SEARCH

SEARCH BY CITATION

Keywords:

  • Tumor microenvironment;
  • stroma;
  • fibroblasts;
  • cancer-associated fibroblasts

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  8. Supporting Information

Normal human and murine fibroblasts can inhibit proliferation of tumor cells when co-cultured in vitro. The inhibitory capacity varies depending on the donor and the site of origin of the fibroblast. It requires direct cell-to-cell contact and is not transferable with supernatant. Here, we show that effective inhibition also requires the formation of a morphologically intact fibroblast monolayer before the seeding of the tumor cells. Interference with the formation of the monolayer impairs the inhibition. Subclones of TERT-immortalized fibroblasts were selected on the basis of differences in the growth pattern and related inhibitory activity. Whereas the well-organized “whirly” (WH) growth pattern was associated with strong inhibition, the disorganized “crossy” (CR) growth pattern was linked to reduced inhibition. Time lapse imaging of tumor-fibroblast co-cultures using extended field live cell microscopy revealed that fibroblast monolayers with growth inhibitory capacity also reduced the motility of the tumor cells whereas noninhibitory monolayers had no effect on tumor cell motility. Gene expression pattern of two isogenic pairs of fibroblasts, WH and CR subclones of the TERT immortalized line (inhibitory, and less inhibitory subsequently) and freshly explanted skin (inhibitory) and hernia (noninhibitory) fibroblasts derived from the same patient, identified a set of genes that co-segregated with the inhibitory phenotype. This suggests that our model system may reveal molecular mechanisms involved in contact-mediated microenvironmental surveillance that may protect the organism from the outgrowth of disseminated tumor cells.

Microenvironmental control is emerging as a major restraining force that can prevent the growth of frankly malignant cells.1–4 It can also nip the outgrowth of disseminated cancer cells in the bud. It may be partially or fully responsible for Peto's paradox, that is, the fact that large mammalians do not have more tumors than small rodents, in spite of the high degree of conservation of oncogenes, tumor suppressor genes and the signal transduction pathways that control cell proliferation.5, 6 Traditionally, this restraint has been attributed to immunological mechanisms. But while they can be involved, particularly in the case of virus induced or virus associated tumors, it is becoming increasingly clear that nonimmune controls play a major role in the protection against nonviral tumors, regarded as “self” by the immune system. It has been shown that reestablishment of normal tissue architecture can override the malignant phenotype.2, 7 Reconstruction of polarity and/or acinar structure in epithelial tumors or induction of differentiation in hematopoietic or germ cell derived tumors may prevent or reverse malignant growth.8–10 In contrast, disruption of the normal tissue equilibrium by inflammation, regeneration or wound healing may promote the growth of potentially or frankly neoplastic cells.7, 11

How is microenvironmental control mediated? This is only incompletely known, but it is obviously multicomponental. Among its known effectors, fibroblasts occupy an important place. Their ability to inhibit tumor cell growth by direct contact, originally discovered by Michael Stoker in the 1960s, has been well-documented.12, 13 On confrontation with tumor cells, they may lose their inhibitory activity, however. Cancer associated fibroblasts can have the opposite effect, as demonstrated by both in vitro and in vivo experiments.14, 15

To assess some parameters associated with the inhibitory effect of fibroblasts on tumor cell growth, we have adopted a high throughput 384 microwell system where color-labeled tumor cells are confronted with fibroblasts.16 In a first study of 107 fibroblast cultures, we have found that the sensitivity of six different tumor lines to the inhibitory effect of fibroblasts differs.17 The efficiency of the fibroblasts differed as well. Skin fibroblasts were more inhibitory than fibroblasts derived from hernias or from cancerous prostates. We also showed that the inhibitory effect required direct contact between the fibroblasts and the tumor cells.

In this study, we show that inhibition was dependent on the architecture (age and confluency) of the fibroblast monolayers. We have also found that inhibitory and noninhibitory fibroblasts differ with regard to their growth characteristics in vitro. Inhibition was associated with a growth pattern designated as “whirly” (WH) while reduced inhibition was linked to a “crossy” (CR) pattern. We show that these differences prevail in two fundamentally different isogenic fibroblast pairs: the inhibitory (WH) and the less inhibitory (CR) subclones of the telomerase immortalized, BjhTERT foreskin fibroblast line as one pair, and the inhibitory (skin) and the noninhibitory (hernia) ex vivo fibroblasts, derived from the same donor, as the other pair. Gene profiling showed concordant expression patterns for the two inhibitory and contrasting expression patterns for the two noninhibitory cell types. Altogether 1,033 genes showed such concordance, permitting the identification of up and down-regulated genes in the inhibitory versus the noninhibitory fibroblasts.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  8. Supporting Information

Patient samples

Collection of patient samples was performed according recommendations of Regional Ethical Review Board. Permission from the Regional and Institutional Ethics Committee of the University of Debrecen [DEOEC RKEB/IKEB No. 2918–2009 (UD MHSC REC/IEC No. 2918–2009)] for skin and hernia samples from normal patient 1. Permission from Regional Ethical Review Board, Uppsala/Örebro (2009/017) for prostate samples of patient 1.

Cell lines

Cancer cell line.

The prostate carcinoma cell line PC-3 (from bone metastasis) was used as the tumor target in our functional assays. To be able to distinguish tumor cells from unlabelled fibroblasts in co-colture, PC-3 cells were transfected with recombinant histone H2A-red fluorescein protein (H2AmRFP) carrying plasmid. PC-3 cells were cultured in Iscove's modified Dulbecco's medium (IMDM) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and penicillin, streptomycin and gentamicin at 100, 100 and 50 μg/ml concentration, respectively (PSG additive).

Primary human fibroblasts.

Primary human fibroblasts were cultivated from diagnostic biopsy samples or from surgically removed tissue pieces from different anatomic locations of pediatric and adult patients. The primary skin fibroblast used in this study was isolated from a pediatric patient wound site skin (Normal patient 1 skin). From the same pediatric patient, fibroblasts were also obtained from material resected during umbilicalis hernia surgery (Normal patient 1 hernia). The prostate fibroblasts were cultivated from a prostate biopsy received from an adult patient diagnosed with prostatic carcinoma (Cancer patient 1 prostate).

Primary cultures were established from mechanically dissociated tissue pieces, in 6-cm tissue culture dishes, cultured in IMDM supplemented with 20% FBS and PSG. Fibroblasts were cultured in IMDM or in IMFG [IMDM + fibroblast growth medium with fibroblast growth supplements (both from 3H Biomedical, Uppsala, Sweden), in 4:1 ratio)], always supplemented with 10% FBS and PSG (figure legends). The fibroblasts were passaged after trypsinization upon near confluency. Multiple vials were frozen after the early passages for repeated experiments. Primary fibroblasts of passage number p < 15 were included in the tumor/fibroblast co-culture assays.

Established fibroblast line.

In addition to primary fibroblasts, we have also included a recombinant telomerase-transfected immortalized human fibroblast line (BjhTERT). It has been generated by introducing a human telomerase encoding vector (MPSV-hTERT) into a human foreskin fibroblast.18 The line is sensitive to contact inhibition and shows no signs of transformation. BjhTERT cells were cultured in IMDM or IMFG supplemented with 10% FBS and PSG.

Co-culture assay

Tumor cell proliferation on fibroblast monolayers was analyzed in 384-well plates. Fibroblasts were plated in 30–100 μl cell culture medium and cultured for 2–8 days to form confluent monolayers. Regular staining with Hoechst 33258 was used to check the absence of mycoplasma contamination. Two hundred (unless otherwise stated in the results section) H2AmRFP labeled PC-3 tumor cells were plated in 50–100 μl cell culture medium IMDM, 10% FBS, PSG) on top of the fibroblast monolayers. The control wells contained labeled tumor cells without fibroblasts. The “Control alone” wells contained 200 labeled tumor cells alone. The “Control mix” wells contained 200 labeled + 400 unlabeled tumor cells.

Automated microscopy

Every well of the 384-well plate was imaged using a modified version of the automated microscope system previously developed by our group.16, 17 In this study, we used the Openlab automation Platefocus 10, developed by us. Images at 2.5× magnification (NA 0.08), covering the entire bottom-area of a well, were captured after seeding of tumor cells (day 0 or day 1) and after 2, 4, 5 or 6 days of co-incubation with fibroblasts. At each time-point both transmitted light and fluorescence images were captured (excitation at 560 nm and emission at 600–620 nm for mRFP labeled cancer cells). The microscope platform was built using a Nikon microscope (Nikon, Tokyo, Japan), a programmable XY-table (Märzhauser, Germany) and a Retiga-4000RV (QImaging, Surrey, BC, Canada). Using binning 2, raw images were 1,024 × 1,024 pixels in size. The system was run on a Mac OS X, version 10.5.6, processor 2 × 3 GHz Quad-Core Intel Xeon.

Image analysis and quantification

Quantification of tumor cell numbers was done at the single cell level, using the find maxima algorithm in ImageJ (NIH, Bethseda, USA). For optimal quantitation of the red-labeled nuclei of the tumor cells, all images were identically processed for quality enhancement using rolling ball background subtraction and 3 × 3 or 5 × 5 median filtering (Image J). The proliferation ratio was calculated by dividing the number of tumor cells on day 4, 5 or 6 with the number tumor cells on day 0 or day 1, and presented as the mean of measurements in at least four individual wells. Controls were grown in at least six wells. All results are presented together with the standard error of the mean (SEM). Image analysis was run on a Mac OS X, version 10.5.6, processor 2 × 3 GHz Quad-Core Intel Xeon with RAM memory 8 GB.

Isolation of subclones of BjhTERT fibroblasts

In two separate experiments, WH and CR subclones were obtained by plating 1,000 BjhTERT cells in a 10 cm tissue culture dish in 10 ml of IMDM medium (10% FBS, PSG). A few colonies started to appear after 2–3 weeks. After 4 weeks, colonies were checked under light microscope and 30 colonies were removed from the dish by the tip of a pipette and cultured further. Out of 30 subclones, we separated four subclones, where two subclones showed the WH and two subclones showed the CR phenotype, respectively. They maintained their phenotypic difference during serial passages.

Obstruction of the fibroblast monolayer

PC-3mRFP tumor cells were seeded, at day 0, into wells where 2,000 inhibitory skin fibroblasts (Normal patient 1) were grown in three different set-ups (a, b and c):

  • a.
    Skin fibroblast cultured alone for three days (from day 3). Confluent monolayer established.
  • b.
    Skin fibroblast cultured alone for one day (from day 3). 500 irradiated unlabeled PC-3 tumor cells were seeded on top of the naïve monolayer (at day 2).
  • c.
    2,500 irradiated PC-3 tumor cells were seeded on the bottom of the well (day 4). Fibroblasts were seeded on top of the irradiated cancer cell layer (day 3).

Extended field live cell movie

Fibroblasts were seeded on round cover slips (30 × 0.17 mm) in a six-well plate; 50,000 BjhTERT fibroblasts were grown for 7 days and 100,000 hernia fibroblasts (Normal patient 1) were grown for 6 days. When monolayers were of comparable confluency, 30,000 PC-3mRFP tumor cells were seeded on top of each monolayer. After 2 hr, when tumor cells attached to the fibroblast monolayer, the cover slip was removed and inserted into a closed poc-mini chamber system.

The proliferation of the tumor cells was followed for 62.5 hr, with images captured every 15 min. For each time-point in the movie a field of 49 images, covering a total area of 4.2 × 5.2 mm2 (22 mm2), were captured using 10× magnification (Ph1 NA 0.25). The total number of tumor cells followed in each area was initially 950 and 1,250 for the BjhTERT and hernia set-up, respectively.

The movie was captured using a program for multifield/extended field capture (Multifield 10×), developed by us using Openlab Automator (Perkin Elmer, Upplands Vasby, Sweden).

RNA-profiling

RNA-profiling was done using Illumina HiSeq (Illumina, San Diego, CA, USA) technology according to the manufactures instruction.19

RESULTS

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  8. Supporting Information

Growth pattern of tumor cells growing on fibroblast monolayers

Our model is based on proliferation measurements of red fluorescent mRFP histone H2A labeled PC3 prostate carcinoma cells (called tumor cells in the text) seeded on top of unlabeled fibroblast monolayers. In the inhibited cultures, the tumor cells grew in small confined islands on either BjhTERT fibroblasts (Fig. 1a) or on normal patient 1 skin fibroblasts (Fig. 1b). In the non-inhibited cultures, the tumor cells grew in bigger islands and showed a more dispersed pattern on either cancer patient prostate 2 fibroblasts (Fig. 1c) or on normal patient 1 hernia fibroblasts (Fig. 1d). A morphological difference between inhibitory and noninhibitory fibroblasts could also be observed in fibroblasts cultivated alone. The inhibitory fibroblasts appeared to grow in an organized WH pattern, while less inhibitory fibroblasts tended to arrange themselves in a CR pattern. Our co-culture data on tumor cells growing on different fibroblast monolayers thus suggested that the growth pattern of the tumor cells reflect the architecture of the underlying fibroblasts.

thumbnail image

Figure 1. Growth pattern of tumor cells and inhibitory capacity of fibroblast monolayers. Fluorescence microscopy images of PC3 mRFP tumor cells, growing on 4 days old fibroblast monolayers. Images show tumor cells after 4–5 days of co-culture with fibroblasts. (a) PC3 mRFP tumor cells growing on an inhibitory fibroblast monolayer (BjhTERT fibroblasts). (b) PC3 mRFP tumor cells growing on an inhibitory fibroblast monolayer (Normal patient 1 skin). (c) PC3 mRFP tumor cells growing on noninhibitory fibroblast monolayer (Normal patient 2 prostate). (d) PC3 mRFP tumor cells on a noninhibitory fibroblast monolayer (Normal patient 1 hernia). (e) Skin fibroblasts (Normal patient 1) were plated at passage number p7, p8 and p14. Data show proliferation ratio of 200 PC3 mRFP cells after 5 days co-culture with fibroblasts (day 5/day 0). Data are represented as mean ± SEM). (f) Three thousand fibroblasts, derived from patients with (Cancer patient 2 prostate) and without cancer (Normal patient 1 skin and hernia), were seeded on day 5 to form a confluent monolayer. At day 0, PC3 mRFP tumor cells (50, 200, 400 and 800 cells) were seeded on top of the fibroblast monolayers. Data show proliferation ratio of PC3 mRFP cells after 5 days co-culture with fibroblasts (day 5/day 0). Data are represented as mean ± SEM).

Download figure to PowerPoint

The inhibitory effect is independent of fibroblast passage number and of tumor cell number

Here, we show that the differences in the inhibitory capacity of normal skin derived fibroblasts, as contrasted to noninhibitory prostate and hernia derived fibroblasts are independent of the fibroblast passage number (passages 7, 8 and 14; Fig. 1e). Moreover, the inhibitory and noninhibitory effects are highly reproducible and do not depend on the initial number of tumor cells seeded (range 50–800 tumor cells; Fig. 1f). Because of saturation of tumor cells the differences in the co-cultures with originally 800 tumor cells were diminished by the end of the experiment. The y-axis in the figure shows the relative proportion of the tumor cell number on day 5/day 0. Both controls with originally 800 labeled tumor cells clearly show diminished expansion of tumor cells alone (around eightfolds compared to 12-folds of 200 tumor cells originally), illustrating the saturated condition.

Phenotype and the Inhibitory effect of BjhTERT subclones

Departing from the WH versus CR growth patterns of the inhibitory versus noninhibitory ex-vivo fibroblasts, subclones of the telomerase immortalized fibroblast line were separated on the basis of corresponding differences in their growth pattern (experimental procedures). Out of 30 subclones, four representative clones were chosen that showed the WH (Figs. 2a and 2c) and the CR phenotype (CR), respectively (Figs. 2b and 2d). Both clones were inhibitory, but the CR subclone was significantly less inhibitory than the WH subclone and the parental non-separated hTERT fibroblasts (Fig. 2e).

thumbnail image

Figure 2. Phenotypes of subclones of BjhTERT fibroblasts and their inhibitory capacity on tumor cell proliferation. (ad) Two morphologically different fibroblast subclones were isolated from BjhTERT fibroblasts by plating a low number of fibroblasts (1 × 103 cells in 10 cm tissue culture dish). Subclones appearing after 3–4 weeks were selected based on their WH and CR morphology and further isolated and expanded to obtain either WH (a and c) or CR (b and d) fibroblast monolayers. Transmitted microscopy images were taken at 10× (a and b) and 20× (c and d) magnification. (e) The inhibitory capacity of WH and CR fibroblast monolayers (4–5 days old) was tested in co-culture with the PC3 mRFP tumor cells. Data show proliferation ratio of PC3 mRFP cells after 4–6 days co-culture with fibroblasts. Data are represented as mean ± SEM of three separate experiments. Each individual column represents co-culture ratios from 41–58 separate wells.

Download figure to PowerPoint

The inhibitory effect of the fibroblasts increases with the age of the monolayer

The age of the fibroblast monolayer and corresponding level of confluency influenced the inhibitory capacity of the fibroblasts. To define this effect, we compared fibroblast monolayers of five different ages. Two thousand BjhTERT fibroblasts were seeded on days 8, 6, 4, 2 and 0. On day 0, 200–400 tumor cells were seeded on top of each monolayer. Fibroblasts cultured for 2 or less than 2 days were named short-term confluency cultures (STCC). Fibroblasts cultured for more than 2 days were named long term confluency cultures. The inhibitory effect was found to increase with the age of the monolayer (Figs. 3a3c). In STCC, where the fibroblasts were in the incipient stage of monolayer formation, tumor cell proliferation was not effectively inhibited, even though the defined WH growth pattern of the tumor cell islands, typical for the inhibitory effect, was visible in the 2 day old STCC (Fig. 3d). At day 0, when the fibroblasts were seeded immediately before the tumor cells they showed no inhibitory effect and the growth pattern of the tumor cells was identical to the controls, where the tumor cells grew without fibroblasts (Fig. 3d).

thumbnail image

Figure 3. The inhibitory effect of inhibitory fibroblast (BjhTERT) monolayer increases with the age of the monolayer, however, interference with the architecture of another inhibitory fibroblast (normal pat1 skin) monolayer impairs fibroblast mediated inhibition. Proliferation ratio data and fluorescence microscopy images of PC-3 mRFP tumor cells co-cultured with fibroblast monolayers of different age. Two thousand BjhTERT fibroblasts were seeded on days 8, 6, 4, 2 and 0. On day 0, 200–400 PC3 mRFP tumor cells were seeded on top the monolayers of different age. (ac) Data show proliferation ratio of PC3 mRFP cells after 2 (a), 4 (b) and 6 (c) days of co-culture with fibroblasts. Data show the average proliferation ratio of four individual experiments (based on the average measurements of four wells in each of the four experiment plates; all plates were collagen coated). Data are represented as mean ± SEM. (d) Representative fluorescence microscopy images of PC-3 mRFP tumor cells seeded on top of fibroblast monolayers of five different ages. Images were captured after 4 days co-culture (×2.5). (e) PC3 mRFP tumor cells were seeded, at day 0, into wells where 2,000 inhibitory skin fibroblasts (Normal patient 1) had been grown in three different set-ups (A, B and C): (A) Skin fibroblasts were cultured alone for three days (from day 3). Confluent monolayer established. (B) Skin fibroblasts were cultured alone for 1 day (from day 3). 500 irradiated PC3 tumor cells were added on top of the naïve monolayer (at day 2). (C) 2,500 irradiated PC3 tumor cells were seeded on the bottom of the well (day 4). Fibroblasts were then seeded on top of the irradiated tumor cell layer (day 3). Data show proliferation ratio of PC3 mRFP cells day 5/day 1 and are represented as mean ± SEM.

Download figure to PowerPoint

Interference with monolayer architecture impairs fibroblast mediated inhibition

In Figures 3a3d, we show that confluent monolayers of fibroblasts are required for effective inhibition of tumor cell proliferation. To study this dependence, we have experimentally interfered with the monolayer formation of the inhibitory skin fibroblasts (normal patient 1). Irradiated (2,000 rad) unlabeled tumor cells were plated at the bottom of the well, one day before the seeding of the fibroblasts, preventing the fibroblasts from forming an organized monolayer. Three days after the seeding of the fibroblasts, 200 labeled tumor cells were seeded on top of fibroblasts alone (Fig. 3e, column A), on top of fibroblasts with irradiated tumor cells seeded on top of their monolayer (Fig. 3e, column B) and on top of the fibroblasts that were hindered to make a confluent monolayer by irradiated tumor cells on the bottom (Fig. 3e, column C). The inhibitory effect was reduced when irradiated nonlabeled tumor cells were plated on the bottom of the well (Fig. 3e, column C), in comparison with unmanipulated monolayers (Fig. 3e, column A). Irradiated nonlabeled tumor cells on the top of an already formed monolayer did not reduce the inhibitory effect (Fig. 3e, column B).

Extended field live cell microscopy reveals differences in tumor cell mobility

Using time lapse imaging combined with extended field live cell microscopy, differences were found in the motility of the tumor cells growing on the top of confluent inhibitory or noninhibitory fibroblast monolayers.

A BjhTERT fibroblast monolayer, with strong inhibitory effect on tumor cell proliferation (Figs. 3a3d), was found to reduce the motility of the tumor cells during a 62.5 hr long filming session (Fig. 4b). The monolayer from hernia fibroblasts (normal patient 1) had no effect on tumor cell motility (Fig. 4b) and allowed continuous tumor cell proliferation throughout the time of filming (Fig. 4a).

thumbnail image

Figure 4. Extended field live cell microscopy. Proliferation and trajectories of tumor cells growing on top of inhibitory (BjhTERT) and noninhibitory (Normal patient 1 hernia) fibroblast monolayers. (a) Proliferation of PC-3mRFP tumor cells during 62.5 hr (250 time-points). (b) Motility of PC-3mRFP tumor cells [length/area of cell trajectories normalized for average number of cells in 12.5 hr intervals (50 time-points)]. (c) Trajectories of PC-3mRFP tumor cells during 12.5 hr intervals. Color-coded images show 50 time-points projection of the red-labeled tumor cells: yellow (0–12.5 hr), green (12.5–25 hr), magenta (25–37.5 hr), blue (37.5–50 hr) and red (50–62.5 hr). (d) A max projection of all five color-coded images shows the total motility (full trajectories) of the PC-3mRFP tumor cells during 0–62.5 hr. See also Supporting Information Figure S1 and Movie S1.

Download figure to PowerPoint

Color-coded trajectories of tumor cells during 12.5 hr intervals showed that the motility of tumor cells on top of the BjhTERT monolayer started to decrease after 25 hr, in the third interval (magenta). In the fifth and last interval (red), the difference in motility between tumor cells on the two types of fibroblast monolayers were striking. The tumor cells growing on the inhibitory BjhTERT monolayer have almost stopped their movement, whereas the tumor cells on the noninhibiting hernia fibroblast monolayer were as motile as at the beginning of the film [Fig. 4c (1–5)]. Projections of the five color-coded trajectories in C, summarizes the motility of the tumor cells over 62.5 hr (Fig. 4d). See also Supporting Information Figure S1 and Movie S1. Supporting Information Figure S1 shows tumor cell motility on different fibroblast monolayers in the first and the last 50 time-points in the movie, illustrating comparable confluency of the two fibroblast monolayers (in phase contrast) at the time of the seeding of the red labeled tumor cells (a and d). Supporting Information Movie S1 shows the time-lapse movie of the first 50 time-points (0–12.5 hr) and the last 50 time-points (50–62.5 hr) of the co-culture experiment.5

thumbnail image

Figure 5. RNA profiling of inhibitory and noninhibitory fibroblasts. Heat-map showing the relative RNA expression levels of 1,033 selected genes. 528 genes were upregulated in the inhibitory fibroblasts and 505 genes were upregulated in the less or noninhibitory fibroblasts (red: high expression, green: intermediate expression and blue: low expression). See also Supporting Information Table S1.

Download figure to PowerPoint

Gene expression in isogenic pairs of inhibitory and noninhibitory fibroblasts

To identify genes that may influence the growth inhibitory phenotype, we have analyzed two isogenic pairs of fibroblasts. The WH (inhibitory) and the CR (less inhibitory) subclones of BjhTERT represented the in vitro derived pair. Fibroblasts isolated from the skin (inhibitory) and from loose connective hernia tissue (non-inhibitory) of the normal patient 1 represented the ex vivo pair. As shown above, the WH BjhTERT subclone and the skin fibroblasts from the normal patient 1 strongly inhibited the proliferation of the tumor cells whereas their isogenic partners BjhTERT CR and the hernia derived fibroblasts were less inhibitory (Figs. 1a1f, 2e, and 4)

For gene profiling, total RNA from confluent cultures of the two fibroblast pairs were analyzed using the Illumina HiSeq technology, identifying 53,196 expressed mRNAs and noncoding regulatory RNAs.19 To identify the differentially expressed transcripts in relation to the inhibitory phenotype, we selected the genes whose expression profile showed the following features:

Genes consistently up-regulated in both types of the inhibitory fibroblasts (i) or in both types of the less or noninhibitory fibroblasts (ii) were identified by the following formulas:

  • (i)
    • equation image
  • (ii)
    • equation image

where the array-based expression values of the 53,196 genes were:

F1 for BjhTERT WH, F2 for BjhTERT CR, F5 for Normal patient 1 hernia, F6 for Normal patient 1 skin.

The above formulas identified 528 genes upregulated in the inhibitory fibroblasts and 505 genes upregulated in the less or noninhibitory fibroblasts, representing approximately 1% of all genes analyzed. The heat-map of relative expression levels of the 1,033 genes are shown in Figure 6 where the genes were sorted according to their expression levels in the WH BjhTERT fibroblasts. By and large, there was good congruency of the expression levels among the selected 1% of the transcriptome of the two fibroblast pairs. The asymmetric color distribution in the heat map also shows that there was more specifically upregulated genes in the inhibitory fibroblasts, compared to the less or noninhibitory fibroblasts (red: high expression, green: intermediate expression and blue: low expression). Among the 1,033 genes, 693 had sufficiently detailed annotations. The DAVID functional annotation system (http://david.abcc.ncifcrf.gov/) revealed that at least one third of the differentially expressed genes coded for membrane associated and/or secreted proteins. Also, numerous genes coded for nuclear proteins, for cancer development, for embryonic differentiation programs, for wound healing and for inflammation (Table 1). The list of all 1,033 individual gene transcripts with their FKPM values is provided as a supplementation data in Supporting Information Table S1.

Table 1. Functional groups of genes, congruently up-regulated in inhibitory or non inhibitory fibroblasts
inline image

DISCUSSION

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  8. Supporting Information

The presented data show, in confirmation of our previous findings, that monolayers derived form fibroblasts of normal skin strongly inhibit the proliferation of tumor cells in co-culture.17 Fibroblasts derived from hernias or from cancerous prostates had less or no inhibitory effect on tumor cell proliferation. Inhibition was dependent on the formation of confluent fibroblast monolayers but independent of the initial number of tumor cells. The inhibitory capacity of the fibroblasts was stable over several passages. We have also shown that the fibroblast effect can be conveniently studied in a well-defined model system, where red fluorescent mRFP histone H2A labeled PC-3 prostate carcinoma cells were seeded on the monolayer of telomerase immortalized human BjhTERT fibroblasts.

Visual observation of mixed cell cultures indicated a correlation between the fibroblast growth pattern and the ability of the tumor cells to grow on top of them. The inhibitory fibroblasts cultures showed a geometrically organized WH growth pattern where the tumor cells were only able to grow out from foci, located at the intersections of the whirls. Subsequently, proliferation occurred mainly inside well circumscribed tumor cell nests, sparing the areas of monomorphic whirls. In contrast, less or noninhibitory fibroblast monolayers were often chaotic and disorganized. The fibroblasts tended to form a CR growth pattern and apoptotic cells were more frequent. The seeded tumor cells grew in large patches or continuous carpets on the top of these noninhibitory fibroblast monolayers.

To establish the possible influence of monolayer architecture on the inhibitory effect, we have isolated isogenic BjhTERT subclones that differed in their growth pattern. Subclones that maintained the WH pattern of the parental line were highly inhibitory whereas monolayers that emerged from single cell clones with CR phenotype were less inhibitory. CR clones grew more slowly, had more spontaneous apoptosis and tended to convert back to the faster growing and more inhibitory WH type.

The biochemical basis of the inhibition is not known but our observations provide some clues. The inhibitory effect was not transferable with fibroblast or mixed cell culture supernatants. Effective inhibition required intact monolayer formation and the effect increased with the age and confluency of the monolayer. When the formation of the monolayer was obstructed by the presence of randomly deposited irradiated tumor cells, the inhibition decreased. These findings could be due to loss of structured accumulation and deposition of extracellular matrix molecules that may provide orientation dependent behavioral cues to the tumor cells in an unmanipulated, inhibitory monolayer.

Induction of cell polarity may provide an important growth inhibitory cue that can override the proliferation driving effect of procarcinogenic mutations and impose a more differentiated phenotype on the tumor cells.20, 21 Cultivation of the tumor cells in a three-dimensional matrix that provides basal membrane components in their immediate surroundings induce apico-basal polarity even in highly dedifferentiated breast carcinoma cells, leading to the formation of acinus like structures.9 Cells that develop apico-basal polarity may be subjected to mechanical constraints imposed by asymmetrically distributed cytoplasmic components of the cytoskeleton, ER, Golgi, adhesion complexes, cell surface receptors, and so forth. Cell division requires the abolition of polarity and homogeneous distribution of cytoplasmic components.22, 23

Conceivably, the organized monolayer may thus induce an asymmetric distribution of tumor cell components, leading to imposed cell polarity that counteracts cell proliferation. This scenario is consistent with our observation that individual, dispersed tumor cells are effectively inhibited in their proliferation as long as they are in direct contact with the fibroblast monolayer underneath, but proliferate without restraint inside a tumor cell nest, surrounded by tumor cells that cannot provide polarity cues. Polarity induction requires static, immobile interaction between the epithelial component and the stroma. Our observation that there is a clear difference between the motility of the tumor cells growing on inhibitory versus noninhibitory monolayers is consistent with this scenario. The finding that growth inhibition is associated with loss of motility along the surface of the inhibitory monolayer but not on a noninhibitory monolayer indicates the occurrence of adhesive interactions in the former case that may impose cellular polarity. Noninhibitory cultures may lack the cell surface molecules or the extra cellular matrix components needed for adhesive interactions, permitting the continuous rolling of tumor cells over the monolayer and preventing the induction of polarity.

The scenario of monolayer imposed inhibition through induced cell polarity is testable. It postulates that asymmetric distribution of cytoplasmic components should be detectable in inhibited but not in noninhibited tumor cells; that interference with polarity regulating genes by siRNA should abrogate the inhibition and that polarity inducing cues may lose their growth inhibitory effect if they approach the tumor cells from several directions at the same time.

The analysis of gene expression patterns in the two inhibitory/noninhibitory fibroblast pairs may bring us closer to the identification of the relevant molecular effectors. Our preliminary data on the two isogenic but functionally different (inhibitory/less or non-inhibitory) fibroblast pairs identified 1,033 (of 53,196) transcripts that co-segregated with the phenotype. They include genes encoding both excreted and cell surface associated proteins that may participate in cell–cell interactions. There is also numerous differentiation and cell motion associated proteins that may contribute to polarity induction in the tumor cells. Characterization of the behavior of these candidate molecules in our labeled PC-3/BjhTERT experimental system may reveal the molecular mechanisms of fibroblast monolayer induced tumor growth inhibition.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  8. Supporting Information

The authors like to express our gratitude to the surgeons who provided us with the material for the outgrowth of primary fibroblasts: Èva Oláh, István Csízy, Tamás Józsa (Department of Pediatrics, Medical and Health Science Center, Debrecen University, Debrecen, Hungary) and Ove Andrén, Jan-Erik Johansson, Swen-Olof Andersson (Clinic of Urology, Örebro University Hospital, Örebro, Sweden). They also want to express their sincere appreciation to Professor Arne Östman and Dr. Martin Augsten (Cancer Centrum Karolinska, Karolinska Institutet, Stockholm, Sweden) for providing us with the BjhTERT cells and to Dr. Waldemar Waldeck (DKFZ Deutsches Krebforschungs-zentrum, Heidelberg, Germany) for the generous gift of the pSV-H2A-mRFP plasmid.

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
IJC_27521_sm_SuppFig1.tif23697KSupporting Information Figure 1
IJC_27521_sm_SuppMovie.avi59855KSupporting Information Movie
IJC_27521_sm_SuppTab1.rtf2619KSupporting Information Table 1

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.