by Dr R. de Pril, Dr T. Perera and Dr A. Lekkerkerker
The propensity of cancer cells to close an open wound in a cell monolayer is thought to predict their migratory ability in metastasis. Using our adenoviral short hairpin shRNA knock-down library we have established a high-throughput wound healing assay to identify novel genes involved in cell migration. We used a 96-pin scratch tool to apply a mechanical scratch-wound in a cellular monolayer. Wound healing was assessed using high-content imaging to measure the remaining scratch wound. In this article we demonstrate that several knock-down constructs targeting known players in motility, such as CXCR4, PIK3CA, ROCK2 and PGF inhibit wound healing, validating our model. We used this model to screen our adenoviral knock-down library, which focuses on drugable targets, and identified a high number of novel genes associated with tumour motility.
Enhanced cell migration is an important hallmark of metastatic cancer cells. Metastasis, the process whereby cancer cells leave the primary tumour and form secondary tumours at a distant site, is the leading cause of death in cancer patients [1]. The development of drugs that specifically inhibit tumour metastasis is therefore crucial for cancer survival.
Cancer cells acquire the ability to metastasise by de-differentiation to a more mesenchymal phenotype. Several triggers, such as intra-tumoural hypoxia, growth factor signalling and cell-cell interaction, underlie the cellular transition of an epithelial to a mesenchymal phenotype [2, 3]. This transition increases the aggressiveness of the tumour. These mesenchymal cells have increased ability to migrate throughout the body, eventually resulting in distant tumours. Cancer cells maintain their migratory capacity in vitro, and this is mediated by the same signalling pathways that affect the tumour in vivo. The basic mechanisms that underlie cellular movement including polarisation, actin remodelling and growth cone formation, are preserved in both three-dimensional and two-dimensional migration [2, 4]. In general, the propensity of cells to close an open wound in a cellular monolayer is thought to predict their migratory ability [5]. To identify new targets that play a role in cancer cell migration, we used this scratch wound assay to screen for druggable genes that affect the motile behaviour of tumour cells.
To that end, we employed Biofocus DPI’s RNAi-based target discovery platform in which we screen our adenoviral SilenceSelect shRNA libraries [Figure 1], [6]. We use human primary cells and cell lines to develop assays that provide the desired phenotypic response characteristic of the disease [7]. By using adenovirus we can efficiently transduce a very wide range of cells, cell lines and primary cells, which otherwise are difficult to transfect or are sensitive to gene or shRNA transfer. Thus, adenoviral transduction enables us to express or knock down individual genes of interest and exploit these cell-systems for our assays. In order to identify novel disease-modifying targets, our libraries are focused on the human druggable genome. This enables us to knock down specific genes in the disease model, yielding hits that can quickly be employed to generate small molecule compounds or antibody therapeutics. We have established a high-throughput wound healing assay that enables us to identify novel genes involved in cell migration using our adenoviral shRNA library. These targets will be employed to develop drugs to specifically tackle the aggressive spread of cancer in patients.
RESULTS
We seeded prostate cancer cells in 96-well plates and transduced them with our adenoviral knock down shRNA library. Transduced cells were cultured for six more days to allow for efficient knock-down on both mRNA and protein levels and to form a confluent monolayer. Next, a wound was formed in each well using a scratch pin. Specifically, we designed a 96-pin scratch tool to apply a consistent mechanical scratch-wound in a cellular monolayer. Pins were generated in such a way that they apply a constant pressure on the monolayer resulting in a reproducible scratch wound of 0.4 mm width [Figure 2A]. The cells were allowed to migrate into the open space over a period of several hours.
Subsequently, the cells were fixed and we imaged the plates on an InCell Analyzer 1000. Transmitted light imaging was used for segmentation and quantification of the scratch wound that remained open. Segmentation is thereby based on the contrast difference between the cell boundaries and the growth surface. We developed an algorithm in-house to measure the wound area that remained open after wound healing [Figure 2; red border]; the algorithm enabled us to identify genes whose knock-down inhibit cell migration. In addition, our algorithm determines open spaces between cells in the monolayer allowing us to identify knock down constructs that affect proliferation, adherence or viability of the cells [Figure 2C]. These constructs will be excluded as they affect motility by their indirect effect on the monolayer rather then by an effect on migration of the cells.
The tumour microenvironment is an important mediator of cancer cell behaviour and therefore we evaluated the effect of the extracellular matrix (ECM) coating on migration to optimise our wound healing assay. We found that, compared to uncoated plates, collagen-IV coating [Figure 2B] as well as collagen–I and fibronectin coating (data not shown) increased the motility of prostate cancer cells. Furthermore, a differential effect on wound healing was detected for specific genes using uncoated versus collagen-IV coated plates. This clearly demonstrates that matrix proteins affect the motile behaviour of cancer cells through specific pathways. Based on these results, we selected collagen-IV coating for further assay development.
To benchmark our screening assay and to control plate-to-plate and batch-to-batch variation, we selected a series of positive and negative controls, which were incorporated on each screen plate. Disease-relevant genes were selected as putative positive controls and tested for their effect on the phenotypic response in the respective screen. We seeded prostate cancer cells on collagen-IV coated plates and transduced the cells with adenoviral knock-down constructs targeting different genes. Two independent knock-down constructs targeting CXCR4, a known player in motility [8], demonstrated a clear reduction in motility of the prostate cancer cells. Furthermore, we demonstrated that knock-down constructs for PIK3CA, ROCK2 and PGF inhibited motility as well. Constructs directed against unrelated genes had no effect on migration [Figure 2D]. These constructs were subsequently used as positive and negative controls for further assay development and screening.
Using these selected controls we tested different time intervals from the moment of scratch formation [Figure 3]. We determined the optimal screening window to be at eight hours, demonstrating a clear separation between negative and intermediate and strong positive controls. At eight hours after scratch formation, prostate cancer cells transduced with negative knock-down controls had closed the scratch wound almost completely. Knock-down of genes involved in motility clearly result in inhibition of the motile behaviour of the cancer cells. Additionally, the wound healing kinetics of the cells demonstrate that the cells are still able to close the scratch wound within a time frame of up to 24 hours with reduced motility. This indicates that the effects on migration are not due to general toxicity but reflect a functional involvement in motility of the cancer cells.
Using collagen-IV coated plates and a time interval of eight hours after scratch formation, we screened a shRNA library of 7000 viruses targeting a selected subset of the druggable genome in biological duplicate. Consecutively, we excluded toxic hits that were identified by the algorithm and performed hit-calling on a plate-to-plate basis. Based on the non-parametric distribution of the samples, duplicate hits were defined on the basis of inter-quartile statistics. As would be expected, most constructs did not affect the motile behaviour of cells and resulted in a nearly complete closure of the wound.
Subsequently, we re-propagated the original hit viruses in combination with negative and positive controls and rescreened these in the wound healing assay. We found a hit confirmation of 85 % of our primary hits that were positive in duplicate within the rescreen. We identified a total of 250 genes, among which were 56 proteases, 45 kinases and 22 GPCRs whose knock down resulted in reproducible inhibition of wound healing. In parallel with other migration screens we have identified targets involved in several pathways that include cell adhesion and growth factor signalling [4, 9]. In combination with results from other migration screens this demonstrates a complex regulation of cell migration that involves polarisation, protrusion formation and cell-cell adhesion [2]. Disruption of these processes thereby results in changes in cellular morphology and loss of directionality of cell movement. Previous studies used epithelial cells that migrate as a cellular sheet maintaining contact between the cells [4, 9]. In contrast, in our screen, we made use of a mesenchymal cell-type limiting the effect of cell-cell adhesion. Aggressive spread of cancer cells is attributed to mesenchymal cells that have overcome the inhibitory effect of cell adhesion within the tumour. We thus identified a set of genes that is specifically involved in the migratory pathways involved in metastasis.
In this high-content screen we have identified a high number of novel genes associated with tumour motility. We will further validate these targets in secondary biological assays in order to determine the role of these targets in cell motility. As our adenoviral knock-down library focuses on druggable targets, this screen has yielded hits that can quickly be employed to generate small molecule compounds or antibody therapeutics.
In summary, our results demonstrate the strength of combining relevant biological assays with adenoviral functional genomics and high-content screening.
REFERENCES
1. Loberg RD, Bradley DA, S. Tomlins A, Chinnaiyan AM, Pienta KJ. The lethal phenotype of cancer: The molecular basis of death due to malignancy. CA Cancer J Clin 2007; 57(4): 225-241.
2. Friedl P, Wolf K. Tumour-cell invasion and migration: Diversity and escape mechanisms. Nat Reviews 2003; 3(5): 362-374.
3. Jiang P, Enomoto A, Takahashi M. Cell biology of the movement of breast cancer cells: Intracellular signalling and the actin cytoskeleton. Cancer Letters 2009. doi:10.1016/j.canlet.2009.02.034 (In Press).
4. Simpson KJ, Selfors LM, Bui J, Reynolds A, Leake D, Khvorova A, Brugge JS. Identification of genes that regulate epithelial cell migration using an siRNA screening approach. Nat Cell Biology 2008; 10(9): 1027-1038.
5. Liang C.-C, Park AY, Guan J-L. In vitro scratch assay: A convenient and inexpensive method for analysis of cell migration in vitro. Nat Protoc 2007; 2(2): 329-333.
6. Arts GJ, Langemeijer E, Tissingh R, Ma L, Pavliska H, Dokic K, Dooijes R, Mesic E, Clasen R, Michiels F, van der Schueren J, Lambrecht M, Herman S, Brys R, Thys K, Hoffmann M, Tomme P, van Es H. Adenoviral vectors expressing siRNAs for discovery and validation of gene function. Genome Research 2003; 13(10):
2325-2332.
7. Bortone K, Michiels F, Vandeghinste N, Tomme P, van Es H. Functional screening of viral siRNA libraries in human primary cells. Drug Discovery World 2004; Fall: 20-27.
8. Liang Z, Wu H, Reddy S, Zhu A, Wang S, Blevins D, Yoon Y, Zhang Y, Shim H. Blockade of invasion and metastasis of breast cancer cells via targeting CXCR4 with an artificial microRNA. Biochemical and Biophysical Research Communications 2007; 363(3): 542-546.
9. Collins CS, Hong J, Sapinoso L, Zhou Y, Liu Z, Micklash K, Schultz PG, Hampton GM. A small interfering RNA screen for modulators of tumor cell motility identifies MAP4K4 as a promigratory kinase.” Proceedings of the National Academy of Sciences of the United States of America 2006; 103(10): 3775-3780.
THE AUTHORS
Remko de Pril*, Tim Perera#, Annemarie Lekkerkerker*;
*Biofocus DPI, P.O. Box 127, 2300 AC Leiden, the Netherlands; #Ortho Biotech Oncology R&D, Turnhoutseweg 30, B-2340 Beerse, Belgium.
