SIRT1 regulates macrophage self‐renewal

Abstract

Mature differentiated macrophages can self‐maintain by local proliferation in tissues and can be extensively expanded in culture under specific conditions, but the mechanisms of this phenomenon remain only partially defined. Here, we show that SIRT1, an evolutionary conserved regulator of life span, positively affects macrophage self‐renewal ability in vitro and in vivo. Overexpression of SIRT1 during bone marrow‐derived macrophage differentiation increased their proliferative capacity. Conversely, decrease of SIRT1 expression by shRNA inactivation, CRISPR/Cas9 mediated deletion and pharmacological inhibition restricted macrophage self‐renewal in culture. Furthermore, pharmacological SIRT1 inhibition in vivo reduced steady state and cytokine‐induced proliferation of alveolar and peritoneal macrophages. Mechanistically, SIRT1 inhibition negatively regulated G1/S transition, cell cycle progression and a network of self‐renewal genes. This included inhibition of E2F1 and Myc and concomitant activation of FoxO1, SIRT1 targets mediating cell cycle progression and stress response, respectively. Our findings indicate that SIRT1 is a key regulator of macrophage self‐renewal that integrates cell cycle and longevity pathways. This suggests that macrophage self‐renewal might be a relevant parameter of ageing.

Introduction

Macrophages reside in essentially all tissues of the body and contribute to diverse functions such as tissue development, homeostasis, repair and immunity (Wynn et al, 2013). Under challenge conditions, during inflammation or in tissues with high turnover, they can be replaced by infiltrating monocytes originating from hematopoietic stem cells (HSCs) (Geissmann et al, 2010; Shi & Pamer, 2011; Mildner et al, 2013; Ginhoux & Jung, 2014). More recently, it has been demonstrated that resident tissue macrophages can also be embryo‐derived and self‐maintain long term by local proliferation independently of HSCs (Sieweke & Allen, 2013; Gentek et al, 2014; Lavin et al, 2015; Ginhoux & Guilliams, 2016; Perdiguero & Geissmann, 2016; Prinz et al, 2017). These observations indicated that macrophages might have inherent self‐renewal mechanisms that classically have been associated with stem cells (Sieweke & Allen, 2013). Indeed, we could recently show that a network of genes governing macrophage self‐renewal overlaps substantially with that controlling self‐renewal in embryonic stem (ES) cells (Soucie et al, 2016). These genes are under the control of macrophage‐specific enhancers, already present in the quiescent state, but repressed by the macrophage transcription factors c‐Maf and MafB. The self‐renewal gene network can be activated in macrophage populations that express low levels of these Maf transcription factors either constitutively, as is the case for alveolar macrophages (AMs) or upon cytokine stimulation, as observed for example in M‐CSF‐induced peritoneal macrophages (Soucie et al, 2016). As a consequence, monocyte or bone marrow‐derived macrophages with a genetic deletion of MafB and c‐Maf (hereafter referred to as Maf‐DKO macrophages) mimic this process and self‐renew indefinitely in culture in the presence of macrophage colony‐stimulating factor (M‐CSF), without transformation or loss of their mature functional phenotype (Aziz et al, 2009; Soucie et al, 2016).

Here, we further investigated the molecular mechanisms of the extended, possibly indefinite, replicative life span of mature differentiated macrophages by focusing on sirtuin family proteins, which affect similar processes in other cell types and organisms. The founding member of this family, silent information regulator 2 (SIR2) from Saccharomyces cerevisia, was found to extend the replicative life span of yeast (Sinclair & Guarente, 1997; Kaeberlein et al, 1999). Mammals have seven sirtuin homologues (SIRT1‐7) in different subcellular compartments (Frye, 2000; North & Verdin, 2004; Guarente, 2013). The closest mammalian homologue SIRT1, residing in nucleus and cytoplasm, has been shown to play important roles in physiological processes affecting organismal longevity as well as stem cell function and self‐renewal (Chung et al, 2010; Boutant & Canto, 2014).

Although initial reports of organismal longevity prolonging effects of sirtuins in D. melanogaster and C. elegans models remain controversial, it appears that sirtuins participate in many processes that affect life span, such as inflammation, cellular senescence, apoptosis, cell cycle control and changes in energy and oxygen metabolism occurring during ageing and anti‐ageing regimens such as caloric restriction (reviewed in Houtkooper et al, 2012; Guarente, 2013; Chang & Guarente, 2014; Poulose & Raju, 2015).

Tissue homeostasis is dependent on somatic stem cells and age‐dependent tissue degeneration is associated with successive loss of stem cell function and numbers (Liu & Rando, 2011; Oh et al, 2014). Remarkably in this context, SIRT1 appears to be important for the function and self‐renewal capability of a variety of stem cells. SIRT1 protein and mRNA levels are elevated in mouse and human ES cells compared to differentiated tissues (Calvanese et al, 2010; Saunders et al, 2010) and are required for stem cell maintenance (Han et al, 2008). SIRT1 inactivation was also shown to reduce cellular proliferation and to accelerate senescence of human mesenchymal stem cells (MSCs; Yuan et al, 2012). Furthermore, SIRT1 deletion in HSCs was shown to cause characteristic changes normally associated with HSC ageing, such as loss of genomic stability, increased sensitivity to DNA damage, myeloid lineage skewing and premature stem cell exhaustion in serial transplantation assays (Singh et al, 2013; Rimmele et al, 2014). Reversely, overexpression of SIRT1 increased cell proliferation and cell cycle progression in MSCs (Yuan et al, 2012), in spermatogonial stem cells (Niu et al, 2016) and in skeletal muscle stem and progenitor cells (Rathbone et al, 2009). Consistent with this, replenishment of the SIRT substrate nicotinamide adenine dinucleotide (NAD⁺) in old mice reverted neuronal, melanocyte and muscle stem cell exhaustion in a SIRT1‐dependent fashion (Zhang et al, 2016a). Besides other sirtuins that might also influence stem cell self‐renewal, as has been reported for SIRT3 in HSCs (Brown et al, 2013), these observations establish SIRT1 as a key regulator for the maintenance of self‐renewal capacity in stem cells.

Sirtuins are class III histones deacetylases (HDACs) that catalyse the deacetylation of acetyl‐lysine residues of histone proteins in a reaction cleaving the co‐factor NAD⁺ (Haigis & Sinclair, 2010). In addition to histones, SIRT1 also deacetylates non‐histone substrates, including p53, Myc and FoxO transcription factors (Cheng et al, 2003; Motta et al, 2004; Frescas et al, 2005; Kitamura et al, 2005; Mao et al, 2011) and cell cycle regulators such as retinoblastoma protein (Rb; Wong & Weber, 2007), which promotes G1‐S phase transition by release of E2F transcription factors from an inhibitory complex with Rb (Sherr, 1995, 1996; Sherr & Roberts, 1999). The effect of SIRT1 on Myc and FoxO1/FoxO3 transcription factors can regulate several cellular pathways including cell cycle control and stress response (Yuan et al, 2009; Sharma et al, 2012). FoxO family proteins are involved in important cellular processes such as stress resistance, metabolism, cell cycle arrest, apoptosis and longevity (Martins et al, 2016).

Based on these observations and our own data that macrophages share a self‐renewal gene network with stem cells (Soucie et al, 2016), we investigated whether SIRT1 could regulate proliferation and replicative life span in differentiated macrophages. To address this question, we studied the role of SIRT1 in alveolar macrophages that display unlimited expansion potential and in Maf‐DKO macrophages in cell culture, as well as in peritoneal and alveolar macrophages in vivo. We performed loss of function analysis of SIRT1 by shRNA inactivation, CRISPR/Cas9 deletion and chemical inhibition with Inauhzin, a SIRT1 antagonist (Zhang et al, 2012) or with the natural inhibitor nicotinamide (NAM). This amide form of vitamin B3 is generated as a by‐product of NAD⁺ cleavage in the sirtuin catalysed deacetylation reaction and inhibits sirtuin activity by negative feedback (Jackson et al, 2003; Sauve & Schramm, 2003; Avalos et al, 2005).

Using these diverse and complementary approaches, we demonstrate that inhibition of SIRT1 limits the self‐renewal ability of terminally differentiated macrophages in culture and in different macrophage populations in vivo. Reversely, the overexpression of SIRT1 in bone marrow‐derived macrophages improved their proliferative capacity. Furthermore, we show that the positive role of SIRT1 in macrophage self‐renewal involves inhibition of FOXO1 and de‐repression of E2F and Myc pathways.

Results

SIRT1 is required for macrophage self‐renewal

To address the potential role of SIRT1 in regulating self‐renewal in macrophages, we first assessed its expression in Maf‐DKO macrophages. SIRT1 showed threefold higher protein levels in Maf‐DKO cells as compared to wild‐type bone marrow‐derived macrophages (WT BMMs) (Fig 1A and B). Based on its reported role in stem cell self‐renewal (Han et al, 2008; Rathbone et al, 2009; Yuan et al, 2012; Niu et al, 2016; Zhang et al, 2016a), we further proceeded to knockdown SIRT1 by means of specific short hairpin RNA (shRNA) to evaluate the functional importance of SIRT1 for Maf‐DKO macrophage self‐renewal. Two retroviral shRNA constructs against SIRT1 were able to specifically induce a twofold to fourfold down‐regulation of target transcripts (Fig 1C). To test the effect of this down‐regulation on macrophage self‐renewal, we infected Maf‐DKO macrophages and analysed colony‐forming ability in M‐CSF containing semi‐solid medium. SIRT1 shRNA‐infected Maf‐DKO macrophages showed a strong reduction in the colony‐forming ability that correlated with the reduction in sirtuin expression levels (Fig 1D and E). We further focused on anti SIRT1 shRNA‐construct #2, which showed the strongest phenotype in terms of colony size and number (Fig 1D and E) as well as reduction in mRNA (Fig 1C) and protein expression (Fig 1F an G). Using this construct we further analysed the role of SIRT1 in cell cycle regulation. DNA content analysis by propidium iodide staining revealed a reduction of the fraction of cycling cells in SIRT1‐knockdown macrophages from 34% to 23% in S/G2 and an increase from 64 to 75% of cells in G1 phase (Fig 1H and I). We also used CRISPR/Cas9‐mediated deletion of SIRT1 to analysed the impact on alveolar macrophages, a macrophage population that naturally expresses low Maf levels and exhibits self‐renewal capacity (Soucie et al, 2016). When we infected Cas9‐expressing AMs with SIRT1‐specific CRISPR gRNA vectors, we observed 30% less colonies in SIRT1 deleted AMs compared to controls (Fig 1J). Since tide analysis (Appendix Fig S1), revealed SIRT1 deletion in 60% of the cells, this indicated a ~50% reduction in colony formation in SIRT1‐deficient cells. Finally, we performed gain of function experiments by infecting bone marrow‐derived macrophages from mice with constitutive expression of reverse tet transactivator (rtTA) with vectors containing a GFP reporter and a sirt1 gene under the control of a tet responsive element (TRE). This allows doxycycline‐inducible SIRT1 expression and constitutive GFP expression during macrophage differentiation (Fig 2A). Using Ki67 staining, we observed a significant enhancement of proliferative capacity in SIRT1‐expressing macrophages compared to empty vector or uninfected control cells 5 days after infection and doxycycline induction (Fig 2B and C). Taken together, these data show that SIRT1 is a critical mediator of self‐renewal capacity in differentiated macrophages.

Figure 1. SIRT1 inactivation inhibits macrophage self‐renewal.

1. Immunoblot for SIRT1 protein comparing bone marrow‐derived wild‐type (WT BMM) and MafB/c‐Maf double knockout (Maf‐DKO) macrophages. Grb2 antibody was used as loading control.
2. Quantification of panel (A). Shown are Sirt1/Grb2 ratios (arbitrary units, A.U.), normalized to Grb2. Error bars indicate the standard error of the mean. Each condition was done in duplicate; data represent the pool of two independent experiments.
3. Quantitative PCR for the expression of SIRT1 comparing Maf‐DKO macrophages infected with indicated shRNA vectors to non‐infected Maf‐DKO and wild‐type (WT) macrophages. Shown are fold changes of the average values normalized to HPRT of two independent experiments and standard error of the mean.
4. Effect of SIRT1 inactivation on the colony formation potential of Maf‐DKO macrophages. Phase contrast magnification ×10. Each condition was done in duplicate; the results shown are representative of two independent experiments. Scale bars = 50 μm.
5. Quantification of panel (D). Data represent the pool of two independent experiments. Error bars indicate SEM.
6. Immunostaining for SIRT1 (red) on Maf‐DKO macrophages infected with shRNA vectors against LacZ or SIRT1. DAPI (blue) was used to stain DNA. Each condition was done in duplicate; the results shown are representative of two independent experiments. Scale bars = 20 μm.
7. Quantification of panel (F). Error bars indicate SEM.
8. DNA content analysis of Maf‐DKO macrophages infected with shRNA vectors against SIRT1 or LacZ. Each condition was done in duplicate; the results shown are representative of two independent experiments. Table indicates the percentage of cells in indicated cell cycle phases.
9. Quantification of panel (H), represented as ratio between proliferating (S+G2) and resting cells (G1). Data represents the pool of two independent experiments.
10. Analysis of colony formation potential after SIRT1 deletion by CRISPR gRNA vector infection of Cas9 expressing alveolar macrophages. Each condition was done in duplicate. Deletion efficiency of Sirt gRNA_1 and sirt gRNA_2 was evaluated by TIDE analysis (Appendix Fig S1) and corresponds to 60.9 and 44.7%, respectively. Error bars indicate SEM.

Source data are available online for this figure.

Figure 2. SIRT1 overexpression increases macrophage proliferation.

1. Experimental scheme of rtTA BMM differentiation and infection with tetO SIRT1 retrovirus. FACS shows purity of macrophage population as 99.8% F4/80⁺/CD11b⁺ cells.
2. Intracellular FACS staining of Ki67 on rtTA bone marrow‐derived macrophages (CD11⁺ F4/80⁺) infected with retrovirus expressing doxycycline‐inducible SIRT1 and constitutive GFP or empty vector (EV), 5 days post‐infection.
3. Quantification of panel (B). The results shown are representative of two independent experiments. Error bars indicate the standard error of the mean.

Pharmacological SIRT1 inhibition abrogates macrophage self‐renewal capability

We next tested whether pharmacological inhibition with NAM, a natural sirtuin inhibitor, also resulted in the functional impairment of macrophage self‐renewal. NAM is generated as a by‐product of sirtuin deacetylase activity and blocks further catalysis in a negative feedback loop (Bitterman et al, 2002; Avalos et al, 2005). As shown in Fig 3A and B, NAM inhibited the colony formation ability of Maf‐DKO cells in a strictly dose‐dependent manner, as revealed by reduced colony size and numbers, with the highest concentration of NAM (10 mM) completely neutralizing colony formation in M‐CSF. We also found that NAM‐treated Maf‐DKO macrophages showed 16‐fold reduction of 5‐bromo‐2′‐deoxyuridine (BrdU) incorporation (Fig 3C and D), demonstrating impairment of DNA synthesis compared to non‐treated cells. Consistent with this, NAM also resulted in a threefold decrease in the number of cells expressing the nuclear Ki67 antigen, a marker of cycling cells (Fig 3E and F). Finally, analysis of DNA content by propidium iodide staining revealed that NAM treatment reduced the fraction of macrophages in G2/S phase from 28 to 6% and increased the percentage in G0/G1 from 71 to 94% (Fig 3G–I). Together the inhibition of macrophage self‐renewal and S phase transition by NAM were similar to the effects observed for SIRT1 inactivation (Fig 1) and were thus consistent with the ability of NAM to inhibit sirtuin deacetylase activity.

Figure 3. Pharmacological inhibition of macrophage self‐renewal and cell cycle progression.

1. Effect of NAM on the M‐CSF‐dependent colony formation ability of Maf‐DKO macrophages. Phase contrast magnification ×10. Each condition was done in duplicate; the results shown are representative of five independent experiments. NAM was added at day 0 and colonies were counted at day 14. Scale bars = 50 μm.
2. Quantification of panel (A). Data represent the pool of five independent experiments.
3. BrdU incorporation analysis on Maf‐DKO macrophages treated or not with 10 mM NAM for 48 h. Each condition was done in duplicate; the results shown are representative of two independent experiments.
4. Quantification of panel (C). Data represent the pool of two independent experiments.
5. Immunostaining for Ki67 antigen (red) on Maf‐DKO macrophages treated or not with 10 mM NAM for 48 h. DAPI was used to stain DNA. Each condition was done in duplicate; the results shown are representative of two independent experiments. Scale bars = 20 μm.
6. Quantification of panel (E). At least 3,000 cells were counted per condition.
7. DNA content analysis of Maf‐DKO macrophages treated or not with 10 mM NAM for 48 h. Each condition was done in duplicate; the results shown are representative of two independent experiments.
8. Quantification of panel (G), represented as ratio between proliferating (S+G2) and resting cells (G1). Data represent the pool of two independent experiments.
9. Table indicates the percentage of cells in indicated cell cycle phases shown in panel (G). Error ranges indicate SEM.
10. Effect of different compounds NAM, the NAD⁺ precursor nicotinamide mononucleotide (NMN), Olaparib, a PARP inhibitor, and Inauhzin, a SIRT1‐specific antagonist on colony formation ability of Maf‐DKO macrophages. Each condition was done in duplicate; the results shown are representative of two independent experiments.
11. Enhanced histone H3 (Lys9) acetylation after SIRT1 inhibition with NAM. Total cell extracts were prepared from untreated or SIRT1 inhibitor (10 mM NAM, 20 h) treated Maf‐DKO macrophages. Protein blots were incubated with anti‐histone H3 (H3) or anti‐histone H3 K9acetyl (H3K9ac), as indicated. Detection of histone H3 served as controls.

Data information: Error bars indicate the standard error of the mean.Source data are available online for this figure.

However, NAM is not only a sirtuin inhibitor but also the amide form of vitamin B3 and a metabolic precursor of NAD⁺, a co‐factor of the sirtuin deacetylase activity. We therefore wanted to confirm that the observed NAM effect was mediated by sirtuin inhibition and not by effects on NAD⁺ pools. To address this point, we directly compared NAM and the NAD⁺ precursor nicotinamide mononucleotide (NMN) on cell cycle progression. In contrast to NAM, NMN did not show any inhibitory effect on cell cycle progression (Fig 3J). This result was consistent with previous observations showing that NAD⁺ depletion using the drug FK866 also had no effect on Maf‐DKO macrophage proliferation (Venter et al, 2014). NAD⁺ is also a co‐factor of poly(ADP—ribose)polymerases (PARPs), enzymes that regulate DNA damage sensing, apoptosis, cellular stress and ageing, (Saldeen et al, 2003) and similarly release NAM as a by‐product of the enzymatic reaction. In addition to its inhibitory role on sirtuins, NAM also acts as a potent inhibitor of these enzymes via a negative feedback loop. However, we did not observe any effect on macrophage self‐renewal with Olaparib (Fig 3J), a specific inhibitor of PARP with no inhibitory activity on SIRT1, 2, 3 and 6 (Ekblad & Schuler, 2016).

Since NAM can inhibit several sirtuin family members, we also tested Inauhzin, a specific SIRT1 antagonist (Zhang et al, 2012). With this inhibitor, we also observed a similar decrease in colony‐forming activity as with NAM, suggesting that NAM principally acts by SIRT1 inhibition (Fig 3J).

Finally, we verified that NAM affected SIRT1 enzymatic activity by evaluating the lysine 9 acetylation status of histone 3, a well‐known target of SIRT1 deacetylase activity (Imai et al, 2000). Specifically, it has been demonstrated that 10 mM NAM had the same effect on inducible gene expression of WT fibroblasts as genetic SIRT1 inactivation, whereas no additional effect was seen with NAM on SIRT‐deficient cells (Nakahata et al, 2008). In Western blot analysis, we observed a similar increase of H3K9 acetylation in Maf‐DKO macrophages that were treated with NAM for the same time and concentration (Fig 3K). Together with our results on genetic inactivation of SIRT1 (Fig 1), these results indicated that NAM impairs macrophage self‐renewal principally by inhibition of SIRT1 activity.

Reversible inhibition of cell cycle arrest in self‐renewing macrophages

We further investigated whether the inhibition of macrophage self‐renewal by NAM involved an irreversible process, such as apoptosis. However, NAM addition to Maf‐DKO macrophages did not result in increased apoptosis, as judged by the absence of detectable levels of activated Caspase‐3, a classical marker of apoptosis that was easily observed in apoptotic beta islets used as a positive control (Fig 4A and B). To further investigate whether the cell cycle arrest observed upon NAM treatment was a reversible process, we treated Maf‐DKO macrophages with NAM, then washed out NAM and continued the culture in NAM‐free medium prior to cell cycle analysis. As observed before, NAM addition reduced the fraction of cycling cells (Fig 4C–E), but depletion of NAM from the culture medium induced cycle re‐entry of NAM pre‐treated cells (Fig 4C–E). The transitory nature of the NAM inhibitory effect was further confirmed by the re‐acquisition of colony‐forming ability of Maf‐DKO cells upon NAM depletion (Fig 4F and G). Together these data show that NAM treatment only transiently blocked differentiated macrophages in a reversible non‐proliferative state and that self‐renewal could be re‐activated by NAM withdrawal.

Figure 4. Reversibility of the NAM inhibition of macrophage self‐renewal.

1. Immunostaining for activated Caspase‐3 (green) on Maf‐DKO macrophages treated with NAM for 12, 24 or 48 h. Apoptotic beta islets (β) from streptozotocin‐treated mice were used as positive control. DAPI was used to stain DNA. Each condition was done in duplicate; the results shown are representative of two independent experiments. Scale bars = 20 μm.
2. Quantification of panel (A). At least 3,000 cells were counted per condition.
3. DNA content analysis of Maf‐DKO cells pre‐treated with 10 mM NAM (during 48 h) and restimulated with NAM‐free medium. Each condition was done in duplicate; the results shown are representative of three independent experiments.
4. Quantification of panel (C) represented as ratio between proliferating (S+G2) and resting cells (G1). Data represent the pool of three independent experiments.
5. Percentage of cells in indicated cell cycle phases of samples shown in panel (D). Error ranges indicate SEM.
6. Effect of NAM pre‐treatment (48 h) and removal on colony formation ability of Maf‐DKO macrophages. Phase contrast magnification ×10. Each condition was done in duplicate; the results shown are representative of three independent experiments.
7. Quantification of panel (F). Data represent the pool of three independent experiments.

Data information: Error bars indicate the standard error of the mean.

NAM abrogates macrophage proliferation in vivo

We further investigated whether NAM‐mediated SIRT1 inhibition also affected endogenous tissue resident macrophage self‐renewal in vivo. We first investigated the effect of NAM inhibition on steady state proliferation of peritoneal macrophages. The peritoneal cavity contains several macrophage subsets that can be distinguished based on F4/80, CD64, TIM4 and Ly6C staining (Appendix Fig S2A; Bain et al, 2016). Ly6C⁻, F4/80^hi, CD64⁺, Tim4⁻ and Ly6C⁻, F4/80^hi, CD64⁺, Tim4⁺ represent resident peritoneal macrophage populations that in the absence of challenge contain cells of embryonic origin that may autonomously self‐renew for at least several months (Bain et al, 2016). F4/80^lo/−, CD64⁺ and Ly6C⁺ represent monocyte‐derived macrophages. Using Ki67 staining as a marker of cycling cells, we observed that the low rate of steady state proliferation of ~5–7% was reduced by about threefold after intra‐peritoneal NAM injection in both resident peritoneal macrophage populations, whereas monocyte‐derived macrophages showed no significant change (Fig 5A and B, and Appendix Fig S2B). During infection or inflammation peritoneal macrophages may significantly increase the proliferative rates over the steady state (Davies et al, 2011). This can be mimicked by intra‐peritoneal M‐CSF injection that induces increased proliferation of peritoneal macrophages (Davies et al, 2013; Jenkins et al, 2013; Gow et al, 2014; Soucie et al, 2016) by down‐regulation of Maf transcription factors and activation of a self‐renewal gene network (Soucie et al, 2016). We assessed the effect of NAM on this induced proliferative response (Fig 5C and D) by EdU incorporation as a measure for DNA synthesis and observed that in the presence of NAM the M‐CSF‐induced proliferation increase was abolished and reduced to background levels. Finally, we have analysed the effect of NAM on alveolar macrophages (gating strategy in Appendix Fig S3), a population of resident macrophages that is dependent on GM‐CSF for autonomous self‐renewal in vivo (Guilliams et al, 2013; Hashimoto et al, 2013). We also observed a significant twofold reduction of steady state proliferation in this macrophage population in response to NAM (Fig 5E and F).

Figure 5. SIRT1 inhibition by NAM reduces macrophage proliferation in vivo .

1. Intracellular FACS staining of Ki67 on peritoneal macrophages from mice 48 h after I.P. NAM (10 mM) or control injections showing TIM4⁺ and TIM4⁻ resident macrophages.
2. Here is the representation of two‐pooled experiments. Quantification of indicated cycling peritoneal macrophage populations expressed as a fraction of Ki67⁺ cells in control mice (100%). Data were pooled from two independent experiments. Percentage of Ki67⁺ in the respective macrophage populations are shown in Appendix Fig S2.
3. EdU incorporation analysis of M‐CSF stimulated MHCII⁺ CD11b⁺ F4/80⁺ peritoneal macrophages 48 h after NAM or control i.p. injections. Representative samples for each condition are shown (n = 2).
4. Quantification of panel (C). Data shown are pooled from two independent experiments using 4–6 mice per group and experiment.
5. Intracellular FACS staining of Ki67 of alveolar macrophages from mice 48 h after 20 mM NAM or control i.p. injections. Representative samples for each condition are shown (n = 2).
6. Quantification of percentage of Ki67⁺ cells of alveolar macrophages shown in panel (E).

Data information: Statistical significance was tested using a two‐tailed, unpaired, nonparametric Mann–Whitney test. Error bars correspond to the interquartile range (median values). Symbols represent individual mice.

Together our results thus demonstrated that NAM abrogated both steady state and induced proliferation of different resident M‐CSF‐ and GM‐CSF‐dependent macrophage populations, suggesting that SIRT1 is of general importance for macrophage proliferation in vivo.

SIRT1 inhibition affects regulators of macrophage cell cycle progression

In order to identify the molecular pathways affected by SIRT1 in self‐renewing Maf‐DKO macrophages, we further investigated the global changes in gene expression occurring 1 and 10 h after NAM‐induced SIRT1 inhibition (Fig 6A). Besides general signalling and immune function pathways, an enrichment map (Merico et al, 2010) based on GSEA analysis of gene sets derived from the biological process ontology group (BP:GO Biological process; http://software.broadinstitute.org/gsea/msigdb/collections.jsp#C5) also revealed a down‐regulation of genes that are positively regulated in cellular proliferation (Fig EV1). In order to characterize in more detail which regulatory pathways of cellular proliferation might be specifically affected by SIRT1 inhibition, we used the BubbleGUM strategy of gene set enrichment analysis (GSEA; Spinelli et al, 2015) to visualize multiple gene sets from the open source and curated Reactome database of reactions, pathways and biological processes (Haw et al, 2011; Fig 6A). Consistent with the biological data and the gene ontology gene set analysis, this also identified down‐regulation of cell cycle regulators and checkpoints 10 h after NAM addition (Fig 6B). More detailed comparison to gene sets affecting distinct cell cycle phases revealed that NAM caused depletion in particular gene sets of S phase and G1 to S as well as M to G1 transitions (Fig 6C). Consistent with this, analysis of individual cell cycle regulatory genes derived from the descriptive gene ontology term “Cyclins” revealed that 10 h after NAM addition central positive regulators of cycle progression during G1 phase, such as Cdk4, cyclin D1 (CcnD1) and in M phase, like cyclin B1 (CcnB1) were down‐regulated and critical cell cycle inhibitors like p27/CDKN1B, Rb1, RbL1 and RbL2 were up‐regulated (Fig 6D). In particular, Rb is a critical master regulator of the G₁/S transition of the cell cycle through its inhibition of E2F transcription factors (Harbour & Dean, 2000) that control gene targets important for cell cycle progression (Stevens & La Thangue, 2003). Supporting the fact that the activity of E2F factors is not regulated by changes in expression levels but by phosphorylation in late G₁ phase of the cell cycle (Fagan et al, 1994), NAM treatment of Maf‐DKO macrophages did not result in any changes in mRNA expression of E2F factors or its DP dimerization partners (Appendix Fig S4). By contrast, immunofluorescence revealed that NAM‐mediated sirtuin inhibition did not change in E2F1 protein expression levels but strongly reduced levels of its phosphorylated transcriptionally active form (Fig 6E and F). Again, we confirmed that this effect appeared to be mediated mainly by SIRT1, since its shRNA inactivation resulted in nearly identical results to NAM inhibition. This is consistent with the ability of SIRT1 to deacetylate RB and inhibit its ability to sequester the transcription factor E2F1 into an inactive form (Wong & Weber, 2007). The functional importance of this was further indicated by the observation that SIRT1 inhibition resulted in down‐regulation of the E2F target gene set in the hallmark collection (http://software.broadinstitute.org/gsea/msigdb/collections.jsp#H) (Fig 6G). Together these observations demonstrated that the requirement of sirtuin activity for cell cycle progression involved a direct effect of SIRT1 on key regulators of cycle progression that lead to activation of E2F and its downstream targets.

Figure 6. SIRT1 activity controls regulators of macrophage cell cycle progression.

1. Experimental scheme of microarray Maf‐DKO macrophage sample generation and legend for BubbleGum analysis.
2. BubbleGUM analysis on microarray samples (Maf‐DKO macrophages treated or not with NAM for 1 or 10 h) using REACTOME mitosis gene sets.
3. BubbleGUM analysis on microarray samples (Maf‐DKO macrophages treated or not with NAM for 1 or 10 h) using REACTOME cell cycle phase gene sets.
4. Heat‐map showing the scaled expression levels (z‐score) of cyclin and cell cycle regulator related genes that were identified as coherently down‐regulated (blue) or up‐regulated (red) upon the administration of NAM (10 h).
5. Immunostaining for transcriptionally active phosphorylated E2F1 (p‐E2F, green) and inactive non‐phosphorylated E2F (E2F, red). 10 mM NAM was added for 48 h. DAPI was used for DNA staining. Scale bar, 20 μm. The results shown are representative of three independent experiments.
6. Quantification of p‐E2F1 positive cells shown as percentage of total DAPI positive cells (panel E, top). Data shown are pooled from three independent experiments. Error bars indicate SEM.
7. GSEA analysis of untreated versus 10‐h NAM‐treated Maf‐DKO macrophages with HALLMARK E2F transcription factor target gene set.

Figure EV1. GO analysis of cellular processes modulated by NAM addition in Maf‐DKO macrophages.

Enrichment map (Merico et al, 2010), collection of using C5 gene sets (BP:GO Biological process; http://software.broadinstitute.org/gsea/msigdb/collections.jsp#C5) comparing untreated and Maf‐DKO macrophages treated for 10 h with NAM. Node size represents the number of genes in the gene set; edge thickness is proportional to the overlap between gene sets. The enrichment score (enrichment P‐value) is mapped to the node colour as a colour gradient from blue (enrichment in “−NAM”), to red (enrichment in “+10 h NAM”).

SIRT1 inhibition perturbs macrophage self‐renewal pathways

We further investigated whether the effect of SIRT1 on cell cycle progression was associated with mechanisms that control long‐term replicative life span. Towards this end, we focused on a network of self‐renewal genes shared by self‐renewing macrophages and ES cells that we recently identified (Soucie et al, 2016). Indeed, SIRT1 inhibition with NAM resulted in the down‐regulation of ES cell signature gene sets (Fig 7A) that we had found to be enriched in self‐renewing macrophages (Soucie et al, 2016), suggesting that shared self‐renewal mechanism might be affected. Indeed, we identified two significantly modulated clusters of self‐renewal genes: one “early” gene set down‐regulated 1 h after NAM treatment (Klf2, Klf4, Nfya, Cited2, Myc) and a second “late” group of genes (Nfyb, Eed, Cebpz, Ube2f, Cirh1a, Rhoj, Akt1) down‐regulated 10 h after NAM treatment (Fig 7B). These observations thus revealed the majority of previously identified self‐renewal genes (12/16) to be down‐regulated by NAM, including the two genes, Myc and Klf2, that form the central nodes of a cross‐regulatory network of these genes (Soucie et al, 2016; Fig 7B and C). Since no Klf2 pathway genes were annotated in the public Molecular Signatures Database (MSigDB, http://www.broadinstitute.org/gsea/msigdb/index.jsp), we focused on Myc‐regulated genes to further analyse the downstream targets of these central factors. In line with the reduction of Myc expression, SIRT1 inhibition by NAM also resulted in the down‐regulation of Myc‐activated genes annotated in hallmark, canonical pathway and genetic perturbation genes sets (Figs 7D and EV2). Collectively, transcriptomic analysis indicated that in macrophages with long‐term replicative life span SIRT1 activity is required to maintain the expression of a self‐renewal gene network and its downstream target genes.

Figure 7. NAM‐mediated SIRT1 inhibition perturbs macrophage self‐renewal pathways.

1. BubbleGUM analysis on microarray samples of Maf‐DKO macrophages treated or not with NAM for 1 h or 10 h using embryonic and tissue stem cells gene sets (Bhattacharya et al, 2004; Wong et al, 2008).
2. Heat‐map showing the scaled expression levels (z‐score) of self‐renewal core genes (Soucie et al, 2016) in Maf‐DKO macrophages treated or not with NAM for 1 or 10 h.
3. Quantitative PCR for the expression of c‐Myc, Klf2 and Klf4 in Maf‐DKO macrophages treated for the indicated times with 10 mM NAM. Average values of three independent experiments normalized to HPRT. Error bars indicate the standard error of the mean. Each condition was done in duplicate.
4. BubbleGUM analysis on microarray samples of Maf‐DKO macrophages treated or not with NAM for 1 or 10 h using Myc gene set data from hallmark, canonical pathways and chemical and genetic perturbations collections of the Broad Institute MSig database.

Figure EV2. NAM‐mediated SIRT1 inhibition affects Myc target genes.

GSEA analysis with Myc/Max and USF transcription factor target gene sets. V$MYCMAX_01; V$MYCMAX_02; V$USF_C gene sets used.

Reciprocal regulation of E2F1, Myc and FoxOs activities in SIRT1 inhibited Maf‐DKO macrophages

Based on our results showing the inhibition of Myc and E2F pathways by SIRT1 inhibition with NAM, we more specifically analysed annotated gene sets containing consensus‐binding sites for these transcription factors in the promoter regions within ± 2 kb of the transcriptional start site (TSS; C3 TFT collection gene sets; http://software.broadinstitute.org/gsea/msigdb/genesets.jsp?collection=TFT#;). Consistent with our previous results, upon NAM addition we observed depletion of gene sets with E2F and E‐box elements recognized by Myc, Myc/Max or the related USF transcription factors (Figs 8A and EV2). Among the other transcription factors, target gene sets affected by sirtuin inhibition, those containing FoxO binding sites, were prominently represented and enriched rather than depleted (Figs 8A and EV3). When analysing significant changes in expression levels of these genes, we observed that most E2F and Myc target genes were down‐regulated upon sirtuin inhibition, whereas the majority of FoxO target genes were up‐regulated (Fig 8B), in line with the observation that FoxO transcription factors have been shown to induce cell cycle arrest (Schmidt et al, 2002; Hill et al, 2014).

Figure 8. NAM‐mediated SIRT1 inhibition in macrophages results in reciprocal regulation of E2F1, Myc and FoxOs activities.

1. GSEA analysis of untreated versus 10‐h NAM‐treated Maf‐DKO macrophages with E2F1, FoxO1 and Myc transcription factor target gene sets. V$FOXO1_01; V$FOXO1_02; V$E2F_01; V$MYC_Q2 gene sets used.
2. Volcano plot analysis on microarray samples (Maf‐DKO cells treated or not with NAM for 10 h) highlighting E2F1, Fox1 and Myc target genes by GSEA (respectively, green, red and blue dots).

Figure EV3. NAM treatment induces nuclear shuttling of FoxO3 transcription factor.

1. GSEA related to FOXO 4 transcriptional activity. V$FOXO4_01; V$FOXO4_02 gene sets used.
2. GSEA related to FoxO3 transcriptional activity (10 mM NAM 10 h).
3. Immunostaining for FoxO3 (red) in Maf‐DKO macrophages without or after 6 h of 10 mM NAM treatment. DAPI (blue) was used for DNA staining. Asterisks and arrows indicate cytoplasmic and nuclear localization of FoxO3, respectively. The results shown are representative of two independent experiments. Scale bars = 20 μm.
4. Quantification of panel (B). Averages and standard error of the mean of two independent experiments are indicated.

The transcriptional activity of FoxO factors is regulated by post‐translational modifications that regulate shuttling between the cytoplasm and the nucleus, binding ability and the expression of target genes (Barthel et al, 2005; Vogt et al, 2005; Hoekman et al, 2006). Indeed, the transcriptional activation of FoxO target genes correlated with cytoplasmic‐nuclear shuttling upon SIRT1 inhibition (Fig 9A–C). Immunofluorescence analysis revealed that in cycling Maf‐DKO macrophages, transcriptionally inactive FoxO1 was localized in the cytoplasm, whereas SIRT1 inhibition induced FoxO1 nuclear translocation in over 95% NAM‐treated cells (Fig 9A–C). Comparable results were obtained for FoxO3 that also showed enhanced nuclear shuttling upon NAM‐mediated SIRT1 inhibition (Fig EV3B–D). The knockdown of SIRT1 by shRNA induced comparable regulatory effects on cytoplasmic/nuclear shuttling of FoxO1 (Fig 9A–C). Finally, this observation was also confirmed biochemically, showing a significant increase of FOXO1 proteins in nuclear extracts after NAM treatment (Fig 9D).

Figure 9. SIRT1 inactivation in macrophages induces nuclear localization of FOXO1.

1. Immunofluorescence staining of FoxO1 for nuclear and cytoplasmic localization (red) in untreated Maf‐DKO macrophages or after 6‐h treatment with 10 mM NAM or after infection with shRNA SIRT1 #2 vector. FcγR (green) immunofluorescence and DAPI (blue) staining was used to define cellular and nuclear perimeters, respectively. The results shown are representative of three independent experiments. Scale bars = 20 μm.
2. Line profile analysis with ImageJ software of FoxO1 localization within the cells indicated with dashed line in (A). X‐ and y‐axes represent the fluorescence intensity and the position along the line used for the analyses, respectively (unit = pixels).
3. Quantification of nuclear FoxO1 positive cells shown as percentage of total DAPI positive cells. Data shown are pooled from three independent experiments. Error bars indicate the standard error of the mean.
4. FoxO1 nuclear localization after SIRT1 inhibition with NAM. Nuclear extracts or total cell extracts were prepared from untreated or SIRT1 inhibitor (10 mM NAM, 20 h) treated Maf‐DKO macrophages. Protein blots were incubated with anti‐FoxO1 as indicated. Separation of total lysates probed with anti‐FoxO1 served as controls.

Source data are available online for this figure.

Both FoxO and Myc transcription factors have also previously been shown to regulate apoptosis (Hoffman & Liebermann, 2008; Zhang et al, 2011), but consistent with the results shown in Fig 4A and B, no significant enrichment for apoptotic program gene sets could be detected (Appendix Fig S5A). We also did not detect any significant regulation of p53 target genes, another transcription factor that can be regulated by sirtuins and controls cell cycle, DNA repair and apoptosis pathways (Appendix Fig S5B).

Together these data indicated that NAM‐dependent SIRT1 inhibition reciprocally blocks positive transcriptional regulators of cell cycle progression and self‐renewal, while activating inducers of cell cycle arrest.

Discussion

In the present study, we revealed a previously unknown function of SIRT1 in the regulation of macrophage self‐renewal. We have shown that SIRT1, the mammalian sirtuin homologue of the yeast Sir2 protein, is required for the self‐renewal ability and M‐CSF‐ and GM‐CSF‐dependent cell cycle progression of differentiated macrophages in culture and in vivo. We have characterized the phenomenon in three prototypic macrophage populations, alveolar macrophages, peritoneal macrophages and bone marrow‐derived macrophages. Inactivation of SIRT1 revealed its essential role in macrophage self‐renewal, which involved positive regulation of E2F‐ and Myc‐dependent transcriptional pathways, while keeping FoxO transcription factors in a cytoplasmic and transcriptionally inactive state. Furthermore, we observed that macrophage cell cycle progression could be transiently inhibited by NAM, a component of vitamin B3 and natural inhibitor of SIRT1 deacetylase activity, which might be relevant to the integration of metabolism and the regulation of macrophage proliferation in vivo. In addition, our demonstration that a known regulator of organismal longevity is required for macrophage self‐renewal capacity suggests a potential link between macrophage replicative life span and ageing.

Mechanistically, we have shown that the requirement of SIRT1 for macrophage proliferation depends on effects on cell cycle controlling pathways and previously identified self‐renewal mechanisms. We recently discovered that self‐renewing macrophages and ES cells share a network of self‐renewal genes (Soucie et al, 2016). These genes are under the control of macrophage‐specific enhancers that are also present in the quiescent state, but are repressed by c‐Maf and MafB transcription factors. Macrophage self‐renewal therefore depends on low levels of MafB and c‐Maf transcription factors that can occur constitutively in alveolar macrophages or can be induced transiently by cytokines in other tissue macrophage populations (Soucie et al, 2016). Genetically engineered MafB/c‐Maf double‐deficient (Maf‐DKO) macrophages permanently mimic this state and, similar to AMs, can be expanded indefinitely in culture (Aziz et al, 2009; Soucie et al, 2016). In this study, we found increased expression levels of SIRT1 in self‐renewing Maf‐DKO macrophages compared to quiescent wild‐type bone marrow macrophages. Both inactivation with shRNA and CRISPR/Cas9 or pharmacological inhibition with different SIRT1 antagonists revealed that SIRT1 is required for macrophage self‐renewal capacity. Reversely, overexpression of SIRT1 increased the low proliferative capacities of bone marrow‐derived macrophages. Interestingly, SIRT1 has also been shown to positively affect the self‐renewal capacity of several different types of stem cells (Rathbone et al, 2009; Yuan et al, 2012; Niu et al, 2016; Zhang et al, 2016b), including ES and iPS cells (Calvanese et al, 2010; Saunders et al, 2010). These data suggest that high levels of SIRT1 are important for active self‐renewal potential in macrophages and stem cells. Consistent with this, our data demonstrate that SIRT1 inhibition in macrophages results in the reduced expression of ES cell enriched gene sets and specifically the down‐regulation of the majority of self‐renewal genes shared between macrophages and ES cells (Soucie et al, 2016), including the central key factors Myc and KLF2 and downstream targets of Myc. This indicates that SIRT1 holds important control in the regulation of the self‐renewal gene network shared between macrophage and ES cells.

Although the SIRT1 homologue SIR2 has originally been characterized as a histone deacetylase enzyme, it has been subsequently shown that SIRT1 can also deacetylate a wide range of signalling molecules, including transcription factors such as Myc (Haigis & Sinclair, 2010), which might explain the strong effect on the self‐renewal gene network. For example, SIRT1‐dependent deacetylation of Myc has been described to enhance Myc/Max association (Mao et al, 2011) and cell proliferation in other cell systems (Mao et al, 2011; Marshall et al, 2011; Menssen et al, 2012). The direct enzymatic activity of SIRT1 on signalling molecules of cell cycle control might similarly affect cell cycle regulation in self‐renewing macrophages. SIRT1 can deacetylate RB, the master regulator of the G₁/S transition of the cell cycle (Harbour & Dean, 2000), which inhibits its ability to sequester and inactivate the transcription factor E2F1 (Wong & Weber, 2007). This provides an explanation why in our experiments SIRT1 inhibition prevented E2F phosphorylation and the transcriptional activation of E2F1 gene targets that drive cell cycle progression. FoxO transcription factors have also been shown to be subject to diverse post‐translational modifications, including acetylation, that regulate cytoplasmic/nuclear shuttling (Calnan & Brunet, 2008; Webb & Brunet, 2014). Lysine‐deacetylation of FoxO factors by SIRT1 has been shown to prevent their transcriptional activity (Motta et al, 2004; Kitamura et al, 2005). Although E2F1 can activate FoxO gene transcription (Nowak et al, 2007), the mechanism of concomitant sirtuin‐induced retention of FoxO factors in an inactive state in the cytoplasm might prevent FoxO‐mediated target gene activation and cell cycle arrest in self‐renewing macrophages. Consistent with this, we showed that the inhibition of SIRT1 activity induced nuclear accumulation of FoxO1 and target gene activation in Maf‐DKO macrophages. Coordinated SIRT1‐mediated deacetylation of key signalling molecules like Rb, Myc and FoxO transcription factors thus appears to result in the coordinated activation of cell cycle progression, self‐renewal pathways and the inactivation of stress response induced cell cycle arrest.

SIRT3, another member of the sirtuin family, has also been reported to influence stem cell self‐renewal in HSCs (Brown et al, 2013). Since Inauhzin, an antagonist of SIRT1 but not SIRT3 (Zhang et al, 2012), showed a slightly weaker effect than NAM, an inhibitor of both sirtuins, we also tested the effect of SIRT3 shRNA inactivation in Maf‐DKO macrophages. We indeed observed a reduction in the colony‐forming ability but to a lesser degree than the one observed after SIRT1 depletion (Appendix Fig S6). Since Inauhzin and shRNA inactivation of SIRT1 recapitulated to a large extent the phenotype of NAM inhibition, this indicates that SIRT1 is the main sirtuin regulator of macrophage self‐renewal.

Interestingly, the temporary inhibition of SIRT1 by NAM permits an adapted and short‐term restriction of macrophage proliferation without compromising their long‐term self‐renewal capacity. Since NAM is the amide form of the vitamin B3 that is also taken up with food, diet might have an important impact on macrophage proliferation and homeostasis in vivo. The concentration of NAM appears to be critical, with a concentration of 10 mM achieving maximal inhibition of macrophage proliferation in our experiments in culture and in vivo. In human patients, a serum concentration of NAM ranging from 2 to 10 mM was measured after buccal absorption (Spector, 1987). The range of reported normal physiological concentration of NAM in rodents (0.9–2.2 mM) already showed moderate effects on macrophage proliferation in our experiments but lies well below the value of full inhibition (10 mM). This suggests that the normal concentrations of NAM are just at the sensitivity threshold where both reduction and increase of NAM would have significant effects on macrophage proliferation. This might be relevant and important for potential pharmacological intervention.

Here, we demonstrated effects of SIRT1 inhibition on macrophage self‐renewal in several prototypic macrophage populations, including alveolar and peritoneal macrophages in vivo. Although we have not specifically analysed other resident macrophage populations such as microglia, Kupffer cells or Langerhans cells, the observed broad effect of SIRT1 inhibition suggests that SIRT1 activity might also be more generally important in other macrophage subtypes. Given that self‐renewal has been observed in many macrophage populations, NAM and SIRT1 activity might thus be relevant for multiple disease conditions with important macrophage contribution and macrophage proliferation. For example, vitamin B3 deficiency or malnutrition for tryptophan, a precursor for in vivo synthesis of nicotinic acid and NAM, leads to pellagra or related symptoms including weight loss and diarrhoea. This is associated with massive intestinal infiltration of inflammatory cells (Hashimoto et al, 2012) and production of the pro‐inflammatory cytokines IL‐22 and IL‐17 cytokines. In this system, inflammation could be reduced by administration of tryptophan or NAM. Although the target cell of this effect was not described, it is interesting to speculate that the inflammatory phenotype might involve deregulated proliferation of pro‐inflammatory monocytes/macrophage and that its repression by NAM might be beneficial. Atherosclerosis is another example of a pro‐inflammatory condition where macrophage proliferation in the arterial wall drives pathogenic macrophage accumulation (Robbins et al, 2013). NAM‐mediated SIRT1 inhibition and repression of excessive proliferation of macrophages might also be beneficial under these conditions. Consistent with this, genetic myeloid deletion of FoxO factors increased monocyte numbers and exacerbated atherosclerotic lesions (Feinberg, 2013; Tsuchiya et al, 2013). Targeting the SIRT1/FoxOs pathway could thus be promising strategy in these pathologies by suppressing proliferation of inflammatory macrophages and monocytes. Finally, the control of SIRT1 activity on macrophage proliferation could be relevant for neuroinflammation and neurodegenerative diseases, such as Alzheimer disease or multiple sclerosis, where microglia, the macrophages of the central nervous system (CNS), have been shown to be highly disease relevant (Perry et al, 2010; Prinz et al, 2011; Prinz & Priller, 2014; Gomez‐Nicola & Perry, 2016). Microglia cells can undergo massive proliferation and expansion during diverse CNS pathologies and in response to genetic ablation or injury (Ajami et al, 2007; Prinz et al, 2011; Waisman et al, 2015; Gomez‐Nicola & Perry, 2016), but it has not been conclusively determined whether microglia proliferation is beneficial or detrimental in various disease settings.

Conversely, reduced macrophage proliferation capacity might be relevant to various ageing processes. Declining self‐renewal of stem cells with age has been linked to the decreasing capacity for tissue regeneration and general age‐related tissue attrition in multiple organ systems (Oh et al, 2014). The role of sirtuins in longevity mechanisms might thus be related to their more recently described role in stem cell self‐renewal, which has been demonstrated for SIRT3 in HSCs (Brown et al, 2013) and for SIRT1 in a variety of tissue specific (Rathbone et al, 2009; Yuan et al, 2012; Singh et al, 2013; Rimmele et al, 2014; Niu et al, 2016; Zhang et al, 2016a) and embryonic stems cells (Han et al, 2008; Calvanese et al, 2010; Saunders et al, 2010). The yeast sirtuin homologue sir2 was originally shown to extend replicative life span of yeast, and overexpression of the metazoan homologues had been reported to extend life span in worms and flies (Tissenbaum & Guarente, 2001; Kenyon, 2005). Although these earlier claims remained contested, recent findings tend to resolve many discrepancies (Guarente, 2013). Sirtuins mediate effects of caloric restriction that results in life span extending metabolic changes and several tissue‐specific SIRT1‐deficient mouse models indicate that SIRT1 can impact on age‐related physiological decline (Boutant & Canto, 2014). Furthermore increased life span was observed upon brain‐specific overexpression of SIRT1 (Satoh et al, 2013) or repletion of the SIRT substrate NAD⁺ in old mice that improved tissue stem cell function in a SIRT1‐dependent fashion (Zhang et al, 2016a).

SIRT1 activity on histone or non‐histone substrates affects multiple cellular processes relevant to ageing, such as inflammation, apoptosis, adipogenesis, autophagy, glucose homeostasis, genome stability, telomere length, DNA repair, stem cell activity and stress resistance. Among the non‐histone substrates of sirtuins, FoxO transcription factors are known to affect longevity and health span via several mechanisms (Martins et al, 2016). Besides their role in cell cycle arrest, FoxO transcription factors act as oxidative sensors and regulate intracellular redox status by modulating the expression of anti‐oxidant enzymes and DNA damage responses (Charitou & Burgering, 2013; Klotz et al, 2015) as well as autophagy and mitophagy (Pietrocola et al, 2013; Webb & Brunet, 2014). In C. elegans, the Sirtuin gene sir‐2.1 regulates life span extension by deacetylation of the DAF‐16 protein, a FoxO family member (Berdichevsky et al, 2006). Under stress conditions, such as deprivation of growth factors, cells react by inducing nuclear localization of FoxO proteins, leading to (G₁‐S) cell cycle arrest by p27 and p21 activation and cyclin D gene inhibition (Medema et al, 2000; Schmidt et al, 2002; Seoane et al, 2004). In the murine hematopoietic system, FoxO proteins have been demonstrated to restrict proliferation potential and repopulating activity of HSCs (Tothova et al, 2007). Myeloid‐specific FoxO triple knockout mice exhibited increased granulocytes–macrophages progenitor proliferation, monocytosis and increased proliferation with decreased apoptosis of cultured peritoneal macrophages (Tsuchiya et al, 2013). Furthermore, a direct link has been described between FoxO transcription factors and the Akt kinase that is part of a shared gene network regulating self‐renewal both in macrophages and ES cells (Soucie et al, 2016). Several mitogenic stimuli act by Akt‐dependent phosphorylation, nuclear exclusion and degradation of FoxO proteins (Biggs et al, 1999; Accili & Arden, 2004) and Akt signalling is important in macrophage proliferation (Ruckerl et al, 2012). The link between sirtuins and this pathway is further supported by our study, where in proliferating macrophages SIRT1 inhibition induced nuclear localization of FoxO transcription factors, up‐regulation of FoxO targets, cell cycle arrest and down‐regulation of Akt.

It has been proposed that the attrition of stem cell activity and regenerative capacity is a hallmark of ageing (Oh et al, 2014). Here, we have demonstrated that SIRT1 activity is not only required for self‐renewal in stem cells but also in macrophages. Some tissue resident macrophage populations are long‐lived, have self‐renewal capacity (Sieweke & Allen, 2013) and have recently been demonstrated to be required for tissue regeneration (Godwin et al, 2013; Aurora et al, 2014; Petrie et al, 2014). It is therefore tempting to speculate that the general life and health span extending activities of sirtuins might not only depend on counteracting ageing in stem cells but also in macrophages.

Materials and Methods

Cells and culture media

Maf‐DKO macrophages were cultured in DMEM (Life Technologies^®/Invitrogen) containing 10% heat‐inactivated FCS (GE Healthcare/PAA, A15‐101), 1% penicillin–streptomycin, 1% sodium pyruvate and 1% of l‐glutamine (Life Technologies^®/Invitrogen), supplemented with 20% supernatant of M‐CSF producing L‐929 cells (L‐cell sup; Stanley, 1985). Maf‐DKO macrophages were cultured with medium changes every 2–3 days. For functional assays, Maf‐DKO macrophages were seeded in six‐well plates and cultured overnight in M‐CSF conditioned medium at a density of 2 × 10⁵ cells/well. The morning after, cells were stimulated with fresh medium supplemented or not with NAM (Sigma‐Aldrich, N0636) for 48 h. WT BMM macrophages were differentiated from total mouse bone marrow for 15 days in macrophage growth medium (L‐cell sup), selecting for adherent cells every 4 days. rtTA macrophages were differentiated from total rtTA mouse bone marrow (The Jackson Laboratory: Gt(ROSA)26Sor^{tm1(rtTA*M2)Jae}) for 5 days in macrophage growth medium (L‐cell sup) and then infected with SIRT1 expressing retrovirus. For culture of AM, alveolar lavages were pooled from ten 1 ml 37°C BAL washes (PBS without Mg²⁺/Ca²⁺, 2 mM EDTA, 2% FBS; GE Healthcare) per mouse and stored on ice. RBC lysis was then performed at room temperature (RT) for 3 min (RBC Lysis Buffer, Invitrogen). Cells were plated at a density of 1.1 million cells per 10 cm bacterial petri dish in complete medium (RPMI, 10% FCS,1% Pen/Strep, 1% Pyruvate, 1% Glutamate) supplemented with 1% GM‐CSF supernatant from J558L cells transfected with murine GM‐CSF cDNA (a kind gift of Dr. D. Gray, London, UK).

Peritoneal and alveolar macrophages stimulation and isolation

Six‐ to eight‐week‐old female C57BL/6 mice were obtained from Janvier laboratories. All mouse experiments were performed under specific pathogen‐free conditions in accordance with institutional guidelines. For baseline proliferation measurement of alveolar and peritoneal macrophages, mice were injected into the peritoneal cavity with either PBS or NAM (10 mM in 200 μl for PM, 20 mM in 200 μl for AM; Sigma‐Aldrich N0636) or Inauhzin (30 mg/kg; Calbiochem 566332). NAM and PBS control injections were repeated after 24 h. The proliferation capacity of alveolar and peritoneal macrophages was assessed after 48 h. Peritoneal macrophages were harvested by peritoneal wash with PBS containing EDTA (2 mM), incubated with Fc receptor blocking antibody (clone 2.4 G2, BD 553142) and stained for FACS analysis using the following antibodies: anti‐B220 (Clone RA3‐6B2, BioLegend 103221); anti‐Ly6C (Clone AL‐21, BD 553104); anti‐F4:80 (Clone BM8, BioLegend 123141); anti‐CD64 (Clone X54‐5/7.1, BioLegend 139308); anti‐CD11b (Clone HL3, BD 563057); and anti‐Tim4 (Clone 54 (RMT4‐54), Ebioscience 12‐5866‐80). Zombie UV Fixable Viability Kit (BioLegend 423108) was used to stain dead cells. Whole lungs were macerated and incubated with 1 mg/ml Collagenase‐2 (Worthington Biochemical Corporation CLS‐2) and 0.15 mg/ml DNase1 (Roche 10104159001) at 37°C for 30 min during constant agitation. The resulting cell suspension was filtered through a 70‐μm mesh and erythrocytes were removed by ACK lysis. Cells were incubated with Fc receptor blocking antibody (clone 2.4 G2, BD 553142) and stained for FACS analysis using the following antibodies: anti‐SiglecF (Clone E50‐2440, BD 552126) and anti‐CD11c (Clone N418, BioLegend 117349). Zombie UV Fixable Viability Kit (BioLegend 423108) was used to stain dead cells. After extracellular staining, permeabilization was performed using BD Cytofix/Cytoperm™ solution. The proliferation capacity of alveolar and peritoneal macrophages was assessed using antibody directed against Ki67 (Clone 16A8, BioLegend 652411) or a corresponding isotype control (Rat IgG2a kappa Isotype Ctrl, BioLegend 400549). For induced M‐CSF peritoneal macrophage proliferation, mice were injected into the peritoneal cavity with either PBS or pig CSF1‐Fc (1 μg/g body weight; a kind gift from David Hume; Gow et al, 2014) supplemented with or without NAM (10 mM, Sigma‐Aldrich N0636). NAM and PBS control injections were repeated after 24 h. The proliferation capacity of M‐CSF‐induced resident peritoneal macrophages was assessed after 48 h, with intra‐peritoneal injection of EdU (5‐ethynyl‐2′‐deoxyuridine, 2 mg/g body weight; Life Technologies) 4 h before analysis. Cells were harvested by peritoneal wash with PBS containing EDTA (2 mM) and stained with fluorochrome‐conjugated antibodies. Detection of EdU incorporation was performed using the Click‐iT^® Plus EdU Flow Cytometry Assay Kit (Life Technologies) following the manufacturers’ instructions. Cells were acquired on a LSRII UV Flow Cytometer DIVA™ software (both BD) and analysed using FlowJo software (TreeStar).

sgRNA viral vector infections

sgRNAs (designed by CRISPRGold; Chu et al, 2016; targeting Sirt1) were cloned into BbsI sites of the pKLV‐U6gRNA(BbsI)‐PGKpuro2ABFP vector (Addgene #50946). Lentiviral particles were produced by the SFR facility (Lyon, France). Alveolar macrophages were harvested by bronchoalveolar lavage from Rosa26‐Cas9p2aGFP transgenic mice (Chu et al, 2016) and transduced with lentiviral supernatant for 4 h. After 1 day, transduced cells were selected with 2 μg/ml puromycin for 2 days. Selected alveolar macrophages were subjected to fluorescence‐activated cell sorting (FACS) to enrich for BFP^hi macrophages and cultured in absence of puromycin.

shRNA viral vector infections

Plasmids encoding shRNA‐targeting SIRT1 (TRCN0000306512; TRCN0000326966) and SIRT3 (TRCN0000306513; TRCN0000039331) were purchased from Sigma, while the plasmid encoding shRNA‐targeting LacZ (TRCN0000072226) was obtained from The RNAi Consortium (TRC; Broad Institute, Cambridge, MA, USA; Moffat et al, 2006). High‐titer lentivirus encoding shRNA‐targeting genes of interest were produced by the SFR facility (Lyon, France) by triple co‐transfection of plasmids harbouring the packaging construct, the transfer vector (Amit et al, 2009) and the envelope‐expressing construct into producer cells using calcium chloride transfection. Virus was concentrated after transfection, and viral supernatants were harvested and stored at −80°C. Prior to infection, Maf‐DKO macrophages were seeded in 6‐well plate and cultured overnight at a density of 5 × 10⁵ cells/well. The morning after, viral supernatants were incubated with 8 μg/ml of polybrene for 90 min at 37°C, 5% CO₂ and a volume of 4 ml/well/infection was used to replace media on Maf‐DKO macrophages. Cells were transduced by spin infection in virus‐containing medium, by centrifugation at 1,200 g for 2.5 h at 25°C. Viral supernatants were removed immediately after spin infection and replaced with medium, and Maf‐DKO cells were then further incubated at 37°C, 5% CO₂ for 48 h prior to harvesting and divided into fractions for knock down efficiency test (qRT–PCR and immunostaining for the target genes), and functional colony formation assay.

SIRT1 overexpression in rtTA BMMs

Puro^r was replaced by GFP in pRetroX‐tight‐Pur (Clontech, 632104). The pRetroX‐Sirt1‐PGK‐GFP was used for inducible expression of SIRT1 by doxycycline administration (Sigma). Phoenix^® E cells were used for production of retroviral particles following the protocol of the Laboratory of Nolan (www.stanford.edu/group/nolan/) 48 h after transfection. Phoenix cells were cultured in M‐CSF conditioned medium (L‐cell sup), in a 32°C incubator overnight, and supernatant was used to infect rtTA BMMs cells plated at 5 × 10⁵ cells/well in six‐well dish 1 day before infection. Cells were infected by two spin infections in virus‐containing medium (supplemented with 8 μg/ml of Polybrene), by means of centrifugation at 1,200 g for 2 h at 32°C after 5 days of differentiation. Freshly after the last infection round, the viral supernatant was replaced by fresh medium. Five days after infection, cells were FACS analysed for Ki67 as described for in vivo assays. All the infection steps are performed with supplement of 500 ng/ml doxycycline for the induced conditions.

Colony assays

500 Maf‐DKO macrophages were plated in semi‐solid methylcellulose medium (M3231, StemCell™ Technology) containing 100 ng/ml of recombinant murine M‐CSF (Peprotech) and supplemented with different concentrations of NAM (Sigma‐Aldrich, N0636), 1 μM Inauhzin (Merck Chemicals, 566332), 1 μM Olaparib (Euromedex, AZD2281) or 200 μM NMN (Sigma‐Aldrich, N3501). Number and types of colonies were scored between 14 days after plating. Transduced alveolar macrophages were plated in duplicates at a density of 10,000 cells per 1 ml of MethoCult medium (M3231, Stem Cell™ Technologies), with the addition of 100 ng/ml of recombinant GM‐CSF (Peprotech). The number of CFUs was counted on day 14 after plating.

Cell cycle analysis

Maf‐DKO macrophages were harvested with trypsin‐EDTA (Life Technologies^®/Invitrogen), transferred to a V‐bottomed tube, washed once with PBS, fixed with 70% ethanol for 1 h at 4°C, washed twice with PBS, treated with 100 μg of RNase A for 30 min at 37°C, washed once with PBS, and finally stained with 20 μg of propidium iodide (Sigma‐Aldrich, P4864) in PBS. Cells were analysed upon FACS acquisition using linear amplification setup. For BrdU incorporation assay, Maf‐DKO macrophages were cultured as described earlier. During the last 24 h of culture, cells were supplemented with 10 μM BrdU (BD) and incubated at 37°C. BrdU incorporation was detected using the BD BrdU Flow Cytometry^® Kit (BD) according to manufacturer's protocol. Briefly, after collection with trypsin‐EDTA, cells were fixed, permeabilized and washed. DNA was digested with DNase provided with the kit and anti‐BrdU antibody conjugated to FITC was added to the cells. For FACS acquisition, Maf‐DKO macrophages were resuspended in FACS medium (0.5% FCS and 1 mM EDTA in PBS), at a concentration of 1 × 10⁶–1 × 10⁷ cells/ml. For Ki67 intracellular staining, cells were harvested and treated as for BrdU staining but treated with an antibody directed against Ki67 (Clone 16A8, BioLegend 652411) or a corresponding isotype control (Rat IgG2a kappa Isotype Ctrl, BioLegend 400549). Cells were analysed on FACSCanto II (BD) or LSRII and by FlowJo software (Tree Star).

Immunohistochemistry

Maf‐DKO macrophages were seeded in LabTek‐II chamber slides (Nunc) at 1 × 10⁵ cells/well and cultured overnight. The next day, cells were stimulated with fresh medium supplemented or not with 10 mM NAM for 48 h. Cells were fixed immediately with PBS/4% formaldehyde for 10 min at RT. After fixation, samples were washed twice in PBS, and then incubated in PBS/0.25% Triton X‐100 for 10 min. Cells were then incubated for 30 min in PBS/2% BSA/1:100 anti‐CD16/32 (BD, clone 2.4G2) for blocking non‐specific interaction and then overnight with the primary antibodies against Ki67 (Dako, M7249; 1:50), activated Caspase‐3 (Cell Signaling Technology^®, #9644; 1:200), SIRT1 (Cell Signaling Technology^®, #8469; 1:100), phospho‐E2F1 (Abcam, ab55325; 1:200), FoxO1 (Cell Signaling Technology^®, #2880; 1:200) or FoxO3 (Cell Signaling Technology^®, #2497; 1:100). Samples were washed twice in PBS and then incubated 45 min with AlexaFluor^® dye conjugated secondary antibodies (Molecular Probes^®/Invitrogen; 1:100). Finally, cells were washed twice in PBS, dried and mounted with ProLong Gold antifade/DAPI reagent (Molecular Probes^®/Invitrogen). Photomicrographs were taken with multi‐fluorescence Zeiss Axioplan 2 microscope and acquired with SmartCapture 2 software. For positive controls of apoptosis, sections of pancreas from streptozotocin (STZ)‐treated mice (McEvoy et al, 1984) were used. In short, for multi‐low‐dose streptozotocin (STZ) experiments, adult males (6‐ to 8‐week‐old) were fasted for 4 h prior to i.p. injection with 40 mg/kg of body weight of STZ prepared in Na‐citrate buffer 0.1 mol/l (pH 4.5) over five consecutive days and diabetes was monitored for 2 weeks after the last injection. Pancreases were fixed overnight at 4°C in 4% formalin (v/v) and for 1–2 days in sucrose 20% (w/v) at 4°C, before embedding in Tissue Tek (OCT Compound, Sakura Finetek Europe, Zoeterwoude, the Netherlands). Whole pancreases were serially sectioned and stained as described above. Percentage of positive stained cells was evaluated using Cell Counter Plug‐in of ImageJ software.

Preparation of cell extracts and nuclear extracts

Cells were washed twice with phosphate‐buffered saline (PBS) and resuspended in lysis buffer [20 mM HEPES pH 7.8, 150 mM NaCl, 1 mM EDTA pH 8, 10 mM MgCl₂, 0,5% NP40, 10% Glycerol, 20 U/ml Benzonase (Sigma), protease inhibitor cocktail (Roche), 1 mM DTT, 1 mM PEFA bloc (Böhringer), 1 μM trichostatin A (TSA), 10 mM NAM (Sigma), 10 mM Na‐butyrate (Sigma)]. After incubation on ice for 30 min cell debris was removed by centrifugation at 16,000 g for 20 min. Nuclear extracts were prepared as described (Schreiber et al, 1989). Briefly, PBS‐washed cells were incubated with hypotonic buffer (20 mM HEPES pH 7.8), 1.5 mM MgCl₂, 10 mM KCL, protease inhibitor cocktail, 1 mM DTT, 1 mM PEFA, 1 μM trichostatin A (TSA), 10 mM NAM, 10 mM Na‐butyrate), 0.1% NP40 was added and the cytoplasmic fraction was removed by short centrifugation. Nuclei were resuspended in hypertonic buffer (20 mM HEPES pH 7.8, 400 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA pH8, 20% glycerol, protease inhibitor cocktail (Roche), 1 mM DTT, 1 mM PEFA bloc, 1 μM trichostatin A (TSA), 10 mM NAM, 10 mM Na‐butyrate), incubated on rocker on ice for 30 min, and supernatants were recovered as nuclear extracts after centrifugation at 16,000 g for 20 min. For Western blotting experiments, the primary antibodies used were Anti‐FoxO1 (Cell Signaling #C29H4), Anti‐H3 (Abcam #ab 1791) and Anti‐H3 K9ac (Abcam #ab 10812).

RNA extraction and quantitative RT–PCR

Lentivirally infected Maf‐DKO macrophages were seeded in six‐well plates and cultured overnight at a density of 2 × 10⁵ cells/well. The morning after, cells were stimulated with fresh medium for 4 h prior to RNA isolation. RNAs were extracted using Qiagen RNeasy Mini Kit (Qiagen). 500 ng of RNA from each sample was used for Reverse Transcription (Superscript II^®, Invitrogen). Oligo(dT) was used to perform reverse transcription. For quantitative real‐time PCR, we used Power SYBR Green PCR Master Mix and a 7,500 Fast Real‐Time PCR System sequence detection system (both Applied Biosystem), following the manufacturer's instructions. Amplification of the cDNA was performed by using primers for HPRT (5′‐GGCCCTCTGTGTGCTCAAG‐3′, Fwd; 5′‐ CTGATAAAATCTACAGTCATAGGAATGGA‐3′, Rev), sirt1 (5′‐AACAATTCCTCCACCTGAGC‐3′, Fwd; 5′‐TCCCACAGGAGACAG AAACC‐3′, Rev), sirt3 (5′‐ ACAGCTACATGCACGGTCTG‐3′, Fwd; 5′‐ACACAATGTCGGGTTTCACA‐3′, Rev), c‐myc (5′‐CAGAGGAGGAACAACGAGCTGAAGCGC‐3′, Fwd; 5′‐TTATGCACCAGAGTTTCGAAGCTGTTCG‐3′, Rev), klf2 (5′‐ACCAAGAGCTCGCACCTAAA‐3′, Fwd; 5′‐GTGGCACTGAAAGGGTCTGT‐3′, Rev), klf4 (5′‐ATGGTCAAGTTCCCAGCAAG‐3′, Fwd; 5′‐GGGCATGTTCAAGTTGGATT‐3′, Rev).

Microarray analysis

Maf‐DKO macrophages were seeded in six‐well plates and cultured overnight at a density of 2 × 10⁵ cells/well. The next day, cells were stimulated with fresh medium supplied (or not) with 10 mM NAM for 1 or 10 h prior to RNA isolation. Total RNA of Maf‐DKO cells (n = 3 biological replicates for each sample) was extracted as described earlier. The quality of RNA was assessed using Agilent 2100 Bioanalyzer using Eukaryote Total RNA Nano Chip platform (Agilent Technologies). The extracted RNA had a RIN > 9. For each sample, 200 ng of RNA was processed by the Microarray and Sequencing Platform (IGBMC, Strasbourg, France), and the obtained biotinylated cDNAs were hybridized on Mouse Gene 1.0 ST array Chip (Affymetrix). Normalization of raw Affymetrix expression data (extracted from CEL files) was performed by Robust Multi‐chip Analysis (Irizarry et al, 2003) through Bioconductor (release 2.9) in the R statistical environment (version 2.14) via the Affy package. Probe sets over a normalized expression value of 6 (log2 scale) in at least one sample across all arrays were considered as expressed and were included in the analysis. At the end, 15,363 probes were kept for the subsequent analysis. In order to select the regulated probes between conditions, LiMMA package (Linear Model for Microarray Data) using an empirical Bayes method was used (Smyth, 2004). Genes having an adjusted P‐value inferior to 0.05 were considered as regulated. The Gene Set Enrichment Analysis (GSEA) method from the Massachusetts Institute of Technology (http://www.broad.mit.edu/gsea) is based on a Kolmogorov–Smirnov test (Subramanian et al, 2005). This method was used to statistically test whether a set of genes of interest was distributed randomly in the large list of genes (corresponding to the 15,363 probes) sorted on the basis of the log difference of their expression level between “10 h NAM‐treated” and “no NAM control” culture conditions. Transcription factor targets gene sets (C3) from MSigDB Collections (TFT collection gene sets; http://software.broadinstitute.org/gsea/msigdb/genesets.jsp?collection=TFT#;) were used for these analyses. Each gene set of interest (V$FOXO1_01; V$FOXO1_02; V$FOXO3_01; V$FOXO4_01; V$FOXO4_02; V$E2F_01; V$USF_C; V$MYC_Q2; V$MYCMAX_01; V$MYCMAX_02) contains genes that share a transcription factor‐binding site defined in the TRANSFAC database (version 7.4, http://www.gene-regulation.com). For heat‐map representation, multiple probe sets corresponding to each gene were aggregated by their median expression value. BubbleGUM is an open‐source software performing numerous GSEA analysis with multiple testing correction (Spinelli et al, 2015). The Reactome pathway database (http://www.reactome.org) was chosen as gene set compendium for its accuracy to describe cell cycle phases and events. Microarray data are available at (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE60187).

Author contributions

FI designed and performed the majority of experiments, analysed data, contributed to writing of the article and prepared figures. JM analysed bioinformatics data. SVA performed in vivo proliferation assays and viral vector infections. CJB and JF performed CRISPR/Cas9 and CFC assays. WC provided technical help for in vivo experiments, viral vector infections, cell culture and RT–QPCR. EK‐L performed biochemical acetylation and nuclear localization assays and RG performed some of the in vivo proliferation assays. PP provided technical help. AL provided constructive discussions. CB contributed to experiments, analysed data, designed and supervised the study, and wrote the article. MHS designed and supervised the study, analysed data and wrote the article.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Review Process File

Source Data for Figure 1

Source Data for Figure 3

Source Data for Figure 9

The EMBO Journal (2017) 36: 2353–2372