Subcutaneous ulcer/air sac model

One of the most widely used animal models for the analysis of S. pyogenes virulence factors is the murine subcutaneous ulcer model and its air sac derivative. This model is particularly attractive because it reproduces many of the core features that are held in common by a broad range of S. pyogenes diseases. These include that the infection results in a highly inflammatory localized lesion in a soft tissue, that there is a robust level of bacterial proliferation, and that the lesion elicits the extensive recruitment of inflammatory cells. Introduced in its modern form by the group of Neil Barg (Bunce, Wheeler, Reed, Musser, & Barg, 1992), it originally involved injecting a mixture of bacteria and cytodex beads into the subcutaneous tissue of a mouse flank. However it was subsequently found that the beads were not necessary to produce an infection (Schrager, Rheinwald, & Wessels, 1996). Typically, 10⁶–10⁸ CFU of the S. pyogenes strain of interest is injected into the tissue, which results in a well-defined area of induration that is apparent by 8–12 hours post-infection. The lesion that forms is characterized by the recruitment of a large number of inflammatory cells, of which the majority are neutrophils that form a loose abscess-like structure that surrounds the area of bacterial growth (Figure 2). Macrophages and the cytokines IL-12, Interferon-γ and TNF-α appear to be important for the formation of this structure, which limits dissemination of bacteria from the site of the lesion (Mishalian, et al., 2011; Raeder, Barker-Merrill, Lester, Boyle, & Metzger, 2000). By 18–24, hours the region of induration ulcerates and develops an eschar, followed by the gradual expansion of the margins of the ulcer to reach a maximum at about Day 3 (Figure 2). At approximately Day 8, the lesion begins to resolve, and by Day 14, it is typically healed. Many different S. pyogenes strains have been shown to be virulent in this model, and these fall into two general classes: those that remain localized to the lesion for the duration of the infection (such as those found in (Brenot, King, Janowiak, Griffith, & Caparon, 2004)), and those that have a propensity to disseminate into the vasculature to cause systemic disease after a period of several days (such as those found in (Limbago, Penumali, Weinrick, & Scott, 2000)). Interestingly, mutations have been described, most notably in the CsrR/S (CovR/S) two-component transcription regulator, which can convert a primarily resident strain into a highly invasive strain (Engleberg, Heath, Miller, Rivera, & DiRita, 2001). Many strains of mice are sensitive to this infection, including strains like the BALB/c and C57BL6 strains, as well as their various transgenic and knock-out variants that are commonly used in immunological analyses. However, a commonly used mouse strain is the outbred SKH1 mouse that has a defect in the Hr gene located on chromosome 14 that renders it hairless (Benavides, Oberyszyn, VanBuskirk, Reeve, & Kusewitt, 2009), which eliminates the requirement for depilation for injection and observation of the lesions. This model also offers a large number of different infection parameters to assess virulence. A commonly used approach is to quantitate the surface area of the resulting ulceration and how it changes over time. Typically, a conclusion regarding virulence between the wild-type S. pyogenes strain and its mutant derivatives is made by a comparison of the lesion area at the time point when the wild-type strain’s lesion reaches a maximum area (usually by Day 3). Since the resulting ulcer will have an irregular border, its area is most accurately measured by the use of digital photography and image processing software (Brenot, King, Janowiak, Griffith, & Caparon, 2004). Other parameters that can be assessed include change in weight over time, numbers of colony forming units (CFU) that are recoverable from the lesion, the number of mice that do or do not develop a ulcer following infection, the time to maximal lesion area, and the time to healing of the ulcer. For S. pyogenes strains that disseminate from the lesion, additional parameters can be measured, which typically include CFU recovered from the spleen and inguinal lymph nodes and lethality. This model has also proven its utility for the monitoring of the behavior of the S. pyogenes transcriptome during growth in a soft-tissue environment. The fact that the bacteria remain confined to a necrotic lesion that is not conducive for maintaining the integrity of host RNAs has allowed the efficient recovery of S. pyogenes RNA directly from the infected tissue for analyses of gene expression (Brenot, Weston, & Caparon, 2007; Loughman & Caparon, 2006; Kietzman & Caparon, 2011).

Figure 2.

Murine subcutaneous ulcer model. This figure shows the histopathology of infection by S. pyogenes HSC5 following the subcutaneous inoculation of 1 x 10⁶ CFUs into the flank skin of SKH1 mice. Shown are hematoxylin & eosin (H&E)-stained (more...)

The subcutaneous air sac model is a derivative of the subcutaneous ulcer model. The same subcutaneous tissue compartment is infected in both models. A distinction for the air sac model is that prior to infection, an air pouch is created under the skin by the injection of air into which the bacteria are subsequently injected (Raeder & Boyle, 1993). This method’s principal advantage is that it facilitates the recovery of host inflammatory cells, which migrate into the sac lumen and can be recovered by lavage for analyses in vitro by FACS and other methods that require a suspension of single cells.

Like any of the animal models for analysis of S. pyogenes pathogenesis, the subcutaneous ulcer model has a number of important strengths and limitations that must be taken into account when designing experiments to provide insight into the role of a specific S. pyogenes gene product in virulence. As discussed, the model reproduces features of a highly inflammatory infection caused by S. pyogenes growing extracellularly in the types of cutaneous tissue that S. pyogenes infects in humans. On the other hand, the histopathology of the infection does not closely reproduce that observed for human cutaneous diseases, including impetigo, pyoderma, erysipelas, cellulitis, or necrotizing fasciitis. These observations can be associated with differences in the anatomy of the murine cutaneous tissues as compared to humans, as the mouse has a much higher density of flank skin hair follicles and a much thinner epidermal layer. Also, there is a considerable variation among the virulence of S. pyogenes strains in this model, and the virulence of any particular strains in humans is not predictive of its virulence in this model. Certain virulence factors may play a more important role in the mouse rather than the human, since mice lack the human-specific targets of factors, like streptokinase and the superantigens. In association, mutations of some S. pyogenes virulence factors do not always result in the same degree of attenuation in different strains. A classic example is the case of the SpeB cysteine protease, whose loss of function in different strains has been associated with significant attenuation or no attenuation at all (Ashbaugh, Warren, Carey, & Wessels, 1998; Lukomski, et al., 1999). This has suggested that different S. pyogenes strains may have unique virulence profiles that are comprised of overlapping, redundant and/or combinatorial sets of virulence genes (Saouda, Wu, Conran, & Boyle, 2001; Sierig, Cywes, Wessels, & Ashbaugh, 2003). In examining the immune system, the subcutaneous model has proven valuable for analysis of the role of innate immunity in the host’s response to S. pyogenes and how specific virulence factors modulate innate immunity. It is less clear that this model will be useful for analyses of the adaptive arm of the immune response, since a protective immune response apparently does not develop after the resolution phase of the infection (Bisno & Gaviria, 1997). However, the limitations of this model are counter-balanced by a number of important strengths. These include its simplicity, its relatively high-throughput and low cost, and the ability to apply the power of mouse genetics to test the role of various host factors-virulence factor relationships (for one example, see (Cusumano, Watson, & Caparon, 2014)). In addition, because the bacteria grow in a defined lesion, it reproduces the dynamic host environment that is remodeled both by host immunity and by microbial metabolism, which consumes host-derived substrates and accumulates metabolic end-products, like organic acids (Loughman & Caparon, 2006). Alterations to the transcriptome that accompanies adaptation to this dynamic environment are closely linked to changes in virulence factor expression (Kreikemeyer, McIver, & Podbielski, 2003). These many advantages have made the subcutaneous ulcer model one of the most versatile and useful models for the analysis of S. pyogenes pathogenesis.

Impetigo model

The histological differences between the human and murine epidermal tissues noted above have made the mouse unsuitable for modeling impetigo and other S. pyogenes infections of superficial cutaneous tissues. Bessen and colleagues have developed an innovative solution to this problem (Scaramuzzino, McNiff, & Bessen, 2000). Their approach was to create a humanized mouse (hu-mouse) by engrafting human epidermal tissue from neonatal foreskin onto the flanks of the SCID mouse. The absence of an adaptive immune response in this mouse line prevents the rejection of the tissue grafts. Once the grafts are established, they are superficially damaged by cross-wise cuts with a scalpel blade, bacteria are topically applied, and are then occluded with a bandage. Both damage and occlusion were required for establishing the infection, similar to what has been observed in experimental infections conducted in humans (Leyden, Stewart, & Kligman, 1980). This results in an impetigo-like lesion that is characterized by erosion of the stratum corneum, with infiltration of murine polymorphonuclear leukocytes (PMNs) that eventually contribute to the formation of pus, which is surrounded by clumps of proliferating streptococci. Virulence is measured by a semi-quantitative visual assessment of histopathology or by the determination of colony forming units to monitor bacterial growth. The degree of damage generally correlates with the magnitude of streptococcal growth (Scaramuzzino, McNiff, & Bessen, 2000). Other characteristics that reproduce elements of the human disease include the finding that S. pyogenes strains of lineages that are most likely to be associated with impetigo tended to cause more severe tissue damage than those lineages that are more likely to cause pharyngitis (Scaramuzzino, McNiff, & Bessen, 2000), and that virulence requires known S. pyogenes virulence factors, which include the virulence gene transcription regulator Mga (Luo, Lizano, Banik, Zhang, & Bessen, 2008).

Despite these strengths, the hu-mouse model does have some important limitations. It is technically complex and requires both a source of human tissue and surgical skills for reproducible production of the hu-mice themselves. The effort required to generate the hu-mice limits the throughput of the assay, as there is a limited quantity of the animals themselves. Also, although human tissue is infected, the model still lacks the human-specific targets of several important virulence factors, including streptokinase and superantigens, which are the serum protein plasmin and HLA on bone marrow-derived T cells, respectively. The use of the immunocompromised SCID mouse means that the infection proceeds in the absence of adaptive immunity. However, because the model can faithfully reproduce many features of human impetigo, the hu-mouse model is one of the most useful and accurate models available for analysis of S. pyogenes infection of the superficial skin.

Models of systemic disease

The common element of the various models of systemic disease is that the major mechanism of damage to the host arises from the presence of streptococci actively multiplying in the bloodstream. With the exception of puerperal sepsis (childbed fever), which is now a rare disease, sepsis is a relatively rare manifestation of infection by S. pyogenes, and often occurs after a localized infection in soft tissue (Reglinski & Sriskandan, 2014; Wong & Stevens, 2013; Rantala, Vuopio-Varkila, Vuento, Huhtala, & Syrjänen, 2009). The murine models of systemic infection do have utility for modeling human disease, since growth in human blood in vitro is the classic assay for assessment of S. pyogenes virulence, for testing its ability to evade phagocytosis, and for testing the protective efficacy of antibodies (Lancefield, 1962). In the mouse, the different types of systemic disease models are distinguished from each other by their route of inoculation, which may occur intravenously (IV), intraperitoneally (IP), intranasally (IN), or intratracheally (IT). The IN route has also been adapted for analysis of oropharyngeal colonization (see below). The IV route introduces the bacteria directly into the bloodstream, while the IN and IT routes require that the streptococci invade across a tissue barrier to reach the vasculature. Each of these routes involves a direct injection or instillation of the bacterial inoculum and is technically simple to perform. It has been proposed that the IN and IT routes may yield more useful information, as they bypass the least number of the steps of infection since they require the bacteria to invade across the lung epithelium in order to reach the bloodstream (Roberts, Scott, Husmann, & Zurawski, 2006). However, to improve upon the upper airway model, Husmann et al. infected mice via IT inoculation to bypass the initial site of upper airway colonization and found that infected mice more quickly developed signs of illness within a few days following infection, as compared to IN infection (Husmann, Dillehay, Jennings, & Scott, 1996). Furthermore, examination of these IT-infected mice found a higher frequency of acute, suppurative bronchopneumonia with two-thirds of these mice developing bacteremia (Husmann, Dillehay, Jennings, & Scott, 1996). Therefore, IT inoculation may be a more efficient mechanism to investigate the role of streptococcal virulence factors in inducing significant airway and systemic disease (Roberts, Scott, Husmann, & Zurawski, 2006; Husmann, Dillehay, Jennings, & Scott, 1996; Husmann, Yung, Hollingshead, & Scott, 1997). Virulence of strains is typically determined by monitoring lethality, by using a single concentration of bacteria to infect a group of mice. Differences in virulence can be tested for significance by the Kaplan-Meier survival probability estimate model (Brenot, King, Janowiak, Griffith, & Caparon, 2004). A typical dose for a strain that is virulent in these assays is approx. 10⁵-10⁶ CFU and numerous strains have been found to be virulent, although it has long been recognized that there is no correlation between the virulence of strains in humans and in systemic models (Lancefield, 1962). This likely indicates that the mouse and human have a differential sensitivity to certain S. pyogenes virulence factors. For those strains that are less virulent, useful information can often be acquired by examination of the kinetics of clearance from the vasculature by enumeration of CFUs recovered from highly perfused organs like the spleen, liver, and lungs. The choice of mouse strain and gender are important considerations, as mice with different genetic backgrounds can have significant differences in sensitivity to systemic infection, with male mice being generally more susceptible (Medina, Goldmann, Rohde, Lengeling, & Chhatwal, 2001). The genes responsible for these differences are apparently located outside of the MHC locus (Goldmann, et al., 2005), and an absence of adaptive immunity does not alter the course of infection (Medina, Goldmann, Rohde, Lengeling, & Chhatwal, 2001). In addition to existing strains, transgenic technologies have been employed to expand the types of strains that can be useful for the analysis of pathogenesis in systemic models. The most interesting application of this approach has been to overcome the limitation noted above that results from certain virulence factors of S. pyogenes having evolved to recognize exclusively human targets. In the case of streptokinase, a transgenic hu-mouse has been developed that expresses human plasminogen (Sun, et al., 2004). Analysis of these mice conclusively establishes streptokinase as virulence factor for systemic infection, as these mice have a markedly increased susceptibility to lethal infection, which is dependent on the expression of human streptokinase (Sun, et al., 2004).

Implanted chamber models

In the models discussed so far, pathology results from streptococci damaging the tissue that are directly associated with or by dissemination through the vasculature. However, many S. pyogenes diseases are caused by toxins or other bacterial products that are released by streptococci growing in a localized lesion that then traffic to tissues and organs at a distant site to cause damage (Reglinski & Sriskandan, 2014; Cunningham, 2000; Ralph & Carapetis, 2013; Wong & Stevens, 2013). In some cases, these diseases can be modeled in mice by using an implanted chamber model. The approach is to surgically-implant a “cage” made of steel or Teflon into the subcutaneous tissues on the mouse flank. The incision is allowed to heal over the course of several weeks to allow a fibrous capsule to form around the implant. The infection is initiated by injection of the bacteria through the skin and into the inside of the cage, which allows the influx of nutrients and the release of streptococcal products, but the retention of streptococcal cells. Implanted chambers have been used to produce a glomerulonephritis-like disease that is characterized by deposition of streptokinase and host complement components in the glomerular basement membrane (Nordstrand, Norgren, Ferretti, & Holm, 1998) and to monitor toxin production in an in vivo environment (Kazmi, et al., 2001). A similar approach using rabbits has been employed for modeling streptococcal toxic shock syndrome (Schlievert, 2007).

Footpad model

Most animal models of infection quantitatively assess events that are associated with the very late or terminal stages of disease, including bacterial proliferation, damage to tissue and lethality. However, many virulence factors act during the very early stages of infection to help establish a foothold for the bacteria in the infected tissue. Since there are relatively few bacteria present at this stage, and often no gross alterations to tissue, it can be a challenge to observe the streptococcal cells or to measure any parameter of early infection to assess the contribution of specific virulence factors. Recent advances in imaging have begun to open a window into this critically important aspect of S. pyogenes pathogenesis. One approach is to use the power of two-photon microscopy, a technology with the resolution to image streptococcal bacteria in live un-fixed tissue at a depth of several millimeters, to examine bacteria over the first hour following their introduction into tissue. In one application, fluorescent dyes were used to label bacterial cells and blood vessels in combination with LysM-eGFP mice, whose macrophages and neutrophils express eGFP. Two-photon microscopy was used to examine the behavior of neutrophils immediately following the injection of wild-type and mutant bacteria into the mouse footpad (Lin, Loughman, Zinselmeyer, Miller, & Caparon, 2009). In addition to positional information both inside and outside of vessels, imaging in real-time allows for the quantitation of neutrophil velocity and meandering for assessment of chemotactic behavior. The application of this method revealed a previously unrecognized role for streptolysin S in altering neutrophil extravasation (Lin, Loughman, Zinselmeyer, Miller, & Caparon, 2009). While this is a promising technique, its general adoption has been slowed by its requirement for expensive and highly specialized equipment.

Oropharyngeal colonization models

A major reservoir for S. pyogenes in humans is the posterior oropharynx, especially the mucosal-associated lymphoid tissue known as Waldeyer’s ring, which includes the tonsils. This tissue is the site of one of the most common manifestations of S. pyogenes infection (pharyngitis), and as a result, there has been considerable interest in developing animal models to attempt to replicate and study this disease process (Cunningham, 2000). However, rodents lack a Waldeyer’s ring homologue in the pharynx, and experimental colonization of the rodent oropharynx has been both challenging and limited primarily to only a few select S. pyogenes strains. Early experiments by Rebecca Lancefield’s group (Hook, Wagner, & Lancefield, 1960) took advantage of a mouse-pathogenic strain of S. pyogenes, the M type 50 strain B514/33 that was originally isolated from an epizootic infection of a mouse colony. More recent experiments using this lineage have used strain B514-Sm, a spontaneous streptomycin-resistant derivative of the original strain, whose high level of antibiotic resistance allows direct selection for the determination of recovered CFUs. Strain B514-Sm seems exceptionally capable of colonizing the murine oropharynx, and approximately 60% of inoculated mice remain colonized for over a month. Mice are infected by IN inoculation with 10⁷ CFU in a 10 μl normal saline drop that is placed into one nostril and inhaled, which distributes the inoculum throughout the upper airway. Colonization is monitored by throat swabs that are resuspended and diluted in saline and plated onto selective media (THY agar with 1000 μg/ml streptomycin) (Roberts, Scott, Husmann, & Zurawski, 2006). As discussed above, with IN inoculation some mice will develop bronchopneumonia and die; however, the dose-response ratio is non-linear, given the random distribution of the inoculum in the airway. More recently, the group of Alam et al. refined delivery by the IN route to expand the number of strains that can be analyzed in the model, and found that the volume of the inoculum is a critical variable. They found that IN inoculation of a serotype M18 strain in a volume of 2.5 μl to each nare of a mouse resulted in consistent nasopharyngeal deposition without lung involvement or significant nasal clearance (Alam, Turner, Smith, Wiles, & Sriskandan, 2013). In addition, they used bioluminescence imaging of a strain that was engineered to express luciferase as a non-invasive method for monitoring the progress of infection (Alam, Turner, Smith, Wiles, & Sriskandan, 2013). The strengths of these models include the ability to colonize the murine upper airway and oropharynx, which is a site that is highly relevant to human streptococcal disease. This advantage has been used to show that mutation of the CovR/S virulence regulator in a serotype M18 strain attenuates infection of the oropharynx, in contrast to the hypervirulence of mutants in subcutaneous models (Alam, Turner, Smith, Wiles, & Sriskandan, 2013). Some disadvantages of these models include a limited number of S. pyogenes strains capable of colonizing the mouse nasopharynx, that strain B514 has an atypical pattern of virulence gene expression (Yung & Hollingshead, 1996), and the significant differences between human and murine nasopharyngeal lymphoid tissue architecture. However, despite these differences, the murine oropharyngeal model has proven to be useful for analysis of both passive and active mucosal immunization using various streptococcal antigens. For example, passive administration of lipoteichoic acid (LTA) or anti-M protein secretory IgA can significantly reduce colonization in the model (Bessen & Fischetti, 1988b; Dale, Baird, Courtney, Hasty, & Bronze, 1994). Active immunization of the nasal mucosa with several surface proteins, including the hemoprotein receptor Shr (Huang, Fisher, Nasrawi, & Eichenbaum, 2011), the fibronectin-binding protein SfbI (Guzmán, Talay, Molinari, Medina, & Chhatwal, 1999), and various M protein-based vaccines, including peptides conjugated to cholera toxin b subunit or diphtheria toxoid (Bessen & Fischetti, 1988a; Olive, et al., 2006; Olive, Clair, Yarwood, & Good, 2002), have proven to be efficacious in reducing mucosal colonization after subsequent IN challenges of vaccinated mice.

Nasopharyngeal-associated lymphatic tissue colonization model

Despite a lack of Waldeyer’s ring tissue in the pharynx, mice and other rodents have a pair of lymphoid lobes located along the lateral nasopharyngeal wall, termed nasal-associated lymphoid tissue (NALT), which function analogously to human tonsils (Asanuma, et al., 1997; Koornstra, de Jong, Vlek, Marres, & van Breda Vriesman, 1991). To investigate whether NALT may be a more appropriate model of S. pyogenes airway colonization, Pat Cleary’s group infected 8- to 11-week-old BALB/c mice IN with 5 x 10⁸ CFU of S. pyogenes clinical isolate strain 591 (serotype M49), strain 90-226 (serotype M1), or derivatives (Park, Francis, Yu, & Cleary, 2003; Park, et al., 2004). Their group determined NALT to be a significant site of S. pyogenes colonization following intranasal inoculation, and by using strain 591, was able to recover viable CFUs, both intra- and extra-cellularly, for about 8 days post-inoculation (Park, Francis, Yu, & Cleary, 2003). They also introduced bioluminescent imaging as a non-invasive method for the assessment of colonization, using a S. pyogenes strain that was modified to express luciferase (Park, Francis, Yu, & Cleary, 2003). An obvious advantage of this model is the ability to directly examine adaptive immune responses following S. pyogenes colonization of the NALT tissue, which has proven useful for studies that have investigated potential vaccine candidates. The Cleary group has shown that following intranasal inoculation with S. pyogenes, NALT develops a rapid and prolonged activation of CD4⁺ T cells of the IL-17-producing Th17 subclass, and that antigens of S. pyogenes, including sortase A, are capable of inducing an antibody-independent, neutrophil-based immune response that promotes the clearance of NALT colonization (Park, et al., 2004; Wang, et al., 2010; Dileepan, et al., 2011; Fan, et al., 2014). Some disadvantages of this model include a short duration of colonization, a limited number of S. pyogenes strains with demonstrated activity in the model, a higher degree of technical difficulty in isolating NALT tissue for experimental analysis, and a requirement for specialized equipment when using bioluminescent imaging.

Vaginal colonization model

Given the limitations of the oropharyngeal and NALT models for carriage, with their limited number of S. pyogenes strains capable of causing infection and relatively short duration of infection, a model capable of sustaining a longer period of colonization is required. As a result, the estradiol-primed mouse vaginal colonization model was developed in C57BL/6 and BALB/c mice and was based on previously developed vaginal colonization models for group B streptococcus, Neisseria gonorrhoeae, and Candida albicans (Watson, Nielsen, Hultgren, & Caparon, 2013; Sheen, et al., 2011; Jerse, 1999; Fidel, Lynch, & Sobel, 1993). While not as prevalent in human female genitourinary disease as some of the other previously mentioned pathogens, S. pyogenes is nonetheless a significant cause of bacterial vulvovaginitis and puerperal sepsis (or “childbed fever”), and asymptomatic vaginal carriage has been linked to nosocomial transmission from healthcare workers to patients, causing wound infections (Stamm, Feeley, & Facklam, 1978; Berkelman, et al., 1982; Stricker, Navratil, & Sennhauser, 2003; Anteby, Yagel, Hanoch, Shapiro, & Moses, 1999). In each of the murine vaginal models, pathogen colonization is enhanced and carriage is extended by one or more administrations of up to 0.5 mg ethinyl estradiol dissolved in sterile sesame oil given 24 to 48 hours prior to vaginal inoculation of ~1 x 10⁶ CFU S. pyogenes in 20 μl of PBS. Estradiol supplementation syncs all of the mice into the estrous phase of the estrous cycle, a period associated with epithelial cell proliferation and minimal inflammatory cell infiltrate; this phase seems to be when mice are the most receptive to colonization with bacterial or fungal pathogens in the vaginal tract, which is likely due to a combination of a hospitable epithelial cell bed and estradiol-induced anti-inflammatory properties. The estrous phase can be monitored by microscopic examination of vaginal lavages with sterile PBS stained with a modified Wright-Giemsa stain and differential counting of epithelial cells and leukocytes (Watson, Nielsen, Hultgren, & Caparon, 2013; Marcondes, Bianchi, & Tanno, 2002). Vaginal lavages can be serially diluted and plated onto selective media to which the S. pyogenes strain is resistant (such as Todd Hewitt yeast extract (THY) agar, supplemented with 1000 μg/ml streptomycin); this allows for the determination of streptococcal colony counts over time (Figure 3). With S. pyogenes, many strains from a diversity of backgrounds will continue to colonize the murine vagina over a period of several weeks to months, with strains exhibiting variability in the overall rate of clearance from the vaginal mucosa (Watson, Nielsen, Hultgren, & Caparon, 2013) (Figure 3). The significant duration of mucosal carriage this model permits allows for the investigation of the role of streptococcal virulence factors and regulatory mechanisms important for chronic mucosal persistence. The model previously demonstrated attenuated vaginal mucosal carriage of strains that lacked the carbon catabolite repressor, CcpA, and the arginine deiminase (Arc) operon, which suggests these two important streptococcal factors are involved in promoting mucosal survival (Cusumano, Watson, & Caparon, 2014; Watson, Nielsen, Hultgren, & Caparon, 2013). A major strength of the vaginal model is that carriage is asymptomatic; mice do not exhibit signs of distress or pain and do not lose weight or fur ruffles, among others (Watson, Nielsen, Hultgren, & Caparon, 2013). Furthermore, the ability to sample vaginal lavages for CFU counts, cytokine levels by ELISA, and other inflammatory cell markers implies that fewer mice are required to complete the experiment, given that mice can be sampled serially over time without requiring sacrifice. Disadvantages of this model include the need for a S. pyogenes strain carrying a selectable marker (namely, a spontaneous mutation in rpsL that produces streptomycin resistance) to eliminate normal vaginal flora, and the requirement for mice to have received prior estrogenization, which may interfere with certain immune responses. Overall, the murine vaginal carriage model should permit investigation of factors that are critical to influencing host-pathogen interactions and persistence at the mucosal surface, a key immunologic barrier to disease.

Figure 3.

Vaginal model of asymptomatic mucosal carriage. A. Presented is an H&E-stained cross-section of vaginal wall from an estradiol-treated mouse colonized with S. pyogenes HSC12 for 8 days (magnification, 40X). Anatomy landmarks include vaginal lumen (more...)

Oropharyngeal colonization

A variant of the murine models described above has been developed using rats to expand the number of streptococcal strains accessible to investigation. In this model, 8- to 12-week-old Fischer CDF344 rats were treated with streptomycin (1000 μg/mL) in their drinking water to disrupt their normal flora and then inoculated either IN or orally for pharyngeal colonization (Hollingshead, Simecka, & Michalek, 1993). IN infection utilizes 20 μl inocula, while oral inoculations are 50 μl, with both containing ~10⁸ CFU. The Fischer rat was found to support colonization with additional strains of S. pyogenes, as compared to the mouse oropharynx model. For example, Hollingshead et al. colonized Fischer rats with the M6 serotype strain JRS4, isogenic M-negative mutant derivatives of JRS4, and an additional unrelated M6 serotype strain S43-29R; in these experiments, strains that expressed M protein exhibited significantly longer oropharyngeal carriage than isogenic M-negative strains (Hollingshead, Simecka, & Michalek, 1993). The duration of colonization in the rat model was several weeks for some individual rats, although the numbers of CFUs recovered from throat swabs (which ranged from a few to ~200 CFUs per rat) remained relatively small over the course of the experiment. Mice colonized with the M6 serotype strain developed serum and salivary antibody responses directed against the M protein, which indicates that the model may have utility for the analysis of the development of serotype-specific immunity. A limitation to the model is that due to streptomycin treatment of the rats, it requires that the infecting S. pyogenes strain be resistant to streptomycin, although spontaneous streptomycin-resistant derivatives of many S. pyogenes strains can readily be obtained with a single-hit frequency in rpsL, the gene that encodes the 30S ribosomal protein S12 (Watson, Nielsen, Hultgren, & Caparon, 2013).

Autoimmune carditis

In addition to the many diseases it causes by active multiplication and direct damage to tissues or toxigenic diseases, S. pyogenes can also cause a plethora of diseases with an autoimmune character (Cunningham, 2000; Ralph & Carapetis, 2013). One of the most prominent of these is acute rheumatic fever (Cunningham, 2012). The heart is particularly sensitive and damage to it is manifested as carditis and valvulitis. The latter is the cause of most morbidity and mortality that results from mitral valve stenosis and aortic regurgitation. Most evidence indicates that this pathology results from autoimmunity that is triggered by a response against S. pyogenes antigens that cross-react with heart tissue. For modeling these diseases in the laboratory, the Lewis rat has been the species of choice. Since the disease is a sequela to an active infection and is driven by adaptive immunity, rats are not infected, but are immunized with S. pyogenes antigens that are implicated as cross-reactive with human tissues. Immunization with streptococcal M protein, a surface protein that is the serotype-specific antigen of S. pyogenes, results in the development of a myocarditis and a valvulitis. The pathology that develops resembles human rheumatic disease, including the formation of Anitschkow cell-containing granulomas (called Aschoff bodies) and infiltration of CD4+ and CD8+ T cells into the valvular lesions (Gorton, Govan, Olive, & Ketheesan, 2006; Kirvan, Galvin, Hilt, Kosanke, & Cunningham, 2014; Quinn, Kosanke, Fischetti, Factor, & Cunningham, 2001). Cross-reactive epitopes have been mapped by using peptides that correspond to defined regions of the M protein, both by direct immunization and by the adoptive transfer of peptide-reactive T cells to naïve rats to induce valvulitis (Kirvan, Galvin, Hilt, Kosanke, & Cunningham, 2014).

PANDAS

Acute rheumatic fever is only one of several autoimmune-type diseases that are caused by S. pyogenes (Cunningham, 2000). Of these, Sydenham’s chorea, obsessive-compulsive disorder, Tourette’s syndrome, and pediatric autoimmune neuropsychiatric disorders associated with streptococcus (known as PANDAS) all have a prominent neurological or neuropsychiatric component (Esposito, Bianchini, Baggi, Fattizzo, & Rigante, 2014; Macrì, Onori, Roessner, & Laviola, 2013). As with rheumatic fever, it is hypothesized that autoantibodies that react with neuronal tissue in the brain are produced against cross-reactive S. pyogenes antigens during an active infection. These antibodies persist, long after the initial streptococcal infection is cleared, to produce chronic symptomatology. The Lewis rat has also been used to model the pathological consequences of streptococcal-induced neuronal autoantibodies (Lotan, et al., 2014). In this model, rats that are immunized with a whole cell lysate of S. pyogenes proceed to develop quantifiable alterations to several behavioral and motor functions, as compared to sham-immunized animals. Serum from these animals reacts with brain tissue in vivo and to dopamine and serotonin receptors in vitro. Purified IgG from these cross-reacting sera could at least partially reproduce symptomatology, following its direct perfusion into the striatum region of the brain (Lotan, et al., 2014). Further analysis using this model may help to identify those neuronal structures that are susceptible to streptococcal-induced autoimmunity.

Model of Otitis Media

Interestingly, while S. pyogenes causes less than 10% of the cases of culture-positive acute otitis media (ear infection), it was the most common agent of this disease prior to the 1950s and remains the fourth most common cause of bacterial culture-positive otitis media today (Segal, et al., 2005). Modern analyses have strongly implicated biofilm formation in the pathogenesis of ear infections caused by numerous bacterial species (Post, 2001), and the most powerful analysis of biofilm formation in otitis media has come from the use of the chinchilla ear infection model (Post, 2001; Hong, Pang, West-Barnette, & Swords, 2007). This model has been adapted for analysis of the contribution of biofilm to the pathogenesis of S. pyogenes otitis media (Roberts, Connolly, Doern, Holder, & Reid, 2010). In this model, approximately 10⁵ CFU of the strain of interest are inoculated via transbullar injection into the ear of a chinchilla. Infection is monitored by otoscopy over the course of 7 days. By 2 days, the majority of ears show evidence of tympanic membrane and inner ear inflammation with an associated serous middle ear effusion. Biofilm formation is apparent as macroscopic structures that contain bacterial microcolonies in localized three-dimensional communities (Roberts, Connolly, Doern, Holder, & Reid, 2010). Numerous parameters can be quantitated, including histopathology, the enumeration of CFUs recovered from the macroscopic structures and from the serous effusion, and by monitoring survival of the animals over the 7-day course of infection. Biofilm ultrastructure can be further assessed by electron and fluorescent microscopic analyses of the recovered macroscopic material (Roberts, Connolly, Doern, Holder, & Reid, 2010). This model has been used to show that the transcription regulator Svr enhances biofilm formation in vivo (Roberts, Connolly, Doern, Holder, & Reid, 2010).

Non-human primate model of pharyngitis

Despite the relatively low cost and accessibility of rodent models of S. pyogenes pathogenesis, there has long been a desire to develop non-human primate models of disease to more closely mimic the human condition of pharyngitis. Rodent models have failed to establish significant oropharyngeal colonization with S. pyogenes of any sufficient duration and do not develop evidence of symptomatic infection. Non-human primates offer advantages of a more similar development and components of immune responses to humans, as compared to rodents, as well as easier access to the site of infection, larger blood volumes to conduct biochemical analyses, and the ability to use multiple different strains of S. pyogenes. Initial reports of primate models for group A streptococci began in the early twentieth century and have evolved over the past 100 years to several different non-human primate species, including the baboon, rhesus monkey, chimpanzee, and the cynomolgus macaque (Taranta, Spagnuolo, Davidson, Goldstein, & Uhr, 1969; Vanace, 1960; Watson, Rothbard, Swift, & de Mello, 1946; Krushak, Zimmerman, & Murphy, 1970; Sumby, Tart, & Musser, 2008). S. pyogenes effectively colonizes the oropharynx of these primates, and there is an associated characteristic humoral immune response with type-specific M-protein antibodies detectable in serum (Taranta, Spagnuolo, Davidson, Goldstein, & Uhr, 1969; Vanace, 1960; Watson, Rothbard, Swift, & de Mello, 1946; Krushak, Zimmerman, & Murphy, 1970; Ashbaugh, et al., 2000). For example, in the baboon model, inoculation with ~3 x 10⁵ CFU of a serotype M3 S. pyogenes strain establishes pharyngeal colonization for at least 6 weeks; in contrast, acapsular derivatives or M-protein–deficient mutants of the parent strain were significantly attenuated for carriage (Ashbaugh, et al., 2000). A hyaluronic acid capsule was further shown to be important for oropharyngeal colonization, as expression of the hyaluronic acid synthase (has) operon was significantly upregulated following inoculation into the baboon pharynx (Gryllos, et al., 2001), which likely served to inhibit phagocytosis by approaching leukocytes at sites of inflammation. Transcriptome analysis indicates similar expression profiles of a subset of S. pyogenes genes in the macaque pharynx, as compared with the human pharynx (Virtaneva, et al., 2003). Advantages of the primate model include the extended persistence of S. pyogenes in the primate pharynx, as compared to rodents, and the subsequent development of type-specific humoral immune responses—both features that are believed to be necessary for the development of rheumatic fever. Limitations of this model include the prohibitive expense and limited availability of primate colonies and the fact that baboon and rhesus monkeys do not develop symptomatic pharyngitis following inoculation. On the other hand, chimpanzees and macaques do develop significant pharyngeal erythema, fever, tonsillar enlargement, and palatal petechiae following infection, which is reminiscent of human S. pyogenes pharyngitis (Virtaneva, et al., 2003; Friou, 1950). Given the limited ability of S. pyogenes to colonize the rodent oropharynx, the non-human primate has become, and currently remains, the gold-standard model for upper airway and oropharyngeal S. pyogenes colonization.

Non-human primate model of sepsis

Hypotension and multi-organ failure are the signature features of streptococcal toxic-shock syndrome. This is a devastating disease with mortality rates that can be as high as 80%, despite the availability of aggressive antibiotic and supportive therapies (Stevens, et al., 1996). The disease has a very rapid time-course and patients often succumb even before an S. pyogenes etiology can be ascertained. Thus, a priority for understanding the pathogenesis of this disease is to elucidate the critical early checkpoints that can be exploited to develop therapeutic strategies for blocking the subsequent onset of multi-organ failure. As discussed above, many of the toxins that are critical for the pathogenesis of this disease, including the superantigen toxins, only recognize species-specific targets. Thus, animal models sensitive to these toxins are required for a critical analysis of the pathogenesis of this disease. This requirement has made the non-human primate model of sepsis (Creasey, et al., 1993) particularly attractive for this analysis (Stevens, et al., 1996). In this model, baboons (Papio cynocephalus cynocephalus) under light anesthesia are intravenously infused with a high dose of S. pyogenes (approx. 10¹⁰ CFU of log-phase, washed S. pyogenes cells) over a 2 hour period (Stevens, et al., 1996). Analyses of physiology and blood chemistry are then conducted over a 10-hour period, which results in a mortality rate that approaches 100% (Stevens, et al., 1996). Parameters of infection assessed include body temperature, heart rate, mean systemic arterial blood pressure and serum chemistry, and cytokine profiles (Stevens, et al., 1996). In this model, blocking the cytokine TNF-α using a therapeutic monoclonal antibody significantly improved both mean arterial blood pressure and survival (Stevens, et al., 1996).

Silkworm and wax worm models

The principal advantages of infection models using easily bred invertebrates include that they are simple to perform, are inexpensive, do not require extensive animal husbandry, and offer a relatively high-throughput. One invertebrate model that has been used to analyze the virulence of human pathogens is the silkworm (Bombyx mori). Fatal infections occurred with inoculation of pathogenic pathogens (S. aureus, P. aeruginosa, V. cholerae), but not by non-pathogenic strains (Hamamoto, et al., 2004; Kaito, Akimitsu, Watanabe, & Sekimizu, 2002). Inoculation into the silkworm is through injection into the haemolymph (comparable to the bloodstream) of approximately 9 x 10⁸ CFU of S. pyogenes, followed by monitoring survival for 5 days at 27^oC (Kaito, et al., 2005). A similar model uses the wax worm (Galleria mellonela), whose principal advantage is that unlike for other invertebrates, infections can be conducted at 37^oC (Loh, Adenwalla, Wiles, & Proft, 2013; Olsen, et al., 2010). Both of these models have proven sensitive to known S. pyogenes virulence factors and have been used to identify novel factors (Kaito, et al., 2005; Loh, Adenwalla, Wiles, & Proft, 2013; Olsen, et al., 2010).

C. elegans model

The well-known Caenorhabditis elegans invertebrate animal model has also been used to analyze S. pyogenes virulence (Bolm, Jansen, Schnabel, & Chhatwal, 2004; Jansen, Bolm, Balling, Chhatwal, & Schnabel, 2002). While most C. elegans virulence assays involve growing the worms on a plate inoculated with the pathogen, a liquid assay was developed to increase reproducibility. Two hundred sterile worms were added to ~2 x 10⁷ CFU of S. pyogenes in a 24-well plate and monitored every 24 hours. More than 80% of the C. elegans organisms were killed by S. pyogenes in under 24 hours. However, colonization of the intestine does not occur with S. pyogenes, as when the worms are infected with Enterococcus faecalis strains. Interestingly, killing is not toxin-mediated, as strains with mutations in the genes encoding SLO and SpeB did not inhibit killing. Further analysis determined that killing was mediated by the production of hydrogen peroxide by S. pyogenes, which could be abrogated by the addition of catalase to the cultures (Jansen, Bolm, Balling, Chhatwal, & Schnabel, 2002).

Myonecrosis model

The zebrafish (Danio rerio) infectious disease model has been particularly successful for analysis of S. pyogenes acute, deep tissue, myonecrotic infections. The symptoms observed mimic those reported for S. pyogenes deep tissue infections in humans, including lack of inflammatory cell infiltrates, large aggregates of bacteria dissecting along tissue planes, and extensive areas of necrotic tissue damage. While S. pyogenes does disseminate to the spleen, its systemic spread to other organs is not observed (Neely, Pfeifer, & Caparon, 2002). Experimental evidence suggests that fatality results from toxic shock, and most likely occurs through the systemic dissemination of streptococcal toxins (Benavides, Oberyszyn, VanBuskirk, Reeve, & Kusewitt, 2009; Rosch, Vega, Beyer, Lin, & Caparon, 2008). Inoculation is usually by IM injection into the dorsal muscle, although IP injections have also been analyzed (Phelps, Runft, & Neely, 2009). The IP LD₅₀ for a serotype M14 strain of S. pyogenes in adult zebrafish was 2.5 x 10² CFU, while the IM route resulted in an LD₅₀ of 3 x 10⁴ CFU (Neely, Pfeifer, & Caparon, 2002). Over the course of 24 hours, an IM injection results in the formation of a hypo-pigmented lesion at the site of injection; this is indicative of extensive muscle necrosis that is apparent when the tissues are examined by histology (Figure 4). The lesions continue to enlarge until the death of the animal at 36 to 96 hours post injection. When analyzing pathogenesis with strains that carry mutations in known or putative virulence genes, a dose of 10 to 100 times the LD₅₀ is used to determine survival, in comparison to fish injected with the wild-type strain using the Kaplan-Meier survival probability estimate model to test for significant differences (Neely, Pfeifer, & Caparon, 2002). Colonization can also be quantified over time by euthanization and dissection of the muscle tissue to determine replication and/or clearance of the pathogen at the site of injection. Dissemination to the spleen can be determined by aseptic dissection of the organ, followed by serial dilution plating (Phelps, Runft, & Neely, 2009).

Figure 4.

Zebrafish myonecrosis model. Zebrafish-infected IM with 10⁵ CFU of wild type S. pyogenes strain HSC5 and a HSC5 salY mutant (ΔSalY), as indicated. At 24 hours post-infection, fish are euthanized and further processed as described (98). SalY is (more...)

Multiple strains of S. pyogenes have been successfully used in the zebrafish model, including M59 (Bates, Toukoki, Neely, & Eichenbaum, 2005; Fisher, et al., 2008; Montañez, Neely, & Eichenbaum, 2005), M14 (Brenot, King, Janowiak, Griffith, & Caparon, 2004; Benavides, Oberyszyn, VanBuskirk, Reeve, & Kusewitt, 2009; Rosch, Vega, Beyer, Lin, & Caparon, 2008; Neely, Lyon, Runft, & Caparon, 2003), and M1 strains (Neely M. , 2015). Analyses of these strains have been used to examine multiple characteristics of S. pyogenes pathogenesis, including nutrient acquisition (Bates, Toukoki, Neely, & Eichenbaum, 2005; Fisher, et al., 2008; Montañez, Neely, & Eichenbaum, 2005), virulence gene regulation (Neely, Lyon, Runft, & Caparon, 2003), protein secretion (Rosch, Vega, Beyer, Lin, & Caparon, 2008), tissue-specific gene expression (Cho & Caparon, 2005) and toxin-mediated inhibition of host responses (Benavides, Oberyszyn, VanBuskirk, Reeve, & Kusewitt, 2009; Lin, Loughman, Zinselmeyer, Miller, & Caparon, 2009; Rosch, Vega, Beyer, Lin, & Caparon, 2008; Phelps & Neely, 2007) (Figure 4). However, like any animal model, inoculation with some strains does not cause disease and a notable example is the commonly used M6 strain JRS4 (Neely, unpublished research).

One of the major advantages of this model is that the zebrafish is one of the smallest animals that has both an adaptive and an innate immune system. In fact, analysis of its completely sequenced genome (Genome Research Limited, 2015) reveals a high degree of synteny with the human genome (Postlethwait, Amores, Force, & Yan, 1999) and similarity to the mammalian immune system (for further review, see (Hsu, et al., 2004)). This includes homologs to all of the mammalian Toll-like receptors (Jault, Pichon, & Chluba, 2004; Meijer, et al., 2004) and the complement system (Hsu, et al., 2004). The small size and fecundity of the zebrafish and the relatively low cost and simplicity of natural breeding allows for the production and housing of thousands of animals in a small space. This facilitates large-scale screens, including pathogen-centric signature tagged transposon mutagenesis screens (Kizy & Neely, 2009), and host-centric genetic or small molecule-based screens using embryos or larvae arrayed in 96-well plates or via automated microinjection (Wang, Liu, Gelinas, Ciruna, & Sun, 2007). There is an extensive methodology available for analysis of different life stages of zebrafish, and several of these have been successfully employed for analysis of the effect of S. pyogenes toxins on immune cell infiltration (Lin, Loughman, Zinselmeyer, Miller, & Caparon, 2009). Zebrafish genes can be selectively and transiently knocked-down using morpholino technology (Nasevicius & Ekker, 2000) and extensive collections of modified and mutant zebrafish lines are available. These include lines that are transparent (Ren, McCarthy, Zhang, Adolph, & Li, 2002; White, et al., 2008) and lines where cells of the myeloid lineage express GFP (for one example, see (Gray, et al., 2011)). The Zebrafish International Resource Center (ZIRC) (The Zebrafish International Resource Center, 2006) maintains a searchable database for finding zebrafish strains with specific mutations. Recent developments in genetic technology has made creating transgenic zebrafish with specific genetic knock-outs highly successful, including the use of Zinc-finger nucleases, TALENs, and the CRISPR/Cas9 systems (for further review, see (Schulte-Merker & Stainier, 2014)).

As with other models, the zebrafish model is not without its weaknesses. These include that the development of reagents, like monoclonal antibodies for cells and markers of interest, lags behind those available for murine models, and it has been difficult to establish zebrafish-specific cell cultures in vitro. While zebrafish do have an immune system, some important differences remain. For example, one interesting difference is a lack of discernable lymph nodes, although evidence of a zebrafish lymphatic system was recently reported (Küchler, et al., 2006; Yaniv, et al., 2006). Lastly, the optimal temperature for the incubation of zebrafish is 27–29^oC, which is problematic for those S. pyogenes virulence genes that may be temperature-regulated.