Molecular analysis of antibiotic tolerance in pneumococci

Int. J. Med. Microbiol. 292, 75 ± 79 (2002) ¹ Urban & Fischer Verlag http: // www.urbanfischer.de/journals/ijmm

Molecular analysis of antibiotic tolerance in pneumococci Lauren S. Mitchell, Elaine I. Tuomanen Department of Infectious Diseases, St. Jude Children's Research Hospital, Memphis, TN, USA

Abstract Widespread pneumococcal resistance and the emergence of tolerance underscores the need to develop new antimicrobials. Uncovering the mechanisms of autolysin activation could yield not only new antibacterial targets but also ways to eradicate a pool of bacteria facilitating the spread of resistance. Although several genes contributing to antibiotic tolerance among pneumococci have been identified, those important in the clinical arena thus far are in a single gene cluster, vex/pep27/vncS/vncR. Mutations within this signal transduction system represent at least one mechanism, which explains tolerance to both penicillin and vancomycin. Since mutations in this locus do not result in tolerance to penicillin alone, there must be other, yet unknown, mutations which account for tolerance to a single antibiotic. In the case of pneumococci, there exist two more autolysins other than LytA suggesting our understanding of how bacteria die is currently only at the beginning. Key words: pneumococci ± antibiotic tolerance ± antibiotic resistance ± two-component systems

Tolerance and resistance: a synergistic relationship One in two persons carry Streptococcus pneumoniae in the nasopharynx at any one time. Thus, encounters with this pathogen are common and any changes in its behavior are likely to have important health implications. Pneumococcus is a major cause of serious bacterial infection among children and older adults. It remains a leading cause of morbidity and mortality causing 1.2 million child deaths each year worldwide. In the United States the annual incidence of invasive disease is 26 cases/100,000, half of which die or suffer permanent morbid conditions; pneumococcal meningitis has the highest case fatality rate at 21% (Schuchat et al., 1997). An example of the rapid spread of a new trait in the circulating population of pneumococci is antibiotic resistance.

The first recognition of resistance was made 30 years ago. Over the past decade, there has been a dramatic increase in the incidence of resistance from < 10% to > 40% of isolates in many localities (Kaplan et al., 1998) (Figure 1). This has forced the use of vancomycin as the agent of last resort for the treatment of invasive disease (AAP, 1997). Is this epidemic spread of resistance accelerated by an underlying trait that has emerged to facilitate the development of resistance? Beta-lactam resistance arises by importation of DNA encoding changes in the active site of cell wall synthetic enzymes (PBP's). These changes exact a price in bacterial metabolism since exclusion of the antibiotic from the active site (i.e., resistance) forces the bacterium to have a mechanism to supply altered cell wall building blocks that still fit the altered active site (GarciaBustos and Tomasz, 1990). The ability to make

Corresponding author: Elaine I. Tuomanen, Department of Infectious Diseases, St. Jude Children's Research Hospital, 332 North Lauderdale Street, Memphis, TN 38105, USA. Phone: ‡ ‡ 901 495 3486, Fax: ‡ ‡ 901 495 3099, E-mail: [email protected]

1438-4221/02/292/2-75 $ 15.00/0

76

L. S. Mitchell, E. I. Tuomanen

Fig. 1.

Penicillin resistance in pneumococci.

alternative building blocks must be in place before bacteria can survive under a new PBP ™regime∫. This is clearly demonstrated when unprepared sensitive pneumococci are transformed with resistance cassettes and become quite sick if they survive (Rieux et al., 2001). A rapid rise in the prevalence of resistance suggests that a genetic background suitable to become resistant has spread through the community of pneumococci. Recently, part of this genetic background has been suggested to be ™tolerance∫. By definition, tolerant pneumococci survive but do not multiply during antibiotic therapy, and will renew replication upon removal of antibiotic (Tuomanen et al., 1988; Liu and Tomasz, 1985) (Figure 2). Thus, tolerant bacteria survive antibiotics, a trait intermediate to susceptibility and resistance (ability to grow in the presence of antibiotics). It stands to reason that tolerance can precede and facilitate resistance by creating survivors. Experimental evidence confirms that tolerant pneumococci more easily acquire antibiotic resistance cassettes. Novak et al. (1999) showed that a pneumococcal mutant tolerant to penicillin was able to acquire a large step up in resistance (from 6 to 50 ng/ml) in a single round of transformation while the parent strain was unable to survive such a challenge. Thus, it is not surprising that nearly all resistant bacteria are also tolerant (Liu and Tomasz, 1985).

Frequency of tolerance Like resistance, antibiotic tolerance can develop to different classes of antibiotics. Clinical isolates of

Fig. 2. Resistance vs. tolerance. Schematic diagram of the reaction of tolerant and resistant bacteria to antibiotic exposure. MIC: Minimum inhibitory concentration.

pneumococci that are tolerant to penicillin have been recognized since 1985 (Liu and Tomasz, 1985; Tuomanen et al., 1988). Recent estimates suggest that as many as 20% of pneumococci are tolerant to penicillin (Henriques Normark et al., 2001). It was not until 1999 that vancomycin-tolerant pneumococci were recognized in the clinical setting (Novak et al., 1999). Four vancomycin-tolerant (and penicillin-resistant) clinical strains of pneumococcus have been reported: two strains were isolated from the nasopharynx, one from the bloodstream, and one from the cerebrospinal fluid (Novak et al., 1999; McCullers et al., 2000). Two recent abstracts indicate a frequency of vancomycin tolerance of 3 ± 8% depending on the collection studied (Atkinson et al., 2000; Rodriguez et al., 2000). Acknowledging that tolerance may facilitate the development of resistance, emergence of vancomycin-tolerant pneumococci may be a warning. To date, there is only one report of vancomycin-resistant pneumococci (Hoban et al., 2001).

Clinical significance of tolerance Tolerant bacteria are clinically significant because the inability to sterilize the infection may lead to prolonged therapy and, possibly, failure of antibiotic treatment. In the rabbit model of pneumococcal meningitis, a tolerant strain grew as well as its wild-

Antibiotic-tolerant pneumococci

type parent, indicating there was no reduction in virulence. However, infection with the tolerant strain resulted in a failure to cure the infection by intravenous vancomycin (Novak et al., 1999). This may translate to treatment of patients, since penicillin-tolerant clinical strains have caused persistent meningitis and complicated bacteremia among HIVinfected patients (Tuomanen et al., 1988). McCullers et al. (2000) reported a case of a 10-month-old healthy female who developed recrudescent meningitis due to a vancomycin-tolerant pneumococcus, despite 8 days of vancomycin therapy and 10 days of cephalosporin therapy. One preliminary study suggests that mortality from meningitis caused by tolerant pneumococci is 30% compared to < 10% for non-tolerant strains (Rodriguez et al., 2000).

Detection of tolerance Tolerant bacteria have the same MIC as sensitive strains. Thus, while resistance is easy to detect by the increased MIC, tolerance is not. Currently the only method of detecting a tolerant strain is a lysis-kill curve. Treatment of a growing culture of wild-type pneumococci with 10 the MIC of an antibiotic results in at least a 3-log decrease in viability within 3 ± 4 hours. In contrast, tolerant strains show little to no decline in viability after the same time period (Handwerger and Tomasz, 1985). Currently, clinical labs are not equipped to carry out such a laborintensive test and thus are unable to detect antibiotic-tolerant clinical strains. Hopefully, tracking the tolerance trait in the future can be done by following underlying genetic changes.

Mechanism of tolerance Antibiotic tolerance was first described in 1970 in a laboratory mutant of pneumococcus that lacked an active cell wall hydrolase/autolysin (Tomasz et al., 1970). This original description identified the essential role of cell wall hydrolases, like the major pneumococcal autolysin LytA, in the bactericidal activity of penicillin. Autolysins are constitutively expressed extracellular enzymes that reside on the cell wall but do not hydrolyze it until a signal triggers them to become active. Under normal growth conditions, the only known physiological trigger is stationary phase lysis. However, unnatural triggers, such as a cell wall synthesis-inhibiting antibiotic (like penicillin or vancomycin), are able to dereg-

77

ulate the mechanisms controlling LytA activation. Although the binding of such an antibiotic to its cell wall target has been well characterized, the genetic elements responsible for subsequent deregulation of the major autolysin leading to lysis of the cell wall remain largely unknown. This ™black box∫ between drug binding to PBP and triggering autolysin is the heart of the mechanism of tolerance. Absence of autolysin activity can be due to either loss of autolysin expression or loss of autolysin activation. The simplest explanation of tolerance is the failure to express the major autolysin, as exemplified by the original lytA deletion mutant. The recently identified laboratory mutants, psaA/B/C/D (Novak et al., 1998), zmpB (Novak et al., 2000b), and clpC (Charpentier et al., 2000), like the lytA mutant, also fail to produce LytA and are tolerant. These bacteria stop growing in the presence of normal amounts of penicillin. They just do not lyse and die since there is no autolysin to trigger. The psa locus encodes an ABC type mangenese (Mn) permease complex required for Mn uptake by pneumococci. ClpC is a chaperone and might interact with LytA by promoting folding, assembly and translocation to the cell membrane. ZmpB is a metalloprotease believed to be involved in targeting proteins for export. Mutagenesis of zmpB reveals a phenotype including adhesion deficiency, transformation deficiency, and tolerance to antibiotics because the choline-binding family, which includes LytA, is absent from the bacterial surface. While these tolerant mutants are informative as to bacterial physiology, clinical isolates of tolerant pneumococci have normal amounts of autolysin that show normal activity when extracted from the tolerant strain and tested for the ability to hydrolyze normal cell wall. This implies that, in the clinical setting, tolerance arises not from absence of autolysin, such as in these mutants, but rather from an alteration in the ability to trigger autolysin activity. Thus, the mechanisms involved in regulating and activating autolysin are the key to understanding clinical tolerance. The first gene cluster associated with triggering autolysin and therefore relevant to tolerance in a clinical setting is the vex/pep27/vncS/R locus (Novak et al., 1999; Novak et al., 2000a) (Figures 3 and 4). Two-component regulatory systems are one way in which bacteria sense and respond to their environment. Typically, two-component transduction systems are composed of two signaling proteins, a transmembrane sensor histidine kinase and a cytoplasmic response regulator. Basically, an external signal binds the kinase resulting in a conformational change. This conformational change is associated

78

L. S. Mitchell, E. I. Tuomanen

Fig. 3. Genes involved in the expression of LytA. A. Mutation of psa/A/B/C/D blocks transcription/translation of LytA. B. Mutation of clpc blocks folding/transport of LytA. C. Mutation of zmpB blocks export of LytA. D. Mutation of vex/pep27/vncS/R blocks activation of LytA.

with either dephosphorylation or phosphorylation of the regulator, which then changes its binding to DNA leading to a change in gene expression. For this locus, VncS is the histidine kinase and VncR is the response regulator. After sensing a specific signal, VncS dephosphorylates VncR leading to changes in gene expression. Mutagenesis of vncS results in tolerance to multiple antibiotics, including penicillin and vancomycin. Conversely, the vncR mutant is not tolerant. It has been proposed that when VncR is present in a phosphorylated state (e.g. during log growth or in a tolerant strain), autolysin activity is

repressed but when either present in a dephosphorylated state (such as when VncS dephosphorylates VncR) or absent (vncR mutant) normal activation of autolysin ensues. Supporting this hypothesis, overexpression of vncS resulted in tolerance to penicillin and vancomycin suggesting that VncS negatively regulates LytA (Novak et al., 1999). Discovery of this locus indicates that signal transduction is essential to the activation of the major autolysin LytA and thus necessary for the bactericidal activity of penicillin and vancomycin. Upstream of vncS/R are genes encoding a putative ABC transporter, Vex, and a 27-amino-acid peptide, Pep27. The Vex transporter consists of three components, Vex 1 and 3 form a transmembrane protein and Vex2 is an ATP-binding cassette protein. It has been hypothesized that Pep27 is exported into the supernatant via the dedicated transporter Vex. Pep27, acting as a quorum-sensing molecule, accumulates to a critical density at which point it is able to signal VncS, eventually leading to activation of LytA. Pep27 (0.1 mM) added to a growing culture of wild-type pneumococci induces > 2-log loss of viability, an effect comparable to the decline in viability observed in response to penicillin and vancomycin. When Pep27 is added to a growing culture of a mutant lacking LytA, viability still decreases although not to the extent seen with wildtype pneumococci. This same mutant shows minimal loss of viability after the addition of antibiotics. These results suggest that LytA participates in

Fig. 4. Proposed mechanism of vancomycin tolerance among clinical isolates of pneumococci. A. Vancomycin binds the D-alanine-Dalanine residue on the cell wall leading to expression of the vex/pep27/vncS/R locus. B. Pep27 is transported to the extracellular space via Vex where it activates VncS leading to dephosphorylation of VncR. C. VncR mediates changes in gene expression resulting in the activation of the major autolysin, LytA.

Antibiotic-tolerant pneumococci

peptide-induced killing but that alternative autolysin-independent pathways exist which can be activated by Pep27.

Mutations underlying clinical tolerance Evaluation of sequences of genes known to be involved in tolerance among clinical isolates has revealed genetic defects believed to be responsible for the development of tolerance. Three vancomycin tolerant strains demonstrated the same amino acid substitution in VncS, a valine for an alanine. Correction of the defect by expression of wild-type VncS reversed the tolerant phenotype (Novak et al., 1999). Thus, at least some tolerant clinical isolates have defects in the two-component system associated with autolysin triggering. In strains with a normal vncS sequence, other mutations must underlie the phenotype. These possibilities are currently not well established but some prospects have been suggested. There are two alleles each for Vex2 (the transporter) and Pep27 (the peptide being transported). If Pep27 must bind Vex2 in order to be exported, correct matching of alleles between Vex2 and Pep27 may be required. If a mismatch of alleles occurs, Pep27 may not be exported and might not be available for signaling via VncS. Therefore, a pneumococcal isolate with a mismatch could be tolerant. Such mismatching of alleles has been identified in 7 clinical isolates tolerant to penicillin and vancomycin (Atkinson et al., 2000).

References American Academy of Pediatrics. Committee on Infectious Diseases: Therapy for children with invasive pneumococcal infections. Pediatrics 99, 289 ± 299 (1997). Atkinson, R. et al.: ICAAC Sept 2000. Abstract no. 1776. Charpentier, E., Novak, R., Tuomanen, E.: Regulation of growth inhibition at high temperature, autolysis, transformation and adherence in Streptococcus pneumoniae by ClpC. Mol. Microbiol. 37, 717 ± 726 (2000). Garcia-Bustos, J., Tomasz, A.: A biological role of antibiotic resistance: Major changes in the peptidoglycan structure of penicillin-resistant pneumococci. Proc. Natl. Acad. Sci. USA 87, 5415 ± 5419 (1990). Handwerger, S., Tomasz, A.: Antibiotic tolerance among clinical isolates of bacteria. Rev. Infect. Dis. 7, 368 ± 386 (1985). Henriques Normark, B., Novak, R., Ortquist, A., Kallenius, G., Tuomanen, E., Normark, S.: Clinical

79

isolates of Streptococcus pneumoniae that exhibit tolerance to vancomycin. Clin. Infect. Dis. 32, 552 ± 558 (2001). Hoban, D. J., Doern, G. V., Fluit, A. C., Roussel-Delvallez, M., Jones, R. N.: Worldwide prevalence of antimicrobial resistance in Streptococcus pneumoniae, Haemophilus influenza, and Moraxella catarrhalis in the SENTRY antimicrobial Surveillance Program, 1997 ± 1999. Clin. Infect. Dis. 32, S81 ± S93 (2001). Kaplan, S. L., Mason, Jr., E. O., Barson, W. J., Wald, E. R. Arditi, M., Tan, T. Q., Schutze, G. E., Bradley, J. S., Givner, L. B., Kim, K. S., Yogev, R.: Three-year multicenter surveillance of systemic pneumococcal infections in children. Pediatrics 102, 538 ± 545 (1998). Liu, H., Tomasz, A.: Penicillin tolerance in multiply drug-resistant natural isolates of Streptococcus pneumoniae. J. Infect. Dis. 152, 365 ± 372 (1985). McCullers, J., English, B. K., Novak, R.: Isolation and characterization of vancomycin tolerant Streptococcus pneumoniae from the cerebrospinal fluid of a patient who developed recrudescent meningitis. J. Infect. Dis. 181, 369 ± 373 (2000). Novak, R., Braun, J., Charpentier, E., Tuomanen, E.: Penicillin tolerance genes of Streptococcus pneumoniae: the ABC-type manganese permease complex PsaA. Mol. Microbiol. 29, 1285 ± 1296 (1998). Novak, R., Henriques, B., Charpentier, E., Normark, S., Tuomanen, E.: Emergence of vancomycin tolerance in Streptococcus pneumoniae. Nature 399, 590 ± 593 (1999). Novak, R., Charpentier, E., Braun, J. S., Tuomanen, E.: Signal transduction by a death signal peptide: Uncovering the mechanism of bacterial killing by penicillin. Mol. Cell 5, 49 ± 57 (2000a). Novak, R., Charpentier, E., Braun, J. S., Park, E., Murti, S. Tuomanen, E., Masure, R.: Extracellular targeting of choline-binding proteins in Streptococcus pneumoniae by a zinc metalloprotease. Mol. Microbiol. 36, 366 ± 376 (2000b). Rieux, V., Carbon, C., Azoulay-Dupuis, E.: Complex relationship between acquisition of beta-lactam resistance and loss of virulence in Streptococcus pneumoniae. J. Infect. Dis. 184, 66 ± 72 (2001). Rodriguez, C. et al.: ICAAC Sept 2000. Abstract no. 1777. Schuchat, A., Robinson, K., Wenger, J. D., Harrison, L. H., Farley, M., Reingold, A. L., Lefkowitz, L., Perkins, B. A.: Bacterial meningitis in the United States in 1995. Active Surveillance Team. N. Engl. J. Med. 337, 970 ± 976 (1997). Tomasz, A., Albino, A., Zanati, E.: Multiple antibiotic resistance in a bacterium with suppressed autolytic system. Nature 227, 138 ± 140 (1970). Tuomanen, E., Pollack, H., Parkinson, A., Davidons, M., Facklam, R., Rich, R., Zak, O.: Microbiological and clinical significance of a new property of defective lysis in clinical strains of pneumococci. J. Infect. Dis. 158, 36 ± 43 (1988).