Lateral trapezoid microfluidic platform for investigating mechanotransduction of cells to spatial shear stress gradients

Accepted Manuscript Title: Lateral Trapezoid Microfluidic Platform for Investigating Mechanotransduction of Cells to Spatial Shear Stress Gradients Authors: Rebecca Soffe, Sara Baratchi, Mahyar Nasabi, Shi-Yang Tang, Andreas Boes, Peter McIntyre, Arnan Mitchell, Khashayar Khoshmanesh PII: DOI: Reference:

S0925-4005(17)30968-1 http://dx.doi.org/doi:10.1016/j.snb.2017.05.145 SNB 22429

To appear in:

Sensors and Actuators B

Received date: Revised date: Accepted date:

30-1-2017 1-5-2017 24-5-2017

Please cite this article as: Rebecca Soffe, Sara Baratchi, Mahyar Nasabi, Shi-Yang Tang, Andreas Boes, Peter McIntyre, Arnan Mitchell, Khashayar Khoshmanesh, Lateral Trapezoid Microfluidic Platform for Investigating Mechanotransduction of Cells to Spatial Shear Stress Gradients, Sensors and Actuators B: Chemicalhttp://dx.doi.org/10.1016/j.snb.2017.05.145 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Lateral Trapezoid Microfluidic Platform for Investigating Mechanotransduction of Cells to Spatial Shear Stress Gradients Rebecca Soffe,1,2* Sara Baratchi,3 Mahyar Nasabi,2 Shi-Yang Tang,2 Andreas Boes,2 Peter McIntyre,3 Arnan Mitchell,2 Khashayar Khoshmanesh2 1

Department of Electrical and Computer Engineering, University of Canterbury,

Christchurch, New Zealand. 2

School of Engineering, RMIT University, Melbourne, Australia.

3

School of Health and Biomedical Sciences, RMIT University, Melbourne, Australia.

E-mail: [email protected]

We developed a platform, to investigate the influence spatial shear stress gradients on intracellular calcium signalling of non-adherent cells.

Abstract Microfluidic platforms enable the influence of a variety of chemical and physical gradients on single or multiple cells to be examined and monitored in real-time. Research into chemical gradients has been more prevalent in literature; however, with the interest in mechanotransduction, research investigating the influence of physical gradients, such as force from shear stress, are appearing. Shear stress studies, to date have been focused on using parallel plate flow chambers, or shear stress gradients spanning the length of the microchannel. In this paper, we designed a Trapezoid microchannel, which enables production of a unique lateral spatial shear stress gradient, in order to examine the responses it evokes in cells, using microscopy. We investigated the effect of a lateral spatial shear stress gradient on intracellular

calcium signalling in HEK-293-TRPV4 cells. The Trapezoid microchannel was designed to produce a spatial shear stress gradient, spanning the width of the microchannel. Additionally, we investigated the effect of temporal shear stress using the microfluidic platform, to further examine the complexity of transduction of shear stress force that mimic pathological and physiological conditions into cellular signalling. We show that the temporal pattern of shear stress, modulated intracellular calcium signalling, with slower response times (responding to stimulus and reaching maximum Calcium influx), compared to those observed in the absence of temporal shear stress. Our experiments highlight the importance of considering the nature of shear stress under physiological and pathological conditions – with microfluidic platforms providing a useful means to do so.

Keywords: Microfluidics, Spatial Shear Stress, Temporal Shear Stress, Discontinuous Dielectrophoresis, Intracellular Calcium Signalling, HEK-293

1. Introduction Microfluidic-based solutions enable of the effect of physiological and pathological conditions on cells to be investigated [1-5]. Researchers have demonstrated the potential of microfluidic platforms in these areas, from single cells analysis to tissues/organs- on chip [4-10]. Microfluidic platforms offer an alternative platform for biological analysis compared to those currently readily available. Advantages of microfluidics platforms, include an increased range and diversity of possible experimental conditions, better control of the extracellular environment, reduced reagent volumes, decreased sample size, and high-throughput [3, 1116]. In this paper, we present a microfluidic platform, for the study of mechanical (i.e. deformation) and biochemical (i.e. intracellular signalling) responses of loosely adherent or non-adherent cells upon exposure to a spatial shear stress gradient. With interest growing in mechanotransduction, microfluidic platforms are being utilised to measure the response of cells to mechanical stimulus from their microenvironments, such as shear stress [15-18]. Mechanotransduction is of interest, as cells within the human body experience mechanical stimulus for example: endothelial cells, experience constant shear stress from blood flow; muscle and epithelial cells, experience stretch forces; joints and cartilage experience gravitational forces; and mechanical pain via sensory neurons [19-21]. The cellular response to mechanical stimulus is crucial for cellular homeostasis and function and influences a range of cellular functions, for instance, proliferation, differentiation, and migration [22, 23]. Mechanosensitive ion channels, are a group of mechanoreceptors expressed on the plasma membrane of different cell types. Mechanotransduction through these channels plays an

important role in maintaining cellular homeostasis and function [24]. Additionally, abnormalities in the functioning of mechanosensitive channels can result in health related issues, such as polycystic kidney, neuronal, muscular degeneration, and cardiac diseases [25, 26]. Mechanical stimulation comes in various forms, such as shear stress, tensile stress (stretch and strain), cell compression, and gravitational forces; with various microfluidic platforms have been designed to replicate these natural forces [13, 15, 27-29]. The influence of shear stress is of interest, as cells are exposed to shear stress in their microenvironments which vary in complexity [30-33]. Commonly used shear devices include cone and plate apparatus as well as parallel-plate chambers. There devices however, only generate uniform shear stress throughout an experiment [34-36]. Microfluidic platforms, offer an alternative to these devices, as they offer control and variability of the microenvironment, whether it be temporal, spatial, or chemical/drug investigation [13, 36-38]. Examples of microfluidic platforms which investigate the influence of shear stress on biological responses, include: (1) uniform or singular shear stress, in which one shear stress level is present throughout the channel, and is produced using a straight channel. For instance, these platforms include a singular straight channel or multi-channel devices with channels of varying widths to produce discrete levels of shear stress [34, 36, 38-42]. However, fluid flow in the human body does not produce uniform shear stress, which highlights a need for more complicated shear stress profiles to be investigated. (2) Spatially resolved shear stress gradient, examples of these devices include: contracting-expanding geometries, tapered microfluidic channels, and utilising a Braille display, to generate shear levels capable of fluid mechanical regulation [11, 37, 43-46]. The vast majority of these devices, can be limited in the extent of the shear stress range that can be investigated in one experiment. In addition, the limited field of view of most microscopes used for these studies means that cells experiencing different shear stress levels may not be able to be simultaneously viewed under the microscope. More complicated microfluidic platforms encompass intersecting channels or flexible membranes, which influence the flow profile over time, and are commonly used in organ-on-chip studies [13, 47-49]. These platforms tend to be more focused on replicating and examining tissues, with their more complex functions, such as organ-on-chip or interstitial-like flow, which attempt to closely replicate living organs. Here, we present and characterise a Trapezoid microchannel, to produce a spatial shear stress gradient, whilst taking advantages of microfluidics, such as small reagent/cell volumes. Several advantages in comparison to the aforementioned microchannel geometries, include:

(1) the inability to measure different shear stress levels simultaneously, and (2) limitations in the microchannel design. (1) In comparison to tapered microchannels (which taper along the length of the microchannel), or multiple microchannels it can provide difficult to observe different levels of shear stress simultaneously [11, 36, 44, 45]. As the shear stress range does not fit in one point of view, for instance multiple channels can be at different positions on the platform [34, 42]. For a tapered microchannel, the tapering is often extend out of view, or is designed to have a small area in which shear stress changes, which could cause difficulty in determining the shear stress that the cell is experiencing with the shear stress change being rapid across this small area. However, the trapezoid microchannel, on the other hand enables cells to always be located at one level of shear stress throughout the entire channel, so it can be readily identified what level of shear stress the cell experiences. (2) In comparison to expanding-contracting microchannels, these often only enable a large shear stress range to be observed in one small section of the microchannel – thus, limiting the sample population size [44]. Furthermore, the change in shear stress may not necessarily be linear due to the shape of the expanding-contracting microchannel. Furthermore, some platforms are limited in the available immobilisation strategies. Limitations in current immobilisation strategies, such as surface coating, precludes high shear stress analysis, as cells often detach under high levels of shear stress [50, 51]. We developed the Discontinuous Dielectrophoresis technique to overcome these limitations [15, 39]. The Discontinuous Dielectrophoresis approach uses an electric field to immobilise cells, followed by electric field deactivation with biologically relevant buffers flushed through the microchannel, to attach cells to their substrate, which enables high shear stress analysis of loosely adherent cells. The interdigital microelectrodes were required to be perpendicular to the flow profile to ensure a uniform distribution of cells over the entire spatial shear stress gradient, which resulted in a lateral trapezoidal design spanning the width of the microchannel. Warkiani et. al., introduced the concept of a trapezoidal microchannel, where they used a slanted spiral microchannel for the microfiltration of blood plasma and isolation of circulating tumour cells [52, 53]. Here, in contrast we use the concept of a trapezoid microchannel for producing a spatial shear stress gradient, to facilitate physiological investigations. We demonstrate its utility for monitoring intracellular calcium responses in HEK-293-TRPV4 cells, and compare the cellular response to uniform shear stress produced by a straight microchannel. The intracellular calcium level, is meticulously controlled by: efflux from the cell through ion transporters, sequestration in intracellular compartments, buffering high affinity cytosolic Calcium-binding proteins, and influx through regulated ion

channels [54]. The Calcium ion is a versatile and potent intracellular messenger, which regulates a wide range of spatial and temporal cellular signals. The TRPV4 is a mechanosensitive ion channel expressed in vascular endothelial cells, and is a non-selective, Ca2+ permeable, cation channel, which responds to various environmental stimuli, such as mechanical, thermal, and osmotic stimuli, and plays important roles in homeostasis and regulation of cellular processes [26, 55]. The trapezoid microchannel offered the ability to classify cells as being located in different regions in the microchannel, and thus enable shear stress comparisons to be undertaken concurrently. To demonstrate the efficacy of our microfluidic platform, we investigated the intracellular calcium signalling response of HEK293-TRPV4 to spatial shear stress gradients, using imaging of the Calcium sensitive dye, Fluo-4AM [44]. We also conducted a dynamic analysis by applying temporal shear stress, to observe this influence on intracellular calcium signalling. Our results revealed that spatial shear stress gradients sensitise the response of TRPV4 channels to shear stress – as was observed by examining the normalised intensity profiles. This paper is structured as follows: in the Materials and Methods section, we discuss cell preparation, data analysis, device fabrication, numerical simulations, and an outline of the Discontinuous Dielectrophoresis procedure. In the Results and Discussion section, we discuss elements of the Trapezoid microchannel and subsequent simulations, and experimental results pertaining to the investigation of the intracellular calcium signalling response to spatial shear stress gradients, including results for temporal shear stress, with appropriate discussions.

2. Materials and Methods 2.1 Cell Preparation HEK-293 T-REx (Life Sciences) cell lines stably expressing human TRPV4 (Transient Receptor Potential Vaniloid type 4) were produced as reported elsewhere [56]. For each experiment, cells were grown in tetracycline-free DMEM media supplemented with 10% FBS, blasticidin (5 µg/ml) and hygromycin (50 µg/ml). Secondly, TRP channels expression was induced using 0.1 µg/ml of tetracycline for four hours before each experiment. For a negative control non-transfected HEK-293 T-REx cells were used. To enable dielectrophoretic immobilisation, cells were diluted in a low electrical

conductivity (LEC) buffer; such that 20 µl of loaded cells were suspended into 1000 µl of LEC comprising of 8.5% w/v sucrose and 0.3% dextrose in deionised water – taking care to minimise pre-exposure of cells to shear stress when mixing.

2.2 Measuring of Calcium Sensitive Dyes Intracellular calcium levels were measured by microfluorimetry of Fluo-4AM loaded cells, as previously described [44]. Assays were performed in HEPES buffered saline (HBS) containing 140 mmol/l NaCl, 5 mmol/l KCl, 10 mmol/l HEPES, 11 mmol/l D-glucose, 1 mmol/l MgCl2, 2 mmol/l CaCl2, and 2 mmol/l probenecid, adjusted to pH = 7.4 [57]. Intensity measurements were obtained using an inverted microscope (Nikon Eclipse, TE 2000) equipped with a: photomultiplier tube, near infrared camera (QuantEM:512SC, Photometrics), 10× objective (CFI Plan Apo Lambda 10×), green and blue filters, and a Hg precentred fibre illuminator (Nikon Intensilight). Utilising the inbuilt time-lapse function of NIS elements, microscope imaging software (Basic Research, Nikon Instruments), snapshot images were acquired every two seconds throughout the entire experiment with the cells illuminated with the blue filter equipped.

2.3 Data Analysis Fluorescence intensity was obtained using NIS elements (Nikon Instruments) by analysing areas by defining regions of interest (ROIs) encapsulating single cells. Fluorescent images were acquired through the use of NIS elements and equipped near infrared camera, with images taken every two seconds throughout the experimental durations. Selected ROIs were defined and measured to quantify changes in the average intensity of [Ca2+]i fluorescence was reported as a ratio of F1/F0, where F0 is the fluorescence at time zero and F1 is fluorescence at the time of reading. For analysis, activation was defined as an increase in the summed fluorescent signal, from test groups which exceeded three standard deviations from the mean of the control. Results are presented as mean ± SEM (standard error mean). For statistical analysis, student t test or ANOVA was performed using the Graph pad prism software, in addition, P value of less than 0.05 were considered significant (*P<0.05, **P<0.01, and ***P<0.001).

2.4 Viability Assay At the completion of each experiment, the cell viability was verified on chip, using propidium iodide (PI) stain (10 µg/ml); with the non-viable cells excluded from analysis of intracellular

results. Furthermore, the viability of cells with both spatial and temporal shear stress were examined using GSK1016790A.GSK1016790A (Sigma-Aldrich), a selective and cellpermeable agonist of the TRPV4 ion channel, was dissolved in DMSO. To obtain a desired concentration of 12.5 nM the GSK1016790A was diluted with HBS [57].

2.5 Microelectrode Fabrication Indium tin oxide (ITO) coated glass slides (Sigma Aldrich), with a surface resistivity of 8 to 12 Ω/sq were used as a base substrate, to produce the ITO microelectrodes. An interdigital microelectrode configuration was utilised to maximise the immobilised cell population, such that the width of the microelectrode fingers and spacing between the fingers were 40 µm, with the fingers of the two electrodes overlapping by 1180 µm to produce the active field. The microelectrodes were patterned using standard photolithography techniques, and then etched using adaptations of previously reported processes for etching ITO microelectrodes [39, 58, 59]. The microelectrodes were etched using concentrated Hydrochloric acid (HCl) at room temperature, for 11 minutes. To ensure that the microelectrodes were correctly etched, a continuity test was undertaken at the completion of etching, for every substrate/microelectrodes etched.

2.6 Microchannel Fabrication To produce a lateral shear stress gradient, several simulations were carried out to determine a suitable trapezoid microchannel design. The short sidewall, long sidewall, and channel width were set to 15, 150, and 300 µm, respectively. The design and simulation results for the microchannel are described in the Results and Discussion section in further detail. The mold master was fabricated using direct laser fabrication and the Polydimethylsiloxane (PDMS) microfluidic channel was produced using standard soft replica molding techniques [60]. To produce the mold master a 3D (three-dimensional) femtosecond direct laser writing technique, which is also known as two photon polymerisation (Photonic Professional GT, Nanoscribe) was employed - which enables the fabrication of nearly arbitrary 3D structures in suitable photoresists [61, 62]. The laser used for this process had a repetition rate of 80 MHz, which allows for fast scanning speeds and an average power of less than 180 mW. The centre wavelength (λ = 780 nm) of the laser light pulses (tP approximately 100 fs) lies well below the absorption edge of the photoresist; however, by tightly focusing the laser pulses, the intensity in the focal spot is high enough to expose the photoresist by multi-photon absorption, which is a nonlinear optical effect. Due to the scale of the microchannel, and

limitations of the Nanoscribe laser scanning volume of 300 µm by 300 µm by 300 µm in the X, Y, and Z direction, a stage movement stitching technique was invoked using the Nanoscribe. Such that the laser would move to the new area, with a partial overlap between the consecutive exposures; the structures were then stitched together, with a stitching error of 1 µm. For the fabrication of the microchannel, we used the IP-PH photoresist (Microchip), which was applied to a Silicon wafer substrate. After the 3D exposure with the direct laser writing technique, the unexposed photoresist was then developed by immersing the mold master in PGMEA (Propylene Glycol Monomethyl Ether Acetate, Dow Chemical Company), a chemical developer, revealing the trapezoid microchannel. This process was selected above other fabrication techniques, due to the rapid fabrication process of the mold master, from design to prototype; in addition, minimal limitations in terms of resolution, especially in the case of the incline of the top of the microfluidic channel between the short and long sidewalls of the micro channel.

2.7 Numerical simulations To determine the flow induced shear stress within the trapezoid microfluidic channel, Computational Fluid Dynamic techniques were used. For a flow rate of 60 µl/min, the highest flow rate we used throughout experiments has a corresponding Reynolds number of approximately five, indicating that the flow was inherently laminar. Furthermore, the buffer media was considered as a Newtonian fluid. Navier-Stokes equations were solved using ANSYS Fluent 6.3 software package (Canonsburg, PA, USA), and the pressure-velocity terms were coupled using the SIMPLE algorithm. For discretization of convection terms, the Second-Order Upwind Scheme was used. Boundary conditions that were considered for simulation, included, an ambient pressure at the inlet (due to presence of a large inlet reservoir), and the application of desired flow rates at the outlet (as the buffer was withdrawn from the outlet). In addition, the surface of glass substrate, microchannel, and immobilised cells were considered as rigid walls with no-slip boundary condition applied. All simulations were undertaken in three-dimensions, in both the absence and presence of immobilised cells, as further detailed in Spatial Shear Stress Gradient: Simulation Analysis section.

2.8 Discontinuous Dielectrophoresis Discontinuous Dielectrophoresis, was utilised to immobilise the HEK-293-TRPV4 cells, which we have previously used to immobilise cells and investigate the intracellular calcium

response of HEK-293-TRPV4 cells to uniform high shear stress, and concurrent chemical and shear stress stimulation [17, 39, 63]. In summary, cells were applied to the microfluidic platform at a flow rate of 1.5 µl/min, producing a maximum shear stress of 0.87 dyn/cm2. Once the cells were flowing over the microelectrodes, the electric field was activated to immobilise the cells along the microelectrodes; activation was achieved by applying an alternating-current sinusoidal signal operating at 10 MHz, 5 Vpk-pk. After a duration of 120 s, the electric field was deactivated; consequently, enabling a suitable population of cells to be immobilised for analysis. In turn, the LEC buffer was exchanged for HBS. The HBS was then washed through the microfluidic platform for ten minutes, to allow the cells to equilibrate. We have previously explained our procedure of Discontinuous Dielectrophoresis, in detail in ‘Analysing the calcium signalling of cells under high shear flows using discontinuous dielectrophoresis,’ using HEK-293-TRPV4 cell [39]. This investigation revealed that the Discontinuous Dielectrophoresis procedure does not influence the HEK-293-TRPV4 cells’ ability to respond to shear stress stimulus, or produce significant shear stress to induce an intracellular calcium signalling response.

3. Results and Discussion To examine the influence of spatial shear stress gradients on intracellular calcium signalling of non-adherent or loosely adherent cells, we took advantage of Discontinuous Dielectrophoresis (Figure 1). This approach was developed by our group, for immobilising loosely adherent cells, such as HEK-293 cells to withstand high levels of shear stress. Discontinuous Dielectrophoresis has enabled a trapping efficiency greater than 90%, at a (single) high shear stress level of 63 dyn/cm2 for HEK-293 cells and has minimal effects on cell viability (greater than 90%) [39]. The microfluidic platform comprises of interdigital Indium Tin Oxide (ITO) microelectrodes, and a trapezoid microchannel. The ITO microelectrodes, did not affect the trapping efficiency of cells (Figure 1c(ii)). Figure 1c(iii) displays a photograph of the trapezoid microchannel, designed to produce the spatial shear stress gradient. We have previously reported results of the response of HEK-293-TRPV4 cells to a uniform shear stress level (present throughout the entire straight microchannel)[39]; we now expand on these results and report the response of HEK-293-TRPV4 cells to spatial shear stress gradient.

3.1 Trapezoid Microchannel Design and Simulations The trapezoid microchannel was designed to produce a lateral shear stress gradient, along the width of the channel, such that cells across the channel would experience different shear stress levels (Figure 2) - see Supplementary Movie 1 to visualise the flow going over the immobilised cells, at different applied shear stress levels. The ability to produce a spatial shear stress gradient is of importance for gaining a better understanding of the signalling responses of cells, which are subjected to shear stress in their microenvironments that vary in complexity [30]. For simplicity, we classify shear stress profiles, in reference to the lateral cross-section, as: (1) uniform shear stress, which occurs in a straight microchannel with a singular level of shear stress present throughout the channel and (2) a spatial shear stress gradient, in which the shear stress varies spatially throughout/across the channel in some form or another. The trapezoid channel produces a spatial shear stress gradient across the width of the microchannel. An extensive numerical analysis via stimulation was carried out, to observe the shear stress profile across the trapezoid microchannel, in the event no cells are present, for different combinations of channel width and sidewall heights (Supplementary Information S1). In conclusion of this analysis, the microchannel width was set to 300 µm, with sidewall heights of 15 and 150 µm (Figure 2a) - with resulting shear stress profile and velocity contours presented in Figure 2b and Figure 2c, respectively (for an applied flow rate of 60 µl/min). The main contributing factors for the selected channel dimensions were, as follows: (1) The short sidewall height needed to consider the diameter of the HEK-293 cell, which is approximately15 µm. (2) The long sidewall height needed to consider the height of the short sidewall, such that the shear stress range across the width of the channel was maximised. (3) The channel width was selected to achieve the largest possible shear stress range across the width of the channel along the glass substrate, whilst ensuring to minimise the width of the microchannel to limit the fabrication time of the microchannel mold master – which was completed on the Nanoscribe. For analysis purposes, the microchannel was arbitrarily divided into four different regions across the width of the microchannel, with average shear stress ratios of 1, 2, 2.9, 3.5, respectively at a flow rate of 60 µl/min Figure 2b. The regions spanned the following areas: 5 to 50, 60 to 110, 120 to 160, 180 to 240 µm, with the origin centred at the left (short) sidewall, corresponding to Regions 1 to 4, respectively. Initially, the Discontinuous Dielectrophoresis procedure trapping efficiency was examined in the presence of the spatial

shear stress gradient (Figure 2d). The trapping efficiency was the number of cells remaining after the application of shear stress divided by the total number of cells initially immobilised; with the shear stress examined over an extensive range (i.e. 0 to 58 dyn/cm2 for Region 4), with the shear stress increased every 300 s. A trapping efficiency of 73% in Region 4 was observed at a shear stress of 58 dyn/cm2. In comparison to the case of uniform shear stress, which resulted in a trapping efficiency greater than 90% for a shear stress of 63 dyn/cm2, as previously reported [39]. We conjectured this difference was attributed to a different behaviour of cell adhesion molecules (CAMs) under spatial shear stress gradient, in comparison to a uniform shear stress. However, for investigating the spatial shear stress gradient on intracellular calcium signalling of HEK-293-TRPV4, the maximum shear stress exposure was limited to 35 dyn/cm2, to ensure that the trapping efficiency remained around 90%. Figure 2e, provides an experimental spatial shear stress gradient application, using the trapezoid microchannel platform, for the investigation of intracellular calcium signalling of HEK-293-TRPV4 cells to a spatial shear stress gradient. To ensure the trapezoid microchannel produces a spatial shear stress gradient in the presence of cells, a further analysis was carried out, using shear stress contours, as explained in the subsequent section.

3.2 Spatial Shear Stress Gradient: Simulation Analysis To determine the characteristic influence of the spatial shear stress gradient on the cells during the application of high levels of shear stress, simulations were conducted, for immobilised cells located within the trapezoid microchannel (Figure 3). Several aspects of the Discontinuous Dielectrophoresis procedure and the application of high levels of shear stress on HEK-293 cells were considered when conducting the simulations, as follows: (1) Discontinuous Dielectrophoresis does not influence the shape of the cell, both during and after immobilisation of the cells, due to a short period the cells are exposed to the electric field. (2) The application of the electric field did not influence the behaviour of either the cells or liquid flow, as the electric field was only activated during the initial immobilisation process, whilst, carrying out the Discontinuous Dielectrophoresis procedure (Materials and Methods). Thus, negating effects which occur on cells in the presence of electric fields, such as cell deformation and formation of secondary flows, for example electro-thermal vortices, due to Joule heating effects. (3) The immobilised cells influence the spatial distribution of flow variables, for example, the velocity gradient and flow velocity, in close proximity of the cells. Consequently, the spatial distribution influences the magnitude of flow-induced shear stress imposed on the cell, compared to the maximum shear stress experienced on a flat surface in the absence of cells – such a shear stress is denoted as the Reference Shear, throughout the paper. For analysis, when describing the Reference Shear for a region, this implies the average Reference Shear experienced over the entire region. Furthermore, the term Base Reference Shear implies the Reference Shear experienced in Region one – the other regions are determined by the aforementioned ratios of 2, 2.9, and 3.5, for Regions 2 to 4, respectively. In simulations, an applied flow rate of 60 µl/min was used (the highest flow rate used during calcium signalling experiments), which corresponds to a Base Reference Shear of 10 dyn/cm2. (4) The cell volume was maintained to account for contact with the glass substrate and the presence of adjacent cells [17]. We investigated configurations of five independent equidistant immobilised cells spanning the width of the microchannel and an immobilised cell cluster containing nine cells located in Region 4. In the case of five (independent) equidistant immobilised cells spanning the width of the microchannel (Figure 3a(i)), the extent of the maximum shear region (indicated in red) on the top surface of the cells increases from Cell 1 to 4, which are located in Regions 1 to 4, respectively. Cell 5 experiences a higher shear stress than Cell 3, due to the steeper shear stress gradient experienced over the entire cell. Furthermore, for all five cells low shear regions are experienced along the cell surface in close proximity to the glass substrate. The

presence of the spatial shear stress gradient was more apparent in the spatial distribution of shear stress over the cell surface (Supplementary Information S2), especially for Cells 1 to 3,. Consequently, the average shear stress experienced over Cells 1 to 5 is higher than the Reference Shear (purple line) at that location in the microchannel Figure 3b(i) – which on average is approximately 1.47 times higher. The results are consistent with our previously reported findings for the case of a single cell, located on the substrate in the centre of a straight microchannel producing uniform shear stress, which was 1.4 times the magnitude of the relative Reference Shear [17]. Further simulations were conducted for the case of an immobilised cell cluster containing nine cells located in Region 4 (Figure 3a(ii)), the more applicable case when using Dielectrophoresis to immobilise cells; where often cells are immobilised in row-like arrangements along the interdigital microelectrodes, consequently forming cell clusters of various depth and width. The cells experienced maximum shear stress along their top surface, whilst minimum shear stress was experienced at interfaces with adjacent cells and the glass substrate. Which was consistent with our previously reported simulation results, for a cell cluster experiencing a uniform shear stress - cells located at the end of a cluster experience the highest shear compared to cells located within the middle of the row [17]. These variations influence the average shear stress experienced on each of the cells located within the cluster. Of note that the average shear stress experienced by cells positioned in a cluster are suitably approximated by the Reference Shear (purple line) Figure 3b(ii) – which on average is approximately 1.02 times higher. The results indicate that the Reference Shear, is a suitable estimation for the shear experienced by the cells in our experiments, as cell clusters are formed during experiments, rather than several single cells. These contours indicate that the trapezoid microfluidic device is suitable for investigating the response of non-adherent cells to a spatial shear stress gradient, at different shear stress levels, with the purpose of gaining a better understanding of the response of cells subjected to shear stresses that vary in complexity.

3.3 Spatial Shear Stress Gradient: Experimental To examine the influence of a spatial shear stress gradient on intracellular calcium signalling of HEK-293-TRPV4 cells (see Supplementary Information S3 for experimental set-up), we initially immobilised cells using Discontinuous Dielectrophoresis (Methods and Materials) [39]. After cells were immobilised using Discontinuous Dielectrophoresis we were able to

conduct the spatial shear stress gradient experiments. Experiments were conducted at three different Base Reference Shear stresses, being 0.25, 3.3, and 10 dyn/cm2 – which correspond to applied flow rates of 1.5, 20, and 60 µl/min, respectively. For ease of analysis, four regions were designated, with the areas defined across the width of the channel (Figure 2b) – with the left sidewall defined as the origin or 0 µm. The areas were defined as Region 1: 5 to 50 µm, Region 2: 60 to 110 µm, Region 3: 120 to 160 µm, and Region 4: 180 to 240 µm. With the assistance of NIS Elements, each cell was defined as a region of interest (ROI), and if more than 50% of the cell was located within Region 1,2,3, or 4, it was assigned to the appropriate region. See section 2.3 for a detailed explanation on how the fluorescent intensity was quantified for each cell. Calcium influx was quantified by measuring changes in the intensity of the Fluo-4AM (as described in the Materials and Methods Section); which was normalised to the basal level, for the entire experimental duration of 900 s (Figure 4) (see Supplementary Information S4 for experimental fluorescent images). In Figure 4, an increase in intracellular Calcium was observed prior to the application of the shear stimulus – time 0 to 60 s, which we attribute to the presence of spatial shear stress gradient even at low shear stress levels.

In order to understand the influence of spatial shear stress gradient on the characteristics of Calcium influx, we compared four parameters (Figure 5a): (1) maximum fold increase of intracellular calcium level ([Ca2+]i); (2) percentage of activated cells; and the (3) cellular (time in which it takes the cell to reach the activation threshold) and (4) peak response times. Viability was 96 ± 4%; with no significant difference in viability observed between applied shear stress and between the four regions (Supplementary Information S4). Additionally, we measured the responses of non-transfected HEK-293 cells under the highest shear stress of 35 dyn/cm2, which we treated as our negative control, and produced a negligible response. No statistically significant difference was observed between the different applied Base Reference Shear, for the maximum fold increase of [Ca2+]i, or between each of the regions (Figure 5b). However, slight differences were observed, for example, between Region 3 and 4 for an applied Base Reference Shear of 10 dyn/cm2, an increase of 0.25 ± 0.06 (P>0.01, N=3) was observed for [Ca2+]i. Furthermore, other differences that can be observed are primarily in comparison to an applied shear stress of 35 dyn/cm2, which indicates that the intracellular calcium signalling response is more sensitive at higher shear stress [39].

For the percentage of activated cells, with an increase in the applied Base Reference Shear an increase in the percentage of the activated cells was observed (Figure 5c). However, with increasing shear stress no significant difference is observed between the percentage of activated cells, with the percentage of activated cells “plateauing” – as can be observed in Figure 5c. For example, for an applied Base Reference Shear of 0.25 dyn/cm2, the percentage of cell activation increased from Region 1 (0.25 dyn/cm2) through to Region 4 (0.87 dyn/cm2), by 43 ± 12% (P>0.01, N=3) fold – however, no significant difference is observed with for Regions 2 and 3 with Region 4. For both the cellular and peak response times, a reduction was observed for Base Reference Shear of 0.25 and 3.3 dyn/cm2 across Region 1 to 4; however, for the Base Reference Shear of 10 dyn/cm2 no clear reduction was observed across the regions. For example, for a Base Reference Shear of 3.3 dyn/cm2, the cellular response time decreased by 450 ± 66 s, (P<0.001, N=3), from Region 1 (3.3 dyn/cm2) to Region 3 (9.6 dyn/cm2) (Figure 5d). Furthermore, for a Base Reference Shear of 3.3 dyn/cm2, the peak response time decreased by 549 ± 53 s, (P<0.001, N=3), from Region 1 (3.3 dyn/cm2) to Region 4 (12 dyn/cm2) (Figure 5e). Of interest is that for Regions 3 with a Base Reference Shear of 3.3 dyn/cm2, the corresponding Reference Shear is 9.6 dyn/cm2 Figure 4c(ii), which is similar to the Base Reference Shear of 10 dyn/cm2 (corresponding to Region 1, Figure 4a(iii)), no significant difference in results for any of the four parameters was observed. Thus, indicating that a spatial shear stress gradient can be compared between different applied flow rates (Base Reference shear), and at different locations along the width of the trapezoid microchannel. We have previously examined the influence of uniform hear stress on intracellular calcium response of cells uniform high shear stress throughout the entire microchannel [17]. We compared the uniform shear stress results of 10 dyn/cm2, against the spatial shear stress gradient results presented in this paper of a similar shear stress level. A spatial shear stress gradient decreased the cellular and peak response times by 667 ± 41 s (P<0.001, N=4) and 622 ± 34 s (P<0.001, N=4) compared to the case of uniform shear stress. However, the maximum fold increase of [Ca2+]i did not improve in the presence of a spatial shear stress gradient, the maximum fold increase was 0.24 ± 0.06 (P<0.05, N=4) lower than in the presence of a uniform shear stress. Consequently, due to decreased response times, and a decrease in the maximum fold increase of [Ca2+]i in the presence of spatial shear stress compared to uniform shear stress, provide an indication that the response of cellular behaviour, is considerably influence by small variations in shear stress. Moreover that

microfluidic platforms provide an opportunity to investigate the influence of shear stress profiles on mechanostransduction – with shear stress profiles that exhibit varying degrees of complexity [27, 64].

3.4 Temporal Shear Stress: Experimental To extend on the investigation of spatial shear stress gradients, temporal shear stress was applied to the Discontinuous Dielectrophoresis immobilised HEK-293-TRPV4 cells, on the trapezoid platform, with the intracellular calcium signalling response observed. In essence, cells were subjected to spatial shear stress gradients, of different Reference Shear over the experimental duration, i.e. temporal shear stress. The temporal shear stress was applied to the cells, with the applied shear stress being varied in a cyclic manner, with each cycle having a duration of 900 s. In the initially 300 s, the Base Reference Shear was set to 0.25 dyn/cm2, essentially the lowest possible shear stress capable on the microfluidic platform; to ensure sufficient consistent flow to the immobilised cells by means of the syringe pump. Secondly, for a period of 300 s, a Base Reference Shear of 3.3 dyn/cm2 was applied, followed by an increase in the Base Reference Shear to 10 dyn/cm2 for a period of 300 s. This cycle was repeated twice more (Figure 6). The cycles are referred to in the manuscript by, first, second, and third cycle, with each cycle containing each of the applied Base Reference Shear. For a Reference Shear of 10 dyn/cm2 and higher, HEK-293 cells had a cellular response time less than 300 s, such that idealistically the Base Reference Shear of 0.25 dyn/cm2 could essentially be a period of minimal shear stress, or “rest” for shear experienced along the cell surface.

The intracellular Calcium increase observed upon application of temporal shear stress was quantified by measuring changes in the intensity of the Fluo-4AM; which was normalised to the basal level, for the entire experimental duration (Figure 6) (see Supplementary Information S5 for experimental fluorescent images). We examined the results at the end of each of cycle, for the experiment duration carried out to that point in time. Two parameters throughout the three cycles were investigated (Figure 7a): (1) the maximum fold increase of [Ca2+]i, and (2) percentage of activated cells; to determine if the presence of temporal shear stress influences the intracellular calcium signalling response. Cell viability was 96 ± 3%;

with no significant difference in viability observed between different applied shear stresses during temporal shear stress experiments (Supplementary Information S5). The temporal shear stress sensitised the response of TRPV4 channels to shear stress, as can be observed through inspecting the maximum fold increase of [Ca2+]i as well as percentage of activated cells. For example, in Region 1 the maximum fold increase of [Ca2+]i increased by 0.18 ± 0.03 (P>0.001, N=4) after the second cycle in comparison to the first cycle (see Supplementary Information S6). In addition, for Region 2 the percentage of activated cells increased by 58 ± 7% (P>0.01, N=4) fold, after the third cycle compared to the first cycle (Figure 7a (i)). Little change in responsiveness was observed between the second and the third cycles, suggesting that the response was already saturated. On closer inspection of the results for Region 4, the percentage of activated cells in response to Reference Shear of 35 dyn/cm2, during the first cycle, by 45 ± 6% (P<0.001, N=3) fold compared to the Reference Shear of 12 dyn/cm2. In the second cycle, cellular responses followed a similar trend, increasing by 25 ± 6% (P>0.001, N=3) fold, however, for the third cycle no difference was observed between the two Reference Shear (12 and 35 dyn/cm2). Furthermore, one may observe in Figure 6, that cells located in Region 4 have a more gradual response, we conjecture this is attributed to that, cells are activated in the second and third cycle, which occurs more in Region 4 (Figure 7a (i)) than the other regions; thus, the overall trend is that the response appears more gradual. In summary, these parameters indicate that temporal shear stress sensitised the response of TRPV4 channels, with observed saturation of the maximum fold increase, and the gradual increase of the percentage of activated cells. In addition, we compared the influence of the presence of the temporal shear stress, compared to the absence of temporal shear stress – experimental durations were comparable, when considering a single cycle of temporal shear stress. Parameters we considered were: (1) the maximum fold increase of [Ca2+]i; (2) percentage of activated cells; and the (3) cellular and (4) peak response times – which are displayed for Regions 2 and 4 in Figure 7b. For the maximum fold increase of [Ca2+]i, a statistically significant difference in Region 4 was observed with the presence of temporal shear stress. The maximum fold increase of [Ca2+]i was higher in the absence of temporal shear stress, by 0.40 ± 0.05 (P<0.001, N=3), and 0.21 ± 0.05 (P<0.001, N=3), in comparison to the first and third cycles in the presence of temporal shear stress, respectively (see Supplementary Information S6). Minimal differences were observed for Regions 1 to 3, whether or not temporal shear stress was

present. A significant difference in the percentage of activated cells was observed between the absence of temporal shear stress and the first cycle of shear stress; however, differences were not significant after the third cycle of temporal shear stress. For example, for Region 2, in the absence of temporal shear stress the percentage of activated cells was higher by 71 ± 8% (P<0.001, N=3) fold, compared to the first cycle. Just considering the first cycle of temporal shear stress, a comparison was undertaken for the cellular and peak response times (Figure 7b (i)). Both the cellular and peak response times were faster for a spatial shear stress gradient in the absence of temporal shear stress. For example for Region 2, in the absence of temporal shear stress, cellular and peak response times were lower by 436 ± 40 s (P<0.001, N=3), and 440 ± 33 s (P<0.001, N=3), respectively. These results, indicate that the presence of temporal shear stress, provided a more controlled intracellular calcium signalling response, due to the slower response, and gradual increase of the percentage of activated cells, in comparison to the absence of temporal shear stress. Thus, providing more evidence that slight changes in the shear stress profile, can causes significant differences in the cellular response characteristics, and requires further attention by researchers [27, 64].

4. Conclusions Here we have demonstrated the effect of spatial shear stress on intracellular calcium levels of HEK-293-TRPV4 cells. The spatial shear stress gradient was achieved through using a specifically designed trapezoid microchannel. The ability to produce a trapezoid microchannel, one with a slopping top was realised with advances in 3D printing technologies. The HEK-293-TRPV4 cells were immobilised using Discontinuous Dielectrophoresis, which allowed for intracellular calcium response analysis to be undertaken at high levels of shear stress. The effect of spatial shear stress gradient was conducted at different levels of shear stress, achieved by applying the appropriate flow rate to the microfluidic platform. With, our experiments revealing that a spatial shear stress gradients effect intracellular calcium signalling of cells, as observed when comparing cellular and peak response times, percentage of activated cells, and maximum fold increase of [Ca2+]i. Additionally experiments to examine the effect of temporal shear stress on calcium signalling. The experiments revealed that in the presence of temporal shear stress the percentage of activated cells was observed to be lower, in comparison to the absence of temporal shear stress. Furthermore, the cellular and

peak response times were slower (i.e. a more gradual response) for HEK-293-TRPV4 cells experiencing temporal shear stress. This suggests that the role of TRPV4 may be underestimated in endothelial cells, which experience high levels of shear stress that vary in the degrees of complexity. We observed that responses to temporal shear stress were sensitised by exposure to a prior cycle of shear stress and then became saturated to a subsequent cycle of exposure. This suggests that the pattern of shear stress exposure is important in eliciting a response in HEK-293-TRPV4 cells. The microfluidic system described facilitated the investigation of intracellular calcium signalling responses of cells that do not normally adhered well to surfaces to a spatial shear stress gradient. This system enables cells to be exposed to different shear stress levels and temporal shear stress configurations. This capability was highlighted through examining the influence of pulsatile shear stress on intracellular calcium signalling. The microfluidic platform can to examine the influence of a range of spatial shear stress gradients, providing more insight the influence of both physiological and pathological cellular microenvironments on the response of cells to shear stress.

Author Biographies: Rebecca Soffe is a Postdoctoral Research Fellow, at the University of Canterbury, New Zealand. She completed her Ph.D. in Electrical and Electronic Engineering in 2016, from RMIT University. Her research interests include, micro-/nano-fabrication, and lab-on-a-chip. Sara Baratchi is a RMIT Vice-Chanellor’s Postdoctoral Research Fellow in the School of Health and Biomedical Sciences, at RMIT University. She obtained her Ph.D. in Neurobiology in 2011, from Deakin University, Australia. Her research interests include, analysis of cellular mechanotransduction using microfluidic technologies. Mahyar Nasabi is a Project Manager, at Australian Research Infrastructure Network (ARIN). He obtained his Ph.D. in Electrical and Electronic Engineering from RMIT University, in 2013. His research interests include, micro-/nano- fabrication techniques, for microfluidics and microelectronics. Shi-Yang Tang is a Vice-Chancellor’s Postdoctoral Research Fellow, at the University of Wollongong, Australia. He obtained his Ph.D. in Electrical and Electronic Engineering from RMIT University, in 2015. His research interests include, microfluidics, and liquid metals. Andreas Boes is a Research Fellow in the School of Engineering, at RMIT University. He recieved his Ph.D. in Electrical and Electronic Engineering from RMIT University, in 2016. His research interests include, nanofabrication, and integrated optics. Peter McIntyre is a Professor in the School of Health and Biomedical Sciences, and also a Deputy Director of the Health Innovations Research Institute, at RMIT University. He obtained his Ph.D. in Biochemistry and Molecular Biology in 1985 from LaTrobe University, Australia. His longstanding research interest is in the mechanisms of chronic pain. Arnan Mitchell is a Distinguished Professor in the School of Engineering, and also a Director of the Micro Nano Research Facility, at RMIT University. He obtained his Ph.D. from RMIT University, in 2000. His research interests range from integrated optics to lab-on-a-chip technologies. Khashayar Khoshmanesh is a Senior Research Fellow in the School of Engineering, at RMIT University. He obtained his Ph.D. in Biomechanical Engineering from Deakin University, Australia, in 2010. His research interests include manipulation of bio-particles in microfluidics under mechanical and electrical forces

5. References [1] E. Berthier, D.J. Beebe, Gradient generation platforms: new directions for an established microfluidic technology, Lab on a Chip, 14(2014) 3241-7. [2] S. Nahavandi, S. Baratchi, R. Soffe, S.-Y. Tang, S. Nahavandi, A. Mitchell, et al., Microfluidic platforms for biomarker analysis, Lab on a Chip, 14(2014) 1496-514. [3] F. Guo, J.B. French, P. Li, H. Zhao, C.Y. Chan, J.R. Fick, et al., Probing cell–cell communication with microfluidic devices, Lab on a Chip, 13(2013) 3152-62. [4] D. Huh, Y.-s. Torisawa, G.A. Hamilton, H.J. Kim, D.E. Ingber, Microengineered physiological biomimicry: Organs-on-Chips, Lab on a Chip, 12(2012) 2156-64. [5] T.-Y. Tu, C.-Y. Chen, D.-S. Jong, A.M. Wo, An integrated electrophysiological and optical approach for ion channel study in a microfluidic system enabling intra- and extracellular solution exchange, Sensors and Actuators B: Chemical, 185(2013) 496-503. [6] D.F. Schaffhauser, O. Andrini, C. Ghezzi, I.C. Forster, A. Franco-Obregón, M. Egli, et al., Microfluidic platform for electrophysiological studies on Xenopus laevis oocytes under varying gravity levels, Lab on a Chip, 11(2011) 3471-8. [7] S. Nahavandi, S.-Y. Tang, S. Baratchi, R. Soffe, S. Nahavandi, K. Kalantar-zadeh, et al., Microfluidic Platforms for the Investigation of Intercellular Signalling Mechanisms, Small, 10(2014) 4810-26. [8] L.Y. Yeo, H.-C. Chang, P.P.Y. Chan, J.R. Friend, Microfluidic Devices for Bioapplications, Small, 7(2011) 12-48. [9] J. Suehiro, M. Shutou, T. Hatano, M. Hara, High sensitive detection of biological cells using dielectrophoretic impedance measurement method combined with electropermeabilization, Sensors and Actuators B: Chemical, 96(2003) 144-51. [10] Y.N. Luo, D.Y. Chen, Y. Zhao, C. Wei, X.T. Zhao, W.T. Yue, et al., A constriction channel based microfluidic system enabling continuous characterization of cellular instantaneous Young's modulus, Sensors and Actuators B: Chemical, 202(2014) 1183-9. [11] M. Rossi, R. Lindken, B.P. Hierck, J. Westerweel, Tapered microfluidic chip for the study of biochemical and mechanical response at subcellular level of endothelial cells to shear flow, Lab on a Chip, 9(2009) 1403-11. [12] I.A. Ges, R.L. Brindley, K.P. Currie, F.J. Baudenbacher, A microfluidic platform for chemical stimulation and real time analysis of catecholamine secretion from neuroendocrine cells, Lab on a Chip, 13(2013) 4663-73. [13] J. Shemesh, I. Jalilian, A. Shi, G.H. Yeoh, M.L.K. Tate, M.E. Warkiani, Flow-induced stress on adherent cells in microfluidic devices, Lab on a Chip, 15(2015) 4114-27. [14] T. Yasukawa, J. Yamada, H. Shiku, F. Mizutani, T. Matsue, Positioning of cells flowing in a fluidic channel by negative dielectrophoresis, Sensors and Actuators B: Chemical, 186(2013) 9-16. [15] W.J. Polacheck, R. Li, S.G. Uzel, R.D. Kamm, Microfluidic platforms for mechanobiology, Lab on a Chip, 13(2013) 2252-67. [16] D.-H. Kim, P.K. Wong, J. Park, A. Levchenko, Y. Sun, Microengineered platforms for cell mechanobiology, Annual Review of Biomedical Engineering, 11(2009) 203-33. [17] R. Soffe, S. Baratchi, S.-Y. Tang, A. Mitchell, P. McIntyre, K. Khoshmanesh, Concurrent shear stress and chemical stimulation of mechano-sensitive cells by discontinuous dielectrophoresis, Biomicrofluidics, 10(2016) 024117. [18] P. Honarmandi, H. Lee, M.J. Lang, R.D. Kamm, A microfluidic system with optical laser tweezers to study mechanotransduction and focal adhesion recruitment, Lab on a chip, 11(2011) 684-94. [19] D. Ingber, Mechanobiology and diseases of mechanotransduction, Annals of Medicine, 35(2003) 564-77. [20] J. García-Añoveros, D.P. Corey, Mechanosensation: touch at the molecular level,

Current Biology, 6(1996) 541-3. [21] G. Scherrer, N. Imamachi, Y.-Q. Cao, C. Contet, F. Mennicken, D. O'Donnell, et al., Dissociation of the opioid receptor mechanisms that control mechanical and heat pain, Cell, 137(2009) 1148-59. [22] P. Delmas, J. Hao, L. Rodat-Despoix, Molecular mechanisms of mechanotransduction in mammalian sensory neurons, Nature Reviews Neuroscience, 12(2011) 139-53. [23] C. Urbich, E. Dernbach, A. Reissner, M. Vasa, A.M. Zeiher, S. Dimmeler, Shear stress– induced endothelial cell migration involves integrin signaling via the fibronectin receptor subunits α5 and β1, Arteriosclerosis, Thrombosis, and Vascular Biology, 22(2002) 69-75. [24] B. Martinac, Mechanosensitive ion channels: molecules of mechanotransduction, Journal of Cell Science, 117(2004) 2449-60. [25] B. Nilius, J. Vriens, J. Prenen, G. Droogmans, T. Voets, TRPV4 calcium entry channel: a paradigm for gating diversity, American Journal of Physiology-Cell Physiology, 286(2004) C195-C205. [26] G.P. Ahern, Transient receptor potential channels and energy homeostasis, Trends in Endocrinology & Metabolism, 24(2013) 554-60. [27] F. Kurth, K. Eyer, A. Franco-Obregón, P.S. Dittrich, A new mechanobiological era: microfluidic pathways to apply and sense forces at the cellular level, Current Opinion in Chemical Biology, 16(2012) 400-8. [28] G. Bao, S. Suresh, Cell and molecular mechanics of biological materials, Nature Materials, 2(2003) 715-25. [29] Kshitiz, D.-H. Kim, D.J. Beebe, A. Levchenko, Micro-and nanoengineering for stem cell biology: the promise with a caution, Trends in Biotechnology, 29(2011) 399-408. [30] W. Liu, N.S. Murcia, Y. Duan, S. Weinbaum, B.K. Yoder, E. Schwiebert, et al., Mechanoregulation of intracellular Ca2+ concentration is attenuated in collecting duct of monocilium-impaired orpk mice, American Journal of Physiology-Renal Physiology, 289(2005) F978-F88. [31] R.M. Nerem, M.J. Levesque, J. Cornhill, Vascular endothelial morphology as an indicator of the pattern of blood flow, Journal of Biomechanical Engineering, 103(1981) 1726. [32] K.A. Barbee, Role of subcellular shear–stress distributions in endothelial cell mechanotransduction, Annals of Biomedical Engineering, 30(2002) 472-82. [33] R. Köhler, J. Hoyer, Role of TRPV4 in the mechanotransduction of shear stress in endothelial cells, TRP Ion Channel Function in Sensory Transduction and Cellular Signaling Cascades, 1st ed.2007. [34] L. Chau, M. Doran, J. Cooper-White, A novel multishear microdevice for studying cell mechanics, Lab on a Chip, 9(2009) 1897-902. [35] P.F. Davies, C.F. Dewey Jr, S.R. Bussolari, E.J. Gordon, M.A. Gimbrone Jr, Influence of hemodynamic forces on vascular endothelial function. In vitro studies of shear stress and pinocytosis in bovine aortic cells, Journal of Clinical Investigation, 73(1984) 1121. [36] L. Wang, Z.-L. Zhang, J. Wdzieczak-Bakala, D.-W. Pang, J. Liu, Y. Chen, Patterning cells and shear flow conditions: convenient observation of endothelial cell remoulding, enhanced production of angiogenesis factors and drug response, Lab on a Chip, 11(2011) 4235-40. [37] L.K. Chin, J.Q. Yu, Y. Fu, T. Yu, A.Q. Liu, K.Q. Luo, Production of reactive oxygen species in endothelial cells under different pulsatile shear stresses and glucose concentrations, Lab on a Chip, 11(2011) 1856-63. [38] R. Booth, S. Noh, H. Kim, A multiple-channel, multiple-assay platform for characterization of full-range shear stress effects on vascular endothelial cells, Lab on a Chip, 14(2014) 1880-90.

[39] R. Soffe, S. Baratchi, S.-Y. Tang, M. Nasabi, P. McIntyre, A. Mitchell, et al., Analysing the calcium signalling of cells under high shear flows using discontinuous dielectrophoresis, Scientific Reports, 5(2015) 11973. [40] Y. Yang, K. Kulangara, J. Sia, L. Wang, K.W. Leong, Engineering of a microfluidic cell culture platform embedded with nanoscale features, Lab on a Chip, 11(2011) 1638-46. [41] E. Gutierrez, B.G. Petrich, S.J. Shattil, M.H. Ginsberg, A. Groisman, A. Kasirer-Friede, Microfluidic devices for studies of shear-dependent platelet adhesion, Lab on a Chip, 8(2008) 1486-95. [42] S. Kou, L. Pan, D. van Noort, G. Meng, X. Wu, H. Sun, et al., A multishear microfluidic device for quantitative analysis of calcium dynamics in osteoblasts, Biochemical and Biophysical Research Communications, 408(2011) 350-5. [43] J.W. Song, W. Gu, N. Futai, K.A. Warner, J.E. Nor, S. Takayama, Computer-controlled microcirculatory support system for endothelial cell culture and shearing, Analytical Chemistry, 77(2005) 3993-9. [44] S. Baratchi, F.J. Tovar-Lopez, K. Khoshmanesh, M.S. Grace, W. Darby, J. Almazi, et al., Examination of the role of transient receptor potential vanilloid type 4 in endothelial responses to shear forces, Biomicrofluidics, 8(2014) 044117(1-13). [45] S. Usami, H.-H. Chen, Y. Zhao, S. Chien, R. Skalak, Design and construction of a linear shear stress flow chamber, Annals of Biomedical Engineering, 21(1993) 77-83. [46] M. Li, D.N. Ku, C.R. Forest, Microfluidic system for simultaneous optical measurement of platelet aggregation at multiple shear rates in whole blood, Lab on a Chip, 12(2012) 135562. [47] D. Huh, G.A. Hamilton, D.E. Ingber, From 3D cell culture to organs-on-chips, Trends in Cell Biology, 21(2011) 745-54. [48] D. Huh, B.D. Matthews, A. Mammoto, M. Montoya-Zavala, H.Y. Hsin, D.E. Ingber, Reconstituting Organ-Level Lung Functions on a Chip, Science, 328(2010) 1662-8. [49] I.K. Zervantonakis, S.K. Hughes-Alford, J.L. Charest, J.S. Condeelis, F.B. Gertler, R.D. Kamm, Three-dimensional microfluidic model for tumor cell intravasation and endothelial barrier function, Proceedings of the National Academy of Sciences, 109(2012) 13515-20. [50] C.S. Simmons, J.Y. Sim, P. Baechtold, A. Gonzalez, C. Chung, N. Borghi, et al., Integrated strain array for cellular mechanobiology studies, Journal of Micromechanics and Microengineering, 21(2011) 54016-25. [51] K. Yamamoto, J. Ando, Endothelial cell and model membranes respond to shear stress by rapidly decreasing the order of their lipid phases, Journal of Cell Science, 126(2013) 1227-34. [52] M.E. Warkiani, G. Guan, K.B. Luan, W.C. Lee, A.A.S. Bhagat, P.K. Chaudhuri, et al., Slanted spiral microfluidics for the ultra-fast, label-free isolation of circulating tumor cells, Lab on a Chip, 14(2014) 128-37. [53] M. Rafeie, J. Zhang, M. Asadnia, W. Li, M.E. Warkiani, Multiplexing slanted spiral microchannels for ultra-fast blood plasma separation, Lab on a Chip, 16(2016) 2791-802. [54] E. Carafoli, Intracellular calcium homeostasis, Annual Review of Biochemistry, 56(1987) 395-433. [55] D.E. Jaalouk, J. Lammerding, Mechanotransduction gone awry, Nature Reviews: Molecular Cell Biology, 10(2009) 63-73. [56] D.P. Poole, S. Amadesi, N.A. Veldhuis, F.C. Abogadie, T. Lieu, W. Darby, et al., Protease-activated receptor 2 (PAR2) protein and transient receptor potential vanilloid 4 (TRPV4) protein coupling is required for sustained inflammatory signaling, Journal of Biological Chemistry, 288(2013) 5790-802. [57] F. Vincent, M. AJ Duncton, TRPV4 agonists and antagonists, Current Topics in Medicinal Chemistry, 11(2011) 2216-26.

[58] B. Sankaran, M. Racic, A. Tona, M.V. Rao, M. Gaitan, S.P. Forry, Dielectrophoretic capture of mammalian cells using transparent indium tin oxide electrodes in microfluidic systems, Electrophoresis, 29(2008) 5047-54. [59] M. Suzuki, T. Yasukawa, H. Shiku, T. Matsue, Negative dielectrophoretic patterning with different cell types, Biosensors and Bioelectronics, 24(2008) 1043-7. [60] G.M. Whitesides, E. Ostuni, S. Takayama, X. Jiang, D.E. Ingber, Soft lithography in biology and biochemistry, Annual Review of Biomedical Engineering, 3(2001) 335-73. [61] W. Xiong, Y.S. Zhou, X.N. He, Y. Gao, M. Mahjouri-Samani, L. Jiang, et al., Simultaneous additive and subtractive three-dimensional nanofabrication using integrated two-photon polymerization and multiphoton ablation, Light: Science and Applications, 1(2012) e6. [62] B.H. Cumpston, S.P. Ananthavel, S. Barlow, D.L. Dyer, J.E. Ehrlich, L.L. Erskine, et al., Two-photon polymerization initiators for three-dimensional optical data storage and microfabrication, Nature, 398(1999) 51-4. [63] R. Soffe, S. Baratchi, S.-Y. Tang, P. McIntyre, A. Mitchell, K. Khoshmanesh, Discontinuous Dielectrophoresis: A Technique for Investigating the Response of Loosely Adherent Cells to High Shear Stress, 9th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2016), SCITEPRESS – Science and Technology Publications, Rome, Italy, 2016, pp. 23-33. [64] J. Wang, J. Heo, S.Z. Hua, Spatially resolved shear distribution in microfluidic chip for studying force transduction mechanisms in cells, Lab on a Chip, 10(2010) 235-9.

Figure 1. Microfluidic Platform used to Investigate Spatial Shear Stress Gradients. (a) Photograph of the assembled microfluidic platform; (b) Schematic representation of the microfluidic platform, emphasizing the trapezoid microfluidic channel and the immobilisation of cells; and (c) Overview of the (i) device setup, (ii) ITO microelectrode dimensions, and (iii) trapezoid microchannel cross-section photograph.

Figure 2. Trapezoid Microchannel Overview. (a) Dimensioned trapezoid microchannel with designated Regions 1 to 4. (b) Shear stress profile across the trapezoid microchannel for a flow rate of 60 µl/min. Taking note, four regions were defined Regions 1 to 4, which correspond to the shear stress ratios of 1, 2, 2.9, and 3.5, respectively. (c) Contours of velocity magnitude for a flow rate of 60 µl/min. (d) Trapping efficiency, with the applied flow rate increased every 300 s, up to 100 µl/min. (e) Spatial shear stress gradient experimental procedure, displayed for investigating the intracellular calcium signalling response of HEK-293-TPV4 cells.

Figure 3. Contours of Shear Stress over Immobilised HEK-293 Cells Obtained via Numerical Simulations, Corresponding to an Applied Flow Rate of 60 µl/min. (a) Contours of shear stress, for: (i) five equidistant immobilised cells spanning the width of the microchannel, and (ii) an immobilised cell cluster containing nine cells located in Region 4. (b) The average shear stress experienced on the cell surface, for each of the cells in both the aforementioned configurations. Numbers are designated to each of the cells within each of the two configurations, from left to right, are numbered 1 to 5 and 9. The purple line in each of the graphs represents Reference Shear experienced at that location in the microchannel, in the absence of cells. With the average shear stress for each of the individual cells denoted by the black circle. Regions 1 to 4, are coloured as follows: blue, orange, red, and green, respectively – Regions 1 to 4 are displayed for (i) from left to right, and Regions 3 and 4 are displayed for (ii). Corresponding spatial shear stress graphs are presented in Supplementary Information S2.

Figure 4. Intracellular Calcium Signalling Responses of HEK-293-TRPV4 Cells to a Spatial Shear Stress Gradient. (a-d) Normalised mean and single cell intensity profiles for 30 to 50 cells, for each of the four regions, at three different shear stress values (denoted above each of the graphs). (i-iii) The three applied flow rates of 1.5, 20, and 60 µl/min which correspond to Base Reference Shear values of 0.25, 3.3, and 10 dyn/cm2, respectively (see Supplementary Information S4 for experimental fluorescent images).

Figure 5. Response Characteristics for HEK-293-TRPV4 Cells Experiencing Spatial Shear Stress Gradient. (a) Normalised mean and single cell intensity profile, showing different response parameters. Comparison of the response characteristics: (b) maximum fold increase of [Ca2+]i; (c) percentage of activated cells; and (d) cellular and (e) peak response times (see Supplementary Information S5 for experimental fluorescent images). Data are presented as the mean ± SEM, *P<0.05, **P<0.01, and ***P<0.001.

Figure 6. Intracellular Calcium Signalling Response of HEK-293-TRPV4 Cells Experiencing Temporal Shear Stress. (a-d) Normalised mean and single cell intensity profiles for 30 to 50 cells, for each of the four regions. The applied shear stress (dyn/cm2) is denoted on the graphs, for the period of time between the dashed lines (see Supplementary Information S5 for experimental fluorescent images).

Figure 7. Response Characteristics for HEK-293-TRPV4 Cells Experiencing Temporal Shear Stress. (a) Comparison of the response characteristics for temporal shear stress over the three cycles for the (i) percentage of activated cells across the three cycles; and (ii) percentage of activated cells for region four, for closer inspection of the influence of the different applied Reference Shear. (b) The comparison of response characteristics between spatial shear stress gradients in the absence and presence of temporal shear stress for: (i) percentage of activated cells; and (ii) peak response time (see Supplementary Information S5 and Supplementary Information S6 for experimental fluorescent images, and maximum fold increase of [Ca2+]i and cellular response time graphs). Data are presented as the mean ± SEM, *P<0.05, **P<0.01, and ***P<0.001.