Article pubs.acs.org/jpr
Label-Free Quantitative Proteomics Reveals Differentially Regulated Proteins in Experimental Gingivitis Nagihan Bostanci,† Per Ramberg,‡ Åsa Wahlander,§ Jonas Grossman,§ Daniel Jönsson,⊥ Virginia Monsul Barnes,∥ and Panos N. Papapanou*,⊥ †
Oral Translational Research, Institute of Oral Biology, University of Zurich, Switzerland Department of Periodontology, Institute of Odontology, The Sahlgrenska Academy at Göteborg University, Göteborg, Sweden § Functional Genomics Center, Zurich, Switzerland ∥ Colgate-Palmolive Technology Center, Piscataway, New Jersey 08854, United States ⊥ Division of Periodontics, Section of Oral and Diagnostic Sciences, Columbia University College of Dental Medicine, New York, New York 10032, United States ‡
S Supporting Information *
ABSTRACT: We investigated the sequential protein expression in gingival crevicular fluid samples during the induction (I) and resolution (R) of experimental gingivitis. Periodontally and systemically healthy volunteers (n = 20) participated in a three-week experimental gingivitis protocol, followed by debridement and two weeks of regular plaque control. Gingival crevicular fluid (GCF) samples were collected at baseline, Day 7, 14, and 21 (induction; I-phase), and at Day 21, 25, 30, and 35 (resolution; R-phase). Liquid chromatography−tandem mass spectrometry (LC−MS/MS) for label-free quantitative proteomics was applied. A total of 287 proteins were identified including 254 human, 14 bacterial, 12 fungal, and 7 yeast proteins. Ontology analysis revealed proteins primarily involved in cytoskeletal rearrangements, immune response, antimicrobial function, protein degradation, and DNA binding. There was considerable variation in the number of proteins identified, both among subjects and within subjects across time points. After pooling of samples between subjects at each time point, the levels of 59 proteins in the Iphase and 73 proteins in the R-phase were quantified longitudinally. Our data demonstrate that LC−MS/MS label-free quantitative proteomics is valuable in the assessment of the protein content of the GCF and can facilitate a better understanding of the molecular mechanisms involved in the induction and resolution of plaque-induced gingival inflammation in humans. KEYWORDS: gingival crevicular fluid, inflammation, biomarker, periodontal disease, pathogenesis
■
INTRODUCTION Periodontal diseases are among the most common chronic, inflammatory conditions in humans.1 They are triggered by bacterial biofilms (dental plaque) that adhere to the tooth surfaces and elicit a local inflammatory response in the gingival tissues (“gingivitis”). Under the influence of genetic predispositions and certain acquired and environmental risk factors, gingivitis may progress into an inflammatory lesion that encompasses additional tooth-supporting structures (the periodontal ligament and the alveolar bone), resulting in periodontal tissue destruction (“periodontitis”) which may ultimately lead to tooth loss. Accumulating evidence suggests that periodontal infection/inflammation may also contribute to adverse general health outcomes including atherosclerosis and diabetes mellitus.2,3 Dental plaque-induced gingivitis is a fully reversible condition that has been used extensively as an appropriate experimental model for the investigation of the pathobiology of the incipient periodontal lesion in humans.4−7 Early studies © XXXX American Chemical Society
focused on the clinical, microbiological, and histopathological features of experimental gingivitis,7−9 while more recent studies have utilized microarray approaches to decipher gene expression signatures in the gingival tissues during the development and the resolution of gingival inflammation.10,11 Although these studies have pointed to novel pathways and molecules that are potentially involved in the pathobiology of the early stages of periodontal disease, they cannot disclose unequivocally the actual effector molecules involved in gingival inflammatory processes, as most biological functions are regulated on the protein level. Furthermore, collection of gingival tissue samples is invasive and cannot be used routinely for diagnostic or prognostic purposes. The gingival exudate known as “gingival crevicular fluid” (GCF) is considered a suitable, easily accessible alternative source of biological material that can be used in the study of gingival inflammation. Received: August 13, 2012
A
dx.doi.org/10.1021/pr300761e | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Article
of gingival tissue samples harvested from each individual to a minimum. Thus, 10 participants were included in a “gingivitis induction” group and another 10 in a “gingivitis resolution” group.
Both the volume of the GCF and its protein content have been shown to increase during the induction of experimental gingivitis.12,13 These proteins are primarily gingiva- and serum-derived, although bacterial proteins are also present. However, protein analysis in the GCF has been challenging due to the limited amount of fluid that can be harvested from each individual gingival site. As a result, the majority of available studies have focused on the identification of a single or a limited number of proteins by using ELISA or immunoblotting14−17 and are thus unable to adequately reflect the plethora of mediators that participate in the local inflammatory response in the dentogingival niche. More recently, bead-based multiplex approaches have been used to capture inflammatory signatures in GCF by mainly targeting inflammatory cytokines.18 Given that proteins do not exert their biological functions as single species but rather as part of multiprotein complexes,19 largerscale proteomic approaches have the potential to provide more comprehensive information than the conventional protein detection assays.20 Indeed, ongoing developments in mass spectrometry instrumentation and the accompanying separation techniques allow accurate assessment of expression levels of large numbers of proteins in complex samples.21 To date, two studies have carried out cross-sectional analyses of GCF samples derived from healthy and diseased periodontal conditions using proteomic approaches,22,23 and only one has assessed quantitatively the changes in GCF proteome signatures in the experimental gingivitis model using iTRAQlabeled samples and Fourier transform ion cyclotron resonance (FTICR) mass spectrometry.20 In this study, we hypothesized that the proteomic profiles in the GCF during the course of experimental gingivitis (i.e., the “gingival exudatome”) would be consistent with known elements of the pathobiology of the reversible gingival lesion but would potentially also disclose the involvement of novel, yet unrecognized proteins and processes. We used a label-free quantification LC−MS approach24−29 that requires lower protein amounts than isotope labeling methodologies and thus allows processing of individual samples that can serve as biological replicates.30 Our specific aims were to systematically investigate the sequential protein expression in the GCF that parallels (i) the gradual conversion from pristine periodontal health to a state of established gingivitis and (ii) the resolution of gingival inflammation during reinstitution of periodontal health.
■
Experimental Gingivitis Protocol
The experimental gingivitis protocol was described in detail in an earlier publication.10 All participants received dental prophylaxis by a dental hygienist. They were instructed in proper tooth brushing and interdental flossing and received fullmouth cleaning and polishing until they showed no or only minimal signs of gingival inflammation. Maxillary impressions were obtained, and acrylic stents that covered the palatal gingival tooth surfaces were fabricated. After establishment of the absence of gingival inflammation, experimental gingivitis was induced over a 21-day period at the maxillary palatal surfaces. During that time, the participants were asked to abstain from brushing of these surfaces and from any means of interproximal cleaning at the area. To prevent accidental removal of plaque from the experimental sites, the individually fabricated stents were always put in place during the regular brushing of the maxillary buccal surfaces and the mandibular teeth. After completion of the three-week gingivitis induction phase, all participants received thorough oral prophylaxis by the same dental hygienist, including full-mouth debridement and polishing. Oral hygiene measures including tooth brushing and dental flossing at least twice daily were reinstituted in the entire dentition. The “gingivitis resolution” phase was completed two weeks after reinstitution of regular oral hygiene. Clinical Examination
All volunteers were prescreened to ensure the absence of pockets with a probing pocket depth >4 mm. Gingival Index (GI)31 assessments were carried out bilaterally at the mesiopalatal and disto-palatal aspects of each interdental papilla between the first and second maxillary bicuspids and between the first maxillary bicuspid and cuspid, using a periodontal probe, at each time point described below. Collection of Gingival Crevicular Fluid Samples
Pairs of GCF samples were obtained from the mesiopalatal and distopalatal aspect of each interproximal papilla included in the experimental protocol at the following time points: in the induction group, at baseline (Day 0), Day 7, Day 14, and Day 21 of experimental gingivitis, and in the resolution group, at the completion of three weeks of experimental gingivitis at Day 21 and at Day 25, Day 30, and Day 35 after the provision of fullmouth prophylaxis and reinstitution of oral hygiene procedures. Each selected site was isolated from salivary contamination using cotton rolls, and a sterile Periopaper strip (OraFlow Inc., Amityville, NY) was gently inserted into the periodontal crevice and left in place for 30 s. Mechanical irritation was avoided, and strips contaminated with blood were discarded. GCF volumes were measured using a Periotron 8000 device (OraFlow Inc., Amityville, NY), and each pair of GCF strips corresponding to the same papilla and obtained at the same time point were placed into a single Eppendorf tube and stored at −80 °C.
EXPERIMENTAL PROCEDURES
Study Population
The study was approved by the Regional Ethical Review Board, Göteborg, Sweden (#005-09). Informed consent was obtained from each subject prior to enrollment in the study. Twenty systemically healthy volunteers (ten female and ten male) with no history of periodontal disease were recruited among the undergraduate students attending the Faculty of Odontology, Sahlgrenska Academy, Göteborg University, Sweden. Complete medical and dental histories were taken from all subjects. None of the subjects had a history of systemic disease or cigarette smoking or had taken medications such as antibiotics or contraceptives that could have affected their periodontal status for at least 3 months prior to enrollment. As this study was designed to concomitantly examine transcriptomic changes in the gingival tissues during the induction and resolution of experimental gingivitis,10 the 20 participants were divided into two groups comprising 10 individuals each, to keep the number
Processing of Gingival Crevicular Samples for Proteomic Analysis
Proteomic analyses of GCF samples was carried out at the laboratory of the Oral Translational Unit, Institute of Oral Biology, University of Zurich, Switzerland, to which the frozen GCF samples were shipped in dry ice. On the day of the analyses, the GCF samples (n = 80) were re-eluted in 100 μL of B
dx.doi.org/10.1021/pr300761e | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Article
and reversed sequences (a total of 197 099 sequences from human, bacterial, and fungal proteins including those from Streptococcus oralis, Streptococcus anginosus, Fusobacterium nucleatum, Campylobacter rectus, Prevotella intermedia, Porphyromonas gingivalis, Tannerella forsythia, Treponema denticola, Actinomyces naeslundii, Veillonella dispar, Aggregatibacter actinomycetemcomitans, and Candida albicans). Carbamidomethylation of cysteines was set as fixed modification, and methionine oxidation was set as variable modification in the database search. Additional search parameters were: maximum number of missed cleavages, 1; peptide mass tolerance, 20 ppm; and fragment mass tolerance, 0.8 Da. Only peptides with ion scores of ≥40 were considered, and only one peptide per protein was accepted with this high score threshold. The false discovery rate (FDR) at the peptide level was estimated to be 3.1%.
PBS and then centrifuged at 13 000g for 15 min. The total protein concentration of each of the sample extracts was measured and calculated using the Qubit fluorometer assay kit (Invitrogen, Life Technologies, Zug, Switzerland) according to the manufacturer’s instructions. Individual sample variation ranged from 20 to 148 μg/mL in the induction group and from 42 to 1620 μg/mL in the resolution group. The maximum amount of protein to be used from each sample was set by the sample of lowest amount; hence, an estimated amount of 2 μg of total protein from each sample was used. This amount was split in two to give 1 μg of protein in each of two technical replicate runs for standardization purposes across all samples. In addition, two sets of pooled samples were prepared: (i) eight pooled samples across all subjects, one at each time point, four from the induction (at Day 1, 7, 14, and 21), and another four from the resolution phase (at Day 21, 25, 30, and 35) that were used for quantification, and (ii) two pooled samples, one from all subjects and time points in each group that served as an alignment reference in the quantification analyses. Each sample was “spiked” with a known amount of two standard proteins: Alcohol Dehydrogenase (ADH, Saccharomyces cerevisiae) and Fetuin (Bovine) to an approximate ratio of 1:300, standard protein/sample total extracted protein amount (w/w). The samples were diluted with ammonium bicarbonate buffer which allowed for a maintained pH value of approximately 7 and then reduced with dithiothreitol (DTT, 45 min, 50 °C) and carbamidomethylated using iodoacetamide (45 min, dark in RT). The samples were digested overnight with trypsin (1:100 w/w) at 37 °C and then desalted with ziptips (Millipore, Zug, Switzerland). After concentration using a Speedvac (Thermo Savant SPD121P, Thermo Scientific, Wohlen, Switzerland), each sample was reconstituted in 3% acetonitrile (ACN) and 0.1% formic acid (FA).
Label-Free Quantification in the Pooled GCF Samples
For label-free quantification, the acquired raw data files corresponding to the eight pooled samples across time points and the two pooled reference samples were imported into the Progenesis LC−MS software (Nonlinear Dynamics, Newcastle upon Tyne, UK) for feature detection, alignment, and quantification. All sample features were aligned according to retention times by manually inserting up to seven landmarks followed by automatic alignment to maximally overlay all the two-dimensional (m/z and retention time) feature maps. Singly charged peptides and peptides with higher charge states than three were excluded from analysis. Furthermore, all features present in the 2D map before and after the applied gradient time were deleted since the void volume and the wash phase should contain no elements of interest. After alignment, samples were divided into the appropriate groups (I-phase, Day 0, 7, 14, and 21, and R-phase, Day 21, 25, 30, and 35), and raw abundances of all features were normalized against the externally spiked in reference peptides (ADH and Fetuin). The peak lists using the most intense 200-fragment deisotoped and deconvoluted ion peaks generated by the Progenesis LC−MS software were used for protein identification as described above and then reimported into the software. To maximize the number of quantifiable proteins, but at the same time keep the false discovery rate at an acceptable level, the Mascot ion score cutoff was adjusted accordingly. Only peptides with ion scores of ≥20 and with a rank of 1 were considered. The FDR at the peptide level was estimated to be 5%. For quantification, only unique peptides were included, and the total cumulative abundance was calculated by summing the individual abundances of all peptides assigned to each protein.
Liquid Chromatography−Tandem Mass Spectrometry (LC−MS/MS)
Each sample was divided into two technical replicates. Samples were injected onto an in-house pulled and packed tip column (length 8 cm) carrying Magic C18 AQ beads (3 μm bead size, 200 Å pore size; Bishoff Chromatography, Leonberg, Germany), 75 μm ID, 375 OD capillary, coupled to an Eksigent nanoLC-1D device (ABSciex, Zug, Switzerland). Samples were separated using a binary solvent system with a flow rate of 200 nL/min and eluted using a gradient from 2% B to 30% B over 60 min (A: 1% ACN, 0.1% FA; B: 100% ACN, 0.1% FA) and acquired using an LTQ Orbitrap (ThermoScientific, Wohlen, Switzerland) equipped with a nanospray ion source, running a standard collision-induced, data-dependent (CID-DDA) method of one survey (MS) scan, followed by five dependent scans (MS/MS) looped throughout the run. The survey scan was acquired from 300 to 2000 m/z units in profile mode with a resolution of 6000 in the Orbitrap. The dependent scans were acquired in centroid mode in the ion trap with a normalized collision energy of 28, activation energy of 0.25, and 30 ms activation time excluding singly charged ions for fragmentation. Dynamic exclusion was applied with a list size of 500 that was repeated every 30 s and with a duration of 120 s.
Data Clustering and Heat Maps
To obtain a global visualization and assessment of GCF protein expression profiles in the pooled samples, cluster analysis and heat maps were generated using the R software (R: A Language and Environment for Statistical Computing, R Development Core Team). Fold changes in abundance intensity were calculated based on value log-ratios (log2-transformed) between pairs of consecutive time points, Day 7/Day 0, Day 7/Day 14, and Day 14/Day 21 in the I-group and Day 25/Day 21, Day 30/Day 25, and Day 35/Day 30 in the R-group. In addition, log ratios were calculated in comparison to Baseline (Day 0) in the I-group and to Day 21 in the R-group.
Database Searching and Protein Identification
Proteins in both the individual GCF samples (n = 80) and the pooled samples (n = 8) were identified using a local installation of Mascotserver 2.3 search engine (Matrix Science, London, UK) against an in-house database constructed using both human and bacterial species, including common contaminants
Statistical Analysis
Significant differences between groups for gingival index, GCF volume, and total protein concentrations were determined C
dx.doi.org/10.1021/pr300761e | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Article
using the repeated measures analysis of variance. The Bonferroni post hoc test was used to compare differences between groups in the GraphPad Software. Differences were considered statistically significant at a p value of
