Research Article

Tissue damage and senescence provide critical signals for cellular reprogramming in vivo

See allHide authors and affiliations

Science  25 Nov 2016:
Vol. 354, Issue 6315, aaf4445
DOI: 10.1126/science.aaf4445

For cell reprogramming, context matters

Differentiated cells in a culture dish can assume a new identity when manipulated to express four transcription factors. This “reprogramming” process has sparked interest because conceivably it could be harnessed as a therapeutic strategy for tissue regeneration. Mosteiro et al. used a mouse model to study the signals that promote cell reprogramming in vivo. They found that the factors that trigger reprogramming in vitro do the same in vivo; however, they also inflict cell damage. The damaged cells enter a state of senescence and begin secreting certain factors that promote reprogramming, including an inflammatory cytokine called interleukin-6. Thus, in the physiological setting, cell senescence may create a tissue context that favors reprogramming of neighboring cells.

Science, this issue p. 10.1126/science.aaf4445

Structured Abstract

INTRODUCTION

The ectopic expression of transcription factors OCT4, SOX2, KLF4, and cMYC (OSKM) enables reprogramming of adult differentiated cells into pluripotent cells, known as induced pluripotent stem cells (iPSCs), that are functionally equivalent to embryonic stem cells. Expression of OSKM in vivo leads to widespread cell dedifferentiation and reprogramming within tissues and eventually to the formation of teratomas (tumors arising from iPSCs). The molecular mechanisms operating during in vitro OSKM-driven reprogramming have been extensively characterized; however, little is known about in vivo reprogramming.

RATIONALE

The process of OSKM reprogramming is inefficient both in vitro and in vivo. A number of cell-intrinsic barriers have been identified in vitro, most of which are activated by cellular damage and are particularly prominent in aged cells. Mechanistically, these cell-intrinsic barriers for reprogramming are primarily mediated by the tumor suppressors p53, p16INK4a, and ARF (the latter two are encoded by the Ink4a/Arf gene locus). In this work, we have investigated the effect of these tumor suppressors, cellular damage, and aging on in vivo reprogramming.

RESULTS

We found that the expression of OSKM in vivo not only triggers reprogramming of some cells but also inflicts extensive damage on many other cells, driving them into a state known as cellular senescence. Senescent cells are characterized by their inability to proliferate and by their secretion of inflammatory cytokines. We have observed a positive correlation between senescence and OSKM-driven reprogramming. For example, tissues lacking p16INK4a/ARF do not undergo senescence, and their ability to reprogram is severely compromised. By contrast, in tissues lacking p53, damage is rampant; this leads to maximal levels of senescence, exacerbated cytokine production, and increased in vivo reprogramming.

To explore the connection between senescence and reprogramming, we manipulated these processes in vivo through pharmacological interventions. In particular, an increase in senescence produced by palbociclib (a drug that functionally mimics p16INK4a) results in higher levels of reprogramming. Conversely, a reduction in senescence achieved by navitoclax (a proapoptotic drug with selectivity against senescent cells) leads to decreased in vivo reprogramming. We found that the cross-talk between senescence and reprogramming is mediated by the cytokine-rich microenvironment associated with senescent cells. This is based, among other evidence, on the observation that pharmacological inhibition of NFκB, a major driver of cytokine production, reduces in vivo reprogramming. Analysis of the inflammatory cytokines produced by senescent cells, both in vivo and in vitro, led us to identify interleukin-6 (IL-6) as a critical secreted factor responsible for the ability of senescent cells to promote reprogramming. In support of this, blockade of IL-6 or its downstream kinase effector PIM potently reduced in vivo reprogramming. These observations can be recapitulated in vitro, where reprogramming efficiency is strongly enhanced by the presence of damaged cells or by the conditioned medium derived from damaged cells. Moreover, immunodepletion of IL-6 from the conditioned medium abolished reprogramming.

Having established that senescence promotes reprogramming, we studied whether tissue injury leading to senescence has a positive effect on OSKM-driven reprogramming. In particular, we show that bleomycin-induced tissue damage strongly promotes reprogramming in the lung. Finally, aging, which is associated with higher levels of cellular senescence, also favors OSKM-driven reprogramming both in progeric and in physiologically aged mice.

CONCLUSION

The expression of OSKM in vivo triggers two different cellular outcomes: reprogramming in a small fraction of cells, and damage and senescence in many other cells. There is a strong positive association between these two processes, due to the fact that cellular senescence creates a tissue context that favors OSKM-driven reprogramming in neighboring cells. The positive effect of senescence on reprogramming is mediated by secreted factors, of which IL-6 is a key player. This also applies to tissue injury and aging, where there is an accumulation of senescent cells that send signals to surrounding cells to promote OSKM-driven dedifferentiation and reprogramming. A similar conceptual interplay may occur in physiological conditions, where damage-triggered senescence could induce cell dedifferentiation to promote tissue repair.

Interplay between cellular senescence and OSKM-driven reprogramming.

Expression of OSKM in vivo, apart from inducing the reprogramming of a small population of cells, also induces damage and senescence in many other cells. Senescent cells release factors that promote the reprogramming of neighboring cells, with IL-6 being a critical mediator. Tissue injury and aging, through the accumulation of senescent cells, favor in vivo reprogramming.

Abstract

Reprogramming of differentiated cells into pluripotent cells can occur in vivo, but the mechanisms involved remain to be elucidated. Senescence is a cellular response to damage, characterized by abundant production of cytokines and other secreted factors that, together with the recruitment of inflammatory cells, result in tissue remodeling. Here, we show that in vivo expression of the reprogramming factors OCT4, SOX2, KLF4, and cMYC (OSKM) in mice leads to senescence and reprogramming, both coexisting in close proximity. Genetic and pharmacological analyses indicate that OSKM-induced senescence requires the Ink4a/Arf locus and, through the production of the cytokine interleukin-6, creates a permissive tissue environment for in vivo reprogramming. Biological conditions linked to senescence, such as tissue injury or aging, favor in vivo reprogramming by OSKM. These observations may be relevant for tissue repair.

Cell identity can be manipulated in vitro, and this has opened new therapeutic possibilities (1). A major achievement in this field was the discovery that four transcription factors—OCT4, SOX2, KLF4, and cMYC (OSKM)—can transform differentiated cells into pluripotent cells, known as induced pluripotent stem cells (iPSCs) (2). These cells have the capacity to differentiate into all the cell types of the adult organism. Recent studies have shown that OSKM overexpression in mice results in cell dedifferentiation in multiple tissue types and reprogramming into iPSCs in vivo (3, 4). However, little is known about the molecular mechanisms or cellular contexts that regulate in vivo reprogramming.

Ink4a/Arf promotes in vivo reprogramming

To explore the mechanisms involved in reprogramming in vivo, we focused on the tumor suppressors p53 (encoded by Tp53), p16INK4a, and ARF (the latter two proteins are encoded by the genetic locus known as Cdkn2a or Ink4a/Arf). These proteins inhibit in vitro reprogramming, and their deletion or down-regulation increases the efficiency of the process (511). To investigate their role in reprogramming in vivo, we generated reprogrammable mice (carrying a ubiquitous doxycycline-inducible OSKM transgene, abbreviated as i4F for “inducible four factors”) combined with null alleles for p53 or Ink4a/Arf, and we confirmed that the OSKM transgene was efficiently induced in mice of the three genotypes (fig. S1A). We evaluated p53 or Ink4a/Arf-heterozygous mice, all in the same inbred pure C57BL6 genetic background, for the development of OSKM-driven teratomas (an assay for pluripotency) after treatment with doxycycline (0.2 mg/ml in the drinking water for 8 days). We used heterozygous mice because the life span of homozygous mice was too short to measure teratoma formation. Unexpectedly, i4F;Ink4a/Arf-heterozygous mice were highly resistant to teratoma formation compared to i4F control mice (Fig. 1A). This was in contrast to i4F;p53-heterozygous mice, which had a higher rate of teratoma formation compared to controls. The organ distribution of teratomas was similar in all three i4F mouse strains (fig. S1B). Notably, the kinetics of spontaneous (nonteratoma) tumor development characteristic of p53 and Ink4a/Arf-heterozygous mice was not affected by the induction of OSKM (fig. S1C).

Fig. 1 p53 limits and Ink4a/Arf promotes in vivo reprogramming.

(A) Incidence of teratomas in mice of the indicated genotypes. Mice were treated with doxycycline (0.2 mg/ml in the drinking water) for 8 days, at 20 to 28 weeks of age. Time (weeks) refers to age. Only mice that died with teratomas were considered; mice that died without teratomas or that were alive appear censored and indicated with ticks. Cohorts were as follows: i4F, n = 33; i4F;p53-het, n = 58; i4F;Ink4a/Arf-het, n = 40. Statistical significance was evaluated using the log-rank test: ***P < 0.001. (B) NANOG immunohistochemistry and hematoxylin and eosin (H&E) staining of pancreas of the indicated genotypes. Mice were treated with doxycycline (0.2 mg/ml) for 7 days and analyzed at the end of the treatment. Images are representative of at least five mice (n ≥ 5). (C) Relative number (per thousand) of NANOG+ cells in the tissues of i4F mice of the indicated genotypes treated with doxycycline as in (B). Quantifications were done in a completely automated manner. Graph represents average ± SD (n = 5 for pancreas; n = 4 for stomach; n = 3 for kidney); statistical significance relative to control (i4F) was assessed by the unpaired two-tailed Student’s t test with Welch’s correction: *P < 0.05. (D) Teratoma formation in the kidney of WT syngeneic C57BL6 mice injected with i4F MEFs of the indicated genotypes and induced in vivo with doxycycline (2 mg/ml) for 14 days. Teratomas appeared within 6 to 8 weeks after treatment. Statistical significance relative to control (i4F) was evaluated using two-tailed Fisher’s exact test: **P < 0.01.

We next investigated which step of teratoma formation—reprogramming, expansion, or differentiation—was affected by the absence of p53 or Ink4a/Arf. We first assessed whether iPSC expansion and differentiation were different between the three genotypes. Subcutaneous injection of i4F, i4F;p53-null, and i4F;Ink4a/Arf-null iPSCs into nude mice produced teratomas with similar efficiency (fig. S1D), suggesting that the differences were not due to a differential ability of the iPSCs from the three genotypes to form teratomas. We next examined the process of reprogramming. Previously, we reported the presence of iPSCs in the blood of i4F mice after treatment with doxycycline (3). Using this assay, we observed the same trend in reprogramming efficiency as observed for the generation of teratomas—that is, i4F;p53-null > i4F > i4F;Ink4a/Arf-null (fig. S1E). We then focused on in situ reprogramming by studying mice 7 days after treatment with doxycycline and by looking for the expression of the pluripotency marker NANOG in tissues. We found that the pancreatic tissue of i4F;p53-null mice contained more NANOG+ cells than that of i4F mice, whereas the pancreases of i4F;Ink4a/Arf-null mice were essentially devoid of NANOG+ cells (Fig. 1, B and C). In line with these findings, tissue architecture was globally dysplastic in the i4F;p53-null pancreas, whereas the pancreas of i4F mice showed focal areas of dysplasia. In contrast, tissue architecture was largely unaffected in i4F;Ink4a/Arf-null pancreas (Fig. 1B). Similar differences among the three genotypes were also observed in the stomach and the kidney (Fig. 1C and fig. S1, F and G). Together, these results suggest that the absence of p53 favors reprogramming in vivo, whereas the absence of Ink4a/Arf limits reprogramming.

The lower efficiency of in vivo reprogramming of i4F;Ink4a/Arf-null mice is in contrast to the higher efficiency of in vitro reprogramming of cells deficient in Ink4a/Arf (610). To further investigate this, we obtained mouse embryonic fibroblasts (MEFs) from the three i4F strains and compared their reprogramming efficiency in vitro and in vivo. As expected, both i4F;p53-null and i4F;Ink4a/Arf-null MEFs had a higher efficiency of in vitro reprogramming compared to controls (fig. S1H). However, when the same set of reprogrammable MEFs was injected into the kidney of wild-type (WT) mice, the efficiency of teratoma formation after doxycycline treatment was i4F;p53-null > i4F > i4F;Ink4a/Arf-null (Fig. 1D and fig. S1, I and J), which is the same pattern observed in whole-body reprogrammable mice (Fig. 1, A to C). When the reprogrammable MEFs were injected into immunocompromised hosts, such as athymic nude mice (lacking T lymphocytes) and NSG mice (lacking T, B, and natural killer lymphocytes), the kinetics of teratoma development maintained the same trend as in immunocompetent C57BL6 mice (fig. S1I). Therefore, the adaptive immune system is not a critical determinant of the differences in reprogramming among genotypes. These results, using the same set of reprogrammable MEFs, indicate that the Ink4a/Arf locus plays a dual role during reprogramming, acting as a barrier in vitro but as a promoter in vivo.

To gain insight into the role of Ink4a/Arf during in vivo reprogramming, we profiled the transcriptome of the pancreas after treatment with doxycycline (0.2 mg/ml for 7 days) in mice of the three genotypes [Gene Expression Omnibus (GEO) database accession number GSE77722]. Notably, most of the pathways that followed the pattern i4F;p53-null > i4F > i4F;Ink4a/Arf-null (fig. S1K) were related to the immune system and extracellular matrix remodeling, both characteristically associated with cellular senescence (1216). Given the well-established involvement of the Ink4a/Arf locus in cellular senescence (17), we hypothesized that the Ink4a/Arf locus may contribute to in vivo reprogramming in a cell-extrinsic manner through the generation of a senescent tissue environment.

Reprogramming in vivo coexists with senescence

To explore the possible association between in vivo reprogramming and senescence, we examined whether both processes coexist within the same tissue context. We performed double stainings for NANOG together with indicators of senescence, such as senescence-associated β-galactosidase (SAβG) or the cell cycle inhibitor p21 (18). We found that, upon induction of reprogramming, NANOG+ cells in the stomach (Fig. 2A) and the pancreas (fig. S2A) generally appeared in close proximity to clusters of SAβG+ or p21+ cells. Moreover, there was a positive correlation between the extent of senescence and the number of reprogrammed cells. Specifically, i4F;p53-null pancreas showed widespread SAβG+ staining, whereas i4F pancreas had focal SAβG+ clusters, and i4F;Ink4a/Arf-null pancreas had few SAβG+ cells (Fig. 2B). This trend (i4F;p53-null > i4F > i4F;Ink4a/Arf-null) was also observed for p21, infiltration of macrophages (F4/80), proliferation [bromodeoxyuridine (BrdU) incorporation], DNA damage (γH2AX), and apoptosis (active caspase-3) (Fig. 2B). Stomach and kidney also presented the same pattern (fig. S2, B and C).

Fig. 2 Senescence and reprogramming coexist.

(A) (Top) Double staining of NANOG (dark brown) and SAβG (light blue) in the stomach of i4F mice. (Bottom) Double staining of NANOG (magenta) and p21 (dark brown) in the stomach of i4F mice. Mice were treated with doxycycline (1 mg/ml in the drinking water) for 7 days and analyzed at the end of the treatment. (B) Staining of the pancreas. Mice were treated with doxycycline (0.2 mg/ml) for 7 days and analyzed at the end of the treatment. All quantifications were done in a completely automated manner and correspond to either the relative percentage of stained surface (SAβG, F4/80, γH2AX, and caspase-3) or the relative percentage of positive cells (p21 and BrdU). Values correspond to average ± SD (n = 3 to 5 mice per group). Statistical significance relative to i4F control was evaluated using the unpaired two-tailed Student’s t test with Welch’s correction: *P < 0.05; **P < 0.01; ***P < 0.001. (C) Staining of the pancreas, as in (B). The staining of p65 (NFκB) is quantified as relative percentage of positive surface, and the staining of pSTAT3 and pSMAD2 is quantified as relative percentage of positive cells. (D) mRNA levels of the indicated cytokines in the pancreas. Mice were treated with doxycycline as in (B). Values are relative to WT. Bars correspond to average ± SD; statistical significance was evaluated using the unpaired two-tailed Student’s t test with Welch’s correction. Comparison of each i4F genotype with its own non-i4F control is indicated as follows: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Comparisons of i4F;p53-null or i4F;Ink4a/Arf-null with i4F are indicated in the same manner but using the symbol “#.”

Moreover, transcription factors responsible for cytokine production upon tissue damage, such as NFκB (p65RelA), STAT3 (phospho-STAT3), and SMAD2 (phospho-SMAD2), also followed the same pattern as reprogramming, damage, and senescence (i4F;p53-null > i4F > i4F;Ink4a/Arf-null) (Fig. 2C). In line with this, the expression levels of mRNAs corresponding to Ink4a and Arf and a panel of cytokines, chemokines, growth factors, and tissue remodeling proteins were consistent with a strong senescence and inflammatory response upon induction of reprogramming in i4F;p53-null pancreas, a moderate response in i4F pancreas, and a minimal response in i4F;Ink4a/Arf-null pancreas (Fig. 2D). In addition, serum levels of cytokines interleukin-6 (IL-6) and tumor necrosis factor–α (TNFα) increased upon OSKM activation, and this increase was most prominent in i4F;p53-null mice (fig. S2D).

We speculate that, in the absence of p53, cells cannot protect themselves from the effects of OSKM overexpression, leading to unrestrained cell proliferation, DNA damage, senescence, inflammation, and high levels of secreted factors. Moreover, it has been reported that the secretory activity of senescent cells [known as senescence-associated secretory phenotype (SASP)] is exacerbated in the absence of p53 (14). In contrast, in the absence of Ink4a/Arf, OSKM does not induce cellular senescence, which translates into lower levels of secreted factors and inflammation. Thus, OSKM triggers two divergent cellular outcomes, damage-induced cellular senescence and cellular reprogramming, that coexist in close proximity within tissues.

Senescence promotes in vivo reprogramming

To explore the link between senescence and in vivo reprogramming, we used small molecules to decrease or increase the number of senescent cells. To eliminate senescent cells, we took advantage of the hypersensitivity of senescent cells to the inhibition of Bcl-family proteins (1921). Simultaneous treatment of i4F mice with doxycyline and the potent Bcl-2/Bcl-xL/Bcl-w–inhibitor navitoclax (also known as ABT263) reduced the amount of senescent cells and their associated secreted factors in pancreatic tissue undergoing reprogramming (Fig. 3, A and B). The reduction in the number of senescent cells was accompanied by a decrease in reprogramming, as measured by the number of NANOG+ cells (Fig. 3A). Conversely, to increase senescence, we used the (CDK4/CDK6)-inhibitor palbociclib (also known as PD0332991), which can be considered as a p16INK4a functional mimetic (22). Treatment with palbociclib increased the number of senescent cells, the production of their associated cytokines, and reprogramming in the pancreas (fig. S3, A and B). These observations suggest that senescence positively regulates in vivo reprogramming.

Fig. 3 Senescence enhances in vivo reprogramming.

(A) Staining of the pancreas of i4F mice treated with doxycycline (1 mg/ml in the drinking water) and, where indicated, with navitoclax (25 mg/kg daily by oral gavage) or BAY 11-7082 (20 mg/kg daily by intraperitoneal injection). Mice were analyzed at the end of the treatment (7 days). Quantifications were done as explained in Fig. 2B. Values for SAβG indicate relative percentage of stained surface, and those for NANOG indicate relative ‰ of positive cells (NANOG is measured by the number of positive cells per thousand). Images are representative of at least three mice (n ≥ 3). (B) mRNA levels of the indicated genes in the pancreas of the same mice as in (A). (C) Staining of the pancreas of i4F mice treated with doxycycline (1 mg/ml in the drinking water) and, where indicated, with anti–IL-6 (1 mg, three times per week, intraperitoneally) or PIM inhibitor (PIMi) (50 mg/kg daily by oral gavage). Mice were analyzed at the end of the treatment (7 days). Quantifications were done as explained in Fig. 2B. Values for SAβG indicate relative percentage of stained surface, and those for NANOG indicate relative ‰ of positive cells. Images are representative of at least five mice (n ≥ 5). (D) mRNA levels of the indicated genes in the pancreas of the same mice as in (C). Values in all panels correspond to average ± SD. Statistical significance relative to control was assessed by the unpaired two-tailed Student’s t test with Welch’s correction: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

The master transcription factor NFκB plays a critical role in promoting the secretory activity of senescent cells (16, 23, 24). To evaluate the possible role of NFκB as a link between senescence and reprogramming, i4F mice were simultaneously treated with doxycycline and with an inhibitor of the NFκB-activating kinases IKKs (BAY 11-7082) (25). BAY 11-7082 diminished the expression of SASP factors, which led to decreased in vivo reprogramming (Fig. 3, A and B). BAY 11-7082 also reduced the extent of senescence (Fig. 3A), likely reflecting the loss of SASP-mediated paracrine senescence (26).

We next investigated which of the SASP factors play a dominant role during in vivo reprogramming. It was recently reported that IL-6 enhances in vitro reprogramming and can replace leukemia inhibitory factor (LIF), a related cytokine often used for reprogramming in vitro (27). We found that among the SASP factors tested, IL-6 showed the most marked up-regulation upon OSKM induction and a good correlation with the degree of reprogramming (~300-fold in i4F;p53-null mice, ~20-fold in i4F mice, and ~4-fold in i4F;Ink4a/Arf-null mice) (see Fig. 2D). We therefore tested the role of IL-6 by treating reprogrammable mice simultaneously with doxycycline and anti–IL-6 antibodies. We found that the anti–IL-6 antibodies abolished the increase in IL-6 serum levels characteristic of OSKM activation (fig. S3C). Blockade of IL-6 in vivo reduced not only reprogramming, as measured by the abundance of NANOG+ cells, but also senescence and the expression of SASP factors (Fig. 3, C and D). The reduction of senescence upon anti–IL-6 is consistent with the known paracrine role of IL-6 in senescence (15). Reciprocally, injection of recombinant IL-6 (rIL-6) in i4F mice increased the extent of reprogramming (fig. S3, C and D). Furthermore, treatment of i4F;p53-null mice with anti–IL-6 antibodies also reduced reprogramming, senescence, and SASP (fig. S3, E to G). These results indicate that IL-6 is one of the critical cytokines linking senescence and reprogramming.

To reinforce the role of IL-6, we tested the involvement of the PIM kinases, which are activated by IL-6 through the JAK/STAT pathway (28), and reinforce NFκB activity through phosphorylation of p65RELA (29, 30). Moreover, PIM1 is known to mediate the effects of IL-6 during in vitro reprogramming (27). Simultaneous treatment of i4F mice with doxycycline and with a PIM inhibitor (ETP-995, abbreviated as PIMi) (fig. S3, H to J, and supplementary text) resulted in a profound reduction of senescence, SASP factors, and reprogramming (Fig. 3, C and D). We conclude that senescence promotes in vivo reprogramming through the SASP, with IL-6 being a critical mediator.

Ink4a/Arf promotes cytokine production

To investigate the mechanism linking senescence to reprogramming, we examined the impact of senescent cells and their secretome on in vitro reprogramming. We used γ-irradiation (20 Gy) to induce a robust senescence response in primary fibroblasts (14). Coculture of reprogrammable i4F MEFs with γ-irradiated WT or p53-null MEFs increased in vitro reprogramming, whereas coculture with γ-irradiated Ink4a/Arf-null MEFs had no effect (Fig. 4A). To determine whether the positive effect of irradiated MEFs on reprogramming was mediated by secreted factors, we added conditioned medium (CM) from γ-irradiated MEFs to reprogrammable i4F MEFs. We found that CM from γ-irradiated WT MEFs enhanced reprogramming and that CM from γ-irradiated p53-null cells was even more potent in promoting reprogramming. In contrast, CM from γ-irradiated Ink4a/Arf-null MEFs had little effect (Fig. 4B).

Fig. 4 Senescence enhances in vitro reprogramming.

(A) Reprogramming efficiency of i4F MEFs cocultured with γ-irradiated (γIR) MEFs (ratio, 1:3) from the indicated genotypes. (B) Reprogramming efficiency of i4F MEFs treated with CM from γ-irradiated MEFs from the indicated genotypes. CM was collected every day for 3 days after γ-irradiation. (C) Cytokine profile of the CM from γ-irradiated MEFs of the indicated genotypes. All cytokines are detected in duplicate. IL-6 and TNFα are marked with red rectangles. (D) Tnf and Il6 mRNA levels in nonirradiated and γ-irradiated MEFs of the indicated genotypes at day 3 after γ-irradiation. (E) Reprogramming efficiency of i4F MEFs treated with CM from γ-irradiated WT MEFs in the absence or presence of increasing amounts of anti–IL-6 (+, 0.03 mg of antibodies per milliliter of CM; ++, 0.1 mg of antibodies per milliliter of CM). In (A) to (E), graphs represent average ± SD; statistical significance relative to control was assessed by the two-tailed Student’s t test with Welch’s correction. A total of three biological replicates were performed, each with independent MEF isolates (n = 3). In (A) and (B), comparisons of each genotype of γ-irradiated MEFs to the non–γ-irradiated control are indicated as follows: *P < 0.05; **P < 0.01. Comparisons of γ-irradiated p53-null or γ-irradiated Ink4a/Arf-null with γ-irradiated WT MEFs are indicated in the same manner but using the symbol “#.” In (D), comparisons of each genotype with its own non–γ-irradiated control are indicated as follows: *P < 0.05; **P < 0.01. In (E), comparisons of each condition to the control without CM are indicated as follows: *P < 0.05; **P < 0.01. Comparisons of each condition to the control without anti–IL-6 are indicated in the same manner but using the symbol “#.”

To identify candidate mediators, we analyzed the CMs from the previous experiment in an array with immobilized antibodies for 40 cytokines. Two cytokines present in the CM from γ-irradiated WT or p53-null cells were noticeably absent in the CM from γ-irradiated Ink4a/Arf-null MEFs, namely, IL-6 and TNFα (Fig. 4C). Analysis of mRNA levels confirmed that a functional Ink4a/Arf locus is necessary for the expression of Il6 and Tnf mRNAs upon γ-irradiation (Fig. 4D). We next added anti–IL-6 antibodies to the CM from γ-irradiated WT cells and found that they abolished the effect of CM and almost completely prevented reprogramming (Fig. 4E). This result supports the idea that IL-6 has a pivotal role in reprogramming. Finally, we explored whether the requirement of Ink4a/Arf for IL-6 production upon γ-irradiation is due to Ink4a or Arf. We infected WT MEFs with retroviruses encoding short hairpin RNAs (shRNAs) against Ink4a or Arf. We found that shArf-MEFs retained normal induction of Il6 mRNA, whereas shInk4a-MEFs failed to up-regulate Il6 upon γ-irradiation (fig. S4). These observations recapitulate in vitro the concept that senescent cells promote reprogramming through the Ink4a/Arf locus and the production of IL-6.

Tissue injury and aging promote reprogramming

Given the link between senescence and reprogramming, we wondered whether biological conditions characterized by higher levels of senescence, such as tissue injury, would also promote in vivo reprogramming. To address this, we focused on the lung because, despite expression of the OSKM transgene (3), we have never observed in vivo reprogramming in the lung under our experimental conditions. To induce cellular senescence in the lung, we chose the DNA-damaging agent bleomycin, which is a well-known model of lung injury and fibrosis (31, 32). Two weeks after treatment with bleomycin, the lungs of WT mice showed signs of fibrosis and senescence (Fig. 5A and fig. S5A). Notably, the lungs of i4F mice treated with bleomycin and subsequently induced with doxycycline showed an abundant number of senescent cells, enhanced expression of p21, clusters of NANOG+ cells, and dysplastic foci (Fig. 5A and fig. S5B). They also showed an increase in the mRNA levels of SASP factors, including Il6 (Fig. 5B). Consistent with our previous observations, doxycycline-treated i4F mice that had not been treated with bleomycin did not show evidence of reprogramming or senescence in the lungs (Fig. 5A) and showed only modest up-regulation of SASP (Fig. 5B). This experiment indicates that the infliction of tissue damage by an exogenous factor renders the lung permissive to OSKM-induced in vivo reprogramming.

Fig. 5 Tissue damage favors in vivo reprogramming.

(A) Staining of the lungs of the indicated mice treated with a single intratracheal dose of bleomycin or with PBS. After 1 day, mice were treated with doxycycline (0.2 mg/ml in the drinking water) for 14 days and analyzed at the end of the treatment. Quantifications were done as explained in Fig. 2B. Values for SAβG indicate relative percentage of stained surface, those for NANOG indicate relative ‰ of positive cells, and those for p21 indicate relative percentage of positive cells. Statistical significance (n ≥ 3) relative to i4F controls was assessed by the unpaired two-tailed Student’s t test with Welch’s correction: *P < 0.05; &P < 0.001. (B) mRNA levels of the indicated genes in the lungs of the same mice as in (A). One-way analysis of variance (ANOVA) and Bonferroni post hoc test indicated significant differences in Il6 and Pai1 between i4F + bleo and WT mice: *P < 0.05.

Following on the concept that tissue injury facilitates OSKM-induced reprogramming, we examined whether aging—a condition characterized by systemic accumulation of cellular damage and senescence (17)—also promotes OSKM-driven reprogramming. We first analyzed in vivo reprogramming in a mouse model of progeria (Terc-null mice of the second generation or G2Terc-null). The premature aging phenotype in these mice is due to the absence of telomerase activity that leads to critically short telomere length (33, 34). Upon activation of OSKM, the pancreas of i4F;G2Terc-null mice displayed substantially higher levels of senescence, SASP factors, and reprogramming, as compared with control i4F mice derived from the same parental strains (Fig. 6, A and B). These results are consistent with the hypothesis that tissue damage, in this case due to short telomeres in a mouse model of premature aging, promotes OSKM-induced in vivo reprogramming.

Fig. 6 Aging promotes in vivo reprogramming.

(A) mRNA levels of the indicated genes in the pancreas of i4F and i4F;G2Terc-null mice treated with doxycycline (0.2 mg/ml in the drinking water) for 14 days and analyzed at the end of the treatment. Statistical significance relative to control was assessed by the unpaired two-tailed Student’s t test with Welch’s correction: *P < 0.05; **P < 0.01; ****P < 0.0001. (B) Stainings of the pancreas of the same mice shown in (A). Quantifications were done as explained in Fig. 2B. Images are representative of at least four mice (n ≥ 4). (C) Stainings of the pancreas of young (7 to 9 weeks) and old (>65 weeks) i4F mice treated with doxycycline (0.2 mg/ml in the drinking water) for 7 days and analyzed at the end of the treatment. Images are representative of at least three mice (n ≥ 3). (D) mRNA levels of the indicated genes in the pancreas of the same mice shown in (C). Statistical significance was evaluated using one-way ANOVA and Bonferroni post-hoc test. Comparisons of each i4F genotype with its own control are indicated as *P < 0.05 and **P < 0.01. Comparisons of old i4F to young i4F are indicated in the same manner but with the symbol “#”. Values in (A) to (D) correspond to average ± SD. In (B) and (C), values for SAβG indicate relative percentage of stained surface, and those for NANOG indicate relative ‰ of positive cells.

Finally, we examined the effect of physiological aging on reprogramming. We compared side by side the impact of OSKM expression in old (>65 weeks of age) and young (7 to 9 weeks of age) i4F mice. Consistent with our observations in Terc-null progeric mice, activation of OSKM in old mice resulted in higher levels of senescent cells in the pancreas and increased levels of NANOG+ cells and teratoma formation, as compared with young mice treated in parallel (Fig. 6C and fig. S6A). As expected (35), the mRNA levels of Ink4a were higher in old WT mice than in young WT mice (Fig. 6D). Moreover, upon induction of OSKM, the mRNA levels of Ink4a and SASP-related factors were higher in old i4F mice than in the corresponding young controls (Fig. 6D). To further strengthen the effect of aging in reprogramming, we injected i4F MEFs into the kidneys of old (>95 weeks of age) WT mice, and this resulted in a higher and faster teratoma formation after treatment with doxycycline, as compared with i4F MEFs injected into young (10 weeks of age) WT mice (fig. S6B). Together, these results indicate that senescent tissue environments, such as those associated with exogenously induced tissue damage or with aging, favor in vivo reprogramming.

Discussion

Here, we provide insight into the mechanism of in vivo reprogramming. We show that, apart from inducing cellular reprogramming, OSKM produces damage and senescence in many other cells. Damaged cells secrete signals that strongly promote in vivo reprogramming in neighboring cells. This process is mediated by the induction of cellular senescence and the ensuing production of cytokines, with IL-6 being of particular relevance. It is well established that Ink4a/Arf, DNA damage, and short telomeres are intrinsic barriers for in vitro reprogramming and, accordingly, in vitro reprogramming efficiency decreases in aged cells (7, 8, 36). In vivo, however, Ink4a/Arf, DNA damage, and short telomeres play an additional positive and dominant extrinsic role in reprogramming by creating a senescent tissue context that strongly promotes reprogramming through the production of factors, such as IL-6. In the absence of p53, OSKM-induced damage is unrestrained and the secretory activity of damaged cells is markedly exacerbated. Therefore, p53 is a strong barrier for reprogramming through both intrinsic and extrinsic mechanisms. We have also uncovered a number of pharmacologically actionable targets that can modulate in vivo reprogramming. Finally, there is growing support for the hypothesis that tissue damage can provoke cellular plasticity conducive to tissue repair (37). In this context, our results suggest that senescence may contribute to reprogramming-like cellular plasticity upon tissue damage.

Materials and methods

Reprogrammable mice

To generate reprogrammable mice combined with null alleles for p53, Ink4a/Arf, or Terc, we used the reprogrammable mouse line known as i4F-B, which carries a ubiquitous doxycycline-inducible OSKM transgene, abbreviated as i4F, and inserted into the Pparg gene (3). Reprogrammable i4F-B mice were crossed with mice deficient in p53 (38), Ink4a/Arf (39), or Terc (33). The i4F;p53-null and i4F;Ink4a/Arf-null mice are in a pure C57BL/6J.Ola.Hsd genetic background and were compared to i4F mice of the same genetic background; the i4F;Terc-null mice are in a mixed genetic background enriched for C57BL/6J.Ola.Hsd and were compared to i4F mice of the same mixed genetic background derived from the same parental mice.

Animal procedures

Animal experimentation at the CNIO, Madrid, was performed according to protocols approved by the CNIO-ISCIII Ethics Committee for Research and Animal Welfare (CEIyBA). To check the proper induction of the OSKM transgene, mice of the three genotypes, i4F, i4F;p53-null, and i4F;Ink4a/Arf-null, were intraperitoneally injected with 500 μl of doxycycline (4 mg/ml; Sigma) dissolved in physiological serum (0.9% w/v NaCl). After 9 hours, mice were sacrificed and samples were taken for analysis. In general, mice of 10–13 weeks of age of both sexes were treated with 0.2 mg/ml doxycycline in the drinking water (supplemented with 7.5% sucrose) for 7 days. For the simultaneous treatment with chemicals (except in the case of palbociclib), mice were treated with the corresponding inhibitor (see Chemical treatments) and with 1 mg/ml of doxycycline in the drinking water for 7 days. In the case of palbociclib, doxycycline was administered by intraperitoneal injection (500 μl of doxycycline 2 mg/ml dissolved in physiological serum every second day for 5 days) to avoid severe dysplasia in the colon. In the case of the i4F;G2Terc-null mice and bleomycin-treated mice, together with their corresponding controls, treatment with 0.2 mg/ml of doxycycline was performed for 2 weeks. For the reprogramming experiment in aged mice, we used old i4F mice that were >65 weeks old and young i4F mice that were 7–9 weeks old. For proliferation analysis, bromodeoxyuridine (5-bromo-2′-deoxyruridine), purchased from Sigma #B5002, was daily injected intraperitoneally at 50 μg/g during the first 5 days of treatment. For survival analyses (death due to teratomas and death due to tumors), i4F, i4F;p53-het, and i4F;Ink4a/Arf-het mice, together with their respective non-reprogrammable controls, were treated with 0.2 mg/ml of doxycycline for 8 days, at the age of 20–28 weeks. Mice were monitored and sacrificed and the cause of death was determined by histological analysis.

Teratoma formation by exogenous MEFs

The kidney was the only location where injected reprogrammable MEFs produced teratomas upon doxycycline treatment, which could be related to the reported permissivity of the kidney to the transplantation of differentiated iPSCs (40) (the other locations that we tested and did not produce teratomas were liver, spleen, muscle, dermis, and intraperitoneal). For teratoma formation in the kidney, MEFs of the three genotypes (i4F, i4F;p53-null, and i4F;Ink4a/Arf-null) were trypsinized and 5 × 105 cells, resuspended in iPSC medium (see below), were injected into the kidney of wild-type C57BL/6J, nude (Hsd: athymic Nude/Nude from Harlan Ibérica) and NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ from Charles River) mice of 10 weeks of age, previously treated for 10 days with 2 mg/ml of doxycycline. We used C57BL/6J mice of > 95 weeks of age for the teratoma formation in old hosts. After injection, mice were treated for 14 days with the same dose of doxycycline. Mice were sacrificed when teratomas were palpable and teratomas were processed for histological analysis.

Teratoma formation by exogenous iPSCs

For subcutaneous teratomas, iPSCs were trypsinized and 1 × 106 cells were subcutaneously injected into the flanks of athymic nude mice (Hsd: athymic Nude/Nude; Harlan Ibérica). Teratomas were isolated when the diameter reached >1.5 cm and processed for histological analysis.

Isolation of iPSCs from the blood

Whole peripheral blood was collected directly from the heart of doxycycline induced i4F, i4F;p53-null and i4F;Ink4a/Arf-null mice at the time of necropsy, and was subjected to two rounds of erythrocyte lysis in ammonium chloride solution (Stem Cells). First round of lysis with 10 ml, for 15 min r.t., followed by centrifugation, and a second round of lysis with 3 ml, for 15 min at r.t., followed by neutralization with 12 ml of iPSC medium. Cells were resuspended, plated on feeders and cultured in iPSC medium in the absence of doxycycline.

Chemical treatments of the mice

Treatment of i4F mice with the Bcl-2/Bcl-xL/Bcl-w–inhibitor navitoclax, also known as ABT263 (Active Biochem, #A-100) (19, 20), was performed by daily oral gavage for 7 days at 25 mg/kg, dissolved in 15% DMSO/PEG400. Treatment with the CDK4/CDK6 inhibitor palbociclib, also known as PD033299 (Selleckchem #S1579) (22), was performed by oral gavage for 5 days at 100 mg/kg, dissolved in 50 mM sodium lactate. Treatment with the inhibitor of NFκB-activating kinases IKKs, also known as BAY 11–7082 (Selleck Chemicals, S2913) (25), was performed by daily intraperitoneal injection during one week at 20 μg/g, dissolved in 10% DMSO/PBS. For inhibition of IL-6, we intraperitoneally injected 1 mg of an anti–IL-6 antibody (BioXCell, BE0046) formulated in PBS, three times per week to the i4F mice and daily to the i4F;p53-null mice, during one week. For treatment with recombinant IL-6 (ABYNTEK BIOPHARMA, S.L, #AI081) mice were intraperitoneally injected with 5 μg of rIL-6 every other day, over the course of 6 days. Treatment with the PIM inhibitor was performed by daily oral gavage for 7 days at 50 mg/kg, dissolved in 10% NMP (N-methylpyrrolidone)/90% PEG300. PIMi was developed at CNIO; for a detailed description, see the supplementary text. Control mice were treated with the corresponding vehicle. All mice were simultaneously treated with 1 mg/ml of doxycycline in the drinking water during the indicated time of treatment with the drugs, except in the case of palbociclib, in which mice were intraperitoneally injected with 500 μl of doxycycline (2 mg/ml) dissolved in physiological serum every second day for 5 days. To induce pulmonary damage, bleomycin (Sigma #15361) was inoculated intratracheally in i4F mice at 2U/kg in males and 1.5U/kg in females and the following day mice were treated with doxycycline (0.2 mg/ml) for 14 days.

Cell culture

Primary MEFs were obtained from embryos at E13.5 and cultured in DMEM supplemented with 10% of FBS and penicillin-streptomycin. IPSCs were cultured over mitomycin-C inactivated feeder cells on gelatin-coated plates and in “iPSC medium”: high-glucose DMEM supplemented with KSR (15%, Invitrogen), LIF (1000 U ml−1), non-essential amino acids, penicillin-streptomycin, Glutamax and β-mercaptoethanol. Cultures were routinely tested for mycoplasma and were always negative. For retroviral transduction, we transfected HEK293T (5 × 106) cells with retroviral vectors expressing mouse shRNA against Ink4a, Arf and Ink4a/Arf, and the corresponding empty vector LMP, all kindly provided by Scott Lowe (41), and packaging vectors using Fugene HD (Roche). Viral supernatants were collected twice a day on two consecutive days starting 36 h after transfection and were used to infect WT MEFs, previously plated at a density of 8 × 105 cells per 10 cm plates. Previous to infection, polybrene was added to the viral supernatants at a concentration of 8 μg/ml. For in vitro reprogramming, i4F, i4F;p53-null and i4F;Ink4a/Arf-null MEFs were plated at a density of 3 × 105 cells per well in 6-well gelatin-coated plates and were cultured in iPSC medium with doxycycline (1 μg/ml). Medium was changed every 48 h. Reprogramming plates were stained for alkaline phosphatase activity (AP detection kit, Sigma Aldrich) and AP+ colonies were scored. For coculture reprogramming experiments, 3 × 105 γ-irradiated MEFs (20 Gy) were plated in 35-cm-diameter plates and, on top of them, 1 × 105 i4F MEFs of the three studied genotypes were plated. For in vitro reprogramming with conditioned medium (CM) from γ-irradiated MEFs, i4F MEFs were cultured in iPSC medium or CM with doxycycline (1 μg/ml) for 14 days. To produce the CM, 1.5 × 106 of γ-irradiated MEFs (20 Gy) were plated in 10-cm-diameter gelatin-coated plates, and cultured in 10 ml of iPSC medium. The medium was collected and filtered (0.22 μm) everyday for 3 days. CM preparations were used freshly or stored frozen at −20°C. To block IL-6 during in vitro reprogramming, anti–IL-6 antibody (BioXCell, BE0046) or control IgG (Santa Cruz Biotechnology sc-2027) were added to the CM at 0.1 mg/ml or 0.03 mg/ml.

Cytokine array

Cytokine levels in conditioned medium (CM) were analyzed using the Mouse Cytokine Array (ProteomeProfiler mouse Cytokine Array Panel A from R&D Systems), following the manufacturer’s instructions. Pixel density was determined using the Image J Software.

Determination of cytokine levels in serum

Cytokine levels in serum were analyzed using the BD Cytometric Bead Array (CBA) Mouse Soluble Protein Master Buffer Kit from BD (#558266), following the manufacturer’s instructions. For IL-6 and TNF detection, specific Mouse Flex Sets from BD were used (#558301 and #558299, respectively). Fluorescence beads detection was determined by Flow Cytometry using the FACSCanto II Cytometer.

RNA-seq methods

Total RNA from pancreas was extracted from i4F, i4F;p53-null and i4F;Ink4a/Arf-null mice (n = 5 per genotype). Samples of total RNA (1 μg/ml) with RNA Integrity Numbers in the range 7.4 to 9.4 (Agilent 2100 Bioanalyzer) were used. PolyA+ fraction was purified and randomly fragmented, converted to double stranded cDNA and processed through subsequent enzymatic treatments of end-repair, dA-tailing, and ligation to adapters, following Illumina’s “TruSeq Stranded mRNA Sample Preparation Part # 15031047 Rev. D” protocol. Adapter-ligated library was completed by PCR with Illumina PE primers (8 cycles). The resulting purified cDNA library was applied to an Illumina flow cell for cluster generation and sequenced on an Illumina HiSeq2000, following manufacturer’s protocols. The complete set of reads has been deposited in the GEO repository (accession number GSE77722). Reads were aligned to the mouse genome (GRCm38/mm10) with TopHat-2.0.4 (42) using Bowtie 0.12.7 (43) and Samtools 0.1.16 (44), allowing two mismatches and five multihits. Estimation of transcript abundances and differential expression were calculated with Cufflinks 1.3.0 (42) using the mouse genome annotation data set GRCm38/mm10 from the UCSC Genome Browser. GSEA (45) was used to perform a gene set enrichment analysis of Reactome, NCI and KEGG pathways. RNA-seq gene list preranked by statistic was used, setting ‘gene set’ as the permutation method and we run it with 1000 permutations. Only those gene sets with significant enrichment levels (FDR q-value<0.05) were considered.

Analysis of mRNA levels

Total RNA was extracted from MEFs with Trizol (Invitrogen), following the provider’s recommendations. For pancreas samples, total RNA was isolated by acid guanidinium thiocyanate-phenol-chloroform extraction. Up to 5 μg of total RNA was reverse transcribed into cDNA using iScriptTM Advanced cDNA Synthesis Kit for RT-qPCR (BioRad #172-5038). Quantitative real time-PCR was performed using GoTaq® qPCR Master Mix (Promega #A6002) in a QuantStudio 6 Flex thermocycler (Applied Biosystem). Primers used are listed in Table 1. For input normalization, we used the housekeeping gene Actin.

Table 1 List of primers used for mRNA expression analyses.

View this table:

Immunohistochemistry

Tissue samples were fixed in 10% neutral buffered formalin (4% formaldehyde in solution), paraffin-embedded and cut in 3 μm sections, which were mounted in superfrost®plus slides and dried. For different staining methods slides were deparaffinized in xylene and re-hydrated through a series of graded ethanol until water. Serial sections were stained with hematoxylin and eosin (HE) and Masson’s trichrome staining was used to assess the presence of fibrotic areas in the lungs. For immunohistochemistry, an automated immunostaining platform was used (Ventana discovery XT, Roche). Antigen retrieval was first performed with high pH buffer (CC1m, Roche), endogenous peroxidase was blocked and slides were then incubated with the appropriate primary antibodies as detailed: NANOG (Cell Signalling Technology, 8822); p65 (NFκB) (Santa Cruz Biotechnology, sc-372); phosphorylated STAT3 (Tyr705) (Cell Signalling Technology, 9145); phosphorylated SMAD2 (phospho-Ser465 and phospho-Ser467) (Cell Signaling Technology, 3101); p21 (HUGO-291 CNIO); F4/80 (D10, CNIO); phosphorylated histone 2A.X (Ser139) (Millipore, 05–636); activated caspase-3 (Cell Signalling Technology, 9661); MYC (Abcam, ab32072) and bromodeoxyuridine (GE Healthcare, RPN202). After the primary antibody, slides were incubated with the corresponding secondary antibodies and visualization systems (OmniRabbit, Ventana, Roche) conjugated with horseradish peroxidase (Chromomap, Ventana, Roche). Immunohistochemical reaction was developed using 3,30-diaminobenzidine tetrahydrochloride (DAB) as a chromogen and nuclei were counterstained with hematoxylin. Finally, the slides were dehydrated, cleared and mounted with a permanent mounting medium for microscopic evaluation. Whole digital slides were acquired with a slide scanner (Mirax Scan, Zeiss), and images captured with the Pannoramic Viewer Software (3DHISTECH). Image analysis and quantification was performed in a completely automated manner using the AxioVision software package (Zeiss). For each staining, several slides were quantified per mouse (at least 3 mice per group).

SAβG staining of histological sections.

For pancreas samples, SAβG staining was performed in tissue cryosections preserved in OCT freezing medium using the Senescence β-Galactosidase Staining Kit (Cell Signaling, #9860). Briefly, 12-μm tissue cryosections were fixed at room temperature for 5 min, with a solution containing 2% formaldehyde and 0.2% glutaraldehyde in PBS, washed three times with PBS, and incubated 48 h at 37°C with the staining solution containing X-gal in N-N-dimethylformamide (pH 6.0). Sections were counterstained with nuclear fast red. For stomach samples, whole-mount SAβG staining was performed. Tissue samples were fixed with the same fixative solution for 45 min and then washed and incubated for 24 h at 37°C with the staining solution. After incubation, tissues were dehydrated for 30 min with 50% ethanol and overnight with 70% ethanol, after which they were paraffin-embedded. Image analysis and quantification was performed in a completely automated manner as indicated above for the immunohistochemistry.

Statistical methods

Samples (cells or mice) were allocated to their experimental groups according to their pre-determined type (cell type or mouse genotype) and therefore there was no randomization. Investigators were not blinded to the experimental groups (cell types or mouse genotypes). Quantitative PCR data were obtained from independent biological replicates (n values correspond to the biological replicates, that is number of mice or number of independent MEF preparations; technical replicates of the PCR were not considered in the n value). Statistical significance was assessed using Student’s t-test (two-tailed, unpaired) with Welch’s correction, Fisher’s exact test (two-tailed), or one-way ANOVA with Bonferroni post-hoc test, as indicated in the figure legends.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/354/6315/aaf4445/suppl/DC1

Supplementary Text

Figs. S1 to S6

Reference (46)

REFERENCES AND NOTES

  1. Acknowledgments: We are grateful to R. Serrano, L. Martínez, O. Domínguez, P. González, M. Gómez, M. Udriste, Z. Vega, M. Lozano, and G. Hernández for technical support. L.M. was a recipient of an FPU contract from the Spanish Ministry of Education (MECD). N.A. was a recipient of an FPI contract from the Spanish Ministry of Economy (MINECO). D.C. and M.R. were recipients of a fellowship from La Caixa. P.J.F.-M. was funded by the Spanish Association Against Cancer (AECC). Work in the laboratory of M.S. is funded by the CNIO and by grants from the MECD cofunded by the European Regional Development Fund (SAF project), the European Research Council (ERC Advanced Grant), the Regional Government of Madrid cofunded by the European Social Fund (ReCaRe project), the European Union (RISK-IR project), the Botín Foundation and Banco Santander (Santander Universities Global Division), the Ramón Areces Foundation, and the AXA Foundation. Work in the laboratory of M.A.B. is funded by the CNIO, MINECO, ERC Advanced Grant, the European Union, the Botin Foundation, and Banco Santander, WRC, and the AXA Research Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. M.S. is a paid adviser for UNITY Biotechnology Inc., a company developing senolytic medicines.
View Abstract

Stay Connected to Science

Navigate This Article