Mitochondrial transfer to granulocytic myeloid-derived suppressor cells augments immunosuppressive activity

SUMMARY

The anti-inflammatory properties of granulocytic myeloid-derived suppressor cells (G-MDSCs) promote Staphylococcus aureus (S. aureus) biofilm persistence. Evidence suggests that G-MDSC activity is shaped not only by S. aureus products but also by intrinsic metabolic programs. This study explores whether G-MDSC activity can be modulated by increasing mitochondrial abundance using a co-culture paradigm with macrophages as a mitochondrial donor. Macrophages transfer mitochondria directly to G-MDSCs via tunneling nanotubes, enhancing G-MDSC respiration, as reflected by increased basal, maximal, and spare respiratory capacity. Augmenting mitochondrial abundance in G-MDSCs enhances T cell-suppressive activity and reduces tumor necrosis factor (TNF) and interleukin 6 (IL-6) production. In a mouse model of S. aureus prosthetic joint infection, adoptively transferred macrophages deliver mitochondria to G-MDSCs, enhancing their suppressive activity and increasing bacterial burden, which is reversed when macrophages with nonfunctional mitochondria are introduced. These findings support the theory that G-MDSCs exploit mitochondria to augment their anti-inflammatory properties in response to S. aureus biofilm.

Graphical Abstract

In brief

Arumugam et al. show that macrophages transfer mitochondria to G-MDSCs via tunneling nanotubes, enhancing their anti-inflammatory activity. Acquisition of functional mitochondria promotes transcriptional and metabolic reprogramming of G-MDSCs and biofilm infection in vivo, whereas transfer of non-functional mitochondria does not, revealing the importance of active mitochondria in altering G-MDSC activity.

INTRODUCTION

Granulocytic myeloid-derived suppressor cells (G-MDSCs) are pathologically activated neutrophils (PMNs) that promote anti-inflammatory responses in several diseases, including cancer and bacterial infections.^1,2 Prior work has demonstrated that G-MDSCs accumulate during Staphylococcus aureus prosthetic joint infection (PJI) in both mouse models and humans.^3-6 PJI is a devastating complication of hip or knee arthroplasty, where S. aureus is a common cause of biofilm formation.⁷ The recalcitrance of biofilm to antibiotics and immune-mediated clearance complicates treatment.^8,9 Therefore, clinical management of PJI often requires a two-stage revision and long-term antibiotics, which are associated with increased healthcare costs and, in some cases, permanent disability.¹⁰ Therefore, alternative treatment strategies are needed to reduce the burden of multiple surgeries and extended antibiotic use that can lead to antimicrobial resistance, which is an expanding health care crisis.¹¹

S. aureus biofilms elicit an anti-inflammatory response that supports bacterial survival. For example, G-MDSCs suppress the proinflammatory activity of PMN, macrophage (MΦ), and T cell infiltrates at the site of biofilm infection.^3,4,12 Interleukin 10 (IL-10) production by G-MDSCs promotes biofilm persistence,⁶ which is dependent, in part, on lactate released from S. aureus biofilm, which induces changes in histone deacetylase activity, leading to epigenetic remodeling to augment IL-10 production.¹³ Furthermore, granulocyte depletion improved bacterial clearance in a mouse model of PJI by enhancing proinflammatory responses,⁴ highlighting the maladaptive role of G-MDSCs, which constitute approximately 60% of the leukocyte infiltrate.

Metabolism plays a critical role in regulating leukocyte activity, which can be targeted to treat various diseases.¹⁴ Single-cell transcriptomics identified an enrichment in glycolytic and hypoxia-inducible factor 1a (Hif1a)-dependent genes in G-MDSCs at the site of infection in a mouse model of S. aureus PJI and human tissues, and both pathways are critical for their immunosuppressive activity.¹⁵ Lactate production by MDSCs has also been shown to inhibit CD4⁺ T cell proliferation during chronic S. aureus infection.¹⁶ These responses share similarities with MDSC metabolic programming in cancer models that also exhibit a glycolytic signature to drive their anti-inflammatory activity.^17,18 This high reliance on glycolysis makes it an attractive target to mitigate G-MDSC activation, and although our prior work revealed that attenuating glycolysis during S. aureus biofilm infection was beneficial, additional mechanisms are at play, since only partial reductions in bacterial burden were observed.¹⁵ Another intriguing approach is to augment G-MDSC oxidative capacity. Recent studies have reported that mitochondria can be directly transferred between cells via tunneling nanotubes (TNTs) to increase oxidative metabolism.^19-21 However, little is known about the role of mitochondria in regulating G-MDSC activity, and we surmised that increasing mitochondrial abundance may alter their metabolic set point and effector function.

This study revealed that MΦs transfer mitochondria to G-MDSCs to increase their respiratory capacity and activity. Direct contact was required for mitochondrial transfer, which was partially dependent on TNTs. Mitochondrial transfer increased G-MDSC oxidative metabolism but unexpectedly enhanced suppressive activity. Evidence of mitochondrial transfer from MΦs to G-MDSCs was also observed in vivo during S. aureus PJI using Cx3cr1^Cre+/− MitoTag^fl/fl mice and following adoptive transfer of MitoTracker Green-labeled MΦs. MΦ adoptive transfer enhanced G-MDSC anti-inflammatory properties, leading to increased bacterial burden during S. aureus PJI, which was attributed to mitochondrial activity, since phenotypes were ablated when MΦs with non-functional mitochondria were introduced. Collectively, these results demonstrate that G-MDSCs can acquire mitochondria from MΦs via TNTs to enhance their anti-inflammatory attributes.

RESULTS

G-MDSCs are typified by limited mitochondrial activity

Although G-MDSCs are considered a pathological activation state of PMNs,²² little information is available regarding their mitochondrial content and function. We first assessed these parameters in G-MDSCs and PMNs vs. MΦs, which have abundant mitochondria.²³ G-MDSCs displayed ~10-fold less mtDNA than MΦs but similar levels as PMNs (Figure 1A). Assessment of mitochondrial activity by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction (Figure 1B), oxygen consumption rate (OCR) (Figure 1C), and SCENITH (single-cell energetic metabolism by profiling translation inhibition)²⁴ (Figure 1D) revealed that G-MDSCs exhibit relatively low mitochondrial activity with high glucose dependence and glycolytic capacity. In comparison, PMNs displayed lower mitochondrial activity than G-MDSCs, as measured by both MTT and OCR (Figures 1B and 1C). In contrast, MΦs utilized both glycolysis and mitochondrial respiration, as previously described.^25,26 To characterize the metabolic pathways required for ATP production, G-MDSCs were treated with 2-deoxyglucose (2-DG), oligomycin, or etomoxir to inhibit glycolysis, oxidative phosphorylation (OxPhos), and fatty acid oxidation. Only 2-DG reduced ATP levels (Figure 1E), revealing G-MDSC reliance on glycolysis for energy production. Cell viability remained unaffected with 2-DG treatment (Figure S1A), likely due to the short interval examined (i.e., 6 h) and because ATP production was not completely inhibited. Exposure of G-MDSCs to S. aureus biofilm increased glucose uptake, as previously described¹⁵; however, ATP generation was reduced (Figures S1B and S1C). This suggests an uncoupling of ATP production that may be an early indicator of impending cell death, as S. aureus biofilm produces numerous cytotoxins.²⁷ To characterize the metabolic profile of G-MDSCs during biofilm infection in vivo, leukocyte infiltrates in a mouse model of S. aureus PJI were analyzed using SCENITH. G-MDSCs displayed a robust glycolytic signature and minimal mitochondrial respiration, in agreement with in vitro assessments, whereas monocyte and MΦ infiltrates utilized both pathways (Figure 1F). Collectively, these results demonstrate that G-MDSCs have fewer mitochondria and depend on glycolysis for their energy demand.

Figure 1. G-MDSCs preferentially utilize glycolysis with minimal mitochondrial activity.

(A and B) Relative mitochondrial DNA content (A) and estimation of mitochondrial reductase activity using an MTT assay (B) in granulocytic myeloid-derived suppressor cells (G-MDSCs), neutrophils (PMNs), and macrophages (MΦs) (n = 3–7 per group; 2.5 × 10⁵ cells/well).

(C) Oxygen consumption rate (OCR) of G-MDSCs, MΦs, and PMNs, determined by Seahorse Mito stress test (n = 7–8 per group; 5 × 10⁵ cells/well) and

(D) metabolic profiles identified by SCENITH (n = 6 per group; 2.5 × 10⁵ cells/well).

(E) ATP levels in untreated control (Co) G-MDSCs and following treatment with 2-deoxy-glucose (2-DG), oligomycin (O), or etomoxir (Et) for various intervals (n = 9 per group; 10⁵ cells/well).

(F) Metabolic profile of leukocytes recovered from the soft tissue surrounding the infected joint on day 7 following S. aureus PJI was determined by SCENITH (n = 10 mice per group).

Data represent mean ± SEM of 2–3 independent experiments. Significance was calculated by unpaired, two-tailed t test (A and B) or two-way ANOVA with Tukey’s correction (E). ***p < 0.001; ****p < 0.0001; ns, not significant.

MΦs transfer mitochondria to G-MDSCs via TNTs

Since G-MDSCs exhibit a glycolytic bias that is linked to their anti-inflammatory activity during S. aureus PJI,¹⁵ we explored whether increasing mitochondrial abundance would shift G-MDSCs to a more oxidative state and alter their activity. MΦs were explored as a potential mitochondrial donor based on their high mitochondrial abundance and presence in infected tissues during PJI in both a mouse model and humans.^3,12 In addition, MΦs have been shown to transfer mitochondria to neurons to modify their metabolism, function, and survival.²⁸ To assess mitochondrial transfer, unlabeled G-MDSCs were co-cultured with MitoTracker Green-labeled MΦs, where mitochondrial exchange was quantified by flow cytometry. More than 90% of G-MDSCs were MitoTracker Green⁺ following MΦ co-culture, which increased with a higher MΦ ratio (Figures 2A and S2A). Mitochondrial transfer was also demonstrated by heightened mtDNA content in G-MDSCs following MΦ co-culture (Figure 2B). As an independent method to validate mitochondrial transfer, MΦs from C57BL/6J mice were co-cultured with BALB/c mouse-derived G-MDSCs. A single nucleotide polymorphism in the BALB/c mt-Co3 gene disturbs an AspI restriction site that is conserved in C57BL/6J mtDNA, allowing endogenous from transferred mtDNA to be distinguished.²⁰ While BALB/c G-MDSCs produced a single 385 bp band, G-MDSCs co-cultured with MΦs exhibited a mixture of BALB/c and C57BL/6J mouse mtDNA, indicating the presence of both endogenous and transferred mitochondria (Figure 2C). To confirm mitochondrial transfer independent of dye labeling approaches, MΦs from Cx3cr1^Cre+/− MitoTag^fl/fl mice were used, where nearly 45% of MΦs were GFP⁺ (Figure S2B) and were fluorescence-activated cell sorting (FACS) purified to enrich for GFP⁺ MΦs for mitochondrial transfer experiments (Figure S2C). When co-cultured with Cx3cr1^Cre+/− MitoTag^fl/fl MΦs, approximately 2.5% of G-MDSCs were GFP⁺ (Figures S2D and S2E). Signals following Cx3cr1^Cre+/− MitoTag^fl/fl transfer were less robust compared with MitoTracker Green labeling, which likely resulted from heterozygous Cx3cr1^Cre expression and/or the potential decay of GFP signal that can occur with oxidative activity.²⁹ G-MDSCs also internalized exogenous mitochondria in a concentration-dependent manner (Figure S2F and S2G). However, higher mitochondrial levels (50 μg) significantly decreased G-MDSC viability (Figure S2H), likely via enhanced reactive oxygen species (ROS) production.³⁰ Based on this and the fact that lower concentrations of exogenous mitochondria did not dramatically alter G-MDSC respiratory capacity (Figure S2I) only cell-cell mitochondrial transfer was pursued further. Prior reports have described that mitochondrial transfer between cells can occur by several mechanisms, including extracellular vesicles, mitochondrial extrusion, gap junctions, and TNTs.³¹ Mitochondrial transfer from MΦs to G-MDSCs was completely inhibited when cells were separated by Transwells, reflecting the requirement for direct cell-cell contact (Figure 2D). Transfer of mitochondria by secreted vesicles or extrusion was excluded, since spent medium from MitoTracker Green-labeled MΦs did not translate to signals in G-MDSCs (Figure 2E). TNTs have recently been described to facilitate mitochondrial transfer^31,32; therefore, the formation of F-actin nanotubes was examined by confocal microscopy. TNT formation was observed between MΦs and G-MDSCs, where G-MDSCs contained MitoTracker Deep Red⁺ mitochondria originating from MΦs (Figure 2F), and mitochondria could be visualized in nanotube structures (Figure S2J inset). Treatment of MΦs with the microtubule inhibitor cytochalasin D significantly decreased mitochondrial transfer (Figure 2G), which, when combined with the Transwell data, suggest that TNTs partially mediate this process. Gap junctions can also facilitate mitochondrial transfer³³; however, the gap junction inhibitor Gap26 had no effect (Figure S2K). We next investigated whether mitochondrial transfer can occur between MΦs or G-MDSCs themselves, using cells from B6.SJL (CD45.1⁺) and C57BL/6J (CD45.2⁺) mice to allow for discrimination between donor and recipient cells. Mitochondrial transfer was observed between MΦs (Figure S2L) as well as G-MDSCs (Figure S2M), suggesting a dynamic process of organelle trafficking. Taken together, these studies demonstrate that MΦs donate mitochondria to G-MDSCs, which is partially facilitated by TNTs.

Figure 2. Mitochondrial transfer from MΦs to G-MDSCs requires physical contact.

(A) Transfer of MitoTracker Green (MTG)-labeled mitochondria from MΦs to G-MDSCs co-cultured at a 1:1 or 5:1 ratio (n = 8–15 per group).

(B) Relative mtDNA content of G-MDSCs with or without MΦ co-culture (n = 4 per group).

(C) mtDNA from BALB/c G-MDSCs with or without C57BL/6J MΦ co-culture was PCR amplified across the mt-Co3 gene and subjected to restriction enzyme digestion.

(D and E) Percentage of MTG⁺ G-MDSCs when (D) separated from MΦs during co-culture with Transwells (n = 7–15 per group) or (E) exposed to spent medium from MTG-labeled MΦs (n = 9 per group).

(F) MΦs and G-MDSCs were stained with MitoTracker Deep Red and CellTracker Violet, respectively. After a 16 h co-culture period, cells were stained with phalloidin-rhodamine to identify tunneling nanotubes (TNTs; arrow). G-MDSCs and MΦs are indicated by asterisk and hatch signs, respectively, and the inset depicts a magnified image of a TNT. Images are representative of 2 independent experiments (scale bar: 10 μm).

(G) MΦs were treated with cytochalasin D for 1 h prior to G-MDSC addition and throughout the co-culture period (n = 6 per group).

Data represent mean ± SEM of 2–4 independent experiments. Significance was calculated by one-way ANOVA with Tukey’s correction (A, D, E, and G) or paired, two-tailed t test (B). *p < 0.05, **p < 0.01, ****p < 0.0001.

Mitochondrial transfer to G-MDSCs enhances oxidative metabolism

Given that mitochondria can be transferred from MΦs to G-MDSCs, we next investigated whether this altered oxidative respiration. Non-adherent G-MDSCs were collected following the co-culture period for Seahorse assays, which contained approximately 50,000 MΦs (<10% of the population) (Figure S3A). Importantly, this number of MΦs produced minimal OCR values (Figure S3B); therefore, the presence of residual MΦs did not confound G-MDSC OCR measurements following MΦ mitochondrial transfer. Basal OCR increased more than 10-fold in G-MDSCs following mitochondrial transfer, with a concomitant 20-fold elevation in both maximal respiration and spare respiratory capacity (Figure 3A). However, no changes in non-mitochondrial OCR were observed, suggesting that heightened OCR results from increased functional mitochondria in G-MDSCs. To validate that respiring mitochondria were responsible for increasing G-MDSC oxidative activity, MΦs were treated with rotenone and antimycin A (RA) to inhibit mitochondrial respiration (Figure S3C). The heightened respiratory capacity of G-MDSCs after MΦ co-culture was abolished when MΦs were pre-treated with RA (Figure 3B), which had no effect on mitochondrial transfer (Figure S3D), revealing the requirement for active mitochondria. MΦ OCR was not affected following mitochondrial transfer to G-MDSCs (Figure S3E), likely due to their abundance of mitochondria, which can maintain respiratory capacity. This was also supported by minimal changes in MitoTracker Green signal in MΦs after mitochondrial transfer to G-MDSCs (Figures 2A-2D).

Figure 3. Mitochondrial transfer enhances G-MDSC oxidative activity.

(A and B) MΦs were (A) left untreated (n = 12 per group) or (B) pre-treated with rotenone and antimycin A (RA) (n = 7 per group) for 2 h prior to G-MDSC co-culture, whereupon OCR and mitochondrial parameters of G-MDSCs were quantified.

(C) OCR and associated parameters for G-MDSCs with or without MΦ mitochondrial transfer following a 2 h exposure to S. aureus biofilm (n = 12 per group).

(D–F) G-MDSCs with or without MΦ mitochondrial transfer were exposed to S. aureus biofilm for 2 h, whereupon (D) mitochondrial ROS (MitoSOX; n = 11 per group), (E) mitochondrial H₂O₂ (MitoPY1; n = 11 per group), and (F) cell viability (n = 33 per group) were quantified.

Results represent mean ± SEM of 2–3 independent experiments. Significance was calculated by two-way ANOVA with Bonferroni’s correction (A) or two-way ANOVA with Tukey’s correction (B–F). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Unstim., unstimulated; ns, not significant.

Our prior work established that S. aureus biofilm augments G-MDSC glycolysis to program their anti-inflammatory activity.¹⁵ Therefore, we determined whether the heightened oxidative metabolism observed following mitochondrial transfer affected G-MDSC metabolic programming upon biofilm exposure. Although basal respiration, maximal respiration, and ATP production were significantly decreased in mitochondrial-transferred G-MDSCs exposed to S. aureus biofilm vs. resting cells (Figure 3C), all three were significantly elevated in G-MDSCs receiving MΦ mitochondria compared to normal G-MDSCs in response to biofilm, reflecting increased oxidative metabolism. In addition, both mtROS and mtH₂O₂ levels were significantly higher in G-MDSCs receiving MΦ mitochondria following biofilm exposure (Figures 3D and 3E). These changes were not observed with a shorter biofilm co-culture period (Figures S3F and S3G), suggesting cumulative action of biofilm-derived signals for triggering maximal G-MDSC oxidative metabolism following mitochondrial transfer. Interestingly, despite the increased oxidative footprint in G-MDSCs receiving MΦ mitochondria, glycolysis and glycolytic capacity were also significantly elevated at baseline (Figure S3J), likely to support the substrate demand of increased mitochondrial abundance following transfer, and these glycolytic metrics were also affected by biofilm (Figure S3J), similar to OCR findings (Figure 3C). Glucose uptake was increased to an equivalent extent in G-MDSCs receiving mitochondrial transfer and normal G-MDSCs at both time points (Figure S3H), and cell viability decreased following biofilm co-culture in a time-dependent manner (Figures 3F and S3I). Non-glycolytic acidification was increased in G-MDSCs after mitochondrial transfer, suggesting a role for a glycolytic-independent factor mediating extracellular acidification (Figure S3J). Collectively, these results demonstrate that increased oxidative metabolism in G-MDSCs following mitochondrial transfer is dependent on active mitochondria from donor MΦs, which persists following S. aureus biofilm exposure.

Mitochondrial transfer to G-MDSCs leads to transcriptional reprogramming

To examine how increased mitochondrial abundance in G-MDSCs programs cellular activation status, bulk RNA sequencing (RNA-seq) was performed on G-MDSCs receiving MΦ mitochondria vs. normal G-MDSCs following S. aureus biofilm exposure. Principal-component analysis (PCA) revealed distinct transcriptional profiles between the groups (Figure S4A). A total of 1,199 genes were significantly upregulated, and 409 genes were decreased, in unstimulated G-MDSCs following MΦ mitochondrial transfer compared to normal G-MDSCs (Figure 4A), reflecting a major impact of mitochondrial acquisition on cellular programming under resting conditions. Ingenuity Pathway Analysis (IPA) of differentially expressed genes revealed that interferon and IL-10 signaling, tumor microenvironment, and Hif1a pathways were significantly upregulated in mitochondrial-transferred G-MDSCs (Figure 4B), which have previously been implicated in G-MDSC function.⁶ In response to S. aureus biofilm, 1,116 and 347 genes were significantly increased and downregulated, respectively, between G-MDSCs receiving MΦ mitochondrial transfer vs. normal G-MDSCs (Figure 4C). IPA revealed significant differences in cytokine and interferon signaling, the tumor microenvironment, Hif1a signaling, and GTPase signaling pathways following biofilm exposure (Figure 4D) that were also observed in the unstimulated comparison (Figure 4B), indicating that G-MDSCs receiving mitochondria maintain their transcriptional footprint after S. aureus biofilm stimulation. This was reflected by similar expression patterns of genes involved in tumor microenvironment (Figure 4E), IL-10 signaling (Figure 4F), and cancer (Figure 4G). Metabolic Flux Balance Analysis (METAFlux)³⁴ was used to calculate oxygen consumption, glucose uptake, and H⁺ release flux scores using transcriptomics data, which revealed an increase in all three parameters in unstimulated G-MDSCs receiving MΦ mitochondria (Figures S4B-S4D), suggesting that mitochondrial transfer also alters metabolic gene expression in G-MDSCs. Collectively, these results establish that mitochondrial transfer to G-MDSCs reprograms their transcriptional landscape, augmenting the expression of genes previously implicated in regulating G-MDSC anti-inflammatory properties during S. aureus biofilm formation.^6,35

Figure 4. Mitochondrial transfer to G-MDSCs induces transcriptional reprogramming.

G-MDSCs with or without MΦ mitochondrial transfer were unstimulated or exposed to S. aureus biofilm for 2 h in Transwells, whereupon RNA was isolated for bulk RNA-seq (n = 3 biological replicates per group).

(A) Volcano plot of significantly upregulated (red) and downregulated (blue) genes in mitochondrial-transferred vs. control G-MDSCs at baseline (i.e., no stimulus), with (B) significantly expressed genes grouped into pathways by IPA.

(C) Volcano plot of significantly upregulated (red) and downregulated (blue) genes in mitochondrial-transferred G-MDSCs vs. control G-MDSCs following a 2 h coculture period with S. aureus biofilm that were (D) grouped into pathways by IPA.

(E–G) Heatmaps representing the Z scores of normalized gene counts for (E) tumor microenvironment, (F) IL-10 signaling, and (G) molecular mechanism of cancer pathways.

G-MDSC anti-inflammatory activity is enhanced following mitochondrial transfer

Since our transcriptomics data suggested that mitochondrial transfer to G-MDSCs enhanced anti-inflammatory gene expression, we next determined whether this translated to altered function. Indeed, S. aureus intracellular survival was significantly increased in G-MDSCs receiving MΦ mitochondria (Figure S4E), along with enhanced T cell-suppressive activity, compared to normal G-MDSCs (Figure 5A). Several factors have been implicated in mediating G-MDSC anti-inflammatory activity, including ROS, nitric oxide (NO), arginase-1 (Arg-1), and prostaglandin E2 (PGE2).² Treatment of mitochondrial-transferred G-MDSCs with inhibitors of each target revealed that NO plays a significant role in the suppressive function of mitochondrial-transferred G-MDSCs (Figure 5B). Given that G-MDSC action is linked to inflammatory mediator production,⁶ cytokine levels were examined in G-MDSCs with or without mitochondrial transfer after exposure to either heat-killed S. aureus (HKSA) or live biofilm. Mitochondrial-transferred G-MDSCs produced less tumor necrosis factor (TNF) and IL-6 in response to HKSA and biofilm (Figures 5C and 5D, respectively), concomitant with significantly more IL-10 following S. aureus biofilm exposure (Figure 5D). Surprisingly, CCL2 levels were significantly increased in G-MDSCs after mitochondrial transfer under resting conditions, which was augmented in response to S. aureus biofilm (Figure 5E). This was also seen at the transcriptional level, where normalized read counts revealed increased Ccl2 expression in G-MDSCs following mitochondrial transfer under both resting and biofilm-stimulated conditions compared to normal G-MDSCs (Figure S5A). Given that mitochondrial components can serve as danger-associated molecular patterns (DAMPs),^36,37 and type I IFN and cyclic GMP-AMP synthase-stimulator of interferon gene (cGAS-STING) signaling pathways were identified by RNA-seq, we next investigated whether MΦ-derived mitochondria stimulated CCL2 production in G-MDSCs via TLR- or STING-mediated signaling. CCL2 secretion was unaffected in both Myd88^−/− and Sting^−/− G-MDSCs following mitochondrial transfer (Figures 5F and S5B), suggesting that transferred mitochondria are not acting as DAMPs. Byproducts of mitochondrial function can also regulate inflammatory mediator production,³⁸ and G-MDSCs receiving mitochondria from RA-treated MΦs produced significantly less CCL2 at baseline and following S. aureus biofilm exposure (Figure 5G), confirming that MΦ mitochondrial activity is important for promoting CCL2 levels in G-MDSCs. Of note, treatment with exogenous mitochondria did not affect CCL2 production by G-MDSCs (Figure S5C). Although CCL2 has been reported to promote G-MDSC immunosuppressive activity,³⁹ CCL2 treatment of G-MDSCs did not impact their ability to inhibit T cell activation regardless of mitochondrial transfer status (Figure S5D).

Figure 5. Mitochondrial acquisition promotes G-MDSC suppressive activity.

(A and B) G-MDSCs with or without MΦ mitochondrial transfer were co-cultured with CD4⁺ T cells at a 1:1 ratio for 72 h (A) (n = 14–16 per group) or (B) pre-treated with the ROS inhibitor N-acetyl cysteine (ROS), the iNOS inhibitor L-NIL (NO), the arginase-1 inhibitor Nor-NOHA (Arg), or the cyclooxygenase-2 inhibitor CAY10452 (PGE) for 30 min prior and throughout the 48 h T cell co-culture period, whereupon T cell proliferation was analyzed by flow cytometry (n = 6 per group).

(C and D) TNF, IL-6, and IL-10 production was assessed by intracellular staining in G-MDSCs with or without MΦ mitochondrial transfer following exposure to (C) heat-killed S. aureus (HKSA) for 5 h (n = 3–6 per group) or (D) S. aureus biofilm for 4 h (n = 6 per group).

(E and F) CCL2 production by (E) wild-type (WT) G-MDSCs (n = 6 per group) and (F) Myd88^−/− G-MDSCs (n = 6–9 per group) with or without MΦ mitochondrial transfer or (G) MΦs treated with rotenone and antimycin A (RA) (n = 12 per group).

Data represent mean ± SEM of 2–4 independent experiments. Significance was calculated by one-way ANOVA with Tukey’s correction. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. ns, not significant.

Since leukocyte inflammatory properties are intimately linked with metabolism,^40,41 untargeted metabolomics was performed to assess the metabolic landscape of G-MDSCs following mitochondrial transfer and how this is impacted by S. aureus biofilm. Approximately 65 metabolites were identified, including amino acids, nucleotides, glycolytic and TCA cycle intermediates, and fatty acid metabolites (Figure S5E). Several nucleotides (thymidine and 2-deoxycytidine) and mitochondrial metabolites (phosphocreatine, propionyl carnitine, and itaconate) were significantly increased in resting G-MDSCs following mitochondrial transfer and potentiated by biofilm exposure (Figures S5F-S5H). Itaconate is synthesized from the tricarboxylic acid (TCA) cycle intermediate cis-aconitate by aconitate decarboxylase 1 (Acod1), which can promote anti-inflammatory responses.^42,43 To determine whether itaconate influenced CCL2 production, mitochondrial transfer was examined in G-MDSCs from Acod1^−/− mice. CCL2 levels were equivalent in both conditions (Figure S5I), and Acod1 had no effect on bacterial burden in a mouse model of S. aureus PJI (Figure S5J). Arginine and tryptophan, which play important roles in T cell suppression,^44,45 were differentially affected in G-MDSCs following mitochondrial transfer (Figure S5K). Although no changes in arginine were detected, tryptophan was significantly elevated in G-MDSCs after mitochondrial transfer, in agreement with increased T cell inhibition (Figure 5A). Taken together, these findings indicate that mitochondrial acquisition by G-MDSCs augments T cell-suppressive activity and CCL2 production, the latter of which occurs in an Acod1-independent manner.

MΦ adoptive transfer supports biofilm growth by enhancing G-MDSC anti-inflammatory activity during S. aureus PJI

Given that mitochondria can be transferred from MΦs to G-MDSCs, we next investigated whether this occurs in vivo and, if so, how this shapes G-MDSC function. This was examined using a mouse model of S. aureus PJI, where G-MDSCs are detrimental.⁴ To evaluate mitochondrial transfer in vivo, MitoTracker Green-labeled CD45.1 MΦs were injected into the soft tissue surrounding the infected joint of C57BL/6J mice (CD45.2) on day 3 post infection, whereupon the MitoTracker Green signal in endogenous CD45.2⁺ cells was assessed by flow cytometry (Figure 6A). Mitochondrial transfer was evident across all leukocyte infiltrates of CD45.2 recipient mice but was most prominent in G-MDSCs and PMNs (Figure 6B). Surprisingly, only ~7,000 CD45.1⁺ MΦs were recovered from infected tissues, reflecting 0.001% of the initial inoculum (Figures 6C and S6A). This indicates that the biofilm environment is averse to MΦ survival, likely due to robust toxin production, as we previously reported.⁴⁶ G-MDSCs recovered from mice with MΦ adoptive transfer were more effective at inhibiting T cell proliferation than G-MDSCs from PBS-treated animals (Figure 6D), similar to in vitro findings (Figure 5A), whereas PMNs had no effect (Figure 6D). Bacterial burden was significantly elevated in the joint following MΦ adoptive transfer compared to PBS-treated mice, with trending increases in the soft tissue surrounding the infected joint and implant, whereas no differences were observed in the femur (Figure 6E). These changes coincided with a significant increase in the percentage of G-MDSCs and a reduction in PMN infiltrates at the site of infection (Figure 6F), which was generally reflected by absolute cell counts (Figure S6B). CCL2 levels were significantly increased in both the joint and surrounding soft tissue following MΦ adoptive transfer compared to PBS-treated animals (Figure 6G), similar to in vitro findings (Figure 5E).

Figure 6. MΦ adoptive transfer increases bacterial burden during S. aureus PJI.

(A–C, E, and F) C57BL/6J mice (CD45.2) received one injection of either vehicle (PBS) or MTG-labeled MΦs from B6.SJL animals (CD45.1) (5 × 10⁶ cells per mouse) on day 3 following S. aureus PJI, whereupon (B) MTG⁺ cells (n = 18 mice per group), (C) the abundance of adoptively transferred CD45.1⁺ MΦs (n = 30 mice per group), (E) bacterial burden (n = 30 mice per group), and (F) endogenous (CD45.2⁺) leukocyte infiltrates (n = 30 mice per group) were assessed 24 h post injection.

(D) G-MDSCs and PMNs were purified from infected tissues 24 h following MΦ adoptive transfer by FACS (day 4 post-infection), whereupon T cell suppression assays were performed (n = 3–8 per group).

(G) CCL2 levels in tissue and joint homogenates were quantified 24 h following MΦ adoptive transfer (day 4 post-infection) (n = 12 mice per group).

(H) Percentage of GFP⁺ MΦs expanded from the bone marrow of Cx3cr1^Cre+/− MitoTag^fl/fl vs. WT mice.

(I and J) S. aureus PJI was established in Cx3cr1^Cre+/− MitoTag^fl/fl and WT animals, whereupon the (I) percentage of GFP⁺ and (J) GFP median fluorescent intensity (MdFI) of leukocytes was quantified using an anti-GFP antibody (n = 10–14 mice per group).

(K) Tissues were collected on day 3 post-infection from Cx3cr1^Cre+/− MitoTag^fl/fl mice and stained with Ly6G (red) and GFP (green) to visualize the presence of MΦ-derived mitochondria in granulocytes. The inset shows a magnified image of the boxed area. Ly6G⁺ and Ly6G⁻ cells are indicated by asterisk and hatch signs, respectively (scale bar: 10 μm).

Data represent mean ± SEM of 2–5 independent experiments. Significance was calculated by unpaired, two-tailed t test (B, C, E–G, I, and J) or one-way ANOVA (D). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

To evaluate mitochondrial transfer from MΦs to G-MDSCs in the PJI milieu independent of dye labeling and adoptive transfer approaches, mice were generated where MΦ mitochondria were tagged with GFP (Cx3cr1^Cre+/− MitoTag^fl/fl) (Figure 6H). Bacterial burden and leukocyte populations were similar between Cx3cr1^Cre+/− MitoTag^fl/fl and wild-type mice, confirming no impact from GFP expression in MΦs (Figures S7A and S7B). The abundance of GFP⁺ G-MDSCs and PMNs was significantly increased in Cx3cr1^Cre+/− MitoTag^fl/fl mice (Figures 6I, 6J, and S7C), reflecting mitochondrial acquisition from endogenous MitoTag⁺ MΦs. This was corroborated by immunofluorescence staining, where several Ly6G⁺ granulocytes in Cx3cr1^Cre+/− MitoTag^fl/fl mice were GFP⁺, reflective of mitochondrial transfer (Figure 6K inset). Collectively, these results support the phenomenon of mitochondrial transfer from MΦs to G-MDSCs and other leukocytes in vivo, which increases CCL2 production, G-MDSC abundance, and suppressive activity concomitant with elevated bacterial burden. Therefore, the acquisition of mitochondria by G-MDSCs may represent a mechanism to promote biofilm infection by enhancing G-MDSC anti-inflammatory activity.

Transfer of functional mitochondria is required to augment anti-inflammatory activity during PJI

To address the importance of bioactive mitochondria for promoting anti-inflammatory activity during S. aureus PJI, MΦs were exposed to low-dose ethidium bromide (EtBr).² EtBr-treated MΦs (EtBr-MΦs) displayed minimal mtDNA and OxPhos activity and failed to augment G-MDSC OxPhos following co-culture (Figures 7A-7C). Both EtBr-MΦs and normal MΦs were detected following adoptive transfer in the mouse PJI model (Figures 7D and S8B); however, the bacterial burden was significantly decreased in animals receiving EtBr-MΦs in the soft tissue, joint, and femur (Figure 7E), concomitant with a reduction in G-MDSCs and increased PMN, monocyte, and MΦ infiltrates, compared to normal MΦs (Figures 7F and S8A), revealing a key role for functional mitochondria. Adoptive transfer of normal MΦs revealed similar trends in both bacterial burden and G-MDSC abundance, as previously shown (Figures 6E and 6F); however, the degree of biological variability across samples in these experiments (Figures 7E and 7F) precluded statistical significance. The reduction in bacterial burden following EtBr-MΦ adoptive transfer coincided with significant decreases in IL-1β, IL-6, and CXCL2 production compared to normal MΦs, whereas other cytokines were not dramatically altered (Figures 7G, 7H, S8C, and S8D). Taken together, these results demonstrate that the metabolic activity of donor mitochondria from MΦs is important for biofilm persistence during S. aureus PJI.

Figure 7. Adoptive transfer of mitochondrial DNA-depleted MΦs attenuates bacterial burden during S. aureus PJI.

(A and B) Relative mtDNA content (n = 10 per group) and (B) OCR of untreated MΦs and MΦ-EtBr (n = 7 per group).

(C) OCR of G-MDSCs co-cultured with MΦs or MΦ-EtBr (n = 8 per group).

(D–H) C57BL/6J mice (CD45.2) received one injection of B6.SJL-derived (CD45.1) normal MΦs, MΦs-EtBr, or PBS directly into the soft tissue surrounding the infected joint, whereupon (D) the percentage of CD45.1⁺ adoptively transferred MΦs, (E) bacterial burden, and (F) endogenous CD45.2⁺ leukocytes (n = 17 mice per group) were assessed 24 h post injection.

(G–H) Cytokine (G) and chemokine and growth factor (H) levels in the soft tissue surrounding the infected joint were quantified using a Milliplex MAP 25-plex assay (n = 11 mice per group)

(I–K) Vehicle (PBS), MΦs, or MΦs-EtBr derived from B6.SJL mice (CD45.1) were injected directly into the soft tissue surrounding the infected joint of C57BL/6J animals (CD45.2) on day 3 following S. aureus PJI (5 × 10⁶ cells per mouse), whereupon viable CD45⁺ cells were recovered 24 h later and subjected to scRNA-seq.

(I) UMAP of 12,489 CD45⁺ cells separated by cluster identity.

(J) G-MDSC marker genes were used to identify G-MDSC-like subsets from the granulocyte clusters. (K) Differently expressed genes were analyzed by IPA to identify upregulated (red) and downregulated (blue) pathways in G-MDSC cluster 3 from the normal MΦ group.

Data represent mean ± SEM of 2–3 independent experiments. Significance was calculated by unpaired, two-tailed t test (A), one-way ANOVA (D–H), or Wilcoxon signed-rank test (J). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. ns, not significant.

To determine how the transfer of functional mitochondria affected G-MDSC transcriptional programs in vivo, scRNA-seq was performed on CD45⁺ cells recovered from the infected tissue of mice 24 h after adoptive transfer of EtBr-MΦs or normal MΦs compared to vehicle (PBS). Cell type identity was assigned using the SingleR⁴⁷ package, with the Immunological Genome Project (ImmGen)⁴⁸ serving as the reference database. Uniform manifold approximation and projection (UMAP) analysis revealed an equal distribution of cells across the groups (Figure S8E). Cell type annotation identified PMNs as the most abundant cell type (Figure S8F), and further graph-based clustering identified 7 unique clusters (Figure 7I) that were used with cell identities to characterize PMN populations. Using our G-MDSC gene score pipeline,^49,50 since the ImmGen database lacks the ability to differentiate PMNs from G-MDSCs, PMN cluster 3 was identified as G-MDSC like, whereas clusters 1, 2, and 7 were more similar to PMNs (Figure 7J). This differed from our flow cytometry findings, where G-MDSCs were the main leukocyte population at the infection site (Figures 6F and 7F). This disconnect may reflect the fact that granulocytes represent an activation continuum, and scRNA-seq was performed relatively early post infection (day 4), so that transcriptional profiles may not directly correlate with surface marker expression. Differential gene expression analysis of G-MDSC cluster 3 identified 168 and 62 genes that were upregulated and downregulated, respectively, in mice receiving MΦ adoptive transfer compared to vehicle (Figure S8G). IPA of differentially expressed genes identified pathogen-induced cytokine storm and IL-1 family signaling as significantly upregulated following MΦ adoptive transfer (Figure 7K), both of which are linked to G-MDSC activity,^51,52 and pathogen-induced cytokine storm signaling was also significantly increased in mitochondrial-transferred G-MDSCs in vitro (Figures 4B-4D). Adoptive transfer of EtBr-MΦs did not dramatically alter G-MDSC transcriptional profiles (Figure S8H), indicating that functional mitochondria are important for these changes. Collectively, these findings support the theory that mitochondrial transfer impacts G-MDSC transcriptional programming in vivo.

DISCUSSION

Prior work has demonstrated that reducing G-MDSC abundance or inhibiting their anti-inflammatory properties is beneficial for controlling infection.^4,6,13,15 While glycolysis has recently been shown to be important for promoting G-MDSC anti-inflammatory activity,¹⁵ the potential of other approaches to reprogram cellular metabolism have not yet been investigated. One example is mitochondrial activity, which has remained relatively underexplored in G-MDSCs. This study aimed to determine whether increasing mitochondrial abundance in G-MDSCs would reprogram their functional attributes by promoting oxidative metabolism. Mitochondrial transfer has been observed between cells during both physiological and disease conditions.^31,53 Both damaged and healthy mitochondria can be transferred between cells, where stressed cells export damaged mitochondria to promote their survival,^54,55 whereas healthy mitochondria can be transferred to impaired cells to enhance effector functions during pathological conditions.^20,28,31

Our study demonstrates that G-MDSC respiratory capacity can be enhanced by mitochondrial transfer from MΦs that requires cell-cell contact and is partially facilitated by TNTs. Although an increase in oxidative metabolism was predicted to attenuate G-MDSC anti-inflammatory activity by diminishing glycolysis, which is critical for their function,¹⁵ this was not observed. Instead, mitochondrial transfer from MΦs to G-MDSCs enhanced anti-inflammatory gene expression and immunosuppressive activity both in vitro and in vivo, leading to increased bacterial burden in a mouse model of S. aureus PJI. These effects were dependent on functional mitochondria, since effects were mitigated when G-MDSCs were exposed to MΦs where mitochondrial DNA was depleted using low-dose EtBr in vitro and in vivo. Functional mitochondrial transfer also enhanced IL-1β, IL-6, CCL2, CXCL1, and CXCL2 production, which coincided with G-MDSC accumulation at the site of PJI. Importantly, the use of Cx3cr1^Cre+/− MitoTag^fl/fl mice revealed that endogenous MΦs transfer mitochondria to G-MDSCs during S. aureus PJI, corroborating our MΦ adoptive transfer studies and validating the accuracy of dye labeling approaches.

Transcriptomic analysis revealed that mitochondrial transfer to G-MDSCs induced numerous pathways, including pathogen-induced cytokine storm signaling, interferon, IL-1β, and hypercytokinemia/hyperchemokinemia in influenza. Although counterintuitive, this finding aligns with existing literature demonstrating that G-MDSC activity is induced by proinflammatory cytokines,^2,51,52 and we and others have shown that G-MDSCs express Il1b.^15,50,56,57 In addition, many of the differentially expressed pathways detected also contribute to immune suppression in the tumor microenvironment.^22,58 For example, we found that S100A8 and S100A9 were significantly enriched in G-MDSCs following mitochondrial transfer, which has been reported to promote MDSC accumulation, anti-inflammatory cytokine production, and suppressive activity.^59-61 In addition, IL-10 and HIF1a signaling were significantly increased in mitochondrial-transferred G-MDSCs, both of which have been identified as regulators of G-MDSC immunosuppressive activity.^6,15 Recently, extracellular mtDNA has been shown to enhance the immunosuppressive function of G-MDSCs through cGAS-STING.⁶² While the cGAS-STING signaling pathway was also upregulated in G-MDSCs following mitochondrial transfer, no phenotypes were observed with Sting^−/− G-MDSCs, suggesting that STING-dependent signaling does not play a significant role in the anti-inflammatory response of G-MDSCs, at least for the readouts examined in this study.

Mitochondrial metabolites such as itaconate play a key role in modulating immune responses.⁶³ In PMNs, itaconate enhances Nrf2 expression to promote antioxidant production and suppress cellular ROS levels.⁶⁴ Irg1 is induced in PMNs during S. aureus lung infection, where itaconate suppresses bacterial killing by inhibiting ROS production.⁴² Recently, β2-adrenergic receptor (β2-AR)-dependent metabolic reprogramming has been reported to increase itaconate production in MDSCs in response to doxorubicin, which activates antioxidant machinery and promotes MDSC survival.⁶⁵ Our metabolomic analysis found that itaconate is increased in mitochondrial-transferred compared to normal G-MDSCs. However, the functional implication of this is unclear, since CCL2 production was not affected in Acod1^−/− G-MDSCs, and the bacterial burden was similar in Acod1^−/− and wild-type (WT) mice during S. aureus PJI. Therefore, although itaconate plays a key role in S. aureus lung infection,^42,43 the metabolite appears to be less critical for dictating infection outcome in the S. aureus PJI model studied here. This aligns with our recent report, which demonstrated that the transcriptional and metabolic profiles elicited by biofilm vary based on the tissue milieu.⁶⁶

CCL2 was dramatically induced in G-MDSCs following MΦ mitochondrial transfer, and although classically recognized for its role in monocyte chemotaxis,⁶⁷ CCL2 has also been reported to recruit G-MDSCs into tumors.^68,69 A prior report described a role for CCL2 in T cell suppression by G-MDSCs via STAT3 signaling,³⁹ and CCL2 production following Siglec receptor engagement by sialoglycans promoted MDSC suppressive activity.⁷⁰ We also identified a correlation between CCL2 production and G-MDSC abundance in response to S. aureus biofilm in vivo, which was specific to scenarios where mitochondrial transfer occurred. However, CCL2 blockade did not abrogate G-MDSC suppressive activity; therefore, further investigation is needed to understand the link between mitochondrial function and CCL2 production during S. aureus biofilm infection and functional implications. Of note, a prior report showed that leukocyte recruitment was not affected in CCR2-deficient mice during S. aureus PJI, although G-MDSCs were not examined in this study, and increased bacterial burden was only observed during chronic infection (i.e., day 28).⁷¹ Therefore, it is possible that the acute infection interval examined here was not sufficient to detect a functional role of CCL2 in G-MDSC activity.

Conclusion

These findings reveal that increasing mitochondrial abundance in G-MDSCs via TNT-mediated transfer from MΦs enhances G-MDSC anti-inflammatory gene expression and suppressive activity. Our study also provides evidence that MΦs can transfer mitochondria to G-MDSCs in vivo, which increased G-MDSC abundance and biofilm growth during S. aureus PJI. Further investigation will be required to identify the factors responsible for promoting G-MDSC anti-inflammatory activity following mitochondrial transfer, and our findings highlight the complexity of G-MDSC metabolism in programming effector function.

Limitations of the study

This study has several limitations. First, although we determined that MΦ OxPhos was not altered following mitochondrial transfer to G-MDSCs, additional changes in MΦ metabolism or functional activity may occur that were not examined here. Second, mitochondrial transfer from MΦs to G-MDSCs was only partially mediated by TNTs, supporting the involvement of an alternative route(s) for mitochondrial transfer, the identity of which remains unknown. Cytochalasin D has been widely used to implicate TNT involvement in mitochondrial transfer between multiple cell types^72-74; however, other cellular functions, such as phagocytosis, are also affected, which could influence findings. Although mitochondrial transfer from MΦs to G-MDSCs was observed in vivo, MΦ-derived mitochondria were also detected in other leukocyte infiltrates, and the functional implications of this for cellular function, if any, remains to be determined. Another interesting finding was that enhanced T cell-suppressive activity in G-MDSCs following mitochondrial transfer was NO dependent. Although NO-mediated pathways are typically associated with M-MDSCs,⁷⁵ NO production has also been ascribed to G-MDSCs.⁷⁶ Our earlier report revealed that G-MDSCs can acquire MΦ markers during chronic S. aureus PJI⁷⁷; therefore, future studies are warranted to explore a possible transition of G-MDSCs into a monocyte-like subset.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

Animal studies were conducted according to the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and comply with the Animal Research: Reporting of In Vivo Experiments guidelines. The animal use protocol (18-013-03) was approved by the University of Nebraska Medical Center Institutional Animal Care and Use Committee. All mice were group-housed at 21°C–23°C and 55% average humidity under a 12 h light/dark cycle with free access to food and water. Both male and female animals were used for these studies between 8 and 12 weeks of age, and no sex-dependent effects were observed for the phenotypes reported. Experiments were conducted with C57BL/6J (RRID:IMSR_JAX:000664), BALB/c (RRI-D:IMSR_JAX:000651), B6.SJL (RRID:IMSR_JAX:033076), Myd88^−/− (RRID:IMSR_JAX:009088), Sting^−/− (RRID:IMSR_JAX:025805), Acod1^−/− (RRID:IMSR_JAX:029340), and Cx3cr1^Cre+/− MitoTag^fl/fl mice created by crossing MitoTag^fl/fl (RRID:IMSR_JAX:032290) with Cx3cr1^Cre (RRID: IMSR_JAX:025524) strains. Sample sizes for individual in vivo studies ranged from 5 to 6 mice/group, with most findings replicated in 2-3 independent experiments. Details regarding the number of animals used in each experiment are provided in the figure legends.

Cell culture and preparation

Bone marrow-derived MΦs and G-MDSCs were generated as previously described.^5,78 Bone marrow was passed through a 70 μm filter and RBCs lysed using sterile H₂O followed by rapid correction to isotonic conditions with 10x PBS. Macrophages were expanded from bone marrow in RPMI-1640 containing 10% FBS, 2 mM L-glutamine, 10 mM HEPES, 1x antimycotic/antibiotic, 50 μM β-mercaptoethanol, and 10% conditioned medium from L929 fibroblasts as a source of M-CSF. Culture medium was replaced on days 3 and 5 with MΦs harvested on day 7 for experiments. Mitochondrial DNA-depleted MΦs (EtBr-MΦs) were generated by culturing bone marrow in the presence of 200 ng/mL EtBr during the 7-day culture period. G-MDSCs were obtained by culturing bone marrow in RPMI-1640 with 10% FBS, 2 mM L-glutamine, 10 mM HEPES, 1x antimycotic/antibiotic, and 50 μM β-mercaptoethanol with G-CSF and GM-CSF (40 ng/mL each) for 3 days, whereupon IL-6 (40 ng/mL) was added on day 3. G-MDSCs were isolated on day 4 of culture using anti-mouse Ly6G microbeads (Miltenyi Biotec). PMNs were directly recovered from the bone marrow with anti-Ly6G microbeads.

METHOD DETAILS

Bacterial strain

The USA300 clinical isolate S. aureus LAC-13C was used in this study.⁷⁹ Bacteria were streaked onto Tryptone Soy Agar plates with 5% sheep blood prior to each experiment. For biofilm studies, a single colony was grown in biofilm medium (RPMI-1640 with 10% FBS, 2 mM L-glutamine, and 10 mM HEPES) overnight at 37°C with constant shaking at 250 rpm. The overnight culture was diluted to an OD600 of 0.05 in biofilm medium and added to 96- or 24-well plates that had been pre-coated with 20% human plasma in 10x PBS overnight at 4°C. Bacteria were incubated at 37°C under static conditions for 4 days during biofilm formation with approximately 50% of spent medium replaced daily. Mature biofilms were used on day 4 of growth for experiments.

Mouse model of S. aureus PJI

A mouse model of PJI was used as previously described.^13,15,66 S. aureus was grown as defined above, whereupon the overnight culture was washed twice and diluted to 5 x 10⁵ CFU/mL in 1x PBS. Mice were anesthetized with ketamine and xylazine, whereupon a parapatellar arthrotomy was performed on the right leg to expose the distal femur. A hole was created in the femoral intercondylar notch using a 26-gauge needle to insert a 0.6 mm orthopedic-grade K-wire. A total of 1,000 CFU S. aureus in a 2 μL volume was added to the tip of the K-wire before the inner and outer skin incisions were closed using 6-0 polyglycolic acid and nylon sutures, respectively. Buprenorphine extended release was administered immediately following surgery to provide 72 h of pain relief. Animals received supplemental heat during the post-surgical period and were closely observed until ambulatory. Mice were monitored daily until sacrifice and no infection-associated mortality occurred with any of the mouse strains examined.

For adoptive transfer experiments, MΦs were generated from the bone marrow of CD45.1 congenic mice as previously described.⁷⁸ The use of MΦs from CD45.1 animals allows identification from endogenous MΦs in C57BL/6J recipients (CD45.2) based on differential CD45 isoform expression that can be discriminated using antibodies specific for each allele. Briefly, untreated and CD45.1 EtBr-MΦs were labeled with MitoTracker Green (200 nM), whereupon 5 x 10⁶ cells per mouse were injected directly into the soft tissue surrounding the joint at day 3 post-infection in 25 μL of 1x PBS. Mice were euthanized 24 h later to quantify bacterial burdens and leukocyte populations.

Antibodies, flow cytometry, and FACS

For quantifying leukocyte populations in PJI tissues, cells were stained with CD45.1-BV711(RRID: AB_2562605), CD45.2-APC (RRID: AB_389210), Ly6G-PE (RRID:AB_1186099), F4/80-PE-Cy7 (RRID:AB_893478), CD11b-AF700 (RRID: AB_493705), MHC class II-BV605 (RRID: AB_2565894), and Ly6C-APC-Cy7 (RRID: AB_1727555), with direct (anti-GFP-APC; RRID:AB_2650669) or indirect (anti-GFP biotin; RRID:AB_305631 with Alexa Fluor 488 Streptavidin; RRID:AB_2337249) staining for GFP to quantify mitochondrial transfer from Cx3cr1^Cre+/− MitoTag^fl/fl mice. Cells were also evaluated for mitochondrial content (MitoTracker Green; Cell Signaling Technology), mtROS (MitoSOX; Invitrogen), mtH₂O₂ (MitoPY1; Tocris), and glucose uptake (2-NBDG; Cayman Chemical). Dead cells were identified using Zombie UV or Zombie NIR dyes (Biolegend) and TruStain FcX (RRID: AB_1574973) was used to block non-specific Ab binding. Samples were analyzed using BD LSR II or BD Fortessa X50 cytometers and UltraComp eBeads (Invitrogen) were used for compensation in all experiments. Data was analyzed using FlowJo v.10.10.0 (RRID:SCR_008520) using the gating strategy in Figure S9.

Examination of mitochondrial transfer

To assess the extent of mitochondrial transfer from MΦs to G-MDSCs, two independent approaches were utilized. For the first, MΦs were labeled with 200 nM MitoTracker Green (Cell Signaling) for 15 min at 37°C and cells were washed with complete RPMI medium to remove excess dye. MitoTracker Green-labeled MΦs were co-cultured with unlabeled G-MDSCs in 24-well plates in G-MDSC medium without IL-6 for 16–18 h, whereupon MitoTracker Green signal in G-MDSCs was assessed by flow cytometry. In some experiments, cells were physically separated using Transwells (0.4 μm) or MΦs were pre-treated with the connexin 43 gap junction inhibitor Gap26 or scrambled control peptide (sGap26) (both at 10 and 100 μM), or the microtubule inhibitor cytochalasin D (0.1 μM) for 1 h prior to G-MDSC co-culture with compounds remaining during the 16–18 h incubation period. When inhibiting MΦ mitochondrial activity prior to G-MDSC co-culture, MΦs were pre-treated with rotenone and antimycin A (0.1 μM each) for 2 h. MΦs or G-MDSCs were also labeled with MitoTracker Green using the same parameters described above for assessing the extent of mitochondrial transfer between the same cell type by leveraging cells isolated from C57BL/6J (RRID:IMSR_JAX:000664) and B6.SJL (RRID:IMSR_JAX:033076) mice.

As a second approach to evaluate mitochondrial transfer independent of dye labeling, G-MDSCs were co-cultured with MΦs from Cx3cr1^Cre+/− MitoTag^fl/fl mice, whereupon cells were washed and fixed in Cyto-Fast Fix/Perm Buffer for 20 min at room temperature followed by washing 2X with Intracellular Staining Perm Wash Buffer (both from BioLegend). GFP signals were quantified by flow cytometry using direct (anti-GFP-APC; RRID:AB_2650669) or indirect (anti-GFP biotin; RRID:AB_305631 with Alexa Fluor 488 Streptavidin; RRID:AB_2337249) staining approaches.

Mitochondrial isolation and ATP quantification

Mitochondria were purified from MΦs using a Mitochondria Isolation Kit (Abcam) following the manufacturer’s protocol where concentrations were estimated using a Pierce bicinchoninic acid (BCA) Protein Assay Kit. Purified mitochondria were stained with MitoTracker Green as described above before adding to G-MDSCs for 16–18 h, whereupon mitochondrial uptake was assessed by flow cytometry. ATP production was quantified using an ATPlite Luminescence Assay System (Revvity Health Sciences) according to the manufacturer’s instructions.

Confocal microscopy

To assess TNT formation, G-MDSCs and MΦs were stained with CellTracker Violet (Invitrogen) and MitoTracker Deep Red (Cell Signaling Technology), respectively, and co-cultured on poly-D-lysine-treated 18 mm diameter glass coverslips (Fisher Scientific) for 16 h. After the 16 h incubation period, cells were fixed on ice with 4% methanol-free paraformaldehyde for 15 min, washed 3X with 1x PBS, and permeabilized with 0.5% Triton X-100 for 10 min at room temperature. For labeling nanotubes, cells were stained with 5 U/mL Rhodamine Phalloidin (Biotium) at room temperature for 20 min and washed 3X with 1x PBS. Coverslips were mounted on glass slides, and images were acquired with a ZEISS LSM 710 (RRID:SCR_018063) or ZEISS LSM 800 with Airyscan (RRID:SCR_015963) using Zen software in the UNMC Advanced Microscopy Core Facility (RRID:SCR_022467). For examining mitochondrial transfer in vivo, Cx3cr1^Cre+/− MitoTag^fl/fl mice were euthanized using an overdose of isoflurane and immediately perfused with PBS, whereupon the infected leg was fixed in 4% paraformaldehyde. After overnight fixation, legs were embedded in optimal cutting temperature medium and 10 μm cryostat sections were prepared. Tissue sections were stained with Ly6G-Alexa Fluor 647 (RRID:AB_1134162) and biotin anti-GFP (RRID: AB_305631) followed by streptavidin-Alexa Fluor 488 (RRID: AB_2337249).

Quantification of mtDNA content and transfer

Total DNA was isolated from MΦs, G-MDSCs, and PMNs using a Wizard Genomic DNA Purification Kit (Promega), according to the manufacturer’s protocol. Real-time PCR was performed using TaqMan probes specific for mtDNA and nuclear DNA (mt-Atp6 and Actb, respectively; both from Thermo Fisher) using 10 ng input DNA, 1x TaqMan probe, and 1x TaqMan Gene Expression Master Mix (Applied Biosystems). Results are expressed as relative mtDNA content by normalizing mtDNA levels to genomic DNA.

As an independent approach to demonstrate mitochondrial transfer, G-MDSCs from BALB/c mice were co-cultured with MΦs from C57BL/6J animals for 18 h, whereupon G-MDSCs were purified using anti-mouse Ly6G microbeads and total DNA was isolated using a Wizard Genomic DNA Purification Kit (Promega). Total DNA was recovered from BALB/c G-MDSCs and C57BL/6J MΦs as a control. A 385 bp fragment of the mt-Co3 gene, encompassing the A9348G polymorphism in C57BL/6J mice, was PCR amplified with forward mt-Co3-F, (CGAAACCACATAAATCAAGCCC) and reverse mt-Co3-R (CTCTCTTCTGGGTTTATTCAGA) primers. The PCR product was purified using a Monarch Spin PCR & DNA Cleanup Kit and digested with PflFI (both from New England Biolabs) that recognizes the AspI restriction site for 30 min at 37°C. DNA fragments were resolved on a 2% agarose gel containing 0.5 μg/mL EtBr and visualized using an Azure Imaging System (Azure Biosystems).

Mitochondria and glycolysis stress tests

OCR and ECAR in G-MDSCs, PMNs, and MΦs was determined using a Seahorse XFe96 analyzer (Agilent, RRID:SCR_019545). For the Mito stress test, cells were collected after the 16–18 h co-culture period and washed with Mito stress test medium (DMEM with 10 mM glucose, 2 mM L-glutamine, 1 mM sodium pyruvate, and 1x antibiotics, pH 7.4) before adding to a 96-well Seahorse plate coated with 10 mg/mL poly-D-lysine at a density of 5 x 10⁵ cells per well in Mito stress test medium. After a 1 h incubation at 37°C, OCR was measured at steady state and after sequential injection of oligomycin (1 μM), FCCP (2 μM), and rotenone (0.5 μM) + antimycin A (0.5 μM). For the Glycolysis stress test, cells were washed with Glycolysis stress test medium (DMEM with 2 mM L-glutamine and 1x antibiotics, pH 7.4) and then added to poly-D-lysine-coated Seahorse plates at a density of 5 x 10⁵ cells per well in Glycolysis stress test medium. After incubation at 37°C for 1 h, ECAR was measured at baseline and after sequential injection of glucose (10 mM), oligomycin (1 μM), and 2-DG (100 mM).

The ability of G-MDSCs, PMNs, and MΦs to metabolize the tetrazolium salt 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was assessed as another readout of mitochondrial activity. Cells were incubated in serum-free medium containing 0.1 mg/mL MTT at 37°C for 1 h, whereupon the supernatant was removed, and cells were lysed in 100 μL DMSO by incubating at 37°C for 30 min on a plate shaker. Plates were read using a spectrophotometer at 570 nm, with results presented as OD₅₇₀.

SCENITH

A SCENITH kit (www.gammaomics.com) was used as an independent method to assess cellular metabolism as previously described.²⁴ Cells were treated with 2-DG (100 mM), oligomycin (1 μM), or a sequential combination of 2-DG and oligomycin for 45 min in complete medium, whereupon 10 μg/mL puromycin was added during the last 30 min of inhibitor treatment. Cells were washed in ice-cold 1x PBS and stained with surface markers for 20 min at 4°C in 1x PBS followed by fixation and permeabilization using a Foxp3/Transcription Factor Staining Kit (Tonbo Biosciences) following the manufacturer’s instructions. Intracellular staining of puromycin was performed with an anti-Puromycin-Alexa Fluor 647 mAb for 1 h at 4°C in Foxp3 permeabilization buffer with cells analyzed using a BD LSRII Flow Cytometer.

Intracellular cytokine staining

G-MDSCs were exposed to S. aureus biofilm for 4 h in the presence of 5 μg/mL Brefeldin A using a Transwell system followed by staining for surface markers (CD45, Ly6G, Ly6C, F4/80, and MHCII) and Zombie UV viability dye for 20 min. Cells were washed and fixed in Cyto-Fast Fix/Perm Buffer for 20 min at room temperature and washed 2X with Intracellular Staining Perm Wash Buffer (both from BioLegend) followed by staining with IL-10-PE/Cy7 (RRID:AB_11149682), IL-1β-APC-eFluor 780 (RRID:AB_2573996), TNF-PerCP/Cy5.5 (RRID:AB_961434), and IL-6-APC (RRID:AB_10694868) for 60 min at room temperature in Intracellular Staining Perm Wash Buffer. Cells were washed and resuspended in 1x PBS for analysis using a BD LSRII Flow Cytometer.

CCL2 quantification

CCL2 production by G-MDSCs was assessed using either a mouse Cytometric Bead Array (BD Biosciences) or ELISA (BioLegend). CCL2 levels in tissue homogenates were normalized to total protein and are reported as pg/mg protein. For CBAs, IFN-ɣ, IL-6, IL-10, and IL-12 production by G-MDSCs was below the limit of detection and not reported.

T cell suppression assay

The effect of mitochondrial transfer on G-MDSC activity was assessed using T cell proliferation assays. For in vitro studies, G-MDSCs were purified following MΦ co-culture using either anti-Ly6G microbeads (Miltenyi) or FACS with CD45-APC (RRID:AB_312977), Ly6G-PE (RRID:AB_1186099), F4/80-PECy7 (RRID:AB_893478), and Ly6C-PerCP-Cy5.5 (RRID:AB_1727558). For isolating G-MDSCs and PMNs from PJI tissues, cells from 6 mice per group were pooled, whereupon CD45⁺Ly6G⁺Ly6C⁺CD11b^high G-MDSCs and CD45⁺Ly6G⁺Ly6C⁺CD11b^low PMNs were collected by FACS using CD45-APC (RRID:AB_312977), Ly6G-PE (RRI-D:AB_1186099), F4/80-PECy7 (RRID:AB_893478), and CD11b-Pacific Blue (RRID:AB_755985). Cells were isolated using either FACSAria, FACSAria II, or Thermo Scientific Bigfoot sorters. For inhibitor studies, mitochondrial-transferred G-MDSCs were pretreated with 500 μM N-acetyl-L-Cysteine (NAC), 500 μM L-NIL, 100 μM nor-NOHA, or 20 μM CAY10452 (all from Cayman Chemical) 30 min before and throughout the T cell suppression assay. CD4⁺ T cells were purified from the spleens of naive mice using a MojoSort Mouse CD4 T cell Isolation Kit (BioLegend) and immediately labeled with eFluor 670 cell proliferation dye (eBioscience) according to the manufacturer’s instructions. After staining, T cells were washed and resuspended in RPMI medium containing 10% FBS, 2 mM L-glutamine, 10 mM HEPES, 1x antimycotic/antibiotic, 50 μM β-mercaptoethanol, and recombinant mouse IL-2 (100 ng/mL). T cell proliferation was induced with either CD3/CD28 Dynabeads (Thermo Fisher) or plate bound anti-CD3 (RRI-D:AB_2616673) and soluble anti-CD28 (RRID:AB_11147170). T cells were co-cultured with G-MDSCs or PMNs at a 1:1 ratio for 48–72 h at 37°C, whereupon the extent of proliferation was assessed with a BD LSRII Flow Cytometer.

MILLIPLEX multi-analyte bead arrays

Cytokine, chemokine, and growth factor levels were quantified in tissue and joint homogenates from the mouse PJI model using a MILLIPLEX MAP kit that measures 25 mediators (Millipore Sigma) and analyzed using a MAGPIX xMAP instrument (Luminex). Results were determined using Belysa Analyst software (Millipore Sigma) with values normalized to total protein content quantified by a Pierce BCA Protein Assay Kit (Thermo Fisher).

Gentamicin protection assay

PMNs and G-MDSCs were exposed to live S. aureus at a MOI of 10:1 (bacteria:cell) for 1 h, whereupon remaining extracellular bacteria were killed by high dose gentamicin (100 μg/mL) for 30 min. Cells were maintained in medium containing 1 μg/mL gentamicin to prevent bacterial outgrowth throughout the culture period. At respective timepoints, cells were lysed in sterile water for 15 min, serially diluted in PBS, and plated on blood agar to quantify intracellular bacterial burden. Percent S. aureus survival was calculated by comparing the bacterial count at 24 h to initial bacterial uptake after high-dose gentamicin treatment.

Bulk RNA-seq

Total RNA was isolated from G-MDSCs using a Quick-RNA Microprep Kit (Zymo Research) with RNA concentration and quality assessed using an Agilent 5200 Fragment Analyzer (RRID:SCR_019417). A SMART-Seq v4 Ultra Low Input RNA Kit (Takara Bio) was used for cDNA synthesis with 10 ng of input RNA and cDNA libraries were prepared using a Nextera XT DNA Library Prep kit (Illumina). Libraries were quantified with a Qubit 3.0 Fluorometer (RRID:SCR_020311) and assessed using a fragment analyzer before sequencing on a NovaSeq 6000 (RRID:SCR_016387) with S1-200 flow cell (100 pair-end reads) at a sequencing depth of ~100 million paired reads per sample. Fastq files were imported into Partek Genomics Suite (RRID: SCR_011860) for subsequent data processing and analysis.^49,50 Read quality was assessed and Phred scores less than 35 were trimmed. Sequencing reads were aligned with STAR (2.7.8a) using the mouse reference genome mm10.⁸⁰ Gene counts were generated and normalized using Partek E/M, with DESeq2 used for differential analysis. Pathway analysis of differentially expressed genes was performed using IPA (Qiagen, RRID:SCR_008653). The resulting dataset has been deposited in the GEO database (GSE295742).

Single-cell RNA-seq

To determine how functional mitochondria affected leukocyte transcriptional programs during S. aureus PJI, 5 x 10⁶ EtBr-treated or normal MΦs were injected into the soft tissue surrounding the infected joint at day 3 post-infection, whereupon single cells were recovered 24 h later. Cells were stained with Zombie UV and anti-mouse CD45-APC (RRID:AB_312977) (both from BioLegend) for 20 min at 4°C. Viable CD45⁺ cells were collected by FACS and analyzed with a Luna automated fluorescent cell counter (Logos Biosystems) to assess viability and density before single cell capture from mice receiving EtBr-MΦs, normal MΦs, or PBS vehicle (3,123, 4,114, and 5,252 captured cells, respectively) with a 10X Genomics instrument. RNA was barcoded and cDNA libraries prepared using a GEM-X Universal 3′ Gene Expression v4 4-plex kit (10x Genomics) according to the manufacturer’s instructions. The 4-plex library was sequenced on an Illumina Novseq 6000 (RRID:SCR_016387) with an SP-100 flow cell resulting in ~700 million reads.

Sequencing data were demultiplexed and aligned to the Mus musculus genome (GRCm39) with single cell count files generated using 10X Genomics Cell Ranger (RRID: SCR_017344). The counts file for each sample was imported into Partek Genomics Suite (RRID:SCR_011860) for further analysis as previously described.⁵⁰ Single cell counts were normalized as counts per million and log transformed in Partek Genomics Suite. Cell type identification was performed with the SingleR package using the Immgen database as reported.⁴⁹ Graph-based clustering was performed for cluster assignment. G-MDSC-like clusters were identified as previously described using mouse G-MDSC marker genes.⁴⁹ Gene-specific analysis was used to identify differentially expressed genes between clusters, and the pathways altered between clusters were assessed using IPA (Qiagen, RRID:SCR_008653). The resulting dataset has been deposited in the GEO database (GSE313764).

Metabolomics

The intracellular metabolome of G-MDSCs ± mitochondrial transfer was profiled by LC-HRMS. Cells were washed twice in 1x PBS, resuspended in 80% MeOH containing a ¹³C¹⁵N-CAA mix as the internal standard, and frozen at −80°C for 15 min. Resulting cell lysates were pelleted, supernatant collected and dried by vacuum centrifugation, and stored at −80°C until LC-HRMS analysis. The resulting pellet was saved for protein quantification and standardization of metabolite peak areas. Prior to LC-HRMS, samples were reconstituted in 100 μL of 50% MeOH, pelleted, and the resulting supernatant was submitted for analysis. For quality control (QC), 20 μL from each sample was pooled and injected as five technical replicates before and after the sample run. The ¹³C¹⁵N-CAA internal standards in these pooled QC samples were used to determine instrument stability during the sample run (Relative standard deviation, RSD <15%). The remaining metabolites in the pooled samples were used to filter out metabolites in the individual samples that did not meet the QC threshold (>25% RSD). Untargeted metabolomics was performed using a High-resolution (Hybrid) mass spectrometer (Orbitrap viz., Exploris 480; Thermo Fisher Scientific) connected with an ultra-high-performance liquid chromatography (UHPLC) system (Thermo Fisher Scientific). Metabolite separation was performed by liquid chromatography using a XBridge Amide analytical column (150 × 2.1 mm ID; 1.7 μm particle size; Waters Corporation) and a binary solvent system infused at a flow rate of 0.3 mL/min. A guard XBridge Amide column (20 × 2.1 mm ID; 1.7 μm particle size; Waters Corporation) was connected in front of the analytical column. Mobile phase A was composed of ammonium acetate and ammonium hydroxide (10 mM each) containing 5% acetonitrile in LC-MS grade water; mobile phase B was 100% LC-MS grade acetonitrile. The pH of mobile phase A was adjusted to 8.0 using glacial acetic acid. The UHPLC pumps were operated in gradient mode. The amide column was maintained at 40°C, and the autosampler temperature was held at 5°C throughout data acquisition.

An HRMS Orbitrap (Exploris 480; Thermo Fisher Scientific) was operated in polarity switching mode and used for untargeted metabolomics in a data-dependent MS/MS acquisition mode (DDA). Electrospray ionization (ESI) parameters were optimized as follows: electrospray ion voltage of −2700V and 3500V in negative and positive mode respectively, ion transfer tube was maintained at 300°C, and m/z scan range was 70–1050 Da. Orbitrap resolution for precursor ion and fragment ion scans was maintained at 120,000 and 60,000, respectively. Normalized collision energies at 30, 50, and 150% were used for fragmentation and data was acquired in profile mode.

Metabolites were detected within a mass tolerance limit of 5 ppm. Precursor ions generating MS/MS spectra using Data Dependent Acquisition (DDA) for preferred ions were selected for further analysis. Identification and detection of all metabolites was aided by the Compound Discoverer (CD) software (Thermo Fisher Scientific) using a workflow that included Predicted composition, Metabolika search, mzCloud, and an internal MassList database containing 4,400 metabolites from Thermo Fisher Scientific. Additionally, the HMDB and KEGG database plug-ins within the CD software were utilized to screen for endogenous metabolites. The generated data was normalized to the internal standard and protein concentration before analysis by MetaboAnalyst 5.0.^81,82 The resulting dataset has been deposted in the NIH Common Fund’s National Metabolomics Data Repository (https://doi.org/10.21228/M84P2S).

QUANTIFICATION AND STATISTICAL ANALYSIS

Sample sizes for each experiment were determined based on standard practices in the field. Significant differences between groups were determined using GraphPad Prism (RRID:SCR_002798). All statistical details, including the exact value of n, what n represents (individual mice or independent biological replicates), and the statistical tests used are provided in the figure legends. For RNA-seq analysis, statistical tests were determined using Partek Flow.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2026.117057.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
CD45-APC	BioLegend	Cat#103112; RRID:AB_312977
Ly6G-PE	BioLegend	Cat#127608; RRID:AB_1186099
F4/80-PECy7	BioLegend	Cat#123114; RRID:AB_893478
Ly6C-PerCP-Cy5.5	BD Biosciences	Cat#560525; RRID:AB_1727558
CD11b-Pacific Blue	BioLegend	Cat#101224; RRID:AB_755985
CD45.1-BV711	BioLegend	Cat#110739; RRID: AB_2562605
CD45.2-APC	BioLegend	Cat#109814; RRID: AB_389210
CD11b-AF700	BioLegend	Cat#101222; RRID: AB_493705
MHC Class II-BV605	BioLegend	Cat#107639; RRID: AB_2565894
Ly6C-APC-Cy7	BD Biosciences	Cat#560596; RRID: AB_1727555
IL-10-PE/Cyanine7	BioLegend	Cat#505026; RRID:AB_11149682
IL-1β-APC-eFluor 780	Invitrogen	Cat#47711482; RRID:AB_2573996
TNF-PerCP/Cyanine5.5	BioLegend	Cat#506322; RRID:AB_961434
IL-6-APC	BioLegend	Cat#504508; RRID:AB_10694868
CD45-Pacific Blue	BioLegend	Cat#103126; RRID:AB_493535
Anti-Puromycin Alexa Fluor 647	Gammaomics	RRID:AB_2827926
Ly6G-Alexa Fluor 647	BioLegend	Cat#127609; RRID:AB_1134162
Biotin Anti-GFP	Abcam	Cat#ab6658; RRID:AB_305631
Alexa Fluor 488 Streptavidin	Jackson ImmunoResearch	Cat#016-540-084; RRID:AB_2337249
Anti-mouse CD3ε	BioLegend	Cat#100359; RRID:AB_2616673
Anti-mouse CD28	BioLegend	Cat#102116; RRID:AB_11147170
APC Anti-GFP	BioLegend	Cat#338010; RRID:AB_2650669
Bacterial and virus strains
Staphylococcus aureus USA300 LAC 13C	Laboratory stock	N/A
Chemicals, peptides, and recombinant proteins
TaqMan Gene Expression Master Mix	Applied Biosystems	4369016
Seahorse XFe96 FluxPak mini	Agilent	103793–100
Invitrogen eBioscience Cell Proliferation Dye eFluor™ 670	Thermo Fisher Scientific	50-112-8751
UltraComp eBeads Compensation Beads	Invitrogen	01-2222-42
TruStain FcX (anti-mouse CD16/32) Antibody	BioLegend	101320
Zombie UV Fixable Viability Kit	BioLegend	423108
Zombie NIR Fixable Viability Kit	BioLegend	423106
2-NBDG	Cayman Chemical	11046
Mitotracker Green FM	Cell Signaling Technology	9074S
MitoSOX	Invitrogen	M36008
MitoPY1	Tocris	4428
MitoTracker Deep Red FM	Cell Signaling Technology	8778S
CellTracker Violet BMQC	Invitrogen	C10094
Rhodamine Phalloidin	Biotium	00027
Paraformaldehyde 16% Aqueous Solution EM Grade	Electron Microscopy Sciences	15710
Gap26	Cayman Chemical	36625
CD3/CD28 Dynabeads	Thermo Fisher Scientific	11452D
Anti-Ly6G beads	Miltenyi Biotech	130-120-337
2-deoxy-glucose	Cayman Chemical	14325
Oligomycin	Cayman Chemical	11342
FCCP	Cayman Chemical	15218
Rotenone	Cayman Chemical	13995
Antimycin A	Cayman Chemical	34799
RPMI-1640	Cytiva HyClone	SH30027.02
L-Glutamine	Cytiva HyClone	SH30034.01
Antibiotic/Antimycotic Solution	Cytiva HyClone	SV30079.01
HEPES	Cytiva HyClone	SH30237.01
G-CSF	BioLegend	574608
GM-CSF	BioLegend	576308
IL-6	BioLegend	575708
N-Acetyl-L-cysteine	Thermo Fisher Scientific	160280250
L-NIL	Cayman Chemical	80310
nor-NOHA	Cayman Chemical	10006861
CAY10452	Cayman Chemical	10075
Critical commercial assays
SMART-Seq v4 Ultra Low Input RNA Kit	Takara Bio	634888
Nextera XT DNA Library Prep kit	Illumina	FC-131-1024
Mitochondria Isolation Kit	Abcam	ab110170
Quick-RNA Microprep Kit	Zymo Research	R1050
Wizard Genomic DNA Purification Kit	Promega	A1120
MojoSort Mouse CD4 T cell Isolation Kit	BioLegend	480006
ELISA MAX Standard Set Mouse MCP-1	BioLegend	432701
Foxp3/Transcription Factor Staining Buffer Kit	Tonbo Biosciences	TNB-0607-KIT
ATPlite Luminescence Assay System	Revvity Health Sciences	6016943
Cytometric Bead Array	BD Biosciences	552364
MILLIPLEX MAP Mouse Cyto/Chemo MAG 25plex	Millipore Sigma	MCYTMAG70PMX25BK
Deposited data
G-MDSC bulk-RNA seq	This paper	GEO: GSE295742
PJI Tissue scRNA-seq	This paper	GEO: GSE313764
Metabolomics data	This paper	https://doi.org/10.21228/M84P2S
Source data for figures	This paper	https://doi.org/10.17632/rr8fscwk9v.1
Experimental models: Organisms/strains
C57BL/6J	The Jackson Laboratory	RRID:IMSR_JAX:000664
C57BL/6J-Ptprcem6Lutzy/J	The Jackson Laboratory	RRID:IMSR_JAX:033076
Myd88 −/−	The Jackson Laboratory	RRID:IMSR_JAX:009088
Sting −/−	The Jackson Laboratory	RRID:IMSR_JAX:025805
Acod1 −/−	The Jackson Laboratory	RRID:IMSR_JAX:029340
MitoTag	The Jackson Laboratory	RRID:IMSR_JAX:032290
Cx3cr1 Cre	The Jackson Laboratory	RRID: IMSR_JAX:025524
Oligonucleotides
Beta-actin	Thermo Fisher Scientific	Mm02619580_g1
mt-Atp6	Thermo Fisher Scientific	Mm03649417_g1
mt-Co3-F (CGAAACCACATAAATCAAGCCC)	IDT	N/A
mt-Co3-R (CTCTCTTCTGGGTTTATTCAGA)	IDT	N/A
Software and algorithms
FlowJo v.10.10.0	BD Life Sciences	https://www.flowjo.com/
Partek Flow	Illumina	https://www.illumina.com/products/by-type/informatics-products/partek-flow.html
ZEN Microscopy Software	Carl Zeiss	https://www.zeiss.com/microscopy/en/products/software/zeiss-zen.html
GraphPad Prism 10	GraphPad Software	https://www.graphpad.com
Seahorse Wave Desktop Software	Agilent	https://www.agilent.com/en/product/cell-analysis/real-time-cell-metabolic-analysis/xf-software/seahorse-wave-desktop-software-740897
Compound Discoverer Software	Thermo Fisher Scientific	https://www.thermofisher.com/us/en/home/industrial/mass-spectrometry/liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/multi-omics-data-analysis/compound-discoverer-software.html?erpType=Global_E1
R studio	RStudio	https://www.r-project.org/
BioRender	BioRender	https://www.biorender.com/
Cell Ranger Multi v9.0.1	10X Genomics	https://www.10xgenomics.com/support/software/cell-ranger/latest
Belysa® Analyst software	Millipore Sigma	https://www.sigmaaldrich.com/US/en/services/software-and-digital-platforms/belysa-immunoassay-curve-fitting-software?srsltid=AfmBOopRWb9BlGfvLvXH2pSmbcbc3h4W7Ucp6AJEPf8Y3pg1CdQCSSxD

Highlights.

• Macrophages transfer mitochondria to G-MDSCs via tunneling nanotubes

• Mitochondrial acquisition by G-MDSCs increases their suppressive activity

• G-MDSC metabolic reprogramming requires transfer of respiring mitochondria

• Macrophages transfer mitochondria to G-MDSCs in vivo to promote biofilm infection