Proteomic and secretomic comparison of young and aged dermal fibroblasts highlights cytoskeleton as a key component during aging

Fibroblast culture

Dermal fibroblasts were obtained from healthy sun-protected skin of either young (<35 years-old) or aged (>55 years-old) healthy women undergoing breast reduction mammoplasty in our Department of Plastic Surgery, Saint-Louis Hospital. Dermal fibroblasts were cultured as previously described [18] in Roswell Park Memorial Institute 1640 (RPMI) supplemented with 10% foetal bovine serum (FBS), 1% penicillin/streptomycin, 2 mM L-Glutamine, 25 μg/mL amphotericin B and 5 μg/mL of plasmocin (complete RPMI medium) and were used at the 3^rd passage to avoid any exhaustion-replicative senescence [11, 28].

Sample preparation for secretome and proteomic assay

Fibroblasts were either stimulated or not stimulated with TGF-β₁ for 24h. Then supernatants were centrifuged (400g, 8 min, 4°C) to remove pellets, before addition of protease inhibitors (Sigma-Aldrich, USA; P8340). Adherent fibroblasts were washed, scraped with a spatula, and centrifuged (400g, 8 min, 4°C). Pellets were lysed with 80 μL RIPA buffer (Rockland, USA; MB-030-0050) including protease inhibitors (Sigma-Aldrich; P8340). Protein measurement was performed by Bradford assay (Bio-Rad, USA; #500-0006) and cell lysates were kept at -80°C until they underwent direct digestion by trypsin before analysis at Jacques Monod Institute (JMI, Paris, France).

Sample preparation prior to LC-MS/MS analysis

50 μg of proteins were precipitated with acetone overnight at -20°C. Samples were centrifuged at 11,000 rpm and 4°C, the pellets were resuspended with 100 μL of NH₄HCO₃ 25mM, then heated at 95°C for 10 minutes. The samples were digested overnight at 37°C with 0.4μg of trypsin (Promega, Madison, WI, USA) in a 25 mM NH₄HCO₃ buffer per sample. Then peptides were desalted using ZipTip μ-C18 Pipette Tips (Pierce Biotechnology, Rockford, IL, USA).

LC-MS/MS acquisition

MS grade Acetonitrile (ACN), MS grade H2O and MS grade formic acid (FA) were obtained from Thermo Chemical (Waltham, MA, USA). Samples were analyzed by a Q-Exactive Plus coupled to a Nano-LC Proxeon 1000 (Thermo Fisher Scientific, Waltham, MA, USA).

Peptides were loaded with an online preconcentration method and separated by chromatography using a Pepmap-RSLC C18 column (0.75 x 500 mm, 2 μm, 100 Å) from Thermo Fisher Scientific, equilibrated at 50°C and operating at a flow rate of 300 nl/min. Peptides were eluted by a gradient of solvent A (H₂O, 0.1 % FA) and solvent B (ACN, 0.1% FA), the column was first equilibrated for 5 min with 95 % of A, then B was raised to 35% in 98 min. Finally, the column was washed with 80% B for 20 min and re-equilibrated at 95% A giving a total run time of 120 min. Peptide masses were analyzed in the Orbitrap cell in full ion scan mode, at a resolution of 70,000, a mass range of m/z 375-1500 and an AGC target of 3×10⁶. MS/MS were performed in the top 20 mode. Peptides were selected for fragmentation by Higher-energy C-trap Dissociation (HCD) with a Normalized Collisional Energy of 27% and a dynamic exclusion of 60 seconds. Fragment masses were measured in the Orbitrap, at a resolution of 17,500, with an AGC target of 2×10⁵, an isolation window of 1.4 Da and a mass range of m/z 200-2000. Monocharged peptides and unassigned charge states were excluded from the MS/MS acquisition. The maximum ion accumulation times were set to 50 ms for MS and 45 ms for MS/MS acquisitions respectively.

Data analysis

Label-free quantification was done on Progenesis QI for Proteomics (Waters, Milford, MA, USA). Between runs, alignment of peptide features and normalization based on all peptide ions were both performed by the Progenesis software. An MGF peak file exported from Progenesis was processed on Proteome Discoverer 2.1 with the mascot node (Mascot version 2.5.1). MS/MS spectra were searched against the Swissprot protein database release 2017_04 with the Homo sapiens taxonomy and a maximum of 2 missed cleavages. Precursor and fragment mass tolerances were set to 10 ppm and 0.02 Da respectively. The following post-translational modifications were included as variables: Oxidation (M), acetyl (N-term), phosphor (STY). Spectra were filtered using a 1% FDR with the percolator node. Multivariate statistics on protein measurements were performed using the Progenesis software and protein abundance was inferred using the top 3 method. A two-group comparison parametric t-test was used to determine differential proteins between the following groups: quiescent young versus quiescent old and TGFβ-stimulated young versus TGFβ-stimulated old cells. Proteins expressed in significant different ways were determined with fold>2, p-value<0.05. A p-value better than 0.05 was used to filter differential significant candidates.

The expression-based heat maps were done using Heatmapper website, following these characteristics: scale type (select direction to scale values) was “row” type, clustering methods (group the data by similar expression levels) was “average Linkage”, and the distance measurement method was “Eucledian”. The schematic molecular pathways were done using the “PathVisio” Software^®, after GOrilla and Panthere website analysis.

Molecular biology

RNA extraction, reverse transcription, and quantitative PCR were performed as previously described [17] and primer sequences were purchased from Eurogentec (Seraing, Belgium).

Western blot (WB)

Fibroblasts were lysed in Tris-Triton lysis buffer supplemented with protease inhibitor cocktail (Sigma-Aldrich; P8340) and centrifuged (20800g, 15 min, 4°C) before protein concentration determination by Bradford assay. Migration of 50 μg of denatured proteins was performed in 10% Bolt Bis-Tris Plus polyacrylamide gel containing sodium dodecyl sulfate, followed by transfer to a nitrocellulose membrane using horizontal gel transfer device iBlot™ 2. Membranes were blocked with TBS1X-5% nonfat milk-0.1% Tween for 1h and incubated overnight at 4°C with specific primary antibodies (Abcam, UK): CORO1C (ab15719, 1/500); CFL1 (ab11062, 1/10000); ARP2/3 (ab49671, 1/2200); FLNB (ab97457, 1/500); ACTC1 (66125-lg, 1/5000). Secondary antibodies (anti-mouse, Bio-Rad 70-65k, 1/2000 or anti-rabbit, ECL-NA1934VS, 1/1000) coupled to horseradish peroxidase were incubated for 1h. Signal was detected using an enhanced chemiluminescence detection kit (Bio-Rad; #170-5061) and ImageQuant chemiluminescent. Blot quantification was performed using ImageJ software.

Immunofluorescence

Immunostaining was performed on fibroblasts seeded on culture chamber slides. Briefly, fibroblasts were seeded at 3500 cells/well on chamber slide, deprived in 0.5% FBS-containing RPMI medium overnight and then activated with 5ng/ml of TGF-β for 24 hours. Fibroblasts were fixed with 4% paraformaldehyde, then permeabilized with Triton X100-0.1%. Staining was performed by overnight incubation with primary antibodies: CORO1C (ab15719, 1/200); CFL1 (SAB2702206, 1/200); FLNB (ab97457, 1/100); ACTC1 (66125-lg, 1/50), followed by an incubation with AF594 secondary antibodies (anti-mouse, A-11001, 1/2000 or anti-goat, A-11058, 1/1000). Imaging was assessed using an Axiovert 200M-fluo microscope and quantification was processed on ImageJ analysis software.

Wound closure by collective migration

The IncuCyte^®S3 Live-Cell Analysis System was used for quantitative label-free collective migration measurements over a 5 day-long lasting culture. Fibroblasts were seeded at 10000 cells per well into 96-well plate to obtain a whole confluency after 24h of culture in complete medium. For siRNA transfection, fibroblasts were transfected with Dharmafect-siRNA mixtures for 5h before cell scratching. For inhibiting the part of cell proliferation during fibroblast migration, mitomycin C (10 μg/ml) was added for 2h before cell scratching. Plates were scratched with a semi-automatized scratch system, creating reproductive regular wounds in each well. Cells were washed to remove cellular debris and complete RPMI medium was added. Plates were placed into the IncuCyte ZOOM^® system for scanning every hour until the scratch wound was completely closed. Captured images were analysed using the integrated software by calculating the cell confluence in the wound.

Collagen gel (lattice) contraction assay

Fibroblasts (100000 cells/mL) were added to a solution containing Type I Collagen (Corning™), 0.006N acetic acid, 0.005N NaOH and incubated in a 6-well plate for 1 h at 37°C to induce gelation. Two lattices were made for each condition. Dermal collagen lattices were detached allowing floatation. OPTIMEM 1% FBS was added at 2X concentration and pictures were taken daily for one week. The lattice area was measured twice by ImageJ.

Transfection assay

Fibroblast siRNA transfection was carried out using Lipofectamine 2000 Transfection Reagent (Life Technologies, USA) according to the manufacturer’s protocol. Briefly, Lipofectamine was diluted in OPTIMEM medium (1/25) as well as the siRNA stock solution (1/50) and incubated for 5 min at RT. Lipofectamine mix was added to the siRNA mix and incubated for 20 min at RT. The resultant mix was then diluted with OPTIMEM to reach a final concentration of 100 nM of siRNA. Cells were transfected with 100 nM of siRNA targeting p16 (Thermo Fisher Scientific) for 24h before RNA and protein extraction. As control siRNA (SiControl), an irrelevant siRNA sequence pool (Thermo Fisher Scientific) was used.

Statistical analysis

Results were expressed as mean ± SEM from independent experiments. Data were symmetrically distributed in a normal distribution. One Way ANOVA Multiple comparisons or Wilcoxon Rank-Sum test were performed using Prism^® (GraphPad 9.1.1) analysis. Differences were considered as significant at *: p<0.05; **: p<0.01; ***: p<0.001.

Secretome study reveals a decrease in protein secretion with age

Senescent cells—and more importantly SASP, the factors they secrete—are widely accepted as drivers of skin aging and multiple age-related cutaneous pathologies. In the present analysis of fibroblast secretome, a total of 945 different proteins between young and old cells was detected in quiescence condition with a significant differential expression of only 16 proteins according to the chosen criteria of selection: fold change>2, p-value<0.05. Under TGF-β₁ addition, 944 different proteins between young and old fibroblasts were identified by HPLC with a significant differential expression of 11 proteins (fold change>2, p-value<0.05). The secretome expression-based heat map (Figure 1A) showed that the differentially-secreted proteins were highly expressed in young fibroblasts compared to those from old ones in either quiescent (13/16) or TGF-β₁-stimulated (8/13) conditions. The role and function of each identified protein was deeply investigated by a global literature review leading to a general secretory pattern associated with skin aging and correlated with cellular pathways (Figure 1B). We underlined that 19% of the proteins differentially-expressed in quiescence condition between young and old fibroblasts are associated with either actin structure, ECM organization, or enzymatic function, whereas other biological processes including proliferation, inflammation, metabolism, or senescence represent less than 8% each. TGF-β₁-treatment induced alterations mainly related to enzymatic function (19%), proliferation/senescence (18%), or actin and ECM organization for 9% each (Figure 1B). A schematic representation of proteins linked to a specific pathway was obtained for quiescence and activation conditions (Figure 1C, 1D, respectively). Noteworthy, the organization of actin and ECM structures appeared to be the major categories linked to the differentially-expressed proteins in both secretome and proteome.

Proteome study revealed a gain in protein production with age

Proteomic study identified a total number of 4169 proteins between quiescent young and old fibroblasts in quiescence condition from which only 63 proteins were statistically differentially-expressed (fold change>2, p-value<0.05). Under TGF-β1 treatment, the young versus old fibroblast comparison of proteomes revealed a difference of 4169 proteins from which only 73 were retained as differentially-expressed according to the criteria of fold change>2 and p-value<0.05. As revealed by the proteome expression-based heat map (Figure 2A) most of these proteins were detected in old fibroblasts rather than in young ones, either in quiescent (34/63) or TGF-β1-treatment (56/73) conditions. Similar to the secretome comparison, actin organization/cytoskeleton function represents 27% of the 63 differentially-expressed proteins in old fibroblasts (Figure 2B) and the same fields were detected under TGF-β₁-treated condition: actin with a 18% rate and ECM organization for 8%, although a 23% group of proteins with undefined functions were detected in old fibroblasts (Figure 2B). The schematic pathway representations (Figure 2C, 2D) reported that most of the proteins were linked to actin-cytoskeleton regulation, inflammation (due to the TGF-β₁ treatment), and ECM maintenance.

Actin-related proteins are a key role in fibroblast features during aging

More in depth, we selected four differentially-expressed proteins, either secretome or proteome, based upon their actin-relation and high expression levels (Table 1): Cofilin (CFL1) and Coronin 1C (CORO1C) which are two major proteins involved in actin-filament polymerization and might share co-activity in cell motility; noteworthy, they were detected in both quiescent and TGF-β1-treated proteome condition; Filamin B (FLNB) which is involved in organization of networks of filamentous actin and stress fibers [29, 30] and Actin alpha cardiac muscle 1 (ACTC1), as one of the six actin isoforms.

CFL1 is an actin filament-binding monomer [31]. It is regulated by Actin-related protein 2/3 complex subunit 3 – ARP2/3 (ARPC3) which has been described as a major actin-branched protein [32] and although its abundancy level was twice lower in the proteome of aged fibroblasts as compared to that of young cells, this was without significance. The proteome analysis showed that CFL1 was decreased in old fibroblasts with a fold change (Old/Young) of 0.30 and 0.23 in quiescence and TGF-β₁-treated condition, respectively (Table 1). This CFL1 decrease with aging was confirmed for mRNA (Figure 3A) and protein expression detected by western blot (WB) and Immunofluorescence (IF) analysis (Figure 3B, 3G). The transfection of CFL1-specific siRNA in young fibroblasts induced an effective down-regulation of CFL1 at mRNA level (42% and 69% in quiescence and stimulated condition, respectively) and protein expression (Figure 3D–3F). However, CFL1 down-regulation did not alter the contractibility capacity of young fibroblasts as shown by gel contraction assay (Figure 3I), whereas it induced some cellular morphological changes, including a compaction of cytoplasm in spindles associated with rounder cell shapes (Figure 3G). CFL1 depletion by siRNA transfection in young fibroblasts also induced a significant decrease in wound closure compared to wild-type young cells (Figure 3H) as shown by using a wound healing-scratching assay in which fibroblasts dynamically migrate to close a linear scratch-induced wound in vitro. The present results confirmed our previous results showing that the migration of young fibroblasts was significantly higher than that of old ones (Figure 3K). This significant difference was not due to a higher proliferative capacity of young fibroblasts, since it was still observed in the presence of mitomycin-C used as a well-known proliferation inhibitor (Figure 3I).

CORO1C, also designed as coronin-3, is an actin cytoskeleton regulator/modulator active in lamellipodia, protrusion formation and invasion [33]. In contrast to CFL1, CORO1C expression was upregulated in old fibroblasts in both quiescence and TGF-β₁-activated proteomic condition, with a fold change (Old/Young) of 35.40 and 3.29, respectively (Table 1). However, CORO1C mRNA expression and the protein detected by WB analysis did not show a significant difference between both young and old fibroblasts (Figure 4A–4C). As previously suggested by Waldera-Lupa et al. (2014), this absence of significant changes of the corresponding mRNA transcripts might suggest that some age-associated alterations detected at the proteome level are likely caused by other processes, such as post-transcriptional regulation, translation efficiency, protein stability or modifications, rather than by differential regulation of gene expression. This is confirmed by the slight, non-significant data depicted by the immunoblot of old and young cells, TGF-β1-treated or not (Figure 4B). CORO1C expression was significantly depleted by siRNA transfection in old fibroblasts in quiescence and stimulated condition for both mRNA and protein expression, with an inhibition >98% and >60%, respectively (Figure 4D–4F). The CORO1C immunofluorescence analysis confirmed the increased expression in old cells and its depletion after siRNA transfection (Figure 4G, 4H). Although such siRNA-induced depletion of CORO1C did not alter old fibroblast contraction ability in collagen lattices (Figure 4I), it did induce a significant alteration in their migratory capacities (Figure 4J).

CORO1C might act together with CFL1 which could be regulated by the actin-associated protein ARP2/3 complex, ARPC3 or ARP2/3. Noteworthy, a twice lower expression of ARPC3 was detected in the proteome analysis of quiescent old fibroblasts although with no significance and a significant lower level of this protein (fold ratio of 0.48 between old and young cells) was measured in the quiescent secretome (cfFigure 1A). The secretion of this key factor in actin filament branching and polymerization was quite surprising and was not detectable in cell supernatants by any method we used. As presented in Supplementary Figure 1, the ARP2/3 mRNA (Supplementary Figure 1A) and protein expression (Supplementary Figure 1B, 1C, 1G, 1H) were not different in young and old fibroblasts, but the efficient ARP2/3 depletion (more than 96%) at mRNA and protein level by siRNA transfection (Supplementary Figure 1E, 1F) did significantly decrease the migration of young fibroblasts in scratching assays (Supplementary Figure 1I).

Altogether, these data lead us to suggest that the three molecules CFL1, CORO 1C and ARP2/3 act together to regulate the actin cytoskeleton and the cytoskeleton regulation itself, as illustrated by the schematic representation of the dynamic actin-polymerization process we propose. This could explain how these molecules alter fibroblast function with aging by acting together for actin organization and consequent ECM disorganization by decreasing/increasing in aged cells (Supplementary Figure 1J). This might induce defective migration during wound healing or dermal defective appearance of aged skin.

Besides the actin-associated proteins, FLNB was retained as one isoform of the superfamily of cytoskeletal filamins that organize filamentous actin in networks and stress fibers [29, 30]. FLNB acts as an actin-binding protein that interacts with multiple receptors and intracellular proteins for regulation of cytoskeleton-dependent cell proliferation, differentiation, and migration [34, 35]. According to the proteome of quiescent fibroblasts, an increase of FLNB was observed under aging with a 2.07-fold change (Old/Young) (Table 1). However, no difference of FLNB mRNA levels was measured between young and old fibroblasts, either activated or not by TGF-β1 (Figure 5A). Depletion of FLNB mRNA expression by siRNA (>87%) (Figure 5B) induced a significant decrease in migratory capacities of old fibroblasts (Figure 5C), although it did not alter their contractibility (Figure 5D).

ACTC1 was also chosen as one of the six actin isoforms [36], with actins playing a major role in the shape and cell movement processes [37]. ACTC1 is predominantly localized in the heart and its alteration by methylation has been linked to cardiopathies [38]. The proteome analysis showed a decrease in the expression of ACTC1 in old fibroblasts with a 0.01-fold change (Old/Young) (Table 1). As for FLNB expression, ACTC1 mRNA levels were not significantly higher in young versus old fibroblasts, either activated or not (Figure 5E) although mRNA detection in young cells was downregulated by 83% after siRNA transfection (Figure 5F). Such ACTC1 depletion in young fibroblasts did significantly impact their migratory capacities (Figure 5G), but did not alter their contractibility (Figure 5H).