Metaproteomics analysis of cervicovaginal swab samples
Analysis of the vaginal microbiome and secretome was performed on cervicovaginal fluid samples collected from young women in the Women’s Initiative in Sexual Health (WISH) study conducted in South Africa. A full description of the study design and cohort have been detailed previously [47] but included sexually active, HIV-negative, non-pregnant women aged 16-22 years. Lateral vaginal wall swabs were collected by the study nurse by rotating Dacron swabs 360º on the lateral vaginal wall before being placed in 1 ml phosphate buffered saline (PBS), transported to the laboratory at 4ºC and stored at -80°C. Samples were computational randomised and the eluted material subjected to metaproteomic analysis using liquid chromatography-tandem mass spectrometry (LC-MS/MS) on a Q-Exactive quadrupole-Orbitrap MS (Thermo Fisher Scientific, MA, USA) coupled with a Dionex UltiMate 3000 nano-UPLC system (120min per sample) [45]. Proteomic data were used to infer vaginal microbiome composition and host secretome as described previously [45]. Briefly, proteins were identified using two different databases: (i) UniProt database restricted to human and microbial entries (73,910,451, release August-2017) and filtered using the MetaNovo pipeline [48]; (ii) human proteins combined with a vaginal metagenome-based database obtained from the study of Afiuni-Zadeh et al. [49]. Taxonomy was assigned using UniProt and relative abundance of each taxon was determined by aggregating the intensity-based absolute quantification (iBAQ) values of all proteins identified for each taxon. The taxonomic analysis of the proteins identified using both of the databases was similar; however, the first database resulted in the identification of a greater number of proteins compared to the vaginal metagenome database. The first database also showed a high degree of similarity compared to 16S rRNA gene sequence data from a subset of the same women and was thus used for downstream analysis [45]. As the majority of taxa identified had <2 proteins detected, a more stringent cut-off was applied to include only taxa with >3 detected proteins, or 2 proteins detected in multiple samples. For LDH analysis, log2-transformed iBAQ values for bacterial L-LDH and D-LDH were aggregated and the participants categorized into low (<median) or high (>median) groups. The presence of BV was assessed by Nugent score (Nugent-BV) [5], with a score ≥ 7 considered as BV, 4-6 as an intermediate and ≤ 3 as non-BV[50]. This study evaluated samples from 113 women recruited from the Cape Town study site who had sufficient sample and metadata available for analysis.
The limma R package [51] was then used to identify differentially abundant host proteins between these categories. Logistic regression was used to adjust for potential confounders including Lactobacillus spp. relative abundance, STIs, prostate-specific antigen and contraceptives.
Culture and LA treatment of cervicovaginal epithelial cells
The human ectocervical (Ect1/E6E7) and vaginal (VK2/E6E7) cell lines (purchased from the ATCC) were cultured and seeded for treatment on transwell supports as previously described [41]. Average LA concentration in cervicovaginal fluid is 1% (+/- 0.2%) [36]; however the multilayered structure of the epithelium and the presence of mucus means viable epithelial cells in basal layers are likely to be exposed to concentrations <1%. Indeed, previous optimisations indicated 0.3% LA is the highest concentration tolerated by monolayers of cervicovaginal epithelial cells without eliciting cytotoxicity [41] and was thus utilised for these analyses. Cells were treated for 1 h with media containing 0.3% L-LA or D-LA (Sigma-Aldrich) at either pH 3.9 or 7, or low pH media control (pH 3.9, media acidified by hydrochloric acid, HCl) added to the apical compartment. HCl-containing media was replenished during the 1 h treatment to maintain a pH<4.5 [41]. Following treatment, media was removed and cells washed with calcium and magnesium-free PBS (PBS-, Life Technologies, Carlsbad, CA) before returning to culture for either a further 4 h (for RNA analysis) or 24 h (for all other experiments). Epithelial barrier integrity was assessed by transepithelial electrical resistance (TEER) using a Millicell-ERS Voltohmmeter and probe (Millipore, MA). The unit area resistance (Ω*cm2) was calculated by multiplying the sample resistance by the membrane area.
Impact of LA treatment on epithelial barrier gene expression in FRT epithelial cells
RNA was extracted using the RNeasy mini kit (Qiagen, Hilden, Germany) as previously described [41]. To determine gene expression in LA treated cervicovaginal epithelial cells, next generation RNA-Seq analysis was performed at the Australian Genome Research Facility (AGRF, Melbourne, Australia) using the HiSeq 2500 NGS platform (Ilumina, San Diego, CA) and an Illumina bcl2fastq 2.20.0.422 pipeline with an output of 100 bp paired-end reads. Sequence read datasets were checked for quality using FastQC [52]. Reads were trimmed based on quality scores using the Trim Reads tool with default parameters within CLC Genomics Workbench 11.0 (CLC; Qiagen). Overlapping paired reads were merged, and then all reads were mapped against the Homo sapiens reference genome (build version hg38) using the RNA-Seq tool in CLC. Differentially expressed gene (DEGs) analysis was performed in Degust (http://degust.erc.monash.edu/), using the Voom/Limma method [53, 54]. A false discovery rate (FDR) cut-off of < 0.05 and a Log2 fold change (FC) ≥ 0.5 was used to identify genes with significantly altered expression. Targeted interrogation of the expression of genes related to the tight junction barrier was performed using a gene list derived from the RT2 ProfilerTM Human Tight Junctions PCR Array (Qiagen). To identify and visualise significantly overrepresented gene ontologies among the DEGs, we utilised the BiNGO application [55] within Cytoscape v.3.8.2 [56] to perform a hypergeometric test using the Benjamini and Hochberg FDR correction. Unranked lists of DEGs were tested against the 2021-07-02 Gene Ontology release (http://geneontology.org)[57].
Altered expression of junctional molecules was confirmed by qRT-PCR using RT2 SYBR Green Mastermix (Qiagen) and primers specific for the following tight junction factors and housekeeping genes; claudin-1 (CLDN1, Cat# PPH02779A), claudin-4 (CLDN4, Cat# PPH07330D), occludin (OCLN, Cat# PPH02571B), junctional adhesion molecule A (F11R, Cat# PPH02605A), zona occludens 2 (TJP2, Cat# PPH09978B) and the housekeeping genes glyceraldehyde 3-phosphate dehydrogenase (GAPDH, Cat# PPH00150F), hypoxanthine phosphoribosyltransferase 1 (HPRT1, Cat# PPH01018C) and ribosomal protein lateral stalk subunit P0 (RPLP0, Cat# PPH21138F; all from Qiagen). Amplification was performed with a thermocycle of 95°C for 10 min, 40 cycles of 95°C for 15 sec and 40°C for 60 sec, followed by melt curve analysis. Primer specificity was verified by melt curve analysis, which indicated a single amplification product with a unique melting temperature for each gene target. Ct values were analysed to determine relative gene expression, which was standardised to the average expression of the housekeeping genes and calculated using the 2−ΔΔCT method [58].
Analysis of tight junction protein expression and localisation
Levels of tight junction proteins in ectocervical epithelial cells were determined by Western blot using primary antibodies against claudin-1, claudin-4, tight junction protein (TJP) 1 and TJP2 (Cat# 51-9000, 36-4800, 33-9100 and 71-1400, respectively, all from Thermo Fisher Scientific), β-actin (Cat# ab8224, Abcam) and secondary antibodies Alexa Fluor 680-goat anti-mouse IgG (Thermo Fisher Scientific) and Alexa Fluor 800- donkey anti-rabbit IgG (Li-COR, Lincoln, NE). Proteins were visualised with an Odyssey Infrared Imaging system and analysed with ImageStudio Lite software (both from Li-COR). Proteins levels were normalised to β-actin protein intensity.
Immunofluorescence and confocal microscopy
Claudin-4 protein location in Ect cells was visualised by immunofluorescence and confocal microscopy. Epithelial cells were fixed with ice-cold methanol, rehydrated with 1% foetal bovine serum (FBS) in PBS- and permeabilised with 0.2% Triton X-100 in PBS- for 10 min. After blocking with 3% BSA in PBS-/0.2% Triton X-100, cells were incubated with rabbit polyclonal anti-claudin-4 antibody (Invitrogen, Cat #36-4800, 1:25 dilution) then Alexa Fluor 488-goat anti-rabbit IgG secondary antibody (Invitrogen, Cat #A-11008, 1:200). Nuclei were visualised using Hoechst 33342 stain (Thermo Scientific). Transwell membranes and attached cells were excised and mounted on slides with ProLong Gold mounting medium (Invitrogen). Confocal Z-stacks were captured at 0.2µm per section, 25 sections in total using a Nikon A1r confocal microscope (Nikon Corporation, Japan) at 60x magnification with an oil-immersion lens. Images were selected at random (at least 3 stacks per slide) and captured for quantitative analysis with Image J software (National Institute of Mental Health). Images were thresholded to eliminate background fluorescence. The sum of fluorescence intensity was calculated for the stack and mean fluorescence intensity (MFI) was determined.
Transcriptomic analysis of vaginal epithelial cells exposed to bacterial culture supernatants
Bacterial culture supernatants were generated by inoculating 10 ml of culture media with 1 ml of 1x107 colony forming units (CFU) of L. crispatus (ATCC 33197), L. jensenii (ATTC 25258), L. iners (ATTC 55195) or G. vaginalis (ATCC 14018). NYC-III media was used for propagation of L. crispatus, L. jensenii and L. iners while Tryptic soy broth [TSB] was used for G. vaginalis. CFU to OD600 measures were pre-calculated for each strain. Cultures were grown anaerobically for 48 h to late stationary phase. L. crispatus, L. jensenii, L. iners and G. vaginalis reached 5.2 x108 CFU/ml, 5.8 x 108 CFU/ml, 4.1 x 108 CFU/ml and 1.2 x 109 CFU/ml, respectively. The 10 ml of culture supernatant was clarified by centrifugation at 3,000 x g for 10 min, sterile filtered (0.2 μm filter) and stored at -20°C until use. Bacterial supernatants were diluted to 20% (v/v) in complete VK2 cell culture medium and added to VK2 cells for 13 h. VK2 cells were serum starved for 24 h in keratinocyte serum-free base media only, prior to addition of 20% bacterial culture supernatants. After the exposure, media was removed, cells were washed once with 1X PBS and 300 μl RNAlater (QIAGEN) was added to wells. Cells were mechanically detached and stored at -80°C before total RNA extraction. Total RNA was extracted using the MasterPureTM Complete DNA and RNA purification kit (Lucigen, WI, USA) according to manufacturer’s instructions. Methodology for transcriptomic analyses by RNASeq including library preparation, RNA sequencing and read mapping has been detailed previously [29]. LA concentration within bacterial supernatant-containing media was quantified by using the D/L lactic acid assay kit (R-Biopharm AG, Darmstadt, Germany, Cat # 11112821035). Concentration of the protonated from of LA within treatment media was calculated from the sample pH and LA concentration using the Henderson-Hasselbalch equation as previously described [36]. Total lactate and protonated LA were normalised to the final CFU/ml for L. crispatus, L. jensenii, L. iners and G. vaginalis achieved at the stationary phase.