We designed a continuous culturing apparatus for use with optogenetic systems to illuminate cultures of microbes and regularly image cells in the effluent with an inverted microscope. The culturing, sampling, imaging, and image analysis are fully automated so that dynamic responses to illumination can be measured over several days.
Optogenetic systems utilize genetically-encoded proteins that change conformation in response to specific wavelengths of light to alter cellular processes. There is a need for culturing and measuring systems that incorporate programmed illumination and stimulation of optogenetic systems. We present a protocol for building and using a continuous culturing apparatus to illuminate microbial cells with programmed doses of light, and automatically acquire and analyze images of cells in the effluent. The operation of this apparatus as a chemostat allows the growth rate and the cellular environment to be tightly controlled. The effluent of the continuous cell culture is regularly sampled and the cells are imaged by multi-channel microscopy. The culturing, sampling, imaging, and image analysis are fully automated so that dynamic responses in the fluorescence intensity and cellular morphology of cells sampled from the culture effluent are measured over multiple days without user input. We demonstrate the utility of this culturing apparatus by dynamically inducing protein production in a strain of Saccharomyces cerevisiae engineered with an optogenetic system that activates transcription.
Optogenetic systems use light to control a growing list of cellular processes including gene expression,1,2,3,4,5 protein localization,6 protein activity,6,7,8 protein binding,8,9,10 and protein degradation.11 A method for culturing cells in a controlled environment with programmed optical stimulation and for measuring their response over biologically relevant timescales is necessary to exploit the potential of these tools for research in cell biology and biotechnology. Our method takes advantage of chemostasis to maintain a constant cell growth rate in a well-mixed, aerated, and temperature-controlled glass culturing vessel12,13 that is exposed to programmed illumination. We image individual cells in the culture effluent with an inverted microscope to measure the response of the culture to programmed illumination. The culturing, sampling, imaging, and image analysis are fully automated so that the fluorescence intensity and cellular morphology of the effluent cell culture can be measured over multiple days without user input.
This protocol can be implemented in most labs familiar with growing cell culture and microscopy, and the apparatus used is inexpensive and made of readily available components. A transparent culturing vessel is placed above a matrix of light emitting diodes (LEDs) capable of emitting 1 µW/cm2-10 mW/cm2 of light. Microbes are grown in the culturing vessel continuously; one peristaltic pump is used to add media at the dilution rate, another is used to withdraw culture at a lesser rate to the microscope, and the difference escapes through an overflow outlet. A heating pad maintains the temperature. Air is continually pumped into the culturing vessel to maintain a positive pressure as well as to mix and aerate the culture. Except for the air pump, power to these devices is regulated by a microcontroller that also receives input from a thermometer and a connected desktop computer. The effluent cell culture is pumped to a microfluidic device on the stage of an inverted microscope. Non-fluorescent and fluorescent images are automatically acquired. The cells in the images are characterized by an algorithm that locates each cell as a region of interest (ROI) and measures the properties of each ROI.
To demonstrate an application of this protocol, we measured the response to varying light intensities of Saccharomyces cerevisiae cells engineered with a blue-light responsive optogenetic system which controls the transcription of fluorescent protein. S. cerevisiae, commonly known as baker's yeast, was selected because multiple optogenetic systems for controlling gene expression in this system already exist14,15,16. Furthermore, this model organism is commonly used for studies in systems biology17 and as a chassis for biotechnological applications18,19,20. Our representative results demonstrate that this protocol can be used to control transcription of a culture over multiple days by varying input light intensities and measuring the production of a fluorescent reporter.
Figure 1: The continuous culturing apparatus. This simplified diagram shows how the apparatus should be assembled when it is used to culture, illuminate, and measure optical properties of the microbes. Please click here to view a larger version of this figure.
Figure 2: Overview of the protocol. The steps in the shaded region must be repeated every time the protocol is used. Closed loop control is possible34, but is not implemented in this protocol.
1. Assemble the Thermometer to the Circuit Board
Figure 3: Connections to read thermometer values. This diagram shows how the digital thermometer should be connected to the PCB so that the microcontroller can get feedback to control the temperature of the culture. Please click here to view a larger version of this figure.
2. Connect the Power Control Components to the Circuit Board
Figure 4: Connections to control power to the heating pad. This diagram shows how the PCB and accessory parts should be assembled in order to control power to the heating pad. The parts to control the peristaltic pumps are connected in a similar manner. Note that the components for the thermometer have been removed for clarity. Please click here to view a larger version of this figure.
3. Connect the LED Matrix to the Circuit Board
Figure 5: Connections to control the LED matrix. This diagram shows how the LED matrix should be connected to the PCB. It also shows that the PCB can be stacked on the microcontroller. Note that the components for the thermometer and for controlling DC power to other devices have been removed for clarity. Please click here to view a larger version of this figure.
4. Install Software and Connect to Hardware
5. Make and Characterize the Light-proof Enclosure for the Culturing Vessel
6. Prepare the Culturing Vessel
Figure 6: Vessel connections. This diagram shows how the vessel and tubing of the apparatus should be connected prior to being autoclaved. Please click here to view a larger version of this figure.
7. Prepare the Media Flask
8. Prepare the Effluent Flask
9. Prepare the Microfluidic Channel
10. Fill the Media Flask
Figure 7: Adding media. This diagram shows how media should be vacuum filtered into the media flask. It ensures that the media remains sterile. Please click here to view a larger version of this figure.
11. Assemble the Apparatus Around the Microscope
12. Calibrate the Pumping Rates
13. Collect Microscope Images of Cultured Microbes
14. Post-experiment
This apparatus was used to stimulate a culture of S. cerevisiae expressing yellow fluorescent protein (YFP) in response to blue-light via an inducible optogenetic transcription system based on the CRY2/CIB1 protein pair30. Cells were grown chemostatically in phosphate-limited media with an average dilution rate of 0.2 ± 0.008. Phosphate limitation is commonly used in S. cerevisiae chemostat experiments to control growth rate and the effects of phosphate limitation are well characterized.31,32,33 Effluent from the continuously-diluted culturing vessel was sampled to a microfluidic device on an inverted microscope. The images were automatically analyzed as shown in Figure 8. Individual cells were identified in background-subtracted phase-contrast images and their YFP concentration was estimated from their fluorescence as measured from the background-subtracted fluorescent images.
The fluorescence of 169,677 cells was analyzed from 33,600 images of effluent acquired from 28 locations in the microfluidic device over the 70 h of the experiment. We used an inverted microscope equipped with a fluorescence illumination system and a CMOS camera. The YFP images were acquired with a 500/20 nm excitation filter, a 535/30 nm emission filter, and a T515lp dichroic. Images were taken with a 40X phase contrast objective, under Köhler illumination. Images were recorded in 16-bit color depth with 87 pixels per micrometer squared. The culture was exposed to varying intensities of blue light for 6 h intervals, followed by complete darkness for 6 h intervals. The culture had been exposed to light prior to the first measurement, which is why its fluorescence intensity is decreasing during the first dark period.
Figure 9 shows the fluorescence due to YFP production upon activation of the optogenetic system in response to illuminating the culturing vessel. It demonstrates the utility of single cell measurements of fluorescence over a population average-single-cell measurements reveal the population distribution of fluorescence intensity. Note the outlier at 42 h 26 min. After reviewing the corresponding images, it was apparent that there were clumps of cells in the majority of the images used to generate the composite image of the background, resulting in artifacts that resembled cells in the background-subtracted image. Also, a bubble from the culturing vessel was pumped to the microfluidic channel and parts of its edges were mistaken as cells by the image analysis algorithm. Since neither the bubble nor the artifacts from subtracting the background corresponded to actual cells, their measured fluorescence intensity is lower than the auto-fluorescence of the cells. This figure demonstrates the quality of data that can be automatically acquired over 3 d. Please click here to view a larger version of this figure.
Figure 8: Visual depiction of the image analysis algorithm. These images have been cropped and expanded for ease of viewing. (A) Six images are acquired to generate the background-subtracted phase contrast image of the cell culture. After the main image is acquired, five additional images are acquired with the effluent-sampling pump briefly turned on between each acquisition to ensure that cells are displaced. A composite image of the background is generated from the five component images. The value of each pixel in the background image is the median value of that same pixel across the 5 component images. (B) The background-subtracted image is then converted to a binary. The binary is then dilated and the holes within continuous sections of the binary are filled. The yellow outlines correspond to selected regions of interest (ROI), based on size and circularity criteria. (C) Those ROI are mapped onto the fluorescent image, where the fluorescence of each cell is measured as the value of the brightest pixel within the ROI. Please click here to view a larger version of this figure.
Figure 9: Population distribution of fluorescence intensities over time. In these subfigures the logarithm of measured fluorescence is displayed, following standard practice in flow cytometry. The line in A and B indicates the intensity of light absorbed or diffracted by the culture, which is measured as the difference in intensity between light transmitted through sterile media and light transmitted through cell culture in the culturing vessel. It is plotted against the second ordinate axis. (A) A box and whisker plot (5th percentile, 25th percentile, median, 75th percentile, 95th percentile) of the logarithm of the measured fluorescence of the population over time. (B) A 2-dimensional histogram of the same data, where color corresponds to the normalized frequency of cells with a measured fluorescence in the range of the corresponding bin. The colors were scaled to the range of data without the outlier at 42 h 26 min included. The normalized frequency of the outlier in the lowest fluorescence bin is 0.7. (C) One dimensional histograms of the logarithm of the measured fluorescence of the population before the first recorded exposure to light and after the greatest exposure to light. It demonstrates that the light-induced expression of YFP in this strain is bimodal. Please click here to view a larger version of this figure.
We designed this apparatus with flexibility in mind. All the code used is free and open-source. The default image analysis process to segment cells is simple and runs quickly. Custom analysis could be implemented by recording user input while analyzing a representative image with the FIJI graphic user interface, converting the input to a beanshell script, and then setting the plugin to call the script. When it is called, this script will be sent a String array called "images" containing the file paths to the most recent set of background-subtracted images. Images and data are saved as they are collected so that they are not lost if an experiment ends abruptly. A blue LED array was chosen for inducing our optogenetic system but it could be replaced with LEDs of different colors. There are additional input and output pins available on the microcontroller as well as through-holes for an additional MOSFET switch and relay switch on the PCB that make it easy for this system to be adapted for more complex purposes. For example, to make this apparatus operate as a turbidostat, use the additional pins on the microcontroller to power an LED and read values from a light-sensor strapped to the culturing vessel, and then modify the software to measure turbidity and dilute the vessel accordingly.
Care must be taken to ensure that the culture is not contaminated, to protect sensitive microscopy equipment, and to ensure that the fluid flow rates are consistent. Contaminations prior to when the culturing vessel is intentionally inoculated can be detected by waiting a few days before inoculating the vessel and verifying that there is no growth inside. To avoid spilling cell culture on the microscope objective, check that the microfluidic device will not leak by first pumping cell culture through it over an absorbent cloth. If the flow of media into the culturing vessel is inconsistent, it is usually because the tube around the rollers of the peristaltic pump has become too loose. If the flow of air from the aquarium pump is inconsistent, ensure that there are no U-shaped bends in the effluent tube where liquid may pool and inconsistently resist air flow. If tubing becomes clogged after an experiment because media has dried up inside of it, soak the clogged tubes in a hot water bath to dissolve the clog.
There are intrinsic limitations to this protocol to consider when planning an experiment or analyzing results. Depending on the dilution rate and length of tube used, there is a delay of roughly 10 min during which the cell culture is pumped from the culturing vessel to the microscope. Therefore, it is not well suited to studying events occurring over shorter timescales. Also, while an experiment can run continuously for about ten days, limited only by the volume of media, the evolution of the cell culture must be considered as the duration of the experiment increases. The limiting nutrient of the media has profound effects on metabolism and gene expression35,36,37,38,39 and should therefore be determined based on the aspect of physiology under study. When analyzing results, the images-especially the images from outlying data points-should be reviewed for errors in the way that ROIs were identified by the image analysis routine. It is possible for bubbles or artifacts to be mistaken for cells, thereby skewing the measured fluorescence. One simple way to eliminate such erroneous measurements is to discard all ROIs with fluorescence intensities below the auto-fluorescence intensity.
This protocol will be useful for measuring fluorescence intensity and/or cellular morphology of effluent cell culture in response to light in organisms that can grow in continuous culture, settle to a consistent focal plane when not stirred, and be automatically identified by an image analysis algorithm. The default cell identification algorithm used here will be most directly applicable to roughly spherical microbes that resemble budding yeast. One alternative40 to this protocol is to grow microbes in a set of batch cultures, and then sample one batch for each time point and characterize the sample by flow cytometry. However, the advantages of the protocol described here are that samples are taken from the same culture, many samples can be taken, and the process is automated. This protocol also improves upon a similar method in which optogenetic yeast were continuously cultured and imaged under a microscope34 by acquiring multiple images quickly and not biasing ROIs identification by fluorescence. A great advantage of this protocol is that many measurements of individual cells can be regularly acquired over several days without user input. After measuring the response of the microbial culture to light exposure, a next step may be to implement in silico closed-loop control of the response.
The authors have nothing to disclose.
We would like to acknowledge Molly Lazar and Verónica Delgado for assistance in testing the protocol, Kieran Sweeney for helpful discussions and editing, and Taylor Scott, My An-adirekkun, and Stephanie Geller for critical reading of the manuscript. Megan Nicole McClean, Ph.D. holds a Career Award at the Scientific Interface from the Burroughs Wellcome Fund.
Extensive lab manual | GitHub | NA | An extensive, regularly updated lab manual is available in the “Optogenetic Chemostat Files” GitHub repository (https://github.com/McCleanResearch/Optogenetic-Chemostat-Files). This also includes a description of the microfluidic mold used to generate the representative results. |
Fritzing Design Viewer | Fritzing | NA | The free, open-sourced software to view and edit the .fzz type circuit board designs is available at "http://fritzing.org/download/" |
Arduino Uno R3 (Atmega328 – assembled) | Adafruit | 50 | Microcontroller. 1 required. |
Arduino Stackable Header Kit | SparkFun Electronics | 10007 | Female pin headers for connecting PCB to microcontroller. 1 required. |
Adjustable 30W 110V soldering iron – XY-258 110V | Adafruit | 180 | For making electrical connections to the PCB. 1 required. |
Soldering iron stand | Adafruit | 150 | For making electrical connections to the PCB. 1 required. |
Mini Solder spool – 60/40 lead rosin-core solder 0.031" diameter – 100g | Adafruit | 145 | For making electrical connections to the PCB. 1 required. |
0.1 μF capacitor | SparkFun Electronics | COM-08375 | Stabilizes voltage in PCB. 1 required. |
10 μF capacitor | SparkFun Electronics | COM-00523 | Stabilizes voltage in PCB. 1 required. |
MAX7219CNG LED Matrix/Digit Display Driver – MAX7219 | Maxim | MAX7219CNG | LED driver. 1 required. |
8 pin IC Socket | Mouser Electronics | 575-144308 | 16 required. These will be stacked on top of each other to support the culture vessel above the LED matrix. |
24 Pin IC socket | Mouser Electronics | 535-24-3518-10 | Optional. Use this to reversibly attach the MAXIM 7219CNG driver to the PCB. |
Digital multimeter | Adafruit | 2034 | For troubleshooting electronics. 1 required. |
Break Away Headers – 40-pin Male (Long Centered, PTH, 0.1") | SparkFun Electronics | PRT-12693 | Male pin headers for connected LED matrix to printed circuit board. Ends can be trimmed with wire cutters. 1 set required. |
Flush diagonal wire cutters | Adafruit | 152 | For trimming long pin headers and cutting power cables. 1 required. |
Premium Female/Female Jumper Wires – 40 x 12" (300mm) | Adafruit | 793 | Wire ribbon for connecting breadboard to LED matrix. Can be connected end-to-end with male pin-headers to be longer. 1 required. |
Half-size breadboard | Adafruit | 64 | The LED matrix will connect to this and the culturing vessel will rest above it. |
Miniature 8×8 Blue LED Matrix | Adafruit | 956 | Light source. Dominant wavelength is 470nm (blue). 1 required. Alternative miniature LED matrices from the same vendor are available with dominant wavelengths: 624 nm (red), 588 nm (yellow), 525 nm (green), 572 nm (yellow-green), and white. |
Stackable header-3 pin | SparkFun Electronics | 13875 | 8 required. |
Resistor Kit – 1/4W (500 total) | SparkFun Electronics | 10969 | For electronics. 1 required. |
IRL520N MOSFET | International Rectifier | IRL520N | Voltage regulating switch for controlling DC current. 4 required. |
Hook-Up Wire – Assortment (Solid Core, 22 AWG) | SparkFun Electronics | PRT 11367 | Wire for electronics. 1 required. |
5V 2A (2000mA) switching power supply – UL Listed | Adafruit | 276 | Power supply for the heating pad and Arduino. 2 required. |
12 VDC 1000mA regulated switching power adapter – UL listed | Adafruit | 798 | For peristaltic pumps. 2 required. |
Electric Heating Pad – 10cm x 5cm | Adafruit | 1481 | For heating the bioreactor. 1 required. |
Low flow variable flow peristaltic pump | Fisher Scientific | 13-876-1 | For pumping media. 1 required |
Medium flow variable flow peristaltic pump | Fisher Scientific | 13-876-2 | For pumping culture. 1 required. |
9 VDC 1000mA regulated switching power adapter – UL listed | Adafruit | 63 | For microcontroller power supply. Order 1. |
High Temp Waterproof DS18B20 Digital temperature sensor + extras | Adafruit | 642 | Thermometer for the bioreactor. 1 required. |
Micromanager | Micromanager | NA | The free, open-sourced microscope control software is available at "https://micro-manager.org/wiki/Download_Micro-Manager_Latest_Release" |
FIJI | ImageJ | NA | The free, open-sourced image analysis software is available at "http://fiji.sc/" |
Arduino Integrated Development Environment | Arduino | NA | The free, open-sourced IDE is available at "https://www.arduino.cc/en/Main/Software" |
Custom code | GitHub | NA | The custom microcontroller code and "Bioreactor Controller" plugin are available in the “Optogenetic Chemostat Files” GitHub repository (https://github.com/McCleanResearch/Optogenetic-Chemostat-Files). |
USB Cable A to B – 6 Foot | SparkFun Electronics | CAB-00512 | Used to download data to microcontroller. 1 required. |
bioreactorTimecourse_example.csv | GitHub | NA | The advantage of loading LED matrix values from a CSV file is that a program can be called by the plugin to update those values based on image analysis results, and those values can be reloaded to the microcontroller, enabling feed-back control. It is available from the “Optogenetic Chemostat Files” GitHub repository (https://github.com/McCleanResearch/Optogenetic-Chemostat-Files). |
Tota-frost gels (diffusion paper) | B&H | B&H # LOFSFTL MFR # T1-72 |
For LED matrix. 1 required. |
Kitting Sheet Crosslink 1/4x12x24in | Grainger, inc | 20JL37 | Black foam for culturing vessel enclosure. 4 required. |
Standard Photodiode Power Sensor, Si, 200 – 1100 nm, 50 mW | Thorlabs | S120VC | For measuring light intensity. 1 required. |
Labelling Tape | Fisher Scientific | 159015N | For labelling and securing loose components. 1 required. |
Compact Power and Energy Meter Console, Digital 4" LCD | Thorlabs | PM100D | For measuring light intensity. 1 required. |
100mL GL45 hybridization glass bottle | Bellco Glass, Inc. | (7910-40150) | Bioreactor vessel. 1 required. |
Six port assembly | Bellco Glass, Inc. | Custom | For the bioreactor vessel. Tubing Specs: .125" OD x .055"ID. Port A: 1.0" long above cap slug and to bottom of tube. Ports B,C,E,F: 1.0" long above cap slug, 33 mm long below. Port D: 1.0" long above cap slug, 65 mm long below. 1 required. Includes 45 mm diameter polypropylene open top screw cap and a white silicone gasket to ensure a tight seal between the cap and the vessel. |
Scotch Magic Tape 3105, 3/4 x 300 Inches, Pack of 3 | Amazon | B0009F3P3U | Clear scotch tape. This is available from many other vendors. It is used to cover markings on the culturing vessel and to secure the coverglass with the PDMS channel to the aluminum support frame. |
1/16" ID x 3/16" OD x 1/16" Wall Tygon Sanitary Silicone Tubing | United States Plastic Corp. | 57288 | Tubing. ~25' required. |
Cole-Parmer Twistit white rubber stopper, size 10 | Cole-Parmer | EW-62992-32 | Media flask stopper and effluent flask stopper. 2 required. |
2L Laboratory Flask | Pyrex | 4980 | Media flask and effluent flask. 2 required. |
Day pinchcock | Fisher Scientific | 5867 | For pinching tubes shut. 3 required. |
Replacement tubing assembly 1/16" ID | Traceable Products | 3372 | The peristaltic pumps come with a set of tubes, but they wear out after weeks of use. |
Replacement tubing assembly 1/50" ID | Traceable Products | 3371 | The peristaltic pumps come with a set of tubes, but they wear out after weeks of use. |
Male luer with lock ring x 1/16" hose barb, Nylon, 25/pk | Cole-Parmer | EW-45505-00 | Connectors. ~10 luers are required. |
Male luer with lock ring x 1/8" hose barb, Nylon, 25/pk | Cole-Parmer | EW-45505-04 | Connectors. 5 required, one for each rubber stopper hole to fill with tubing. |
Female luer x 1/16" hose barb adapter, Nylon, 25/pk | Cole-Parmer | EW-45502-00 | Connectors. ~10 luers required. |
Female luer x 3/16" hose barb adapter | Cole-Parmer | EW-45502-08 | Connectors. ~10 luers required. |
Cole-Parmer Luer Accessory, Female Luer Cap, Nylon, 25/Pk | Cole-Parmer | SC-45502-28 | |
Cole-Parmer Luer Accessory, Male Luer Lock Plug, Nylon, 25/Pk | Cole-Parmer | EW-45505-56 | |
Microbore PTFE Tubing, 0.022"ID x 0.042"OD, 100 ft/roll | Cole-Parmer | EW-06417-21 | Tubing. 1 roll required. |
Masterflex platinum-cured silicone tubing, L/S 13, 25 ft | Cole-Parmer | EW-96410-13 | Tubing. ~25' required. |
3/16" ID x 1/4" OD x 1/32" Wall Tygon Sanitary Silicone Tubing | United States Plastic Corp. | 57293 | Tubing. ~1' required. |
Vacuum filter | Fisher Scientific | 974107 | Nalgene vacuum filter for sterile filtering media. |
Aquel Oxy-Boost 200 | Rena Aquatic Supply | AP200 | Dual diaphram adjustable flow air pump for aerating and mixing media. 1 required. |
0.2 μm pore syringe filter | Corning International | 431229 | This ensures that air from the aquarium pump does not contaminate the apparatus. 1 required. |
Slygard 184 Silicone Elastomer Kit | Dow Corning | Slygard 184 | For microfluidic device. 1 required. |
American Safety Razor GEM Scientific Single-Edge Razor Blades | Fisher Scientific | 17989000 | For cutting tubes and PDMS. 1 blade required. |
Harris Uni-Core hole puncher 1.2mm ID | Sigma-Aldrich | WHAWB100028 ALDRICH | For punching inlet/outlet in microfluidic device. 1 required. |
Microscope cover glass 22×60-1.5 | Fisher Scientific | 12-544-G | For microfluidic device. 1 required. |
Rectangular aluminum frame with a square window | Custom | Custom | To support the microfluidic channel. Outer dimensions: 3 inches x 1.25 inches. Inner dimmensions (cut out portion): 7/8 inches x 7/8 inches Thickness: ~1/32 inches |