Team:UC Davis/Hardware


SELECTING OPTOGENETIC HARDWARE FOR SAMPLE ILLUMINATION

The Light Plate Apparatus (LPA) from Jeff Tabor’s Lab is “ a general purpose, customizable, user-friendly instrument for performing optogenetics...experiments in batch and adherent cell culture systems” (http://www.taborlab.rice.edu/hardware). The design revolves around a 24-well tissue culture plate and is comprised of a custom printed circuit board (PCB) populated with 48 LEDs and surrounded by a 3D-printed chassis. For our project, we decided to build two LPAs so that we could run two experiments that were independent of each other in duration, LED settings, and cell lines. As a team, we also decided to have both LPAs built completely in-house, despite the fact that the Tabor Lab has provided commercial assembly instructions for the LPA (http://taborlab.github.io/LPA-hardware/fab/index.html).

BUILDING THE HARDWARE (LPAs)

We utilized most of the parts listed in the Bill of Materials (BOM) provided by the Tabor Lab, and found suitable replacement parts for those that were discontinued. Our team made our own updated BOM that covers all of the components we used in our version of the device. Using the open-source .stl and CAD files, we were able to purchase the PCBs from OshPark and 3D-print a modified version of the LPA chassis in-house that Hyunsoon Kim had previously designed. We populated the PCB with all the required components and used a digital multimeter (DMM) to test connectivity of the PCB before powering up. Following the Tabor Lab’s instructions, we used an AVR ISP mkII programmer to program our two circuit boards. The LPA’s customizable design allows for LEDs of different wavelengths to be used and easily switched around into different configurations. However for our experiments we only utilized blue (472nm) and red (630nm) LEDs.

SELECTING THE SOFTWARE AND PROGRAMMING THE LPAs

Using the Iris software, created by the Tabor Lab, we were able to write specifically designed programs for our two LPAs (http://taborlab.github.io/Iris/documentation/documentation.html). Iris allowed us to independently control the specific LED intensity, duration of illumination, and waveform of light incident on each tissue culture well. For example, for an experiment where we compared a registry part to ours system, we had to have different LED intensities per well. Using the Iris program, we were able to set triplets of wells to 1000GS, 1500GS, 1750GS, and 2000GS while also flashing each well on and off every 20s for 26 hours (see below). This was achieved easily by creating multiple experiments within a single plate and defining the parameters for each one, independent of all other experiments.

CALIBRATING THE LPAs

On their website, the Tabor Lab has outlined two different protocols to effectively ensure that all LEDs within the LPA are calibrated correctly (http://taborlab.github.io/LPA-hardware/initializing/calibration_led.html?highlight=calibrating). However, due to unforeseen difficulties with reading image formats into MatLab, we decided to calibrate our LPA using a method that involved measuring the illumination in each well with a commercially available lux meter and than manually editing the individual dot correction (dc) and greyscale calibration (gcal) files. These two text files are responsible for the overall emitted intensities of each individual LED; “Coarse adjustments can be made by setting the LED dot correction, while fine adjustments can be made setting the greyscale value” (http://taborlab.github.io/LPA-hardware/initializing/calibration_led.html?highlight=gcal). To improve our lux meter, we designed and 3D printed an adapter to help create a light-tight and stable interface between the meter and the surface of the LPA. The adapter so that we could more accurately measure the light coming from each individual well. Pictured below, from left to right, is our 3D-printed adapter, lux meter, and our lux meter taking measurments.

Calibration Workflow

Before calibrating, we measured the overall lux of each well across three different intensities. Using the Iris software, we produced three constant waveform program files at 500, 1000, and 1500GS respectively, in order to test the effectiveness of our lux meter and adapter. From this test, we realized that each well had a noticeable differences in overall lux and showed us that the LEDs had noticeable variability.

  1. We began calibration by measuring the lux of each well in both LPAs and recording them in a spreadsheet
  2. Next, we observed the gcal and dc .txt files for their corresponding values to each LED
    1. We decided to keep all of dc.txt values constant (x=24) and only adjust the gcal.txt values, since gcal corresponded to more fine-tuned adjustments
  3. After some testing with the lux meter, we concluded that 2-3 gcal integer values were approximately equal to 1 lux. Therefore, we changed each LED’s gcal value by multiples of three in the direction of the closest multiple of ten.

    4. So after measuring one of the LPAs, the average value for lux was 115 with a standard deviation of 3.54. We then corrected the gcal values so that they gave us an average value of 120 lux and a standard deviation of 0.76. If our null hypothesis was that all wells were equal in lux values (STD=0) and we were able to calibrate our LEDs to be within two standard deviations from the null, therefore we could conclude that our method of calibration was statistically significant.

    Text files for dot correction (above) and greyscale calibration (below)

    CHARACTERIZING THE LIGHT-TEMPERATURE RELATIONSHIP IN THE LPA

    Motivation

    Due to some design properties of the LPA, we were concerned that the heat generated by the unit would be sufficient to change the temperature of the samples. Even small changes in temperature - particularly if related to the incident light intensity - could trigger changes in endogenous gene expression and contribute to noise and/or unintended gain or loss of signal for our target genes. Neither the Tabor Lab’s website nor their GitHub had documentation on how different intensities affected temperature. Therefore, we wanted to know if and how much the temperature of samples might be influenced by the operation of the LPA.

    Experimental Setup

    To test how illumination might influence the temperature of the sample, we conducted a number of experiments in which we explored the influence of three key variables: light intensity, illumination waveform, and the use of fans for cooling . We used the Iris program (http://taborlab.github.io/Iris/) to create illumination programs for each experiment. Iris uses hardware greyscale units (GS), from 0 to 4095GS, to report all light intensities (amplitudes) and based on this intensity system, we decided to not only test the extremes, but also constant light versus flashing.

    Measurement Setup

    We used a Fluke 51 II Handheld Digital Probe Thermometer to monitor changes in temperature in culture wells. We also wanted to observe the temperature uniformity of the 24 well plate and decided to gather information on inner wells versus outer wells. To accomplish this we:

    1. Drilled holes into the lid above an “external” and “internal” well. These holes enable us to simultaneously thread two independent temperature probes into culture wells located in the interior and periphery of the plate
    2. Filled each well with 1mL of preheated water to simulate media
    3. Inserted probes into the water of each well
    4. Placed unit into 37°C CO2 incubator
    5. Initiated illumination program
    6. Read temperature of incubator, internal and external well at regular intervals

    Experiment #1: Constant waveform, maximum illumination (4000GS), no fan

    First we tested the LPA at 4000GS with a constant waveform for a duration of 5 hours. The inside of the LPA increased in temperature by ~3.6°C (40.6°C) from the incubator temperature of 37°C. Even if this intensity is optimal for our LACE system, this temperature would be too hot and stressful for the cells.

    Experiment #2: Constant waveform, changing illumination (500GS to 4000GS)

    Another test we ran was to identify how fast the wells could heat up if programmed with different intensities. For this test, we programmed the LEDs to be at 500GS for 250 minutes and then switch to 4000GS for the duration of the experiment. After 900 minutes (15 hrs), we observed that the wells heated up to ~40°C after 150 minutes. Two important conclusions were drawn from this experiment. First, different intensities most certainly affected overall sample temperature. Secondly, after a period of time, the temperature reaches a plateau point and only revolves around that given point.

    Experiment #3: Box waveform (20s on, 20s off), maximum illumination (4000GS), no fan

    We hypothesized that flashing the light source ½ on, ½ off might generate less heating (though at the expense of fewer photons hitting our sample). To test this hypothesis, we tried illuminating our culture plates at 4000GS and flashed the LEDs on and off every 20s; we found that after 1240 minutes (20.6 hours) the max temperature reached was 38.9°C. This was indeed cooler than constant illumination (40.6°C) but still likely to induce some stress on cells experiencing this perturbation.

    Experiment #4: Constant waveform, maximum illumination (4000GS), plus fan

    Another test we decided to run was setting the LPA to 4000GS and flashing, but then having a small USB fan sucking hot air away from the unit. Placing the fan above the LPA yielded results of about ~1°C decrease from the original 40.6°C we were receiving. These results proved to us that maybe with the addition of a built-in fan or heat sinks, the LPA could be redesigned to better combat temperature constraints.

    Experiment #5: Constant waveform, below maximum illumination, no fan

    Next, we tested the lower end of the intensity spectrum at 1500GS with a constant waveform. From this test, we observed the temperature rise to ~38.6°C after 1000 minutes (16.7 hrs). The LPA was also tested at 1000GS constant for 980 minutes (16.3 hrs) and was observed to reach a maximum temperature of ~37.8°C.

    Experiment #6: Box waveform (20s on, 20s off) , below maximum illumination, no fan

    At 1500GS flashing on and off every 20s, we observed a decrease in temperature of about ~1°C after 16.7 hrs with a final temperature of 37.7°C. Again, this reinforces the idea that flashing our LEDs will keep the sample cooler than constant light conditions. Testing at 1000GS flashing as well gave us a maximum temperature of ~37.1°C after 330 minutes. From this test, we concluded that with the addition of a fan, 1500GS and 1000GS flashing would be safe intensities to work with. However, we decided to go with 1500GS as our main intensity for better activation of our system.

    Analysis of Temperature Tests

    From these tests, it was observed that it took the culture plate approximately 240 minutes to heat up to the point at which the temperature would plateau. It was also noted that the greatest amount of heating was occuring from t=0 to t=150 minutes. These experiments also illustrated a common occurrence for the outer wells to heat up faster than the inner wells, but for the inner wells to have an overall higher temperature after the plateau point. This heating difference across wells was noted and in response, we made sure that each experiment was randomized for well designation. For future tests, it may be interesting to set the LPA to the illumination maximum (4000GS) for a period of time, but then decrease the illumination and observe the duration of the cooldown period.

    Tips and Modifications

    • With the black culture plates that we used, the cell culture plate adapter piece was to large and our culture plates had too much room to wiggle. To prevent this, we super glued thin pieces of nitrile rubber to either side of the adapter so that the plate could not move around undesirably
    • The LPA plate lid inner lining was a few millimeters too short for our culture plate. Therefore, we extended the CAD file for LPA plate lid and kept the ratios of the bolt holes the same for compatibility
    • LED sockets needed to be replaced and resoldered many times due to being very frail. The PCB copper pads would also become removed from the board due to excess heat from the soldering iron

    Preparing the LPA for an illumination experiment

    1. Upload the experiment-specific Iris files to the SD card
    2. Using a Kim wipe sprayed with iso-propyl alcohol, gently wipe down the LPA’s surfaces and place the LPA in the BSC
      1. Turn on the UV light and let the LPA sit for ~5 mins
    3. Load the 24-well plate into the LPA and secure the lid properly (such that the lid is not loose from the 24-well plate, but also not secured so tightly that the LPA lid starts to bend)
    4. Once secured, spray the LPA and USB fan (experiment dependent) with iso-propyl alcohol so that all surfaces are lightly covered (do not soak the LPA as the 3D printed material is porous and can prevent evaporation) and place in the incubator
    5. After a 15 minute drying period, allowing all iso-propyl to evaporate, plug in the LPA and wait for the green “ON” LED for confirmation that the program is being run
      1. If using a USB fan, switch on to the “High” setting