See Supplementary Information.

Animal experiments were approved by the Service des Affaires Vétérinaires, Direction Générale de l'Agriculture, de la Viticulture et des Affaires Vétérinaires (DGAV), état de Vaud (Epalinges, Switzerland; authorizations 876.9 and 877.9) and performed according to Swiss and ARRIVE guidelines (http://www.nc3rs.org.uk/arrive-guidelines).

C57BL/6J mice were from Charles River Laboratories (Saint-Germain-sur-l'Arbresle, France). SIRT3^−/− and SIRT5^−/− C57BL/6J mice were described (41, 42) and obtained from Prof. Johan Auwerx, Laboratory for Integrative and Systems Physiology, Ecole Polytechnique Fédérale de Lausanne, Lausanne, Switzerland. SIRT3^−/− males were crossed with SIRT5^−/− females. Thirty-two SIRT3/5^+/− females were crossed with 16 SIRT3/5^+/− males. Among the 205 F2 mice, 4 males and 8 females were double knockout mice and used to establish the SIRT3/5^−/− mouse line. All mice used in this study were 7–14-week old, housed under specific pathogen-free conditions and exempt of mouse hepatitis virus and murine norovirus. For genotyping purposes, DNA was extracted and analyzed by PCR using the Mouse Direct PCR Kit (Bimake, Houston, TX) and primers pairs described in Supplementary Information.

Bone marrow (BM) cells were cultured 7 days in IMDM or RPMI 1640 supplemented with 100 IU/ml penicillin, 100 μg/ml streptomycin (Invitrogen, San Diego, CA), 10% heat inactivated fetal bovine serum (Biochrom GmbH, Berlin, DE) and 50 U/ml macrophage colony-stimulating factor (ImmunoTools, Friesoythe, Germany) or 30% L929 cell supernatant to generate BM-derived macrophages (BMDMs) (43, 44). Cells were seeded in half-area 96-well plates (2.5 × 10⁴ cells/well), 96-well plates (2 × 10⁵ cells/well) and 6-well plates (3 × 10⁶ cells/well) without growth factors. Neutrophils were isolated from the bone marrow using the Neutrophil isolation kit (Miltenyi, Bergisch Gladbach, Germany) and plated in 96-well plates (10⁵ cells/well). Salmonella minnesota ultra pure lipopolysaccharide (LPS) was from List Biologicals Laboratories (Campbell, CA), Pam₃CSK₄ from EMC microcollections GmBH (Tübingen, Germany), and CpG ODN 1826 (CpG) and poly(I:C) from Invivogen (San Diego, CA). Monosodium urate (MSU) crystals were prepared as described (45). Listeria monocytogenes 10403 s was grown in brain heart infusion broth (BD Biosciences, Erembodegem, Belgium). Bacteria were washed with 0.9% NaCl and adjusted at 10¹⁰ cfu/ml. When required, bacteria were heat-inactivated for 2 h at 70°C.

RNA was extracted (RNeasy kit) and reverse transcribed (QuantiTect reverse transcription kit) (Qiagen, Hilden, Germany). PCRs were performed in triplicate with 1.25 μl cDNA, 1.25 μl H₂O, 0.62 μl primers [Supplementary Information and (46)] and 3.12 μl KAPA SYBR Green Fast (Kapa Biosystems, Wilmington, MA) using a QuantStudio™ 12K Flex system (Life Technologies, Carlsbad, CA). Gene expression was normalized to actin expression.

Total and nuclear proteins were extracted, submitted to PAGE and transferred onto membranes as described (47, 48). Membranes were incubated with primary and secondary HRP-coupled antibodies and revealed by chemiluminescence [Supplementary Information and (49)]. Images were recorded with a Fusion Fx system (Vilber Lourmat, Collégien, France). Full length blots are presented in Supplementary Figure S1.

Single cell suspensions from thymus, spleen and BM were incubated with 2.4 G2 to block Fc receptors and stained with antibodies described in Supplementary Information (50). For hematopoietic stem cells (HSC) and progenitor cells, lineage cocktail contained antibodies directed against CD45R (B220), CD3e, CD11b, CD19, Ly6C/G, Ter119/Ly-76. Data were acquired using an Attune Nxt flow cytometer (ThermoFisher, Waltham, MA) and analyzed using FlowJo version 10.2 (FlowJo LLC, Ashland, OR). Gating strategies are presented in Supplementary Figure S2.

BMDMs were plated in half-area black 96-well plates in RPMI without phenol red (Invitrogen). Cells were incubated for 10 min at 37°C with 5 μM MitoSOX (Thermofisher). Stimuli were added and fluorescence (Ex₅₁₀, Em₅₈₀) recorded using a Synergy plate reader (BioTek, Winooski, VT). Neutrophils in HBSS without calcium and magnesium (ThermoFisher) were incubated for 1 h with 100 nM PMA (Enzo Life Sciences, Farmingdale, NY) and 5 μM MitoSOX during the last 10 min of incubation. ROS were measured by flow cytometry.

Cytokines were quantified by ELISA (Supplementary Information) or Luminex using a custom ProCarta kit (ENA-78/CXCL5, G-CSF, IFNγ, IL-1α, IL-1β, IL-3, IL-6, IL-10, IL-12p40, IL-17A, IL-18, IP-10/CXCL10, KC/CXCL1, MCP-1/CCL2, MIP-1α/CCL3, MIP-2/CXCL2, TNF) (Invitrogen, Carlsbad, CA) and a bioplex 200 system (Bio-Rad, Hercules, CA) (51).

The metabolic activity of BMDMs was measured using the XF Cell Mito Stress, Glycolysis Stress and Mito Fuel Flex Test Kits on a 96-well format Seahorse XFe^® system (Agilent Technologies, Santa Clara, CA) (46).

Neutrophils were incubated with live L. monocytogenes for 1 h in RPMI medium. Serial dilutions of reaction mixtures were plated on blood agar plates (BD Biosciences). Twenty-four hours later, colonies were enumerated. To measure NETosis, neutrophils were incubated for 3 h with 100 nM PMA and 5 μM of the cell impermeable dye Sytox green. Fluorescence (Em₅₀₄, Ex₅₂₃) was recorded using the Synergy plate reader.

Listeriosis was induced by challenging intravenously (i.v.) age and sex-matched mice with a low (7.3 × 10³ cfu) or a high (0.9–1 × 10⁵ cfu) inoculum of L. monocytogenes. Blood and organs were collected 1–3 days post-infection to quantify bacteria and cytokines and analyze cell populations. A model of endotoxemia was developed by challenging mice intraperitoneally (i.p.) with 10 mg/kg LPS. Body weight loss, severity score and survival were registered at least twice daily (52, 53) by 2–3 operators. The severity score was graded from 0 to 4 based on the mobility, the posture, the appearance and the weight loss of mice (detailed criteria were approved by the Service des Affaires Vétérinaires, DGAV, and are available upon request). Mice were sacrificed when they met a severity score of 4. A mice found dead was assigned a score of 5.

Groups were compared by variance analysis followed by two-tailed unpaired Student's t-test or a Mann-Whitney test when appropriate. Survival was analyzed using the Kaplan-Meier method. P < 0.05 was used to indicate statistical significance. Analyses were performed using PRISM 8.0.1 (GraphPad Software, San Diego, CA).

We generated a SIRT3/5 double knockout mouse line (SIRT3/5^−/−, see Materials and Methods) to study the interaction between SIRT3 and SIRT5. Genomic-DNA based PCR genotyping (Figure 1A) and Western blotting analyses (Figure 1B) confirmed the truncation of the Sirt3 and Sirt5 genes and the absence of SIRT3 and SIRT5 protein expression in SIRT3/5^−/− mice. Fecundity and development were normal. The size (Figure 1C) and the female/male sex ratio (Figure 1D) of the litters as well as the weight of adult female and male mice (Figure 1E) were like those of SIRT3/5^+/+, SIRT3^−/−, and SIRT5^−/− mouse lines. Autopsy did not reveal gross abnormalities in SIRT3/5^−/− mice.

The dual deletion of SIRT3 and SIRT5 was not compensated by an increased expression of mRNA encoding for SIRT1, SIRT2, SIRT4, SIRT6, and SIRT7 in bone marrow derived macrophages (BMDMs) (Figure 1F). There was around 20% increase of total protein acetylation in SIRT3/5^−/− BMDMs (Figure 1G). SIRT3/5 deficiency did not alter the mitochondrial mass, evaluated by measuring mitochondrial ribosomal protein L19 (Mrpl19) mRNA levels and the fluorescence intensity of the mitochondrial dye MitoTracker (Figure 1H). The oxygen consumption rate (OCR), which reflects mitochondrial respiration, was weakly increased in SIRT3/5^−/− BMDMs (Figure 1I). The extracellular acidification rate (ECAR), a readout of the glycolytic activity, was not affected in resting and LPS stimulated SIRT3/5^−/− BMDMs (Figure 1I and Supplementary Figure S3A). OCR and ECAR were similarly affected in resting SIRT3^−/− BMDMs (Supplementary Figure S3B) and SIRT5^−/− BMDMs (38). The dependency, capacity, and flexibility of BMDMs to oxidize the mitochondrial fuels glucose, glutamine and fatty acids were identical for SIRT3/5^+/+ and SIRT3/5^−/− BMDMs (Figure 1J). SIRT3/5^+/+ and SIRT3/5^−/− BMDMs expressed similar levels of a number of genes encoding for molecules involved in the fatty acid metabolism, i.e., Acadl (acyl-CoA dehydrogenase long chain), Cpt1 (carnitine palmitoyltransferase 1), Fabp4 (fatty acid binding protein 4), Hadha (hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit alpha), Hmgcr (3-hydroxy-3-methylglutaryl-CoA reductase), Mvd (mevalonate diphosphate decarboxylase) and Sqle (squalene epoxidase) (Figure 1K). Finally, the OCR of SIRT3/5^+/+ and SIRT3/5^−/− BMDMs was not different before and after addition of etimoxir, an inhibitor of Cpt1 (88.0 vs. 85.5% inhibition in SIRT3/5^+/+ vs. SIRT3/5^−/− BMDMs; n = 6; P = 0.3). Hence, SIRT3/5 deficiency had no strong impact on basic metabolic parameters of BMDMs.

The bone marrow is the main source of hematopoietic stem cells (HSC) and progenitors of immune cells during adulthood (54, 55). The number of CD45⁺ hematopoietic cells per leg (femur + tibia) was identical in SIRT3/5^+/+, SIRT3^−/−, SIRT5^−/−, and SIRT3/5^−/− mice (Figure 2A), as well as the composition of the HSC pool which is made of lineage negative, Sca1 positive, c-kit positive (LSK) cells, long-term (LT)-HSC, short-term (ST)-HSC, and multipotent progenitors (MPP) (Figure 2B). Accordingly, the percentage and the absolute number of T cells, B cells, neutrophilic granulocytes and monocytes were similar in SIRT3/5^+/+ and SIRT3/5^−/− mice, while some minor changes were observed in SIRT3^−/− and SIRT5^−/− mice (decreased CD3⁺ T cells and CD19⁺ B cells and increased monocytes) (Figure 2C and Table 1).

SIRT3/5^−/− mice, but not single knockout mice, showed a slight reduction of thymus cellularity when compared to SIRT3/5^+/+ mice (SIRT3/5^+/+ vs. SIRT3/5^−/− mice: 12.7 ± 1.8 vs. 8.3 ± 1.2 million cells; P = 0.03). However, the proportion of CD4/CD8 single positive, double positive and double negative (DN1-DN4) thymocytes was comparable in all mouse lines (Table 2). The size of the spleen (SIRT3/5^+/+, SIRT3^−/−, SIRT5^−/−, and SIRT3/5^−/− mice: 6.1 ± 1.5, 5.3 ± 1.1, 7.3 ± 1.6, and 7.0 ± 1.8 × 10⁷ cells; P > 0.05 for all) and the proportion of total T cells, B cells, dendritic cells (DCs), neutrophilic granulocytes and monocytes were not affected in SIRT3/5^−/− mice (Table 3). Small, statistically significant, differences were noticed between SIRT3/5^+/+ and SIRT3/5^−/− mice, i.e., reduced percentages of effector memory CD4⁺ T cells (23.1 ± 3.3 vs. 17.2 ± 2.3% of CD4⁺ T cells), CD4⁻CD8⁻ T cells (4.2 ± 0.9 vs. 3.0 ± 0.2% of CD3⁺ T cells) and conventional DC1 (cDC1, 35.4 ± 6.3 vs. 27.2 ± 3.0% of CD11c⁺ DCs) (Table 3). Like SIRT3/5^−/− mice, SIRT3^−/− mice showed a reduced percentage of cDC1, while SIRT5^−/− mice showed an increased percentage of Ly6C^low (alternative) monocytes at the expense of Ly6C^high (inflammatory) monocytes. Collectively, these results suggested that the dual deletion of SIRT3 and SIRT5 had a subtle impact on the development of immune cells.

Macrophages and neutrophils are proficient at sensing microbial products and play key defense roles during infections. SIRT3 and SIRT5 single deficiency did not influence antimicrobial host defenses (37, 38). Therefore, we asked whether dual deficiency of SIRT3 and SIRT5 would reveal a phenotype unseen in single knockouts. SIRT3/5^+/+ and SIRT3/5^−/− BMDMs were exposed to LPS, CpG, and poly(I:C), which are sensed through TLR4, TLR9, and TLR3, respectively. SIRT3/5^−/− BMDMs produced higher levels of TNF, IL-6, and IL-12p40 (as a trend for CpG-induced IL-12p40) and lower levels of IL-10 than SIRT3/5^+/+ BMDMs in response to LPS and CpG, while SIRT3/5^+/+ and SIRT3/5^−/− BMDMs produced similar levels of TNF and IL-6 in response to poly(I:C) (Figure 3A).

To address whether the increased inflammation driven by SIRT3/5-deficiency was linked to increased intracellular signaling, we quantified by Western blotting the nuclear translocation of NF-κB p65 and the phosphorylation of ERK1/2 and p38 MAPKs in BMDMs exposed for 0, 10, 30, and 60 min to LPS (Figure 3B). In SIRT3/5^−/− BMDMs, there was an increased NF-κB p65 nuclear content at baseline and, albeit not statistically significant, after 30 and 60 min of stimulation. The level of phospho-ERK1/2 was also increased after 30 min of exposure to LPS. Overall, SIRT3/5 deficiency increased inflammatory intracellular signaling pathways and inflammatory cytokine production by BMDMs.

ROS activate the NOD-like receptor pyrin domain-containing-3 (NLRP3) inflammasome that cleaves pro-IL-1β into mature IL-1β that is secreted. Considering that SIRT3 and SIRT5 activate enzymes playing a role in the detoxification process of ROS (i.e., IDH1, IDH2, SOD1, and SOD2) (17–21), we questioned whether SIRT3/5^−/− BMDMs produce increased levels of mitochondrial ROS (mtROS) and IL-1β. SIRT3/5^+/+, SIRT3^−/−, SIRT5^−/−, and SIRT3/5^−/− BMDMs were exposed to monosodium urate (MSU) crystals, a commonly used activator of the NLRP3 inflammasome before measuring mtROS (Figure 3C). MSU crystals induced mtROS equally in SIRT3/5^+/+ and SIRT5^−/− BMDMs, 1.5-fold more in SIRT3^−/− BMDMs and 2-fold more in SIRT3/5^−/− BMDMs. mRNA levels of Idh1, Idh2, Sod1, and Sod2 were similar in SIRT3/5^+/+ and SIRT3/5^−/− BMDMs (Figure 3D), in agreement with the fact that sirtuins target the activity rather than the expression of IDH1, IDH2, SOD1, and SOD2. As expected from the above, SIRT3/5^−/− BMDMs secreted higher levels of IL-1β than SIRT3/5^+/+ BMDMs (Figure 3E).

The role of sirtuins in the development and functions of granulocytes is scarce and has not been reported for SIRT5 (56, 57). The mRNA expression levels of SIRT3 and SIRT5 decreased gradually from common myeloid progenitors (CMP) to granulocyte-monocyte progenitor (GMP; 1.6-fold) and from GMP to neutrophilic granulocytes (5.3–5.7-fold) (Figure 4A). Compared to SIRT3/5^+/+ mice, SIRT3/5^−/− mice expressed in the bone marrow statistically significantly more CMP (P = 0.03) but not GMP (P = 0.06) (Figure 4B). Accordingly, SIRT3/5^−/− mice expressed normal numbers of neutrophilic granulocytes in the bone marrow and spleen (Figure 2C and Table 3). We then addressed whether SIRT3/5 deficiency affected neutrophil functions. We setup a killing assay in which neutrophils were incubated for 1 h with Listeria monocytogenes before quantifying bacteria. As shown in Figure 4C, 115, 96, 77, and 66% of the starting inoculum were recovered from assays using SIRT3/5^+/+, SIRT3^−/−, SIRT5^−/−, and SIRT3/5^−/− neutrophils, respectively. Hence, the dual deletion of SIRT3 and SIRT5 promoted the killing of L. monocytogenes by neutrophils. This prompted us to analyze two main mechanisms through which neutrophils kill bacteria, i.e., the production of ROS and the release of neutrophil extracellular traps (NETs). SIRT3/5^−/− neutrophils, and to a lesser extent SIRT3^−/− and SIRT5^−/− neutrophils, produced increased levels of ROS when compared to SIRT3/5^+/+ neutrophils (Figure 4D). In contrast, SIRT3/5^+/+, SIRT3^−/−, SIRT5^−/−, and SIRT3/5^−/− neutrophils produced similar amounts of NETs (Figure 4E). Thus, the proficient killing of L. monocytogenes by SIRT3/5^−/− neutrophils was more likely related to an increased generation of ROS than NETs. Lastly, we measured by Luminex the cytokines released by whole blood exposed to heat-killed L. monocytogenes. Fifteen of 17 mediators were produced at measurable levels: CCL2, CCL3, CXCL1, CXCL5, CXCL10, G-CSF, IFNγ, IL-1α, IL-1β, IL-6, IL-10, IL-12p40, IL-17A, IL-18, and TNF. Going well along with an increased inflammatory response of BMDMs (Figure 3), SIRT3/5^−/− whole blood produced more G-CSF and showed a trend toward producing more IL-1α, IL-6, and IFNγ than SIRT3/5^+/+ whole blood (Figure 4F). The production of other cytokines was not affected.

Myeloid cells play a crucial role in protecting from L. monocytogenes infection (58, 59). Considering that SIRT3/5 deficiency increased L. monocytogenes killing by neutrophils (Figure 4C) and cytokine production by macrophages and to some extent by whole blood (Figures 3A, 4F), we tested the relevance of these observations in vivo using a model of listeriosis. Mice challenged intravenously with a high inoculum of L. monocytogenes (0.9–1.5 × 10⁵ cfu) exhibited signs of disease 36–40 h post-infection and were severely sick after 65 h as shown by elevated severity scores in most animals. SIRT3/5^−/− mice were not as ill as SIRT3/5^+/+ mice (P = 0.01; Figure 5A) and had 2.3-fold less L. monocytogenes in blood collected 48 h after infection (SIRT3/5^+/+ vs. SIRT3/5^−/−: 3.5 ± 0.9 × 10³ cfu/ml vs. 1.5 ± 0.3 × 10³ cfu/ml; median ± SEM; P = 0.005) (Figure 5B). In line with these observations, SIRT3/5^−/− mice had a modest, statistically significant, delayed mortality rate compared to SIRT3/5^+/+ mice (median survival of SIRT3/5^+/+ vs. SIRT3/5^−/−: 3.0 vs. 3.12 days; P = 0.01) (Figure 5C). Going well along with an improved response to infection, SIRT3/5^−/− mice had significantly higher blood concentrations of TNF at day 1 (P = 0.001), TNF and IL-1β at day 2 (P = 0.004 and 0.03) and KC/CXCL1 at day 3 (P = 0.05). Albeit not statistically significant, the levels of G-CSF, IL-1α, MCP-1/CCL2, MIP-2/CXCL2 at day 1, G-CSF, KC, MCP-1, and MIP-2 at day 2, and IL-1α, IL-1β, IL-6, IL-10, MCP-1, MIP-2, and TNF at day 3 post-infection were 1.5–4.4-fold higher in SIRT3/5^−/− than in SIRT3/5^+/+ mice (Figure 5D). As expected, 2 days after the onset of infection, listeriosis induced a dramatic drop of circulating leukocytes including B cells, T cells, neutrophils and Ly6C^high inflammatory monocytes that rebounded at day 3 (Figure 5E). No differences were observed between SIRT3/5^+/+ and SIRT3/5^−/− mice, suggesting that the reactivity rather than the number of blood leukocytes conferred some protection to SIRT3/5^−/− mice during listeriosis. Interestingly, there was no statistically significant difference of mortality when SIRT3/5^−/− and SIRT3/5^+/+ mice were challenged with a low inoculum (7.3 × 10³ cfu) of L. monocytogenes responsible for <25% death (Figure 5C). Finally, we tested the mouse lines in a model of endotoxemia induced by an intraperitoneal challenge with 10 mg/kg LPS. Surprisingly, there was no statistically significant difference between the SIRT3/5^+/+ and SIRT3/5^−/− groups (n = 21–22 mice per group; P = 0.1).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.