Accepted Article
FGF-2 antiproliferative stimulation induces proteomic dynamic changes and high expression of fosB and junB in K-Ras-driven mouse tumor cells
Francisca Nathalia de Luna Vitorino1, Fabio Montoni1, Jaqueline Neves Moreno1, Bruno Ferreira de Souza1, Mariana de Camargo Lopes1, Barbara Cordeiro1, Cecilia Sella Fonseca1,2, Joshua M Gilmore3, Mihaela I Sardiu3, Marcelo Silva Reis1, Laurence A Florens3, Michael P Washburn 3,4, Hugo Aguirre Armelin1, Julia Pinheiro Chagas da Cunha1 *.
Laboratório Especial de Ciclo Celular - Center of Toxins, Immune-Response and Cell Signaling - CeTICS, Instituto Butantan, São Paulo, SP, 05503-900, Brazil.
Departamento de Bioquímica, Instituto de Química, Universidade de São Paulo, São Paulo, SP, 05508-000, Brazil.
Stowers Institute for Medical Research, Kansas City, MO, 64110, USA.
Department of Pathology and Laboratory Medicine, University of Kansas Medical Center, Kansas City, KS, 66045, USA
Running title
Proteomics of FGF2 antiproliferative stimulation
Keywords
proteomics, fosB, junB, FGF2, src, antiproliferative, and DNA replication.
Footnotes
*Correspondence and Lead Contact: julia.cunha@butantan.gov.br. Phone: +55 11- 26279731
Av. Vital Brasil, 1500. Sao Paulo-SP. Brazil. +55-11-2627-9731
Received: 05 11, 2018; Revised: 06 28, 2018; Accepted: 07 16, 2018
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/pmic.201800203 .
Accepted Article
Financial support: Grants 22619-7 and 07467-1 from Sao Paulo Research Foundation (FAPESP) and CAPES. JMG, MES, LF, and MPW were supported by the Stowers Institute for Medical Research.
Abbreviations:
SCX, Stage Tip Fractionation;
LFQ, Label-Free Quantification; GO, Gene Ontology;
WCE, Whole Cell Extract;
PPI, Protein-Protein Interaction;
FGF2, Fibroblast Growth Factor 2;
FBS, Fetal Bovine Serum;
MudPIT, Multidimensional Protein Identification Technology PLGEM, Power Law Global Error Model
CDK, Cyclin Dependent Kinase
Abstract
FGF-2 is a well-known cell proliferation promoter; however, it can also induce cell cycle arrest. To gain insight into the molecular mechanisms of this antiproliferative effect, we investigated, for the first time, the early systemic proteomic differences induced by this growth factor in a K-Ras-driven mouse tumor cell line using a quantitative proteomics approach. More than 2900 proteins were quantified, indicating that terms associated with metabolism, RNA processing, replication, and transcription were enriched among proteins differentially expressed upon FGF2 stimulation. Proteomic trend dynamics indicated that, for proteins associated mainly with DNA replication and carbohydrate metabolism, an FGF2 stimulus delayed their abundance changes, whereas FGF2 stimulation accelerated other metabolic programs. Transcription regulatory network analysis indicated master regulators of FGF2 stimulation, including two critical transcription factors, fosB and junB. We investigated their expression dynamics both in the Y1 cell line (a murine model of adenocarcinoma cells) and in two other human cell lines (SK-N-MC and UM-UC-3) also susceptible to FGF2 antiproliferative effects. Both protein expression levels depended on FGFR and src signaling. JunB and fosB knockdown did not rescue cells from the growth arrest induced by FGF2; however,
fosB knockdown rescued cells from DNA replication delay, indicating that fosB expression underlies one of the FGF2 antiproliferative effects, namely, S-phase progression delay.
Statement of significance of the study
Fibroblast growth factor 2 (FGF2) is a well-known cell proliferation promoter, but in some cellular contexts, such in some breast tumors (mainly MDA-MB-134 and MCF7) and neuroepithelial cell lines as well as in mice adrenocortical tumor cell line (such as Y1 lineage mainly used in this study), among others; FGF2 stimulation induces proliferation arrest (1, 2] [3, 4). This FGF2-induced cell cycle arrest likely has important implications in cancer biology. However, so far, the molecular mechanisms underlying these FGF2 antiproliferative effects have not been fully elucidated. Here, we provide timely molecular mechanistic results regarding FGF2 antiproliferative effects, an area overlooked in past years. In addition, our data suggest that fosB expression underlies one of the FGF2 antiproliferative effects, namely, S-phase progression delay. To our knowledge, this report is the most extensive analysis of a systemic response of a tumor cell line to FGF2 antiproliferative stimulation.
Accepted Article
Introduction
The GO→G1→S transition is regulated by transcription waves that properly induce changes in gene expression to commit cells to DNA replication. Dysregulations of these events play a key role in oncogenesis. Growth factors induce activation of cyclin-dependent kinases (CDKs) that, in turn, activate downstream effectors to promote cell cycle progression. C-fos, c-jun, and c-myc are well-known transcription factors that belong to the immediate early gene response, which is rapidly and transiently turned on in response to mitogenic stimulation. Once activated, they propagate a wave of stimulation of other factors termed secondary response genes that may or may not include other transcription factors [5].
JunB is part of the JUN protein family, which includes c-Jun, junB and junD, which in turn act as transcription factors involved in the regulation of gene activity followed by a response to growth factors. JunB acts as a negative regulator of cell division, an inducer of senescence, a tumor suppressor, and an antagonist of cyclin D1 induction by c-jun. In the absence of c-jun, junB can act as
Accepted Article
a positive regulator of replication, whereas in the presence c-jun, the formation of c-jun/junB heterodimers occurs; consequently, it is possible to observe antiproliferative activities [6].
FosB is part of the FOS family of proteins, which also includes c-Fos, Fra-1 and Fra-2. Specifically, the truncated form of FosB is well studied for its involvement in neuroplasticity and drug addiction 7. The activity of all members of this family is modulated by phosphorylation generated by different kinases (MAPK, cdc2, PKA, PKC) [8].
Fibroblast growth factors (FGFs) are part of a family of mitogenic factors, which in mammals, comprises 23 members that are involved in regulating various activities such as angiogenesis, proliferation, differentiation, and survival (for a review, see (91). The binding of FGF to its receptor induces the activation of four major pathways: Ras-Raf-MAPK; PI3K-Akt; STATS; and phospholipase-C [9]. FGF2, also called basic FGF [10], is known as a mitogenic and survival factor and has a crucial role in maintaining the undifferentiated state and self-renewal of human embryonic stem cells [11]. Although it is constantly associated with oncogenesis due to its proliferative actions, in some cellular contexts, FGF2 acts as an anti-proliferative and tumor suppressive agent. For example, in human neuroepithelial cells (SK-N-MC), FGF2 generates an accumulation of cells in G2 mainly through interference in the phosphorylation cascade, which begins with cdc25 and leads to the activation of the mitotic complex cyclin B-CDK1 (2). In breast tumor cell lines, including MDA-MB-134 and MCF7, FGF2 arrests growth proliferation (1, 12). In the latter, FGF2 increases the levels of cyclin D1, cyclin E, and cdk4 (which are important for cell cycle progression in G1) and increases the levels of p21, a cdk inhibitor [1, 13]
In a mouse adrenocortical carcinoma cell line (Y1 lineage), physiological levels of FGF2 induce S-phase delay progression and cell cycle arrest at G2/M, leading to the appearance of senescence markers. Neither apoptosis nor necrosis is observed. These phenotypic effects occur through FGF receptors, and neither depends on the MEK/ERK or PI3K/Akt mitogenic pathways. However, RhoA and src signaling are activated upon FGF2 stimulation and play a key role in cell cycle arrest (3, 4). This peculiar action of FGF2 as an antiproliferative agent is dependent on high levels of active ras because the effects described above are also observed in Balb3T3 cells transformed by the H-rasV12 oncogene [3].
Proteomic analyses, mainly phosphoproteomics, have been used to study FGF2 signaling [14]. It was observed that cyclin D2 is highly phosphorylated by FGF2 signaling in a src-dependent way [15]. However, to our knowledge, no in-depth proteomic analysis has been performed to explore FGF2-
Accepted Article
antiproliferative effects; thus, the molecular mechanisms involved in these phenomena are poorly understood. Understanding this mechanism is crucial to shed light on FGF2 downstream activation, as deregulation of FGF signaling is associated with many diseases, including cancer (16).
In the present work, we investigated global proteomic alterations induced by FGF2 in comparison to serum stimulation. We found that many proteins were altered and were mainly involved in metabolism, the cytoskeleton, replication, and transcription. Transcription regulatory networks indicated master regulators of FGF2 signaling, and here, we further investigated some signaling cascades and knockdown effects of two of these transcription factors, fosB and junB, both overexpressed upon FGF2 stimulus. We discuss in detail both the proteomic alterations detected, as well as the participation of these transcription factors in the phenotypic alterations induced by FGF2.
Materials and Methods
1. Cell culture, serum-starvation, growth factor stimulation and inhibitory assays
Y1 murine adrenocortical carcinoma cells (17) and two human (SK-N-MC and UM-UC-3) cell lines were obtained from ATCC and were grown at 37℃ with a 5% CO2 atmosphere in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (FBS), ampicillin (25 mg/L), streptomycin (100 mg/L) and sodium bicarbonate (1.2 g/L). SK-N-MC was derived from human brains (from a metastatic site at supra-orbital area) whereas UM-UC-3 was derived from a transitional cell carcinoma from urinary bladder. Cells were plated (4 x 105) and cultured until 40% confluency when they were starved by depletion of FBS for 48 h. Next, the cells were stimulated with DMEM medium supplemented with 10% FBS with or without 10 ng/ml FGF2. For inhibitory assays, 10 uM PP1 inhibitor, and 150 nM PD173074 were added to cells 10 min or 1 h before growth factor stimulation. Samples were harvested at the indicated time points. Sub-lines of Y1 RhoA-V14 (constitutively active), RhoA-N19 (dominant-negative), and cells carrying the empty vector (PCM) were kindly provided by Dr. Fabio Forti (USP) (3). The RhoA activity in each subline was previously demonstrated 3 Eight was the general number of passages between collection and thawing. Cells were tested against Mycoplasma every year.
2. Total protein extracts, SDS-PAGE and Western blot
Cells were harvested, washed with ice-cold PBS, and lysed in RIPA buffer containing protease, phosphatase and deacetylase inhibitors (0.1 mM PMSF, 2 mg/L leupeptin, 0.1 mM sodium orthovanadate, 1 uM pepstatin, 10 mM pyrophosphate, 1 mM glycerophosphate, 5 mM sodium
Accepted Article
butyrate and 1 mM NaF). The extracts were quantified using a BCA Protein Assay Kit (Thermo Scientific) following the manufacturer’s instructions. Proteins were fractionated by 12% SDS-PAGE and transferred to nitrocellulose membranes (GE). Membranes were blocked for 1 h with TBS-T (Tris-Buffer-Saline with 0.1% Tween 20) containing 7% non-fat milk, then incubated for 1 h at room temperature or overnight at 4℃ with the antibodies diluted in the TBS-T milk solution. Membranes were washed three times with TBS-T and then incubated with the secondary antibody conjugated with horseradish peroxidase for 50 min, followed by washes with TBS-T only. Immunoreactive signals were visualized using the ECL Prime WB Detection Reagent Kit (GE Healthcare) and detected in a UVITEC Cambridge photodocumentation system.
3. LC-MS/MS data acquisition, processing, and analysis
Total protein extracts from Y1 cells stimulated (as described above) with serum or serum plus FGF2 (10 ng/ml) for 0 h, 3 h, and 5 h were obtained in 3 biological replicates. One hundred and fifty micrograms of the proteins was TCA precipitated and dissolved in 100 mM Tris-HCL, pH 8.5, 8 M urea. The sample was reduced with 5 mM TCEP, incubated at room temperature for 30 min, and carboxymethylated by the addition of 10 mM of iodoacetamide. Proteins were digested with endoproteinase Lys-C for 4 h and then digested with trypsin overnight at 37 ℃. The peptides were loaded onto a 100 um three-phase capillary column packed with 8 cm of 5 um C18 reverse phase (Aqua; Phenomenex) particles followed by 3 cm of 5 um SCX resin (Partisphere SCX; Whatman) and an additional 1.5 cm of reverse phase. Columns loaded with the peptides were placed in-line with an Agilent 1100 quaternary HPLC connected with an LTQ-XL (Thermo) mass spectrometer equipped with a nanospray source. Twelve automatic steps were used in the MudPIT run with increasing concentrations of ammonium acetate. Each full MS scan (400 m/z to 1600 m/z) was followed by 5 MS/MS events fragmented by CID. RAWDistiller v.1.0, an in-house software [18] was used to create the peak lists in ms2 format. The MS/ MS spectra were searched using SEQUEST v.27 (rev.9) with a mass tolerance of 3 amu for precursor and +0.5 amu for fragment ions. No enzyme specificity was imposed for the search against a protein database combining 26,214 mouse proteins (NCBI, 2012-03- 09 release), as well as 167 usual contaminants. To estimate false discovery rates (FDRs), each protein sequence was randomized, leading to a total search space of 52,762 sequences. To account for carboxamidomethylation, 57 Da were added statically to cysteines, while a differential search of +16 Da on methionine accounted for oxidation (the maximum number of modified residues per modification was set at 2). Spectra/peptide matches (PSMs) were filtered using conservative criteria using DTASelect 191: PSMs were only retained if they had a DeltCn of at least 0.08; minimum XCorr values of 1.8 for singly, 2.0 for doubly, and 3.0 for triply charged spectra; maximum Sp rank of 10. In
Accepted Article
addition, peptides had to be fully tryptic and at least 7 amino acids long. All 15 analyses were compared using CONTRAST; combining all runs, proteins had to be detected by at least 2 peptides and/or 2 spectral counts. Proteins that were subsets of others were removed using the parsimony option in DTASelect on the proteins detected after merging all runs. Proteins that were identified by the same set of peptides (including at least one peptide unique to each protein group to distinguish between isoforms) were grouped together, and one accession number was arbitrarily considered representative of each protein group. NSAF7 [20] was used to create the final report on all detected peptides and nonredundant proteins identified. Spectral and protein-level FDRs were, on average, 0.07 ± 0.03% and 0.49 ± 0.17%, respectively. To estimate relative protein levels, distributed Normalized Abundance Factors (dNSAFs) were calculated for each non-redundant protein or protein group, as described in 201, where spectral counts for shared peptide j (sSpC) were distributed based on spectral counts unique to each protein/protein group i (uSpC) divided by the sum of all unique spectral counts for the M protein isoforms that shared peptide j with protein i. All mass spectrometry data are available from ftp://MSV000081638@massive.ucsd.edu, with password = FNdLV30550. The Proteome Xchange accession number is: PXD008026.
4. Experimental Design and Statistical Rationale
Whole cell extracts from cells stimulated for 3 h and 5 h with serum and serum plus FGF2 as well as from cells arrested at G0/G1 were obtained in three biological replicates, totaling 15 samples (5 types of samples) and analyzed independently by MudPIT (Figure S1A). The total number of proteins passing the selection criteria described above were close among the biological replicates and similar across the 5 sample types (Figure S1B). Pearson correlation coefficients for the pairwise analysis of biological replicates (calculated in Excel; bottom of Table S1) were above 0.945, on average (Figure S1C). We concluded that three biological replicates were enough to reach a stationary number of identified proteins, since good agreement between biological replicates was observed. To be subjected to further statistical analysis, proteins had to be identified in at least 2 of 3 replicates in at least one of the 5 sample types (Table S1). We used the Power Law Global Error Model (PLGEM package in R version 3.2.1) to evaluate significant differences between time points with and without FGF2 stimulation. Pavelka et al (21) demonstrated that this model can be applied to label-free quantitation based on spectral-counts. PLGEM (21) was used to calculate signal-to-noise (STN) ratios between FGF2- and serum-treated samples and derive p-values for significant enrichment of proteins between time-matched samples (Table S3). Proteins with p<0.05 were considered differentially expressed. The resulting lists of proteins were analyzed by DAVID 6.8
Accepted Article
(http://david.abcc.ncifcrf.gov). Coomassie-stained gels, western blotting, and electropherograms results reported in this study are representative images of (at least) 3 replicates obtained from independent biological replicates.
5. Fuzzy c-means clustering
Average dNSAF values for each protein were used to generated fuzzy c-means clustering using http://computproteomics.bmb.sdu.dk/Apps/FuzzyClust/.
6. Cytoscape analysis
To perform an interactome network analysis for each condition, all detected proteins were subjected to id mapping and network topology analyses. To this end, we coded Jupyter Notebooks (v.6) with the Python3 kernel relying on pandas and NetworkX libraries. Id mapping: We used BLAST to apply the reciprocal best hit approach to map our protein database to the Mus musculus database available at STRING (version 10.5). The obtained ids were then mapped to the UniProt database to retrieve relevant metadata such as “Gene Name” and associated GO terms. Network topology analyses: Interactome networks were designed for each condition, using protein-protein interactions extracted from the STRING database. We removed edges with a STRING experimental score equal to zero and combined scores less than or equal to 0.4. Finally, for each network, only the largest component (i.e., the largest maximal connected graph) was kept, discarding all the remaining components. Each network node was annotated with three attributes: 1) Relative expression: either differentially expressed with statistical significance, exclusive to one condition, or unaffected; 2) Node betweenness centrality; and 3) Number of connections. Each network edge was annotated with two attributes: 1) Edge betweenness centrality; 2) Whether it was a connection exclusive to a condition or not. Subnetworks were generated to filter proteins of interest, such as the top 20 differentially expressed proteins with highest betweenness centrality. All network analyses were performed using NetworkX, and Cytoscape (version 3.4.0 or 3.6.0) was used for visualization. The PPI network analysis is also shown in Table S4. The apps Reactome FI and iRegulon (v1.3) were used to identify enriched pathways and master regulators, respectively. For iRegulon, all differentially expressed proteins upon 3 h of stimulation (ups and downs - 261 proteins) were used as the input. The search for “predict for regulators and targets” was performed searching for 10 K Motif selection of Mus musculus and all other default options.
Accepted Article
7. Growth curve
For growth curves, 5x104 cells were cultivated in six-well plates, starved for 48 h and stimulated at the indicated timepoints. Cells were harvested and counted in a Beckman Coulter cell counter.
8. EdU incorporation assay
The EdU incorporation assay was performed using the Click-iT Edu Alexa Fluor 647 imaging kit (Invitrogen) and flow cytometry kit (Invitrogen) according to the manufacturer’s directions. Cells were stimulated (as previously described) with FBS or FBS plus 10 ng/ml of FGF2 for 16 h followed by incubation with 2 mM EdU for 1 h. Cells were fixed, permeabilized, and incubated with Alexa Fluor 647 azide. The cell cycle was estimated by propidium iodide incorporation. For slides, Alexa Fluor 488 azide and DAPI were used for EdU and nuclear staining, respectively. Images were acquired using a 10X objective on a Nikon microscope. ImageJ was used to calculate the number of positive EdU cells relative to the number of total cells (DAPI staining).
9. FosB and junB knockdown
Y1 cells (2.5x105) plated onto a 6-well plate were transfected with 5 µg of the shRNAs together with transfection agent Lipofectamine 3000 (Invitrogen) following the manufacturer’s instructions. shRNA for Fosb 85198, Junb 232241, and GFP were purchased from Sigma. After 24 h of transfection, cells were cultured with medium supplemented with 5 µg/L of puromycin. A stably transfected population was obtained after approximately 12 days of antibiotic selection. The following siRNAs were purchased by Qiagen: FosB_9 (CTGGAGCGCTTTATACTGTGA); Fosb_4 (CAGGCGGAAACTGATCAGCTT) and AllStars siRNAs (control). For siRNA assays, six hours (according to Qiagen, 6 hours is the minimum time after transfection to observe gene knockdown). prior to growth factor stimulation, 10 nM concentrations of each siRNA were added to the cell culture together with 4.5 ul of the HiPerFect transfection reagent (Qiagen). After the indicated time of growth factor stimulation, cells were harvested and lysed in SDS-PAGE sample buffer. Alternatively, the EdU incorporation assay was performed using the protocols described above.
10. qPCR assays
mRNA was obtained using Illustra RNAspin Mini RNA Isolation (GE Healthcare Life Sciences) followed by a reverse transcription reaction using 2 ug of RNA following SuperScript III Reverse Transcriptase (Thermo), as per the manufacturer’s instructions. qPCRs were performed by using SYBR Green Master mix (Thermo) and 300 nM concentrations of forward and reverse primers (Thermo) targeting genes FOSB and RPL27. The latter was used as a housekeeping gene. Reactions were performed in both technical and biological triplicates on a StepOnePlus System Real-Time PCR
Accepted Article
(Applied Biosystems). Relative expression levels of fosB mRNA were calculated according to the Pfaffl method (22]
11. Antibodies, Inhibitors and Primers
Santa Cruz Biotechnology: C-Fos (sc52), JunB (sc73), c-Jun (H79).
Cell Signaling Technology: FosB (2251), JunB (3753), p44/42 MAPK(ERK1/2) (9102), phospho- p44/42MAPK(ERK1/2) (9101), c-Fos (4384), phosphor-c-Fos (5348).
KPL: anti-rabbit IgG (474-1506), anti-mouse IgG (074-1806), anti-goat IgG (141306).
Abcam: HPRT (ab109021).
Inhibitors: PP1 567809 (Merck), PD173074 (Sigma).
FosB F: CCG AGA AGA GAC ACT TAC CC
FosB R: CTC TTC AAG CTG ATC AGT TTC C
RPL27 F: GGA AAG TGG TGC TGG TCC T
RPL 27 R: ACC AGG GCA TGG CTG TAA G
Results
Quantitative proteomics and trend dynamics analysis upon proliferative and anti-proliferative stimulus
FGF2 treatment induces cell cycle arrest at G2/M and S-phase progression delay in a ras- driven mouse tumor cell (3, 4); however, little is known about how these antiproliferative effects are established as well as the molecules directly involved in these effects. To shed light on these issues, we analyzed early proteomic alterations differentially induced by FGF2. To this end, Y1 cells were harvested at G0/G1, and their whole cell extracts (WCEs) were obtained from cells cultivated in DMEM medium plus serum, supplemented or not with physiological levels of FGF2 were obtained after 0, 3, and 5 h of stimulation. We assume that global proteomic changes would initiate at 3 - 5 h upon stimulus based on previous results reviewed and discussed at Rocha et al [23]. As antiproliferative effects are observed both by stimulation by FGF2 alone and in addition to serum [3] (Figure S1A), we believed that a more reliable comparison to the proliferative effects of the serum would be to compare it with serum supplemented with FGF2, as serum contains different growth factors. Proteins were trypsinized and analyzed by MudPIT (Figure S1A). After filtering, on average,
Accepted Article
over 2000 proteins (Figure S1B) were identified for each time point and quantified with reproducible quantitative values (dNSAFs) (Figure S1C), totaling non-redundant 2966 proteins (Table S1).
Hierarchical clustering of the quantitative values clearly separated clusters of proteins that were only expressed at a particular time point (Figure S1D). To assess proteome dynamics, protein abundance changes were also evaluated by c-fuzzy means clustering 24, allowing us to clearly identify different and interesting trends of protein dynamics upon growth factor stimulation (Figure 1). To gain insight into the biological functions among clusters, we grouped them into four, according to their protein abundance dynamics, as follows: Group A: clusters where the FGF2 stimulus did not change the protein abundance (clusters 2 and 5); ii. Group B: clusters whose protein abundance was changed by FGF2 but not by serum stimulation (clusters 8 and 9); iii. Group C: clusters whose abundance pattern was changed similarly by serum and FGF2 stimulation (cluster 4 and 7); and iv. Group D: clusters that contained proteins whose expression occurred earlier or later upon FGF2 stimulation (cluster 1, 3 and 6).
The top ten enriched Reactome pathways from each group are shown in Table S2. Group A was enriched in two main pathways, namely, the “rRNA processing pathway” and the “mRNA splicing pathway”; group B, in “mRNA splicing” and “protein folding”; group C, in “S-phase”, “rRNA processing pathway”, and “metabolism of amino acids and derivatives”; and group D, in “metabolism of amino acids and derivatives”, “translation” and “S-phase”. Although some pathways coincided in more than one group, a detailed analysis of each pathway indicated some interesting specificities from each group (Figure S2). For instance, comparing the S-phase diagram of group C and D it was possible to observe that a key difference between them concerned pathways involving cdk2 at S-phase entrance, ubiquitin-dependent degradation of cyclin D and the primase component of DNA polymerase (Figure S2A). These pathways appeared to be enriched, particularly in group D, suggesting important pathways where FGF2 stimulation may be either advanced or delayed regarding serum stimulation and that are directly associated with S-phase entrance and progression. Significant differences were also observed in “mRNA splicing” between groups A and B (Figure S2B).
Group D was particularly interesting, as it contained proteins whose expression changes either occurred earlier or were delayed upon FGF2 stimulation. Particularly, cluster 3 contained proteins whose abundance increased at 3 h with serum stimulation (compared with G0/G1 samples), but only after 5 h with FGF2 stimulation. GO analysis indicated that these proteins were enriched in terms associated with DNA replication, corroborating the phenotypic findings, namely, S-phase
Accepted Article
progression delay, observed in our model 4). In addition to DNA replication, terms associated with proteasome, splicing, DNA repair, and metabolism were also increased among proteins from cluster 3. Cluster 1 described the opposite-direction fitting proteins whose abundance was decreased first upon 3 h of serum stimulation (compared with G0/G1 samples) and, second, upon FGF2 stimulation. Only after 5 h of stimulation did this group of proteins reach the abundance previously detected in the 3 h serum-stimulated samples. In agreement with FGF2 morphological alterations, GO analysis indicated that regulation of the actin cytoskeleton was enriched at cluster 1. Although the morphological changes were only observed after at least 6-8 hours of FGF2 stimulation (data not shown), our data indicated that expression changes in the proteins associated with both replication and cytoskeleton had already occurred. Cluster 6 is also interesting because it contained proteins whose expression first changed upon FGF2 stimulation. GO analysis indicated that many proteins associated with metabolism (e.g., TCA cycle, glutathione metabolism) were enriched, suggesting that FGF2 stimulation accelerates metabolic changes in the Y1 cell line. Whether this dynamic change plays a role in FGF2 cytotoxicity needs to be further evaluated.
Terms associated with DNA replication, mRNA processing, and metabolism are enriched among differentially expressed proteins
Several differentially expressed proteins were statistically identified by PLGEM (Power Law Global Error Model) (21) in pairwise analyses of samples stimulated with serum and serum plus FGF2. Of these, 261 proteins were identified as differentially expressed upon 3 h of stimulation; 243 proteins, after 5 h of stimulation. (Figure 2A and Table S3). Eight proteins were identified as up- regulated after both 3 h and 5 h of FGF2 stimulation. One of them was Mycbp (c-myc binding protein), which is known for its ability to bind to the N-terminus of c-myc, activating the transcription of E-box genes [25]. Interestingly, it has been shown previously that FGF2 stimulation induces c-myc stabilization via ERK signaling (26). Ten proteins were identified as down-regulated after both 3 h and 5 h of FGF2 stimulation. Importantly, two of them are associated with the cell cycle progression: mitotic spindle organizing protein 2 (mzt2) and thymidine kinase 1 (tk1). The latter catalyzes the addition of a gamma-phosphate group to thymidine, which is essential for the biosynthesis of dNTPs and fundamental to the G1→S transition [27].
To obtain an overview of protein functional pathways induced by serum or FGF2 stimulation, DAVID (http: //david.abcc.ncifcrf.gov) functional analysis was performed (Figure 2B). We filtered the terms that were overrepresented in both the up- and down-regulated lists defined by PLGEM (such as those associated with mitochondria, ribosome, and translation). Therefore, terms associated with
Accepted Article
“cell division”, “regulation of cell cycle”, “nucleosome assembly”, “mRNA splicing”, “mRNA processing”, and “RNA splicing” were overrepresented among the down-regulated proteins, while terms associated with “spliceosomal snRNP assembly”, “protein folding”, “cytokinesis”, “vesicle- mediated transport”, “glutathione metabolic process”, “smalIGTPase mediated signal transduction”, and “oxidation-reduction process” were overrepresented among the up-regulated proteins.
Terms associated with DNA replication were particularly identified as down-regulated after 3 h of FGF2 treatment. Ten differentially expressed proteins associated with the term “DNA Replication” (GO: 0006260) from Gene Ontology were identified, including gins2, tk1, nfic, rfc2, r2, fen1, and p21.
Protein-protein interaction and transcription network analyses
The interactome analysis could provide hints on structural changes in biological pathways due to proliferative or anti-proliferative signals induced by serum or serum supplemented with FGF2, respectively. Thus, the exclusively and differentially expressed proteins were subjected to protein- protein interaction (PPI) and network analyses using the STRING database and Cytoscape software (Figure S3) (Table S4). Considering only experimental evidence for PPI and after filtering proteins with some node connectivity criteria (e.g., minimum node degree and betweenness centrality), networks composed of 150 proteins and 495 connections were mapped. In accordance with mitogenic stimulation induced by serum, the cyclin dependent kinase 2 (cdk-2), which is essential for the G1→S transition, was shown as an important hub in serum interactome (blue nodes and edges).
The protein network of antiproliferative stimulation induced by FGF2 identified src, ppie, and supt5h as important hubs (orange nodes and edges). Src is a non-receptor protein tyrosine kinase that is involved in many different pathways, ranging from the immune response to the cell cycle (for a recent review, see (28]). It was previously shown that FGF2 antiproliferative effects are dependent on src signaling [3]. The fact that our proteomics analyses detected src as an important hub in FGF2- stimulated samples reinforces the quality of our data and the importance of this type of analysis to identify key players in the anti-mitogenic response. Ppie, or peptidylprolyl isomerase E, has protein- folding activities and exhibits RNA-binding activity. Supt5h is a DSIF Elongation Factor Subunit that regulates mRNA processing and transcription elongation by RNA polymerase II. The interactome of the 3 h FGF2 stimulation also identified hubs involved in cytoskeleton remodeling, such as actb, in agreement with the profound changes in cell morphology of Y1 cell after FGF2 stimulation.
Transcription factors (TFs) play key roles in cell controlling gene expression, leading to proteomic changes that, ultimately, induce phenotypic alterations. From the differentially expressed
Accepted Article
proteins detected in our proteomic data, we found 32 proteins classified as TFs (Table S3) according to Gene Ontology (GO:0001071 - nucleic acid binding transcription factor activity and GO:0000988 - transcription factor activity). To gain further insight into the transcriptional regulatory network differentially induced by FGF2 stimulation, up- and down-expressed proteins were analyzed by iRegulon (29), which identifies the enriched TFs motifs in cis-regulatory sequences of given list, ranking the TFs that may be associated with them. From the 3 h-differentially expressed proteins (261 proteins), iRegulon yielded 8 candidate TFs in which, fosB, gabpp and srf were the top three TFs with 92, 37 and 69 targets, respectively (Figure 3 and Figure S1E). Motifs associated with fosB were also shared by other members of the activator protein 1 (AP-1) family, which is a transcription factor complex involved in different regulatory processes such as cell proliferation, differentiation, and death [30]. Interestingly, of the AP-1 family, only fosB and junB were detected as differentially expressed by proteomics analysis (Table S3). Fos and jun families mediate the signaling that ultimately leads to cell proliferation. Understanding this pathway may elucidate and unravel the mechanism that leads to cell proliferation arrest induced by FGF2. From the network analysis, it is possible to note that fosB may be involved in (positive or negative) regulation of more than 90 differentially expressed proteins. In addition, fosB can also be further regulated by itself and by srf, while junB may also be regulated by srf, gabbp and fosB.
FosB and JunB, two transcription factors, are overexpressed upon FGF2 stimulus
The above analyses indicate that fosB and junB may act as master regulators of FGF2 stimulation, we hence investigated their putative role in FGF2 stimulation. First, we analyzed in more detail their expression after different time points of FGF2 stimulation compared to serum stimulation. G0/G1-arrested Y1 cells were stimulated with serum with or without the addition of FGF2. JunB expression started after 1 h of FGF2 stimulation and peaked after 5 h (Figure 4A) but was expressed at lower levels upon serum stimulation. FGF2 induced the expression of fosB from 3 h to 24 h after stimulation, peaking at 3 h to 8 h. On the other hand, serum stimulation barely induced the expression of this protein (Figure 4A).
We investigated whether fosB and junB would also be overexpressed in other cell lines that are susceptible to FGF2 antiproliferative stimulation 2, such as SK-N-MC and UM-UC-3, a neuroepithelial and a urinary bladder human cell line, respectively. The growth arrest was confirmed by counting the number of cells until 72 h after stimulation (data not shown). Then, both cells were G0/G1 arrested by serum removal and stimulated with either serum or serum plus FGF2 for 0, 1, 3, 5
Accepted Article
and 8 h. As seen in Figure S4A and S4B, FGF2 also causes overexpression of both fosB and junB in these cell lines.
For comparative purposes, we also analyzed the expression of c-fos and c-jun proteins (Figure 4A), which are the canonical members of the fos and jun gene families, respectively; c-fos and c-jun are typical immediate response genes, whose expression inductions are rapid, intense and transient (5, 31). We detected that both serum or serum plus FGF2 treatments rapidly induced c-fos and c-jun expression.
We estimated the dynamics of fosB, junB, c-fos, and c-jun expression upon serum or serum plus FGF2 stimulation based on western blot intensities normalized to the hprt loading control (Figure 4B); c-fos and c-jun expression peaked after 1 h of serum plus FGF2 stimulation (positive Log2(FGF2/serum)), followed by an increase in samples stimulated with serum (negative Log2(FGF2/serum)). However, the levels of both proteins remained higher in FGF2 after 8 h but were considerably reduced in serum-stimulated cells. On the other hand, the expression of fosB and junB was consistently higher at all time points of FGF2 stimulation, while the fosB and junB levels were completely abolished after the serum stimulus. Serum stimulation appeared to induce a rapid and transient expression of c-fos and c-jun, while FGF2 induced a rapid and constant stimulation of both. The permanent expression of these two members of the fos and jun family may be involved in the antiproliferative effect of FGF2, as will be further discussed.
JunB and FosB expression occurs through the FGF receptor and depends on the MEK and SRC signaling pathway
It is known that the antiproliferative effect of FGF2 is dependent on high levels of k-Ras-GTP, src, and rhoA signaling (3, 4). To investigate if these signaling pathways were also involved in the induction of fosB and junB expression, Y1 cells were stimulated with serum or serum plus FGF2 in the presence or absence of specific inhibitors. In all cases below, the inhibitory effect was confirmed either by evaluating the phosphorylation of a downstream protein and/or by evaluating the abrogation of the antiproliferative effect induced by FGF2 (data not shown).
To evaluate if junB and fosB expression occurs through FGFR tyrosine kinase activation, the PD173074 inhibitor, which is considered a pan-inhibitor of receptors of this class, was used. This compound inhibited the antiproliferative and cytotoxic effects induced by FGF2 [3], and in agreement, we observed that PD173074 decreased the activation of ERK1/2 in FGF2-treated
Accepted Article
samples. Blocking by PD173074 also abrogated the expression of most of junB and fosB, indicating that their expression occurs through FGFR tyrosine kinase activation (Figure S4C and Figure S4F).
Src inhibitors promote survival in clonogenic assays of cells stimulated with FGF2, as well as prevent FGF2-induced G2/M arrest, indicating that the src pathway is crucial for the antiproliferative effects of FGF2 4). Thus, we evaluated whether the expression of fosB and junB would depend on src using the PP1 inhibitor, which is a competitive and reversible ATP inhibitor of Src tyrosine kinase family proteins. Both fosB expression and junB expression were affected by the inhibition of src (Figure S4D and Figure S4G); c-jun expression was unaffected, while c-fos expression was partially affected by the PP1 inhibitor. Given the importance of Src protein kinase in FGF2 cytotoxicity and its antiproliferative effects 4), the results suggest that expression of junB and fosB may be a downstream effector of src signaling and could be involved in antiproliferative events.
The antiproliferative response induced by FGF2 is dependent on RhoA-GTP [3]. Cells ectopically expressing a constitutively active RhoA (RhoA-V14) or a dominant-negative RhoA (RhoA- N19) were stimulated with FGF2 or serum, and expression of fosB and junB was evaluated by western blotting (Figure S4E and Figure S4H). However, regardless of the activity of RhoA, neither fosB nor junB were affected, suggesting that their expression is independent of Rho signaling.
FosB knockdown increases DNA replication
To directly evaluate the participation of fosB and junB in the growth arrest induced by FGF2 in Y1 cells, we assessed the effects of the knockdown of these two proteins by RNA interference assays, using both short-interfering RNAs (siRNAs) and short hairpin RNA (shRNA). Both fosB and junB expression were decreased upon exposure to shRNA (Figure 5A and Figure S5A and S5B). To analyze if these proteins were involved with FGF2 growth arrest, we performed clonogenic assays (data not shown) and growth curves. Neither junB nor fosB knockdown rescued the cell growth blockage induced by FGF2. In fact, junB and fosB knockdown cells showed a reduction in cell growth even upon serum stimulation (Figure 5B).
As fosB and junB are considered immediate early genes, we tested if their knockdown effects would be evident earlier than the proliferation arrest, such as during DNA replication. Thus, we performed EdU incorporation in Y1 cells stimulated with serum with or without the addition of FGF2 after 16 h (when most cells are in S-phase) (Figure 5C and 5D). In accordance with previous findings, FGF2 treatment induced a reduction in the percentage of cells in S-phase (Figure 5C: Y1
Accepted Article
cells: 34.8% to 18.8% and Figure S5C: shGFP cells: 32.8% to 9.9%). However, shRNA for fosB followed by FGF2 stimulation increased the number of cells in S-phase (31%), in the same range as the number of cells obtained by serum-stimulation (34.8%) (p-value < 0.05). No major differences were found for junB knockdown (Figure S5C). Taken together, these data suggest that fosB expression may be involved in the S-phase progression delay induced by FGF2.
To confirm that fosB knockdown increases the number of cells undergoing DNA replication, siRNAs for fosB were used to transfect Y1 cells (Figure S5D). DNA replication was evaluated by EdU incorporation and visualized by fluorescence microscopy. EdU-labeled cells were counted after 16 hours of serum or serum plus FGF2 stimulation (Figure S5E). In agreement with shRNA assays, siRNA assays also showed that fosB knockdown promotes an increased number of cells in S-phase, suggesting that fosB overexpression may be involved with S-phase cytotoxicity induced by FGF2.
Discussion
Growth factors trigger an intricate signaling cascade that culminates in gene expression and, therefore, proteomic alterations, promoting cell cycle progression. The FGF2 response starts from ligation at the tyrosine kinase receptor, inducing many responses by activating classical pathways such as MAPK, PI3K and AKT (9). Deregulation of FGF signaling is associated with many diseases and is a target for the development of many therapeutic strategies in cancer 161. Here, we have identified early-proteomic alterations upon growth factor stimulation comparing proliferative (serum stimulation) and antiproliferative (serum plus FGF2 stimulation) signals in a mouse tumor cell aiming to explore FGF2-antiproliferative effects. By quantitative proteomics analysis, the abundance of more than 2000 proteins was assessed in two early time points upon growth factor stimulation (3 h and 5 h). Regulatory networks indicated transcription factors, such as two members of the AP-1 complex, fosB and junB, as putative master regulators involved in differential FGF2 stimulation; possibly involved in the global proteomics alterations induced by this growth factor.
Although the early detectable effects of FGF2 stimulation (namely, S-phase progression delay and morphological changes) are only observed after 6-8 h of stimulation, we clearly detected many proteomic alterations upon FGF2 stimulation in early time points. Expression of proteins associated with DNA replication is delayed upon FGF2, even though DNA replication starts only after 8 h of stimulation. Important proteins associated with S-phase regulation, such as gins2, nfic, rfc2,
Accepted Article
tk1, were down-regulated upon FGF2 stimulation. The Gins2 protein belongs to the GINS complex, whose formation is essential for the initiation of DNA replication [32]. Rfc2 (replication factor C) participates, along with proliferating cell nuclear antigen (Pcna) and other accessory proteins, in the DNA elongation process during replication (33). Importantly, cdk2, which is involved in the G1→S transition, appears as an important hub in the interactome network analysis. In agreement with S- phase progression delay, this protein appears mainly upon a serum stimulus. Cdks activation occurs mainly by phosphorylation at specific residues; therefore, it is possible that, due to our quantitative proteomics approach, highly modified peptides were inaccurately quantified, leading to the very low amounts of cdk2 in FGF2-stimulated samples. Taken together, FGF2 stimulation changes the abundance of important proteins that are involved in DNA replication and, thereby, may participate in the S-phase progression delay induced by FGF2. It remains to be studied whether these expression changes are a cause or a consequence of the S-phase progression delay induced by FGF2.
Quantitative trends of proteomic abundance show four different patterns. The cluster of proteins whose abundance is changed first upon serum stimulation and then upon FGF2 stimulation (cluster 3) is enriched in terms associated with DNA replication, proteasome, splicing, and DNA damage (Figure 1). Another important observation is that serum stimulation and FGF2 stimulation change the abundances and dynamics of metabolic proteins in a completely different manner, as seen in clusters 1 and 6. While terms associated mainly with carbohydrate metabolism are enriched in cluster 1, terms associated with lipid and amino acid metabolism are enriched in cluster 6. In this regard, it is interesting that Gsk3b (Glycogen synthase kinase 3 beta), a protein that acts as a negative regulator in the hormonal control of glucose metabolism, appears as an important hub in the interactome analysis as exclusively expressed upon serum stimulation. This protein is involved in a surprisingly large number of cellular processes and diseases and is also able to phosphorylate transcription factors (54). In chondrocytes, this protein has been shown to play an essential role in FGF1 signaling 55]. Taken together, these results suggest that FGF2 stimulation can change not only the abundance of proteins associated with DNA replication but also the abundance and dynamics of proteins associated with cell metabolism.
We have used the iRegulon framework (29), which has been validated and used to predict important TFs in different models (29, 36). iRegulon analysis clearly indicates that members of AP-1 (mainly fosB), as important master regulators of FGF2 signaling, are associated with the direct regulation of 35% of differentially expressed proteins. It was previously shown that ACTH and FGF2
Accepted Article
can induce the expression of fos and jun families in the Y1 cell line, in primary cultures (37) and in vivo infused adrenal glands (38). However, no detailed analysis of their expression, their associated signaling cascades or their association with DNA replication arrest has been investigated before.
Herein, we showed that fosB and junB expression occurs through FGFR and depends on src activation. The src protein appears as the second largest hub in the interactome network, suggesting that it indeed may play an important role in FGF2 signaling. In addition, many members of the src family appear exclusively upon 3 h of FGF2 stimulation, as shown by hierarchical clustering analysis (Figure S1D). Src is a member of a family of tyrosine kinase-like proteins that are involved in the growth and differentiation of eukaryotic cells. Src is usually in an inactive state but can be transiently activated by growth factors, including by FGF2 [15, 39]. Of note, the src pathway has been shown to play important roles in FGFR trafficking 40. Thus, the fact that we found that the inhibition of src signaling by PP1 inhibitors also abrogates the overexpression of junB and fosB is very interesting and indicates that junB and fosB may be indeed master regulators of FGF2 antiproliferative effects.
FosB knockdown (both by siRNA and shRNA) increased the number of cells under DNA replication, suggesting that fosB may be involved in the FGF2 S-phase progression delay. There is evidence that FGF2 stimulation results in replicative stress in the Y1 cell line 41. Therefore, it is possible that the overexpression of fosB may play a role in this replicative stress by overstimulating the expression of genes involved in G1→S transition and DNA replication. In fact, it was shown that AP-1 dimers bind directly to cyclin D1 promoters, thereby activating cyclin D1 and promoting G1→S transition (42). Thus, lowering the levels of fosB (by knockdown) would avoid replicative stress and therefore promote DNA replication. Nevertheless, knockdown assays for both fosB and junB failed to recover the cell proliferation induced by FGF2, indicating that other downstream players may be involved in proliferation arrest.
It is well known that the nucleotide pool is essential to proper DNA replication (43) and that an alteration in its levels may induce replicative stress. Therefore, the levels of enzymes involved in dNTP synthesis are highly regulated as cells enter the S phase 1441. Strikingly, two enzymes involved in dNTP biosynthesis (Tk1 and Rrm2) were detected as differentially expressed in our proteomics analysis. Thymidine kinase 1 (tk1), which is down-regulated upon FGF2 stimulation, is an enzyme essential for the biosynthesis of dNTPs once it catalyzes the addition of a gamma-phosphate group to thymidine and is fundamental to the G1→S transition [27]. RRM2, a ribonucleotide reductase involved in the formation of deoxyribonucleotides from ribonucleotides, is up-regulated upon FGF2
stimulation. Buisson et al (45) observed that an increase in RRM2 counteracts DNA damage and replicative stress. Whether the increase in RRM2 abundance is a mechanism to circumvent DNA replicative stress induced by FGF2 may be further evaluated. Nevertheless, our data suggest that FGF2 stimulation may change the nucleotide pool (by altering these proteins levels) and, therefore, could promote replicative stress.
FGF2 is a member of the family of heparin binding growth factors, which bind with high affinity to heparan sulfate proteoglycans (HSPG) on cells and within the extracellular matrix 1461. It was shown that this binding may impact on signaling response of FGF2 possibly tuning cellular response toward growth, migration or differentiation 4. Whether the FGF2 antiproliferative effects observed here also depend on different HSPGs types (and amounts) need to be further evaluated.
Finally, our proteomic dataset together with an in-depth analysis of expression dynamics, identification of regulons and validation of some signaling cascades related to junB and fosB expression, allowed us to conclude that the antiproliferative FGF2 stimulation induces proteomic alterations in a broad group of functional proteins involved mainly in metabolism, DNA replication and splicing. Moreover, fosB overexpression may be involved in S-phase progression delay, suggesting that this TF may be an important regulator of antiproliferative signals induced by FGF2.
Accepted Article
Author Contributions Section
Investigation, F.N.L.V., F.M., J.N.M., M.C.L., B.C., C.S. F. and J.M.G .;
Conceptualization, F.N.L.V., M.S.R., H.A.A. and J.P.C.C .;
Validation, F.N.L.V. and J.P.C.C .;
Formal Analysis, F.N.L.V., J.P.C.C., J.M.G. and M.S .;
Writing - Original Draft, F.N.L.V. and J.P.C.C .;
Software, B.F.S. and M.S.R .;
Supervision and Resources, H.A.A., J.P.C.C., L.A.F. and M.P. W .;
Project Administration, J.P.C.C.
Funding acquisition, M.P.W, H.A.A. and J.P.C.C.
Data curation, M.I.S, J.M.G, L.A.F, M.S.R. and B.F.S.
20
Accepted Article
Acknowledgments
The authors are grateful to Ismael Feitosa Lima, Ivan Novaski Avino, and Mariana Morone for technical assistance. We thank Dr. Fabio Forti for providing Y1-RhoA lineages and Dr. Matheus Dias, Dr. M. Carolina Elias and Dr. Vincent Noel for important discussions and for reading the manuscript. This work was supported by fellowships from CNPq and by grants 22619-7 and 07467-1 from the São Paulo Research Foundation (FAPESP). JMG, MES, LF, and MPW were supported by the Stowers Institute for Medical Research. FACs analysis was performed using a flow cytometer acquired with FAPESP grant 2015-10037-4.
The authors declare no conflicts of interest.
References
[1] H. Wang, M. Rubin, E. Fenig, A. DeBlasio, J. Mendelsohn, J. Yahalom, R. Wieder, Cancer research 1997, 57, 1750.
[2] V. A. Smits, M. A. van Peer, M. A. Essers, R. Klompmaker, G. Rijksen, R. H. Medema, The Journal of biological chemistry 2000, 275, 19375.
[3] E. T. Costa, F. L. Forti, T. G. Matos, A. Dermargos, F. Nakano, J. Salotti, K. M. Rocha, P. F. Asprino, C. K. Yoshihara, M. M. Koga, H. A. Armelin, Cancer Res 2008, 68, 6215.
[4] J. Salotti, M. H. Dias, M. M. Koga, H. A. Armelin, PLoS One 2013, 8, e72582.
[5] T. Fowler, R. Sen, A. L. Roy, Molecular cell 2011, 44, 348.
[6] L. Bakiri, D. Lallemand, E. Bossy-Wetzel, M. Yaniv, Embo j 2000, 19, 2056; E. Passegue, E. F. Wagner, Embo j 2000, 19, 2969.
[7] E. J. Nestler, M. Barrot, D. W. Self, Proceedings of the National Academy of Sciences of the United States of America 2001, 98, 11042.
[8] M. C. Gruda, K. Kovary, R. Metz, R. Bravo, Oncogene 1994, 9, 2537.
[9] D. M. Ornitz, N. Itoh, Wiley interdisciplinary reviews. Developmental biology 2015, 4, 215.
[10] H. A. Armelin, Proceedings of the National Academy of Sciences of the United States of America 1973, 70, 2702.
[11] R. H. Xu, R. M. Peck, D. S. Li, X. Feng, T. Ludwig, J. A. Thomson, Nature methods 2005, 2, 185.
Accepted Article
[12] S. W. McLeskey, I. Y. Ding, M. E. Lippman, F. G. Kern, Cancer research 1994, 54, 523.
[13] M. R. Johnson, C. Valentine, C. Basilico, A. Mansukhani, Oncogene 1998, 16, 2647.
[14] H. Zhao, D. L. Cunningham, A. J. Creese, J. K. Heath, H. J. Cooper, Journal of proteome research 2015, 14, 5077; A. M. Hinsby, J. V. Olsen, M. Mann, The Journal of biological chemistry 2004, 279, 46438.
[15] A. Vercoutter-Edouart, J. Lemoine, C. E. Smart, V. Nurcombe, B. Boilly, J. Peyrat, H. Hondermarck, FEBS letters 2000, 478, 209.
[16] I. S. Babina, N. C. Turner, Nature reviews. Cancer 2017, 17, 318.
[17] Y. Yasumura, V. Buonassisi, G. Sato, Cancer research 1966, 26, 529.
[18] Y. Zhang, Z. Wen, M. P. Washburn, L. Florens, Analytical chemistry 2011, 83, 9344.
[19] D. L. Tabb, W. H. McDonald, J. R. Yates, 3rd, Journal of proteome research 2002, 1, 21.
[20] Y. Zhang, Z. Wen, M. P. Washburn, L. Florens, Analytical chemistry 2010, 82, 2272.
[21] N. Pavelka, M. L. Fournier, S. K. Swanson, M. Pelizzola, P. Ricciardi-Castagnoli, L. Florens, M. P. Washburn, Molecular & cellular proteomics : MCP 2008, 7, 631.
[22] M. W. Pfaffl, Nucleic acids research 2001, 29, e45.
[23] K. M. Rocha, F. L. Forti, A. P. Lepique, H. A. Armelin, Microsc Res Tech 2003, 61, 268.
[24] V. Schwammle, O. N. Jensen, Bioinformatics (Oxford, England) 2010, 26, 2841.
[25] T. Taira, J. Maeda, T. Onishi, H. Kitaura, S. Yoshida, H. Kato, M. Ikeda, K. Tamai, S. M. Iguchi-Ariga, H. Ariga, Genes to cells : devoted to molecular & cellular mechanisms 1998, 3, 549.
[26] A. P. Lepique, M. S. Moraes, K. M. Rocha, C. B. Eichler, G. N. Hajj, T. T. Schwindt, H. A. Armelin, J Mol Endocrinol 2004, 33, 623.
[27] P. Reichard, Annual review of biochemistry 1988, 57, 349.
[28] J. Espada, J. Martin-Perez, International review of cell and molecular biology 2017, 331, 83.
[29] R. Janky, A. Verfaillie, H. Imrichova, B. Van de Sande, L. Standaert, V. Christiaens, G.
Hulselmans, K. Herten, M. Naval Sanchez, D. Potier, D. Svetlichnyy, Z. Kalender Atak, M. Fiers, J. C. Marine, S. Aerts, PLoS computational biology 2014, 10, e1003731.
[30] R. Eferl, E. F. Wagner, Nat Rev Cancer 2003, 3, 859.
[31] S. Bahrami, F. Drablos, Advances in biological regulation 2016, 62, 37.
[32] M. Ueno, M. Itoh, L. Kong, K. Sugihara, M. Asano, N. Takakura, Molecular and cellular biology 2005, 25, 10528.
[33] U. Hubscher, Methods in molecular biology (Clifton, N.J.) 2009, 521, 19.
[34] A. Ali, K. P. Hoeflich, J. R. Woodgett, Chemical reviews 2001, 101, 2527.
Accepted Article
[35] J. R. Chapman, O. Katsara, R. Ruoff, D. Morgenstern, S. Nayak, C. Basilico, B. Ueberheide, V. Kolupaeva, Molecular & cellular proteomics : MCP 2017, 16, 1126.
[36] A. Qiao, Z. Zhao, H. Zhang, Z. Sun, X. Cui, Pathology, research and practice 2017, 213, 548; M. Menotta, S. Biagiotti, S. Orazi, L. Rossi, L. Chessa, V. Leuzzi, D. D’Agnano, A. Plebani, A.
Soresina, M. Magnani, Molecular and cellular biochemistry 2018, 438, 153; L. Li, G. Wang, N. Li, H. Yu, J. Si, J. Wang, Experimental and therapeutic medicine 2017, 14, 1065; H. Jung, S. Han, S. Kim, Genomics & informatics 2015, 13, 76; R. Janky, M. M. Binda, J. Allemeersch, A. Van den Broeck, O. Govaere, J. V. Swinnen, T. Roskams, S. Aerts, B. Topal, BMC cancer 2016, 16, 632.
[37] E. Kimura, M. H. Sonobe, M. C. Armelin, H. A. Armelin, Molecular endocrinology (Baltimore, Md.) 1993, 7, 1463; C. F. Lotfi, H. A. Armelin, J Endocrinol 2001, 168, 381.
[38] R. B. Baccaro, P. O. Mendonca, T. E. Torres, C. F. Lotfi, Cell and tissue research 2007, 328, 7.
[39] V. M. Ding, P. J. Boersema, L. Y. Foong, C. Preisinger, G. Koh, S. Natarajan, D. Y. Lee, J. Boekhorst, B. Snel, S. Lemeer, A. J. Heck, A. Choo, PloS one 2011, 6, e17538.
[40] E. Sandilands, S. Akbarzadeh, A. Vecchione, D. G. McEwan, M. C. Frame, J. K. Heath, EMBO reports 2007, 8, 1162.
[41] M. H. Dias, C. S. Fonseca, J. D. Zeidler, L. L. Albuquerque, M. S. da Silva, E. Cararo-Lopes, M. S. Reis, V. Noel, I. A. Prior, H. A. Armelin, bioRxiv 2017.
[42] J. R. Brown, E. Nigh, R. J. Lee, H. Ye, M. A. Thompson, F. Saudou, R. G. Pestell, M. E. Greenberg, Molecular and cellular biology 1998, 18, 5609.
[43] M. Anglana, F. Apiou, A. Bensimon, M. Debatisse, Cell 2003, 114, 385.
[44] P. Nordlund, P. Reichard, Annual review of biochemistry 2006, 75, 681; A. Hofer, M. Crona, D. T. Logan, B. M. Sjoberg, Critical reviews in biochemistry and molecular biology 2012, 47, 50.
[45] R. Buisson, J. L. Boisvert, C. H. Benes, L. Zou, Molecular cell 2015, 59, 1011.
[46] A. Yayon, M. Klagsbrun, J. D. Esko, P. Leder, D. M. Ornitz, Cell 1991, 64, 841; A. C. Rapraeger, S. Guimond, A. Krufka, B. B. Olwin, Methods in enzymology 1994, 245, 219.
[47] K. Forsten-Williams, C. C. Chua, M. A. Nugent, Journal of theoretical biology 2005, 233, 483.
Figure Legends
Figure 1. Trends of protein expression upon G0/G1, serum and serum plus FGF2 stimulation. Fuzzy c-means clustering analysis highlighted nine important clusters (subdivided into 4 groups, A-D) enriched in the indicated REACTOME pathway (right). Clusters from group D were individually analyzed by KEGG pathways as indicated below each cluster. Average dNSAF values for each protein were employed to generate clusters using http://computproteomics.bmb.sdu.dk/Apps/FuzzyClust/.
Accepted Article
Accepted Article
Cluster 2
Cluster 5
1.5-
Group A
expression changes
0.0
-1.5-
GO/G1
3h
5h
GO/G1
3h
5h
- mRNA Splicing
- rRNA processing
Cluster 8
Cluster 9
1.5-
Group B
expression changes
0.0
treatment
FGF2 serum
- mRNA Splicing
- Protein folding
-1.5-
GO/G1
3h
5h
GO/G1
3h
5h
Cluster 4
Cluster 7
1.5-
Group C
expression changes
- S-phase
- rRNA processing pathway
0.0
- Metabolism of amino acids and derivatives
-1.5-
GO/G1
3h
5h
GO/G1
3h
5h
Cluster 1
Cluster 3
Cluster 6
1.5
Group D
expression changes
- Metabolism of amino acids and derivatives
0.0
- Translation
- S-phase
-1.5-
GO/G1
3h
5h
GO/G1
3h
5h
GO/G1
3h
5h
mmu03050:Proteasome
mmu03050:Proteasome
mmu01210:2-Oxocarboxylic acid metabolism
mmu00020:Citrate cycle (TCA cycle)
mmu03040:Spliceosome
mmu00020:Citrate cycle (TCA cycle)
mmu00620:Pyruvate metabolism
mmu03030:DNA replication
mmu00010:Glycolysis /
mmu03015:mRNA surveillance pathway
mmu00630:Glyoxylate and dicarboxylate metabolism
Gluconeogenesis
mmu04810:Regulation of actin cytoskeleton
mmu03420:Nucleotide excision repair
mmu00480:Glutathione metabolism mmu00620:Pyruvate metabolism
mmu01212:Fatty acid metabolism
mmu03430:Mismatch repair
mmu00350: Tyrosine metabolism
mmu00520:Amino sugar and nucleotide sugar metabolism
mmu00071:Fatty acid degradation
mmu00360:Phenylalanine metabolism
mmu00900:Terpenoid backbone biosynthesis
mmu00062:Fatty acid elongation
A.
47.
Log, dNSAF 5h Serum
?
-10
-10
-15
-15
8
&
20
-15
-10
-5
20
-15
-10
-5
Log2 dNSAF 3h FGF2
Log2 dNSAF 5h FGF2
B. Log2 dNSAF 3h Serum GO terms_Biological Processes Accepted Article Ran
cell division
regulation of cell cycle
Down
nucleossome assembly
expressed
mRNA splicing, via spliceosome
proteins
mRNA processing
RNA splicing
spliceosomal snRNP assembly
protein folding
cytokinesis
Up
vesicle-mediated transport
expressed proteins
glutathione metabolic process
small GTPase mediated signal transduction
oxidation-reduction process
0
2
4
6
8
10
12
14
16
-log (p-value)
17 up 16 down
3 up 10 down
Fach
Trait
Srt
10 up 5 down
18 up
Tapo 18 down
Bet
Laha
2up 4 down
3 up
4 down
Down regulated
Up regulated
Regulon of the list
Regulon on the list
Fosb regulation
Gabpa regulation
Snf regulation
8up
A.
1h
3h
5h
8h
12h
24h
+
+
+
+
+
+
+
+
+
+
+
+ Serum
+
+
+
+
+
+
FGF2
- 40 kDa
JunB
-40 kDa
c-Jun
Hprt
-25 kDa
1h
3h
5h
8h
12h
24h
+
+
+
+
+
+
+
+
+
+
+
+
Serum
+
+
+
+
+
+
FGF2
FosB
- 40 kDa
c-Fos
-70 kDa
Hprt
- 25 kDa
B.
3.5
log2(FGF2/serum)
2.5
1.5
0.5
-0.5
4
9
14
19
24
hours
-1.5
-2.5
FosB
c-Fos
JunB ☒ c-Jun
Accepted Article
Accepted Article
A.
shFosB
shGFP
Y1
shJunB
shGFP
Y1
+
+
+
+
+
+
Serum
Serum
+
+
+ FGF2
+
+
+
-
+
+
+
+
+
FGF2
FosB
- 40 kDa
JunB
- 40 kDa
- 25 kDa
-25 kDa
Hprt
Hprt
C.
shFosB
Y1
10°
B.
2.5
S - phase cells
x 106 cells per mL
Y1+Serum
& shJunB+Serum
S - phase cells
shJunB+S+FGF2
105
29.0
34.8
shGFP+Serum
0- shFosB+Serum
Serum
1.5
+- shFosB+S+FGF2
RL1-H
104
shGFP+Serum
30-0-00-003-COM
shJunB and
shGFP+S+FGF2
0.5
shFosB
-o Y1+Serum
10ª
Y1+S+FGF2
1
2
3
102
Days
20k
40k
60k
80k
100k
Ok 40k
60k
80k
100k
BLH-3
BLH-3
10º
D.
2.0
*
105
S - phase cells
S - phase cells 18.8
31.0
Relative units (SF/S)
Serum
1.5
+
02
10€
FGF2
1.0
103
0.5
102
20k
40k
60k
80k
100k
20k
k 40k 60k
80k
100k
BLH-3
BLH-3
0.0
Propidium lodide
shFosB
shGFP
no Vector
Manuscript length: 33 pages and 5 figures; Supporting material: 5 figures and 4 tables.