Research ArticleInfectious Disease

IVIG-mediated protection against necrotizing pneumonia caused by MRSA

See allHide authors and affiliations

Science Translational Medicine  21 Sep 2016:
Vol. 8, Issue 357, pp. 357ra124
DOI: 10.1126/scitranslmed.aag1153

Stemming the tide of MRSA-induced pneumonia

Controversies persist about the use of human intravenous immunoglobulin (IVIG) as an adjunctive treatment for severe Staphylococcus aureus pneumonia because empirical evidence supporting its use is lacking. Diep and colleagues now show that, of the myriad antibodies contained in IVIG, only two specific antibodies that neutralize the toxic effects of α-hemolysin (Hla) and Panton-Valentine leukocidin (PVL) are necessary and sufficient to confer protection against necrotizing pneumonia caused by MRSA in a rabbit model. Preexposure prophylaxis with IVIG, or postexposure treatment with IVIG in combination with either vancomycin or linezolid, improved survival outcomes in this preclinical animal model.

Abstract

New therapeutic approaches are urgently needed to improve survival outcomes for patients with necrotizing pneumonia caused by Staphylococcus aureus. One such approach is adjunctive treatment with intravenous immunoglobulin (IVIG), but clinical practice guidelines offer conflicting recommendations. In a preclinical rabbit model, prophylaxis with IVIG conferred protection against necrotizing pneumonia caused by five different epidemic strains of community-associated methicillin-resistant S. aureus (MRSA) as well as a widespread strain of hospital-associated MRSA. Treatment with IVIG, either alone or in combination with vancomycin or linezolid, improved survival outcomes in this rabbit model. Two specific IVIG antibodies that neutralized the toxic effects of α-hemolysin (Hla) and Panton-Valentine leukocidin (PVL) conferred protection against necrotizing pneumonia in the rabbit model. This mechanism of action of IVIG was uncovered by analyzing loss-of-function mutant bacterial strains containing deletions in 17 genes encoding staphylococcal exotoxins, which revealed only Hla and PVL as having an impact on necrotizing pneumonia. These results demonstrate the potential clinical utility of IVIG in the treatment of severe pneumonia induced by S. aureus.

INTRODUCTION

Community-associated pneumonia caused by methicillin-resistant Staphylococcus aureus (MRSA) has been reported worldwide (1). The best treatment for this potentially life-threatening infection has not been clearly defined. Controversies persist among national clinical guidelines on the treatment of severe community-associated MRSA pneumonia. Intravenous immunoglobulin (IVIG), which contains pooled human polyclonal antibodies that neutralize lung-damaging toxins produced by MRSA (2), is recommended as adjunctive therapy for treatment of severe community-associated MRSA pneumonia by the U.K. and Canadian clinical practice guidelines (3, 4). In contrast, IVIG is not routinely recommended by the Infectious Diseases Society of America (IDSA) clinical practice guidelines because empirical evidence supporting its use is lacking (5).

Current use of IVIG for the treatment of community-associated MRSA pneumonia is supported only by in vitro data showing that it contains antitoxin antibodies (2) and by anecdotal case reports (69). Randomized clinical trials to test the efficacy of IVIG as adjunctive therapy may be impractical because community-associated MRSA necrotizing pneumonia occurs sporadically and is rapidly fatal; moreover, it may be unethical to withhold a potentially life-saving intervention for a severe infection with a mortality rate of 30 to 75% (69). However, it should be noted that protective effects of IVIG as adjunctive therapy in patients with bacterial sepsis or septic shock have been conflicting, with some clinical trials showing significant reductions in mortality in adults with sepsis when treated with IVIG compared to placebo or no intervention, but this benefit was not observed in trials with low risk of bias or in neonates (10). Direct evidence that IVIG protects against severe pneumonia caused by community- or hospital-associated MRSA strains is needed.

Here, we conducted a series of experiments in a rabbit model of S. aureus–induced acute, necrotizing pneumonia to determine whether treatment with IVIG reduced mortality. This model mimics the rapidly fatal course of infection that occurs in humans, including severe hypoxemia, leukopenia, lung necrosis, pulmonary edema, alveolar hemorrhage, hemoptysis, and death (11). Rabbits are similar to humans in their exquisite sensitivity to two key lung-damaging staphylococcal toxins, α-hemolysin (Hla) and Panton-Valentine leukocidin (PVL), whereas other species, including mouse, rat, and monkey, are resistant to the leukolytic effects of PVL (1113). We found that IVIG improved survival of rabbits infected by each of the major epidemic clones of Hla- and PVL-producing community-associated MRSA, including the North American clone USA300, U.S. Midwest clone USA400, Asian/Pacific clone USA1000, Southwest Pacific clone USA1100, and European/North African clone ST80. IVIG also improved survival of rabbits infected with the Hla-producing (but not PVL-producing) hospital-associated MRSA clone USA100 (14). Hla and PVL were shown to be critical for the pathogenesis of necrotizing pneumonia, and small amounts of anti-Hla– and anti-PVL–neutralizing antibodies in IVIG were necessary and sufficient for its protective efficacy in rabbits.

RESULTS

IVIG-mediated protection against lethal necrotizing pneumonia in a rabbit model

The optimal dosage of IVIG for the treatment of community-associated MRSA pneumonia is not clearly defined, although a high-dose IVIG of 2 g/kg is thought to be needed to achieve protective titers because staphylococcal toxins are not neutralized as efficiently as streptococcal toxins by IVIG (15). Here, we show that a commercial preparation of human IVIG, ClairYg, contained antibodies that neutralized the cytotoxic effects of PVL against human polymorphonuclear leukocytes (PMNs) and Hla against rabbit red blood cells (Fig. 1, A to E). Two different lots of ClairYg exhibited similar neutralizing activities against PVL (Fig. 1, B and C) and Hla (Fig. 1E).

Fig. 1. Human IVIG contains neutralizing antibodies to PVL and Hla.

(A) Kinetics of pore formation in human PMNs determined by uptake of propidium iodide (PI) in response to the S. aureus toxins LukS-PV and LukF-PV (PVL) at biologically relevant concentrations (see also Fig. 5G). (B and C) Two different lots of IVIG {ClairYg lot nos. 11L00258 [IVIG(L8)] and 11L00443 [IVIG(L3)]} blocked PVL-induced pore formation in human PMNs. (D) Kinetics of neutralization of PVL-induced membrane pore formation in response to human IVIG, which was added at the indicated times after PVL was mixed with human PMNs. (E) Two different lots of IVIG exhibited similar inhibition of rabbit red blood cell lysis in response to varying concentrations of Hla. Results in (A) to (D) are representative of two independent experiments with three human blood donors, and those in (E) are representative of three independent experiments with a single batch of rabbit erythrocytes.

A pharmacokinetic study in rabbits further showed that intravenous administration of 1/10 of the recommended human dose, 200 mg/kg, of IVIG achieved serum concentrations of 3.4 ± 1.0 mg/ml at 2 hours and 1.8 ± 0.1 mg/ml at 48 hours after dosing. These concentrations were three- to sixfold higher than 0.5 mg/ml of IVIG that was sufficient to neutralize the cytotoxic effects of PVL and Hla in vitro (Fig. 2, A to C).

Fig. 2. Human IVIG, either alone or in combination with vancomycin or linezolid, confers protection in a rabbit model of necrotizing pneumonia.

(A) Total human IgG, (B) anti-PVL human IgG, and (C) anti-Hla human IgG in the serum of rabbits taken at the indicated time after intravenous administration of human IVIG (200 mg/kg). Error bars indicate SEM. AU, arbitrary units. (D) Kaplan-Meier survival curves, (E) LW/BW (×103) ratio, and (F) bacterial densities, log10 (CFU per lung) for animals treated intravenously with IVIG (200 mg/kg) once at 1.5 hpi (n = 14 rabbits); vancomycin (30 mg/kg) four times at 1.5, 13, 25, and 37 hpi (n = 15 rabbits); IVIG + vancomycin (n = 13 rabbits); or saline at 1.5 hpi (n = 14 rabbits) with 5.4 × 109 CFU of the SF8300 WT strain. One-sided log-rank (Mantel-Cox) test was used to test hypothesis that survival of animals treated with saline is shorter than survival of those treated with IVIG, vancomycin, or the combination of IVIG + vancomycin, as well as the hypothesis that survival of animals treated with IVIG alone or with vancomycin alone is shorter than those treated with the combination of IVIG + vancomycin, with P < 0.010 (significance level of 0.05 divided by five different comparisons) being considered statistically significant to account for multiple comparisons using Bonferroni method. (G) Kaplan-Meier survival curves, (H) LW/BW (×103) ratio, and (I) bacterial densities, log10 (CFU/lung) for animals treated with IVIG(L3) (200 mg/kg) intravenously once at 1.5 hpi (n = 15 rabbits), linezolid (50 mg/kg) subcutaneously four times at 1.5, 10, 18, and 26 hpi (n = 15 rabbits), IVIG + linezolid (n = 15 rabbits), or IVIG (200 mg/kg) depleted of anti-Hla IgG and anti-LukS/LukF IgG intravenously once at 1.5 hpi (n = 15 rabbits) with 5.3 × 109 CFU of the SF8300 WT strain. One-sided log-rank (Mantel-Cox) test was used to test the hypothesis that survival of animals treated with saline is shorter than survival of those treated with IVIG, linezolid, or the combination of IVIG + linezolid, as well as the hypothesis that survival of animals treated with IVIG alone or with linezolid alone is shorter than those treated with the combination of IVIG + linezolid, with P < 0.010 (significance level of 0.05 divided by five different comparisons) being considered statistically significant to account for multiple comparisons using Bonferroni method. (E and H) LW/BW (×103) ratio and (F and I) bacterial densities, log10 (CFU per lung), for saline-treated animals were compared to those of each of the other three treatment groups by nonparametric one-way analysis of variance (ANOVA) with Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Filled symbols represent data from dead animals, and open symbols represent data from surviving animals that were euthanized at 48 hpi (E and F) or 96 hpi (H and I).

To evaluate the protective efficacy of IVIG, either alone or in combination with vancomycin, which is recommended in clinical practice guidelines as a first-line agent for the treatment of MRSA pneumonia (35), rabbits were randomized for treatment with (i) saline; (ii) vancomycin (30 mg/kg, twice daily), a dosing regimen that yields peak serum concentrations of 36.1 ± 4.2 μg/ml at 1 hour after dosing (16); (iii) IVIG (200 mg/kg) alone; or (iv) an IVIG + vancomycin combination, which was administered at 1.5 hours postinfection (hpi) with the USA300/SF8300 epidemic clone of community-associated MRSA in the rabbit model of necrotizing pneumonia (11). Overall survival rates were 7% for animals treated with saline, compared to 7% for those treated with vancomycin alone (P = 0.23 versus saline by one-sided log-rank test), 50% for those treated with IVIG alone (P = 0.010 versus saline), and 71% for those treated with the IVIG-vancomycin combination (P < 0.001 versus saline) (Fig. 2D). Animals treated with the IVIG-vancomycin combination showed greater survival than those treated with IVIG alone (P = 0.14) or vancomycin alone (P < 0.001). All infected animals had severe pulmonary edema, as evidenced by lung wet weight–to–body weight (LW/BW) ratios exceeding the normal range of 4 to 5 for uninfected lungs, although there were no between-group differences (Fig. 2E). A significant reduction in bacterial count in the lungs was observed in animals treated with the IVIG-vancomycin combination but not IVIG alone or vancomycin alone (Fig. 2F).

Linezolid, a protein synthesis inhibitor recommended in clinical practice guidelines as an acceptable alternative to vancomycin for treatment of MRSA pneumonia (35), has been shown to be superior to vancomycin for the treatment of necrotizing pneumonia in the rabbit model (16). Here, we evaluated the protective efficacy of IVIG alone or in combination with linezolid. Rabbits were randomized for treatment with (i) a negative control human IVIG antibody preparation depleted of anti-PVL and anti-Hla antibodies (200 mg/kg; IVIG-depleted); (ii) linezolid (50 mg/kg, three times daily), a dosing regimen that yielded a peak serum concentration of 10.5 ± 2.3 μg/ml at 1 hour after dosing (16); (iii) IVIG (200 mg/kg) alone; or (iv) an IVIG + linezolid combination, which was administered at 1.5 hpi with the USA300/SF8300 MRSA strain (11). Overall survival rates were 0% for animals treated with IVIG-depleted, compared to 47% for those treated with IVIG alone (P < 0.001 versus IVIG-depleted), 67% for those treated with linezolid (P < 0.001 versus IVIG-depleted), and 93% for those treated with the IVIG-linezolid combination (P < 0.001 versus IVIG-depleted) (Fig. 2G). Animals treated with the IVIG-linezolid combination showed greater survival than those treated with IVIG alone (P = 0.003) or linezolid alone (P = 0.037 by one-sided log-rank test), although the latter was not statistically significant because the Bonferroni-corrected significance threshold after accounting for multiple comparisons was 0.01 (Fig. 2G). The various treatment modalities did not have any effect on LW/BW ratios (Fig. 2H), suggesting that IVIG and/or linezolid did not reduce acute lung injury when compared to LW/BW ratios for the IVIG-depleted control, although they prevented respiratory dysfunction and death to at least 96 hpi (Fig. 2G). Consistent with our previous findings in the same rabbit model that linezolid reduced bacterial counts in lungs of infected animals (16), linezolid, when administered either alone or in combination with IVIG at 1.5 hpi, also significantly reduced log10 [colony-forming unit (CFU) per lung] (P < 0.001; Fig. 2I).

To further test whether prophylaxis with IVIG protected against necrotizing pneumonia, IVIG (200 mg/kg) was administered 24 hours before challenge with not only USA300/SF8300 but also four other community-associated MRSA strains: USA400/MW2, USA1000, USA1100, and ST80 (Fig. 3). Compared with saline control, prophylaxis with IVIG resulted in a statistically significant mortality reduction of 63% for challenge with USA300/SF8300 (P = 0.022 by one-sided log-rank test; Fig. 3, A to C), 75% for USA400/MW2 (P = 0.003; Fig. 3, D to F), 50% for USA1000 (P = 0.013; Fig. 3, G to I), 75% for USA1100 (P = 0.001; Fig. 3, J to L), and 63% for strain ST80 (P < 0.001; Fig. 3, M to O). Preexposure prophylaxis or postexposure treatment with IVIG [Figs. 2 (F and I) and 3 (C, F, I, and O)] did not influence bacterial count in the lungs of animals challenged with USA300/SF8300, USA400, USA1000, and ST80 (the exception being that IVIG resulted in a reduction in bacterial count in the lungs of animals challenged with USA1100; Fig. 3L). These data suggested that the mechanism of action of IVIG-mediated protection was not due to enhanced opsonophagocytic killing.

Fig. 3. Prophylaxis with IVIG protects against lethal necrotizing pneumonia caused by five clinical community-associated MRSA strains.

Eight rabbits per experimental group were randomized to receive either saline or IVIG(L8)/IVIG(L3) (200 mg/kg) at 24 hours before infection (hbi) with 5.8 × 109 CFU of USA300/SF8300 (A to C), 5.7 × 109 CFU of USA400 (D to F), 1.5 × 109 CFU of USA1000 (G to I), 6.3 × 109 CFU of USA1100 (J to L), and 6.8 × 109 CFU of ST80 (M to O). One rabbit randomized for prophylaxis with IVIG(L3) had anesthesia-related death before challenge with the ST80 strain (M to O). Kaplan-Meier survival curves were compared with one-sided log-rank (Mantel-Cox) test to evaluate the hypothesis that survival of animals pretreated with saline is shorter than survival of those pretreated with IVIG, with P < 0.05 being considered statistically significant. LW/BW ratio and log10 (CFU per lung) for saline-pretreated animals were compared to those pretreated with IVIG by nonparametric Mann-Whitney U test. Filled symbols represent data from dead animals, and open symbols represent data from surviving animals that were euthanized at 48 hpi.

Characterization of the relative contributions of toxins in the pathogenesis of community-associated MRSA pneumonia

In clinical settings, IVIG (1000 to 2000 mg/kg) is typically used because this high-dose regimen is needed to exert an anti-inflammatory response in the patient. Because even a low-dose regimen of IVIG (200 mg/kg) was protective in the rabbit pneumonia model, it seemed unlikely that IVIG’s protective effect involved dampening of acute lung inflammation. Instead, on the basis of our data (Fig. 1) and other in vitro studies (2, 15, 17), the mechanism of IVIG-mediated protection seemed to be due to the presence of specific antibodies in IVIG that neutralized some of the toxins produced by S. aureus. One approach to determine the mechanism of protection would be to affinity-purify the different antitoxin antibodies from IVIG and evaluate individual protective effects of each specific antibody preparation in the rabbit model. However, this approach would be cost prohibitive because S. aureus produces many toxins. Thus, we took a genetic approach to systematically characterize those toxins that have been implicated in S. aureus necrotizing pneumonia in our rabbit model to identify those that were most potent. Then, we evaluated whether affinity-purified antibodies that neutralized those specific toxins were protective in our rabbit model.

In-frame deletions of 17 different genes at eight different chromosomal loci that encode staphylococcal exotoxins were constructed in a USA300/SF8300 clinical strain (tables S1 and S2). Although these toxin genes have demonstrated roles in disease pathogenesis in one or more animal models, only the PVL mutant has been shown previously to be severely attenuated in the New Zealand white rabbit model of necrotizing pneumonia (11). In this rabbit model, deletion of the gene encoding PVL (ΔlukSF) or Δhla attenuated the capacity of the mutant bacteria to cause acute lung injury at 9 hpi, as evidenced by significantly decreased LW/BW ratios (P < 0.001; Fig. 4A). In contrast, deletion of genes encoding phenol-soluble modulin α types 1 to 4 (Δpsm-α), γ-hemolysin (ΔhlgABC), leukocidin E and D (ΔlukED), leukocidin G and H (ΔlukGH) (also known as leukocidin A and B), staphylococcal enterotoxin Q and K (Δseq/sek), and staphylococcal enterotoxin-like X (ΔselX) did not result in significant reduction in LW/BW ratios (Fig. 4, A to C). Bacterial counts in lungs of animals at 9 hpi with the SF8300 wild-type (WT) strain did not differ from those infected with any of the other mutant strains (Fig. 4, D to F). Lungs of animals infected with Δhla and ΔlukSF mutants were grossly different from those infected with the SF8300 WT parental strain, with extensive areas of necrosis and severe pulmonary edema (Fig. 4K). Compared to the SF8300 WT parental strain, the Δpsm-α, Δhla, and ΔlukSF mutants provoked a reduction in interleukin-8 (IL-8) in rabbit lungs (Fig. 4G), which resulted in a concomitant reduction in leakage of IL-8 into the blood (Fig. 4H). In contrast, only the ΔlukSF mutant, but not Δpsm-α and Δhla mutants, provoked a reduction in monocyte chemotactic protein 1 (MCP-1) (Fig. 4, I and J). Reduced IL-8 in rabbit lungs and blood has been shown previously to result in a reduced influx of neutrophils, which are known to play a central role in mediating acute lung injury and inflammation in the rabbit pneumonia model (11).

Fig. 4. Effects of major staphylococcal exotoxins in acute lung injury.

Comparison of (A to C) LW/BW (×103) and (D to F) log10 CFU per lung for rabbits (n = 9 animals per experimental group) euthanized at 9 hpi with SF8300 WT or isogenic mutants containing deletion of genes encoding PVL (ΔlukSF), Δpsm-α, Δhla, ΔhlgABC, ΔlukED, ΔlukGH (also known as leukocidin A and B), Δseq/sek, and Δselx. (G to J) Concentrations of IL-8 and MCP-1 in the lungs and plasma of rabbits euthanized at 9 hpi with SF8300 WT, or Δpsm-α, Δhla, and ΔlukSF mutant strains. Nonparametric one-way ANOVA with Kruskal-Wallis test followed by Dunn’s multiple comparisons test were used to evaluate differences between WT and each of the mutant strains. (K) Photographs depict gross pathology of representative lungs harvested from rabbits at 9 hpi with the four isogenic SF8300 strains: WT, Δpsm-α, Δhla, and ΔlukSF mutant strains.

Consistent with their reduced capacity to cause acute lung injury, the Δhla mutant, the ΔlukSF mutant, as well as the ΔhlaΔlukSF double mutant caused significantly lower mortality rates compared to the SF8300 WT parental strain (P = 0.007 and P < 0.001; Fig. 5A). The Δpsm-α mutant also exhibited reduced capacity to cause lethal infection compared to SF8300 WT, although P = 0.024 by one-sided log-rank test is considered not statistically significant using the Bonferroni-corrected significance threshold of 0.0125 to account for multiple comparisons. The attenuated virulence of the ΔlukSF and ΔhlaΔlukSF was evident also in the significantly reduced LW/BW ratio and bacterial counts in lungs (all significant, P < 0.01; Fig. 5, B and C).

Fig. 5. Staphylococcal necrotizing pneumonia is mediated principally by two toxins, PVL and Hla.

Comparison of Kaplan-Meier survival curves (A), LW/BW ratio (×103) (B), and log10 CFU per lung (C) for rabbits challenged with USA300/SF8300 WT strain, or isogenic mutant strains Δpsm-α, Δhla, ΔlukSF, or the double mutant strain ΔhlaΔlukSF (n = 12 animals per experimental group). One-sided log-rank (Mantel-Cox) test was used to test the hypothesis that survival of animals challenged with SF8300 WT is shorter than survival of those challenged with each of the four mutant strains Δpsm-α, Δhla, ΔlukSF, or ΔhlaΔlukSF with P < 0.0125 (significance level of 0.05 divided by four different comparisons) being considered statistically significant to account for multiple comparisons using the Bonferroni method. Comparison of Kaplan-Meier survival curves (D), LW/BW ratio (×103) (E), and log10 CFU per lung (F) of rabbits challenged with WT strains belonging to the epidemic community-associated MRSA strains USA300/SF8300 (ST8), USA400/MW2 (ST1), USA1000 (ST59), USA1100 (ST30), and ST80 (n = 7 animals per experimental group). Amounts of LukS-PV (G) and Hla (H) produced in the lungs of infected animals. Two-sided log-rank (Mantel-Cox) test was used to test the hypothesis that survival of animals challenged with SF8300 WT is not different from survival of those challenged with each of the other four community-associated MRSA strains, with P < 0.0125 (significance level of 0.05 divided by four different comparisons) being considered statistically significant to account for multiple comparisons using the Bonferroni method. The following log-rank test P values are for comparison of SF8300 WT versus each of the other community-associated MRSA strains: P = 0.20 for SF8300 versus USA400, P = 0.19 for SF8300 versus USA1000, P = 0.46 for SF8300 versus USA1100, and P = 0.006 for SF8300 versus ST80. For both studies, the LW/BW ratio (×103), bacterial densities in lungs, and toxin concentrations in lungs for animals challenged with SF8300 were compared to each of the four other strains by nonparametric one-way ANOVA with Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Filled symbols represent data from dead animals, and open symbols represent data from surviving animals that were euthanized at 36 hpi.

Because SElX was previously shown to be important in an American Dutch belted rabbit pneumonia model (18) using the USA300/LAC strain, we constructed an in-frame deletion of the selx gene in this strain and then compared survival of animals infected with the isogenic strains in our New Zealand white rabbit model of necrotizing pneumonia. No difference in mortality was observed between the LAC WT and LACΔselx mutant (fig. S1, A to D), which is consistent with the lack of any differences in acute lung injury at 9 hpi with SF8300 WT and SF8300Δselx (Fig. 4C) in our rabbit model. The discrepancies between the null result in our New Zealand white rabbit model and the positive result in the American Dutch belted rabbit model (18) could be due to the use of different rabbit strains, an explanation that may be supported by the fact that more recent studies by the same group did not show any differences between USA400/MW2 WT and the MW2Δselx mutant in a New Zealand white rabbit model of infective endocarditis (19).

Inasmuch as the major clones of community-associated MRSA are known to up-regulate production of Hla and PVL (20), we compared here their pathogenic capacity in the rabbit model and determined the amounts of toxin produced in vivo. USA300, USA1000, USA1100, USA400, and ST80 strains were relatively similar in their capacity to cause rapidly lethal necrotizing pneumonia, with 34 of 35 rabbits challenged with these five strains succumbing to infection (Fig. 5, D to F). The ST80 strain caused more rapidly lethal infection than the USA300/SF8300 strain (P = 0.006 by two-sided log-rank test), although both strains caused lethality in 100% of challenged animals (Fig. 5D). These community-associated MRSA clones produced PVL and Hla above the enzyme-linked immunosorbent assay (ELISA) detection limit in 94 and 69% of the 35 infected lungs, respectively (Fig. 5, G and H). PVL production in the lungs varied among the different community-associated MRSA strains, from a low concentration of 0.53 μg per lung by USA1000 to a high concentration of 5.07 μg per lung by USA1100 (Fig. 5G). Hla production in the lungs also exhibited strain-to-strain variability, from a low concentration of 0.28 μg per lung by USA400 to a high concentration of 3.24 μg per lung by USA1000 (Fig. 5H). Concentrations of these toxins produced in the lungs of many of the animals exceeded the toxic concentrations of PVL and Hla produced against mammalian cell targets in vitro (see Fig. 1, A and E).

Mechanisms of protection by IVIG

Having shown that Hla and PVL are the principal toxins mediating lethal necrotizing pneumonia in the rabbit model, we next tested whether the protective efficacy of IVIG was due to specific antibodies that neutralized these toxins. Recombinant Hla and PVL subunits, LukS-PV and LukF-PV, were used in affinity column chromatography to selectively capture toxin-specific antibodies. Affinity-purified anti–LukS-PV and anti–LukF-PV antibodies neutralized PVL-induced pore formation in neutrophils (fig. S2, A and B), whereas the anti-Hla antibodies neutralized Hla-induced lysis of red blood cells (fig. S2C). Because IVIG also has anti-inflammatory activities that could confer protection independent of specific antitoxin antibodies, we controlled for this possibility by preparing IVIG depleted of its anti–LukS-PV, anti–LukF-PV, and anti-Hla antibodies through negative selection on affinity column chromatography. The IVIG-depleted preparation was deficient in its toxin-neutralizing activity against PVL-induced pore formation in neutrophils and Hla-induced red blood cell lysis (fig. S2, A to C).

Prophylaxis with affinity-purified anti-LukS/F-PV antibodies (1 mg/kg), anti-Hla antibodies (1 mg/kg), or IVIG (200 mg/kg) demonstrated a similar protective efficacy in reducing mortality after challenge with the USA300/SF8300 WT strain compared to saline control (Fig. 6, A to C). Using a Bonferroni-corrected significance threshold of 0.0125 to account for multiple comparisons, P = 0.010 for saline versus IVIG and P = 0.006 for saline versus anti-Hla immunoglobulin G (IgG), by one-sided log-rank test, were considered statistically significant, whereas P = 0.014 for saline versus anti-LukF/S IgG was not statistically significant (Fig. 6A). Administration of depleted IVIG (200 mg/kg) did not confer significant protection compared to the saline-treated animals (P = 0.12; Fig. 6A). Sera from animals treated with IVIG exhibited similar neutralization activities to sera from animals treated with specific antitoxin antibodies in the PVL-induced human neutrophil pore formation assay and the Hla-induced rabbit red blood cell hemolysis assay (fig. S2, D to F).

Fig. 6. Anti-Hla and anti-PVL antibodies affinity purified from human IVIG protect against USA300 community-associated MRSA necrotizing pneumonia.

Comparison of Kaplan-Meier survival curves (A), LW/BW ratio (×103) (B), and log10 CFU per lung (C) of rabbits pretreated at 1.5 hours before infection with 6.4 × 109 CFU of SF8300 WT strain with the following regimens: saline (n = 10 rabbits), IVIG(L8) (200 mg/kg; n = 12 rabbits), IVIG(L8) (200 mg/kg) depleted of anti-Hla IgG and anti-LukS/LukF IgG (n = 9 rabbits), anti-Hla IgG (1 mg/kg) affinity-purified from IVIG (n = 8 rabbits), anti-LukS IgG (1 mg/kg), and anti-LukF IgG (1 mg/kg) affinity-purified from IVIG (n = 8 rabbits). One-sided log-rank (Mantel-Cox) test was used to test the hypothesis that survival of animals pretreated with saline is shorter than survival of those pretreated with each of four antibody preparations, with P < 0.0125 (significance level of 0.05 divided by four different comparisons) being considered statistically significant to account for multiple comparisons using the Bonferroni method. LW/BW ratio (×103) and log10 CFU per lung for animals challenged with SF8300 were compared to each of the four other strains by nonparametric one-way ANOVA with Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Comparison of Kaplan-Meier survival curves (D), LW/BW ratio (×103) (E), and log10 CFU per lung (F) of rabbits pretreated at 1.5 hours before infection with 5.1 × 109 CFU of SF8300 WT with depleted IVIG(L3) (200 mg/kg) or IVIG(L3) (200 mg/kg) (eight rabbits per experimental group). Comparison of Kaplan-Meier survival curves (G), LW/BW ratio (×103) (H), and log10 CFU per lung (I) of rabbits pretreated at 1.5 hours before infection with 5.3 × 109 CFU of SF8300 WT with depleted IVIG(L3) (200 mg/kg) or anti-Hla IgG (4 mg/kg) (eight rabbits per experimental group). Comparison of Kaplan-Meier survival curves (J), LW/BW ratio (×103) (K), and log10 CFU per lung (L) of rabbits pretreated at 1.5 hours before infection with 5.4 × 109 CFU of SF8300 WT with depleted IVIG(L3) (200 mg/kg) or anti-LukS/LukF IgG (4 mg/kg) (eight rabbits per experimental group). One-sided log-rank (Mantel-Cox) test was used to test the hypothesis that survival of animals pretreated with depleted IVIG is shorter than survival of those pretreated with the other antibody preparations, with P < 0.05 being considered statistically significant. Filled symbols represent data from dead animals, and open symbols represent data from surviving animals that were euthanized at 48 hpi. LW/BW ratio (×103) and bacterial densities in rabbit lung for the different pairwise comparisons were evaluated with the nonparametric Mann-Whitney U test.

To further confirm the mechanisms of IVIG-mediated protection against lethal pneumonia, three additional independent studies were conducted. Prophylaxis with IVIG (200 mg/kg) resulted in 63% reduction in mortality when compared to depleted IVIG (200 mg/kg) (P = 0.002 by one-sided log-rank test; Fig. 6, D to F). Similarly, prophylaxis with anti-Hla antibodies (4 mg/kg) resulted in 88% reduction in mortality (P < 0.001 by one-sided log-rank test; Fig. 6, G to I) or anti-LukS/F antibodies (4 mg/kg) (P < 0.001 by one-sided log-rank test; Fig. 6, J to L) when compared to depleted IVIG (200 mg/kg).

The generalizability of these results on the mechanisms of action of IVIG was further strengthened by comparing protective efficacies of IVIG-depleted and IVIG using two additional clinical strains, the lesser community-associated MRSA strain USA400/MW2 found in the Midwest of the United States (14) and Alaska (21) and the predominant multidrug-resistant hospital-associated MRSA strain USA100/NRS382 found in health care facilities throughout the United States (14). When compared to IVIG-depleted (200 mg/kg), prophylaxis with IVIG (200 mg/kg) resulted in 100% reduction in mortality with the USA400 strain (P < 0.001 by one-sided log-rank test; Fig. 7, A to C) and 50% reduction in mortality with the USA100 strain (P = 0.022; Fig. 7, D and E). Together, these data indicate that IVIG may protect against lethal infection through its specific antitoxin antibodies.

Fig. 7. Human IVIG protects against pneumonia caused by community-associated MRSA strain USA400 and hospital-associated MRSA strain USA100.

Comparison of Kaplan-Meier survival curves (A), LW/BW ratio (×103) (B), and log10 CFU per lung (C) of rabbits pretreated at 1.5 hours before infection with 5.5 × 109 CFU of USA400 WT strain with IVIG(L3) (200 mg/kg) depleted of anti-Hla IgG and anti-LukS/LukF IgG or complete IVIG(L3) (200 mg/kg) (eight rabbits per experimental group). Comparison of Kaplan-Meier survival curves (D), LW/BW ratio (×103) (E), and log10 CFU per lung (F) of rabbits pretreated at 1.5 hours before infection with 5.6 × 109 CFU of USA100/NRS382 WT strain with depleted IVIG(L3) (200 mg/kg) or IVIG(L3) (200 mg/kg) (eight rabbits per experimental group). One-sided log-rank (Mantel-Cox) test was used to test the hypothesis that survival of animals pretreated with depleted IVIG is shorter than survival of those pretreated with the other antibody preparations, with P < 0.05 being considered statistically significant. Filled symbols represent data from dead animals, and open symbols represent data from surviving animals that were euthanized at 96 hpi. LW/BW ratio (×103) and bacterial densities in rabbit lung for the different pairwise comparisons were evaluated with the nonparametric Mann-Whitney U test.

DISCUSSION

Necrotizing pneumonia caused by community-associated MRSA strains is associated clinically with a rapid and high mortality rate, despite appropriate antibiotic treatment (1). There is an urgent need for new therapeutic approaches that would supplement current antibiotic therapy to improve survival during the acute phase of the infection. A number of case reports have documented the use of IVIG as an effective adjunct to antimicrobial therapy to treat severe cases of staphylococcal pneumonia (69). By demonstrating that IVIG administration protects against death in a rabbit model of necrotizing pneumonia using five different community-associated MRSA clinical strains and one predominant hospital-associated MRSA clinical strain and by unraveling the mechanism of this protective effect, our study provides preclinical evidence that IVIG may have a role to play in the treatment of severe MRSA pneumonia.

Several hypotheses regarding the mechanism of action of IVIG have been proposed. Our working hypothesis was that IVIG protects against MRSA necrotizing pneumonia by neutralizing S. aureus toxins that are involved in disease pathogenesis. It was based on previous observations that IVIG contains neutralizing antibodies against several S. aureus virulence factors, such as superantigens (22), pore-forming toxins (2, 23), and microbial surface components recognizing adhesive matrix molecules (24). Thus, we first attempted to identify which S. aureus virulence factors may be involved in the rabbit model of necrotizing pneumonia. A comprehensive analysis of the relative contributions of various toxins secreted by USA300 revealed that only Hla and PVL, and to a lesser extent PSM-α, had a significant impact on acute lung injury and survival outcomes in the rabbit model of necrotizing pneumonia (Figs. 4 and 5). Hla and PVL were produced at toxic levels in the lungs of many of the rabbits infected with the five major epidemic community-associated MRSA strains, indicating that they also contributed to the pathogenesis of necrotizing pneumonia caused by not only USA300 but also other prevalent CA-MRSA clones. Finally, the absence of lethality induced by the USA300 ΔhlaΔlukSF double mutant confirmed the importance of these two toxins in the pathogenesis of necrotizing pneumonia (Fig. 5A).

The presence of neutralizing antibodies against Hla and PVL in IVIG was shown previously (2, 23) and confirmed herein for the two different batches of IVIG used in the rabbit studies (Fig. 1, A to E). Pharmacokinetic analysis in rabbits indicated that administration of IVIG (200 mg/kg) resulted in serum titers sufficient to neutralize Hla and PVL secreted by the bacteria in rabbit lungs for at least 48 hours (Fig. 2, A to C). Prophylaxis with IVIG reduced mortality compared to saline control in the rabbit pneumonia model using five different community-associated MRSA strains that produced both Hla and PVL (Fig. 3, A to O). The roles of neutralizing antibodies against these two toxins were confirmed by demonstrating that affinity-purified anti-Hla and anti-PVL antibodies protected against lethal infection, whereas IVIG depleted of these neutralizing antibodies did not significantly improve animal survival [Figs. 2 (G to I), 6 (A to L), and 7 (A to F)]. Given that the bicomponent toxins PVL (LukSF), LukED, HlgABC, and LukGH/LukAB are antigenically related, we suspect that most antibodies against PVL in IVIG could be attributed to previous exposure to one of these toxins, and we cannot rule out possible cross-neutralization afforded by affinity-purified antibodies. However, it may have little impact in our rabbit model because the deletion of genes encoding the other bicomponent toxins did not have a significant impact on acute lung injury (Fig. 4).

Given that the distinctions between community-associated MRSA and hospital-associated MRSA strains are blurred by cross-transmission of these strains into and out of health care facilities (25), it was important to determine whether IVIG also protected against a hospital-associated MRSA strain. Clinical studies are conflicting as to whether S. aureus strains with high cytotoxic activities are associated with more severe disease. Rose et al. (26) reported paradoxically that MRSA strains with low cytotoxic activity were associated with increased mortality rates because of nosocomial pneumonia. Stulik et al. (27) reported that high α-hemolysin cytotoxic activity was associated with ventilator-associated pneumonia for infection with methicillin-susceptible S. aureus strains but not MRSA strains. It is not clear whether the USA100/NRS382 clinical strain used in our study would be classified as having high or low cytotoxic activity because this strain belongs to clonal complex 5 that tends to exhibit wide variation in cytotoxicity (26). Nonetheless, the USA100/NRS382 clinical strain is representative of the hospital-associated MRSA lineage that is widespread in health care facilities across the United States (14). In the rabbit pneumonia model, USA100/NRS382 caused rapidly fatal pneumonia in animals administered depleted IVIG, whereas those animals administered IVIG had a 50% reduction in overall mortality rate (P < 0.022; Fig. 7D), indicating that IVIG protected against pneumonia caused by not only community-associated MRSA strains but also a hospital-associated MRSA strain.

IVIG has been used clinically for its autoimmune-inhibiting and anti-inflammatory therapeutic effects (28). In the rabbit model of necrotizing pneumonia, it is possible that IVIG-mediated protection was due to its anti-inflammatory effects that lessened the severity of host-mediated acute lung injury (11). However, high-dose IVIG, 2 g/kg, is generally required for IVIG’s anti-inflammatory effects, whereas protection in the rabbit model of necrotizing pneumonia was achieved using 1/10 of the human dose, 200 mg/kg. If there was an anti-inflammatory effect in the rabbit model, then it was likely to be minimal because 59 of 64 (92%) rabbits administered IVIG (200 mg/kg) depleted of anti-Hla and anti-PVL antibodies did not survive (Figs. 2G, 6 (A, D, G, and J), and 7 (A and D)] compared to a similar mortality rate of 84% (53 of 63) for rabbits administered saline [Figs. 2D, 3 (A, D, G, J, and M), and 6A].

The role of IVIG in enhancing in vitro opsonophagocytic killing of S. aureus has been debated (2933). Notably, rabbits administered IVIG depleted of its antitoxin antibodies did not differ in lung bacterial counts from those treated with saline (Fig. 6C), suggesting that the myriad other antibodies in IVIG did not enhance in vivo opsonophagocytic killing of S. aureus. In support of this, six of seven studies in which IVIG was compared to saline in the rabbit model also showed no significant differences in lung bacterial counts [Figs. 2F and 3 (C, F, I, and O)], with the exception being the one study in which rabbits were challenged with the USA1100 strain, which yielded mean log10 CFU of 8.39 and 5.52 (P = 0.007) for those animals pretreated with saline and IVIG, respectively (Fig. 3L). Although this exception could be due to IVIG enhancing bacterial killing of USA1100 strain, another potential explanation is that IVIG protected all infected animals from acute death (Fig. 3J), thereby allowing those animals that lived longer a correspondingly greater period of time for clearance of bacteria from the lungs (Fig. 3L). Accordingly, the two USA1100-infected rabbits, which were treated with saline and survived, also had fewer lung bacterial counts (log10 CFU of 5.04 and 7.27; Fig. 3L).

Linezolid, a protein synthesis inhibitor, is recommended in the IDSA clinical practice guidelines as an acceptable alternative to vancomycin for treatment of MRSA pneumonia (5). We have shown previously (16) as well as herein (Fig. 2, D to I) that linezolid, but not vancomycin, protected against lethal necrotizing pneumonia in rabbits. Combination treatment with IVIG-vancomycin resulted in an overall survival rate of 71% (Fig. 2D), whereas combination treatment with IVIG-linezolid resulted in a higher overall survival rate of 93% (Fig. 2G). Linezolid was shown previously to inhibit production of PVL and Hla in the rabbit lungs (16). The combined antitoxin properties of linezolid, which inhibits production of bacterial toxins, and IVIG, which neutralizes preformed toxins and prevents their cytotoxic effects on host cells, conferred the greatest protection in the rabbit model. Comparative data such as these are important for demonstrating a potential benefit of using a protein synthesis inhibitor, like linezolid, over vancomycin in combination with IVIG.

Our study has certain limitations. IVIG, either alone or in combination with vancomycin or linezolid, was administered at a single time point, 1.5 hpi. We have shown previously in the same rabbit model of necrotizing pneumonia that delaying treatment with linezolid to 4 or 9 hpi decreased or abolished its protective effects (16). Whether these narrow windows of efficacy in the model translate into a time frame within which a patient is clinically likely to present and be diagnosed with staphylococcal pneumonia is not known, a practical limitation of the rabbit model due to the acute nature of this experimental infection (16). The combination of linezolid and IVIG demonstrated the greatest protective effects (Fig. 2), yet whether it can extend the treatment window and confer protection when administered at a later stage of infection remains to be determined. Our study is also limited by the fact that only a single commercial IVIG preparation, ClairYg produced by Laboratoire Français du Fractionnement et des Biotechnologies (LFB), was evaluated for protective efficacies in the rabbit model. Although different commercial IVIG preparations may vary in their concentrations of specific antitoxin antibodies, it is beyond the scope of this work to determine the in vitro and in vivo antitoxin properties of IVIG from the different companies. However, the two lots of ClairYg evaluated herein showed similar concentrations of anti-Hla and anti-LukS/F neutralizing antibodies (Fig. 1) and protective efficacies in vivo (Figs. 2, 3, and 6). A different commercial IVIG preparation, Tégéline produced by LFB, was shown previously to contain high titers of anti-LukS/F antibodies (2). The difference in neutralizing activities of different commercial IVIG preparations against staphylococcal and streptococcal superantigens has been shown to be relatively small, ranging from 6 to 11% (15). It should be noted that even a low dose of IVIG (200 mg/kg), which is 1/10 of the recommended dose for use in humans, demonstrated protective efficacies in our rabbit studies. It remains to be seen whether any differences in neutralization potencies of different commercial IVIG preparations, especially when used at the recommended high dose of 2 g/kg, would affect clinical efficacy.

In conclusion, IVIG was protective against lethal S. aureus necrotizing pneumonia in a rabbit model because of the presence of specific polyclonal antibodies that neutralized the two key lung-damaging toxins, Hla and PVL. The results demonstrate the potential clinical utility of IVIG as an adjunct therapy for treating S. aureus necrotizing pneumonia.

MATERIALS AND METHODS

Study design

Protective effects of IVIG, alone or in combination with vancomycin or linezolid, were determined in preexposure prophylaxis studies and postexposure treatment studies in a rabbit model of necrotizing pneumonia. IVIG dosing in rabbits was determined empirically on the basis of serum concentrations of human IVIG in rabbits and in vitro toxin neutralization titers. To determine whether specific antibodies in IVIG conferred protection, we affinity-purified anti-PVL and anti-Hla antibodies from IVIG. IVIG also has opsonophagocytic killing and anti-inflammatory activities that could confer protection independent of specific antitoxin antibodies. To control for these possibilities, IVIG depleted of its anti-PVL and anti-Hla antibodies was also used. The various IVIG-derived antibody constructs were compared for protective efficacies in the rabbit pneumonia model using five different community-associated MRSA clinical strains and one predominant hospital-associated MRSA clinical strain. Unraveling the mechanism of IVIG-mediated protection was facilitated by comparing virulence of S. aureus WT and isogenic mutants deficient in various secreted toxins (table S1 and S2). The rabbit experimental pneumonia protocol was reviewed and approved by the University of California, San Francisco (UCSF) Institutional Animal Care and Use Committee and conducted in a facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. Infected animals were monitored every 2 to 3 hours for the first 30 hpi and then three times daily thereafter for clinical signs of pulmonary dysfunction, defined as abnormal respiratory rate >75 breaths/min, cough, and cyanosis. Animals euthanized for pulmonary dysfunction were recorded as nonsurvivors. Sample sizes were estimated using log-rank test with 5% type I error rate and 80% power. All pathogenesis studies for the comparison of WT and isogenic mutant strains were conducted as blinded, randomized experiments. Although animals were randomized for treatment studies, these studies cannot be conducted in a blinded manner because of differences in administration of various treatment modalities.

Rabbit model of necrotizing pneumonia

To establish necrotizing pneumonia in the rabbit model, a 1.5-ml instillate containing various bacterial strains was delivered directly into the lungs of anesthetized New Zealand white outbred rabbits through a 2.5-mm pediatric endotracheal tube as previously described (11). We used 8- to 12-week-old rabbits, weighing 2.0 to 2.8 kg. For the mortality studies, rabbits were monitored every 2 to 3 hours for the first 30 hpi and then three times daily thereafter, and survivors were euthanized at either 36, 48, or 96 hpi. For the time-killed studies, rabbits were euthanized at 9 hpi to determine the extent of acute lung injury. Lungs were removed aseptically from euthanized rabbits or those that were found dead. Lungs were cut into <0.5-cm pieces. Part of the lung sample was homogenized in 0.9% saline and titered by plating serial dilutions on blood agar to determine the number of CFUs.

The following regimens were administered intravenously at the indicated doses via the marginal ear vein of the rabbits for either preexposure prophylaxis (24 and 1.5 hours before induction of experimental pneumonia) or postexposure treatment (1.5 hours after induction of experimental pneumonia): IVIG (200 mg/kg; ClairYg, LFB), vancomycin (30 mg/kg), IVIG (200 mg/kg) depleted of anti-LukS/F-PV and anti-Hla IgG, affinity-purified anti-LukS/F-PV IgG (1 or 4 mg/kg), and affinity-purified anti-Hla IgG (1 or 4 mg/kg). Two different lots of IVIG, IVIG(L3) and IVIG(L8), were used in various experimental studies. Linezolid was dissolved in 5% cyclodextrin to 12 mg/ml and administered at 50 mg/kg subcutaneously.

Statistical analyses

Survival curves were generated using the Kaplan-Meier method, and significance was assessed by means of the log-rank (Mantel-Cox) test, with Bonferroni correction for multiple comparisons where appropriate (GraphPad version 6.0). Normal distribution was not assumed, so LW/BW (×103), log10 CFU, and concentrations of toxins and cytokines were compared using a nonparametric two-sided Mann-Whitney U test or one-way ANOVA with Kruskal-Wallis test, followed by Dunn’s multiple comparisons post hoc test where appropriate.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/8/357/357ra124/DC1

Materials and Methods

Fig. S1. SElX does not contribute to lethal infection with USA300/LAC strain in the New Zealand white rabbit model of necrotizing pneumonia.

Fig. S2. Anti-Hla IgG and anti-LukF/S IgG affinity-purified from IVIG have potent neutralization activities.

Table S1. Bacterial strains used in the present study.

Table S2. Oligonucleotides used for construction of in-frame gene deletions using pKOR1 allelic replacement system.

References (3436)

REFERENCES AND NOTES

  1. Acknowledgments: We thank the LFB for providing IVIG (ClairYg) and J.-M. Dugua, F. Bettsworth, C. Lamotte, and B. Le Levreur for their technical assistance. Funding: This work was supported in part by the U.S. Public Health Service grant NIH R01 AI087674 to B.A.D., a UCSF Research Evaluation and Allocation Committee grant to B.A.D., and MedImmune, a member of the AstraZeneca group. Author contributions: B.A.D., H.F.C., and G.L. designed the studies and obtained funding. B.A.D., V.T.M.L., H.N.L., M.G.P., A.H.D., X.W., E.C.D., F.A.-A., L.B., H.M., T.T.M., and H.F.C. constructed the isogenic mutants, performed the rabbit studies, and/or analyzed those data. C.B., M.N.S., J.-P.R., and G.L. prepared affinity-purified antibodies and depleted IVIG, performed in vitro toxin neutralization assays and toxin quantification, and/or analyzed those data. O.K. and G.M.-B. performed cytokine quantification and/or analyzed those data. C.T. and B.R.S. performed toxin quantification and/or analyzed those data. B.A.D. performed statistical analyses. B.A.D., V.T.M.L., B.R.S., H.F.C., and G.L. wrote the manuscript. Competing interests: B.A.D. received research funding from Arsanis Biosciences, Cubist (now a part of Merck), Genentech (now a part of Roche), Integrated BioTherapeutics, MedImmune (now a part of AstraZeneca), and Pfizer. C.T. and B.R.S. are employed by MedImmune, a member of the AstraZeneca group, and may hold AstraZeneca stocks or stock options. C.T. and B.R.S. are coinventors on patent WO/2012/109285, entitled “Antibodies that specifically bind Staphylococcus aureus alpha toxin and methods of use,” which was filed by MedImmune. H.F.C. is a paid consultant for Cubist, AstraZeneca, Theravance, Allergan, Pfizer, and Genentech. The other authors declare that they have no competing interests. Data and materials availability: Patent WO/2012/109285 describing MEDI4893* has been filed by MedImmune. MEDI4893*, an anti-Hla human monoclonal antibody, was used as an ELISA reagent for quantification of toxin levels in rabbit lungs.
View Abstract

Navigate This Article