The Role of T Helper 17 (Th17) and Regulatory T Cells (Treg) in the Pathogenesis of Vulvovaginal Candidiasis among HIV-Infected Women

Division of Medical Microbiology, Department of Pathology & Laboratory Medicine, Faculty of Health Sciences, Walter Sisulu University, Private Bag X1, Mthatha 5117, Eastern Cape, South Africa National Health Laboratory Services (NHLS), Nelson Mandela Academic Hospital, Mthatha, South Africa Faculty of Health Sciences, Walter Sisulu University, Private Bag X1, Mthatha 5117, Eastern Cape, South Africa Department of Internal Medicine, Faculty of Medicine, University of Kinshasa, Kinshasa, Democratic Republic of the Congo


Introduction
In HIV-infected patients, de Repentigny et al. in  in 2006 reported on the role played by mucosal and cell-mediated immunity against both oral and oesophageal Candida infections [1,2]. Unfortunately, the systemic cellmediated immune response was found to be of no protective value against vulvovaginal candidiasis (VVC) as demonstrated by in vitro studies and animal models [3]. Studies have shown that TGF-β, a well-known downregulatory cytokine, is constitutively expressed in the mucosal surfaces of the vagina independently of the presence of 1 specific anti-Candida response [3,4].
A study by Shacklet and Anton in 2010 established the presence of 17 as a subset of CD4+ T cells, which are able to secrete IL-17 and IL-22. e authors have underlined that these secreted cytokines induce the occurrence of peptides with anti-Candida activity, hence having the potential to inhibit the development of VVC [5]. It is not well elucidated whether or not the action of 17 cells is reduced due to CD4+ T cells depletion observed during advanced HIV infection, leading to recurrent episodes of VVC. What is known is that numerous HIV particles are further expressed through toll-like receptors when infected leukocytes (neutrophils and macrophages) are attracted into the vaginal mucosa following the initial HIV infection [6,7]. It is however not well understood, whether the secreted IL-17 and IL-22 interfere with the activity of the polymorphonuclear (PMN) cells and macrophages and whether this interference facilitates the development of VVC. e present study sought to quantify plasma and genital cytokines/chemokines associated with increased susceptibility for vulvovaginal candidiasis and to determine which cytokines/chemokines discriminate between HIV-infected and HIV-uninfected women with and without VVC.

Study Participants (Patients).
During a cross-sectional study, which enrolled 200 black African women between June and December 2011, presented each with a presumptive diagnosis of lower genital tract infection (LGTI), only 51 women were finally eligible for participation into this present study after signing a written informed consent [8,9]. e study was conducted in Umlazi D Clinic-a primary healthcare center (PHC) in the South African province of KwaZulu-Natal. Inclusion criteria into the study included age of 18 years and above as well as the presence of symptoms such as vaginal discharge, hypogastric pain, vulval itching, and soreness, as well as dyspareunia. Signs included vulval erythema with or without oedema, excoriations, and/or the presence of thick curdy vaginal discharge. Exclusion criteria included patients' age below 18 years and presence of visible blood in genital area including menses [8]. Commonest causes of LGTIs were screened from genital samples using various microbiological and molecular tests as published elsewhere [8,9]. Briefly, Chlamydia trachomatis and Neisseria gonorrhoeae were tested from cervical swabs using a strand displacement amplification technology (Becton Dickinson Probetec Assays, Sparks, Maryland, USA) [9]. QIAamp DNA mini kits (Qiagen Ltd., Chatsworth, CA) were used to extract DNA of Mycoplasma genitalium, Trichomonas vaginalis, and herpes simplex virus type 2 followed by amplification techniques using in-house PCR method and analysis of the amplicons by gel electrophoresis as previously described [9]. Nugent's score, following Gram stain performed on smears made from vaginal swabs obtained from women's anterior fornix, was used to exclude the presence of bacterial vaginosis (BV), when the score was from 7 to 10 [8,9]. e 51 women retained in this present study were eligible because they were free from Chlamydia trachomatis, Neisseria gonorrhoeae, Mycoplasma genitalium, Trichomonas vaginalis, herpes simplex virus type 2, and bacterial vaginosis, the commonest causes of nonulcerative LGTIs in women. Of the 51 women, 16/26 HIV infected and 15/25 HIV uninfected women were diagnosed with symptomatic VVC as described in the section below, but free of other causes of LGTIs. Additional patients' characteristics were published elsewhere [9]; of the 26 HIV infected women, 11 (42.3%) had CD4+ T cell count <200 cells/mm 3 (of which 8 patients had plasma viral load ≥4 Log 10 copies/mL and 3 had plasma viral load between 1.3 Log 10 and 3.99 Log 10 copies/mL) and 15 (57.7%) had CD4+ T cell count between 200 and 349 cells/mm 3 (of which 14 patients had plasma viral load between 1.3 Log 10 and 3.99 Log 10 copies/mL and 1 patient had plasma viral load <1.3 Log 10 copies/mL). All HIV-infected patients were receiving regimen 1 antiretroviral therapy (ART) made of two nucleoside reverse-transcriptase inhibitors (NRTIs) and one nonnucleoside reverse-transcriptase inhibitor (NNRTI). None of these patients received a protease inhibitor (PI). e mean duration (±standard deviation) on ART for 25/26 HIVpositive patients was 4.2 (±1.6) months. e remaining 1 patient with undetectable plasma viral load (VL < 1.3 Log 10 copies/mL) received ART for 10.7 months. Of 26 HIV-positive women, all 11 with CD4+ T cells <200 cells/mm 3 and 5 with CD4+ T cells between 200 and 349 cells/mm 3 had confirmed symptomatic VVC. T cells for HIV negative patients were not measured.

Diagnostic Criteria of Symptomatic Vulvovaginal Candidiasis (VVC).
From each of the eligible 51 study participants, vaginal swab (Becton Dickinson) was taken from the anterior fornix and subsequently plated onto Sabouraud Dextrose agar with chloramphenicol (BBL TM Becton Dickinson) and incubated at 29°C for 48 h. e numbers of yeast colonies per plate were determined and the 2011 European guidelines were applied for the final diagnosis of symptomatic VVC [8,10,11]. Briefly, the absence of symptomatic VVC was established when women whose genital specimens were collected showed negative microscopy and culture results for Candida or had microscopy negative with less than 10 colonies of Candida per plate [11]. e identification of the Candida isolates at species level as well as the antifungal susceptibility testing of these cultured isolates was not the focus of this study.

Specimen Collection and Measurement of Cytokines/
Chemokines. Blood tubes containing ethylenediaminetetraacetic acid (K 2 EDTA, Becton Dickinson) and non-EDTA reagents were used for the collection of blood samples by venipuncture. A vaginal tampon, Tampax Regular ® (Compak), was used to collect vaginal fluid and then subsequently placed into a sterile container with 10 mL of phosphate buffered saline (PBS; Oxoid Limited Basingstoke, UK) (pH � 6.9), stored at 4°C for a period of 4 hours prior to transport to the laboratory. Tampon containing fluid was expressed by means of an autoclaved wooden tongue depressor and, thereafter, filtered using a 0.22 μm cellulose acetate filter membranes (Sigma) [8].

Statistical Analysis.
Means ± standard error of the mean (SEM) and proportions (percentages) were used to summarize continuous and categorical variables, respectively. Analysis of the variance (ANOVA) differentiated between groups for normally distributed data whilst Mann-Whitney U or Kruskal-Wallis nonparametric tests were used for asymmetrically distributed data. Multiple comparisons were performed using post hoc Bonferroni pairwise tests with a p value of less than 0.05 being considered significant. Canonical discriminant analysis (DA) was used to discriminate group memberships and Mahalanobis distances were used for supporting the classification of variates into distinct groups, with values of Wilks' lambda closer to zero being the evidence for well separated groups. In DA analysis, each variate that was more than 1.96 Mahalanobis distance units from the centroid group had a less than 5% chance of belonging to that group. Box's Test of Equality of Covariance Matrices (with p < 0.001) was used to ascertain the assumption of group homogeneity. Data analysis was performed using SPSS® statistical software version 21.0 (SPSS Inc., Chicago, IL).

Concentrations of Plasma Cytokines/Chemokines Associated with Increased Odds for Symptomatic Vulvovaginal Candidiasis in HIV-Infected and Uninfected Women.
In univariate analysis, Table 1 depicts the mean concentrations of all plasma cytokines/chemokines that displayed significant differences across the 4 groups. Five potential predictor variables (IL-1β, MIP-1β, TGF-β1, TGF-β2, and TGF-β3) were identified by ANOVA as meeting the initial DA assumption of being normally distributed (Table 1). Higher mean levels of plasma IL-1β, TGF-β1, and TGF-β2 were significantly observed in HIV + VVC− group whilst higher mean levels of plasma MIP-1β and TGF-β3 were seen in HIV + VVC + group.
After adjusting for multiple comparisons, those 5 potential predictor variables were confirmed to display significant differences across the study groups using post hoc Bonferroni pairwise tests ( Figure 1): the HIV + VVC + group had a significantly lower mean level of IL-1β compared to the mean level of the HIV + VVC− group (p < 0.0001) and the HIV−VVC− group (p � 0.023), suggesting that lower concentrations of IL-1β were strongly associated with symptomatic VVC in HIV-infected women (p � 0.0003) (Figure 1(a)).
When comparing the mean levels of MIP-1β, the HIV−VVC− group had a significantly lower mean concentration than the mean value of the HIV + VVC− group (p � 0.012) and more importantly of the HIV + VVC + group (p < 0.0001), suggesting that lower concentrations of MIP-1β were strongly associated with the absence of symptomatic VVC in HIV-uninfected women (p � 0.0004) (Figure 1(b)). e highest mean concentration of TGF-β1 was observed in the HIV + VVC− group as compared to the mean level of every other group (p � 0.0001) (Figure 1(c)). A significant difference in mean levels of TGF-β2 was observed when comparing the HIV−VVC− group with the HIV−VVC + group (p < 0.006) and also between the HIVVVC− group and the HIV + VVC− group (p � 0.002) (Figure 1(d)).
e mean level of TGF-β3 was significantly higher (p � 0.013) in the HIV + VVC + group and also in the HIV−VVC + group as compared to the mean level of every other remaining group (p � 0.005), suggesting that higher levels of TGF-β3 were strongly associated with symptomatic VVC irrespective of HIV serostatus (Figure 1(e)).
Canonical discriminant analysis (CDA) was carried out with normally distributed variables that showed significant univariate associations. CDA identified 3 canonical functions and 3 significant variates (MIP-1β, TGF-β1, and TGF-β3) able to discriminate between the 4 study groups.
e obtained 3 discriminant functions had eigenvalues of 3.516, 0.864, and 0.022 with canonical correlations of 0.882, 0.681, and 0.147, respectively. e larger eigenvalue represents more shared variance in the linear combination of variables. Of the 3 identified canonical functions, the first 2 functions contributed a total of 99.5% of the total variance, which easily stratifies into the 4 study groups (HIV−VVC−; HIV−VVC+; HIV + VVC; and HIV + VVC+). e multivariate Box's M test was used and confirmed that variables followed a multivariate normal distribution, and the variance-covariance matrices were equal (homogeneity of variances/covariances) across the 4 study groups (Box's M (5.49), p (0.57) > α (0.001)). Wilks' lambda value was significant: 0.116; χ2 � 95.78; p < 0.0001 (Table 2).
Functions' coefficients were computed and used in deciding which variable predicted group membership. Comparing the values between groups, the higher coefficient means the predictor variable is attributed more to that group (Table 3). TGF-β1 strongly predicted HIV + VVC− group while TGF-β3 and MIP-1β predicted HIV + VVC + group as shown in Table 3.
Standardized canonical discriminant function coefficients were used for predicting functions defining group centroids ( Figure 2). e canonical group means (also called group centroids) are the mean for each group's canonical observation scores. e larger the difference between the canonical groups, the better the predictive power of the canonical discriminant function in classifying observations.
Predictor variables maximizing Mahalanobis distances between two closest groups are depicted in Table 4.
Taking into account that each variate that was more than 1.96 Mahalanobis distance units away from a specific centroid group had a less than 5% chance of belonging to that group, from Table 4, it can be concluded that MIP-1β was International Journal of Microbiology     Table 3 that showed MIP-1β and TGF-β3 as significantly predicting the HIV + VVC + group while TGF-β1 predicting the HIV-VVC + group.
Furthermore, after performing the canonical discriminant analysis from all normally distributed variables that displayed significant univariate associations, only one canonical function composed of one cytokine (IL-13), an immune-allergic mediated anti-inflammatory cytokine, was shown to significantly predict a case classification into HIV + VVC + group. at one discriminant function had eigenvalue of 0.48 with a canonical correlation of 0.57. is function contributed 100% of the total variance: Box's M

Discussion
Findings from this study showed that more than one pathogenic mechanism can explain the occurrence of symptomatic VVC in HIV-infected women.

New Insights about Treg and 17 Cells in HIV-Infected
Women with Symptomatic VV. Using discriminant analysis, it was found that mean levels of circulatory TGF-β3 were significantly higher in women with VVC regardless of their HIV serostatus. In addition, after adjusting for multiple comparisons using post hoc Bonferroni pairwise tests, mean concentrations of genital IL-6, IL-8, IL-10, IL-17, and TGF-β3 were significantly higher in HIV-infected women coinfected with VVC. Previous studies have shown that, in the presence of TGF-β3 and IL-6, naïve CD + T cells will differentiate into 17 cells [12][13][14][15][16]. Newly differentiated 17 cells will then subsequently produce more IL-6 in addition to IL-8 and IL-17 as effector cytokines [12][13][14][15][16].
ese authors have also shown that functional 17 cells were able to produce IL-21 and IL-23 as effector cytokines [12][13][14][15][16], but these cytokines were not measured in this present study.
As a major effector cytokine for 17 functionality, IL-17 induces a strong inflammatory response via IL-6 and IL-8 [11][12][13][14][15]. Besides inflammation, IL-6 in return is involved in regulating acute-phase and immune responses as well as hematopoietic functions [17]. On the other hand, IL-8 has a strong chemotactic activity for neutrophils at the site of the infection in addition to the recruitment and activation of other cells such as NK cells and GM-CSF [18]. e production of classical 17 cells obtained via the presence of TGF-β favors subsequent production of higher levels of IL-10 [12][13][14][15][16], as also shown in our present study. Furthermore, the same authors have demonstrated that, in the presence of TGF-β and IL-10, CD4+ T cells will differentiate to regulatory T cells (Treg cells) resulting in the production of more IL-10 that blocks 17-mediated inflammation in return [12][13][14][15][16]. From these findings of the previous authors and our own findings, we can then hypothesize that the obtained Treg cells via IL-10 represent regulatory mechanisms in order to prevent excessive tissue damage from 17 cells via IL-6 and IL-8, suggesting the existence of a new paradigm of 17/Treg mechanisms in HIV infected women with VVC.
In this study, although some 1 cytokines (IL-2 and IL-12) were significantly higher among HIV-infected women with VVC, IFN-c, which is known to be a major 1 effector cytokine, was shown to be significantly low. Liu et al. have shown that when myeloid cells are infected, there is a strong production of IL-12 that subsequently leads to the differentiation of 1 cells, hence the production of IFN-c [19]. However, during some studies on 1-inducing effector cytokines, it was reported that, during various pathological conditions, the role played by IFN-c was not critically important as compared to IL-12 [14,20]. ese discrepant findings regarding CD4-mediated immune responses were only reconciled after 17 cells were discovered [14]. It became evident that IL-12 and IL-23 share the p40 subunit of IL-12, and also, IL-12Rβ1 is a subunit of IL-23R [21]. is was subsequently confirmed by animal studies that showed that mice lacking these IL-12 associated subunits were deficient to both IL-12 ( 1 effector cytokine) and IL-23 ( 17 effector cytokine) [21].
In conclusion, there is 1/ 2 paradigm shift in the pathogenesis of VVC among HIV infected women [12][13][14][15][16]. Of major importance was the observed significant level of TGF-β3, a member of TGF-β family of cytokines. It is well established that the immunoregulatory properties of these cytokines are associated with the absence of anti-Candida cell-medicated immunity in the vagina [22][23][24]. In addition, 17 cells rather than 1/ 2 paradigms were demonstrated to be key players for immune responses in HIV   [12][13][14][15][16]. Studies have also shown that, during the course of HIV infection, there are progressive declines of 17 cells as well as a decrease of 17/Treg ratio [16]. is unbalance seems to be more important in advanced HIV disease [16]. Most of these studies however have been assessing the loss of 17 cells in the gastrointestinal mucosa of HIV-infected patients [16].  [25]. Whilst our findings are keeping with some previously published studies which have shown that inhibition of 17 cell differentiation led to decrease in production of IL-17 with consequent exacerbation of VVC, other authors have suggested that inflammatory response during VVC occurs independently of 17 cell lineage [26][27][28].

4.2.
e Role of IgE-Mediated Hypersensitivity Response. Findings from this study showed that 2-type cytokines (IL-4, IL-5, and IL-13) were significantly increased in VVC women coinfected with HIV. Previous analyses of 2 cytokines in clinical studies showed only a small amount of IL-4, IL-5, and IL-13 in vaginal fluid with no effect on fungal burden [6]. However, controversies were raised by Babula et al. who showed that women with recurrent VVC had increased vaginal levels of IL-4 polymorphism, possibly resulting in reduced production of anticandidal compounds such as nitric oxide and mannose-binding lectin (MBL) [29]. A similar controversy from the literature was from a study that showed that increased IL-4 and IL-10 were associated with higher fungal burden and VVC severity [30,31]. e authors have then hypothesized that the associations between recurrent VVC and the vaginal presence of significant levels of IL-4, IL-5, and IL-13 could be the result of an allergic reaction due to the presence of Candida-specific IgE production and eosinophil stimulation in HIV infected women [30,31]. Moreover, a study by  has shown that the vaginal S100 alarmin response to Candida does not require 17 cell lineage and effector cytokines, suggesting that the inflammatory response during VVC may occur independently of the 17 pathway [27]. erefore, in addition to the hypothesized 17 altered immune response as the pathogenic mechanisms in VVC patients coinfected with HIV, the role played by local hypersensitivity cannot be neglected. Our cohort of HIV infected women with symptomatic VVC had significantly higher levels of IL-4, IL-5, and IL-13 in their genital . Our previous reports using the same cohort of women have shown that HIV infected women whose CD4+ T cells were less than 200/ mm 3 had more than 7-fold high chance of developing VVC as compared to their HIV-negative counterparts as well as HIV-infected women whose CD4+ T cells are above 200 [8]. However, when comparing only HIV-infected women stratified by levels of immunosuppression, the odds increased by 9-fold and 60-fold higher for women with advanced and severe immunosuppression, respectively [11].

Conclusions and Limitations
In HIV-infected women, VVC was primarily and strongly predicted by the presence of circulatory and genital TGF-β3, a key factor for 17 differentiation in the presence of IL-6, which can also differentiate into Treg cells and becomes a strong downregulatory and anti-inflammatory response in the presence of IL-10. In addition, the roles played by Candida-induced inflammatory but nonprotective 17 induced responses in the lower genital tract should not be ignored. Secondary pathogenic mechanisms included the presence of vaginal IL-13, an allergic mediated 2 cytokine, and HIV-induced severe immunosuppression. e latter was supported by the presence of circulatory MIP-1β, a marker of chronic immune activation observed during severe HIV-induced immunosuppression. e limitation of this study is its cross-sectional design. Although our findings on cytokine profiles are suggestive of 17/Treg responses, other cell types such as NK cells or macrophages could have also secreted similar cytokines, and the final confirmation should come from measuring cytokine profiles from post-T cell stimulation and from culture supernatants. Nevertheless, we cannot neglect the value of the obtained findings on cytokines from clinical specimens since they provide predictive clinical information and insights into disease mechanisms. In addition, the performed discriminant function analysis did not take into account CD4 counts of the study participants. e two closest centroid groups were HIV + VVC+ and HIV−VVC+. It is possible that HIV + VVC + women on HAART with CD4 count closer to 350 cells/mm3 could have an immune response relatively comparable to the HIV−VVC + group, thus bringing group centroids for the two study groups closer. IgE and eosinophil levels in the genital tract of women were not measured in order to back up the theory of hypersensitivity. Raw data for this manuscript can be obtained on request. Data Availability e datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethical Approval
Ethical approval was obtained from the Biomedical Research Ethics Committee of the University of KwaZulu-Natal (ref. BE 224/11). Confidentiality was maintained throughout the study. Minor patients were excluded as mentioned before. Permissions to conduct the study were obtained from the clinic management. In order to protect privacy and confidentiality of patients, no names were recorded, and instead, a personalized research number was used for each patient, and only investigators had access to the collected data.

Consent
All patients signed written informed consent.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
TA conceived the study and its design, collected the data, participated in data analysis, and wrote the study proposal and the final draft of this manuscript. BL performed data analysis and interpretation and edited the final draft of the manuscript. 8 International Journal of Microbiology