The authors have declared that no competing interests exist.
Conceived and designed the experiments: LKM LW SLH. Performed the experiments: LW CC BJM. Analyzed the data: LW SK LKM. Contributed reagents/materials/analysis tools: CC BJM. Wrote the paper: LW LKM. Edited the manuscript: CC BJM SLH.
The cervicovaginal fluid (CVF) coating the vaginal epithelium is an important immunological mediator, providing a barrier to infection. Glycosylation of CVF proteins, such as mucins, IgG and S-IgA, plays a critical role in their immunological functions. Although multiple factors, such as hormones and microflora, may influence glycosylation of the CVF, few studies have examined their impact on this important immunological fluid. Herein we analyzed the glycosylation of cervicovaginal lavage (CVL) samples collected from 165 women under different hormonal conditions including: (1) no contraceptive, post-menopausal, (2) no contraceptive, days 1-14 of the menstrual cycle, (3) no contraceptive, days 15-28 of the menstrual cycle, (4) combined-oral contraceptive pills for at least 6 months, (5) depo-medroxyprogesterone acetate (Depo-Provera) injections for at least 6 months, (6) levonorgestrel IUD for at least 1 month. Glycomic profiling was obtained using our lectin microarray system, a rapid method to analyze carbohydrate composition. Although some small effects were observed due to hormone levels, the major influence on the glycome was the presence of an altered bacterial cohort due to bacterial vaginosis (BV). Compared to normal women, samples from women with BV contained lower levels of sialic acid and high-mannose glycans in their CVL. The change in high mannose levels was unexpected and may be related to the increased risk of HIV-infection observed in women with BV, as high mannose receptors are a viral entry pathway. Changes in the glycome were also observed with hormonal contraceptive use, in a contraceptive-dependent manner. Overall, microflora had a greater impact on the glycome than hormonal levels, and both of these effects should be more closely examined in future studies given the importance of glycans in the innate immune system.
The mucosal lining of the female genital tract provides a robust barrier to infection from pathogens such as HIV-1 [
Multiple factors may influence the glycomic composition of the CVF including hormones and vaginal microflora. Oral contraceptives have been shown to regulate the glycosylation of serum glycoproteins such as α1-acid glycoprotein [
To date, no systematic study has examined the effects of exogenous hormones and microflora on the vaginal glycome. Herein, we utilize lectin microarrays [
Following Institutional Review Board approval by the University of Pittsburgh (#PRO11020218), written informed consent was obtained from subjects enrolled in our study. Women were excluded if: they were breastfeeding or pregnant; presented vaginal symptoms; had a hysterectomy; had been diagnosed with any cervical or vaginal infections or had used any antimicrobials in the past 14 days; had used any vaginal devices or vaginally-applied products (excluding tampons) in the past week. Upon enrollment the women had: an OraQuick advance rapid HIV test; a pregnancy test; their demographic information was recorded; height and weight taken and medical, gynecologic and sexual histories taken. Cervicovaginal lavage (CVL) was collected from 169 women. 4 of 169 samples were excluded.
These women were characterized into six groups: (1) post-menopausal; (2) not contracepting on days 1–14 of the menstrual cycle; (3) not contracepting on days 15–28 of the menstrual cycle; (4) combined oral contraceptive; (5) DMPA (medroxyprogesterone acetate, Depoprovera); and (6) Mirena IUD (levonogerestrol Intrauterine Device) usage. These data were also stratified according to microflora status using the Nugent score [
First a vaginal swab was taken for a Gram Stain evaluation for microflora status using the Nugent method of scoring [
The CVL samples (100 μg protein based on Lowry) were labeled with NHS-Cy5 dye (6 μg GE Healthcare Life Sciences, Piscataway, NJ) in 100 mM NaHCO3 buffer, pH 9.3. A pooled sample of the CVL was labeled with NHS-Cy3 dye (60 μg per mg of protein) for use as the reference. Samples were incubated for 45 min at room temperature with gentle shaking to allow labeling and the free dye was then quenched by addition of 2 M Tris buffer, pH 6.8 (final concentration 250 mM). Samples were then dialyzed overnight against PBS at 4°C using a Microdialyzer (3,000 MW membrane, Spectrum Laboratories). All the steps were performed in the dark. Protein concentrations of samples were obtained after dialysis using the microBCA assay (Pierce).
Lectin microarrays were printed as previously described [
Printed slides were blocked with 25 mM ethanolamine in 100 mM sodium borate (pH 8.5) for 1 hr at RT in a coplin jar with gentle shaking. Slides were washed 3 × with PBST (PBS with 0.005% Tween 20) and 1 × with PBS for 5 min each and then dried using a slide spinner (Labnet International). An ArrayIt multi-well hybridization cassette (Arrayit Corporation, Sunnyvale, CA) was used to create 24 distinct arrays for hybridization. 2.5 μg of Cy5-labeled CVL sample was incubated with equal amounts of Cy3-labeled reference in 100 μl total volume PBST. Samples were added to each array and incubated for 2 h at RT with gentle agitation. Sample solutions were then aspirated and slide chambers were washed 5 × with PBST (100 μL, 3 min, RT), followed by removal of the hybridization chamber and a final wash with PBS (5 min in a coplin jar). The slide was spun-dry and scanned in the Cy3 (ex/em 532/550–600 nm) and Cy5 (635/655–695 nm) channels with a Genepix 4300A slide scanner (Molecular Devices, Sunnyvale, CA). Data were extracted using Genepix 7 (Molecular Devices) and processed with Microsoft Excel 2011.
The background-subtracted median fluorescence of the three replicate spots per protein was tested for outliers using the Grubbs outlier test with α = 0.05. The average value was determined for the three replicates after excluding the outliers if there were any. The log2 values of the average signals were median-centered over the array in each channel to account for differences in labeling efficiency [
Cervicovaginal lavage (CVL) samples, which represent a comprehensive collection of fluid and mucus from the lower reproductive tract [
In lectin microarray analysis, we label the glycoproteins in our samples with Cy5-dye through lysine coupling to the corresponding NHS-ester. The labeled glycoproteins are then mixed with a Cy3-labeled biological reference and incubated with our microarrays to obtain a semiquantitative dual-color analysis that is accurate for relative glycomic composition (
(A) Experimental scheme. Equal amounts of Cy5-CVL sample and Cy3-pooled CVL reference were analyzed on the lectin microarray. (B) Heat map of hierarchical clustering. Median-normalized log2 ratios (Sample (S) /Reference (R)) of 165 CVL samples were hierarchically clustered using Pearson correlation coefficient as the distance metric and average linkage analysis (yellow, log2(S) > log2(R); blue, log2(R) > log2(S)). Samples are color coded by hormonal and microbial status across the top: Postmenopausal (orange), Days 1–14 (black), Days 15–28 (light teal), Oral contraceptive (purple), Depo-Provera (gray), IUD (pink), BV (green), Intermediate (dark teal), Normal (blue). (C) Select sample and lectin clusters showing BV vs. Normal patterns. Pearson correlation coefficient scale is shown.
Initial examination of the heatmap suggested that both microbial status and hormonal levels impact the glycome. Samples most clearly segregated by microbial status (
To more closely examine the effects of bacterial vaginosis (BV) on CVL glycosylation in this cohort, we performed ANOVA analysis on a combined cohort of BV (n = 23), intermediate (n = 23) and normal (n = 90) samples, disregarding hormonal status, for each lectin on our microarray. In line with our earlier findings, we observed statistically significant decreases in lectins corresponding to α2,6-sialic acid, and high mannose epitopes and an increase in β-Gal, β-GalNAc binding (Figs
(A) Representative
(A) High mannose glycan structure. Lectin binding determinants are shaded in grey. (B-C) Notched boxplot representation of binding levels of (B) GRFT and (C) SVN. Significance: *, 0.01<
Several of these changes are consistent with the known biological effects of BV on the glycome of the vagina. In bacterial vaginosis higher levels of sialidase, an enzyme that cleaves sialic acid molecules from underlying β-Gal and β-GalNAc structures, are observed [
Our data also shows a loss of high mannose residues on glycoproteins of the CVL from women with BV (
Previous studies have shown that both HIV infection risk and urinary tract infection risk increase with bacterial vaginosis [
We reanalyzed our lectin microarray data for hormonal effects, examining the data for the 90 women with normal microflora. We excluded postmenopausal women, as we could not evaluate the status of their microbiome. We examined the effects of normal hormonal status on the CVL glycome by comparing women in days 1–14 of their menstrual cycle to women in days 15–28. We observed several changes associated with the menstrual cycle including a loss of GRFT binding in the later stages (p = 0.036,
(A) Bi-plot of lectin microarray data for CVL from women in days 1–14 (x-axis) versus days 15–28 (y-axis). Graph shows average data for each lectin. Lectins and antibodies showing significant differences (p < 0.05) between the two groups are labeled with diamonds. Lectins with similar binding glycans are labeled in the same color (yellow: Gal/GalNAc; green: high mannose; purple: sialyl Lewis A). Lectins above the red dashed line showed increased expression levels during days 15–28 of the menstrual cycle compared to days 1–14. (B) Visual representation of glycans showing significant differences between the two groups.
(A) Representative
The glycome of fluids from the cervico-vaginal tract plays an important role in the innate immune system, preventing pathogenic interactions and modulating immune activation. The study described herein and the accompanying work by Moncla et al (PONE-D-15-01714), which use cervical vaginal lavage samples (CVL) to represent the fluid and mucus from the lower reproductive tract, are the first to examine the role of hormonal contraceptives and microflora on glycosylation in this critical immunological fluid. Our data demonstrates that pathogenic microflora, such as that observed in bacterial vaginosis, have a profound effect on the CVL glycome overriding hormonal effects. These alterations may lower the immunological function of the cervico-vaginal fluids, enhancing the ability of secondary pathogens, such as HIV-1, to infect the host. Hormonal contraceptives also alter this glycome, with currently unknown implications. Both the route of contraceptive administration and the type of synthetic hormone used may play a role in these changes. These differences point out the need for further paired-patient studies examining the direct effects of different hormonal contraception agents on the glycome and on the innate immune protection provided by cervico-vaginal fluids.
Raw data is available at
(XLSX)
Lectin microarray analysis of matched dialyzed and undialyzed Cy3-labeled samples. The fluorescence was inhibited by free sugar in the CVL. Sample images are shown.
(TIFF)
(A-H) Notched boxplot representation of binding levels of (A) SNA, (B) TJA-I, (C) ECA, (D) RCA, (E) AIA, (F) MNA-G, (G) GRFT and (H) SVN for women on no hormonal contraceptives (days 1–14 and 15–28) and on hormonal contraceptives (oral contraceptives (Oral), Depo-Provera (Depo) or IUD). Significance: *, 0.01<
(TIF)
(A-H) Notched boxplot representation of binding levels of (A) SNA, (B) TJA-I, (C) ECA, (D) RCA, (E) AIA, (F) MNA-G, (G) GRFT and (H) SVN for women at days 1–14 of the menstrual cycle with different flora states. The same effects caused by microflora were observed in these women, matching the trends seen in combined cohorts. Significance: *, 0.01<
(TIF)
(A-H) Notched boxplot representation of binding levels of (A) SNA, (B) TJA-I, (C) ECA, (D) RCA, (E) AIA, (F) MNA-G, (G) GRFT and (H) SVN for women on Depo with different flora states. The same effects caused by microflora were observed in these women, matching the trends seen in combined cohorts. Significance: *, 0.01<
(TIF)
Notched boxplot representation of binding levels of MAL-I to normal, intermediate and BV samples is shown. Outliers are marked red. The observed difference is not statistically significant (
(TIFF)
Notched boxplot representation of binding levels for HPA, a lectin that binds α-GalNAc, to normal, intermediate and BV samples is shown. None of the differences were statistically significant. Outliers are marked red.
(TIF)
Notched boxplot representation of binding levels for lectins that bind Man5-Man6 to normal, intermediate and BV samples are shown. (A) AMA, (B) ASA, (C) GNA, (D) NPA and (E) ConA. None of the differences were statistically significant. Outliers are marked red.
(TIF)
(DOCX)
We would like to acknowledge Boval Biosolutions for lyophilized protease- and IgG-free bovine serum albumi (no. LY-0081) and Dr. B. O’Keefe (NCI, Frederick) for recombinant GRFT and SVN.