Testing Cellular Communication Pathways
As a proof of concept for our novel therapeutic discovery hardware device, APUS, we need to show two cells communicating with a measurable output. If we could see that cells could communicate with each other in our device, then APUS can effectively house and maintain a synthetic pathway. We chose a 2-strain pathway that communicated with the quorum signals, C4HSL and 3-OHC14HSL (1). Either cell can act as the sender and receiver, so we have two potential pathways to test. As the pathway that has CFP acting as the receiver is less complex, that is the first pathway we planned to measure (Fig. 1a). We will be able to measure the success of this pathway by the receiver’s fluorescence, since it can only be triggered by the sender’s output.
Figure 1: Two 2-pathway options within the CFP/YFP synthetic pathway. (A) Pathway 1 involves the CFP cell acting as the receiver. When activated with IPTG, the receiver will be on, constitutively fluorescing CFP. Only when the YFP sender cell produces 3-OHC14HSL will the sender cell turn off. (B) Pathway 2 involves the YFP cell acting as the receiver. The CFP sender cell will produce C4HSL, triggering the receiver cell to produce YFP fluorescence oscillations.
Under the guidance of the DAMP laboratory, I successfully transformed plasmids pC165 (CFP) and pC210 (YFP) into E. coli (Image 1). This required miniprep, creating competent cells, and transformation. When performing initial plate reader and APUS experiments with the two cells, CFP remained fluorescing, indicating no communication. We tested different growing conditions, switching media (LB to M9), trying a more reliable incubator, and using a new batch of antibiotic. There was still no difference in communication. At this point, we returned to the paper from which we found our cells. After looking at the bacteria more in depth, we realized the researchers had knocked out native genes: LacI, araC, sdiA. If these genes had remained in the native genome, there would have been communication. Instead, we had to reintroduce the native genes by transforming both YFP sender and CFP receiver cells with an inverter plasmid: pC239 (Figure 2)
Image 1: The plates of the newly transformed YFP and CFP cells. The YFP cell is E.coli transformed with the pC210 plasmid, selected with kanamycin resistance. The CFP cell is E.coli transformed with the pC165 plasmid, also selected with kanamycin resistance.
Figure 2: The design-build-test-learn cycle of choosing a synthetic biological pathway. We found a pathway that represents cell communication via fluorescence.
Back to Top
Dual-Syringe Continuous Pumping Mechanism (DSCPM)
The design-build-test-learn cycle of fine tuning our continuous pumping mechanism. With guidance from Harvard Apparatus, We found a way to minimize the disruption of flow laminarity at very small flow rates when using solenoid pinch valves.
Designing the DSCPM: Meeting Our Objectives
In the very beginning of the design process of APUS, we knew that we would need to make the right choice when it came to our fluidics accessories: in particular, the pumps. The PDMS chips of APUS are designed with specific dimensions that facilitate of exponential cell growth and proliferation, and the optimal flow rate of media through the PDMS chips is 0.5 - 20 µL/min. In addition to very small flow rates, APUS’s pumps must be able to pump continuously from a reservoir, since one of the defining objectives of APUS is to be able to fully automate days long experiments.
High-resolution flow (able to achieve laminarity at very small flow rates) is generally achieved with traditional syringe pumps. The downside is that syringe pumps are not continuous and need to be refilled once the syringe is depleted, and tend to be very expensive. Continuous flow is usually achieved with peristaltic pumps, however achieving very small flow rates with peristalsis requires mechanical precision that is very difficult to acheive.
The Dual-Syringe Continuous Pumping Mechanism (DSCPM) is designed as a combination of existing pumping mechanisms and electronics principles: two syringes dispense and refill inversely to one another, producing a (square wave) "alternating current" (AC) flow. This flow is then rectified into forward flow. The DSCPM’s ability to automate flow rectification means that it can pump continuously and for a theoretically unlimited time.
Designing the DSCPM: Choosing a Flow Rectification Method
There were two possibilities when it came to the means of flow rectification: check valves, or solenoid operated pinch valves. Check valves are the fluidic equivalent of a diode, and can theoretically rectify flow passively, which would be optimal, if not for the minimum cracking pressure of the valve, which is the minimum pressure required to "crack", or open the valve in the forward direction. Check valves can rectify flawlessly at most flow rates, however the minimum cracking pressure of the valve can take some time to be met at very small flow rates (µL/min scale). When testing check valves with a cracking pressure of 0.5psi (one of the smallest cracking pressures that is market available) it was observed that they would sometimes take multiple minutes to open, and then a bolus of fluid would be dispensed. This is extremely damaging to the laminarity of the flow, which is why the only viable rectification method is manual/active rectification by solenoid operated valves. The electronic equivalent to this mechanism would be to manually rectify flow with switches.
Building the DSCPM
The DSCPM prototype was comprised of a High-Torque Servo Motor, 10µL syringes, 2 Three-Way Solenoid Pinch Valves, and three 3D Printed parts (printed with PLA filament on a Creality Ender).
Additional views of our pumping system (DSCPM) in AutoCAD.
The initial DSCPM Prototype was a proof of concept for flow rectification and used 1mL syringes, so the flow rates demonstrated were much higher than the flow rates we would use in APUS. When we fitted the DSCPM with their final syringes (10µL), there was a logical fallacy that became an emergent issue at very small flow rates. Three way solenoid valves work by pinching one branch of tube, and allowing flow through the other. When the solenoid switches states, there is an instant, when the solenoid is mid push, that there is freeflow allowed between the three branches of the junction of tube. At large flow rates, any flow that occurs in this instant is negligible, however at very small flow rates (eg. 10µL/min), the instant of free flow can have a proportionally larger effect on the laminarity of the flow. For example, if fluid is being pushed upward in altitude at a very small flow rate (eg. 10µL/min), almost all of the progress made in one sweep of the motor is lost on the switch of the solenoid valves due to gravity.
Improving DSCPM - What we’ve learned
There are two possible new valve configurations to solve the freeflow issue, which are explained in the two waveforms. The first configuration (The DSCPM 2-3/1-2) adds a two-way (on/off) solenoid valve (Valve C) that is placed in series with the previous solenoid valves (Valves A and B) and pulses off on the switch of A and B, so that in the instant when A and B would allow freeflow, C prevents any flow. Valves A and B are direct inverses of one another, and can theoretically be minimized to one valve in the final product, however the off state of a three-way valve is maintained by a spring inside the solenoid. The spring is not strong enough to fully pinch two of the tubes that we used in APUS, so two three-way valves are needed in the final product. With less stiff tubing, the valves can theoretically be minimized to one valve.
The second configuration (The DSCPM 2-2) uses two two-way solenoids with staggered switching to prevent any free flow, since implementing rectification logic with two-way solenoids will only allow freeflow if they switch at the same time.
While the DSCPM 2-2 is the monetarily optimal choice since it only requires two solenoids, it is less optimal in terms of laminarity. The issue that is introduced with the two new configurations is that when a valve pinches, the fluid in the pinched location is displaced. In the original configuration, the solenoids switching at the same time meant that the fluid was just displaced into the segment of tube that was unpinched. In both of these configurations, the displaced fluid has nowhere to go but forward, so this “unmatched” volume propagates forward. With the DSCPM 2-2, there are two cross sections of “unmatched” volume, however the DSCPM 2-3/1-2 only has one cross section. This means that theoretically, the pressure spikes that occur at the solenoid switch will be twice as big with the DSCPM 2-2 as the spikes with the DSCPM 2-3/1-2. Because of this, despite the extra $100 or so, the DSCPM 2-3/1-2 was the final configuration used in APUS due to its optimal laminarity.
The CAD design for the Dual-Syringe Continuous Pumping Mechanism (DSCPM) 2-3/1-2 featuring solenoid pinch valves.
Performance Metrics- Measuring Flow Laminarity and its Affect on the Cells
To quantify the disruption of laminarity due to the pinching of Valve C, pressure was measured versus time at different speeds with a Honeywell ABP2 Series Differential Pressure Sensor. A Harvard Apparatus PhD 2000 Syringe Pump was used as the control.
In the above figures, the solenoid clicks are visible on the DSCPM Pressure vs. Time plots.
From the data, it is clear that the laminarity of the DSCPM is not as consistent as the laminarity of the Harvard Apparatus Pumps, but with the next metric shown, we show that the cells are not affected by the pressure spikes whatsoever.
The figure below shows a video of cells in a monolayer chamber (MC) during a pressure spike. A MC in the first channel is shown, as well as an MC from the second one. Early in APUS experimentation, it was observed that sometimes, when there was a problematic pressure gradient in our fluidic system caused by an obstruction, the cells in the first channel were “jolted” during the pressure spikes, while the second channel MCs were “protected”. We hypothesized that this protection may be due to attenuation from the turns in the channel of the PDMS chip. In between now and then, we have added a bubble trap into our fluidic system (bubbles cause very problematic pressure gradients), and we have gotten more skilled at the fluidics assembly required for our experiment protocol.
In the figures shown, which are from our most recent experiment (as of Oct 11th) there is no discernable disturbance in either MC shown.
Back to Top
PDMS Chips
PDMS Chip Leakage
The largest challenge encountered when housing cells in PDMS chips is the issue of leakage at the port connections. This problem is clearly illustrated in the figure below, where the stainless steel needle connections inserted into the PDMS chips do not firmly adhere in place due to the flexible nature of PDMS. Leakage typically occurs at a rate of 50 µL/min and can even manifest at rates as low as 1.5 µL/min. This leakage results in the contamination of the environment by bacteria and media, leading to inaccurate results during data collection.
Furthermore, beyond the initial hurdle of loading cells into a PDMS chip, the PDMS chip must undergo centrifugation at 1328 rpm to securely trap the cells within chambers where they can grow. Therefore, achieving the concept of zero leakage, which is uncommon both in academia and industry, is imperative for the commercial success of APUS.
We went into solving this problem determined to find a solution that can withstand at least a high flow rate of 500 µL/min, which is ten times the maximum flow rate we will use when running experiments on the PDMS chip.
Cycle 1: Instant Adhesive
During the initial cycle, we applied instant adhesive glue to secure the 1/32’’ OD needles that were vertically inserted into the ports. While this approach provided more robust port connections, a significant drawback emerged when attempting to remove the tubing from the PDMS chips for centrifugation. The needles frequently dislodged, and the adhesive failed to adhere effectively to the PDMS chip surface. This prompted the realization that adhesive alone would not suffice for permanently securing the connections within the PDMS chip.
Instead, it was determined that applying a thin layer of PDMS after inserting the needles into the ports might prevent them from becoming dislodged, offering a more reliable solution for cementing the connections within the PDMS chip.
Cycle 2: 1:9 Curing Agent to PDMS Ratio and 1/32” OD Needles
In the second cycle, we sealed the ports using 1:9 ratio of curing agent to PDMS , and we vertically inserted 1/32” OD needles to the top of the PDMS chip. This sealing method yielded promising results, supporting a sustained flow rate of up to 500 µL/min. However, a significant challenge arose when attempting to detach the tubing from the needle connections after centrifugation, as the connections became disengaged from the ports.
Cycle 3: 1:9 Curing Agent to PDMS Ratio and 1/16” OD Needles
In the third cycle, we once again sealed the ports using a 1:9 curing agent to PDMS ratio, and we vertically inserted straight 1/16” OD needles into the top of the PDMS. In contrast to the 1/32” OD needle connections, these could only sustain a flow rate of up to 100 µL/min. We encountered the same issue of connections coming loose after centrifugation. If anything, the wider and thicker needles proved more challenging to insert into the PDMS chips before sealing and were prone to deformation from their cylindrical shape.
Hence, the preferred solution appeared to revert to using 1/32” OD needles and explore alternative methods for securely affixing the connections into the PDMS chips.
Cycle 4: 2:8 Curing Agent to PDMS Ratio and 90º 1/32” OD Needles
In the fourth and final cycle, we sealed the ports using a 2:8 curing agent to PDMS ratio, resulting in a harder PDMS texture after incubation. Building on our previous findings, we continued to use the 1/32” OD needles for insertion into the ports on top of the PDMS chip prior to sealing. However, instead of inserting the needles vertically, we bent them at a 90-degree angle to prevent the ports from coming loose when removing tubing or during centrifugation.
This innovative approach also yielded promising results, with the connections supporting a sustained flow rate of up to 2000 µL/min in initial tests! The 2:8 ratio held up well, effectively addressing the issue of leakage as anticipated.
To test the performance of PDMS chips with the new sealing method pre and post centrifugation, we tested PDMS chips’ maximum flow rate they can handle before they leak. Before being centrifuged, chips could withstand high flow rates up to 10,000 µL/min. This flow rate is not only exceedingly higher than the PDMS chips will ever need to endure, but is unheard of in the academic and industrial community.
Post-centrifugation, the maximum flow rate decreases; however, not significantly since the mean of the data still centered around the range of 2500 - 3500 µL/min. With these metrics, we demonstrated that PMDS chips with monolayer cell traps is indeed an innovative way to house cells independently, while allowing for cell communication between various strains of bacteria.
Back to Top
Motherboard
Cycle 1: Initial Motherboard Design
When we first designed the motherboard, our aim was to make it compatible with three chips, each featuring two channels for housing different cells, allowing for two separate experiments to run concurrently. Here are the initial design requirements:
- Running Media to Clean Chips and Main Pathways
- Running 10% Bleach/ DI Water through main pathways (not PDMS Chips)
- Running IPTG + Media through BOTH experiments
- Running IPTG + Media through ONE experiment
- Running IPTG + Media through and SWITCHING experiments (one or both)
- Running IPTG + Media through and disposing parts of it
- Running glycerol at the end of the experiment before PDMS chip storage
Using a 16-solenoid system, we constructed the following design. The blue layer represents the flow path through which the liquid travels, while the red layer represents the control layer responsible for opening and closing valves to regulate flow within the flow layer.
While the fabrication process, involving laser cutting and plasma curing, proceeded without any issues, we observed a problem when multiple channels were connected to a single flow port. This configuration led to backflow and contamination in other channels. To address this issue, we developed a motherboard with spread-out ports.
Cycle 2
In cycle 2, we designed a motherboard connected to a disposal system. This setup enabled experiments to work seamlessly with 1-2 PDMS chips. It allowed fluid exiting the first PDMS chip to be directed straight to disposal without the need to change tubing. Essentially, our goal was to emphasize the plug-and-play capability of the motherboard, enabling interchangeable experiments without the necessity of tubing replacement.
Despite spacing the ports further apart, they were still tightly packed within the limited space on the acrylic surface where they were intended to fit. As depicted in the image above, this cramped arrangement hindered the adhesive from securely adhering to the silicone, resulting in numerous interlayer bubbles and leaks originating from the motherboard. Eventually, liquid began to seep out from the sides of the motherboard.
Cycle 3 : Testing for Safe Effective Sterilization Method
In cycle 3, we conducted experiments with 2-cell bacterial pathways while retaining the option for 2-channel experiments. This time, the adhesive help up better, with no visible gaps or bubbles in the silicone layer, ensuring fluid flow without backflow into other channels.
Our next step involved testing various cleaning solutions for the motherboard. We observed that Ethanol damaged the acrylic, while Isopropyl Alcohol degraded the adhesive. As depicted in the figure above, both these solutions caused harm to the motherboard's layers. We proceeded to the next testing stage in search of a more suitable sterilization solution.
Cycle 4
In cycle 4, we personalized the Motherboard for the 2-strain synthetic pathways we were conducting with APUS. This involved the use of two PDMS chips, with media flowing through only one channel in each chip. This simplified the Motherboard compared to the design in cycle 2.
Additionally, we sterilized the Motherboard using a 10% Bleach solution, which effectively eliminated bacteria while preserving the adhesive and acrylic materials. Subsequently, we integrated the Motherboard with the PDMS chips and the bench-top microscope experiments we were conducting. While fluid flowed through the Motherboard and PDMS chips, we encountered an issue with the formation of bubbles. Small bubbles were effectively removed by the bubble trap, but larger bubbles produced by the Motherboard caused the collapse of the monolayer chambers housing cells, destroying the cells within the PDMS chips.
Cycle 5 : Eliminating Bubble Formation
Cycle 5 focused on resolving issues related to the Motherboard's functionality to eliminate potential bubbles. To address this, we constructed smaller pocket-sized motherboards, each utilizing distinct sealing methods. Among these methods, it was the o-rings that successfully established airtight connections at the ports, effectively preventing the entrapment of air inside the Motherboard and the formation of bubbles. The only remaining task was the synthesis of the final Motherboard, ready to be employed for the synthetic pathway and cells we experimented with.
Cycle 6 : Final APUS Motherboard
In cycle 6, we designed and fabricated the APUS Motherboard, incorporating the o-ring method to seal ports and prevent bubble formation. Furthermore, the screws on the Motherboard ensure a tight adhesion between all layers. The design closely resembles that of cycle 4, with the addition of a disposal port to enhance its plug-and-play capability.
Back to Top
Environmental Chamber
Maintaining precise temperature control is essential for optimizing cell growth efficiency. To achieve this, a temperature-controlled chamber has been designed. This chamber not only was tested to ensure the safe housing of cells but also offers adaptability to various microscopes, minimizing abrupt temperature fluctuations during cell observation under the microscope. The primary objective of this hardware is to allow users to select and maintain a user-defined temperature, tailored to the specific cell type being cultured for the experiment.
Designing the Hardware
The design created for the cell temperature chamber was made in order to accommodate various microscopes and hold cells while maintaining a consistent temperature via PID temperature control technology. This adaptable element is positioned beneath the microscope and facilitates data recording without necessitating the removal of cells from the controlled environment. After several testing of the first prototype, the chamber connects to the heater via a thermally insulated tube. Overcoming the task of efficiently delivering the appropriate amount of heat to the chamber posed significant challenges. This was mainly because the heater could not be placed inside the cell chamber directly because the space was too small and could not be increased in order to fit it under the microscope. To fix this issue after several tests, direct airflow was added to the system through a vent, allowing enough heat transfer to achieve the required temperature in the cell chamber. Another challenge encountered while testing the product was delivering accurate heat levels without causing the heater holder to overheat. This problem was solved by recognizing the significance of airflow and employing a thermally insulated tube to regulate heat distribution.
PID temperature controller
After getting some preliminary results another issue encountered was the presence sudden spikes in temperature levels, even if the heater was shut down every time the temperature setpoint was reached. To overcome this issue a PID temperature controller using relays was created by realizing an Arduino library.
The PID controller is a powerful tool, enabling the analysis of current, past, and future values in relation to a designated setpoint and the real-time data from the sensor. For the purpose of this project a PID temperature controller designed for a relay-utilizing system was created.
Fundamentally, this controller operates by evaluating the current, past, and future values, subsequently generating an output that, if surpassing a predefined threshold, triggers the shutdown of the heating system, regardless of whether the desired setpoint temperature has been attained. This approach stems from the fact that the PID controller's output is a composite of the present value, a past value calculated using integrals, and a future value extrapolated through derivatives. When these components are combined, the result not only reflects the immediate state but also incorporates the past and future aspects. Consequently, abrupt fluctuations in temperature readings are effectively mitigated.
The system was initially implemented with equal weights assigned to the present, integral, and derivative constants. The preliminary outcomes indicated a noticeable reduction in temperature spikes. Nonetheless, to fully optimize the PID system's performance, fine-tuning the values of the present, integral, and derivative constants became necessary. This tuning process proved essential as it allowed the PID controller's output to emphasize the present, past, or future, based on the specific requirements of the system. In the scenario where controlling temperature spikes posed the main challenge, a focus on refining the controller by augmenting the derivative value proved effective. Subsequently, through several extended overnight experiment, the temperature was consistently maintained within the desired range, confirming the effectiveness of the chosen approach. In the graph below we can see two different overnight experiments each with a different setpoint, one 37 C and the other 35. Both temperature were well maintained throughout the experiment.
The system was subsequently tested in an alternative scenario. During an ongoing experiment, the setpoint for the environmental chamber was changed to a lower value. This test aimed to gauge the system's rapidity in adapting to a change in setpoint in order to make sure a sudden change would not influence the cells. The graph depicted below clearly illustrates the system's remarkable agility in adjusting to the abrupt temperature shift, effectively maintaining the new setpoint without impacting the cells.
