A Novel Cylindrical Biaxial Computer-Controlled Bioreactor and Biomechanical Testing Device for Vascular Tissue Engineering

Culture chamber

The base of the bioreactor is machined from a solid aluminum plate. Two uniaxial spring-actuated stages (Newport model M-423) with appropriate fittings for mounting linear actuators (Newport model LTA-HS, Irvine, CA) are mounted to the base. Atop both uniaxial stages are three axes manually adjustable stages (Newport model M461) with three manual micrometers to one of allow for alignment of cannula and TEBV. Connected to these stages is a stainless steel rod and custom-machined part to allowing mounting of L-shaped cannula and a load cell (Delta Metrics Inc. model XLU68s, Worthington, OH) to measure axial force. The TEBVs are mounted on the cannula that allow for luminal perfusion. Because this system is capable of being used on a range of vessel sizes, the cannula can range in outer diameter from 1 to 5 mm. The ends of the cannula may be tapered to allow easier mounting of the vessel and given grooves to prevent sliding of the o-rings or sutures that hold the TEBV in place.

A custom polycarbonate top with gas exchange filters is used to prevent contamination during transport and culture. Rubber baffles allow for enclosure of the top around the stainless steel rod while not restricting linear motion. The bath in which the vessel is cultured is made from polycarbonate with a clear cover slip to allow for imaging on an inverted confocal microscope (Zeiss LSM 510, cameleon argon laser, Jena, Germany). A second flow loop allows for circulation and medium exchange in the polycarbonate bath.

Flow control

A peristaltic pump (Ismatec model IPC, Glattbrugg, Switzerland) is used to generate luminal flow with given rate and pattern. The pump communicates with the computer via a data acquisition board allowing a prescribed flow waveform to be generated. The reservoir is a five-port 250-mL Erlenmeyer flask with solid stopper. Medium is pumped out of this flask from a bottom port and travels through the pump before entering the first pressure transducer. Flow continues into the culture chamber and through the vessel before exiting the culture chamber and returning to the second pressure transducer. Once leaving the second transducer, flow reenters the reservoir. The flow loop contains one bypass before the vessel to allow for total removal of air bubbles from the tubing. The bypass is closed during culture to ensure all flow is through the vessel. Measurements of flow rates are obtained using an ultrasonic flow-monitoring system (Transonic model T402, Ithaca, NY) with appropriate probe size. The probe is located on the vessel to directly measure the flow rate through the vessel, but is large enough as to not impose any external traction forces on the vessel. Data were collected using an analog output from the flow-monitoring unit and time-averaged over each second.

Pressure controls

Pressure in the vessel is controlled by regulating static gas pressure in the reservoir. Static pressure in the reservoir is controlled using a custom-designed syringe pump. A syringe is mounted onto a uniaxial stage whose position is controlled via a linear actuator. The syringe plunger is fixed allowing movement of the stage to move the syringe tube relative to the plunger. Tubing and a filter connect the syringe to the reservoir. As the plunger moves into the tube, pressure is increased in the reservoir. Likewise, as the plunger moves out of the tube, pressure is decreased. Because we control static pressure in the reservoir, pressure at the reservoir inlet and outlet are equal, thereby not inducing any pressure gradient nor flow. Pressure in the vessel is monitored using two pressure transducers (Honeywell model FGP, Columbus, OH) located on the bottom rack of the incubator properly calibrated to account for static pressure head. A feedback control was developed in LabView (National Instruments v7.1, Austin, TX) to deliver periodic pressure curves with adjustable set-points for minimum and maximum pressure, and waveform period. Further development with a more sophisticated control algorithm could use a Fourier series approach to deliver a physiological pressure waveform at physiological rates. Due to the risk of vascular leakage or air leakage in the syringe, a three-way solenoid pinch valve is used to reload the piston once it reaches the end of its range of travel. Reloading of the syringe takes approximately 10 s. Analog inputs from the pressure transducers and load cell, as well as analog outputs to the pump and digital outputs to the pinch valves, are communicated with the computer via a data acquisition board (National Instruments model USB-6259).

Axial motion control

Uniaxial sliding stages on the culture device fit linear actuators that are controlled using a Newport motion controller (model XPS). This motion controller is capable of controlling eight linear actuators simultaneously. Two actuators are used per device to prevent translation of the vessel in the camera focus area. Therefore, one motion controller is capable of controlling all four parallel bioreactors. This motion controller is capable of monitoring actuator movement with 0.035 μm resolution. LabView communicates with the motion controller via an Ethernet cable to precisely control movements of the actuators to prescribed spatial locations; hence, axial extensions are referenced to the unloaded length of the vessel. Additionally, using a feedback loop in LabView, oscillatory axial distensions are capable at frequencies of ∼1 Hz.

Vascular geometry monitoring

A digital camera (AVT Guppy IEEE 1394, Newburyport, MA) is mounted on an imaging lens (VZM 200i; Edmund Optics, Barrington, NJ) above the vessel and combined with a fiber optic light source (Dolan-Jenner Fiber-Lite, Boxborough, MA) for illumination. Real-time processing of the digital image is accomplished using LabView vision software to measure outer diameter of the vessel. In addition, markers may be attached to the outer surface of the vessel and tracked in real time to quantify the displacements in a local region of the vessel (Fig. 3a), thereby eliminating end effects on the quantification of strain. Wall thickness is measured using a pulse-echo ultrasonic technique. An ultrasonic pulser-receiver (Olympus NDT model 5900PR Waltham, MA) is used in combination with a 25-MHz, 0.25′′-diameter, flat focused transducer (Olympus NDT model V-324SM). Measurements were made in a fluid medium bath with a distance of approximately 0.5′′ between the transducer and the front wall of the vessel. As the ultrasonic traverses the vessel, at each vessel–medium interface, there is a change in acoustic impedance causing a reflection that is captured by the transducer, and peaks are observable on the ultrasonic signal from all four of these interfaces. The signal is acquired via a 250-megasample/s, 8-bit digitizer (National Instruments model PCI-5114). Approximately 20 waveforms are averaged and manipulated using a hilbert transform in combination with convolution to accentuate the peaks over the noise. A cross-correlation algorithm is then used to determine the location of the four peaks corresponding to the dimensions of the vessel. The specifications of the digitizer allow a theoretical resolution of 3 μm.

Preparation, sterilization, and sterile tissue cannulation

Before sterilization, all components are washed in a soap bath and thoroughly rinsed with water. Pressure transducers are rinsed three times with 70% ethanol and allowed to evaporate overnight. Ethylene oxide is used to sterilize the fully assembled device (with chamber cover removed), pressure transducers, and load cell. Sterilization of additional tubing segments and the reservoir is accomplished in an autoclave. The flow loop is created consisting of the reservoir, pressure transducers, pump, and appropriate tubing. Cannulation of the vessel occurs in a biological safety cabinet. Once securely mounted using sterile sutures or o-rings, the chamber cover is attached. The flow loop is connected to the culture chamber, and lastly the linear actuators are secured to the uniaxial stages. Typically, cannulating the vessel and setting up the bioreactor is accomplished in under 1 h.

Culture capabilities of this device are demonstrated using biological vascular substitutes (BVS) created using the self-assembly method as previously described.³³ Vessels were cultured and tested in Dulbecco's modified Eagle's medium (with 4500 mg/L of D-glucose and L-glutamine; Mediatec, Inc., Manassas, VA) supplemented with 10% fetal bovine serum (Mediatec, Inc.), antibiotic (P/S; 50 IU/mL penicillin and 50 mg/mL streptomycin; Mediatec, Inc.), and additional L-glutamine (5000 g/L; Mediatec, Inc.).

Biomechanical testing

The custom LabView program allows us to prescribe a variety of loading patterns for mechanical testing (e.g., pressure–diameter, force–length, and stress–relaxation tests). To demonstrate the device capabilities, pressure–diameter and force–length tests were performed. Biomechanical testing may be performed in the bioreactor immediately after culture without any required handling of the TEBV. Before testing, the unloaded length of the vessel was measured. Under quasistatic conditions, pressure was varied from 0 to 120 mmHg at fixed axial stretches of λ_z = 1.0, 1.15, and 1.25, where λ_z = l/L, l is the current length and L is the unloaded length, at the time of interest. From these data, pressure–diameter (P − d) and axial force–pressure (f − P) curves were generated. Additionally, λ_z was cyclically varied up to 1.25 at pressures of 20, 40, and 120 mmHg. Axial force–length (f − L) curves were generated from these data.

A second set of mechanical testing data are shown using collagen-gel-based TEBVs. The collagen-gel-derived TEBVs were developed following the methods of Stegemann and Nerem.³⁴ Briefly, SMCs were isolated from rat aortas and combined with reconstituted bovine type I collagen. Controlling the environmental conditions of the collagen, gelation was induced around glass mandrels to form the TEBVs. After 2-day static culture, the TEBVs were mounted on this device and cyclically inflated at λ_z = 1.00 from pressures of 0–10 mmHg for five cycles, then 0–20 mmHg for five cycle, then 0–30 mmHg for five cycles, and so forth until these freshly gelled vessels ruptured; pressure, diameter, and axial force were monitored at a fixed axial length.

Confocal imaging

Decellularized BVS were used to image collagen content and organization via multi-photon microscopy. A Zeiss LSM 510 META inverted confocal microscope with Chameleon Ultra Ti:sapphire femtosecond laser (Coherent) was tuned to 800 nm and used to excite and detect collagen. The laser reached the sample through either a 40 ×/1.3NA or 63 ×/1.4NA oil immersion objectives (Zeiss). The fields of view were 225 × 225 μm and 143 × 143 μm, respectively. The META module of the microscope was configured as a 380–420 bandpass filter to detect backward scattering second harmonic signal from collagen.³⁵ Imaris software (Bitplane) was used for 3D reconstruction of images.

Vascular geometry monitoring

Camera calibration was performed using a glass slide with etched lines spaced at 500 μm. The imaging lens was focused to achieve 20 μm per pixel, representing a field view of 8500 × 11,400 μm on the digital camera. A representative image of a BVS and the outer diameter are shown for a vessel cultured at an axial extension λ_z* = l/L(0), where L(0) is the unloaded axial length at the beginning of culture (t = 0), with cyclic pressure application (32.5 ± 7.5 mmHg, rate of one cycle per minute), and incremental flow (5 min at 3 mL/min once a day) for nutrient and metabolite exchange (Fig. 3). Notice that we have placed markers on the vessel to allow for measurements of local displacement, in parallel to measurements of diameter and overall vessel length. Data were collected once per minute for 96 h. As seen, there is little variation in the diameter measurements (less than 100 μm) between consecutive data points. The discontinuity occurring around 50 h is due to removal of the culture chamber from the incubator and medium exchange in the reservoir.

Wall thickness measurements are accomplished using ultrasonic techniques. One sample digitized waveform is shown (Fig. 4a). To determine the peaks from this waveform (which represent the interfaces between the vessel and the medium), the waveform is subjected to an algorithm that includes crosscorrelation and a Hilbert transform. After the raw data are processed (Fig. 4b), the peaks can be easily determined using a simple peak search algorithm. Sequential waveforms can be processed in the same manner allowing thickness measurements to be determined throughout the course of the test or culture. The thickness measurements for one pressurization cycle (0–150–0 mmHg) can be seen in Figure 4c.

Loading control

To demonstrate pressure–flow independence, two experiments were performed. The first was a low-pressure/high-flow test where pressure was held constant at 20 mmHg while maintaining a constant flow rate of 350 mL/min. The second was a high-pressure and low-flow demonstrated by holding pressure constant at 100 mmHg and flow at 50 mL/min. There is no variation in both pressure (range 3 mmHg) and flow (range 10 mL/min) during low-pressure/high-flow testing (Fig. 5). Likewise, there is little deviance in both pressure and flow during high-pressure/low-flow conditions.

To illustrate the independence between the four bioreactors, two bioreactors were set with mean pressures of 50 and 100 mmHg and ranges 20 and 40 mmHg, respectively, while two bioreactors were set to hold constant pressures of 20 and 40 mmHg. All four reactors are under constant axial stretch and no flow. Due to the independent design of the pressure control system, all four bioreactors are capable of operating in parallel without adverse effects (Fig. 6).

Biomechanical testing

Loading and unloading curves from statically cultured BVS demonstrate the cylindrical biaxial biomechanical testing capabilities of our device. These results, generated from a single BVS, provide representative data regarding the capabilities of our device. As previously described, a standard testing protocol consists of three pressure cycles from 0–120–0 mmHg at a constant axial stretch. During this cycle, we monitor outer diameter as well as axial force (Fig. 7a and b, respectively). Axial force–length curves, generated at three constant pressures, exemplify a typical viscoelastic response of biological materials as seen by the hysterisis occurring between the loading and unloading phases (Fig. 7c). All samples P − d tests were pressurized and depressurized at a rate of 2 mmHg/s. These data combined with ultrasound thickness data can be used to generate mean axial and circumferential stress–strain curves, as well as perform stress analysis; see Equation (1).

Biomechanical tests were also performed on 2-day statically cultured collagen gels. These data were generated from a single collagen construct and provide representative capabilities of the device. At just 2 days after gelling, these vessels are very fragile. Given the fragility of these vessels, it was clear that they would not exhibit high burst strength at these early culture times. Thus, we conducted a testing protocol that gradually increases the cyclic testing pressure from 0–10 mmHg to 0–20 mmHg, and so forth (see Fig. 8a). We were able to successfully mount and mechanically load these vessels and conduct pressure–diameter and axial force–length tests (Fig. 8a and b, respectively). This protocol allows for the collection of several loading and unloading curves for stress analyses before rupture.

Confocal imaging

In multiple loaded configurations, collagen fibers of a BVS were imaged using confocal microscopy. With a single BVS mounted in the bioreactor, two successive images were taken at the same location within the vessel in two different loading scenarios: unloaded (Fig. 9a) and loaded to 20 mmHg (Fig. 9b). Densely arranged large and small bundles of collagen fibers can be seen. Quantification of these fibers at the various loading conditions will provide information relating fiber kinematics to the kinematics of the whole tissue, as well as information on the evolution of fiber content, orientation, and kinematics with tissue growth and remodeling to more accurately generate microstructurally motivated biomechanical and growth and remodeling models.

While Figure 9 simply demonstrates the ability of our system to image a vessel under various loading conditions, a representative sample of the microstructure would have to be determined using multiple images of the same vessel at random locations. When the imaged length (255 μm) is compared with the total unloaded length (30–40 mm), there is sufficient vascular length to attain multiple samples from different locations on the same vessel. Multiple algorithms have been developed to detect fiber angle from images for use across a wide variety of mediums.^36,37 Additionally, these algorithms have been used to quantify structural information from biologics.^32,38–40 These algorithms will be sufficient to extract the quantitative structural information required for microstructurally motivated biomechanical models.