Engineered 3D environments to elucidate the effect of environmental parameters on drug response in cancer

Abstract

Traditional in vitro models used for the development of anti-cancer drugs are based on the monolayer culture of cells, which has a limited predictivity of in vivo efficacy. A number of cell culture platforms have been developed in recent years to improve predictivity and further to elucidate the mechanisms governing the differing responses observed in vitro versus in vivo. One detrimental aspect of current in vitro models is their inability to decouple the effect of different extrinsic factors on the responsiveness of the cells to drug treatment. Here, we have used an engineered poly(dimethylsiloxane) (PDMS) microwell array as a reductionist approach to study the effect of environmental parameters, independently of each other. It is observed for MCF-7 breast cancer cells, that culture within the three-dimensional (3D) environment of the microwells alone had an effect on the response to Taxol and results in a reduction of cell death in comparison to cells cultured on flat substrates. Additionally the microwells allowed the response of single versus multicell clusters to be differentiated. It was found that the formation of cell–cell contacts alters the drug response, depending on the type of adhesive protein present. Thus, with this microwell platform it is revealed that the presence of cell–cell contacts in addition to the dimensionality and the matrix composition of the environment are important mediators of altered drug responses. In conclusion the microwell array can not only serve as a platform to reveal which parameters of the extracellular environment affect drug response but further the interdependence of these parameters.

---

Insight, innovation, integration

Currently available in vitro models fail to recapitulate the characteristics of the in vivo environment and thus many drug candidates are lost in the transition from in vitro to in vivo. Discrepancies between in vitro and in vivodrug response have been postulated to be due to differences in the signaling from the extracellular environment. Therefore, a greater understanding of the signaling pathways by which the extracellular environment can affect drug response is an essential step towards more effective drug development. However, in vitro platforms used to study this phenomenon typically suffer from the drawback of being unable to differentiate between the effects of different extrinsic parameters. A PDMS based microwell array is exploited as a cell culture platform to independently study the effect of several of these parameters on the response of cancer cells to drug treatment. This method is compatible with high resolution imaging, which allows for apoptosis detection on the single cell level in a spatially resolved manner.

Introduction

The prognosis of breast cancer patients has greatly improved during recent years, partly because many new therapies, including new chemotherapeutic drugs and targeted treatments, have been developed.¹ Despite these advances, breast cancer remains a major clinical problem worldwide. Currently one in four deaths in the United States are due to cancer, where breast cancer alone is responsible for 15% of cancer related deaths in women.² One hurdle is the still ineffective development process of new drugs, with many drugs being lost in the transition from in vitrodrug screening to animal models and clinical trials,³ articulating a demand for more relevant preclinical models. Still today the early screening process uses cancer models based on monolayer culture of cells, which do not accurately reflect the physiological environment of the tumor in vivo. On the contrary, it has become common knowledge that conventional monolayer culture induces both phenotypic and genotypic changes in the cells,⁴ for example growth kinetics and metabolic rates of cells are different for cells cultured in 2D and 3D.⁵ Therefore such preclinical in vitro models have had limited predictivity for in vivo efficacy.⁶

A central reason for the failure of current in vitro techniques is their inability to mimic key aspects of the in vivo environment of the tumor. Hence, in order to improve the in vivo to in vitrotranslation efficiency new culture models that better mimic the in vivo environment need to be developed. Therefore, the emphasis in current research has been to elucidate which are the key parameters of the extracellular environment that affect drug response. It has long been known that cells in 3D multicellular tumor spheroids (MCTSs) typically have a lower susceptibility to cytotoxic drugs compared to cells grown on 2D substrates. This difference has been argued to be an effect of lower drug penetration, development of a hypoxic core and decreased growth.⁷ Increased signaling from enhanced cell–cell contacts in 3D culture could be one important factor in the reduced drug responsiveness in spheroids,⁸ which could in turn be the determinant of decreased growth.⁹ Indeed, it has been shown that in the presence of E-cadherin interfering antibody, cells within spheroids were sensitized to several chemotherapeutic drugs.¹⁰

As MCTSs typically are formed on non-adhesive substrates or in hanging drops, the interaction with the extracellular matrix is absent in this model.⁷ However, the composition of the ECM has shown to have an important effect on drug response. In initial work it was found that the growth of cancer cells on flat substrates coated with different ECM proteins affected their response to anti-cancer drugs.¹¹ A recent paper uses a more in vivo like fibronectin matrix, produced by fibroblast cells, to study matrix induced alteration of drug response.¹² The matrices can be tuned to mimic different stages of the tumor progression by altering the fibroblasts that produce them, for example matrices derived from NIH-3T3 fibroblasts are regarded as control (non-cancerous) or early-stage cancerous environments.¹³ Using this model it was possible to show that not only the presence of the matrix but also the organization of the matrix in 2D or 3D influences drug response. The effect of the dimensionality of the environment on the behavior of cancer cells has further been studied using porous polymeric scaffolds. It was revealed that the 3D culture of cells within the scaffolds induced a decreased drug response.^14,15 Therefore dimensionality, the extracellular matrix and cell–cell contacts all play a role in determining drug responsiveness.

However, there are several drawbacks of the aforementioned models. One disadvantage is that most of the models are limited to study the effect of only one or a few parameters of the extracellular environment. An even more important consideration is the inability of many of these models to differentiate between the effects of different extrinsic parameters. In fundamental studies of cell biology, reductionist models of the in vivo environment have shown promise in the deconvolution of the importance of extrinsic factors on cell behavior.^16,17 Such models typically rely on micropatterned substrates, for example protein patterns in the size of a single cell that have been used to study the effect of cell spreading on different cell behaviors.¹⁸ However, until recently there has been a lack of tools that enable the investigation of cells cultured within structurally and biochemically controlled 3D environments. Therefore, within our lab we have developed a poly(dimethylsiloxane) (PDMS) based microwell array for the culture of cells in a 3D environment, in which not only dimensionality but also cell cluster size and protein coating can be controlled independently.¹⁹

This work aimed at evaluating the microwell array as a reductionist approach to study the effect of different extrinsic parameters on the response to anti-cancer drugs. The microwell array has been developed to be compatible with high-resolution microscopy (Fig. 1). This allowed detailed information of cell health at the single cell level to be obtained, eliminating the need for population based measurements, which are typically associated with large errors and can be difficult to adapt for measurements in 3D cultures.²⁰ As a model experiment we have chosen to study the Taxol-induced apoptosis in breast carcinoma cells. In short, this paper presents the use of a microwell array to investigate the individual effects of the 3D culture, matrix composition and cell–cell contacts on cellular responses to Taxol. In a central experiment we compared the results obtained using this model to a more complex 3D fibronectin matrix, which has been previously used as an in vitro model of the tumor stroma.¹²

Fig. 1 The principle of using the microwell array. The microwell array was molded onto a thin glass cover slide to enable high-resolution, inverted-stage microscopy.¹⁹ Before seeding cells, subtractive microcontact printing of fibronectin was used to specifically functionalize the wells, while the plateau was passivated with Pluronics.⁴¹ This setup ensures that it is possible to seed, incubate and analyze the cells on the same platform.

Results

Determining the rate and type of cell death induced by Taxol

The induction of cell death by Taxol, a standard drug in the treatment of breast cancer,^21,22 was initially investigated using two different assays. Taxol acts by suppressing microtubule dynamics, which leads to a mitotic block and apoptosis in cancer cells.²³ In a first experiment we measured the number of both apoptotic and necrotic cells in a population of cells grown on flat fibronectin-coated PDMS substrates and treated with 100 nM Taxol. The LIVE/DEAD assay identifies dead cells on the basis of membrane integrity and was used to differentiate necrotic from apoptotic cells,^24,25 while apoptosis was detected by nuclear fragmentation, which is an easily identifiable characteristic of apoptosis.²⁶ It was observed that the number of apoptotic cells, as determined by nuclear fragmentation, increased by 24 ± 4% from 24 h to 48 h of treatment (57 ± 3% and 81 ± 1% cell death respectively) (Fig. 2). Apoptosis levels of untreated samples, i.e. after incubation with media alone, were less than 5%. The LIVE/DEAD assay revealed that the level of necrosis in these cell populations remained below 10%, even after incubation with Taxol for 48 h. In all of the following experiments the cells were analyzed after 24 h of Taxol treatment, as at this time-point the rate of apoptosis was in a range where differences in drug response would be possible to detect.

Fig. 2 Cell viability of MCF-7 cells cultured on flat fibronectin coated PDMS after Taxol treatment was determined by counting the number of non-viable cells, as determined by the presence of a fragmented nucleus (black bar) or uptake of Ethidium homodimer (white bar), relative to the total number of cells present. MCF-7 cells cultured on flat fibronectin coated substrates showed an increase in fragmented nuclei over 48 h. The low number of cells stained with Ethidium homodimer DEAD stain indicates that most cells undergo apoptosis.

Using the microwell array to study the effect of dimensionality on drug response

Dimensionality was the first extrinsic factor that was evaluated, as it has previously been shown to be an important factor in governing altered drug responsiveness.^14,27 The individual microwells within the array were 34 μm in diameter by 10 μm in depth and MCF-7 breast carcinoma cells cultured in the microwells formed small clusters (≤6 cells per well after 16 h culture; Fig. 3B and supplementary figure†). At this time point the clusters seldom filled the full volume of the well, however from confocal images it was clearly visible that the cells interacted with both the sides and the bottom of the microwell to form 3D clusters. This contrasted to cells cultured on flat substrates, also coated with fibronectin, which were visibly more flat and stretched (Fig. 3A and supplementary figure†). Thus we found it interesting to explore whether the culture of small clusters of cancer cells in the engineered 3D environments of microwells has an effect on drug response. MCF-7 breast carcinoma cells cultured either within microwells or on flat substrates coated with fibronectin showed very low levels of cell death after 24 h, regardless of the substrate. Treatment with 100 nM Taxol significantly increased cell death after culture either on flat PDMS (59 ± 3% death) or within microwells (42 ± 4% death) (Fig. 3C and Table 1). Intriguingly, the response to Taxol was significantly lower for cells cultured within the microwells (17 ± 5% decrease, p < 0.001).

Fig. 3 (A) Confocal images and cross section of z-reconstruction of the cell organization on flat substrates (left) and in the microwells (right). The cells were cultured for 24 h in 34 μm wide fibronectin-coated microwells and then stained for F-actin (green) and nuclei (red). In microwells the cells are inhibited from spreading in comparison to cells cultured on flat substrates, but were observed to interact with both the bottom and sides of the microwells to form small 3D clusters. Scale bar is 10 μm. (B) Microwells coated with fibronectin contained 1 to 6 cells per well, with two thirds of the cells existing in pairs or clusters of 3 cells. (C) MCF-7 cells that are seeded in 34 μm wide microwells show normal nuclei after 40 h cell of culture (A). 24 h Taxol treatment induced apoptosis in the cells. This is shown by the fragmented nuclei in B (white arrows).

Table 1 Summary of the level of cell death determined as percent of fragmented nuclei, on flat substrates and in 34 μm wide microwells after 24 h exposure to Taxol

		Control	Drug	Control	Drug	Significance
Fibronectin	MCF-7	3 ± 1	59 ± 3	2 ± 1	42 ± 3	p < 0.001
	HRE	4 ± 1	20 ± 2	1 ± 1	14 ± 2	Not sign.
Collagen I	MCF-7	3 ± 1	70 ± 2	3 ± 1	59 ± 2	p < 0.05

---

The effect of dimensionality on drug response was also investigated for normal cells, namely human primary renal epithelial (HRE) cells, to determine whether the effect of dimensionality was specific to cancer cells. It was observed that Taxol induced cell death was significantly lower for normal cells than observed for the cancer cells (Table 1). On flat substrates, Taxol induced cell death was 20 ± 2% and within microwells 14 ± 3%, which was significantly lower than the cell death which occurred when MCF-7 breast carcinoma cells were cultured either on flat substrates or in microwells (39 ± 4% lower; p < 0.001 and 28 ± 4% lower; p < 0.001 respectively). Further, the normal cells showed no significant difference in drug response between the flat and the 3D substrates.

Studying the effect of dimensionality in combination with different matrix coating

The effect of dimensionality on drug-induced cell death was further explored in combination with different matrix coatings. Hence, we coated both the microwells and the flat substrates with collagen I, another common cell adhesion protein and assessed the responsiveness of MCF-7 breast carcinoma cells after culture in both 2D and 3D. We found, in coherence with the fibronectin results, that cells grown in collagen I coated microwells showed a small, but significant, reduction in cell death after treatment with 100 nM Taxol compared to cells cultured on flat collagen I coated substrates (12 ± 4% lower, p < 0.01) (Table 1). This difference in drug response between the two substrates was lower than observed for fibronectin-coated substrates. Further, it can be concluded that MCF-7 breast carcinoma cells were more sensitive to Taxol treatment after culture on collagen I, in comparison to fibronectin, on both flat substrates (12 ± 5% higher, p < 0.05) and within microwells (17 ± 5% higher, p < 0.001).

Using the microwells to differentiate between cell death in single cells and in cells forming cell–cell contacts

The microwell array offers a unique opportunity to study the importance of cell–cell contacts on the responsiveness to treatment. By controlling the well size and seeding density it is possible to form cell clusters in the microwells in different size ranges. Here we compared the level of cell death for single cellsversus pairs of cells. The level of cell death within fibronectin coated microwells was 61 ± 7% for single cells and 40 ± 4% for cells with cell–cell contacts, thus revealing that single cells were significantly more sensitive to drug treatment (21 ± 9% higher, p < 0.01) (Fig. 4). In contrast to cells cultured in fibronectin-coated microwells, there was no significant difference in drug response for single cells and cell pairs cultured in collagen I coated microwells (39 ± 6% for single cellsversus 54 ± 4% for pairs of cells). Further, the inter-dependency between cluster size and matrix coating was explored using the microwell platform and revealed that while single cells cultured on fibronectin were more sensitive to drug treatment than those cultured on collagen I, this trend was reversed after the formation of cell–cell contacts and within larger clusters.

Fig. 4 The level of cell deathversus number of cells present within the microwell coated with either fibronectin (black bars) or collagen I (white bar) was plotted. It was found that single cells cultured in fibronectin-coated microwells were more sensitive to drug treatment than cells that had formed cell–cell contacts. This phenomenon was not observed for MCF-7 cells cultured in collagen I coated microwells, as no significant difference in cell death was observed between single and pairs of cells. Further, single cells cultured on fibronectin were more sensitive to drug treatment than those cultured on collagen I. Conversely, this trend was reversed when 2 or more cells were present within the microwells. Key: * p < 0.05: not sign. = not significant, grey symbols refer to collagenversus fibronectin, while black symbols refer to single cellsversus pairs of cells.

Comparison of the Taxol response and proliferation in the 3D environments of the microwell array and in a 3D fibronectin matrix

Fibronectin matrix, derived from fibroblasts, is considered to be a mimic of the 3D environment which cancer cells experience in vivo and has been used previously to assess the response to Taxol for a number of different cancer cell lines.¹² It was therefore interesting to compare cell behavior after culture within the microwells to that of cells cultured in the more complex fibronectin matrices. MCF-7 cells reseeded into the cell-derived matrices interacted with the fibronectin fibers (Fig. 5A) and showed only negligible cell death (2 ± 1% death). Treatment with Taxol induced a significant increase in cell death in the matrices (26 ± 5% death), however cell death was significantly lower in comparison to cells cultured on flat substrates (33 ± 6% lower, p < 0.001) (Fig. 5B). Hence, the cells were less sensitive to Taxol treatment after culture within the cell-derived fibronectin matrix, which correlates well with the results observed in the 3D environment of the microwell array and highlights the importance of dimensionality in governing drug responsiveness. However, while both platforms followed the same biological trend, the cell death observed after culture in the microwells was slightly greater (16 ± 7% higher, p < 0.05) than observed after culture in the matrix.

Fig. 5 (A) MCF-7 seeded into cell-derived fibronectin matrices were observed to interact tightly with the fibronectin fibers. This example image shows F-actin staining of cells cultured within the matrix for 24 h. (B) MCF-7 cells were cultured within cell-derived fibronectin matrices for 16 h before exposure to Taxol for a further 24 h. The level of apoptosis was significantly lower in the fibronectin matrices compared to on flat substrates. In comparison to the microwells the drug response was slightly lower. (C) Levels of proliferating cells were determined by incubation of the cells for 6 h with the DNA-intercalating molecule BrdU. The proliferation of the cells cultured in microwells (black bar) was unchanged in comparison to cells cultured on flat substrates (striped bar). Conversely proliferation in cell-derived matrices (white bar) was significantly reduced. Key: * p < 0.05; *** p < 0.001.

One factor that influences the cellular response to Taxol is cell turn-over rate, as the mode of action of this drug requires cell cycle transit.²⁸ To determine the extent to which the lower drug response observed in the 3D environments was influenced by proliferation rate changes, we determined the proliferation rates in the different environments studied. Proliferation rates were assessed using the BrdU proliferation assay, for MCF-7 cells cultured in fibronectin coated microwells and in cell derived matrices, in comparison to cells grown on fibronectin coated flat substrates. No difference in proliferation was observed between cells cultured in microwells (40 ± 2% proliferation) and on flat substrates (38 ± 3% proliferation) (Fig. 5C). However, in cell-derived matrices the proliferation was significantly lower than on flat substrates (24 ± 6% decrease, p < 0.001).

Discussion

In order to improve the development process of new anti-cancer drugs, we need to understand the effect of the signaling from the extracellular environment on drug response. Typically, in vitro models used to study such effects are based on polymer scaffolds or protein-based matrices. One drawback of such models is that they often constitute a complex environment in which it can be difficult to separate the effect of different environmental parameters. Therefore we have investigated the possibility of using a microwell array to study the effect of the extracellular context on drug response in a controlled 3D environment.

Dimensionality alone affects the response to Taxol

It was observed that the cancer cells cultured within the 3D environment of the microwells were less responsive to treatment with Taxol than cells cultured on 2D substrates. Our findings are in agreement with previous research, which revealed that the 3D culture of cancer cells induces a reduction in drug sensitivity.^14,27 Typically the reduced drug responsiveness in 3D environments has been demonstrated in MCTSs or 3D scaffolds, where cells are present in large clusters. The formation of smaller cancer cell clusters (10–20 cells) in matrigel has previously shown to alter both cell morphology and gene expression in comparison to culture on tissue culture plastic.⁴ However, the effect of these phenotypic and genotypic changes on drug response was not investigated. It is therefore interesting to see that these small cell clusters, with ≤6 cells per cluster, show differences in drug response depending on the dimensionality of the environment. This indicates that dimensionality can alter drug responsiveness even in the absence of dense packing of the cells and oxygen gradients provided by more complex in vitro models, such as MCTS. In contrast, normal renal epithelial cells showed no significant difference in response in 3D versus 2D, even though the response followed the same trend observed for cancer cells. We propose that the drug response in these cells is either not affected by the 3D environment or to a lesser extent. However, it should be borne in mind that these cells are derived from kidney and thus the difference observed between cancer and normal cells may in part be due to the differing tissue origin.

The effect of dimensionality is also observed in a collagen I environment

The reduced drug responsiveness after culture within a 3D environment was observed for two different interfacing proteins; fibronectin and collagen I. Intriguingly, the effect was more dramatic when the cells were adhered to fibronectin. In addition, there was an overall reduced sensitivity to treatment on fibronectin compared to collagen I, observed both on flat substrates and within the microwells. Both proteins are important constituents of the stroma surrounding tumors but are known to engage different integrins^29,30 which could explain the differences observed. The higher cell death observed after interaction with collagen I in comparison to fibronectin could also be due to the increased turn-over rate for MCF-7 on collagen I,³¹ thus rendering the cells more susceptible to the effects of Taxol. Furthermore, in vivo studies have shown that increased β1- and extracellular fibronectin expression has been associated with more aggressive and invasive breast cancer,^32,33 which may in part explain the reduced responsiveness to drug treatment observed after interaction with fibronectin, as revealed in this work.

Reduced apoptosis in cells forming cell–cell contacts is matrix specific

Cell–cell contacts have been hypothesized to be one important parameter in governing reduction in drug response,³⁴ and thus we used the microwell platform to explore the effect of this parameter. Intriguing differences in the effect of forming cell contacts was observed depending on the matrix protein. On collagen I coated substrates the formation of cell–cell contacts had little effect on the drug responsiveness of the cells. Interestingly, previous work has emphasized that crosstalk between adhesion proteins in cell-matrix contacts and in cell–cell contacts may disrupt E-cadherin interaction, as was shown for cancer cells adhering on collagen I.³⁵ It is possible that a similar crosstalk takes place in the MCF-7 cells in this study and therefore would explain why there was no significant effect of cell–cell contacts on drug response for cells adhering on collagen I. Conversely, our experiments could nicely reveal that cells in fibronectin-coated microwells lacking cell–cell contacts were more sensitive to treatment. Thus the formation of cell–cell contacts, leading to reduced drug responsiveness, would appear to be one contributing factor of the increased overall resistance observed when cells were cultured in the presence of fibronectin, as opposed to collagen I. On fibronectin/cadherin patterned surfaces it was found that the dominant role of integrin interactions over E-cadherin interactions in MCF-7 cells was rigidity dependent and could be ruled out on soft substrates.³⁶ Hence, if matrix interactions are reduced in a system, as is typically observed on soft substrates,³⁷ this may strengthen the cell–cell contacts, which could in turn lead to reduced drug responsiveness. Intriguingly this may also explain in part the effect of dimensionality observed. It is probable that cells within the microwells have reduced matrix interactions, and thus increased cell–cell contacts, as they are restricted in both their spreading and migration.

The dimensionality effect in the microwells partly mimics that observed in a 3D fibronectin matrix

If the advantage of a reductionist model is that different extrinsic factors may be studied separately, then the disadvantage is clearly that one fails in mimicking the complexity of the in vivo environment. Therefore, we found it interesting to compare our results to a cell-derived fibronectin matrix, which is a more complex in vivo model that has been previously used to study matrix induced changes in drug response.¹² After culture within the matrix, the cells were less responsive to drug treatment in comparison to cells grown on flat substrates. This correlates well with our results with the microwell platform, and further indicates that we were able to mimic some aspects of the complex environment of the matrices within the 3D environments of this in vitro model. However, even though the drug response was reduced in both 3D systems, the effect was enhanced in the matrices.

In addition we found that the proliferation rate in the two 3D systems was significantly different. In the microwells the proliferation rate was unchanged in comparison to flat substrates, while it was significantly reduced in cell-derived matrices. This result is coherent with previous research using these matrices,¹² however the difference in proliferation was more pronounced in this work, possibly due to differences in the proliferation assays utilized. As Taxol is an anti-mitotic drug it is probable that the lower proliferation rate observed in the cell-derived matrix could at least partially explain the reduced drug response in the matrix. However, this is not the case for cells cultured in the microwells, for which there was no difference in proliferation in comparison to cells on flat substrate. There is not always a correlation between high mitotic frequency and elevated response to chemotherapeutic drugs.^12,38 Instead, the response to taxol should be regarded as a function of both turn-over rate and drug sensitivity. For highly aggressive cancers, low sensitivity to treatment may even rule out an increased susceptibility to anti-cancer drugs due to a high turn-over rate.³⁹ Another aspect to consider when comparing these two 3D systems is the organization of the ECM, as a coating in the microwells or as fibers in the matrix. The organization of the ECM is known to affect integrin signaling,⁴⁰ which could in turn lead to differences in drug response.³⁴ Hence, we suggest that the effect of dimensionality observed in the two systems is not only related to changes in proliferation rate and also not only induced by the 3D characteristic of the environment. Instead, other environmental parameters, such as the scaffold pliability, the type and amount of cell–cell interactions and the conformation of the matrix molecules, may play additional roles in determining responsiveness to drug treatment.

Conclusions

In summary, with this platform it was possible to demonstrate the individual effect of dimensionality, adhesive proteins and cell–cell contacts on the responsiveness of breast cancer cells to drug treatment. Further it was possible to explore the inter-relationship between these environmental parameters and reveal a cumulative effect of dimensionality and matrix composition on drug response. The lowest sensitivity to Taxol was acquired after culture in 3D in the presence of fibronectin. This drug response pattern can be expected to be different for different cancer cell types and would be interesting to explore. In addition, this could be extended to larger clusters of cells, using microwells with a higher volume, to provide more predictive results for cells residing in dense tumors. There is probably not one major extrinsic factor that alters the responsiveness to treatment in vivo and it is clear that a crucial step in the development of more effective treatments must be the understanding of the interdependence of signaling from different constituents of the environment.

Materials and methods

Materials and cells

Fibronectin was isolated from fresh human plasma (Swiss Red Cross) and labeled, when required, with Alexa Fluor 488 (Molecular Probes, Switzerland). Collagen I was purchased from Gibco, Switzerland. The protein solutions were diluted to a working concentration of 25 μg ml⁻¹ for fibronectin and 46 μg ml⁻¹ for collagen I in PBS before use. MCF-7 cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), while HRE cells and REBM™ (Renal Epithelial Cell Basal Medium) and REGM media supplements were purchased from Ionza, Switzerland. Dulbecco's modified Eagle's medium (DMEM) cell culture media, for MCF-7 cells, penicillin/streptomycin and FBS were all obtained from Invitrogen, Switzerland. Taxol was purchased in 1 mg aliquots from Sigma Aldrich and stored as 1 mM stock solutions in DMSO at −20 °C.

Fabrication of the microwell arrays

Arrays of microwells with a diameter of 34 μm and depth of 10 μm were fabricated in PDMS as previously described.¹⁹ Briefly, microstructures were created in SU-8 on a silicon wafer using standard photolithograpy. The structures were transferred into a PDMS master by molding PDMS onto the silicon wafer. After fluorosilanization of the PDMS master, it could be repeatedly used to create thin film PDMS replicates on thin glass cover slips (Menzel-Gläser, Germany, strength 0, approx. 100 μm thickness). The arrays were functionalized with protein only in the wells using a recently developed method⁴¹ based on substractive microcontact printing (μCP). In short, the arrays were first coated with protein, and then substractive μCP was used to remove the protein from the plateau surface using glutaraldehyde functionalized PDMS stamps. In a final step the surface was rendered non-adhesive for cells by backfilling with Pluronic F-127®. 2D samples were prepared by molding thin PDMS films on thin glass cover slips and subsequently rendered hydrophilic in an air-plasma to facilitate coating with protein. All samples were glued into the bottom of a Petri dish into which a hole was previously drilled to facilitate cell culture and imaging of the samples.

Production of cell-derived fibronectin matrices

Fibronectin matrices were produced as previously described.⁴² In short, over-confluent NIH 3T3 fibroblasts were cultured in 96-well plates coated with fibronectin for at least 4 days and subsequently decelluralized. After completed decelluralization the matrices were rinsed with PBS and reseeded with MCF-7 breast carcinoma cells within a week.

Characterization of cell morphology on the substrates

MCF-7 cells were seeded onto the substrates at 3 × 10⁻⁵cells ml⁻¹. After 2 h the microwell substrates were rinsed twice to remove any non-adherent cells. After 24 h the cells were fixed using paraformalehyde (4% for 10 min at RT) and permeabilized with triton-X (0.1% for 20 min at RT). The cells were stained for F-actin with Phalloidin-Alexa Fluor 488 and nuclei were stained with propidium iodide. Imaging was performed on inverted confocal microscopy using at least 40× magnification. (Olympus, Switzerland).

Drug treatment and apoptosis determination

MCF-7 breast carcinoma cells were maintained in DMEM supplemented with foetal bovine serum (10% (v/v)), penicillin (100 U ml⁻¹) and streptomycin (100 μg ml⁻¹) and were grown in a humidified atmosphere (95% (v/v) air, 5% (v/v) carbon dioxide at 37 °C). Human renal epithelial (HRE) cells were cultured in REBM™ supplemented with REGM media supplements, as recommended by the cell supplier. For experimentation MCF-7 or HRE cells were seeded onto the substrates at 3 × 10⁻⁵cells ml⁻¹ and allowed to adhere for ca. 2 h before rinsing twice with cell culture media to remove any non-adherent cells. Cells were cultured on the substrates for 16 h prior to incubation with Taxol (100 nM, including 0.1% (v/v) DMSO) for 24 or 48 h. Control samples consisted of cells cultured in DMEM media with 0.1% (v/v) DMSO to match test samples. Cell viability was determined using the LIVE/DEAD assay (Invitrogen) according to the manufacture's protocol. Apoptosis was detected by nuclear fragmentation, as described previously.¹¹ Nuclei were stained with Hoechst 33 342 (0.3 μg ml⁻¹ for 20 min at 37 °C) and total nuclei and number of fragmented nuclei was determined using an inverted widefield microscope (Zeiss, Germany) with a 20× air objective. See Fig. 3C. Experiments were always performed in duplicates and repeated three times. In every experiment >100 cells were counted, with an exception for the analysis of single cells in microwells, in which a minimum of 20 cells per experiment were counted.

Assessment of proliferation

Proliferation was assessed using the BrdU assay according to the manufacture's instructions. Samples preparation and seeding of cells was performed as described above for the drug treatment experiments. After overnight cell adhesion the media was exchanged for media containing 10 μM BrdU (Sigma Aldrich, Switzerland) and the samples were incubated at 37 °C for 6 h; subsequently all the samples were fixed with ice-cold methanol (70% (v/v) in water for 20 min). The samples were treated with 2 M HCl for 20 min, neutralized with 0.1 M Borax for 2 min and permeabilized with 0.1% (v/v) Triton-X for 10 min. All stages after fixation were performed at room temperature and in between each stage of the treatment the cells were washed three times with PBS. Samples were labeled with primary mouse-anti-BrdU IgG (1 : 100 (v/v); 60 min; BD Biosciences), washed three times with PBS and incubated with goat anti-mouse IgG Alexa Fluor 488 (1 : 400 (v/v); 60 min; Molecular Probes). 0.5% (w/v) BSA was included in all staining buffers to avoid non-specific binding. The samples were counterstained with Hoechst 33 342, to label cell nuclei, and imaged using a 20× objective on an inverted Zeiss widefield microscope. BrdU-labeled versus Hoechst 33 342 labeled nuclei were manually counted with at least 100 cells per experiment. The experiments were performed in duplicates and repeated three times.

Statistic analysis

Quantitative data was plotted as the mean ± standard error of the mean (SEM) and normalised relative to the negative control (cells treated with media alone). Statistical analysis was performed using Student's unpaired two-way t-tests (using MinitabTM). Comparison between data sets was performed using one-way ANOVA (Tukey test) using MinitabTM. Differences were considered as statistically significant when p < 0.05.

Acknowledgements

The authors would like to thank the Competence Center for Materials Science and Research, Switzerland, CCMX, for their kind support of this work.

References

1. C. Mettlin, CA Cancer J. Clin., 1999, 49, 138–144.
2. A. Jemal, R. Siegel, E. Ward, Y. Hao, J. Xu, T. Murray and M. Thun, CA Cancer J. Clin., 2008, 58, 71.
3. G. N. Hortobagyi, Semin. Oncol., 2004, 31, 21–27.
4. P. Kenny, G. Lee, C. Myers, R. Neve, J. Semeiks, P. Spellman, K. Lorenz, E. Lee, M. Barcelloshoff and O. Petersen, Mol. Oncol., 2007, 1, 84–96.
5. A. Ivascu and M. Kubbies, J. Biomol. Screening, 2006, 11, 922–932.
6. L. A. Kunz-Schughart, J. P. Freyer, F. Hofstaedter and R. Ebner, J. Biomol. Screening, 2004, 9, 273–285.
7. J. Friedrich, R. Ebner and L. Kunz-Schughart, Int. J. Radiat. Biol., 2007, 83, 849–871.
8. A. Ivascu and M. Kubbies, Int. J. Oncol., 2007, 31, 1403.
9. B. St Croix, C. Sheehan, J. W. Rak, V. A. Flørenes, J. M. Slingerland and R. S. Kerbel, J. Cell Biol., 1998, 142, 557–571.
10. S. K. Green, G. Francia, C. Isidoro and R. S. Kerbel, Mol. Cancer Ther., 2004, 3, 149–159.
11. F. Aoudjit and K. Vuori, Oncogene, 2001, 20, 4995–5004.
12. I. Serebriiskii, R. Castelló-Cros, A. Lamb, E. Golemis and E. Cukierman, Matrix Biol., 2008, 27, 573–585.
13. R. Castelló-Cros, D. R. Khan, J. Simons, M. Valianou and E. Cukierman, BMC Cancer, 2009, 9, 94.
14. H. Dhiman, A. R. Alok and A. K. Panda, Biomaterials, 2005, 26, 979–986.
15. C. Fischbach, R. Chen, T. Matsumoto, T. Schmelzle, J. S. Brugge, P. J. Polverini and D. J. Mooney, Nat. Methods, 2007, 4, 855–860.
16. K. Bhadriraju and C. S. Chen, Drug Discovery Today, 2002, 7, 612–620.
17. M. Ochsner, M. Textor, V. Vogel and M. L. Smith, PLoS One, 2010, 5, 9445.
18. W. Liu and C. Chen, Adv. Drug Delivery Rev., 2007, 59, 1319–1328.
19. M. Ochsner, M. R. Dusseiller, H. M. Grandin, S. Luna-Morris, M. Textor, V. Vogel and M. L. Smith, Lab Chip, 2007, 7, 1074–1077.
20. K. Ng, D. Leong and D. Hutmacher, Tissue Eng., 2005, 11, 182–191.
21. F. A. Holmes, R. S. Walters, R. L. Theriault, A. D. Forman, L. K. Newton, M. N. Raber, A. U. Buzdar, A. U. Frye and G. N. Hortobagyi, J. Natl. Cancer Inst., 1991, 83, 1797–1805.
22. E. P. Mamounas, J. Bryant, B. Leinbersky, L. Fehrenbacher, S. M. Sedlacek, B. Fisher, D. L. Wickerham, G. Yothers, A. Soran and N. Wolmark, J. Clin. Oncol., 2005, 23, 3686–3696.
23. M. A. Jordan and L. Wilson, Curr. Biol., 1998, 10, 123–130.
24. N. Morley, A. Rapp, H. Dittmar, L. Salter, D. Gould, K. O. Greulich and A. Curnow, Mutagenesis, 2006, 21, 105–114.
25. W. C. Prozialeck, J. R. Edwards, P. C. Lamar, J. Liu, V. S. Vaidya and J. V. Bonventre, Toxicol. Appl. Pharmacol., 2009, 238, 306–314.
26. H. Lövborg, P. Nygren and R. Larsson, Mol. Cancer Ther., 2004, 3, 521–526.
27. J. L. Horning, S. K. Sahoo, S. Vijayaraghavalu, S. Dimitrijevic, J. K. Vasir, T. K. Jain, A. K. Panda and V. Labhasetwar, Mol. Pharmaceutics, 2008, 5, 849–862.
28. D. Henley, M. Isbill, R. Fernando, J. S. Foster and J. Wimalasena, Cancer Chemother. Pharmacol., 2006, 59, 235–249.
29. M. C. Siebers, X. F. Walboomers, J. van den Dolder, S. C. G. Leeuwenburgh, J. G. C. Wolke and J. A. Jansen, J. Mater. Sci.: Mater. Med., 2008, 19, 861–868.
30. J. Heino, Matrix Biol., 2000, 19, 319–323.
31. K. Hirtenlehner, M. Pec, E. Kubista and C. F. Singer, Eur. Cytokine Network, 2002, 13, 234–240.
32. E. Yao, H. Zhang, Y. Chen, B. Lee and K. Chew, Cancer Res., 2007, 67, 659–664.
33. E. Ioachim, A. Charchanti and E. Briasoulis, Eur. J. Cancer, 2002, 38, 2362–2370.
34. L. A. Hazlehurst, T. H. Landowski and W. S. Dalton, Oncogene, 2003, 22, 7396–7402.
35. A. Koenig, C. Mueller, C. Hasel, G. Adler and A. Menke, Cancer Res., 2006, 66, 4662–4671.
36. J. Tsai and L. Kam, Biophys. J., 2009, 96, L39–L41.
37. T. Yeung, P. C. Georges, L. A. Flanagan, B. Marg, M. Ortiz, M. Funaki, N. Zahir, W. Ming, V. Weaver and P. A. Janmey, Cell Motil. Cytoskeleton, 2004, 60, 24–34.
38. T. Aas, S. Geisler, G. E. Eide, D. F. Haugen, J. E. Varhaug, A. M. Bassøe, T. Thorsen, H. Berntsen, A. L. Børresen-Dale, L. A. Akslen and P. E. Lønning, Eur. J. Cancer, 2003, 39, 438–446.
39. M. Meads, R. Gatenby and W. Dalton, Nat. Rev. Cancer, 2009, 9, 665–674.
40. S. Even-Ram and K. M. Yamada, Curr. Opin. Cell Biol., 2005, 17, 524–532.
41. M. Rottmar, M. Hakanson, M. L. Smith and K. Maniura-Weber, J. Mater. Sci.: Mater. Med., 2009, 20, 1–6.
42. K. E. Kubow, E. Klotzsch, M. L. Smith, D. Gourdon, W. C. Little and V. Vogel, Integr. Biol., 2009, 1, 635–648.

---

Footnote

† Electronic supplementary information (ESI) available: Supplemental figure. See DOI: 10.1039/c0ib00074d

---