Multi-omics characterization of partial chemical reprogramming reveals evidence of cell rejuvenation
- Division of Genetics, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, United States
- Department of Cell Biology, Harvard Medical School, United States
- Broad Institute of MIT and Harvard, United States
Figures
Figure 1
Functional effects of partial chemical reprogramming.
(A) Overview of the study. Tail and ear fibroblasts were isolated from young (4-month-old) and old (20-month-old) male C57BL/6 mice, and cryo-stocks were prepared once reaching ~80–90% confluency (passage number P1). All fibroblasts used in this study were ≤ P4. These cells were subjected to partial chemical reprogramming followed by the indicated analyses. (B) AP staining. Young and old fibroblasts were treated with 2c, 7c, or DMSO for 4 days, followed by visualization of cells positive for alkaline phosphatase activity with the StemAb Alkaline Phosphatase Staining Kit II (4X objective). Scale bars: 126.2 µm (100 pixels). (C) TMRM staining. Following treatment for 6 days with 2c, 7c, or DMSO, fibroblasts were stained with 250 nM TMRM and 10 µg/ml Hoechst 33342 for 20 min at 37 °C, 5% CO2, and 3% O2. TMRM fluorescence intensity was normalized to the number of nuclei per field and quantified across four to five images from random fields for each independent biological replicate (n=3). For CCCP treatment, cells were treated with 50 µM CCCP in DMSO for 15 min prior to TMRM staining. Error bars represent means ± standard deviations, and data were quantified based on percent change from control-treated fibroblasts. Scale bars: 31.1 µm (100 pixels). Statistical significance was determined by one-way ANOVA and Tukey’s post-hoc analysis. *p < 0.05, **p < 0.01, ***p < 0.001. (D) Effects on oxygen consumption. Top: representative raw traces of oxygen consumption of cells subjected to the Mito Stress Test protocol (basal, followed by 1 µM oligomycin a, 5 µM FCCP, and 1 µM rotenone / antimycin a) following 6 days of partial chemical reprogramming. Error bars represent means ± standard deviations from three technical replicates per treatment. Bottom: quantified oxygen consumption rates across four independent biological replicates (n=4, error bars represent means ± standard deviations). Statistical significance was determined by one-way ANOVA and Tukey’s post-hoc analysis. *p < 0.05, **p < 0.01, ***p < 0.001.
-
Figure 1—source data 1
Normalized change in TMRM fluorescence.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig1-data1-v1.xlsx
-
Figure 1—source data 2
Oxygen consumption rates.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig1-data2-v1.xlsx
Figure 2 with 2 supplements
Effect of partial chemical reprogramming on gene expression.
(A) PCA of bulk RNA-seq samples. Fibroblasts were treated with 2c, 7c, or control for 6 days followed by RNA-seq analyses. (B) Differentially expressed genes. Differentially expressed genes were determined using edgeR and considered statistically significant at a Benjamini-Hochberg false discovery rate (FDR) cut-off < 0.05. (C) Expression of pluripotency markers. Effect of partial chemical reprogramming on the normalized expression of pluripotency markers and Klf4 and Myc (c-Myc). Statistical inference was performed with edgeR. *p.adjusted < 0.05, **p.adjusted < 0.01, ***p.adjusted < 0.001. (D) Splicing damage. Splicing damage was determined by the proportion of alternative-splicing events that may disrupt protein function. Statistical significance was determined by one-way ANOVA and Tukey’s post-hoc analysis. ^p < 0.1, *p < 0.05. (E) Association of gene expression changes induced by chemical reprogramming with signatures of aging and OSKM reprogramming. Signatures of aging are labeled in red, whereas signatures of OSKM reprogramming are labeled in cyan. NES: normalized enrichment score. Statistical inference was done with the fgsea R package. *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001.
-
Figure 2—source data 1
Expression of pluripotency markers.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig2-data1-v1.xlsx
-
Figure 2—source data 2
Splicing damage values.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig2-data2-v1.xlsx
Figure 2—figure supplement 1
Effect of partial chemical reprogramming on aging-related changes in fibroblast gene and protein expression.
(A) Effect of fibroblast age on gene expression changes. Left panel: association of aging-related changes in fibroblast gene expression with signatures of aging and OSKM reprogramming. Signatures of aging are labeled in red, whereas signatures of OSKM reprogramming are labeled in cyan. NES: normalized enrichment score. Statistical significance was determined using GSEA as implemented in the fgsea R package. *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001. Right panel: Volcano plot of age-related differentially expressed genes (genes with FDR < 0.05 colored in red, differential expression analysis performed using edgeR). (B) Correlation between gene expression changes induced by chemical reprogramming in young and old fibroblasts. Spearman correlation coefficients are shown in bold. Labeled are the genes that are most strongly differentially expressed after 2c and 7c treatment (Benjamini-Hochberg FDR < 0.05) in young and old fibroblasts (shown in green). Also depicted are genes significantly changing only in old fibroblasts (shown in cyan), or only in young fibroblasts (shown in red). (C) GSEA of aging-related changes in protein abundance. Shown here are comparisons between old and young untreated fibroblasts (left), old 2c-treated and old untreated fibroblasts (center), and old 7c-treated and old untreated fibroblasts (right). 'Benjamini-Hochberg FDR < 0.1, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001.
Figure 2—figure supplement 2
Characterization of alternative splicing events.
(A) Observed alternative splicing events. Observed alternative splicing events in control samples (n=4 independent biological replicates) grouped by different types of alternative splicing (A3SS = alternative 3’ splice site, A5SS = alternative 5’ splice site, IR = intron retention, ES = exon skipping) and age. (B) Distribution of alternative splicing changes in old fibroblasts after 7c treatment. Each datapoint represents one splicing event that was differentially spliced (|ΔPsi| > 0.1 and Benjamini-Hochberg FDR < 0.05) following 7c treatment. (C) Distribution of alternative splicing changes in young fibroblasts after 7c treatment. Each datapoint represents one splicing event that was differentially spliced (|ΔPsi| > 0.1 and Benjamini-Hochberg FDR < 0.05) following 7c treatment.
Figure 3 with 2 supplements
Effect of partial chemical reprogramming on protein expression.
(A) PCA of log2 normalized protein abundances. Fibroblasts were treated for 6 days with 2c, 7c, or control followed by mass spectrometry-based proteomic analyses. (B) Global effects of partial chemical reprogramming on the proteome. Scaled heatmap of normalized protein abundances with clustering. (C) Comparisons between effects on protein and gene expression. Log2 fold changes of protein (vertical axis) versus mRNA (horizontal axis) for 20-month-old fibroblasts treated with 7c. Spearman correlation coefficient is shown in bold. Labeled are the genes that are most strongly differentially expressed after 7c treatment (adjusted p-value < 0.05) at both the mRNA and protein levels (shown in green). Also depicted are genes significantly changing only at the protein level (shown in cyan), or at the mRNA level (shown in red). Spearman correlation coefficients between protein and mRNA levels were > 0.7 for all treatment conditions (refer to Figure 3—figure supplement 1). (D) Functional GSEA. Pathways enriched by gene expression changes induced by partial chemical reprogramming (blue: 7c, green: 2c) and aging in primary fibroblasts (red) from the current study, as well as by established signatures of OSKM reprogramming (cyan) and aging (red). ^Benjamini-Hochberg FDR < 0.1, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001. (E) Effect of partial chemical reprogramming on the expression of protein complexes associated with aging. Protein abundances of mitochondrial OXPHOS complexes, chaperones, collagens, and the spliceosome are all affected by partial chemical reprogramming. Each datapoint represents the abundance of an individual protein. Significance was assessed with a linear regression model that included effects for gene, treatment, age group, and a treatment:age group interaction. p-values within each panel were corrected for multiple testing with the default Dunnett correction (‘single-step’) method in the multcomp R package (Hothorn et al., 2008). 'adjusted p-value < 0.1, *adjusted p-value < 0.05, **adjusted p-value < 0.01, ***adjusted p-value < 0.001, n.s. adjusted p-value ≥ 0.1.
-
Figure 3—source data 1
Differential expression analysis of mRNA-seq experiments following data normalization in edgeR.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig3-data1-v1.xlsx
-
Figure 3—source data 2
Differential expression analysis of TMT proteomics experiments following data normalization.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig3-data2-v1.xlsx
-
Figure 3—source data 3
Gene set enrichment analysis (GSEA) of mRNA-seq experiments.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig3-data3-v1.xlsx
Figure 3—figure supplement 1
Additional correlation analyses.
(A) Comparisons between protein differential abundance and differential gene expression. Shown here for all treatment conditions (top: 2c vs. control, bottom: 7c vs. control. Left panels: young fibroblasts, right panels: old fibroblasts). Spearman correlation coefficients are shown in bold. (B) Correlations between protein and mRNA levels. Heatmap of Spearman correlation coefficients across all treatment conditions calculated based on the union of top 1000 genes with the lowest p-value for each pair of signatures (blue: 7c, green: 2c). (C) Correlation between log2 normalized abundances of the proteomics samples. Heatmap of Spearman correlation coefficients.
Figure 3—figure supplement 2
GSEA of protein abundances.
(A) GSEA (GObp, HALLMARK, KEGG, and REACTOME) of functions and pathways most significantly altered by partial chemical reprogramming. Shown are the top 25 most significant ontologies in each comparison (2c and 7c vs. controls, in young and old fibroblasts). 'Benjamini-Hochberg FDR < 0.1, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001. (B) Protein abundances of processes affected by partial chemical reprogramming. Log2 fold changes of proteins involved in alternative mRNA splicing (left), mitochondrial unfolded protein response (UPRmt, center), and RNA methylation (right). 'Benjamini-Hochberg FDR < 0.1, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001.
Figure 4
Correlation of gene expression and protein abundance changes with signatures of aging and OSKM reprogramming.
Spearman correlation of gene expression and protein abundance changes induced by partial chemical reprogramming (blue: 7c, green: 2c) with the signatures of OSKM reprogramming (cyan) and aging (red). Correlation coefficients ρ were calculated by the Spearman method based on the union of top 650 genes with the lowest p-value for each pair of signatures. Statistically significant pairwise correlations with Spearman ρ > 0.1 are labeled with asterisks. ***Benjamini-Hochberg FDR < 0.001.
Figure 5 with 1 supplement
Effect of partial chemical reprogramming on the phosphoproteome.
(A) Targeted GSEA of phosphorylation targets for four selected gene ontologies following partial chemical reprogramming. Bar lengths depict normalized enrichment scores. **Benjamini-Hochberg FDR < 0.01, ***FDR < 0.001. (B) Kinase signaling pathways. z-values for a kinase enrichment analysis are shown for the effects of partial reprogramming with 2c (horizontal axis) vs. 7c (vertical axis) in young (left) and old (right) fibroblasts. Signaling pathways significantly affected (Benjamini-Hochberg FDR < 0.05) by both 7c and 2c treatment (Prkaca) are colored in red, and signaling pathways affected only by 2c in young fibroblasts (Makp1 and Akt1) are colored in green (highlighted in bold for old fibroblasts). Horizontal and vertical red lines indicate the 0.05 Benjamini-Hochberg FDR significance threshold cut-offs for the comparisons 2c vs. control and 7c vs. control, respectively. (C) Role of Prkaca in partial chemical reprogramming. Left: knockdown of Prkaca during 7c partial chemical reprogramming in 20-month-old fibroblasts and effects on mitochondrial membrane potential, as assessed by TMRM fluorescence (n=3 independent biological replicates). p-values were determined by one-way ANOVA and Tukey’s post-hoc analysis. ns p ≥ 0.1, ***p < 0.001. Right: staining of 20-month-old fibroblasts for cellular localization of Tom20 (red) and Prkaca (green) during 7c partial chemical reprogramming. Representative images are shown, and data was collected for n=3 independent biological replicates. Scale bars: 20 µm.
-
Figure 5—source data 1
Differential expression analysis of TMT phosphoproteomics experiments following data normalization.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig5-data1-v1.xlsx
-
Figure 5—source data 2
Normalized change in TMRM fluorescence.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig5-data2-v1.xlsx
-
Figure 5—source data 3
Uncropped blots (image files) against GAPDH and Prkaca.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig5-data3-v1.zip
-
Figure 5—source data 4
Uncropped blots (labeled PDF file) against GAPDH and Prkaca.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig5-data4-v1.zip
Figure 5—figure supplement 1
Effect of partial chemical reprogramming on cellular signaling.
(A) GSEA of phosphopeptide abundances. GSEA of phosphorylated pathways significantly altered by partial chemical reprogramming. We show all ontologies with Benjamini-Hochberg FDR < 0.05 in at least one comparison (2c and 7c vs. controls, in young and old fibroblasts). 'Benjamini-Hochberg FDR < 0.1, *FDR < 0.05, **FDR < 0.01. (B) Knockdown of Prkaca and effects on mitochondrial membrane potential. Upper panels: western blot showing knockdown of Prkaca in 20-month-old fibroblasts relative to GAPDH levels using RNA interference (n=3). NC: negative control (non-targeting siRNA). p-value was determined by a two-tailed t-test. ***p < 0.001. Lower panels: representative images of 20-month-old fibroblasts stained with TMRM (red) and Hoechst 33342 (blue) following 7c partial chemical reprogramming and Prkaca knockdown. Scale bars: 31.1 µm (100 pixels). (C) Effect of 7c partial chemical reprogramming on cellular apoptosis. 20-month-old fibroblasts were treated with 7c or control vehicle for 1,3, or 6 days followed by staining with DAPI and Annexin V FITC. The percentage of cells undergoing apoptosis (FITC positive, DAPI negative) was quantified by flow cytometry (n=3 independent biological replicates). p-values were determined by a two-tailed t-test. ns p ≥ 0.1, ***p < 0.001.
Figure 6 with 1 supplement
Effect of partial chemical reprogramming on the metabolome.
(A) Hierarchical clustering (Ward’s method, Euclidean distance) of polar metabolite samples. Fibroblasts were treated for 6 days with 2c, 7c, or control followed by cell scraping and collection in cold 0.9% saline solution (n=3 independent biological replicates). Polar metabolites were isolated from the frozen cell pellets by chloroform-methanol extraction, and the upper polar phase was analyzed by hydrophobic interaction liquid chromatography (HILIC) coupled to a quadrupole mass spectrometer in both positive and negative ionization modes (Paynter et al., 2018; Poganik et al., 2023). All subsequent analyses were performed using MetaboAnalyst 5.0 (Xia et al., 2009). (B) Metabolites affected by partial chemical reprogramming. A total of 203 metabolites were identified by HILIC-positive and -negative methods, combined. Metabolite peak areas were normalized to the amount (µg) of protein in each cell pellet (determined by BCA assay) and log-transformed. In total, abundances of 109 metabolites were significantly altered by partial chemical reprogramming (Benjamini-Hochberg FDR < 0.05, colored in red). (C) Global effects of partial chemical reprogramming on the metabolome. Scaled heatmap of normalized, log-transformed metabolite abundances. (D) Abundances of top 50 metabolites significantly altered by partial chemical reprogramming with 2c or 7c. Scaled heatmap with labeled metabolites. (E) Abundance of metabolite-related proteins. Abundance of proteins related to S-adenosylhomocysteine biosynthesis (top) and xanthine metabolism (bottom) following partial chemical reprogramming with 2c and 7c vs. control in young and old cells. 'Benjamini-Hochberg FDR < 0.1, *FDR < 0.05, **FDR < 0.01, ***FDR < 0.001.
-
Figure 6—source data 1
HILIC-neg and HILIC-pos normalized metabolomics data.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig6-data1-v1.xlsx
Figure 6—figure supplement 1
Effect of age on fibroblast metabolome.
(A) PCA of metabolite samples. PCA was performed following normalization and log-transformation of metabolite peak areas. (B) PCA of untreated fibroblasts. PCA was performed following normalization and log-transformation of metabolite peak areas. (C) Volcano plots of metabolites affected by partial chemical reprogramming. Shown in red and blue are metabolites that are significantly elevated and decreased, respectively (FDR < 0.1 and |log2 fold change| ≥ 1). Shown in grey are metabolites that are unchanged (FDR ≥ 0.1, or |log2 fold change| < 1).
Figure 7 with 3 supplements
Effects of partial chemical reprogramming on biological age.
(A) Effects of partial chemical reprogramming on transcriptomic age (tAge). tAge was assessed by using mouse multi-tissue transcriptomic clocks to analyze the bulk RNA-seq data presented in Figure 2 (n=4 independent biological replicates). p-values were determined by one-way ANOVA and Tukey’s post-hoc analysis. ns p ≥ 0.1, ^p < 0.1, *p < 0.05, **p < 0.01. (B) Effects of partial chemical reprogramming on epigenetic age (DNAmAge). Levels of mean DNA methylation (DNAm) was assessed by DNAm microarray on the Horvath mammal 320k chip (n=5 independent biological replicates). p-values were determined by one-way ANOVA and Tukey’s post-hoc analysis. *p < 0.05, **p < 0.01, ***p < 0.001.
-
Figure 7—source data 1
Transcriptomic age predictions.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig7-data1-v1.xlsx
-
Figure 7—source data 2
Epigenetic age predictions.
- https://cdn.elifesciences.org/articles/90579/elife-90579-fig7-data2-v1.xlsx
Figure 7—figure supplement 1
Expression of pluripotency markers following OSKM reprogramming of fibroblasts.
Following doxycycline-induced expression of OSKM in young and old fibroblasts, fully-reprogrammed iPSC colonies were stained for pluripotency markers Sox2 (top), Oct4 (middle), and Nanog (bottom). Representative images obtained from reprogrammed old fibroblasts (left: Hoechst 33342, middle: pluripotency markers, right: merged images). Scale bars: 126.2 µm (100 pixels).
Figure 7—figure supplement 2
Aging-related changes in the fibroblast epigenome.
(A) Number of fibroblast CpG sites significantly differentially methylated with age. Shown in cyan and red are sites that experience an increase and decrease in mean methylation with age, respectively. (B) Effect of age on fibroblast mean methylation levels. p-value was determined by a two-tailed t-test. ns p-value ≥ 0.05. (C) Enrichment analysis of significantly differentially methylated positions (DMPs) that change with fibroblast age for specific chromatin states. down: DMPs that are hypomethylated with age. up: DMPs that are hypermethylated with age.
Figure 7—figure supplement 3
Effect of partial chemical reprogramming on the epigenome.
(A) PCA of DNAm microarray samples. PCA was performed after normalization of array data and filtering of probes. (B) Number of fibroblast CpG sites significantly differentially methylated with 2c and 7c treatment. Shown in cyan and red are sites that experience an increase and decrease in mean methylation, respectively. (C) Effect of partial chemical reprogramming on fibroblast mean methylation levels. p-values were determined by one-way ANOVA and Tukey’s post-hoc analysis. ns p-value ≥ 0.05, ***p-value < 0.001. (D) Enrichment analysis of significantly differentially methylated positions (DMPs) that change during partial chemical reprogramming for specific chromatin states. down: DMPs that are hypomethylated with partial chemical reprogramming. up: DMPs that are hypermethylated with partial chemical reprogramming.
Author response image 1
Additional files
-
Supplementary file 1
Raw genecounts.
Raw genecounts from mRNA-seq experiments obtained following STAR mapping.
- https://cdn.elifesciences.org/articles/90579/elife-90579-supp1-v1.xls
-
Supplementary file 2
Proteomics and phosphoproteomics datasets.
Normalized peptide and phosphopeptide abundances obtained from TMT 18plex experiments.
- https://cdn.elifesciences.org/articles/90579/elife-90579-supp2-v1.xlsx
-
Supplementary file 3
Splicing raw data.
Observed alternative splicing events.
- https://cdn.elifesciences.org/articles/90579/elife-90579-supp3-v1.xlsx
-
MDAR checklist
- https://cdn.elifesciences.org/articles/90579/elife-90579-mdarchecklist1-v1.docx
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Multi-omics characterization of partial chemical reprogramming reveals evidence of cell rejuvenation
eLife 12:RP90579.
https://doi.org/10.7554/eLife.90579.3
