Background: Metastasis is a complex process which is difficult to study and model. Experimental ingenuity is therefore essential when seeking to elucidate the biological mechanisms involved. Typically, in vitro models of metastasis have been overly simplistic, lacking the characteristic elements of the tumour microenvironment, whereas in vivo models are expensive, requiring specialist resources. Here we propose a pipeline approach for the study of cell migration and colonization, two critical steps in the metastatic cascade. Methods: We used a range of extracellular matrix derived contexts to facilitate a progressive approach to the observation and quantification of cell behaviour in 2D, 3D and at border zones between dimensions. At the simplest level, cells were set onto collagen-coated plastic or encapsulated within a collagen matrix. To enhance this, a collagen compression technique provided a stiffened, denser substrate which could be used as a 2D surface or to encapsulate cells. Decellularized tissue from the chorioallantoic membrane of the developing chicken embryo was used to provide a more structured, biologically relevant extracellular matrix-based context in which cell behaviour could then be compared with its in vivo counterpart. Results: Cell behaviour could be observed and quantified within each context using standard laboratory techniques of microscopy and immunostaining, affording the opportunity for comparison and contrast of behaviour across the whole range of contexts. In particular, the temporal constraints of the in vivo CAM were removed when cells were cultured on the decellularized CAM, allowing for much longer-term cell colonization and cell-cell interaction. Conclusions: Together the assays within this pipeline provide the opportunity for the study of cell behaviour in a replicable way across multiple environments. The assays can be set up and analysed using easily available resources and standard laboratory equipment. We believe this offers the potential for the detailed study of cell migration and colonization of tissue, essential steps in the metastatic cascade. Also, we propose that the pipeline could be used in the wider arena of cell culture in general with the increasingly more complex contexts allowing cell behaviours and interactions to be explored in a stepwise fashion in an integrated way. Keywords: Metastasis, CAM, 3D culture, Decellularization, 3D model, Extracellular matrix * Correspondence: firstname.lastname@example.org Cell Migration Lab, School of Biological Sciences, University of Reading, Reading RG6 6UB, UK Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Keeton et al. BMC Cancer (2018) 18:622 Page 2 of 12 Background 3D culture Metastasis is the leading cause of cancer-related death Rat Tail Type I Collagen (BD Bioscience) was used for and has as such been an important area of investigation the construction of collagen gels, diluted and adjusted to into the mechanisms and processes involved. Metastasis pH 7.5 according to the manufacturer’s instructions. is, however, a complex, multi-stage process which due to Simple collagen gels were set onto tissue culture plastic its temporal and unpredictable nature is difficult to with cells seeded either over the surface or encapsulated study. The development of suitable models to elucidate within. Where fibronectin was incorporated into the col- the mechanisms and processes involved has been chal- lagen gel, human recombinant fibronectin was added to lenging due to the inherent nature of the process. the collagen mix at a final concentration of 10 μg/ml. The outward spread of cancer from a tumour occurs in Compressed collagen discs were prepared using collagen several stages: cellular escape, invasion, intravasation, ex- at 2 mg/ml set in a 24-well culture dish then compressed travasation, seeding and colonization at distant sites [1, 2]. between two fine nylon mesh layers bounded by layers Both in vitro and in vivo experimental models have been of filter paper and glass plates then weighted to 126 g used to gain further insight into metastatic mechanisms. for 2.5 or 5 min . Compressed collagen discs were ei- Until recently in the main, in vitro models have been rela- ther left to free-float bathed in medium or set into a tively simple, using extracellular matrix (ECM) components 1 mg/ml collagen gel. Where cells were encapsulated and tissue culture approaches to investigate the escape and into compressed collagen, a copper grid (1.7 mm migration of cells [3–5]. With a growing emphasis on the hole-size) was used above the nylon mesh layer to avoid importance of the tumour microenvironment, it has be- cells being crushed during the compression step. come clear that both structural and cellular components of thetissuearchitectureplayacrucialroleinthemetastatic Chick Chorioallantoic membrane (CAM) assay process [6–8]. Models of metastasis, therefore, need to re- The chicken egg chorioallantoic membrane is highly vas- flect the complexity of the tumour microenvironment, the cularized, comprising three layers which together pro- conduits involved in the metastatic processes and the tissue vide an interface between the developing chick embryo architecture and features of sites of metastatic seeding and and shell, allowing gas and calcium exchange. The stro- colonization. Better models will enable not only a more de- mal and epithelial components of the CAM provide an tailed understanding of the processes involved but also pro- ECM similar to human epithelia rendering it suitable for vide improved opportunities for the testing of candidate the in vivo exploration of cell migration [16, 17]. molecules before drugs trials. Fertilized eggs were obtained from Henry Stewart, and Advances in the manufacture and use of biomaterials Co. Ltd., allowed to settle overnight in a holding incuba- in the biomedical field have led to a range of materials tor at 19 °C then placed in a humidified incubator at and approaches that are potentially available for the cul- 37 °C (Day 0). For live assays, eggs were windowed at 2– ture of cells in more relevant biological contexts [9–11]. 3 days by dropping the level of the albumen using a syr- The development of tissue engineering approaches using inge and needle then making a small window in the egg- patient-derived material now also provides the oppor- shell. After 7 days of incubation, CAM invasion assays tunity to generate more natural and complex materials were conducted by seeding permanently transfected as substrates for 3D cell culture and the study of disease. HT1080 or MDA-MB-231 cells expressing Green Fluor- Based on a biomaterials approach, here we propose a escent Protein (GFP) directly onto the CAM surface in a set of in vitro assays of increasing complexity which 1 mg/ml collagen solution (Rat tail Collagen Type 1, BD were used in comparison with a well characterized in Bioscience). Eggs were harvested at different time points vivo assay, the chicken chorioallantoic membrane up to Day 14 of incubation. Live CAM images were (CAM) assay, for the study of cell migration and taken using a Leica MZFLiii stereo microscope with colonization in a 3D tissue context [12–14]. DC500 camera and × 1 Leica lens with × 10 zoom, at room temperature. Harvested whole CAM was fixed and Methods stained with Phalloidin Atto 565 (Sigma), DAPI (Sigma), Cell culture α-rabbit Ki67 (abcam 16,667), α-GFP antibody: GFP HT1080 human fibrosarcoma cells, MCF-7 human rabbit IgG (A1112 Invitrogen) and secondary antibodies: breast cancer cells, MDA-MB-231 human breast cancer Alexafluor 488 and Alexafluor 647 (Life Sciences). Tis- cells and SK-MEL-28 human melanoma cells (HPA sue was embedded in OCT (Fisher) and sectioned using ECCC) were routinely cultured and passaged in Greiner a Kryostat (Bright Model OTF). Bio-one flasks placed in a humidified incubator at 37 °C/ 5% CO , in DMEM (low glucose with glutamine) supple- Decellularized CAM (dCAM) mented with 10% Fetal Bovine Serum and 1% Penicillin/ CAM was decellularized using an adapted protocol Streptomycin (Gibco). based on that described by Medberry . Briefly: CAM Keeton et al. BMC Cancer (2018) 18:622 Page 3 of 12 harvested at Day 9/10 of incubation was flash frozen in /0.17, × 10 Zeiss A Plan 0.25 Ph1 lens and a Zeiss LDA liquid nitrogen, thawed in ddH O at 4 °C for 30 min, Plan × 20/0.35 Ph1 ∞ /1.0 (PS). A Leica DMi8 with drained then stirred for 5 min at 37 °C in 0.02% trypsin/ DF33000G camera, moveable platform and onstage STR 0.05% EDTA (Gibco, Sigma). Tissue was washed in Tokai HIT incubator was used to take individual images ddH O then exposed to the following reagents, with a of live cells using phase contrast microscopy with × 4/ ddH O wash step between each: 3% TritonX-100 for 0.10 PH0 HI PLAN or × 10/0.25 PH1 N PLAN lenses. 5-10 min, 1 M sucrose for 5 min, 4% deoxycholate (Sigma) for 5 min, 0.1% peracetic acid/4% ethanol Statistics (Sigma/Fisher) for 5–15 min, ddH O for 5 min. dCAM A two-way ANOVA with Tukey’s test for multiple com- was then freeze-dried. For cell culture, dCAM was ex- parisons or a non-parametric test with Dunn’s test for posed to UV radiation for 20 min then soaked for at multiple comparisons was conducted using GraphPad least 24 h in PBS in a tissue culture incubator at 37 °C/ Prism 6, dependent on the data distribution. 5% CO . PBS was replaced with DMEM/ 10% FBS/ 1% Pen/Strep and replaced in the incubator for 48 h. The Results medium was aspirated, and cells re-suspended at high Cell morphology and migration speed differs with density were seeded at low volumes, typically 0.5 ml at dimension 1×10 , left to adhere (2–4 h) then additional medium HT1080 fibrosarcoma cells and MDA-MB-231 breast can- added. Samples were prepared for mass spectrometry by cer cells displayed different migratory characteristics when solubilizing dCAM for 3 days according to the Medberry moving on as opposed to in a simple collagen-based con- protocol. Centrifugation to pellet undissolved particles text. When encapsulated in a collagen gel, MDA-MB-231 was conducted, and the pellet was re-suspended in cells adopted a more compact morphology (Fig. 1a)than DMSO. Both supernatant and DMSO were diluted in thosemigrating over thesurfaceofacollagengel (Fig. 1b). formic acid to a final concentration of 0.1%. Mass spec- HT1080 cells, however, became less spread and more elon- trometry was conducted by the Functional Genomics gatedwhenencapsulatedwithin a collagen gel (Fig. 1c, d). and Proteomics Facility at the University of Birmingham Cell aspect ratio (cell length to width) was used to quantify using ORBITRAP MS with CID fragmentation. these morphological differences which were found to be significant when compared with cells moving in 2D and 3D Microscopy for each cell type, (Fig. 1e). Cell migration speed in 3D was A Nikon TiE fitted with a DS-Fi2 camera, Plan × 10/0.25 significantly reduced for MDA-MB-231 cells as the collagen Ph1 DL and Plan Fluor EL WD × 20/0.45 Ph1 DM ∞/0.2 increased from 1 mg/ml to 2 mg/ml (mean values in μm/ WD 7.4 lenses, an environmental chamber and a moveable minute: 0.27, 0.20, difference 0.07, n =3), Fig. 1f. For platform stage (Prior Scientific) was used in conjunction HT1080 cells the migration speed was significantly faster in with NIS Elements software for time-lapse microscopy. the 3D matrix at 1 mg/ml (mean values in μm/minute: Image analysis was conducted using ImageJ MTrackJ plu- 0.25, 0.38, difference 0.13, n = 3) but reduced when the gin (ImageScience) and FIJI ImageJ software . Live im- matrix density was increased from 1 mg/ml to 2 mg/ml aging of chicken embryo and CAM were obtained using a (mean values in μm/minute: 0.38, 0.32, difference = 0.06). Leica MZFLiii stereo microscope with DC500 camera and However, there was a significant difference in migration × 1 Leica lens with × 10 zoom, at room temperature. Laser speed between 2D and 3D conditions for both collagen scanning confocal microscopy was conducted using either concentrations (Fig. 1f). A cell migration assay which pro- a Nikon A1 Plus or A1-R microscope at room temperature, vided both 2D and 3D environments for cells to move on, usinga ×20PlanApo VC ×20 DIC NR,NA 0.75lensand over or into, allowed cell migration to be tracked and ana- ×601.40PlanApo ∞/0.17 WD 0.13, NA 1.4 lens. Images lysed as cells moved within and between different environ- were acquired and prepared using NIS Elements, ImageJ ments. For both MDA-MB-231 (Fig. 1g, h) and HT1080 and/or Photoshop CS6 Extended. Reflectance microscopy cells (Fig. 1i, j) cell migration was slower when cells moved was conducted using a Leica TCS SP2 confocal microscope across and/or into the 3D matrix (MDA-MB-231 mean at room temperature with a Leica HCX PlanApo lbd.BL × values in μm/minute: 2D, 0.53, Border, 0.65, 3D 0.29 and 63 NA 1.4 oil immersion lens. Scanning Electron Micros- for HT1080: 0.25, 0.39 and 0.13 μm/minute respectively, n copy (SEM) was conducted for gold sputter coated samples = 3). At border zones, however, cells moved at greater speed (Edwards S150b) using a Quanta FEI 600F. migrating up and down the borders and appearing to use A Zeiss Axio Vert.A1 epifluorescence microscope with them as 1D migration tracks as well as transition zones. an inverted lens and moveable platform was used to take The results from these in vitro assays demonstrated the individual images of live cells to monitor experiment need for the cellular response to its surrounding environ- progress and check for fluorescent protein expression. ment to be considered when studying cell characteristics Lenses used were: × 5 Planar Plan Neofl Ph1 0.15 ∞ and behaviour. Keeton et al. BMC Cancer (2018) 18:622 Page 4 of 12 Fig. 1 Cells adopt different morphologies and migration characteristics in 2D compared to 3D. MDA-MB-231 (MDA) cells adopted a more compact morphology when migrating in collagen (b) than over a collagen coated surface (a). However, HT1080 cells migrating in collagen were more elongated (d) than when migrating over it (c). Microscopy images were taken using a Nikon TiE phase contrast microscope and DS-Fi2 camera, a moveable stage and environmental chamber set at 37 °C with a continuous CO /O supply. NIS Elements software was used for image 2 2 capture and a Plan × 10/0.25 Ph1 DL lens; scale bars = 50 μm. The differences in aspect ratio are quantified in e where MDA-MB-231 and HT1080 cells are compared both to and in collagen (1 mg/ml). Cell migration speed is compared for cells moving over 2D collagen (2D) compared with cells moving in either 1 mg/ml or 2 mg/ml collagen for both MDA-MB-231 and HT1080 cells in f. Non-parametric Kruskal-Wallis Test with Dunn’s test for multiple comparisons was run for each condition. MDA-MB-231 and HT1080 cells were set up in a 2D/3D assay, and migration speed investigated in three different regions created: 2D, border, 3D. g and h show MDA cells behave differently according to their location and context as do HT1080 cells shown in i and j (2D, two dimensions, 3D, three-dimensional context, B, Border zone between the two contexts). Images show static shots taken from time-lapse movies, scale bars = 100 μm. Statistics were generated using a Two-way ANOVA with Tukey’s multiple comparisons tests using GraphPad Prism 6. Significance is shown: ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001. n = 3 A stiffer more biologically relevant multi-dimensional of cell invasion, migration and colonization. In setting context a compressed collagen disc into a thin layer of un- Using a compressed collagen technique developed for compressed collagen, a multi-dimensional heteroge- stem cell differentiation in biomaterials engineering neous 3D environment was created with multiple , the 2D/3D assay was developed to provide a border zones. Reflectance imaging of the collagen stiffer, more biologically relevant context for the study structure of the simple context (Fig. 2a)compared Keeton et al. BMC Cancer (2018) 18:622 Page 5 of 12 Fig. 2 Compressed Collagen Assay provides a stiffer, more structured growth environment. A compressed collagen assay provides a stiffer, more structured growth environment for cell culture facilitating a greater range of cell morphology in colonizing cells. a and b show the collagen matrix in more detail via reflectance microscopy, a showing uncompressed collagen at 2 mg/ml and b compressed collagen derived from 2 mg/ ml gel. (Image was taken using Leica TCS SP2 using 488 argon laser and Leica HCX PlanApo lbd.BL × 63 NA 1.4 oil immersion lens, at room temperature, scale bars = 20 μm). Cells seeded on compressed collagen (CC) set into a non-compressed collagen matrix (LC) colonized the compressed collagen in preference to migrating away from it. c and d show MDA-MB-231 cells colonizing the compressed collagen set into a lower density 1 mg/ml collagen gel. Few MDA-MB-231 cells emerged at the border zones (B) (scale bars = 100 μm). e and f show a few cells with a mainly rounded morphology which have escaped (yellow arrows) and moved away from the densely colonized compressed collagen into the lower density collagen (scale bars = 20 μm). In f emerging cells followed the contours of the compressed collagen disk (green arrows) while in e they appeared to form chains moving away from it (green arrows). Images b, c, e, f, were generated using a Nikon TiE phase contrast microscope, moveable stage, environmental chamber at 37 °C with a continuous CO /O supply, DS-Fi2 camera, lenses: Plan × 10/0.25 Ph1 DL, 2 2 Plan Fluor EL WD × 20/0.45 Ph1 DM ∞/0.2 WD 7.4. g and h show live GFP+ HT1080 cells colonizing the compressed collagen matrix. Cells had a diverse range of morphologies including rounded (yellow arrows) and elongated shapes (red arrows). Images were taken using a Zeiss Axio Vert.A1, DS-Fi2 camera, a Zeiss LDA Plan × 20/0.35 Ph1 ∞/1.0 (PS) lens at room temperature, scale bars = 100 μm with the structural composition of the compressed both individually (Fig. 2e, f) and collectively (Fig. 2f) collagen (Fig. 2b) showed that the compressed colla- within these contexts. Colonization of compressed gen comprised a denser network of aligned collagen collagen could be visualized more clearly using per- fibres in comparison to the short more random manently transduced HT1080 GFP+ cells (Fig. 2g, h). arrangement of the uncompressed collagen context The HT1080 GFP+ cells in the live images show that (Fig. 2a). Cells seeded onto the compressed collagen cells were able to adopt a range of morphologies from migrated over and colonized the stiffer matrix before spherical to elongated on the compressed collagen. migrating out onto the less dense collagen (Fig. 2c The development of this more complex in vitro assay and d). Cells were also seen to move and transition in which the entire range of cell morphologies adopted with a range of morphologies, including elongated in vivo was observed, demonstrated the need for a stiffer mesenchymal cell migration and spherical, compact more complex environment to support cell culture and morphologies (Fig. 2e, f). Cells appeared to move the investigation of cell migration and colonization. Keeton et al. BMC Cancer (2018) 18:622 Page 6 of 12 CAM as an in vivo model for the study of cell migration dCAM as a 3D context for the investigation of cell and colonization behaviour The chicken egg chorioallantoic membrane (CAM) model, The decellularized CAM provided a simple and easy to use which has been well characterized and used in develop- substrate upon which cancer cells could be seeded. Three mental biology and the investigation of angiogenesis, was different cell lines were used: MCF-7, MDA-MB-231 and explored as an in vivo model for the investigation of meta- HT1080 cells. These were seeded and allowed to proliferate static mechanisms, in particular, those of cell migration and as either a monoculture (Fig. 5b) or as a co-culture (Fig. colonization [16, 20–22]. Following a similar approach to 5a). Populated dCAM was fixed and stained, and 3D im- theinvitro assays,GFP+HT1080orMDA-MB-231cells ages obtained using regular confocal imaging without sec- suspended in a collagen gel were seeded directly onto the tioning, allowing cell-cell and cell-matrix interactions to be CAM surface. Outward migration of cells was observed at visualized in intact tissue. Ki67 staining for cell proliferation suitably chosen time points (Fig. 3a-c)and fixed, and in HT1080 cells cultured on dCAM (Fig. 5c) showed that stained CAM was probed to examine the extent of invasion cells were at different stages in the cell cycle while the and morphology of cells located within the CAM tissue dCAM was being colonized. Comparative Ki67 staining in (Fig. 3d-i). As the CAM is relatively thin (typically seeded CAM (Fig. 5d) showed just a few human cells prolif- 30-100 μm), it was possible to use confocal microscopy to erating amongst the chick cells of the CAM. visualize and examine the spread of cells over and into intact CAM tissue as shown in Fig. 3d. Stained and sec- Testing the pipeline approach tioned CAM was used to examine the timescales of inva- SK-MEL-28 melanoma cells were introduced into each of sion (Fig. 3g-i) and the specifics of cell morphology and cell the in vitro assays of the proposed pipeline: 2D/3D assay, interaction with the surrounding tissue. 3D encapsulation in either collagen or collagen supple- The complexity of the CAM tissue, however, limited mented with fibronectin, Compressed Collagen (CC) or the options for probing exogenous cell properties and Compressed Collagen with fibronectin (CCF) and dCAM interactions. The short developmental timescales of the (Fig. 6). The melanoma cells adopted an elongated morph- chicken embryo model provided only a narrow window ology at border zones, on collagen (Fig. 6a, b) or on colla- of opportunity for cell migration and colonization. How- gen supplemented with fibronectin (Fig. 6c, d)used attwo ever, if an acellular tissue structure could be developed different concentrations (1 mg/ml and 2 mg/ml collagen). offering the benefits of the complex ECM structure pro- However, when encapsulated within collagen gels or colla- vided by the CAM without the complication of the chick gen gels supplemented with fibronectin, cell colonies cells, then this could be used as a platform for cell cul- within the denser matrix (Fig. 6i, j) showed a more com- ture over longer time periods. pact arrangement compared with those observed in the lower density matrix (Fig. 6g, h). SK-MEL-28 cells encap- Decellularized CAM (dCAM) as a 3D growth substrate sulated in compressed collagen or compressed collagen CAM harvested from developing chicken embryos was supplemented with fibronectin were seen to partially decellularized and characterized to assess its suitability populate the stiffened collagen matrix before invading into as an ECM based growth matrix for cell culture and the the surrounding lower density collagen matrix. Cells es- further exploration of cell migration and colonization. caping the compressed collagen adopted an elongated As the membranes are thin and delicate, careful morphology with filopodia extending out into the lower optimization was necessary to ensure that minimal dam- density gel matrix (Fig. 6e, f). When colonizing the decel- age was caused while cells were removed. Phalloidin and lularized chorioallantoic membrane (dCAM), melanoma DAPI staining used for whole CAM (Fig. 3d-i) showed cells formed multiple layers, quite unlike the behaviour that no remaining cell cytoskeleton material or nuclei displayed when they were cultured on 2D. Ki67 staining remained following decellularization (Fig. 4a-d). Scan- indicated that the melanoma cells were actively dividing ning electron microscopy enabled the decellularized within each of the layers observed. CAM surface to be visualized in detail. The images showed that vasculature and acellular surfaces were pre- Discussion served (Fig. 4e-f). Initial results for comparative mass Recent studies focussing on the progression of cancer have spectrometry of CAM versus dCAM showed that foetal highlighted the importance of the tumour microenviron- CAM proteins present in whole CAM had been re- ment in both preventing and facilitating the outward spread moved during production of the decellularized CAM of cancer [7, 23, 24]. In vitro models of metastasis have typ- (Table 1 and Fig. 3). Used as a growth matrix, cells were ically been simple and have lacked the complexity and seeded onto small sections of dCAM in tissue culture structure of the tissue environment, whereas in vivo models dishes and were observed to adhere and proliferate on have been expensive, difficult to set up and limited in their the dCAM (Fig. 4g, h). application. While any model has inherent limitations, a Keeton et al. BMC Cancer (2018) 18:622 Page 7 of 12 Fig. 3 Cell migration and invasion can be explored by seeding GFP+ cells onto CAM. Cells re-suspended in 1 mg/ml collagen were seeded onto the surface of live CAM. a, bright field (BV = blood vessel) and b, c, epifluorescence images of live CAM with cells seeded over the surface (Leica MZFLiii stereo microscope with DC500 camera and × 1 Leica lens with × 10 zoom, at room temperature, scale bars = 1 mm). d, HT1080 GFP+ cells dispersed over fixed and stained whole CAM, scale bar = 100 μm. e, f Invaded HT1080 GFP+ cells show different morphologies in fixed and sectioned CAM. g, h, i: a timeline for the invasion of MCF7 GFP+ cells seeded onto CAM shows that cell invasion was evident 1 day after cell seeding and by day 5, cells could be seen within the vasculature and were well disseminated within CAM tissue (scale bars = 25 μm, image planes marked XY, XZ, YZ). Images D-I were taken using Nikon A1 plus confocal microscope, at room temperature. Image D was taken using a × 20 Plan Apo DIC N2, 0.75 NA lens. Image E-I were taken using a × 60 Plan Apo ∞/0.17 WD 0.13, NA 1.40 lens complex 3D tissue culture model representative of the ap- pipeline for the comparison and contrast of cell behav- propriate native environment which can be manipulated iours in increasingly complex ECM based 3D environ- and controlled under experimental conditions would pro- ments. This pipeline approach is illustrated in the model vide a good platform for the study of cellular and molecular shown in Fig. 7. In the 2D/3D assay, the simplest of the mechanisms [11, 25]. pipeline assays, a range of cell behaviours and morph- There has been much emphasis on the extracellular ologies could be observed and quantified at different matrix as a transitory layer and essential conduit for mi- contextual locations within the same assay. This com- grating cells, as well as being a contributor to the bines conventional approaches to cell migration in tumour microenvironment [8, 26, 27]. Extracellular which cell behaviours can be observed in both 2D and matrix materials have therefore proved a popular start- 3D and additionally introduces a border zone at which ing point for much of the recent research in this area. cell transition between 2D and 3D contexts can be ob- Using collagen, the main constituent of ECM as a start- served. Pleomorphic cell behaviour observed in this ing point, we have developed a set of assays which build assay demonstrated the adaptability of cells to a simple on the existing assays used in the field to provide a context with only limited variability in surface and Keeton et al. BMC Cancer (2018) 18:622 Page 8 of 12 Fig. 4 dCAM provides a collagen-rich 3D substrate for cell culture. Laser scanning confocal spectral unmixing was used to determine the residual components after decellularization of CAM (Nikon A1 Plus at room temperature using a × 60 1.40 Plan Apo ∞/0.17 WD 0.13, NA 1.4 lens). a, combined image, b DAPI only, c CAM background only, d, phalloidin for cellular actin cytoskeleton (scale bar for A = 20 μm). Scanning electron microscopy (SEM) was used to characterize the surface of the decellularized tissue (Quanta FEI), e and f show dCAM surface features including vasculature (yellow arrows) and fibrous extracellular matrix (scale bars: E = 50 μm, F = 5 μm). dCAM used as a growth matrix: g shows a bright field image of dCAM during colonization and h shows MDA-MB-231 GFP+ cells adhering and proliferating over the dCAM (DC, yellow arrows). Images G and H were taken using a Zeiss Axio Vert inverted epifluorescence microscope and × 5 Planar Plan Neofl Ph1 0.15 ∞ /0.17 lens with the DS-Fi2 camera, operating at room temperature, scale bars = 1 mm Table 1 Characterization of solubilized CAM and dCAM using mass spectrometry Id. Description Unique Protein peptides Coverage CAM 1 P84407 Alpha-fetoprotein OS = Gallus gallus GN = AFP PE = 1 SV = 1 - [FETA_CHICK] 3 11.57 1 Q98UI9 Mucin-5B OS = Gallus gallus GN = MUC5B PE = 1 SV = 1 - [MUC5B_CHICK] 3 1.94 2 P01012 Ovalbumin OS = Gallus gallus GN=SERPINB14 PE = 1 SV = 2 - [OVAL_CHICK] 6 18.39 2 P02112 Hemoglobin subunit beta OS = Gallus gallus GN=HBB PE = 1 SV = 2 - [HBB_CHICK] 2 21.09 2 P00698 Lysozyme C OS = Gallus gallus GN = LYZ PE = 1 SV = 1 - [LYSC_CHICK] 3 33.33 2 O93532 Keratin, type II cytoskeletal cochleal OS = Gallus gallus PE = 2 SV = 1 - [K2CO_CHICK] 2 3.86 2 P01013 Ovalbumin-related protein X (Fragment) OS = Gallus gallus GN=SERPINB14C PE = 3 SV = 1 - 2 19.4 [OVALX_CHICK] dCAM 1 Q90617 Lysosome-associated membrane glycoprotein 2 OS = Gallus gallus GN = LAMP2 PE = 2 SV = 1 - 2 6.12 [LAMP2_CHICK] 1 P11722 Fibronectin (Fragments) OS = Gallus gallus GN=FN1 PE = 2 SV = 3 - [FINC_CHICK] 2 3.5 1 P02112 Hemoglobin subunit beta OS = Gallus gallus GN=HBB PE = 1 SV = 2 - [HBB_CHICK] 2 21.09 1 P02467 Collagen alpha-2(I) chain (Fragments) OS = Gallus gallus GN=COL1A2 PE = 1 SV = 2 - [CO1A2_CHICK] 2 1.62 2 P02112 Hemoglobin subunit beta OS = Gallus gallus GN=HBB PE = 1 SV = 2 - [HBB_CHICK] 2 21.09 2 P11722 Fibronectin (Fragments) OS = Gallus gallus GN=FN1 PE = 2 SV = 3 - [FINC_CHICK] 2 3.5 Key: 1, Supernatant; 2, Pellet Keeton et al. BMC Cancer (2018) 18:622 Page 9 of 12 Fig. 5 dCAM provides a structured 3D environment for studying cell proliferation and migration. a, dCAM partially populated with a co-culture of MDA-MB-231 (white arrows) and MCF7 GFP+ (yellow arrows) breast cancer cells stained with phalloidin for actin cytoskeleton (red) and DAPI nuclear stain (blue). b, MDA-MB-231 cells stained with phalloidin (red) and DAPI (blue) appear to have formed layers over the dCAM surface. c, Cells stained with cell proliferation marker Ki67 (Alexafluor 488, green), phalloidin (red), DAPI (blue) on dCAM. Differential Ki67 staining suggests that not all cells were actively proliferating (proliferating cells – white arrows, high Ki67, low Ki67 cells indicated with yellow arrows). d.1-d.4, Ki67 staining of human cells proliferating and migrating amongst chick CAM cells in invaded CAM (section): combined channels D.1, DAPI, blue (D.2); Phalloidin, red (D.3); Ki67/Alexafluor 647, white (D.4). Images were taken using Nikon A1R confocal microscope operating at room temperature, Image A using a × 20 Plan Apo VC DIC NR, NA 0.75 lens and Images B-D using a × 60 1.40 Plan Apo ∞/0.17 WD 0.13, NA 1.4 lens. Scale bars in C, D=20 μm. Image planes for 3D images in C and D are marked XY, YZ, XZ constituents. The second and more complex assay de- dense collagen environment which could be used to fur- scribed here, the compressed collagen assay (CC), pro- ther explore invasive and migratory behaviour in a flex- vided a stiffer and more elastic context for cell study, ible manner. These collagen-based assays were further with the added benefit of multiple regions: the stiff com- augmented with fibronectin when testing the pipeline, to pressed collagen, two different border zones, a simpler demonstrate that additional ECM constituents could be collagen matrix and a two-dimensional planar surface. introduced to further explore cell behaviours. Colonization within this assay took place over a much The chick-derived decellularized ECM (dCAM), pro- longer period than was possible in either a simple colla- vided a still more complex 3D context with the natural gen context or a 2D monolayer. It was possible to ob- features and variation of an in vivo environment. Seeded serve and quantify both cell morphology and migration cells were able to divide and colonize the ECM environ- behaviours in this more complex environment, one ment over a longer time period, extending the potential which not only facilitated the extended observation of culture time to weeks rather than days. Co-culture was cell-cell and cell-ECM interactions but enabled a variety also supported with cell types being introduced at differ- of cell behaviours to emerge. In this context, the cells ent time points within the tissue culture program, and adopted a range of morphologies more closely resem- cell-cell interactions and contribution within the 3D en- bling those seen in vivo. Following the recent develop- vironment observed. Variability in staining for the prolif- ment of a thin high-density fibrillary collagen layer for eration marker Ki67 suggested that while some cells the study of proteolytic invasion  the compressed were actively proliferating, others may have become qui- collagen assay developed here provides an alternative escent indicating that cells may be able to differentiate Keeton et al. BMC Cancer (2018) 18:622 Page 10 of 12 Fig. 6 Using the pipeline to characterize the escape and colonization of SK-MEL-28 melanoma cells. Melanoma cells were introduced into four pipeline assays, providing the opportunity to compare and contrast cell behaviour in each context. In the 2D/3D assay, melanoma cells adopted an elongated morphology on 2D plastic, at border zones (white B) and on top of the 3D matrix. Collagen was used at two different concentrations: a, 1 mg/ml and c, 2 mg/ml and with 10 μg/ml fibronectin: b, 1 mg/ml collagen + fibronectin and d, 2 mg/ml collagen + fibronectin. When melanoma cells were encapsulated in collagen (g, 1 mg/ml, h 2 mg/ml) or collagen with 10 μg/ml fibronectin (i, 1 mg/ml + fibronectin, j, 2 mg/ml + fibronectin), they proliferated to form small tight colonies in the 2 mg/ml gels and looser spread structures at 1 mg/ml. In a 2 mg/ml compressed collagen matrix, e1 - e2, melanoma cells partially colonized the matrix (blue arrows show uncolonized matrix) before escaping into the surrounding lower density matrix (LC). Melanoma cells within the compressed collagen (CC) adopted ovoid groupings (yellow arrows) cells becoming elongated with narrow filopodia upon escape (green arrows). Similar behaviours were observed in compressed collagen with fibronectin (CCF), f1 - f2. Melanoma cells seeded onto dCAM, cultured for 10 days proliferated to form layers. Ki67 staining (green) indicated that cells were actively dividing within each layer – white arrows. Phalloidin – red. Dapi – blue. Images a, b, e1–2, f1–2 were generated using Nikon TiE timelapse system and Plan × 10/0,25 Ph1 DL lens and NIS Nikon Elements software. Scale bars = 100 μm. Images c, d, g-j were generated using a Leica DMi8 inverted microscope, Leica DF33000G camera, Tokai Hit STR stage top incubator, ND × 4/0.10 PH0 HI PLAN and × 10/0.25 PH1 N PLAN achromatic objective lenses. Scale bars = 50 μm. Image k was generated using Nikon A1-R confocal microscope with a × 60 1.40 Plan Apo ∞/0.17 WD 0.13, NA 1.4 lens. Scale bar = 25 μm and settle in the dCAM environment, as opposed to the substrate in a controlled in vitro environment. A signifi- continuous cell division typically observed during cell cant advantage of using the decellularized tissue as a culture on 2D surfaces. Behaviours seen within the live substrate for the study of cell interactions and behaviour CAM model were explored further within dCAM pro- in 3D was that imaging could be conducted of cells in viding the opportunity to move between a complex live situ, either live or fixed but without sectioning, thus model and a structured biologically relevant culture allowing populated tissue to remain intact and therefore Keeton et al. BMC Cancer (2018) 18:622 Page 11 of 12 Fig. 7 A Pipeline Approach to the Experimental Modelling of Metastasis. The models discussed and presented in this paper are shown in order of increasing complexity: 2D, two-dimensional cell migration assay; 3D, three-dimensional cell migration assay; 2D/3D assay; CC, compressed collagen assay; dCAM, decellularized chorioallantoic membrane assay; CAM, chorioallantoic membrane assay. The structural features are listed along with the type of study that the model is suitable for and the time frame for its use minimising the introduction of artefacts. In this way cell preserving the structural integrity of each cultured tissue morphology in both CAM and dCAM could be identi- environment. The inherent flexibility of the models pro- fied and quantified in a similar way to that carried out in vides the opportunity for the manipulation of experimen- the simple 2D/3D assay. Finally, the introduction of mel- tal conditions across and within the assays so that cell anoma cells into each of the in vitro assays of the pipe- behaviours in each context can be compared and con- line demonstrated the differential response of cells to trasted under different experimental conditions. Thus a each environment in this context based approach. pipeline approach can be adopted for the elucidation of We feel this pipeline approach provides a robust set of as- cell migration and colonization, important steps in the says which can be used to explore the escape, invasion and metastatic process. The testing of drug compounds would colonization steps of metastasis. During assay development also benefit from such a pipeline approach employing a we have introduced a variety of cancer cell types including range of versatile tissue culture tools to explore and eluci- cells from a primary sarcoma (HT1080), breast cancer cells date the mechanisms and effects of drug interactions in which are oestrogen positive and considered non-invasive increasingly complex 3D contexts which can be manipu- (MCF-7) and triple-negative breast cancer cells which are lated and then reproducibly replicated. metastatic and highly invasive (MDA-MB-231). We have Abbreviations further tested the pipeline with melanoma cells originating 1D: One-dimensional; 2D: Two-dimensional; 3D: Three-dimensional; from skin carcinoma (SK-MEL-28). CAM: Chick chorioallantoic membrane; CC: Compressed collagen; CCF: Compressed collagen with fibronectin; dCAM: Decellularized chick chorioallantoic membrane; ECM: Extracellular matrix; SEM: Scanning electron Conclusions microscopy Taken together these ECM-based assays provide an op- portunity to study cell interactions and behaviours in con- Acknowledgements texts of increasing complexity, with the ability to observe, We thank Amanpreet Kaur, EM Lab, University of Reading, for technical record and quantify cell behaviours and events while assistance with Scanning Electron Microscopy. Keeton et al. BMC Cancer (2018) 18:622 Page 12 of 12 Funding 12. Cheng CW, Solorio LD, Alsberg E. Decellularized tissue and cell-derived This work was funded by the BBSRC (Biotechnology and Biological Sciences extracellular matrices as scaffolds for orthopaedic tissue engineering. Research Council) grant number BB/J012580/1. Biotechnol Adv. 2014;32(2):462–84. 13. Mazza G, Rombouts K, Rennie Hall A, Urbani L, Vinh Luong T, Al-Akkad W, Longato L, Brown D, Maghsoudlou P, Dhillon AP, et al. Decellularized Availability of data and materials human liver as a natural 3D-scaffold for liver bioengineering and The datasets used and analysed during this study are available from the transplantation. Sci Rep. 2015;5:13079. corresponding author on reasonable request. 14. Theodoridis K, Tudorache I, Calistru A, Cebotari S, Meyer T, Sarikouch S, Bara C, Brehm R, Haverich A, Hilfiker A. Successful matrix guided tissue Authors’ contributions regeneration of decellularized pulmonary heart valve allografts in elderly SK was responsible for the concept and design of the pipeline, for carrying sheep. Biomaterials. 2015;52:221–8. out experiments, for the collection of data and preparation of the 15. Jones RR, Hamley IW, Connon CJ. Ex vivo expansion of limbal stem cells is manuscript. JMD and AB were responsible for the design and development affected by substrate properties. Stem Cell Res. 2012;8(3):403–9. of the CAM assay within the pipeline and assistance with the preparation 16. Lokman NA, Elder AS, Ricciardelli C, Oehler MK. Chick Chorioallantoic and review of the manuscript. MC was involved in the design and membrane (CAM) assay as an in vivo model to study the effect of newly development of the decellularization process and for final review of the identified molecules on ovarian Cancer invasion and metastasis. Int J Mol manuscript. PD was responsible for the overall concept and design of the Sci. 2012;13(8):9959–70. pipeline and for preparation and review of the manuscript. All authors read 17. Nowak-Sliwinska P, Segura T, Iruela-Arispe ML. The chicken chorioallantoic and approved the final manuscript. membrane model in biology, medicine and bioengineering. Angiogenesis. 2014;17(4):779–804. Ethics approval and consent to participate 18. Medberry CJ, Crapo PM, Siu BF, Carruthers CA, Wolf MT, Nagarkar SP, Not applicable. Agrawal V, Jones KE, Kelly J, Johnson SA, et al. Hydrogels derived from central nervous system extracellular matrix. Biomaterials. 2013;34(4):1033–40. 19. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Competing interests Preibisch S, Rueden C, Saalfeld S, Schmid B, et al. Fiji: an open-source The authors declare that they have no competing interests. platform for biological-image analysis. Nat Meth. 2012;9(7):676–82. 20. Gabrielli MG, Accili D. The Chick Chorioallantoic membrane: a model of molecular, structural, and functional adaptation to Transepithelial ion Publisher’sNote transport and barrier function during embryonic development. J Biomed Springer Nature remains neutral with regard to jurisdictional claims in Biotechnol. 2010;2010:12. published maps and institutional affiliations. 21. Ribatti D. The chick embryo chorioallantoic membrane as a model for tumor biology. Exp Cell Res. 2014;328(2):314–24. Author details 1 22. Deryugina EI, Quigley JP. Chapter 2 Chick embryo Chorioallantoic Cell Migration Lab, School of Biological Sciences, University of Reading, 2 membrane models to quantify angiogenesis induced by inflammatory and Reading RG6 6UB, UK. Blizard Institute, Barts & The London School of tumor cells or purified effector. Molecules. 2008;444:21–41. Medicine & Dentistry, The Blizard Building, Centre for Immunology, Queen 3 23. Provenzano PP, Inman DR, Eliceiri KW, Knittel JG, Yan L, Rueden CT, White Mary, University of London, 4 Newark Street, London E1 2AT, UK. Natural JG, Keely PJ. Collagen density promotes mammary tumor initiation and Biosciences SA, Seestrasse 155A, Kilchberg, 8802 Zurich, Switzerland. 4 progression. BMC Med. 2008;6(1):11. Institute of Child Health, University College London, London WC1E 6BT, UK. 24. Quail DF, Joyce JA. Microenvironmental regulation of tumor progression and metastasis. Nat Med. 2013;19(11):1423–37. Received: 7 November 2017 Accepted: 18 May 2018 25. Yamada KM, Cukierman E. Modeling tissue morphogenesis and cancer in 3D. Cell. 2007;130(4):601–10. 26. Frantz C, Stewart KM, Weaver VM. The extracellular matrix at a glance. J Cell References Sci. 2010;123(24):4195–200. 1. Valastyan S, Weinberg RA. Tumor metastasis: molecular insights and 27. Hynes RO. The extracellular matrix: not just pretty fibrils. Science. 2009; evolving paradigms. Cell. 2011;147(2):275–92. 326(5957):1216–9. 2. Gupta GP, Massagué J. Cancer metastasis: building a framework. Cell. 2006; 28. Artym VV, Swatkoski S, Matsumoto K, Campbell CB, Petrie RJ, Dimitriadis EK, 127(4):679–95. Li X, Mueller SC, Bugge TH, Gucek M, et al. Dense fibrillar collagen is a 3. Wolf K, Alexander S, Schacht V, Coussens LM, von Andrian UH, van Rheenen potent inducer of invadopodia via a specific signaling network. J Cell Biol. J, Deryugina E, Friedl P. Collagen-based cell migration models in vitro and 2015;208(3):331–50. in vivo. Semin Cell Dev Biol. 2009;20(8):931–41. 4. Artym VV, Matsumoto K: Imaging cells in three-dimensional collagen matrix. Curr Protoc Cell Biol / editorial board, Juan S Bonifacino [et al] 2010, Chapter 10:Unit 10.18.11–Unit 10.18.20. 5. Kramer N, Walzl A, Unger C, Rosner M, Krupitza G, Hengstschlager M, Dolznig H. In vitro cell migration and invasion assays. Mutat Res. 2013; 752(1):10–24. 6. Joyce JA, Pollard JW. Microenvironmental regulation of metastasis. Nat Rev Cancer. 2009;9(4):239–52. 7. Lu P, Weaver VM, Werb Z. The extracellular matrix: a dynamic niche in cancer progression. J Cell Biol. 2012;196(4):395–406. 8. Nelson CM, Bissell MJ. Of extracellular matrix, scaffolds, and signaling: tissue architecture regulates development, homeostasis, and cancer. In: Annual Review of Cell and Developmental Biology, vol. 22; 2006. p. 287–309. 9. Lutolf MP, Hubbell JA. Synthetic biomaterials as instructive extracellular microenvironments for morphogenesis in tissue engineering. Nat Biotechnol. 2005;23(1):47–55. 10. Loessner D, Holzapfel BM, Clements JA. Engineered microenvironments provide new insights into ovarian and prostate cancer progression and drug responses. Adv Drug Deliv Rev. 2014;79–80:193–213. 11. Gu L, Mooney DJ. Biomaterials and emerging anticancer therapeutics: engineering the microenvironment. Nat Rev Cancer. 2016;16(1):56–66.
– Springer Journals
Published: Jun 1, 2018