Age-related changes in DNA packaging, protein levels, and gene activity in blood vessel muscle cells during different types of aging

Human vascular smooth muscle cells (VSMCs), derived from aorta of at least 3 young male donors (age 29–34 years), were purchased from ATCC or Gibco and cultured in Vascular Cell Basal Medium (ATCC) supplemented as defined by the manufacturer in a humidified atmosphere with 5% CO₂ at 37^oC. The cells were seeded at 4,000 cells/cm² density and passaged when they reached 70–80% confluency until they reached the state of replicative senescence (cells were recognized as senescent if at least 70% of the population were SA-β-gal-positive and no more than 30% were BrdU-positive). Premature senescence was induced in cells on passage 10 or lower by treatment with curcumin (7,5 µM concentration, DNA-damage independent senescence) or doxorubicin (100 nM concentration, DNA-damage dependent senescence) as previously described by us [6–8]. Briefly, to induce senescence, VSMCs were treated 24 h after seeding and analyzed after 7 days of culture with the drug. Replicatively senescent cells were analyzed 24 h after seeding. Cells at early passages were collected 24 h after seeding and served as a control. The efficiency of senescence induction is shown in the Additional files (Additional file 1).

VSMCs derived from atherosclerotic plaques (AP) were obtained from patients who underwent carotid endarterectomies. The explants (1–2 mm² fragments) were placed in DMEM on selective dishes covered with gelatin (Biocoat Gelatin Cellware, Corning) and they remained in culture until cells reached confluency, no more than 30 days. Next, cells were detached from the culture dish, seeded at a density of 4,000 cells/cm² in Vascular Cell Basal Medium, cultured for 24 h, and collected for subsequent analysis. The proliferation potential of cells derived from atherosclerotic plaques is presented in Additional file 1. Cells from atherosclerotic plaques were compared to established VSMC senescence models.

Human skin fibroblasts were purchased from ATCC or Gibco and cultured in DMEM (Sigma-Aldrich or Merck) supplemented with 10% fetal bovine serum (FBS, BioWest), antibiotic and antimycotic mix solution (100 U/ml penicillin, 100 µg/ml streptomycin and 250 ng/ml amfoterycin B, Sigma-Aldrich). Cells were seeded at 4,000 cells/cm² density and passaged when they reached 70–80% confluency until replicative senescence (evaluated in similar manner as in VSMCs). Premature senescence induction was performed in cells on passage 25 or lower in the same manner as in VSMCs; however, different concentrations of curcumin (10 µM) or doxorubicin (150 nM) were used due to lower sensitivity of these cells, which was established experimentally (Additional file 2). Fibroblasts were treated 24 h after seeding and analyzed after 7 days of culture with the drug. Replicatively senescent cells were analyzed 24–48 h after seeding. Young cells collected 24–48 h after seeding, depending on the assay, served as control (detailed description in subchapter 3.8 and Additional file 2).

Cells were fixed with 4% PFA and stored in ethanol at -20^oC. Cells were permeabilized with 0.5% Triton X-100 for 10 min, blocked in 1.5% Goat Serum, 2% Bovine Serum Albumin and 0.5% Triton X-100 for 10 min and incubated with primary antibody for 2 h. Primary antibodies used: H3K9me3 (Diagenode, C15410193, 1:500), H3K27me3 (Diagenode, C15410195, 1:500), H3K4me3 (Diagenode, C15410003, 1:200), H3K9Ac (Cell Signaling, 9649, 1:400). Subsequently, cells were incubated with secondary antibody Alexa Fluor 488 or Alexa Fluor 555 (ThermoFisher Scientific, 1:500) for 1 h. DNA was stained with DAPI (1 µg/ml). Cells were analyzed using a fluorescent microscope Nikon Eclipse Ti.

Whole-cell protein lysates were prepared according to Laemmli [9]. Electrophoretically separated proteins were transferred to a nitrocellulose membrane and blocked in 5% BSA or low-fat milk. Incubation with primary antibodies was conducted overnight at 4^oC. Primary antibodies used: HMGB1 (Abcam, ab79823, 1:500), lamin B1 (Santa Cruz, sc-365962, 1:500), HP1α (Cell Signaling, #2616, 1:1000), H3K4me3 (Diagenode, C15410003, 1:1000), H3K9me3 (Diagenode, C15410193, 1:1000), H3K9Ac (Cell Signaling, #9649, 1:1000), H3K27me3 (Diagenode, C15410195, 1:1000), GAPDH (Millipore, MAB374, 1:150000), Suv39H1 (Cell Signaling, 8729, 1:500), EED (GeneTex, GTX634650, 1:1000), H3 (Abcam, ab1791, 1:10000), H3.3 (Abcam, ab176840, 1:1000). Membranes were then incubated with HPR-conjugated secondary antibody at 1:2000 dilution in low-fat milk (Dako, Agilent), washed and developed with ECL (ThermoFisher Scientific). Protein level was normalized to GAPDH or H3 and presented as a fold change relative to control.

Trypsinized VSMCs and AP cells were fixed in RNAlater solution and stored in -80 °C. Isolation of total RNA was performed in an automated manner using the MagNA Pure Total RNA Isolation kit (Roche) in the MagNA Pure Compact apparatus (Roche) according to the manufacturer’s instructions. The RNA concentration and quality were measured using a NanoDrop 2000 spectrophotometer (ThermoFisher Scientific), while integrity was determined using capillary electrophoresis with a Bioanalyzer 2100 (Agilent Technologies) and a set of dedicated reagents provided by RNA 6000 Nano kit (Agilent Technologies). Isolated RNA was stored at -20 °C (short-term) or -80 °C (long-term). To control the level of contamination with genomic DNA, a PCR reaction was performed to amplify exons 1 and 2 of human β-globin (HBB). The reaction involved PCR reaction buffer (A&A Biotechnology), an appropriate pair of HBB primers (sequence shown in Additional file 5) and the tested sample. The reaction was performed in a Mastercycler^® nexus (Eppendorf) under the following conditions: 4 min at 94^oC, 30 cycles of 45 s at 94^oC, 45 s at 57^oC, 1 min 72^oC. The obtained product was examined using electrophoresis in a 1% agarose gel.

The gene expression profile of VSMCs subjected to different types of senescence (in at least three biological replicates) was examined by microarray analysis using 250 ng of RNA and the GeneChip™ WT PLUS reagent kit (ThermoFisher Scientific) compatible with the Affymetrix Gene Atlas System platform, according to the manufacturer’s instructions. The prepared samples were hybridized to Affymetrix™ HuGene 2.1 ST Array Strips (Affymetrix, Santa Clara, CA, USA). After washing and staining, the arrays were scanned in the GeneAtlas Imaging Station (Affymetrix), generating. CEL files as the data output. Data was analyzed using Transcriptome Analysis Console (TAC) software (ThermoFisher Scientific). Probes signal intensities were normalized with RMA (Robust Mulit-array Average) method and, after quality control, one-way ANOVA was applied to determine differentially expressed genes (DEGs) between treated cells and control. To minimize the variability originating from different sample preparation dates in a comparison analysis, batch effect in TAC was applied. The criteria for selecting DEGs were fold change ≤ − 1.3 and ≥ 1.3, and p value ≤ 0.05. The enrichment analysis was performed using g: Profiler tool [10, 11] with default parameters.

The reverse transcription reaction was performed using the Maxima First Strand cDNA Synthesis Kit (ThermoFisher Scientific) according to the manufacturer’s instructions. The reaction was performed in a Mastercycler^® nexus thermal cycler (Eppendorf) using 1 µg of RNA isolated as described in 2.4. Reaction conditions consisted of 10 min at 25 °C, 15 min at 50 °C and 5 min at 85 °C. The expression of selected genes changing during senescence was assessed by performing real-time polymerase chain reaction (qPCR). For this purpose, the LightCycler^® 480 SYBR Green I Master reagent (Roche), a set of previously validated primers purchased from realtimeprimers.com (primer sequences are shown in Additional file 5), and 20 ng of cDNA template (VSMCs and AP cells) were used. The reaction was performed in triplicates in a volume of 10 µl using the LightCycler^®480 apparatus/instrument (Roche) under the following conditions: 10 min at 95 °C and 40 cycles of 10 s at 95 °C, 10 s at 59 °C and 30 s at 72 °C. GAPDH, EIF2B and DAD1 were used as reference genes and changes in gene expression levels were analyzed using the Pfaffl method [12] and were presented as a fold change relative to control cells.

The chromatin immunoprecipitation procedure was performed using the iDeal kit for histones (Diagenode) according to the manufacturer’s instructions. Briefly, 3 million VSMCs were trypsinized, suspended in PBS, and subjected to the cross-linking process with fresh, 1% methanol-free formaldehyde solution (Carl Roth) at room temperature. The time of reaction depended on the experimental variant (after the optimization process). Control cells were incubated for 1 min, cells treated with doxorubicin and curcumin for 2 min, and replicatively senescent cells for 3 min. The reaction was stopped by adding 0.125 M glycine and further incubation at room temperature for 5 minutes. Cells were then washed twice with cold PBS and centrifuged at 1800 rpm (Eppendorf 5424 R) for 5 min at 4 °C. From this point on, the entire immunoprecipitation procedure was performed at 4 °C using cold reagents. After cell lysis, chromatin was sheared using Bioruptor Plus (Diagenode) with a built-in cooling system set at 4 °C. Cells were sonicated at high power mode for 40 cycles (30s ON, 30s OFF) (CTRL) or 30 cycles (DOX, CUR, and RS). Sonication efficiency was controlled using 2100 Bioanalyzer and the High Sensitivity DNA kit (Agilent). The chromatin immunoprecipitation was performed overnight on a rotating wheel using 7.5 × 10⁵ cells, dedicated buffers, A-coated beads and antibodies recognizing H3K4me3 (1 µg, Diagenode, C15410003), H3K9me3 (1.5 µg, Diagenode, C15410193) and HP1α (2.5 µg, Cell Signaling, #2616). After several subsequent washings, chromatin was eluted and decrosslinked overnight at 65 °C with shaking. Finally, DNA was precipitated with magnetic beads and quantified using Qubit dsDNA HS kit (Invitrogen). Immunoprecipitation was assessed by qPCR using LightCycler^® 480 SYBR Green I Master (Roche) and the following amplification conditions: 10 min at 95 °C and 40 cycles of 30 s at 95 °C, 30 s at 60 °C, 30 s at 72 °C.

The library was prepared from 3 ng of material using the KAPA HyperPrep Kit and compatible KAPA Unique Dual-Indexed Adapter Kit (KAPA Biosystems) according to the manufacturer’s instructions. The library quality was assessed using 2100 Bioanalyzer and a High Sensitivity DNA Analysis kit (Agilent), and the quantity was measured using qPCR using the KAPA Library Quantification kit (KAPA Biosciences).

Raw reads were trimmed using Trimmomatic software [13]. Trimmed sequences were mapped to the human reference genome provided by ENSEMBL, (GRCh38) using Bowtie2 [14] with default parameters. Duplicates reads were identified and removed using the MarkDuplicates tool from the GATK package (version 4.2.3.0) [15] with default parameters except OPTICAL_DUPLICATE_PIXEL_DISTANCE set to 12,000. Peaks were detected with MACS2 [16]. To account for variability between replicates, reads overlapping each peak detected by MACS2 were counted, separately in each replicate, with the summarize Overlaps tool from the Genomic Alignments library. Resulting counts were used by edgeR [17] to select locations with reproducible binding patterns across all replicates. Peaks were visualized using the IGV software. Functional analysis was performed using g: Profiler tool with default settings.

The enzymatic activity of class I, IIb and IV histone deacetylases was determined colorimetrically using the Colorimetric Histone Deacetylase Activity Assay kit (ScienCell). During the first stage of reaction, lysine residues in the substrate included in the kit are deacetylated by HDAC enzymes. The resulting product releases a chromophore detected at 405 nm after incubation in a dedicated development buffer. Histone deacetylase activity was given in pmol of deacetylated product/min/mg of lysate using formulas provided by the manufacturer. The set of tested deacetylases included HDAC1, HDAC2, HDAC3, HDAC6, HDAC8, HDAC10 and HDAC11. Whole-cell protein extracts were obtained using RIPA lysis buffer (Sigma) supplemented with protease inhibitors (Roche). Protein concentration was estimated using the BCA colorimetric method (Sigma); absorbance was measured at 562 nm (Tecan). The results were compared to a standard curve.

Enzyme activity was analyzed according to Dimri [18]. Cells were fixed (2% formaldehyde, 0.2% glutaraldehyde, PBS), washed and incubated overnight at 37^oC without access to light and carbon dioxide in staining buffer (5 mM potassium ferrocyanide, 5 mM potassium ferrycyanide, 150 mM NaCl, 2 mM MgCl2, 0.02 M phosphate buffer and X-Gal at concentration 1 mg/ml, pH 6.0.). Nuclei were stained with DAPI (1 µg/ml). Cells were analyzed using a fluorescent microscope Nikon Eclipse Ti. The results are expressed as % of cells with increased enzyme activity (blue color) to all cells (based on DAPI-stained nuclei).

The ability of cells to replicate DNA was assessed using the BrdU incorporation assay. BrdU (Sigma-Aldrich) at 10 µM concentration was added for 24 h to the medium. Then, cells were fixed with ice-cold 70% ethanol and stored at -20^oC for a minimum of 24 h. Cells were immunostained with anti-BrdU primary antibody (1:100, Becton Dickinson) and a secondary Alexa Fluor 488-conjugated antibody (1:500, ThermoFisher Scientific). DNA was stained with DAPI (1 µg/ml). The percentage of dividing cells was assessed by considering the number of BrdU-positive cells among all cells (DAPI-stained nuclei).

For DNA content analysis, VSMCs and fibroblasts were fixed in 70% cold (-20^oC) ethanol and stored at least 24 h at -20ºC. After washing twice in a 0,05% Tween solution in PBS (1200 rpm, 5 min., Eppendorf 5702) the pellet was resuspended in PBS 1:1 with citrate buffer (4 mM citric acid and 0,2 M Na2HPO4) and incubated for 5 min. After centrifugation (1200 rpm, 5 min), cells were resuspended in staining buffer (3,8 mM sodium citrate, 50 µg/ml RNAse A, 50 µg/ml PI, in PBS) and incubated for 30 min in the dark. Approximately 10,000 cells were analyzed using the LSRFortessa flow cytometer (BD Biosciences).

Statistical analysis was performed on at least three separate biological replicates. Standard deviations were determined using the GraphPad Prism program, and in most cases, one-way ANOVA was used to determine the statistical significance of changes occurring between senescent and young cells. If another statistical test is used, its details are included in the description of the given method and in the figure caption. The obtained values were recorded as follows: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), p < 0.0001 (****). The following software were used to prepare the figures: ImageJ, GraphPad, Excel, Transcriptome Analysis Console.

In VSMCs undergoing senescence in three different manners (premature induced by curcumin - CUR or doxorubicin - DOX and replicative - RS), we analyzed histone H3 methylation and acetylation, key epigenetic modifications that significantly influence chromatin packing and gene expression. Modifications typical of condensed chromatin - heterochromatin, i.e. H3K9me3 and H3K27me3, and markers typical of active euchromatin, such as H3K4me3 and H3K9Ac were selected (Fig. 1). To ensure that changes in the amount of a given modification were not due to alterations in H3 level, we normalized the protein level to both GAPDH (unchanged in all variants) and H3 (the expression of which diminished during senescence, Fig. 1C). Considering heterochromatin markers, we observed more than a 50% decrease in H3K9me3 in PS (CUR, DOX) and its almost complete disappearance in RS (Fig. 1A). This decrease was evident when compared to both GAPDH and H3 level (Fig. 1B). A small decrease was recorded for H3K27me3 when normalized to the GAPDH level, but not if normalized to the H3 level and was similar in both types of senescence. Although chromatin loosening was linked to a reduction in modifications characteristic of heterochromatin, it did not coincide with an increase in modifications typically associated with active euchromatin. Unexpectedly, the levels of H3K4me3 also decreased in both types of senescence (normalization to both GAPDH and H3). H3K4me3 showed an identical profile to H3K9me3, and significant differences between PS and RS were also visible. The H3K9Ac level did not change if normalized to H3 and diminished when GAPDH served as loading control. The alterations were senescence-type independent. We analyzed the chromatin compaction markers by immunocytochemical staining (Fig. 1D). The fluorescence intensity was measured only for nuclei with a surface area greater than 200 µm² for senescent cells or smaller than 200 µm² in the case of young ones. Images of DAPI-stained nuclei were analyzed using CellProfiler, and the nuclear surface area was among the measured parameters. After statistical analysis, the data were visualized using box plots to compare distributions across experimental conditions.

To define a threshold for further analysis, we used the lower edge of the box (Q1) in the CUR condition, which was 200 μm². This value reflects the lower boundary of typical nuclear sizes where curcumin-induced changes were most evident. Notably, it approximately aligns with the upper whisker (highest non-outlier value) of the control condition, supporting its use as a biologically relevant cut-off to identify senescent cells with enlarged nuclei (manuscript with details is in preparation). The results showed that the average staining intensity for all studied modifications decreased.

In addition to a significant loss of histone modifications, H3K9me3, H3K4me3 and histone H3 itself, in both types of senescence, the H3.3 variant (normalized to GAPDH) decreased in PS cells, while RS cells exhibited H3.3 levels comparable to young cells. However, if H3 served as control, no changes were observed between young and PS cells and even an increase in RS cells was noticed (without statistical significance) (Fig. 1A and B). The level of histone H3.3, which is incorporated in a DNA replication-independent manner during transcription, was analyzed due to the association of senescence with cell cycle arrest [17, 19]. H3.3 plays a role in chromatin organization and function and regulates gene expression and genome integrity. It is involved in both transcriptional activation and repression, depending on the genomic location and specific histone chaperones [20, 21]. Microarray data revealed that senescence is associated with the transcript loss of numerous canonical and non-canonical histone variants, regardless of the type of senescence. (Fig. 1E). The observed decrease in expression of genes encoding histone clusters may lie behind senescence-associated chromatin loosening. Our results indicated that H3K9me3 and H3K4me3 decreased in all types of senescence while H3K27me3 and H3K9Ac were not altered.

Due to the gradual decline in H3K4me3 and H3K9me3 in the examined types of senescence, we analyzed where these markers accumulate in the genome and how this may affect the gene expression profile. We analyzed the sites of their deposition using histone immunoprecipitation and next-generation sequencing (ChIP-seq).

Cells treated with doxorubicin (DOX) had the lowest number of H3K4me3 reads, and the total number of enrichment regions was 5-fold lower than in other variants (Fig. 2A). The number of reads, above 30,000, was similar in CTRL, CUR, and RS. Most of the enrichment was recorded at gene promoter regions at the transcription start site (TSS), which is consistent with the literature data for this modification [22]. Exceptionally, in the DOX variant, over 80% of all reads were located mainly in the promoters, and the small spread around the TSS may suggest the appearance of a very narrow peak of enrichment for this histone modification.

For H3K9me3, the lowest number of reads was also recorded for the DOX variant (15454). Slightly more were obtained for RS (27542) and CTRL (47027) cells, while, the highest number of enrichments/reads was obtained for cells treated with curcumin (52611) (Fig. 2B). Moreover, H3K9me3 localized mainly to distal intergenic regions for all variants tested. Occasionally, this modification accumulated at promoter sites, but this was only 10% of all reads, and TSSs were mostly devoid of H3K9me3. In addition, the visual enrichment analysis in IGV software showed a slightly higher density of this modification in pericentromeric and subtelomeric regions (Additional file 3).

Significant reads indicated by edgeR (edgeR < 0.05), located in the promoters of protein-coding genes in all experimental variants, were selected for functional analysis of H3K4me3-rich regions (Fig. 3A). As H3K4me3 usually marks active chromatin, its enrichment in the gene promoter is strongly associated with gene transcription and as a result, activation of biological processes the genes are involved in. The Venn diagram shows intersecting genes with H3K4me3 located in their promoters in both young and senescent cells (i.e. CTRL, CUR, DOX, RS) that were used to identify enriched functional terms from Gene Ontology. The set of genes common to senescence, i.e. the intersection of the CUR, DOX and RS sets, contained 82 genes responsible for development and cell differentiation. However, the presence of only 44 genes characteristic of PS in both doxorubicin (DOX) and curcumin (CUR) treated cells hindered a comprehensive functional analysis. Conversely, the appearance of enriched H3K4me3 peaks at loci specific to a given experimental condition (CTRL: 412, CUR: 218, DOX: 331, and RS: 329) indicates genome-wide reorganization of H3K4me3. The results of functional enrichment analyses for key processes associated with these genes, along with other overlapping gene sets are shown in the graph (Fig. 3A).

Functional enrichment analysis of GO terms connected with protein-coding genes decorated with inactive chromatin mark, H3K9me3, in promoter regions, introns, or intergenic sites was also performed using significant reads (edgeR < 0.05). Due to an insufficient number of reads, a complete analysis of the same groups as in H3K4me3 was not performed (Fig. 3B). Moreover, despite an extensive set of characteristic enrichment sites for CUR (221), DOX (165) and RS (137), no information was obtained on the biological processes in which the genes might be involved.

The observed decline in H3K4me3 and H3K9me3 in senescent cells has prompted us to investigate the involvement of selected histone-modifying enzymes (methyltransferases, demethylases, but also acetyltransferases and deacetylases) specific for lysines 4, 9 and 27 of histone H3. The corresponding list of proteins was downloaded from Gene Ontology database and compared with gene expression data obtained with DNA microarrays. The summary of the gene expression profile for each analyzed enzyme and the applied GO term ID are presented in Additional file. 6

The HP1α protein attaches to trimethylated lysine 9 of histone H3 and is involved in forming transcriptionally inactive heterochromatin [35]. Our study has shown that the expression of HP1α decreases in all analyzed types of senescence (Fig. 6A and B). Immunocytochemical analysis of cell nuclei selected based on their surface area [manuscript in preparation] showed a decrease in the average fluorescence intensity of HP1α (per nuclear area) of senescent cells (Fig. 6D). It was accompanied by altered distribution of this protein in PS (Fig. 6C). In the nuclei of young control cells, few clusters of HP1α appeared (median is 2) (Fig. 6E), and the relatively intense fluorescence signal was mostly evenly distributed. In senescent cells, a decrease in fluorescence intensity and increased foci number were observed; however, the foci were less pronounced in RS than in PS. The highest number of HP1α foci was recorded in cells treated with doxorubicin.

Moreover, the HP1α enrichment sites in the genome were investigated. As HP1α interacts with H3K9me3 and the level of both proteins decreases during senescence, we analyzed the co-occurrence of HP1α sites with H3K9me3. In light of previous reports that HP1a (an HP1α homolog) in Drosophila melanogaster may interact with chromatin independently of H3K9me3 [36], we decided to investigate the potential binding of HP1α to chromatin independently of H3K9me3 (using the ChIP-seq method) (Fig. 7).

The analysis of the enrichment sites of HP1α showed that the total number of reads was significantly lower than for H3K9me3. Moreover, the fewest number of reads was obtained for DOX-treated cells (Fig. 7). Similarly to H3K9me3, the highest number of HP1α reads was recorded in distal intergenic areas, and the percentage differed between variants. The DOX variant appears to have the strongest rearrangement of HP1α enrichment in the genome as the percentage of reads at distal intergenic sites decreased (Fig. 7A). The HP1α distribution around the TSS showed lower reads in each variant, with a significant proportion located in more distant regions of the genes. As the similarities in the distribution of reads between HP1α and H3K9me3 were detected, an analysis of an intersecting set of concurrent sites was illustrated by a Venn diagram (Fig. 7B left). As can be noted, the most significant number of such interaction sites occurred in young cells. Therefore, only genes identified in young cells were subjected to functional enrichment analysis. The concurrent HP1α and H3K9me3 sites were associated with the regulation of transcription and RNA biosynthesis, and the genes involved in these processes belong mainly to the ZNF family of transcription factors. Attempts were made to identify sites of HP1α binding with chromatin not involving H3K9me3. The analysis did not reveal many sites; no genes were common for senescence types.

Functional enrichment analysis of RS and PS-specific genes was performed to identify changes occurring in senescent VSMCs based on the results obtained from microarrays (Fig. 8). Principal component analysis (PCA) showed significant differences in the transcriptome between young (blue) and RS cells (green), young and PS cells (red - DOX, purple - CUR), and between the two types of senescent cells (PS and RS) (Fig. 8A). The results support the existence of differences between stress-induced and replicative senescence. Differentially expressed genes (DEGs), specific to each senescence variant, are also shown in a heatmap form and as hierarchical clustering of both genes and variants (Fig. 8B). The colors on the heatmap represent the recorded signal (log2), where blue indicates the lowest and red is the highest signal. The clustering for experimental variants provided the same conclusions as the PCA analysis. Therefore, it was decided to analyze gene expression changes common to all senescence variants. In the Venn diagram, the intersecting sets of DEGs were marked accordingly, and their expression was normalized to control cells (Fig. 8C). Seven genes with the highest and lowest expression for each set are shown (Fig. 8D). The functional enrichment analysis was then preformed for each set of genes in the studied group (PS, RS, genes common to PS + RS) using the Gene Ontology database, and the results are presented for biological processes and cellular components (Fig. 8E). List of genes characteristic exclusively for senescence induced by DOX or CUR is presented in Additional file 8.

The results showed that genes with altered expression in all types of senescence (referred to genes common to all types of senescence), encode proteins located mainly in the nucleus, where they are involved in cell cycle control, chromosome organization, replication and DNA repair. Significant differences in gene functions can be seen between RS and PS. In the case of RS, gene products are associated with maintaining metabolic processes and protein synthesis and are most often found inside various cell organelles, including the nucleus and the cytoplasm. Cells treated with doxorubicin and curcumin (PS) show changes in the expression of genes related to response to external stimuli. Interestingly, these genes are also involved in cell adhesion and cytoskeletal reorganization. The proteins encoded by these genes localize in the cytoplasm and are associated with actin filaments and cell adhesion sites. Changes in their expression could affect cell migration, as we have previously observed for PS cells (in preparation).

Analysis of senescence markers showed that atherosclerotic plaques (AP) consist of a heterogeneous population of young and senescent cells. Due to different proportions of senescent cells, the expression of senescence markers is strongly donor-dependent. Cells derived from atherosclerotic plaques may preserve young cell morphology, but, at the same time, decreased levels of LMNB1 (lamin B1) and HMGB1 (High mobility group box 1 protein), recognized senescence markers, were detected in whole cell extracts (Additional file 1). About 60% of the cells showed increased SA-β-gal activity, but approximately 60% could still replicate DNA (Additional file 1). The level of histone H3 modifications was strongly donor-dependent and mostly resembled that in young VSMCs cultured in vitro (Fig. 9A). For comparison, we have shown VSMCs senesce in vitro. We normalized all H3 modifications in VSMCs and AP cells to GAPDH level because, in this manner, it is more visible that the level of studied modifications did not change spectacularly in AP cells. As we have shown for VSMCs, in the case of H3K9me3 and H3K4me3 normalization to H3 and GAPDH gave the same results; however, in the case of H3K27me and H3K9Ac their relative level decreased if GAPDH served as a loading control. In AP cells, the level of all studied H3 modifications remained relatively high. The HP1α level was also donor-dependent, and the direction of HP1α changes in individual donors correlated with the heterochromatin marker H3K9me3 (Fig. 9A). mRNA analysis demonstrated a drop in HP1α gene expression in AP cells (Fig. 9B). The immunostaining analysis showed that the average fluorescence intensity of HP1α protein decreased with increasing nucleus area (Fig. 9C). Similarly to the in vitro senescence of VSMCs, there was an increase in the number of HP1α foci in the large nuclei of senescent cells (selected based on the area > 200 μm) compared to young ones (selected based on the area < 200 μm), in which foci appeared in smaller numbers (Fig. 9C and D). Immunocytochemical staining showed that the fluorescence signal intensity of the tested markers was highly variable and donor-dependent, which is consistent with observations at the protein level.

We applied qPCR to measure the level of genes encoding selected histone-modifying enzymes (Fig. 9E). The results are presented as a fold change relative to young VSMCs and compared to senescent VSMCs. The expression of the majority of genes decreased (SUV39H1, EZH2, KDM4D, HDAC1) or remained unchanged (HDAC3), which was consistent with those obtained for VSMCs. Surprisingly, in contrast to VSMCs, there was no significant change in EZH1 expression, and the expression of HDAC9 decreased. This might be a result of a more heterogenous population of AP cells. Nevertheless, the trend in enzyme expression mostly resembled this observed in the in vitro model.

We performed selected analyses on human fibroblasts to recognise whether senescence-associated alterations in H3 modifications and HP1α in VSMCs are cell type-specific or universal. Senescence was induced in a manner similar to that of VSMCs. The percentage of senescent and non-proliferating cells was evaluated, and the expression profile of typical senescence markers, LMNB1 and HMGB1, was analysed (Additional file 2). However, unlike in the case of VSMCs, young fibroblasts that served as control cells were collected 48 h after seeding since preliminary experiments showed that, at this time point, they were proliferating most intensely and were, therefore most suitable for studying differences with respect to senescent cells. In some analyses, we also show results for cells collected 24 h after seeding for comparison. According to the literature, in early passages between 1 and 20, the time for cell division is 30–40 h [37, 38].

In fibroblasts, a significant decrease in H3K9me3 in both senescence types was observed (Fig. 10A), which reflects the alterations observed in VSMCs (Fig. 1A). The H3K27me3 level tended to increase, without statistical significance, if compared to the H3. Diminished H3K27me3 level was noticed in PS when normalized to GAPDH. Euchromatin markers showed a similar trend to those observed in VSMCs - that is a noticeable decrease in H3K4me3 in both types of senescence (normalized to both H3 and GAPDH), no significant alterations in H3K9Ac when compared to H3 and a reduction if normalized to GAPDH. Similarly to VSMCs, neither H3K4me3 nor H3K9Ac diversification between PS and RS was detected. In both senescence types, there was also a decrease in histone H3 itself (Fig. 10B) and its variant H3.3 (irrespective whether normalized to H3 or GAPDH, statistically significant only with GAPDH) (Fig. 10C). In conclusion, when the level of histone modifications in senescent fibroblasts was compared to proliferating ones (control 48 h), changes in the level of H3K9me3, H3K4me3, H3K9Ac and H3K27me3 resembled those observed in VSMCs. Thus, a reduced level of H3K9me3 and H3K4me3 appears to be a universal senescence marker. In addition, there was a senescence-related overall decrease in histone H3, which could result in nucleosome loss and chromatin loosening. Analysis of the level of H3 modifications in control cells as a function of the time of culture showed low levels of all tested modifications 24 h after seeding. This is related to the delayed entry of cells into the cell cycle (Additional file 2). Immunocytochemical analysis showed a decrease in the fluorescence intensity and significant similarities in the distribution of all markers that accompanied the senescence of VSMCs (Fig. 1B).

Analysis of enzymes regulating chromatin structure by modifying histone H3 showed inhibition of SUV39H1 expression (Fig. 10D) in fibroblasts regardless of the senescence type. Similar results were obtained in VSMCs. In contrast to VSMCs, in fibroblasts the level of HDAC1 increased in PS and decreased in RS cells (compared to the 48-hour control) (Fig. 10E). The highest increase of HDAC1 was recorded in cells treated with curcumin, which was, however, not reflected by protein acetylation status (activity of the deacetylase set, Fig. 10F). In contrast, the highest activity of deacetylases was observed in doxorubicin-treated cells.

HP1α synthesis was reduced in all types of senescence, with the most spectacular drop observed during RS (Fig. 10G). This was also confirmed by immunocytochemical staining (Fig. 10H). To eliminate population heterogeneity, a similar criterion to cell nuclei size selection as in VSMCs (in vivo and ex vivo) was applied to fibroblasts. Unlike VSMCs, no accumulation of HP1α foci in the nucleus was observed during fibroblast senescence (not shown).

We characterized senescence type-dependent changes in chromatin-associated proteins and enzymes that influence the nucleus and chromatin structure in VSMCs. This study can serve as a starting point to search for factors that may be used to distinguish between premature and replicative senescence. We mapped the deposition of H3K4me3, H3K9me3 and HP1α in the genome of VSMCs subjected to PS and RS and described transcriptomic alterations that arise depending on the senescence-inducing stimulus. We compared the profile of H3 modifications and HP1α level and distribution characteristic for VSMCs with changes detected in fibroblasts and, to some extent, in VSMCs derived from atherosclerotic plaques. A phenomenon that seems to be universal regardless of cell type and senescence stimulus, is a significant decrease in H3K9me3, H3K4me3 and HP1α levels. In contrast, alterations detected in cells derived from atherosclerotic plaques were more variable and donor-dependent. Our study provides evidence that changes in the structure of the nucleus and chromatin, associated with a gradual loosening of chromatin, are universal for all senescence types. However, the highest extent of deheterochromatinization is a feature of cells undergoing RS. We can also conclude that certain similar alterations occur in senescent VSMCs ex vivo, although inter-individual differences usually obscure them. Our results clearly showed that differences existed not only between young and senescent cells but also between PS and RS ones. Only subtle differences between various PS types were observed, suggesting that different stressors activate the same cellular mechanisms/pathways.