Knockout of adenylosuccinate synthase purA increases susceptibility to colistin in Escherichia coli

Abstract Colistin is a cationic cyclic antimicrobial peptide used as a last resort against multidrug-resistant gram-negative bacteria. To understand the factors involved in colistin susceptibility, we screened colistin-sensitive mutants from an E. coli gene-knockout library (Keio collection). The knockout of purA, whose product catalyzes the synthesis of adenylosuccinate from IMP in the de novo purine synthesis pathway, resulted in increased sensitivity to colistin. Adenylosuccinate is subsequently converted to AMP, which is phosphorylated to produce ADP, a substrate for ATP synthesis. The amount of ATP was lower in the purA-knockout mutant than that in the wild-type strain. ATP synthesis is coupled with proton transfer, and it contributes to the membrane potential. Using the membrane potential probe, 3,3′-diethyloxacarbocyanine iodide [DiOC2(3)], we found that the membrane was hyperpolarized in the purA-knockout mutant compared to that in the wild-type strain. Treatment with the proton uncoupler, carbonyl cyanide m-chlorophenyl hydrazone (CCCP), abolished the hyperpolarization and colistin sensitivity in the mutant. The purA-knockout mutant exhibited increased sensitivity to aminoglycosides, kanamycin, and gentamicin; their uptake requires a membrane potential. Therefore, the knockout of purA, an adenylosuccinate synthase, decreases ATP synthesis concurrently with membrane hyperpolarization, resulting in increased sensitivity to colistin.


Introduction
The emergence of antibiotic-resistant bacteria is a serious medical concern.More than one million people died from antibioticresistant bacterial infections worldwide in 2019 (Antimicrobial Resistance Collaborators 2022 ), indicating the urgent need for dev eloping str ategies to ov ercome antibiotic r esistance.Colistin is used against multidrug-resistant gram-negative bacteria, such as Pseudomonas aeruginosa and Acinetobacter baumanii, as a last-resort antibiotic (Kassamali et al. 2015, El-Sayed Ahmed et al. 2020 ).Colistin has a positive charge, and it electrostatically interacts with LPS, disrupting the outer membrane; in addition, the hydrophobic moiety of colistin disrupts the inner membrane (Velkov et al. 2010, El-Sayed Ahmed et al. 2020 ).The molecular mechanism underlying colistin action remains unclear.LPS modification by twocomponent systems, such as PmrAB and PhoPQ, induces bacterial resistance to colistin by inhibiting the interaction between colistin and the outer membrane (El-Sayed Ahmed et al. 2020 ).
Bacterial membrane potential influences the susceptibility to aminogl ycosides and pol ymyxin B, whic h ar e positiv el y c har ged antimicr obials, because electr ostatic inter actions between these antimicrobial molecules and the bacterial cell membrane are k e y factors for their uptake (Taber et al. 1987, Alteri et al. 2011 ).In Staphylococcus aureus , ATP synthase knoc k out leads to the hyperpolarization of the bacterial membr ane, r esulting in susceptibility to polymyxin B and colistin (Vestergaard et al. 2017 ).T hus , gene m utations affecting membr ane potential can alter the susceptibility to colistin; ho w e v er, the genes affecting both colistin susceptibility and membrane potential are not fully understood.
In this study, we screened colistin-sensitive mutants from an E. coli gene knoc k out libr ary to identify genes that influence colistin susceptibility.The knoc k out of purA , whic h encodes aden ylosuccinate synthase that is responsible for the de novo adenine synthesis from IMP (Jensen et al. 2008 ), results in a colistin-sensitive phenotype; purA knoc k out induces colistin sensitivity by altering the membrane potential.

Bacterial strains and culture conditions
Esc heric hia coli BW25113 and its gene-knoc k out m utant wer e grown in Luria-Bertani (LB) agar plates.A single colony was inoculated into the LB liquid medium and aer obicall y cultur ed at 37 • C with a gitation.Esc heric hia coli str ains tr ansformed with pMW118 were cultured in the LB liquid medium supplemented with 100 μg/mL ampicillin and 1 mM IPTG.The bacterial strains and plasmids used in this study are listed in Table S1 .

Screening of colistin-sensiti v e strains
Esc heric hia coli gene-knoc k out m utants fr om the K eio libr ary (Baba et al. 2006, Yamamoto et al. 2009 ) were cultured in LB medium supplemented with 50 μg/mL kanamycin in a 96-well microplate, with shaking using a microplate shaker (BR-034P, Taitec), at 37 • C o vernight.T he o vernight cultures were inoculated into a fresh LB medium supplemented with 200 ng/mL colistin in a 96-well micr oplate and wer e cultur ed at 37 • C for 24 h without shaking.The OD 595 values of the cultures were measured using a microplate reader (Multiskan FC, Thermo Fisher Scientific).

Bacterial killing assay
Esc heric hia coli overnight culture was diluted to 1 × 10 6 CFU/mL in LB medium.For the colistin-killing assay, the diluted culture was supplemented with 0.8 μg/mL colistin and was incubated at 37 • C for 30 min without shaking.The culture was serially diluted 10-fold and spread on an LB agar plate and incubated overnight at 37 • C. The colonies were counted to calculate the number of surviving bacteria.For the aminoglycoside-killing assay, the diluted culture was supplemented with 17.5 μg/mL kanamycin or 10 μg/mL gentamicin and incubated at 37 • C for 60 min without shaking.The culture was serially diluted 10-fold; 5 μL was spotted onto LB agar plates, which were then incubated overnight at 37 • C.

Genetic manipulation
To obtain the purA -gene knoc k out m utant, tr ansduction with P1 vir was performed using JW4135-KC as the phage donor and BW25113 as the r ecipient str ain.The kanamycin-r esistant marker in the transductant was deleted by introducing pCP20 expressing FLP r ecombinase, r esulting in the purA -gene knoc k out m utant (markerless deletion).To construct a complementation plasmid, a DNA fr a gment carrying the purA gene was amplified via PCR using the genomic DNA of BW25113 as the template and specific primers ( Table S2 ); the amplified gene was inserted into the KpnI and BamHI sites of pMW118, resulting in pMW118-purA.

Measurement of the membrane potential
Membr ane potentials wer e measur ed as described pr e viousl y (Hudson et al. 2020 ) with minor modifications.An aliquot (50 μL) of the overnight culture was inoculated into 5 mL fresh LB medium supplemented with 100 μg/mL ampicillin and 1 mM IPTG and was cultured with shaking until an OD 600 of 0.5 was r eac hed.The cultures (1 mL) were centrifuged to collect bacterial cells; the cells were suspended in PBS.EDTA (10 mM) was added to the cell suspension; the suspension was incubated at room temperature for 5 min and then centrifuged.The cells were suspended in a buffer (500 μL) (130 mM NaCl, 60 mM Na 2 HPO 4 , 60 mM NaH 2 PO 4 , 10 mM glucose, 5 mM KCl, 0.5 mM MgCl 2 ); 30 μM DiOC2(3) with or without 1 mM CCCP was added to the cell suspension.The cells w ere w ashed with the same buffer, and fluorescence (RFU/s) w as measured using a fluorescence microplate reader (Fluoroskan Ascent CF, Thermo Fisher Scientific) with excitation at 485 nm and emission at 678 nm.

Measurement of cell surface charge by cytochrome C binding assay
Cell surface c har ge was measur ed as described pr e viousl y (Gasc h et al. 2013 ) with minor modifications.Ov ernight cultur es (5 mL) were centrifuged to collect the bacterial cells; they were washed twice with buffer (20 mM MOPS, 5 mM sodium citrate, 1 mM EDTA, pH7.0).The cells were suspended in the same buffer and the OD 600 was adjusted to 21; 500 μg/mL cytoc hr ome C was added to the cells and incubated at 37 • C for 15 min.The cells were centrifuged, and the OD 530 value of the supernatant was measured.The concentr ation of cytoc hr ome C in the supernatant was calculated using a standard curve.

Measurement of zeta potential
Ov ernight cultur es of E. coli wer e diluted 500-fold using PBS.The zeta potential was measured using Zetasizer Nano ZSP (Malvern) and disposable folded capillary cells (DTS1070, Malvern).Equilibration was performed for 300 s before the measurement of each sample.

Quantifying ATP
ATP le v els wer e measur ed as described pr e viousl y (Ryuno et al. 2019 ) with minor modifications.Overnight cultures (50 μL) were inoculated into fresh LB medium (5 mL) supplemented with 100 μg/mL ampicillin and 1 mM IPTG and cultured to OD 600 = 0.5 or 0.8.The culture (100 μL) was mixed with an equal volume of ethanol and incubated on ice for 15 min.The sample was centrifuged, and the supernatant was diluted 100-fold with Milli-Q water.The sample (50 μL) was mixed with an equal volume of solution containing firefly luciferase and the substrate (Kikkoman, J apan); the luminescence w as measur ed immediatel y using a luminometer (Promega).

Sta tistical anal ysis
All data were analyzed using one-way ANOVA with post hoc Dunnett's test using Prism 9 (Gr a phP ad softwar e).

purA knockout increased susceptibility to colistin
We searched for colistin-sensitive mutants in the E. coli gene knoc k out libr ary (K eio collection) (Baba et al. 2006 ) and identified 21 gene knoc k out m utants that could not grow in the presence of 200 ng/mL colistin ( Table S3 ).There were mutants of atpB , atpE , and atpH genes encoding ATP synthase subunits ( Table S3 ).We focused on PurA, an enzyme in the de novo purine synthetic pathwa y (Fig. 1 A) that pla ys an important role in bacterial virulence (Ivanovics et al. 1968, Sigwart et al. 1989, Faith et al. 2012, Connolly et al. 2017 ) and analyzed the mechanism of colistin susceptibility.To confirm the increased susceptibility of the purA-knoc k out mutant to colistin, we performed a bactericidal assay.Following treatment with colistin for 30 min, the number of viable bacterial cells was 100-fold lower in the purA -knoc k out m utant compar ed to that in the wild-type strain (Fig. 1 B).The number of viable bacterial cells in the purA -knoc k out m utant was r estor ed by the intr oduction of purA (Fig. 1 B).Ther efor e, purA knoc k out incr eases colistin susceptibility.

purA knockout decreased the amount of ATP
We speculated that purA knoc k out r educes AMP le v els, r esulting in the depletion of ADP and ATP.We examined the growth differences between the wild-type strain and the purA -knoc k out m utant.The gr owth of purA -knoc k out m utant was compar able to that of the wild-type strain until the late exponential phase (OD 600 = 0.8); ho w e v er, the gr o wth w as delay ed, compared to that of the wild-type str ain, fr om the late exponential phase to the stationary phase (Fig. 2 A).We measured the amount of ATP per OD 600 value at the mid and late exponential phases (OD 600 = 0.5 and 0.8).The amount of ATP was lower in the purA -knoc k out m utant than that in the wild-type strain, at both growth time points (Fig. 2 B).The amount of ATP in the purA -knoc k out m utant was r estor ed by the introduction of purA (Fig. 2 B).T herefore , purA knockout decreases ATP levels in bacterial cells.

purA knockout caused hyperpolarization of the bacterial membrane
ATP synthesis is coupled with pr oton tr ansfer, whic h contributes to the membrane potential.We hypothesized that purA knockout decreases ATP synthesis concomitantly with decreased proton transfer, leading to bacterial membrane hyperpolarization.We examined the membrane potential using the membrane potential probe, 3,3 -diethyloxacarboc y anine iodide [DiOC2(3)] (Hudson et al. 2020 ).The purA -knoc k out m utant sho w ed a higher fluorescence intensity than the wild-type strain (Fig. 3 A).The increased fluorescence intensity in the purA -knoc k out m utant was abolished by the introduction of purA (Fig. 3 A).In addition, the fluorescence signal of DiOC2(3) decreased after treatment with the proton uncoupler (CCCP) (Fig. 3 A).T herefore , purA knockout led to membr ane hyper polarization.
Next, we examined whether treatment with CCCP abolished colistin susceptibility in the purA -knoc k out m utant.The CCCP treatment abolished the difference in colistin susceptibility between the purA -knoc k out m utant and wild-type str ains (Fig. 3 B).These results suggested that the colistin susceptibility of the purA -knoc k out m utant could be attributed to membr ane hyperpolarization.

purA knockout increased susceptibility to aminoglycosides
Membrane potential influences the bacterial susceptibility to cationic antimicrobial peptides and aminoglycosides (Taber et al. 1987, Alteri et al. 2011, Liu et al. 2020 ); thus, we examined whether the purA-knoc k out m utant was sensitiv e to aminogl ycosides, using a bactericidal assay.Following treatment with kanamycin or gentamicin, the number of viable bacterial cells was lower in the purA -knoc k out m utant than that in the wild-type strain (Fig. 4 ).The number of viable bacterial cells in the purA -knoc k out m utant was r estor ed by the intr oduction of purA (Fig. 4 ).Ther efor e, purA knoc k out sensitizes E. coli to aminoglycosides.

purA knockout did not alter the cell surface charge
Alteration of cell surface charge following LPS modification influences susceptibility to cationic antimicrobial peptides, such as  colistin (El-Sayed Ahmed et al. 2020 ).We examined the cell surface c har ge in the purA -knoc k out m utant using a binding assay for cytoc hr ome C, a positiv el y c har ged pr otein.The binding of cytoc hr ome C did not differ between the wild-type strain and the purA -knoc k out m utant (Fig. 5 A).Furthermor e, to measur e the cell surface c har ge using another anal ytical method, we examined the zeta potential of bacterial cells and observed no significant difference in the zeta potential between the wild-type strain and the purA -knoc k out m utant (Fig. 5 B).We also tested the w aaGknoc k out m utant as a positiv e contr ol, we found that the m utant exhibited lo w er zeta potential than that of the wild-type strain, consistent with pr e vious r eports (Hyldgaard et al. 2014, Soh et al. 2020 ).These results suggested that the colistin sensitivity in the purA-knoc k out is not r elated to alter ations in the cell surface c har ge.

Discussion
purA knoc k out incr eased bacterial susceptibility to colistin.Membrane potential was hyperpolarized in the purA-knoc k out m utant, and CCCP abolished both membrane hyperpolarization and colistin susceptibility in the purA-knoc k out m utant, indicating that hyper polarization incr eases colistin susceptibility in the purAmutant.In addition, the cell surface charge, which plays an important role in colistin resistance, was not altered in the purA mutant.This study is the first to show that the disruption of de novo purine synthesis pathway increases colistin susceptibility via membrane hyperpolarization.
In the purA -knoc k out m utant, the amount of ATP w as lo w er than that in the wild-type str ain.Knoc k out of purA decreased AMP synthesis, resulting in the depletion of ADP and ATP.F1Fo-A TP synthase couples A TP synthesis and proton transfer; thus, decr eased ATP synthesis decr eases pr oton tr ansfer, leading to hyper polarization (Fig. 6 ).The membr ane hyper polarization could lead to increased uptake of colistin and aminoglycosides (Fig. 6 ).Accor dingly, w e identified the knockout mutants of atpB , atpE , and atpH that encode ATP synthase subunits, as colistin-sensitive mutants ( Table S3 ).The knockout of ATP synthase subunits possibl y decr eases ATP synthesis, leading to the hyper polarization of the membr ane, r esulting in colistin susceptibility.ATP synthase knoc k out leads to a colistin-sensitiv e phenotype in S. aureus (Vestergaard et al. 2017 ), indicating that colistin susceptibility in r esponse to decr eased ATP synthesis is conserv ed between gr amnegative and gram-positive bacteria.Ho w ever, as the purine synthesis pathway is closel y r elated to various cellular processes, we cannot exclude the possibility that factors other than ATP synthesis or membrane hyperpolarization might be involved in colistin susceptibility.The effect of purA -knoc k out on the cellular processes other than the membrane potential should be investigated in future research.
F or Bacillus anthr acis , Salmonella dublin, and Listeria monocytogenes , purA knoc k out decr eases bacterial cells in mouse tissues (Ivanovics et al. 1968, Sigwart et al. 1989, Faith et al. 2012 ).For S. aureus , purA knoc k out attenuates the bacterial killing activity in zebrafish (Connolly et al. 2017 ) and decreases bacterial cells in mouse tissues (Lan et al. 2010 ).For E. coli, purA knoc k out decr eases bacterial growth in human serum (Samant et al. 2008 ).T hus , purA plays an important role in bacterial virulence.Ho w e v er, the mec hanism by which purA contributes to virulence remains unclear.It is possible that the purA -knoc k out m utants hav e decr eased ATP synthesis and are sensitive to the cationic antimicrobial peptides in human serum or animal tissues, resulting in decreased virulence .Further in vestigation is needed to examine the effect of the purA -knoc k out on the bacterial susceptibility to antimicrobial peptides during immunity.
The recent increase in colistin use has resulted in an increase in the number of colistin-resistant bacteria (Dadashi et al. 2022 ).This study suggests that purA knoc k out not onl y decr eases bacterial virulence (Ivanovics et al. 1968, Sigwart et al. 1989, Faith et al. 2012, Connolly et al. 2017 ) but also sensitizes bacteria to colistin.Ther efor e, c hemical molecules that target the de novo purine synthesis pathway could offer two benefits: decreased virulence and susceptibility to cationic antimicrobial peptides and aminoglycosides.In this regard, PurA is a promising antibiotic target.

Figure 1 .
Figure 1.purA knoc k out incr eases susceptibility to colistin.(A) De novo purine synthesis pathway (B) Bacterial cells (1 × 10 6 CFU/mL) wer e tr eated with 0.8 μg/mL colistin for 30 min.The number of viable bacterial cells before and after treatment was measured by plating the samples on LB agar plates and incubating the plates o vernight.T he data are presented as means ± SD of three independent experiments.* * , P < 0.01.ns: not significant.

Figure 2 .
Figure 2. purA knoc k out decr eases the amount of ATP.(A) Gr owth curv es of wild-type and purA-knoc k out m utants.Arr owheads indicate the time point (OD 600 = 0.5 and OD 600 = 0.8) at which samples were taken for analysis in (B).The data are presented as means ± SD of three independent experiments.(B) The amount of ATP in the wild-type strain and purA -knoc k out m utant was measur ed.The data ar e pr esented as means ± SD of six independent experiments at OD 600 = 0.5 experiments and three independent experiments at OD 600 = 0.8 experiments.* * , P < 0.01.ns: not significant.

Figure 3 .
Figure 3. purA knoc k out leads to membr ane hyper polarization.(A) Exponentiall y gr o wing E. coli cells w er e tr eated with or without 1 mM CCCP, stained with DiOC2(3), and the fluorescence was measured.The data are presented as means ± SD of five independent experiments.* * , P < 0.01.ns: not significant.(B) E. coli cells (1 × 10 6 CFU/mL) wer e tr eated with 0.8 μg/mL colistin with or without 50 μM CCCP for 30 min.The number of viable bacterial cells was determined.The data are presented as means ± SD of three independent experiments.* * , P < 0.01.ns: not significant.

Figure 5 .
Figure 5. purA knoc k out does not alter cell surface c har ge.(A) Ov ernight-cultur ed E. coli cells were mixed with cytoc hr ome C, and the amount of cytoc hr ome C bound to the bacterial cells was measured.The data are presented as means ± SD of three independent experiments.ns: not significant.(B) Ov ernight-cultur ed E. coli cells were suspended in PBS and the zeta potential was measured.The data are presented as means ± SD of at least three biological replicates.* * * * , P < 0.0001.ns: not significant.

Figure 6 .
Figure 6.Model of increased colistin-susceptibility caused by purA knoc k out.purA knoc k out decr eased AMP le v els, whic h r educed the le v els of ADP and A TP.A TP synthesis is coupled with proton transfer; ther efor e, decr eased ATP synthesis causes membrane hyperpolarization.Membr ane hyper polarization incr eased the uptak e of positi v el y c har ged antimicr obials suc h as colistin and aminogl ycosides, r esulting in increased susceptibility to these molecules.