Surface patterning and preparation
The nanoscale patterned surface was first prepared by spin-coating a 2:3 mixture of PMMA:chlorobenzene thinner (Microchem, Newton, MA, USA) onto an acetone-cleaned 1 cm × 1 cm p-doped silicon wafer (boron-doped 111 Si, 450-648-μm thick and 4–75 Ω·cm-1, Exsil, Inc., Prescott, AZ, USA). Spin-coating was completed in a two-step process at 400 rpm for 8 s, and 4000 rpm for 25 s, for a measured thickness of 80 nm using a KLA-Tencor Alpha-Step 200 profilometer (Milpitas, CA, USA) and Veeco Dimension 3100 atomic force microscope (AFM; Bruker AXS, Santa Barbara, CA, USA). A series of ten 400 μm × 400 μm array patterns of 100-nm diameter wells spaced at 5 μm × 1 μm were then patterned into this PMMA surface via electron beam lithography, using an FEI Inspec scanning electron microscope (SEM; Hillsboro, OR, USA) equipped with nanometer pattern generating system (NPGS; JC Nabity, Bozeman, MT, USA). This pattern was characterized and validated through SEM and AFM imaging.
Just prior to seeding cells onto the array, the pattern was developed in a three-step process of immersion in 1:3 mixture of methyl isobutyl ketone (MIBK; Microchem Corp., Newton, MA, USA) and isopropyl alcohol (IPA; Honeywell, Morristown NJ, USA) for 1 min, IPA for 30 s, and a deionized water rinse before drying the chips under nitrogen (N2) gas. Onto this developed pattern, 100-nm polystyrene nanoparticles (carboxylated and fluorescent) conjugated with RGD were distributed onto the wells.
Conjugation of RGD to the nanoparticles was carried out according to a revised general protein coupling workflow protocol (Bio-Rad, Hercules, CA, USA). In this, 1 ml of 0.02% (w/v) 0.1 μm FluoSpheres fluorescent polystyrene nanoparticles (Invitrogen, Eugene, OR, USA) were conjugated with 42.7 μl of a 1 mg·ml-1 solution of the GRGDSPK (RGD) peptide sequence (Anaspec, Inc. Fremont, CA, USA) according to the protocol delineated by Tran et al. .
RGD-conjugated nanoparticles were distributed across the array through micropipetting, with three passes of a 1 μl droplet of the RGD-nanoparticle suspension over each developed nanowell array pattern. This was followed by a double rinse of PBS. Entrapment of the negatively charged RGD-conjugated nanoparticles through electrostatic attraction to the positively charged wells was confirmed using an FEI Inspec SEM.
Preliminary cell culturing
Human umbilical vascular endothelial cells (BD Biosciences, San Jose, CA, USA) were grown in M199 medium (Lonza Biowhittaker, Walkersville, MD, USA), supplemented with 15% (v/v) fetal calf serum (Lonza Biowhittaker); 1% (v/v) each of 200 mM L-glutamine (Lonza Biowhittaker), 1 M HEPES (Lonza Biowhittaker), and penicillin:streptomycin (10,000 U∙ml-1 : 10,000 μg∙ml-1, Lonza Biowhittaker); 25 mg of endothelial cell growth supplement (ECGS; Biomedical Technologies, Inc., Stoughton, MA, USA) and 26.4 mg heparin sodium, for 500 ml total mixed medium. HUVECs were cultured within a bioreactor set at 37°C, 5% CO2 until they reached approximately 80% confluency within a gelatin-coated culture flask (dextrose gelatin veronal; Biowhittaker, Lonza, Basel, Switzerland) culture flask, and were used between passages 2–6.
Cell culturing under flow conditions
At 80% confluency, cells were passaged using 2 ml of a 1:1 mixture of trypsin and Hank’s Balances Salt Solution (HBSS, without Ca2+ or Mg2+; Lonza Biowhittaker). Cells were resuspended at 350,000 cells∙ml-1, of which 0.2 ml was seeded onto each patterned chip surface for 10 min. These chips were then fully immersed in the M199 media and left to incubate statically for 24 h in the bioreactor. Following this incubation period, the chips were placed into the designated flow channel which was fed from a system of two 50 ml reservoirs, the first with filtered gas exchange, filled with 20 ml and 30 ml of the mixed M199 media. Media was drawn through the channel and silicone connective tubing (Masterflex® platinum-cured silicone tubing, L/S®; Cole-Parmer, Vernon Hills, IL, USA) by peristaltic pumping action (Cole-Palmer). Cells were then incubated under flow for 24 h. For the 48 h static controls, the static seeded chips were indirectly rinsed once with PBS, and their media was replenished.
Channel design and flow conditions
Flow was applied to the seeded chip surfaces within a two-piece acrylic thin-channel device of dimensions 16 cm × 4 cm × 0.1 cm (Figure 2). Square grooves of dimensions 1 cm × 1 cm × 0.5 cm were machined into the lower interior face, allowing the chips to be positioned in such a way that the seeded surface would be flush with the wall.
The pump was set to produce average flow rates ranging between 89–239 ml∙min-1 for 24 h within the bioreactor. The direction of flow was marked by a small scratch along each chip, away from the pattern, which was later used for reference in analysis.Wall shear stress values for all flow conditions were estimated based on the relationship τ = (6μQ) / (H2W) with μ referring to the dynamic viscosity of the M199 mixed media, μ = 0.782 mPa∙s-1, Q referring to the volumetric rate of media flow through the channel, and H and W being the height and width dimensions of the flow channel’s cross section.
The pulsatile action of the pump (without a damper) caused the flow rate and subsequently the wall shear stress to suffer a time dependency. This time dependency was determined through measuring pressure changes in the channel. These pressure changes were correlated to velocity change through Bernoulli’s principle, then to the flow rate variance and finally to the wall shear stress using the above equation.
Cell adhesion and orientation
At 24 h, the chips were removed from the channel and cells were stained for immunofluorescence using an actin cytoskeleton and focal adhesion staining kit (Millipore, Billerica, MA, USA) according to the standard Millipore protocol. In this, HUVECs were fixed using 4% para-formaldehyde for 15 minutes, and then washed in 0.05% Tween 20 and perforated by applying a 0.05% Triton X solution for 5 minutes. Following a second washing step, non-specific antigen binding was blocked using a 0.5% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO, USA) solution for 15 minutes. Vinculin was then fluorescently labeled using fluorescein isothiocyanate conjugatedmouse anti-immunoglobulin G (mIgG-FITC) for 1 hour, followed by a washing step. F-actin was stained next using tetramethyl rhodamine isothicyanate (TRITC) conjugated Phalloidin. After 2 final washing stages, cells were mounted for imaging in a vector shield containing DAPI (Vector Laboratories, Inc., Burlingame, CA, USA). Cells were then imaged at 10x using epifluorescence and confocal microscopy.
Using ImageJ software (National Institutes of Health, USA), the angles of orientation were measured relative to the flow direction according the system detailed in Figure 1d and the cell numbers were counted per unit patterning area: 400 μm × 400 μm. Over patterned surfaces, this unit area was defined by the evident boundary between the fluorescent green nanoparticle array and the surrounding unpatterned PMMA. Over the +Glass, +Si, and PMMA surfaces, this unit area was defined by a 400 μm × 400 μm grid overlay applied to each image. In each image, only full arrays were considered, and only cells with a visible nucleus and with no visible sign of necrosis or apoptosis that were fixed over the array after 24 h were counted. For analysis of orientation, angle was measured for the long axis of the cell body relative to the flow direction, as determined to be lengthwise along the widest segment of the cell body containing the nucleus, discounting apparent filopodia.
Percent coverage of pattern
Over these same defined array and grid areas, gray scale and red scale histograms were collected and the average pixel intensity was taken for each measurement within the ImageJ software (National Institutes of Health, USA). Similarly, gray scale and red scale histograms were taken for null regions and regions of full cell coverage of each image, and the average pixel intensities were also measured. Data for the array and grid areas were averaged within each image, and were then compared relative to the controls for their respective image. Percent coverage of cells was evaluated for a given array (or grid).