Infection assay in the structured environment

The microfluidic mother machine device, with ten thousands of compartmentalised channels each hosting between 1 and 6 bacteria, was fabricated and assembled as previously reported [89,90]. Briefly, each channel had a cross section comparable to the size of individual E. coli cells (≈1 × 1 μm²) and a length of 25 μm. These channels were connected to a main microfluidic chamber that was 25 and 100 μm in height and width, respectively. A volume of 10 μl of 0.5 mg/ml bovine serum albumin was added to a 50-ml aliquot of an E. coli overnight and centrifuged for 5 min at 4,000 rpm and 20°C. The supernatant was filtered twice (Medical Millex-GS Filter, 0.22 μm, Millipore, Burlington, Massachusetts, USA), and 1 ml of this “spent” LB was used to resuspend the bacterial pellet to an OD_{600 nm} (optical density at 600 nm) of 50. This facilitated filling the dead-end channels of the mother machine, while maintaining the bacteria in stationary phase as previously reported [89]. A 2-μl aliquot of the concentrated bacterial suspension above was loaded into the main delivery channel of the mother machine device, and the chip was incubated at 37°C for approximately 20 min. We aimed for approximately 50% of the mother machine dead-end channels to be loaded with 1 or 2 bacteria to facilitate the assessment of the fate of individual bacteria throughout the single-cell infection assay. The chip was then connected to fluorinated ethylene propylene tubing (1/32"  × 0.008"), with the inlet tubing attached to a flow rate unit (Flow Unit S, Fluigent, Paris, France) controlling the pressure applied by a computerised pressure-based flow control system (MFCS-4C, Fluigent), and the outlet tubing into a separate waste tube allowing for the phage-containing outflow to be collected at regular time points.

The chip was mounted on an inverted microscope (IX73 Olympus, Tokyo, Japan) within an environmental chamber kept at 37°C, and spent LB was flushed through the main channel of the chip at 300 μl/h for 8 min before lowering the flow rate to 100 μl/h. A first set of bright field images was acquired via a 60×, 1.2 N.A. objective (UPLSAPO60XW, Olympus) and a sCMOS camera with an exposure time of 0.01 s (Zyla 4.2, Andor, Belfast, United Kingdom). For experiments with the ompC GFP reporter strain, we also acquired fluorescence images by using a FITC filter and a blue LED at 20% intensity (CoolLED pE200white, maximal power = 200 mW, Andover, UK). About 50 areas of the mother machine were imaged, each area containing 23 channels, typically hosting 1 or 2 bacteria per channel.

Phages were then diluted to a concentration of 10⁴, 10⁷, or 10⁹ PFU ml⁻¹ in LB depending on the specific experiment and added to the mother machine at 300 μl/h for 8 min before lowering the flow rate to 100 μl/h. Bright (and fluorescence) images were then acquired every hour for 24 h. At t = 24 h, propidium iodide (PI, Thermo Fisher Scientific, Waltham, Massachusetts, USA) was introduced into the chip to distinguish between viable but nongrowing bacteria (not stained by PI) and dead bacteria with compromised cell walls (stained by PI). Through the chip, 20 μM PI in LB was flowed at 300 μL/h for 8 min then reduced to 100 μl/h for an additional 10 min prior to imaging with a TRITC filter, a green LED at 100% intensity, and a camera exposure time of 0.01 s.

Since each mother machine channel typically accommodates up to 6 bacteria, some of the bacterial progeny was pushed out from the open end of the hosting channels during the single-cell assay, and the fate of such cells during T4 exposure could not be directly measured. Therefore, we predicted the probabilities of the possible fates of these cells (i.e., killed by T4 or surviving T4) based on the measurable fate for the remaining 95% of the population. Moreover, filamentation was defined as a cell over twice the average length of control bacteria growing in the hosting channels of the mother machine (i.e., longer than 7 μm).

In order to measure the phage population dynamics, the mother machine outflow was collected hourly into tubes containing 1 mg/ml kanamycin to prevent phage replication and serially diluted in LB and spotted on plates prepared as follows: 4 ml of liquid 0.5% LB agar was pipetted onto plates containing 1.5% LB agar. A volume of 200 μl of an overnight E. coli culture (E. coli BW25113 or E. coli B in the case of phage T4 and phage T3, respectively) was added to each plate, followed by 100 μL of each phage dilution. Plates were incubated overnight at 37°C, and plaque counts were enumerated. Only plates containing 30 to 300 plaques were counted.

In order to measure the number of particles reaching individual spatial refuges, we used fluorescent nanoparticles with size similar to bacteriophage T4 (i.e., amine-modified polystyrene beads conjugated to orange fluorescent dye, diameter range 100 to 120 nm, Sigma-Aldrich, St. Louis, Missouri, USA). Such nanoparticles were injected in the mother machine at a concentration of 10⁷ particles ml⁻¹ at a flow rate of 100 μl h⁻¹, and their diffusion in 20 spatial refuges was continuously imaged for 2 h via a Texas red filter, a green LED at 100% intensity, and a camera exposure time of 0.1 s. This experiment was performed in triplicate.