cells showed a high incidence of attachment. (2) Attachment and invasion occurred at a single and predictable locality. The polarity of the coculture made the tumor discernible from the normal epithelium by light microscopy and facilitated fluorescence measurement on living cultures. (3) The tumor cells began the infiltrative process at an easily recognizable starting line. This considerably facilitated the measurement of infiltration. However, the cocultures described in group 4 are not suitable for studying early stages of invasion because they lack a basement membrane at the site of contact with tumor cells. Nevertheless, tumor progression into the stromal tissue that could be easily quantified may be a useful parameter for in vitro study of the effects of some therapeutic modalities, such as radiation and photodynamic therapy.

Serial section analysis and measurements of the lateral infiltration distance revealed that invasion in our model can be easily quantified. The results are reproducible, and a time-dependent progression of tumor cells can be seen clearly. However, we had to exclude about 10% of cultures from this analysis because of high deviations from the horizontal level during sectioning.

This organ coculture model shows marked similarities to the situation in vivo. It contains human bronchial epithelium, connective tissues, and human lung-cancer cells that are growing and infiltrating the normal tissue in a three-dimensional manner. In addition, the proximity of this model to in vivo cancer was suggested by other observations during the study of histologic sections: for instance, the phenomenon of mesenchymal stimulation induced by EPLC 32M1 and NCI H125 cells. This fibrotic reaction with proliferation of connective tissue as collagen types I and III and fibronectin, termed desmoplasia, has been observed in different human tumorsespecially in infiltrating breast carcinomas and in lung cancer (18) and it has been considered a sign of higher malignancy (19). Further, the invasion rate in this model is more comparable with that in human malignancy. In other coculture models using chorioallantoic membranes, heart fragments, or synthetic extracellular matrix substances, deep infiltration was observed after 1 to 2 d of incubation, indicating a much higher invasion rate in comparison to in vivo malignancy (10, 20, 21). Although all cell lines used were originally obtained from malignant and metastasized bronchial tumors, there were remarkable differences in adhesion rates, degree of damage caused by infiltration, and the pattern of invasion between the cell lines. The invasion distances of LCLC 103H and NCI H125 cells were clearly below those of EPLC 32M1 cells, indicating a higher malignant potential of the latter. The cocultures could therefore provide a model for studying the factors that enable tumor cells to degrade extracellular proteins and invade through stromal tissues.

The usefulness of our cultures for studies of fluorescent tumor markers, particularly of 5-ALA-induced PPIX, can be confirmed by the first experiments we performed. The results were highly reproducible and, as shown in Figure 9, there were marked differences in the kinetics and the intensity of PPIX fluorescence in tumor and normal epithelium after incubation with 5-ALA. Such differences could be of great interest for the clinical application of fluorescent dyes in the diagnosis of and therapy for bronchial cancer. The cocultures are suitable for measuring the relative fluorescence intensity of a fluorescent marker in tumor and normal epithelium as a function of the time after incubation, and they permit us to determine the best ratio of tumor fluorescence to normal epithelium fluorescence. However, for measurements of fluorescence distribution at a cellular level, immediately frozen sections or confocal laser scanning microscopy should be performed. The results shown here should only demonstrate the usefulness of the coculture model for studying the kinetics and distribution of fluorescent dyes in tumor and normal epithelium. Experiments are in progress to find the best suitable ratio of tumor fluorescence to normal epithelium fluorescence as a function of the time after incubation and the concentration of the fluorescent dye.