1. Three-dimensional microfabricated bioreactor and closed-loop control system Alex Makowski Michael Hwang Jenny Lu Dr. John Wikswo

2. Ultimate Project Goals Computer modulated growth of tissue structures within microfluidic devices. Possible uses include drug testing and dose determination, bypassing several stages of FDA trials.

3. Problem Statement- Design progress is blocked on three distinct fronts: Sensors are required to provide computer with necessary information. Bioreactor design limits the quantity and quality of cell morphology within the device. Previously used cells (primary human fibroblasts) do not easily form tissue-like structures.

4. Subsequent Requirements: Choose or design an appropriate sensor for pH measurements (most needed to determine cell metabolism and health). Choose a cell line that will exhibit observable morphological change under successful conditions. Redesign the bioreactor to incorporate new cell line and maximize efficiency of pH sensors.

5. Resultant Main Chip Design Channel Layer - PDMS Glass Slide Plexiglas Matrigel Layer with Cells (8mm diam.) Access Ports Top View: Side View: 500 micron

6. Piecewise Breakdown- Bottom Bottom Layer – Thickness: 500 micron- 1mm – Reasoning: Desired by McCawley lab to guarantee flow and cell health Dimensions 1.8 inch square PDMS layer with 8mm radius centered well 1.9x1.9 square glass slide about 500 micron thickness (dimensions needed for already designed clamp) Glass Slide Matrigel Layer with Cells (8mm diam.) 500 micron -1mm 8mm 1.8 in. 1.9 in.

7. Piecewise Breakdown- Top Filter layer – Dimensions: ~10 micron thickness x 1.9 inches square (3 micron pores) Channel Layer – Dimensions (1.9 inch square 3-5mm thickness) – Channels: 60-100 microns deep 6mm across (15 micron wide channels) Plexiglas Layer (machine shop made) – Dimensions (1.9 inch square 5-7mm thick) Port Assemblies-Upchurch Scientific – Interface with Tygon tubing (500micron) Channel Layer - PDMS Plexiglas Note: Thickness of upper layers is due to pressure generated by fluidic system. Layer thickness prevents deformation.

8. Chip Design Notes Filter prevents solid waste from leaving bioreactor and clogging microchannels. Dimensions have been optimized for fluidic perfusion of cell layer and minimal dead volume

9. Cell Line: MCF10A Human breast epithelial cells (MCF10A) Growth- 2 to 3 days to grow to ~90% confluence in culture dish. Acinar morphology-3D hollow cell ball.

10. Acinar morphology of MCF10A cells Matrigel experiment shows it takes 20 days to form into acinar morphology. (Debnath and Brugge 2005)

11. Investigating cell concentration Matrigel experiment on chamber slide with different cell concentrations: 5000 cells/chamber 1000 cells/chamber 500 cells/chamber 250 cells/chamber Illustration courtesy of Cassio Lynm, JAMA

12. Getting the right concentration Show what concentration gives the best images under the confocal microscope. Correlate the optimal concentration in chamber slide with what we will put in the bioreactor.

13. Cell culturing dish Growth Media PDMS layer Glass slide Bottom Matrigel layer Matrigel and cells layer Side View Top View with MCF10A cellsTop View without cells LAYOUT OF EXPERIMENT USING PDMS LAYER WITH CELLS

14. Protocol-Preparation Step Aspirate media in cell culturing dish Wash the cells with 2mL of Trypsin/EDTA then aspirate Add 2mL of Trypsin/EDTA then 8mL of growth media Put in 15mL conical tube Centrifuge the conical tube at 1000RPM, 4°C, and for 5 min. Aspirate the growth media Resuspend the cell pellet Add 2mL of Ca,Mg-free media Do a 1:4 dilution with Trypan blue dye Count the cells in hemacytometer Goal in prep. Step To get the right number of cells for bioreactors and chamber slide wells.

15. Protocol-Matrigel experiment Pre-coat the chamber slide with 30μL of Matrigel, bioreactors with 9 μL of Matrigel. Let the Matrigel polymerize in incubator for 15 min. Mix 5000 cells with 70 μL of Matrigel for each top layer of chamber slide, and 1500 cells with 21 μL of Matrigel for top layer of bioreactor. Let the second layer Matrigel polymerize for 1 hr. Add 400 μL of growth media in chamber slide and 10mL of growth media in culture dish where bioreactor sits. Change growth media every 2-3 days. Goal in Matrigel experiment To make sure the cells can live in PDMS layer of bioreactor for 20 weeks without the added perfusion layer and filter.

16. Cell culturing dish Growth Media PDMS layer Glass slide Bottom Matrigel layer Matrigel and cells layer Side View Top View with MCF10A cellsTop View without cells LAYOUT OF EXPERIMENT USING PDMS LAYER WITH CELLS

17. Top View Magnified Side View Growth Media Matrigel and cells layer Bottom Matrigel layer PARALLEL EXPERIMENT WITH CHAMBER SLIDE

18. Problems and Solutions P1: Matrigel migrated from the well of bioreactor to the surface of PDMS layer. S1: Let the second layer of Matrigel with cells sit for 1 hour instead of 20 minutes to polymerize. P2: Some cells migrated to the surface of PDMS layer and some even to the cell culturing dish. S2: The final bioreactor has a filter that prevents the cells from migrating out of the well.

19. Diagram of the Entire Fluidic System Cell Media for Infusion into Bioreactor and for Differential pH Reference Fresh Media Acidified Media Bioreactor Iridium Oxide pH-Sensing Electrode Iridium Oxide Quasi-Reference Electrode Cavro® XLP 6000 Syringe Pump pH 8 Calibration Solution pH 6 Calibration Solution

20. Valves allow selective input/output through one of three ports Fresh media/solution aspirated through left port Dispense media through one of two other ports

21. Why do we need a reference electrode? Cell Media for Infusion into Bioreactor and for Differential pH Reference Fresh Media Acidified Media Bioreactor Iridium Oxide pH-Sensing Electrode Iridium Oxide Quasi-Reference Electrode Cavro® XLP 6000 Syringe Pump pH 6 Calibration Solution pH 8 Calibration Solution

22. Junction Potentials Exist at Heterogeneous Metal/Metal, Liquid/Liquid, or Liquid/Metal Interfaces For example, at a metal/liquid interface metal cations enter solution while electrons remain restricted to the metal phase. Hence, a potential develops. Charges are confined to interfaces in electrostatic conditions. Junction potentials vary among interfaces between materials of different compositions and are often difficult to determine. + + + + + + + _ _ Metal Solution _ _ _ _ _

23. Absolute Reference Electrodes Remain at a Known, Constant Potential Relative to Sample Solutions Examples include the Ag/AgCl liquid junction reference electrode or Hg/Hg 2 Cl 2 (saturated calomel electrode). Ag/AgCl or Hg/Hg 2 Cl 2 wire remains at a constant potential relative to an internal saturated KCl solution whose composition does not change. Liquid junction potential between internal electrolyte and sample solution at the high resistance porous frit is very small and invariant – High concentration of salt in internal electrolyte dominates the exchange current across the porous frit. – Similar mobility of the cation and anion of the salt causes both charges to cross the frit at an equal rate and thus create virtually no potential difference at the interface.

24. A Quasi-Reference Electrode is Suitable for Our Measurements The quasi-reference iridium oxide exhibits a relatively constant potential relative to sample solutions for short time scales. – The solution bathing the quasi-reference electrode—and therefore the electrode potential—does not change. – If the ionic composition of sample solutions do not vary too much, the liquid junction potential between the solution immersing the quasi-reference and the sample solutions will also be nearly constant. An absolute reference electrode that remains at a constant potential for longer periods of time is superfluous. – The quasi-reference electrode is identical to the pH-sensing electrode and therefore drifts at the same rate. – Recalibration simultaneously compensates for the drift of both electrodes. – Hence, for example, measuring a series of solutions of known pH followed by a solution of unknown pH can yield the unknown pH. A 125-μm-diameter iridium oxide quasi-reference electrode is cheaper (produced in lab) and easier to integrate into the fluidic system.

25. Optimal location of quasi-reference electrode Shorter distance between iridium oxide electrodes: provides more equal noise coupling to both electrodes reduces inter-electrode impedance reduces number of liquid junction potentials between the sensors vs.

26. Further Protection against Noise Cell Media for Infusion into Bioreactor and for Differential pH Reference Fresh Media Acidified Media Bioreactor Iridium Oxide pH-Sensing Electrode Iridium Oxide Quasi-Reference Electrode Cavro® XLP 6000 Syringe Pump pH 8 Calibration Solution Faraday Cage

27. Iridium Oxide Electrode Production Summary of Steps – Prepare electrodeposition solution by heating a solution (pH 10.5) of IrCl 4, K 2 CO 4, and K 2 CO 3 at 90°C for 17 minutes. – Etch 125-μm-diameter Ti wires in 70% H 2 SO 4 at 80°C for 2 minutes. – Apply.5 mA/cm 2 current density to 4-mm immersed tip of etched Ti wire for 2-4 minutes using a two- electrode galvanostat (with a Pt disk counter- electrode). Results – Deposition time affects color and thickness of deposit (which in turn affects slope and lifetime). – Etching with hot sulfuric acid improves uniformity of deposition. 4 mm

28. Iridium Oxide Electrode Production Insulation – Submerge the entire bare portion of Ti wire in red insulating varnish and allow to air dry for 20 min. – Dissolve away varnish from end of wire with acetone. Diagram (right): – Iridium oxide deposit exhibits linear potential response to pH – Red insulating varnish prevents a junction potential between the uncoated Ti wire and solution from confounding the response of the iridium oxide – Bare Ti wire at end allows for electrical connection of electrode to measurement system 4 mm Red Insulating Varnish Bare Titanium Wire Iridium Oxide Deposit

29. Iridium Oxide Electrode Potential (relative to Ag/AgCl ref.) pH Navy Blue Deposit Voltage (mV) Two-point Slope Calibration (mV/pH) 4399-71 5328-70 6258-75 7183-77.5 7.4152-66.6667 8112-61 951--- pH Gold Deposit Voltage (mV) Two-point Slope Cal. (mV/pH) 4190-61 5129-61 668-65.5 72.5-66.25 7.4-24-55 8-57-49 9-106---

30. Integration of Electrodes in Fluidic Lines Tygon Tubing PDMS Glob Fused Silica Capillary Tube Iridium Oxide Electrode Tygon Tubing Flow Tygon Tubing: 500 μm inner diam. Electrode: 125 μm diam.

31. Motivation for Using an Instrumentation Amplifier Superior common mode rejection of noise that couples equally to both electrodes (especially from syringe pumps) High input impedance (G Ω vs. kΩ) Silent, high-gain pre-stage to overcome noisy DAQ device and attain desired limit-of- detection (.01 pH)

32. Instrumentation Amplifier Circuit and Box (Completed) 8-DIP AD621 Precision Instrumentation Amplifier Prototyping board 12 V batteries (x2).1 μF bypass capacitors (x2) Banana jacks Power switch Shielding aluminum box

33. Note: Actual circuit varies slightly from diagram, but connectivity is the same. DAQ 12 V.1 μF Analog GND (Pin 1) Analog Input Channel 0 (Pin 2) Overlay -IN +IN.1 μF Bias Current Return Path Virtual Ground

34. Return Path for Bias Current IA + - Platinum Wire

35. Diagram of Complete System IA + - DAQ Device

36. Unidirectional vs. Reversible Flow Example Schematic of Reversible Flow System—19 valve minimum

37. Illustration of Stop-Flow Operation 1. Flow 2. Stop and allow media acidification 4. Stop and Measure 3. Flow

38. Stop-flow Over Continuous Flow More equal nutrient/metabolite exchange Easier to conceptualize/model exchange Less delay in acquiring current pH measurement If response time of sensor is long, allows measurement of endpoint pH rather than moving average of pH signal (calculation of acidification rate better with former)

39. Automated Sensor Characterization RS-485 communication lines soldered. Power lines remain to be soldered for four more pumps. LabVIEW program for repetitive, long-term sensor characterization (to ensure that slope, sensitivity, and response time do not deteriorate over the course of a few weeks) not yet begun. IA + - DAQ Device