Lung Adenocarcinoma Cell Responses in a 3D in Vitro Tumor

Subscriber access provided by Binghamton University | Libraries

Article

Lung adenocarcinoma cell responses in a 3D in vitro Tumor Angiogenesis Model Correlate with Metastatic Capacity Laila Christine Roudsari, Sydney Jeffs, and Jennifer L. West ACS Biomater. Sci. Eng., Just Accepted Manuscript • Publication Date (Web): 08 Jun 2017 Downloaded from http://pubs.acs.org on June 11, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Biomaterials Science & Engineering is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Biomaterials Science & Engineering

Lung adenocarcinoma cell responses in a 3D in vitro Tumor Angiogenesis Model Correlate with Metastatic Capacity Laila C. Roudsari1, Sydney E. Jeffs1, Jennifer L. West1,2 Duke University, Department of Biomedical Engineering1 and Department of Mechanical Engineering & Materials Science2

Laila C. Roudsari Address: Duke University Department of Biomedical Engineering Room 1427, Fitzpatrick CIEMAS 101 Science Drive Campus Box 90281 Durham, NC 27708-0281

Sydney E. Jeffs Address: Duke University Department of Biomedical Engineering Room 1427, Fitzpatrick CIEMAS 101 Science Drive Campus Box 90281 Durham, NC 27708-0281

Jennifer L. West Email: [email protected] Phone: (919) 660-5458 Address: Duke University Department of Biomedical Engineering Room 1427, Fitzpatrick CIEMAS 101 Science Drive Campus Box 90281 Durham, NC 27708-0281

Keywords: cancer, biomaterial, metastasis, vascular, tumor engineering, endothelial, poly(ethylene glycol)

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract Many tools from the field of tissue engineering can be used to develop novel model systems to study cancer. We have utilized biomimetic synthetic hydrogels, based on poly(ethylene glycol) (PEG) modified with cell adhesive peptides (RGDS) and peptides sensitive to degradation by matrix metalloproteinases 2 and 9 (GGGPQGIWGQGK), as highly controlled 3D substrates for cell culture. We have previously shown that this hydrogel can support growth of tumor cells and also growth and assembly of microvascular networks. Based on this technology, a 3D in vitro tumor angiogenesis model was developed using a dual layer PEG-based hydrogel comprised of vascular cells (endothelial cells, pericytes) and lung adenocarcinoma cells in separate layers to support recapitulation of the vessel recruitment process as it occurs in vivo. This model was previously used to study highly metastatic murine 344SQ cells and in this paper was used to investigate 2 additional types of lung adenocarcinoma cells: non-metastatic murine 393P cells and somewhat metastatic human A549 cells. All three cell types readily formed spheroid structures in the 3D hydrogels. When cultured in the dual layer format, where tumor cell spheroids were adjacent to a hydrogel layer with microvascular tubule networks, all three tumor cell types recruited vascular invasion into the cancer cell layer. Interactions between vessels invading the cancer layer and the cancer cell structures was nearly twice as high for the highly metastatic 344SQ cells as for the other two cell types. Secretion of angiogenic growth factors by the tumor cells was evaluated. 344SQ cells produced the greatest amount of VEGF and FGFb, which probably accounts for the greater degree of vessel recruitment observed. Upon interaction with vessel structures, the 344SQ spheroids underwent a dramatic change in morphology, increasing in size and adopting highly irregular shapes, suggestive of invasive phenotype. This behavior was observed to a much lesser degree for A549 cells and 393P cells.


Page 2 of 39

Page 3 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction Tumor angiogenesis, the recruitment of vasculature to a tumor site, has been described as a hallmark of cancer since cancer cells rely on the delivery of nutrients and oxygen to support exponential tumor growth and metastasis, the leading cause of death from cancer. Animal models are the gold standard for studying cancer. However, to date, outcomes of in vivo studies using anti-angiogenesis therapeutics have not matched clinical outcomes in human patients [1– 4]. Thus there is a need for improved models to study tumor angiogenesis that support identification of important genes, signaling molecules, cell-cell interactions, and cellextracellular matrix (ECM) interactions involved in tumor progression and metastasis. A better understanding will lead to development of more advanced screening tools that will more accurately represent human tumorigenesis and thus better predict clinical outcomes for potential anti-angiogenesis and anti-metastasis therapeutics. 3D in vitro models allow us to recapitulate the in vivo cancer microenvironment by customization of the cells, material components, and signaling molecules to support controlled experimentation to identify the role of each of these components in tumor progression [5]. We previously reported development of a bi-layer 3D tumor angiogenesis model that incorporates vascular cells and cancer cells into separate layers that are in contact in a degradable poly(ethylene glycol) (PEG)-based hydrogel [6]. Cells in the model can interact via paracrine signaling, which is known to drive vascular cell recruitment to a tumor site, as well as cellmediated juxtacrine interactions, because the cells can migrate between the 2 layers of hydrogel [6]. This model supported identification of a potential tumor progression-promoting role for vascular cells in the tumor microenvironment, using 344SQ cells, a mouse metastatic lung



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

adenocarcinoma cell line capable of reversibly undergoing an epithelial-to-mesenchymal transition (EMT) [6]. This paper is aimed at introducing new cancer cell types with varying metastatic capacities into the model to understand whether outcomes in the model correlate to these differing in vivo behaviors. This paper will explore the behavior of two cancer cell lines, both lung adenocarcinoma cells derived from primary tumors: A549 and 393P. In addition, the results associated with these 2 cell types will be compared to previously reported work with 344SQ cells. The 344SQ cell data published previously was included for comparison in key figures in this paper to allow ease of visualization of the data on the same plots. There are a few differences in the origin and genetic mutations of 344SQ, A549, and 393P cells (Table 1). 344SQ cells are derived from a subcutaneous metastatic site, whereas both A549 and 393P cells are derived from a primary tumor site [7,8]. 344SQ and 393P cells are both of mouse origin, while A549 cells are a human cancer cell line. Both 344SQ and 393P cells are derived from mice with Kras and p53 mutations, which are the most common genetic mutations in lung cancer [7,13]. A549 cells only harbor a Kras mutation, with normal p53 [14]. In spite of commonalities in known genetic mutations, 344SQ cells are highly metastatic in vivo and 393P cells are metastasis-incompetent. A549 cells form slow-growing tumors that are not reported to be highly aggressive and they do exhibit the capacity to metastasize [15]. In contrast to 344SQ cells, A549 cells display characteristics of both epithelial and mesenchymal cells and have a variable response to EMTinducing treatment, rather than exhibiting pronounced EMT behavior [9–12]. Additionally, all three cell types have been described to exhibit features of type II alveolar epithelial cells, such as surfactant protein C expression, which are a distal lung epithelial subtype found in the lung alveoli and that function to secrete surfactant [7,16,17]. The similarities and differences between


Page 4 of 39

Page 5 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the cell types in the following studies provide an interesting opportunity for comparison of differing cell behaviors in the tumor angiogenesis model. In this paper, characterization of A549 and 393P behavior in 3D in single layer PEG-based hydrogels over time was performed. Next, A549 and 393P cells were characterized for angiogenic growth factor secretion in PEG hydrogels to assess their angiogenesis-stimulating capacity. Finally, both cell types were incorporated into the bi-layer tumor angiogenesis model and analysis of tumor-vessel interactions and changes in cancer cell morphology was performed.

2. Materials and Methods 2.1 Cell Maintenance Human umbilical vein endothelial cells (HUVEC, Lonza) were cultured in Endothelial Basal Medium-2 (Lonza), with the Endothelial Growth Medium-2 SingleQuot kit and 2 mM LL-glutamine, 100 U ml-1 penicillin, and 100 µg ml-1 streptomycin. Human vascular pericytes (HVP, Sciencell) were cultured in Pericyte Medium (Sciencell). 344SQ and 393P cells were cultured in RPMI 1640 media supplemented with 10% FBS (Atlanta Biologics) and 10 µg/mL gentamicin and 0.25 µg/mL amphotericin-B. 393P cells were used in passages 5-9. A549 cells were cultured in F-12k media (Gibco) supplemented with 10% FBS (Atlanta Biologics). 100 U mL-1 penicillin and 100 µg mL-1 streptomycin were added to the media at day 1 following encapsulation in PEG hydrogels. A549 cells were used in passages 85-87. All cells were cultured in a humidified incubator at 37º C and 5% CO2.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2.2 Materials Synthesis Hydrogel materials were synthesized and precursor solutions were prepared as described previously [6]. RGDS was conjugated to PEG to support cell adhesion to the hydrophilic, protein adsorption-resistant PEG matrix. Conjugation to yield PEG-RGDS was performed by mixing RGDS (American Peptide) with monoacrylate PEG-succinimidyl valerate (PEG-SVA, 3,400 Da, Laysan Bio) in a ratio of 1:1.2 (PEG-SVA:RGDS). The conjugation was performed in HEPBS buffer (20 mM HEPBS, 100 mM NaCl, 2 mM CaCl2, 2 mM MgCl2, pH 8.5) and allowed to react overnight at 4 ºC, which was followed by dialysis to remove unreacted RGDS and lyophilization to dryness. Analysis of PEG-RGDS was performed with gel permeation chromatography with an evaporative light scattering detector (Supplemental Fig. 1, Polymer Laboratories). The same reaction methods were used to conjugate the peptide, GGGPQGIWGQGK (PQ), to PEG, with the exception that the molar ratio used was 2.1:1 (PEG-SVA:PQ). This ratio was chosen to support reaction of PEG chains with both the N-terminus and the lysine residue at the C-terminus of the PQ peptide, thus yielding PEG-PQ-PEG. PQ can be degraded by cellsecreted matrix metalloproteinases-2 and -9, thus supporting cell-mediated degradation of the hydrogels. PQ was synthesized using standard Fmoc chemistry on an APEX 396 solid phase peptide synthesizer (Aapptec). Subsequent analysis was performed with MALDI-ToF mass spectrometry. Analysis of PEG-PQ-PEG was performed with gel permeation chromatography with an evaporative light scattering detector (Supplemental Fig. 1, Polymer Laboratories). PEG-RGDS was fluorescently tagged for incorporation into dual layer hydrogels to support visualization of the separate layers. Synthesis of AF488-PEG-RGDS was performed as previously described [18]. In brief, Alexa Fluor® 488 carboxylic acid, 2,3,5,6-tetrafluorophenyl ester (AF488-TFP; Invitrogen), dissolved in dimethylsulfoxide (DMSO), was mixed with PEGRGDS, dissolved in 0.1 M sodium bicarbonate buffer at pH 9, at a ratio of 10:1 (AF488-


Page 6 of 39

Page 7 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


TFP:PEG-RGDS). The reaction was let go for 2 hours at room temperature, followed by dialysis to remove unreacted AF488-TFP and lyophilization to dryness.

To prepare sterile hydrogel precursor solutions, lyophilized PEG-RGDS was dissolved in 10 mM HEPES-buffered saline with 1.5% v/v triethanolamine (Sigma) (HBS-TEOA) at a concentration of 59.3 mM PEG-RGDS. The same solution was used to dissolve lyophilized PEG-PQ-PEG at 20% w/v. Both solutions were filter sterilized with a 0.2 µm syringe filter (PALL Corporation). Following filtration, solutions were kept at -20°C until use.

2.3 Cell Encapsulations in PEG-based Hydrogels To form cell-laden hydrogels, PEG-PQ-PEG and PEG-RGDS precursor solutions were thawed and further diluted to 4% w/v PEG-PQ-PEG and 3.5 mM PEG-RGDS in HBS-TEOA with 3.5 µL/mL N-vinyl pyrrolidone (NVP, Sigma) and 10 µM eosin Y photoinitiator (Sigma). Cells were then harvested with trypsin, centrifuged (200 x g for 5 min), and then resuspended in the polymer precursor solution at a concentration of 3 x107 cells/mL at a 4:1 ratio of HUVEC:HVP for vascular cell-laden hydrogels (referred to as V-hydrogels) and 1.0 x106 cells/mL for A549 cancer cell-laden hydrogels and 1.5 x106 cells/mL for 393P cell-laden hydrogels (both A549 and 393P single layer hydrogels are referred to as C-hydrogels). Cellladen polymer precursor solution was pipetted onto Sigmacote-treated (Sigma) glass between 2 380 µm-thick polydimethylsiloxane (PDMS) spacers. The polymer solution was then covered with a methacrylate-treated glass coverslide to create a hydrogel disc and photopolymerized via exposure to white light (Fiber-Lite Series 180, 150 Q halogen, Dolan Jenner). Following



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

polymerization, the hydrogel was lifted from the Sigmacote-treated slide via the methacrylated glass coverslip and placed in a well plate for long term culture. To confirm hydrogel formation and assess the mechanical properties of cell-laden hydrogels, C-hydrogels and V-hydrogels were formed as described above, with the modification that 1 mm-thick PDMS spacers were used to reduce the influence of the glass stiffness on the resulting mechanical properties. Cell-laden hydrogels were allowed to grow for 3 days. Compressive testing was performed using a RSA III Microstrain Analyzer (TA Instruments). Uniaxial compressive strain was applied at 0.003 mm/s. Stress-strain data was plotted and the compressive modulus was measured from the linear region of the curve.

2.4 Immunocytochemistry Immunocytochemistry on cells in hydrogels was performed as described previously [6]. In brief, cell-laden hydrogels were fixed with 4% paraformaldehyde for 45 minutes. Blocking in 5% donkey serum was performed overnight at 4°C while rocking. Primary antibodies were then added for 2 nights at 4°C. Primary antibodies used in this work include: goat anti-VE cadherin (Santa Cruz Biotechnology). Hydrogels were rinsed 3 times with 0.01% tween-20 (Sigma) in PBS and one time in PBS only. Secondary antibodies were then added for 2 nights at 4°C. Secondary antibodies used in this work include: donkey anti-goat-647 (Life Technologies).

2.5 Analysis of Growth Factor Secretion from Cells in PEG Hydrogels C-hydrogels were used to assess angiogenic growth factor secretion. A549 and 393P cells were cultured in hydrogels for 16 days. Media was changed on hydrogels on day 1 of culture and every other day after that. Conditioned media was collected 24 hours following the last media change. The conditioned media was assessed for vascular endothelial growth factor (VEGF),


Page 8 of 39

Page 9 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


basic fibroblast growth factor (bFGF), and platelet-derived growth factor BB (PDGF-BB) by ELISA according to the protocol for each assay (all from R&D Systems). Hydrogels were degraded at the time of media collection with 2 mg/mL collagenase (Sigma) for 20 min, followed by addition of RIPA buffer to lyse the cells. Growth factor concentrations were normalized to total protein in the samples assessed via a BCA assay (ThermoFisher Scientific). Data is shown as mean ± standard deviation and an ANOVA followed by pairwise comparisons with a TukeyKramer HSD test using an alpha level of 0.05 to evaluate for statistically significant differences.

2.6 Tumor Angiogenesis Model Construction Tumor angiogenesis model hydrogels were constructed using the same protocol as described previously [6] for 344SQ cells with the exception of using A549 as the cancer cell component at a concentration of 1.0 x 106 cells/mL or 393P cells at a concentration of 1.5 x 106 cells/mL in the hydrogel. In brief, a C-hydrogel spiked with fluorescent PEG-RGDS was polymerized in a PDMS spacer well (500 µm thick) for 20 s. A second PDMS spacer well (500 µm thick) was stacked on top of the first and vascular cell-laden polymer precursor solution was pipetted on top of the first hydrogel. This was followed by a second polymerization for 30 s to form a V-hydrogel on top of the C-hydrogel. Tumor angiogenesis model hydrogels were always constructed with the V-hydrogel on top. The PDMS spacers were removed and the dual layer hydrogel was placed in culture in HUVEC media (EBM-2 with EGM-2 supplements, described in Section 2.1) media for all 3 cancer cell lines. A schematic of the tumor angiogenesis model is included in Supplemental Fig. 2. In order to simplify the naming convention, V-C hydrogels are bilayer hydrogels comprised of vascular cell and cancer cell layers and B-C hydrogels are comprised of an acellular “blank” hydrogel layer and a cancer cell layer. All tumor angiogenesis model hydrogels were fixed at day 10 in culture for analysis. Imaging and image analysis on



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

these hydrogels was performed as described previously [6]. Briefly, imaging was performed on a Zeiss LSM 510 confocal microscope and z-stacks were taken through the hydrogels to support analysis of cancer cell behavior based on distance from the V-hydrogel, as identified by lack of fluorescent PEG-RGDS. The average intensity of 488-PEG-RGDS for each slice was plotted and a defined intensity threshold for the interface was used to identify the interface slice. The interface bin was defined as the interface slice plus 10 slices above and below the interface. The remaining slices were divided into 105 µm-thick bins (Supplemental Fig. 2). Analysis of area and circularity of clusters of cancer cells was performed using ImageJ. Vessel cell-cancer cell interactions were quantified manually by counting the number of times CD31 or VE-cadherin fluorescence overlapped with a phalloidin-stained cancer cluster. Phalloidin fluorescence was much brighter in cancer clusters compared to vascular cells so imaging settings were adjusted to preferentially visualize phalloidin fluorescence in cancer cells. Analysis of vascular cell invasion into hydrogels was performed using Imaris, whereby for each image, the YZ view in surfaces mode was used to visualize 50-µm sections, viewing the vascular cell channel and the fluorescently-tagged PEG-RGDS channel. For each section, the measurement tab was used to draw a line between the interface (as determined by PEG-RGDS-488) and the tip of the vessel invasion into the fluorescently tagged region of the hydrogel. The longest measurement for each image was recorded as the maximum vessel penetration into the hydrogel. Vascular parameters were evaluated for V-C as well as V-B control hydrogels for 344SQ cells. Vascular network parameters were quantified using the Angiogenesis Tube Formation Application Module in MetaMorph software and total tubule length, number of branch points, and average tubule thickness were evaluated. Networks were evaluated adjacent to the hydrogel interface as well in the vascular hydrogel region furthest from the interface in order to determine if there were local


Page 10 of 39

Page 11 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


changes in network parameters based on distance to nearest cancer cells. Alexafluor 488-PEGRGDS was used to determine the location of the interface.

3. Results 3.1 Growth and Morphology in PEG Hydrogels Compressive testing was performed on C-hydrogels and V-hydrogels alone. The compressive modulus of V-hydrogels was 16.0 ± 6.2 kPa and C-hydrogels was 15.1 ± 7.0 kPa (Supplemental Fig. 3). No statistically significant difference was detected between the two hydrogel types, indicating the bulk mechanical properties did not change given the differing cell densities. When comparing this measured stiffness to that of a normal lung, which has been reported in the range of 1 kPa, and a fibrotic lung, which is between 6-20 kPa, the hydrogels fall closer to a diseased lung [19]. Because many reports show that the ECM stiffens in cancerous tissue, this enhanced stiffness in our study emulates cancerous tissue. 393P formed spherical aggregates in PEG hydrogels by day 4 in culture that exhibited an average area of 1,952 ± 144 µm2 (average diameter = 50.8 ± 0.8 µm, Fig. 1). Spheroid growth did not persist over time in culture, with spheroid size essentially constant between days 4 and 16. Single A549 cells also formed spherical aggregates within PEG hydrogels that were evident by day 4 in culture (Fig. 1). A549 were kept in culture for 16 days and analyzed every 4 days for cluster area. A549 clusters increased in size over time, whereby average cluster area was 1050 ± 145, 2171 ± 67, 2267 ± 442, and 2991 ± 168 µm2 on days 4, 8, 12, and 16, respectively. Cluster diameter was also measured, and found to be 37.6 ± 2.1, 52.4 ± 0.8, 55.1 ± 5.5, and 63.7 ± 3.6 µm on days 4, 8, 12, and 16, respectively.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

3D culture of normal lung epithelial cells results in formation of spherical aggregates with cleared lumen and epithelial polarity that recapitulates the structure and function of lung epithelium in vivo [20]. 344SQ lung cancer cells evaluated previously have also been shown to recapitulate this behavior, whereby 344SQ cell spheroids organized into a single layer of cells surrounding a centrally cleared lumen [21]. 393P exhibited lumenization, as depicted in Fig. 2, whereby 28 ± 7% of total clusters evaluated at day 12 in culture were lumenized (Fig. 2b). A549 cluster lumenization was also evaluated. No clusters comprised of a single nuclear ring surrounding a hollow lumen were observed for A549 cells in PEG hydrogels.

3.2 Angiogenic Growth Factor Secretion from PEG Hydrogels To assess the angiogenesis-stimulating capacity of 393P and A549 cells in 3D in PEG hydrogels, enzyme-linked immunosorbent assays (ELISA) were performed on VEGF, PDGFBB, and FGFb from the conditioned media of 393P and A549 cells over time in culture (Fig. 3). For 393P cells very little VEGF secretion was detected at day 4 but secretion significantly increased over time in culture. By day 8, 210.6 ± 72.6 pg VEGF / 106 cells was detected, which increased to 509.9 ± 144.3 and 729.9 ± 220.3 pg VEGF / 106 cells on days 12 and 16 respectively. PDGF-BB secretion did not significantly change over the course of culture (day 4: 75.5 ± 106.8, day 8: 98.6 ± 30.4, day 12: 73.1 ± 8.7, and day 16: 72.9 ± 18.7 pg PDGF / 106 cells). FGFb was not detected in the conditioned media at day 4, but was present in the conditioned media at all other time points (day 8: 22.1 ± 24.7, day 12: 9.1 ± 5.0, and day 16: 1.9 ± 3.3=4 pg FGFb / 106 cells). For A549 cells in PEG hydrogels, VEGF was present in the conditioned media at all four time points at varying concentrations (day 4 653.0 ± 43.8, day 8 405.5 ± 55.4, day 12 384.8 ± 3.3, and day 16 795.4 ± 116.4 pg VEGF / 106 cells). Interestingly, there was a significant


Page 12 of 39

Page 13 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


decrease in VEGF secretion observed for days 8 and 12 (p < 0.01) but secretion increased again at day 16. No PDGF-BB was detected until day 16 in culture (0.2 ± 0.4 pg PDGF-BB / 106 cells). FGFb was detected at all time points (5.0 ± 8.7, 9.4 ± 3.9, 9.1 ±1.9, and 22.1 ± 4.7 pg FGFb / 106 cells at days 4, 8, 12, and 16, respectively) and only exhibited a significant increase at day 16 (p < 0.05).

3.3 Evaluation of Cancer Cluster Morphology Changes in Tumor Angiogenesis Model 393P and A549 cells interacted with vascular cells at the interface of V-C hydrogels (Fig. 4a). Comparing the number of vessel cell-cancer cell interactions in V-C hydrogels in the tumor angiogenesis model, 344SQ had the highest number of interactions compared to A549 and 393P cells, though this difference did not achieve statistical significance (Fig. 4b). This observation, though, correlates to the higher VEGF and FGFb concentrations secreted by 344SQ cells. Changes in 393P cluster morphology were observed at the hydrogel interface (Fig. 5, Supplementary Movie 2). Most clusters observed were slightly larger and exhibited an elongated, oval morphology. Cluster size was also observed to increase, but not uniformly. Most cluster size increases were small, but a few clusters exhibited a drastic size increase (Fig. 5a left versus right images). Quantification of cluster morphology via analysis of 393P cluster area and circularity for 975 clusters in V-C hydrogels that were broken down into bins according to distance from the interface revealed changes in the distribution of clusters at the hydrogel interface (Fig. 5b and c). For cluster area analysis, it was evident that all clusters with area greater than 5,000 µm2 were in the interface or (0,105) bins (Fig. 5b). This trend extended below 5,000 µm2, whereby out of a total of 26 clusters with area greater than 3,000 µm2, only 2 clusters were not in the interface or (0,105) bins. In addition, the only significantly different distribution of cluster area for the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

hydrogel bins compared to the interface was the (0,105) hydrogel bin, which had the highest number of clusters with area greater than 5,000 µm2. Quantification of cluster circularity for 393P V-C hydrogels revealed that all bins were significantly different from the interface bin (Fig. 5c). The circularity distributions for hydrogel regions below the interface followed a similar pattern of left-skewed distribution, with most clusters exhibiting circularity between 0.8 and 1. However, for the interface region, while the highest number of clusters had circularities between 0.8 and 1, the distribution did not taper off at lower circularities as it did for regions below the interface. The (0,105) region exhibited a unique circularity distribution, where the circularity of very few clusters was between 0.8 and 1 and the highest number of clusters were found in the 0.1-0.3 circularity bins. For quantification of cluster area in 393P control B-C hydrogels, in contrast to V-C hydrogels, it was evident that very few clusters were present in the (0,105) hydrogel bin (Supplemental Fig. 4). The distribution of clusters at the hydrogel interface was significantly different than the 3 regions below it, as determined by a Kolmorogov-Smirnov test with p < 0.0125 (Bonferroni corrected). Additionally, the interface region had the highest number of clusters with area greater than 5,000 µm2, with 7 total clusters in that bin. It was also evident that at the interface, there were fewer small clusters (500-1500 µm2) compared to the 3 regions below it. This difference could have also contributed to the significant differences in distribution. For cluster circularity in B-C hydrogels, there was a lower number of clusters with high circularity at the interface, but there was also not an increase in clusters with low circularity as exhibited in VC hydrogels (Supplemental Fig. 4c). Plotting cluster area versus cluster circularity for B-C hydrogels revealed only 1 cluster with large area and low circularity (area > 5,000 µm2, circularity < 0.25) at the hydrogel interface.


Page 14 of 39

Page 15 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Over time in A549 V-C hydrogels, it was observed that some A549 clusters at the hydrogel interface interacted with vascular cells, as identified by VE-cadherin, and lost their round shape in comparison to the morphology observed in C-hydrogels (Fig. 6a, Supplementary Movie 3). To assess regional changes in A549 clusters, z-stacks obtained were used to divide the hydrogel into bins that were 105 micrometers thick. Quantification of cluster area was performed for each of the bins of vascular and cancer hydrogels (Fig. 6b). This analysis revealed that the number of large clusters with projected area greater than 5000 µm2 in the interface bin of the tumor angiogenesis model was not higher than in the other regions of the hydrogel, as was reported for 344SQ cells. In fact, the hydrogel bin with the highest number of clusters with area greater than 5000 µm2 was (-315, -210), which represents the cells that are furthest from the vascular cell source. Additionally, there were no significant differences in the distribution of cluster area for any group compared to the interface bin. Quantification of cluster circularity was also performed for each bin, and this analysis did reveal an enhanced number of clusters with low circularity (< 0.3) for the interface region that was higher than in the other hydrogel bins (Fig.6c). However, there were no significant differences in the distribution of cluster circularity for any region compared to the interface. Thus while there were vascular cell-cancer cell interactions and changes in A549 morphology observed, there were not corresponding drastic increases in cluster area and decreases in cluster circularity for A549 cells in the tumor angiogenesis model. As a control, A549 cluster area and circularity were also evaluated for B-C hydrogels (Supplemental Fig. 5). This analysis revealed a lack of structures with high area and low circularity at the hydrogel interface. For both cluster area and circularity, no significant differences in cluster distribution at the interface were observed. In addition, the (-315,-210) bin



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

for cluster area had the highest number of clusters with area greater than 5,000 µm2, which was also true for the V-C hydrogels, indicating this high number of large clusters was not a result of the presence of vascular cells. Cluster area and circularity plotted together for clusters at the interface of B-C hydrogels further confirmed there were very few large clusters with low circularity. Comparing the V-C area and circularity data for 393P and A549 cells to previously published data with 344SQ cells, only 344SQ cell clusters exhibited a significant difference in distribution of both cluster area and circularity at the hydrogel interface [6]. Area versus circularity was plotted for all clusters in V-C hydrogels for 344SQ, A549, and 393P cells (Fig. 7). For 344SQ clusters, there was a population of clusters at the hydrogel interface that exhibited both large area and low circularity (area > 5,000 µm2, circularity < 0.25, Fig. 7). For 393P cells in V-C hydrogels, no clusters with large area and low circularity were observed at the interface. Very few clusters with high area and low circularity at the interface occurred for A549 V-C hydrogels.

3.4 Evaluation of Vascular Network Changes in Tumor Angiogenesis Model Vascular parameters were evaluated for V-C as well as V-B control hydrogels for 344SQ cells. Maximum vessel penetration length into the cancer hydrogel of V-C hydrogels and compared this invasion to V-B hydrogels (Supplemental Fig. 6). There was higher maximum vessel penetration in the V-B hydrogels compared to the V-C hydrogels. This is likely due to vascular cells being recruited to cancer clusters and are not continuing to invade past their initial interaction with cancer clusters. Vascular network parameters evaluated for 344SQ cells were total tubule length, number of branch points, and average tubule thickness. Total tubule length per image was significantly lower in V-C hydrogels compared to V-B hydrogels at the interface (4079 ± 1508 µm versus


Page 16 of 39

Page 17 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


9904 ± 2462 µm, respectively). Total tubule length far from the interface (6282 ± 1266 µm) was not significantly different from the interface for V-C hydrogels or V-B hydrogels. However, total tubule length for V-B hydrogels far from the interface (4567 ± 815 µm) was significantly lower compared to the same group at the interface. The number of branch points was not significantly different for V-C and V-B hydrogels at the interface (83 ± 32 versus 233 ± 61). The same trend also followed for branch points as did for tubule length, whereby the number of branch points for V-B hydrogels far from the interface (108 ± 33) was significantly lower compared to the same group at the interface. Average tubule thickness was also not significantly different for any groups evaluated (V-C interface: 9.8 ± 2.2; V-B interface: 10.9 ± 0.2 µm; V-C far from interface: 10.3 ± 0.4; V-B far from interface: 8.9 ± 0.3 µm). It has been reported in the literature that addition of VEGF or VEGF and FGFb to HUVEC seeded on polyacrylamide hydrogels led to inhibition of network formation, which was found to be due to an increase in cell migration [33]. The growth factor concentration experienced by vascular cells in the interface region of the hydrogel is likely the highest as they are the closest to cancer spheroids. Thus this increased growth factor concentration could play a role in the lower total tubule length observed in V-C hydrogels at the interface. Additionally, vessel structures associated with tumors have been reported to be irregular and characterized by non-contiguous structures that are tortuous [2]. Even though the structures analyzed here are not within the cancer clusters, the postulated high concentration of angiogenic growth factors in this region of the hydrogel could still contribute to more rapid vascular network formation that is less mature in structure as compared to the control hydrogels.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

4. Discussion When cultured in 2D on tissue culture polystyrene, all three cell types grew rapidly and over the course of culture and did not reach senescence. Additionally, they all grew in colonies and exhibited an epithelial-like morphology (Supplemental Fig. 7). When cultured in 3D, differences in growth and morphology were observed. For 344SQ cells, growth arrest occurred for aggregates between days 4-6 in culture [21]. 393P aggregate area was not significantly different at any of the time points evaluated (days 4, 8, 12, and 16), indicating that clusters reached growth arrest very quickly in hydrogels, as the clusters became densely cellular. In contrast, A549 cell clusters in PEG hydrogels grew over time, with significant increases in cluster area on day 16. These findings corroborate reports in the literature that while cancer cells in monolayer culture exhibit exponential growth, 3D cultures better recapitulate growth dynamics of tumors in vivo, in which there is an early growth phase, followed by a period of growth delay [22–25]. Literature on 393P cells in vivo in mouse models demonstrate these cells are metastasisincompetent, in contrast to 344SQ cells which are highly metastatic in vivo [7]. Interestingly, 393P cells exhibited a lack of epithelial polarity in 3D in Matrigel cultures as well as reduced ability to undergo EMT. These findings contrast those associated with 344SQ cells, which exhibit epithelial polarity in 3D in Matrigel and an ability to reversibly undergo EMT. In comparing these previous findings to the results reported in this paper, 393P clusters in this study did exhibit some lumenal clearing and lumenization. While quantification was not performed in the earlier studies from other groups, differences in behavior could be due to differences in matrix properties such as stiffness and availability of adhesive ligands.


Page 18 of 39

Page 19 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For A549 cells, spherical aggregates similar to those reported previously in Matrigel were observed [12,14]. Fessart, et al. reported an average spheroid diameter of 87 µm as a final size at day 20 in culture, whereas the average diameter we recorded at day 16 was 64 µm [12]. The smaller diameter could be attributed to time in culture. However, this difference in size is more likely a result of differences in the composition of the material used for culture. Matrigel is much softer than the PEG hydrogels employed in this work (E = 2.5 kPa for Matrigel compared to 15 kPa for these PEG hydrogels) and 344SQ lung adenocarcinoma cell spheroids have been previously reported to increase in size with decreasing materials stiffness [21]. Additionally, the growth characteristics for A549 compared to 344SQ and 393P cancer cells in our report matches what has been described in vivo previously. In a xenograft study, A549 tumors grew the slowest, as measured by tumor volume, compared to 3 other tumor cell types (A431 epidermoid carcinoma, Calu-6 anaplastic lung carcinoma, and MLS ovarian carcinoma cells) [15]. Some lung-derived tumor cells have been shown to form aggregates that can lumenize and polarize in 3D culture, mimicking the architecture of native lung alveoli. Comparing the cell types used in this work for their lumenal clearing as a metric of ability to organize and exhibit epithelial polarity in 3D, all three cell types formed aggregates that exhibited some degree of lumenal clearing over time in culture. 344SQ and 393P cells both formed structures with a distinct central lumens. This is in contrast to 393P behavior in Matrigel cultures, as 393P cells grow into disorganized aggregates and do not display a centrally cleared lumen [7]. Differences in behavior could be due to differences in the matrix properties of Matrigel described above. A549 cells, however, only exhibited minor evidence of lumenal clearing and no distinct central lumens surrounded by a layer of organized nuclei. These differing behaviors contrast what might be expected for these cells solely based on our knowledge of their mutations, whereby both



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

344SQ and 393P cells are more mutated but exhibit a greater capacity to organize into an epithelial phenotype in 3D culture. Thus it is evident that Kras and p53 mutations alone do not solely determine cancer cell behavior. The angiogenic capacity of 344SQ, A549, and 393P cells in 3D in PEG hydrogels also varied for the different cell lines. 344SQ cells secreted VEGF and FGFb in the highest concentration compared to A549 and 393P cells. VEGF and FGFb are responsible for enhancing endothelial cell migration and proliferation, indicating 344SQ might have the capacity to stimulate more angiogenesis than A549 and 393P cells. Interestingly, 393P cells secreted the highest concentration of PDGF-BB, indicating vascular structures associated with these tumors might be more mature than the other two tumor cell lines because PDGF-BB plays a role in pericyte recruitment to stabilize nascent vessels. VEGF plays a very important role in tumor angiogenesis and is the most widely studied angiogenic growth factor. Thus concentrations of VEGF in the conditioned media of Chydrogels were compared to those reported in the literature for tumors in vivo. A few studies evaluated the concentration of VEGF in plasma of patients with non-small cell lung cancer (NSCLC), whereby the average VEGF concentrations were reported to be approximately 110 and 88 pg/mL VEGF (107 and 18 patients evaluated, respectively) [26,27]. 344SQ, 393P, and A549 cells in C-hydrogels secreted 360 pg/mL, 250 pg/mL, 202 pg/mL VEGF into the conditioned media on average, respectively. Thus, the VEGF concentration secreted by NSCLC cell lines in C-hydrogels is higher than reported plasma concentrations in vivo. Overall, however, concentrations evaluated were similar to in vivo values. To note, it has also been reported that microdialysis measurements of VEGF interstitial concentrations in normal skeletal muscle and breast tissue were similar to plasma levels of VEGF, indicating using plasma VEGF


Page 20 of 39

Page 21 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


concentrations to estimate concentrations in interstitial tissue is valid [28]. Additionally, in the study with 18 patients, all patients had either Stage III or IV disease and while no statistical association was observed between VEGF levels and disease stage, patients with stage IV disease had a higher mean concentration of VEGF (Stage III mean VEGF: 63.4 pg/mL, Stage IV mean VEGF: 108.7 pg/mL) [26]. This correlates with the finding in this paper that the metastatic cancer cells had the highest VEGF secretion. In addition, when comparing the VEGF plasma levels to healthy subjects, the concentration reported in one study was ~40 pg/mL, indicating the concentration of VEGF detected from C-hydrogels is higher than both NSCLC patients and healthy subjects [29]. Lastly, in our system, VEGF concentrations were evaluated in the conditioned media of C-hydrogels, rather than in the hydrogel. Solving Fick’s 2nd law of diffusion, for 344SQ cells we estimated that approximately 415 pg/mL VEGF is present around cancer clusters at the interface of V-C hydrogels (based on a diffusion distance of 500 µm over 24 hours, diffusivity = 0.75 x 10-6 cm2/s) [30]. For 393P and A549 cells, the estimated VEGF concentration in the hydrogel is 295 pg/mL and 245 pg/mL, respectively. Comparing the number of vessel cell-cancer cell interactions in V-C hydrogels in the tumor angiogenesis model, 344SQ had the highest number of interactions compared to A549 and 393P cells, though this difference did not achieve statistical significance. This observation, though, correlates to the higher VEGF and FGFb concentrations secreted by 344SQ cells. To note, interactions between vascular cells and cancer cells, as well as morphological changes in cancer cell clusters, are likely due to MMP-mediated degradation of the hydrogels, rather than pores in the hydrogels, as mesh size estimated for a similar hydrogel composition was approximately 172.4 Å (17 nm), which is too small for a cell to rely on for migration [31].



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

When looking at overall changes in cancer cluster morphology in V-C hydrogels for 344SQ, A549, and 393P cells, it was evident that only 344SQ clusters exhibited a significant difference in distribution of both cluster area and circularity at the hydrogel interface. This same trend was not evident for A549 or 393P cell clusters (Supplementary Movies 1-3). For 393P cells in V-C hydrogels, while no clusters with large area and low circularity were observed at the interface, some morphological changes were observed for 393P cells interacting with vascular cells and a decrease in lumenization of clusters was observed. It was also evident that large, invasive clusters were observed in the (0,105) hydrogel region for 393P clusters in V-C hydrogels, but these cluster morphological changes were not as drastic as were observed for 344SQ cells. This finding is not surprising for 393P cell clusters, as they have been described as only undergoing partial EMT after 7 days of exposure to transforming growth factor beta 1 (TGF-β1), which was contrast to 344SQ cells that underwent a full EMT under the same conditions within 2 days [7]. Thus this inability to fully undergo EMT could contribute to preventing 393P cluster morphological changes that could result due to the presence of TGF-β1 secreted by vascular cells in the tumor angiogenesis model. In addition, one study showed interesting protein expression differences between 344SQ and 393P cells. Expression of proteins involved in cell-ECM contacts were higher in 344SQ cells and expression of proteins involved in cell-cell contacts was higher in 393P cells [32]. These findings help to explain some of the differences we observed in this study, whereby 344SQ cells might have a greater propensity to invade the surrounding ECM due to the higher cell-ECM proteins expressed. Very few clusters with high area and low circularity at the interface occurred for A549 VC hydrogels (Fig. 6). One factor to note for A549 cells is that fewer overall clusters were analyzed in A549 V-C hydrogels (739 344SQ clusters versus 379 A549 clusters). However,


Page 22 of 39

Page 23 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


large, invasive A549 clusters at the interface were all below 10,000 µm2 in size, whereas for 344SQ cells, several clusters exhibited area greater than 15,000 µm2. Thus regardless of cluster density differences, changes in 344SQ clusters in the V-C hydrogel were more drastic than that of A549 cells. Similar to 344SQ and 393P, A549 clusters at the interface with large area and low circularity also did not have lumens. Additionally, when the number of large, invasive (area > 5,000 µm2, circularity < 0.25) clusters was normalized to overall number of clusters at the interface for 344SQ, 393P, and A549 cells, it was evident that 13%, 1%, and 6% of clusters were large and invasive, respectively, which correlates well with metastatic capacity. Overall, these differential findings in the tumor angiogenesis model correlated to known characteristics of the cells (Table 2). For example, when the cells are grouped by their known genetic mutations, it is apparent that the single mutant A549 cell clusters took longer to reach growth-arrest, did not lumenize, and secreted the lowest concentration of all 3 angiogenic growth factors assessed. These characteristics might represent a less aggressive tumor, in that it would likely grow slower in vivo, would be less plastic in its phenotype, and would not recruit vascular cells to promote growth as well as the double mutant cancer cells might. When the cells are grouped by the tumor site they were derived from whereby 344SQ cells were derived from a metastatic lesion and 393P and A549 cells were derived from a primary tumor, differences are evident in cell behavior in the tumor angiogenesis model. 344SQ cells exhibited the highest percent of large, invasive clusters, with many exhibiting cluster area larger than observed in the other 2 cell types and 344SQ cells in V-C hydrogels had the highest number of vessel cell-cancer cell interactions compared to the other 2 cell types. Finally, when the cells were grouped by their metastatic capacity into highly metastatic 344SQ cells, moderately metastatic A549 cells, and



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

metastasis-incapable 393P cells, it was evident that the percent of large invasive clusters in V-C hydrogels correlated to this finding (Table 2).

5. Conclusion In this paper, we have demonstrated the ability of the tumor angiogenesis model developed previously to support exploration of other cancer cell lines aside from 344SQ cells. We have also acquired additional evidence demonstrating a potential role vascular cells play in tumor progression by expanding our findings to incorporate multiple cancer cell lines. Furthermore, by incorporating a well-studied cancer cell line, we have compared our findings with in vivo data, which supports the notion that in vitro models with enhanced complexity compared to existing oversimplified 2D culture systems could serve as an avenue for accurately predicting cancer cell behavior in vivo. In conclusion, differing behaviors of lung adenocarcinoma cell lines in 3D in PEG hydrogels were observed and correlated to tumor origin, mutations, and previous in vitro and in vivo findings.

Supporting Information Available: The following files are available free of charge. Supplemental Figures 1-3 cover material characterization and model setup. Supplemental Figures 4 and 5 show cluster area and circularity data from control hydrogels for 393P and A549 cells. Supplemental Figure 6 is focused on vascular network quantification. Supplemental Figure 7 shows 2D culture differences between the 3 cancer cell types. Supplemental Movies 1-3 are 3D projections of V-C hydrogels for each of the 3 cancer cell types analyzed.


Page 24 of 39

Page 25 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


6. References [1]

L. Eklund, M. Bry, K. Alitalo, Mouse models for studying angiogenesis and lymphangiogenesis in cancer, Mol. Oncol. 7 (2013) 259–282. doi:10.1016/j.molonc.2013.02.007.

[2]

R.K. Jain, Normalization of tumor vasculature: an emerging concept in antiangiogenic therapy., Science. 307 (2005) 58–62. doi:10.1126/science.1104819.

[3]

D.F. Hayes, Bevacizumab treatment for solid tumors: boon or bust?, JAMA. 305 (2011) 506–508. doi:10.1001/jama.2011.57.

[4]

R.K. Jain, Lessons from multidisciplinary translation trials on anti-angiogenic therapy of cancer, Nat. Rev. Cancer. 8 (2008) 309–316.

[5]

L.C. Roudsari, J.L. West, Studying the influence of angiogenesis in in vitro cancer model systems, Adv. Drug Deliv. Rev. (2015). doi:10.1016/j.addr.2015.11.004.

[6]

L.C. Roudsari, S.E. Jeffs, A.S. Witt, B.J. Gill, J.L. West, D. Hanahan, et al., A 3D Poly(ethylene glycol)-based Tumor Angiogenesis Model to Study the Influence of Vascular Cells on Lung Tumor Cell Behavior., Sci. Rep. 6 (2016) 32726. doi:10.1038/srep32726.

[7]

D.L. Gibbons, W. Lin, C.J. Creighton, Z.H. Rizvi, P. a Gregory, G.J. Goodall, et al., Contextual extracellular cues promote tumor cell EMT and metastasis by regulating miR200 family expression., Genes Dev. 23 (2009) 2140–51. doi:10.1101/gad.1820209.

[8]

M. Lieber, B. Smith, a Szakal, W. Nelson-Rees, G. Todaro, A continuous tumor-cell line from a human lung carcinoma with properties of type II alveolar epithelial cells., Int. J. Cancer. 17 (1976) 62–70. doi:10.1002/ijc.2910170110.

[9]

J.H. Kim, Y.S. Jang, K.-S. Eom, Y. Il Hwang, H.R. Kang, S.H. Jang, et al., Transforming growth factor beta1 induces epithelial-to-mesenchymal transition of A549 cells., J. Korean Med. Sci. 22 (2007) 898–904. doi:10.3346/jkms.2007.22.5.898.

[10]

H. Kasai, J.T. Allen, R.M. Mason, T. Kamimura, Z. Zhang, TGF-beta1 induces human alveolar epithelial to mesenchymal cell transition (EMT)., Respir. Res. 6 (2005) 56. doi:10.1186/1465-9921-6-56.

[11]

A. Aref, R.Y.-J. Huang, W. Yu, K.-N. Chua, W. Sun, T.-Y. Tu, et al., Screening therapeutic EMT blocking agents in a three-dimensional microenvironment, Integr. Biol. 5 (2013) 381–389.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

[12]

D. Fessart, H. Begueret, F. Delom, Three-dimensional culture model to distinguish normal from malignant human bronchial epithelial cells., Eur. Respir. J. 42 (2013) 1345– 56. doi:10.1183/09031936.00118812.

[13]

S. Zheng, a K. El-Naggar, E.S. Kim, J.M. Kurie, G. Lozano, A genetic mouse model for metastatic lung cancer with gender differences in survival., Oncogene. 26 (2007) 6896– 6904. doi:10.1038/sj.onc.1210493.

[14]

M.A. Cichon, V.G. Gainullin, Y. Zhang, D.C. Radisky, Growth of lung cancer cells in three-dimensional microenvironments reveals key features of tumor malignancy, Integr. Biol. 4 (2012) 440. doi:10.1039/c1ib00090j.

[15]

J. Ehling, B. Theek, F. Gremse, S. Baetke, D. Mockel, J. Maynard, et al., Micro-CT imaging of tumor angiogenesis: Quantitative measures describing micromorphology and vascularization, Am. J. Pathol. 184 (2014) 431–441. doi:10.1016/j.ajpath.2013.10.014.

[16]

B.A. Mercer, V. Lemaître, C. a Powell, J. D’Armiento, The Epithelial Cell in Lung Health and Emphysema Pathogenesis., Curr. Respir. Med. Rev. 2 (2006) 101–142. doi:10.2174/157339806776843085.

[17]

K. Vaporidi, C. Tsatsanis, D. Georgopoulos, P.N. Tsichlis, Effects of hypoxia and hypercapnia on surfactant protein expression proliferation and apoptosis in A549 alveolar epithelial cells, Life Sci. 78 (2005) 284–293. doi:10.1016/j.lfs.2005.04.070.

[18]

J.C. Hoffmann, J.L. West, Three-dimensional photolithographic patterning of multiple bioactive ligands in poly(ethylene glycol) hydrogels, Soft Matter. 6 (2010) 5056. doi:10.1039/c0sm00140f.

[19]

A. Marinković, F. Liu, D.J. Tschumperlin, Matrices of physiologic stiffness potently inactivate idiopathic pulmonary fibrosis fibroblasts, Am. J. Respir. Cell Mol. Biol. 48 (2013) 422–430. doi:10.1165/rcmb.2012-0335OC.

[20]

X. Wu, J.R. Peters-Hall, S. Bose, M.T. Pena, M.C. Rose, Human bronchial epithelial cells differentiate to 3D glandular acini on basement membrane matrix, Am. J. Respir. Cell Mol. Biol. 44 (2011) 914–921. doi:10.1165/rcmb.2009-0329OC.

[21]

B.J. Gill, D.L. Gibbons, L.C. Roudsari, J.E. Saik, Z.H. Rizvi, J.D. Roybal, et al., A synthetic matrix with independently tunable biochemistry and mechanical properties to study epithelial morphogenesis and EMT in a lung adenocarcinoma model, Cancer Res. 72 (2012) 6013–6023. doi:10.1158/0008-5472.CAN-12-0895.

[22]

L.B. Weiswald, D. Bellet, V. Dangles-Marie, Spherical cancer models in tumor biology, Neoplasia. 17 (2015) 1–15. doi:10.1016/j.neo.2014.12.004.


Page 26 of 39

Page 27 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


[23]

A.C. Luca, S. Mersch, R. Deenen, S. Schmidt, I. Messner, K.L. Schäfer, et al., Impact of the 3D Microenvironment on Phenotype, Gene Expression, and EGFR Inhibition of Colorectal Cancer Cell Lines, PLoS One. 8 (2013). doi:10.1371/journal.pone.0059689.

[24]

K. Chitcholtan, P.H. Sykes, J.J. Evans, The resistance of intracellular mediators to doxorubicin and cisplatin are distinct in 3D and 2D endometrial cancer., J. Transl. Med. 10 (2012) 38. doi:10.1186/1479-5876-10-38.

[25]

R. Edmondson, J.J. Broglie, A.F. Adcock, L. Yang, Three-dimensional cell culture systems and their applications in drug discovery and cell-based biosensors., Assay Drug Dev. Technol. 12 (2014) 207–18. doi:10.1089/adt.2014.573.

[26]

M. Shingyoji, S. Ando, H. Nishimura, T. Nakajima, VEGF in Patients with Non-small Cell Lung Cancer during Combination Chemotherapy of Carboplatin and Paclitaxel, 2640 (2009) 2635–2639.

[27]

S. Sörenson, H. Fohlin, A. Lindgren, M. Lindskog, B. Bergman, C. Sederholm, et al., Predictive role of plasma vascular endothelial growth factor for the effect of celecoxib in advanced non-small cell lung cancer treated with chemotherapy., Eur. J. Cancer. 49 (2013) 115–20. doi:10.1016/j.ejca.2012.07.032.

[28]

F. Mac Gabhann, A.S. Popel, Model of competitive binding of vascular endothelial growth factor and placental growth factor to VEGF receptors on endothelial cells., Am. J. Physiol. Heart Circ. Physiol. 286 (2004) H153–64. doi:10.1152/ajpheart.00254.2003.

[29]

K.M. Oltmanns, H. Gehring, S. Rudolf, B. Schultes, C. Hackenberg, U. Schweiger, et al., Acute hypoxia decreases plasma VEGF concentration in healthy humans., Am. J. Physiol. Endocrinol. Metab. 290 (2006) E434–E439. doi:10.1152/ajpendo.00508.2004.

[30]

J. Malda, J. Rouwkema, D.E. Martens, E.P. Le Comte, F.K. Kooy, J. Tramper, et al., Oxygen Gradients in Tissue-Engineered PEGT/PBT Cartilaginous Constructs: Measurement and Modeling, Biotechnol. Bioeng. 86 (2004) 9–18. doi:10.1002/bit.20038.

[31]

C. a. Durst, M.P. Cuchiara, E.G. Mansfield, J.L. West, K.J. Grande-Allen, Flexural characterization of cell encapsulated PEGDA hydrogels with applications for tissue engineered heart valves, Acta Biomater. 7 (2011) 2467–2476. doi:10.1016/j.actbio.2011.02.018.

[32]

M.J. Schliekelman, D.L. Gibbons, V.M. Faca, C.J. Creighton, Z.H. Rizvi, Q. Zhang, et al., Targets of the tumor suppressor miR-200 in regulation of the epithelial-mesenchymal transition in cancer, Cancer Res. 71 (2011) 7670–7682. doi:10.1158/0008-5472.CAN-110964.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

[33]

R.L. Saunders, D. a Hammer, Assembly of Human Umbilical Vein Endothelial Cells on Compliant Hydrogels, Cell Mol Bioeng. 3 (2010) 60–67. doi:10.1007/s12195-010-01124.Assembly.

Table 1. Comparison of the behavior of lung adenocarcinoma cells studied. For 344SQ, A549, and 393P cells, the cancer cell type, species of origin, mutations, and in vivo behavior are described.


Page 28 of 39

Page 29 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1. 393P and A549 spheroid formation over time in culture. 393P and A549 cells form spheroids in 3D culture in PEG hydrogels. (a) Images of 393P and A549 on days 4, 8, 12, and 16 (red – phalloidin; blue – DAPI; scale bar = 100 μm). 393P spheroid area (b) quantified at days 4, 8, 12, and 16. For (b), 2-3 images were taken per time point as repeated measures, n=3 hydrogels per time point; no significant differences were determined by a one-way ANOVA (p < 0.05). For (c), 4 images were taken per time point as repeated measures, n=3 gels per time point; letters indicate statistically different groups as determined by a oneway ANOVA and subsequent Tukey-Kramer HSD tests (p < 0.05). All values are reported as mean ± s.d.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 2. 393P lumenization of lung adenocarcinoma spheroids over time in culture. 393P cell clusters exhibit central clearing and lumenal organization in 3D in PEG hydrogels. (a) Image of 393P clusters with centrally cleared lumen (red – phalloidin; blue – DAPI; scale bar = 50 μm); (b) Quantification of % of 393P clusters with central lumenal clearing (100 μm from 3 z-stack images were analyzed per time point as repeated measures, n=3 hydrogels per time point; values are reported as mean ± s.d.; no statistical significances were detected at p < 0.05 as determined by a one-way ANOVA).


Page 30 of 39

Page 31 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Angiogenic growth factor secretion by lung adenocarcinoma cells. VEGF, PDGF-BB, and FGFb secretion from 344SQ, A549, and 393P cells at day 16 in culture. *p < 0.05, **p < 0.01; statistical analysis was evaluated with a one-way ANOVA, followed by subsequent Tukey’s HSD tests; n=6 hydrogels (344SQ), n=3 hydrogels (393P and A549). Angiogenic growth factor secretion for 344SQ was reproduced from ref 6. Copyright 2016 Nature Publishing Group.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. Vessel cell-cancer cell interactions in V-C hydrogels. (a) Image of vessel cells interacting with A549 clusters in the tumor angiogenesis model (red – phalloidin; blue – DAPI; cyan – VE-cadherin; scale bar = 100 μm), (b) Comparison of the number of vessel cell-cancer cell interactions in V-C hydrogels for 344SQ, A549, and 393P cells; no statistically significant differences were observed as determined by a one-way ANOVA with p < 0.05. Vessel-cancer interactions / hydrogel data for 344SQ was reproduced from ref 6. Copyright 2016 Nature Publishing Group.


Page 32 of 39

Page 33 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. Vessel cell-393P cell interactions in V-C hydrogels. (a) Images of vessel cell-cancer cell interactions at the hydrogel interface in V-C hydrogels (red – phalloidin; blue – DAPI; cyan – VE-cadherin; scale bars = 100 μm) and analysis of 393P cluster morphology changes based on location in V-C hydrogels: (b) Cluster area histogram broken down into bins. (c) Cluster circularity histogram broken down into bins. * next to a labeled group in the figure legend indicates statistical significance as compared to the interface group as determined by Kolmorogov-Smirnov tests: p < 0.0125 (Bonferroni corrected); 3 images were taken per hydrogel as repeated measures across 3 hydrogels, 975 total clusters analyzed.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 34 of 39

Page 35 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. Vessel cell-A549 cell interactions in V-C hydrogels. (a) Image of vessel cells interacting with A549 clusters in the tumor angiogenesis model depicting different morphologies of A549 clusters observed (red – phalloidin; blue – DAPI; cyan – VE-cadherin; scale bar = 100 μm) and analysis of A549 morphology changes based on location in V-C hydrogels: (a) Cluster area histogram broken down into bins. (b) Cluster circularity histogram broken down into bins. No statistically significant differences as compared to the interface group were determined by Kolmorogov-Smirnov tests: p < 0.0125 (Bonferroni corrected); 3 images were taken per hydrogel as repeated measures across 3 hydrogels, 279 total clusters analyzed.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 ACS Paragon Plus Environment

Page 36 of 39

Page 37 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 7. Cluster area versus cluster circularity plotted for the interface region of V-C hydrogels for all 3 cell types studied: 344SQ, 393P, and A549. (344SQ – blue X; 393P – red cross; A549 – black circle). Cluster area versus cluster circularity data for 344SQ was reproduced from ref 6. Copyright 2016 Nature Publishing Group.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Table 2. Summary of outcomes for 3 lung cancer cell lines studied.


Page 38 of 39

Page 39 of 39

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


For Table of Contents Use Only


Lung Adenocarcinoma Cell Responses in a 3D in Vitro Tumor

Recommend Documents