Expression of lux Genes in a Clinical Isolate of Streptococcus pneumoniae: Using Bioluminescence To Monitor Gemifloxacin Activity

Abstract

A clinical isolate of Streptococcus pneumoniae was transformed with a plasmid containing the lux operon of Photorhabdus luminescens that had been modified to function in gram-positive bacteria. Cells containing this plasmid produced light stably and constitutively, without compromising the growth rate. Light output was correlated with measurements of optical density and viable counts during exponential growth and provided a sensitive, real-time measure of the pharmacodynamics of the fluoroquinolone gemifloxacin.

The luxCDABE operon of Photorhabdus luminescens has been expressed successfully in a variety of gram-negative bacteria, thereby generating a bioluminescence phenotype that does not require the addition of exogenous substrate (11, 16). Light output from these bioluminescent bacteria is a highly sensitive reporter of metabolic activity (10) and can therefore be used to monitor the real-time effects of antimicrobials on bacterial metabolism (14, 15). Moreover, in experimental systems in which a strong correlation between bioluminescence and viable counts can be demonstrated, measurement of bioluminescence offers a rapid, alternative method for monitoring bacterial viability (10, 14).

Recent publications (3, 13) describe modifications to the lux operon that enable it to be expressed efficiently in the gram-positive bacterium Staphylococcus aureus. In this paper, we describe the construction of a derivative of the Streptococcus spp.-Escherichia coli shuttle vector pVA838 in which the modified lux operon described by Qazi et al. (13) was placed under the control of a constitutive promoter from S. pneumoniae. Introduction of this plasmid into a clinical isolate of S. pneumoniae conferred a highly stable bioluminescence phenotype. We have examined the relationship between bioluminescence and other parameters used to monitor growth and viability and evaluated the use of bioluminescence as a means of monitoring the pharmacodynamic effects of gemifloxacin on S. pneumoniae. Gemifloxacin is a new fluoroquinolone which is potent against a wide spectrum of gram-positive and gram-negative pathogens (12).

Bacterial strains and plasmids.

S. pneumoniae strain SMH 11622 (6) was a clinical isolate from Southmead Hospital (Bristol, United Kingdom). E. coli DH5α was used as a host for propagating plasmids. Plasmid pSB2025 carries the lux operon from P. luminescens, which was modified for expression in gram-positive bacteria and cloned into pSL1190 (13). All other constructs were derivatives of pBluescript or the Streptococcus-E. coli shuttle vector pVA838 (9).

Construction of a plasmid expressing the lux genes in S. pneumoniae.

In a previous attempt to generate a bioluminescent transformant of S. pneumoniae, the cloned luxAB genes from Vibrio harveyi (5) were introduced into pVA838 as a BamHI-SalI fragment via pBluescript (Fig. 1).In order to gain expression from this construct, the constitutive promoter of the S. pneumoniae aminopterin resistance operon (ami) (1) was fused to the 5" end of luxA. This was achieved by subcloning the ami promoter (nucleotide positions 176 to 217; EMBL accession number X17337) as a PvuII-EcoRV fragment (via pBluescript) into the SmaI site of the multiple cloning site immediately upstream of luxAB. Transformants of S. pneumoniae D39 containing this plasmid (pAL1) produced low levels of bioluminescence upon addition of the substrate.

FIG. 1.

Schematic diagram showing the construction of plasmid pAL2. Arrows indicate the locations and orientations of open reading frames. Gram-positive ribosome-binding sites, located upstream of luxA, luxC, and luxE in the modified lux cassette, are represented by shaded regions at the 5" ends of open reading frames. The nucleotide sequence shows the region between the ami promoter and the start of the modified lux operon: the −10 and −35 regions are shown in bold (lowercase), the ribosome binding site is shown in bold and uppercase, the luxA start codon is shown in uppercase, and restriction sites are underlined. Em-R, erythromycin resistance gene; Cm-R, chloramphenicol resistance gene.

To generate a self-bioluminescent transformant of S. pneumoniae that would express higher levels of bioluminescence without the addition of exogenous substrate, we replaced luxAB in pAL1 with the lux operon from P. luminescens that had been modified (and redesignated luxABCDE) for optimal expression in gram-positive bacteria (Fig. 1) (13). To enable the luxABCDE cassette to be subcloned into pAL1, an EcoRI restriction site was introduced downstream of luxE by inserting an oligonucleotide linker (5"-GCGTCGAATTCGACGCTGCA-3" [the EcoRI site is underlined]) into the unique PstI site in pSB2025. The luxABCDE cassette was excised from this construct (pSB2025-1) by digestion with EcoRI and ligated to the 9-kbp EcoRI fragment from pAL1. Following the transformation of E. coli DH5α and selection on Luria-Bertani agar containing erythromycin (500 μg ml⁻¹), several highly bioluminescent transformants were isolated. One of these was found to contain pAL1 carrying the luxABCDE operon in the correct orientation for expression under the control of the ami promoter. This construct was designated pAL2 (Fig. 1). The sequence of the promoter region and its junction with the modified lux operon (Fig. 1) was confirmed by automated DNA sequencing (MWG Biotech, Ebersberg, Germany).

S. pneumoniae SMH 11622 was transformed with pAL2 according to the optimized protocol of Bricker and Camilli (2). Cells were grown in a candle jar at 37°C in Todd-Hewitt broth (Oxoid, Basingstoke, United Kingdom) supplemented with 0.5% yeast extract, and transformants were selected on blood agar plates containing 150 μg of erythromycin ml⁻¹. Cultures of the lux transformant and the wild-type strain were then grown at 37°C in air to select for clones that had lost their CO₂ dependency. These clones were then grown in air in all subsequent experiments.

Evaluation of bioluminescence, growth, and plasmid stability in cultures of S. pneumoniae SMH 11622 containing the modified lux operon.

For an evaluation of this experimental system, S. pneumoniae SMH 11622/pAL2 was grown at 37°C in static cultures in brain heart infusion broth (Oxoid), and bioluminescence was monitored as a function of time and cell growth (optical density [OD] and viable counts). Bioluminescence and the OD at 590 nm (OD₅₉₀) were measured on an automated luminometer-photometer (Lucy1; Anthos), and viable counts were determined by plating cells onto blood agar with a spiral plater (Autoplate model 3000; Spiral Biotech, Bethesda, Md.). The bioluminescence of untreated cultures increased exponentially over the first 6 to 7 h and reached a maximum of approximately 10³ relative light units after 10 h (Fig. 2a and b).Throughout this period, we observed a positive linear correlation between bioluminescence and OD (coefficient of determination, R² = 0.9578) (Fig. 2a and c). However, this relationship was abolished during the late exponential and early stationary phases of growth, when bioluminescence decreased markedly (Fig. 2a). This observation demonstrates that OD and bioluminescence reflect different parameters (biomass and metabolic activity, respectively) and that, under certain conditions, the two measurements are not closely correlated. The growth curve of the wild-type strain did not differ significantly from that of the lux transformant (Fig. 2a), indicating that bioluminescence did not inhibit growth rate and that plasmid pAL2 did not present a significant metabolic load.

FIG. 2.

Relationship between growth (□) and bioluminescence (○) in batch cultures of S. pneumoniae SMH 11622/pAL2. Growth was monitored by measuring OD₅₉₀ (a) or CFU per milliliter (b). Growth of the wild-type strain was monitored by measuring OD₅₉₀ only (▵) (a). A scatterplot (c) shows the correlation between bioluminescence and OD₅₉₀. RLU, relative light units.

We were initially unable to confirm a close correlation between viability and either bioluminescence or OD because the data on the viable counts were consistently erratic, particularly during the early exponential phase, when viability appeared to decrease (Fig. 2 and 3).However, when the data on the untreated culture in the regrowth experiment (Fig. 4)were analyzed, they revealed a positive linear correlation (R² = 0.9261) between bioluminescence and viable counts during exponential growth. It is possible that some of the colony counts in Fig. 1 were artificially low, possibly as a result of the addition of magnesium chloride at a 1% (wt/vol) concentration to the blood agar. When monitoring the bactericidal effects of quinolones by measuring CFU per milliliter, it is standard practice to use MgCl₂ at this concentration to neutralize any of the quinolone that is carried over to the agar plates (6). However, colony counts were subsequently found to be higher in the absence of MgCl₂, which suggests that colony formation in this strain may be hypersensitive to Mg²⁺.

FIG. 3.

Kill curves comparing the effects of gemifloxacin on bioluminescence and OD₅₉₀ (a) and bioluminescence and viable counts (b) in batch cultures of S. pneumoniae SMH 11622/pAL2. Gemifloxacin was added at final concentrations of 0 (□), 0.1 (○), 0.4 (◊), and 1.1 (▵) mg/liter. Open symbols represent growth (OD₅₉₀ or CFU per milliliter), and closed symbols represent bioluminescence. Data points are expressed as the means (± standard deviations) of measurements taken in duplicate (viable counts) or triplicate (bioluminescence and OD₅₉₀).

FIG. 4.

Use of bioluminescence (a), viable counts (b), and OD₅₉₀ (c) to monitor the regrowth of S. pneumoniae SMH 11622/pAL2 following exposure to gemifloxacin. At time zero, gemifloxacin was added at final concentrations of 0 (□), 0.1 (○), and 0.4 (◊) mg/liter and then removed by dilution (1:100) after 1 h. Data points are expressed as the means (± standard deviations) of measurements taken in duplicate (b) or triplicate (a and c).

The genetic stability of pAL2 in S. pneumoniae SMH 11622 was examined in cultures grown for up to 22 h in the absence of erythromycin. We detected no significant loss of pAL2, as determined from the bioluminescence of cultures (with or without erythromycin) or from colony counts on blood agar (with or without erythromycin). Given that pAL2 was maintained stably, erythromycin was omitted from experimental cultures to prevent any interference with the effects of gemifloxacin.

Concentration-dependent killing.

The activity of gemifloxacin towards the bioluminescent derivative of S. pneumoniae SMH 11622 (gemifloxacin MIC, 0.16 mg/liter) (7) was determined by monitoring bioluminescence, OD, and viable counts in static cultures containing gemifloxacin at concentrations of 0.1 mg/liter (minimum concentration in serum), 0.4 mg/liter, and 1.1 mg/liter (maximum concentration in serum). The kill curves (Fig. 3) show that gemifloxacin inhibited bioluminescence, OD, and viability in a concentration-dependent manner and that changes in all three parameters were closely correlated.

Bacterial regrowth after exposure to gemifloxacin.

Regrowth was monitored after log-phase cultures of S. pneumoniae SMH 11622/pAL2 were incubated with gemifloxacin (0.1 and 0.4 mg/liter) for 1 h and then diluted (1:100) in fresh brain heart infusion broth to effectively remove the antibiotic. By using bioluminescence (Fig. 4a), viable counts (Fig. 4b), and OD (Fig. 4c) to monitor recovery, we obtained similar, concentration-dependent regrowth curves. Bacterial regrowth following exposure to antibiotics can be quantified by calculating the postantibiotic effect (PAE). Traditionally, PAE has been determined by viable counting, which represents the difference in time taken for the viable counts of treated and untreated cultures to increase by 1 log₁₀ above the viable counts measured immediately after removal of the antibiotic (8). The laborious nature of this method has led to the development of several alternative procedures for determining PAE, including the measurement of intracellular ATP by bioluminescence (4, 15). We used the data presented in Fig. 4a and b to compare the self-bioluminescence and viable counts as methods for calculating PAE. For cells exposed to 0.1 mg of gemifloxacin/liter, the PAE was calculated as 50 min by measuring bioluminescence and 62 min by viable count; for cells incubated with 0.4 mg of gemifloxacin/liter, the PAE was calculated as 8.1 h by measuring bioluminescence and 8.6 h by viable count. The similarity of the PAE values obtained by these two methods indicates that measuring bioluminescence is a valid alternative method for monitoring the PAEs of quinolones on S. pneumoniae.

Conclusions.

Studies of the pharmacodynamics of quinolones against S. pneumoniae are usually performed by monitoring changes in viable counts (for examples, see references 6 and 7). In this paper, we have described the generation of a self-bioluminescent clinical isolate of S. pneumoniae and provided evidence that measurement of light output from this strain is a valid approach for monitoring the antimicrobial pharmacodynamics.

Constitutive and stable expression of the lux genes in S. pneumoniae was achieved by placing the modified lux operon (13) downstream of the ami promoter (1) in a derivative of pVA838 (9). In evaluating this system, we demonstrated that expression of bioluminescence had a negligible effect upon both growth rate and MIC, which indicates that our interpretation of the pharmacodynamic data should be relevant to the wild-type strain. We observed a strong correlation between bioluminescence, OD, and viable counts during exponential growth in the absence or presence of gemifloxacin. However, bioluminescence was shown consistently to peak before the OD and viable counts reached their maximum values and then to decline sharply. This observation indicates that there is a considerable drop in metabolic activity as cells enter the stationary phase and illustrates that bioluminescence provides an additional parameter to those of OD and viable counts but that the three parameters are not always correlated.

Data from the kill curves and regrowth experiments show that gemifloxacin has similar effects upon metabolic activity and cell replication in S. pneumoniae. It should be stressed, however, that exposure to other classes of antibiotics, or indeed other quinolones, may have effects upon cellular metabolism different from the effects upon cell replication (for an example, see reference 15), and in these situations, measurements of bioluminescence would have to be interpreted together with other growth parameters. This study provides an example of the enormous potential that bioluminescent derivatives of gram-positive pathogens offer in the observation of the real-time pharmacodynamic effects of antimicrobial agents. Light output is noncumulative, reflecting actual metabolic rate, and can be measured directly, continuously, and nondestructively in high-throughput screening or continuous-culture models.