A Continuous Culture Millifluidic Device for the Study of Escherichia coli under Low and Fluctuating Nutrient Conditions

Summary

We present a protocol for the millifluidic continuous culture device (MCCD), which allows the cultivation of bacterial populations under constant low-nutrient conditions (down to hundreds of nanomolar concentrations) or temporal nutrient fluctuations on a minute timescale.

Abstract

Microbial habitats in nature are often characterized by low concentrations of mixed nutrients, spatial heterogeneity, and temporal fluctuations. However, traditional laboratory culturing methods fail to replicate these conditions. Batch cultures cannot sustain growth in low-nutrient environments, while chemostats maintain steady-state growth with a single limiting nutrient but are challenging to implement when the goal is to maintain defined low concentrations of nutrient mixtures or to introduce rapid fluctuations. Microfluidic systems generate dynamic environments but yield insufficient biomass for population-level omic analyses. To address these limitations, we introduce the millifluidic continuous culture device (MCCD), a versatile platform for studying microbial responses to stable and fluctuating nutrient conditions. The MCCD houses bacterial populations inside a Sterivex filter (0.45 µm polyvinylidene fluoride [PVDF] porous filtering membrane), where a continuous flow of media sustains stable culture conditions while preventing nutrient depletion. A three-way solenoid valve system, controlled via custom Matlab software, enables precise, minute-scale nutrient fluctuations. This protocol provides a step-by-step guide to operating the MCCD in two modes: (1) constant low-nutrient conditions and (2) fluctuating-nutrient conditions. Using this system, Escherichia coli grew exponentially in a mixture of amino acids and nucleobases present at tens to hundreds of nanomolar concentrations, reaching cell concentrations on the order of 10⁹ cells/mL. By recreating key features of natural microbial habitats, the MCCD enables the study of bacterial growth and physiology under controlled yet ecologically relevant conditions in E. coli and other microbial species.

Introduction

Bacteria in natural habitats encounter a nutrient landscape that often varies in both composition and concentration and can fluctuate across a broad range of timescales^1,2,3,4. In environments such as the ocean⁵, soil³, and gut⁶, microbial growth is frequently constrained by one or more essential nutrients, and nutrient availability can change due to physical, chemical, and biological processes⁴. These nutrient mixtures and their fluctuations drive microbial strategies for survival and growth, shaping responses across different levels of biological organization⁴.

Microbial adaptations to nutrient dynamics occur at the community, population, and single-cell levels. At the community level, fluctuations can promote species coexistence⁷, coordinate community composition and metabolic activity with temporal variation⁸, and create opportunities for new species to colonize⁹. At the population level, changing nutrient conditions can accelerate evolution by selecting for genetic changes¹⁰ or induce phenotypic variability, including bet-hedging strategies¹¹, that improve survival under unpredictable conditions. At the single-cell level, bacteria adjust their physiology in response to the nutrient environment^12,13. One recently reported example is a specific physiology that confers bacteria a growth advantage under fluctuating conditions¹⁴. Because bacteria frequently encounter nutrient limitation, studying how they respond to low and variable nutrient conditions—both as individual cells and within populations or communities— is essential for understanding their behavior in their natural environments.

Chemostats are well-suited for studying growth under limitation by a single nutrient, but are difficult to use when the goal is to control the concentrations of multiple nutrients in low-concentration mixtures. Chemostats have been valuable tools for studying bacterial adaptation to nutrient limitation by providing a well-controlled, steady environment. However, their operation inherently restricts them to conditions in which growth is limited by a single essential nutrient element. In a chemostat, the dilution rate is adjusted so that cellular growth balances the supply of one limiting nutrient, thereby fixing the growth rate rather than the concentrations of individual nutrients¹⁵. This framework does not readily generalize to environments containing mixtures of substitutable nutrients—such as multiple carbon sources—because the dilution rate controls only the overall growth rate, not the concentrations of each component in the medium^16,17. As a result, the concentrations of individual nutrients and their co-utilization patterns emerge from microbial utilization rather than being externally prescribed^16,17.

Additionally, while nutrient fluctuations can be induced in chemostats, rapid fluctuations or single shifts on the scale of minutes are difficult to achieve due to the typically large culture volumes. Microfluidic devices have emerged as powerful tools for creating precise low-concentration and fluctuating nutrient environments¹⁴. When paired with video microscopy, they enable measurement of single-cell growth. Although microfluidic systems permit quantification of the expression of selected proteins using fluorescent reporters, unlike chemostats, typical microfluidic devices are too small to produce the biomass needed for omic analyses, for example, to study the proteome of a bacterial population. To bridge this gap, we developed the millifluidic continuous culture device (MCCD).

The MCCD enables bacterial growth under controlled nutrient conditions, including constant low-nutrient conditions with minimal or no depletion, as well as fluctuating nutrient conditions, using a continuous flow culture system capable of generating sufficient biomass for downstream analyses. The MCCD uses a 0.45 µm PVDF filter as the culture vessel, in which medium is continuously replenished by a peristaltic pump delivering fresh medium from a reservoir (Figure 1A). The porous membrane of the filter retains bacteria while allowing a prescribed flow of nutrients, which prevents nutrient depletion even at low nutrient concentrations (hundreds of nanomolar to micromolar range). Because biomass is retained while fresh medium continuously flows through the device, the MCCD shares similarities with retentostats¹⁸, in which cells are retained while medium is renewed. However, the MCCD differs from classical retentostats in its architecture and mode of nutrient delivery. The retentostat typically operates as a stirred bulk reactor, where cells are suspended in a well-mixed culture volume. Under such conditions, nutrient availability depends on the balance between mixing, diffusion, and cellular uptake, and local depletion or gradients can arise, particularly at low nutrient concentrations and high cell densities. In contrast, in the MCCD, cells assemble directly on the membrane inside the small-volume Sterivex cartridge, forming layers only a few cells thick that are continuously exposed to flowing medium. No active mixing is imposed within the cartridge; instead, medium is homogenized upstream and delivered by continuous flow. Nutrients are supplied by convective transport through the thin cell layers, and because the diffusion distance is small and the flow rate is high, local nutrient gradients are minimized. As a result, the nutrient conditions experienced by the population are primarily determined by the composition of the incoming medium, enabling more consistent and well-defined low-nutrient conditions than can be ensured in stirred retention systems.

Figure 1: The millifluidic continuous culture device (MCCD) for bacterial culture under low-concentration or fluctuating nutrient conditions. (A) Schematic of the MCCD for constant nutrient conditions. A peristaltic pump drives the flow of fresh medium from a reservoir through a 0.45 µm PVDF filter containing bacterial cells. The porous membrane in the filter retains the bacteria while allowing medium to pass through. Continuous medium flow maintains constant nutrient conditions inside the filter, even at very low nutrient concentrations. (B) Schematic of the MCCD for fluctuating nutrient conditions. A software-controlled three-way solenoid valve upstream of the peristaltic pump alternates medium inflow between two reservoirs (e.g., high-nutrient concentration, C_High, and low-nutrient concentration, C_Low). The software controls the valve's plunger position to determine the active inflow source. (C) Fluidic assembly for constant nutrient conditions. A cap with three silicone tubes that traverse it seals the bottle containing fresh medium. One tube connects to a vent filter for aeration, while the other two carry medium to the filters. Barb-to-Male Luer fittings connect these tubes to the peristaltic tubing, and connect each peristaltic tube to a one-way check valve, serving as the entry point to the filter. The filter outlet is also fitted with a one-way check valve. The image shows two samples, each contained in one filter. (D) Fluidic assembly for fluctuating nutrient conditions. The metal platform houses the microcontroller and electronic components controlling two solenoid valves via custom MATLAB software. Threaded-to-Barb fittings connect both the inlets and outlets of the solenoid valves to the tubing. Each valve has two inlets for the two alternate media, here a high-nutrient concentration medium, C_High, red-labeled tubes, and a low-nutrient concentration medium, C_Low, blue-labeled tubes. The media are contained within capped bottles with a vent filter, as in (C). The outlets from the solenoid valves connect to the filters via peristaltic tubing, as in the steady-state assembly (C). The image shows two samples, each contained in one filter. Please click here to view a larger version of this figure.

Additionally, the incorporation of a software-controlled, three-way solenoid valve upstream of the Sterivex filter enables the user to switch the medium flow between two media contained in different reservoirs (Figure 1B). As a result, the device can be used to study the response of bacteria to either a stable environment at a constant (including low concentration) nutrient condition or to nutrient fluctuations, with fluctuation timescales of several minutes or longer.

This protocol outlines the assembly and operation of the MCCD for studying bacteria under both low-nutrient and fluctuating-nutrient conditions. As an example, this protocol provides a step-by-step guide to conducting a growth experiment using Escherichia coli NCM3722 in two operating modes of the MCCD: constant low-nutrient conditions and fluctuating nutrient conditions. The MCCD setup can be readily adapted to other nutrient conditions and microbial species, making it a versatile platform for studying bacterial physiology under nutrient environments typical of those encountered in nature.

Protocol

1. Bacterial growth in constant nutrient conditions

NOTE: This protocol describes the procedure for performing a growth curve of E. coli NCM3722 in a very dilute Rich-Defined Medium ("low RDM") condition. This growth curve runs for 12 h, with one sample collected every 2 h (a total of 6 time points). The key principle of this protocol is that each time point corresponds to a single Sterivex filter (henceforth referred to as a 0.45 µm PVDF filter). To collect a sample, the culture inside the filter is retrieved by centrifuging it upside down, the cell number is quantified, and the filter is discarded. Each 0.45 µm PVDF filter yields only one sample. Preparing the experiment requires the definition of three key experimental parameters: the flow rate of the peristaltic pump, the duration of the experiment, and the desired number of samples (i.e., the number of 0.45 µm PVDF filters). Set the initial flow rate to 2 mL/min. Adjust this parameter as needed (e.g., increase the flow rate to maintain the supply of a very dilute nutrient).

1. Preparation of medium and materials
   NOTE: Since each filter corresponds to a single time point sample, the experimental runtime of each 0.45 µm PVDF filter varies depending on its collection time (e.g., 2 h for the first time point, 4 h for the second time point, and so on).
   1. Estimate the required culture medium volume using the formula
      ,
      where n is the number of samples (0.45 µm PVDF filters), Q is the flow rate (in mL/min), and t_i is the experimental runtime (in minutes) for each sample.
   2. Multiply by 1.2 to include an additional 20% volume to account for flow-rate variation. For this example, calculate
      
   3. Round the calculated volume up (e.g., prepare 6,200 mL in this example).
   4. Prepare material for sterilization (see Table 1).
      NOTE: Prepare and sterilize the materials the day before the experiment to ensure readiness and minimize delays. Autoclave these materials, except the one-way Luer valves; sterilize the valves with 70% ethanol.
      1. Connect both ends of six silicone peristaltic tubes to "Male Luer to Barb" plastic fittings by inserting the Barb end of the fitting into the tubing. Cover the Male Luer ends with aluminum foil to prevent contamination.
         NOTE: Six peristaltic tubes are prepared because the experiment uses six 0.45 µm PVDF filters. All tubes will later be mounted on the same peristaltic pump head so that all filters receive identical flow.
      2. Place the tubing inside a closed, sterilizable container.
      3. Place the vent filters inside a closed, sterilizable container.
      4. Place the tubed bottle caps with silicone tubing that passes through the caps inside a closed, sterilizable container.
         NOTE: The tubed bottle caps used in this protocol are equipped with three tubes that traverse the caps: two tubes of equal diameter for delivering medium to the 0.45 µm PVDF filters, and a third tube with a wider diameter. These tubes are securely attached to the caps and cannot be removed.
      5. Submerge 12 one-way Luer check valves in a 70% ethanol solution and leave them overnight to sterilize.
         NOTE: (PAUSE POINT) Sterilized materials can be stored submerged and sealed at room temperature for up to 24 h before the experiment. Before use, remove the valves from the ethanol and rinse with sterile ultra-pure water to remove residual ethanol. Alternatively, check valves can be 3D-printed using autoclavable materials to allow sterilization by autoclaving instead of ethanol treatment.
   5. Autoclave the peristaltic tubing, vent filters, bottle caps, graduated cylinder, and bottles using standard labware settings.
   6. Prepare 6,200 mL of low rich-defined media (RDM; see Table of Materials) by mixing the reagents listed in Table 2.
      NOTE: Prepare the medium on the day of the experiment, during the approximately 5 h waiting period after starting the pre-culture (see step 1.2.1). This preserves medium freshness and minimizes contamination risk.
2. Preparation of a bacterial pre-culture
   NOTE: Use a pre-culture in the exponential phase to start the experiment. During this phase, bacteria grow at a constant rate, synthesizing cell constituents in a balanced manner and dividing at a defined cell mass. Use exponential-phase cells to minimize variability and standardize the starting physiological state.
   1. The day before the experiment, prepare an overnight (ON) culture as follows: Inoculate 3 mL of full RDM with a frozen glycerol stock of E. coli NCM3722 in a sterile 15 mL culture tube. Incubate the culture overnight (~16 h) at 37 °C with shaking at 200 rpm in a shaking incubator.
      NOTE: (PAUSE POINT) Once the ON culture is in the incubator, no further steps are followed on this day. Resume the protocol the next day by proceeding to step 1.2.2.
   2. On the day of the experiment, prepare the pre-culture as follows: Add 6 µL of ON culture to 6 mL of full RDM. Divide equally into two 15 mL culture tubes and label the tubes as A and B. Incubate both tubes at 37 °C with shaking at 200 rpm.
      NOTE: The pre-culture typically requires approximately 5 h to reach the exponential phase (OD₆₀₀ ≈ 0.5). Use this period to prepare the fresh medium (see step 1.1.6). Do not proceed to step 1.2.3 until OD₆₀₀ reaches approximately 0.5.
   3. Monitor optical density (OD) at 600 nm using tube B. Proceed with the experiment when OD₆₀₀ reaches approximately 0.5, indicating exponential growth in full RDM for E. coli NCM3722 (see section 1.4).
      NOTE: Start the experiment with exponential-phase cells to improve reproducibility and standardize starting conditions.
3. Preparation of an experiment in the MCCD with constant nutrient conditions
   ​NOTE: Perform steps 1.3.1 to 1.3.5 inside a sterile biosafety cabinet immediately after the pre-culture begins growing.
   1. Allocate the total medium volume to each bottle.
      NOTE: Use the bottles as medium reservoirs. Use bottle caps fitted with two medium-delivery tubes, with each tube delivering medium to one 0.45 µm PVDF filter. Allocate the total medium volume (prepared in step 1.1.6) among the bottles based on the amount required by each timepoint sample (i.e., each 0.45 µm PVDF filter) over the course of the experiment.
      1. Use the formula provided in step 1.1.1 to determine the volume of medium needed for each 0.45 µm PVDF filter. For example, using the following terms:
         
         The total medium volumes required for each time point, in chronological order, are 288, 576, 864, 1152, 1440, and 1728 mL.
      2. Distribute the calculated medium volumes across bottles, grouping the volumes needed for two time points into a single bottle. For example: ~900 mL in one 1 L bottle (for time points 1 and 2), ~2,100 mL in one 2.5 L bottle (for time points 3 and 4), ~3200 mL of the medium in one 4 L bottle (for time points 5 and 6).
   2. Seal each bottle with the tubed bottle caps (Figure 1C). Tighten the caps securely to prevent leaks.
   3. Connect a 0.2 µm vent filter to the widest silicone tube extending from the tubed bottle cap (Figure 1C). This allows sterile air exchange during the experiment.
   4. Attach one end of each peristaltic tube (with fittings already attached; see section 1.1.4 on material preparation) to the smaller-diameter silicone tubes extending from the tubed bottle caps (Figure 1C). Ensure secure connections to maintain sterility. Leave the aluminum foil cover on the free Male Luer lock fitting in place until the experiment begins to prevent contamination.
   5. Place the bottles with their assembled tubing and the peristaltic pump into an incubator set to 37 °C. Secure the tubing into the channels of the peristaltic pump head.
      NOTE: All peristaltic tubes are mounted on the same pump head of a single peristaltic pump, so that all 0.45 µm PVDF filters receive medium from the same pump. Using a single pump ensures that the flow rate is identical across filters and prevents variability that could arise from differences between pumps.
4. Protocol of an experiment in the MCCD with constant nutrient conditions
   1. Dilute the pre-culture (OD = 0.5) 1:10 by adding 600 µL of pre-culture to 5.4 mL of fresh low RDM.
   2. Using a 1 mL pipette, introduce 1 mL of the diluted pre-culture into each of the six 0.45 µm PVDF filters.
   3. Rinse the one-way check valves with autoclaved ultrapure water and attach one to each end of each 0.45 µm PVDF filter.
   4. Remove the aluminum foil covers from the peristaltic tubing fittings.
   5. Connect each of the six 0.45 µm PVDF filters to a Male Luer fitting on the peristaltic tubing (Figure 1C).
   6. Set the flow rate to 2 mL/min on the peristaltic pump and start the flow. Define this moment as time 0.
   7. Place a waste receptacle below the 0.45 µm PVDF filters to collect the outflow.
      NOTE: Figure 2A shows the complete assembly of the experiment inside the incubator box.
5. Sample collection
   1. After 2 h, unplug one of the two 0.45 µm PVDF filters connected to the 1 L bottle.
   2. Unscrew the one-way check valves and place the 0.45 µm PVDF filter into a 50-mL centrifuge tube with the inlet facing downward.
   3. Centrifuge the centrifuge tube containing the 0.45 µm PVDF filter at 2,000 × g for 2 min.
   4. Using a pipe cutter, apply gentle pressure to the plastic encasing at the end closest to the outlet to open the 0.45 µm PVDF filter.
      NOTE: Disinfect the pipe cutter before use (e.g., by wiping with 70% ethanol) to avoid contamination when opening the 0.45 µm PVDF filter.
   5. Release the inner cylinder containing the porous membrane and discard the section of the filter that holds the membrane.
   6. Rinse the inner walls of the 0.45 µm PVDF filter encasing (those perpendicular to the flow direction) with the culture medium collected in the centrifuge tube.
      ​NOTE: This step ensures that any cells trapped on the walls during centrifugation are collected.
   7. Measure the total volume collected using a 5 mL serological pipette.
   8. At subsequent time points (4 h, 6 h, 8 h, 10 h, and 12 h), unplug the corresponding 0.45 µm PVDF filters and repeat steps 1.5.2–1.5.6. Follow this specific order:
      1. At 4 h, unplug the second 0.45 µm PVDF filter connected to the 1 L bottle.
      2. At 6 h and 8 h, unplug the 0.45 µm PVDF filters connected to the 2.5 L bottle.
      3. At 10 h and 12 h, unplug the 0.45 µm PVDF filters connected to the 4 L bottle.
         NOTE: Follow this order to match sampling time points to the medium volumes distributed in step 1.3.1.
6. Downstream analyses
   1. Take three 100 µL aliquots from each sample and fix with glutaraldehyde to a final concentration of 1.5%.
      CAUTION: Glutaraldehyde is toxic, a strong irritant, and a potential sensitizer. Handle inside a chemical fume hood while wearing appropriate personal protective equipment (lab coat, nitrile gloves, and safety glasses). Avoid skin contact and inhalation. Dispose of glutaraldehyde-containing waste according to institutional hazardous chemical waste regulations.
   2. Use the fixed samples to count the number of cells per microliter with flow cytometry after labeling with a fluorescent DNA dye.
      CAUTION: Fluorescent DNA dyes may be mutagenic. Wear gloves and dispose of dye-containing waste according to institutional biosafety regulations.
      ​NOTE: As an alternative to DNA staining and flow cytometry, cell densities can be estimated by plating a small volume of the sample on rich-medium agar plates (e.g., LB agar) and counting colony-forming units (CFUs).
   3. Multiply the cell number per microliter by the total volume measured in step 1.5.7 to calculate the total number of cells contained within the 0.45 µm PVDF filter.
   4. Centrifuge the remaining volume in 1.5-mL microcentrifuge tubes at 10,000 × g for 4 min.
   5. Remove the supernatant.
   6. Store the pellets of cells at -80 °C for future analyses (e.g., transcriptomics).

Figure 2: Experimental assembly of the MCCD. (A) Photograph of the MCCD setup for constant nutrient conditions. The setup is housed in an incubator box to maintain a constant temperature, with a waste receptacle collecting the flow-through medium. (B) Photograph of the MCCD setup for fluctuating nutrient conditions. The bottle labeled with red tape contains a high-nutrient concentration medium (C_High), while the bottle labeled with blue tape contains a low-nutrient concentration medium (C_Low). The metal platform in the center supports the components that control nutrient fluctuations, corresponding to those described in Figure 1D. Please click here to view a larger version of this figure.

2. Bacterial growth in fluctuating-nutrient conditions

1. Assembly of a controller for a solenoid valve
   NOTE: Control the three-way solenoid valve using a microcontroller-based switching circuit connected to a computer running a custom graphical user interface (GUI) developed in MATLAB.
   1. Assemble the solenoid valve control circuit (Figure 3A).
      1. Connect a digital output pin of the microcontroller to the base of an NPN transistor through a 1 kΩ resistor.
      2. Connect the collector of the transistor to the negative terminal of the solenoid valve.
      3. Connect the emitter of the transistor to ground.
      4. Connect the positive terminal of the solenoid valve to an external DC power supply compatible with the valve voltage.
      5. Place a diode across the solenoid terminals. Connect the striped side of the diode to the positive supply line.
         NOTE: The diode suppresses voltage spikes generated when the solenoid is switched off.
   2. Attach plastic "Threaded to Hose Barb" fittings to the solenoid valves, ensuring the Barb end extends outward.
   3. Install MATLAB R2022b or later on a computer with USB connectivity.
   4. Install the MATLAB support package required for serial communication with microcontroller hardware.
   5. Connect the microcontroller board to the computer using a USB cable to establish serial communication.
   6. Open the MATLAB-based GUI used to generate fluctuating nutrient conditions in the MCCD (available in the link https://github.com/juanitaLara/MCCD).
   7. Verify the connection between the solenoid valves and the MATLAB GUI platform:
      1. Open the GUI platform in Matlab and click on the Connect button (Figure 3B). Wait until the message Controller connected successfully appears in the command window.
      2. Input a value for the time step (e.g., 5 s for the High and Low phases) under the Automated operation panel.
      3. Observe the solenoid valves and listen for the click sound they produce during switching between the two media.
      4. Using a watch, confirm that the solenoid valves switch exactly at the specified interval (e.g., every 5 s).
         NOTE: Perform this verification before each experiment to confirm communication and correct valve switching. (PAUSE POINT) After verifying solenoid valve operation, the system can remain assembled inside the incubator for several hours before starting the experiment.
2. Preparation of media and materials
   NOTE: This protocol involves alternating nutrient conditions between low and high RDM with a period of 1 h (30 min in high RDM followed by 30 min in low RDM). The experiment runs for three h of fluctuations, with two samples collected: (1) At 3 h 1 min (high RDM phase, immediately after switching from low to high), and (2) at 3 h 31 min (low RDM phase, immediately after switching from high to low).
   1. Estimate the required culture media volumes by experimental phases.
      1. First, calculate the volumes required for the first 3 h (three full fluctuation periods):
         low RDM volume =1.2 × 2[mL/min] × 90[min] × 2[0.45 µm PVDF filters] = 432[mL]
         high RDM volume =1.2 × 2[mL/min] × 90[min] × 2[0.45 µm PVDF filters] = 432[mL]
      2. Calculate the volume required for the final 30 min phase:
         high RDM volume =1.2 × 2[mL/min] × 30[min] × 1[0.45 µm PVDF filters] = 72[mL]
      3. Round the volumes to convenient values (e.g., 500 mL of low RDM and 600 mL of high RDM).
   2. Prepare materials for autoclave sterilization (see Table 3) by following steps 1.1.4 –1.1.5.
      NOTE: Prepare and sterilize the materials the day before the experiment to ensure readiness and minimize delays on the experiment day (steps 2.2.3 and 2.2.4). This setup, in comparison with the constant-nutrient experiments, requires additional tubing to connect media bottles directly to the solenoid valve (Figure 1B,D). These tubes require only one "Barb to Male Luer" fitting, as the solenoid valve fittings are pre-assembled. Cover the ends of these tubes with aluminum foil before autoclaving.
   3. Sterilize the prepared peristaltic tubing, vent filters, caps, graduated cylinder, and bottles in an autoclave.
   4. Submerge 4 one-way Luer check valves in a 70% ethanol solution and leave them overnight to sterilize.
      NOTE: (PAUSE POINT) Sterilized check valves can remain submerged in ethanol overnight and stored sealed at room temperature for up to 24 h. Before use, rinse the valves with sterile Milli-Q water.
   5. Clean and sterilize the solenoid valves by flushing them with 70% ethanol 3–5 times, followed by ultrapure water 3–5 times.
   6. On the day of the experiment, during the approximately 5-h waiting period after pre-culture preparation (see step 2.3), prepare 500 mL of low RDM and 600 mL of high RDM by mixing the reagents listed in Table 4.
      NOTE: Preparing the media on the day of the experiment ensures freshness and minimizes contamination risks.
3. Preparation of a bacterial pre-culture
   1. Follow the steps described in step 1.2.
4. Preparation of an experiment in the MCCD with fluctuating-nutrient conditions
   NOTE: Perform steps 2.4.1–2.4.4 inside a sterile biosafety cabinet immediately after the pre-culture begins growing.
   1. Place each medium into a 1 L bottle.
   2. Seal the bottles with the pre-assembled caps fitted with three ports and pre-installed silicone tubing. Tighten the caps securely to prevent leaks.
   3. Connect a 0.2 µm vent filter to the wide silicone tube extending from one of the ports on each bottle cap.
   4. Connect the "inlet tubing" (shown in Figure 1D) to the silicone tubes extending from the tubed bottle caps before connecting it to the solenoid valves:
      1. Attach one end of the silicone tube (with only one attached fitting) to the tubing extending from the low RDM bottle cap.
      2. Leave the aluminum foil cover on the free Male Luer lock fitting until the bottles are taken to the incubator to prevent contamination.
      3. Repeat step 2.4.4.1 for the high RDM bottle.
   5. Place the bottles with their assembled "inlet tubing", the peristaltic pump, and the solenoid valves with the controller inside an incubator set to 37 °C.
   6. Connect the "inlet tubing" (attached in step 2.4.4) to the inlets of the solenoid valves in the following way:
      1. Connect the two tubes from the low RDM bottle to one of the two inlets in each solenoid valve (tubes labeled with blue tape in Figure 1D).
      2. Connect the two tubes from the high RDM bottle to one of the two inlets in each solenoid valve (tubes labeled with red tape in Figure 1D).
   7. Connect a peristaltic pump-compatible silicone tube (with only one fitting) to the outlet of each solenoid valve (Figure 1D). Leave the aluminum foil cover on the free Male Luer lock fitting until the experiment begins to maintain sterility.
5. Protocol of an experiment in the MCCD with fluctuating-nutrient conditions
   1. Verify the connection between the solenoid valves and the Matlab software by performing step 2.1.7.
   2. Input the desired time step values for the fluctuations into the Matlab GUI platform: Enter time steps of 1800 s in High RDM (30 min) and 1800 s in Low RDM (30 min).
   3. Select Low under the Start from menu to begin in the Low RDM condition (Figure 3).
   4. Seed the two 0.45 µm PVDF filters as described in step 1.4.
   5. Start the experiment.
      NOTE: Figure 2B depicts the complete fluidic assembly for this experiment.
6. Sample collection
   1. Collect samples as in step 1.5. Harvest the first 0.45 µm PVDF filter at 3 h 1 min and the second at 3 h 31 min.
      NOTE: Unplug either filter first; both filters are fed by the same media.
7. Downstream analyses
   1. Follow the steps described in step 1.6 to count the number of cells via flow cytometry and to process and store samples for downstream analyses (e.g., transcriptomics).

Figure 3: Controller of the solenoid valves to impose nutrient fluctuations. (A) Wiring diagram of the electronic circuit used to control the solenoid valve. The circuit consists of a microcontroller board, a transistor used as a switching element, a resistor connected to the transistor base, and a flyback diode placed across the solenoid valve to protect the circuit from voltage spikes generated during valve switching. The microcontroller controls the transistor, which in turn switches the power supplied to the solenoid valve. Adapted with permission from ref. ²⁰ . (B) After connecting the microcontroller board to both the computer running the GUI and a power source with the voltage required by the solenoid valves, communication is established by clicking the Connect button. The valves can be operated manually using the Manual Operation panel, where the user selects the medium to be flown (e.g., "High" or "Low"). Alternatively, the Automated Operation panel allows the user to specify the initial medium and the time steps (in seconds) for each phase (High and Low), and to start the operation by clicking the Start button. The Real-time Monitor displays a step function (0 = Low, 1 = High) over time, indicating which medium was flowing through the solenoid valves to the filters over the preceding seconds. Please click here to view a larger version of this figure.

Results

In this study, we used the MCCD to quantify the growth rate of E. coli NCM3722 under constant low-nutrient conditions at three RDM concentrations. These concentrations were derived from the rich-defined medium (RDM) formula described by Neidhardt¹⁹, with the amino acid and nucleobase supplements diluted to 0.5%, 0.25%, and 0.01% of the original formulation, respectively.

Replicate growth experiments were performed in each medium to quantify growth rates. For th...

Discussion

Bacterial culturing methods that capture the nutrient complexity of natural environments are essential for advancing the study of microbial physiology beyond simplified laboratory conditions. Natural ecosystems contain mixtures of nutrients spanning a wide concentration range, which often do not support maximal growth rates^1,3,4,21,22,

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
1K Ohm Resistor	BC Robotics	ELC-021	Power Rating: 1/6 W
1N4001 Diode	BC Robotics	ELC-001	DC Reverse Voltage: 50 V; Average Rectified Current: 1 A
3-stop platinum-cured silicone peristaltic tubing	Innofluid	SE-TUB-SIL-SSS-2.4*0.8	Inner diameter 0.8 mm; outer diameter 2.4 mm; compatible with peristaltic pump used
3-way Solenoid valves	Emerson Electric Co.	0055A301L0T00F3	12 V DC, 1 W
Aluminum foil			Suitable for autoclaving
Arduino Uno Rev4	Arduino	SKU ABX00173
Autoclave	Systec	D-200	Or different company
Cap closure with three ports – for 1 and 2 L bottles	Nalgene	2162-0531	Or different company (make sure they fit the selected bottles)
Cap closure with three ports – for 4 L bottle	Nalgene	2162-0830	Or different company (make sure they fit the selected bottles)
Cell density meter CO8000	WPA biowave		OD meter
Centrifuge	Eppendorf	5804R	Centrifuge that fits 50 mL Falcon tubes
Centrifuge	Eppendorf	5424R	Centrifuge that fits 1.5 and 2 mL eppendorf tubes
Conical centrifuge tube - 50 mL	Fischer Scientific	10788561	Falcon tube
Culture tubes (14 mL, sterile)	Greiner bio-one		Culture tubes
Ethanol	Sigma Aldrich	493511	Dilute to 70%
Flow cytometer	Beckman	C09756	CYTOFlex
Glutaraldehyde 25%	Sigma Aldrich	G5882	Or different company
Graduated cylinder - 500 mL	Nalgene	3662-0500	Or different company
Incubator box			Custom-built incubator box – contact corresponding authors for construction details
LabV1 Intelligent Low Flow Rate Peristaltic Pump	Innofluid	SE-LABV1-AMC12(10)	Available at Darwin Microfluidics
Male Luer-to-barb fitting	Aldrich	Z277142	Male Luer-to-barb adapter compatible with 0.8 mm ID tubing (see Figure 1C)
Male Luer-to-threaded fitting	Aldrich	Z277142	Used to connect the solenoid valve to the "inlet tubing" (see Figure 1D)
Matlab	Mathworks	R2022b or later	Software to run the custom-made software for the solenoid valves
MOPS EZ Rich Defined Medium Kit	Teknova	M2105	MOPS medium for Enterobacteria
One-way Luer check valve	Masterflex	30505-92	Prevents backflow into medium reservoirs
Pipe cutter - 0-37 mm	KS Tools	2222051	Or different company
Pipette set	Fisher Scientific	05-403-151	Or different company
Platinum-Cured Silicone Tubing - 1/16 in	Nalgene	8060-0020	Inner diameter 1/16 in (1.6 mm); used for fluctuating-nutrient experiments
Platinum-Cured Silicone Tubing - 1/8 in	Nalgene	8060-0030	Inner diameter 1/8 in (3.2 mm); used for fluctuating-nutrient experiments
Polypropylene copolymer bottle - 1 L	Nalgene	2126-1000	Or different company
Polypropylene copolymer bottle - 2 L	Nalgene	2126-2000	Or different company
Polypropylene copolymer bottle - 4 L	Nalgene	2126-4000	Or different company
Serological pipette - 5 mL	Corning	CLS4487	Or different company
Solderless Breadboard	BC Robotics	PROTO-001
Sterivex-HV filter - 0.45 µm PVDF	Millipore	SVHV01015
SYBR Green I Nucleic Acid Gel Stain	Invitrogen	S7563	Nucleic acid stain for cell counting in flow cytometer
TIP120 Darlington Transistor	BC Robotics	TRN-005
Tubes - 1.5 mL	Eppendorf	3810X	Tubes for centrifuging cell cultures and storing pellets
Vent filter	Millex	SLFG05000	0.2 µm hydrophobic vent filter
Waste receptacle			Volume capacity should be at least the total media volume to be used in the experiment