COPI vesicle formation and N-myristoylation are targetable vulnerabilities of senescent cells

Abstract

Drugs that selectively kill senescent cells (senolytics) improve the outcomes of cancer, fibrosis and age-related diseases. Despite their potential, our knowledge of the molecular pathways that affect the survival of senescent cells is limited. To discover senolytic targets, we performed RNAi screens and identified coatomer complex I (COPI) vesicle formation as a liability of senescent cells. Genetic or pharmacological inhibition of COPI results in Golgi dispersal, dysfunctional autophagy, and unfolded protein response-dependent apoptosis of senescent cells, and knockdown of COPI subunits improves the outcomes of cancer and fibrosis in mouse models. Drugs targeting COPI have poor pharmacological properties, but we find that N-myristoyltransferase inhibitors (NMTi) phenocopy COPI inhibition and are potent senolytics. NMTi selectively eliminated senescent cells and improved outcomes in models of cancer and non-alcoholic steatohepatitis. Our results suggest that senescent cells rely on a hyperactive secretory apparatus and that inhibiting trafficking kills senescent cells with the potential to treat various senescence-associated diseases.

Fig. 1. RNAi screens identify senolytic targets.

a, Experimental design for the RNAi screens to identify senolytic targets. b, Right: quantification of cell survival of senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells three days post-transfection with BCL2L1 siRNA (n = 3). Left: representative DAPI-stained immunofluorescence (IF) images. Scale bar, 100 µm. c, Results of the primary siRNA screen for senolytic targets in OIS. Normalized cell counts are shown as mean B-score, reflecting counts normalized to account for plate positional effects using the B-scoring method. A candidate was considered a hit if the B-score in ≥2 replicates was <−3. d, Summary of the siRNA screen for senolytic targets in OIS, genes in the same pathway are indicated in bold. e, Re-test of OIS screen candidates. A candidate was considered a hit if the change in % cell survival was >20 with siRNAs in ≥2 replicates. f, Percentage cell survival in the context of OIS (4OHT) and control (DMSO) cells (n = 3). The data represent the deconvolution of values shown in e. g, Right: quantification of cell survival of doxorubicin-induced senescent (Doxo) and control (DMSO) IMR90 cells three days post-transfection with BCL2L1 siRNA (n = 3). Left: representative DAPI IF images. Scale bar, 100 µm. h, Results of the primary siRNA screen for senolytic targets in doxorubicin-induced senescence. Normalized cell counts are shown as mean B-score. A candidate is considered a hit if the B-score was <−3 in ≥2 replicates. i, Summary of the siRNA screen for senolytic targets in doxorubicin-induced senescence. j, Re-test of TIS screen candidates. A candidate was considered a hit if the change in % cell survival was >15 with siRNAs in ≥2 replicates. k, Percentage cell survival of doxorubicin-induced senescence (Doxo) and control (DMSO) cells (n = 6 for DMSO- and 4OHT-treated cells, n = 3 for BCLXL siRNA transfected cells). Data represent the deconvolution of values shown in j with additional replicates. l, Common pathways identified in the siRNA screen for senolytic targets. Data in b, f, g and k are presented as mean ± s.d. (unpaired, two-tailed Student’s t-test). n represents independent experiments in b, f, g, k. Data in c and h is representative of three replicates. Data are presented as percentage cell survival in control cells versus the difference in cell survival between control and senescent cells in e and j. Source numerical data are available as source data. Source data

Fig. 2. COPI is a vulnerability of senescent cells.

a, Right: percentage cell survival of the indicated senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells (n = 4, IMR90 Vector+4OHT; n = 5, other groups). Left: representative DAPI-stained IF images. Scale bar, 100 µm. b, Crystal violet staining. An image representative of three independent experiments is shown. c, Senolytic activity of COPB2 in doxorubicin-induced senescence in IMR90 cells (n = 4 shCOPB2.1, n = 5 other shRNAs). d, Senolytic activity of COPG1 depletion during OIS in IMR90 ER:RAS cells (n = 5, IMR90 Vector+4OHT; n = 6, other groups). e, Schematic outlining the strategy of co-culture senolytic testing of COPB2 siRNAs. f, Right: percentage survival in a co-culture experiment of IMR90 green fluorescent protein (GFP) ER:RAS with IMR90 Cherry cells transfected with the indicated siRNAs. Cell numbers were determined from counts of mCherry or GFP-positive cells detected by IF (n = 3). Left: representative IF images. Scale bar, 100 µm. g, Representative images of IMR90 ER:RAS cells seven days post addition of 4OHT and stained for Senescence-Associated (SA)-β-Gal activity 72 h after treatment with 2.5 µM GCA (n = 3). Quantification is shown in Extended Data Fig. 2a. Scale bar, 100 µm. h, Dose–response curves for the senolytic effect of the GBF1 inhibitor GCA in the context of OIS (n = 6). Red, IMR90 ER:RAS +4OHT; Black, IMR90 vector +4OHT; Blue, IMR90 ER:RAS +DMSO. i, Percentage survival of control cells (RAS DMSO) and oncogene-induced senescent cells (RAS 4OHT) transduced with vectors and treated with ABT-263, GCA or BFA (n = 5 for GCA/BFA; n = 3, ABT-263). j, Caspase-3/7 activity in control (DMSO) or senescent (4OHT) cells after treatment with DMSO or 2.5 µM GCA seven days after senescence induction (n = 3). k, Senolytic activity of GCA in senescence induced by irradiation (n = 3), bleomycin and DMSO (n = 6), and doxorubicin (n = 4). Data are presented as mean ± s.d. Comparisons to the corresponding DMSO-treated cells (grey bars) with two-way analysis of variance (ANOVA). l, Right: percentage cell survival of p16^INK4a positive and negative cells in PBECs after treatment with GCA or vehicle (DMSO). Left: representative p16^INK4a (green)-stained IF images. Scale bar, 50 μm. n = 3. Data in all figures are presented as mean ± s.d. n represents independent experiments throughout the figure. Unpaired, two-tailed, Student’s t-test was used unless otherwise stated. Source numerical data are available as source data. Source data

Fig. 3. COPB2 depletion causes Golgi disruption and triggers the UPR in senescent cells.

a, GSEA plot for COPI transport in cells undergoing OIS. NES, normalized enrichment score; FDR, false discovery rate. b, Right: percentage of dispersed trans-Golgi by IF in senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells transfected with the indicated siRNAs (n = 3). Quantification was performed using organelle count (Methods). Left: representative IF images. The white arrow points to a cell with a normal trans-Golgi, and the yellow arrow indicates a cell with dispersed trans-Golgi. Scale bar, 100 µm. c, Right: percentage of dispersed cis-Golgi by IF in senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells transfected with the indicated siRNAs (n = 3). Quantification was performed using the integrated intensity threshold (intensity × area) in the ‘region growing’ collar. Left: representative IF images. Normal cis-Golgi (white arrow) and dispersed cis-Golgi (yellow arrow) are indicated. Scale bar, 100 µm. d, Right: quantification of intracellular levels of IL-8 in senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells after transfection with the indicated siRNAs by measurement of the pixel intensity coefficient of variance (CV) within cytoplasmic collar (n = 4). Left: representative IF images. Scale bar, 100 µm. Statistical tests were performed using two-way ANOVA relative to DMSO-treated cells. e, SASP inhibition caused by treatment with 10 µM glucocorticoids (Bec, beclomethasone; Tri, triamcinolone) prevents senolysis induced by COPB2 depletion. Quantification of cell survival is shown for senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells treated as indicated (n = 3). f, GSEA plot showing that an UPR gene signature is enriched in IMR90 ER:RAS upon COPB2 depletion. g,h, Right: percentage of cells positive for nuclear CHOP (g, n = 5) and nuclear ATF6 (h, n = 4) by IF six days after treating with 4OHT (to induce OIS) or DMSO (as control) for cells transfected with the indicated siRNAs. Staining was performed 72 h later. Left: representative IF images. Scale bar, 100 µm. Unpaired two-tailed Student’s t-test was used for statistical comparison in g. All data are presented as mean ± s.d. n represents independent experiments throughout. Statistical tests were performed using two-way ANOVA against scrambled siRNA unless otherwise stated. Source numerical data are available as source data. Source data

Fig. 4. COPI inhibitors cause Golgi disruption, trigger UPR and result in autophagy defects.

a–c, Percentage fragmented trans-Golgi (a, right), early endosome numbers per cell (b) and intracellular levels of IL-8 (c) by IF in senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells after treatment for 48 h with 1.25 µM GCA, 150 nM BFA or 1 µM ABT-263 (n = 3). Representative IF images are shown in a (left) and Extended Data Fig. 4a (for b) and Supplementary Fig. 7a for c. Scale bar, 100 μm. d, Percentage cell survival of senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells treated with either 1.25 µM GCA or 150 nM BFA (n = 3) following glucocorticoid (10 µM Bec, beclomethasone; 10 µM Tri, triamcinolone) pretreatment four days after senescence induction. e, Right: relative Proteostat signal intensity in senescent (4OHT) or control (DMSO) IMR90 ER:RAS cells after 48 h treatment with 1.25 µM GCA or 150 nM BFA (n = 3). Left: representative IF images. Scale bar, 100 µm. f,g, Right: percentage positive cells for nuclear CHOP (f) or LC3B foci number (g) by IF, 48 h after either control (DMSO) or senescent (4OHT) cells were treated with either 1 µM ABT-263, 1.25 µM GCA or 150 nM BFA (n = 3). Left: representative IF images for CHOP (f) and L3CB (g). Scale bars, 100 µm. h,i, Percentage p62/SQSTM1 (h) and ATF6 (i) positive cells by IF in senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells treated with either 1 µM ABT-263, 1.25 µM GCA or 150 nM BFA following glucocorticoid pretreatment as in d. IF staining was carried out 48 h post senolytic drug addition (n = 3). j, Survival of control (DMSO) or OIS (4OHT) cells pre-treated with 1 µM GSK2656157 or 1 µM GSK2606414 before a 48 h treatment with GCA or BFA at day 7 post senescence induction (n = 3). k, Scheme summarizing how COPI inhibition induces the death of senescent cells. All data throughout the figure are presented as mean ± s.d. n represents independent experiments throughout the figure. Two-way ANOVA was performed for statistical analysis in a, c–e, g, h, i and j. Unpaired, two-tailed Student’s t-test was used for statistical analysis of the data in b and f. Source numerical data are available as source data. Source data

Fig. 5. Therapeutic benefits of inhibiting the COPI pathway.

a, Experimental design for the sequential treatment of cancer cells with chemotherapy and GCA. Etop., etoposide. b,c, Quantification of cell survival of A549 cells (b) or SKHep1 cells (c) after treatment with the indicated drug combinations (n = 6). Unpaired, two-tailed, Student’s t-test. Data are presented as mean ± s.d. d, Experimental design of tumour growth in mice co-injected with 5PT squamous cancer cells and HFFF2 fibroblasts. IR, irradiation. e, Tumour growth curves showing the tumour volume monitored over time. Data are presented as mean ± s.e.m. for all mice in each group (n = 7 mice per group, shCOPB2.1 + IR, n = 6 mice). Repeated Measure (RM) two-way ANOVA with Greenhouse–Geisser correction and Dunnett’s correction was used for statistical analysis of the day-20 timepoint relative to shControl+IR. The AUC analysis for data pooled from two experiments is shown in Extended Data Fig. 5d. All comparisons are to shControl+IR. f, Experimental design of the mouse model of lung fibrosis by intratracheal instillation of human senescent lung fibroblasts into nude mice. All analyses were performed three weeks after cell delivery (except those in Extended Data Fig. 5h, which were performed 48 h post-instillation). g–j, Relative expression of the mRNAs coding for human CDKN2A (g), or mouse Cdkn1a (h), Col3a1 (i) and Pai1 (j) in lung samples from the experiment described in f (n = 5 mice per group). Statistical analysis was performed using ordinary one-way ANOVA. Data are presented as mean ± s.d. k, Lung hydroxyproline content in samples from mice of the experiment described in f (n = 5 mice per group). Ordinary one-way ANOVA. Data are presented as mean ± s.d. l, Ashcroft scoring for alveolar septal thickening in sections from lungs of mice grafted with IMR90 cells treated as indicated (n = 5 mice per group). Ordinary one-way ANOVA. Data are presented as mean ± s.d. m, Representative images of lung sections stained with haematoxylin and eosin (H&E, top) and Masson’s trichrome (bottom) from mice of the experiment described in f. Scale bar, 100 µm. n represents independent experiments or mice throughout the figure. Source numerical data are available as source data. Source data

Fig. 6. NMTi phenocopy COPI inhibition and are senolytic.

a, Western blots of ARF GTPases for control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells 72 h after treatment with 300 nM IMP1088 or 1.5 µM DDD86481, seven days after senescence induction. An immunoblot of GAPDH is included as a loading control. Representative immunoblots from three independent experiments are shown. b–d, Right: quantification of IF staining for trans-Golgi (TG) dispersal (TGN46, b), cis-Golgi dispersal (GM130, c) and intracellular levels of IL-8 (d) in control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells five days after treatment with 300 nM IMP1088 or 1.5 µM DDD86481, seven days after senescence (n = 3). Left: representative IF images. Scale bars, 100 µm. e, GSEA plot of the UPR gene signature in IMR90 ER:RAS treated with the NMTi IMP1088. f, Right: quantification of XBP1 IF staining in control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells treated for five days with 300 nM IMP1088 or 1.5 µM DDD86481 seven days after senescence induction (n = 3). Left: representative IF images. Scale bar, 100 µm. g, Quantification of IF staining for p62/SQSTM1. Control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells were treated with 300 nM IMP1088 or 1.5 µM DDD86481 seven days after senescence induction for five days (n = 3). h–j, Dose–response curves of control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells treated for seven days with NMT inhibitors, seven days after senescence induction, with IMP1088 (h, n = 9), DDD86481 (i, n = 5) and IMP1320 (j, n = 4). k, Right: quantification of dispersed trans-Golgi in C64A or WT IpaJ transduced control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells (n = 3) seven days post senescence induction. Left: representative IF images. Scale bar, 50 μm. l, Percentage survival of C46A or WT IpaJ transduced control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells seven days post senescence induction. Survival is measured relative to vector-transduced cells. Unpaired two-tailed Student’s t-test (n = 3). Data are presented as mean ± s.d. Statistical analysis was performed throughout the figure by ordinary two-way ANOVA unless otherwise specified. n represents independent experiments throughout. Source numerical data and unprocessed blots are available as source data. Source data

Fig. 7. NMTi target senescent cells in cancer models.

a, Experimental design for the sequential treatment of cancer cells with chemotherapy and NMTi. b, Crystal violet staining of control (DMSO) and senescent (treated with doxorubicin or etoposide) HCT116 cells treated with 300 nM IMP1088 for seven days, seven days after senescence induction. The images show the results of two independent experiments. c,d, Dose–response curves in HCT116 (c, n = 4) or MCF7 (d, n = 3 for DMSO and etoposide and n = 2 for doxorubicin) cells treated with either doxorubicin or etoposide and treated with IMP1088 seven days post senescence induction. Data are presented as mean ± s.d. e, Experimental design for f. f, AUC analysis for tumour volume measured over time. Data are presented as mean ± s.e.m. (n = 6 mice, 5PT+veh; n = 8, other groups; also Extended Data Fig. 7e). Ordinary one-way ANOVA. g, Tumoural pituitaries from 18.5dpc Hesx1^Cre/+;Ctnnb1^lox(ex3)/+ embryos were cultured in the presence of NMTi (600 nM IMP1088) or vehicle (DMSO) and processed for histological analysis after 72 h. h, Left: quantification of β-catenin-accumulating cells after NMTi treatment. Right: images with representative IF staining with β-catenin and cleaved caspase-3 (CC3). Main scale bars, 50 μm; inset scale bars, 40 μm. Data are presented as mean ± s.d. n represents the total number of histological sections analysed (n = 22, DMSO; n = 15, NMTi). Unpaired, two-tailed Student’s t-test. i, Quantification of CC3-positive area (percent of the pituitary surface) after NMTi treatment (n = 12 pituitary sections per group). Data are presented as mean ± s.d. Unpaired, two-tailed, Student’s t-test. j, Experimental design of the liver oncogene-induced senescence experiment. k–m, Left: quantification of Nras-positive cells (k), SA-β-Gal staining (l) and p21^CIP1 staining by immunohistochemistry (IHC) (m) in the liver of mice treated with vehicle or IMP1320 (n = 9 mice per group). Data are presented as mean ± s.e.m. Unpaired, two-tailed, Student’s t-test. Right: representative IHC images (k–m). Arrows indicate examples of SA-β-Gal-positive cells. Scale bars, 100 μm. n represents independent experiments or mice unless otherwise specified. Source numerical data are available as source data. Source data

Fig. 8. NMTi eliminate senescent cells and improve NASH-induced liver steatosis and fibrosis.

a, Experimental design for the model of WD-induced NASH. b, Quantification of blood serum levels of cholesterol and ALT in normal, WD mice treated with vehicle (Chow+veh, n = 15, WD+veh, n = 14) or WD mice treated with DDD86481 (WD+NMTi, n = 15). Ordinary one-way ANOVA. c,d, Representative images (c) and quantification (d) of p21^CIP1 staining of liver sections. Yellow arrows in c indicate examples of p21^CIP1-positive cells. Scale bar, 50 μm. Chow+veh, n = 15; WD+veh, n = 14; WD+NMTi, n = 15. e, GSEA plot showing that a senescence signature is downregulated in WD-fed mice treated with NMTi. f–h, Representative images (f) of H&E (top), Oil Red O (middle) (chow+veh, n = 15; WD+veh, n = 14; WD+NMTi, n = 15) and CD68 IHC (bottom) stained liver sections (chow+veh, n = 14; WD+veh, n = 13; WD+NMTi, n = 14) (scale bars, 50 µm (H&E), 20 µm (Oil red O and CD68)) and quantification of Oil Red O staining (g) and CD68 staining (h). i,j, Representative images of Picrosirius Red-stained liver sections (i) and quantification (j). Scale bar, 50 μm. Chow+veh, n = 15; WD+veh, n = 14; WD+NMTi, n = 15. k,l, Levels of Col1a1 (k) and Col4a1 (l) mRNA from bulk liver extracts (chow+veh, n = 15; WD+veh, n = 14; WD+NMTi, n = 15). m,n, GSEA plots showing that senescence signature of Kupffer cells (m) and collagen formation (n) are downregulated in WD-fed mice treated with NMTi. Data are presented as mean ± s.e.m. Ordinary one-way ANOVA. n represents number of mice throughout the figure. Source numerical data are available as source data. Source data

Extended Data Fig. 1. Models for oncogene- and therapy-induced senescence.

a, IMR90 ER:RAS model for oncogene-induced senescence (OIS). Addition of 4-OHT activates ER:RAS, triggering OIS. b, Quantification of IF staining for BrdU incorporation in IMR90 ER:RAS cells 5 days after treatment with 4-OHT or DMSO (n = 3). c, Percentage of cells staining positive for SA-β-galactosidase activity in IMR90 ER:RAS cells 5 days after treatment with 4-OHT or DMSO (n = 3). Representative brightfield image (right). Scale bar, 100 µm. d, Percentage of cells staining positive for p16^INK4a by IF in IMR90 ER:RAS cells 5 days after treatment with 4-OHT or DMSO (n = 3). Representative IF image (right). Scale bar, 100 µm. e, B-score normalization of cell numbers shown for the siRNA screen performed on control, non-senescent (DMSO treated) IMR90 ER: RAS cells. Points show normalized values for 3 technical replicates. Red box indicates samples with a B-score < 3. f, Screen results for control (DMSO-treated IMR90 ER:RAS) and OIS (4OHT-treated IMR90 ER:RAS). Graph displays B-score in OIS versus the negative of the difference in B score between OIS and control screens. Points show mean normalized values for 3 technical replicates with cut-offs shown for OIS B-score < -3 and a difference in B score of > 2 between control and OIS cells. g, Model of therapy-induced senescence (TIS). Senescence was induced by 7 days of doxorubicin treatment in IMR90 cells. h, Quantification of IF staining for BrdU incorporation in IMR90 cells 2 days after treatment with doxorubicin or DMSO (n = 3). i, Percentage of IMR90 cells staining positive for SA-β-galactosidase activity 7 days treating with doxorubicin or DMSO (n = 3). Representative brightfield images (right). Scale bar, 100 µm. j-k, Percentage of IMR90 cells staining positive for p21^CIP1 (j) or p16^INK4a (k), 6 (j) or 3 (k) days by IF after treating with doxorubicin or DMSO (n = 3). Representative IF image (right, j & k). Scale bar, 100 µm. Data represented as mean ± SD unless otherwise stated. Statistical tests performed using unpaired two-tailed t-test unless. N represents independent experiments. Source numerical data are available in source data. Source data

Extended Data Fig. 2. COPI inhibitors are senolytic.

a-c, IMR90 ER:RAS cells were treated 7 days after senescence induction with either DMSO (D), 2.5 µM golgicide A (G), or 150 nM brefeldin A (B) for 72 h, and cells fixed and stained as indicated. Percentage of either total cells or cells positive for SA-β-galactosidase (a), p16^INK4a (b), and p21^CIP1 (c) relative to the total number of DMSO-treated cells. (n = 3 replicates). Two-way ANOVA. The significance of total cell comparisons is shown in black, while the significance for comparisons of positive for SA-β-galactosidase (a), p16^INK4a (b), and p21^CIP1 (c) are shown in blue, green, and red respectively. Percentage cells represent a fraction of cells compared to the total number of cells in the DMSO-treated sample for each replicate. d, IMR90 ER:RAS cells were transfected with the indicated siRNAs 7 days after treatment with 4OHT (senescent) or DMSO (controls). Cells were fixed and stained for SA-β-galactosidase activity 72 h after transfection. Representative images (left). Scale bar, 100 µm. Quantification (right) of a relative number of senescent cells. Shown are the percentage of total or SA-β-gal-positive cells relative to the total number of DMSO-treated cells. (n = 3). Scr. scrambled. Two-way ANOVA. Percentage cells represent a fraction of cells compared to the total number of cells in the DMSO-treated sample for each replicate. e, Dose-response curves for senolytic effect of GBF1 inhibitor brefeldin A in the context of OIS (Dose 20 µM – 40 nM, n = 6, 20 nM – 40pM, n = 3 IMR90 vector + 4OHT, n = 4 IMR90 ER:RAS + DMSO and +4OHT). EC₅₀, half maximal effective concentration. f, Senolytic activity of brefeldin A in the context of senescence induced by irradiation (n = 3, n = 2 for ABT-263 and 40 nM BFA treatment), bleomycin (n = 6) or doxorubicin (n = 4). Data represented as mean ± SD. Comparisons are to the corresponding DMSO-treated cells (gray bars) using two-way ANOVA. Data represented as mean ± SD throughout figure. N represents independent experiments throughout figure. Source numerical data are available in source data. Source data

Extended Data Fig. 3. Effects of COPB2 depletion on senescent cells.

a, Heatmap derived from RNA-Seq data showing the relative mRNA levels of COPI components (in black) in senescent cells. As a reference, relative levels of MMP3 and CXCL8 (upregulated during senescence in red) and MKI67 and COL15A1 (downregulated during senescence, in blue) are included. Represented are the log₂Fold change (FC) in oncogene-induced senescent (RAS) or therapy-induced senescent (BLEO) IMR90 ER:RAS or IMR90 cells transduced with shRNA vector, induced to senesce for 10 days and measured relative to respective cell lines treated with DMSO (n = 3). b, GSEA plot of COPI-mediated transport signature in bleomycin treated IMR90 10 days post senescence induction. NES, normalized enrichment score; FDR, false discovery rate. c, Quantification of intracellular levels of IL6 in senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells after transfection with the indicated siRNAs (n = 4). Representative IF image (left) Scale Bar, 100 µm. Statistical test performed using two-way ANOVA. d, Heatmap derived of RNA-Seq data showing the relative mRNA levels of SASP components in the indicated cells. Data represented as row z-score normalized. (n = 3). e, Quantification of percentage nuclear XBP1 positive by IF in senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells after transfection with the indicated siRNAs (n = 3). Representative IF image (left). Scale Bar, 100 µm. Unpaired two-tailed, student’s t-test. Data is represented throughout the figure as mean ± SD unless otherwise stated. N represents number of independent experiments throughout figure. Source numerical data are available in source data. Source data

Extended Data Fig. 4. Effects of COPI inhibition in the unfolded protein response and autophagy.

a, Representative IF images of EEA1 staining in control (DMSO) and senescent (4OHT) IMR90 ER:RAS treated with either DMSO, 1.25 µM of golgicide A (GCA) or brefeldin A (BFA) for 48 h, 7 days after senescence induction. (n = 3) (Quantification provided in Fig. 4b) Scale bar, 100 µm. b, Percentage cell survival of senescent (4OHT) and control (DMSO) IMR90 ER:RAS cells transduced with indicated shRNA or vector and treated with either ABT-263, 1.25 µM GCA or 150 nM BFA for 72 h, 7 days after senescence induction (n = 4). Ordinary Two-way ANOVA with comparison is to the corresponding IMR90 ER: RAS vector +4OHT condition. c-d, Percentage cells positive for nuclear XBP1 (c) and ATF6 (d) by IF, 48 hours after either control (DMSO) or oncogene-induced senescent (4OHT) cells were treated with either 1 µM ABT-263, 1.25 µM GCA or 150 nM BFA (n = 3). Representative IF images (left, c and d). Scale Bar, 100 µm. Significance was calculated using unpaired, two-tailed, Student’s t-test. e-f, Quantification (e) of the level of BiP protein (relative to α-tubulin) as assessed by western blot of either control (DMSO) or senescent (4OHT) 48 h after treatment with either 1.25 µM GCA or 150 nM BFA (n = 3) and 7 days after senescence induction. Unpaired, two-tailed, Student’s t-test. Representative immunoblot (out of three independent experiments) (f) shown for BiP, α-tubulin, and LC3 is shown. Immunoblot of α-tubulin is included as a loading control. g, Representative IF images for p62/SQSTM1 immunostaining are shown for control (DMSO) and senescent (4OHT) IMR90 ER: RAS treated with 1.25 µM GCA for 48 h 7 days post senescence induction (Quantification provided in Fig. 4h) (n = 3). Scale bar, 100 µm. Data throughout figure represented as mean ± SD where applicable. N represents independent experiments throughout figure. Source numerical data and unprocessed blots are available in source data. Source data

Extended Data Fig. 5. Therapeutic benefits of inhibiting the COPI pathway.

a, Experimental design for the sequential treatment of cancer cells with chemotherapy and brefeldin A (BFA). b-c, Quantification of cell survival of A549 cells (b, n = 6) or SKHep1 cells (c, n = 6) after treatment with the indicated drug combinations. Unpaired, two-tailed, Student’s t-test. Data represented as mean ± SD. d, Area under the curve (AUC) analysis for tumour volume measured over time in two independent experiments (Experiment A and B; see Fig. 5e for the tumour growth curves of Experiment A and Supplemental Figure 12f for Experiment B). Data represented as mean ± SEM (n = 14 mice per group, Vector, Vector + Irrd and shCOPB2.2 + Irrd; n = 13 mice, shCOPB2.1 + Irrd; n = 7 mice per group shCOPA.1 + Irrd and shCOPA.2 + Irrd). Ordinary two-way ANOVA was used for the statistical comparison of groups to Vector + Irrd. e, Experimental design (left) of tumour growth in NSG cancer model with 5PT squamous cancer cells co-injected with HFFF2 fibroblasts (Right) Tumour growth curves showing the tumour volume monitored over time (IR=irradiation). Data represented as mean ± SEM (n = 8 mice per group, 5PT n = 6 mice). RM Two-way ANOVA with Geisser Greenhouse correction and Dunnet’s correction used for comparisons to 5PT + HFFF2 shControl used for the statistical test. P values are shown for the final time point. f, Representative images (of three independent experiments) of SA-β-galactosidase staining in irradiated IMR90s transduced with shRNAs against COPB2 with positive cells staining blue. Scale bar, 200 µm. g, Quantification of the initial engraftment of human fibroblasts in the lungs of nude mice measured 48 hours after the instillation of 5×10⁵ of the indicated cells. Engraftment was assessed by the expression levels of mRNAs coding for human MMP3. Ordinary one-way ANOVA and (shCntl, n = 3 mice, shCOPB2.1, n = 3 mice, shCOPB2.2, n = 3 mice). Data represented as mean ± SD. N in figure (b, c, e, g) represents independent experiments or mice. N in figure (d) represents mice from two pooled experiments (Experiment A and Experiment B) Source numerical data are available in source data. Source data

Extended Data Fig. 6. NMT inhibitors phenocopy the effects of inhibiting COPI in senescent cells.

a, Quantification of IF staining for EEA1 vesicles in control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells were treated with 300 nM IMP1088 or 1.5 µM DDD86481 for 5 days, 7 days after senescence induction (n = 3). Ordinary Two-way ANOVA. Representative IF image (left). Scale bar, 100 µm. b, Quantification of intracellular levels of IL6 in control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells treated with 300 nM IMP1088 or 1.5 µM DDD86481 for 5 days, 7 days after senescence induction (n = 3). Ordinary Two-way ANOVA. Representative IF images (left). Scale bar, 100 µm. c, Relative fold change of IL6, IL8, G-CSF, and CCL2 in the supernatant of (DMSO) or day 7 senescent cells (4OHT + DMSO) treated with NMTi (4OHT + IMP1088/DDD8641) for 5 days. Concentration determined by ELISA normalized to cell counts and shown relative to senescent cells (4OHT + DMSO) (n = 3). Comparisons are to senescent cells (4OHT + DMSO). Significance was calculated using ordinary two-way ANOVA, Dunnett’s correction. d-e, Percentage of cells positive nuclear ATF6 (d) and nuclear CHOP (e) by IF in control (DMSO) or senescent (4OHT) IMR90 ER:RAS cells treated with 300 nM IMP1088 or 1.5 µM DDD86481 for 5 days, 7 days after senescence induction (n = 3). Ordinary two-way ANOVA, Dunnett’s correction. Representative IF images are shown (left). Scale bar, 100 µm. f-g, Dose-response curves for senolytic effect of NMT inhibitors in bleomycin-induced senescence. IMP1088 (f, n = 7), DDD86481 (g, n = 5). EC₅₀, half-maximal effective concentration. Control (DMSO) or senescent (Bleomycin) IMR90 was treated with inhibitors 7 days after senescence induction and fixed for assessing survival 7 days after treatment. Data is represented as mean ± SD throughout the figure. N throughout figure represents independent experiments. Source numerical data are available in source data. Source data

Extended Data Fig. 7. NMT inhibitors target senescent cells in cancer models.

a-b, Dose-response curves for assessing the senolytic activity of DDD86481 treated HCT116 (a, n = 4 for each condition) or MCF7 (b, n = 3 for DMSO, n = 3 for etoposide, n = 2 for doxorubicin) cells induced to senesce with either doxorubicin or etoposide. Data represented as mean ± SD c-d, Dose-response curves for assessing the senolytic activity of IMP1320 treated HCT116 (c, n = 4 for each condition) or MCF7 (d, n = 3 for DMSO, n = 3 for etoposide, n = 2 for doxorubicin) cells induced to senesce with either doxorubicin or etoposide treatment. EC₅₀, half-maximal effective concentration. Data represented as mean ± SD. e, Tumour growth curves showing the tumour volume monitored over time (IR=irradiation). Data represented as mean ± SEM (n = 6 mice 5PT + Veh, n = 8 mice per group all other groups; see the Fig. 7f for the relative AUC analysis. RM Two-way ANOVA with Geisser Greenhouse correction and Dunnet’s correction used for to 5PT/IR HFFF2 + vehicle used for the statistical test. P values are shown as the final timepoint. f-g, Quantification of β-catenin positive (f) and p21^Cip1 positive/ β-catenin positive (g) cells in the pituitary gland at 0, 24, 48, and 72 h after treatment with 600 nM IMP1088 ex vivo. Data is box and whisker plot; box, 25^th to 75^th percentile; whisker minimum to maximum with all points shown; n represents the number of sections. (f, 0 h, n = 42; 24 h, n = 19; 48 h, n = 30; 72 h, n = 15) (g, 0 h, n = 42; 24 h, n = 19; 48 h, n = 30; 72 h, n = 21) Ordinary One-way ANOVA with Dunnett’s correction. h, Synaptophysin is a marker of the normal hormone-producing cells in the pituitary gland, quantification of synaptophysin-positive area (purple; % of the pituitary surface) after NMTi treatment highlights no significant effect of the treatment on normal cells. The left panel shows representative images (n = 6 sections for all groups). Unpaired, two-tailed, Student’s t-test. Scale bar, 50 µm. N in figure (a-e) represent independent experiments or mice. N in figure (f and g, h) represents histological sections. Source numerical data are available in source data. Source data

Extended Data Fig. 8. The therapeutic benefit of NMTi in a model of bleomycin-induced lung fibrosis.

a, Experimental design of the mouse model of bleomycin-induced lung fibrosis. All analyses were performed at 28 days after treatment with bleomycin (intratracheal 0.75 U/kg) into the lungs of 6-8 weeks old, C57BL/6 J male mice. b, Lung hydroxyproline content in samples from mice of the experiment described here. (n = 6 mice, vehicle; n = 8 mice, NMTi). c-j, Relative expression of the indicated mRNAs in lung samples (n = 6 mice, vehicle; n = 8 mice, NMTi). Data represented as mean ± SD throughout figure. Unpaired two-tailed t-test used throughout figure. N represents mice throughout figure. Source numerical data are available in source data. Source data

Extended Data Fig. 9. Treatment with NMTi does not induce toxicities in a model of bleomycin-induced lung fibrosis.

a-h, Bilirubin (a), ASAT/GOT (b), ALAT/GPT (c), LDH (d), Creatine (e), Urea (f), BUN (g) and CK-MB (h) were measured in the blood of mice subjected to the bleomycin-induced lung fibrosis experiment (described in Extended Data Fig. 8), showing no significant increase in toxicity associated with the NMTi treatment. Dotted lines indicate normal range of the different metabolites. (n = 6 mice, vehicle; n = 8 mice, NMTi). Unpaired, two-tailed t-test. Data represented as mean ± SD. Source numerical data are available in source data. Source data

Extended Data Fig. 10. NMT inhibitors eliminate senescent cells and improve NASH-induced liver steatosis and fibrosis.

a, Body weight of chow-fed (n = 15 mice) and western diet (WD)-fed mice treated with vehicle (n = 14 mice) or NMT inhibitor (n = 15 mice) following the last round of treatment. Ordinary One-way ANOVA. Dunnett’s test. b-f, GSEA plots showing the enrichment of the indicated signatures in mice fed WD (WD + veh) as compared with mice fed with chow diet (Chow + veh; b, d, f) or in mice fed with WD and treated with NMTi (WD + NMTi) as compared with mice fed with WD and treated with vehicle (WD + vehicle; c, e). NES, normalized enrichment score; FDR, false discovery rate. Source numerical data are available in source data. Source data