Design and Use of a Low Cost, Automated Morbidostat for Adaptive Evolution of Bacteria Under Antibiotic Drug Selection

Summary

We describe a low cost, configurable morbidostat that enables the characterization of antibiotic drug resistance by dynamically adjusting the drug concentration. The device can be integrated with a multiplexed microfluidic platform. The approach can be scaled up for laboratory antibiotic drug resistance studies.

Abstract

We describe a low cost, configurable morbidostat for characterizing the evolutionary pathway of antibiotic resistance. The morbidostat is a bacterial culture device that continuously monitors bacterial growth and dynamically adjusts the drug concentration to constantly challenge the bacteria as they evolve to acquire drug resistance. The device features a working volume of ~10 ml and is fully automated and equipped with optical density measurement and micro-pumps for medium and drug delivery. To validate the platform, we measured the stepwise acquisition of trimethoprim resistance in Escherichia coli MG 1655, and integrated the device with a multiplexed microfluidic platform to investigate cell morphology and antibiotic susceptibility. The approach can be up-scaled to laboratory studies of antibiotic drug resistance, and is extendible to adaptive evolution for strain improvements in metabolic engineering and other bacterial culture experiments.

Introduction

Since the introduction of the first antibiotic drug penicillin, microbial antibiotic resistance has developed into a global health problem¹. Although the acquisition of antibiotic resistance can be retrospectively studied in vivo, the conditions of these experiments are often not controlled throughout the entire evolution². Alternatively, adaptive laboratory evolution can reveal the molecular evolution of a microbial species under environmental stresses or selection pressure from an antibiotic drug³. Recently, many well-controlled evolutionary experiments of antibiotic drug resistance have elucidated the emergence of antibiotic drug resistance. For example, Austin's group demonstrated rapid emergence in a properly engineered microfluidic compartmented environment⁴. The recently developed morbidostat induces systematic mutations under drug selection pressure^5,6. The morbidostat, a microbial selection device that continuously adjusts the antibiotic concentration to maintain a nearly constant population, is a major advance from the fluctuation test used in microbiology^7,8. In the fluctuation test, an antibiotic drug is injected at high concentration, and the surviving mutants are screened and counted. Instead, microbes in a morbidostat are constantly challenged and acquire multiple mutations.

The morbidostat operates similarly to the chemostat, a culture device invented by Novick and Szliard in 1950 that maintains a constant population by continuously supplying nutrients while diluting the microbial population⁹. Since its introduction, the chemostat has been advanced and improved. Current microfluidic chemostats have reached nanoliter and single-cell capacities. However, these devices are unsuitable for adaptive evolution experiments, which require a large cell population with many mutation events^10,11. Recently, mini-chemostats with working volumes of ~10 ml have also been developed to fill in the gap between liter scale bioreactors and the microfluidic chemostat^12,13.

Here we present the design and use of a low-cost, automated morbidostat for an antibiotic drug resistance study. The proposed module can be employed in a shaker incubator in a microbiology laboratory with minimal hardware requirement. The open-source firmware is also easily tailored to specific applications of adaptive evolution, such as metabolic engineering³. Finally, the morbidostat is integrated into a multiplexed microfluidic platform for antibiotic susceptibility testing¹⁴.

Protocol

1. Assembly and Pretesting of the Morbidostat Device

1. Assembly of the Morbidostat
   1. Punch 3 holes on the cap of the culture vial with an 18 G syringe needle. Cut three pieces of polyethylene tubing ~7 cm in length. Insert these three pieces of polyethylene tubing on the cap.
   2. Use tape to wrap the edge of the cap to serve as the cast for the polydimethylsiloxane (PDMS) mixture. Mix 5 g of A component and 0.5 g of B component of the PDMS in a 150 ml plastic container by stirring manually with a toothpick. Load the mixture into a 10 ml syringe.
      1. Pour the PDMS mixture on the cap with the syringe. Bake the entire cap to cure the PDMS for 8 hr in the oven at 70 °C.
   3. Solder the LED anode with the 24 gauge wire and LED cathode with the carbon resistor with a soldering iron. Solder a 680 Ω carbon resistor with a 24 G wire. Insulate the wire connection using electrical tape.
   4. Solder the collector of the photodetector with the 24 G wire and the emitter of the photodetector with a 100 kΩ carbon resistor. Insulate the wire connection using electrical tape.
   5. Place the light emitting diode in the culture vial holder. By following the circuit diagram in Supplementary Figure 2, connect the light emitting diode and power it with a power supply of 5 V. Ensure that the LED works by using a digital camera that can detect IR (e.g., a mobile phone camera).
   6. Place the photodetector in the culture vial holder. By following the circuit diagram in Supplementary Figure 2, connect the photodetector and power it with a power supply of 5 V. Connect the wire to both sides of the photodetector resistors to measure the voltage across the electronic control board.
   7. Glue a magnet on the shaft of the cooling fan, which serves as the magnetic stirrer unit. Note that the alignment of the shaft of the fan to the culture vial, and the spacing between the culture vial and the magnet is very critical for the proper function of the magnetic stirring unit.
   8. Connect each piece of polyethylene tubing of the culture vial into the corresponding piece of silicone tubing for the micro-pumps, medium bottles, and waste bottle. Turn on the pump for 1 hr and measure the total volume being pumped from the medium bottle to the culture vial to determine the pump rate. Note that the typical pumping rate should be ~4 ml/hr, which corresponds to the dilution rate D = 0.33 hr⁻¹ for a 12 ml working volume of the culture vial.
   9. Place the whole assembly on a mid-sized (34 cm x 34 cm x 43.5 cm) shaker incubator.
2. Pretesting the Morbidostat
   1. Set power supply voltage of cool fan to 5 V to start its motor to test the magnetic stirring bar. Note that the stirring action can be inspected visually. The voltage that drives the cooling fan motor is precisely controlled by pulse width modulation (PWM).
   2. Prepare a series of wild type E. coli samples with optical densities at 600 nm typically ranging from 0.07 to 0.3 for calibration. Place a test E. coli sample in the culture vial and record the photodetector voltage.
      1. Place ~100 µl of the same sample in the plate reader and record the optical density. Use the photodetector voltage versus optical density data to plot the calibration curve. Note that typically, one unit change in optical density corresponds to ~7.6 V change in the photodetector with the bias scheme used here.

2. Running the Morbidostat

1. Preparation Before Starting the Daily Experiment
   1. Prepare the culture vial with three ultra-chemical resistant Tygon tubings with inner diameter 1/32 inch and outer diameter 3/32 inch for inlets to form the device as in section 1.1 and in Supplementary Figure 1.
   2. Prepare the M9 minimal medium (0.2% glucose) and trimethoprim (TMP) drug containing medium. Sterilize the medium, the medium bottle, the drug medium bottle, the waste bottle, and the culture vial in an autoclave at 121 °C.
      NOTE: The TMP is a stock solution that is used to later achieve the concentrations in Table 1 and the TMP drug containing medium is not autoclaved.
2. Running the Morbidostat
   1. On the first day of the experiment, thaw 1 ml of frozen wild type E. coli MG1655 cells for 5 min at room temperature and transfer 120 µl of the cells to the culture vial containing ~12 ml of M9 growth medium. After the first day, thaw 1 ml of frozen E. coli cells from the previous day for 5 min and transfer 120 µl of the cells to a new culture vial containing ~12 ml of M9 growth medium.
   2. Turn on the shaker incubator and set the temperature to 30 °C with no shaking.
   3. Start the morbidostat operation by turning on the micro-pump, the magnetic stirring unit, and the optical density measurement.
      NOTE: During the operation, a feedback threshold algorithm is used to control the drug injection. When the measured optical density exceeds a preset threshold, the drug injection pump would be turned on and the medium pump would be turned off. Otherwise, the drug injection pump remains off and the medium pump is on.
   4. Monitor the voltage from time to time to ensure there is microbial growth.
      Note: Because of the freeze and thaw cycle, E. coli cells exhibit a delay of ~4 hr during each day's growth.
   5. After ~23 hr, stop the micro-pump by turning off its supply voltage and take out the culture vial. Add 15% glycerol into the culture vial and freeze the sample at -80 °C for storage.
   6. Repeat steps 2.2.1 to 2.2.5.

3. Antibiotic Susceptibility Measurement in 96 Well Format

NOTE: Perform the growth rate measurement in a plate reader with a 96 well format to determine the antibiotic resistance level.

1. Thaw the daily frozen sample at room temperature for 5 min and transfer 15 µl of the cell to the culture vial containing ~15 ml M9 medium. Allow them to grow until the optical density at 600 nm reaches ~0.1. Divide the 15 ml into 1 ml bottles and add the TMP drug at this step to obtain the desired drug concentrations as shown in Table 1.
2. Take 200 µl of the sample from step 3.1 and place into each well in the plate reader and allow them to grow for 12 hr⁵. The optical density at wavelength 600 nm is recorded automatically during the measurement.
   NOTE: For each data point, triplicate measurements are recorded and averaged to ensure the consistency of the data. The data on optical density versus time can be used to obtain the growth rate.
3. Plot the growth rate data versus the drug concentration to obtain IC₅₀. Note that IC₅₀ is defined as the drug concentration at which the growth rate is fifty percent of the maximal growth rate⁵.

4. Single Cell Morphology and Antibiotic Susceptibility Measurement in Microfluidic Devices

1. Perform fabrication of the microfluidic devices via multilayer soft lithography from PDMS as described elsewhere¹⁵.
   1. Use photolithography to make the photoresist mold for the control and flow layers. Note that for the control layer, use negative photoresist (~15 µm height) and positive photoresist (~8 µm height) to produce a mold on a bare silicon wafer.
   2. Prepare PDMS mixtures with cross linking ratios of 5:1 and 20:1 for the control layer and the flow layer, respectively. Put the mixtures into a desiccator under vacuum to remove any bubbles, for 1 hr.
   3. Expose the control layer photoresist molds on silicon wafer to trimethylchlorosilane (TMCS) vapor. Make the control layer of ~6 mm thickness by casting the PDMS mixture (with cross linking ratio 5:1) on the silicon wafer on the control layer photoresist mold. Bake the control layer for 40 min at 80 °C for 1 hr in the oven.
   4. Make the flow layer by spin coating a PDMS mixture (with cross linking ratio 20:1, spin speed 3,000 rpm for 60 sec) on a flow layer photoresist mold. Bake the flow layer for 30 min at 80 °C for 1 hr in the oven.
   5. Use a razor blade to cut the control layer into the desired chip size (typically 3 cm x 2 cm) and manually peel off the control layer. Place the control layer on top of the flow layer and align them under a stereomicroscope. Cure the aligned assembly at 80 °C in the oven overnight. Afterwards, soak the assembly in 250 ml of deionized water in a plastic container for 5 hr to dissolve the residual TMCS.
   6. Subsequently, punch the inlet holes with 20 G needles and bond the entire elastomer to a glass slide coated with a blank PDMS layer (PDMS cross linking ratio 5:1, spin speed 2,000 rpm for 30 sec) by air plasma treatment for 40 sec at 1.2 Torr air pressure.
   7. Carry out extensive baking for 36 hr at 80 °C to reduce the cytotoxic effects from PDMS. Note that this baking step is critical to minimize the toxicity to the bacterial cells.
2. On Chip Growth Experiment
   1. Before loading into the microfluidic device, flush it with deionized water for 1 hr and M9 medium for 1 hr.
   2. Thaw the daily frozen sample at room temperature for 5 min and inoculate a new test tube with fresh M9 medium. Place the sample in a shaker incubator at 37 °C overnight.
   3. Dilute the microbial sample 10 fold with M9 medium and load it into the microfluidic device by pressuring the microbial sample container at 5 psi. Then load the TMP drug medium into the chip by pressuring the drug medium container at 5 psi. Execute the mixing between the microbe and drug by actuating the micromechanical valve between the two chambers on the microfluidic chip for 15 min.
   4. Place the microfluidic chip in the microscope incubator mounted on the inverted microscope. Acquire the cell image in the growth chamber every hr for 8 hr with the CCD camera mounted on the inverted microscope.
   5. Use the custom program script to calculate the cell numbers through a threshold algorithm on the cell image data¹⁸. Note that the cell number data are averaged typically over three microfluidic chambers.

Results

The above-described morbidostat is schematized in Figure 1. The common morbidostat operations, including experimental evolution, antibiotic susceptibility test and cell morphology checking, were validated in an E. coli MG1655 culture exposed to trimethoprim (TMP), a commonly used antibiotic drug^5,6. TMP induces very distinctive stepwise increases in drug resistance, and the mutations are clustered around the dihydrofolate reductase (DHFR) gene. Therefo...

Discussion

A low-footprint morbidostat device from low-cost components is demonstrated. The increases in drug resistance level registered by the device are consistent with those of previous reports⁵. Designed for evolutionary studies of drug resistance, the device is potentially applicable to many other experiments. First, a comprehensive database of drug-induced mutations can be established for a large set of clinically relevant antibiotics. For example, the evolutionary pathway of multiple drug resistance can be studie...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Environmental Shaker Incubator	BioSan	ES-20
Arduino Leonardo board	Arduino	Leonardo
680 Ohm Carbon Resistor	Digikey		Bias resistor for LED
100k Ohm Carbon resistor	Digikey		Bias resistor for phototransistor
940 nm light emitting diode	Bright LED Electronic	BIR-BM13E4G-2	Optical density measurement
940 nm phototransistor	Kodenshi	ST-2L2B	Optical density measurement
Darlington pair IC Toshiba	Mouser	ULN2803APG	this IC drives micropumps and magnetic stirring unit
5 V DC brushless fan	ADDA	AD0405LX-G70	spec: 5 V supply voltage and 80 mA; available www.jameco.com
Piezoelectric micropump	CurieJet	PS15I-FT-5L	Pressure > 3 kPa; Flow rate >5 ml/min
Tygon 3350 Tuning	Saint Gobain	ABW00001	ID: 1/32" OD: 3/32" L:50'
Magnetic Stir bar	COWIE		tapered shape dim: 10 mm x 4 mm
Glass scintillation 20 ml vial	DGS		Pyrex glass 28 mm (diameter) x 61 mm (height)
Culture vial holder			Custom made from Polyformaldehyde
Silicone	Dow Corning	Sylgald 184	used to seal the glass vial
Medium bottle	VWR	66022-065
Difco M9 minimal salt 5x	BD		Medium
Cadamino Acid	BD		Medium
glucose	Sigma
Agar Bateriological	Oxoid		for agar plate
Luria Bertani medium
Inverted microscope	Leica Microsystems	Leica DMI-LED	used for microfluidic measurement Use 40X objective NA = 0.55
Microscope Incubator	Live Cell Instrument	CU-109	used for microfluidic measurement
Solenoidal valves	Pneumadyne	S10MM-31-12-3	Normally open 1.3 Watt 12 Vdc
USB interface card	Hobby Engineering	USBIO24-R Digital I/O Module	for microfluidics measurement
Air compressor	Rocker Scientific	ROCKER 440	Pressure source for microfluidcs; Max. Pressure 80 psi
Male luer-lock fittings to 1/8" barb	ValuePlastics.com	MTLL230-1	used for microfluidic control
1/8" barb to 10-32 threaded port	ValuePlastics.com	B-1	used for microfluidic control
Female luer-lock fittings to 10-32 threaded port	ValuePlastics.com	KFTL-1	used for microfluidic control
NPN darlington transistor 500 mA, 40 V (2N6427)	DigiKey.com	2N6427GOS-ND	used for microfluidic control
10 kOhm, carbon film resistor, 0.25 W	DigiKey.com	P10KBACT-ND	used for microfluidic control
Tantalum capacitor, 10 μF, 25 V, 10%	DigiKey.com	478-1841-ND	used for microfluidic control
Andor CCD camera	Andor	Zyla 4.2 Plus SCMOS	used for microfluidic on chip imaging
ELISA plate reader
two component Silicone	Momentive	RTV 615	used for microfluidic chip fabrication
SU-8 photoresist	Micrchem	SU8 2015	used for microfluidic chip fabrication
AZ4620 photoresist	Clariant	AZ 4620	used for microfluidic chip fabrication
Plasma cleaner	Harrick Plasma	PDC 32G	used for microfluidic chip fabrication
20 Gauge Syringe Needle	BD		used for microfluidic chip fabrication
Labcycler	Sensoquest	Labcycler	PCR
DNA polymerase	Toyobo	KDO Plus	PCR amplification
Trimethoprim	Sigma
Plate reader	Biotek	Synergy H1 hybrid	antibiotic resistane measurement