Effects of Shear Stress Gradients on Ewing Sarcoma Cells Using 3D

Feb 1, 2017 - Average shear stress (cPa) per layer in each 3D printed scaffold ...... Karageorgiou , V.; Kaplan , D.Porosity of 3d biomaterial scaffol...
0 downloads 0 Views 3MB Size
Download PDF
Article pubs.acs.org/journal/abseba

Effects of Shear Stress Gradients on Ewing Sarcoma Cells Using 3D Printed Scaffolds and Flow Perfusion Jordan E. Trachtenberg,†,∇ Marco Santoro,‡,∇ Cortes Williams, III,§ Charlotte M. Piard,‡ Brandon T. Smith,† Jesse K. Placone,∥ Brian A. Menegaz,⊥ Eric R. Molina,† Salah-Eddine Lamhamedi-Cherradi,⊥ Joseph A. Ludwig,⊥ Vassilios I. Sikavitsas,§ John P. Fisher,‡ and Antonios G. Mikos*,†,# †

Department of Bioengineering, Bioscience Research Collaborative − MS 142, Rice University, 6500 Main Street, Houston, Texas 77030, United States ‡ Fischell Department of Bioengineering, Jeong Kim Engineering Building, University of Maryland, 8228 Paint Branch Drive, College Park, Maryland 20742, United States § Stephenson School of Biomedical Engineering, University of Oklahoma, 202 West Boyd Street, Norman, Oklahoma 73019, United States ∥ Department of Bioengineering, University of California, San Diego, 9500 Gilman Drive #0412, La Jolla, California 92093, United States ⊥ Department of Sarcoma Medical Oncology, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, Texas 77030, United States # Department of Chemical and Biomolecular Engineering, Rice University, 6100 Main Street, Houston, Texas 77005, United States S Supporting Information *

ABSTRACT: In this work, we combined three-dimensional (3D) scaffolds with flow perfusion bioreactors to evaluate the gradient effects of scaffold architecture and mechanical stimulation, respectively, on tumor cell phenotype. As cancer biologists elucidate the relevance of 3D in vitro tumor models within the drug discovery pipeline, it has become more compelling to model the tumor microenvironment and its impact on tumor cells. In particular, permeability gradients within solid tumors are inherently complex and difficult to accurately model in vitro. However, 3D printing can be used to design scaffolds with complex architecture, and flow perfusion can simulate mechanical stimulation within the tumor microenvironment. By modeling these gradients in vitro with 3D printed scaffolds and flow perfusion, we can identify potential diffusional limitations of drug delivery within a tumor. Ewing sarcoma (ES), a pediatric bone tumor, is a suitable candidate to study heterogeneous tumor response due to its demonstrated shear stress-dependent secretion of ligands important for ES tumor progression. We cultured ES cells under flow perfusion conditions on poly(propylene fumarate) scaffolds, which were fabricated with a distinct pore size gradient via extrusion-based 3D printing. Computational fluid modeling confirmed the presence of a shear stress gradient within the scaffolds and estimated the average shear stress that ES cells experience within each layer. Subsequently, we observed enhanced cell proliferation under flow perfusion within layers supporting lower permeability and increased surface area. Additionally, the effects of shear stress gradients on ES cell signaling transduction of the insulin-like growth factor-1 pathway elicited a response dependent upon the scaffold gradient orientation and the presence of flow-derived shear stress. Our results highlight how 3D printed scaffolds, in combination with flow perfusion in vitro, can effectively model aspects of solid tumor heterogeneity for future drug testing and customized patient therapies. KEYWORDS: 3D printing, pore size gradient, computational fluid dynamics, poly(propylene fumarate), tumor model, insulin-like growth factor-1

■

INTRODUCTION The convergence of tissue engineering and cancer biology has given a precedent toward the development of 3D in vitro models.1−3 In support of these recent reviews, 3D scaffolds mimic the tumor microenvironment more accurately than conventional monolayer cultures, where cells cultured in 2D systems are intrinsically unable to capture the complex 3D © XXXX American Chemical Society

Special Issue: Tissue Engineering and Biomaterials Approaches to Tumor Modeling Received: October 17, 2016 Accepted: February 1, 2017 Published: February 1, 2017 A

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering interactions present within the native tumor niche.4 Both hydrogels and synthetic fiber-based scaffolds have been used as 3D models to culture cancer cells. Fiber-based 3D scaffolds for bone tumor modeling are primarily fabricated by conventional processing methods, such as electrospinning.3 Electrospun scaffolds have been successfully used for tumor modeling, as they mimic the microarchitectural cues present in vivo.5 However, the small pore size and the limited diffusion of nutrients prevent cell migration within these systems.5 These shortcomings have been recently overcome by using 3D electrospun scaffolds in concert with a flow perfusion bioreactor, which induced cell proliferation and a homogeneous cell distribution when compared with static conditions.6,7 Yet, the ability to control the architecture of traditionally fabricated scaffolds, including pore size, pore interconnectivity, and regularity of pores and fibers, is somewhat limited.8 3D printing, however, is an ideal processing technique to control scaffold architecture and incorporate complex internal geometries and interconnected pores. To model the Ewing sarcoma (ES) tumors in vitro, we used a polymer, poly(propylene fumarate) (PPF), which can represent the bone tumor niche due to its suitability for bone-related applications9−16 and robust mechanical properties.17 In addition to providing a suitable material for bone tissue engineering, the combination of this polymer with extrusion-based printing has enabled the fabrication of scaffolds with controlled pore sizes.18 3D printing technologies can regulate the pore size within each layer of a scaffold, permitting the fabrication of 3D gradient structures.19−21 Scaffolds with 3D printed pore size gradients can then be combined with flow perfusion to model the complex heterogeneity of ES tumors. Solid tumors are highly heterogeneous structures, owing in part to their complex organization of cells and varying intratumoral permeability, which can create diffusional barriers for mass transport and drug delivery.22 By modeling physiological levels of biomechanical stimulation present in the bone tumor niche, a recent study has demonstrated a shear stress-dependent secretion of insulin-like growth factor-1 (IGF1) in a 3D culture of ES cells under flow perfusion, resulting in a shear stress-dependent drug response against an IGF-1 receptor (IGF-1R) inhibitor.6 These findings were extremely relevant as the IGF-1/IGF-1R pathway plays a central role in ES cell survival, progression, and drug response.23−25 Although most cancer biologists emphasize genetic mutations and protein expression as key markers for tumor progression and response to therapy, we recognize that the interaction of tumor cells with their microenvironment can impact their phenotype and drug response, as seen in previous work.6,7,26 Furthermore, the factors of fluid shear stress and macro-scale architecture, both of which represent cues from the in vivo microenvironment, may have synergistic effects on tumor phenotype, protein expression, and cell proliferation. In this study, we have leveraged the benefits of flow perfusion and extrusion-based 3D printing to mimic the microenvironmental cues within a ES tumor. 3D printed scaffolds with pore size gradients can mimic the heterogeneous permeability of solid tumors and subsequent organization of tumor cells in response to varying mass transport conditions within a single 3D model. Instead of impeding cell migration with small pore sizes, we intended to model larger pore sizes (0.2−1 mm) found commonly in cortical and trabecular bone,27 which is a common site for primary ES tumors.28 Depending on the location of the solid tumor, pore sizes within the long bones

can range between 0.01 and 0.5 mm for compact bone29 and up to 1 mm for trabecular bone.27 By controlling the scaffold architecture using 3D printing techniques, we can also control the shear stress environment that the cells experience under flow perfusion. In this work, we hypothesized that a 3D gradient scaffold could be used to elicit a shear stress gradient and affect the overall tumor phenotype. Specifically, ES cells would experience different shear stress depending on (1) their location within the scaffold and (2) the orientation of the pore size gradient. The goal of this study was to use 3D printing in order to control the in vitro microenvironment and effectively direct the ES cell phenotype, which would have tremendous benefits in the future for modeling intratumoral heterogeneity in cancer patients30 and testing in vitro tumor response to cancer therapeutics.31

■

MATERIALS AND METHODS

Experimental Design. We designed 3D printed PPF scaffolds with three different gradient orientations (LMS, MMM, and SML), as shown in Figure 1. Pore size gradients were obtained by combining

Figure 1. Cross-sectional view of gradient scaffold orientations (3 distinct pore size layers with 2 arrays per layer) with the following pore sizes: large (L) = 1 mm, medium (M) = 0.6 mm, and small (S) = 0.2 mm. Layers were printed separately and stacked on top of each other within the bioreactor cassette to enable delamination during the biological analyses. three individual 3D printed PPF layers (two arrays per layer) with large (L), medium (M), or small (S) pore size. Computational modeling was used to estimate the shear stress gradients experienced by cells cultured under flow perfusion conditions. Scaffolds seeded with ES cells were cultured under static conditions or within a flow perfusion bioreactor. Half of the medium was replaced on days 3, 6, and 8 of culture. After 10 days of culture, both samples cultured under static or flow perfusion conditions were harvested. Each layer of the sample was delaminated and analyzed for DNA content and protein expression via flow cytometry. After 3, 6, 8, and 10 days of culture, the media from static and flow conditions were collected, frozen, and later evaluated for cumulative IGF-1 release via enzyme-linked immunosorbent assay (ELISA). PPF Fabrication and Scaffold Printing. PPF was synthesized in a step-growth polymerization reaction as previously described.18,32 The number-average molecular weight (Mn, 2,280 ± 23 Da) and polydispersity index (PDI, 1.72 ± 0.02) were characterized by gel permeation chromatography after purification (n = 3). The PPF printing solution was prepared following previous work.18 Briefly, PPF was mixed with diethyl fumarate (DEF, Sigma, St. Louis, MO) in an 85:15 wt % ratio. First, a photoinitiator, phenylbis(2,4,6-trimethylbenzoyl)-phosphine oxide (BAPO, BASF, Florham Park, NJ), was dissolved in DEF (with 1 wt % BAPO) by vortexing and was mixed with warm PPF (∼70 °C) to permit UV cross-linking. 10 × 10 × 0.54 mm3 square prisms were designed (SolidWorks, Waltham, MA), sliced into two layers (Bioplotter RP, EnvisionTEC, Gladbeck, Germany), and printed at the following conditions: extruder temperature (55 °C), platform temperature (5 °C), print head speed (5 mm/s), needle offset (0.25 mm), and syringe tip diameter (0.34 mm). Pressure was varied (0.8−1.6 bar) in order to produce fibers with diameters of 0.320−0.360 mm, which corresponds to an B

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering approximate 6% error compared to the syringe tip diameter. The spacing between fibers was adjusted to achieve distinct pore sizes of 0.2 mm (small, S), 0.6 mm (medium, M), and 1 mm (large, L) based on a biological range of bone pore sizes.27 Each printed layer was cross-linked with UV light (single projection, 10 s exposure, 70 mm height) following previous methods.18 Scaffolds were printed on plastic tape (Scotch, Long Lasting Moving & Storage Packaging Tape) to allow adhesion to the build plate during printing and easy removal after postcuring (Otoflash, EnvisionTEC, Gladbeck, Germany). Postcuring was required to fully cross-link scaffolds and was performed as previously described.18 Previous work also details the settings used in the Bioplotter software.18 Images of scaffolds were taken with a stereomicroscope (MZ6, Leica Microsystems, Wetzlar, Germany) and used to quantify pore size and fiber diameter (n = 12 scaffolds per group, 5 measurements per image) (ImageJ, NIH) following previously reported methods.18,20 Briefly, pore size and fiber diameter measurements were quantified as the average value in the x and y directions (Figure S1). Porosity Measurements. Porosity was quantified (n = 12 per group) using gravimetric analysis following previous methods.20 Briefly, sample dimensions (length, L; width, W; thickness, T) and weight (wscaffold) were measured and used to calculate the porosity (ε) according to eq 1, where ρscaffold is the scaffold density, and ρmaterial is the density of the material. The respective densities of PPF (1.267 g/ mL),33 DEF (1.052 g/mL) (Sigma), and BAPO (1.19 g/mL) (BASF) were used to estimate ρmaterial. ρ wscaffold ε = 1 − scaffold = 1 − ρmaterial LWTρmaterial (1)

Figure 2. Bioreactor cassette design and boundary conditions. (A) Side-view of the bioreactor cassette (one scaffold (yellow) at the bottom of a channel and boundary conditions used for computational simulations). The channel below the scaffold is a cylinder with diameter = 7 mm. For the flow perfusion experiment, O-rings and a stainless steel mesh were placed on top of the bioreactor cassette to prevent scaffolds from floating. (B) Top view of the bioreactor cassette (3D CAD model, diameter = 11.1 cm), showing all six scaffold channels. Each multilayer scaffold (indicated in yellow in panel A) has dimensions of approximately 10 × 10 × 1.2 mm3 (length × width × height) and are press-fit into the bottom of each of six scaffold channels. Six scaffolds were loaded into the bioreactor cassette (B) and then placed in a bioreactor (flow perfusion) or a beaker (static) for the 10 day experiment.

Scanning Electron Microscopy (SEM). Representative images of the 3D printed scaffolds were taken using SEM prior to cell seeding. Pore and fiber measurements were obtained from light microscope images, but SEM provided better image contrast since the fibers were transparent. Samples (n = 1 per formulation) were sputter-coated with 10 nm of gold (Denton Desk V, Moorestown, NJ). SEM (FEI Quanta 400 ESEM FEG, FEICo, Hillsboro, OR) images were obtained at 5.00 kV (high voltage, HV), 6 μs dwell time, and 2.0 mm spot size with 50, 100, and 1,000× magnification following previous methods.34 Computational Modeling of Shear Stress in 3D Printed Scaffolds. Each layer of the scaffolds (L, M, and S) was scanned nondestructively with microcomputed tomography (μCT) using a commercial system to obtain 2D image slices (Quantum FX, PerkinElmer, Waltham, MA; L10101, Hamamatsu Photonics, Japan; PaxScan 1313, Varian Medical Systems, Palo Alto, CA). The images of sequential layers were then filtered, thresholded, and stacked (LMS, MMM, and SML) using the open-source visualization software 3D Slicer (www.slicer.org) to form the 3D reconstructions. For each scaffold orientation, slices were excised from the edges of the reconstruction to avoid edge effects in the computational fluid dynamic (CFD) simulations. The exact size of the resulting digital scaffold was different for each case; however, the typical size was 10 × 10 × 1.2 mm3 (length × width × height). Simulations were then conducted with flat velocity profiles based on a flow rate of 0.6 mL/min, no-slip boundary conditions imposed at the walls, and also an outlet with constant static pressure.35,36 The culture medium was modeled as an incompressible Newtonian fluid with a dynamic viscosity of 1 cP, according to previous studies.6 In order to avoid any unwanted entrance effects and allow for fully developed laminar flow, we added a 4.8 mm long inlet channel in front of the scaffolds (Figure 2A), which satisfies the conditions for laminar flow in a noncircular pipe.37,38 Hydraulic radius (Dh), cross-sectional area (A), perimeter of the channel (P), Reynolds number (Re), average velocity (⟨v⟩), medium density (ρ), medium viscosity (μ), and entrance length (EL) and are defined in (eq 2−eq 4):

Dh =

4A = 10 mm P

(2)

Re =

2Dh⟨v⟩ρ ≪1 μ

(3)

EL = 0.035Re· Dh ≪ 4.8 mm

(4)

Simulations were performed using two methods to ensure the accuracy of the results: Fluent 16.2 (ANSYS, Inc.) and (2) a custom in-house code utilizing the lattice Boltzmann method. Both methods have been extensively utilized for computed surface shear stresses on μCT reconstructions.35,36,39,40 Bioreactor Cassette Design. Bioreactor cassettes were designed (SolidWorks) (Figure 2B) and custom machined (708580 JBM-5, Jet, La Vergne, TN; and Shapeoko 2, www.inventables.com) from a 1/2 × 12 × 12 in3 polycarbonate sheet (8574K32, McMaster Carr, Douglasville, GA). Cassettes were placed in an integrated medium reservoir and flow chamber developed in previous work.41 In order to prevent scaffolds from floating during flow perfusion, a stainless steel mesh was placed in between the O-rings and the top of the cassette (Figure 2A), as previously reported.34 Expansion, Seeding, and Culture of ES Cells on 3D Printed Scaffolds under Static and Flow Perfusion Conditions. The human ES cell line (TC71) was available from our institution (MDACC) and has been previously characterized.6 ES cells were cultured in Roswell Park Memorial Institute (RPMI) medium 1640 (Mediatech, Houston, TX). RPMI medium 1640 was supplemented with 10% FBS (Gemini Bioproducts, West Sacramento, CA) and antibiotics (100 IU/mL penicillin and 100 μg/mL streptomycin; Gibco, Waltham, MA) (referred to as “complete medium”). At the beginning of the experiment, cells were detached with 0.05% trypsinEDTA (Gibco, Waltham, MA) and counted with a hemocytometer. 3D printed scaffolds were sterilized in 24-well ultralow attachment plates (Costar, Corning, Corning, NY) in a 12 h ethylene oxide cycle (Anderson Sterilizers, Haw River, NC) and allowed to degas for 24 h prior to cell culture. The scaffolds were then prewet in a sterile gradient of ethanol (100, 75, 50, 25, and 0%), followed by 3 washes of phosphate-buffered saline (PBS, Gibco, Waltham, MA) and were incubated on a rotating shaker overnight in 1 mL of complete medium. On the following day, 2 × 105 cells were seeded on each scaffold in 1 mL complete medium (ultralow attachment plates) and placed on a rotating table in an incubator overnight to facilitate cell adhesion. Cassettes were sterilized in ethylene oxide and prewet in sterile deionized water prior to cell culture as described above. All other bioreactor components were sterilized with a 20 min autoclave cycle. After 12 h of incubation, seeded layers were press-fit into cassettes in C

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

bioreactor conditions). For ELISA data, Student’s t test was used to make pairwise comparisons between static and bioreactor conditions for specific scaffold/time point combinations. DNA content and protein expression data were compared using both a one-way ANOVA (keeping some factors fixed, such as scaffold orientation (LMS, MMM, and SML) or layer location (top, middle, and bottom)) and fullfactorial analysis (considering all factors simultaneously) with Tukey’s HSD within static and bioreactor conditions following previously reported methods.42,43 In order to compare results within specific scaffold layers across culture conditions (e.g., top layer of LMS orientation under static vs flow perfusion), pairwise comparisons were made using Student’s t test (p < 0.05). All statistical analyses were performed using JMP Pro 10 software (SAS Institute Inc., Cary, NC). Refer to Supporting Information, pp S7−72, for all raw reports of statistical analysis.

an orientation following Figure 1 in either static (loaded bioreactor cassette in a 400 mL beaker and 40 mL of complete medium added with spacers on the bottom) or perfusion (loaded bioreactor cassette with 50 mL complete medium added) conditions. The bioreactor setup has been described in previous work.41 Cassettes under perfusion were subject to a flow rate of 0.6 mL/min and incubated at 37 °C and 5% (v/v) CO2 (HeraCell 150i; ThermoScientific, Waltham, MA). DNA Assay. A DNA assay was conducted (n = 6 samples per group) with a Quant-iT PicoGreen dsDNA kit (Thermo Scientific, Waltham, MA) following previous protocols.6 Each layer was separated and frozen in 1 mL of deionized water prior to processing for the assay. Samples were subjected to three freeze/thaw cycles (15 min each), vortexed, and sonicated for 10 min. Eight lambda DNA standards were prepared by serial dilution (0−2 μg/mL). 100 μL of TE buffer solution, 50 μL of sample, and 150 μL of PicoGreen solution were added to a black 96 well plate, and fluorescence was measured with a fluorescent plate reader (FLx800 Fluorescence Microplate Reader; BioTek Instruments, Winooski, VT). DNA content was calculated based on the standard curve. Flow Cytometry. 3D Detachment. At the end of the experiment, cassettes were disassembled. The top, middle, and bottom layers of each scaffold were delaminated for flow cytometry analysis. Two delaminated layers (biological replicates) were pooled together to achieve three biological replicates (n = 6 total replicates pooled into 3). Samples were pooled in order to ensure the detection of viable cells during the flow cytometry analysis. Layers were placed in 300 μL of PBS during separation. The PBS solution was aspirated into a 96-well plate and centrifuged at 800g for 1 min to collect any immediately dissociated cells. Cells were then dissociated from the scaffolds by alternating two incubations (20 and 10 min) in Accutase buffer (BD Biosciences, San Jose, CA) with centrifugation. Then, cell pellets were resuspended in PBS (200 μL) and centrifuged twice before staining with antibody. Staining was done with 100 μL of PBS. Cells were stained with phycoerythrin (PE)-conjugated mouse antihuman CD221 antibody (BD Pharmingen; BD Biosciences, San Jose, CA) in order to characterize the expression of insulin-like growth factor-1 receptor (IGF-1R). 2D Detachment. TC71 cells lifted from 2D culture flasks were briefly exposed to Accutase (1 mL/T75 flask, 30 s−1 min), centrifuged (600g for 5 min), resuspended in PBS (200 μL), and kept on ice until staining with cells from the 3D detachment. Staining was done with 100 μL PBS. Gating and Analysis. TC71 cells lifted from 2D culture flasks (unstained and single-stained IGF-1R samples) were used as controls to define the values of forward-scattering (FSC) and side-scattering (SSC), which roughly translate to the size and shape of the cells. Once this “gating” was performed, the FSC and SSC boundaries were kept constant among all groups in order to make meaningful comparisons. Following previous protocols for TC71 ES cells,6 only the fraction of cells falling within the “live cell” region were considered (Figure S2). Stained cells were evaluated using a flow cytometer (FACSCanto II; BD Biosciences, San Jose, CA), and data were analyzed using CytoBank software (www.cytobank.org; Mountain View, CA). Enzyme-Linked Immunosorbent Assay (ELISA). After 3, 6, 8, and 10 days of culture, half of the medium from each group (20 mL for static and 25 mL for flow perfusion) was harvested (n = 3 replicates per scaffold orientation and culture condition) to quantify the concentration of IGF-1 ligand following the manufacturer’s instructions (ELISA kit DuoSet; R&D Systems, Minneapolis, MN). Cumulative release was calculated in terms of the amount of soluble IGF-1 per scaffold, taking into account the volume of media and number of scaffolds for each time point and assuming negligible surface adsorption. The optical density of each sample was measured using a microplate reader (DTX880; Beckman Coulter, Brea, CA). Statistical Analysis. A one-way analysis of variance (ANOVA) and Tukey’s Honestly Significant Difference (HSD) test were used to compare the mean ± standard deviation (p-value of 0.05). No statistical differences were observed across the respective top, middle, and bottom layers.

generate a 3D reconstruction. Computational modeling indicated that the average shear stress was higher in the top layer compared to that in middle and bottom layers, regardless of the orientation (p < 0.05). While no statistical significance was found among layers in the LMS orientation, in both MMM and SML orientation the top layers registered the highest values of wall shear stress. Computational models of average wall shear stress within the scaffolds layers for LMS, MMM, and SML orientations are represented in Figure 5. In a few instances, localized regions of higher shear stress were observed near the edges (e.g., SML top

Figure 5. Top view of 3D printed scaffold layers. Average wall shear stress per layer (top, middle, and bottom) for all gradient orientations (LMS, MMM, and SML). Wall shear stress is mapped as a color distribution (heat map) on the top surface of each layer. Heat map indicates the distribution of high (red, 8 cPa) to low (blue, 0 cPa) shear stresses. The heat map legend applies to all layers. E

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Figure 6. Effect of shear stress gradient on Ewing sarcoma (ES) cell proliferation after 10 days of static and flow perfusion culture. DNA content after 10 days of cell culture in static (A, n = 6) and flow perfusion conditions (B, n = 3). Percentage of live cells measured via flow cytometry in static (C, n = 3) and flow perfusion conditions (D, n = 3). Data are reported as the mean + standard deviation. & indicates statistical significance (p < 0.05) by full-factorial analysis within static or flow perfusion conditions. * indicates statistically greater, comparing static and flow perfusion conditions within each orientation/layer tested (p < 0.05) by Student’s t test.

Figure 7. Effect of shear stress gradients on Ewing sarcoma (ES) expression of insulin-like growth factor-1 receptor (IGF-1R). Live ES cells expressing total insulin-like growth factor-1 receptor (IGF-1R) under static (A, n = 6) and flow perfusion conditions (B, n = 3). Data are reported as the mean + standard deviation. & indicates statistical significance (p < 0.05) by full-factorial analysis within static conditions. No statistical differences were observed within flow perfusion conditions nor between pairwise comparisons of static and flow perfusion.

Figure 8. Effect of shear stress gradients on Ewing sarcoma (ES) expression of insulin-like growth factor-1 (IGF-1). Cumulative release of IGF-1 over time under static (A) and flow perfusion conditions (B). Data are reported as the mean + standard deviation (n = 3). * indicates statistical significance between static and flow perfusion conditions within each orientation tested; & indicates significance within each time point (p < 0.05, n = 3).

F

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

gradients (LMS, MMM, and SML) under static or flow perfusion conditions. ES is a pediatric bone sarcoma that has been shown to exhibit shear stress-dependent protein expression,6 serving therefore as a model cell type for our purposes. Scaffold analyses showed that the porosity of each layer was comparable to that of bone tissue engineering scaffolds in previous studies (ranging from 19 to 60%),27 thus demonstrating that 3D printed scaffolds can suitably model primary bone malignancies. Notably, the fiber diameter and surface roughness of the fibers are known to contribute to cell phenotype, spreading, and attachment.46 We focused on mimicking the macro-porous structure of bone, but microand nanoscale porosity within the scaffold fibers would likely influence cell attachment, cytoskeletal organization, and cell proliferation.46 Consistent with other investigations, the enhanced nutrient supply provided by flow perfusion promoted cell proliferation in all experimental groups cultured under flow perfusion conditions compared to static cultures.6,7 Interestingly, scaffold porosity affected cell growth under flow perfusion, with higher cell content in S layers than in either M or L layers. This phenomenon can be ascribed to the smaller pore size and, hence, the higher surface area, available to cell growth. These results corroborate previous reports with the same seeding efficiency and similar pore sizes, where scaffolds with smaller pore size had better seeding due to their lower permeability.45 In a flow perfusion bioreactor under constant flow rates, scaffold architecture (e.g., pore size and porosity) and anisotropy dramatically impact the shear stress profile within scaffolds.7,39,40 Indeed, we found in this study that the average shear stress in 3D printed layers depended on both pores size, as well as scaffold orientation. In each orientation, cells in the top layer experience the highest level of shear stress, and the pressure drop within lower layers likely caused a decline in shear stress. At the same time, wall shear stress should increase with decreasing porosity (i.e., increasing hydraulic resistance to flow) and, hence, be maximized in S layers compared to M and L layers. These conflicting phenomena resulted in homogeneous levels of shear stress in orientation LMS, where the lower layers have smaller pores, and thus, constant shear stress was maintained, despite the loss in kinetic energy. The same was not true for the MMM orientation, where the homogeneous pore size distribution resulted in lower shear stress levels in the lower two layers, again, due to the pressure drop. Conversely, we expected differences in shear stress to be exacerbated in the SML orientation because the S layer is exposed to the highest kinetic energy. The lack of a clear statistical significance among layers in orientation SML can be explained by the presence of edge effects, closed or misaligned pores, and other manufacturing defects that have been shown to affect the shear stress profiles under flow perfusion conditions. 40 The same manufacturing defects contributed to heterogeneous shear stress distribution within single layers, as shown in Figure 5. We observed that scaffold imperfections influenced the shear stress distribution to the extent that they may also affect cell migration, proliferation, and differentiation. While beyond the scope of this study, it would be beneficial to image live cells on the scaffolds over time to see if the cells preferentially migrate to regions of high or low shear stress. Previous investigation in our group already showed that flow perfusion did not affect IGF-1R expression.6,7 This study replicates those data (Figure 7) and shows that shear stress exerts an effect on the IGF-1/IGF-1R axis only via IGF-1, with

higher cell viability under flow perfusion compared to static conditions remain valid. We completed an additional experiment (repeated twice) comparing 2D controls detached with the 2D and 3D detachment protocol and consistently found that the 3D protocol notably decreased cell viability (Figure S4). Effect of Shear Stress Gradient on ES Cell Signaling Transduction of the IGF-1/IGF-1R Pathway. Flow cytometry analysis was then used to identify populations of cells expressing IGF-1R within each layer of LMS, MMM, and SML scaffolds. 2D controls stained for IGF-1R (conjugated to PE) were used to identify IGF-1R+ cells within the live cell populations. We employed a full-factorial analysis to evaluate the main effects of gradient orientation (LMS, MMM, and SML) and layer location (top, middle, or bottom) on fluorescent intensity of PE (IGF-1R histograms), for both static and flow perfusion conditions. As shown in Figure 7, IGF-1R is strongly expressed in all groups regardless of the orientation and/or culture conditions. Under static conditions, IGF-1R expression in group MMM was significantly lower than that in group SML or LMS (p = 0.0443). Yet, >80% of the cells tested positive for IGF-1R (Figure 7A). In Figure 7B (flow perfusion conditions), there are instances where mean + standard deviation exceeds 100%, which is due to the large variation (minimum and maximum IGF-1R+ %) in biological responses within replicates. The IGF-1R pathway activation is, by definition, liganddependent; that is, the presence of IGF-1 ligand is required in order to have phosphorylated (and thus active) IGF-1R within the ES cell culture. We then analyzed the cell culture medium in all groups at different time points via ELISA in order to quantify the IGF-1 ligand concentration (Figure 8). Gradient orientation had a minimal effect on IGF-1 secretion under static conditions. On the contrary, groups LMS and MMM exhibited significantly higher IGF-1 levels than group SML from day 6 onward under flow perfusion conditions, as shown in Figure 8B. The transition from static to flow perfusion conditions resulted in an increased IGF-1 production for all scaffold orientations (LMS, MMM, and SML), mirroring previous data from our laboratory that demonstrated shear stress-mediated IGF-1 secretion in ES cells.6

■

DISCUSSION Tissue-engineered scaffolds with defined architecture and porosity can describe microarchitectural cues present in the native bone as well as in the bone tumor microenvironment.4,27,44 In particular, pore morphology and porosity can impact cell attachment, alter the permeability of media and nutrients, and facilitate cell migration.45 Extrusion-based 3D printing techniques can further facilitate the design of complex scaffolds with spatial gradients in porosity and pore size, which are able to mimic the architecture of cortical and trabecular bones.20,21 When coupled with scaffolds that model the complex geometry of bone, flow perfusion bioreactors can expose bone tumor cells in vitro to physiologically relevant levels of shear stress normally found in vivo in the bone niche.6,7 However, few studies have investigated the combination of scaffolds composed of pore size gradients with flow perfusion bioreactors to achieve complex, intrascaffold shear stress environments, which we hypothesized would elicit a gradient phenotypic response in the tumor cells. In this work, we tested this hypothesis by culturing ES cells on 3D printed PPF scaffolds with three different pore size G

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

stasis.22 Although it was beyond the scope of this study, we expect that drug testing against IGF-1R would also depend on the scaffold orientation. Additionally, the use of 3D printed scaffolds with gradients in porosity and shear stress under flow perfusion would allow observation of a wide spectrum of drug responses4 within a single scaffold. This design strategy would eliminate the need for multiple experimental groups with different scaffold designs, resulting in a platform for future preclinical drug screening. To this end, the effects of pore size and flow rate on drug responsiveness would be an interesting direction for this work.

negligible effect on IGF-1R. Remarkably, the LMS and MMM orientations resulted in higher IGF-1 content than the SML orientation, suggesting that not only the presence of a gradient but also its direction (LMS vs SML) can affect IGF-1 secretion. It is worth emphasizing how both the SML and LMS orientations share the same scaffold porosity and similar levels of shear stress, as shown in Figure 4 and Figure S3. Yet, IGF-1 cumulative release in the LMS group was higher than that in the SML group, suggesting that further mechanisms regulate IGF-1 production differently in these two gradient orientations. Previous work demonstrated that IGF-1 manifests a feedforward behavior that stimulates ES cells to secrete more IGF-1 in an autocrine fashion (i.e., activating the IGF-1R pathway).6,47 Although it remains to be proven, we contend that a similar mechanism might have taken place in the lower layer of the LMS and MMM orientations, where shear stress-mediated IGF-1 secretion from the upper layer promoted an enhanced autocrine release of IGF-1 ligand. ES cells preferentially grew on layers with lower permeability (i.e., S layers), and so the gradient cell distribution within LMS orientation potentially intensified the autocatalytic secretion of IGF-1 as compared to the SML orientation, which had a lower concentration of cells in the bottom layer. Overall, the combination of 3D printed scaffolds and flow perfusion can be leveraged to model the heterogeneous porosity (and permeability) of solid tumors,22,48 which results in the presence of shear stress gradients. Although we could not discern any clear layer-dependent effect in our model, we observed that cell response depended on the scaffold orientation. A possible explanation for this lack of significance among S, M, and L layers in each orientation can be ascribed to the manufacturing defects among layers and among different scaffolds, as well as the lack of significance in porosity between the M and L layers. Together with the fluctuations in biological response that might occur, these confounding effects might have prevented the isolation of specific layer-dependent (and, thus, shear stress-dependent) phenomena. We envision that optimizing the scaffold fabrication process and limiting scaffoldto-scaffold variability would improve the significance of future findings. Printing all layers within a single scaffold could reduce scaffold variability, and it may also promote more uniform shear stress distributions and enhance cell migration and proliferation due to the increased connectivity (i.e., fusion) of layers. However, building a single scaffold would further complicate the separation of sequential layers and maintenance of cell viability and attachment during flow cytometry and biochemical analysis. After minimizing batch-to-batch variability, we envision extending this work to investigate scaffolds with more complex pore size gradients, including a broader range of pore sizes and scaffolds with more than 3 distinct pore sizes (i.e., more than 3 layers). Additionally, different flow perfusion strategies (continuous vs oscillatory flow) could be employed to further enhance the biomimicry of our proposed tumor model. In order to mimic various biomechanical cues in vivo, flow perfusion conditions could be adjusted to investigate cell adaptation49 and attachment50 under oscillatory, pulsatile, or turbulent flows, which would provide other important models for understanding maintenance and progression of tumor phenotype. Previous work has also modeled the effects of flow perfusion rates and scaffold morphology on cell bridging and detachment,51 which, when applied to ES and other tumor cells, may offer insights about cell extravasation and meta-

■

CONCLUSIONS 3D printing was used to fabricate multilayered scaffolds with pore size gradients that mimic the complex heterogeneity of solid tumors. We seeded ES cells on scaffolds with pore size gradients and subjected the cells to physiologically relevant shear stress gradients under flow perfusion. Cell proliferation, protein expression, and ligand secretion were both shear-stressdependent and scaffold-dependent. The coupled observations that both flow perfusion and gradient orientation affect IGF-1 ligand secretion and, therefore, the IGF-1R signaling pathway demonstrates that both culture conditions and scaffold architecture play an interdependent role in directing phenotypic response. Remarkably, local cross-talk among tumor cells exposed to different mechanical stimuli affected the overall IGF-1 ligand production in ES cells. We envision extending this work to investigate scaffolds with more complex pore size gradients (i.e., more than 3 distinct pore size layers) and larger dimensions to enhance the biomimicry of our proposed tumor model. The combination of 3D printed scaffolds with bioreactors offers a unique platform to study cancer biology and drug testing in the future.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsbiomaterials.6b00641. Figures and detailed reports of raw statistical analysis (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Department of Bioengineering, Rice University, P.O. Box 1892 MS 142, Houston, Texas 77251-1892, USA. Tel: +001713-348-5355. Fax: +001-713-348-4244. E-mail: mikos@rice. edu. ORCID

Antonios G. Mikos: 0000-0002-0709-990X Author Contributions ∇

J.E.T. and M.S. contributed equally to the preparation of this manuscript.

Author Contributions

J.E.T. and M.S. designed the study; performed cell culture, bioreactor experiments, and scaffold characterization; performed data analysis; and wrote the manuscript. C.W. designed the study; performed the reconstructions and flow simulations; and wrote the manuscript. C.M.P. and J.K.P. designed the study and fabricated scaffolds. B.A.M., E.R.M., and S.E.L.C. performed the flow cytometry and data analysis; S.E.L.C. H

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

angiogenic injectable bone cement for femoral head osteonecrosis. Biomaterials 2010, 31, 4048−4055. (12) Childers, E. P.; Wang, M. O.; Becker, M. L.; Fisher, J. P.; Dean, D. 3d printing of resorbable poly(propylene fumarate) tissue engineering scaffolds. MRS Bull. 2015, 40, 119−126. (13) Hedberg, E. L.; Kroese-Deutman, H. C.; Shih, C. K.; Crowther, R. S.; Carney, D. H.; Mikos, A. G.; Jansen, J. A. In vivo degradation of porous poly(propylene fumarate)/poly(dl-lactic-co-glycolic acid) composite scaffolds. Biomaterials 2005, 26, 4616−4623. (14) Henslee, A. M.; Spicer, P. P.; Yoon, D. M.; Nair, M. B.; Meretoja, V. V.; Witherel, K. E.; Jansen, J. A.; Mikos, A. G.; Kasper, F. K. Biodegradable composite scaffolds incorporating an intramedullary rod and delivering bone morphogenetic protein-2 for stabilization and bone regeneration in segmental long bone defects. Acta Biomater. 2011, 7, 3627−3637. (15) Henslee, A. M.; Yoon, D. M.; Lu, B. Y.; Yu, J.; Arango, A. A.; Marruffo, L. P.; Seng, L.; Anver, T. D.; Ather, H.; Nair, M. B.; Piper, S. O.; Demian, N.; Wong, M. E.; Kasper, F. K.; Mikos, A. G. Characterization of an injectable, degradable polymer for mechanical stabilization of mandibular fractures. J. Biomed. Mater. Res., Part B 2015, 103, 529−538. (16) Kim, K.; Dean, D.; Wallace, J.; Breithaupt, R.; Mikos, A. G.; Fisher, J. P. The influence of stereolithographic scaffold architecture and composition on osteogenic signal expression with rat bone marrow stromal cells. Biomaterials 2011, 32, 3750−3763. (17) Fong, E. L.; Watson, B. M.; Kasper, F. K.; Mikos, A. G. Building bridges: Leveraging interdisciplinary collaborations in the development of biomaterials to meet clinical needs. Adv. Mater. 2012, 24, 4995− 5013. (18) Trachtenberg, J. E.; Placone, J. K.; Smith, B. T.; Piard, C. M.; Santoro, M.; Scott, D. W.; Fisher, J. P.; Mikos, A. G. Extrusion-based 3d printing of poly(propylene fumarate) in a full-factorial design. ACS Biomater. Sci. Eng. 2016, 2, 1771−1780. (19) Bose, S.; Vahabzadeh, S.; Bandyopadhyay, A. Bone tissue engineering using 3d printing. Mater. Today 2013, 16, 496−504. (20) Trachtenberg, J. E.; Mountziaris, P. M.; Miller, J. S.; Wettergreen, M.; Kasper, F. K.; Mikos, A. G. Open-source threedimensional printing of biodegradable polymer scaffolds for tissue engineering. J. Biomed. Mater. Res., Part A 2014, 102, 4326−4335. (21) Sobral, J. M.; Caridade, S. G.; Sousa, R. A.; Mano, J. F.; Reis, R. L. Three-dimensional plotted scaffolds with controlled pore size gradients: Effect of scaffold geometry on mechanical performance and cell seeding efficiency. Acta Biomater. 2011, 7, 1009−1018. (22) Paulsen, S. J.; Miller, J. S. Tissue vascularization through 3d printing: Will technology bring us flow? Dev. Dyn. 2015, 244, 629− 640. (23) Kolb, E. A.; Gorlick, R. Development of igf-ir inhibitors in pediatric sarcomas. Curr. Oncol. Rep. 2009, 11, 307−313. (24) Tolcher, A. W.; Sarantopoulos, J.; Patnaik, A.; Papadopoulos, K.; Lin, C. C.; Rodon, J.; Murphy, B.; Roth, B.; McCaffery, I.; Gorski, K. S.; Kaiser, B.; Zhu, M.; Deng, H.; Friberg, G.; Puzanov, I. Phase i, pharmacokinetic, and pharmacodynamic study of amg 479, a fully human monoclonal antibody to insulin-like growth factor receptor 1. J. Clin. Oncol. 2009, 27, 5800−5807. (25) Mateo-Lozano, S.; Gokhale, P. C.; Soldatenkov, V. A.; Dritschilo, A.; Tirado, O. M.; Notario, V. Combined transcriptional and translational targeting of ews/fli-1 in ewing’s sarcoma. Clin. Cancer Res. 2006, 12, 6781−6790. (26) Malik, R.; Lelkes, P. I.; Cukierman, E. Biomechanical and biochemical remodeling of stromal extracellular matrix in cancer. Trends Biotechnol. 2015, 33, 230−236. (27) Karageorgiou, V.; Kaplan, D. Porosity of 3d biomaterial scaffolds and osteogenesis. Biomaterials 2005, 26, 5474−5491. (28) Ludwig, J. A. Ewing sarcoma: Historical perspectives, current state-of-the-art, and opportunities for targeted therapy in the future. Curr. Opin. Oncol. 2008, 20, 412−418. (29) Wang, X.; Xu, S.; Zhou, S.; Xu, W.; Leary, M.; Choong, P.; Qian, M.; Brandt, M.; Xie, Y. M. Topological design and additive

wrote the manuscript. J.A.L., V.I.S., J.P.F., and A.G.M. designed the study and wrote the manuscript. Notes

The authors declare the following competing financial interest(s): J.P.F. and J.K.P. are founders of the company 3D Bioworks. All other authors have no competing financial interests.

■

ACKNOWLEDGMENTS The authors acknowledge support from the National Institutes of Health for work in the areas of bone and cartilage tissue engineering (R01 CA180279 and R01 AR068073), as well as from the Armed Forces Institute of Regenerative Medicine (W81XWH-14-2-0004) (to A.G.M). J.P.F. also acknowledges the NIH (R01 AR061460) for funding the Bioplotter printing system. J.E.T. acknowledges funding from the National Science Foundation and the Howard Hughes Medical Institute Graduate Research Fellowships. We would like to thank Dr. Angelo Benedetto and Dr. Dan Harrington for their assistance with flow cytometry and Dr. Josephine Lembong for her helpful feedback on the manuscript.

■

REFERENCES

(1) Loessner, D.; Holzapfel, B. M.; Clements, J. A. Engineered microenvironments provide new insights into ovarian and prostate cancer progression and drug responses. Adv. Drug Delivery Rev. 2014, 79−80, 193−213. (2) Asghar, W.; El Assal, R.; Shafiee, H.; Pitteri, S.; Paulmurugan, R.; Demirci, U. Engineering cancer microenvironments for in vitro 3-d tumor models. Mater. Today 2015, 18, 539−553. (3) Lamhamedi-Cherradi, S.-E.; Santoro, M.; Ramammoorthy, V.; Menegaz, B. A.; Bartholomeusz, G.; Iles, L. R.; Amin, H. M.; Livingston, J. A.; Mikos, A. G.; Ludwig, J. A. 3d tissue-engineered model of ewing’s sarcoma. Adv. Drug Delivery Rev. 2014, 79−80, 155− 171. (4) Fong, E. L. S.; Lamhamedi-Cherradi, S.-E.; Burdett, E.; Ramamoorthy, V.; Lazar, A. J.; Kasper, F. K.; Farach-Carson, M. C.; Vishwamitra, D.; Demicco, E. G.; Menegaz, B. A.; Amin, H. M.; Mikos, A. G.; Ludwig, J. A. Modeling ewing sarcoma tumors in vitro with 3d scaffolds. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 6500−6505. (5) Burdett, E.; Kasper, F. K.; Mikos, A. G.; Ludwig, J. A. Engineering tumors: A tissue engineering perspective in cancer biology. Tissue Eng., Part B 2010, 16, 351−359. (6) Santoro, M.; Lamhamedi-Cherradi, S. E.; Menegaz, B. A.; Ludwig, J. A.; Mikos, A. G. Flow perfusion effects on threedimensional culture and drug sensitivity of ewing sarcoma. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 10304−10309. (7) Santoro, M.; Menegaz, B. A.; Lamhamedi-Cherradi, S. E.; Molina, E. R.; Wu, D.; Priebe, W.; Ludwig, J. A.; Mikos, A. G. Modeling stroma-induced drug resistance in a tissue-engineered tumor model of ewing sarcoma. Tissue Eng., Part A 2017, DOI:10.1089/ten.tea.2016.0369 (8) Trachtenberg, J.; Kasper, F.; Mikos, A., Polymer Scaffold Fabrication. In Principles of Tissue Engineering; Lanza, R., Ed.; Academic Press: Waltham, MA, 2013; pp 423−440. (9) Fisher, J. P.; Vehof, J. W.; Dean, D.; van der Waerden, J. P.; Holland, T. A.; Mikos, A. G.; Jansen, J. A. Soft and hard tissue response to photocrosslinked poly(propylene fumarate) scaffolds in a rabbit model. J. Biomed. Mater. Res. 2002, 59, 547−556. (10) Alge, D. L.; Bennett, J.; Treasure, T.; Voytik-Harbin, S.; Goebel, W. S.; Chu, T.-M. G. Poly(propylene fumarate) reinforced dicalcium phosphate dihydrate cement composites for bone tissue engineering. J. Biomed. Mater. Res., Part A 2012, 100, 1792−1802. (11) Chang, C. H.; Liao, T. C.; Hsu, Y. M.; Fang, H. W.; Chen, C. C.; Lin, F. H. A poly(propylene fumarate)–calcium phosphate based I

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering manufacturing of porous metals for bone scaffolds and orthopaedic implants: A review. Biomaterials 2016, 83, 127−141. (30) Murrell, D. H.; Hamilton, A. M.; Mallett, C. L.; van Gorkum, R.; Chambers, A. F.; Foster, P. J. Understanding heterogeneity and permeability of brain metastases in murine models of her2-positive breast cancer through magnetic resonance imaging: Implications for detection and therapy. Transl Oncol 2015, 8, 176−184. (31) Ludwig, J. A.; Lamhamedi-Cherradi, S. E.; Lee, H. Y.; Naing, A.; Benjamin, R. Dual targeting of the insulin-like growth factor and collateral pathways in cancer: Combating drug resistance. Cancers 2011, 3, 3029−3054. (32) Kasper, F. K.; Tanahashi, K.; Fisher, J. P.; Mikos, A. G. Synthesis of poly(propylene fumarate). Nat. Protoc. 2009, 4, 518−525. (33) Wang, S.; Lu, L.; Yaszemski, M. J. Bone-tissue-engineering material poly(propylene fumarate): Correlation between molecular weight, chain dimensions, and physical properties. Biomacromolecules 2006, 7, 1976−1982. (34) Levorson, E. J.; Santoro, M.; Kasper, F. K.; Mikos, A. G. Direct and indirect co-culture of chondrocytes and mesenchymal stem cells for the generation of polymer/extracellular matrix hybrid constructs. Acta Biomater. 2014, 10, 1824−1835. (35) Cioffi, M.; Boschetti, F.; Raimondi, M. T.; Dubini, G. Modeling evaluation of the fluid-dynamic microenvironment in tissue-engineered constructs: A micro-ct based model. Biotechnol. Bioeng. 2006, 93, 500− 510. (36) Maes, F.; Van Ransbeeck, P.; Van Oosterwyck, H.; Verdonck, P. Modeling fluid flow through irregular scaffolds for perfusion bioreactors. Biotechnol. Bioeng. 2009, 103, 621−630. (37) Bird, R.; Stewart, W.; Lightfoot, E. Transport Phenomena, 2nd ed.; John Wiley: New York, 2002. (38) Holman, J. Heat Transfer, 10th ed.; McGraw-Hill Series in Mechanical Engineering; McGraw-Hill: New York, 2010. (39) Porter, B.; Zauel, R.; Stockman, H.; Guldberg, R.; Fyhrie, D. 3-d computational modeling of media flow through scaffolds in a perfusion bioreactor. J. Biomech. 2005, 38, 543−549. (40) Voronov, R.; Vangordon, S.; Sikavitsas, V. I.; Papavassiliou, D. V. Computational modeling of flow-induced shear stresses within 3d salt-leached porous scaffolds imaged via micro-ct. J. Biomech. 2010, 43, 1279−1286. (41) Dahlin, R. L.; Meretoja, V. V.; Ni, M.; Kasper, F. K.; Mikos, A. G. Design of a high-throughput flow perfusion bioreactor system for tissue engineering. Tissue Eng., Part C 2012, 18, 817−820. (42) Lam, J.; Kim, K.; Lu, S.; Tabata, Y.; Scott, D. W.; Mikos, A. G.; Kurtis Kasper, F. A factorial analysis of the combined effects of hydrogel fabrication parameters on the in vitro swelling and degradation of oligo(poly(ethylene glycol) fumarate) hydrogels. J. Biomed. Mater. Res., Part A 2014, 102, 3477−3487. (43) Henslee, A. M.; Gwak, D. H.; Mikos, A. G.; Kasper, F. K. Development of a biodegradable bone cement for craniofacial applications. J. Biomed. Mater. Res., Part A 2012, 100A, 2252−2259. (44) Gomes, M. E.; Holtorf, H. L.; Reis, R. L.; Mikos, A. G. Influence of the porosity of starch-based fiber mesh scaffolds on the proliferation and osteogenic differentiation of bone marrow stromal cells cultured in a flow perfusion bioreactor. Tissue Eng. 2006, 12, 801−809. (45) Van Bael, S.; Chai, Y. C.; Truscello, S.; Moesen, M.; Kerckhofs, G.; Van Oosterwyck, H.; Kruth, J. P.; Schrooten, J. The effect of pore geometry on the in vitro biological behavior of human periosteumderived cells seeded on selective laser-melted ti6al4v bone scaffolds. Acta Biomater. 2012, 8, 2824−2834. (46) Trachtenberg, J. E.; Mountziaris, P. M.; Kasper, F. K.; Mikos, A. G. Fiber-based composite tissue engineering scaffolds for drug delivery. Isr. J. Chem. 2013, 53, 646−654. (47) Kashima, H.; Shiozawa, T.; Miyamoto, T.; Suzuki, A.; Uchikawa, J.; Kurai, M.; Konishi, I. Autocrine stimulation of igf1 in estrogeninduced growth of endometrial carcinoma cells: Involvement of the mitogen-activated protein kinase pathway followed by up-regulation of cyclin d1 and cyclin e. Endocr.-Relat. Cancer 2009, 16, 113−122.

(48) Fukumura, D.; Jain, R. K. Tumor microenvironment abnormalities: Causes, consequences, and strategies to normalize. J. Cell. Biochem. 2007, 101, 937−949. (49) Stolberg, S.; McCloskey, K. E. Can shear stress direct stem cell fate? Biotechnol. Prog. 2009, 25, 10−19. (50) Alvarez-Barreto, J. F.; Linehan, S. M.; Shambaugh, R. L.; Sikavitsas, V. I. Flow perfusion improves seeding of tissue engineering scaffolds with different architectures. Ann. Biomed. Eng. 2007, 35, 429− 442. (51) McCoy, R. J.; Jungreuthmayer, C.; O’Brien, F. J. Influence of flow rate and scaffold pore size on cell behavior during mechanical stimulation in a flow perfusion bioreactor. Biotechnol. Bioeng. 2012, 109, 1583−1594.

J

DOI: 10.1021/acsbiomaterials.6b00641 ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX