Metallohelices that kill Gram-negative pathogens using intracellular antimicrobial peptide pathways

Iron-based self-assembled optically pure compounds mimic the mechanisms of small peptides, according to biophysical, genomic, transcriptomic and other analyses.

Diphenyl sulfide (2.8 ml, 3.1 g, 16.6 mmol, 1.0 eq.) and paraformaldehyde (1.8 g, 59.9 mmol, 3.6 eq.) were added to a solution of 33 wt% HBr in glacial acetic acid (18 ml). The mixture was stirred under reflux at 50 °C overnight. Upon cooling, the addition of water (100 ml) caused a white precipitate to form, which was collected by filtration, washed with water (100 ml) and 2:1 n-hexane/ethyl acetate (80 ml), and dried in air. The product was recrystallised from hot toluene/n-hexane, to give a white crystalline solid, which was collected by filtration and dried overnight at 50°C in vacuo.

bis-4-(bromomethyl)phenylmethane 3
Diphenylmethane (5.0 g, 29.7 mmol, 1.0 eq) and paraformaldehyde (5.35 g 0.178 mol, 6.0 eq.) were suspended in a mixture of aqueous 48 wt% HBr solution (80 ml) and glacial acetic acid (25 ml). Tetradecyltrimethylammonium bromide (0.16 g) was added and the suspension was stirred under reflux at 125 °C overnight. Upon cooling, the yellow solid formed was collected by filtration, washed with water (100 ml) and dried in air. The product was recrystallised from hot S6 toluene/n-hexane, to give a white powder, which was collected by filtration and dried overnight at 70 °C in vacuo.
Tetradecyltrimethylammonium bromide (0.16 g) was added and the suspension was stirred under reflux at 125 °C overnight. Upon cooling, the yellow solid formed was collected by filtration, washed with water (100 ml) and dried in air. The product was recrystallised from hot DCM/nhexane, to give a yellow powder, which was collected by filtration and dried overnight at 50 °C in vacuo.

2,8-bis(bromomethyl)dibenzofuran
Dibenzofuran (1.0 g, 5.9 mmol, 1.0 eq) and paraformaldehyde (0.78 g, 26.0 mol, 4.38 eq) were suspended in a solution of 33 wt% HBr in glacial acetic acid (10 ml). 90 wt% phosphoric acid (5 ml) was added and the mixture stirred under reflux at 65 °C for 1 h, then at ambient temperature overnight, before cooling. The reaction contents were added to ice-cold water (150 ml), and the white solid collected by filtration and dried in air. The crude product was dissolved into a minimal volume of hot toluene, before adding an excess (~100 ml) of n-hexane. The product was recrystallised from hot toluene/n-hexane, to give a white powder, which was collected by filtration and dried overnight at 50°C in vacuo.
The crude product was extracted using diethyl ether (3 × 100 ml), dried over sodium sulfate, S12 filtered through celite, and the solvent removed under reduced pressure to leave a yellow oil.

Synthesis and characterisation of [Zn2L3][ClO4]4 complexes -General Procedure
The appropriate optically pure diamine (3.0 eq.) and 2-pyridinecarboxaldehyde (6.0 eq.) were dissolved in acetonitrile (30 ml) and stirred for 30 min at ambient temperature to form a yellow solution containing the ligand. Zinc(II) perchlorate hexahydrate (2.0 eq.) was added was added and no colour change was observed as the solution was stirred at ambient temperature for 4 h.
The volume of the solution was reduced to ~10 ml under reduced pressure, and ethyl acetate was added dropwise to cause precipitation of a white solid, which was collected by filtration, washed with ethyl acetate (20 ml), and dried in air.

Synthesis and characterisation of [Fe2L3]Cl4 complexes -General procedure
The appropriate optically pure diamine (3.0 eq.) and 2-pyridinecarboxaldehyde (6.0 eq.) were dissolved in methanol (50 ml) and stirred for 2 h at ambient temperature to form a yellow solution containing the ligand. Anhydrous iron (II) chloride (2.0 eq.) was added, and an immediate colour change to deep purple was observed. The solution was then heated at reflux S28 (80 °C) for 48 h. After filtering through fluted filter paper, the solvent was removed under reduced pressure to give the desired product as a dark purple solid, which was dried overnight at 50 °C in vacuo. Note that the presence of water of crystallisation in these compounds was confirmed by NMR and IR spectroscopy and the absence of other solvents was confirmed also by NMR. The hydration number was then determined by thermogravimetric analysis (section 1.6) and the relevant mass loss was correlated with microanalytical data.

Optical purity of amines: synthesis and analysis of Mosher amides
The diamines (R/S)-3a and (R/S)-3b were synthesised using racemic phenylglycinol. These mixtures could not be distinguished from the nominally optically pure products e.g. (R,R)-3a, by NMR spectroscopy. Therefore a chiral derivatisation method was employed 4 whereby samples of nominally S,S-, R,R-and racemic (i.e. S,S/R,R/S,R) diamine were converted to the corresponding diamides using (R)-(+)-Mosher's acid (α-methoxy-α-trifluoromethyl-phenylacetic acid) and the products analysed by 1 H-NMR spectroscopy as follows.
After stirring for 1 hour the reaction mixture was concentrated in vacuo and the residue was suspended in hexane (10 ml The spectra for 3b isomers are shown in Figure S1. The diamide derivative of the racemic diamine gave two resonances in the ratio 1:1 (green line) indicating again that the stereogenic centres are isolated as far as can be determined by 1 H-NMR spectroscopy at 400 MHz, but also that local diastereomeric units containing S-and R-amines could be distinguished via the 1 H chemical shift of the methoxy group in (R)-(+)-Mosher's acid. The diastereomeric diamides of (R,R)-3b (blue line) and (S,S)-3b (red line) gave essentially a single resonance for the OMe unit.
Thus, the total enantiomeric excess of all amine-derived chiral centres could be measured for a given sample (found to be >98 % e.e.).
Measurements were collected from a 1 cm path-length quartz cuvette and, unless otherwise mentioned, the standard parameters used were: bandwidth 1 nm, response time 1 sec, wavelength scan range 200-700 nm, data pitch 0.2 nm, scanning speed 200 nm/min, with four accumulation taken per sample to give an average spectrum with reduced noise. Circular Dichroism CD spectra (Fig. S3) were measured on a Jasco J-815 spectrometer calibrated conventionally using 0.060% ACS a holmium filter. Measurements were collected using a 1 cm path-length quartz cuvette and unless otherwise mentioned the standard parameters used were: bandwidth 1 nm, response time 1 sec, wavelength scan range 200-700 nm, data pitch 0.2 nm, scanning speed 100 nm/min, with four accumulations.

Thermogravimetric analysis
Thermogravimetric analysis (TGA) was performed using a Mettler Toledo TGA/DSC 1 STAR® system instrument. Samples were weighed accurately into a pre-weighed 40 μl TGA/DSC aluminium crucible (DSC consumables Inc.) and heated 298 to 573 K (25 to 300 °C), at 5 K/min under a nitrogen atmosphere. The mass of the sample was recorded at various temperature points along this range. Mass decrease due to loss of water of crystallisation was correlated with microanalytical data. S51 Fig. S4. Thermogravimetric analysis (TGA) of metallohelices 5a-h mass loss due to water of crystallization at lower temperatures was correlated with results of microanalysis.

Crystallography
Crystallographic data is uploaded to the Cambridge Crystallographic Data Centre (CCDC): 1904782, 1904783 and 1904784 with outline comments below. While the Zn(II) metallohelices with perchlorate counter-ions generally gave higher quality crystals, many challenges in refinement nevertheless resulted from disorder, twinning etc. In two instances, higher quality although still problematic crystals were grown by deliberately mixing the two enantiomers (as synthesised) in acetonitrile, then crystallising.

Compound 6b
Single crystals of C112H108Cl4N14O22Zn2 were grown from an optically pure sample in acetonitrile using a diffusion chamber with ethyl acetate. A suitable crystal was selected and mounted on a glass fibre with Fomblin oil and placed on an Xcalibur Gemini diffractometer with a Ruby CCD area detector. Using Olex2, 5 the structure was solved with the ShelXS 6 structure solution program using Direct Methods and refined with the ShelXL 7 refinement package using Least Squares minimisation.
Given the importance of this structure type, many crystals were investigated. Diffraction was weak and the data were recorded in the orthorhombic cell but refined finally in the lower symmetry monoclinic, and while this leads to a lower Friedel pair coverage the handedness of the sample was already known. The structure is twinned about the b axis swapping the a and c axis. The twin components refine to 0.5276 (14). Since this is about a two-fold axis it does not affect the absolute configuration. The asymmetric unit contains the Zn helix, four ClO4counter ions and two molecules of acetonitrile.
The structure was extensively disordered. Occupancy of the disordered components was originally linked to free variables and fixed at set values at the latter stages of the refinement.
Xylenyl group C51 to C58 was modeled as disordered over two positions related by a small shift.
The occupancy of the two components was fixed at 60:40. Similarly C87 to C94 was modeled over two positions at 50% occupancy. Perchlorates Cl30-O34 and Cl40-O44 where modeled over two positions related by a small shift in the anion. The occupancy of the two components of Cl30/Cl3A was fixed at 60:40. The occupancy of the two components of Cl40/Cl4A was fixed at 50:50. All atoms were refined anisotropically. Several DFIX, DANG and SIMU restraints were Fig. S5. Structure of the cation unit of Δ-6b. Ellipsoids modelled at 50% probability. Hydrogen atoms, solvent molecules and counterions removed for clarity. Zn(II) ions are shown in pink, nitrogen atoms in blue, oxygen atoms in red, and carbon atoms in dark grey.

Compound 6e
We are grateful to the EPSRC UK National Crystallography Service (NCS) for recording the data for this compound. 8 Single crystals of C137.5H129.25Cl4N17.75O23S3Zn2 were grown from an acetonitrile solution of a deliberate mixture of the two enantiomers, layered with ethyl acetate. A suitable crystal was selected and mounted on a Mitegen head with Fomblin oil and placed on an AFC10 goniometer with a Rigaku FRE+ with VHF Varimax confocal mirrors and equipped with an HG Saturn 724+ CCD detector. The crystal was kept at 100(2) K during data collection. Using Olex2 5 , the structure was solved with the ShelXT 9 structure solution program using Direct Methods and refined with the ShelXL 7 refinement package using Least Squares minimisation.
The asymmetric unit contains the metallohelix, four perchlorates and seven molecules of acetonitrile. Acetonitrile molecules N1A-C2A, N1B-C2B andN1C-C2C were modelled at full occupancy. Acetonitrile N1D-C2D was modelled at 75% occupancy. Acetonitrile N1F-C2F and N1G-C2G were modelled at a half occupancy and refined iostropically. Another acetonitrile was modelled over two closely related positions (N1E-C2E and N1H-C2H) at 50% occupancy and refined isotropically. Perchlorate Cl20 was modelled as disordered over two positions about Cl20. The occupancies were linked to a free variable which refined to a ratio of 82:18. The minor component oxygen atoms were refined isotropically. Perchlorate Cl30 was modelled as disordered over two positions about Cl30. The occupancies were originally linked to a free variable but once this had settled were fixed at a ratio of 75:25 for the final stages of the refinement. The minor component oxygen atoms were refined isotropically. Perchorate Cl40 was modelled over two positions. The occupancy of these two perchlorate positions (Cl40 and l4A) was fixed at 50%. This disordered model was also composed of water molecules that shared that space when a perchlorate was not present. One perchlorate component Cl40 occupied the same location as water O1C (50% occupancy). The other perchlorate Cl4A occupied the same location as two closely related waters O1A (25% occupancy) and O1B (50%) occupancy. To clarify, one part of the disordered model has perchlorate Cl40 and waters O1A and O1B and the other part has perchlorate Cl4A and water O1B. No hydrogen atoms were located on these water molecules and the oxygen atoms were refined isotropically. Several DFIX, DANG and SIMU restraints were used to give the disordered components reasonably bond lengths, angles and thermal S56 parameters. Further, the structure contained large solvent accessible voids; the solvent masking algorithm in Olex2 was used to treat the electron density from solvent molecules that were not located as a diffuse contribution to the overall scattering without specific atom positions.

Compound 6h
Single crystals of C257H223.5Cl6N26.5O42Zn4 were grown from acetonitrile solution of a deliberate mixture of the two enantiomers, layered with ethyl acetate. A suitable crystal was selected and mounted on a glass fibre with Fomblin oil and placed on an Xcalibur Gemini diffractometer with a Ruby CCD area detector. The crystal was kept at 100(2) K during data collection. Using Olex2 5 , the structure was solved with the ShelXS 6 structure solution program using Direct Methods and refined with the ShelXL 7 refinement package using Least Squares minimisation. The asymmetric unit contains two crystallographically independent metallohelices of opposite handedness. Six perchlorate anions and three acetonitrile molecules were located and refined but the other two perchlorates and additional solvent were not located (see below). Acetonitrile N900-C902 was refined at half occupancy. Several of the perchlorates and part of a ligand were modeled as disordered. The occupancy of the disordered components was originally assigned to a free variable but fixed for the final refinement cycles. Perchlorate Cl50 was modeled as disordered over two positions related by a rotation about the O51-Cl50 bond. The occupancies were fixed at 50:50. Perchlorate Cl70 was modeled as disordered over two positions related by a rotation about the O71-Cl70 bond. The occupancies were fixed at 50:50. Perchlorate Cl60 was modeled over two closely related positions. The occupancy was fixed at 70:30 Cl60-O64 to Cl6A to O64A.
The minor component oxygen atoms were refined isotropically. Phenyl C218-C223 was modeled as disordered over two positions related by a small displacement. The disorder was traced back to the benzylic carbon C217. The occupancy of the two components was fixed at 60:40 C217-C223 to C1A-C1G. Several DFIX, DANG and SIMU restraints were used to give the disordered components reasonably bond lengths, angles and thermal parameters. The solvent masking algorithm in OLEX2 was used to treat the electron density from those anions and solvent molecules that were not located as a diffuse contribution to the overall scattering without specific atom positions.
Weak data led to a B alert in the cifcheck. The fragile crystals were very weakly diffracting containing voids filled with disordered solvent. Crystals were measured at 150K with long exposure times but had little diffraction intensity out further than 165 degrees 2 Theta. This does not affect the structure determination. Fig. S7. Structure of the cation unit of 6h. Ellipsoids modelled at 50% probability. Hydrogen atoms, solvent molecules and counterions removed for clarity. Zn(II) ions are shown in pink, nitrogen atoms in blue, oxygen atoms in red, and carbon atoms in dark grey.

Intermetallic distance
In Figure 2 of manuscript the horizontal axis has the metallohelices in order of increasing Fe-Fe distance. These were determined by crystallography, or where this was unavailable, by calculation (vide infra).
The crystallographic data comes exclusively from the Zn(II) structures, as described above. While there will be slight differences between the inner-sphere structures of Fe(II) and Zn(II) compounds, the size and shape of the metallohelix is determined by the nature and folding of the ligand strands.
Starting points for geometry optimisations were taken from adapted crystallographic structures. So as to capture possible conformational local minima, structures were optimised using ligand field molecular mechanics (LFMM) 10 as implemented in the DommiMOE program 11 before being annealed at 500 K for 1 ns prior to re-optimisation. Structures were further optimised using the Firefly quantum chemistry package 12 which is partially based on the GAMESS(US) source code 13 using B3LYP-D3(BJ) 14 functional and the 6-31g basis set with convergence criteria of 0.0001 a.u. For the compounds 5c and 5g a number of conformations were located.

Stability testing
No decomposition was observed for 5b in aqueous solution over a period of weeks by NMR spectroscopy. In order to give an indication of stability, the compound was dissolved in KCl/HCl buffer (pH 1.5) in a UV-vis cell and the MLCT band at 550 nm was measured over time. While decomposition was still very slow, at 5 h ca 2% loss in signal was observed reproducibly (see figure). This corresponds to a first-order rate constant of 4.2 × 10 -5 min -1 , or t1/2 = 11 d.

Ethidium bromide displacement
A Varian Cary Eclipse spectrofuorophotometer with a 1 cm quartz cell was used to perform fluorescence measurements at room temperature. Excitation of the CT-DNA-ethidium bromide complexes was performed at 546 nm and fluorescence emission was measured at 595 nm.
Aliquots of 1mM metallohelix stock solutions were added to a solution of 1.3 µM ethidium bromide and 3.9 µM CT-DNA in 10 mM Tris buffer (pH 7.4) containing 1 mM EDTA, in a total volume of 2.5 mL. Fluorescence was measured after each addition until it was reduced to 50 %.

UV melting assays
A Varian Cary 4000 UV-Visible spectrophotometer, equipped with a Peltier controlled 6-sample cell-changer, was used to conduct the DNA stability assays. The absorbance at 260 nm was measured as a function of temperature using a heating rate of 0.4 ˚C/min, a 1 nm bandwidth and an average time of 2 s. The experiments were performed using masked 1.2 ml cuvettes of 1 cm pathlength. The Tm of CT-DNA was measured using a concentration of 7.5×10 -5 M (per base) in 10 mM Tris buffer (pH 7.4).

Linear dichroism spectroscopy
A Jasco J-720 spectrometer adapted for LD spectroscopy was used to record flow LD spectra using a large volume (1 ml

Atomic Force Microscopy
NdeI digestion was used to linearize plasmid pSP73, which was then purified using

Topoisomerase I and gyrase inhibition assays
For the enzyme inhibition assays, the compound Λ-5b was dissolved in 10 % DMSO at a concentration of 500 µM, and 3 µl of that were used in a total reaction volume of 30 µl, thus resulting in a final compound concentration of 50 µM. For the Topoisomerase I inhibition assay, supercoiled plasmid pBR322 was purified using the Qiagen Plasmid Maxi kit and 480 ng of plasmid were mixed with either 10 % DMSO or compound. The buffer used was 1 x Cutsmart buffer (NEB) and the E. coli enzyme was obtained from NEB (catalogue number M0301S). The enzyme was diluted in 1x Cutsmart buffer and 3 µl, corresponding to 1 U of enzyme, were added to the reaction mix. The reaction was allowed to proceed for 30 min at 37 ˚C. For the gyrase supercoiling assay, the enzyme and relaxed plasmid pBR322 were obtained from Inspiralis (catalogue number K0001). 500 ng of plasmid were again mixed with either compound or 10 % S69 DMSO and added to 1 x Inspiralis gyrase assay buffer. The enzyme was diluted 1/6 in dilution buffer and 5 µl of that were added to the reaction mix. The reaction was allowed to proceed for 30 min at 37 ˚C. The reactions were terminated by addition of 30 µl of loading buffer (40 % (w/v) sucrose, 100 mM Tris-HCl pH8, 10 mM EDTA, 0.5 mg/ml bromophenol blue), followed by 30 µl of chloroform-isoamyl alcohol (v/v, 24:1). The mixtures were shaken and the organic and aqueous phases were separated by centrifugation. The upper phase was then transferred to a 1 % agarose gel in TAE buffer. The gels were run at 85 V for 1 h 50 min, stained with SybrSafe and visualised using a GeneSys Syngene gel imager. Topo I -

Λ-5b activity in stationary phase cells
Cultures of E. coli Sakai were grown overnight to stationary phase, collected by centrifugation, and re-suspended to an OD600 of 0.1, in 20 ml PBS containing either 8 μg/ml Λ5b or 20 % isopropanol before incubation for a further three hours at 37 °C. Cells were subsequently harvested by centrifugation and washed twice in clean sterile PBS. The final cell pellet was resuspended to an OD600 of 0.01 in 50 ml CAMHB. These cultures were incubated at 37 °C and monitored for growth by taking periodic OD600 measurements and viability counts over five hours.

Bacterial sub-cellular compound localization-further details
For visualisation of the sub-cellular localisation of the target compound we used the Click-IT cell reaction buffer kit (Invitrogen) as per the manufacturer's instructions and following the procedure described below. To observe the dividing cells, overnight culture of EHEC Sakai ∆stx 1-2 was diluted in cation-adjusted Mueller Hinton broth and grown to mid-exponential (OD600 = 0.52). Then, Λ-5b', at the MIC of 8 µg/ml or at a quarter of the MIC (2 µg/ml), or methanol were added to the culture and incubated for 30 min at 37 ˚C. Ten minutes before the end of the incubation period, 5 µg/ml of the membrane dye FM™ 4-64FX (Invitrogen) was added to stain the cell membrane. Then, the cells (1 ml for each treatment) were collected by centrifugation and fixed with 4 % PFA for 15 min at 4 ˚C. They were then washed with PBS by centrifugation and permeabilised with 0.5 % Triton X-100 in PBS by incubating at room temperature for 30 min.
Cells were washed with PBS followed by 2 % BSA by centrifugation at 5000 g and resuspended in 180 µl of the Click reaction mix (containing 5 µM AF-488 azide). They were incubated at room temperature in the dark for 30 min, washed with 2% BSA in PBS and stained with 1 µg/ml DAPI for 1 min. They were finally washed with PBS and used for microscopy. For stationary S80 phase cells, the same procedure was followed with the following modifications. A 24 h culture of EHEC Sakai ∆stx 1-2 at an OD600 of 2.9 was used. Cells were treated with the compound only at the MIC concentration of 8 µg/ml, or methanol. Following incubation with the membrane dye, 700 µl of the culture were collected (an OD600 equivalent of 2.0) and 360 µl of the Click reaction mix were used instead, to account for the higher number of bacteria present. Slides were prepared using either 1 % agarose pads (in PBS) or Poly-L-lysine (0.01 % solution) solution.
Samples (3 µl) were then immobilised onto the slides, allowed to dry and then 4 µl SlowFade Gold antifade reagent (ThermoFisher Scientific) were added and coverslips were mounted for viewing.

Microscopy
Imaging was performed on a Leica DMi8 Inverted Microscope equipped with a Hamamatsu

Image analysis
Fluorescence microscopy images were analysed using ImageJ/Fiji. For quantitation, individual bacterial cells were selected using the red channel images showing the membrane stain. In brief, the red channel images were processed using the Filter  Convolve command with the default kernel and an automatic median threshold was applied. Then, cells were automatically detected using particle analysis with a specified size range of 0.5-6.0. The regions of interest (ROIs) were manually confirmed and those that did not correspond to individual cells were discarded. These ROIs were then used on the green and blue channels to measure the corresponding intensities per bacterial cell. Based on a histogram of the mean blue intensity values, the cells in the sample treated with the lead compound were split into two populations (of DAPI mean intensity/cell either above or below 11000). A Kolmogorov-Smirnov test was then performed to evaluate if the mean AF-488 (green) intensity values of these populations were significantly different from each other.  (upper) or stationary phase (lower) were treated with 8 µg/ml Λ-5b' and stained for membrane , nucleic acid (DAPI) and the Λ-5b' (via click reaction with AF-488 azide). Note that in stationary phase, only cells lacking membrane staining showed weak Λ-5b' staining.

Whole genome analysis
Fastq reads from the MiSeq whole genome sequencing of WT and tolerant clones were uploaded to EnteroBase 15 (http://enterobase.warwick.ac.uk) and assembled into contigs. The genomes of the tolerant clones were then aligned against the WT assembly whilst all genomes including the WT were also aligned against the reference E. coli O157:H7 Sakai (Assembly: GCF_000008865.1) for SNP and insertion/deletion (indel) detection. The EnteroBase analysis containing the assembled genomes and the two SNP projects is available online and can be accessed at http://enterobase.warwick.ac.uk/species/ecoli/search_strains?query=workspace:4462, while raw reads have been deposited to NCBI under project PRJNA517036. In addition, for confirmation of the EnteroBase analysis, alignment of raw reads to the reference E. coli O157:H7 Sakai genome (Assembly: GCF_000008865.1) including plasmids pO157 and pOSAK1 (Accession numbers: NC_002695 for the chromosome, NC_002128 for pOSAK1 and NC_002127 for pO157) was performed using Bowtie2. 16 For the purposes of mapping, the chromosome and two plasmids were used as reference in a single multifasta file. Bam files were sorted and indexed using Samtools 17 and visualised in Artemis genome browser 18 for manual validation. Qualimap 19 was used to calculate mapping statistics. Finally, SNP and small indel detection was also performed using VarScan2. 20 To confirm the mutations by PCR and Sanger sequencing the following primers were used: btuBF aggaaaggtgcgatgattg btuBFin gttgttgggcgattatgc btuBR cgctcaacaataaacgcttc galUF tcggctggtggtactatc galUR tccctcgacgatttcctg waaGF atgagatgtatctttcggttattcc waaGR tgcccgccatttcaaatc waaQF cttgtggataagccatttcg S84

Transcriptomic analysis
The Fastq read outputs from the MiSeq instrument were mapped using Bowtie2 against the E. coli O157:H7 Sakai genome. As for the genome analysis, the chromosomal and pO157 and pOSAK1 plasmid sequences were introduced in a single fasta file, which was used as the reference for Bowtie2. Alignment statistics are shown below. Only mapped reads were then used to create bam files and read counts per gene were calculated using BEDtools coverageBed. 21 Differential expression analysis was performed using DESeq2. 22 For increased accuracy we used multi-factor design to account for sample pairing and hence differences between the samples, whilst measuring the effect of the treatment. The design formula used was thus dds <-DESeqDataSet (se, design = ~replicate + condition), whereby condition refers to treatment).
Results show differentially expressed genes with a false discovery rate (FDR) cutoff of 0.05 and a P-adjusted value less than 0.05. No fold-change cut off was employed as we argue this would introduce an artificial assumption bias based on the importance of relative mRNA levels to phenotype. To identify overrepresented biological pathways in the dataset, the KEGG mapper Search&Color Pathway tool 23 was used and STRING 24 was used to visualize connections between differentially expressed genes. Raw reads and normalised count data can be found at the NCBI GEO database under entry GSE125633.
Data S1. (separate file) DESeq2 analysis of the transcriptomic response of E. coli Sakai to Λ-5b. The worksheet named "all genes" contains the result of DESeq2 analysis on the expression of all the genes in the E.
coli Sakai genome (Assembly: GCF_000008865.1; including plasmids pO157 and pOSAK1) in the presence of the compound compared to bacteria treated with solvent only (control). The genes with significantly higher transcript levels (padj < 0.05) when treated with the compound are listed in the worksheet named "up-regulated", while genes with significantly lower transcript levels in the presence of the compound are listed in the worksheet termed "down-regulated".