OUP user menu

The effects of Escherichia coli capsule, O-antigen, host neutrophils, and complement in a rat model of Gram-negative pneumonia

Thomas A. Russo , Bruce A. Davidson , Ulrike B. Carlino-MacDonald , Jadwiga D. Helinski , Roger L. Priore , Paul R. Knight III
DOI: http://dx.doi.org/10.1016/S0378-1097(03)00636-0 355-361 First published online: 1 September 2003

Abstract

Gram-negative enteric bacilli are agents of life-threatening pneumonia. The role of the bacterial capsule and O-antigen moiety of lipopolysaccharide in the pathogenesis of Gram-negative pneumonia was assessed. In a rat model of pneumonia the LD50 of a wild-type extraintestinal pathogenic Escherichia coli strain (CP9) was significantly less than its isogenic derivatives deficient in capsule (CP9.137), O-antigen (CP921) or both capsule and O-antigen (CP923) (P≤0.003). Studies using complement depleted or neutropenic animals established that both neutrophils and complement are important for the pulmonary clearance of E. coli. Data from these studies also support that capsule and O-antigen serve, at least in part, to counter the complement and neutrophil components of the pulmonary host defense response. Lastly, the contribution of E. coli versus neutrophils in causing lung injury was examined. Findings suggest that E. coli virulence factors and/or non-neutrophil host factors are more important mediators of lung injury than neutrophils. These findings extend our understanding of Gram-negative pneumonia and have treatment implications.

Keywords
  • Pathogenesis
  • Capsule
  • O-antigen
  • Pneumonia
  • Neutrophil
  • Complement
  • Lung injury
  • Escherichia coli

1 Introduction

Gram-negative enteric bacilli, such as Escherichia coli, Pseudomonas aeruginosa, Klebsiella pneumoniae, Enterobacter sp., Acinetobacter, Serratia marcescens, are the most common cause of hospital-acquired pneumonia, being implicated in 55–85% of cases [16]. Pneumonia accounts for 15% of hospital-acquired infections [57] with an associated crude mortality and estimated attributable mortality ranging from 24 to 76% and from 20 to 50% respectively [5,6,8].

Microbial pathogens have acquired a variety of virulence factors that enable successful infection. In contrast to healthy ambulatory individuals, approximately 75% of critically ill patients are colonized with Gram-negative bacilli [9,10]. A combination of bacterial adherence factors (e.g. pili), endogenous host factors (e.g. age, underlying disease), and exogenous environmental factors (e.g. cross contamination, invasive devices, antibiotic use) are contributory to both colonization and subsequent aspiration [2,7]. However, to cause pneumonia, bacterial pathogens need to be able to survive within the lung. Capsular polysaccharides have been established as virulence traits that protect against pulmonary clearance in several respiratory pathogens, including Streptococcus pneumoniae and Haemophilus influenzae. Presumably, the capsular polysaccharides present on enteric Gram-negative bacilli play a similar role, however, this assumption has not been clearly established [11,12]. Lipopolysaccharide also appears to be important in protecting against pulmonary clearance, as demonstrated in a Pseudomonas pneumonia model [13].

A variety of innate and acquired host factors contribute to protection against lung infection [14,15], including resident macrophages, complement components, antimicrobial peptides, and an influx of neutrophils followed by blood monocytes [16]. Lastly, a specific acquired immune response may be required for the clearance of pathogens not cleared by the innate inflammatory response.

Although the host's goal is to eradicate invading pathogens, an over-exuberant or prolonged proinflammatory phase may result in host-mediated lung injury [1720]. Therefore, the ideal host response is to maximize bacterial clearance and minimize host-mediated collateral damage to host tissue. Likewise, Gram-negative bacilli virulence factors may also contribute to pulmonary damage [2124]. Remarkably, data on the relative importance of bacterial versus host factors in directly mediating pulmonary damage are limited.

Our long-term goal is to identify strategies to decrease morbidity and mortality from Gram-negative pneumonia. To accomplish this, an increased understanding of bacterial virulence traits that contribute to the pathogenesis of Gram-negative pneumonia and the relative importance of bacterial versus host factors in mediating acute lung injury is needed. In this study, a rat model of Gram-negative pneumonia was used for testing the following hypotheses: (1) capsule and the O-antigen contribute to the survival of E. coli in pneumonia, (2) neutrophils and complement contribute to the clearance of E. coli in the lung, and (3) both host factors and bacterial factors are responsible for host tissue damage in Gram-negative pneumonia.

2 Materials and methods

2.1 Bacterial strains

A human bacteremic isolate of E. coli (CP9, O4/K54/H5) and isogenic derivatives were used as model pathogens for these studies [25,26]. CP9 is a well-characterized extraintestinal pathogenic E. coli strain [27]. It is a human blood isolate and is virulent in a variety of in vivo infection models [28,29]. Transposon mutagenesis (TnphoA for capsule genes and TnphoA′1 for O-antigen genes) was used to generate proven isogenic derivatives of CP9 deficient in the K54 capsule alone (CP9.137), the O4 O-antigen alone (CP921), or both (CP923) [25,26]. Sequence analysis established the precise location of the transposon insertions in the expected capsule and O-antigen gene clusters [26,30]. As a result of their locations within these gene clusters, polar effects of the transposon insertions only affected genes involved in capsule or O-antigen biosynthesis, transport and assembly [26,30,31].

2.2 Pulmonary infection model

Animal studies were reviewed and approved by the University at Buffalo Institutional Animal Care Committee. An established rat (Long–Evans) model for studying pulmonary damage was used as reported [32]. In brief, Long–Evans rats (250–300 g) were anesthetized with approximately 3% halothane in 100% oxygen until unconscious and then maintained at 1.5% halothane. The trachea was exposed surgically, and a 4 inch piece of 1–0 silk was slipped under the trachea to facilitate instillation of the inoculum. The animals were suspended in a supine position on a 60° incline board. The bacterial challenge inoculum was prepared in normal saline introduced intratracheally (1.2 ml kg−1) via a 1 ml syringe and a 26 gauge needle and the incision closed with surgical staples. Various challenge inocula (2.0×105–6.0×108 CFU) of CP9 (wild-type), CP9.137 (capsule minus), CP921 (O-antigen minus), and CP923 (capsule and O-antigen minus) were used to establish the consequences of infection for each of these strains. Animals that died within 15 min after bacterial challenge were excluded from the analysis, since these deaths were attributed to the challenge procedure, not the infecting bacteria themselves. Each experimental animal was prospectively assigned a time for subsequent sampling.

  1. Lung homogenate harvest technique: At harvest, halothane anesthesia was induced and a midline incision made through the peritoneum and thoracic cavity. Blood for bacterial culture was obtained from the abdominal aorta. 100 µl of blood was cultured on Luria–Bertani (LB) plates in duplicate at 37°C. The pulmonary vasculature was flushed of residual blood by injecting the right ventricle with 10 ml of 5 mM ethylenediamine tetraacetic acid (EDTA)/normal saline using a 26 gauge needle. The lungs were weighed, suspended in normal saline to a total weight of 10 g (assumed to equate to 10 ml), and homogenized, on ice, for 3 s three times using a Polytron PT-2000 homogenizer (Brinkman Instruments, Westbury, NY, USA). Aliquots of the homogenate were removed for E. coli titer.

  2. Determination of lung homogenate bacterial titers: Serial 10-fold dilutions of the lung homogenate were performed in 1× phosphate-buffered saline and cultured on LB plates, in duplicate, at 37°C. E. coli titers were determined (CFU ml−1) and multiplied by the 10 ml total volume of the homogenate to yield the total lung titer in CFU per lung.

  3. Assessment of alveolar capillary integrity: The details of this technique have been previously reported. In brief, this technique measures the leakage of [125I]albumin from the plasma into the lung [33]. This measurement, designated as the pulmonary permeability index (PI), provides a sensitive indication of injury to the alveolar capillary wall.

  4. Assessment of neutrophils influx: Total lung myeloperoxidase (MPO) activity was used to quantify pulmonary neutrophils as described [33]. In brief, lung homogenates were assayed for MPO activity by measuring the rate of absorbance change at 460 nm after adding 50 µl sample to 1.5 ml assay buffer (50 mM phosphate buffer, pH=6.0, 0.004% v/v H2O2 and 525 µM o-dianisidine hydrochloride (Sigma Chemical, St. Louis, MO, USA)). Units are calculated based on the rate of absorbance increase over time and normalized to total lung homogenate volume.

  5. Complement depletion: Complement depletion was accomplished by intraperitoneal injection of 20 µg of anti-complement cobra venom factor, suspended in 1 ml of 10 mM EDTA plus 1% (v/v) glycerol, (Sigma Chemical, St. Louis, MO, USA) into the rat 24 h prior to initiation of injury. Adequacy of complement depletion was assessed in all animals by determining the CH50 in serum, derived from a sample of blood taken just prior to bacterial challenge, as developed by Mayer [34]. Only animals in which CH50 was <20 U ml−1 were considered complement depleted.

  6. Depletion of neutrophils: Neutrophil depletion was accomplished by intraperitoneal injection of 1 ml of rabbit, anti-rat neutrophil antiserum (Accurate Chemical Co., Westbury, NY, USA). Normal rabbit serum (Accurate Chemical Co.) was used for controls. This induced selective depletion of blood neutrophils to <500 neutrophils ml−1 of blood, 16–18 h after injection of antiserum. Before bacterial challenge, a white blood cell count was performed to insure adequate depletion of the neutrophils [35].

2.3 Statistical analysis

2.3.1 LD50 analyses

The probability that an animal would die within the first 24 h was modeled using the logistic function, p=1/(1+exp(−yi)), where: p is the probability of death within 24 h, yi=bi (xmi), x is the common log of the challenge inoculum, mi is the common log of the LD50 for strain i, and bi is the slope parameter of the logistic function for strain i. The logistic model was fitted by: (1) using a single b and m, (2) using a different b and m, and (3) using a single b and different m. Analyses established that by increasing the numbers of parameters from two to five by using a separate LD50 for each strain significantly improved the fit of the model (P<0.001), but that further increasing the number of parameters from five to eight by using different slope parameters for each strain did not significantly improve the fit of the model (P=0.297). To put 95% confidence limits on the LD50 for individual strains the log likelihood was recomputed, changing each LD50 in small increments until twice the log likelihood decreased by 4.

2.3.2 Effect of depleting neutrophils and complement on bacterial growth/clearance

A two-sample, two-tailed t-test assuming equal variance was used for these analyses.

2.3.3 Pulmonary damage

A Fisher Z transformation test was used to test correlations between MPO levels, bacterial titers and pulmonary PI. A two-sample, two-tailed t-test assuming equal variance was used to test for differences in pulmonary PI when complement was repleted or depleted.

3 Results

3.1 Evaluation of capsular polysaccharide and the O-antigen as virulence factors in the rat model of Gram-negative pneumonia

Long–Evans rats were used in a rat model of pneumonia [32]. The roles of the bacterial capsule and O-antigen in the pathogenesis of Gram-negative pneumonia were assessed by comparing the LD50 of the wild-type parent CP9 (challenge inocula: 2.4×105–6.0×107 CFU, n=26) with its isogenic derivatives CP9.137, capsule minus (challenge inocula: 1.1×106–3.7×108 CFU, n=18); CP921, O-antigen minus (challenge inocula: 1.6×106–5.8×108 CFU, n=16); and CP923, capsule and O-antigen minus (challenge inocula: 9.6×105–3.8×108 CFU, n=18). Six different challenge inocula for each strain, at one-half log intervals, were used for this determination. The animals were followed post-bacterial challenge for 24 h. Previously, we established that all deaths occurred within this time frame. The probability that an animal would die was modeled by a logistic function described in Section 2. The LD50 was significantly greater for each of the isogenic mutants (CP9.137, CP921, CP923) than for their wild-type parent (CP9) (Table 1). Therefore, the presence of capsule or O-antigen increases bacterial virulence and adversely affects host survival.

View this table:
1

LD50 for CP9, CP9.137, CP921, and CP923 within 24 h in the rat pneumonitis model

Strain (n)LD50 (CFU)95% Lower limit (CFU)95% Upper limit (CFU)
CP9 (wild-type) (26)8.0×1065.9×1061.1×107
CP9.137 (K54 minus) (18)3.0×1072.0×1074.6×107
CP921 (O4 minus) (16)3.1×1072.0×1074.8×107
CP923 (O4, K54 minus) (18)2.0×1081.3×1083.2×108
  • P<0.001 relative to CP9.

3.2 Evaluation of neutrophils and complement on E. coli's pulmonary clearance in vivo

Neutrophils and complement are important components of the host's innate defense response. Capsule and O-antigen have been shown to protect extraintestinal pathogenic E. coli against the bactericidal effects of complement and professional phagocytes [36]. Therefore, the effect of neutrophils and complement on the pulmonary clearance of E. coli in vivo was assessed. Normal rats, neutropenic rats and complement depleted rats were challenged with CP9 (wild-type) (vehicle treated n=23, complement depleted n=9, neutrophil depleted n=9), CP9.137 (capsule minus) (vehicle treated n=15, complement depleted n=5, neutrophil depleted n=9), and CP921 (O-antigen minus) (vehicle treated n=15, complement depleted n=8, neutrophil depleted n=6), with a challenge inoculum of approximately 1.0×107 CFU. 6 h post-challenge whole lung bacterial titers were determined (Table 2).

View this table:
2

The effect of complement and neutrophil depletion on the pulmonary clearance of CP9 (wild-type), CP9.137 (capsule minus), and CP921 (O-antigen minus) at 6 h post-bacterial challenge in the rat pneumonia model

Strain/CITotal lung CFU in vehicle treated animals (n)Total lung CFU in cobra venom factor treated animals (n)Total lung CFU in anti-rat neutrophil antiserum treated animals (n)
CP9 (wild-type)
8.6×1069.6×105±3.9×105 (6)ND9.1×107±1.8×107 (5)
1.1×1078.5×106±2.7×106 (9)3.0×108±1.6×108 (6)4.0×108±5.0×107 (2)
2.3×1078.3×108±2.0×108 (8)1.0×109±3.2×108 (3)4.5×108±3.2×107 (2)
CP9.137 (capsule minus)
9.3×1065.5×105±7.2×104 (9)2.9×107±1.2×107 (2)1.8×106±2.3×104 (6)
1.5×1077.5×105±1.5×105 (6)5.0×108±3.6×108 (3)3.0×106±4.8×104 (3)
CP921 (O-antigen minus)
1.0×1077.2×105±2.4×105 (9)3.5×108±1.6×108 (5)2.4×107±1.2×107 (3)
1.3×1072.1×105±3.8×104 (3)1.5×108±3.3×107 (3)ND
1.8×1072.3×106±6.4×105 (3)ND7.0×106±2.0×106 (3)
  • P≤0.05, all comparisons are between cobra venom factor or anti-rat neutrophil antiserum treated animals with vehicle treated animals challenged with the same strain and challenge inoculum.

  • Challenge inoculum (CFU).

  • Lungs harvested at 6 h.

  • Number of animals.

  • Standard error of the mean.

The pulmonary growth of CP9 (wild-type), CP9.137 (capsule minus), and CP921 (O-antigen minus) was increased in the absence of either complement or neutrophils in vivo (Table 2), thereby establishing that both of these innate host defense factors are important for the pulmonary clearance of E. coli. In the presence of complement and neutrophils, CP9.137 (capsule minus) and CP921 (O-antigen minus) are cleared more readily than their wild-type parent CP9 (Table 2) [32]. However, complement depletion eliminates and neutropenia partially eliminates this survival advantage of CP9 (wild-type) (Table 2). Therefore, these data also support the concept that capsule and O-antigen serve, at least in part, to counter the complement and neutrophil components of the pulmonary host defense response.

3.3 Evaluation of E. coli and pulmonary neutrophils in mediating pulmonary damage in the rat model of pneumonia

The pertinent features of Gram-negative pneumonia occurred in the rat pneumonia model; namely bacterial growth (or clearance), an ensuing inflammatory response, and with progressive bacterial proliferation pulmonary damage, as manifested by large increases in lung weight, broncho-alveolar lavage protein content, the pulmonary PI, and a highly significant decrease in oxygenation [32]. Bacteremia was uncommon (2.8%) [32]. Although a role of the systemic inflammatory response and disseminated infection cannot be excluded as contributors to death, lung injury parameters strongly support that respiratory failure is a central feature in the pathogenesis of this infection, similar to untreated Gram-negative pneumonia in humans.

Therefore an important question in the pathogenesis of Gram-negative pneumonia is the relative role of host factors (e.g. reactive oxygen species) versus the infecting bacterium in causing pulmonary damage. To begin to address this question experiments were designed to determine the relative role of neutrophils and E. coli in mediating pulmonary injury. This was accomplished by determining the pulmonary PI (a measure of the integrity of the alveolar capillary wall), MPO, and bacterial titer at 6 h after challenge with the wild-type strain CP9 (n=37). The correlation, or lack thereof, between pulmonary damage, pulmonary neutrophil influx and bacterial load was determined by plotting pulmonary PI versus MPO and bacterial CFU.

Increases in pulmonary PI correlated (Fisher Z transformation test) with increasing bacterial titer (correlation=0.809, P<0.001; R2=0.6538) (Fig. 1A), whereas there was a lack of correlation between pulmonary PI and MPO activity (correlation=0.175, P=0.3021; R2=0.0307) (Fig. 1B).

1

Pulmonary damage correlated with bacterial load, not neutrophil recruitment, in the rat model of Gram-negative pneumonia. Rats (n=37) were challenged with various inocula of CP9 (wild-type). After 6 h the pulmonary PI (a measure of the integrity of the alveolar capillary wall), MPO activity, and bacterial titer were measured as described in Section 2. The correlation between pulmonary damage and bacterial load was determined by plotting PI versus bacterial CFU. Increases in PI correlate (Fisher Z transformation test) with increasing bacterial titer (correlation=0.809, P<0.001; R2=0.6538) (A). In contrast, there is a lack of correlation between PI and MPO activity (correlation=0.175, P=0.3021; R2=0.0307) (B).

These findings suggest that E. coli virulence factors and/or non-neutrophil host factors are more important mediators in mediating lung injury than neutrophils. Taken together with data demonstrating that neutrophils contribute to the clearance of E. coli, these findings support the contention that neutrophils are primarily beneficial to the host in this model of E. coli pneumonia.

4 Discussion and conclusions

Enteric Gram-negative bacilli are the most common cause of hospital-acquired pneumonia and result in significant morbidity and mortality. A rat model of Gram-negative pneumonia was used to increase our understanding of the pathogenesis of this infection. LD50 experiments demonstrate that both the K54 capsule and O4-antigen moiety of LPS from the extraintestinal pathogenic E. coli strain CP9 are pulmonary virulence factors. As predicted, experiments using neutropenic and complement depleted rats demonstrated that both neutrophils and complement contribute to the pulmonary clearance of E. coli. Further, capsule and O-antigen, at least in part, serve to counter these pulmonary defense components. Perhaps most importantly, in this model system, lung injury correlates with bacterial load, not pulmonary neutrophils. This finding supports neutrophils as being primarily beneficial and either non-neutrophil host factors or E. coli are the primary mediators of pulmonary damage.

The lung has evolved a coordinated response to physical and infectious insults designed to minimize injury and clear invading pathogens. Three major variables impact on the likelihood of whether the protective host response will be overwhelmed and pneumonia will develop: (1) bacterial challenge inoculum, (2) bacterial virulence, and (3) dysfunction of host response. Increases in any of these factors enhance the probability that pneumonia and acute lung injury will develop. The mechanism by which capsule and O-antigen enhance the pulmonary pathogenic potential of extraintestinal pathogenic E. coli is probably multi-factorial. An obvious mechanism is affording the bacterium protection against host bactericidal factors.

Previous in vitro studies have demonstrated that the K54 capsule of CP9 is the primary determinant that protects against the bactericidal activity of complement [37,38]. Previous in vivo studies have demonstrated that capsule conferred protection to CP9 in systemic infection models (after both intraperitoneal or intravascular challenge), in an abscess model, but not in the urinary tract [28,29,37]. This study extends these observations and demonstrates the importance of capsule in protecting against clearance in host sites where complement is a critical host defense component, such as the lungs. This is further demonstrated by the enhanced growth of the isogenic capsule deficient derivative of CP9, CP9.137, in complement depleted animals. However, CP9.137 also demonstrated a modest increase in growth in neutropenic animals (Table 2), suggesting that the K54 capsule may also protect E. coli against the bactericidal activity of neutrophils.

In CP9, in vitro, the O4-antigen is the major bacterial determinant for protecting against the bactericidal effects of antimicrobial peptides, but only a minor determinant in protecting against the bactericidal activity of complement [26,37]. However, within the lungs, in vivo, enhanced growth of the isogenic O-antigen deficient derivative of CP9, CP921, occurred in complement depleted animals. It is unclear whether the O-antigen serves to protect E. coli against complement's bactericidal properties, opsonic properties, or both within the lungs. CP921 also demonstrated an increase in growth in neutropenic animals (Table 2), supporting the concept that the O4-antigen also protects E. coli against the bactericidal activity of neutrophils.

Remarkably, our knowledge of the mediators of pulmonary damage during Gram-negative pneumonia is limited. Neutrophils have been incriminated in mediating pulmonary damage in a variety of other clinical settings including aspiration [17], systemic sepsis [18,19], and adult respiratory distress syndrome [20]. While host responses are clearly capable of producing pulmonary damage, our data support the contention that any deleterious effects of neutrophils are far out-weighed by their protective effects. Although our data do not suggest that neutrophils are detrimental to the host in natural infection, a contribution of neutrophils to pulmonary damage cannot be excluded. Given that pulmonary injury correlated with bacterial titer, but not MPO, a reasonable interpretation of our results is that the bacterial and/or non-neutrophil host factors are the primary mediators of pulmonary injury. Bacterial exotoxins are the likely candidates for mediating host tissue damage, however, host factors, such as proteases and cytokines, are additional candidates that may cause tissue damage [2224,39].

The promise for using biologics in the successful treatment or modulation of Gram-negative pneumonia exists. However, it is becoming clear that an understanding of host–pathogen interactions, whether host and/or pathogen factors mediate damage, and the respective targets is needed to realize improved treatment outcomes. Without this knowledge, inappropriate inhibition of a host process or bacterial component will not only be ineffective but may actually be deleterious to the host [16,40,41].

Acknowledgements

Financial support: VA Merit Review from the Department of Veterans Affairs (T.A.R.); National Institutes of Health grants HL69763, AI42059 (T.A.R.), and HL48889, AI46534 (P.R.K.) and the The John R. Oishei Foundation (T.A.R.).

References

  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
  34. [34].
  35. [35].
  36. [36].
  37. [37].
  38. [38].
  39. [39].
  40. [40].
  41. [41].
View Abstract