- Email: [email protected]
Osmotic Stress in ViableEscherichia colias the Basis for the Antibiotic Response by Polymyxin B
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS ARTICLE NO.
246, 619–623 (1998)
RC988682
Osmotic Stress in Viable Escherichia coli as the Basis for the Antibiotic Response by Polymyxin B Joon-Taek Oh,* Tina K. Van Dyk,† Yolanda Cajal,* Prasad S. Dhurjati,‡ Myron Sasser,§ and Mahendra K. Jain*,1 *Department of Chemistry and Biochemistry and ‡Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716; †DuPont Company, P.O. Box 80173, Wilmington, Delaware 19889-0173; and §MIDI Inc., Newark, Delaware 19713
Received April 14, 1998
Cationic antimicrobial peptides, such as polymyxin B (PxB), below growth inhibitory concentration induce expression of osmY gene in viable E. coli without leakage of solutes and protons. osmY expression is also a locus of hyperosmotic stress response induced by common food preservatives, such as hypertonic NaCl or sucrose. High selectivity of PxB against Gram-negative organisms and the basis for the hyperosmotic stress response at sublethal PxB concentrations is attributed to PxB-induced mixing of anionic phospholipid between the outer layer of the cytoplasmic membrane with phospholipids in the inner layer of the outer membrane. This explanation is supported by PxB-mediated rapid and direct exchange of anionic phospholipid between vesicles. This mechanism is consistent with the observation that genetically stable resistance against PxB could not be induced by mutagenesis. q 1998 Academic Press
Remarkable selectivity of antimicrobial cationic peptides, produced by virtually all organisms (1, 2); remains an enigma. For example, defensins and other proteins from human plasma, magainins from frog skin, cecropins from insect larvae, and thionins from plants do not have a deleterious effect on the organisms that produce them. This fact alone rules out nonspecific mechanisms of action, such as leakage of the cytoplasmic content. Evolutionary success of such antimicrobials also suggests strategies for overcoming drug resistance. 1
Corresponding author. Fax: /302-831-6335. E-mail: mkjain@ udel.edu.
AcylrL-DabrL-ThrrL-DabrL-DabrL-DabrD-Phe L-Leu L-ThrRL-DabRL-Dab Polymyxin B or PxB (Dab Å a, g-diaminobutyric acid) PxB-nonapeptide or NP is devoid of the acyl chain and the first residue Polymyxin B, the focus of this study, is produced by Gram-positive Paenibacillus polymyxa and it is highly selective against Gram-negative organisms. Polymyxins have been in prevalent use for several decades, however there is no indication of the antibiotic resistance in clinical isolates. They act on membranes, yet the antibacterial mechanism is not established. Binding of PxB to surface lipolysaccharides, and consequent disruption of the outer membrane of Gram-negative organisms (3) to permit their entry into periplasmic space of Gram-negative organisms, is a necessary but not a sufficient condition for the antimicrobial effect; for example, PxB-nonapeptide (NP) disrupts the outer membrane, yet it is not an antibiotic (3, 4). Studies with model membranes have provided compelling evidence that these cationic peptides interact with anionic phospholipid prevalent in bacterial membranes, yet they are not ionophores. They induce leakage of the cytoplasmic contents at significantly higher concentrations than the antimicrobial concentrations. Note that other peptides that induce comparable nonspecific membrane changes are not antibiotic in vivo. In this paper we report a strategy to monitor the stress induced by sublethal PxB in viable and growing E. coli. Results show that PxB induces a highly selective expression of osmY gene, as is also induced by hyperosmolar saline and sugar solution.
619
0006-291X/98 $25.00 Copyright q 1998 by Academic Press All rights of reproduction in any form reserved.
AID
BBRC 8682
/
6953$$$221
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS TABLE 1
Response of the Stress Sensitive Strains of E. coli to PxB and Other Additives Response to Fusion gene::lux
Strain #
Stress inducer (conc.)
PxB
NaCl
Sucrose
CCCP
NP
grpE* recA* katG* inaA* lac osmY*
TV1061 DPD2794 DPD2511 DPD2146 TV1048 DPD2170
Protein/ethanol (0.5 M) DNA/mitomycin C (0.3 mM) Oxidative/H2O2 (15 mM) Proton leak/salicylate (1 mM) Limited carbon source (0) Osmotic/ sucrose (0.5 M)
0 0 0 0 0 /
0 0 0 0 0 /
0 0 0 0 0 /
0 0 0 / 0 0
0 0 0 0 0 0
MATERIALS AND METHODS Strains of E. coli. Construction, characterization, and the growth and luminescence response assay conditions are as described before (6-15). Typically, the inducer or peptide (passed through 0.2 mm-pore filter) was added to 5 ml aliquot of cells grown to OD 0.2 at 307 C in LB medium containing 50 mg/ml ampicillin or kanamycin to maintain the plasmid. Cell growth was monitored as OD change at 600 nm. Other specific conditions are described in figure legends. Assay and detailed characterization of the PxB-mediated intervesicle exchange of monoanionic phospholipid has been described before (16, 17). Purity check of commercial PxB (Sigma) and NP (Boehringer) revealed that the trace impurities (õ5% by HPLC) were inactive in the assays described in this paper.
expression of osmY. These results, in conjunction with those described below show that PxB, without entering the cytoplasm, induces transcriptional expression of osmY gene, and that this effect is not due to the disruption of the outer membrane or due to induction of proton leakage by the uncoupler through the cytoplasmic membrane. The growth curve of the lac-lux strain (Figure 1A) was monitored as an increase in the turbidity as well
RESULTS AND DISCUSSION Results in Table 1 show that osmotic stress, without leakage of the cytoplasmic content, is the primary locus for the antimicrobial effect of PxB in Escherichia coli. The physiological basis of the method is that organisms often respond and adapt to sublethal environmental adversities by increased expression of stress proteins to restore homeostasis (5). Thus by coupling transcription initiated at specific promoters to the expression of lux genes it is possible to monitor luminescence in real time as the read-out for the transcriptional response to stress in viable cells (6-9). Each E. coli strain in Table 1 contains a plasmid with a specific stress promoter coupled to the expression of luxCDABE. Thus each strain responds to a specific stress, such as oxidative damage, or proton leakage, or osmotic changes (1015). Since most stress genes are not normally highly expressed under normal growth conditions, the luminescence response to stress in real time typically yields an excellent ratio of signal to background noise. As supported by detailed evidence developed below, results in Table 1 clearly show that, below the minimum inhibitory concentrations (MIC) for growth in 60 minutes, PxB induces only the expression of osmY promoter, as also induced by hyperosmolar NaCl or sucrose. Note that both NP and CCCP (carbonylcyanide m-chlorophenylhydrazone, a proton translocator or uncoupler that depletes the proton gradient) do not induce
FIG. 1. [A] Growth profile for E. coli strain TV1048 in the absence (open symbols) or presence of 0.25 mM polymyxin B (closed symbols) monitored as a change in turbidity at 600 nm (circles) or luminescence (squares). [B] Short-time effect of CCCP, PxB, and NaCl on the luminescence of TV1048. Each aliquot was prepared by dilution of overnight cultured broth with fresh medium followed by shaking 1 hour at 30 7C. Luminescence response (in arbitrary units on Turner luminometer at standard settings) to additives was measured after adding: (inverted triangles) 5 or (diamonds) 30 mM CCCP, (circles) 0.1, (squares) 2, and (triangles) 25 mM PxB and (hexagons) 0.5 M NaCl to a fresh aliquot of the culture.
620
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
as the increase in the luminescence. In TV1048, lac promoter drives expression of the lux gene, and luminescence is observed under normal growth conditions in a moderately carbon limited medium. Note that 0.25 mM PxB shows partial growth inhibition seen not only as lower turbidity, but also as lower luminescence. Specificity of the effect of PxB can be judged from the fact that even up to 20 mM NP there was no noticeable effect on the growth curve monitored by either of the two responses (data not shown). The possibility that PxB depletes proton gradient is ruled out by results in Table 1, as also supported by the short term effect of PxB and CCCP on the intrinsic luminescence from TV1048 strain (Figure 1B). The significance of these results is best evaluated in terms of their MIC: 0.25 mM for PxB and 10 mM for CCCP. Results in Figure 1B show that the decrease in luminescence induced by 5 to 30 mM CCCP is much more pronounced than that induced by near 1001MIC of PxB. Since CCCP depletes the proton gradient (13), it reduces the ATP levels available for the luminescence reaction. As expected, such a response, due to depletion of the proton gradient by CCCP, is seen with all strains in Table 1 because they are derived from the same parent and their growth characteristics are identical; they differ only in the luminescence response to a particular stress promoter. As also shown in Figure 1B, 0.5 M NaCl causes a rapid but partial decrease in the luminescence. This is attributed to lowering of ATP levels, presumably as a short term response to adapt to the osmotic stress. Specific luminescence response to PxB on carbonstarvation stress sensitive TV1048, uncoupler sensitive DPD2146, and hyperosmolarity sensitive DPD2170 were compared. As shown in Figure 2, the response with hyperosmolarity sensitive DPD2170 is biphasic, i.e. luminescence increases up to MIC, and then decreases. The other strains exhibit a monotonic response where a decrease is seen only above the MIC. This is expected because as shown in the upper panel for the OD change, PxB lowers the growth rate in a concentration dependent fashion above the MIC. The biphasic stress response is typical of that seen for lux fusions (9, 11, 15): the increase reflects the specific response that promotes expression of osmY by PxB, whereas the decrease above the MIC is a nonspecific response associated with the loss of viability due to reduced ATP levels. These results show that PxB mediates induction of osmY gene product, normally induced by hyperosmotic stress (14), and the concentration window at or below the MIC is useful for monitoring the stress-response. The response of DPD2170 to hyperosmolar sucrose (Figure 3A) and to PxB (Figure 3B) in M9 minimal medium was compared. A biphasic response is observed in both cases. The MIC for PxB in M9 medium is about
FIG. 2. (Top panel) Effect of [PxB] on the change in OD in 60 minutes in the log phase of growth of TV1048. OD inhibition at 60 minutes was calculated by 100 1 (ODc 0 OD)/ODc , where ODc is OD of control at 60 minutes. (Bottom panel) Effect on the luminescence in 60 minutes after the addition of indicated concentration of PxB during the log growth phase of TV1048 (squares), DPD2170 (open triangles) and DPD2146 (closed triangles) strains of E. coli. The luminescence increase at 60 minutes was calculated as 100 1 (L 0 Lc)/Lc , where Lc is luminescence of control at 60 minutes.
0.2 mM, modestly lower than 0.25 mM seen in the LB medium. However the absolute magnitude of the luminescence response is due to a higher level of expression in the M9 medium, which is consistent with higher levels of the promoter per cell. In the presence of 0.75 M sucrose the luminescence response to [PxB] is a monotonic decrease from the maximum response level. This is an expected result if the maximum level for the expression of the osmY promoter is saturated at 0.5 M sucrose. Results with sucrose also rule out the possibility that the effect of PxB and NaCl is on an ion-channel or some such osmo-regulatory mechanism. The possibility that PxB modulates cytoplasmic regulators of osmY (14) can also be discounted by the observation that PxB immobilized on agarose beads is antibacterial, and also induces osmotic response; agarose alone has no such effect (results not shown). Collectively, results of the transcriptional response induced by PxB below its MIC clearly show that, without entering the cytoplasm, PxB induces the osmY promoter, as is also induced, not only by other peptide antibiotics (such as cecropins and magainins), but also by hyperosmolar NaCl or sucrose. Specificity of this effect is attested by results in Table 1, which show that none of the strains selective for other stresses have the biphasic response profiles for PxB, NP, NaCl and sucrose: for example, the uncoupler sensitive DPD2146 does not respond to PxB or hyperosmolar stress, and
621
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 3. [A] Effect of [Sucrose] on the change in the luminescence in 120 minutes after the addition of sucrose to osmotic stress sensitive strain, DPD2170, in the log phase of growth on M9 minimal medium. [B] Effect of [PxB] on the change in the luminescence in 120 minutes after the addition of PxB in the absence (circles) or presence of 0.75 M sucrose (squares) to DPD2170 in the log phase of growth on M9 minimal medium. Luminescence increase at 120 minutes was calculated as 100 1 (L 0 Lc)/Lc , where Lc is luminescence of control at 120 minutes.
the hyperosmolarity sensitive DPD2170 does not respond to CCCP. These observations rule out that the hyperosmolarity response may have its origin in the loss of the energy requiring mechanisms that maintain the osmotic balance. It is remarkable that micromolar PxB induces the hyperosmotic end-effect leading to bacteriostasis, similar to that induced by two of the most commonly and traditionally used food preservative, hypertonic brine or sucrose. This similarity suggests practical applications. Possible mechanism of induction of osmY by PxB invites speculation from which we refrain. However, note that control of osmolarity in eukaryotes (19) and prokaryotes (20) is finely tuned, and it is controlled by a host of genes with roles in feed-back metabolic loops. From our perspective, the key element of this puzzle, i.e. the suggestion that the phospholipid bilayer is the primary and early locus of action, has significant implications. PxB forms vesicle-vesicle contacts through which anionic phospholipid exchange between the vesicles (16, 17). In fact, a strong correlation is seen in the efficacy of several natural antimicrobial peptides to promote phospholipid exchange versus the growth inhibition or the induction of osmY (results not shown). Most organisms appear to maintain a remarkably
specific phospholipid composition in their membranes (18), and microorganisms adapt to changing environment by changing their phospholipid composition. Thus it is likely that PxB could form contacts between the two anionic phospholipid monolayer surfaces that enclose the periplasmic space in Gram-negative organisms. Resulting loss of phospholipid compositional specificity in the membranes in contact could be deleterious for the viability of the Gram-negative organism. An important prediction of the lipid exchange as the basis for the antimicrobial action is that organisms can not become genetically resistant by mutations altering the phospholipid targets for the antimicrobial. Not only that we have not found any reported instance for such target-based resistance against PxB, but our repeated attempts to isolate resistant strains from chemically or UV mutagenized E. coli and Pseudomonas aeruginosa have failed. Also the fact that PxB does not enter the cytoplasm suggests that the drug efflux mechanisms may not overcome the antibiotic action. This assertion is consistent with the observation that we have not been able to adopt our strain of E. coli to grow at higher concentration of PxB above the MIC. To recapitulate, the very existence of natural antimicrobial peptides suggests evolutionary strategies towards target selectivity, putatively without entry into the cytoplasm. The underlying mechanism offers possible evolutionary solutions to the problem of antibiotic resistance. The osmY-lux response provides a basis for a unique assay for antimicrobials whose targets may not develop resistance. Growth inhibition and bacteriostasis induced by such agents, without bactericidal effect, also offers possibilities for managing infection through normal immune system. ACKNOWLEDGMENT This work was supported by NIH (GM29703).
REFERENCES 1. Hancock, R. E. W., Falla, T., and Brown, M. (1995) Adv. Microbial Physiology 37, 135–175. 2. Hoffman, J. A. (1995) Curr. Opinion. Immunol. 7, 4–10. 3. Storm, D. R., Rosenthal, K. S., and Swanson, P. E. (1977) Ann. Rev. Biochem. 46, 723–763. 4. Kubesch, P., Boggs, J., Luciano, L., Maass, G., and Tummler, B. (1987) Biochemistry 26, 2139–2149. 5. Welch, W. J. (1993) Scientific American May, 56–64. 6. Dukan, S., Dadon, S., Smulski, D. R., and Belkin, S. (1996) Appl. Environ. Microbiol. 62, 4003–4008. 7. Belkin, S., Smulski, D. R., Vollmer, A. C., Van Dyk, T. K., and LaRossa, R. A. (1996) Appl. Environ. Microbiol. 62, 2252–2256. 8. LaRossa, R. A., Majarian, W. R., and Van Dyk, T. K. (1997) U.S. Patent 5,683,868. 9. Van Dyk, T. K., Belkin, S., Vollmer, A. C., Smulski, D. R., Reed, T. R., and LaRossa, R. A. (1994) in Bioluminescence and Chemi-
622
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
luminescence: Fundamentals and Applied Aspects (Cambell, A. K., Kricka, L. J., and Stanley, P. E., Eds.), pp. 147–150, Willow, Chichester. 10. Belkin, S., Van Dyk, T. K., Vollmer, A. C., Smulski, D. R., and LaRossa, R. A. (1996) Environ. Toxicol. Water Qual. 11, 179– 185. 11. Van Dyk, T. K., Majarian, W. R., Konstantinov, K. B., Young, R. M., Dhurjati, P. S., and LaRossa, R. A. (1994) Appl. Environ. Microbiol. 60, 1414–1420. 12. Vollmer, A. C., Belkin, S., Smulski, D. R., Van Dyk, T. K., and LaRossa, R. A. (1997) Appl. Environ. Microbiol. 63, 2566–2571. 13. Heytler, P. G., and Prichard, W. W. (1962) Biochem. Biophys. Res. Comm. 7, 272–275.
14. Yim, H. H., Brems, R. L., and Villarego, M. (1994) J. Bacteriol. 176, 100–107. 15. Van Dyk, T. K., Ayers, B. L., Morgan, R. W., and LaRossa, R. A. (1998) J. Bacteriol. 180, 785–792. 16. Cajal, Y., Rogers, J., Berg, O. G., and Jain, M. K. (1996) Biochemistry 35, 299–308. 17. Cajal, Y., Ghanta, J., Easwaran, K., Surolia, A., and Jain, M. K. (1996) Biochemistry 35, 5684–5695. 18. Hancock, R. E. W., Karunaratne, D. N., and Bernegger-Egli, C. (1994) in Bacterial Cell Wall, (Ghuysen, J. M., and Hakenbeck, R., Eds.), pp. 263–279, Elsevier, Amsterdam. 19. Waldegger, S., and Lang, F. (1998) J. Membrane Biol. 162, 95– 100. 20. Csonka, L. N. (1989) Microbiol. Rev. 53, 121–147.
623
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC
246, 619–623 (1998)
RC988682
Osmotic Stress in Viable Escherichia coli as the Basis for the Antibiotic Response by Polymyxin B Joon-Taek Oh,* Tina K. Van Dyk,† Yolanda Cajal,* Prasad S. Dhurjati,‡ Myron Sasser,§ and Mahendra K. Jain*,1 *Department of Chemistry and Biochemistry and ‡Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716; †DuPont Company, P.O. Box 80173, Wilmington, Delaware 19889-0173; and §MIDI Inc., Newark, Delaware 19713
Received April 14, 1998
Cationic antimicrobial peptides, such as polymyxin B (PxB), below growth inhibitory concentration induce expression of osmY gene in viable E. coli without leakage of solutes and protons. osmY expression is also a locus of hyperosmotic stress response induced by common food preservatives, such as hypertonic NaCl or sucrose. High selectivity of PxB against Gram-negative organisms and the basis for the hyperosmotic stress response at sublethal PxB concentrations is attributed to PxB-induced mixing of anionic phospholipid between the outer layer of the cytoplasmic membrane with phospholipids in the inner layer of the outer membrane. This explanation is supported by PxB-mediated rapid and direct exchange of anionic phospholipid between vesicles. This mechanism is consistent with the observation that genetically stable resistance against PxB could not be induced by mutagenesis. q 1998 Academic Press
Remarkable selectivity of antimicrobial cationic peptides, produced by virtually all organisms (1, 2); remains an enigma. For example, defensins and other proteins from human plasma, magainins from frog skin, cecropins from insect larvae, and thionins from plants do not have a deleterious effect on the organisms that produce them. This fact alone rules out nonspecific mechanisms of action, such as leakage of the cytoplasmic content. Evolutionary success of such antimicrobials also suggests strategies for overcoming drug resistance. 1
Corresponding author. Fax: /302-831-6335. E-mail: mkjain@ udel.edu.
AcylrL-DabrL-ThrrL-DabrL-DabrL-DabrD-Phe L-Leu L-ThrRL-DabRL-Dab Polymyxin B or PxB (Dab Å a, g-diaminobutyric acid) PxB-nonapeptide or NP is devoid of the acyl chain and the first residue Polymyxin B, the focus of this study, is produced by Gram-positive Paenibacillus polymyxa and it is highly selective against Gram-negative organisms. Polymyxins have been in prevalent use for several decades, however there is no indication of the antibiotic resistance in clinical isolates. They act on membranes, yet the antibacterial mechanism is not established. Binding of PxB to surface lipolysaccharides, and consequent disruption of the outer membrane of Gram-negative organisms (3) to permit their entry into periplasmic space of Gram-negative organisms, is a necessary but not a sufficient condition for the antimicrobial effect; for example, PxB-nonapeptide (NP) disrupts the outer membrane, yet it is not an antibiotic (3, 4). Studies with model membranes have provided compelling evidence that these cationic peptides interact with anionic phospholipid prevalent in bacterial membranes, yet they are not ionophores. They induce leakage of the cytoplasmic contents at significantly higher concentrations than the antimicrobial concentrations. Note that other peptides that induce comparable nonspecific membrane changes are not antibiotic in vivo. In this paper we report a strategy to monitor the stress induced by sublethal PxB in viable and growing E. coli. Results show that PxB induces a highly selective expression of osmY gene, as is also induced by hyperosmolar saline and sugar solution.
619
0006-291X/98 $25.00 Copyright q 1998 by Academic Press All rights of reproduction in any form reserved.
AID
BBRC 8682
/
6953$$$221
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS TABLE 1
Response of the Stress Sensitive Strains of E. coli to PxB and Other Additives Response to Fusion gene::lux
Strain #
Stress inducer (conc.)
PxB
NaCl
Sucrose
CCCP
NP
grpE* recA* katG* inaA* lac osmY*
TV1061 DPD2794 DPD2511 DPD2146 TV1048 DPD2170
Protein/ethanol (0.5 M) DNA/mitomycin C (0.3 mM) Oxidative/H2O2 (15 mM) Proton leak/salicylate (1 mM) Limited carbon source (0) Osmotic/ sucrose (0.5 M)
0 0 0 0 0 /
0 0 0 0 0 /
0 0 0 0 0 /
0 0 0 / 0 0
0 0 0 0 0 0
MATERIALS AND METHODS Strains of E. coli. Construction, characterization, and the growth and luminescence response assay conditions are as described before (6-15). Typically, the inducer or peptide (passed through 0.2 mm-pore filter) was added to 5 ml aliquot of cells grown to OD 0.2 at 307 C in LB medium containing 50 mg/ml ampicillin or kanamycin to maintain the plasmid. Cell growth was monitored as OD change at 600 nm. Other specific conditions are described in figure legends. Assay and detailed characterization of the PxB-mediated intervesicle exchange of monoanionic phospholipid has been described before (16, 17). Purity check of commercial PxB (Sigma) and NP (Boehringer) revealed that the trace impurities (õ5% by HPLC) were inactive in the assays described in this paper.
expression of osmY. These results, in conjunction with those described below show that PxB, without entering the cytoplasm, induces transcriptional expression of osmY gene, and that this effect is not due to the disruption of the outer membrane or due to induction of proton leakage by the uncoupler through the cytoplasmic membrane. The growth curve of the lac-lux strain (Figure 1A) was monitored as an increase in the turbidity as well
RESULTS AND DISCUSSION Results in Table 1 show that osmotic stress, without leakage of the cytoplasmic content, is the primary locus for the antimicrobial effect of PxB in Escherichia coli. The physiological basis of the method is that organisms often respond and adapt to sublethal environmental adversities by increased expression of stress proteins to restore homeostasis (5). Thus by coupling transcription initiated at specific promoters to the expression of lux genes it is possible to monitor luminescence in real time as the read-out for the transcriptional response to stress in viable cells (6-9). Each E. coli strain in Table 1 contains a plasmid with a specific stress promoter coupled to the expression of luxCDABE. Thus each strain responds to a specific stress, such as oxidative damage, or proton leakage, or osmotic changes (1015). Since most stress genes are not normally highly expressed under normal growth conditions, the luminescence response to stress in real time typically yields an excellent ratio of signal to background noise. As supported by detailed evidence developed below, results in Table 1 clearly show that, below the minimum inhibitory concentrations (MIC) for growth in 60 minutes, PxB induces only the expression of osmY promoter, as also induced by hyperosmolar NaCl or sucrose. Note that both NP and CCCP (carbonylcyanide m-chlorophenylhydrazone, a proton translocator or uncoupler that depletes the proton gradient) do not induce
FIG. 1. [A] Growth profile for E. coli strain TV1048 in the absence (open symbols) or presence of 0.25 mM polymyxin B (closed symbols) monitored as a change in turbidity at 600 nm (circles) or luminescence (squares). [B] Short-time effect of CCCP, PxB, and NaCl on the luminescence of TV1048. Each aliquot was prepared by dilution of overnight cultured broth with fresh medium followed by shaking 1 hour at 30 7C. Luminescence response (in arbitrary units on Turner luminometer at standard settings) to additives was measured after adding: (inverted triangles) 5 or (diamonds) 30 mM CCCP, (circles) 0.1, (squares) 2, and (triangles) 25 mM PxB and (hexagons) 0.5 M NaCl to a fresh aliquot of the culture.
620
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
as the increase in the luminescence. In TV1048, lac promoter drives expression of the lux gene, and luminescence is observed under normal growth conditions in a moderately carbon limited medium. Note that 0.25 mM PxB shows partial growth inhibition seen not only as lower turbidity, but also as lower luminescence. Specificity of the effect of PxB can be judged from the fact that even up to 20 mM NP there was no noticeable effect on the growth curve monitored by either of the two responses (data not shown). The possibility that PxB depletes proton gradient is ruled out by results in Table 1, as also supported by the short term effect of PxB and CCCP on the intrinsic luminescence from TV1048 strain (Figure 1B). The significance of these results is best evaluated in terms of their MIC: 0.25 mM for PxB and 10 mM for CCCP. Results in Figure 1B show that the decrease in luminescence induced by 5 to 30 mM CCCP is much more pronounced than that induced by near 1001MIC of PxB. Since CCCP depletes the proton gradient (13), it reduces the ATP levels available for the luminescence reaction. As expected, such a response, due to depletion of the proton gradient by CCCP, is seen with all strains in Table 1 because they are derived from the same parent and their growth characteristics are identical; they differ only in the luminescence response to a particular stress promoter. As also shown in Figure 1B, 0.5 M NaCl causes a rapid but partial decrease in the luminescence. This is attributed to lowering of ATP levels, presumably as a short term response to adapt to the osmotic stress. Specific luminescence response to PxB on carbonstarvation stress sensitive TV1048, uncoupler sensitive DPD2146, and hyperosmolarity sensitive DPD2170 were compared. As shown in Figure 2, the response with hyperosmolarity sensitive DPD2170 is biphasic, i.e. luminescence increases up to MIC, and then decreases. The other strains exhibit a monotonic response where a decrease is seen only above the MIC. This is expected because as shown in the upper panel for the OD change, PxB lowers the growth rate in a concentration dependent fashion above the MIC. The biphasic stress response is typical of that seen for lux fusions (9, 11, 15): the increase reflects the specific response that promotes expression of osmY by PxB, whereas the decrease above the MIC is a nonspecific response associated with the loss of viability due to reduced ATP levels. These results show that PxB mediates induction of osmY gene product, normally induced by hyperosmotic stress (14), and the concentration window at or below the MIC is useful for monitoring the stress-response. The response of DPD2170 to hyperosmolar sucrose (Figure 3A) and to PxB (Figure 3B) in M9 minimal medium was compared. A biphasic response is observed in both cases. The MIC for PxB in M9 medium is about
FIG. 2. (Top panel) Effect of [PxB] on the change in OD in 60 minutes in the log phase of growth of TV1048. OD inhibition at 60 minutes was calculated by 100 1 (ODc 0 OD)/ODc , where ODc is OD of control at 60 minutes. (Bottom panel) Effect on the luminescence in 60 minutes after the addition of indicated concentration of PxB during the log growth phase of TV1048 (squares), DPD2170 (open triangles) and DPD2146 (closed triangles) strains of E. coli. The luminescence increase at 60 minutes was calculated as 100 1 (L 0 Lc)/Lc , where Lc is luminescence of control at 60 minutes.
0.2 mM, modestly lower than 0.25 mM seen in the LB medium. However the absolute magnitude of the luminescence response is due to a higher level of expression in the M9 medium, which is consistent with higher levels of the promoter per cell. In the presence of 0.75 M sucrose the luminescence response to [PxB] is a monotonic decrease from the maximum response level. This is an expected result if the maximum level for the expression of the osmY promoter is saturated at 0.5 M sucrose. Results with sucrose also rule out the possibility that the effect of PxB and NaCl is on an ion-channel or some such osmo-regulatory mechanism. The possibility that PxB modulates cytoplasmic regulators of osmY (14) can also be discounted by the observation that PxB immobilized on agarose beads is antibacterial, and also induces osmotic response; agarose alone has no such effect (results not shown). Collectively, results of the transcriptional response induced by PxB below its MIC clearly show that, without entering the cytoplasm, PxB induces the osmY promoter, as is also induced, not only by other peptide antibiotics (such as cecropins and magainins), but also by hyperosmolar NaCl or sucrose. Specificity of this effect is attested by results in Table 1, which show that none of the strains selective for other stresses have the biphasic response profiles for PxB, NP, NaCl and sucrose: for example, the uncoupler sensitive DPD2146 does not respond to PxB or hyperosmolar stress, and
621
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 3. [A] Effect of [Sucrose] on the change in the luminescence in 120 minutes after the addition of sucrose to osmotic stress sensitive strain, DPD2170, in the log phase of growth on M9 minimal medium. [B] Effect of [PxB] on the change in the luminescence in 120 minutes after the addition of PxB in the absence (circles) or presence of 0.75 M sucrose (squares) to DPD2170 in the log phase of growth on M9 minimal medium. Luminescence increase at 120 minutes was calculated as 100 1 (L 0 Lc)/Lc , where Lc is luminescence of control at 120 minutes.
the hyperosmolarity sensitive DPD2170 does not respond to CCCP. These observations rule out that the hyperosmolarity response may have its origin in the loss of the energy requiring mechanisms that maintain the osmotic balance. It is remarkable that micromolar PxB induces the hyperosmotic end-effect leading to bacteriostasis, similar to that induced by two of the most commonly and traditionally used food preservative, hypertonic brine or sucrose. This similarity suggests practical applications. Possible mechanism of induction of osmY by PxB invites speculation from which we refrain. However, note that control of osmolarity in eukaryotes (19) and prokaryotes (20) is finely tuned, and it is controlled by a host of genes with roles in feed-back metabolic loops. From our perspective, the key element of this puzzle, i.e. the suggestion that the phospholipid bilayer is the primary and early locus of action, has significant implications. PxB forms vesicle-vesicle contacts through which anionic phospholipid exchange between the vesicles (16, 17). In fact, a strong correlation is seen in the efficacy of several natural antimicrobial peptides to promote phospholipid exchange versus the growth inhibition or the induction of osmY (results not shown). Most organisms appear to maintain a remarkably
specific phospholipid composition in their membranes (18), and microorganisms adapt to changing environment by changing their phospholipid composition. Thus it is likely that PxB could form contacts between the two anionic phospholipid monolayer surfaces that enclose the periplasmic space in Gram-negative organisms. Resulting loss of phospholipid compositional specificity in the membranes in contact could be deleterious for the viability of the Gram-negative organism. An important prediction of the lipid exchange as the basis for the antimicrobial action is that organisms can not become genetically resistant by mutations altering the phospholipid targets for the antimicrobial. Not only that we have not found any reported instance for such target-based resistance against PxB, but our repeated attempts to isolate resistant strains from chemically or UV mutagenized E. coli and Pseudomonas aeruginosa have failed. Also the fact that PxB does not enter the cytoplasm suggests that the drug efflux mechanisms may not overcome the antibiotic action. This assertion is consistent with the observation that we have not been able to adopt our strain of E. coli to grow at higher concentration of PxB above the MIC. To recapitulate, the very existence of natural antimicrobial peptides suggests evolutionary strategies towards target selectivity, putatively without entry into the cytoplasm. The underlying mechanism offers possible evolutionary solutions to the problem of antibiotic resistance. The osmY-lux response provides a basis for a unique assay for antimicrobials whose targets may not develop resistance. Growth inhibition and bacteriostasis induced by such agents, without bactericidal effect, also offers possibilities for managing infection through normal immune system. ACKNOWLEDGMENT This work was supported by NIH (GM29703).
REFERENCES 1. Hancock, R. E. W., Falla, T., and Brown, M. (1995) Adv. Microbial Physiology 37, 135–175. 2. Hoffman, J. A. (1995) Curr. Opinion. Immunol. 7, 4–10. 3. Storm, D. R., Rosenthal, K. S., and Swanson, P. E. (1977) Ann. Rev. Biochem. 46, 723–763. 4. Kubesch, P., Boggs, J., Luciano, L., Maass, G., and Tummler, B. (1987) Biochemistry 26, 2139–2149. 5. Welch, W. J. (1993) Scientific American May, 56–64. 6. Dukan, S., Dadon, S., Smulski, D. R., and Belkin, S. (1996) Appl. Environ. Microbiol. 62, 4003–4008. 7. Belkin, S., Smulski, D. R., Vollmer, A. C., Van Dyk, T. K., and LaRossa, R. A. (1996) Appl. Environ. Microbiol. 62, 2252–2256. 8. LaRossa, R. A., Majarian, W. R., and Van Dyk, T. K. (1997) U.S. Patent 5,683,868. 9. Van Dyk, T. K., Belkin, S., Vollmer, A. C., Smulski, D. R., Reed, T. R., and LaRossa, R. A. (1994) in Bioluminescence and Chemi-
622
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC
Vol. 246, No. 3, 1998
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
luminescence: Fundamentals and Applied Aspects (Cambell, A. K., Kricka, L. J., and Stanley, P. E., Eds.), pp. 147–150, Willow, Chichester. 10. Belkin, S., Van Dyk, T. K., Vollmer, A. C., Smulski, D. R., and LaRossa, R. A. (1996) Environ. Toxicol. Water Qual. 11, 179– 185. 11. Van Dyk, T. K., Majarian, W. R., Konstantinov, K. B., Young, R. M., Dhurjati, P. S., and LaRossa, R. A. (1994) Appl. Environ. Microbiol. 60, 1414–1420. 12. Vollmer, A. C., Belkin, S., Smulski, D. R., Van Dyk, T. K., and LaRossa, R. A. (1997) Appl. Environ. Microbiol. 63, 2566–2571. 13. Heytler, P. G., and Prichard, W. W. (1962) Biochem. Biophys. Res. Comm. 7, 272–275.
14. Yim, H. H., Brems, R. L., and Villarego, M. (1994) J. Bacteriol. 176, 100–107. 15. Van Dyk, T. K., Ayers, B. L., Morgan, R. W., and LaRossa, R. A. (1998) J. Bacteriol. 180, 785–792. 16. Cajal, Y., Rogers, J., Berg, O. G., and Jain, M. K. (1996) Biochemistry 35, 299–308. 17. Cajal, Y., Ghanta, J., Easwaran, K., Surolia, A., and Jain, M. K. (1996) Biochemistry 35, 5684–5695. 18. Hancock, R. E. W., Karunaratne, D. N., and Bernegger-Egli, C. (1994) in Bacterial Cell Wall, (Ghuysen, J. M., and Hakenbeck, R., Eds.), pp. 263–279, Elsevier, Amsterdam. 19. Waldegger, S., and Lang, F. (1998) J. Membrane Biol. 162, 95– 100. 20. Csonka, L. N. (1989) Microbiol. Rev. 53, 121–147.
623
AID
BBRC 8682
/
6953$$$222
05-12-98 20:27:18
bbrcgs
AP: BBRC