the growth and survivability of streptococcus pneumoniae clinical isolates subjected to various...

Bacteriology

The growth and survivability of Streptococcus pneumoniae clinicalisolates subjected to various environmental conditions

G.J. Mazzolaa,*, J.E. Mortensenc, L.A. Millerb, J.A. Poupardb

aGlaxoSmithKline Pharmaceuticals, King of Prussia, PA, USAbGlaxoSmithKline Pharmaceuticals, Collegeville, PA, USAcChildren’s Hospital Medical Center, Cincinnati, OH, USA

Received 4 September 2002; accepted 11 November 2002

Abstract

Historically, it has been hypothesized that environmental stress would favor the survival of antibiotic susceptible bacteria over resistantones; however, there is little direct evidence to support this theory. Clinical isolates of S. pneumoniae were chosen and categorized as:penicillin susceptible, quinolone susceptible (PSQS, n � 3); penicillin resistant, quinolone susceptible (PRQS, n � 3); and penicillinresistant, quinolone resistant (PRQR, n � 5). Baseline growth of each isolate was measured by optical density for 24 h. The resulting opticaldensity curves were compared to those obtained for the same isolates subjected to changes in environmental conditions, such as varioustemperature, pH, and diluted media. In addition, each isolate was inoculated onto cotton fiber disks, held at room temperature, and therecoverable CFU measured over 144 h. In comparison to controls grown under ideal conditions, the density of PSQS isolates wassignificantly lower than PRQR isolates after 24 h for the following conditions (p � 0.01): incubation at 40°C (1.3 log10 lower); at pH 6.5(1.6 log10 lower); and in limited nutrient conditions (1.36 log10 lower). When inoculated onto cotton fiber disks, the PRQR isolates decreasedan average of 5.0 log10 after 72 h as compared to controls. In contrast, PSQS isolates decreased an average of 8.1 log10 (p � 0.01). Resultsof this study support the concept that antibiotic resistant isolates may not be at a competitive disadvantage in comparison to susceptibleisolates when subjected to some adverse environmental conditions. © 2003 Elsevier Science Inc. All rights reserved.

1. Introduction

Streptococcus pneumoniae is a leading cause of morbid-ity and death throughout the world with an estimated mor-tality rate of 3–5 million deaths per year (Tomasz, 1997).Since the first isolate of penicillin resistant S. pneumoniaewas discovered in Australia in 1967, reports of resistantisolates have risen dramatically and have caused concernover the effective treatment for pneumococcal infections(Krisher, 1995). Although a definitive prevalence of pneu-mococcal resistance in the U.S. may vary based on locationand age of patients, a recent surveillance study reported that34.2% of clinical isolates were classified as penicillin non-susceptible, and 22.4% classified as multiresistant (Doern,2001).

This increase in resistance has been occurring in directproportion to increased use of antimicrobials (Austin et al.,

1999). Therefore, the current use of antimicrobials is beingclosely scrutinized to develop guidelines to curtail the over-use of these agents. A seemingly obvious strategy for re-ducing the emergence of resistance would be to limit theunnecessary use of antimicrobials. However, the assump-tion that the reduction in the use of antimicrobials willdiminish the existing pool of resistant bacteria is in ques-tion. For example, despite relatively little use in 30 years,streptomycin resistance remains frequent in Gram-negativebacteria (Chiew et al., 1998). It is likely that even totalcessation of specific antimicrobials will not result in a re-version to susceptibility in many cases (Morell, 1997).

Contemporary theory holds that emerging resistance oc-curs according to Darwinian tenants of natural selection andthe introduction of resistant clones into the population. Inthis scenario, those populations of bacteria that developrandom mutations conferring resistance gain a competitivesurvival advantage over those which do not (Livermore etal., 2000). An assumption has also developed which holdsthat the reverse must also be true and forms the basis ofseveral studies aimed at correlating reduction in antimicro-

* Corresponding author. Tel.: �1-610-270-7342; fax: �1-610-270-4572.

E-mail address: [email protected] (G.J. Mazzola).

Diagnostic Microbiology and Infectious Diseasewww.elsevier.com/locate/diagmicrobio45 (2003) 153–164

0732-8893/03/$ – see front matter © 2003 Elsevier Science Inc. All rights reserved.doi:10.1016/S0732-8893(02)00526-6

bial use and reduction in resistant isolates in the environ-ment as reviewed in Monroe et al., and Guillemot (Guil-lemot, 1999; Monroe et al., 2000). That is, in the absence ofdrug pressure, there is actually a cost of fitness to theorganism for carrying a gene coding for resistance. There-fore, the resistant bacterial populations will undergo a lossof resistance genes, and hence be susceptible to the antimi-crobial(s) again. There are several studies demonstratingthat indeed, resistant strains are at a growth disadvantage tosusceptible strains (Helling et al., 1981; Jones et al., 1980).For example, in one study, the plasmid, pBR322 coding fortetracycline-resistance was transformed into E. coli. Thegrowth of this plasmid-containing population under glu-cose-limited conditions as well as in chemostat cultures wascompared to the control group, which did not receive theplasmid. Results showed that the plasmid-containing cellsdid not reproduce as rapidly and fell to about 1 cell perthousand at the end of 100 generations of growth. Theseresults were reproducible in both glucose rich and limitedculture (Lee et al., 1985). However, there is little directevidence to support reversion to susceptibility. In fact,adopting these assumptions ignores the remarkable adapt-ability of microorganisms and their ability to retain resis-tance and gain additional mutations to compensate for anyloss of fitness, thereby gaining an even greater survivaladvantage.

Evidence for advantageous mutation during prolongedstress is abundant (Bennett and Lenski, 1997; Cairns et al.,1988; Leroi et al., 1994; Moore et al., 2000; Papadopouloset al., 1999; Travisano and Lenski, 1996). Also, there isfurther evidence of compensatory mutations which negatethe cost of fitness due to antimicrobial resistance (Anders-son and Levin, 1999; Bjorkman et al., 1998; Bottger et al.,1998; Lenski, 1998). However, most work has been carriedout using Gram-negative bacteria, which were not isolatedfrom patients presenting with clinical infection. Also, thereis evidence that the molecular mechanisms, rate, virulence,and relative fitness of cells undergoing compensatory evo-lution is highly dependent on the specific growth environ-ment (Bjorkman et al., 2000). This naturally leads to thequestion of whether or not these processes actually occurwithin a human host, and if so, how they contribute to theworld wide spread of antimicrobial resistance.

If demonstrated, enhanced survival of resistant isolatesof important clinical pathogens such as S. pneumoniae hasimplications for the widespread dissemination of antimicro-bial resistant strains and could impact policies regarding therationale for appropriate drug selection, environmental dis-infection, and housekeeping procedures in hospital and daycare settings. However, there is little documentation de-scribing long-term S. pneumoniae survivability other than invarious atypical culture media, cotton swabs, or in sputum(Ross and Lough, 1978; Wasas et al., 1998; Williams andKauffman, 1978). In addition, studies regarding survivabil-ity on dry surfaces using species other than S. pneumoniaeshow mixed results when correlating survival and antimi-

crobial resistance (Hirai, 1991; Jawad et al., 1998; Tuo-manen et al., 1986; Wendt et al., 1998).

Therefore, the goal of this study was to describe thesurvival of S. pneumoniae on cotton fiber as well as underunfavorable pH, temperature, and diluted media conditions.Data from this study also supports the notion that there areways in which bacteria can circumvent the physiologic costof antibiotic resistance or adapt to adverse, diluted mediaenvironments

2. Materials and methods

2.1. Isolate selection and preparation

All isolates were selected from the GlaxoSmithKlineClinical Microbiology isolate repository. The clinical iso-lates were isolated from patients and collected as part of theAlexander Project (Gruneberg and Felmingham, 1996) orA.L.E.R.T. (Mortensen et al., 1998) surveillance studies. S.pneumoniae was isolated from clinical specimens, sus-pended in a glycerol solution, and frozen at �70°C.

Isolates included in this study were chosen at randomfrom each of three resistance categories based on previouslyobtained minimal inhibitory concentration (MIC) determi-nations conducted according to NCCLS guidelines (NC-CLS, 1997). Five clinical isolates of penicillin resistant,quinolone resistant (PRQR) S. pneumoniae, and three iso-lates of penicillin resistant, quinolone susceptible (PRQS) S.pneumoniae isolates were included. To serve as the suscep-tible clinical isolate controls, three penicillin susceptible,quinolone susceptible (PSQS) isolates were also tested. S.pneumoniae ATCC 49619 was also used for this study as anoverall experimental control. Isolate MIC values deter-mined in the original study are listed in Table 1. It should benoted that, due to a MIC value of 1 �g/ml, isolate #205229is technically penicillin intermediate, but was included withthe PRQR category during analysis due to its quinoloneresistance profile. As a result, isolate 5052S was added tothe study as an assurance that at least 3 isolates in the PRQRcategory would be penicillin resistant. Isolate 98-641-124Swas also added to include a highly multi-drug resistantisolate.

Prior to use in each experiment of this study, each isolatewas incubated for 12-16 h on Mueller-Hinton plates con-taining 5% sheep blood, pH 7.2 (BBL) and incubated at37°C in 5% CO2. This procedure was then repeated forcolonies scraped from the initial growth plate. Colonieschosen from these plates were then used in the growth andsurvival studies outlined below.

2.2. Baseline growth of each isolate by viable countmethod

Inocula were prepared by suspending growth from sub-culture plates in 10 mL of sterile saline to the turbidity of a

154 G.J. Mazzola et al. / Diagnostic Microbiology and Infectious Disease 45 (2003) 153–164

0.5 McFarland standard. These suspensions were then di-luted 1:200 in cation adjusted Mueller-Hinton broth supple-mented with 5% lysed defribrinated horse blood (GSK Me-dia Prep. Department) to obtain an approximateconcentration of 5 � 105 CFU/mL consistent with NCCLSguidelines (NCCLS, 1997). These tubes were incubated at37°C in 5% CO2 and viable counts performed in triplicate at0, 2, 4, 6, 8, and 24 h. The baseline growth of each isolateas measured by viable plate count was repeated three times.The generation time of each isolate was calculated andcompared by analysis of variance (ANOVA) to demonstrateany statistical differences.

2.3. Validation of growth measurements by optical density

The growth of each isolate over a 24 hour period wasmeasured by inoculating the wells of a microtiter plate(Transgalactic, Inc.) with 100 uL of a 5 � 105 CFU/mLsuspension of organism selected from plates and prepared inMueller-Hinton broth, with 5% lysed horse blood, pH 7.2.Therefore, each well contained a total of 5 � 104 CFU oforganism at the start of testing. A total of 5 wells per isolatewere inoculated. The microtiter plates were incubated at37°C in ambient air and the turbidity of each well wasmeasured by a Bioscreen C® turbidity reader (Transgalac-tic, Inc.) in 30 minute increments for 24 h at a wavelengthof 600 nm. This wavelength was chosen to minimize theinterference of blood absorbance. The entire procedure wasrepeated in triplicate.

To determine that the environmental conditions in theBioscreen device would yield comparable growth curves toa standard incubator, a duplicate microtiter plate was pre-pared and treated as above, but samples were taken fromeach well and plated at 2 hour increments for viable cellcounts. Plots of optical density (O.D.) vs. time were drawnfor each isolate and raw data were exported into Excel(Microsoft) spreadsheets for further analysis.

2.4. Correlation between O.D. and CFU/mL

Raw ASCII data of time and optical density were ex-ported from the Bioscreen software into Excel (Microsoft).The viable counts (all data points without averaging) inCFU/mL were also entered into the same spreadsheet man-ually at the same corresponding timepoints as the O.D. data.These data were then copied into the non-linear estimationmodule of Statistica (StatSoft). The user-specified regres-sion function, was used to approximate the four parameternon-logistic function. In the case of each isolate, the finalequation arrived at to use for the O.D. correlation toCFU/ml was accepted only if the R2 value was greater than0.95. The final equation for each isolate was then used inlater studies to convert O.D. to CFU/mL. For the baselinegrowth study, plots of observed versus predicted valueswere evaluated to determine accuracy of this method.

2.5. The effect of pH on growth

The growth of each isolate was measured as describedabove in section 2.3 with the exception that suspensionswere prepared in Mueller-Hinton broth, with 5% lysed horseblood, at pH 5.0, 6.0, 6.5, 7.2, or 8.0. The pH of 500 mL ofMueller-Hinton, 5% lysed horse blood stock media waslowered using 2.5N HCl (Malinkrodt) or raised using 0.5NNaOH (J.T. Baker), as measured with an Orion pH 920Ameter and semi-micro pH probe. The isolates inoculatedinto the pH 7.2 broth served as the control.

2.6. The effect of temperature on growth

The growth of each isolate was measured as describedabove in section 2.3 with the exception that the microtiterplates were incubated at 25, 37, 40, and 42°C in ambient air.The isolates incubated at 37°C served as the control.

Table 1Minimal inhibitory concentration for the S. pneumoniae isolates used in the study

S. pneumoniae MICs mcg/mL Source

Gemifloxacin Trovafloxacin Ciprofloxacin Ofloxacin Penicillin Erythromycin Source Year

Pen R, Quin R1 205118 0.03 1 8 8 2 0.03 Alex Proj 19922 205229 0.125 4 32 32 1 0.03 Alex Proj 19923 403346 0.25 4 32 32 2 0.03 Alex Proj 19944 50525 0.25 4 32 16 2 0.03 ALERT 19975 98-641-124S 0.125 4 16 16 4 �16 Alex Proj 1998

Pen R, Quin S6 10070S 0.008 0.06 1 2 2 2 ALERT 19977 5049S 0.016 0.125 1 2 8 �16 ALERT 19978 2033S 0.016 0.06 2 2 4 1 ALERT 1997

Pen S, Quin S9 6052S 0.004 0.06 0.25 1 �0.008 �0.016 ALERT 199710 34025S 0.016 0.125 1 2 0.016 �0.016 ALERT 199711 701085 0.016 0.125 1 2 0.016 �0.016 Alex Proj 1997

155G.J. Mazzola et al. / Diagnostic Microbiology and Infectious Disease 45 (2003) 153–164

2.7. The effect of diluted media on growth

The growth of each isolate was measured as describedabove in section 2.3 with the exception that the growthmedium in each well contained a 1:10, or 1:100 salinedilution of Mueller-Hinton broth, with 5% lysed horseblood, pH 7.2, or Mueller-Hinton broth, without 5% lysedhorse blood, pH 7.2. The isolates prepared in Mueller-Hinton broth, containing 5% lysed horse blood, pH 7.2served as the control.

2.8. Survivability on cotton fiber in dry environment

Inocula were prepared by suspending growth from sub-culture plates in 10 mL of sterile saline to the turbidity of a0.5 McFarland standard. These suspensions were then di-luted 1:10 in cation adjusted Mueller-Hinton broth supple-mented with 5% lysed defribrinated horse blood to obtain anapproximate concentration of 1 � 107 CFU/mL. A 50 uLsample was then inoculated onto one of nine 1.27 cm cottondisks (S&S Grade 740-E Special Purpose cotton filterdisks). Each disk corresponded to a given time-point atwhich a plate count was performed. The starting density waschosen due to an anticipated loss of recovery of approxi-mately 0.5–1.0 Log10 at the time of inoculation onto thedisks. Therefore, the maximum recovery from the original50 uL inoculum would be 1.0 � 105–5.0 � 105 CFU.Samples were stored in a sealed falcon tube and maintainedat room temperature. Plate counts were conducted as de-scribed below at time 0, 2, 8, 24, 48, 72, 96, 144 h afterdrying. Controls for each isolate were prepared by dilutingthe original inocula 1:200 (approximately 5 � 105 CFU/mLin order to compare with the original T0 recovery) andstoring at room temperature.

Recovery of organisms from the dry disks was per-formed by adding 2 mL of adjusted Mueller-Hinton brothsupplemented with 5% lysed defribrinated horse blood toeach test tube containing a disk, and vortexing on a lowsetting for 10 s. Appropriate dilutions were made fromsamples taken from each tube and 50 uL were inoculatedonto Mueller-Hinton plates with 5% sheep blood by thespread plate method and incubated overnight at 37°C in 5%CO2. The CFU were counted and the viable cell countrecovered from the dry sample calculated and compared tothe count at initial drying time (T0). Plots of viable cellcount over time were drawn for each isolate and the datawere exported into Excel (Microsoft) for further analysis.

2.9. Statistical analysis

The data from the individual growth curves were con-verted to CFU/mL for each time-point using the equationsderived for each isolate as described above in section 2.4.The 5 well replicates were averaged to form a run average

at each time-point for each of the 3 repeats in each exper-iment. This was completed for each experiment of temper-ature, pH, diluted media, and survival on cotton fiber.

The sorted data were then imported into SAS for “MixedModel” Analysis (Proc Mixed procedure) as outlined inLittell et al. (1996). In order to truncate the data and stillretain useful information about the general shape of eachcurve, only the data points at 6, 8, and 24 h were used in theSAS analysis to highlight the similarities and differencesbetween isolates and resistant categories.

The Proc Mixed procedure allows the modeling of vari-ation between the variable levels tested using the individualaverages from each separate repeated experiment. Contraststatements were used to analyze the differences betweenresistance categories at each time-point, and within thatanalysis, to compare the differences from the experimentalcontrols. These comparisons were grouped by resistancecategory and the categorical differences were then com-pared for statistical significance.

For example, this procedure supplied answers to suchquestions as: Is the difference of PRQR Isolates at 40°C vs.37°C the same as the difference of the PSQS Isolates at40°C vs. 37°C? This analysis was conducted for each timechosen, i.e., 6 h, 8 h, 24 h.

3. Results

3.1. Baseline growth of each isolate by viable countmethod

The growth of each isolate over a 24-h period was mea-sured by viable cell counts in triplicate. The mean log10

CFU/mL for each time point is displayed in Fig. 1. As canbe seen in Fig. 1, the growth curves are characteristic ofthose usually obtained for S. pneumoniae grown under idealconditions. None of the curves are statistically significant (p� 0.05) in regard to their differences from one another withthe exception of isolate 50495 (PRQS), which is signifi-cantly lower at the 8 hour timepoint.

Additionally, the 2 to 8 hour generation time was calcu-lated by the following equation: Generation time (tgen) mea-sured in minutes � 1/k � 60. k � (logT2 � logT1)/0.301*T2-T1. The resulting range of generation time was24–36 min.

An ANOVA was conducted to prove any statisticallysignificant differences in generation time between resistancecategories at an alpha of 0.05. The variation in generationtime between the groups is marginally larger than the vari-ation within the groups, resulting in an F ratio of 1.12, anda p-value of 0.37, which is not sufficient to prove a differ-ence at an alpha level of 0.05. Therefore, there is no sig-nificant difference in generation time.


3.2. Validation of growth measurements by optical density

To test that the environmental conditions in the Bio-screen device would yield comparable growth curves to astandard incubation, samples were taken from a microtiterplate during 24-h growth at 37°C in the Bioscreen device.To minimize the effect that opening the device every 2 hduring the growth cycle would have on the isolates, sampleswere only taken from a total of 6 of the isolates. Theresulting CFU/mL measurements were then plotted alongwith the same measurements taken from the controls incu-bating at 37°C with 5% CO2. The resulting plots showed ahigh degree of correlation between growth of 10 mL ofinocula in 15 mL test tubes at 37°C with 5% CO2 with the100 uL samples incubating at 37°C in the Bioscreen device.

3.3. Correlation between O.D. and CFU/mL

Each isolate was incubated at 37°C for 24 h in theBioscreen device measuring optical density every 30 min.This was repeated on three separate occasions, with eachincluding 5 replicates per isolate. Specific equations foreach isolate were derived from the best line fit analysisusing Statistica software after plotting optical density as a

function of the CFU/mL. These equations were then used inthe subsequent conversion of O.D. readings from the Bio-screen device to CFU/mL for further analysis.

3.4. The effect of temperature on growth

Isolates incubated at 42°C did not increase in density by24 h as measured by optical density. Therefore, only thedata for growth at 25, 37, and 40°C was used in the analysis.In most cases, isolates incubated at 40°C also did not sig-nificantly increase in density at all.

No appreciable growth was observed for the isolatesincubated at 25°C, from time 0 to 8 h. However, by 24 h, thePSQS, PRQS, and PRQR isolates had reached log10 levelsof 8.7, 8.8, and 7.7, respectively. Again, little difference wasobserved between 0 and 8 h when isolates were incubated at40°C. By 24 h, the PSQS, PRQS, and PRQR isolates hadreached log10 levels of 5.0, 6.5, and 6.5, respectively.

Although this information illustrates the differences be-tween categories at each temperature, the more importantcomparison is the difference of each isolate to it’s control,then making comparisons by resistance category. Thesecomparisons are illustrated in Fig. 2.

The growth of PSQS isolates was not as affected by a

Fig. 1. Baseline growth of each isolate using standard plate count method.


reduction in temperature as the PRQS or PRQR isolates. At8 h, the PSQS isolates were on average 3.2 logs lower thanthe control group, as compared to 4.0 lower for the PRQSand 3.8 lower for the PRQR isolates. By 24 h, the PSQS and

PRQS isolate were almost identical to their controls, how-ever, the PRQR were approximately 1.0 log lower.

Incubation at a higher temperature has a different effecton the isolates than incubation at a lower temperature. When

Fig. 2. Isolate growth grouped by resistance category when incubated at 25°C and 40°C as compared to the 37°C control.


incubated at 40°C, there was no difference between catego-ries at 6 h with an average of 2.2–2.6 log10 lower growthcompared to their respective controls. However, at 24 h thePRQS and PRQR isolates showed growth that was only 2.2logs lower than their controls versus that of 3.5 logs lowerfor the PSQS isolates. Therefore, in comparison to controlsgrown under ideal conditions, the density of PSQS isolateswas an average of 1.3 log10 lower than PRQR isolates after24 h at 40°C.

3.5. The effect of pH on growth

Isolates in the pH 5.0 and 6.0 media did not increase indensity as measured by optical density. Therefore, only thedata for growth at pH 6.5, 7.2, and 8.0 was used in theanalysis.

When isolates were incubated in pH 6.5 media, littledifference was observed from time 0 to 8 h in the PSQS andPRQS categories, yet the PRQR isolates displayed an in-crease from 6.2 to 7.4 log10 CFU/mL. By 24 h, the PSQS,PRQS, and PRQR isolates had reached log10 CFU/mL lev-els of 6.6, 8.4, and 7.4, respectively. When incubated inmedia at pH 8.0, the PSQS and PRQR displayed a morerapid increase in density, reaching 6.6 and 6.8 log10 CFU/mL, respectively at 6 h and 8.6 and 8.4 log10 CFU/mL,respectively at 8 h. By 24 h, the PSQS, PRQS, and PRQRisolates had reached log10 CFU/mL levels of 8.7, 8.3, and8.7, respectively.

Although this information illustrates the differences be-tween categories at each pH, the more important compari-son is the difference of each isolate to it’s control, thenmaking comparisons by resistance category. These compar-isons are illustrated in Fig. 3.

When incubated in media at pH 6.5, the growth rate ofthe PSQS isolates was lower than the PRQS and PRQRisolate growth rate in comparison to categorical controls. Inaddition, at 24 h, the PSQS growth was 1.6 log10 lower thanboth the PRQS and PRQR isolate growth, due to an ex-tremely long lag time before any increase in density wasobserved.

Growth in pH 8.0 media resulted in lower rates of growthcompared to controls. However, the observed reduction ingrowth was not as low as that observed for incubation at pH6.5. In fact, the PSQS and PRQR isolates were relativelyunaffected in contrast to the longer lag time observed for thePRQS isolates; and by 24 h, there was no significant dif-ference among the three categories.

3.6. The effect of limited nutrients on growth

Isolates incubated in 1:100 diluted media did not in-crease in density after 24 h as measured by optical density.Therefore, only the data for growth at 1:10 diluted mediaand the Mueller-Hinton without blood was used in theanalysis in comparison to controls.

No significant growth was observed for any of the isolate

categories incubated in media diluted 1:10, from time 0 to8 h. However, by 24 h, the PSQS, PRQS, and PRQR isolateshad reached log10 CFU/mL levels of 6.1, 6.2, and 7.7,respectively.

Although this information illustrates the differences be-tween categories in each media type, the more importantcomparison is the difference of each isolate to it’s control,then making comparisons by resistance category. Thesecomparisons are illustrated in Fig. 4.

When incubated in a 1:10 dilution of Mueller-Hintonwith 5% horse blood, an increased lag time was observedfor all isolates. However, by 24 h, the PRQR isolates hadincreased to a density that was close to levels achieved bythe PRQR controls. In contrast, the PSQS and PRQS iso-lates were 2.4 and 2.6 log10, respectively, lower than theircorresponding controls.

Therefore, in comparison to controls grown under idealconditions, the density of PSQS and PRQS isolates at 24 hwas on average 1.36log10 lower than the PRQR isolates.When incubated in Mueller-Hinton broth without blood, noappreciable growth took place for any of the isolates. There-fore, no significant differences were noted.

3.7. Survival on cotton fiber in dry environment

The CFU recovered from the cotton disks at each time-point was measured by viable plate count methodology. Attime 0, just after isolate inoculation onto the disks, thenumbers of recovered CFU immediately decrease approxi-mately 1 Log10 regardless of resistance category. However,as shown in Fig. 5, after 36 h higher numbers of CFU wererecovered in the PRQS (3.5 Log10) and PRQR (3.6 Log10)categories than the PSQS category (3.1 Log10). After 72 h,the recovered CFU from the PSQS, PRQS, PRQR isolates,had decreased to 0.3, 2.5, and 2.3 Log10, respectively, show-ing a substantial difference between resistant and suscepti-ble isolates. Furthermore, this relationship holds true whencomparing the recovered CFU from the cotton fiber disks tothe control isolates grown in Mueller-Hinton with 5% horseblood at room temperature. This can be seen in Fig. 5, whichillustrates the 8.1 log10 decrease from controls in the PSQScategory in contrast to the 5.0 and 5.4 log decrease, respec-tively, in the PRQR and PRQS categories at 72 h.

4. Discussion

The results of this study are not consistent with theassumption that the genetic and physiologic traits that con-fer antibiotic resistance in S. pneumoniae come at a costwhich render the isolates less capable of dealing with ad-verse environmental changes. For most of the experimentsincluded, the PSQS isolates displayed the greatest decreasein growth and survivability in comparison to the PRQS andPRQR isolates under the same conditions. In fact, the onlygrowth condition under which the PSQS isolates reached a


significantly higher density was growth at 25°C. The great-est differences between resistant and susceptible isolateswere observed when isolates were grown in low pH media,1:10 diluted media, and also survival on cotton fiber. In each

of these cases, the PRQR isolates reached a density that wasbetween 1.3 and 3.0 log10 higher at the terminal time-pointthan the PSQS isolates when compared to controls. It isinteresting to note that there were no significant differences

Fig. 3. Isolate growth grouped by resistance category when incubated in pH 6.5 and pH 8.0 media as compared to the pH 7.2 control.


between the PRQS and PRQR isolates with the exception ofgrowth in 1:10 diluted media.

These differences were due to an increased lag timeobserved for all isolates whenever the growth conditionsdeviated from the ideal set points. However, the lag time

observed for the PRQR and PRQS isolates was, on average,shorter than the PSQS isolates. It is possible that the PRQRand PRQS isolates have developed compensatory mecha-nisms that ameliorate the effects of environmental change.This is consistent with experimental results illustrating both

Fig. 4. Isolate growth grouped by resistance category when incubated in 1:10 diluted media and media without blood to the standard media control.


theories of adaptive mutagenesis and compensatory muta-tion (Cairns et al., 1988; Lenski and Bouma, 1988; Leroi etal., 1994; Papadopoulos et al., 1999; Schrag et al., 1997).These theories hold that while there is a cost of fitness ofresistance, these physiologic detriments can be amelioratedover time either by direct mutation or natural selection overseveral generations.

Results of this study also show an immediate advantage

of resistance after a delayed lag phase. This may seem tosupport the notion of adaptive mutagenesis and contradictthe majority of findings in literature of compensatory mu-tation which point out that beneficial mutations are usuallyobserved after approximately 100 generations (Phillips,1996). Unfortunately, since these are clinical isolates, thereis no way of determining what, if any, mutations had pre-viously occurred in each population during the course of

Fig. 5. Decrease in recovered CFU grouped by resistance category when inoculated onto cotton fiber disks and comparison to isolate controls which wereinoculated into Mueller-Hinton broth, 5% blood, pH 7.2, kept at room temperature for the same time course.


infection. However, it stands to reason that, had there beenany detrimental effects of resistance in the PRQR and PRQSisolates prior to infection, then mutations which amelioratethose effects may have already taken place prior to exper-imentation. Also, the number of generations necessary toproduce advantageous mutations may be dependent on themechanism by which they occur. For instance, under star-vation conditions, E.coli carrying the Tn5 Bleomcin resis-tance gene have an immediate survival advantage over sus-ceptible E.coli. It is hypothesized that bleomycin resistanceplays a role in DNA repair and may therefore explain itsadvantageous effect during starvation (Blot et al., 1991). Itmust also be noted that since the organisms chosen for thisstudy were not derived from isogenic strains, there may befactors other than resistance contributing to the enhancedsurvival of some isolates that reflect the genetic diversity ofthe species.

The conclusion from this study is that the drug-resistantS. pneumoniae (DRSP) isolates chosen here were not at acompetitive disadvantage compared to the susceptible iso-lates. This suggests that DRSP isolates may be able torespond and adapt to adverse environmental conditions,leading to enhanced survival over susceptible isolates. Themechanisms of possible adaptation were not studied. Thereare, however, several studies of E.coli showing advanta-geous mutation during prolonged stress (Cairns et al., 1988;Leroi et al., 1994; Papadopoulos et al., 1999). Furthermore,there is evidence for compensatory mutations which negatethe cost of fitness due to antimicrobial resistance(Anderssonand Levin, 1999; Andersson and Levin, 1999; Bottger et al.,1998; Lenski and Bouma, 1988). Although most work hasbeen carried out using E.coli, there is evidence that envi-ronmental adaptation in S. pneumoniae occurs by many ofthe same recA dependent mechanisms and incorporation ofinsertion elements (Claverys et al., 2000; Havarstein et al.,1997). A competitive advantage of DRSP clinical isolatesover susceptible isolates has widespread implications for thespread of community acquired pneumonia, nosocomial in-fection, and fomite transmission in day care centers.

Curtailing the use of a specific drug may slow the rate atwhich resistance will arise and spread, but the data pre-sented here casts doubt that the costs of fitness to bacteriathat carry resistance genes will cause their demise in com-petition to susceptible ones in an environment lacking theselective agent. This leads to the rather pessimistic conclu-sion that controlling the spread of resistance may be evenmore difficult than originally anticipated.

Acknowledgments

The authors thank Dr. Paul McAllister, Director Statis-tical Sciences North America, GSK for his help in designingthe statistical analysis.

References

Andersson, D. I., & Levin, B. R. (1999). The biological cost of antibioticresistance. Curr Opin Microbiol, 2, 489–493.

Austin, D. J., Kristinsson, K. G., & Anderson, R. M. (1999). The relation-ship between the volume of antimicrobial consumption in human com-munities and the frequency of resistance. Proc Natl Acad Sci USA, 96,1152–1156.

Bennett, A. F., & Lenski, R. E. (1997). Phenotypic and evolutionaryadaptation of a model bacterial system to stressful thermal environ-ments. EXS, 83, 135–154.

Bjorkman, J., Hughes, D., & Andersson, D. I. (1998). Virulence of anti-biotic-resistant Salmonella typhimurium. Proc Natl Acad Sci USA, 95,3949–3953.

Bjorkman, J., Nagaev, I., Berg, O. G., Hughes, D., & Andersson, D. I.(2000). Effects of environment on compensatory mutations to amelio-rate costs of antibiotic resistance. Science, 287, 1479–1482.

Blot, M., Meyer, J., & Arber, W. (1991). Bleomycin-resistance genederived from the transposon Tn5 confers selective advantage to Esch-erichia coli K-12. Proc Natl Acad Sci USA, 88, 9112–9116.

Bottger, E. C., Springer, B., Pletschette, M., & Sander, P. (1998). Fitnessof antibiotic-resistant microorganisms and compensatory mutations.Nat Med, 4, 1343–1344.

Cairns, J., Overbaugh, J., & Miller, S. (1988). The origin of mutants.Nature, 335, 142–145.

Chiew, Y. F., Yeo, S. F., Hall, L. M., & Livermore, D. M. (1998). Cansusceptibility to an antimicrobial be restored by halting its use? Thecase of streptomycin versus Enterobacteriaceae. J Antimicrob Che-mother, 41, 247–251.

Claverys, J. P., Prudhomme, M., Mortier-Barriere, I., & Martin, B. (2000).Adaptation to the environment: Streptococcus pneumoniae, a paradigmfor recombination-mediated genetic plasticity? Mol Microbiol, 35,251–259.

Doern, G. V. (2001). Antimicrobial use and the emergence of antimicrobialresistance with Streptococcus pneumoniae in the United States. ClinInfect Dis, 33 (Suppl 3), S187–S192.

Gruneberg, R. N., & Felmingham, D. (1996). Results of the AlexanderProject: a continuing, multicenter study of the antimicrobial suscepti-bility of community-acquired lower respiratory tract bacterial patho-gens. Diagn Microbiol Infect Dis, 25, 169–181.

Guillemot, D. (1999). Antibiotic use in humans and bacterial resistance.Curr Opin Microbiol, 2, 494–498.

Havarstein, L. S., Hakenbeck, R., & Gaustad, P. (1997). Natural compe-tence in the genus Streptococcus: evidence that streptococci can changepherotype by interspecies recombinational exchanges. J Bacteriol, 179,6589–6594.

Helling, R. B., Kinney, T., & Adams, J. (1981). The maintenance ofplasmid-containing organisms in populations of Escherichia coli. J GenMicrobiol, 123, 129–141.

Hirai, Y. (1991). Survival of bacteria under dry conditions; from a view-point of nosocomial infection. J Hosp Infect, 19, 191–200.

Jawad, A., Seifert, H., Snelling, A. M., Heritage, J., & Hawkey, P. M.(1998). Survival of Acinetobacter baumannii on dry surfaces: compar-ison of outbreak and sporadic isolates. J Clin Microbiol, 36, 1938–1941.

Jones, I. M., Primrose, S. B., Robinson, A., & Ellwood, D. C. (1980).Maintenance of some ColE1-type plasmids in chemostat culture. MolGen Genet, 180, 579–584.

Krisher, K. (1995). Antibiotic-resistance in Streptococcus pneumoniae.Adv Exp Med Biol, 390, 123–140.

Lee, S. W., & Edlin, G. (1985). Expression of tetracycline resistance inpBR322 derivatives reduces the reproductive fitness of plasmid-con-taining Escherichia coli. Gene, 39, 173–180.

Lenski, R. E., & Bouma, J. E. (1988). Stability of recombinant DNA andits effects on fitness. Trends in Ecology & Evolution, 3, 18–20.


Lenski, R. E. (1998). Bacterial evolution and the cost of antibiotic resis-tance. Int Microbiol, 1, 265–270.

Leroi, A. M., Bennett, A. F., & Lenski, R. E. (1994). Temperature accli-mation and competitive fitness: an experimental test of the beneficialacclimation assumption. Proc Natl Acad Sci USA, 91, 1917–1921.

Littell, R. C., Milliken, G. A., Stroup, W. W., & Wolfinger, R. D. (1996).SAS System for Mixed Models. SAS Institute Inc., Cary, NC.

Livermore, D. M., & Dudley, M. N. (2000). Antimicrobials: better use,better drugs, or both? Curr Opin Microbiol, 3, 487–488.

Monroe, S., & Polk, R. (2000). Antimicrobial use and bacterial resistance.Curr Opin Microbiol, 3, 496–501.

Moore, F. B., Rozen, D. E., & Lenski, R. E. (2000). Pervasive compen-satory adaptation in Escherichia coli. Proc R Soc Lond B Biol Sci, 267,515–522.

Morell, V. (1997). Antibiotic resistance: road of no return. Science, 278,575–576.

Mortensen, J., Jacobs, M. R., & Koeth, L. (1998). Susceptibility of Strep-tococcus pneumoniae and Haemophilus Influenza in 1997: results of ahospital based study. San Diego, CA, 38th Interscience Conference onAntimicrobial Agents and Chemotherapy (ICAAC).

National Commitee for Clinical Laboratory Standards. (1997). Methods forDilution Antimicrobial Susceptibility Tests for Bacteria that Grow Aero-bically-Fourth Edition; Approved Standard. NCCLS, Villanova, PA.

Papadopoulos, D., Schneider, D., Meier-Eiss, J., Arber, W., Lenski, R. E.,& Blot, M. (1999). Genomic evolution during a 10,000-generationexperiment with bacteria. Proc Natl Acad Sci USA, 96, 3807–3812.

Phillips, P. C. (1996). Waiting for a compensatory mutation: phase zero ofthe shifting-balance process. Genet Res, 67, 271–283.

Ross, P. W., & Lough, H. (1978). Survival of upper respiratory tractbacteria on cotton-wool swabs. J Clin Pathol, 31, 430–433.

Schrag, S. J., Perrot, V., & Levin, B. R. (1997). Adaptation to the fitnesscosts of antibiotic resistance in Escherichia coli. Proc R Soc Lond BBiol Sci, 264, 1287–1291.

Tomasz, A. (1997). Antibiotic resistance in Streptococcus pneumoniae.Clin Infect Dis, 24 (Suppl 1), S85–S88.

Travisano, M., & Lenski, R. E. (1996). Long-term experimental evolutionin Escherichia coli. IV. Targets of selection and the specificity ofadaptation. Genetics, 143, 15–26.

Tuomanen, E., Cozens, R., Tosch, W., Zak, O., & Tomasz, A. (1986). Therate of killing of Escherichia coli by beta-lactam antibiotics is strictlyproportional to the rate of bacterial growth. J Gen Microbiol, 132,1297–1304.

Wasas, A. D., Huebner, R. E., De Blanche, M., & Klugman, K. P. (1998).Long-term survival of Streptococcus pneumoniae at room temperatureon Dorset egg medium. J Clin Microbiol, 36, 1139–1140.

Wendt, C., Wiesenthal, B., Dietz, E., & Ruden, H. (1998). Survival ofvancomycin-resistant and vancomycin-susceptible enterococci on drysurfaces. J Clin Microbiol, 36, 3734–3736.

Williams, S. G., & Kauffman, C. A. (1978). Survival of Streptococcuspneumoniae in sputum from patients with pneumonia. J Clin Micro-biol, 7, 3–5.


the growth and survivability of streptococcus pneumoniae clinical isolates subjected to various...

Documents