Rapid Killing of Streptococcus pneumoniae with a Bacteriophage Cell Wall Hydrolase

See allHide authors and affiliations

Science  07 Dec 2001:
Vol. 294, Issue 5549, pp. 2170-2172
DOI: 10.1126/science.1066869


Nasopharyngeal carriage is the major reservoir forStreptococcus pneumoniae in the community. Although eliminating this reservoir would greatly reduce disease occurrence, no suitable intervention has been available for this purpose. We show here that seconds after contact, a purified pneumococcal bacteriophage lytic enzyme (Pal) is able to kill 15 common serotypes of pneumococci, including highly penicillin-resistant strains. In vivo, previously colonized mice revealed undetectable pneumococcal titers 5 hours after a single enzyme treatment. Pal enzyme had little or no effect on microorganisms normally found in the human oropharynx, and Pal-resistant pneumococci could not be detected after extensive exposure to the enzyme.

Streptococcus pneumoniae remains one of the most challenging human pathogens because of the morbidity and mortality it causes in young children, the elderly, and in immunocompromised patients. The asymptomatic carrier state, particularly in children, is thought to be the major reservoir of the pathogen. Pneumococci account for several million cases of acute otitis media and an estimated 60,000 cases of invasive disease in the United States each year, with a mortality of 10% (1). Because of the worldwide increase of resistance to multiple antibiotics in pneumococci, treatment has become more difficult than in the past.

Prevention of pneumococcal disease relies today on vaccination of the susceptible population. However, vaccination encounters a number of problems such as the limited quantity of serotypes in the pediatric formulation, incomplete protection against colonization, and selection of nonvaccine serotypes (2, 3). There is need for an alternative preventive strategy for situations where vaccination is insufficient, impossible, or inefficient. Eradication or even reduction of nasopharyngeal carriage is likely to have a major impact on the transmission of S. pneumoniae and the incidence of infection. Antibiotic prophylaxis in controlled surroundings has shown limited success but carries the risk of selective pressure resulting in an increase of resistant strains (4). Until now, there has been no substance that can specifically reduce the number of pneumococci carried on human mucous membranes without affecting the normal indigenous mucosal flora.

Since its discovery, the destructive effect of bacteriophage on their host organisms has been exploited to develop strategies to kill infecting bacteria (5, 6). Recently, Nelson et al. developed a method in which the purified lytic enzyme from streptococcal phage C1 was able to kill group A streptococci in vitro and to prevent and eliminate pharyngeal colonization in mice (7). We continued this line of investigation and explored the lytic action of a pneumococcal phage enzyme and its potential to kill S. pneumoniae. All known pneumococcal phages contain a cell wall lytic system consisting of a holin that permeabilizes the cell membrane, and either anN-acetylmuramoyl-L-alanine amidase (amidase) or a lysozyme, capable of digesting the pneumococcal cell wall (8). Both types of enzymes contain a COOH-terminal choline-binding domain and an NH2-terminal catalytic domain. The lytic system allows the virus to escape the host cell after successful replication. We obtained Escherichia coli DH5α (pMSP11) expressing the amidase Pal of phage Dp-1 (9) from R. Lopez. We produced the enzyme in E. coli and purified it by affinity chromatography in a single step as described, with some modifications (10, 11). We then defined a unit for the enzyme, using lysis of exponentially growingS. pneumoniae serogroup 14 with serial dilutions of purified Pal (12).

In the first series of experiments, we measured the killing ability of Pal in vitro by exposing 15 clinical strains ofS. pneumoniae, 2 pneumococcal mutants (R36A, Lyt 4-4), and 8 species of oral commensal streptococci (S.gordonii, S. mitis, S. mutans,S. oralis, S. salivarius, S. intermedius, S. crista,and S. parasanguis) to purified enzyme at a final concentration of 100 U/ml, and in the case of the oral streptococci, to 1000 and 10,000 U/ml (13). The pneumococcal strains, obtained from various sources [Web table 1 (11)], included nine serogroups that most frequently cause invasive disease in North America, Europe, Africa, and Oceania (14). Furthermore, three highly penicillin-resistant strains were included, which represent the internationally spread clones Sp9-3, Sp14-3, and Sp23-1, that account for a majority of penicillin-resistant pneumococci in daycare centers and hospitals (15, 16). Within 30 s, 100 U of Pal decreased the viable titer of the 15 strains of exponentially growing S. pneumoniae by log10 4.0 colony-forming units (cfu)/ml (median, range 3.3 to 4.7) as compared to controls incubated with the enzyme buffer alone (Fig. 1). Pneumococci with intermediate (n = 1) and high penicillin resistance (n = 3) were killed at the same rate as penicillin-sensitive strains [median (range) log10 4.0 (3.7 to 4.7) versus log10 4.1 (3.3 to 4.7) cfu/ml,P = not significant]. Moreover, the capsule-deficient laboratory strain R36A and the mutant Lyt 4-4, deficient in capsule and lacking the major pneumococcal autolysin LytA, showed identical susceptibility to Pal as the clinical pneumococcal isolates (decrease of log10 4.2 and 3.9 cfu/ml, respectively,P = not significant). The latter results suggest that the pneumococcal capsule does not interfere with the enzyme's access to the cell wall and that autolysin does not contribute significantly to cell lysis caused by Pal. One-hundred units of Pal also killed exponentially growing S. oralis and S. mitis, but at a significantly lower rate (log10 0.8 and log10 0.23 cfu/ml, respectively, P < 0.05). Both strains are known to incorporate choline in their cell walls and therefore provide a binding site for the enzyme (17). The remaining oral streptococcal strains were unaffected at enzyme concentrations as high as 10,000 U/ml and up to 10 min of exposure.

Figure 1

In vitro killing of 15 clinical S. pneumoniae strains, 2 pneumococcal mutants, and 8 oral streptococcal species in log-phase with 100 U/ml Pal during 30 s, expressed as the decrease of bacterial titers in powers of 10. Numbers above “S. pneumoniae” indicate serogroups/serotypes; bold print designates the nine most frequently isolated serogroups. Error bars show standard deviation of triplicates. I, intermediate susceptibility to penicillin [minimum inhibitory concentration (MIC) = 0.1 to 1.0]; R, highly penicillin-resistant (MIC ≥ 2.0).

In vitro, S. pneumoniae, including the R36A and Lyt 4-4 mutants, in stationary phase were more resistant to the lethal action of Pal. Nevertheless, exposure to 10,000 U/ml resulted in killing of log10 3.0 cfu/ml (median, range 3.0 to 4.0) in 30 s. The mechanism responsible for the decrease in susceptibility to hydrolysis by Pal in nongrowing pneumococci is likely to be a change in the cell wall structure (18), such as an increase in peptidoglycan cross-linking.

Electron microscopy of S. pneumoniaeserogroup 14 exposed to only 50 U/ml of Pal for 1 min (19) revealed protrusions of the cell membrane and the cytoplasm through single breaks in the cell wall, which appeared predominantly near the septum of the dividing diplococci (Fig. 2B). After 5 min, empty cell walls remained, retaining their original shape, indicating that digestion of amide bonds in a restricted location within the cell wall is sufficient for cell death (Fig. 2D).

Figure 2

Electron micrographs of S. pneumoniaeexposed to 50 U of Pal. (A) shows unexposed control cells. Higher magnification of enzyme-treated cells after 1 min exposure shows cell membrane protrusion (B) or cytoplasmic leaks from membrane rupture (C) through isolated breaks in the cell wall. After 5 min (D), killing is virtually complete and only empty cell walls are left [magnification as in (A)]. Bars, 0.5 μm.

The ability of Pal to eradicate S. pneumoniaefrom a mucosal surface was then tested in vivo in a mouse model of nasopharyngeal colonization (20) with minor modifications (21). In the first experiment, 18 mice were colonized intranasally with 108 cfu of S. pneumoniaeserogroup 14. Animals were randomized 42 hours later for nasal and pharyngeal treatment with a total of 1400 U of Pal enzyme or buffer. Five hours after the treatment, all animals were killed, and pneumococcal titers were analyzed in the nasal wash. Treatment with Pal reduced S. pneumoniae to undetectable levels (log10 0 cfu/10 μl nasal wash) as opposed to treatment with buffer only [median (range) log10 3.0 (2.0 to 3.0) cfu/10 μl, P < 0.001] (Fig. 3A). We repeated the experiment with a lower dose of enzyme, randomizing the animals (n = 16) for treatment with 700 U of Pal or buffer. Enzyme treatment here completely eliminated pneumococci from five of eight animals and significantly decreased titers in the remaining three (P < 0.001) (Fig. 3B). Each experiment included three uncolonized control animals that revealed no S. pneumoniae. These results indicate that pneumococci on mucosal surfaces are highly susceptible to the action of the lytic enzyme.

Figure 3

Elimination of S. pneumoniae serotype 14 in the mouse model of nasopharyngeal carriage. (A) After nasal and pharyngeal treatment with a total of 1400 U of Pal, no pneumococci were retrieved in the nasal wash, compared to buffer-treated colonized mice (P < 0.001). No pneumococci were isolated from noncolonized control mice. (B) After treatment with a total of 700 U of Pal, pneumococci were completely eliminated in five of eight colonized mice (P < 0.001) and overall were significantly reduced. Bars show median.

We further addressed the question whether surviving pneumococci are able to recolonize the nasopharynx of mice after a single dose of enzyme. Thirty mice were colonized as before (day –2) and randomized 42 hours later for treatment with 1400 U of Pal or buffer. Three mice of each group were killed on days 0, 2, 4, 6, and 8 and the nasal wash was analyzed. All buffer-treated mice remained colonized throughout the experiment (mean log103.0) with titers decreasing progressively from day 6. Pal-treated mice had sterile nasal cultures on day 0 (5 hours after treatment), and thereafter, there was a discrete reappearance of pneumococci in one or two of the three animals tested, with titers significantly lower throughout day 6 and disappearance on day 8 [P < 0.0001 for the comparison of both curves, two-way analysis of variance (ANOVA)] [Web fig. 1 (11)]. Surviving pneumococci, therefore, were unable to reestablish titers sufficient for successful recolonization in mice.

Repeated exposure to low concentrations of Pal (<1 U) on agar plates or increasing concentrations in liquid assays did not lead to the recovery of resistant strains (22). This may be related to the fact that the cell wall receptor for Pal and other pneumococcal phage lytic enzymes is choline, a molecule that is necessary for pneumococcal viability (9, 23, 24). While not yet proven, it is possible that during a phage's association with bacteria over the millennia, to avoid being trapped inside the host, the binding domain of lytic enzymes has evolved to target a unique and essential molecule in the cell wall, making resistance to these enzymes a rare event.

We believe that with Pal and similar bacteriophage lytic enzymes, we may have the opportunity to control or eliminate nasopharyngeal colonization by S. pneumoniae and thus significantly reduce or prevent infection by these bacteria. Because bacteriophage have been described for nearly all bacteria, this targeted approach to control and/or prevent infection may be applied to other pathogens (particularly Gram-positive bacteria) whose reservoir or site of infection is the human mucous membrane.

  • * To whom correspondence should be addressed. E-mail: vaf{at}


View Abstract

Navigate This Article