Article pubs.acs.org/jpr
Sulfated Hyaluronan Containing Collagen Matrices Enhance CellMatrix-Interaction, Endocytosis, and Osteogenic Differentiation of Human Mesenchymal Stromal Cells Stefanie Kliemt,†,# Claudia Lange,‡,# Wolfgang Otto,† Vera Hintze,§ Stephanie Möller,∥ Martin von Bergen,†,⊥,# Ute Hempel,*,‡,# and Stefan Kalkhof*,†,# †
Department of Proteomics, Helmholtz-Centre for Environmental Research-UFZ, Permoserstrasse 15, 04318 Leipzig, Germany Institute of Physiological Chemistry, TU Dresden, Fiedlerstrasse 42, Dresden 01307, Germany § Institute of Material Science, Max-Bergmann-Centre of Biomaterials, TU Dresden, 01069 Dresden, Germany ∥ Biomaterials Department, INNOVENT e. V., 07745 Jena, Germany ⊥ Department of Metabolomics, Helmholtz-Centre for Environmental Research-UFZ, 04318 Leipzig, Germany # Department of Biotechnology, Chemistry and Environmental Engineering, Aalborg University, Sohngaardsholmsvej 49,DK-9000 Aalborg, Denmark ‡
S Supporting Information *
ABSTRACT: Inorganic−organic composite implant materials mimicking the environment of bone are promising applications to meet the increasing demands on biomaterials for bone regeneration caused by extended life spans and the concomitant increase of bone treatments. Besides collagen type I (Col-I) glycosaminoglycans (GAG), such as hyaluronan, are important components of the bone extracellular matrix (ECM). Sulfated GAGs are potential stimulators of bone anabolic activity, as they are involved in the recruitment of mesenchymal stromal cells (MSCs) to the site of bone formation and support differentiation to osteoblasts. Nevertheless, no consecutive data is currently available about the interaction of hyaluronan or sulfated hyaluronan derivatives with hMSCs and the molecular processes being consequently regulated. We applied quantitative proteomics to investigate the influence of artificial ECM composed of Col-I and hyaluronan (Hya) or sulfated hyaluronan (HyaS3) on the molecular adaptation of osteogenic-differentiated human MSCs (hMSCs). Of the 1,370 quantified proteins, the expression of 4−11% was altered due to both aECM-combinations. Our results indicate that HyaS3 enhanced multiple cell functions, including cell-matrixinteraction, cell-signaling, endocytosis, and differentiation. In conclusion, this study provides fundamental insights into regulative cellular responses associated with HyaS3 and Hya as components of aECM and underlines the potential of HyaS3 as a promising implant-coating-material. KEYWORDS: artificial extracellular matrix, human mesenchymal stromal cell, hyaluronan, glycosaminoglycan, SILAC, quantitative proteomics
■
INTRODUCTION Because of the high regenerative capacity of bone, especially with younger patients, the majority of skeletal fractures will heal without the imperative of intervention. However, large criticalsized bone defects as a result of bone tumor resections and severe nonunion fractures need guided regeneration and surgical correction in order to heal properly.1 Due to extended life spans, an increasing number of elderly patients will suffer from diseases with impaired bone healing, such as diabetes mellitus and osteoporosis.2 Within the past few years a huge variety of biomaterials appeared on the market, including osteoinductive biomaterials, that induce bone formation and natural materials which have © 2012 American Chemical Society
the advantage of facilitating cell attachment or maintaining differentiation functions.3 On the other hand, there are synthetic polymers which allow precise control of a variety of parameters, including molecular weight, degradation time, hydrophobicity, and surface structure/roughness.4 To combine these advantages, many implants are made of composite materials, containing both an organic and an inorganic component, and possess porous three-dimensional structures.5,6 Macromolecules such as the extracellular matrix (ECM) components collagen type I (Col-I) and glycosaminoglycans Received: July 13, 2012 Published: November 21, 2012 378
dx.doi.org/10.1021/pr300640h | J. Proteome Res. 2013, 12, 378−389
Journal of Proteome Research
Article
available or proprietary software solutions have been developed for signaling and metabolic pathway analysis. Freely available software, such as Protein Analysis Through Evolutionary Relationships (PANTHER), is widely used,31 and therefore, the results can be easily reproduced and compared with those of many other “omics” studies. On the other hand, commercial software, such as Pathway Analysis (IPA), is based on a manually created pathway database that allows for a more dedicated data analysis.32 On the basis of 110 proteins which were found to be differentially expressed using Col-I and Hya as references, it was shown that, in particular, the HyaS3-aECM enhanced cellular functions associated with the actin cytoskeleton, cell-matrixinteraction, cell signaling, stress, endocytosis, matrix remodeling, and synthesis. The results were confirmed by Western blot analyses and immunofluorescence staining of selected proteins.
(GAGs) are promising candidates for tissue engineering as coatings for implant materials, such as titanium or ceramic, to increase osseointegration.6−10 Resulting inorganic−organic composites, mimicking the natural environment of the ECM of bone, have been shown to influence parameters such as healing time and implant stability by affecting osteoclasts and osteoblasts and their respective progenitors.11 GAGs are involved in a wide range of physiological processes, including cell proliferation and cell migration, inflammatory response modulation, and angiogenesis.12−14 Those effects are often dependent on the GAGs chemical structure as well as the degree and pattern of sulfation.15 To avoid side effects and to control specific properties of biomaterials, a consistent preparation and detailed characterization of the structure, including the arrangement of repeating units and functional groups, is required. Hyaluronan is a GAG made up of repeating disaccharide units (alternating D-glucuronic acid and D-Nactetylglucosamine units linked by alternating β-1,3- and β-1,4glycosidic bonds). This ordered structure and the ability to have chemical modifications without structure disruption makes it a promising candidate. Hyaluronan is readily available and can be chemically sulfated.16 Moreover, unmodified hyaluronan is known to initialize and regulate developmental processes during chondrogenesis and osteogenesis as well as to affect osteoclastogenesis and osteoblast differentiation.17,18 Sulfated hyaluronan derivatives have been shown to efficiently mimic heparin, promote cell adhesion, inhibit osteoblasts proliferation, increase osteoblast differentiation, and decrease osteoclastic resorption activity.18−22 The potential of improving the biocompatibility of biomaterials by applying sulfated hyaluronan was demonstrated by Cen et al.,23 who analyzed a modified electrically conductive polypyrrole with either hyaluronan or sulfated hyaluronan. In comparison, sulfated hyaluronan prolonged the recalcification time and reduced the adhesion of platelets. In this study we aim to investigate the influence of artificial ECM (aECM) composed of Col-I mixed with either nonsulfated hyaluronan (Hya) or sulfated hyaluronan (HyaS3) with an average sulfation degree of three on the adaptation and cellular response of human bone marrow derived mesenchymal stromal cells (hMSCs). The data was analyzed focusing mainly on the effects of HyaS3 relative to Hya and Col-I. The hMSCs were cultured onto different aECMs, differentiated into osteoblast lineage by established protocols, and analyzed by quantitative proteomics. Osteogenic differentiation of hMSCs can be interpreted as a series of events that results in a shift of the cellular gene expression profile due to external stimulation. Though microarrays or modern sequencing techniques are frequently used to measure changes in the mRNA levels during differentiation,24 proteins are the effector molecules in most cellular processes and, thus, are suitable to identify molecular processes occurring during differentiation. The combination of stable isotope labeling and modern mass spectrometry (MS) allows for an accurate quantification of hundreds of proteins.25,26 We used stable isotope labeling by amino acids in cell culture (SILAC), which is a simple but powerful technique for metabolic labeling of proteins during the cultivation of cells.27 Since the cellular homeostasis is tightly regulated, precise quantification of as many proteins as possible is needed in order to detect slight changes in molecular pathways based on the annotation of proteins to certain and specific pathways, as shown elsewhere.28−30 These pathways are the key for biological interpretation. Several either freely
■
EXPERIMENTAL PROCEDURES
Preparation and Characterization of Artificial Extracellular Matrices
Sulfated hyaluronan derivatives and aECMs were prepared and characterized as described in detail by Hempel et al.21 Artificial ECM composed of Col-I, Hya, and HyaS3 was used for in vitro experiments with hMSCs. Environmental Scanning Electron Microscopy
Artificial ECM was investigated with an environmental scanning electron microscope (ESEM, FEI-Company, Oregon, USA) in low vacuum at a working distance of 10 mm and with an electron beam energy of 10 kV. The resulting images reveal the surface topography of the matrices due to the detection of secondary electrons emitted from the matrix surface using a secondary electron detector. Isolation of Human Mesenchymal Stromal Cells
Bone marrow aspirates were collected from healthy donors (males, average age 33 ± 5.5 years) at the Bone Marrow Transplantation Centre of the University Hospital Dresden. MSCs were isolated from the aspirates according to Oswald et al.33 The study was approved by the local ethics commission (ethic vote No. EK114042009). All donor cells were treated as independent biological replicates without pooling. MSC preparations were chosen according to similar osteogenic differentiation potentials as assessed by alkaline phosphatase activity.34 For SILAC labeling, cells were used in the second passage. Cell Culture Media Composition
The SILAC protein ID and quantitation media kit (DMEM Flex) was purchased from Invitrogen (Karlsruhe, Germany), and the isotopes L-lysine and L-arginine labeled “heavy” medium was reconstituted according the manufacturer’s instructions. The SILAC medium contained either 0.1 mg (U-13C6)-L-lysine/mL and 0.1 mg (U−‑13C6, 15N4)-L-arginine/ mL or nonlabeled arginine and lysine in Dulbeccós Modified Eagle Medium (DMEM, Biochrom KG, Berlin, Germany) for both “heavy” and “light” SILAC medium, respectively. Both media were also supplemented with 200 g glucose/L, 10% heatinactivated fetal calf serum (HI-FCS), 200 mM glutamine, and 0.2% penicillin/streptomycin. SILAC Label Incorporation of hMSCs
To ensure a label incorporation of more than 95% for most rapidly dividing cell lines, five cell divisions are needed. 379
dx.doi.org/10.1021/pr300640h | J. Proteome Res. 2013, 12, 378−389
Journal of Proteome Research
Article
pieces, digested with trypsin, and measured by online reversedphase nanoscale liquid chromatography tandem mass spectrometry on a NanoAcquity UPLC system (Waters Corporation, Milford, MA, USA). This UPLC system was connected to an Velos-Orbitrap XL ETD (Thermo Fisher Scientific, Waltham, MA, USA) equipped with a nano-ESI source (TriVersaNanoMate, Advion, Ithaca, NY, USA) as described earlier22,36 using a 3 h acetonitrile gradient (2−85% phase B, acetonitrile/0.1% formic acid; phase A, water/0.1% formic acid) with a flow rate of 300 nL/min. MS analysis was performed in positive ion mode using a continuous scanning of eluted peptide ions in a mass range of 300−1600 m/z. The system automatically switched to CID-MS/MS mode on the six most intense ions with an intensity of more than 2,000 counts and a charge state higher than 1. For protein identification and relative quantification, MaxQuant software (version 1.1.1.25, www.maxquant.org) was used. A database search was carried out by the integrated search engine Andromeda against a concatenated database containing normal and reverse sequences of all proteins in the IPI human database (version 3.68, http://www.ebi.ac.uk/IPI/IPIhelp.html).37 False discovery rates (FDR) for peptide as well as protein identification of 0.01 were applied, and the precursor mass tolerance was set to 10 ppm and refined during MaxQuant processing as described.38 Protein quantification was based on at least two peptides. Oxidation (methionine), acetylation (protein Ntermini), and deamination (glutamine) were used as variable modifications. Carbamidomethylation (cysteine) was set as fixed modification, lysine-6 (13C-labeled) and arginine-10 (13C-, 15 N-labeled) were set as heavy SILAC labels, and a maximum of two tryptic missed cleavages were allowed. The option “match between runs” was applied with a time window of 3 min. For other settings, including ratio normalization, MaxQuant default parameters were used. MS raw files and MaxQuant parameter files are available at the PRIDE (PRoteomicsIDEntifications database) file sharing server (http://www.ebi.ac.uk/pride/). Normalized ratios were log2-transformed for further analysis, including the t-test (paired, two sided t-test).
However, hMSCs show a very slow cell division rate of 0.15− 0.25 cell divisions per day. For determination of the label incorporation rate, we cultured hMSCs as described in Stable Isotope Labeling of hMSCs for 10, 34, 48, and 64 days (corresponding to 2.2, 3.1, 5.1, and 5.4 cell divisions, respectively) in media containing heavy labeled lysine and arginine. Afterwards, cells were lysed and digested as described in Cell Lysis and Protein Extraction and Protein Separation, Liquid Chromatography/Tandem Mass Spectrometry and Data Analysis. Samples were measured by LC-MS/MS using a 30 min gradient. MaxQuant analysis (for detailed description see Protein Separation, Liquid Chromatography/Tandem Mass Spectrometry, and Data Analysis) of the raw files was then used to determine the ratio of heavy labeled peptides to the remaining nonlabeled ones. The incorporation efficiency was calculated as follows: 100[(Ratio(H/L))/(Ratio(H/L) + 1)] on the protein level, leading to the finding that, despite the fact of less than 5 divisions, the heavy/light ratio was constantly about 97% from day 34 on (data not shown). Stable Isotope Labeling of hMSCs
MSCs were expanded for 34 days (corresponding to 3−4 cell divisions) in the “heavy” medium at 37 °C, 5% CO2 to gain a stable isotope incorporation rate of more than 95% or in the “light” medium using the same conditions. Of those, 7,000 hMSC/cm2 (stable isotope labeled or nonlabeled) were cultured onto aECM-coated tissue culture polystyrene (TCPS) plates (6 cm, Greiner Bio-One GmbH, Frickenhausen, Germany). Osteogenic differentiation of hMSCs was induced at day 4 of culture by adding osteogenic differentiation medium (ODM; DMEM basal medium with 10% Hl-FCS and 0.2% penicillin/streptomycin, supplemented with 10 nM dexamethasone, 300 μM ascorbic acid, and 10 mM β-glycerophosphate containing either labeled or nonlabeled arginine and lysine). Medium change with ODM was carried out twice a week. Cell Lyses and Protein Extraction
At day 11 after seeding, hMSCs were collected using trypsin/ 0.25% EDTA for 20 min. After centrifugation at 221g for 4 min at room temperature, the cell pellets were resuspended in cold PBS and then centrifuged at 395g for 4 min at 4 °C. Next, the pellet was resuspended in cold PBS and 100 mM Tris-HCl with 50 mM MgCl2 (pH 8), followed by centrifugation at 11,500g for 6 min at 4 °C. The cell pellet was lysed using urea lysis buffer containing 2 M thiourea, 6 M urea, and 100 mM NH4HCO3. The extracts were clarified by centrifugation at 16,060g for 30 min at 4 °C. Protein concentration was determined using the Bradford protein assay (RotiQuant, Carl Roth GmbH, Karlsruhe, Germany).
SILAC Control Experiments
The effects of technical variance during cell culture (labeling effect) and the biological variance caused by different donors (donor effect) were determined in biological triplicates as recently described.39 Briefly, we mixed heavy and light Col I samples of either the same donor or different donors to estimate the variances caused by technical or biological variances. The samples were measured and analyzed analogously to the main SILAC experiments. Pathway Analysis and Functional Annotation
Protein Separation, Liquid Chromatography/Tandem Mass Spectrometry, and Data Analysis
An evaluation of our data showed that >66% of all proteins had a peptide variance below 50% and that the biological variance of >82% of all proteins was below 50% (Supporting Information Figure 1). Thus, proteins which were differentially expressed and displayed a log2(ratio (H/L)) ± 0.5 in at least two out of three biological replicates were considered to be regulated and were used for cluster analysis with the web-delivered commercial application IPA (“Spring Release (2012)”, Ingenuity Systems, www.ingenuity.com) and the PANTHER (www.pantherdb.org) classification system according their molecular functions.40 The “functional analysis” tool of IPA was used to extract biological functions and canonical pathways that were significantly altered within the data sets. Biological networks
For SDS-PAGE, equimolar amounts of the lysates (heavy/ light) were mixed according to the protein quantification. Protein mixtures, which contained 30 μg of protein in total, were precipitated with acetone and dissolved in 0.5 M Tris-HCl buffer (pH 6.8) containing 40% (v/v) SDS, 20% (v/v) glycerol, 2% (v/v) bromophenol-blue, and 10% (v/v) 2-mercaptoethanol. After heating for 5 min at 95 °C, a 1D-SDS-PAGE (12% resolving gel, 4% stacking gel) was carried out using Laemmli’s protocol.35 Gel electrophoresis was stopped after proteins entered approximately 3 cm in the gel. Gel lanes were excised without prior staining, cut into five gel slices, minced into smaller 380
dx.doi.org/10.1021/pr300640h | J. Proteome Res. 2013, 12, 378−389
Journal of Proteome Research
Article
are built upon generating interactions between the “focus genes” and all other genes stored in the knowledge base. The score for each network was calculated according to the set of the user’s significant genes. It indicates the likelihood of the genes in a network being found together due to random chance and is derived from a p-value calculated by Fisher’s exact test which is further used for ranking the functions according to their significance within the network.
using a RotorGene PG3000 (Corbett, Wasserburg, Germany). The relative expression values were determined by the RotorGene software release 6.0 for three different donors.
■
RESULTS
Artificial ECMs Reveal Homogeneous and Smooth Surface Structures in the Wet State
The surface morphology of the aECMs, both in a dry and wet state was analyzed by using an environmental scanning electron microscope (ESEM) (Figure 1a and b). ESEM images revealed
Western blot Analysis
For Western blot analysis, 14,000 hMSC/cm2 were seeded onto the aECM in 3 cm plates and cultured for 11 days, analogous to the aforementioned SILAC experiments. Cells were lysed using lysis buffer containing 10 mM Tris/HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 1% (v/v) Triton X-100, 1 mM Na3CO4, 1 mM PMSF, and 0.1 mM aprotinin. The cells were then scraped off from the plates and centrifuged at 16,060g for 30 min at 4 °C. The protein concentration in the cell lysates was determined using the Bradford Protein Assay. Western blot analysis were performed as previously described using the antibodies in Supporting Information Table 1a.41 Immunofluorescence
For immunofluorescence investigations, hMSCs were fixed with 4% paraformaldehyde (Merck, Darmstadt, Germany) after 11 days of cell culture and treated as previously described.42 In brief, after blocking nonspecific binding sites with 1% bovine serum albumin (Carl Roth GmbH, Karlsruhe, Germany), the cells were incubated with the primary and, subsequently, the secondary antibody (Supporting Information Table 1a) for 45− 60 min each. The nuclei of the cells were stained with DAPI. Finally, the samples were embedded in MOWIOL 4-88 (Carl Roth GmbH, Karlsruhe, Germany) and investigated using an Axiophot microscope and an AxioCam camera (Zeiss, Jena, Germany). Figure 1. Surface structure of aECM and matrix influence on hMSCs morphology and activity. Characterization of aECM surface topography by ESEM in dry (a) and wet (b) states, revealing strong differences in surface roughness in the dry state but smooth and homogeneous surface structures in the wet state. Investigation of the cell morphology of hMSCs grown on different aECMs for 11 days by immunofluorescence of F-actin (green) and nuclei (blue) (c), and analysis of the metabolic activity of the cells by MTS assay (d), representing properly grown cells and similar cell numbers on all three matrices, with increased metabolic activity on HyaS3 (n = 3). Investigation of processes involved in matrix remodeling (e). Procollagen I α1 mRNA was determined by real time PCR to show collagen synthesis (n = 3); MMP2 activity was assessed by zymography to reveal matrix degradation (n = 4). Both processes were decreased on HyaS3 compared to Hya.
MTS Assay
MTS assays of cells cultured on aECMs for 11 days were performed to determine the metabolic activity of the cells. The assays were conducted according the manufacturer’s instructions of the Promega Kit (CellTiter 96 AQueous One Solution Cell Proliferation Assay; Promega Corporation, Madison, USA). The cells were incubated with the MTS-solution, and the metabolic activity was measured spectrophotometrically (Benchmark Plus, Bio-Rad Laboratories, München, Germany). Zymography
Zymography was used to determine the activity of matrix metalloproteinase 2 (MMP2), using gelatinase and collagenase to degrade the organic matrix in the media of hMSCs cultured on different aECMs for 11 days. Proteins in the medium were separated in a gel containing 0.05% gelatin using SDS-PAGE. After activation of the MMP2 overnight in 1 M Tris (pH 7.4), 1 M CaCl2, 1 mM ZnCl2, the gel was incubated with a coomassie staining solution of coomassie brilliant blue R250 (Merck, Darmstadt, Germany) for 30 min. Quantification of bright bands in the stained gel was performed using the ImageQuant (GE Healthcare, Freiburg, Germany) software.
a highly decreased surface roughness of the matrices, with the roughest being Col-I, followed by Hya and HyaS3 in the dry state (Figure 1a). By examining the surface of dry Col-I matrices, collagen fibrils become observable, resulting in a fibrillar surface structure with a roughness in the micrometer range. This was less pronounced in the case of dry Hya, as the fibrils seemed to be more embedded in the matrix material. In contrast, the surface of dry HyaS3 on the micrometer scale shows a less rough and a more homogeneous morphology compared to Col-I and Hya, but reveals an increased surface structure on the nanometer scale when it was viewed with higher magnification. In the wet stage, the analysis of surface topography revealed a decreased surface roughness of all three matrices compared to dry matrices (Figure 1b). Due to swelling
Collagen Expression
The procollagen Iα1 mRNA expression was investigated by real time polymerase chain reaction (PCR) as previously described.21 All of the RNA was isolated from hMSCs cultured on different aECM for 11 days, and cDNA was synthesized to perform real time PCR (Supporting Information Table 1b) 381
dx.doi.org/10.1021/pr300640h | J. Proteome Res. 2013, 12, 378−389
Journal of Proteome Research
Article
ensure comparability to the other two experiments and in line with the standard significance estimation in proteomics experiments, the results of a paired t-test were added. The broader distribution (for plots including outliers, see Supporting Information Figure 2) of expression ratios indicates a stronger effect of HyaS3 (Figure 3a). For Hya, the expression of 47 proteins (4% of quantified proteins, Figure 3b) was either up- or down-regulated by more than 50% (Supporting Information Table 4). For HyaS3, the expression of 145 proteins (11% of quantified proteins, Figure 3b) was altered compared to the control Col-I (Supporting Information Table 5). Further analysis revealed how the altered proteins from both Hya and HyaS3 overlap and whether the main effect on the proteome of hMSCs was caused by hyaluronan itself or by sulfated hyaluronan. To estimate which effects were mainly induced by sulfated hyaluronan, we calculated the ratio of HyaS3 to Hya (HyaS3/Hya) for all proteins (in total 1,056) being quantified in both experiments. This comparison revealed the regulation of 110 proteins, which represent about 7.9% of the quantified proteins in at least two out of three biological replicates (Supporting Information Table 6, Figure 3b). In total, 235 proteins were consistently found to be differentially expressed in at least one of the three analyses (Figure 3c). Of those, 46 proteins (20% of the regulated proteins) were found to be altered by Hya (hyaluronan effect) and 110 proteins (47%) were altered when comparing HyaS3/ Hya (sulfation effect). Thus, on the basis of the assumption that those proteins are regulated due to the different degrees of sulfation of hyaluronan, it can be concluded that the sulfation effect (110 regulated proteins) is stronger than the hyaluronan effect (46 proteins). However, for 101 proteins which were only found to be affected in the HyaS3 data set, a large fraction of these could only be explained as the result of a combination of the sulfation and the hyaluronan effect.
processes in water, the surface structures of the wet aECMs were more homogeneous; however Col-I still displayed a slightly increased surface roughness in the micrometer range. Those results show that the surface structures of the three aECMs are highly different in the dry state, but they are smoother and quite homogeneous in the wet state, which represents the state in cell culture. Immunofluorescence staining of the cell nuclei and F-actin of cells grown on different aECMs for 11 days showed that the cells grew properly on all three matrices, displaying a similar number of cells (Figure 1c). Nevertheless, the arrangement of the F-actin stress fibers was varied and seemed to have a more regular structure on Col-I (Figure 1c). A MTS-assay of the cells cultured on different aECMs for 11 days showed that hMSCs cultured on HyaS3 displayed increased metabolic activity compared to cells cultured on Hya (Figure 1d). These results support the finding that the cells grew properly on the matrices, with slightly increased metabolic activity on HyaS3. Protein Quantification Using SILAC and LC-MS/MS Analysis
Proteomic analysis using SILAC technology was applied for the relative quantification of protein expression levels between Hya and HyaS3 in comparison to Col-I (Figure 2).
Pathway Analysis and Clustering According to the Molecular Function
The regulated proteins observed in the three different comparisons were initially clustered according the assigned biological processes given in the gene ontology (GO) database, which uses the PANTHER classification system. The most prevalent terms were cytoskeletal organization, cell adhesion/ GTPase, cellular stress, and metabolism. However, differences in the number of proteins assigned to these terms were observed in the three analyses (Figure 4a). For Hya, cytoskeletal proteins (38%, 16 proteins, fold enrichment 2.8) formed the biggest cluster followed by the clusters metabolism (24%, 10 proteins, fold enrichment 1.0), stress (19%, 8 proteins, fold enrichment 1.4), and cell adhesion (7%, 3 proteins, fold enrichment 2.4). Altered proteins on HyaS3 were mainly involved in stress (42%, 63 proteins, fold enrichment 3.3) followed by cell adhesion (14%, 21 proteins, fold enrichment 2.0), metabolism (9%, 14 proteins, fold enrichment 1.5), and the cytoskeleton (5%, 7 proteins, fold enrichment 0.4). Similar to the HyaS3 data, a large number of regulated proteins in comparison of HyaS3 to Hya were stress-related (25%, 26 proteins, fold enrichment 2.2) and metabolismassociated (24%, 25 proteins, fold enrichment 1.5). Proteins were also involved in cytoskeleton rearrangement (19%, 20 proteins, fold enrichment 1.5) and in cell adhesion (12%, 13 proteins, fold enrichment 1.3).
Figure 2. Schematic workflow of the qualitative and quantitative proteomics analysis of osteogenic differentiated hMSCs on aECMs consisting of Col-I, Hya, and HyaS3.
Though 1,488 proteins were identified, of which 1,370 were quantified on the basis of at least two peptides (Supporting Information Table 2), of those, 1,017 were found in at least two out of three biological replicates (Supporting Information Table 3). Of those proteins, only the ones that had an average log2ratio of more than +0.5 or less than −0.5 were considered to be differentially expressed; these criteria were only fulfilled by 1, respectively, 4 proteins in analogously conducted control experiments, which were conducted to estimate the donor as well as the labeling effect. For both experiments, a Kolmogorow−Smirnow-test was performed. The log2-values of the measured ratios were found to be normally distributed for Hya (p-value 0.6396) and not normally distributed (p-value 0.0000), but instead gamma distributed for HyaS3. However, to 382
dx.doi.org/10.1021/pr300640h | J. Proteome Res. 2013, 12, 378−389
Journal of Proteome Research
Article
Figure 4. (a) Regulated proteins clustered in distinct biological processes using PANTHER. Pie charts of processes associated with the cytoskeleton, cell adhesion, cellular stress, and metabolism are shown for Hya, HyaS3, and HyaS3/Hya. (b) Affected signaling pathways determined by IPA. The diagram shows the top 15 canonical pathways predicted by IPA. Blue bar charts illustrate the −log(p-value) for each pathway and, in addition, the sum of the ratios (log2ratio(H/L)) of proteins clustered in a specific signaling pathway.
significantly regulated pathways. Due to the positive effects in wound healing resulting from hyaluronan sulfation, the focus of the analysis was laid on the effects of HyaS3 compared to Hya. We used the calculated ratios of HyaS3 and Hya to eliminate the impact of the reference Col-I. On the basis of the list of 110 regulated proteins, 15 pathways were determined to be affected (Figure 4b). Among those, actin cytoskeleton signaling (p-value 1.14 × 10−4), integrin signaling (p-value 2.51 × 10−3), and calcium signaling (p-value 2.74 × 10−7) were significantly altered. All regulated pathways which are discussed below were seen to be significant (Fisher’s exact test right-tailed p-value
