Levels of genital tract defensins and cytokines differ between HIV-uninfected US and African women

Abstract

Problem

To explore the impact of race and geographic region on biomarkers of HIV risk and vaginal health, differences in soluble immune mediators were measured in US vs. African and US white vs. US black women at enrollment into a phase 2 microbicide trial.

Methods

Levels of soluble mucosal immune mediators and inhibitory activity against E. coli, which may serve as biomarkers of risk for HIV and other genital tract infections, were quantified in cervicovaginal lavage (CVL) collected from HIV-uninfected women in the US (n=73) and Africa (n=73). Differences between groups were analyzed with multivariable logistic regression models for dichotomous variables and linear regression models for continuous variables.

Results

Secretory leukocyte protease inhibitor, lactoferrin, human beta defensins, interleukin (IL)-8 and interferon-gamma induced protein-10 were significantly higher in US compared to African women in multivariable analysis, but only IL-1β was significantly different between US white and black women. E. coli inhibitory activity did not differ among groups in adjusted analyses.

Conclusion

Differences in soluble mucosal immunity between US and African women may play an important role in women’s risk for HIV and other genital tract infections and response to prevention strategies including vaginal microbicides and should be considered in future studies.

Introduction

Racial disparities in the incidence and prevalence of common sexually transmitted infections (STIs) including HIV, HSV-2, gonorrhea, and chlamydia are well described^1,2 and may be attributed to several factors. Variations in socioeconomic status, access to health care and behavioral practices may contribute; however genetic, biologic and other environmental factors may also be important.^3–5 These disparities exist both globally, with 70% of new HIV infections in 2012 occurring in sub-Saharan Africa, and in the US, where black women are 19-fold more likely than white women to be diagnosed with HIV.^6,7

This study, a secondary analysis of samples and data obtained from a phase 2 microbicide trial (Microbicides Trial Network (MTN)-001),⁸ sought to compare the mucosal immune environment in the female genital tract of US and African as well as US white and black women. The goal was to explore the relationship between self-reported race and candidate biomarkers of HIV risk and genital tract mucosal health. Only two prior studies have compared levels of immune mediators between US and African women. One study of HIV-uninfected women from Kenya and the US demonstrated significantly lower levels of interleukin (IL)-8 and secretory leukocyte protease inhibitor (SLPI) in the genital tract of African compared to US women.⁹ The other study of HIV-infected women found that African women had higher levels of the pro-inflammatory mediators IL-8, IL-1β and IL-6 as well as SLPI when compared to their US counterparts.¹⁰ Increased levels of pro-inflammatory mediators in the genital tract of HIV-negative and HIV-positive women suggest increased risk for HIV acquisition or HIV transmission, respectively. Currently, it is unclear why there are disparate results in the genital milieu between US and African women and the epidemiological associations of HIV risk. Neither of the prior studies examined the inhibitory activity of genital tract secretions against E. coli.

Prior work suggests that E. coli inhibitory activity may reflect contributions from antimicrobial peptides and other immune molecules secreted by epithelial and immune cells as well as vaginal microbiota including Lactobacillus species that may be protective for HIV acquisition.^11–15 In healthy US adult women (n=99), the median E. coli inhibitory activity was ~68% and was associated with proteins secreted primarily by Lactobacillus (L.) crispatus and L. jensenii.¹² Women with high E. coli inhibitory activity had a predominance of L. crispatus and culture supernatants from L. crispatus exhibited greater E. coli inhibitory activity compared to supernatants obtained from L. iners or G. vaginalis.¹⁶ In contrast, at-risk US adolescents had E. coli inhibitory activity that was significantly lower than adults (48%, n=23 adolescents) and there was a relative paucity of L. jensenii in vaginal swabs from the adolescents, but a modest correlation between E. coli inhibitory activity and L. jensenii in both groups.¹⁷ Similarly, in African women at-risk for HIV (HPTN 035, n=132), the mean E. coli inhibitory activity was 29.5 ± 35.7% and higher activity was associated with an increased risk of HIV acquisition.¹⁸ Thus, in populations with a Lactobacillus-dominant vaginal microbiome, E.coli inhibitory activity may reflect a healthy mucosal immune environment, whereas in populations with a paucity of Lactobacillus, the antimicrobial activity may be mediated primarily by inflammatory proteins and associated with an increased risk of HIV. Building from prior work, the current study explored whether levels of immune mediators and E. coli inhibitory activity differ by race with the hypothesis that these differences may contribute to baseline risk for HIV and STI acquisition.

Methods

Study Protocol

MTN-001 was a 21-week, phase 2, open-label, crossover study with participants randomized to different sequences of oral and vaginal tenofovir to assess pharmacokinetics at various anatomic sites as well as adherence and acceptability of the drugs. The results from this study have been published⁸ and the protocol is available at http://www.mtnstopshiv.org/studies/71. An exploratory objective of this study was to assess the impact of oral and vaginal tenofovir on mucosal immunity, evaluating levels of mucosal immune mediators and anti-microbial activity at enrollment and the end of each sequence during the study.¹⁹ This analysis compared differences in E. coli inhibitory activity and levels of select cytokines, chemokines and antimicrobial peptides in CVL between US and African women as well as US white and black women at the baseline (pre-drug) visit. Out of 168 total women enrolled in MTN-001, 144 had complete data and follow up, the results of which were published in the original study.⁸ This analysis includes results from 146 women with available baseline mucosal immune mediator and anti-microbial activity data.

Study Participants and Sample Collection

Women were recruited from clinical research sites and via community outreach in 4 US cities (Birmingham, AL; Cleveland, OH; New York, NY; and Pittsburgh, PA) and 3 sites in Africa (Botha’s Hill and Umkomaas in Durban, South Africa and Kampala, Uganda). All women provided written informed consent prior to study participation. Women were age 18–45 years, HIV negative, sexually active, and using contraception. Race/ethnicity was categorized by self-report and women could report more than one race. US women who reported their race was black in addition to any other race were classified as black. One woman reported her race was white as well as Asian and was classified as white. Women were excluded if they were pregnant or breast feeding, had a gynecologic procedure within 90 days of enrollment or had a clinically apparent gynecologic abnormality on pelvic exam including grade 2 or higher genital lesions, erythema or edema, graded according to the female genital grading table for use in microbicide studies. Women using acyclovir, valacyclovir or tenofovir disoproxil fumarate/emtricitabine were excluded, as were women with uncontrolled active or chronic medical conditions including immunologic or infectious diseases. Women were also excluded if they reported recent or intended use of a spermicide, vaginal ring, diaphragm or non-study vaginal product or if they were diagnosed with a UTI or an active STI at screening including symptomatic bacterial vaginosis (BV) or vulvovaginal candidiasis, trichomoniasis, chlamydia, gonorrhea, syphilis, or evident genital herpes. BV was diagnosed by Amsel criteria,²⁰ candidiasis and trichomoniasis by saline microscopy of vaginal swabs, chlamydia and gonorrhea by strand displacement amplification (SDA), active genital herpes by culture of lesions and syphillis by serology. CVL was collected at enrollment by washing the cervix and vagina with 10 ml of normal saline and supernatants were aliquotted and stored at −80°C for 1–3 years until assays were performed as previously described.^8,21

Measurement of Genital Tract Immune Mediators and E. coli inhibitory activity

Levels of IL-6, IL-8, IL-1β, interferon-gamma induced protein 10 (IP-10) and macrophage inhibitory protein (MIP)-1α and 3α were measured using a multiplex proteome array with beads from Chemicon International (Billerica, MA), measured using Magpix and analyzed using StarStation (Applied Cytometry Systems, Sacramento, CA). Human beta defensins (HβD) 1, 2 & 3, human neutrophil peptides (HNP) 1-3, lactoferrin, myeloperoxidase (MPO) and SLPI were measured using commercial ELISA kits as previously described¹⁹. Levels below the lower limit of detection (LOD) were set at the midpoint between zero and the LOD. Protein level was determined using a MicroBCA assay (Thermo Fisher Scientific, Rockford, IL). E. coli (ATCC strain 43827) was grown overnight to stationary phase and ~10⁹ cfu/ml were mixed with CVL supernatants or control buffer (20 mmol/L potassium phosphate, 60 mmol/L sodium chloride, 0.2 mg/ml albumin, pH 4.5) and incubated at 37°C for two hours. Mixtures were further diluted in buffer to yield 800–1000 colonies on control plates and added in duplicate to agar enriched with trypticase soy broth. Colonies were counted after an overnight incubation at 37°C and results presented as the mean percentage inhibition relative to control wells.²² Samples from all sites including samples from US white and black and African participants were batched and run together on the same multiplex plates in a central laboratory.

Statistical Analysis

Differences in baseline variables between groups were assessed using Pearson’s Chi-square or Fisher’s exact tests for dichotomous variables and t-tests for continuous variables. Immune mediator and cytokine values were log transformed with the exception of MIP-1α, which was dichotomized at the lower LOD as the majority of the values were below the LOD. Multivariable logistic regression models were used for dichotomous variables and linear regression models were used for continuous variables. Multivariable models comparing US vs. African women and US white vs. US black women controlled for contraceptive method (oral contraceptive pills, injectables, other), presence of HSV-1 and HSV-2 antibody and the presence of ASCUS (atypical cells of undetermined significance) on Pap smear as these are potential confounders of mucosal immunity.

Results

US vs. African women

Eighty four percent (61/73) of the African women self-identified race as either Zulu or Bantu (Table I). Fifty one percent (37/73) of the US women self-identified as white and 48% (35/73) self-identified as Black/African American. US women had significantly higher educational attainment, were more likely to be married and have ASCUS on their Pap smear compared to African women. African women were significantly more likely to be HSV-1 and/or HSV-2 seropositive, more likely to use injectable contraceptives and report a significantly higher number of vaginal sex acts in the past week compared to US women.

Table I.

Baseline Characteristics of African vs. US Women and US Black/AA^* vs. US White

Variable	African	US	p-value	US Black/AA*	US White	p-value
	Data presented as mean (sd) or % (n)
	N=73	N=73		N=35	N=35
Age	31.0 (6.4)	31.1 (7.7)	0.9	33.5 (7.5)	30.0 (7.3)	0.01
BMI	29.5 (6.9)	30.7 (7.2)	0.9	33.3 (7.2)	28.7 (6.6)	0.7
Race/Ethnicity1
Am Indian/Alaska	0% (0)	3% (2)		6% (2)	0% (0)
Asian	0% (0)	4% (3)		0% (0)	3% (1)
Black/AA	0% (0)	48% (35)		100% (35)	0% (0)
White	0% (0)	51% (37)		6% (2)	100% (35)
Latina/Hispanic	1% (1)	8% (6)		9% (3)	6% (2)
Zulu	52% (38)	0% (0)
S Afr Indian	8% (6)	0% (0)
Colored	7% (5)	0% (0)
Bhaca	1% (1)	0% (0)
Bantu	32% (23)	0% (0)
Finished secondary school or higher	22% (20)	88% (64)	<0.001
Attended college				63% (22)	89% (31)	0.04
Monthly Income2				$1704 ($1231)	$3386 ($2151)	<0.001
Contraception
Pills	16% (12)	41% (30)	<0.001	29% (10)	51% (18)	0.02
Injectables	74% (54)	14% (9)		23% (8)	3% (1)
Other	10% (7)	47% (34)		49% (17)	46% (16)
Married	36% (31)	64% (56)	<0.001	86% (30)	66% (23)	0.09
Number of vaginal sex acts in past week	3.0 (2.2)	2.1 (2.0)	0.008	2.0 (2.4)	2.1 (1.5)	0.8
Any unprotected sex in past week	51% (37)	52% (38)	0.9	49% (17)	54% (19)	0.6
BV – Amsel’s criteria	11% (8)	4% (3)	0.1	9% (3)	0% (0)	0.1
Vaginal pH	4.4 (0.5)	4.3 (0.6)	0.6	4.6 (0.7)	4.1 (0.5)	0.002
ASCUS (at screening)	0% (0)	16% (7)	0.001	9% (3)	11% (4)	1
HSV-1 positive	85% (62)	49% (36)	<0.001	60% (21)	37% (13)	0.02
HSV-2 positive	66% (48)	36% (26)	<0.001	57% (20)	17% (6)	<0.001

¹Women could provide more than one answer

²Income provided in local currency therefore comparisons not performed

^*AA African American

In univariate analysis, US women had significantly higher levels of the majority of immune mediators measured in the CVL including SLPI, lactoferrin, MPO, HβD-1, 2 & 3, IL-1β, IL-6, IL-8, IP-10, MIP-3α and protein (Table II). US women also had significantly higher E. coli inhibitory activity (31.7% vs. 16.8%) than the African women. In multivariable analysis controlling for contraceptive use, HSV-1 and HSV-2 serostatus and ASCUS on Pap smear, SLPI, lactoferrin, HβD-1, 2, & 3, IP-10, IL-8, and protein remained significantly higher in US women. However, the differences in E. coli activity and the inflammatory mediators IL-6, IL-1β and MIP-3α were no longer significant (Table II). In additional analyses controlling for the variables above as well as marital status and BV by Amsel’s criteria, SLPI, lactoferrin, HβD-1, IL-6 and IL-8 remained higher in US vs. African women. To assess whether the differences in mediators between US and African women were associated with study site, we analyzed log₁₀ mediator levels by site and found that levels from the individual US sites were higher than in the individual African sites for several mediators (Figure 1). When only US black women were included in the comparison, the US vs. African differences remained the same, with US black women having higher levels of the majority of immune mediators (SLPI, lactoferrin, HβD 1 & 3, IP-10, IL-8 and protein) compared to the African women (data not shown).

Table II.

Levels of Mucosal Immune Mediators and E. coli Inhibitory Activity in US vs. African Women [% (n) or mean (sdev)]

Variable	US (n=73)	African (n=73)	Univariate p-value	Multivariate p-value*
Log10 MPO	2.72 (0.93)	2.38 (0.82)	0.02	0.06
Log10 SLPI	5.37 (0.40)	4.90 (0.50)	<0.001	0.004
Log10 Lactoferrin	3.04 (0.52)	2.56 (0.60)	<0.001	0.003
Log10 HBD-1	3.45 (0.39)	2.92 (0.50)	<0.001	<0.001
Log10 HBD-2	3.27 (0.53)	3.02 (0.84)	0.04	0.03
Log10 HBD-3	3.12 (0.45)	2.72 (0.51)	<0.001	0.005
Log10 HNP1-3	4.54 (0.65)	4.37 (0.67)	0.1	0.2
Log10 IL-1β	0.94 (0.88)	0.54 (0.78)	0.005	0.3
Log10 IL-6	0.94 (0.71)	0.51 (0.67)	<0.001	0.09
Log10 IP-10	2.24 (0.83)	1.84 (0.77)	0.004	0.009
MIP-1α > lower limit	51% (37)	53% (39)	0.9	0.6
Log10 IL-8	2.93 (0.53)	2.56 (0.56)	<0.001	<0.001
Log10 MIP-3α	1.71 (0.78)	1.33 (0.82)	0.006	0.1
Log10 protein	2.76 (0.31)	2.28 (0.35)	<0.001	<0.001
E. coli activity %	31.74 (35.73)	16.81 (22.73)	0.003	0.2

^*Model controls for contraceptive method, HSV-1, HSV-2, and Pap smear

Figure 1.

Genital tract levels (log₁₀) of SLPI, lactoferrin, HβD-1, IL-8 and several other mediators were significantly higher in US compared to African women; the majority of US sites (left) had higher median levels than African sites (right). Boxes represent the 25^th to 75^th percentiles, horizontal lines within the boxes represent the median values and the whiskers represent the range from minimum to maximum levels of each mediator.

US White vs. US Black

White US women had significantly higher educational attainment and monthly income as compared to their black US women counterparts (Table I). Black US women were older and more likely to use injectable contraceptives as compared to white US women who were more likely to use oral contraceptive pills. Black US women were also significantly more likely to be HSV-1 and/or HSV-2 seropositive and had a higher mean vaginal pH. In multivariable analysis controlling for type of contraceptive, HSV-1 and 2 serostatus and ASCUS, only IL-1β was significantly higher in US black as compared to US white women (Table III). There was no difference in E. coli inhibitory activity (mean of 32% in both groups).

Table III.

Levels of Mucosal Immune Mediators and E. coli Inhibitory Activity in US White vs. US Black Women [% (n) or mean (sdev)]

Variable	US White Women (n=35)	US Black Women (n=35)	Univariate p-value	Multivariate p-value*
Log10 MPO	2.99 (0.78)	2.45 (1.02)	0.02	0.1
Log10 SLPI	5.39 (0.39)	5.36 (0.42)	0.3	0.7
Log10 Lactoferrin	3.03 (0.52)	3.05 (0.53)	0.9	0.1
Log10 HBD-1	3.51 (0.40)	3.37 (0.38)	0.1	0.5
Log10 HBD-2	3.28 (0.42)	3.32 (0.42)	0.7	0.9
Log10 HBD-3	3.04 (0.40)	3.21 (0.44)	0.1	0.2
Log10 HNP1-3	4.62 (0.63)	4.45 (0.69)	0.3	0.9
Log10 IL-1β	0.84 (0.91)	1.04 (0.90)	0.4	0.04
Log10 IL-6	1.02 (0.81)	0.86 (0.63)	0.4	0.7
Log10 IP-10	2.19 (0.61)	2.23 (1.00)	0.8	0.5
MIP-1α > lower limit	59% (20)	43% (15)	0.2	0.5
Log10 IL-8	2.93 (0.50)	2.94 (0.58)	1.0	0.2
Log10 MIP-3α	1.80 (0.82)	1.62 (0.76)	0.4	0.9
Log10 protein	2.69 (0.27)	2.83 (0.35)	0.07	0.1
E. coli activity %	32.46 (37.77)	32.20 (33.06)	0.3	0.9

^*Model controls for contraceptive method, HSV-1, HSV-2, and Pap smear

Discussion

Despite epidemiologic data documenting increased rates of acquisition of HIV and other STIs in African and African American women compared to US white women, few studies have explored potential biologic mechanisms that might contribute to these disparities. This study found significant differences between US and African women in concentrations of several mucosal immune molecules in the genital tract, including higher levels in US women of antimicrobial peptides such as HβD-1, 2, & 3 and SLPI, all of which have in vitro activity against HIV and other STI and may be protective. Conversely, US women also had significantly higher levels of IL-8 and IP-10, which may promote HIV infection through the recruitment of target cells via their function as chemoattractants for T cells.

Our findings are similar to results obtained previously comparing healthy HIV-negative women in the US (San Francisco, CA), (n=18) and Kenya (n=36)⁹. Consistent with our data, the US women had higher levels of SLPI, IL-8 and GM-CSF. Defensins, lactoferrin and IP-10 were not measured in the previously published study. Conversely, the Kenyan women also had higher numbers of activated CD4+ CCR5+ T cells and activated CD8+ T cells from cervical cytobrush samples as compared to the US women. These contradictory observations (e.g. lower IL-8 and GM-CSF in African women but increased numbers of activated T cells) underscore the complexity of the mucosal immune environment and the need for larger longitudinal studies to better understand HIV/STI risk against a background of a wide variety of soluble and cellular immune mediators.

While there were significant differences in our study between all US women and black US women when compared to African women, the only difference between US black and white women was higher IL-1β in the black US women. These findings suggest that the two US study groups may have greater similarities with respect to genetics, environment, and behaviors than between the US and African groups studied. Racial and geographic differences in the prevalence of the CCR5 Δ32 mutation and several HLA haplotypes have been shown to confer protection from or progression of HIV infection.²³ There are known polymorphisms in the gene for HβD-1 that are associated with susceptibility to HIV; however, whether these polymorphisms are more prevalent in persons of African descent is not known^{24,25 26}. Due to significant genetic admixture in the US population, US women who self-identify as black may have a significant percentage of genes of European (white) ancestry, (up to 20% in some studies);^27–29 therefore, depending on the gene of interest, these women may have more genetic similarity to US whites than Africans. Contributions from genetic, geographic and environmental factors that alter risk for infection have been described in Ethiopian immigrants to Israel³⁰ and in the Italian-Ugandan AIDS Project, which found that a pattern of immune activation promoting HIV infection occurred in both Italian and Ugandan persons living in Africa but not those living in Italy.³¹

Although we detected a 2-fold greater inhibitory activity against E. coli in the US as compared to the African women, the significance did not persist in multivariable analysis. Notably, the overall activity in this study was lower than previously observed in low risk US women¹² but consistent with the low levels observed in other studies of at-risk women including a subset of women who participated in HPTN-035¹⁸ and US adolescents¹⁷. We speculate that the lower E. coli inhibitory activity observed in both groups here may reflect low levels of hydrogen peroxide producing lactobacilli and sexual activity. We did not have quantitative cultures to test this, but higher levels of Lactobacillus species were observed in women of European (white) descent compared to women of African descent who had higher levels of anaerobes and BV associated species.^32–34 We observed a trend toward higher rates of BV in the African (11%) compared to US (4%) women (p=0.1); however, this was assessed by Amsel criteria, which may not be as sensitive as Gram stain or molecular measures.^35,36 In addition to the vaginal microbiome, sexual activity may also contribute to low E. coli inhibitory activity in both groups. In ongoing studies, we found that postcoital CVL has significantly less E. coli inhibitory activity reflecting in part the enhancing effects of seminal plasma on E. coli growth (Nakra et al, mspt submitted). Larger studies with well-defined diverse populations are needed to more accurately define the factors contributing to E. coli inhibitory activity and in doing so better assess its utility as a biomarker of HIV/STI risk. Future studies may also allow us to identify newer biomarkers that may more accurately predict HIV/STI risk or could be exploited to augment host defense.

There was a significant difference in the type of contraceptive used between groups with 74% of African women reporting use of injectable contraceptives compared to 14% of US women. Injectable progesterone containing hormonal contraceptives such as depot medroxyprogesterone acetate (DMPA) are widely used especially in populations at high risk for HIV infection such as sub-Saharan Africa and some epidemiologic studies suggest that women using DMPA have an increased risk of HIV acquisition and transmission^37–39. DMPA-induced alterations in mucosal immune mediators, immune cell populations and the vaginal microbiome have been postulated as potential biologic mechanisms linking DMPA use to a potential increased risk of HIV,⁴⁰ however few studies have specifically examined the impact of DMPA on these factors. Decreased expression of SLPI has been demonstrated in endometrial biopsies after treatment with DMPA compared to baseline⁴¹ and immortalized cervical cells treated with MPA and TNF had decreased RANTES and increased IL-8.⁴² In vitro studies have also demonstrated significant upregulation of proinflammatory cytokines after treatment of immortalized vaginal epithelial cells with high doses of MPA and culture supernatants harvested from MPA treated cells were able to increase HIV p24 expression from a latently infected HIV (U1) cell line.⁴³ DMPA may also alter the vaginal microbiome as evidenced by a study which demonstrated a significant reduction in H₂O₂ producing lactobacilli in women on long-term DMPA.⁴⁴ Notably, in the current study, despite the higher use of injectable contraceptives, the African women had lower levels of proinflammatory mediators in CVL compared to US women, and these differences persisted when contraception was included in the model.

This study had a number of limitations. First, in addition to not having samples to assess the microbiome or immune cell populations, the study is limited by its cross-sectional design and limited behavioral, demographic and clinical data. While the African women reported a significantly greater number of sex acts, we did not test for semen and are unable to assess the impact of semen on levels of genital tract immune mediators. There were significant differences in protein levels recovered in CVL between US and African women; protein was not reported in the two prior studies comparing immune mediators between US and Africans.^9,10 We did not “correct” for protein as the individual proteins measured reflect <1% of the total protein in CVL. The differences in CVL protein are unlikely to reflect sampling, processing, or shipping differences since the concentrations of some mediators did not differ between groups. Sampling was performed according to standard operating procedures and assays performed after all samples were shipped to a central laboratory. Due to the smaller sample size in the US white vs. black analysis, we also had lower power to detect differences between groups. Three US women reported more than one race but were classified as either black or white and this should not have significantly impacted our findings. Finally, as the mean age of study participants was 30–33, the results are not necessarily applicable to adolescents and older women.

Conclusions

The findings presented here demonstrate different levels of immune mediators in HIV-uninfected US compared to African women, which may play an important role in defining risk for HIV and STI acquisition. Larger prospective studies of US and African women are needed to accurately measure soluble and cellular mucosal immune mediators, antimicrobial activity, the vaginal microbiome and genetic polymorphisms. The factors contributing to mucosal immunity are complex and dynamic and may be altered by many other factors including age, geography, genetics, hormones, microbiome, concurrent infections and sexual activity and thus measurements of a single molecule must be interpreted cautiously. Defining population differences may contribute to the development of biomarkers that identify women at-risk for HIV and other STI and to the tailoring of prevention efforts.