Antibiotic Korormicin A Kills Bacteria by Producing Reactive Oxygen Species

As multidrug antibiotic resistance in pathogenic bacteria continues to rise, there is a critical need for novel antimicrobial agents. An essential requirement for a useful antibiotic is that it selectively targets bacteria without significant effects on the eukaryotic hosts. Korormicin is an excellent candidate in this respect because it targets a unique respiratory enzyme found only in prokaryotes, the Na⁺-pumping NADH:quinone oxidoreductase (Na⁺-NQR). Korormicin is synthesized by some species of the marine bacterium Pseudoalteromonas and is a potent and specific inhibitor of Na⁺-NQR, an enzyme that is essential for the survival and proliferation of many Gram-negative human pathogens, including Vibrio cholerae and Pseudomonas aeruginosa, among others. Here, we identified how korormicin selectively kills these bacteria. The binding of korormicin to Na⁺-NQR promotes the formation of reactive oxygen species generated by the reaction of the FAD and the 2Fe-2S center cofactors with O₂.

ABSTRACT

Korormicin is an antibiotic produced by some pseudoalteromonads which selectively kills Gram-negative bacteria that express the Na⁺-pumping NADH:quinone oxidoreductase (Na⁺-NQR.) We show that although korormicin is an inhibitor of Na⁺-NQR, the antibiotic action is not a direct result of inhibiting enzyme activity. Instead, perturbation of electron transfer inside the enzyme promotes a reaction between O₂ and one or more redox cofactors in the enzyme (likely the flavin adenine dinucleotide [FAD] and 2Fe-2S center), leading to the production of reactive oxygen species (ROS). All Pseudoalteromonas contain the nqr operon in their genomes, including Pseudoalteromonas strain J010, which produces korormicin. We present activity data indicating that this strain expresses an active Na⁺-NQR and that this enzyme is not susceptible to korormicin inhibition. On the basis of our DNA sequence data, we show that the Na⁺-NQR of Pseudoalteromonas J010 carries an amino acid substitution (NqrB-G141A; Vibrio cholerae numbering) that in other Na⁺-NQRs confers resistance against korormicin. This is likely the reason that a functional Na⁺-NQR is able to exist in a bacterium that produces a compound that typically inhibits this enzyme and causes cell death. Korormicin is an effective antibiotic against such pathogens as Vibrio cholerae, Aliivibrio fischeri, and Pseudomonas aeruginosa but has no effect on Bacteroides fragilis and Bacteroides thetaiotaomicron, microorganisms that are important members of the human intestinal microflora.

IMPORTANCE As multidrug antibiotic resistance in pathogenic bacteria continues to rise, there is a critical need for novel antimicrobial agents. An essential requirement for a useful antibiotic is that it selectively targets bacteria without significant effects on the eukaryotic hosts. Korormicin is an excellent candidate in this respect because it targets a unique respiratory enzyme found only in prokaryotes, the Na⁺-pumping NADH:quinone oxidoreductase (Na⁺-NQR). Korormicin is synthesized by some species of the marine bacterium Pseudoalteromonas and is a potent and specific inhibitor of Na⁺-NQR, an enzyme that is essential for the survival and proliferation of many Gram-negative human pathogens, including Vibrio cholerae and Pseudomonas aeruginosa, among others. Here, we identified how korormicin selectively kills these bacteria. The binding of korormicin to Na⁺-NQR promotes the formation of reactive oxygen species generated by the reaction of the FAD and the 2Fe-2S center cofactors with O₂.

INTRODUCTION

Korormicin is an antibiotic produced by marine bacteria (Fig. 1). It was first isolated from Pseudoalteromonas sp. and synthesized chemically shortly thereafter (1–6). However, after initial characterizations by Hayashi and coworkers, korormicin ceased to be available until recently when it was isolated from a different strain of Pseudoalteromonas (strain J010) by Tebben et al. (7). Korormicin has been shown to act against a wide range of Gram-negative pathogens, including Vibrio cholerae, Aliivibrio fischeri, Vibrio alginolyticus, etc. (5).

FIG 1.

Structure of korormicin A.

Yoshikawa et al. established that the target of korormicin is the Na⁺-pumping NADH-quinone oxidoreductase (Na⁺-NQR), a respiratory enzyme found only in bacteria (Fig. 2). Korormicin does not affect any of the other respiratory complexes from bacteria or mitochondria (5). This enzyme catalyzes electron transfer from NADH to quinone, and, uniquely for a respiratory complex, it uses the energy released to pump Na⁺ rather than H⁺ across the cell membrane (8–10). Na⁺-NQR is an integral membrane complex made up of six subunits, NqrA to NqrF, and at least five redox cofactors including flavin adenine dinucleotide (FAD), 2Fe-2S, two flavin mononucleotides (FMNs), and riboflavin (10, 11).

FIG 2.

Na⁺-NQR electron transfer scheme illustrating probable sites of side reactions with O₂ in the presence of korormicin (green). The structure model was drawn in Swiss-PdbViewer; the structure was obtained from PDB 4P6V.

Hayashi et al. determined that korormicin is a noncompetitive inhibitor with respect to quinone, which indicates that it acts at a distinct binding site (6). They discovered that, in contrast to its strong inhibition of Na⁺-NQR from Vibrio alginolyticus, korormicin is much less effective against the Na⁺-NQR from Haemophilus influenzae (12). They also discovered a random mutant of V. alginolyticus Na⁺-NQR that, like the enzyme from H. influenzae, is much less sensitive to korormicin. They determined that this mutant carries the substitution G140A in NqrB (NqrB-G140A). Hayashi et al. noted that Na⁺-NQR from H. influenzae differs from the enzyme in most other bacteria at two locations with the following substitutions: NqrB-G140C, the same sequence position as the spontaneous mutation in the V. alginolyticus enzyme, and NqrB-G141T (12, 13). To further study the roles of the amino acids at these two sequence positions, we have recently constructed and characterized individual site-directed mutants in the Na⁺-NQR from V. cholerae: NqrB-G140A and NqrB-G141A (14–16). We confirmed that these two glycine residues are essential for korormicin inhibition and also found that mutations at these sites can affect quinone binding, as judged by changes in the apparent K_m [K_m(app)] for quinone during steady-state turnover (14).

We also recently carried out a complete and stereospecific synthesis of korormicin and tested its efficacy on the isolated Na⁺-NQR from V. cholerae (expressed in V. cholerae). We reported 50% inhibitory concentration (IC₅₀) values of 5 nM for the wild-type (WT) enzyme and higher than 50 μM for the NqrB-G140A and NqrB-E144C mutants (16). We also tested the efficacy of korormicin on the H. influenzae Na⁺-NQR (expressed in V. cholerae) and determined that the IC₅₀ value is higher than 50 μM.

The redox cofactors and the pathway for electrons through Na⁺-NQR have been largely defined: electrons move sequentially from NADH to a non-covalently bound FAD in NqrF, a 2Fe-2S center, also in NqrF, a covalently bound FMN in NqrC (FMN_C), a covalently bound FMN in NqrB (FMN_B), and to a riboflavin cofactor and from there to the final electron acceptor, ubiquinone (16–25) (Fig. 2). A newly proposed Fe center located in between NqrD and NqrE may also play a role in the electron transfer pathway (11). Two of these intramolecular electron transfer steps have been shown to be linked to Na⁺ transport and/or generation of membrane potential: 2Fe-2S → FMN_C and FMN_B → riboflavin (22). Also, a number of negatively charged residues that are involved in Na⁺ transport have been identified (26). However, the mechanism that connects the redox reaction to the pumping of Na⁺ is still unknown. Despite the availability of a crystallographic structure (11), the binding site for quinone has also not been completely defined.

The ability of Na⁺-NQR to conserve the energy released in its redox reaction, by pumping Na⁺ across the cell membrane, depends on both facilitating and controlling electron transfer through the cofactors of the enzyme (24). In many respiratory enzymes, correct control of electron transfer apparently also serves to minimize the occupancy of intermediates that could react directly with O₂, thus preventing side reactions that produce reactive oxygen species (ROS) such as superoxide and peroxide. Anything that interferes with correct electron transfer, such as inhibitors or structural perturbations, can potentially cause production of ROS, leading to destruction of the bacterial cell (27–29).

Here, we show that the mechanism by which korormicin inhibits cell growth involves more than mere inhibition of the activity of Na⁺-NQR. When korormicin binds to Na⁺-NQR, this leads to the production of ROS, which in turn can kill the cell. We have found that korormicin is effective in inhibiting the growth of wild-type Vibrio cholerae and Pseudomonas aeruginosa but ineffective against mutant strains that lack Na⁺-NQR. We also determined that the FAD and the 2Fe-2S center are the sites that likely react with O₂ after the binding of korormicin, leading to the formation of ROS. Interestingly, we show that Pseudoalteromonas J010, the marine bacterium that produces korormicin expresses Na⁺-NQR; but the activity of this enzyme is insensitive to the inhibitor, and no ROS is produced even when korormicin is added exogenously.

RESULTS

Korormicin is an effective antibiotic.

(i) Determining the MIC of korormicin in three species of bacteria. Since korormicin has not been commercially available for a number of years, the first step in this project was to secure supplies of this compound. To this end, we isolated korormicin from a natural source, Pseudoalteromonas J010, as described previously (16).

To test the potency of korormicin as an antibiotic, we determined the MICs for three species of bacteria, two of which contain Na⁺-NQR, Vibrio cholerae (O395N1) and Pseudomonas aeruginosa (PAO1), and, as a negative control, Escherichia coli (DH5α) that does not. Tests were carried out with bacteria grown in liquid medium and on plates (see Materials and Methods). MIC values were determined as the lowest concentrations of korormicin that resulted in no colonies on LB plates or in colorless wells after the addition of 0.2 mg/ml p-iodonitrotetrazolium chloride (INT) (30). Values of 20 ± 10 μM were obtained for both Vibrio cholerae and Pseudomonas aeruginosa (in liquid medium and/or on plates), while the growth of E. coli was not inhibited even by much higher concentrations (1,000 μM) of korormicin. These results are qualitatively similar to the ones from Yosikawa et al. and Tebben et al. (5, 7), except that, in contrast to our results, Yosikawa et al. found that Pseudomonas aeruginosa was insensitive to korormicin (7).

(ii) Testing korormicin antibiotic effectiveness on a range of bacteria including strains with and without Na⁺-NQR. In agreement with previous reports (5–7), the above described results suggest that the antibiotic activity of korormicin is effective only when the target cells express Na⁺-NQR. There are two reports in which korormicin was tested in bacteria that naturally lack the nqr operon, including E. coli and Staphylococcus aureus (1, 7). To explore this further, we used the MIC values determined for V. cholerae and P. aeruginosa to test the effectiveness of korormicin against a number of other bacteria, some of which have Na⁺-NQR and some that do not (Table 1). We first examined the effects of korormicin on the growth of mutant strains of V. cholerae and P. aeruginosa from which the genes coding for Na⁺-NQR were deleted (Δnqr strains). In both cases, the growth of the Na⁺-NQR deletion strain was almost completely unaffected by korormicin even at concentrations as high as 100 μM, whereas growth of the corresponding wild-type strain was almost completely suppressed at 20 μM. The V. cholerae Δnqr strain normally grows at less than half the rate of the wild type, but this growth rate is almost completely unaffected by korormicin. As a control, we tested the effect of 20 μM korormicin on a complementation strain of V. cholerae Δnqr. In this strain, Na⁺-NQR is expressed in the pBAD plasmid using 0.05% arabinose as an inducer. Korormicin was as effective in killing the recombinant strain as it was the wild-type cells but only when arabinose was added to the culture (Table 1).

TABLE 1.

Effect of korormicin on growth of various bacterial strains including ones with mutant Na⁺-NQR^a

Strain	Growth (OD600) in liquid medium			Growthc on plates
	−Korormicin A	+Korormicin A	Ratiob	−Korormicin A	+Korormicin A
Vibrio cholerae O395N1 (WT)	1.3 ± 0.1	0.04 ± 0.02	0.03
Δnqr Vibrio cholerae O395N1	0.6 ± 0.1	0.6 ± 0.2	1.0
Pseudomonas aeruginosa PAO1 (WT)	1.7 ± 0.1	0.04 ± 0.01	0.02
Δnqr Pseudomonas aeruginosa PAO1	1.3 ± 0.2	1.2 ± 0.2	0.92
Nqr-pBAD expressed in Δnqr Vibrio cholerae O395N1	1.2 ± 0.2	1.1 ± 0.2	0.92
NqrB-G141A-pBAD expressed in Δnqr Vibrio cholerae O395N1	1.1 ± 0.1	1.1 ± 0.2	1.0
Vibrio cholerae O395N1-Δnqr expressing the H. influenzae Na+-NQR	0.94 ± 0.2	1.5 ± 0.2	1.6
Pseudoalteromonas citrea	2.1 ± 0.3	0.19 ± 0.2	0.09
Pseudoalteromonas J010	1.9 ± 0.2	1.8 ± 0.1	0.95
Aliivibrio fischeri				+	−
Pseudoalteromonas luteoviolacea				+	−
Escherichia coli DH5α				+	+
Escherichia coli Nissle				+	+

^aThe OD₆₀₀ was measured after 7.5 h of growth starting with the same number of cells with (+) or without (−) korormicin A. The plate experiments were done plating 5 × 10⁸ cells on the corresponding plates and locating a disk soaked in korormicin. In all experiments 20 μM korormicin A was used. All measurements were repeated at least five times from three different independent experiments. Errors indicate one standard deviation.

^bValues represent the OD₆₀₀ with korormicin A/OD₆₀₀ without korormicin A. Instances of killing by korormicin are represented by values higher than 1; resistance to killing is represented by values equal to or lower than 1.

^c+, growth; −, inhibition of growth.

We then determined the effect of 20 μM korormicin on a number of other bacterial strains, including Aliivibrio fischeri, Pseudoalteromonas citrea, and Pseudoalteromonas luteoviolacea that naturally have Na⁺-NQR and two E. coli strains that lack Na⁺-NQR. In all of these cases, only the strains that express Na⁺-NQR were susceptible to korormicin (Table 1).

(iii) Testing korormicin antibiotic effectiveness in bacteria with korormicin-insensitive variants of Na⁺-NQR. The mutant strain of V. cholerae (described above) in which the operon for Na⁺-NQR from H. influenzae replaced that of the native enzyme was not affected by korormicin (Table 1). This is apparently because the Na⁺-NQR from H. influenzae is itself resistant to korormicin, as described above. This suggests that in order for korormicin to kill bacteria, they not only must have an active Na⁺-NQR but also must express an enzyme that is susceptible to inhibition by korormicin.

As described earlier (14–16), we have made single-point mutants of the V. cholerae Na⁺-NQR enzyme that show almost no sensitivity to korormicin, including NqrB-G140A and NqrB-G141A. Consistent with the above results, V. cholerae cells that express genes for these mutant enzymes instead of the wild-type Na⁺-NQR are also able to grow well in the presence of korormicin (Table 1).

We noticed that the Na⁺-NQR sequences from both Bacteroides fragilis and Bacteroides thetaiotaomicron have leucine rather than glycine at the sequence position corresponding to NqrB-G141 (Fig. 3), suggesting that these enzymes are likely to also be insensitive to korormicin. In both cases, membrane preparations from the bacteria exhibited NADH-quinone oxidoreductase activity that was dependent on the presence of NaCl (100 mM versus no added NaCl), confirming the presence of active Na⁺-NQR. In both, this activity was insensitive to korormicin up to 400 nM, or a concentration about 100 times higher than the IC₅₀ for Na⁺-NQR from V. cholerae.

FIG 3.

Structural origin of korormicin sensitivity. Sequence alignment of a portion of the NqrB subunit shows the G140 and G141 residues (V. cholerae numbering) present in every known korormicin-sensitive Na⁺-NQR, with the table below highlighting this relationship.

Thus, for korormicin to be effective as an antibiotic, it apparently needs the target cells to have an active Na⁺-NQR enzyme that it can inhibit. However, the fact that the Na⁺-NQR deletion strains are not sensitive to korormicin indicates that the antibiotic activity of korormicin on cells is not a direct result of the inhibition of Na⁺-NQR per se since these cells already lack Na⁺-NQR activity. That is, in the presence of korormicin, cells without Na⁺-NQR grow better than cells with Na⁺-NQR.

Antibiotic activity of korormicin is accompanied by elevated production of ROS.

(i) ROS measurements. If the antibiotic activity of korormicin is not due to the direct effects inhibiting electron flow through Na⁺-NQR, the next most likely explanation is that korormicin binding alters Na⁺-NQR in some way that promotes a side reaction in which electrons are diverted to O₂, producing superoxide. A process of this kind has been described for HQNO (2-heptyl-4-hydroxyquinolone-N-oxide) inhibition of the cytochrome bc₁ complex in P. aeruginosa (27, 31).

We therefore measured production of reactive oxygen species (ROS) in V. cholerae cells growing in the presence and absence of korormicin. We used the ROS indicator 2',7'-dichlorodihydrofluorescein diacetate (H2DCFDA), a cell-permeant, chemically reduced derivative of fluorescein. As shown in Table 2, there is a dramatic increase in ROS production in the presence of korormicin.

TABLE 2.

Reactive oxygen species formation in the presence and absence of korormicin as shown by H2DCFDA fluorescence

Strain	Relative H2DCFDA fluorescencea
	−Korormicin A	+Korormicin A	−Korormicin A/+korormicin A
V. cholerae O395N1 (WT)	1.0 ± 0.2	8.4 ± 0.7	0.12
Δnqr V. cholerae O395N1	1.8 ± 0.6	1.9 ± 0.2	0.95
NqrB-G141A (pBAD) expressed in Δnqr V. cholerae O395N1	0.35 ± 0.1	0.42 ± 0.1	0.83
V. cholerae O395N1-Δnqr expressing the H. influenzae Na+-NQR (+0.05% arabinose)	0.3 ± 0.1	0.4 ± 0.1	0.75
Pseudoalteromonas J010	0.9 ± 0.18	0.6 ± 0.1	1.5

^aCultures were treated with korormicin A (+koromicin; final concentration, 19.4 μM) or an equal volume of dimethyl sulfoxide (−korormicin). ROS was measured using the probe H2DCFDA. One hour after korormicin A treatment, the cultures were treated with H2DCFDA at a concentration of 9.6 μM for 2 h. Fluorescence intensity was measured on a Tecan plate reader at 495 nm, and values represent H2DCFDA fluorescence relative to that of WT Vibrio cholerae. Instances of high ROS production in the presence of korormicin are boldface; low ROS production is underlined.

(ii) Elevated ROS is not observed in bacteria that lack Na⁺-NQR or in strains that have a korormicin-insensitive Na⁺-NQR. We used the same method to measure ROS production in the Na⁺-NQR deletion strain (Δnqr strain) of V. cholerae in the presence and absence of korormicin (Table 2). Consistent with our hypothesis, this strain shows relatively low production of ROS, which does not increase significantly in the presence of korormicin. Similarly, when the V. cholerae strain that expresses the Na⁺-NQR from H. influenzae is grown in the presence of korormicin, there is essentially no effect on growth rate and only a small increase in ROS production. Also, the NqrB-G141A mutant, which is resistant to korormicin, produced relatively smaller amounts of ROS. This is consistent with the fact that the Na⁺-NQR from H. influenzae and the mutant NqrB-G141A are not susceptible to korormicin inhibition (see above).

Antibiotic activity of korormicin is not observed in the absence of O₂.

Production of ROS almost certainly occurs through a reaction in which electrons from Na⁺-NQR are donated to O₂. Without O₂, this reaction should not take place, thus halting the mechanism by which korormicin kills bacteria. To test this, we repeated the treatment of V. cholerae cells with korormicin under anaerobic conditions (using the anaerobic system GasPak) with fumarate as a final electron acceptor (32). The recombinant V. cholerae Na⁺-NQR was grown under anaerobic conditions, and expression, induced by arabinose, was verified by Western blotting against a histidine tag located at the end of NqrF (33). The presence of active Na⁺-NQR in these anaerobically grown cells was confirmed by measuring Na⁺-dependent NADH-quinone oxidoreductase activity in the cell membranes. Under these anaerobic conditions, korormicin caused no inhibition of growth either in cells growing on plates or in liquid medium. These results confirm that korormicin requires O₂ for its antibiotic activity.

The cofactor(s) in Na⁺-NQR responsible of the generation of ROS.

So far, we have shown that inhibition of Na⁺-NQR by korormicin in the presence of O₂ leads to generation of ROS. The next step was to determine which cofactor or cofactors in Na⁺-NQR are responsible for the reaction with oxygen. To this end, we measured the intrinsic ROS as described above in mutants of Na⁺-NQR that lack one or more specific redox cofactors. The basis of this experiment is that, in keeping with the pathway of electron transfer through the enzyme shown in Fig. 2, when a mutation removes one of the cofactors, this introduces a block to electron flow, and the reduced form of upstream cofactors will tend to become populated, promoting their ability to react with O₂. If the reduced form of a given cofactor is able to react directly with O₂, deletion of downstream cofactors would be expected to promote the oxygen reaction and the formation of ROS, whereas deletion of that cofactor itself should suppress ROS formation.

Table 3 shows fluorescence-based ROS measurements for a series of cofactor deletion mutants of Na⁺-NQR. We examined ROS formation in mutants that lack the following cofactors (in order of the electron transfer pathway): the non-covalently bound FAD (NqrF-S245A), the 2Fe-2S center (NqrF-C70A), the proposed single Fe coordinated by cysteines in NqrD and NqrE (NqrD-C20A, and NqrE-C120A), the covalently bound FMN_C (NqrC-T225Y), and the covalently bound FMN_B (NqrB-T236Y). The FMN_C and FMN_B deletions showed the highest levels of fluorescence, and the single-Fe center mutants showed elevated fluorescence, while the FAD and 2Fe-2S cofactor mutants both showed relatively low fluorescence. This suggests that the FAD and 2Fe-2S cofactors are most likely to be responsible for the reaction with O₂ and the generation of oxygen radicals.

TABLE 3.

Reactive oxygen species formation in WT Vibrio cholerae and mutants that lack specific Na⁺-NQR redox cofactors

Strain	Normalized fluorescencea
Vibrio cholerae O395N1 (WT)	1.0
Δnqr Vibrio cholerae O395N1	1.8 ± 0.6
Δnqr Vibrio cholerae O395N1 with recombinant nqr expressed in pBAD (+0.05% arabinose)b	0.9 ± 0.2
ΔFAD (NqrF-S245A)	1.2 ± 0.1
ΔFeS (NqrF-C70A)	1.9 ± 0.1
ΔFMNC (NqrC-T225Y)	6.6 ± 0.1
ΔFMNB (NqrB-T236Y)	9.3 ± 1.1
ΔFe (NqrD-C112A)	4.2 ± 0.2
Vibrio cholerae O395N1-Δnqr expressing the Haemophilus influenzae Na+-NQR in pBAD (+0.05% arabinose)	0.9 ± 0.3

^aROS was measured using the probe H2DCFDA. One hour after arabinose induction the cultures were treated with H2DCFDA at a concentration of 9.6 μM for 2 h. Fluorescence intensity was measured on a Tecan plate reader at 495 nm. Instances of high ROS production are boldface; low ROS production is underlined.

^bRecombinant WT.

The Na⁺-NQR of Pseudoalteromonas J010 that produces korormicin is resistant to korormicin inhibition while Na⁺-NQRs from some nonproducing Pseudoalteromonas strains are sensitive.

Bacteria of the genus Pseudoalteromonas are ubiquitous and diverse marine organisms that have attracted interest because they synthesize many bioactive compounds, some of which can act as antibiotics against other bacteria. The strain J010 was isolated from the crustose alga Neogoniolithon fosliei; the bacteria and the algae form an association which seems to protect the algae against colonization by other bacteria in the ocean, most likely Vibrio species. The various species and strains of Pseudoalteromonas seem to have different capabilities of producing secondary metabolites. Tebben et al. found that the J010 strain produces seven different korormicins and a new bromopyrrole, among other compounds. They tested all seven korormicins and found that korormicin A is the most abundantly produced and the most effective as an antibiotic (7). For the current study, only korormicin A was used.

Activity of Pseudoalteromonas Na⁺-NQR.

We sequenced the genome of Pseudoalteromonas J010 and found that it contains the nqr operon, as do many marine bacteria. This raised the question of how these bacteria are able to survive if they simultaneously produce korormicin and Na⁺-NQR. To confirm that Na⁺-NQR is actually active in the bacteria at the same time that korormicin is present, the J010 strain was grown to early stationary phase, the point where the bacteria are typically harvested to obtain korormicin. Membranes were then isolated, and NADH-quinone oxidoreductase activity was measured in the presence and absence of Na⁺. The quinone reductase activity was clearly accelerated by 20 to 50 mM NaCl, confirming that it was due to Na⁺-NQR. This activity was unaffected by 20 μM korormicin A, an amount that completely abolishes activity in V. cholerae, indicating that the Na⁺-NQR from J010 is insensitive to the antibiotic. We noticed that the J010 sequence (Fig. 3) includes the substitution NqrB-G141A (V. cholerae numbering). This is the same sequence position as the NqrB-G141A mutation that renders the V. cholerae Na⁺-NQR (and, thus, also the bacterium) insensitive to korormicin. This sequence substitution almost certainly accounts for the korormicin A insensitivity of the J010 Na⁺-NQR and explains why this strain is itself able to produce korormicin(s).

Interestingly, most other strains of Pseudoalteromonas do not produce korormicin. We hypothesized that these strains would have no need for a korormicin-insensitive Na⁺-NQR. Indeed, when the effect of korormicin was tested on two such strains, P. citrea and P. luteoviolacea, both were sensitive to the antibiotic (7) (Table 1). In both cases, the sequence shows that NqrB has the double glycine, corresponding to NqrB-G140/NqrB-G141 in V. cholerae, that is typical of an Na⁺-NQR that is sensitive to korormicin (Fig. 3). Thus, the NqrB-G141A (V. cholerae numbering) substitution in Pseudoalteromonas J010 is clearly an adaptation that allows the strain to produce korormicin.

Conclusions.

The ability of korormicin to kill bacteria depends on the presence of an active Na⁺-NQR in the cells. Importantly, for korormicin to work as an antibiotic, the Na⁺-NQR must itself be sensitive to korormicin. The reason for this turns out to be that korormicin kills cells by an indirect mechanism in which binding to Na⁺-NQR leads to the production of reactive oxygen species (ROS) that cause damage to the cells. Korormicin binding disturbs the normal electron transfer processes in Na⁺-NQR, which apparently promotes a side reaction in which some of the redox cofactors of the enzyme give electrons directly to O₂, producing ROS. Measurements on mutants of Na⁺-NQR that lack specific cofactors indicate that the production of ROS takes place at the FAD and/or 2Fe-2S cofactors. Our measurements show that, under all conditions where korormicin is working as an effective antibiotic, significantly higher levels of ROS are present. Korormicin is ineffective under anaerobic conditions, confirming the role of O₂ in this proposed mechanism. The antibiotic activity of korormicin does not work in bacteria that lack Na⁺-NQR. Korormicin is also ineffective in bacteria that have a version of Na⁺-NQR that is insensitive to the antibiotic. These korormicin-resistant bacteria include B. fragilis and B. thetaiotaomicron, which are important members of the human intestinal microflora. Interestingly, the bacteria that produce korormicin, Pseudoalteromonas J010, also have a korormicin-insensitive Na⁺-NQR (Tables 1 and 2). This work was carried out with korormicin A, but Pseudoalteromonas J010 produces at least six other korormicins (B to G). Dibrov et al. reported that korormicin A can have cytotoxic effects on mammalian cells whereas some synthetic analogs do not (40). The present investigation is a foundation for a much wider study of this entire family of promising antibiotic agents.

MATERIALS AND METHODS

Isolation of korormicin from Pseudoalteromonas J010.

Pseudoalteromonas J010 was grown in a 30-liter fermentor in marine broth at 28°C with aeration at 20 liters/min and constant agitation at 200 rpm. Cells were grown for 48 h and harvested in stationary phase.

Bacterial strains and growth conditions.

Table 4 shows the bacterial strains used in this study.

TABLE 4.

Strains used in this work

Strain	Genotype and/or characteristic(s)	Reference or source
Escherichia coli strains
DH5α	λ− ϕ80dlacZΔM15 Δ(lacZYA-argF)U169 recA1 endA1 hsdR17(rK− mK−) supE44 thi-1 gyrA relA1	Laboratory stock
Nissle 1917	Serovar O6:K5:H1	Laboratory stock
Pseudoalteromonas J010	Isolate	C. Motti (7)
Pseudoalteromonas citrea		ATCC 29719
Pseudoalteromonas luteoviolacea		ATCC 29581
Pseudomonas aeruginosa strains
PAO1	Wild type	38
Transposon deletion of PA2994 (nqrF)	PA2994::ISlacZ-hah	38
Aliivibrio fischeri		ATCC 7744
Vibrio cholerae strains
O395N1 (WT)	O1 serotype, classical biotype, parental	39
Δnqr O395N1	Deletion of the nqr operon in O395N1	39
nqr-pBAD-O395N1	Recombinant nqr operon cloned and expressed in the pBAD-TOPO vector	33
NqrB-G141A	Site-directed mutant of the glycine 141 to alanine in the nqrB gene, cloned into the pBAD vector and expressed in the Δnqr strain	14
Haemophilus influenzae nqr-pBAD	nqr operon from Haemophilus influenzae (strain RD [KW20], ATCC 51907), cloned into the pBAD vector and expressed in the Δnqr strain from Vibrio cholerae	Laboratory strain
Bacteroides fragilis strain ADB77	TM400 ΔthyA	M. Malamy (34)
Bacteroides thetaiotaomicron strain VP5482		M. Malamy

MIC.

To determine the MICs of korormicin against V. cholerae O395N1, the V. cholerae Δnqr strain, Pseudomonas aeruginosa (PAO1), the Δnqr P. aeruginosa strain, and E. coli strains DH5α and Nissle, a korormicin stock solution was serially diluted (2-fold dilution per step) into Difco Antibiotic Medium 3 in 96-well microplates. Different initial concentrations of korormicin (10, 20, 50, 100, and 1,000 μM) were tested, and a similar 2-fold serial dilution of erythromycin was used as a positive control. Each well containing 100 μl of test solution was inoculated with 10 μl of an appropriately diluted fresh culture to reach 5 × 10⁵ CFU/ml per well. The plates were covered and incubated at 37°C and 225 rpm for 20 h, and bacterial growth was detected by adding 40 μl of 0.2 mg/ml p-iodonitrotetrazolium chloride (INT) to each well, followed by a 30-min incubation under the same condition. Development of a pinkish appearance indicated bacterial growth, whereas colorless wells indicated growth inhibition. MIC values were recorded as the lowest concentrations of korormicin showing colorless wells. The assays were performed in triplicate, and the entire procedure was repeated twice. For all subsequent experiments, korormicin was used at the determined MIC value or higher.

Bacterial growth.

Vibrio cholerae WT (O395N1) and mutant derivatives, including the strain expressing recombinant Na⁺-NQR from Haemophilus influenzae, were grown aerobically in LB (Miller) medium in the presence of 50 μg/ml streptomycin and, when needed, 100 μg/ml ampicillin. Pseudomonas aeruginosa WT (PAO1) and ΔnqrF mutants were obtained from the Pseudomonas Collection. The mutation in the nqrF gene is an insertion of a transposon which was verified by DNA sequencing. The wild type and the mutant were grown in LB (Miller) medium. Escherichia coli strains DH5α and Nissle were also grown in LB (Miller) medium at 37°C with constant agitation (200 rpm.) Pseudoalteromonas J010, Pseudoalteromonas citrea, and Pseudoalteromonas luteoviolacea were grown in marine broth at 28°C with constant agitation (200 rpm). Bacteroides fragilis and Bacteroides thetaiotaomicron were grown in an anaerobic chamber (5% H₂, 10% CO₂, and the balance N₂), at 37°C with constant agitation (150 rpm) in supplemented brain heart infusion (BHIS) medium and basal medium, respectively (34, 35). For all experiments testing korormicin, cells were inoculated in liquid cultures. Korormicin at 20 μM was added in early exponential phase (3 h), and growth continued to approximately 7.5 h after inoculation. At this point, the optical density at 600 nm (OD₆₀₀) was measured, and cells were plated to measure the number of CFU (data not shown).

Anaerobic growth of Vibrio cholerae to test korormicin effectiveness.

Wild-type V. cholerae and the strain expressing the recombinant Na⁺-NQR (in the presence of arabinose) were grown in an anaerobic jar in the presence of a GasPak anaerobic system in LB (Miller) medium in the presence of 40 mM sodium fumarate, pH 8 (32).

Growth for Na⁺-NQR activity measurements.

V. cholerae cells expressing recombinant Na⁺-NQR variants were grown in LB medium using arabinose as inducer of the nqr operon as reported previously (33). Cells were harvested at the end of the exponential phase of growth and washed with 50 mM Tris-HCl, pH 8, 300 mM NaCl, and 1 mM MgSO₄ (Tris buffer).

Pseudoalteromonas J010 was grown in marine broth at 28°C with constant agitation. Cells were harvested at the stationary phase (48 h). Cells were harvested with Tris buffer.

Bacteroides fragilis and Bacteroides thetaiotaomicron were grown as reported above. Cells were washed anaerobically with 40 mM KH₂PO₄, pH 7, and 5 mM dithiothreitol (DTT; KPi buffer) (see below).

Membrane preparation and Na⁺-NQR activity measurements.

Cells were broken using a microfluidizer (model 110S) operated at a gas input pressure of 80 lb/in² in Tris buffer. Broken cells were separated by centrifugation, and the supernatant was centrifuged in an ultracentrifuge at 100,000 × g for at least 5 h. In the case of the Bacteroides strains, these procedures were carried out in anaerobic conditions using 40 mM KH₂PO₄, pH 7, 5 mM DTT (KPi buffer), instead of Tris buffer. Cells were broken using a French pressure cell operated at 25,000 lb/in². Membranes were washed with Tris buffer or deaerated KPi buffer and then stored at −80°C until used.

Na⁺-NQR activity was measured spectrophotometrically in membrane fractions as reported previously (18) using NADH as an electron donor and ubiquinone-1 as the electron acceptor with and without the addition of 100 mM NaCl.

ROS detection.

V. cholerae cells were grown overnight in LB medium (Miller) with the appropriate antibiotics. After approximately 16 h of growth, cells were diluted 50-fold in fresh LB medium with antibiotics in a 48-well plate. In the case of recombinant Na⁺-NQR expression, arabinose was added to the cultures at the time of inoculation. After 3 h of growth (mid-exponential phase), cells were treated with 19.4 μM korormicin, and growth was continued for 30 min. At this time, H2DCFDA was added to a final concentration of 9.6 μM. The incubation continued in the dark, and at 7.5 h after the start of growth, the OD was measured. Fluorescence was measured in a Tekan Infinite M1000 Pro plate reader (excitation, 495 nm; emission, 527 nm). In the case of Pseudoalteromonas, cells were grown in marine broth at 28°C and treated in the same way as the V. cholerae cells, but Pseudoalteromonas cells were harvested at 9.5 h for fluorescence measurement.

Genomic sequencing and analysis of Pseudoalteromonas J010.

Genomic DNA from Pseudoalteromonas J010 was isolated from a 20-ml 48-h culture using a PureLink genomic DNA kit (ThemoFisher). A Nextera XT kit was used to shear the genomic DNA to fragments with an average size of approximately 1.2 to 1.5 kb and to prepare the DNA library for sequencing. The DNA was then sequenced on an Illumina sequencer (iSeq 100 Sequencing System). Sequence reads were assembled using an A5-miseq pipeline (version 20160825) (36), and the assembled contig sequences were subjected to gene annotation using PROKKA (version 1.12) (37) in order to identify the genes in the nqr operon.

All measurements were repeated at least five times from each of three different biological replicas.

Data availability.

The sequence of Pseudoalteromonas J010 was deposited in GenBank under accession number RDBW00000000 and BioProject number PRJNA497110.