Modification of bacteraemia by specific antibodies and relation with mortality in a pneumococcal mouse sepsis model

Abstract

The relationship between mortality and the bacteraemic profile was investigated in a pneumococcal (serotype 6B) sepsis BALB/c mouse model where animals received protection by specific hyperimmune serum. A single intraperitoneal dose of hyperimmune serum obtained from mice immunized with the heat-inactivated strain was administered (non-diluted or diluted to 1/4 or to 1/16) to 5-mice study groups 1 h prior to intraperitoneal inoculation with the infective inoculum (3·57 × 10⁸ cfu/ml). Blood cultures were performed daily over 15 days, with 8 μl of blood being collected from the tail vein; the samples were resuspended in Todd-Hewitt broth containing 10% trisodium citrate and plated onto blood agar for colony counting. Animals included in the control group received placebo (PBS). Mortality was 100% in control animals within the first 48 h. Hyperimmune serum decreased and delayed mortality in a dose-related trend, producing 100%, 80%, 60% and 40% survival rates at 72, 96, 144 and 360 h, with non-diluted serum. Bacteraemic profiles with maximum colony counts ≥5 × 10⁷ cfu/ml in blood during the follow-up period were related to ≥65% probability of death, regardless of the serum dilution administered.

INTRODUCTION

In the pre-antibiotic era, antibody-based immunotherapy was used to treat pneumococcal diseases [1]. Immunogenicity depends on the pneumococcal serotype [2], and evidence of its participation in outcomes is based on the spontaneous resolution of fever in the absence of treatment at the time when capsular antibodies appear [3], and the increase in severity of infections when IgG₂[4] or C₃ complement [5] deficiencies are present. The moderate effectiveness of administering serum containing type-specific antibodies in the pre-antibiotic era was probably due to differences in serotype immunogenicity. Serotype 6B, which is one of the most common serotypes isolated in bacteraemia [2,6], is poorly immunogenic [2].

The capability of a strain to invade the host should be related to the bacteraemic profile and, if the strain is pathogenic, its bacteraemic profile (bacterial capability of invasion) should be related to death. To our knowledge, no previous studies have investigated this relationship. This model attempts to explore the effect that the presence of specific antibodies (obtained by pre-inoculation administration of specific hyperimmune serum) has on mortality and the bacteraemic profile produced by a serotype 6B Streptococcus pneumoniae in a sepsis mouse model.

MATERIALS AND METHODS

The study was performed in accordance with regulations in force on the care and use of laboratory animals in the European Community.

Infecting strain

A serotype 6B Streptococcus pneumoniae strain (MIC/MBC penicillin = 2/4 μg/ml) was grown until an absorbance of 0·3 (UV-VIS spectrophotometer Shiimadzu UV-1203, Japan) was obtained in Todd-Hewitt broth supplemented with 0·5% yeast extract (THYB) (Difco, Detroit, MI, USA), aliquoted and stored at –70°C in 15% glycerol. These bacterial aliquots were used in all the following experiments.

Animals

Female BALB/c mice of 8–12 weeks old and weighing 19–22 g were used.

Determination of the minimal lethal dose and challenge dose

Groups of 10 mice per dilution were intraperitoneally inoculated with 0·2 ml of a 10², 10⁴, 10⁶, 10⁷ and 10⁸ colony-forming units (cfu)/ml suspension (spectrometrically measured) to determine the minimal dose causing 100% mortality over a 15 day follow-up period. Bacteria in a logarithmic phase of growth in THYB were centrifuged. The pellet was washed three times and resuspended in phosphate-buffered saline (PBS), pH 7·2, to reach 10⁸ cfu/ml (spectrometrically measured); the suspension was adequately diluted to obtain the different bacterial concentrations. The inocula were confirmed by culture of serial dilutions onto blood Mueller-Hinton agar incubated at 37°C in 5% CO₂ air. The number of dead mice was recorded daily. The minimal lethal dose (MLD) was calculated from the results obtained in three independent experiments. Once the MLD had been determined, 2 × MLD/ml was used as the infective inoculum (challenge dose).

The bacteraemic profile was investigated as described below in groups of five animals inoculated with the challenge dose and two 100-fold lower bacterial dilutions.

Hyperimmune serum

The bacteria in a logarithmic phase of growth were inactivated at 60°C for 1 h. Animals were inoculated weekly with 200 μl of the inactivated bacterial suspension (10⁸ cfu/ml in PBS) intraperitoneally for 5 weeks. Animals were exsanguinated by cardiac puncture to obtain the serum. Specific IgG antibodies to capsular 6B polysaccharide were determined for pre- and immune sera by enzyme-linked immunoassay (ELISA), according to the standardized ELISA protocol (Workshop at the Centers for Disease Control, Atlanta, Georgia, 1996). Neutralization of antibodies to cell-wall polysaccharide was carried out with CWPS (Statens Serum Institute, Copenhagen, Denmark). For detection, horseradish peroxidase-conjugated goat anti-mouse IgG (Bio-Rad, Richmond, CA, USA) was used.

Hyperimmune serum was diluted in PBS to obtain the different dilutions to be administered.

Dose-ranging study

The survival rates and bacteraemic profiles of animals inoculated with the challenge dose over a 15 day follow-up period were determined in a dose-ranging study with administration of non-diluted serum, or dilutions of 1/4 and 1/16. A single intraperitoneal injection of 200 μl of hyperimmune serum was administered 1 h prior to intraperitoneal inoculation with 200 μl of the challenge dose. Animals in the control group received PBS as placebo in order to minimize the number of animals used in the study. In addition, 20 animals received non-immune serum (from a pool of sera collected from mice prior to immunization) as placebo, to rule out a protective effect of normal serum. Blood samples were obtained daily (except on day 1 when they were collected at 2, 6 and 24 h) over the 15 day follow-up period from five animals per study group, to study the bacteraemic profile. Samples were taken from live animals and from those that had died since the previous sampling time. In order to rule out overgrowth after the death of animals, which could mislead the results, the bacteraemic profile was also investigated up to 72 h after death, in five placebo-treated animals, with sampling times at 8 h intervals. To collect blood samples, tails were disinfected and anaesthetized (local anaesthesia; ethyl chloride, Cloretilo Cheminosa, Ern, Barcelona, Spain), and the terminal portion of the tail was eliminated with scissors. An 8 μl blood sample was collected by pressing the tail and using a calibrated loop; this was then resuspended in Todd-Hewitt broth containing 10% trisodium citrate, and plated onto blood agar for colony counting of the first sample. To obtain the subsequent blood samples, the crust was removed and again, by pressing the anaesthetized tail, an 8 μl blood sample was collected and processed following the microbiological procedure described. This collection procedure ensured that blood collection did not compromise mouse survivability. The lower limit of detection was 1·25 × 10² cfu/ml.

Statistical analysis

Survival curves were obtained by the Kaplan-Meier method. An ordinal log-rank test was used to compare different study groups. A Cox regression analysis was also used to compare survival with each dilution tested when dilutions were cube root transformed. A Probit regression analysis was used to study the relationship between the maximum cfu/ml and outcome (death/survival).

RESULTS

Immunoglobulin antibodies of the hyperimmune serum, titrated in parallel with the corresponding pre-serum to 6B polysaccharide by ELISA, were 1:400.

Mortality ranged from 0% with 10² cfu/ml to 40% with 10⁷ cfu/ml as infective inocula. The minimal lethal dose confirmed by culture was 5 × 10⁷ cfu/ml, and the challenge dose was set at 1 × 10⁸ cfu/ml. The inoculum, confirmed by culture in the dose-ranging study, was 3·57 × 10⁸ cfu/ml. Figure 1 shows the median bacterial counts in blood samples over 144 h with the challenge dose (3·57 × 10⁸ cfu/ml) and the two 100-fold lower inocula tested: 3·57 × 10⁴ and 3·57 × 10⁶ cfu/ml. Mortality was 20% with 3·57 × 10⁴ (one animal dead at 216 h) and 3·57 × 10⁶ cfu/ml (one animal dead at 168 h), and 100% with the challenge dose (four animals dead at 24 h and one animal at 48 h) as expected. The bacteraemic profiles of the five animals inoculated with the challenge dose were similar; all animals had colony counts ≥10⁸ cfu/ml for at least 50% of the 48 h survival period (all animals died within 48 h). The bacteraemic profiles with 10⁴ and 10⁶ cfu/ml showed a great variability between animals, but colony counts were always <10⁸ cfu/ml, except in the two dead animals mentioned above where colony counts at death were ≥10⁸ cfu/ml.

Fig. 1.

Median bacteraemic profile (median colony counts; n = 5) over 144 h after inoculation with 10⁴ (▴), 10⁶ (▪) or 10⁸ (•) cfu/ml.

The survival curves over the 360 h follow-up period in the placebo (saline) and the three serum-administered groups are shown in Fig. 2. Over the first 72 h, 100% and 80% survival rates were obtained with non-diluted serum and serum diluted to 1/4, respectively; on the contrary, 80% and 100% mortality was obtained in serum diluted to 1/16 and control groups, respectively, both at 48 h and 72 h. The effect of non-immune serum as placebo (95% mortality, 19 out of 20 mice, within 48 h and 100% at 72 h) was not different to that of the saline placebo. A statistically significant difference (P = 0·001) was found with respect to mortality over time between the four study groups using the ordinal log-rank test. The effect of serum dilutions was also statistically significant (P = 0·0081) when the Cox regression was used.

Fig. 2.

Survival curves over the 360 h follow-up period in the placebo and the three serum-administered groups. (▴) Control; (▪) serum 1/16; (•) serum 1/4; (*) non-diluted.

Visually, the bacteraemic profiles of animals receiving serum diluted to 1/16 (100% mortality) were similar to the bacteraemic profile of control animals infected with the challenge dose and shown in Fig. 1. Figures 3 and Figures 4 show the individual bacteraemic profile over 144 h of dead and surviving mice, respectively, all of which received either non-diluted serum or serum diluted to 1/4. As seen in the figures, the bacteraemic profile of dead animals generally consisted of a persistent bacteraemia between 10⁶ and 10⁸ cfu/ml from 24 to 120 h. All dead animals had at least one blood sample with colony counts ≥10⁷ cfu/ml in the period 24 to 48 h, and colony counts ≥10⁸ cfu/ml at death. No overgrowth was observed in the post-mortem bacteraemic profile of the five animals where this parameter was investigated, with colony counts ranging from 2·48 × 10⁷ to 3·93 × 10⁸ cfu/ml over the 72 h sampling time (two animals died at 24 h, two animals at 36 h and one animal at 48 h). The bacteraemic profile of the animals that survived after the serum protection generally consisted of a decreasing bacteraemia <10⁶ cfu/ml over the sampling period. Figure 5 shows the relation of the maximum log₁₀ cfu/ml obtained over the sampling interval and the probability of death. Maximum bacterial counts <10⁷ cfu/ml obtained over the sampling interval were related to <10% probability of death, whereas a 50% probability was related to colony counts of 3·5 × 10⁷ cfu/ml, and colony counts ≥5 × 10⁷ cfu/ml were related to ≥65% probability of death, irrespective of the administration of hyperimmune serum or of the dilution administered.

Fig. 3.

Individual bacteraemic profile over 144 h of dead mice after protective administration of non-diluted (*) or 1/4 diluted (•) specific hyperimmune serum. Times of death not shown in the figure were 336 h (non-diluted serum) and 360 h (serum 1/4).

Fig. 4.

Individual bacteraemic profile over 144 h of surviving mice after protective administration of non-diluted (*) or 1/4 diluted (•) specific hyperimmune serum.

Fig. 5.

Relationship between maximum log₁₀ cfu/ml and probability of death, irrespective of the administration or not of protective serum or the dilution administered.

DISCUSSION

This study attempts to explore the biological features that may be modified by the administration of specific immunoprotection: mortality decrease, mortality delay, and their relation with blood bacteraemic profile. The difference in virulence of pneumococcal strains in mice is dependent on the specific capsular types of the pneumococci. Although a previously published study reported that only 60% of serotype 6 isolates were virulent for mice [7], a penicillin-resistant serotype 6B was used because of its frequency in bacteraemic infections [2,6], poor immunogenicity [2], high prevalence, and the fact that it is representative of the resistance pattern in Spain [6,8]. Inoculation was performed intraperitoneally because, although previous studies have indicated that pneumococci can be more virulent by the intraperitoneal route than by the intravenous route, suspension of bacteria in saline (as in this study) has revealed similar virulence by both routes of infection [7].

The effect of administering protective serum is assessed by two end-points commonly used for efficacy evaluation in animal models, bacterial counts in tissue fluids and survival rates [9], since the reduction of bacterial counts could not be related to animal survival [10]. The observed effect could be entirely attributed to the hyperimmune nature of the serum, since experiments in which control animals received non-immune serum showed the same results as in those where control animals received PBS as placebo. These results contrast with those obtained in a previous study, in which administration of normal human serum that lacked anti-capsular or anti-cell-wall polysaccharide antibodies had some protective effect in mice [11]. The reason for the difference could be the presence of human complement in that particular study (but not in our experiments), which was responsible for the modest protective effect since mice are complement poor [11].

Three main effects are produced by hyperimmune serum (diluted to 1/4 or non-diluted): a significant decrease in mortality, a delay in mortality (when it occurs), and a variation of the bacteraemic profile. A decrease in mortality had been previously demonstrated by parenteral administration of human immunoglobulins in a pneumonia model at short term (48 h) [12], and in a protection model at day 10 [13] with other serotypes. In this study, the administration of specific hyperimmune serum produced a mortality decrease with both serum diluted to 1/4 and non-diluted specific antiserum, but not with dilution to 1/16. Other studies investigating passive immunization against S. pneumoniae have demonstrated that an anti-capsular antibody protective threshold exists [14,15], which could explain the lack of protection when the hyperimmune serum was diluted to 1/16 in this study.

With respect to mortality delay, we used a 15 day follow-up period to clearly discriminate mortality decrease from mortality delay. If a short-term follow-up period (i.e less than 144 h) had been used, the mortality delay produced by administration of dilution to 1/16 would have been regarded as a decrease in mortality compared with the control. This suggests that in this kind of study, long-term follow-up is needed. In this study, a significant mortality delay over the first 144 h was found with specific non-diluted, 1/4- and 1/16-diluted sera compared with the control. This delay in mortality was dilution-dependent, since the delay was greater for 1/4 than for 1/16, while the non-diluted serum showed the longest delay.

In relation to the effect on the bacteraemic profile, two facts could be described: (a) the decrease in the bacteraemic curve by administration of hyperimmune serum; and (b) the relation between bacteraemia and mortality. With regard to the first fact, a decrease in the bacteraemic curve occurred when serum diluted to 1/4 or non-diluted serum was administered, irrespective of whether animals survived or not; the bacteraemic profile of persistent colony counts ≥10⁸ cfu/ml in control animals versus counts of 10⁶–10⁸ cfu/ml in serum-administered animals that died and ≤10⁶ cfu/ml in serum-administered animals that survived. With regard to the relation between bacteraemia and mortality, bacteraemia with persistent colony counts ≥10⁶ cfu/ml, and/or with colony counts ≥10⁶ cfu/ml from 24 to 48 h, were related to death and, from a statistical perspective, maximum colony counts ≥5 × 10⁷ cfu/ml were related to a ≥65% probability of death, irrespective of the administration of hyperimmune serum or of the dilution administered.

In conclusion, in this study, protection with specific hyperimmune serum caused a change in the bacteraemic profile; this variation implied a decrease in mortality over time, thus decreasing the bacterial virulence (i.e. mortality and decrease in blood colony counts) of a serotype 6B penicillin-resistant S.pneumoniae quantitatively.