The human cerebral microvascular endothelial cell (hCMEC) line (Merck, London, UK) was cultured in EndoGRO-MV Complete Media (Merck) supplemented with 1% penicillin–streptomycin and grown at 37°C with 5% CO₂ in culture vessels coated with 5 µg/cm² Collagen Type 1 (Merck). Prior to the experiment, the cells were seeded onto coverslips in a 6-well plate or a fluorodisc in culture media at a density of 4 × 10⁴ cells/ml and left for 24 h to settle. The culture media was then replaced with Leibovitz L-15 media (Lonza), and the cells were moved to a 37°C incubator suitable for bacterial infection.

Bacterial cultures were grown from a single colony in lysogeny broth (LB), supplemented with the appropriate antibiotic if needed, and grown in a 37°C shaking incubator. The seven bacterial strains used in this study are listed in Table 1. During experiments, bacterial cells were added during the exponential growth phase at OD₆₀₀ 0.3 (unless otherwise stated as stationary) and incubated for 1 h prior to fixation or phage addition.

The bacteriophages used in this study are listed in Table 2. E. coli EV36 was used as a host to grow and purify K1F-GFP phage, and K12/FtsZ-mNeon was used as a host to grow and purify T7 phage, T7-mCherry phage, and T7Δ0.4 phage. After host clearance, each phage propagation culture was centrifuged at 3,220 × g for 15 min. The resulting supernatant was filter sterilised and incubated on ice with 0.2 M of NaCl for 1 h and then centrifuged at 3,220 × g for 15 min at 4°C. The supernatant was then incubated with 10% w/v PEG8000 overnight at 4°C to precipitate the phage and then centrifuged at 3,220 × g for 15 min at 4°C. The resulting phage pellet was resuspended in an SM buffer 1 (1 M of NaCl, 8 mM of MgSO₄•7H₂O, and 25 mM of Tris-HCl). A CsCl density gradient was set up with three solutions of densities of 1.7, 1.5, and 1.4 g/ml, along with a phage solution with CsCl added to give a density of 1.3 g/ml. The solutions were added in equal volumes to a centrifuge tube, starting with the heaviest, and centrifuged at 125,000 × g for 20 h at 4°C. The resulting phage band was extracted and placed in dialysis tubing (molecular weight cutoff (MWCO) of 14 kDa) and dialysed overnight in SM buffer 1 at 4°C, followed by dialysis in SM buffer 2 (100 mM of NaCl, 8 mM of MgSO₄•7H₂O, and 25 mM of Tris-HCl) for 2 h at room temperature twice. After dialysis, the purified phage was retrieved and stored at −20°C, and the phage titre was found by plaque assay. WT T7, T7-mCherry, and K1F-GFP phage stocks were diluted to approximately 10⁹ pfu/ml for use in experiments. For the experiments, phages were incubated for 1 h prior to fixation.

Cultures of hCMECs were incubated with the relevant bacterial strain at OD₆₀₀ of 0.2–0.4, which were added to the Leibovitz media for 60 min (unless otherwise stated) along with antibiotics or plasmid induction if needed, and if phages were being used, then the phage was added at 1 × 10⁷ pfu/ml and incubated for a further 60 min. Control cultures were incubated with bacteria or phage alone for 60 min in parallel.

To investigate if strains acted intracellularly, hCMEC cultures were infected with bacterial cultures at OD₆₀₀ 0.3 for 1 h and then incubated with 100 μg/ml of gentamicin for a further 2 h to kill extracellular bacteria, performed in triplicate for each condition. For each condition, a minimum of 300 human cells were counted to calculate the internalised bacteria percentages.

To quantify human cell death after bacterial infection, hCMEC cultures were infected with bacterial cultures at OD₆₀₀ 0.3 for 1 h, in triplicate conditions. For each condition, a minimum of 500 human cells were counted.

After bacterial and/or phage infection, hCMEC cultures were fixed with 4% paraformaldehyde (Thermo Fisher Scientific) in phosphate-buffered saline (PBS) for 15 min and then washed in PBS. Cells were then permeabilised in ice-cold PEM/0.05% saponin for 5 min, washed, incubated with 50 mM of NH₄Cl in PBS for 15 min, and then washed in PBS/0.05% saponin. The fixed cells were stained overnight at 4°C with anti-Prokaryotic Cell Division GTPase (FtsZ) antibody (Agrisera, Vännäs, Sweden) or rabbit anti-E. coli (strain K12) FtsI Polyclonal antibody if no fluorescent proteins were present and FtsZ or FtsI staining was needed, with diluted 1:100 in 0.05% saponin in PBS. This was followed by conjugation with secondary Donkey Anti-Rabbit IgG H&L (Alexa Fluor^® 488) diluted 1:500 in 0.05% saponin in PBS at room temperature for 45 min. For samples with fluorescent proteins, the fixed cells were stained with the following antibodies diluted in PBS/0.05% saponin for 60 min at room temperature: 5 µg/ml of GFP-booster (ChromoTek, Planegg, Germany), 5 µg/ml of RFP-booster (ChromoTek), and 5 µg/ml of phalloidin CF680R conjugate (Biotium, Fremont, CA, USA).

After staining, the coverslips were washed, mounted on slides with DAPI-containing Fluoroshield Mounting Medium (Abcam, Cambridge, UK), and sealed. The slides were then imaged using a Zeiss LSM800 confocal microscope using the following excitation wavelengths: DAPI at 405 nm, GFP/mNeon at 488 nm, mCherry at 561 nm, and phalloidin at 633 nm.

Quantification was performed by manually counting Z-rings of 75 bacterial cells per condition and measuring cell length using the Fiji (ImageJ) software for phage filamentation experiments. Data were plotted with error bars showing one standard deviation of uncertainty. For significant difference comparisons, Tukey’s tests were performed, and the calculated probability values (p-values) are displayed as p < 0.05 (*), p < 0.01 (**), and not statistically significant p > 0.05 (ns).

Cultures of bacteria alone expressing fluorescent proteins were imaged live on 2% agarose pads. Cultures of K12/FtsZ-mNeon were grown to OD₆₀₀ 0.5 and imaged immediately. For EV36/FtsZ-mCherry, cultures were grown to OD₆₀₀ 0.3 and induced with 0.2% arabinose for 5 min (stopped using 0.2% glucose) and incubated for 45 min to allow time for protein folding before imaging. Agarose pads were made by dissolving 0.2 g of agar into 10 ml of sterile water by heating, and a drop was placed onto a glass microscope slide and left to dry. Culture measuring 10 µl was then placed onto the agarose layer and covered with a coverslip once dry. Cells were imaged using the Zeiss LSM 880 confocal microscope, with mNeon excitation at 488 nm and mCherry at 561-nm wavelengths, and transmitted light was used to visualise the cell outline.

T7-mCherry phage was added to a K12/FtsZ-mNeon culture (OD₆₀₀ 0.3) and incubated at 37°C with shaking for 60 min; 300 mesh copper grids (FC300Cu) were hydrophilised by glow discharge. The sample was added to the grid and incubated for 2 min, and then 2% uranyl acetate was dropped into the grid and incubated for 2 min. A wash step was performed three times and blotted; 0.1% trehalose was added to the grid and incubated for 2 min. This was left to air-dry. The sample was imaged using Zeiss LSM 880 confocal microscope (565 nm for T7-mCherry and 488 nm for K12/FtsZ-mNeon) and Jeol 2100 transmission electron microscope, and ImageJ was used for overlay.

We first sought to observe Z-rings in bacterial cells, using the E. coli K12/FtsZ-mNeon (Moore et al., 2017) strain, which has its genomic copy of FtsZ gene replaced with FtsZ-mNeon. A crucial aspect of the design of this FtsZ construct is that mNeon protein has been placed on an internal loop of FtsZ so that the head-to-tail mode of polymerisation required for the function was not impaired (Moore et al., 2017), and the mNeon proteins were prevented from forming aggregates with each other. The Z-rings were visualised by placing exponential-phase bacterial cultures onto 2% agarose pads for visualisation by confocal microscopy. FtsZ-mNeon assembled at the cell midbody as a single Z-ring (Figure S2A). This visualisation was repeated in the E. coli K1/K12 hybrid strain EV36 (Vimr and Troy, 1985), by transforming it with an arabinose-inducible FtsZ-mCherry plasmid (Galli and Gerdes, 2010), and FtsZ-mCherry forms a Z-ring at the EV36 midbody (Figure S2B).

We next observed the phenotype of the Z-rings in a meningitis infection model to see if FtsZ localisation changed during bacterial infection of human cells. hCMEC brain cells were infected with K12/FtsZ-mNeon culture for 1 h, fixed, and stained before visualising by confocal microscopy. During human cell infection, FtsZ rings were still visible at the midbody of K12 cells as a single ring per cell (Figure 1A), suggesting cells can divide normally, and this was also found for the EV36/FtsZ-mCherry strain (Figure 1B). These data confirm that FtsZ assembles as a single ring at the cell midbody under normal bacterial growth conditions and during bacterial infection of human cells. The EV36 strain has previously been shown to have an infection rate of 20%–50%, depending on the human cell line (Moller-Olsen et al., 2018; Moller-Olsen et al., 2020). Since the K12 strain is non-pathogenic and not expected to act intracellularly (Meier et al., 1996; Sahu and Kar, 2012), an internalisation experiment was done using gentamicin to kill extracellular bacteria and to confirm if K12 would be present inside human cells. This showed that only 4.1% of hCMECs infected with K12/FtsZ-mNeon had intracellular bacteria. Since the infection rate is low, for the following experiments, we chose to focus on both intracellular and extracellular bacteria in the infection model for both strains. Nevertheless, both extracellular and intracellular bacteria are toxic to human cells and have a variety of mechanisms to cause disease (Mak et al., 2014). We confirm this by quantifying human cell death by confocal microscopy after bacterial infection of human cells and found that 5.3% of human cells died after infection with the K12 strain, and 12.1% of human cells died after infection with the EV36 strain (Figure S3). Recent studies have presented the efficiency of bacteriophages in targeting both intracellular and extracellular bacteria in the human cell environment (Śliwka et al., 2022), and bacteriophages have been shown to be more efficient in the presence of human cells (Shan et al., 2018). It is therefore necessary to test the proposed system in infection human cell models, in order to understand better the interplay of phage and bacteria in the presence of human cells.

We then applied the known FtsZ inhibitor Kil peptide to the model system to investigate how FtsZ inhibition would appear in these conditions. Kil stops FtsZ from polymerising, leading to filamentation of cells that are unable to divide, so a Kil peptide plasmid (Haeusser et al., 2014) was transformed into both strains. For these experiments, the plasmid was induced with 0.2% arabinose for 2 h. Imaging in Figure S4A confirmed that induction of Kil in K12/FtsZ-mNeon resulted in long filamentous cells, with varying FtsZ phenotypes that ranged from distinct multiple rings (42.2% of cells) to diffuse FtsZ spread (30.1%) and cells in an intermediate phenotype (27.7%) of faint rings within a diffuse background. Previous studies have shown a diffuse FtsZ spread upon Kil inhibition using antibody staining (Haeusser et al., 2014), which was confirmed in Figure S5; however, here we present that when using a fluorescent protein stably expressed, multiple distinct Z-rings still form along the cell body, potentially due to Kil not having fully acted on these cells yet.

To support these findings, images for the induction of Kil were quantified in K12/FtsZ-mNeon under the following conditions: control (no Kil induction), 1-h induction, and 2-h induction. The averages (Figures S4C , D) showed an increase in the number of Z-rings and cell length after 1 h of Kil induction, with a further increase after 2 h, confirming that the cells are not dividing and continually elongating in the presence of Kil, and formation of multiple Z-rings suggests cell division had been unsuccessfully attempted. The increase seen after 2 h compared to 1 h suggests that the effect and phenotypes seen may be dependent on the amount of Kil peptide present in the cell.

These data support previous findings that the phage λ Kil peptide causes filamentation of E. coli cells by blocking cell division with diffuse FtsZ spread by antibody staining, as FtsZ is unable to polymerise to form rings and further show that multiple Z-rings can form along the body of the filamentous cell, as it fails to divide at new division sites when using fluorescent proteins.

The K1F-GFP phage (Moller-Olsen et al., 2018) has been previously engineered to express GFP, and this was first visualised in human cell vacuoles (Figure 2A) to confirm fluorescence compared to no-phage controls as a visual control for use in confocal experiments. In order to visualise the T7 phage in the K12/FtsZ-mNeon model system, we attached a fluorescent protein to the C-terminal of the phage’s minor capsid protein so that it is displayed at the surface, using homologous recombination from an engineered plasmid. The presence of mCherry gene in the phage lysate was confirmed by PCR screening of the recombinant phage genome using the primers in Table S1 and associated plaque assays (Figures S6A –D), prior to purification of positive lysate by caesium chloride gradient purification (Figures S6E , F). Fluorescence of the mCherry tag was further tested by confocal microscopy (Figure 2B) showing mCherry fluorescence inside human cells after phage infection, alongside no-phage controls (Figure S7), performed to show the fluorescence seen was from the phage label, confirming successful engineering of the T7-mCherry phagemid, which was able to be taken up by human cells.

K1F-GFP phage was confirmed to target the EV36/FtsZ-mCherry strain (Figure S8A), and T7-mCherry phage was confirmed to target the K12/FtsZ-mNeon strain (Figure S8B) by using growth curves. This was further confirmed by correlative light and electron microscopy (CLEM) showing the formation of a septum at the midbody of cells, as well as colocalisation of FtsZ-mNeon and T7-mCherry signals within K12 cells (Figure S9), confirming the specificity of the phage.

Next, the effect of T7-mCherry on FtsZ was examined by infecting hCMEC with K12/FtsZ-mNeon, followed by incubation with T7-mCherry. Confocal microscopy showed that T7 phage infection caused a change from the single midbody ring previously seen and led to the formation of filamentous, multi Z-ring E. coli cells (Figures 3A , B), which were unable to divide, suggesting that T7 phage can block cell division. We were then interested to see how the Z-rings and cell length of E. coli changed over time in control conditions (no phage) and the presence of T7 phage by using a time-series experiment. For the control, hCMECs were infected with exponential K12/FtsZ-mNeon and fixed at 15-min intervals for 2 h (Figure S10). For the phage time series, hCMECs were infected with K12/FtsZ-mNeon for 1 h, and then T7 phage was added (Figure S11). For each condition, bacterial cells were quantified using confocal microscopy (Figures 3C , D). Since K12/FtsZ-mNeon cells were incubated prior to phage addition, this meant phage time of 0 min was equivalent to control 60 min, so these were plotted to overlap. Larger error bars were due to the variation in phenotypes, as some cells were fully filamentous, whereas others were yet to be infected. In the first 60 min of the control series, each cell averages one Z-ring at the midbody as seen earlier in Figure 1, but later we unexpectedly observed two 2 polar rings in the majority of cells. Upon phage addition, the number of Z-rings started to increase after 15–30 min, and after 1 h, they reached an average of four rings, double compared to the control. After 1 h of incubation with phage, there was a notable increase in cell length showing filamentation, but interestingly, this did not start until about 30–45 min after infection, suggesting that the number of rings started to increase first and then filamentation followed.

To investigate the cause of the filamentation, confocal microscopy was performed using a K12ΔminC strain from the Keio collection with minC gene knocked out, alongside a K12 control. Cultures were incubated for 2 h in no-phage conditions, then stained with an FtsZ antibody, and imaged (Figure S12). The images were quantified, and average cell length was plotted (Figure 3E), showing a significant increase in cell length of K12ΔminC compared to the control, suggesting that inhibition of MinC could be causing the filamentation seen upon phage infection of cells.

Taken together, these data confirm the specificity of engineered T7-mCherry for K12 strains and show that during infection, the phage causes a filamentous (potentially due to MinC inhibition) non-dividing phenotype with multiple Z-rings formed where the division had been attempted, similar to what was seen with the known FtsZ inhibitor Kil, suggesting that T7 phage can inhibit FtsZ to block division.

To investigate the effect of K1F-GFP phage on FtsZ localisation, hCMECs were infected with host EV36/FtsZ-mCherry and K1F-GFP phage. The EV36/FtsZ-mCherry cells showed a long filamentous phenotype with multiple Z-rings and DNA along the cell after K1F-GFP infection (Figures 4A , B), unlike control samples, suggesting that the K1F phage can also inhibit FtsZ to block host cell division.

To further confirm this, bacterial cells were quantified for control conditions and K1F-infected conditions (Figures 4C, D). The control showed an average of 1.027 rings per cell as expected, and after K1F-GFP infection, this increases to 4.120 rings per cell, suggesting cells had attempted to divide two to three times unsuccessfully. There was an increase in average cell length, showing cells were starting to elongate to confirm filamentation. These data show that similar to phages λ and T7, the K1F phage is able to inhibit FtsZ to block host cell division.

We hypothesised that the observed effect of FtsZ inhibition by phages may not be specific to just FtsZ and that other divisome proteins may be involved in the infection process. To test this, a K12ΔzapA strain from the Keio collection with zapA gene knocked out was used, alongside a K12 control. Cultures were incubated for 1 h, then infected with T7 phage for a further hour, alongside a no-phage control, and imaged by light microscopy to observe cell filamentation (Figure 5A). The images were quantified, and average cell lengths for the conditions were plotted (Figure 5B). As previously shown, there was an increase in cell length upon T7 phage infection of K12 cells as they became filamentous, but this phenotype appears to be absent during T7 phage infection of K12ΔzapA cells, suggesting that ZapA is essential for cell filamentation and the phenotype seen during phage inhibition of cell division.

Next, the localisation of FtsI protein in this system was investigated. Cultures of K12 were grown for 1 h, infected with T7 phage, alongside a no-phage control, and stained with an FtsI antibody. Under normal growth conditions in the no-phage control, we observed that FtsI is localising to the midbody of the cell as a single ring (Figure 5C) as expected after recruitment by FtsZ as the divisome forms. After T7 phage infection of K12, we observed mislocalisation of FtsI (Figure 5D). This suggests that other divisome proteins are mislocalised upon phage infection as cell division is inhibited, potentially being recruited by mislocalised FtsZ.