Animal Models of Streptococcus pneumoniae Disease

Abstract

Summary: Streptococcus pneumoniae is a colonizer of human nasopharynx, but it is also an important pathogen responsible for high morbidity, high mortality, numerous disabilities, and high health costs throughout the world. Major diseases caused by S. pneumoniae are otitis media, pneumonia, sepsis, and meningitis. Despite the availability of antibiotics and vaccines, pneumococcal infections still have high mortality rates, especially in risk groups. For this reason, there is an exceptionally extensive research effort worldwide to better understand the diseases caused by the pneumococcus, with the aim of developing improved therapeutics and vaccines. Animal experimentation is an essential tool to study the pathogenesis of infectious diseases and test novel drugs and vaccines. This article reviews both historical and innovative laboratory pneumococcal animal models that have vastly added to knowledge of (i) mechanisms of infection, pathogenesis, and immunity; (ii) efficacies of antimicrobials; and (iii) screening of vaccine candidates. A comprehensive description of the techniques applied to induce disease is provided, the advantages and limitations of mouse, rat, and rabbit models used to mimic pneumonia, sepsis, and meningitis are discussed, and a section on otitis media models is also included. The choice of appropriate animal models for in vivo studies is a key element for improved understanding of pneumococcal disease.

INTRODUCTION

Streptococcus pneumoniae (the pneumococcus) is a major human pathogen that colonizes the upper respiratory tract and causes both life-threatening diseases such as pneumonia, sepsis, and meningitis and milder but common diseases, like sinusitis and otitis media (91). Colonization of the nasopharynx by S. pneumoniae begins early on in life and peaks (up to 55%) at the age of 3 years (31). Colonization not only provides the basis for pneumococcal horizontal spread but is also necessary for invasive disease (31). The pneumococcus produces a plethora of virulence factors that allow bacteria to spread from the upper to the lower respiratory tract, leading to pneumonia and invasive disease (126). The latest report from the Centers for Disease Control and Prevention in the United States estimated that in 2006 pneumococci caused 41,400 invasive infections and 5,000 deaths, most of which were due to bacteremic pneumonia (55). However, each year at least 1 million children younger than 5 years of age die of pneumonia and invasive diseases in developing countries (177). Mortality rates for pneumonia and meningitis are especially high for young children, the elderly, and immunocompromised individuals (55), including patients with human immunodeficiency virus (135).

Pneumococcal diseases are treated with β-lactams (penicillin G, amoxicillin, cephalosporins), respiratory fluoroquinolones, macrolides, and vancomycin (159). However, efforts to treat pneumococcal diseases have been complicated by increasing resistance to antibiotics. A recent work reported high rates of resistance for some antimicrobials such as penicillin (34.2%), trimethoprim-sulfamethoxazole (31.9%), and erythromycin (29.5%) (74). For the prevention of pneumococcal disease, there is a licensed 23-valent polysaccharide vaccine which is effective against 23 out of 90 pneumococcal serotypes but does not confer protection in children younger than 2 years of age or in the elderly. A 7-valent conjugate vaccine has also been available since 2000 and is effective in children, although its cost is high and protection is induced only against serotypes included in the vaccine formulation (32). Both 9- and 11-valent conjugate vaccines will be licensed in the near future (16). Protein-based or protein subunit vaccines common to different capsular serotypes represent attractive alternatives to improve protection and to address the current limitations of polysaccharide vaccines (16).

ANIMAL MODELS OF PNEUMOCOCCAL DISEASE

Animal experimentation is an essential tool for the study of infectious diseases. Numerous animal models of diseases caused by S. pneumoniae are currently available for clarifying mechanisms of disease pathogenesis, testing novel drugs and vaccine candidates, and characterizing the role of bacterial and host factors. Recent review articles have specifically addressed the value of animal models to test pneumococcal protein vaccines (225) and the utility of murine models of pneumonia to assess the efficacy of antimicrobials (175). Some authors have examined mouse genetic susceptibility to pneumococcal disease (121), while others have discussed the value of knockout mice to investigate the pathophysiology of bacterial meningitis (191). In 1999, Giebink effectively reviewed the utility of the chinchilla model for the study of middle ear (ME) pneumococcal infection (90), while Sabirov and Metzger have lately evaluated the importance of the mouse to test mucosal vaccines against otitis media (211). A recent review on animal models of invasive pneumococcal disease focused on the importance of sepsis and pneumonia in experimental animal models for studying vaccine efficacy and also provided very useful considerations on technical aspects (39). Meningitis and otitis media models were mentioned by the authors, although to a lesser extent (39). The present report is a more comprehensive review of animal models that have significantly contributed to our knowledge of all the clinically relevant pneumococcal diseases, highlighting the importance of readouts, defining the advantages and the drawbacks, and suggesting potential applications of each one of these systems. Models of colonization are briefly mentioned throughout the text but not discussed in detail. The reader interested in animal models of pneumococcal colonization should refer to a recently published review (157). We fully appreciate the fundamental importance of nasopharyngeal carriage as a crucial step for invasive disease, but we have chosen to focus on animal models of acute disease, such as pneumonia, sepsis, meningitis, and otitis media. For easier reading, major experimental animal models are listed in tables according to the pneumococcal disease and the animal species. Representative models were chosen based on historical importance, innovative improvement of preexisting models, widespread use within the pneumococcus research community, and usefulness for future studies.

Considerations of Animal and Pneumococcal Strains

The choice of both animal and bacterial strains should be carefully considered before approaching the study of pneumococcal disease in vivo. It is beyond the scope of this review to include a detailed discussion on animal or S. pneumoniae strains to be used in experimental models of disease; however, we provide general considerations on the subject.

Mouse and rat inbred strains afford more-uniform responses to experimental treatments due to their tightly controlled immune system. For this reason, they have extensively been employed as models for infection, vaccination, and drug efficacy studies. Outbred mouse strains are instead generated to maintain maximum heterozygosity. Their phenotypic diversity, despite being lower than that of humans, may be important to mimic the natural variation in response to infection. In addition, their lower cost makes them attractive alternatives to inbred strains.

Experimental pneumococcal disease has been investigated by use of different rodents and the rabbit, but undoubtedly the mouse represents the most commonly used animal model. Inbred mouse strains, including BALB/c, C57BL/6, DBA, and CBA mice, have largely been exploited for the analysis of pneumococcal pneumonia (12, 113, 131, 176, 186, 201, 234), sepsis (15, 42, 43, 163, 186), meningitis (86, 95, 136, 173, 230), and otitis media (162, 164). Experimental sepsis has also been induced in immunodeficient mice, such as the CBA/N strain, which carries the Xid-linked immune deficiency (38, 43). CBA/N mice are unable to produce antibodies against pneumococcal polysaccharides and phosphocholine and are thereby highly susceptible to infection by S. pneumoniae (41). However, outbred mice, such as MF1, CD-1 (Swiss), Swiss-Webster, and NMRI, have increasingly become more popular. They have widely been used for analyzing pathogenicity mechanisms of pneumonia (50, 58, 114, 120, 125, 131), sepsis (109, 131, 251), and meningitis (59, 95, 160, 268). In addition, outbred mice have also been employed to evaluate active and passive protection against pneumonia and sepsis (4, 115, 163) and the efficacy of antibiotic therapy against invasive pneumococcal disease (13, 141, 169, 220, 223). The host genetic factors controlling the response to infection by S. pneumoniae are still unknown. Mouse susceptibility and resistance to pneumococcal disease have been investigated in more detail in several studies (72, 92, 130, 215) and discussed in a few reviews (41, 121). The rat has also been exploited for studying invasive pneumococcal disease and otitis media. Outbred strains, including Wistar, Sprague-Dawley, and CD (derived from Sprague-Dawley) rats, have been extensively utilized as experimental models (2, 49, 71, 79, 101, 103, 145, 147, 166, 193, 214, 221, 242, 253). The New Zealand White rabbit is the most commonly used strain to induce pneumococcal pneumonia, sepsis, and meningitis (26, 57, 69, 98, 106, 119, 196, 246). Finally, the chinchilla (only outbred strains are available), gerbil, and guinea pig are commonly used animal models of experimental otitis media (81, 89, 90, 260).

There is large variation in the outcomes of infection by S. pneumoniae in experimental animal models, depending on the pneumococcal strain used. It is well accepted that the virulence of S. pneumoniae strains is strongly influenced by the capsular serotype in both humans and mice (38, 129, 215); however, the genetic background of the strain also seems to be relevant for disease development (29, 129, 215). Certain human isolates are poorly virulent in vivo, whereas other pneumococcal strains that are highly virulent in mice have less relevance for human disease (38, 39). Typically, strains of capsular serotypes 2, 3, 4, 5, and 6 are virulent in mice, whereas strains of types 14, 19, and 23 are relatively avirulent (21, 39). The low virulence of many pneumococcal strains in rodents can be balanced by either using larger challenge doses (40) or neutropenic hosts (19, 49, 169, 221, 223).

Experimental pneumococcal disease has been induced with both historical laboratory strains and clinical isolates. The use of classic strains, able to generate consistent features of disease, allows comparison of results among research groups. Of particular relevance is the type 2 D39 reference strain (142), which several research groups have employed to study the pathogenesis of pneumococcal disease and test vaccine antigens over the years (25, 40, 50, 92, 103, 104, 109, 123, 124, 131, 180-182, 186). Serotype 3 strains, in particular A66 (9) and WU2 (43), are also widely used for inducing sepsis in the mouse and the rat (2, 40, 43, 109, 129, 163) and to study invasive pneumococcal disease in the rabbit (246, 247). More recently, laboratory strains with sequenced genomes, such as the type 4 TIGR4 strain (238) and the type 19F G54 strain (75), have also been employed by different research groups (58, 59, 80, 131, 187, 197, 215). The availability of clinical isolates is also important in attempting the reproduction of human disease in laboratory animals. Several groups have preferred serotype 3 and 6 strains isolated from patients with pneumococcal meningitis and have successfully induced this disease in the mouse, the rat, and the rabbit (26, 35, 69, 76, 85, 95, 145, 192, 268, 269). Clinical isolates of serotypes 3, 6, 9, and 19 have also been employed in laboratory animal models to elucidate features of pneumococcal pneumonia and sepsis (12, 14, 40, 49, 196, 223, 234, 251). For experimental otitis media, strains of serotypes 3, 4, 6, 7, 9, 11, 14, 19, and 23, which are the ones most commonly isolated from children with acute otitis, have been used (18, 80, 81, 89, 101, 154).

ANIMAL MODELS OF PNEUMONIA

S. pneumoniae is the leading causative agent of community-acquired pneumonia (159). In 2006, bacteremic pneumonia constituted more than 66% of invasive pneumococcal disease in the United States (55). Typically, pneumococcal pneumonia is lobar. However, other radiographic patterns are also commonly found, including lobular bronchopneumonia and occasionally interstitial or mixed types of pneumonia (127, 188). The proportion of pneumococcal pneumonia associated with bacteremia is estimated to be approximately 20 to 25% (7). The pathogenesis of pneumococcal pneumonia is a complex interplay between pneumococcal virulence and the host immune response (91). Since the late 1970s, when an important model of pneumococcal pneumonia was set up for the rat (14), many different animal models of pneumonia have been developed. Pneumonia due to S. pneumoniae has been studied largely by use of the mouse, probably the most commonly employed system today for the characterization of this disease, due to the ease of manipulation and data reproducibility in large numbers of animals.

Mouse Models

Different research groups have reproduced the features of pneumococcal pneumonia in the mouse (Table 1). Mouse models of pneumonia allow the analysis of different parameters, including animal survival after infection, the presence of bacteria in lungs and blood, levels of inflammation, and histology of lung tissue. Additionally, quantification of antibody titers and antimicrobials performed in vaccine and drug pharmacokinetic studies, respectively, are also feasible. Two main routes of infection are currently used to induce pneumonia: the intratracheal (i.t.) and the intranasal (i.n.). The i.t. model requires a complex and invasive technique for disease induction but offers the advantage of allowing 99% delivery of the bacterial inoculum to the lungs (208). The i.n. route includes both the standard aspiration method (50) and the aerosol nebulizer system (176). The model of infection through i.n. aspiration is the most commonly used, as it is fast and easy to perform without invasive surgical procedures and also because it mimics the natural route of infection in humans. The i.n. aerosol model instead requires an exposure chamber with a nebulizer, but it is suitable for the simultaneous infection of many mice. A key aspect of i.n. aspiration models is the need for anesthesia to allow infection of the lower respiratory tract (58), in comparison to absence of anesthesia in both colonization (157, 258) and i.n. aerosol (113, 176) pneumonia models. When animal sedation is required, the choice of anesthetic should also be evaluated for experimental design and result interpretation. It has been reported that pentobarbital and halothane have distinct effects on the pathogenesis of pneumococcal pneumonia in mice (207).

TABLE 1.

Mouse models of pneumonia

Route of infection	Mouse strain(s)	Pneumococcal serotype(s)	Aim(s) of study	Readouts	Utility	Reference(s)
i.t.	C57BL/6	3	Antibiotic efficacy, model description	CFU, histology, survival, pharmacokinetics	Drug efficacy and pharmacokinetics	12
i.t.	ddY	NSa	Gastric juice in pneumonia	Survival, CFU, PMNb, lung histology	Pathogenesis of pneumonia in the elderly	114
i.n.	MF1	2	Role of pneumolysin and autolysin, model description	CFU, lung histology	Host-parasite interaction, disease pathogenesis, vaccine and drug studies	50
i.n.	Neonatal and infant NMRI	1	Immunization in early life	Antibody titers, CFU	Neonatal and early-life vaccinology	115
i.n.	BALB/c and OF1	14 (for BALB/c) and 3, 9 (for OF1)	Therapy in neutropenic mice	CFU, survival, pharmacokinetics	Antibiotic efficacy, immunotherapy studies	201, 223
i.n.	CBA/J	3, 11, 19	Model of penicillin-resistant pneumonia	Survival, CFU, histology, pharmacokinetics	Antibiotic efficacy in immunocompetent mice	234
Aerosol nebulizer	CBA/J	28	Model of chronic obstructive pneumonia	CFU, serum immunoglobulin G	Disease pathogenesis	113
Aerosol nebulizer	CBA/J	3	Low-dose model, drug efficacy	Survival, CFU, histology	Efficacy of tissue-directed antimicrobials	176

^aNS, not specified.

^bPMN, polymorphonuclear cells.

i.t. models.

The i.t. pneumonia model was first described by Azoulay-Dupuis et al. to test the efficacies of different antibiotics (12). It is based on an inoculation technique previously described for the study of pneumonia caused by Staphylococcus aureus and Klebsiella pneumoniae (78). Briefly, mice were suspended vertically on a board by supporting the lower incisors on a wire loop and retaining the lower incisors with a rubber band. The oropharynges were transilluminated, and the tracheas were cannulated with needles attached to microliter syringes containing the pneumococcal suspension (12). This model is still used in several studies on drug efficacy, host response to infection, and the role of pneumococcal virulence factors in the disease (1, 11, 19, 168, 169, 208, 237). Another i.t. mouse model, based on pneumococcal inoculation via the oropharynx, was developed by Iwasaki et al. to investigate the combined effects of bacteria and gastric juice commonly observed for elderly patients with aspiration pneumonia (114). Finally, studies on neutrophil migration after induction of pneumonia were carried out by injection of bacteria directly into the exposed trachea (167, 232).

All i.t. infection models have the advantage of allowing the delivery of the entire bacterial inoculum into the lower respiratory tract and directly causing pneumonia without intermediate pathogenic steps. By using different mouse strains, it is possible to induce either acute or subacute forms of disease (13). Swiss mice develop acute pneumonia and die rapidly (2 to 4 days), while C57BL/6 animals develop a subacute progressive disease leading to death in 8 to 10 days (13).

i.n. models.

In the late 1980s, two studies demonstrated the importance of pneumolysin (25) and autolysin (24) in pneumococcal virulence by using in vivo models, including an i.n. pneumonia model based on bacterial instillation into the mouse nostrils. However, the first detailed i.n. aspiration model was described by Canvin et al. in 1995, who introduced the practice of infecting animals with mouse-passaged pneumococci (Table 1) (50). Pneumococci recovered from the blood of intraperitoneally (i.p.) infected animals were inoculated into the nostrils of anesthetized mice, which rapidly developed bronchopneumonia with bacteremia (50). This model has been employed largely in studies on host immune response to pneumococcal lung infection (23, 70, 122, 123, 130), mouse susceptibility and resistance to the disease (92), invasion by S. pneumoniae of bronchoepithelial cells in vivo (124), the role of virulence determinants in pneumonia (30, 58, 120, 125, 158), and the efficacy of antibiotics and anti-inflammatory drugs (250, 252). Furthermore, many investigations on protection from i.n. challenge after immunization with vaccine candidates have been performed. Pneumolysin (4), PspA (40, 257), PspC (15), and phosphorylcholine (241) have successfully been employed in vaccine studies followed by i.n. challenge. Whereas PsaA has been able to confer protection against colonization (37, 183), it did not elicit protection against focal pneumonia (40). PsaA has also been reported to be unable to provide significant protection against i.p. challenge (180). In passive-immunization studies, capsular polysaccharide antibodies obtained from vaccinated infants were also able to protect mice against lung infection (213). The protective efficacies of novel protein targets and live attenuated pneumococci (with deletions in genes encoding the capsule, pneumolysin, and PspA) have also been tested as vaccine candidates in this type of pneumonia challenge model (100, 205). Of importance for vaccine studies was also an early-life infection model (115). As children younger than 2 years of age are major targets of pneumococcal vaccination, the mouse model of i.n. infection with aspiration set up for adult animals (50) was adjusted to neonatal (1-week-old) and infant (3-week-old) mice, which mimic the immune systems of human neonates and infants, respectively. By using this approach, several aspects of vaccinology in early life were investigated, including the efficacies of polysaccharide conjugate vaccines delivered by the mucosal route (115), the role of T-cell responses to pneumococcal conjugates (116), and the importance of maternal antibodies for protecting offspring against S. pneumoniae (204).

Other i.n. aspiration models of pneumonia have been developed for mice rendered neutropenic after treatment with cyclophosphamide: Ramisse et al. described the role of human immunoglobulins in passive and active immunotherapy against pneumonia (201), while Soriano et al. focused on the pharmacokinetics of penicillins and cefotaxime in the disease (223). While both systems described above employ immunosuppressed animals, Tateda et al. established an i.n. model of penicillin-resistant pneumococcal pneumonia in immunocompetent CBA/J mice (234). This experimental system represents the first example of penicillin-resistant pneumonia in uncompromised mice, which are generally quite insensitive to infection by penicillin-resistant strains (169). The last model was employed both for therapeutic studies (229, 233) and for analyzing the host inflammatory response during pneumonia (228).

Finally, experimental pneumococcal pneumonia in mice can also be induced by the i.n. aerosol route using an exposure chamber and a nebulizer. The technique less closely resembles the development of pneumococcal pneumonia in humans, where the disease generally follows aspiration of bacteria from the upper respiratory tract. The first example of aerosol-induced pneumococcal pneumonia originates from an early work by Coil et al., who described that splenectomized mice were more susceptible to type 3 pneumococci delivered by means of an aerosolized suspension (61). Another attempt at an i.n. aerosol pneumonia model was made to analyze the development of pneumonia after airway obstruction in mice persistently colonized by S. pneumoniae (113). This model intends to mimic the occurrence of pneumonia in humans with altered mucociliary clearance mechanisms, chronic obstructive pulmonary disease, or concomitant viral infections (113). However, a more exhaustive model was established by Nuermberger et al. (176). They used a low bacterial inoculum to generate a murine disease with a resemblance to human pneumonia that was closer than those seen for traditional mouse i.n. and i.t. models, which rely on larger doses to produce fulminant infections due to concomitant sepsis (12, 50). This model is especially suitable for evaluating the efficacy of antimicrobials at the site of infection rather than in the bloodstream, as the experimental approach allows the development of an indolent subacute pneumonia (175). The main disadvantage of this model is the use of neutropenic mice, which might not optimally represent the human host (175, 176).

Remarks on murine pneumococcal pneumonia.

i.t. models generally cause lobar pneumonia (12), while systems based on i.n. aspiration of bacteria induce bronchopneumonia (50). In both cases, there are concomitant pneumonia and sepsis: infection readily leads to death of animals, which most likely succumb to septic shock rather than to pulmonary disease. Clearly, models of focal pneumonia in the absence of generalized sepsis are more helpful to study the pathogenesis of pulmonary disease, vaccine protection, and antibiotic efficacy exclusively in the lungs. In addition, they better reproduce the clinical manifestations in humans, where pneumonia is usually not associated with bacteremia (7). The use of pneumococcal strains of serotypes 14, 19, and 23, which are poorly virulent in mice, is a necessary requisite for causing this type of disease. So far, type 19 strains have been employed to induce progressive focal pneumonia without sepsis (40, 234). Alternatively, neutropenic mice infected with low doses of virulent pneumococcal strains (e.g., type 3) also develop primary pneumonia with delayed dissemination (approximately 4 days after infection) (176).

Rat and Rabbit Models

Rat and rabbit models are uncommonly employed compared to the mouse. Their larger size allows the collection of more-substantial samples, but one drawback is that experimental groups are normally smaller, thus potentially compromising statistical significance. Sophisticated surgical procedures are required to induce pneumonia via i.t., intrabronchial, or intrapulmonary routes of infection, and to our knowledge, no i.n. models of pneumococcal pneumonia have been set up with these animals.

Table 2 summarizes the principal characteristics of rat models of pneumococcal pneumonia. Main readouts include animal survival, observation of gross pathology, lung histology, and bacterial counts in lungs and blood. The rat has been employed mainly for clarifying features of pneumonia occurring in asplenic, neutropenic, cirrhotic, or alcoholic subjects, thus representing a valuable experimental system for studying pneumonia in the immunocompromised host. Alcoholism is a predisposing factor for the development of pneumococcal pneumonia (140), and mortality rates from bacteremic pneumonia double (up to 40%) in alcoholic patients with hepatic insufficiency (8). The models in ethanol-intoxicated (71) and cirrhotic (166) rats were developed over 15 years ago by using the i.t. route of infection to deliver pneumococci into the left lobe of the lung following intubation of the main stem bronchus (14). Infection caused lobar pneumonia and sepsis (14). The model in ethanol-intoxicated animals was based on continuous ethanol feeding leading to chronic intoxication (71), while experimental cirrhosis was induced in rats by administration of a hepatotoxin that caused cirrhosis and ascites (166). In both cases, rats showed increased susceptibility to pneumococcal pneumonia. These models have been used by the same research group in studies on phagocytosis and antibiotic efficacy (84, 184, 199). A sophisticated surgical method based on the delivery of bacteria into the apical lobe bronchi of adult rats was described in the early 1990s (79). Briefly, rats were anesthetized and intubated before the thoracic cage was opened to reveal the apical lobe of the lung. The apical lobe bronchus was exposed and partially sutured to impair mucociliary clearance, and the inoculum was injected through the bronchial wall into the lumen toward the lung periphery (79). Histological examination of rat lungs showed clear consolidation of the apical lobes (lobar pneumonia) (79). Two other models of lobar pneumonia induced either intrabronchially in infant rats (221) or intrapulmonarily in adult rats (49) rendered neutropenic were employed to study the efficacy of antibiotics against penicillin-resistant lobar pneumonia (49, 221, 222). As already described for the mouse, the use of immunocompromised animals is necessary to produce pneumonia by penicillin-resistant strains of S. pneumoniae in rats. The above-described infant rat model is based on a nonsurgical intrabronchial instillation of bacteria resuspended in cooled melted agar (as an adjuvant and growth medium) (221). Another group developed a refined intrapulmonary pneumonia model for infant rats based on the nonsurgical inoculation of a low bacterial inoculum (1 to 400 CFU per rat) (214). The right chests of pups were prepared with ethanol, and rat-passaged bacteria were delivered transthoracically into the mid-right lungs. Prior to inoculation, pneumococci were entrapped in cooled agar particles to provide both protection from clearance and adjuvant activity. The virulences of 10 different serotypes prevalent in children were compared (214). Despite being a model of focal pneumonia, it differs from the one described for the mouse by Briles et al. (40) in that experimental pneumonia may be accompanied by bacteremia and occasionally by meningitis, mimicking the disease seen for children of developing countries. The model was utilized in both neonatal (4-day-old) and infant (3-week-old) rats to assess passive protection against intrapulmonary challenge with S. pneumoniae (155, 156, 224).

TABLE 2.

Rat models of pneumonia

Route of infection	Rat strain	Pneumococcal serotype(s)	Aim(s) of study	Readouts	Utility	Reference(s)
i.t.	R strain albino	3	Antimicrobial therapy, model description	CFU, histology	Pathogenesis of lobar pneumonia, drug efficacy studies	14
i.t.	Sprague-Dawley	3	Model in cirrhotic and chronically intoxicated rats	Survival, CFU, hematology and chemistry tests	Pneumonia in alcoholic and cirrhotic patients	71, 166
Intrabronchial	Wistar	3	Effect of pneumolysin on lungs	Lung histology	Pathogenesis of lobar pneumonia, detailed surgery information	79
Intrabronchial	Infant CD	6	Model of penicillin-resistant pneumonia in neutropenic rats	Gross pathology and CFU	Drug efficacy studies	221
Intrapulmonary	CD	2, 6B	Model of penicillin-resistant pneumonia in neutropenic rats	CFU, lung histology, pharmacokinetics	Antibiotic efficacy studies	49
Intrapulmonary	Infant Sprague-Dawley	1, 3, 4, 5, 6B, 7F, 9V, 14, 19F, 23F	Low-dose pneumonia model	Survival, CFU, lung histology	Pathogenesis of pneumonia in children, vaccine studies	214

The rabbit has also been employed to study pneumococcal pneumonia, although fewer models are available. Readout parameters include animal survival, determination of leukocyte numbers in lungs and blood, histology of the lungs, and assessment of drug concentration in serum (Table 3). All models described below are based on i.t., intrabronchial, or intrapulmonary routes and are useful systems to study disease pathogenesis and perform drug efficacy studies. However, all of them necessitate surgical expertise for disease induction. An early i.t. model was established in 1987 and it is based on the instillation of pneumococci in rabbits by means of a catheter inserted into the animal trachea (246), as previously described for experimental pneumonia caused by S. aureus (185). By using this approach, analyses of the roles of pneumococcal cell surface components in the induction of pulmonary inflammation (246), of host oligosaccharides in preventing pneumococcal colonization of the nasopharynx and subsequent lung infection (110), and of the platelet-activating factor in the pathogenesis of pneumonia (47) were performed. Two further models were developed for immunocompetent rabbits based on intrabronchial instillation of bacteria (57, 196). The former reproduces ventilator-associated pneumonia (57), while the latter mimics penicillin-resistant pneumonia (196). In the model of antibiotic-resistant pneumonia, pathology—if untreated—was characterized by initial unilobar lesions and subsequent consolidation of other lobes and confluent bronchopneumonia with bacteremia (196). The system mimics lethal pneumococcal pneumonia in patients and allows simulation in the rabbit of human antibiotic pharmacokinetics. As this model simulates “human-like treated pneumonia,” it constitutes an extremely valuable tool for assessing drug efficacy and pharmacokinetics in humans (67, 195, 196), and it may also be useful for evaluating the selection of resistant pneumococcal strains in vivo (68). As aforementioned, the same research group also developed a novel infection model based on the induction of ventilator-associated pneumococcal pneumonia in adult rabbits (57). This approach proved to be instrumental to assess antibiotic efficacy against pneumonia associated with mechanical ventilation, which is a serious health concern in intensive care units (56). Finally, positron emission tomography was also exploited to study intrapulmonary experimental pneumonia localized in the upper lung lobes of rabbits (119).

TABLE 3.

Rabbit models of pneumonia

Route of infection	Rabbit strain	Pneumococcal serotype(s)	Aim of study	Readouts	Utility	Reference
i.t.	New Zealand White	2, 3	Inflammation induced by cell wall	CFU, protein content, leukocytes in lungs	Disease pathogenesis, efficacy of antimicrobial and adjunctive therapy	246
Intrabronchial	New Zealand White	9, 19	Model of penicillin-resistant pneumonia	Survival, CT scanb, lung histology, pharmacokinetics	Efficacy and pharmacokinetics of drugs (model of human therapy)	196
Intrabronchial	New Zealand White	NSa	Ventilator-associated pneumonia	CFU, lung histology	Pneumonia in ICUc	57
Intrapulmonary	New Zealand White	NS	Neutrophil activity in lungs	FDGd uptake in lungs	Acute and chronic lung inflammation in pneumonia	119

^aNS, not specified.

^bCT scan, computerized tomography scan.

^cICU, intensive care units.

^dFDG, ¹⁸F-labeled fluoro-2-deoxy-d-glucose.

ANIMAL MODELS OF SEPSIS

The definition of sepsis and related terms (e.g., bacteremia, septicemia, severe sepsis, septic shock) has been controversial. In 1989, sepsis was defined as the host systemic response to infection according to specific measurements of clinical parameters (e.g., blood leukocyte counts, body temperature, and heart and respiratory rates), in contrast to bacteremia, which is defined only according to the presence of bacteria in the blood (33). S. pneumoniae is an important cause of sepsis in humans, generally occurring as a form of an advanced stage of pneumonia (60). Predisposing conditions are splenectomy (219), hepatic cirrhosis, respiratory insufficiency, and cardiovascular disease (140). In children of less than 3 years of age, a common clinical manifestation is occult bacteremia (without a focus) (153). In discussions of experimental animal models developed to study pneumococcal disease, the terms sepsis and bacteremia are commonly found and used interchangeably. Experimental sepsis by S. pneumoniae can be generated either by inoculating bacteria directly into the bloodstream or by injection into the peritoneal cavity. While sepsis induced by the i.p. route is secondary to peritoneal infection, with the concomitant strong peritoneal inflammation, intravenous (i.v.) inoculation produces a “cleaner” model, where the clearance of bacteria by the reticuloendothelial system can be easily studied. In addition, i.p. infection represents neither an important clinical manifestation of invasive pneumococcal disease nor the natural route of infection by S. pneumoniae in humans.

Mouse Models

Besides a few early studies carried out in guinea pigs to analyze the role of spleen and complement in experimental pneumococcal sepsis (44, 45), the mouse is the most extensively used animal model for inducing sepsis by S. pneumoniae. Most research groups have analyzed the occurrence of murine sepsis principally by determining the presence of bacteria in the blood and by observing animal survival after infection (Table 4). Sepsis in the mouse can be induced by either the i.p. or the i.v. route of infection. Sepsis can also develop secondary to pneumonia induced by i.n./i.t. routes or to meningitis following intracranial injection, but the kinetics of bacterial transit from lungs or brain into the bloodstream cannot be controlled experimentally. Consequently, performing viable counts in the blood after i.n./i.t. or intracranial inoculations does not constitute a direct measure of bacterial clearance operating in those districts and it is not the finest approach to evaluate immune mechanisms and efficacy of drugs or vaccines in experimental pneumonia or meningitis. As above mentioned, the i.v. model is a direct and valuable system for inducing infection in the bloodstream and thus for studying the mechanisms of bacterial clearance, although the technique can be time-consuming and difficult to perform compared to i.p. infection. To facilitate i.v. injection of bacteria into the mouse tail vein, researchers have used an infrared lamp to allow vasodilatation (109). i.p. inoculation is technically easier but presents disadvantages including the risk of causing tissue damage or pain to the animal and may not allow the study of bacterial clearance by the reticuloendothelial system (38). Briles et al. observed that certain pneumococcal strains that were lethal following injection of mice by the i.p. route were avirulent by the i.v. route, indicating that i.p.-induced sepsis is much more severe (38, 43). For this reason, the i.p. route is widely used to mouse passage pneumococci in order to render them more virulent (50).

TABLE 4.

Mouse models of sepsis

Route of infection	Mouse strain(s)	Pneumococcal serotype	Aim(s) of study	Readouts	Utility	Reference
i.v.	CBA/N, DBA/2, BALB/c	3	Protection by anti-PCa antibodies, model description	Survival, CFU	Bacterial clearance mechanisms, passive and active protection	43
i.v.	CD-1	3	Host response to sepsis	CFU, hematological parameters, cytokines, histology	Pathogenesis of sepsis	251
i.p.	CBA/N, BALB/c	3	Protection by anti-PC antibodies, model description	Survival, CFU	Passive and active protection, drug efficacy and virulence studies	43

^aAnti-PC, anti-phosphocholine.

Studies performed by the group of Briles et al. in the early 1980s employed the i.v. mouse sepsis model mainly for vaccine studies. These works largely contributed to demonstrate S. pneumoniae clearance from the bloodstream promoted by naturally occurring anti-phosphocholine antibodies (43, 163, 261). The model was employed and is still used by this group and others, resulting in several investigations on the role of pneumococcal virulence factors during sepsis and in studies aimed at analyzing the efficacy of vaccines (15, 20, 30, 38, 42, 109, 131, 186, 200, 202, 227, 245). For instance, thanks to the use of the i.v. infection route, it was possible to unambiguously establish the role of PspC as a key virulence factor in pneumococcal sepsis (109). In a work by Loeffler et al., the therapeutic value of a novel antimicrobial, the Cpl-1 phage enzyme, was proven by using an i.v. mouse sepsis model (149). One of the most detailed studies of the pathogenesis of sepsis in the mouse was made by Wang et al., who investigated the systemic host responses to sepsis in normal and immunosuppressed mice infected by the i.v. route (251). A complete analysis of different parameters, such as bacterial loads in blood and lungs, lung injury, biochemical markers, and hematological and inflammatory mediators, was carried out (251).

The i.p. route of infection is one of the earliest techniques employed to induce experimental sepsis. This method was used by Briles et al. in early studies of blood clearance mechanisms (43). i.p. challenge models are still widely employed in studies on the efficacies of vaccine candidates, including the surface proteins PspA, PspC, and PsaA; a toxoid derivative of pneumolysin (PdB); the caseinolytic protease (ClpP); the components of two iron uptake transporters, PiaA and PiuA; and the more recently discovered pilus protein subunits (46, 51, 87, 118, 180, 181). The i.p. model also proved to be useful to clarify the role of pneumococcal virulence factors in systemic infections (182, 262) and to evaluate efficacy of antibiotics (53, 141). The i.p. sepsis model was also employed to assess passive protection conferred by a mouse hyperimmune serum (263) and by human anticapsular (117) and anti-pneumolysin antibodies (171). Passive immunization with anticapsular antibodies in infant (13- to 15-day-old) mice demonstrated that anticapsular opsonophagocytic titers are better predictors of protection than are enzyme-linked immunosorbent assay immunoglobulin G titers (117). Very few studies focused on the host response to pneumococcal sepsis by describing the function of tumor necrosis factor alpha in controlling systemic infection (22, 178) and increasing the permeability of the blood-brain barrier (BBB) (244).

Rat and Rabbit Models

Experimental models of sepsis in the rat and the rabbit have been used mainly from the beginning of the 1970s throughout the 1980s, but presently they have mostly been replaced by the mouse due to the latter's greater ease in manipulation and smaller size. Similarly to that induced in the mouse model, sepsis induced in the rat and rabbit models is studied by observing animal survival over time and by determining the presence and number of bacteria in the blood (Table 5). As in the case of rat models of pneumococcal pneumonia (Table 2), rat models of sepsis are particularly valuable for studying the disease in immunocompromised animals. The first model of sepsis in the rat was set up in 1972 by Leung et al., who infected splenectomized rats by injecting pneumococci into the tail veins (147). Despite being an early work, the authors made significant observations, including the importance of (i) the use of i.v. rather than i.p. challenge to show a difference between normal and immunocompromised animals in susceptibility to infection and (ii) vaccination to protect splenectomized rats from fatal sepsis (147). The model was also employed to evaluate the efficacies of penicillin and steroid drugs (99), of human gamma globulins (179), and of a type 6 capsular polysaccharide vaccine (111) to treat and/or prevent pneumococcal postsplenectomy sepsis. Starting from a previously described model of pneumococcal pneumonia in cirrhotic rats (166), a similar experimental strategy was applied to investigate pneumococcal sepsis in cirrhotic rats (2). Patients with alcohol-induced hepatic cirrhosis are predisposed to pneumococcal infection, partially due to decreased liver production of complement factors. In this model, cirrhotic rats are also hypocomplementemic compared to healthy animals (2, 166). By using this approach, a detrimental role of pneumolysin was observed in both depleting complement components and reducing serum opsonic activity (3), resulting in enhanced pneumococcal virulence in the cirrhotic host (2). To our knowledge, there are no reports of i.p. models of sepsis in the rat.

TABLE 5.

Rat and rabbit models of sepsis

Route of infection	Animals	Pneumococcal serotype	Aim(s) of study	Readouts	Utility	Reference
i.v.	Sprague-Dawley rats	25	Model description for splenectomized rats	Survival, CFU	Sepsis in immunocompromised hosts	147
i.v.	Sprague-Dawley rats	3	Role of complement in cirrhotic rats	Survival, CFU	Sepsis in cirrhotic patients	2
i.p.	New Zealand White rabbits	1	Role of lysosomes in sepsis	Lysosomal enzymes in plasma and organs, CFU, vital parameters	Clinical and metabolic parameters during sepsis	98
Intracardiac	New Zealand White rabbits	3	Role of spleen in clearance, model description	CFU, splenic mass, and blood flow	Sepsis in splenectomized patients	106
i.v.	Chinchilla rabbits	NSa	Pharmacokinetics of cephalosporins	CFU, drug concentrations in plasma	Drug efficacy studies	82

^aNS, not specified.

One of the first models of pneumococcal sepsis in the rabbit was described in 1970 by Guckian et al., and it represents a fine example of analysis of clinical parameters (e.g., body temperature, cardiac output, arterial blood pressure) detected during the course of this disease (98). Infection was established by the i.p. route, and the model was employed to analyze the roles of lysosomes and cathepsin inhibitor in plasma (98), metabolic alterations in the course of disease (96), and the effects of splenectomy and drug therapy in the disease (97). An intracardiac model of experimental sepsis in New Zealand White rabbits was utilized to investigate the key role played by the spleen during the disease (106). This represents another example of refined surgery and measurements of clinical parameters in the rabbit in the course of infection. Effects of hemisplenectomy, total splenectomy, spleen transplant, and spleen repair on the clearance of S. pneumoniae during experimental sepsis were evaluated (62, 106). Finally, an i.v. model of sepsis was established in chinchilla rabbits to study the pharmacokinetics of cephalosporins (82).

ANIMAL MODELS OF MENINGITIS

S. pneumoniae is one of the major infectious agents responsible for acute bacterial meningitis, which can be fatal in 5% to 40% of patients (137, 146). Up to 30% of survivors suffer from neurological sequelae (e.g., learning, hearing, and memory impairment, seizures, and motor deficits) due to brain damage (146). In patients with meningitis, brain injury can manifest with several patterns, including vasculitis, cortical necrosis, and neuronal apoptosis in the hippocampus (137, 146). Development of meningitis generally initiates from the colonization of the nasopharynx by S. pneumoniae, which reaches the lungs and then invades the bloodstream with subsequent crossing of the BBB (146). Meningitis can also be caused either by contiguous spread of pneumococci infecting the sinuses or the ME or by accidental traumatic inoculation of bacteria into the central nervous system (CNS) (137). Experimental animal models are essential tools to study the pathogenesis of meningitis and the efficacies of different drugs against the disease. If the mouse is a key laboratory animal for the induction of experimental pneumonia and sepsis, rabbits and rats represent the counterparts for the study of experimental meningitis.

Mouse Models

As previously mentioned, the mouse has become an experimental system for studying pneumococcal meningitis only recently, while most studies between the mid-1970s and the 1990s were carried out in the rat and the rabbit. Two major types of S. pneumoniae murine meningitis models exist: (i) direct infection by the intracerebral or the intracisternal (i.cist.) route and (ii) infection induced by the i.p. or i.n. route. Direct bacterial inoculation into the CNS mimics the contiguous spread of bacteria from the nasopharynx or the ME or the inoculation of pneumococci into the brain due to trauma. This system allows the study of host-pathogen interactions once infection is established in the meninges but does not enable the analysis of the different pathogenic steps occurring from colonization to disease in the CNS (137). In contrast, meningitis induced via i.n. or i.p. routes is useful for the analysis of pathogenesis according to the natural way of infection but has the drawback that approximately 50% of infected animals will die due to sepsis without ever developing meningitis (243, 268). Characterization of meningitis in the mouse model has been performed by analyzing different parameters, including animal survival; assessment of clinical scores; viable bacterial counts in the brain, cerebrospinal fluid (CSF), and other organs; histological analysis of the brain tissue; and determination of leukocyte and cytokine levels in CSF and serum (Table 6). One disadvantage presented by the mouse model is the difficulty of collecting CSF samples (only up to 10 μl can be obtained) due to the small size of the animal (52).

TABLE 6.

Mouse models of meningitis

Route of infection	Mouse strain(s)	Pneumococcal serotype(s)	Aim of study	Readouts	Utility	Reference
Intracerebral (ventricular)	CD-1	NSa	Drug efficacy	CFU in brain and organs	Therapeutic studies	220
Intracerebral (right forebrain)	C57BL/6	3	Model description	Immunohistochemistry, CFU, clinical scores, behavioral tests	Meningitis pathophysiology, drug and virulence studies	86
i.cist. (cisterna magna)	C57BL/6	3	Role of eNOSb	Survival, clinical scores, cytokines, and histology	Meningitis pathophysiology, hearing loss, therapeutic studies	136
Intracerebral (subarachnoidal)	MF1	2, 3, 4	Model description	Histology, survival, CFU	Virulence studies, host-parasite interaction	59
i.cist. (cisterna magna)	Infant BALB/c C57BL/6, CD-1	3	Model of brain damage	Survival, CFU in CSF, histology, and apoptosis	Meningitis in children, brain damage	95
i.p.	Infant C57/129	6	Role of ICAM-1	Survival, CFU in organs, cytokines	Hematogenous meningitis, drug studies	230
i.n.	Swiss	6	Model description	Brain histology, CFU in CSF, cytokines	Pathogenesis via natural infection, drug studies	268
i.n.	Swiss	2	Model description	CFU in brain and other organs	Disease pathogenesis via natural infection	160

^aNS, not specified.

^beNOS, endothelial nitric oxide synthase.

Direct induction of meningitis.

The meningitis model of Gerber et al., described in 2001 and based on a previously established Listeria monocytogenes meningoencephalitis model (218), is an attractive system for analyzing pneumococcal meningitis following the infection of inbred mice into the right frontal lobe of the brain (86). Meningitis was studied by performing bacterial counts for different organs, histology of the brain, clinical score calculations, and behavioral tests. This model has been employed to study the efficacy of rifampin against meningitis (173) and has also been used for analyzing host and pneumococcal factors involved in disease pathogenesis (35, 76, 198, 255). In another work by the same research group, mice surviving pneumococcal meningitis presented with clear deficits in motor skills, spatial memory, and learning when subjected to behavioral tests (e.g., Morris water maze test, tightrope test) (254). Therefore, this mouse model also enables the assessment of postinfectious sequelae (254). The induction of experimental meningitis in the mouse was also accomplished by inoculating pneumococci directly into the cisternae magnae (i.cist.) of anesthetized C57BL/6 mice (136). By use of this model, Koedel et al. described the role of nitric oxide synthase in the disease, and the same system has been used to study innate immune responses to infection, including cytokine expression patterns and the role of caspase-1, MyD88, and complement factors (133, 134, 136, 138, 209). As all mice infected by the i.cist. route presented with hearing loss, this system allowed the establishment of the first experimental example of pneumococcal meningitis-associated hearing loss in the mouse (133). The two models described above finely resemble pneumococcal meningitis in humans and offer the opportunity to study the pathophysiologic changes during meningitis, the effects of drug administration, and postinfection neurological sequelae.

Most mouse models of meningitis employ inbred mice, but studies employing outbred strains also exist. Shapiro et al. induced meningitis by the intracerebral-ventricular route in outbred CD-1 mice and studied the therapeutic efficacy of clinafloxacin (220). Another model of meningitis in outbred MF1 mice was developed by our research group by inoculating S. pneumoniae strains of different capsular serotypes via the intracranial-subarachnoidal route (59). The establishment of meningitis was confirmed by histological analysis of the brain, and animal survival and bacterial counts in different organs were also evaluated. Because levels of disease development and outcomes were comparable among strains of serotypes 2, 3, and 4 (59), this model may be instrumental for screening the virulences of different serotypes and possibly for assessing attenuation of isogenic mutants. The model is based on a modified version of a method describing the induction of cryptococcal meningoencephalitis in the mouse (28). Recently, a novel model of brain damage in pneumococcal meningitis has been described for infant (11-day-old) mice of both inbred (C57BL/6, BALB/c) and outbred (CD-1) strains: all mice infected via the i.cist. route rapidly developed the disease and some animals also exhibited brain damage (95). Among the mouse strains tested, BALB/c mice showed the largest susceptibility to infection, while C57BL/6 mice were the most resistant and the ones with major neuronal injury both in the cortex and in the hippocampus. The model was established from previous studies of brain injury to infant rats (143) and is promising for the study of neuronal damage in children (95).

Hematogenous meningitis.

Tan et al. developed one of the first i.p. meningitis model for inbred infant mice in order to study the role of intercellular adhesion molecule 1 (ICAM-1) in this disease (230). Similarly, two additional murine models of meningitis via the i.p. route were described later on by Iizawa et al., who analyzed the efficacy of cefozopran against the disease (112), and by Tsao et al., who focused on elucidating the role of tumor necrosis factor alpha in regulating the opening of the BBB during experimental meningitis (243). In the latter model, antibiotics were given soon after infection to avoid mouse death due to septic shock before the development of meningitis.

The model of meningitis induced by the i.n. route mimics the natural way of infection. The work of Kostyukova et al. represents one of the first attempts to analyze the roles of different pneumococcal virulence factors (e.g., hyaluronidase, capsule, and pneumolysin) in the pathogenic steps leading to meningitis following i.n. infection (139). However, the first detailed model of hematogenous meningitis should be accredited to a work conducted by Zwijnenburg et al., who infected Swiss mice via the i.n. route using a mixture of bacteria and purified pneumococcal hyaluronidase to favor systemic invasion from the nasopharyngeal mucosa (268). As aforementioned, the disadvantage of this system relies in the fact that only 50% of animals developed the disease (268). The same research group largely used the above-described model to study the role of interleukin-1 (IL-1), IL-10, and IL-18 in the disease (264, 266, 267). Recently, the model was also employed to assess the therapeutic effects of the inhibitor of complement factor C1, which was effective at promoting bacterial clearance and reducing both mouse illness and inflammation in the CNS (265). Finally, the model described by Marra and Brigham is also based on i.n. infection of Swiss mice with pneumococci (without hyaluronidase) (160). The course of meningitis and bacterial counts in brains, lungs, and blood following infection with different challenge doses were evaluated in that study (160).

Nonhematogenous meningitis.

A study conducted by van Ginkel et al. reported a novel in vivo route for pneumococci to enter the CNS along olfactory nerves: nasal colonization by S. pneumoniae resulted in bacterial entry into the brain via retrograde axonal transport without causing bacteremia (249). To our knowledge, this is the first report describing a nonhematogenous meningitis caused via i.n. infection of mice.

Rat and Rabbit Models

Rat and rabbit models are still currently employed by several research groups to study pneumococcal meningitis. They rely on the induction of disease via the i.cist. route, with the exception of two models of either hematogenous or otologic meningitis developed in the rat (206, 253). As already mentioned for the pneumonia models in rats and rabbits, more-sophisticated surgical procedures are here employed to cause the disease; however, animal size has the advantage of allowing large samples of and multiple sampling for CSF and blood compared to what is possible with the mouse model. Except for an otologic model (253), none of the methods used to inoculate these animals mimic the natural route of infection occurring in humans, but they enable the study of disease once infection is established in the meninges (Table 7). The diseases induced in the two animal species differ in some aspects, including brain damage. Apoptosis in the dentate gyrus and cortical necrosis are the predominant features of damage in the rabbit (269) and in the adult rat (103), respectively. Conversely, the infant rat is unique, as it reproducibly manifests both hippocampal apoptosis and cortical necrosis (143), similar to what is observed for patient brain damage (137, 146).

TABLE 7.

Rat and rabbit models of meningitis

Route of infection	Animal	Pneumococcal serotype(s)	Aim(s) of study	Readouts	Utility	Reference(s)
i.cist.	Infant Sprague-Dawley rat	3	Drug efficacy, MMPa in brain damage	Drug pharmacokinetics, histology, CFU, MMP in CSF	Therapeutic studies, host immune response, brain damage	145, 242
i.cist.	Wistar rat	3, 6	Microvascular changes	ICPb, rCBFc, brain edema	Meningitis pathophysiology	193
i.cist.	Wistar rat	2	Damage to brain ependyma	Ciliary beat frequency, CFU, CSF parameters, SEMd of brain	Virulence studies, host response to infection, brain damage	103
i.p.	Infant Sprague-Dawley rat	3	Pathogenesis of inner ear invasion	CFU in blood and CSF, histology of ME and brain	Meningitis secondary to ear infection, drug efficacy studies	206
Otologic, i.p.	Wistar rat	2	Cochlear implantation in meningitis	Gram staining, histology, CFU counts	Post-cochlear implantation meningitis, therapeutic studies	253
i.cist.	New Zealand White rabbit	3	Model description, drug study	β-Lactam concentrations in CSF	Disease pathophysiology, host response, drug and virulence studies	69
i.cist.	New Zealand White rabbit	3	Hearing assessment	Hearing loss, histology of inner ear and brain	Hearing loss due to meningitis, drug efficacy	26

^aMMP, metal metalloproteinases.

^bICP, intracranial pressure.

^crCBF, regional cerebral blood flow.

^dSEM, scanning electron microscopy.

i.cist. models for rats.

Both infant and adult rats have been employed extensively in different studies on pneumococcal meningitis and represent valuable tools for studying disease characteristics, host immune response to infection, and efficacy of antimicrobials and anti-inflammatory drugs (Table 7). Infant (6- to 12-day-old) rats are generally more susceptible than adults to pneumococcal infection, and despite their smaller size, repeated collection of CSF and blood is still feasible. As aforementioned, another advantage of infant compared to adult rats is the possibility of inducing brain damage both in the hippocampus (apoptosis) and in the cortex (necrosis) (143). Thus, the infant rat represents a robust and reliable laboratory animal to study brain damage (27, 143, 145).

In the early 1990s, the i.cist. infant rat meningitis model was used by different groups, mainly for therapeutic studies (242, 259). In 1990, Tsai et al. established an important i.cist. model in infant Sprague-Dawley rats for the study of cefepime pharmacokinetics (242). However, the inoculation technique was fully described later in studies conducted by Leib et al. (145) and is based on a previously described model for meningitis caused by group B streptococci (132). Briefly, pups were infected i.cist. with 10 μl of bacterial suspension by use of a 32-gauge needle and then monitored for seizures due to injection (132, 145). Development of meningitis was proven by culturing bacteria present in the CSF (132, 145). The meningitis model in the infant rat has largely been used by Leib and coworkers in studies on the involvement of matrix metalloproteinases (143, 145, 165) and endothelin (194) in neuronal damage. By means of this model, the efficacies of several therapeutic approaches against meningitis and/or postinfectious sequelae, including antioxidants (6), adjunctive therapy drugs (27, 143, 144, 194), and antimicrobials such as daptomycin (94) and phage lysins (93), have been evaluated. It is interesting that in this experimental system, the use of dexamethasone as an adjuvant therapy increased hippocampal cell injury and reduced the learning capacities of rats (144). More recently, the same model was also employed to analyze the key role of choline in the pathogenesis of meningitis and brain damage (83).

The adult rat model of pneumococcal meningitis was originally described in detail in a work carried out to study the microvascular alterations during the establishment of meningitis (193). Wistar rats were anesthetized, tracheotomized, and artificially ventilated before the insertion of catheters into the cisternae magnae. A craniotomy was then performed in the right parietal bone to place a probe for measuring regional cerebral blood flow and intracranial pressure, and once these parameters reached stable baseline levels, meningitis was induced by injecting live pneumococci in a volume of 75 μl (192, 193). The method was instrumental to assess the detrimental effect of the oxidant peroxynitrite in the disease (128) and the function of transforming growth factor β2 in suppressing cerebrovascular alterations and brain edema in early experimental meningitis (192). Another adult rat i.cist. model was developed by Hirst et al., who investigated the effects of pneumococcal infection on cerebral ciliated ependyma during experimental meningitis. An accurate description of the animal surgical preparation is given, based on the implantation of a 19-gauge cannula into the cisterna magna, which allowed both pneumococcal infection and CSF sampling during animal follow-up. A detailed report on treatment and analysis of CSF, blood, and brain tissue is also provided (103). This technique was successfully employed to show that S. pneumoniae caused loss of and damage to ependymal cells and ciliary functions, which may contribute to several neuropathological aspects of meningitis (103). Recently, the same research group also showed that both autolysin and pneumolysin play key roles in the pathogenesis of pneumococcal meningitis (104).

Hematogenous and otologic models for rats.

Hematogenous and otologic models for rats were established to study meningitis secondary to sepsis or otitis media. In the model of Rodriguez et al., infant Sprague-Dawley rats were inoculated by the i.p. route, and some of the animals infected developed meningitis (about 50%) and cochlear inflammation (20 to 80%). This model may be useful to study the pathogenesis of inner ear invasion by S. pneumoniae during hematogenously acquired experimental meningitis (206). Recently, another model was set up in adult rats inoculated via both otologic (inner ear and ME) and i.p. routes of infection to establish whether cochlear implantation increases the risk of pneumococcal meningitis (253). The use of three infection routes covers the potential routes used by S. pneumoniae to reach the CNS: direct spread from the inner ear, contiguous or hematogenous spread from the ME, and invasion of the CNS from the bloodstream. Meningitis was induced in all animals with cochlear implants, regardless of the infection route (253).

i.cist. infection in rabbits.

Rabbit models of pneumococcal meningitis have contributed substantially to our present knowledge of the disease and have significantly helped the evaluation of antimicrobial drug efficacy. The i.cist. rabbit model is an advantageous system that allows accurate analysis of disease features, as well as multiple sampling and large samples of CSF and blood (235). Due to the large quantity of CSF available from this animal, many types of analysis can be performed, providing information on bacterial and leukocyte counts and concentrations of glucose, proteins, and other metabolites (Table 7). A limitation of the model is that experimental groups are generally small due to animal size, representing an issue for the performance of dependable statistical analysis. Another obvious drawback relates to the unnatural method of infection, based on the instillation of relatively large numbers of bacteria into the cisterna magna.

The adult rabbit model, originally developed by Dacey and Sande in 1974, is one of the first important models of pneumococcal meningitis based on an extremely accurate technique for inducing the disease. Briefly, animals were secured in a frame through a dental helmet attached to the skull: their cisternae magnae were then punctured with needles mounted in a geared electrode introducer, and S. pneumoniae was inoculated (69). The use of this model greatly helped the understanding of several aspects of pneumococcal meningitis, including bacterial replication in the CSF, host defenses and inflammation, BBB alteration, and also the systemic effects of meningitis on animal respiration and circulation (236). Numerous studies have been undertaken to evaluate antibiotic efficacy (34, 64, 73, 85, 172, 189, 203, 226, 231) and the role of host and bacterial factors in disease pathogenesis (10, 36, 247). Interestingly, the effectiveness of a bactericidal but nonbacteriolytic antibiotic, such as daptomycin, in clearing pneumococcal infection from the CSF and in reducing inflammation was proven by means of both this rabbit model (226) and a previously described rat model (94). Also, of no less importance is the fact that this rabbit model is a useful tool to analyze damage in the hippocampus due to meningitis (269). Similarly, Lindquist et al. described the establishment of a rabbit model based on the use of a conventional head holder and on the percutaneous puncture by hand of the cisternae magnae of rabbits (148). Finally, Bhatt et al. developed a model of meningogenic labyrinthitis via i.cist. infection of New Zealand rabbits with the aim of assessing hearing loss secondary to the induction of meningitis. Data obtained from this model indicate that the severity of hearing loss strongly correlates with the duration of infection (26). The model largely mimics hearing loss due to meningitis in humans and may be instrumental for testing the efficacies of therapeutic drugs.

ANIMAL MODELS OF OTITIS MEDIA

Otitis media is one of the most common diseases in childhood. It manifests in different forms depending on the duration and type of exudate (e.g., purulent, mucoid, and serous) and it may lead to sequelae, including hearing loss, learning disabilities, and speech impediments (65). S. pneumoniae is the main cause of ME disease in children; other common causes include Haemophilus influenzae and Moraxella catarrhalis. The pathogenesis of the disease depends on different factors including concomitant bacterial and/or viral (e.g., influenza) infection of the upper respiratory tract, impaired immunity, age, genetic predisposition, and dysfunction of the Eustachian tube (65). Different models of pneumococcal otitis media have been developed using the chinchilla, gerbil, guinea pig, rat, and, more recently, the mouse in order to understand different aspects of the disease and test novel antimicrobials and potential vaccine candidates. Description of disease by use of these animal models has been done by observing different parameters, including the presence of bacteria, polymorphonuclear cells, and inflammatory molecules in ME fluid as well as histological analysis of Eustachian tube, the ME, and the inner ear (Table 8).

TABLE 8.

Animal models of otitis media

Route of infection	Animal(s)	Pneumococcal serotype(s)	Aim(s) of study	Readouts	Utility	Reference
Intratympanic	Chinchilla	23	Model description, antibody response	Exudation in ME, antibody titers, CFU	Disease pathogenesis, host response, vaccine and drug studies	89
Intrabullar	Mongolian gerbil, chinchilla	3	Model description	Otoscopy, IEa histology, CFU	Antibiotic efficacy and kinetics, vaccine and virulence studies	81
Intratympanic	Guinea pig	23F	Model description	CFU in ME fluid, ME histology	Drug efficacy studies, disease pathogenesis	239
Intrabullar	Sprague-Dawley rat	3, 6A	Model description	Survival, CFU in ME and blood	Evaluation of drugs and vaccines	101
i.n.	Rivm:WU (CBP) rat	14	Model description, vaccine study	ME histology, TEMb, antibody titers	Immunization and correlates of protection in a natural-route model	248
Intrabullar	BALB/c, C57BL/6 mice	3, 6, 9, 19	Model description	CFU in blood and ear, survival, otomicroscopy	Host inflammatory response, drug and vaccine studies	164
i.n.	BALB/c mice	19F	Prevention of otitis media by phage lysin	Bacterial multiplication in vivo	Otitis media in children, vaccine studies	162

^aIE, inner ear.

^bTEM, transmission electron microscopy.

Chinchilla, Gerbil, and Guinea Pig Models

Historically, the chinchilla has been the preferred model for mimicking pneumococcal otitis media and studying its characteristics (Table 8). Giebink very elegantly reviewed the importance of this animal model, emphasizing how infection can be induced by small bacterial inocula administered either directly into the ME or i.n. (90). One major advantage of this experimental system is its feasibility for the study of disease pathogenesis, the immunogenicity of vaccine antigens, and the efficacy of antimicrobials at the mucosal site without causing disseminated disease (90). In addition, chinchillas have large bullae, which are easily accessible for both inoculation and repetitive sampling of ME fluid (211). Two disadvantages relate to the unavailability of inbred chinchilla strains and the high costs of the animal compared to those of other rodents. The chinchilla model was first described in 1976 by Giebink and colleagues. Acute otitis media was induced by inoculation of pneumococci directly into the ME cavity through the dorsal bulla (89). Pathogenic aspects of otitis media, as well as the immune response evoked after infection, were analyzed (89). Both this group and other research teams have employed this model in several investigations, including characterization of the inflammatory response to infection (174, 216), analysis of the role of pneumococcal virulence factors in disease (217, 240) and of the efficacies and kinetics of antimicrobials against purulent and serous otitis media (48, 107, 108), and vaccine studies (90, 150). As seen for other pneumococcal disease agents, the virulence of S. pneumoniae in experimental otitis media induced in the chinchilla varies largely by serotype (90). A comparative study was recently conducted to analyze the virulences of different pneumococcal strains in chinchillas inoculated via the transbullar route. Fourteen strains of serotypes 3, 4, 6A, 9V, 11, 14, 19F, and 23F were tested in the model, and data showed how even within a single serotype, strains differ in their capabilities to cause local and systemic disease (80). Besides the classic intratympanic route, Giebink et al. also established a chinchilla model of otitis media based on simultaneous i.n. infection of S. pneumoniae and influenza A virus: approximately 70% of animals developed the disease, while only 20% of chinchillas presented with otitis after inoculation with bacteria only (88). This last model more closely resembles the clinical manifestations in humans and is valuable for the study of pathological aspects of the disease as well as the efficacies of drugs and vaccines (90).

The acute otitis media model in the gerbil is also a well-established system for studying disease pathogenesis and comparing the efficacies of different drugs. The anatomies and histologies of Eustachian tubes and MEs in gerbils and chinchillas are similar, with both allowing easy inoculation through the tympanic membrane or the bulla. The model described by Fulghum et al. in a comparative study between gerbils and chinchillas (81) was developed following the method originally used by Giebink et al. (89). Advantages of using the gerbil as a model of ME infection include the availability of inbred strains, the reasonable cost, and the susceptibility of this animal to common pathogens of otitis media, including S. pneumoniae, H. influenzae, and M. catarrhalis (81). Infection was carried out by instilling bacteria directly into the ME by transbullar challenge (81). This system was employed in investigations on the efficacy and pharmacokinetics of β-lactams against otitis media caused by penicillin- and cephalosporin-resistant S. pneumoniae (17, 18, 54, 190). Intrabullar challenge of gerbils with virulent pneumococci was also carried out to investigate the pathogenesis of pneumococcal meningitis secondary to otitis media (160, 170).

Finally, it is also worth mentioning the guinea pig as a model of pneumococcal otitis media (239). Although less used than the chinchilla and the gerbil, it represents a reliable and relatively inexpensive otosurgical animal for which inbred strains and experimental reagents are also available (211). Investigations on the cytotoxic effects triggered by live pneumococci in the cochlea (63) and also therapeutic (105) and vaccine (260) studies were performed.

Rat and Mouse Models

In recent years, interest in both rat and mouse models of otitis media has increased, since these animals—especially the mouse—are less expensive and more characterized in terms of immunological and genetic information than are chinchillas, gerbils, and guinea pigs.

Otitis media has been induced in the rat by two different routes of infection: direct inoculation into the ME and i.n. infection. In a model described in 1988, S. pneumoniae was directly instilled into the tympanic cavities of rats to investigate bacterial replication in the ME and blood (101). Inoculation of rat ME with pneumococci caused otitis media with either purulent or serous effusion, closely resembling the disease in the human host (101). This system has been used to evaluate the efficacy of penicillin for the treatment and prevention of permanent mucosal changes due to pneumococcal acute otitis (5, 102) and to assess protection against experimental otitis media conferred either by anticapsular antibodies (256) or by active immunizations with PspA (256) or whole killed bacteria (66). Moreover, the virulences of pneumococcal strains with different opacity phenotypes (151) and various levels of susceptibility to penicillin (154) were also evaluated by intratympanic challenge of rats. van der Ven et al. described a novel rat model based on the i.n. instillation of bacteria, thus mimicking the natural route of infection of otitis media in humans (248). Histamine was inoculated into Wistar rats [strain Rivm:WU (CBP)] by the intratympanic route in order to impair ciliary activity and contribute to Eustachian tube dysfunction, and bacteria were then delivered i.n. with an infusion catheter. Histamine pretreatment facilitated the ascent of bacteria from the nasopharynx to the ME, allowing assessment of mucosal immunity directly in the Eustachian tube and ME. The model was utilized in a vaccination study to analyze protection against otitis media following immunization with a tetanus toxoid-conjugated type 14 capsular polysaccharide (248). The importance of this model was also revealed in the work of Eriksson and Hellström, who provided clear evidence of acute otitis media development in the rat after multiple i.n. inoculations of pneumococci (77).

Less than 10 years ago, only a few mouse models of pneumococcal otitis media were available (210). Clearly, the small size of the animal together with the poor accessibility of the ME for inoculation and sampling makes the mouse a tricky choice to study otitis media. Two different routes of infection are used in the mouse model: either direct injection (intrabullar or intratympanic) into the ME or i.n. inoculation. Direct microbial injection into the ME warrants reproducibility in disease initiation and development and allows the use of small bacterial inocula, whereas the i.n. route mimics the natural mode of ME infection, but induction of experimental otitis is more sporadic and requires larger doses of bacteria. In 2003, Melhus and Ryan established a model of otitis media caused by S. pneumoniae and H. influenzae in BALB/c and C57BL/6 mice inoculated via the intrabullar route (164). The course of disease was dependent upon both bacterial and mouse strains, with BALB/c being the most susceptible mouse strain. The authors also found that highly virulent pneumococcal serotypes (types 3 and 6) consistently induced lethal sepsis with or without ME infection, while less virulent strains (types 9 and 19) produced otitis (164, 210). Direct inoculation of bacteria into the ME can also be instrumental in assessing inflammation in the ME following the injection of heat-killed pneumococci (152, 161). As acute bacterial otitis media is an inflammatory response to replicating microorganisms, the use of heat-killed bacteria does not mimic the natural disease but still allows measurements of inflammatory parameters in the ME, such as the amount and type of effusions, the thickness of the tympanic membrane, and the recruitment of polymorphonuclear cells (152). The mouse intratympanic/intrabullar model of pneumococcal otitis media found applications in both vaccine (211, 212) and therapy (161) research fields. Finally, a recent work elegantly described the induction and prevention, by means of phage lysins, of otitis media in BALB/c mice stably colonized by bioluminescent S. pneumoniae (162). In the model, recurrent otitis media was induced by infecting the colonized mice via the i.n. route with influenza virus (to favor bacterial spread from the nasopharynx to the ME) and was efficiently treated with the Cpl-1 lysin (162). This novel model of naturally occurring otitis media mimics infection in children and may prove useful in studies on the prevention and treatment of this important disease.

CONCLUSIONS

Pneumococcal disease in humans is varied and multifaceted, and no experimental animal model is able to completely mimic human disease. However, despite its limitations, animal experimentation remains an unquestionably valuable tool to clarify the pathogenic mechanisms of disease. This review provides a perspective on the development and applications of different animal models used to mimic and study the major diseases caused by S. pneumoniae. We have highlighted the progress made over the years, indicating how certain historical models represented milestones for developing novel and more-effective methods, while others are still in use largely to investigate pneumococcal diseases. Some of the techniques for inducing disease were described to point out how thoroughly these animal models were set up. Characteristics of pneumococcal pneumonia, sepsis, and meningitis have successfully been reproduced in the mouse, the rat, and the rabbit. In the case of otitis media, the chinchilla, the gerbil, and the rat have been the preferred animal models throughout the years, although recently researchers have also reproduced the disease in the mouse. All these models were and continue to be helpful tools for elucidating aspects of disease pathogenesis, characterizing innate and adaptive immunities to S. pneumoniae, and testing the efficacies of antibiotics and other therapies as well as potential vaccine candidates. Various infection routes and methods have been employed to induce disease, and we have discussed the importance of different readouts and emphasized the main advantages and limitations of the models. The utility of each animal model is also highlighted to facilitate researchers in the choice of the appropriate system. The mouse is comparatively inexpensive and easy to handle and allows the screening of drugs and vaccines with high statistical power. The commercial availability of plentiful murine experimental reagents and strains, including transgenic and knockout strains, makes the mouse an attractive model to address the interplay between bacteria and host factors. Furthermore, the completion of the mouse genome sequence and the availability of technologies to introduce targeted mutations into the genome provide a powerful and extensive toolkit for studying mouse susceptibility to pneumococcal disease. In contrast, larger animals present the advantage of obtaining larger samples in addition to an increased ease in performing inoculation and surgical procedures, but they are more expensive, and the results obtained may lack statistical significance due to the use of smaller animal groups.

The development of pneumococcal disease depends upon both bacterial (e.g., capsular serotype and other virulence determinants) and host (e.g., genetic background, immune response, age, sex) factors. Thus, in addition to the selection of the appropriate in vivo model, the use of several animal and pneumococcal strains is highly recommended to test a research hypothesis. We also wish to emphasize how in vivo experimentation should preferably be combined with in vitro assays (e.g., cell culture assays) to understand and reproduce essential characteristics of the pathogenesis of infectious diseases and to elucidate key features of host-pathogen interaction. Finally, worth considering is the fact that certain host-microbe interactions are species specific, thus suggesting additional caution to scientists who wish to mimic, in an animal model, pathogenicity mechanisms which occur only in the human host.