[001] Introduction

[002] In load bearing tissues, cellular response is balanced between healthy physiological mechanical strain and over straining that leads to damage and degeneration. For the intervertebral disc and other annular or vessel-like structures in the body (e.g., cardiovascular system, lymphatic system, and cervix) these strains are three dimensional and heterogeneous due to spatial variations in tissue composition and fiber alignment. As a result, these strains, and the resulting cell response, can only be partially replicated with uni- or bi-axial strains applied through traditional cell stretching systems. Therefore, the true mechanical environment in vessel-like structures has not been replicated in vitro, despite a substantial need to better understand the degenerative cascade in tissues like the intervertebral disc, which is a leading cause for physical disability. ¹ ’ ² ’ ³ ’ ⁴

[003] Mechanical overloading can lead to greater rates of degeneration or herniation as well as additional overloading, thereby initiating a self-perpetuating cycle or degenerative cascade. ⁹ ’ ¹⁰ ’ ¹¹ Early stages of disc degeneration are marked, in part, by tears in the structurally critical annulus fibrosus (AF), ⁵ ’ ⁶ where full thickness tears result in bulging or herniation, which has been directly linked to lower back pain. More advanced stages of degeneration include significant tissue remodelling, with a decrease in disc height and water content, and increased tissue fibrosis, resulting in both an overall increase and shift in intradiscal loading from the nucleus pulposus to the AF. ’ Current treatment strategies focus on treating symptoms of end-stage degeneration through invasive means. To date, biological treatments to delay or prevent progression of advanced degeneration have been limited.

[004] Development of biological treatment strategies have been challenging due to translatability from small animal models to human cells and tissues. To this end, there has been an increase in organ-on-a-chip platforms that simulate in situ loading conditions and study the cellular response to biological and mechanical stimuli. However, to date, there are no organ-on-a-chip models that replicate the hoop strains experienced by cells within a pressurized vessel structure like the AF. ¹³ For example, under physiological loading, the intervertebral disc experiences axial compression in combination with bending or rotation. This organ level loading is transferred as intradiscal pressure from the nucleus pulposus to the outer AF tissue as tensile hoop stresses. ¹⁴ ’ ¹⁵

[005] Meanwhile strains from the AF to the cell are transferred yet attenuated by upwards of 70%. ¹⁶ In the AF, mechanical overloading or changes in internal pressure within the nucleus pulposus may result in AF tissue remodelling. ¹⁷ Moreover, AF fissures may result in concentrated regions of higher stress and strain. ¹⁸ While cells make up less than 1% of the disc volume, DNA content is greatest in the outer AF, where nerves and blood vessels have been shown to grow within fissures located in the outer third of the AF. ^19,20,21,22 Most loading systems developed for cell-laden constructs represent in situ loading that primarily experience uniaxial loading, such as tendons and ligaments, rather than the complex loading conditions representative of the AF. [006] Organ-chips with more complex mechanical activity, such as hoop tensile strains, are needed to study mechanobiology. [007] Relevant Literature [008] US4,839,280; US5,122,470; US6,048,723: Banes AJ. [009] WO2013086502: Ingber, DE. Organ chips and uses thereof. [010] Huh D, Matthews BD, Mammoto A, Montoya-Zavala M, Hsin HY, Ingber DE. Reconstituting Organ-Level Lung Functions on a Chip. Science (80- ).2010 Jun 25;328(5986):1662. [011] Vazquez, M., Evans, B. A., Riccardi, D., Evans, S. L., Ralphs, J. R., Dillingham, C. M., & Mason, D. J. (2014). A New Method to Investigate How Mechanical Loading of Osteocytes Controls Osteoblasts. Frontiers in Endocrinology, 5. [012] Electroforce Mechanical Test Instruments, https://www.tainstruments.com/wp- content/uploads/BROCH-EF-LF-EN.pdf [013] Summary of the Invention [014] The invention provides for a mechanically active organ-on-a-chip with geometrically defined actuation, including related methods and compositions. [015] In an aspect, the invention provides a mechanically active microfluidic system configured to mimic a mechanical stimulus, the system comprising: (a) a flexible, planar, optically transparent microfluidic device comprising an embedded microfluidic channel comprising live cells; and (b) a rigid dieform structure, wherein the microfluidic device is deformed against the dieform structure, imposing on the channel a geometrically governed mechanical stimulus. [016] In embodiments: [017] the microfluidic device comprises a bottom layer with top surface shaped, casted or molded to form the bottom and sides of the channel, and a top layer closing and forming the top side of the channel. [018] the device comprises a surface 10- 100mm or 20-50mm across, such as each side of a rectangle;

[019] the device is 2-5 mm or 3-5mm in thickness;

[020] the channel is 5-80 or 10-50 or 10-20 mm long;

[021] the channel is 20-200 um or 50-150 um wide/high;

[022] the device is elongate parallel with the channel

[023] the channel is made using soft lithography;

[024] the device is in a miniature format, such as wherein the device comprises a surface 1-10 or 3-5 mm across, the channel is 0.5-8 or 2-4 mm in length, and 1-10 or about 5 um wide/high. [025] the device is in a multiplex format comprising multiple embedded microfluidic channels comprising live cells, such as a 96 channel format configured to the dimensions of a microtiter plate, such as about 127.8 mm x 85.5 mm;

[026] the top layer is 0.5-1.5mm thick, the bottom later 2.5-3.5mm thick, the surface is about 25 x 45 mm and the channel is about 150um wide x 50um high, and about 10-20 mm long;

[027] the mechanical stimulus is selected from: (a) hoop strain (circumferential strain), (a) hoop strain in combination with axial strain and radial strain which can be called multiaxial, (c) cyclic or static versions of hoop strain and/or axial/radial strain, and (d) tension or compression for components of hoop strain and/or axial/radial strain;

[028] the system is configured to apply to the channel a tensile strain which can be first order approximated: E _x = Y/(T/2 + R), wherein E _x is tensile strain at the channel, T is thickness of the device, R is radius of curvature of the dieform, and Y is distance between channel and device midplane thickness;

[029] the mechanical stimulus is a tensile hoop-strain, and/or radial compressive strain;

[030] the system is configured to mimic strains of a hard or soft tissue selected from bone, cartilage, brain, intervertebral disc, cervix, bladder, and cardiovascular walls;

[031] the system is configured to mimic tensile hoop-strains in annulus fibrosus;

[032] the system is configured to apply to the channel a strain simulating physiological strain on healthy and degenerative annulus fibrosus;

[033] the cells are configured in a 2D or 3D cell culture;

[034] the device is made of a polysiloxane, such as poly dimethylsiloxane PDMS, poly(methyl methacrylate) (PMMA), polystyrene and poly(lactic-co-glycolic acid) (PLGA), or a thiol-ene polymer such as PDMS chains with vinyl- siloxane or thiol end groups, tri-allyl-tri-azine: tri- thioltriacine and tri-allyl-tri-azine: tetra-thiolpentaerythritol; and/or

[035] the device is made of polydimethylsiloxane PDMS [036] In an aspect the invention provides use of a mechanically active microfluidic system disclosed herein configured to mimic a strain comprising: (a) a flexible, planar, optically transparent microfluidic device comprising an embedded microfluidic channel comprising live cells; and (b) rigid dieform structure, wherein the microfluidic device is deformed over the dieform structure, imposing on the channel a geometrically governed mechanical stimulus such as strain; the method comprising: deforming the microfluidic device with the dieform structure to impose on the channel the strain.

[037] In embodiments the method further comprises determining a resultant change in gene expression of the cells.

[038] The invention encompasses all combinations of the particular embodiments recited herein, as if each combination had been laboriously recited.

[039] Brief Description of the Drawings

[040] Fig. 1. A) Fabrication process of the spine-on- a-chip using standard lithography. Fig. IB) Schematic of experimental design to apply tensile hoop strains to cells in the channel through bending. The channel is viewed from below with an inverted microscope. Fig. 1C) Flexion of the spine results in complex strains, where the anterior AF experiences compressive axial strains and the posterior AF experiences tensile axial strains. The location of the channel within the thickness of the chip can control whether uniaxial (hoop) strains are applied to the cell monolayer. Moving the channel below or above the neutral axis can be used to apply biaxial (hoop and axial) strains to cells in the channel.

[041] Fig. 2A) Physical SpOC with bending applied manually. Fig. 2B) Experimental SpOC set up. The chip is loaded between a 3D printed barrel structure and a 1 mm slide on top of the microscope tray. At all times micropipette tips with cell culture media remain plugged into ports at either end of the channel. Polypropylene ’packing’ tape cradles the chip while being connected to servo motor horns with pin joints. As the servo horns rotate in tandem upward the tape pulls tight which forces the chip to conform to the 3D printed barrel structure. Scale bar: 4 mm.

[042] Fig. 3 A). Finite element analysis of SpOC: color map of hoop strains across the SpOC channel in its largest bent state. Fig. 3B) Based on the model, the centroidal axis (inset) was offset from the neutral axis by 0.2 mm. Hoop strains were calculated with respect to the centroidal axis. Fig. 3C) Components of three-dimensional strains, hoop, circumferential and axial, are modeled at the midpoint of the channel as the chip is bent starting initially in a flat configuration. Fig. 3D) Uniformity of strain along the channel is demonstrated with the model- predicted components of three-dimensional strain at locations along the channel. Data is presented only for half of the channel due to symmetry.

[043] Figs. 4A-4B. Cells were imaged with a 10X objective lens on an inverted brightfield and fluorescent microscope before and after loading respectively. A live/dead assay indicated that cells remained viable hours after loading.

[044] Fig. 5. Beam theory, or geometric model, used to predict applied strains along the x-axis.

[045] Description of Particular Embodiments of the Invention

[046] Unless contraindicated or noted otherwise, in these descriptions and throughout this specification, the terms “a” and “an” mean one or more, the term “or” means and/or and polypeptide sequences are understood to encompass opposite strands as well as alternative backbones described herein. It is understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview of this application and scope of the appended claims. All publications, patents, and patent applications cited herein, including citations therein, are hereby incorporated by reference in their entirety for all purposes.

[047] The invention provides a microfluidic device for sustaining cell and tissue life while providing physiologically relevant environments for recreating organs-on-a-chip. Advantages of our device include (1) High accuracy and uniformity of strain without auxiliary equipment. (2) Simplicity and low-cost (3) Adaptability to many microfluidic chip designs (4) Ability to apply tension or compression. In particular embodiments the devices are used as or in organs-on-a- chip bioreactors.

[048] A dieform acts as a rigid structure over which a microfluidic chip is deformed for the purposes of geometrically governed mechanical stimulus such as strain. The geometry of the design allows for high accuracy and uniformity of strain without auxiliary equipment that is responsible for measuring and controlling as seen in other bioreactors. A reversible structure means compression and or tension can be applied. Holes in the dieform allow for microscopy. The devices may be designed such that standard microfluidic manufacturing processes are utilized and that any selection of materials could be used. The device can accommodate microfluidic chips for all tissues allowing for organ-on-a-chip advancement. The devices can be actuated by hand or with any form of mechanical actuation. The devices can be miniaturized for integration on standard 96-well plate configurations.

[049] The devices provide simple, low cost, and confident mechanical stimulus for organ-on-a- chip devices in research and industry including hard and soft tissues such as bone, cartilage, brain, intervertebral disc, cervix, bladder, cardiovascular walls etc. These organ-on-a-chips can be patient specific platforms to test the effects of disease as well as drug efficacy (drug screening). The devices can also augment existing bioreactor technology necessary for stimulating cells for sustained life. The device could add mechanical stimulus to an existing microfluidic device.

[050] Example 1: Mechanically Active Microfluidic Device to Mimic Tensile Hoop- Strains in the Annulus Fibrosus

[051] In this example we demonstrate an easily controlled, mechanically actuated organ-chip that can apply low (5%) and high (10%) tensile hoop- strains, representing intradiscal strains observed during compression and bending of the disc joint .

[052] Materials and Methods: Devices were manufactured by bonding two sheets of poly-di- methyl- siloxane (PDMS) (25 x 45 mm) to enclose a chamber with a rectangular cross section (1500 x 40 x 1500 pm) ⁷ . Strain applied to the chamber was controlled by altering the location of the chamber within the device and with respect to a barrel-like structure by varying the thickness of the two sheets while maintaining an overall thickness of 4 mm (ti + t ₂ = 4 mm). Therefore, tensile hoop-strains (ε _x ) were increased by positioning the chamber further from the center of the barrel (greater ti, Fig. 5). Strains were applied by deforming the device around a barrel (R = 10.6 mm radius). Two sets of devices were manufactured to apply low (5%) or high-tensile (10%) strain. Tensile strain across the chamber was predicted using beam theory with an ideal geometrical model as a function of the total device thickness (T = ti + t ₂ ), the radius of curvature (R), and the distance (Y) between the chamber and the midway thickness (T 12) called the neutral axis of the device. Strain was actuated with servo motors capable of being placed in an incubator. Experimentally, strain was measured by imaging the chamber before and after bending with a 20X objective lens on an Olympus CKX31 microscope. Strains along the x-axis were calculated by tracking movement of the chamber walls (MATLAB, MathWorks Inc.). Thirty sequential measurements were acquired and tests were performed in triplicate (90 data points to assess repeatability). Between each test, the device was uninstalled and reinstalled to mimic laboratory use. Additional testing for extended use was performed (100 cycles).

Accelerated life testing was also performed to analyze fatigue life of the device.

[053] Results and Discussion: Measured strains were 7.38% %0.88% for the low-strain group and 11.90% ± 1.69% for the high-strain group. These strains were greater than the predictive model by 2.65% and 2.30%, respectively (absolute values). The standard error ranging between -10-20% was just above that of a comparable device, yet the range of strains applied still fell within the range reported for native disc tissues ⁹ . Drift in strain values over 100 cycles was negligible We also demonstrate robustness with device fatigue analysis during multi- week cellular studies and radial compressive strain measurements due to Poisson’s effect. Live-dead analysis demonstrated that bovine fibroblast-like cells were viable after 7-days in culture (Live/Dead ThermoFisher, Waltham, MA). We also demonstrated and evaluated changes in gene expression with tensile strains.

[054] Our disclosed organ-chip can apply tensile hoop-strains and provides a platform to evaluate annulus fibrosus cell mechanobiology.

[055] Example 2. Flexing organ-on-a-chip to model in situ loading of the intervertebral disc

[056] Here we disclose the design, fabrication, and enabling proof-of-concept for a novel microfluidic device for replicating complex strains needed for microphysiological systems of the disc and other annular or vessel-like structures. The disclosed flexing Spine-on-a-chip (SpOC) system enables evaluation of the effects of dynamic physiological hoop strains on cell mechanobiology. We show strong agreement between measured strains in the channel and a corresponding computational model. We also exemplifed cell viability of bovine AF cells seeded within the channel following cyclic loading at 10% strain for 3 hours (1.5Hz).

[057] Device design principles: SpOC is a mechanically actuated, beam- like organ-chip that is flexed in a C-shape to apply a range of strains mimicking intradiscal strains in the AF. Leveraging beam theory, the SpOC applies a scalable and uniform strain to a channel with embedded cells. In other words, tensile hoop strain (epsilon) applied to the cells can be approximated utilizing the total device thickness (T), device radius of curvature (D), and the distance between the channel and the mid plane thickness of the device (‘y’), which is assumed to be the neutral axis. By placing the channel further from the neutral axis along the y-direction or by bending the chip about a tighter radius, the applied hoop strains on the channel increases linearly. While hoop strain increases linearly, a component of compressive strain is also applied, representing biaxial strains observed in degenerated discs in the circumferential and axial directions of the organ. In embodiments we keep the chip radius of curvature constant and instead adjusted the dimensions of the chip to match strains in the degenerated disc.

[058] SpOC is a deformable microfluidic chip constructed from a polymer like polydimethylsiloxane (PDMS), and enclosing a single microchannel that can apply mechanical strain to seeded cells via bending. In embodiments the chip is bent over a stiff cylinder stand, which may be 3D-printed, using for example, packing tape and servo motors. This channel may be embedded within PDMS by way of standard soft lithography. The channel width (e.g. 300 um) and height (e.g. 50 um) were chosen to minimize the strain gradient across embedded cells while providing space for protein accumulation, access to media, and any preferential reorientation of cells (23).

[059] The chip can act like a beam in pure bending, leading to predictable and uniform strain along the channel length, resembling in vivo AF tissue under hoop strain. For example, when the disc is flexed, there are regions of tension and compression in the anterior and posterior AF leading to hoop and axial strains within circumferential sections of the AF. By ‘unravelling’ these circumferential sections, we can re-assign the directions of hoop and axial strains within a linear assembly, resembling the loading modality induced by the applied bending on the chip. SpOC’s bending mechanism can apply uniaxial hoop strains to cell monolayers and biaxial strains in cell-gel constructs. [060] Spine-on-a-chip design and fabrication

[061] The positioning of the microchannel with respect to the neutral surface (NS) of the chip dictates the strain magnitude of cells housed within the channel. For our initial estimation, we assumed that the NS was co-planar with the centroidal plane. When SpOC is bent around a known radius, depth-dependent strains can be estimated using Euler beam theory. Employing brightfield microscopy and image processing, the motion of the channel walls can be optically tracked to measure the applied strain in one direction. We fabricated “low” and “high” strain configurations corresponding to 5% and 10% target strains, respectively. We measured the wall strains to be 7.4% + 0.8% and 11.9% + 1.5% for the “low” and “high” strain configurations, respectively. The small variation in measured strain showcased the repeatability of the strains within a single chip and between chips. While we measured the channel movement optically in one direction, a computational model verified the extent of strain uniformity and established a mathematical relationship to describe strain in three dimensions relative to channel position within the device.

[062] The applied strains measured within the channel align well with previous literature on in situ strains in the AF. Desrochers and Duncan measured a mean of 4.7 + 2.9% (moderate flexion) and 5.7 ± 4.0% (high flexion) surface fiber oriented cell strains from in situ bovine caudal (CC1-CC2) samples (24). Peak fiber-oriented tissue strains can reach 17% (8)(25). The range of peak strains measured on average using magnetic resonance imaging on the outer rings of cadaveric anterior AF samples under sustained compression were between -2.9 to 8.1% radially and -11.3 to 1.8% axially (8). In the same study, peak strains within the posterior AF tissue were measured on average as -3.4 to 6.7% radially and -6.1% to 1.0% axially. [063] The chip can produce tensile and compressive strains that are comparable to other micro-physiological systems that are used to load AF cells, like commercial cell stretchers and hydrostatic pressure chambers (26), with strains ranging from 3 to 20%. Typically, lower strains (3-10%) were used to simulate healthy tissue loading and larger strains (greater than 15%) were used to simulate pathologic loading. However, these systems mainly focus on applying strains in a uniaxial fashion, either tension or compression, while our chip is able to supply both in concert. More recently, microfluidic chips have been developed to apply similar loads to cells within tailored microenvironments. Dai et al. (27) studied the effects of nutrient replenishment and fluid shear stress on mice discs in a microfluidic chamber and Lee et al. (28) proposed a chip that supplies uniaxial compression to chondrocytes suspended in alginate gel. As with these chips, ours can apply long-term static or cyclic loads, but with added versatility of applying them to either monolayers or 3D AF cell cultures. Compared to ex situ and in vitro studies, the chip achieves physiologically relevant loads observed in the AF. By reproducing the loading conditions within the AF, we can be more confident that the biological effects we observe in vitro could translate in vivo. However, while the applied channel strain and measured cell strains from the literature are similar (29), there remains a need to quantify the strain transferred from the channel to the encased cells. We addressed this by later measuring the average strain in vitro using brightfield microscopy and image processing.

[064] Finite element simulation of Spine-on-a-chip bending

[065] We developed finite element models of three chip configurations (0%, 5%, and 10% target strain) to validate our uniaxial strain measurements and establish the relationship between specific target strains along the chip’s bending cycle. We visualized the models’ strain in the x direction, which we considered as ‘hoop strain’ relative to the AF. These results corroborated our one-dimensional strain measurements taken under brightfield microscopy. Then, we selected the ‘high strain’ chip model (10% target) and varied the bending angle to achieve specific strain magnitudes within the channel. We defined the angle from beneath the chip to its outer edge as the bending angle. For 5% and 10% target strain, the model informs us to apply a bending angle of 9.8° and 19.2°, respectively. Moreover, due to the linear relationship between the bending angle and strain below 20°, we can apply controlled loading sequences of varying strains without varying the channel position. For example, we can alternate between 5% and 10% strain for each consecutive cycle or provide a 5% strain for 100 cycles before transitioning to applying a 10% strain for 100 cycles, and so on. Beyond 20°, the strain begins to taper as the bending angle continues to increase. This is because the upper surface of the chip contacts the cylinder, the chip’s radius of curvature is limited to that of the barrel and further lifting end of the chip does not significantly change its curvature. Since strain is directly related to the radius of curvature, it also remains unchanged. For the subsequent loading train, we utilized the model to modify the servo motor control sequence to ensure that the chip applies a 10% strain.

[066] In addition to enabling us to simplify manufacturing, the model provides an inside look at the strain uniformity along the channel length. We observed a change in strain away from the center of the chip, towards the ends of the channel. These changes are likely due to the tendency of the chip to bend into a hyperbolic paraboloid, imposing a unique strain distribution that cannot be estimated with the pure beam bending assumption. The curvature becomes more pronounced towards edges of the chip, resulting in strain deviations in each direction at ~2 mm away from the center of the chip. A more uniform strain distribution may be obtained by redesigning the chip with a 13 mm channel to standardize the applied strain along the full length of the channel. Alternatively, the chip may be widened and the channel length maintained to create a uniform strain distribution while avoiding the reduction of the cell count within the channel. Using the model, the SpOC design is readily iterated to achieve target cell populations and applied strain distributions without the need for extensive lab time or fabrication.

[067] AF cell strain and viability under cyclic load

[068] Prior to strain experiments, we confirmed that cells could be seeded and grown within the PDMS channel. The AF cells remained alive and adhered for more than two weeks in static conditions with a change in media every third day. Then, to establish the feasibility of SpOC as a platform for studying cell mechanobiology within load bearing tissues, we filled the chip’ s microchannel with bovine AF cells and allowed cells to proliferate for 7 days before applying cyclic loading of 10% strain at 1.5 Hz for 3 hours. We took brightfield images of the channel when unbent and bent to obtain ‘unstretched’ images and ‘stretched’ images. By processing the brightfield images with Image J, we later determined the cells’ circularity and average intracellular strains parallel (hoop) and perpendicular (circumferential) to the applied strain along the cell’s major axis. Immediately after completing loading train, we verified cell viability using live/dead cell imaging. Upon visual inspection, the chip showed no signs of wear or fracture after straining for more than 66K cycles at 1.5 Hz at 10% strain. The packing tape cradling the device exhibited discoloration but did not fracture. Servo motors and the apparatus to anchor each component remained intact.

[069] Conclusion

[070] We produced a microfluidic platform, SpOC, to study cell mechanobiology within load bearing tissues found in the spine. The chip design and loading mechanism was able to apply controllable strains to a cell monolayer embedded within its microchannel. In the future, 3D constructs could occupy the channel in the form of hydrogels. The chip can study response to loading such as orientation, gene expression, and response to drugs added to the supernatant.

[071] From the computational modelling perspective, we can quantify the strains at localized regions within the cell monolayer encased within its microchannel using image processing, specifically digital image correlation. This high-resolution strain data enables us to correlate the strain distribution across the microchannel to cell behaviour such as spreading area, proliferation, or apoptosis. Moreover, it allows us to create descriptive computational models to investigate cell biomechanics, including the potential to estimate cell elasticity (30).

[072] Methods

[073] Chip fabrication: Each chip consists of a 17 mm-long straight rectangular channel (300 microns by 50 microns) connecting an inlet and outlet port (0.5 mm diameter). A mylar photomask printed at 10K dpi was created after designing the channel in AutoCAD (AUTODESK, San Rafael). Standard single step photolithography was used to create the master mold with a channel comprised of SU-8 (3050) spin coated onto a four-inch silicon wafer at a thickness of 50 microns (first 500 rpm for 10s with 300 rpm/s, then 3300rpm for 30s with 300 rpm/s, Kayaku Advanced Materials, Westborough, MA). The SU-8 was baked for 14 minutes at 95C before being exposed to 160mJ/cm ^A 2 of UV (Karl Suss MA6) and baking for another 4.5 minutes at 95C. A SU-8 developer was used for 5.5 minutes to develop the partially cured SU-8 on a lab shaker. The SU-8 was then hard baked at 200C for 30 min. By replica molding polydimethylsiloxane (PDMS; Sylgard 184 kit, Dow Chemicals, Midland, MI) at a 10:1 ratio of base-to- crosslinker, a slab of PDMS was created with a channel featured on one side.

[074] After punching 0.5 mm ports on either ends of the channel, the molded PDMS slab was bonded to a second flat PDMS slab using oxygen plasma. Within 10 seconds of the oxygen plasma treatment, PDMS slabs were pressed together with light pressure spread evenly across the chip and bonded using oxygen plasma (30s, RF power 21W, and 70% oxygen flow). Deionized water was added to the channel within 10 minutes of oxygen plasma bonding. The thicknesses of both slabs were controlled to dictate the position the channel (+/- 0.05 mm) and therefore the magnitude of applied strain. The PDMS layer thicknesses ranged from 0.5 to 3.5 mm depending on the desired strain magnitude.

[075] Device actuation and characterization: Polypropylene ’packing’ tape cradling the chip and connected the chip to servo motor horns with pin joints to apply a force to SpOC. As the servo horns rotate upward in tandem the tape pulls tight, forcing the chip to conform to a three- dimensional (3D) printed barrel structure. When the servo horns rotate downward, the chip relaxes to a flat configuration with no applied strain. Throughout actuation the chip remains sandwiched between the 3D printed barrel and a standard 1 mm glass slide to limit movement. Thus, the channel is kept at a constant focal length from the objective lens. The magnitude and frequency of applied strain provided by the servo motors was set to apply 10% strain in the channel at 1.5 Hz using an Arduino UNO microcontroller (motor: Power HD 3001HB connected to an external 3 A, 5V power supply). However, these settings can be adjusted to create an upward limit of hoop strains near 15%. The entire apparatus to deliver applied strains consisting of the two servo motors, 3D printed barrel structure and glass slide were anchored to a metal fixture that could be moved on the microscope table without disrupting the position of the chip. All cyclic loading was applied at room temperature (22C).

[076] Experimentally, hoop strain was measured by imaging the channel before and after bending with a 10X objective lens on an Olympus CKX31 microscope which peers through a 5x5mm aperture in the barrel structure. Images were acquired before and after applied strain and the channel width was measure in the reference and deformed configuration. Hoop strains were calculated as engineering strain by dividing the change in channel width with loading by the initial channel width measured in the reference configuration (MATLAB, MathWorks Inc.). To assess experimental repeatability, thirty sequential measurements were acquired to assess changes in loading over an extended period of time. This loading sequence was considered to be a single set, which was repeated three times (total of 90 data points) to assess repeatability of measurements acquired after uninstalling and reinstalling the SpOC.

[077] To confirm that chips could physically withstand extended cyclic loading, an accelerated life test was conducted by cyclically loading chips the equivalent of three hours of loading a day for a week. The chips and the device were examined before and after for signs of wear or crack propagation. Given the likelihood of fast fracture occurring during a single cycle of strain, cracks of 1 mm or larger were searched for with the naked eye.

[078] Device simulation: Experimental measurements of channel movement could determine the tensile hoop strains during loading, but not 3D strains. A finite element (FE) model was developed to validate laboratory measurements and determine 3D strain components. 3D strains were compared to native AF tissue strains to confirm physiological relevance. FE models were also used to assess strain heterogeneity, which can be used to alter the channel design for maximizing channel size.

[079] Models were created in Solidworks (Dassault Systemes) with the channel positioned for three target strain configurations, including 0%, 5%, and 10% according to beam theory. For these configurations, the channel was placed at 0.00 mm, 0.63 mm, and 1.26 mm below the neutral axis, respectively. Each model was imported into FEBio Studio 1.3 (31) and meshes were constructed using first order tetrahedral elements for the chip. A mesh convergence analysis was conducted for the 10% strain configuration to confirm stability of the model results. [080] A Y-displacement of 7 mm was applied to the end of the chip using a rigid deflector, replicating the function of the tape in a more computationally stable manner. This enforced the chip’s curvature during the test cycle that starts when the chip begins to bend until it reaches maximum curvature. Model predicted chip curvature was verified with the experimental setup by overlaying images of the chip in the deformed condition with the model output. Strains in the x, y, and z directions were evaluated along the length of the channel. Lab-based strain measurements from the center of the chip were compared to model results at the same location. The percent error between the model’s prediction and empirical measurements of hoop strain was calculated. Lastly, the applicability of the Neo-Hookean constitutive model was confirmed for the purposes of small strains in the PDMS chip by using FEBio Studio to model the ASTM D412 tensile testing conducted by Johnston et al. (32)

[081] References

[082] 1. Showalter BL, DeLucca JF, Peloquin JM, Cortes DH, Yoder JH, Jacobs NT, et al. Novel human intervertebral disc strain template to quantify regional three-dimensional strains in a population and compare to internal strains predicted by a finite element model. Journal of Orthopaedic Research. 2016 Jul 1; 34(7): 1264-73.

[083] 2. Amin DB, Moawad CM, Costi JJ. New Findings Confirm Regional Internal Disc Strain Changes During Simulation of Repetitive Lifting Motions. Annals of Biomedical Engineering. 2019.

[084] 3. O’Connell GD, Elliott DM. DEGENERATION AFFECTS THE STRUCTURAL AND TISSUE MECHANICS OF THE INTERVERTEBRAL DISC Grace Divine O’Connell.

[085] 4. Yang H, Haldeman S, Lu ML, Baker D. Low Back Pain Prevalence and Related Workplace Psychosocial Risk Factors: A Study Using Data From the 2010 National Health Interview Survey. Journal of Manipulative and Physiological Therapeutics. 2016 Sep l;39(7):459-72.

[086] 5. Schwarzer AC, Aprill CN, Derby R, Fortin J, Kine G, Bogduk N. The Prevalence and Clinical Features of Internal Disc Disruption in Patients With Chronic Low Back Pain. Spine. 1995 Sep;20(17).

[087] 6. Shahraki NM, Fatemi A, Agarwal A, Goel VK. Prediction of clinically relevant initiation and progression of tears within annulus fibrosus. Journal of Orthopaedic Research. 2017 Jan 1;35(1): 113-22.

[088] 7. Rahyussalim AJ, Zufar MLL, Kurniawati T. Significance of the association between disc degeneration changes on imaging and low back pain: A review article. Asian Spine Journal. 2020; 14.

[089] 8. O’Connell GD, Johannessen W, Vresilovic EJ, Elliott DM. Human Internal Disc Strains in Axial Compression Measured Noninvasively Using Magnetic Resonance Imaging. Spine. 2007 Dec;32(25).

[090] 9. Hoy D, March L, Brooks P, Blyth F, Woolf A, Bain C, et al. The global burden of low back pain: Estimates from the Global Burden of Disease 2010 study. Annals of the Rheumatic Diseases. 2014;73(6):968-74.

[091] 10. Hosea TM, Hannafin JA. Rowing Injuries. Vol. 4, Sports Health. 2012. p. 236-45.

[092] 11. Reid DA, McNair PJ. Factors contributing to low back pain in rowers. Vol. 34, British Journal of Sports Medicine. 2000. p. 321-2.

[093] 12. Daly C, Ghosh P, Jenkin G, Oehme D, Goldschlager T. A Review of Animal Models of Intervertebral Disc Degeneration: Pathophysiology, Regeneration, and Translation to the Clinic. Vol. 2016, BioMed Research International. Hindawi Limited; 2016.

[094] 13. Kaarj K, Yoon JY. Methods of delivering mechanical stimuli to Organ-on-a- Chip. Vol. 10, Micromachines. MDPI AG; 2019.

[095] 14. Yang B, O’Connell GD. Intervertebral disc swelling maintains strain homeostasis throughout the annulus fibrosus: A finite element analysis of healthy and degenerated discs. Acta Biomaterialia. 2019 Dec l;100:61-74.

[096] 15. White A, Panjabi M. Clinical Biomechanics of the Spine. Philadelphia: Lippincott; 1978.

[097] 16. latridis JC, Mente PL, Stokes IAF, Aronsson DD, Alini M. Compression- Induced Changes in Intervertebral Disc Properties in a Rat Tail Model. Vol. 24, SPINE. [098] 17. Molinos M, Almeida CR, Caldeira J, Cunha C, Goncalves RM, Barbosa MA. Inflammation in intervertebral disc degeneration and regeneration. Vol. 12, Journal of the Royal Society Interface. Royal Society of London; 2015.

[099] 18. Shahraki NM, Fatemi A, Agarwal A, Goel VK. Prediction of clinically relevant initiation and progression of tears within annulus fibrosus. Journal of Orthopaedic Research. 2017 Jan 1 ;35(1): 113-22.

[0100] 19. Urban JPG, Roberts S, Ralphs# JR. The Nucleus of the Intervertebral Disc from Development to Degeneration 1. Vol. 40. 2000.

[0101] 20. Bezci SE, Werbner B, Zhou M, Malollari KG, Dorlhiac G, Carraro C, et al.

Radial variation in biochemical composition of the bovine caudal intervertebral disc. JOR SPINE. 2019 Sep;2(3).

[0102] 21. Rudert M, Tillmann B. Lymph and blood supply of the human intervertebral disc: Cadaver study of correlations to discitis. Acta Orthopaedica. 1993;64(l):37-40 [0103] 22. Benoist M. Natural history of the aging spine. Vol. 12, European Spine Journal. 2003.

[0104] 23. Abbott RD, Howe AK, Langevin HM, latridis JC. Live free or die: Stretch- induced apoptosis occurs when adaptive reorientation of annulus fibrosus cells is restricted. Biochemical and Biophysical Research Communications. 2012 May 4;421(2):361-6.

[0105] 24. Desrochers J, Duncan NA. Strain transfer in the annulus fibrosus under applied flexion. Journal of Biomechanics. 2010 Aug;43(ll):2141-8.

[0106] 25. Heuer F, Schmidt H, Wilke HJ. The relation between intervertebral disc bulging and annular fiber associated strains for simple and complex loading. Journal of Biomechanics. 2008;41(5): 1086-94.

[0107] 26. Molladavoodi S, McMorran J, Gregory D. Mechanobiology of annulus fibrosus and nucleus pulposus cells in intervertebral discs. Vol. 379, Cell and Tissue Research. Springer; 2020. p. 429-44.

[0108] 27. Dai J, Xing Y, Xiao L, Li J, Cao R, He Y, et al. Microfluidic Disc-on-a-Chip Device for Mouse Intervertebral Disc - Pitching a Next- Generation Research Platform to Study Disc Degeneration. ACS Biomaterials Science and Engineering. 2019 Apr 8;5(4):2041-51.

[0109] 28. Lee D, Erickson A, You T, Dudley AT, Ryu S. Pneumatic microfluidic cell compression device for high-throughput study of chondrocyte mechanobiology. Lab on a Chip. 2018;18(14). [0110] 29. O’Connell GD, Vresilovic EJ, Elliott DM. Human intervertebral disc internal strain in compression: The effect of disc region, loading position, and degeneration. Journal of Orthopaedic Research.2011 Apr;29(4):547–55. [0111] 30. Chen C-T, Gu GX. Learning hidden elasticity with deep neural networks. Available from: https://www.pnas.org/lookup/suppl [0112] 31. Maas SA, Ellis BJ, Ateshian GA, Weiss JA. FEBio: Finite elements for biomechanics. Journal of Biomechanical Engineering.2012;134(1). [0113] 32. Johnston ID, McCluskey DK, Tan CKL, Tracey MC. Mechanical characterization of bulk Sylgard 184 for microfluidics and microengineering. Journal of Micromechanics and Microengineering.2014;24(3). [0114] 33. Jin L, Liu Q, Scott P, Zhang D, Shen F, Balian G, et al. Annulus fibrosus cell characteristics are a potential source of intervertebral disc pathogenesis. PLoS ONE. 2014 May 5;9(5). [0115] 34. Dogru S, Aydemir D, Salman N, Ulusu NN, Alaca BE. Impact of PDMS surface treatment in cell-mechanics applications. Journal of the Mechanical Behavior of Biomedical Materials.2020 Mar 1;103.