Pneumolysin Induces the Salient Histologic Features of Pneumococcal Infection in the Rat Lung In Vivo C. Feldman, N. C. Munro, P. K. Jeffery, T. J. Mitchell, P. W. Andrew, G. J. Boulnois, D. Guerreiro, J. A. L. Rohde, H. C. Todd, P. J. Cole, and R. Wilson Host Defence Unit, Departments of Thoracic Medicine and Lung Pathology, National Heart and Lung Institute, Brompton Hospital, London, and Department of Microbiology, University of Leicester, Leicester, United Kingdom

Streptococcus pneumoniae infections are common, but how they cause host tissue injury and death is incompletely understood. Immunization with pneumolysin, a thiol-activated toxin produced by the pneumococcus, partially protects animals during subsequent infection. The mechanism by which pneumolysin contributes to disease is not known. The aim of the present investigation was to determine the histologic changes induced by recombinant pneumolysin in the rat lung and to compare them with the changes induced by live organisms. Injection of either toxin (200 or 800 ng) or bacteria into the apical lobe bronchus was associated with the development of a severe lobar pneumonia restricted to the apical lobe. The changes induced by the toxin were greater at the higher concentration, and changes were most severe in those animals in which there was partial ligation of the apical lobe bronchus. The pneumonitis was less severe following injection of a modified toxin with decreased hemolytic activity, generated by site-directed mutagenesis of the cloned pneumolysin gene, indicating that this property of the toxin was important in generating pulmonary inflammation. There was still considerable pneumonitis after injection of a modified toxin with decreased capacity to activate complement.

Infection with Streptococcus pneumoniae remains a common cause of morbidity and mortality in humans (1). The capsule of the pneumococcus has traditionally been described as its most important virulence factor, and is thought to protect the organism from phagocytosis (2), thereby contributing to its invasiveness. However, the capsule itself is not toxic to the host, and the mechanisms of host tissue injury and associated mortality are at present unclear (2). Pneumococcal cell wall components, and enzymes and toxins produced by the bacterium, have been suggested as having roles in the pathogenesis of disease. Pneumolysin is a thiol-activated toxin that is found in the cytoplasm of the pneumococcus and is released on autolysis (3, 4). Mice immunized with pneumolysin are partly but significantly protected from subsequent pulmonary challenge with the intact organism (5). More recently, a genetically engineered pneumolysin-negative mutant was found to be more rapidly cleared from the bloodstream, and was significantly less virulent in mice, than its pneumolysin-positive wild-type parent. The

virulence of the mutant was restored by reinstatement of its ability to produce the toxin (6). The exact mechanism by which the toxin contributes to the virulence of the organism is unclear, but pneumolysin has been shown to affect the activity of human polymorphonuclear leukocytes, lymphocytes, and platelets (7-10) and to activate the classic pathway of complement independently of specific pneumococcal antibody (11). The toxin also disturbs the structure and function of ciliated respiratory epithelium in vitro (12), suggesting that the effects of the toxin may be diverse depending upon where it is released in the body. Recombinant pneumolysin has been derived from the cloned pneumolysin gene and has been shown to be structurally and functionally identical to native toxin (13, 14). Modified toxins with different biologic properties have been generated by site-directed mutagenesis (15). The aim of the present study was to investigate the histologic changes induced by the injection of recombinant pneumolysin into the rat lung and compare them with the changes induced by injection of live organisms into the same animal model.

(Received in original form December 10, 1990 and in revised form April

18, 1991)

Materials and Methods

Address correspondence to: Dr. R. Wilson, Host Defence Unit, Department of Thoracic Medicine, National Heart and Lung Institute, Emmanuel Kaye Building, Manresa Road, London 8W3 6LR, United Kingdom.

Organisms The organism used in this study was a serotype 3 laboratory strain of S. pneumoniae (GB05), which was known to produce pneumolysin. The organism was cultured on 5 % horse blood agar plates incubated at 37° C overnight. The injectate

Abbreviations: colony-forming units, CFU; hemolytic units, HU; phosphate-buffered saline, PBS. Am. J. Respir. Cell Mol. BioI. Vol. 5. pp. 416-423, 1991

Feldman, Munro, Jeffery et al.: Effect of Pneumolysin in the Rat Lung In Vivo

was prepared by resuspending one colony of the bacterium from the blood agar plate in 100 ml of broth constituted by 25 g/liter of nutrient broth number 2, supplemented by yeast extract 3 g/liter, liver digest 3 g/liter (Oxoid Ltd., Basingstoke, Hants, UK), magnesium chloride (BDH Ltd., Dagenham, Essex, UK), and glucose 0.15% (wt/vol) (Phoenix Pharmaceuticals Ltd., Gloucester UK), which was incubated at 37° C with gentle rocking for 4 h (mid-log phase growth). Thereafter, a 20-ml aliquot of the preparation was centrifuged (5,000 X g) for 10 min at 4 C, the supernatant aspirated and discarded, and the residual bacterial pellet resuspended in 20 ml phosphate-buffered saline (PBS). The organism was triple-washed in PBS and then resuspended in the same medium for use as injectate. This methodology gave a reproducible viable count of bacteria in the final injectate, although five animals were rejected because the subsequent viable count was too low. A viable count of the injectate (colony-forming units [CFU]/ml) was performed before and after the operating session according to standard dilutional techniques. Preliminary studies (data not shown) in a dose-range from 104 to 108 CFU/ml did not show any obvious difference in the histology produced; however, mortality was higher in rats injected with the larger viable count. Therefore, animals were included in the study providing the result of the subsequent viable count was 5 x 105 to 5 x 1Q6 CFU/m!. 0

Recombinant Pneumolysin The recombinant toxin was prepared from Escherichia coli strain MC1061 pJW208, as previously described (14). The toxin was purified using high performance liquid chromatography, and the recombinant preparation was found to be identical to wild-type pneumolysin by sodium dodecyl sulfate polyacrylamide gel electrophoresis and N-terminal amino acid sequencing as described previously (14). The hemolytic titer of the purified toxin was measured by a semiquantitative technique as described previously (12) and was approximately 1.5 x 106 hemolytic units (HU)/mg protein. The toxin was diluted in PBS as required. The hemolytic activity of the toxin was measured before and at the end of each operating session to confirm accuracy and stability. Modified Pneumolysin Preparations Three additional pneumolysin preparations were studied. In one, heat inactivation of the original recombinant preparation was undertaken. Heat inactivation was carried out following dilution of the toxin in PBS to 40 j.tg/ml, by heating the preparation in a hot water bath at 56° C for 30 min. After heating, there was complete loss of measurable hemolytic activity and loss of ability to activate complement. The second and third modified toxins were prepared by oligonucleotide-mediated, site-directed mutagenesis as described previously (15). One modified toxin studied was derived by substituting tryptophan at position 433 with phenylalanine (Trp 433> Phe). This toxin had a markedly lowered hemolytic activity (1 X 103 HU/mg protein) but a normal complement-activating activity. The third modified toxin, tyrosine 384 to phenylalanine (Tyr 384 > Phe), was derived by the same methodology and had normal hemolytic activity, but its capacity to activate complement was diminished by

417

about 70%, as assessed by the method of Paton and associates (11) and by radioimmunoassay. Animals A total of 150 male specific pathogen-free Wistar rats (Charles River, Margate, UK), weighing 180 ± 40 g were randomly selected from an established breeding colony over a 9-mo period. The rats were fasted for 12 h preoperatively but were allowed free access to water. The rats were operated on in groups of six to 12 animals per session. Anesthesia Each rat was anesthetized with 1.7 j.tl/g (of rat weight) Hypnorm (0.818 mg fentanyl acetate and 10 mg fluanisone per milliliter; Janssen Pharmaceuticals Ltd., Marlow, UK) intramuscularly. Endotracheal intubation was performed under direct vision using a 5-cm-long 5FG cannula (Portex Ltd., Hythe, UK). The rat was placed on its left side, connected to a small animal ventilator (Harvard Apparatus Ltd., Edenbridge, UK), and ventilated using 100% O2 with a tidal volume of 1.6 ml and a rate of 90 cycles/min. Operative Procedure The operative procedure was similar to that used previously by our group to facilitate chronic bacterial colonization of the rat bronchial tree by P. aeruginosa (16, 17). A 2-cm incision, parallel to the line of the ribs, extending forward from the right shoulder was made and continued with blunt dissection to the rib cage. The thoracic cage was opened through the fourth intercostal space to reveal the apical lobe of the lung. The apical lobe bronchus was exposed by gentle retraction of the lung. In experiments in which partial ligation of the apical lobe bronchus was part of the study design, a 7/0 prolene suture (Ethicon, Edinburgh, UK) was tied around the apical lobe bronchus as close as possible to its origin from the right main bronchus. The suture was tightened to narrow but not occlude the apical lobe bronchial lumen. Twenty microliters of test solution was injected into the lung through the bronchial wall into the lumen of the apical lobe bronchus toward the lung periphery, distal to the suture when present and in the same area when not. The apical lobe was reflated by gently manually inflating the lungs with 5 to 7 ml of air via the endotracheal tube, using a 10-ml syringe. Successful inflation of the lung, as well as subsequent visualization of the apical lobe moving in synchrony with the ventilator, confirmed the ligation to be partial. The thoracotomy was repaired in two layers using 3/0 Ethibond (Ethicon). A 3FG cannula (Portex) was left in the thoracic cavity protruding through the wound. Before discontinuing ventilation, air was manually aspirated from the chest cavity through this drain, which was then removed. The animal was extubated, and 0.01 ml naloxone hydrochloride (Narcan 0.4 mg/ml; Du Pont, Stevenage, UK) was administered intramuscularly. Spontaneous respiration resumed in less than 1 min, and the rat breathed high-concentration oxygen via a face mask for 5 min. After return to the cage, food and water were allowed freely until the animal was killed. The rats were killed at 1, 7, or 30 days after the operative procedure. Death occurred 3 to 5 min after an overdose of Hypnorm.

418

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 5 1991

Blood Culture Direct cardiac puncture was performed in 14 animals infected with S. pneumoniae (four animals with ligation at day 1; four animals without ligation at day 1; six animals without ligation at day 7) prior to opening the chest. One hundred-microliter aliquots were plated out directly on 5 % horse blood agar plates incubated aerobically overnight. A further 100-J.t1 aliquots of blood were incubated in 5 ml of the broth described above at 37° C for 6 h, and then 100-J.t1 aliquots were plated out on 5% horse blood agar at 37° C overnight. All isolates were identified by Gram-staining and by standard identification techniques. Removal of Lung for Histology A vertical incision was made in the neck of the animal and extended to divide the sternum longitudinally allowing full exposure of the thoracic cavity. The heart and lungs were removed en bloc, care being taken to avoid damage to the lungs. The lungs were fully inflated with 10% formal saline via the tracheal cannula at a pressure of 15 em water. The trachea was then tied to prevent back leakage of fixative, and the whole preparation was immersed in formal saline for at least 96 h before routine processing and paraffin wax embedding for histology. Histology A single longitudinal section exposing the main axial bronchi of the right apical lobe of each animal was prepared and stained with hematoxylin and eosin. Further sections were stained with alcian blue (pH 2.6) and periodic acid Schiff applied in combination to detect acidic or neutral mucins, respectively. Slides were coded before histologic assessment by the observer (P. K. 1.), who was unaware of their treatment. The overall appearance and tissue response was assessed by examination of 22 parameters (see Table 1) each scored on a semiquantitative scale of 0 to 3. The histopathologic scores for alveolar edema, alveolar infiltrate, presence of interstitial and alveolar neutrophils (not macrophages), and a value reflecting the extent of spread of the lesion throughout the lobe were taken as the salient features of lobar pneumonia and summed to give a pneumonitis score. As judged by the extent of bronchus-associated lymphoid tissue, the animals began the experiment "clean" and free of respiratory infection. Study Design The study was continued until there were six surviving rats in each assigned group. The different groups are summarized in Table 2. S. pneumoniae injection (20 pi GB05 5 X J(Y to 5 X 1(}1 CFUlml in PBS; groups 1 to 4). Two groups of rats were studied with partial ligation of the apical lobe bronchus; one group was killed at 1 day, the other at 1 wk. Two other groups ofrats were studied without ligation; one group was killed at 1 day, the other at 1 wk. Recombinant pneumolysin injection (groups 5 to 13). Two amounts of pneumolysin were studied: a 20-J.t1 bolus of PBS containing 200 or 800 ng pneumolysin. Two groups were studied with, and two without, ligation. Animals were killed

TABLE 1

Histologic parameters assessed Category

Parameter

Bronchial

Dilatation Secretions* Epithelial ulceration Goblet cell/epithelial hyperplasia Loss of cilia

Alveolar

Congestiont Hemorrhages Pneumonitist

Peribronchial/vascular

Lymphoedema Interstitial edema Alveolar edema

Infiltratet

Broncho-associated lymphoid tissue Perivascular Interstitial Alveolar Interstitial neutrophilia Intra-alveolar neutrophil Macrophages Foamy macrophages Lymphocytes

Alveolar

Cuboidal metaplasia Fibrosis

* Grades 0 to 3 = none,acellular or macrophages alone,neutrophilia, or containing monocytes, respectively. t Grades 0 to 3 = none, mild/patchy, moderate, or severe. t Grades 0 to 3 = none, focal, 25 to 50% of lobe, or > 50% of lobe.

TABLE 2

The design of the study Streptococcus pneumoniae (GB05) Groups 1 and 2, with ligation Groups 3 and 4, without ligation Pneumolysin 800 ng Groups 5 through 7, with ligation Groups 8 and 9, without ligation 200 ng Groups 10 and 11, with ligation Groups 12 and 13, without ligation Modified toxin Trp 433 > Phe Groups 14 and 15, with ligation Tyr 384 > Phe Groups 16 and 17, with ligation Control animals Groups 18 and 19, heat-inactivated pneumolysin with ligation Groups 20 through 22, PBS with ligation Group 23, cage controls

1 Day

1 Wk

+ +

+ +

+ +

+ +

+ +

+ +

+

+

+

+

+ +

+ + +

1 Mo

+

+

Definition ofabbreviations: Trp 433 > Phe = tryptophanat position433 substituted withphenylalanine; Tyr 384 > Phe = tyrosineat position 384 substituted with phenylalanine; PBS = phosphate-buffered saline.

Feldman, Munro, Jeffery et al .: Effect of Pneumolysin in the Rat Lung In Vivo

at 1 day and at 1 wk. A further group with ligation, given pneumolysin 800 ng, was studied at 1 mo. Modified pneumolysin (groups 14 to 17). The modified toxins (with lowered hemolytic activity or lowered ability to activate complement) were prepared in a final concentration of 40 ILg/ml (a 20-ILl bolus contains 800 ng of toxin), based on the known protein concentration of the preparations. Trp 433 > Phe was confirmed to have a lowered hemolytic activity of approximately 1 X io HU/mg protein, and Tyr 384 > Phe was confirmed to have 70 % reduced ability to activate complement. TWo groups of animals prepared with partial ligation were studied for each toxin; one group was killed at 1 day and one at 1 wk. Control-operated animal groups (groups 18 to 23) . Two groups ofrats (groups 18 and 19) with ligation were studied after administration of heat-inactivated recombinant pneumolysin (800 ng); one was killed at I day and the other at 1 wk. Three groups of rats were injected with 20 JLl of sterile PBS after partial ligation of the apical lobe bronchus and were killed at I day, 1 wk, and 1 mo (groups 20 through 22).

Figure 1. Upper panel: Partialligationof the apical lobe bronchus and intrabronchial injection of either 2 x 10" pneumococci or 800 ng active pneumolysin alone(illustrated) induced an inflammatoryinfiltrate of both lung parenchyma and airwaysecretions (arrows) and the widespread flooding of alveolar spaces characteristic of lobar pneumonia. (Hematoxylin and eosin stain; original magnification: X200.) Lower panel: Histology of ligated apical lobe from an animal given 800 ng heat-inactivated pneumolysin as a control procedure. Alveoli and major conducting airways (C) are clear of secretionsand inflammatory cells. (Hematoxylin and eosin stain; original magnification: x200.)

419

Normal rats whose age was equivalent to 1 wk postoperation, but on which no operative procedure had been performed, were also killed (group 23) . Statistics The data approximated to a normal distribution, and Student's t test was used to compare the results of pneumolysininjected animals with those from heat-inactivated pneumolysin controls, and pneumococcal-injected animals with PBS control. For each group, animals with partial ligation of the apical lobe bronchus were compared with animals without ligation .

Results The lungs from 138 rats were entered into the study (23 groups of six animals). A severe pneumonia, which was localized to the apical lobe, developed in those animals injected with S. pneumoniae (GB05). Pneumolysin injection alone induced identical histologic changes in the rat lung

420

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 5 1991

(Figure 1, upper panel), while the alveoli and major conducting airways remained clear of fluid, secretions, andinflammatory cells (Figure 1, lower panel) after injection of heatinactivated pneumolysin. Five animals died within 24 h of the operation, having not fully recovered from the anesthetic. This was probably due to the operative procedure rather than the treatment administered. S. pneumoniae (GB05) infection was frequently fatal. Of 31 animals infected by S. pneumoniae (20 p.l of 5 x 105 to 5 X 106 CFU/ml), seven died as a result of the infection (two with partial ligation, five with no ligation). Thus, there was no difference between the partial ligation and nonligation groups with respect to animal mortality. The mortality rate was higher with an organism inoculum increased to 20 p.l of 1 x lOs CFU/ml (six animals lived, four died). The infection was frequently associated with an initial bacteremia (four of four animals at 1 day with partial ligation and two of four animals at 1 day without ligation). No animals surviving 1 wk or longer had positive blood cultures. No animals died after the injection of pneumolysin into the apical lobe bronchus. The changes induced in all ex-

periments appeared to be confined to the apical lobe, and all other lobes appeared to remain macroscopically normal. The mean values (± SEM) of histopathologic score for the "overall tissue response" and for "pneumonitis" are shown in Tables 3 through 5. Mean values for overall tissue response (Table 3) and pneumonitis (Table 4) were highest in the animals either injected with bacteria or the highest concentration of the pneumolysin. The maximum overall histopathologic score achieved for anyone section was 32 out of a possible 66. The histopathologic scores were consistently more severe in those animals with partial ligation of the airway than in those without ligation (Table 5). The changes induced by pneumolysin in animals with partialligation of the bronchus were greater at the higher concentration of toxin. The changes induced were resolving by 1 wk, and in the animals injected with 800 ng of pneumolysin there had been complete resolution with no residual damage after 1 mo. In the PBS control group, there were mild histologic changes present at 1 day. Twoof the animals in the group had high values (of 10 and 15) due to perivascular lymphoedema,

TABLE 3

Mean histologic scores for overall tissue response* PL Heat-inactivated PL PL PL HA PL c' SP PBS Cage Controls (2 X ](1 +L; (+L; Day of (800 ng +L; (800 ng +L; (200 ng +L; (800 ng +L; (800 ng +L; (-L; Sacrifice Groups 18 and 19) Groups 5 to 7) Groups 10 and 11) Groups 14 and 15) Groups 16 and 17) Groups 1 and 2) Groups 20 to 22) Group 23)

1 day I wk

5.8 (1.0)

20.8 t (5.1)

I5.5t (2.0)

12.2* (2.7)

14.3§ (3.9)

15.511~

(2.9)

8.8 (1.3)

6.9 (0.9)

11.3 (3.9)

6.0 (1.8)

5.5 (1.0)

7.8 (1.6)

6.7 (0.7)

4.8 (0.8)

1 mo

5.4 (0.7)

Definition of abbreviations: PL = mutant pneumolysin (Trp 433

=

3.0 (0.4)

5.2 (0.7)

pneumolysin; SP

=

Streptococcus pneumoniae; + L = partial ligation of right upper lobe bronchus; - L = no ligation; HA c' = mutant pneumolysin (Tyr 384 > Phe) with reduced ability to activate complement;

> Phe) with reduced hemolytic activity;

= phosphate-buffered saline. * Each score represents the mean score (SEM) of six rats. t p < 0.05 compared with PL heat-inactivated group.

PBS

*p

= 0.06 compared with PL heat-inactivated group.

§

=

p

0.07 compared with PL heat-inactivated group.

II P = 0.06 compared with PBS control. , P

< 0.01

compared with cage control.

TABLE 4

Mean histologic scores for pneumonitis * PL Heat-inactivated PL PL PL HA PL c' SP PBS Cage Controls (+L; (-L; Day of (800 ng +L; (800 ng +L; (200 ng +L; (800 ng +L; (800 ng +L; (2 x 1(j4 +L; Sacrifice Groups 18 and 19) Groups 5 to 7) Groups 10 and 11) Groups 14 and 15) Groups 16 and 17) Groups 1 and 2) Groups 20 to 22) Group 23)

1 day 1 wk 1 mo

0.8 (0.5)

6.7t (2.1)

4.0t (1.0)

1.5 (1.0)

5.7* (2.3)

4.2 (1.6)

1.3 (0.8)

0 (0)

2.3 (1.9)

0.5 (0.5)

0 (0)

0.8 (0.5)

0.3 (0.2)

0 (0)

0.1 (0.2)

Definition of abbreviations: see Table 3. * Each score represents the mean score (SEM) of six rats. t P < 0.05 compared with PL heat-inactivated group. P = 0.07 compared with PL heat-inactivated group.

*

0 (0)

o (0)

Feldman, Munro, Jeffery et al.: Effect of Pneumolysin in the Rat Lung In Vivo

421

TABLE 5

Mean histologic scores for overall tissue response and pneumonitis in animals without partial ligation of the bronchus* Groups 8 and 9

Groups 12 and 13

Groups 3 and 4

PL

PL

PL

PL

SP

(800 ng, OTR)

(800 ng, P)

(200 ng; OTR)

(200 ng; P)

(2 x ]()4, OTR)

1 day

8.0t (2.1)

3.3 (1.1)

8.2 t (1.2)

2.2 (0.7)

7.8 t (1.7)

2.2 (0.8)

1 wk

4.5 (0.6)

0 (0)

5.8 (0.7)

0 (0)

2.7t (0.9)

0.2 (0.2)

Day of Sacrifice

SP

(2

x ur. P)

Definition of abbreviations: OTR =: overall tissue response; P =: pneumonitis. For other abbreviations, see Table 3. * Each score represents the mean score (SEM) of six rats. t P < 0.05 compared with animals with partial ligation, all of which had consistently higher values than those shown above.

focal alveolar congestion, edema, and hemorrhage. There was evidence of goblet cell hyperplasia, luminal secretions, and mild perivascular cell infiltrate. These changes had resolved completely by 1 wk. Heat-inactivated toxin did not produce significant changes compared with the PBS or untreated controls at any of the time points. Significant changes were seen at 1 day in both the Trp 433 > Phe modified toxin and the Tyr 384 > Phe modified toxin, but these changes were less than the wild-type recombinant toxin preparation at the same concentration and the changes had completely resolved at 1 wk. This was particularly the case in the pneumonitis score of the group treated with the mutant pneumolysin of reduced hemolytic activity. The relatively low pneumonitis score but high overall tissue response seen in this group was due to more perivascular and peribronchial infiltrate and the presence ofluminal bronchial secretions and intra-alveolar foamy macrophages. The alcian blue/periodic acid Schiff combined stain demonstrated the presence of a moderate excess of mucins in the airway lumen of test animals, usually acidic rather than neutral. The majority of airway and alveolar content did not stain and is likely to represent exudate. Airway epithelium was invariably intact, pseudostratified, ciliated, and columnar in the pneumococcus-infected and pneumolysin-treated groups.

Discussion Studies of pneumococcal pneumonia in laboratory animals have suffered from an inability to reproduce consistently a model of severe infection (18). Several techniques have been used in an attempt to increase the severity of the disease, including the instillation of chemical irritants into the airways and mechanical obstruction of the airways (18). The most practical, and therefore most commonly employed, methods have used both deposition of the organisms as far distally as possible into the airway and the concomitant administration of mucin into the airway (18-20). These techniques have been used successfully in rats; however, the exact mechanism by which they work is uncertain. Although some investigators have suggested that it may be due to obstruction of the bronchus with subsequent atelectasis, this has not been demonstrated histologically in all cases nor was it thought to be the most important mechanism by all researchers (19). It has also been suggested by others that protection of the

pneumococci against defense mechanisms was the reason for the success of the technique (19). However, it is also likely that foreign material, such as mucin (often porcine mucin), may produce inflammation itself, thereby confounding the histologic picture. In the present study using specific pathogen-free male Wistar nits, an animal model was used in which we could successfully reproduce pneumonia without the need to inject additional potentially irritating foreign material into the bronchus. The infection in the animals was frequently fatal, and the illness was often associated with an initial bacteremia. The histologic features appeared to remain localized to the single lobe of the lung, and all other lobes remained macroscopically normal. A number of factors may have contributed to the success of this model. First, we attempted to inject both bacteria and toxin distally into the lung. In separate experiments using the same methodology (results not reported), india ink was visualized on the pleural surface of the apical lobe immediately after injection into the bronchus. This suggests rapid dispersal of the intrabronchial injectate to the small airways and alveoli. Second, it is likely that mucociliary clearance from the lobe was impaired because of narrowing of the bronchus by the external compression of partial ligation, although the apical lobe remained inflated unless the substances under test produced consolidation. The ligation may have delayed clearance of the organism, allowing it to become established. The presence of delayed mucociliary transport may be an important factor in the generation of significant infection and may be one mechanism by which well-known factors (e.g., viruses) predispose to bacterial infective respiratory illnesses. Although the histology in the animals with partial ligation of the airway was significantly more severe than that without ligation, partial ligation was not associated with a higher mortality rate. This suggests that mortality may relate more to systemic invasion by the organism than to the degree of pneumonic changes in the lung, and that partial ligation of the bronchus and the subsequent pneumonia does not influence bacteremia. The injection of pneumolysin induced histologic features identical to those of the bacterial infection, whereas control animals injected with heat-inactivated pneumolysin were unaffected. Interestingly, PBS alone induced mild changes particularly associated with congestion, perivascular lymphoedema, transitory cell infiltrate, and goblet cell hyper-

422

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 5 1991

plasia. These changes subsided by I wk. There was complete resolution of the pneumolysin-induced inflammatory response in those animals killed at 1 mo. It is known that in pneumococcal infections in humans there is usually complete resolution of the inflammatory damage in the lungs, with little residual damage, and it has recently been shown in the rat lung model with Streptococcus sanguis that this will occur, provided the type II pneumocytes escape significant damage (21). Previous work has suggested that pneumolysin can induce inflammation in tissues. Based on its cytolytic effect on eukaryotic cells, the toxin has been shown to cause ocular inflammation in animals (22) and has been suggested as a possible cause of cellulitis in pneumococcal infections (23). There is substantial in vitro evidence that pneumolysin has the potential to induce inflammation such as we have shown in the rat lung, but this is the first in vivo demonstration of this potential. Pneumolysin had a wide range of complex, dose-related interactions with macrophages and monocytes, polymorphonuclear leukocytes, and lymphocytes (7-10). The toxin has been shown to be able to activate the classic complement pathway (11), independent of specific pneumococcal antibodies, and thiol toxins have been shown to cause the generation of leukotrienes from polymorphonuclear leukocytes (24). Although we have shown that the toxin does induce inflammation in the rat lung, the mechanism by which it did so is unknown. In order to confirm that it was not a nonspecific effect of a foreign protein and to investigate its mechanism of action, we studied the toxin after heat inactivation and studied two modified toxins with lowered biologic activity. The heat-inactivated toxin had no residual activity in vitro and did not cause any significant histologic changes. Substituting tryptophan with phenylalanine at position 433 produced a toxin with reduced hemolytic activity but normal ability to activate complement. Although the latter toxin produced less severe changes in the rat lung than did its parent molecules, as demonstrated by the low pneumonitis score, it did cause statistically significant changes at 1 day. This may have been a consequence of its residual hemolytic activity or of its ability to activate the complement cascade, which was identical to that of the native toxin. The latter capacity may represent a further important mechanism for the induction of inflammation by pneumolysin. However, the second mutant toxin (Tyr 384 > Phe) with reduced ability to activate complement produced substantial inflammation. This may be due to either its residual capacity to activate complement or, more likely, to the fact that the other properties of the toxin can induce inflammation independent of this property. In terms of the pathogenesis of pneumococcal disease, recent studies have suggested that bacterial cell wall components are able to elicit inflammation in animal models of both pneumonia and meningitis (25, 26). The mechanism is related to interaction of these cell wall components with both complement-mediated and complement-independent host responses (27). Products of arachidonic acid metabolism, in particular the lipoxygenase pathway, are thought to play an important role in this inflammatory process, and inhibitors of prostaglandin metabolism may reduce inflammation (27). Even components of the cell wall released by the action of penicillin are capable of eliciting such inflammation, and it

has been suggested that in vivo treatment of patients with highly lytic antibiotics may be associated initially with increased inflammation and a worsening of their general condition, despite death of the invading organism (28). Although the role of pneumolysin in vivo in human infections is unknown, the fact that pneumolysin is produced in these situations in significant amounts is suggested by the appearance of antibodies in the serum of such patients (29). Pneumolysin is an intracellular toxin released upon lysis of the organism (3). It has recently been demonstrated that the pneumococcal autolysin may be a further virulence determinant of the pneumococcus (30). Although the mechanisms underlying virulence are likely to be complex and multifactorial, lysis of the organism may be accompanied by release of potentially toxic factors, including both cell wall components and pneumolysin. Large amounts of pneumolysin may also be expected to be released soon after the treatment of patients with lytic antibiotics. It is tempting to suggest that pneumolysin may contribute to inflammation mediated by cell wall components in this situation (28). It is thus possible that the contribution of pneumolysin to virulence of the organism may lie, at least in part, in its ability to produce inflammation in host tissues. The mortality and morbidity from pneumococcal infections remains significant, particularly in patients with certain well-documented negative prognostic features (31). This is the case despite the use of potent antimicrobial agents and the introduction of intensive care unit facilities (32, 33). It has been suggested that the only means of further decreasing the mortality would be via the use of a pneumococcal vaccine to prevent infection (32) or otherwise modify the disease process. The currently available vaccines work efficiently in healthy young adults, but there is some debate as to their efficacy in other categories of high-risk patients (e.g., the elderly or patients after splenectomy) (34). In addition, children younger than 2 yr of age are known to respond poorly to polysaccharide vaccines (35, 36). The current vaccine only contains 23 of the possible serotypes, and there is a need to continually survey those serotypes causing disease. A mechanism of increasing the efficacy of the vaccine may be by conjugation with a protein antigen (37). It would be an advantage if the protein chosen was common to all pneumococci. Pneumolysin is a protein molecule that has excellent antigenicity (38); vaccination with it significantly protects animals from pneumococcal infection (5). If, as we suggest in the present study, it is in itself making an important contribution to the pneumococcal disease process, it appears to be an ideal candidate for a conjugate pneumococcal vaccine. Acknowledgments: Dr. Feldman is supported by the Medical Research Council of South Africa and by the Zoutendyk Scholarship Trust. Dr. Munro is supported by Lilly Industries, Basingstoke, United Kingdom. Dr. Boulnois is a Lister Institute-Jenner Research Fellow. Work in London was supported by the National Fund for Research into Crippling Diseases (Action Research), and work in Leicester was supported by a grant from the Medical Research Council of the United Kingdom (Drs. Boulnois and Andrew).

References 1. Austrian, R. 1986. Pneumococcal pneumonia. Diagnostic, epidemiologic, therapeutic and prophylactic considerations. Chest 90:738-743. 2. Austrian, R. 1987. Pneumococcal infections. In Harrison's Principles ofInternal Medicine. lith ed. E. Braunwald, K. J. Isselbacher, R. G. Peters-

Feldman, Munro, Jeffery et al.: Effect of Pneumolysin in the Rat Lung In Vivo

3. 4. 5. 6. 7. 8. 9. 10.

11. 12. 13.

14.

15. 16. 17. 18. 19.

dorf, J. D. Wilson, J. B. Martin, and A. S. Fauci, editors. McGraw-Hill, New York. 533-537. Johnson, K. K. 1987. Cellular location ofpneumolysin. FEMS Microbiol. Lett. 2:243-245. Bernheimer, A. W. 1976. Sulphydryl activated toxins. In Mechanisms in Bacterial Toxinology. A. W. Bernheimer, editor. John Wiley and Sons, New York. 85-97. Paton, J. C., R. A. Lock, and D. J. Hansman. 1983. Effect ofimmunization with pneumolysin on the survival time of mice challenged with Streptococcus pneumoniae. Infect. Immun. 40:548-552. Berry, A. M.,J. Yother, D. E. Briles, D. Hansman, andJ. C. Paton. 1989. Reduced virulence of a defined pneumolysin-negative mutant of Streptococcus pneumoniae. Infect. Immun. 57:2037-2042. Johnson, M. K., D. Boese-Marrazzo, and W. A. Pierce, Jr. 1981. Effects of pneumolysin on human polymorphonuclear leukocytes and platelets. Infect. Immun. 34:171-176. Paton, J. C., and A. Ferrante. 1983. Inhibition of human polymorphonuclear leukocyte respiratory burst, bactericidal activity, and migration by pneumolysin. Infect. Immun. 41:1212-1216. Ferrante, A., B. Rowan-Kelly, andJ. C. Paton. 1984. Inhibition of in vitro human lymphocyte response by the pneumococcal toxin, pneumolysin. Infect. Immun. 46:585-589. Nandoskar, M. A., A. Ferrante, E. J. Bates, N. Hurst, and J. C. Paton. 1986. Inhibition of human monocyte respiratory burst, degranulation, phospholipid methylation and bactericidal activity by pneumolysin. Immunology 59:515-520. Paton, J. C., B. Rowan-Kelly, and A. Ferrante. 1984. Activation of human complement by the pneumococccal toxin pneumolysin. Infect. Immun. 43: 1085-1087. Steinfort, C., R. Wilson, T. Mitchell et al. 1989. The effect of Streptococcus pneumoniae on human respiratory epithelium in vitro. Infect. Immun. 57:2006-2013. Walker, 1. A., R. L. Allen, P. Falmagne, M. K. Johnson, and G. J. Boulnois. 1987. Molecular cloning, characterisation, and complete nucleotide sequence of the gene for pneumolysin, the sulphydryl-activated toxin of Streptococcus pneumoniae. Infect. Immun. 55: 1184-1189. Mitchell, T. J., J. A. Walker, F. K. Saunders, P. W. Andrew, and G. J. Boulnois. 1989. Expression of the pneumolysin gene in Escherichia coli: rapid purification and biological properties. Biochim. Biophys. Acta 1007:67-72. Saunders, F. K., T. J. Mitchell, J. A. Walker, P. W. Andrew, and G. J. Boulnois. 1989. Pneumolysin, the thiol-activated toxin does not require a thiol-group for in vitro activity. Infect. Immun. 57:2547-2552. Lapa e Silva, J. R., D. Guerreiro, B. Noble, L. W. Poulter, and P. J. Cole. 1989. Immunopathology of experimental bronchiectasis. Am. J.Respir. Cell Mol. Bioi. 1:297-304. Guerreiro, D., J. Rohde, H. Todd, M. Sheppard, and P. J. Cole. 1990. Proliferation ofbronchial epithelium in experimental bronchiectasis. Thorax 45:781. Pennington, J. E. 1985. Animal models of pneumonia for evaluation of antimicrobial therapy. J. Antimicrob. Chemother. 16:1-6. Nungester, W. J., and L. F. Jourdonais. 1936. Mucin as an aid in the experimental production of lobar pneumonia. J. Infect. Dis. 59:258-265.

423

20. Wood, W. B., Jr. 1941. Studies on the mechanism of recovery in pneumococcal pneumonia. I. The action of type specific antibody on the pulmonary lesion of experimental pneumonia. J. Exp. Med. 73:20~-222. 21. Rhodes, G. C., J. W. Tapsall, and A. W. J. Lykke. 1989. Alveolar epithelial .responses in experimental streptococcal pneumonia. J. Pathol. 157:347-357. 22. Johnson, M. K., andJ. H. Allen. 1971. Ocular toxin of the pneumococcus. Am. J. Ophthalmol. 72:175-180. 23. Peters, N. S., S. J. Eykyn, andA. G. Rudd. 1989. Pneumococcal cellulitis: a rare manifestation ofpneumococcaemia in adults. J. Infect. 19:57-59. 24. Bremm, K. D., W. Konig, P. Pfeiffer et al. 1985. Effect ofthiol-activated toxins (streptolysin 0, alveolysin, and theta toxin) on the generation of leukotrienes and leukotriene-inducing and -metabolizing enzymes from human polymorphonuclear granulocytes. Infect. Immun. 50:844-851. 25. Tuomanen, E., H. Liu, B. Hengstler, O. Zak, and A. Tomasz. 1985. The induction of meningeal inflammation by components of the pneumococcal cell wall. J. Infect. Dis. 151:859-868. 26. Tuomanen, E., R. Rich, andO. Zac. 1987. Induction of pulmonary inflammation by components of the pneumococcal cell surface. Am. Rev. Respir. Dis. 135:869-874. 27. Tuomanen, E., B. Hengstler, O. Kak, and A. Tomasz. 1986. The role of complement in inflammation during experimental pneumococcal meningitis. Microb. Pathog. 1:15-32. 28. Tuomanen, E. 1988. Partner drugs: a new outlook for bacterial meningitis. Ann. Intern. Med. 109:690-692. 29. Jalonen, E., J. C. Paton, M. Koskela, Y. Kerttula, and M. Leinonen. 1989. Measurement of antibody responses to pneumolysin - a promising method for the presumptive aetiological diagnosis of pneumococcal pneumonia. J. Infect, 19:127-134. 30. Berry, A.IM., R. A. Lock, D. Hansman, andJ. C. Paton. 1989. Contribution of autolysin to virulence of Streptococcus pneumoniae. Infect. Immun. 57:2324-2330. 31. Seaton, A., D. Seaton, and A. G. Leitch. 1989. Pneumonia. In Crofton and Douglas's Respiratory Diseases. 4th ed. Blackwell Scientific Publications, Oxford. 285-345. 32. Austrian, R., and J. Gold. 1964. Pneumococcal bacteremia with especial reference to bacteremic pneumococcal pneumonia. Ann. Intern. Med. 60:759-776. 33. Hook, E. W., III, C. A. Horton, and D. R. Schaberg. 1983. Failure of intensive care unit support to influence mortality from pneumococcal bacteremia. JAMA 249: 1055-1057. 34. Austrian, R. 1984. A reassessment of the pneumococcal vaccine. N. Engl. J. Med. 310:651-653. 35. Siber, G. R., P. H. Schur, A. C. Aisenberg, S. A. Weitzman, and G. Schiffman. 1980. Correlation between serum IgG-2 concentration and the antibody response to bacterial polysaccharide antigens. N. Engl. J. Med. 303:178-182. 36. Mayon White, R. T. 1989. Pneumococcal vaccine. Thorax 43:345-348. 37. Lin, K. T., and C. J. Lee. 1982. Immune response of neonates to pneumococcal polysaccharide-protein conjugates. Immunology 46:333-342. 38. Lock, R. A., J. C. Paton, and D. Hansman. 1988. Comparative efficacy of pneumococcal neuraminidase and pneumolysin as immunogens protective against Streptococcus pneumoniae. Microb. Pathog. 5:461-467.