Experimental Models of Vaginal Candidiasis and Their Relevance to Human Candidiasis

The rat model.The rat model of vaginal candidiasis has been in use since early 1960, mostly with the purpose of assessing the anticandidal activity of new drugs. A substantial contribution to the physiology of the model and its strong dependence on pseudoestrus maintenance was provided first by Sobel and Muller (14) and then by Kinsman and Collard (15), who showed the chronic nature of the infection and provided details about ovariectomy and the estrogen usage that is required for infection to occur. The burden of infection was assessed by enumeration of CFU in the vaginal fluid or in excised vaginal tissue following an intravaginal Candida challenge. In the experiments conducted by the investigators referred to above, there was no evidence of strong leukocyte infiltration in pseudoestrus animals challenged by the fungus, and there were no appreciable or measurable clinical signs of disease. Candida pathogenicity and host responses at the vaginal level were mostly addressed by Cassone and collaborators, who highlighted the nature of the various host immunoeffectors involved in the acquired protection and showed the critical role played by the secretory aspartyl proteinases (Sap) in this model (7, 16–18). An intense hyphal growth with biofilm adhering to the vaginal epithelial cells (VEC) was the dominant cytological aspect of the infection, with evidence for macrophages rather than neutrophil infiltration (19). A detailed description of the model has been reported elsewhere (20). Among the immunoeffectors shown to provide anticandidal protection in this model, antibodies against a major Sap (Sap2) and surface mannoproteins were prevalent, although a degree of protection could be conferred by adoptive transfer of both B and T cells from Candida-immunized rats to naive animals. Anti-Sap2 antibodies were shown to inhibit adherence and biofilm formation by Candida on the vaginal epithelial surface (7, 19). Together with some clinical data from RVVC subjects (16), these studies supported the notion of anti-Candida vaccination in women affected by RVVC (reviewed in reference 4).

The mouse model.A second and more popular model of vaginal candidiasis is a murine model, particularly that developed and adopted by Fidel and collaborators (21–26). Advantages of this model over the rat one are lower cost, easy handling, and great availability of genetically modified animals. In addition, no ovariectomy is necessary for infection, as estrogen administration alone is sufficient for pseudoestrus induction. As in rats, the infection is evaluated and monitored by determination of colony-forming counts (CFU) of fungal cells in the vaginal fluid or in the tissue of excised vagina following intravaginal Candida administration. Fidel and collaborators showed the typical estrogen dependence of Candida infection, the changes of the surface vaginal epithelium caused by the pseudoestrus, the fungal biofilm on vaginal epithelium, and the partial refractoriness to a subsequent reinfection, as in rats (22–25). Striking differences from the rat model regarding mechanisms of immune responses and inflammation have been reported in most of the studies mentioned above. In mice, the fungal intravaginal challenge causes a strong inflammatory exudate with dominance of polymorphonuclear cells (neutrophils) and the production of neutrophil-chemoattractive chemokines and cytokines by the VEC (25). No signs of induction of adaptive immune responses and protective T cells or antibodies have been reported; rather, a protective state could be attributed only to the absence of inflammatory response (26). Some authors, using a variant mouse model, have reported the induction of Th17 and of a particular subset of CD4 T cells and have associated the production of cytokines interleukin-17A (IL-17A) and IL-22, beta-defensin, and anti-Candida defensive peptides with at least partial control of the infection, with some similarity to oral candidiasis (11, 27, 28). More recently, the activation of a typical Nod-like receptor protein 3 (NLRP3) inflammasome-mediated cytokine response has been reported to occur (29–31), and Sap production by C. albicans has been associated with vaginal inflammation, indicating that one or more of these enzymes could indeed be the direct or indirect cause of inflammasome activation in the epithelial cells (29, 30, 32). Interestingly, the data on the relevant pathogenetic role exerted by those enzymes in the mouse model are probably the only data that largely match those reported in the rat model and in women (see above), hence collectively suggesting that C. albicans Sap could indeed play a dominant pathogenic role in vaginal candidiasis. Data reported by Naglik's research group, using a similar mouse vaginal infection model, also highlighted the role of some Sap enzymes in contributing to infection, which is in keeping with SAP gene expression in infected women (33, 34). Overall, among the many putative virulence factors of C. albicans, including the transition from dimorphic yeast to hypha, some adhesins (in particular, those of the agglutinin-like sequence [ALS] family) (6, 35), and biofilm formation, Sap enzymes have been those most intensely studied and advocated for as having a role in vaginal candidiasis and perhaps also in some forms of oral infection (36).

Estrogens.The most salient, common aspect that generates confidence in the reliability of the models is that rodent models and human infection are stringently estrogen dependent. Vaginal candidiasis is extremely infrequent in prepubertal and, to some extent, in postmenopausal women, and women with a history of infrequent or rare VVC episodes become susceptible following hormone replacement therapy (3). Estrogens exert a multifunctional permissive role for vaginal candidiasis through a number of both host- and Candida-directed effects (reviewed in reference 4). Both rats and mice must be placed under stable pseudoestrus conditions to establish experimental vaginal infection. In rats, this is usually achieved by ovariectomy followed by estrogen treatment, although recent studies showed that a unique regimen of estrogen treatment may be sufficient to allow C. albicans infection to occur (37). Of note, ovariectomy has been shown to cause a reduction in the number of lactobacillus components of the vaginal microbiota (38; see also below). In mice, estrogen administration is the standard treatment, taking into account the differing susceptibilities of different animal strains to estrogens (39, 40). In both rodents, pseudoestrus induces stage-dependent, profound changes in the vaginal epithelium, with a keratinized surface to which fungal cells strongly adhere and upon which they multiply by hyphal growth and by eventually forming a biofilm (19, 24). Besides these structure-modifying properties, other estrogen-induced biological activities appear to be of special relevance for VVC/RVVC. In particular, estradiol has been reported to inhibit the differentiation of Th17 cells, a subset of CD4 cells supposedly involved in the mucosal antifungal defense; the evidence is stronger, however, in oral and intestinal than in vaginal infection (11, 27, 41, 42). Overall, there is evidence that the responses to estrogen have similarities in rodents and humans. However, the marked differences between rodent models and humans in other relevant biological aspects need to be considered, as shown below.

Commensalism and immune priming.C. albicans is a normal, though minor, component of the human microbiota, whereas it is absent from microbiota of rodents. Candida commensalism shapes human anti-Candida immunity by priming/boosting a strong memory immunity as evidenced by both T and B cell recall responses to fungal antigens in normal, healthy subjects, perhaps also involving a persistent activation of innate anti-Candida immunoeffectors (43–45). It is logical to think that the broad-spectrum immunity described above has a role in the resistance of healthy subjects to candidiasis, but how and to what degree this role is played at the vaginal level is unknown. Women with neutropenia or a low level of CD4⁺ T cells, as in the case of those living with AIDS, who suffer severe systemic or oroesophageal candidiasis, respectively, are not at major risk of vaginal candidiasis in the absence of the known typical risk factors for vaginitis such as pregnancy, diabetes, hormone replacement therapy, and antibiotic therapy, to mention only the major ones (3, 9, 10). Therefore, if commensalism-induced immunity has a functional role in vaginal candidiasis, this must be sought for in the unique immunity of the female reproductive tract rather than in systemic or other nonvaginal mucosal site-derived factors. As outlined elsewhere (4), the vagina is indeed a strongly tolerant body site as it must accept both constantly present (the vaginal microbiota) and occasionally present (semen and fetal) non-self-material. Hence, a delicate balance is maintained that must take into account two equally essential needs for the health of the vagina—immune defense (resistance) and immune tolerance (46). Disease may well occur by loss of tolerance rather than by loss of resistance. Commensalism-shaped immunity is likely to participate in specific mechanisms of C. albicans vaginal immune tolerance through a complex set of immune-regulatory responses. Among them, an experimentally supported concept is that tolerance is regulated by IL-22 and IDO (indoleamine 2,3-dioxygenase)-dependent tryptophan metabolism (kynurenins) (8). Importantly, tolerance can be overwhelmed or bypassed by a high burden of fungal cells, particularly in the hyphal forms, in the presence of the aforementioned typical factors predisposing subjects to candidal vaginitis. On this basis, a reasonable assumption is that in women with RVVC, the tolerance “threshold” is lower and is more easily surpassed than in women not prone to VVC, possibly because of various disease-predisposing gene polymorphisms or other unknown factors (47). A demonstration that the commensalism-immune priming-tolerance axis of the human vagina can be overwhelmed by high burdens of C. albicans cells was provided by studies performed by Fidel's research group in human volunteers (48; see below for further consideration of these studies).

On the other hand, the role of commensalism in vaginal candidiasis is clearly appreciated when the source of infection is considered. There is sufficient evidence to suggest that most of the cases of VVC/RVVC are of an endogenous nature, due to vaginal colonization from the intestine or the cervico-vaginal-vulvar areas or skin or even the persistence of a low number of fungal cells in the vagina (endogenous infection in Candida—primed) (3). This makes an important, perhaps critical, difference from experimental rodent models in which the infection is achieved by direct intravaginal inoculation of high burdens of fungal cells, no signs of infection other than the fungus burden are clinically appreciable, and the animals are immunologically naive to the fungus (exogenous infection in Candida—unprimed). While not impossible, it is unlikely that endogenous infection occurs through the introduction of a single bout of fungal cells (in some studies, millions) from the intestine/perivaginal area. Exogenous infection in animals without any preexisting immunity to Candida can be naturally “proinflammatory,” particularly under neutral pH conditions and in the absence of lactobacilli (as in mice), all conditions which favor rapid and extensive hyphal development. Recent data suggest that lactate, an abundant product of sugar fermentation in a lactobacillus-dominated vaginal environment, downmodulates proinflammatory cytokine production (49). From this point of view, the mouse model would appear to be the more “proinflammatory” model and, in fact, rapid and sustained intravaginal influx of neutrophils accompanied by high levels of inflammasome-dependent IL-1β is typical of this model (29–31). A similar polymorphonuclear leukocyte (PMN) influx is not seen in the rat model, which, under conditions of estrogen treatment, may have an acidic rather than neutral vaginal pH and hyphae that are formed more slowly. Moreover, lactobacilli are present and can play an anti-inflammatory role (50, 51). Only limited macrophage infiltration is seen in these Candida-infected animals, cell-mediated, Th1-type responses to C. albicans are present, and both proinflammatory and anti-inflammatory cytokines are produced during the course of infection (17, 18, 52). This picture is somewhat closer to that of human infection, particularly in the Th1-type response (53), provided that there is no concomitant bacterial infection.

Despite all the findings described above, strong support for the concept that mouse rather than rat models could be a proxy of vaginal candidiasis in women was derived from the already-mentioned studies by Fidel's research group (47) in human volunteers intravaginally challenged with live fungal cells (exogenous infection). In those studies, a correlation was found between fungal intravaginal burden and inflammation, as shown by leukocyte infiltration and induction of a proinflammatory cytokine-rich environment. However, the caveat remains that a direct intravaginal challenge with high fungal loads would not reflect the natural source of infection or the immune responses that follow a naturally acquired infection, particularly in RVVC subjects. In the above-reported exogenous infection, there was no influence of the stage of the menstrual cycle, whereas it is well known that in naturally infected women, VVC essentially occurs or worsens in the luteal phase of the cycle, when estrogen and progesterone levels rise (3). Importantly, and in contrast to rodent models, no keratinization of vaginal epithelial surface has ever been reported in women (3).

The vaginal microbiota and pH issues.It is generally assumed that the vaginal microbiota exerts a role in the vaginal healthy state and protects against Candida infection, although it is unknown to what extent and by which mechanisms this occurs. In healthy premenopausal women, this role is dominantly played by several Lactobacillus spp. which produce lactic acid from sugar substrates and so render the vaginal microenvironment typically acidic (54). In a recent study (55), the vaginal microbiota of mice was seen to be mostly composed of Proteobacteria, Firmicutes, Actinobacteria, Bacteroidetes, and Cyanobacteria, with no evidence of dominant lactobacilli. In contrast, Lactobacillus spp. have been reported to be a major component of rat vaginal microflora and can contribute to make the vaginal pH lower than neutral under estrus conditions, although not as uniformly as in humans (56–58). Ovariectomized rats lost the lactobacillus component of the vaginal microbiota but regained its full composition, and its acidic condition, as a consequence of estrogen treatment (38). Using an ex vivo model of the cervicovaginal epithelium, Abramov and collaborators (50) have recently shown that Lactobacillus crispatus, a major component of the vaginal microbiota, binds to epithelial cells and induces NF-κB activation but does not induce expression of innate immunity mediators and proinflammatory cytokines such as IL-8, IL-1β, IL-1α, and tumor necrosis factor alpha (TNF-α). L. crispatus 2029 inhibited IL-8 production in epithelial cells and increased production of IL-6 working as anti-inflammatory cytokine, hence maintaining the homeostasis of female reproductive tract.

Overall, the studies performed so far highlight important differences between rats and mice treated with estrogens with regard to the pH of both the vaginal microenvironment and the vaginal microbiota and overall immune responses. However, rats appear to be less distant than mice from women in some of those vaginal parameters. In particular, it is difficult to conceive that similar immune-pathogenetic mechanisms are taking place at such different pH values as 4.5 and 7. Nonetheless, we should recognize that the “real” pH of the vaginal microniches where infectious foci develop and immunoeffectors are recruited is not known and could be quite different and dynamically variable from the “static” one measured in the vaginal fluid.

Table 1 presents a short summary of the main similarities and differences between rat and mouse models, as well as between rodent models and human candidiasis.