J Antimicrob Chemother 2013; 68: 2648 – 2659
doi:10.1093/jac/dkt244 Advance Access publication 24 June 2013
Proteome profiles of vaginal fluids from women affected by bacterial
vaginosis and healthy controls: outcomes of rifaximin treatment
Federica Cruciani1, Valerie Wasinger2, Silvia Turroni1, Fiorella Calanni3, Gilbert Donders4,
Patrizia Brigidi1 and Beatrice Vitali1*
1
Department of Pharmacy and Biotechnology, University of Bologna, Bologna, Italy; 2Bioanalytical Mass Spectrometry Facility, Mark
Wainwright Analytical Centre, The University of New South Wales, Sydney, Australia; 3Alfa Wassermann SpA, Bologna, Italy; 4Departments
of Obstetrics and Gynaecology, General Hospital Heilig Hart, Tienen, and University of Antwerp, Antwerp, Belgium
*Corresponding author. Tel: +39-051-2088750; Fax: +39-051-2099734; E-mail: [email protected]
Received 20 February 2013; returned 6 April 2013; revised 17 May 2013; accepted 18 May 2013
Objectives: This study was designed to characterize the proteome of vaginal fluid (VF) from women with bacterial
vaginosis (BV) in comparison with that from healthy women, and to evaluate the effect exerted by rifaximin vaginal
tablets.
Methods: Women with BV (n¼ 39) and matched healthy controls (n¼ 41) were included in the study. BV patients
were distributed among four groups receiving different doses of rifaximin. Vaginal rinsings were collected at the
screening visit from all the participants and at a follow-up visit from BV-affected women. The VF proteome was
analysed by tandem mass spectrometry using an Orbitrap mass analyser.
Results: A large number of human proteins were differentially expressed in women with BV in comparison with
healthy women (n ¼118) and in BV-affected women treated with rifaximin (n¼284). In both comparisons, a
high proportion of the dysregulated proteins (20%) were involved in the innate immune response. Twenty-one
of 24 proteins increased in abundance in women with BV versus healthy women and 31/59 proteins decreased
after rifaximin treatment, suggesting a general reduction of the immune response resulting from the therapy.
Major changes in protein abundance were found following treatment with 25 mg of rifaximin once daily for 5 days.
Conclusions: BV is associated with a massive change in the VF proteome, mainly regarding the abundance of proteins involved in the innate immune response. Rifaximin at a dosage of 25 mg for 5 days modulated the vaginal
proteome, counteracting the alterations associated with the BV condition.
Keywords: antibiotics, vaginal fluid proteome, MS/MS analysis, innate immune response
Introduction
Vaginal fluid (VF) is a complex biological fluid consisting of water,
electrolytes, low molecular weight organic compounds (glucose,
amino acids and lipids), a vast array of proteins and proteolytic
enzymes arising from plasma transudate, vaginal epithelial cells
and vaginal microbiota.1 – 3 VF forms the first line of defence
against external pathogens, signals fertility and aids insemination,
pregnancy and labour.4 Collection of VF according to standardized
procedures5 is minimally invasive and safe, and therefore it is especially interesting and potentially useful as a source of biomarkers
for the diagnosis of pathological conditions as well as for the development of prevention strategies.1 – 3
Bacterial vaginosis (BV) is a common vaginal disorder among
women of childbearing age. BV is an imbalance in the ecology of
the normal vaginal microbiota that is characterized by the depletion of lactobacilli and the proliferation of anaerobic bacteria
such as Gardnerella vaginalis, Mobiluncus species, Prevotella
species, Mycoplasma hominis and Atopobium vaginae.6,7 BV
affects 10% –15% of women of reproductive age and is associated
with genital tract infections8 and pregnancy complications such as
preterm birth.9
Conventional treatments for BV are represented by metronidazole and clindamycin, but their effectiveness seems to be limited
due to BV recurrence10 and adverse effects associated with their
systemic absorption.11 The limits of conventional antibiotics raise
the question for alternative therapeutics. Rifaximin is a semisynthetic rifamycin derivative with a broad antimicrobial spectrum and a good safety profile because of its negligible systemic
absorption.12 On the basis of these pharmacological features,
together with the emerging evidence that rifaximin does not
dramatically affect the gut microbiota,13,14 it has recently been
proposed as a promising candidate for BV cure and remission
maintenance.11,15
# The Author 2013. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved.
For Permissions, please e-mail: [email protected]
2648
JAC
Impact of rifaximin on vaginal proteome
Proteomics is a powerful discipline that has become increasingly important in the study of biological processes by defining cellular
functions and networks at the protein effector level. In particular,
advances in mass spectrometric technologies have made it possible to comprehensively analyse proteins in very complex mixtures, making feasible the resolution of human biological fluid
proteomes.16 Furthermore, advancements in multidimensional
protein separation techniques have allowed the identification of
proteins present in trace amounts, thereby increasing the
dynamic range of detection in complex biological samples.17 To
date, studies on the identification of potential biomarkers in the
VF of women affected by BV have been restricted to the use of
antibody-dependent techniques.18 – 20 Only a limited number of
proteomic studies using high-throughput mass spectrometry
techniques have been performed on VF, and the majority of them
have been focused on the search for potential markers for
pregnancy-associated conditions such as preterm labour and
intra-amniotic infections.1,16,21
The present study was designed to characterize the proteome
profile of VF from BV-affected women in comparison with
healthy women, as well as to evaluate the effect exerted by rifaximin administered through vaginal tablets. The VF proteome was
analysed by tandem mass spectrometry (MS/MS) using an Orbitrap
mass analyser, which is an instrument of choice for many proteomics applications owing to its high mass accuracy, high resolving
power and high dynamic range.22 In addition, the ProteomeSep
(MF10) fractionation system was used in order to increase the
proteome coverage, particularly by enriching low-mass and lowabundance proteins and peptides.23,24
Methods
Subjects and sample collection
A total of 80 Belgian pre-menopausal, non-pregnant women, aged
between 18 and 50 years were included in the study (Figure 1). At a
screening visit (V1), 39 patients were diagnosed for BV as they presented
with a Nugent score .3 and were positive for at least three of four
Amsel’s criteria [BV (n ¼39)]. These BV-affected women were included in
a multicentre, double-blind, randomized, placebo-controlled study
(EudraCT: 2009-011826-32) that was performed to compare the efficacy
of different doses of rifaximin vaginal tablets versus placebo for the treatment of BV.11 The patients underwent a randomization visit and were distributed among four treatment groups: group A received a 100 mg
rifaximin vaginal tablet once daily for 5 days (n¼10), group B received a
25 mg rifaximin vaginal tablet once daily for 5 days (n¼10), group C
received a 100 mg rifaximin vaginal tablet once daily for the first 2 days
and a placebo vaginal tablet for the remaining 3 days (n¼9), group D
received a placebo vaginal tablet once daily for 5 days (n¼10). Study medication was administered intravaginally at bedtime. A follow-up visit (V3)
was performed after 7– 10 days from the end of therapy. Remission was
evaluated at V3 according to Amsel’s criteria (,3) and Gram’s stain
Nugent score (≤3). A control group consisted of 41 age-matched healthy
subjects who had no signs of vaginal tract infection at the screening visit
V1 [H (n¼41)]. These patients were selected from the same patient population, presenting at the same clinical outpatient service, but who had never
had BV and were negative on Amsel and Nugent testing. All the enrolled
women were subjected to tests to exclude sexually transmitted infections,
Candida infection and abnormal findings on cervical Pap smears. All
patients signed informed consent in accordance with the approval from
the Ethics committee of the Heilig Hart Hospital of Tienen, Belgium.
Standardized vaginal rinsings with 2 mL of saline were collected for analysis at V1 and V3 by flushing and reaspirating the fluid through a 22 gauge
needle in the left, central and right upper vaginal vaults as described elsewhere.25 The vaginal rinsings were subsequently stored at 2808C until use.
Protein extraction
From each vaginal rinsing, 1 mL was centrifuged at 9500 g for 15 min to
separate the pellet from the supernatant, which was used for protein
extraction.
Nine volumes of acetone:HCl (10 :1) were added to the supernatant of
the vaginal rinsing and proteins were precipitated by centrifuging at
12000 g for 10 min. The protein pellet was dissolved in 1 mL of 70%
ethanol and the sample was spun at 12 000 g for 10 min. Acetone (1 mL)
was added and the proteins were further precipitated by centrifugation at
12000 g for 5 min. After removing the supernatant, the pellet was air
dried and stored at 2208C. Each protein extract was resuspended in
50 mM ammonium bicarbonate, 2 M urea and 10 mM dithiothreitol pH 8
and quantified using the 2-D Quant Kit (GE Healthcare, Uppsala, Sweden)
according to the manufacturer’s instructions.
Proteomic study design
Proteins extracted from VF of the enrolled women were collected into nine
pools (Table S1, available as Supplementary data at JAC Online): H pool, containing proteins from healthy women; BV pool, containing proteins from
BV-affected patients collected at V1; A-R pool, containing proteins from
BV-affected women belonging to treatment group A who were in remission
at V3; A-N pool, containing proteins from BV-affected women belonging to
treatment group A who were not in remission at V3; B-R pool, containing
proteins from BV-affected women belonging to treatment group B who
were in remission at V3; B-N pool, containing proteins from BV-affected
women belonging to treatment group B who were not in remission at V3;
C-R pool, containing proteins from BV-affected women belonging to treatment group C who were in remission at V3; C-N pool, containing proteins
from BV-affected women belonging to treatment group C who were not
in remission at V3; D-N pool, containing proteins from BV-affected
women belonging to treatment group D who were not in remission at V3.
In order to identify low-abundance proteins and peptides, the Microflow
MF10 system was used to fractionate protein pools H and BV (Table S1, available as Supplementary data at JAC Online). The fractionation was not performed on A-R, A-N, B-R, B-N, C-R, C-N and D-N pools due to the low protein
content of samples. Two comparisons were carried out: (i) fractionated pool
of proteins from VF of healthy women (HF) versus fractionated pool of proteins from women affected by BV (BVF); and (ii) whole protein pools from
patients affected by BV treated with different doses of rifaximin or
placebo before (BV) and after (A-R, A-N, B-R, B-N, C-R, C-N, D-N) treatment
(Table S1, available as Supplementary data at JAC Online).
MF10 fractionation of proteins
Prior to fractionation, pools H and BV, containing 1 mg of protein each, were
prepared. To constitute these pools, equal quantities of protein from each
vaginal sample were mixed, dried down and resuspended in 280 mL of
90 mM Tris/10 mM epsilon aminocaproic acid (EACA) and 1 M urea buffer
pH 10.2. MF10 fractionation of proteins was performed using a 5-cartridge
assembly. The cathode-end cartridge was fitted with a 5 kDa restriction
membrane followed by 125 kDa, 50 kDa, 25 kDa and 5 kDa separation
membranes. The anode-end cartridge was fitted with a 1 kDa membrane
(regenerated cellulose, Millipore) and a 5 kDa membrane facing the anodecirculating buffer. The resulting assembly generated five chambers. Cartridge assemblies had two lanes of chambers that allowed fractionation
of two samples in one run. One hundred mL of 90 mM Tris/10 mM EACA
and 1 M urea buffer pH 10.2 was added to the buffer reservoir and circulated
around the electrodes. Protein pools (140 mL) were added to the chamber
closest to the cathode for separate runs. Fractionations were performed
2649
Cruciani et al.
Enrolled women (n = 80)
Screening visit V1
Healthy women
H (n = 41)
BV-affected women
BV (n = 39)
Randomization visit
Distribution of BV-affected women in four treatment groups and administration of
rifaximin/placebo vaginal tablets once daily at bedtime
A (n = 10)
B (n = 10)
C (n = 9)
D (n = 10)
100 mg of rifaximin
for 5 days
25 mg of rifaximin
for 5 days
100 mg of rifaximin
for 2 days
placebo
for 5 days
Follow-up visit V3
(after 7–10 days from the end of the therapy)
Women in remission at V3
R (n = 11)
A-R (n = 2)
B-R (n = 5)
C-R (n = 4)
Women not in remission at V3
N (n = 28)
A-N (n = 8)
B-N (n = 5)
C-N (n = 5)
D-N (n = 10)
Figure 1. Study design and participants throughout the study.
at 250 V for 30 min. Following fraction collection, the lower fractions (1–
5 kDa and 5 – 25 kDa) were further concentrated and desalted using
Stage tipsw C18, 200 mL (Proxeon Biosystems, Odense, Denmark), according to the manufacturer’s instructions (pools HF and BVF).
MS/MS analysis
Liquid chromatography (LC) – MS/MS analysis was carried out for the HF and
BVF pools and for the unfractionated BV, A-R, A-N, B-R, B-N, C-R, C-N and D-N
pools containing 50 mg of protein each (Table S1, available as Supplementary data at JAC Online). Each fraction or pool was resuspended in 50 mL of
50 mM ammonium bicarbonate, 2 M urea and 10 mM dithiothreitol pH 8.
Trypsin (1 mg) was added and the reaction was incubated at 378C overnight.
The digestion was halted by addition of 5 mL of formic acid and the samples
dried. Digested samples were resuspended in 10 mL of buffer A (0.1% formic
acid), and 0.2 mL of each sample in triplicate was analysed with blanks in
between (buffer A).
Digested peptides were separated by nano-LC using an Ultimate 3000
HPLC and autosampler system (Dionex, Amsterdam, the Netherlands).
Samples (0.2 mL) were concentrated and desalted onto a micro C18
2650
precolumn (500 mm×2 mm, Michrom Bioresources, Auburn, CA, USA)
with H2O:CH3CN (98 :2, 0.05% trifluoroacetic acid, v/v) at 10 mL/min. After
a 4 min wash, the precolumn was switched (Valco 10 port valve, Dionex)
into line with a fritless C18 nano column (75 mm i.d.×10 cm containing
5 mm, 200 Å media, Michrom Bioresources) manufactured according to
Gatlin et al.26 Peptides were eluted using a linear gradient of H2O:CH3CN
(98 :2, 0.1% formic acid, v/v) to H2O:CH3CN (64 : 36, 0.1% formic acid, v/v)
at 250 nL/min over 30 min. High voltage (2000 V) was applied to low
volume tee (Upchurch Scientific, Oak Harbor, WA, USA) and the column
tip positioned 0.5 cm from the heated capillary (T ¼2808C) of an
LTQ-Orbitrap Velos (Thermo Electron, Bremen, Germany) mass spectrometer. Positive ions were generated by electrospray and the Orbitrap operated
in data-dependent acquisition mode. A survey scan of 350–1750 m/z was
acquired (resolution¼30000 at 400 m/z). Up to the 10 most abundant ions
(.5000 counts) with charge states ≥2 were sequentially isolated and
fragmented within the linear ion trap using collisionally induced dissociation. Mass-to-charge ratios selected for MS/MS were dynamically excluded
for 45 s.
MS peak intensities were analysed using Progenesis LC –MS data analysis software v4 (Nonlinear Dynamics, Newcastle upon Tyne, UK). Ion intensity maps from each run were aligned to a reference sample and ion feature
Impact of rifaximin on vaginal proteome
matching was achieved by aligning consistent ion m/z and retention times.
The peptide intensities were normalized against total intensity (samplespecific log-scale abundance ratio scaling factor) and compared between
groups by one-way analysis of variance (P≤ 0.05 for statistical significance)
and post hoc multiple comparison procedures. Type I errors were controlled
for by false discovery rate with q value significance set at 0.01.27,28 Results
are reported as mean+SD (normalized ion intensity score). Peptides exhibiting a statistically significant 1.5-fold or greater difference in abundance
between groups were identified using the database search program
Mascot (Matrix Science, London, UK, www.matrixscience.com). MS/MS
spectra of differentiating peptides were searched against the Swiss-Prot
database (version 15) using Mascot. Parent and fragment ions were
searched with tolerances of +4 ppm and +0.5 Da, respectively. Peptide
charge states were set at +2 and +3. ‘No enzyme’ was specified. Proteins
and peptides were considered confidently identified when matches had a
high ion score .20 and were statistically significant and at least semitryptic. Following identification, a filter was applied to select proteins of
human origin and those produced by microorganisms associated with
the vaginal environment.
Gene ontology (GO) and network analysis
Identified proteins were submitted for GO analysis (AmiGO version 1.8,
database release 3 November 2012; http://amigo.geneontology.org) to
define biological processes, molecular functions and subcellular localizations. Protein accession numbers and their corresponding fold changes
were imported into MetaCoreTM, a web-based computational platform
(v6.8 build 30387; Thomson Reuters, St Joseph, MI, USA) for pathway enrichment and network analysis. The network building among dysregulated
proteins and the MetaCore database proteins was performed using the
shortest path algorithm and its variants. Networks were ranked according
to their statistical significance (P, 0.001) and interpreted in terms of GO.
Major hubs were identified based on the connections and edges within
the networks.
Results
Clinical outcome
The clinical outcome of rifaximin treatment is shown in Figure 1.
The highest percentage of therapeutic remission was found for
treatment group B (5/10, 50%), while lower percentages were
found among the other treatment groups (A: 2/10, 20%; C: 4/9,
44%). No women treated with placebo were found in remission
at V3. These data are in agreement with the clinical outcomes recently obtained in a larger group of women.11
Multivariate analysis of MS data
A multivariate analysis (principal component analysis) was performed on the extracted ions of differentially expressed peptides
(P,0.05) obtained from MS analysis (Figure 2). According to the
x-axis, which explained 93.04% of the overall variance in the
dataset, the proteomic profiles from healthy women and
BV-affected women were dramatically different from each other
(Figure 2a). A less clear separation was shown by the peptides of
BV-affected women before and after treatment with rifaximin or
placebo (Figure 2b). Along the x-axis, which accounted for
39.14% of the total variance, the pools of peptides of VF samples
collected after rifaximin treatment were closely grouped but segregated from the pools of peptides of BV-affected women and
women treated with placebo. According to the y-axis, which
JAC
showed a variation of 18.36%, the peptide pools of VF collected
after cure were distinct from each other, on the basis of the dose
of antibiotic and the response to treatment. Interestingly, pools
A-N and B-N were in line with the BV pool and near to the C-R
and C-N pools, suggesting a similarity among the proteomic profiles of BV-affected women, women who were not in remission
after rifaximin treatment and women who received the antibiotic
for only 2 days. Moreover, according to the y-axis, the B-R pool
was the most distant from the BV pool, demonstrating significant
changes in protein pattern. The A-R and D-N pools occupied an
intermediate position between the B-R and BV pools.
Comparison of protein profiles between healthy
and BV-affected women
Following a Mascot database search using acquired MS/MS data, a
total of 131 human and microbial proteins were successfully identified in the HF and BVF pools (Table S2, available as Supplementary
data at JAC Online). The vast majority of the identified human proteins (84/118, 71%) were increased in the BVF pool, with a median
5.5-fold ratio (range 1.5-fold to 521.1-fold). A significant reduction,
ranging from 21.5 to 25645.4-fold (median, 27.0), occurred for
34/118 (29%) human proteins.
Each human protein was assigned to a biological process, a cellular localization and a molecular function based on information
from the GO database (Figure 3). Most of the differentially
expressed proteins (24/118, 20%) were involved in the innate
immune response and complement activation (Figure 3a). Interestingly, this GO category grouped 14 immunoglobulin chain
regions that were almost all over-represented in BV (median
7.1-fold ratio). Epidermis development and keratinization
accounted for 15/118 (13%) identified proteins whereas 14/118
(12%) were classified as involved in small-molecule metabolic
process. Only 6/118 (5%) proteins were involved in the inflammatory response. More than half of the identified proteins were localized in the extracellular space (41/118, 35%) or associated with
the plasma membrane (18/118, 15%) (Figure 3b). Nearly a
quarter of the identified proteins were cytoplasmic (26/118,
22%). According to molecular function (Figure 3c), as many as
48% (57/118) of the differentially expressed proteins were classified as having binding activity. Among these, protein binding (22/
118, 19%) was the most represented GO category, followed by
calcium ion (15/118, 13%) and antigen binding (14/118, 12%).
Nineteen (22/118) and 16% (19/118) of identified proteins were
related to enzymatic and structural molecule activity, respectively.
MetaCore enrichment analysis revealed that the majority of
enriched pathways were related to cytoskeleton remodelling, complement activation (classical, alternative and lectin-induced pathways) and blood coagulation (data not shown). Based on the
functional subnetworks built using the ‘analyse network’
algorithm, the proteins differentially expressed in the HF and BVF
pools were primarily involved in developmental processes
(P ¼ 1.22×10231), immune system processes (P¼ 3.93×10222)
and response to chemical stimulus (P ¼ 1.71×10220) (Table S3,
available as Supplementary data at JAC Online). Figure 4 shows a
high-significance subnetwork linked to immune response and
complement activation. Most of the identified proteins mapped
in this network were immunoglobulins and the majority of them
were significantly overexpressed in BV (range 3.7-fold to
12.6-fold; median, 7.6). Notably, in this network as well as in
2651
Cruciani et al.
Principal component 2 (3.27%)
(a) 0.6
0.4
0.2
HF
BVF
0.0
–0.2
–0.4
–0.6
–0.8
–1.0
–0.4
–0.2
0.0
0.2
Principal component 1 (93.04%)
0.4
0.6
Principal component 2 (18.36%)
(b) 0.5
B-R
0.4
0.3
0.2
A-R
D-N
0.1
0.0
C-R
–0.1
–0.2
–0.3
C-N
A-N
B-N
–0.2
–0.1
BV
0.0
0.1
0.2
0.3
Principal component 1 (39.14%)
0.4
0.5
0.6
Figure 2. Multivariate analysis of MS data. (a) Principal component analysis of the peptides in the fractionated pools of healthy women (HF) and women
affected by BVat V1 (BVF). (b) Principal component analysis of the peptides in the unfractionated pools of women affected by BV before treatment (BV) and
after treatment with 100 mg of rifaximin once daily for 5 days (remission, A-R; no remission, A-N), 25 mg of rifaximin once daily for 5 days (remission, B-R;
no remission, B-N), 100 mg of rifaximin once daily for 2 days (remission, C-R; no remission, C-N) and placebo for 5 days (D-N). Samples were run in triplicate.
other top-scoring networks, a highly linked hub was represented by
the transcription factor SP1. According to the transcriptional regulation networks, SP1 was ranked #1, with 42 targets among the 118
identified human proteins (P¼ 7.10×102115). Among the 13 microbial proteins that were differentially expressed between the HF
and BVF pools, 9 (69%) were derived from Lactobacillus strains
(belonging to Lactobacillus acidophilus, Lactobacillus casei, Lactobacillus gasseri and Lactobacillus helveticus) and were mainly
involved in glucose metabolism and protein synthesis (Table S2,
available as Supplementary data at JAC Online). Three proteins
from Staphylococcus aureus, Staphylococcus epidermidis and
Candida glabrata were significantly increased in the BVF pool,
even though none of these bacteria is known to be associated
with BV.
Impact of rifaximin on the proteome
of BV-affected women
A total of 314 human and microbial proteins were successfully
identified in VF samples of BV-affected women before and after
rifaximin/placebo treatment (Table S4, available as Supplementary
2652
data at JAC Online). In general, most of the proteins differentially
expressed in response to rifaximin/placebo treatment were
reduced (192/314, 61%) rather than increased in abundance
(122/314, 39%). Of the 284 identified human proteins, 48 (17%)
were dysregulated in all pools from rifaximin-treated women compared with the BV pool, regardless of both antibiotic dosage and
clinical outcome. Interestingly, the greatest variation for these dysregulated proteins occurred in the B-R pool, followed by the A-R
pool, while few or no changes were observed after placebo administration. Opposite trends of expression for BV versus H and rifaximin versus BV were observed for 46 of the 89 proteins that
differentially changed their abundance in both datasets. In particular, 26 of these proteins were up-regulated (6) or downregulated (20) in at least four of the six pools from rifaximin-treated
women, contrary to what was found in the BV versus H comparison.
For 17/26 (65%) proteins, the greatest fold changes were associated with the B-R pool. Interestingly, group B showed the
largest total number of differentially expressed human proteins,
with 214 and 155 dysregulated proteins in the B-R and B-N pools,
respectively. Moreover, the fold changes of 83 proteins in the B-R
pool were the highest among all pools, suggesting a major
JAC
Impact of rifaximin on vaginal proteome
(a)
NA; 13%
acute-phase and
inflammatory
response; 5%
blood coagulation;
5%
cell adhesion; 3%
others; 7%
cytoskeleton
organization; 2%
epidermis
development and
keratinization; 13%
transport; 11%
small molecule
metabolic process;
12%
innate immune
response and
complement
activation; 20%
response to
oxidative stress and
external stimulus;
5%
(b)
regulation of
endopeptidase
activity; 4%
NA; 6%
nucleus; 2%
cytoplasm; 22%
keratin filament;
5%
internal organelle;
8%
cytoskeleton
constituent; 3%
integral or
associated with
plasma membrane;
15%
haptoglobinhaemoglobin
complex; 2%
granules and
vesicles; 2%
(c)
extracellular space;
35%
NA; 8%
others; 2%
transporter activity;
5%
structural molecule
activity; 16%
receptor activity;
2%
enzymatic activity;
19%
antigen binding;
12%
calcium ion
binding; 13%
ion binding; 2%
lipid binding; 2%
protein binding;
19%
Figure 3. Pie charts showing the GO categorization of the MS/MS-identified proteins differentially expressed between healthy and BV-affected women.
Classification was performed according to keyword categories [(a) biological process, (b) cellular component, (c) molecular function]. When proteins
were associated with more than one functional category, one GO term was chosen arbitrarily. NA, not available.
2653
Cruciani et al.
TOSO
SCF (Fb×4/aB Crystallin) E3
ligase
IGHM
SKP1
Siah1/SIP/Ebi E3 ligase
IL1R2
IgM
C1
SIAH1
IL1RN
Plastin
NEFL
C1q
IL-1RI
S100P
BMP1
Keratin 5/14
PDK1
Actin cytoplasmic 2
IgG2A
IgG1
IgG3
IgG2B
IgG2
CD14
SP1
LDHA
IGHG2
SP-D
NPY
PDGF-B
CNK1
IGHG1
IgG
Amyloid beta 42
Renin
PPL(periplakin)
Actin cytoskeletal LSP1
IgG4
Fc gamma RII beta
OAS1
F protein (RSV)
HGF
LRP2 (Megalin)
PDGF-AB
GSK3 alpha/beta
2'–5'–oligoadenylate
synthetase
ESR
IgJ
Generic enzyme
Transcription factor
Generic binding protein
Positive effect
Protein kinase
Generic receptor
Protein
Negative effect
Generic protease
Receptor ligand
A complex or a group
Unspecified effect
Metalloprotease
Figure 4. Protein network of differentially expressed proteins in VF of BV-affected women (BVF) in comparison with healthy women (HF). MetaCore pathway
analysis software was used to generate a network of connections between all identified proteins with altered expression levels. Network proteins and
connections are visualized by symbols that specify the functional nature of the protein or interaction. Blue or red circles flank experimentally identified
proteins and indicate, respectively, a significant down- or up-regulation of the protein in the BVF pool compared with the HF pool. The mixed coloured
circle for IgM indicates mixed protein expression between files. This figure appears in colour in the online version of JAC and in black and white in the
print version of JAC.
impact of this treatment regimen on the BV-related proteome.
Conversely, placebo administration was associated with the
lowest number of differentially expressed proteins (207) and the
expression variation was often in the opposite direction with
respect to the trend observed in the other pools.
Each human protein was assigned to a biological process, a cellular localization and a molecular function based on information
from the GO database (Figure 5). Similar to the BV versus H comparison, most proteins were involved in the innate immune
response and complement activation (59/284, 21%) and smallmolecule metabolic process (41/284, 14%), whereas only 9/284
(3%) were involved in the inflammatory response (Figure 5a).
Interestingly, the most represented GO category grouped 17
2654
proteins that were identified as dysregulated also in the BVF
versus the HF pool. Ten of them exhibited a trend toward underexpression, contrary to what was found in the BV versus the H dataset.
As expected, a large number of proteins were localized in the
extracellular space (108/284, 38%) and plasma membrane
(34/284, 12%) (Figure 5b). As many as 19% (55/284) of the differentially expressed proteins were cytoplasmic. The main represented
molecular functions were structural molecule activity (50/284,
18%), antigen binding (39/284, 14%) and protein binding (36/284,
13%) (Figure 5c).
According to the MetaCore enrichment analysis, the most
enriched pathways were related to cytoskeleton remodelling,
blood coagulation and complement activation (lectin-induced
JAC
Impact of rifaximin on vaginal proteome
(a)
NA; 7%
acute-phase and
inflammatory
response; 3%
others; 6%
blood coagulation;
5%
cell adhesion; 5%
transport; 6%
cytoskeleton
organization; 6%
small molecule
metabolic process;
15%
epidermis
development and
keratinization; 7%
signal transduction;
5%
response to
oxidative stress and
external stimulus;
5%
response to drug;
2%
(b)
nucleus; 3%
proteolysis and
regulation of
endopeptidase
activity; 7%
innate immune
response and
complement
activation; 21%
NA; 4%
cornified envelope;
2%
molecular
complexes; 2%
cytoplasm; 19%
keratin filament;
5%
internal organelle;
6%
cytoskeleton
constituent; 7%
integral or
associated with
plasma membrane;
12%
granules and
vesicles; 2%
extracellular space;
38%
(c)
NA; 6%
others; 3%
transporter activity;
4%
antigen binding;
14%
calcium ion
binding; 8%
structural molecule
activity; 18%
ion binding; 1%
protein binding;
13%
other enzymatic
activities; 12%
(inhibition of)
peptidase activity;
13%
small molecules
binding; 8%
Figure 5. Pie charts showing the GO categorization of the MS/MS-identified proteins differentially expressed between BV-affected women before and after
rifaximin/placebo treatment. Classification was performed according to keyword categories [(a) biological process, (b) cellular component, (c) molecular
function]. When proteins were associated with more than one functional category, one GO term was chosen arbitrarily. NA, not available.
2655
Cruciani et al.
C1qRp
CR1
C3aR
alpha-X/beta-2 alpha-M/beta-2
integrin
integrin
C5AR
CD21
1 2 3 4 5 6 7
C1 inhibitor
B
CRP
1 2 3 4 5 6 7
B
B
B
B
B
B
1 2 3 4 5 6 7
C1q
1 2 3 4 5 6 7
C3dg
1 2 3 4 5 6 7
C1r
IgG1
B
C3c
iC3b
C5a
B
C1s
1 2 3 4 5 6 7
B
C3a
CS
CS
CS
1 2 3 4 5 6 7
C
IgM
Factor 1
B
1 2 3 4 5 6 7
C2
C1
C4
C
DAF
1 2 3 4 5 6 7
B
C3b
C4BP
B
C
C8alpha
B
1 2 3 4 5 6 7
C2b
B
CS
C
1 2 3 4 5 6 7
C2a
C8beta
MCP
C4a
C8gamma
C7
CS
1 2 3 4 5 6 7
C4b
CS
C5
C3
C6
CS
C9
CS
CS
CS
CS
CS
CS
C
CS
1 2 3 4 5 6 7
Antigen
C3 convertase
(C2aC4b)
C5b
1 2 3 4 5 6 7
C5 convertase
(C2aC4bc3b)
CS
B
CD59
B
Clusterin
Membrane attack
complex
Lysis of target cell
Figure 6. Classical complement pathway associated with proteins differentially expressed in response to rifaximin treatment. Mapped on the pathway are
the set of proteins differentially expressed in the A-N (1), A-R (2), B-N (3), B-R (4), C-N (5), C-R (6) and D-N (7) pools, in comparison with the BV pool.
Experimental data are visualized on the map as upward (red) and downward (blue) thermometer-like icons that indicate, respectively, up-regulated
and down-regulated signals. The height of the coloured bar corresponds to the relative expression value for the particular protein. This figure appears
in colour in the online version of JAC and in black and white in the print version of JAC.
and classical pathways) (data not shown), similar to the previous
analysis of the HF and BVF pools. Based on the functional subnetworks built using the ‘analyse network’ algorithm, the proteins differentially expressed in response to rifaximin/placebo treatment
were primarily involved in cell differentiation (P ¼ 1.44×10245),
complement activation (P ¼ 4.10×10239) and response to chemical stimulus (P ¼ 4.34×10234) (Table S5, available as Supplementary data at JAC Online). Figure 6 shows the MetaCore map of five
dysregulated proteins (complement C3, complement C4-A, immunoglobulin g1, immunoglobulin m and plasma protease C1 inhibitor) in the classical complement pathway. All the proteins
were generally down-regulated after antibiotic administration
and the highest decreases were observed in the B-R pool, confirming the major impact exerted by group B treatment on the vaginal
environment.
2656
More than half (16/30, 53%) of the microbial proteins that were
differentially expressed in BV-affected women before and after
rifaximin/placebo treatment were from Lactobacillus species
(L. acidophilus, Lactobacillus brevis, L. casei, Lactobacillus delbrueckii
subsp. bulgaricus, L. gasseri, L. helveticus, Lactobacillus johnsonii),
and were mainly involved in glucose metabolism, replication and
protein synthesis. Interestingly, only trigger factor from L. brevis
was found to be down-regulated in all pools after rifaximin treatment, with a median 22.5-fold ratio. Contrasting expression
patterns among pools were observed for nearly half (7/16, 44%) of
the proteins from lactobacilli, suggesting a lack of correlation with
the antibiotic treatment. Fourteen (47%) dysregulated microbial
proteins were from other microorganisms that are associated with
the vaginal environment, Oenococcus oeni, Pichia guilliermondii,
Bifidobacterium longum subsp. infantis, Saccharomyces cerevisiae,
Impact of rifaximin on vaginal proteome
S. epidermidis, Ureaplasma parvum, Mycoplasma genitalium,
Escherichia coli and S. aureus.
Discussion
In the last decade, many studies have focused on the complex
pathophysiological processes underlying BV, with the aim of
finding efficient strategies for prevention and cure of this vaginal
condition.11,15,29 Nonetheless, the root cause of BV remains
poorly understood, even if the polymicrobial nature of the vaginal
microbiota has been shown to be implicated in the disease aetiology.30 – 33 The present study reports for the first time, to our
knowledge, the proteome profiling of VF from BV-affected
women in comparison with healthy women, and discusses the potential of rifaximin in restoring a healthy condition.
Profound changes were detected in the VF proteome of
BV-affected women compared with healthy women, as well as following rifaximin/placebo treatment. In particular, according to
principal component analysis, the major separation from the BV
condition was observed following treatment with 25 mg of rifaximin once daily for 5 days, suggesting that this dosage exerted
the major impact on the vaginal proteome. This finding is in agreement with both the clinical11 and microbiological response,15
which has demonstrated that this dosage is the most effective in
inducing a decline of BV-associated bacteria without affecting
the normal population of lactobacilli.
The MS/MS analysis of protein fractions allowed the identification of 118 human proteins differentially expressed between VF
from BV-affected and healthy women. A high percentage of proteins (100/118, 85%) had already been identified in human VF
and cervical mucus, indicating good consistency with the data
reported in the literature.3 The vast majority of the differentially
expressed proteins were up-regulated in BV-affected women, suggesting that BV is characterized by global stimulation of protein expression by the vaginal mucosa. Based on GO classification, a high
proportion of these proteins are involved in the innate immune response and metabolic processes and are localized in the extracellular space and cytoplasm, in line with the classification of VF
proteins reported by Zegels et al.3 Interestingly, almost all immunoglobulins and other immune molecules were up-regulated in
the presence of BV, suggesting a major role for the immune
system in the pathophysiological process underlying this vaginal
condition, as previously hypothesized.34,35 By contrast, only 5%
of identified proteins were involved in the inflammatory response,
in accordance with the assumption that BV is a non-inflammatory
vaginal infection,36 as compared with aerobic vaginitis,37 even
if few studies correlate BV with altered levels of certain
pro-inflammatory cytokines.38 A high number of intracellular and
cytoskeleton proteins and keratins were identified, probably resulting from the disruption of the epithelial cell layer.3 Interestingly,
according to MetaCore analysis of the networks involving the dysregulated proteins, the transcription factor SP1 was shown to be a
highly linked hub in top-scoring networks, built with the shortest
path algorithm. SP1 can activate or repress transcription in response to physiological and pathological stimuli, regulating the expression of a large number of genes involved in a variety of
processes, such as cell differentiation, cell growth, apoptosis, the
immune response, response to DNA damage and chromatin remodelling.39 However, whether and to what extent SP1 may
JAC
participate in the onset and progression of BV need to be further
explored.
To date, studies aimed at identifying potential protein markers
for BV have been conducted using traditional ELISA techniques.
In these studies, different markers, such as antimicrobial peptides18,19 and cytokines19 have been evaluated. By contrast, our
proteomic study, conducted with a high-throughput approach
without seeking specific protein targets, allowed us to characterize
the entire protein profile of VF under BV conditions and provided
novel information that could be integrated with data coming
from traditional analyses.
Following rifaximin/placebo treatment, 284 human proteins in
VF were differentially expressed compared with the BV condition.
More than 200 proteins (223/284, 79%) had already been identified in previous reports.3 Most of the dysregulated proteins were
down-regulated in patients treated with rifaximin, suggesting a
role for the antibiotic in counteracting the protein profile alterations
observed in BV-affected women. Similarly to the BV versus H comparison, the main categories resulting from GO classification referred to the innate immune response and complement
activation, and small-molecule metabolic process, whereas only
a small percentage (3%) was involved in the inflammatory response. Notably, immunoglobulins and other immune molecules
exhibited a trend towards under-representation, contrary to findings in the BV versus H dataset comparison, indicating a general
shutdown of the immune response after antibiotic treatment.
Our proteomic study also highlighted a different modulation of
the vaginal proteome according to the antibiotic dosage. The
largest number of differentially expressed proteins and the greatest fold changes in protein expression were indeed identified following treatment with 25 mg of rifaximin once daily for 5 days,
further confirming the major impact of this treatment regimen
on the BV-related proteome. This dosage has also recently been
shown to result in the best clinical response in a larger group of
women.11 Conversely, placebo administration was associated
with the lowest number of differentially expressed proteins and
the expression variation was often in the opposite direction to
the trend observed in rifaximin-treated women.
With regard to the microbial proteins whose expression was
found to be dysregulated in the BV versus the H group or following
rifaximin treatment, .50% were of Lactobacillus origin, as
expected considering the abundance of lactobacilli in the vaginal
microbiota. However, no correlation was observed between the expression levels of these proteins and the development of the pathology or the effects of antibiotic treatment. Despite this, the
identification of microbial proteins in VF described in our work
represents an important step forward in the knowledge of the
vaginal ecology because until now only one microbial protein has
been detected in VF.2
In conclusion, our study demonstrates for the first time that BV
is associated with profound changes in the VF proteome, mainly
with respect to the innate immune response, and suggests the
ability of rifaximin, especially at a dosage of 25 mg once daily for
5 days, to modulate the vaginal proteome by counteracting the
alterations associated with the BV condition. The proteomic data
reported in this work support the clinical and microbiological
results previously reported11,15 and open the way to further exploration of the real advantages of rifaximin in comparison with metronidazole and clindamycin in the treatment of BV.
2657
Cruciani et al.
Acknowledgements
We are grateful to the staff of the Bioanalytical Mass Spectrometry Facility
(University of New South Wales, Australia) where the proteomic
experiments were carried out.
Funding
This work was supported by a research grant provided by Alfa Wassermann
SpA (grant number 468/2012).
Transparency declarations
F. Calanni is an employee of Alfa Wassermann SpA and G. D. has received
advisory fees, lecture fees and grant support from Alfa Wassermann SpA
All other authors: none to declare.
Author contributions
F. Cruciani, V. W. and S. T. performed the experiments and statistical analysis
of the data. G. D. enrolled the subjects and collected the vaginal
samples. F. Calanni and P. B. supervised the study. B. V. conceived and
designed the experiments.
Supplementary data
Tables S1 to S5 are available as Supplementary data at JAC Online (http://
jac.oxfordjournals.org/).
11 Donders GG, Guaschino S, Peters K et al. A multicenter, double-blind,
randomized, placebo-controlled study of rifaximin for the treatment of
bacterial vaginosis. Int J Gynaecol Obstet 2013; 120: 131– 6.
12 Rivkin A, Gim S. Rifaximin: new therapeutic indication and future
directions. Clin Ther 2011; 33: 812– 27.
13 Bajaj JS, Heuman DM, Sanyal AJ et al. Modulation of the metabiome by
rifaximin in patients with cirrhosis and minimal hepatic encephalopathy.
PLoS One 2013; 8: e60042.
14 Maccaferri S, Vitali B, Klinder A et al. Rifaximin modulates the colonic
microbiota of patients with Crohn’s disease: an in vitro approach using a
continuous culture colonic model system. J Antimicrob Chemother 2010;
65: 2556 –65.
15 Cruciani F, Brigidi P, Calanni F et al. Efficacy of rifaximin vaginal tablets in
the treatment of bacterial vaginosis: a molecular characterization of the
vaginal microbiota. Antimicrob Agents Chemother 2012; 56: 4062 –70.
16 Kolialexi A, Mavrou A, Spyrou G et al. Mass spectrometry-based
proteomics in reproductive medicine. Mass Spectrom Rev 2008; 27: 624–34.
17 Fang X, Zhang WW. Affinity separation and enrichment methods in
proteomic analysis. J Proteomics 2008; 71: 284– 303.
18 Balu RB, Savitz DA, Ananth CV et al. Bacterial vaginosis, vaginal fluid
neutrophil defensins, and preterm birth. Obstet Gynecol 2003; 101: 862– 8.
19 Fan SR, Liu XP, Liao QP. Human defensins and cytokines in vaginal lavage
fluid of women with bacterial vaginosis. Int J Gynaecol Obstet 2008; 103:
50– 4.
20 Valore EV, Wiley DJ, Ganz T. Reversible deficiency of antimicrobial
polypeptides in bacterial vaginosis. Infect Immun 2006; 74: 5693 –702.
References
21 Shah SJ, Yu KH, Sangar V et al. Identification and quantification of
preterm birth biomarkers in human cervicovaginal fluid by liquid
chromatography/tandem mass spectrometry. J Proteome Res 2009; 8:
2407– 17.
1 Dasari S, Pereira L, Reddy AP et al. Comprehensive proteomic analysis of
human cervical-vaginal fluid. J Proteome Res 2007; 6: 1258– 68.
22 Scigelova M, Makarov A. Orbitrap mass analyzer—overview and
applications in proteomics. Proteomics 2006; 6 Suppl 2: S16–21.
2 Tang LJ, De Seta F, Odreman F et al. Proteomic analysis of human
cervical-vaginal fluids. J Proteome Res 2007; 6: 2874 –83.
3 Zegels G, Van Raemdonck GA, Tjalma WA et al. Use of cervicovaginal fluid
for the identification of biomarkers for pathologies of the female genital
tract. Proteome Sci 2010; 8: 63.
4 Bigelow JL, Dunson DB, Stanford JB et al. Mucus observations in the fertile
window: a better predictor of conception than timing of intercourse. Hum
Reprod 2004; 19: 889–92.
5 Donders GG, Van Calsteren C, Bellen G et al. Association between
abnormal vaginal flora and cervical length as risk factors for preterm
birth. Ultrasound Obstet Gynecol 2010; doi:10.1002/uog.7568.
6 Hillier SL, Holmes KK, Marrazzo JM. Bacterial vaginosis. In: Holmes KK,
Sparling PF, Mårdh PA et al., eds. Sexually Transmitted Diseases. New York:
McGraw– Hill, 2008; 737– 68.
7 Vitali B, Pugliese C, Biagi E et al. Dynamics of vaginal bacterial
communities in women developing bacterial vaginosis, candidiasis, or no
infection, analyzed by PCR-denaturing gradient gel electrophoresis and
real-time PCR. Appl Environ Microbiol 2007; 73: 5731– 41.
23 Ly L, Wasinger VC. Peptide enrichment and protein fractionation using
selective electrophoresis. Proteomics 2008; 8: 4197– 208.
24 Wasinger V, Ly L, Fitzgerald A et al. Prefractionation, enrichment,
desalting and depleting of low volume and low abundance proteins and
peptides using the MF10. Methods Mol Biol 2008; 424: 257– 75.
25 Donders GG, Van Calsteren K, Bellen G et al. Predictive value for preterm
birth of abnormal vaginal flora, bacterial vaginosis and aerobic vaginitis
during the first trimester of pregnancy. BJOG 2009; 116: 1315 –24.
26 Gatlin CL, Kleemann GR, Hays LG et al. Protein identification at the low
femtomole level from silver-stained gels using a new fritless electrospray
interface for liquid chromatography-microspray and nanospray mass
spectrometry. Anal Biochem 1998; 263: 93–101.
27 Karp NA, McCormick PS, Russell MR et al. Experimental and statistical
considerations to avoid false conclusions in proteomics studies using
differential in-gel electrophoresis. Mol Cell Proteomics 2007; 6: 1354– 64.
28 Storey JD, Tibshirani R. Statistical significance for genomewide studies.
Proc Natl Acad Sci USA 2003; 100: 9440–5.
8 Atashili J, Poole C, Ndumbe PM et al. Bacterial vaginosis and HIV
acquisition: a meta-analysis of published studies. AIDS 2008; 22:
1493– 501.
29 Fredricks DN, Fiedler TL, Thomas KK et al. Changes in vaginal bacterial
concentrations with intravaginal metronidazole therapy for bacterial
vaginosis as assessed by quantitative PCR. J Clin Microbiol 2009; 47: 721–26.
9 Schwebke JR. New concepts in the etiology of bacterial vaginosis. Curr
Infect Dis Rep 2009; 11: 143–7.
30 Drell T, Lillsaar T, Tummeleht L et al. Characterization of the vaginal
micro- and mycobiome in asymptomatic reproductive-age Estonian
women. PLoS One 2013; 8: e54379.
10 Bradshaw CS, Morton AN, Hocking J et al. High recurrence rates of
bacterial vaginosis over the course of 12 months after oral metronidazole
therapy and factors associated with recurrence. J Infect Dis 2006; 193:
1478– 86.
2658
31 Ling Z, Kong J, Liu F et al. Molecular analysis of the diversity of vaginal
microbiota associated with bacterial vaginosis. BMC Genomics 2010; 11:
488.
Impact of rifaximin on vaginal proteome
JAC
32 Ravel J, Gajer P, Abdo Z et al. Vaginal microbiome of reproductive-age
women. Proc Natl Acad Sci USA 2011; 108 Suppl 1: 4680– 7.
36 Donders GG. Definition and classification of abnormal vaginal flora. Best
Pract Res Clin Obstet Gynaecol 2007; 21: 355–73.
33 Srinivasan S, Hoffman NG, Morgan MT et al. Bacterial communities in
women with bacterial vaginosis: high resolution phylogenetic analyses
reveal relationships of microbiota to clinical criteria. PLoS One 2012; 7: e37818.
37 Donders GG, Vereecken A, Bosmans E et al. Definition of a type of
abnormal vaginal flora that is distinct from bacterial vaginosis: aerobic
vaginitis. BJOG 2002; 109: 34 –43.
34 Cauci S, Scrimin F, Driussi S et al. Specific immune response against
Gardnerella vaginalis hemolysin in patients with bacterial vaginosis. Am J
Obstet Gynecol 1996; 175: 1601 –5.
38 Beigi RH, Yudin MH, Cosentino L et al. Cytokines, pregnancy, and bacterial
vaginosis: comparison of levels of cervical cytokines in pregnant and
nonpregnant women with bacterial vaginosis. J Infect Dis 2007; 196:
1355– 60.
35 Eade CR, Diaz C, Wood MP et al. Identification and characterization of
bacterial vaginosis-associated pathogens using a comprehensive
cervical-vaginal epithelial coculture assay. PLoS One 2012; 7: e50106.
39 Chu S, Ferro TJ. Sp1: regulation of gene expression by phosphorylation.
Gene 2005; 348: 1 – 11.
2659