The Route Less Taken: Pulmonary Models of Enteric Gram-negative Infection

Abstract

Many pathogens are capable of causing a fulminant infection in pulmonary tissues of mammals. Animal models have provided an extensive understanding of the genetic and molecular mechanisms of bacterial pathogenesis as well as host immune response in the lungs. Many clinically relevant Gram-negative bacteria are host restricted. Thus, the powerfully informative tools of mouse models are not available for study with these organisms. However, over the past 30 years, enterprising work has demonstrated the utility of pulmonary infection with enteric pathogens. Such infection models have increased our understanding host-pathogen interactions in these organisms. Here we provide a review and comparison of lung models of infection with enteric Gram-negative bacteria relative to naturally occurring lung pathogens.

Introduction

Animal models of infection make possible the study virulence determinants of important human pathogens in the context of an active host immune response. Murine model systems in particular offer the advantages of genetic tractability, and a well characterized immune response. Additionally, a multitude of reagents have been developed for these models including mutant mouse strains, antibodies, and cytokines to fully characterize host-pathogen interactions. However, mouse model systems can pose challenges in that infection via a specific route does not always mimic the disease or pathology elicited by an infectious agent in a human host. Therefore, lung models have been developed to provide an alternate approach for investigating enteric bacterial pathogenesis. The rationale for these studies is that, like the intestine, the lung is a site of mucosal immunity. Obviously, there are profound differences between these two mucosal environments. For instance, a bottleneck of colonization in an orogastric infection is the gastric barrier, whereas in intranasal infection the ciliated mucosa lining the upper respiratory tract acts as a physical barrier. Additionally, the pH encountered by a typical gastrointestinal pathogen ranges greatly from acidic in the stomach to basic in the large intestine. The lungs, however, remain uniform in pH. Importantly, lungs are generally a sterile environment and therefore do not mimic the conditions of the GI mucosal environment where competing flora abound, and influence the course of infection (Logsdon & Mecsas, 2006, Magalhaes, et al., 2007). Despite these and other differences, lung models of infection have has served to model some aspects of enteric infections in cases where there is no ideal model system. These models complement cell culture studies of host-pathogen interaction, particularly in identifying virulence factors, determining host response, and developing vaccine strategies.

Naturally occurring lung pathogens models have been developed for genera including Yersinia, Francisella, Bordetella, Pseudomonas, Klebsiella, and Burkholderia and have been characterized extensively (Hoffmann, et al., 2005, Lathem, et al., 2005, Lawlor, et al., 2005, Wiersinga, et al., 2006, Rick Lyons & Wu, 2007, van Gent, et al., 2011). Generally, infection models of natural pulmonary pathogens are characterized by a low lethal dose, systemic spread, and an increase in bacterial burden (Table 1). Characterizations of infections with these organisms provide an important comparator for non-pulmonary pathogens. Here we discuss the models that have been developed for studying enteric pathogens in the context of a pneumonic infection.

TABLE 1.

Characteristics of respiratory bacterial pathogens in murine lung models

Respiratory Organism	LD50	Time of Death	Routea	Disseminationb	Inputc	CFU/lung (24hr)	CFU/lung (End dpi)d	Refs.
F. tularensis	<1×101	5 days	IN/A	Lv, Sp, Ln	2×102	1×105	1×109	(Conlan, et al., 2003, Wu, et al., 2005)
B. pseudomallei	<3×101	3 days	IN/A	Sp, Lv	3×103–1×104	1×106	1×108	(Jeddeloh, et al., 2003, Cuccui, et al., 2007)
Y. pestis	3×102	3 days	IN	Lv, Sp, Ln	Ix104	1×103	1×109	(Smith, 1959, Lathem, et al., 2005)
K. pneumoniae	1×103	3 days	IN/IT	Sp, Tr	1×104	1×106	ND	(Lawlor, et al., 2005, Aujla, et al., 2008)
B. pertussis	>1×105	ND	IN	NDe	1×105	1×105	1×102 (30 dpi)	(Khelef, et al., 1994)
B. bronchiseptica	>5×105	ND	IN	ND	1×103–5×105	5×105	<1 (50 dpi)	(Harvill, et al., 2000)
P. aeruginosa	1×107	1–3 days	IN/IT	Sp	1×107–5×107	1×107	ND	(George, et al., 1993, Power, et al., 2004)

^aA=aerosol, IN=intranasal, IT=intratracheal

^bdissemination to organs after infection: Lv (liver), Sp (spleen), LN (lymph nodes), SI (small intestine), LI (large intestine), PP (Peyer’s patches), B (Blood)

^cInput is in CFU/mouse for a typical experiment

^ddpi=day post-inoculation

^eND=not determined

Naturally Occurring Lung pathogens

Y. pestis

Plague has been extensively studied over the past several decades both by epidemiology of human infection and a wide variety of animal models (Smith, 1959, Smith, 1959, Fingold, 1969, von Reyn, et al., 1977, Davis, et al., 1996, Lathem, et al., 2005, Begier, et al., 2006). Progression of murine pneumonic plague was described in mice in 1959 (summarized in Table 1) (Smith, 1959, Smith, 1959). However, only recently has a pneumonic infection in mice been thoroughly characterized in terms of host and bacterial response at the molecular level (Lathem, et al., 2005). As a natural host for Yersinia, the mouse model of infection closely mimics that seen human disease. After intranasal inoculation with 1×10⁴ CFU, 75% of the inoculum is recovered from the lungs 24 hours post inoculation. Whether this reduction is due to a physical bottleneck in bacteria reaching the lungs or active killing by the host is unclear. Regardless, by 72 hours post inoculation bacteria rapidly replicate to 10×10¹⁰ CFU/lung (Smith, 1959, Lathem, et al., 2005), at which point mice were moribund or had already succumbed to infection. Despite the presence of bacteria, lung histology is indistinguishable from uninfected mice until 48 hours post infection. This is likely due to the lack of inflammatory response elicited by the bacteria in the early course of infection (Lathem, et al., 2005, Bubeck, et al., 2007). By 48 hours post inoculation, however, severe histopathology is observed, including influx of neutrophils, edema, and tissue consolidation. No colonization of the spleen is observed until 36 hours post inoculation indicating that the bacteria do not disseminate before this or that the host is able to quickly kill those that do. By this time viable bacteria are detectable at levels of 100 CFU/spleen and increase to 1×10⁹ CFU/spleen by 72 hours post inoculation. All virulence and pathology is dependent on the Yersinia virulence plasmid as a plasmid minus strain is avirulent in mice (Lathem, et al., 2005).

The enormous bacterial burdens associated with this model of infection allow for detailed analysis of the bacterial transcriptome during infection (Lathem, et al., 2005). These analysis revealed that a number of metabolic and virulence associated genes were upregulated in lung tissue 48 hours post inoculation compared to growth in broth cultures (Lathem, et al., 2005). Similar analysis of splenic colonization would be greatly informative if tissue specific bacterial transcriptomes were available. For instance, it was noted that the psa operon was down-regulated in the lungs. Since psaA is important for colonization of the spleen following intranasal inoculation, it would be interesting to know if up-regulation took place there (Cathelyn, et al., 2006). Such analysis would enhance our understanding of how Yersinia is able to colonize different host tissues, perhaps allowing for novel therapeutics targeting virulence factors necessary in multiple organs.

Another feature of the Y. pestis intranasal infection model is that mice can be protectively immunized, allowing for a greater understanding of host response (Smith, 1959). Much of the early work in vaccine development focused on subcutaneous injection of heat killed bacteria or live attenuated vaccines such as the EV76 strain (for review see (Smiley, 2008)). The EV76 strain lacks an iron acquisition locus rendering it greatly attenuated, even in the presence of the virulence plasmid (Jackson & Burrows, 1956, Une & Brubaker, 1984). However, EV76 has not proven to be very efficacious as a live attenuated vaccine and is associated with severe side effects (Meyer, 1970, Bartelloni, et al., 1973). On the other hand, recombinant vaccines using the F1 component of capsule and LcrV of the TTSS generate a strong protective response against intranasal challenge in mice (Williamson, et al., 1997, Eyles, et al., 2000, Tripathi, et al., 2006, Smiley, 2008). However, new interest has been generated in live attenuated vaccines as avirulent, protective mutants continue to be uncovered. For instance, mice infected with LpxL⁺ Y. pestis are protected from subsequent intranasal or subcutaneous challenge (Montminy, et al., 2006). Additionally, Y. pestis ΔyopH is avirulent in pneumonic infection (Bubeck & Dube, 2007). Y. pestis ΔyopH infected mice are better able to survive subsequent pneumonic or bubonic infections (Bubeck & Dube, 2007). Thus models of pneumonic plague inform as to the role of adaptive immunity for vaccine development.

Francisella tularensis

Like Y. pestis, F. tularensis is a highly infectious respiratory and lymphatic pathogen maintained in a sylvatic reservoir and transmitted through an arthropod vector (deer flies), by aerosol, or handling of infected animals. As little as 10 CFU are capable of causing life-threatening respiratory illness, which can lead to systemic colonization. The symptoms of tularemia vary with the route of inoculation, but generally include pneumonia, fever, headache, and swollen glands (Matyas, et al., 2007, Sjostedt, 2007). Unlike plague, however, tularemia is not known to be transmissible from person to person and has adopted an intracellular lifestyle (Sjostedt, 2007).

The mouse model of infection closely mimics human disease. Both F. tularensis Types A and B can cause mouse and human disease, but Type A infection has a much higher prevalence in human infection (Rick Lyons & Wu, 2007). Mice intranasally inoculated with as little as 8 CFU of Type A succumb to infection within 7 days in various mouse backgrounds (Wu, et al., 2005). In fact, only DBA/2 mice showed modest resistance to this low dose, while BALB/c, C57BL/6, and C3H/HeN mice were exquisitely sensitive, suggesting that this model of infection is not strain dependent (Wu, et al., 2005). Similar experiments were conducted with two separate Type A strains with similar results, suggesting that this is not a bacterial strain specific phenomena either (Conlan, et al., 2003). Following aerosol infection with 200 CFU, lungs were colonized to >1×10¹⁰ CFU/lung by 5 days (Conlan, et al., 2003). Systemic organs of mice were colonized 48 hours post inoculation and remained colonized until mice succumbed. As expected mice inoculated with the vaccine strain, LVS, were able to clear systemic infection after 2 weeks, although survival is dose dependent (Wu, et al., 2005). The fact that Type A causes disease while the vaccine strain does not is consistent with human studies, further supporting the idea that this infection model mimics human infection (Rick Lyons & Wu, 2007). A working murine model of infection for tularemia facilitated discovery of specific bacterial virulence factors, and the role of the innate immune system for countering them (Rick Lyons & Wu, 2007).

Vaccination studies using the murine lung model of F. tularensis infection shed light both on the role of adaptive immunity and on disease prevention (Wu, et al., 2005). Such a model prevents the necessity of human volunteers such as the prisoner “volunteers” employed for the development of the LVS vaccine strain (Hornick & Eigelsbach, 1966). Intranasal inoculation with the LVS vaccine strain can protect BALB/c mice from intranasal or subcutaneous challenge with virulent F. tularensis (Wu, et al., 2005). However, C57BL/6 mice are less protected in these assays. This result may suggest a mixed efficacy in human populations. Using purified antigen is another approach being utilized to develop and efficacious vaccine. For instance, in TraSH studies many F. tularensis attenuated mutants were identified (Weiss, et al., 2007). Amongst the genes identified were LPS and O-antigen biosynthetic genes. Subunit vaccines to these antigens have been used with some success (Wayne Conlan & Oyston, 2007); however, they are generally not protective for all Francisella subspecies. By making subunit vaccines from novel virulence factors identified in these studies a more broad range pulmonary vaccine may be developed.

Bordetella pertussis and Bordetella bronchiseptica

Bordetella pertussis is the causative agent of whooping cough, a severe respiratory disorder characterized by a contagious chronic cough, which can be fatal in infants (Markov & Crowcroft, 2007). In humans limited histopathology available suggests that bacteria line the lung bronchioles and form lesions in the epithelium (Mallory & Horner, 1912, Elahi, et al., 2007). B. pertussis is an obligate human pathogen and has no natural animal host. Animal models-from mice to pigs to puppies-have been employed with some success, yet none completely mimic the human disease (Elahi, et al., 2007). Although B. bronchiseptica has only been shown to affect severely immunocompromised people, it induces a more severe infection in mice and has been used in murine infections to model B. pertussis human disease (Elahi, et al., 2007). After receiving 1×10⁵ CFU, wild type mice can clear most of a B. pertussis infection within 30 days (Table 1) (Khelef, et al., 1994, Vandebriel, et al., 2003). This process takes almost 60 days in a B. bronchiseptica infection (Harvill, et al., 2000). Additionally, infection with B. bronchiseptica induces a much more intense histopathology than B. pertussis in mice (Khelef, et al., 1994, Harvill, et al., 1999, Vandebriel, et al., 2003). Histopathology during B. bronchiseptica infection generally consists of inflammation, of the lungs and recruitment of neutrophils around the bronchioles (Harvill, et al., 1999). As with human infection, murine infections are not lethal by intranasal routes of inoculation, however, infection of SCID mice or infection by the intracranial route (a meningitis model) can be lethal (Harvill, et al., 1999, Burns, et al., 2003). Similarly, IL-4^−/− and IFN-γR^−/− intranasally inoculated mice clear B. pertussis as well as wild type mice, while Ig^−/− could not clear the infection even after 3 months (Mahon, et al., 1997). By using mutant mice in such experiments the host factors important in immunity will continue to be elucidated.

Although Bordetella murine lung infections do not model all symptoms of whooping cough (e.g. unlike piglets, mice are not able to cough (Elahi, et al., 2007)) such experiments have been invaluable in determining the role of virulence factors in causing disease as well as in identifying potential vaccine components. For example, a B. pertussis Δfim mutant strain which lacks the fimbrial adhesin, is cleared more rapidly than a wild type strain following intranasal inoculation (Vandebriel, et al., 2003). Additionally a B. parapertussis O-antigen mutant is more rapidly cleared from the lung (Burns, et al., 2003). Since Bordetella species all differ in O-antigen structure comparing differential antigens may lead to a better understanding of how pathogens evolve to broaden host range.

Pseudomonas aeruginosa

P. aeruginosa is an important opportunistic pathogen associated with nosocomial infection. Pseudomonas forms biofilms, making it highly resistant to antibiotics, on catheters and other medical devices, such as ventilator tubes. Like Yersinia, P. aeruginosa employs a type three secretion system to intoxicate host cells. Chronic infection can develop in exposed immunocompromised individuals (Sadikot, et al., 2005). Burn victims are especially vulnerable to bacteremia (Ressner, et al., 2008), while ventilated or cystic fibrosis patients are at higher risk for contracting chronic pneumonia caused by P. aeruginosa (Moreau-Marquis, et al., 2008).

Since neither healthy humans nor rodents are natural hosts for Pseudomonas, it is not surprising that murine lung models do not efficiently reproduce human disease (which is generally limited to immunocompromised individuals). Additionally, infections tend to be dependent on both the mouse and bacterial strains. C57BL/6 mice infected with 1×10⁷ CFU P. aeruginosa clear the majority of CFU by 24 hours post inoculation (Table 1) (Power, et al., 2004). Infected mice show a marked increase in neutrophil recruitment at this time, suggesting that unlike Y. pestis (Lathem, et al., 2005, Bubeck, et al., 2007), P. aeruginosa does not actively inhibit migration of PMNs at early time points. In fact, P. aeruginosa infected mice show an increase in inflammatory cytokines such as MIP-2 and TNF production four hours post inoculation (Power, et al., 2004). CH3/HeJ mice infected with a non-lethal dose (LD₅₀ in this model is 1×10⁷ CFU) of P. aeruginosa did not develop chronic pneumonia, and cleared the bacteria over the course of two weeks (George, et al., 1993). Interestingly, this was true for 7 different bacterial strains tested, suggesting that the lack of chronic infection is not due to the strain of bacteria, but rather the host background.

Mutant mice which mimic cystic fibrosis allow for the investigation of human infection. Infections of CFTR^−/− mice better mimic what is observed in human infection with Pseudomonas. Upon infection with 4×10⁶ CFU, bacteria replicate over the course of 7 days, over which time 65% of mice succumb to pneumonia (Hoffmann, et al., 2005). This infectious dose is sublethal to wild type BALB/c mice.

Although efforts in vaccine development are underway for CF patients, P. aeruginosa is a nosocomial pathogen, and therefore vaccine development is not a high priority, or perhaps even not advisable (Doring & Pier, 2008). However, pulmonary models of infection will allow for testing novel therapeutics and treatment regimens. For instance, in TTSS mutants which cannot secrete effectors are outcompeted by wild type bacteria in the lungs and spleens of neutropenic mice following intranasal inoculation (Vance, et al., 2005). Targeting virulence factors such as the TTSS is an emerging paradigm in antimicrobial chemotherapy (Clatworthy, et al., 2007).

Klebsiella pneumoniae

Klebsiella pneumoniae is a nosocomial, human pathogen associated with urinary tract infection, pneumonia, and the infection of burn wounds (Podschun & Ullmann, 1998, de Macedo, et al., 2003). The vast majority of infection with K. pneumoniae occurs in immunocompromised patients, including chronic alcoholics (Dorff, et al., 1973). In fact, at least 35% of nosocomial Klebsiella pneumonia has been reported in alcoholic patients, who can rapidly succumb to infection (Dorff, et al., 1973, Jong, et al., 1995). Additionally, burn victims can be colonized by K. pneumoniae, which can readily progress into bacteremia (Sharma, et al., 2006, Ressner, et al., 2008). K. pneumonia infections are becoming increasingly complicated with the rise in antibiotic resistant strains around the world (Paterson, 2006, Hosoglu, et al., 2007, Oteo, et al., 2008, Tokatlidou, et al., 2008).

Due to the variety of infections caused by K. pneumoniae several animal model systems including non-human primate intratracheal, mouse burn, rat urinary and systemic infections have been developed (Uehling & Wolf, 1969, Berendt, et al., 1978, Cryz, et al., 1984). To study K. pneumoniae induced pneumonia in mice, both intratracheal and intranasal models of infection have been utilized (Shankar-Sinha, et al., 2004, Lawlor, et al., 2005). During intratracheal infection, mice are anesthetized and the trachea is surgically exposed for the delivery of bacteria. The LD₅₀ for both infection models are ~2×10³ (Yadav, et al., 2003, Shankar-Sinha, et al., 2004, Lawlor, et al., 2005). Twenty-four hours post intranasal inoculation with ~2×10⁴ CFU bacterial burdens in the lungs and trachea reach levels of ~1×10⁷ CFU/g organs and by 24 hours splenic colonization had been established with an average burden of 2.5×10² CFU/g spleen (Lawlor, et al., 2005). Mice succumbed to infection between 72 and 96 hours. At this time mice harbored burdens of ~1×10⁹ CFU/g organ in the trachea, lungs, and spleen (Lawlor, et al., 2005).

A major virulence factor associated with K. pneumoniae is a polysaccharide capsule, which protects the bacteria from phagocytosis, antimicrobial peptides, and neutrophil mediated killing (Williams, et al., 1983, Domenico, et al., 1994, Podschun & Ullmann, 1998, Campos, et al., 2004). Bacteria lacking capsule are attenuated for virulence in pneumonic, subdermal, and renal murine infection models (Yoshida, et al., 2001, Regue, et al., 2004, Lawlor, et al., 2005, Lawlor, et al., 2006). Additionally, mice intranasally inoculated with a capsule deficient mutant (cps⁻) are protected from subsequent challenge with wild type K. pneumoniae (Lawlor, et al., 2006). Capsule is not the only factor important for causing pneumonia. In fact, signature tagged mutagenesis has revealed 106 loci necessary for colonization of the lung and spleen (Lawlor, et al., 2005). As antibiotic resistance becomes more prevalent, these data may help identify new targets for therapeutics to complement the dwindling number of efficacious drugs.

Burkholderia pseudomallei

There are several species of Burkholderia, three of which are important human pathogens most commonly found in the moist soil of Southeast Asia and northern Australia (Choy, et al., 2000, Wiersinga, et al., 2006). B. cepacia (formerly Pseudomonas cepacia) generally infects the lungs of cystic fibrosis patients, but is not thought to cause disease in healthy individuals (Wiersinga, et al., 2006, Harrison, 2007). B. thialandensis is normally considered a saprophyte not normally associated with disease (Wiersinga, et al., 2006). However, human infection with this organism has been reported (Glass, et al., 2006). B. pseudomallei is an opportunistic intracellular pathogen. It is of interest to note that B. mallei is believed to have recently evolved from B. pseudomallei through reductionist evolution as Y. pestis formed from Y. pseudotuberculosis (Achtman, et al., 1999, Godoy, et al., 2003). Although B. pseudomallei is a much more prevalent human pathogen, this may represent a common theme in the evolution of pathogens. In addition, both B. pseudomallei and B. mallei have been listed by the CDC as category B agents (Cong, et al., 2001). As B. pseudomallei is more common human pathogen, and thus prominently studied, it will be the focus of this discussion.

B. pseudomallei has a broad host range and can infect animals such as goats and sheep as well as humans (Choy, et al., 2000). Infection is thought to occur through inhalation or open cuts can lead to severe pneumonia; and, subsequently, highly lethal septicemia can develop (Glass, et al., 2006). Seroprevalence in exposed populations are as high as 80% by four years of age (Kanaphun, et al., 1993). However, the vast majority of clinically relevant B. pseudomallei infections are in individuals between 40–60 years old generally with an underlying medical condition such as diabetes mellitus (White, 2003, Wiersinga, et al., 2006). These data highlight the fact the B. pseudomallei is an opportunistic pathogen.

Once in a host, B. pseudomallei enters host cells, exits the phagocytic vacuole, and replicates intracellularly (Harley, et al., 1998). Spread to neighboring cells is accomplished by actin polymerization based motility (Harley, et al., 1998, Wiersinga, et al., 2006). B. pseudomallei (but not B. thailandensis (Harley, et al., 1998)) contains at least three sequence clusters with high homology to the TTSS of S. typhimurium and S. flexneri (Rainbow, et al., 2002, Stevens, et al., 2002). In the absence of several components of the putative TTSS B. pseudomallei showed diminished ability to form actin tails (Stevens, et al., 2002).

Aerosol or intranasal inoculation of mice with B. pseudomallei results in fulminant infection. Mice that received 4.5×10³ CFU demonstrated systemic colonization by 48 hours post inoculation with splenic and hepatic burdens reaching 1×10³–1×10⁴ CFU/g organ in the spleen and liver and 1×10⁶ CFU/g lung (Jeddeloh, et al., 2003, Cuccui, et al., 2007). By three days post inoculation colonization of the spleen and liver increased to 1×10⁸ CFU/g organ and 1×10⁹ CFU/g lung (Jeddeloh, et al., 2003). Mice infected with as little as 196 CFU succumbed to infection within 4 days (Jeddeloh, et al., 2003). STM analysis of B. pseudomallei identified 39 signature tagged mutants attenuated in lung following intranasal inoculation; 22 were in capsular biosynthesis genes (Cuccui, et al., 2007). Seven other genes were involved in amino acid biosynthesis including, leucine, serine, and threonine (Cuccui, et al., 2007). Other mutagenesis studies have shown that amino acid auxotrophs for purine and histidine are also attenuated following intranasal infection (Pilatz, et al., 2006). These mutants were also attenuated for intracellular replication in cell culture, suggesting the intracellular lifestyle is essential for B. pseudomallei survival during the course of infection (Pilatz, et al., 2006). Interestingly, neither of these screens identified TTSS components as essential virulence factors, indicating that the screens did not saturate the genome, that the putative TTSS is not a required virulence mechanism in B. pseudomallei in this model of infection, or that they function redundantly.

Lung models of enteric Gram-negative pathogens

Salmonella Typhi

Salmonella Typhi is the causative agent of typhoid. Clinical manifestations include fever, delirium, diarrhea, and hepatosplenomegaly. According to the World Health Organization, there are an estimated 21 million cases of Salmonella Typhi worldwide each year (2008). Much has been learned my modeling the molecular pathogenesis of Typhoid on animal models of S. typhimurium (Ohl & Miller, 2001). Because Salmonella Typhi is host restricted to humans, orogastric infection in mice does not cause disease. However, because of its importance in global health, much emphasis has been placed on vaccine design and improvement. To this end, a pneumonic model of infection with Salmonella Typhi vaccine strain has been well characterized (Table 2) (Galen, et al., 1997, Pickett, et al., 2000, Shi, et al., 2010).

TABLE 2.

Characteristics of enteric bacterial pathogens in murine lung models

Enteric Organism	LD50	Time of Death	Routea	Disseminationb	Inputc	CFU/lung (24hr)	CFU/lung (End dpi)d	Refs.
Y. pseudotuberculosis	18	8 days	IN	Lv, Sp,	4×102	1×103	1×107 (4 dpi)	(Fisher, et al., 2007)
C. jejuni	~5×109	6 days	IN	B, Lv, Sp, LI, SI, LN	1×109	2×103–1×107	<1 (5 dpi)	(Baqar, et al., 1996, Al-Banna, et al., 2008)
V. cholerae	ND	<1–3 days	IN	Lv, Sp	4×107	2×103–2×104	ND	(Fullner, et al., 2002)
S. flexneri	ND	1–6 days	IN	ND	2×107	9×105	1×105	(Voino-Yasenetsky & Voino-Yasenetskaya, 1962, Mallett, et al., 1995, Phalipon, et al., 1995)
S. typhi	ND	ND	IN	PP	1×109	1×109	<10 (dpi)	(Pickett, et al., 2000)
Y. enterocolitica	ND	ND	IN	ND	5×106	ND	2×106 (2 dpi)	(Di Genaro M.S., et al., 1998)
Y. pseudotuberculosis	18	8 days	IN	Lv, Sp,	4×102	1×103	1×107 (4 dpi)	(Fisher, et al., 2007)

^aA=aerosol, IN=intranasal, IT=intratracheal

^bdissemination to organs after infection: Lv (liver), Sp (spleen), LN (lymph nodes), SI (small intestine), LI (large intestine), PP (Peyer’s patches), B (Blood)

^cInput is in CFU/mouse for a typical experiment

^ddpi=day post-inoculation

^eND=not determined

Mice intranasally infected with 1×10⁹ CFU of a Salmonella Typhi vaccine strain cleared bacteria from the lungs after 72 hours. When an isogenic vaccine strain expressing tetanus toxin was used, mice generated robust immune responses to the protective antigen (Galen, et al., 1997). Additionally, Galen et. al. intranasally or orogastrically infected mice with a Salmonella Typhi vaccine strain expressing a fusion protein of tetanus toxin. Intranasally infected mice developed a stronger immune response than those infected via the oral route. Similarly, mice intranasally infected with a Salmonella Typhi vaccine strain expressing the exogenous antigen pspA from S. pneumoniae developed a robust immune response to this antigen (Shi, et al., 2010). Together, these data demonstrate the validity and importance of pneumonic models of infection for vaccine development of host restricted pathogens.

Shigella flexneri

Shigella spp. such as S. flexneri, S. dysenteriae, and to a lesser extend S. sonneii cause a debilitating bacterial dysentery in humans (Philpott, et al., 2000). Symptoms of Shigellosis generally include cramps, fever, in addition to purulent and bloody stool (Philpott, et al., 2000). Shigella is highly contagious and spreads via the fecal oral route of infection (Phalipon & Sansonetti, 2007). As few as 100 organisms of S. flexneri are capable of causing disease (DuPont, et al., 1989, Phalipon & Sansonetti, 2003). Animal models of pathogenesis, however, present difficulties in studying this tissue specificity and pathogenesis in general, as mice do not develop disease upon oral infection with Shigella (Philpott, et al., 2000). For instance, the rabbit ileal loop model is useful for studying S. flexneri invasiveness, but does not examine infection in the colon. Additionally, animals do not exhibit symptoms of dysentery. Despite this limitation much has been learned about the progression of disease from the available animal models and human studies (Sereny, 1955, Mathan & Mathan, 1991, Menard, et al., 1993, Philpott, et al., 2000).

Shigella traverse the intestinal wall through specialized M cells (Wassef, et al., 1989, Philpott, et al., 2000). Transmigrated bacteria then enter epithelial layer through the basolateral side (Philpott, et al., 2000). Shigella move within the cytoplasm of host cells by polymerizing actin in an IcsA dependent manner (Bernardini, et al., 1989). Shigella remains intracellular and spreads from cell to cell via actin based motility dependent on IcsA (Sansonetti, 1998, Philpott, et al., 2000). Work in the rabbit ileal loop model shows those bacteria that traverse M cells often encounter macrophages, which phagocytose the bacteria. Shigella activates the apoptotic pathway and kills the engulfing host macrophage in an IpaB dependent manner (Zychlinsky, et al., 1994, Philpott, et al., 2000). These host-pathogen interactions lead to the release of IL-8, an important chemokine for neutrophil recruitment (Sansonetti, et al., 1999).

In an effort to learn more about the host-pathogen interaction during Shigella infection, and in the absence of a suitable model to mimic human dysentery, work has been invested in the development of a pneumonic model of infection. Seminal work on this model began in the late 1950’s (Voino-Yasenetsky & Voino-Yasenetskaya, 1962). In these studies 113 white mice (strain not specified) were given ~1×10⁷ CFU of S. sonneii and 173 mice were inoculated with S. flexneri. By 48 hours post-inoculation 83% of S. sonneii infected mice succumbed to infection as did 74% of those mice infected with S. flexneri (Voino-Yasenetsky & Voino-Yasenetskaya, 1962). Bacterial loads in the lungs of surviving mice by 48 hours had diminished to ~4×10³ CFU/organ. Histopathological analysis of infected mice revealed an influx of neutrophils. Large numbers of the bacteria remained extracellular, but phagocytized bacteria were also observed. Importantly, bacteria were also noted to exist intracellularly within the epithelial cells of the lung.

It was proposed that IgA opsonization of Shigella in the intestinal lumen would facilitate the bactericidal activity of infiltrating neutrophils and resident macrophages and reduce bacterial infection (Phalipon, et al., 1995). Supporting this model, vaccination via the intranasal or orogastric route with proteosome-LPS complex from S. sonneii cross protects against subsequent intranasal challenge with S. flexneri or S. sonneii (Mallett, et al., 1995) and that protection was associated with production of IgA and IL-6 (Mallett, et al., 1993, Phalipon, et al., 1995). More recent work has revealed that transgenic BALB/c mice expressing IgA antibodies to Shigella LPS from an implanted hybridoma are less susceptible to subsequent intranasal infection (Phalipon, et al., 1995). These data highlight how the mucosal environment can be predictive of other mucosal immune environments. Additionally, they demonstrate the importance of the pneumonic models in vaccine development.

Campylobacter jejuni

Campylobacter jejuni is the causative agent of a self-limiting diarrhea in humans, which can last for several days to weeks and range in severity from loose stool to watery stool with mucus or blood. In rare cases, neuropathy can develop as the host immune responds to C. jejuni lipooligosaccarides. Our understanding of C. jejuni pathogenesis has been hampered by the lack of a mammalian animal model which mimics human disease. As part of the natural flora of avian species, chick models have allowed a much greater understanding of colonization of the gastrointestinal tract (Young, et al., 2007). However, experimental infection in this system does not induce disease. Infections in other hosts such as ferrets do induce disease (Fox, et al., 1987), but costs and lack of reagents make this model less tenable (Young, et al., 2007). Mice orally infected with C. jejuni clear the bacteria quickly. Like Shigella, C. jejuni infection induces IL-8 in humans. However, recently a model has been developed in which SCID mice cleared of their natural flora were chronically colonized and inflammation of the gastrointestinal tissues induced, which will aid in future understanding of host response (Chang & Miller, 2006).

To complement existing models of pathogenesis, an intranasal model of C. jejuni infection was developed (Table 2) (Baqar, et al., 1996). After intranasal inoculation of 1×10⁹ CFU lungs, stomach, large intestine, small intestine, liver, mesenteric lymph nodes, and spleen were all colonized by 6 hours and declined over the course of 7 days. Whether the colonization of gastrointestinal tissues was due to direct colonization via post-nasal drip or systemic spread cannot be determined. However, the fact that both the liver and spleen were colonized at early time points supports an argument for the latter. Six days after intranasal inoculation, 70% BALB/c mice, 50% C3H/HeJ mice, and 30% CBA/CAJ mice succumbed to infection. C58/J mice were more resistant and showed no mortality at this dose, and only 20% of mice succumbed with a dose of 1×10¹⁰ (Baqar, et al., 1996). However in a subsequent study, mortality in BALB/c mice was not observed even though the same strain of C. jejuni was used. This suggests there may have been technical differences between the two studies or lab specific mutations (Al-Banna, et al., 2008). The lethality in C. jejuni infected mice, however, correlates with the severity of the symptoms of the donor patient (Baqar, et al., 1996).

Recruitment of neutrophils in the gut is an important aspect of infection with Campylobacter in the chicken model of enteric infection (Fox, et al., 1987). Similarly, histopathology of the lung following intranasal infection includes the influx of neutrophils as well as foamy macrophages, bronchopneumonia, and the formation of granulomas (Al-Banna, et al., 2008). Intranasal infection with C. jejuni may allow for the study of interactions between bacteria and neutrophils in the context of a mucosal immune environment. Further work will allow for the elucidation of specific virulence factors or novel functions for those already known.

Vibrio cholerae

V. cholerae is the causative agent of cholera, dysentery associated with profuse, watery diarrhea. Massive fluid loss can lead to dehydration and result in death without medical intervention. The study of V. cholerae has led to a great appreciation for the virulence factors necessary for causing disease. The rabbit ileal loop model, neonate mouse model, and studies from naturally or experimentally infected human cohorts have elucidated much about the pathogenesis of V. cholerae (De & Chatterje, 1953, Freter & Gangarosa, 1963, Merrell, et al., 2002, Butler & Camilli, 2005, Butler, et al., 2006). These models demonstrate the roles of classic virulence factors, such as cholera toxin (CTX) and toxin co-regulated pilus, as well as appreciating the role of physiological functions, such as chemotaxis in disease progression (Lee, et al., 1999, Butler & Camilli, 2004). In adult mice V. cholerae is cleared from the gut without mimicking acute infection suffered by infected humans (Knop & Rowley, 1975, Knop & Rowley, 1975), thus putting many tools made available by murine models out of reach. Fullner el. al. developed a lung model of V. cholerae infection to determine infectivity. They tested the dependence of infectivity on known virulence factors, and monitored the host immunological response (Table 2) (Fullner, et al., 2002). Upon intranasal inoculation with ~5×10⁷ CFU of a CTX⁻ strain, mice were reported to have demonstrated diffuse pneumonia associated with influx of neutrophils, hemorrhage, fibrination, and epithelial sloughing (Fullner, et al., 2002). This pathology could be alleviated by the removal of the gene encoding the endotoxin, RTX. This suggests that the inflammatory pneumonia observed is due to Vibrio specific virulence factors and not simply toxicity due to the high bacterial input. However, removal of other known virulence factors such as the hemmagglutin/protease hapA or the hemolysin hlyA had no effect. Intranasal Vibrio infection also resulted in an increase in TNF, IL-6, and MIP-2. This inflammatory response was reduced in a stain lacking RTX. Although clearly not an exact model for human disease, these studies promise greater understanding of innate mucosal immune effectors in mature mice. It should be noted, however, that the these studies were done in the absence of functional CT, which has been shown to play an anti-inflammatory role tissue culture models (Cong, et al., 2001). It may be that in the presence of CT the inflammatory pneumonia would not be as evident following intranasal inoculation.

Enteric Yersinia

Robust orogastric and pneumonic animal models for Yersiniosis have been developed and characterized extensively (Lathem, et al., 2005, Logsdon & Mecsas, 2006, Fisher, et al., 2007). Of the pathogens reviewed here, Yersinae is the only genus for which both pneumonic and enteric modes of infection exist. We review these infection models to demonstrate the similarities and differences between established pneumonic model and enteric models of infection.

Three Gram-negative Yersinia bacteria cause various diseases in animals and humans. Y. pseudotuberculosis and Y. enterocolitica transmit themselves via a fecal-oral route of infection. Resulting infection leads to self-limiting gastroenteritis and lymphadenitis. Y. pestis, causes bubonic and pneumonic plague. All three Yersinia human pathogens harbor a 70Kb virulence plasmid, which encodes a TTSS and associated translocated toxins, called Yops.

There are several reported human cases of pneumonic infection with Y. enterocolitica (Bigler, et al., 1981, Girszyn, et al., 2007) and one possible infection with Y. pseudotuberculosis (Hagiwara, et al., 1995). Furthermore the protective properties of cellular extracts of Y. enterocolitica on subsequent intranasal challenge has been investigated (Di Genaro M.S., et al., 1998). Following intranasal inoculation, with 5×10⁶ CFU of serotype 0:8 Y. enterocolitica bacterial loads fell to 2×10⁶ by 48 hours, but were associated with histopathology consistent with pneumonia (Di Genaro M.S., et al., 1998). Although the LD₅₀, mean time to morbidity, or ability to spread systemically in this model were not established, they did observe a twofold decrease in bacterial burden and no histopathology when mice were immunized with cellular extract prior to infection, suggesting a role for adaptive immunity in protecting against pneumonic Yersiniosis.

Experimental infection with Y. pseudotuberculosis with an initial inoculum of less than 100 CFU colonized both BALB/c and Swiss Webster mice (Fisher, et al., 2007). This infection model was associated with high mortality. Furthermore, histopathology of the lungs in this model was consistent with pneumonia, and included an influx of neutrophils and macrophages and tissue damage. However, dissemination to the spleen, liver and blood occurred sporadically and at lower burdens those observed in the lungs or by infection with Y. pestis (Table 1). However, tissue colonization was dependent virulence factors known to be important for infection for Y. pests. Although its efficacy was not tested, a humoral immune response was generated by mice surviving as demonstrated by IgG production. Additionally, this model has been subsequently used to screen small molecule inhibitors of the TTSS (Garrity-Ryan, et al., 2010). From this screen, two inhibitors were identified that reduced bacterial burden and limited the course of infection.

Concluding Remarks

The pathogens discussed here can be divided into three categories: naturally occurring lung pathogens, opportunistic lung pathogens, and enteric pathogens. It is interesting that the first group is associated with a very low LD₅₀, while the latter two groups (with the exception of Y. pseudotuberculosis) are associated with a relatively high LD₅₀. These differences highlight the fact that although the lungs can act as a surrogate for sites of mucosal immunity, there are important differences between the lung and the GI tract in physical structure and local immune responses. With that in mind, murine lung models of pathogenesis allow for a greater understanding of the basic biology by which pathogens survive in pulmonary tissues as well as an understanding of the host response. Such insights offer invaluable information for the development of vaccines and novel therapeutics. Additionally, the pneumonic route of infection can be used as an alternate site of mucosal immunity for enteric pathogens for which there is not suitable models. Continued experimentation in this venue will inform as to the how these pathogens elicit an immune response, and in some cases can predict immune responses during gastrointestinal infection.