Non-genetic adaptation by collective migration

Cell populations must adjust their phenotypic composition to adapt to changing environments. One adaptation strategy is to maintain distinct phenotypic subsets within the population and to modulate their relative abundances via gene regulation. Another strategy involves genetic mutations, which can be augmented by stress-response pathways. Here, we studied how a migrating bacterial population regulates its phenotypic distribution to traverse diverse environments. We generated isogenic Escherichia coli populations with varying distributions of swimming behaviors and observed their phenotype distributions during migration in liquid and porous environments. We found that the migrating populations became enriched with high-performing swimming phenotypes in each environment, allowing the populations to adapt without requiring mutations or gene regulation. This adaptation is dynamic and rapid, reversing in a few doubling times when migration ceases. By measuring the chemoreceptor abundance distributions during migration towards different attractants, we demonstrated that adaptation acts on multiple chemotaxis-related traits simultaneously. These measurements are consistent with a general mechanism in which adaptation results from a balance between cell growth generating diversity and collective migration eliminating under-performing phenotypes. Thus, collective migration enables cell populations with continuous, multi-dimensional phenotypes to flexibly and rapidly adapt their phenotypic composition to diverse environmental conditions.

Methods
Growth rates measurements

Growth measurements were done using a microplate reader (BioTek Epoch 2 microplate spectrophotometer). Overnight cultures were diluted by 100-fold into fresh media supplemented with appropriate inducer concentrations. 200 μL of diluted cells were aliquoted into each well of polystyrene 96-well plates (Falcon 96-wells REF 353072), and the plates were loaded into the plate reader. The plates were incubated at 30oC while continuously shaking linearly at 567 cpm. The OD600 of each well was measured every 7 minutes for 36 hours. To extract the growth rates of each sample, the slopes of the linear region (at the mid-exponential growth phases) of the natural log of OD600 versus time (hours) curves were calculated.

Formation of propagating waves and expansion speed measurements using swim plate and capillary tube assays

For the swim plate assay, 3 µL of cells in exponential phase were inoculated at the center of semi-solid agarose plates for RP437-derived strains or 10  of cells in exponential phase for MG1655-derived strains. For RP437-derived strains, the semi-solid agarose plates were made using M9-based media, as described above, supplemented with 0.14% agarose (American Bioanalytical Agarose GPG/LE) and appropriate concentrations of aTc. Similarly, for MG1655-derived strains, the semi-solid agarose plates were made using H1 minimal media, as described above, supplemented with 0.14% agarose. The plates were incubated at 30oC for 15 hours (RP437-derived strains) or at 33.5oC for 48 hours (MG1655-derived strains). The edges of the migrating colonies were picked using pipette tips and diluted into chemotaxis buffer (for RP437-derived strains) or minimal chemotaxis buffer (for MG1655-derived strains) for subsequent tracking and imaging analyses. To measure the expansion speeds of bacterial populations in swim plates, images were taken with a Canon DS126291 camera every 30 minutes. The diameters of the expansion rings as a function of time and expansion speeds were extracted from the images using a customized MATLAB code.

Overnight cultures were inoculated 100-fold into 5 mL of media with appropriate inducers and grown at 30oC until OD600 = 0.2. Cells were concentrated to an OD600 = 6. A 12-inch capillary tube was filled with growth media (M9 glycerol with casamino acids, MgSO4, and PVP-40) and 6-hours aged aTc, and one side of the tube was plugged by stabbing the tube into a plate filled with solidified agar (1.5% agar in distilled water). The concentrated cell mixture was filled on the other side of the tube using a 1 mL syringe, and another agar plug was added into that side. Both sides of the tube were then plugged with clay. The tubes are incubated horizontally at 30oC for 9 hours. To isolate the wave, a 3-cm section of the tube that contained the wave was fractionated, and the mixture inside the 3-cm section was diluted into chemotaxis buffer for subsequent tracking analysis. To measure the expansion speeds of bacterial populations in capillary tubes, images were taken with a Canon DS126291 camera every 10 minutes. Positions of the populations as a function of time and expansion speeds were extracted from images using ImageJ.

Tracking single cells and tumble detection

Washed cells were diluted with chemotaxis buffer to an OD600 = 0.0005. The diluted mixture was injected inside the microfluidic device using a 1 mL syringe. To prevent evaporation, holes on both sides of the microfluidic device were taped. The device was placed on an inverted microscope (Nikon Eclipse Ti-U) equipped with a custom environmental chamber (50% humidity and 30oC). A custom MATLAB script was used to record 3-minute phase contrast videos at 4X magnification and 20 frames-per-second. Data from video recordings were stored as .bin files.

A customized MATLAB code was used to detect tumble and extract tumble bias of single cells, as previously described58. Only trajectories longer than 10 seconds were considered for analysis. Each tumble bias distribution was generated using 4 3-minute videos that contained ~1000 total trajectories.

Measuring the time scales of relaxation of TB distribution to the standing batch distribution

E. coli RP437 or HE205 were grown to mid-exponential phases and let to migrate in capillary assays for 9 hours, as mentioned above. The sections of the capillary tubes containing the migrating bacterial populations were fractioned, and the cells were added into 1 mL of growth media.  For every 15- or 30-minutes during growth in the batch cultures, the TB distributions of the populations were measured. The mean of TB distributions normalized by the trajectory lengths were determined for each time points.

The equation to model for the relaxation of mean TB is:

where  is the mean TB over time,  is the mean TB of the batch culture,  is the mean TB during collective migration in the capillary tubes, and  is the relaxation timescale. To estimate , we fitted the means of the mean TBs across replicates and the standard error of the means across replicates to the equation using maximum likelihood estimation. To estimate the error of , we sampled the posterior distribution using Markov Chain Monte Carlo simulation to get 100 simulated parameter values and calculate the standard deviation of those values.

Protein expression measurements in swim plates

Non-motile MG1655-derived cells expressing labeled Tsr or non-motile RP437-derived cells expressing labeled CheY and CheZ used as biosensors for gene expression were grown as described above. Subsequently, cells were mixed with the appropriate media and 0.14% liquified agarose to a final OD600 = 0.01. Swim plates were left to solidify at room temperature for 2-3 hours, and afterward, 10 μL of OD600 = 0.01 motile WT MG1655 (labeled Tsr experiments) or motile WT RP437 (labeled CheY and CheZ experiments) cells were inoculated in the center of the plate. The motile population of cells shapes the gradient of the plate and ensures that the non-motile biosensor strain experiences the same local environment as motile cells. After a colony of migrating cells has formed, cells from the center and the edge of the colony are picked and diluted into minimal chemotaxis buffer for subsequent imaging.

Single-cell protein copy number quantification using fluorescent microscopy

Washed cells were diluted 10-fold and plated on agarose pads (3% agarose in minimal chemotaxis buffer). The cell suspension was left to dry for 10 minutes, and cells were immediately imaged afterward. Imaging was performed using an inverted microscope (Nikon Eclipse Ti-E) equipped with an oil immersion 100x phase-contrast objective lens. To measure the fluorescence intensity of the strain expressing cheY-mRFP and cheZ-mYFP, fluorescent proteins were excited using a light-emitting diode illumination system (pE-4000, CoolLED). The RFP and YFP channels were imaged consecutively using a multi-band filter (Semrock), and the fluorescence emissions were led into a 2-camera image splitter (TwinCam, Cairn), leading to two identical sCMOS cameras (ORCA-Flash4.0 V2, Hamamatsu). The same setup was used to measure the fluorescence intensity of the strain expressing Tsr-mYFP. In a typical experiment, 20 fields of view, containing approximately 1000 cells in total, are imaged.

Fluorescent image analysis for protein copy number quantification

For each field of view, one phase contrast, one YFP, and one RFP channel images were acquired. Using a custom MATLAB script, images from the two cameras were aligned using an affine transformation, and cells were segmented on the phase-contrast channel using a modified Otsu algorithm. Then, the image background was subtracted from each image, and the fluorescent intensity of every cell was extracted. Finally, the area of each cell was calculated by fitting an ellipsoid. The fluorescent intensity of each cell was divided by the cell area as a measure of protein concentration.