Molecular profiling of laser microdissected cells and stroma from synthetic, tissue-engineered colorectal cancer models represents a novel and manipulatable approach to characterize and dissect the distinctive biology at the interface between tumor cells at the invasive front and cancer associated stromal cells.
Invading colorectal cancer (CRC) cells have acquired the capacity to break free from their sister cells, infiltrate the stroma, and remodel the extracellular matrix (ECM). Characterizing the biology of this phenotypically distinct group of cells could substantially improve our understanding of early events during the metastatic cascade.
Tumor invasion is a dynamic process facilitated by bidirectional interactions between malignant epithelium and the cancer associated stroma. In order to examine cell-specific responses at the tumor stroma-interface we have combined organotypic co-culture and laser micro-dissection techniques.
Organotypic models, in which key stromal constituents such as fibroblasts are 3-dimentioanally co-cultured with cancer epithelial cells, are highly manipulatable experimental tools which enable invasion and cancer-stroma interactions to be studied in near-physiological conditions.
Laser microdissection (LMD) is a technique which entails the surgical dissection and extraction of the various strata within tumor tissue, with micron level precision.
By combining these techniques with genomic, transcriptomic and epigenetic profiling we aim to develop a deeper understanding of the molecular characteristics of invading tumor cells and surrounding stromal tissue, and in doing so potentially reveal novel biomarkers and opportunities for drug development in CRC.
Organotypic co-cultures are tissue-engineered models which aid in reconstructing the in vivo tumor microenvironment in 3-dimensions by juxtaposing malignant epithelial cells and stromal cells in a collagen gel containing essential extracellular matrix components1-3. Principally conceived as a method of measuring tumor invasion, organotypics reduce reliance on in vivo animal models, and avoid the shortcomings of other in vitro techniques such as Transwell assays, in which invading cells are forced artificially into a mono-dispersed state2-4. The inclusion of stromal cells in these models, such as fibroblasts, reflects the essential role of the tumor microenvironment in regulating malignant invasion and metastasis3-4, however the phenotypic heterogeneity of the stroma is such that in order to maximize the physiological relevance of organotypic models, organ specific, anatomically accurate ex vivo fibroblasts should be included wherever possible3,6.
Organotypics are a versatile platform to study interactions between tumor and stromal cells and are increasingly used in novel ways to examine the impact on tumor invasion of chemical inhibitors and targeted gene alterations7,8. Studying the invasive tumor margin is a particularly exciting prospect. In organotypic models, established colorectal cancer (CRC) cell lines typically produce a well stratified epithelial layer, and in cross section, cells that have acquired the capacity to invade the extracellular matrix (ECM) are easily discernible. In light of the established micro-topographic heterogeneity of tumor and tumor associated stromal cells in vivo, extracting these cells using Laser microdissection and studying them in isolation, may reveal important biological insights regarding the origins of colorectal cancer metastasis and how it may be more efficiently targeted. In the following methodology, all patients who donated tissue samples provided written informed consent, and the study was approved by the institutions regional research ethics committee.
1. Establishment of Primary Fibroblast Cultures from Colonic Explants
2. Organotypic Preparation
Preparation of the fibroblast impregnated organotypic gel:
Preparation of gel coated nylon sheets:
Raising organotypic gels onto steel grids for invasion:
3. Organotypic Fixation
4. Laser Microdissection of the Invasive Margin
We have applied the above method to multiple combinations of CRC cell lines and stromal cells. One example presented here is SW480 epithelial CRC cell lines with primary human ex vivo colonic fibroblasts. 3-dimensional co-cultures are constructed (Figure 1), subjected to microscopy and laser capture microdissection (Figure 2), and analyzed by microRNA (miRNA) profiling (Figure 3) for comparison of differentially expressed miRNAs between cells at the invasive tumor front and those in the stratified (non-invading) tumor zones.
Figure 1. Schematic and photographic representations of the organotypic model. CRC cells are seeded onto a synthetic stromal layer containing fibroblasts and essential extracellular matrix components. Cells are co-cultured at an air-liquid interface created by elevating the organotypic gel onto a sterile stainless-steel grid. Click here to view larger image.
Figure 2. LMD of SW480 CRC cells invading the organotypic stroma. Cresyl Violet is used to highlight CRC epithelial cells (A); invasive tumor islands are then defined (B); before laser dissection occurs (C) and the cut piece is transported to a collection device (D). Non-invading cells and stromal cells are identified and isolated using the same methodology. Click here to view larger image.
Figure 3. MicroRNA microarray data. Arrays were scanned using a GenePix Pro 3.0.5 scanner to detect Cy5 at a wavelength of 635 nm. Mean fluorescent intensities were normalized and expression ratios calculated between laser microdissected invasive margin CRC cells and non-invasive margin CRC cells in the stratified epithelial layers. Data was sorted by fold change and data values +/- 2 fold change from a representative experiment are presented showing differential miRNA gene expression between the invasive margin cells and non-invasive margin cells. Click here to view larger image.
Here we describe a method to specifically isolate and characterize tumor cells which have acquired the capacity to invade the cancer associated stroma during the earliest stages of metastatic progression in a 3-dimensional co-culture construct of epithelial and stromal cells.
Co-culture models comprised of CRC epithelial cells juxtaposed with a synthetic stroma containing ex vivo human colonic fibroblasts were used to study CRC invasion in 3-dimensions. This physiologically relevant system which mimics in vivo conditions can be adapted for other cancer scenarios by substituting cells in the epithelium and constructing the stroma using ex vivo fibroblasts of various anatomical origins. One limitation of this approach is the oversimplification of the stromal compartment using fibroblast cells only, as the in vivo colonic stroma is a complex and dynamic tissue with varying cell constituents. One example is the immune-derived cells, which are present in significant quantities in primary tumor material but absent from this model. Despite this, the reliable reproducibility of this model coupled with the ease with which the cell constituents can be experimentally manipulated (e.g. by over expression or knockdown of individual genes in the epithelial and stromal compartments), make this a valuable and cost-effective model compared to conventional in vivo experimentation.
In this manuscript we also demonstrated that CRC cells at the tumor invasive front in this model system express different miRNA expression profiles to identical cells not at the invasive front and situated in stratified epithelial layers. This represents a proof of concept that laser microdissected tissue from organotypic co-culture models is a valid approach to study early processes in metastasis. The focus here was exclusively on miRNA expression, since miRNAs are strongly implicated in the pathogenesis of numerous malignancies and deregulated expression both in epithelial and cancer associated stromal cells have important consequences in CRC2,3,5. Moreover miRNAs are stable to extraction and analysis from conventional archival formalin-fixed paraffin-embedded tissue, making them powerful biomarkers for diagnosis and prognostication on human tissue subjected to conventional processing in routine histopathological laboratories9,10. However, our methodology could very easily be adapted for genomic and proteomic analysis, further emphasizing the versatility of this approach.
The authors have nothing to disclose.
MB is supported by grant funding from an MRC fellowship. KP and AHM are supported by grant funding from Wessex Medical Research and Cancer Research UK/RCS (England) (C28503/A10013). We are thankful for the support of the University of Southampton Histochemistry Research Unit.
Colorectal cancer cell lines – example shown SW480 | ATCC | ATCC CCL-228 | |
Collagen | BD Biosciences | 354265 | |
Matrigel | BD Biosciences | 354234 | |
Nylon membrane | Merck Millipore | VVLP01300 | |
Metal grid | The Mesh Company | WSS20-A4 | themeshcompany.com |
Laser microdissection platform | Leica Microsystems | Leica AS LMD | |
Membrane mounted slides | Molecular devices | ||
Cresyl Violet | Merck Millipore | 1052350025 |