Loss of mTORC2‐induced metabolic reprogramming in monocytes uncouples migration and maturation from production of proinflammatory mediators

Monocyte migration to the sites of inflammation and maturation into macrophages are key steps for their immune effector function. Here, we show that mechanistic target of rapamycin complex 2 (mTORC2)‐dependent Akt activation is instrumental for metabolic reprogramming at the early stages of macrophage‐mediated immunity. Despite an increased production of proinflammatory mediators, monocytes lacking expression of the mTORC2 component Rictor fail to efficiently migrate to inflammatory sites and fully mature into macrophages, resulting in reduced inflammatory responses in vivo. The mTORC2‐dependent phosphorylation of Akt is instrumental for the enhancement of glycolysis and mitochondrial respiration, required to sustain monocyte maturation and motility. These observations are discussed in the context of therapeutic strategies aimed at selective inhibition of mTORC2 activity.


INTRODUCTION
Extravasationand tissue infiltration [1][2][3] and their further maturation into tissue macrophages 4,5 are crucial to monocyte function. Once in the tissue, a variety of soluble mediators and signaling pathways determine monocyte differentiation into proinflammatory or prohealing macrophages, which contribute to progression and resolution of the inflammatory process, respectively.
In myeloid cells, the PI3K-mechanistic target of rapamycin (mTOR) signaling pathway is activated by TLR ligands and cytokines and controls functional differentiation of monocytes and macrophages. [6][7][8][9] Importantly, macrophage differentiation is intrinsically linked to proinflammatory phenotype. In monocytes, β-glucansstimulate the AKT-mTORC1-HIF1α axis, resulting in a shift from oxidative phosphorylation (oxphos) toward glycolysis. 15 mTORC1-depedent hexokinase (HK1)-mediated glycolysis is indispensable for NLRP3 inflammasome activation and subsequent IL-1β and IL-18 expression in macrophages. 16 Like mTORC1, mTORC2 has been linked to the regulation of glycolysis and oxphos in myeloid cells. mTORC2 activity in myeloid cells, however, appears to promote an anti-inflammatory phenotype.
In macrophages, deletion of Rictor led to reduced glycolysis and oxphos and inhibition of M2 activation upon IL-4 stimulation. 17 More recently, TCA cycle-related genes were shown to be up-regulated in IL-4-treated monocyte-derived macrophages from mTORC2-deficient mice. 18 Upon activation, Rictor-deficient macrophages have been reported to undergo a dramatic shift toward a proinflammatory phenotype, and secrete high levels of cytokines such as TNF-α and IL-6. 19 Despite these observations, the effect of mTORC2 on the inflammatory response in vivo remains unclear. 19 In this study, we investigated the contribution of mTORC2 to distinct early events of monocyte/macrophage-mediated inflammation, namely cell migration and functional maturation. Our data are consistent with an essential role of mTORC2, but not mTORC1, in sustaining migration of monocytes to inflammatory sites and their subsequent maturation via metabolic reprogramming. Importantly, loss of mTORC2-regulated functions outweighs the proinflammatory differentiation of Rictor-deficient macrophages in vivo, resulting in an impaired inflammatory response.
The littermates that are CSF1R-cre ER +/− (referred to here as CSF1R-creER Rictor WT ) were used as control WT mice. For all mice used, the genotypes were determined by PCR analysis of earpiece by a commercial vendor (Transnetyx). To obtain the CSF1R-cre Rictor KO macrophages, following intraperitoneal tamoxifen administration for 3 days, bone marrows (BM) were extracted, and macrophages were matured for 7 days ( Figure S1(A)) with 20 ng/ml M-CSF. All in vivo experiments were conducted with strict adherence to the Home Office guidelines following approval by the Queen Mary University of London Ethics committee.

Reagents
All reagents and resources used in this study are summarized in Table S1.

BM-derived macrophage culture
Bone marrow derived macrophages (BMDMs) originated from hematopoietic progenitors in the BM and are differentiated into mature macrophages in the presence of Refs. 20

Phagocytosis in BMDMs
BMDMs were grown to confluence and then plated at 1 × 10 6 cells/well in 6-well plates. BMDMs were treated with 100 ng/ml LPS and 20 ng/ml

Migration assays
BMDMs were resuspended at 1 × 10 6 cells per ml in DMEM with 10% FBS. To measure chemotaxis 0.1 ml of cells was added to 6.5 mm diam- oligomycin before being exposed to 50 ng/ml CCL2 gradient for 2 h.

Confocal microscopy
Cells were allowed to adhere onto poly-L-lysine coated coverslips and fixed in 4% paraformaldehyde (Thermo Fisher Scientific; cat# 28906) for 5-10 min at room temperature. Where mentioned, permeabilization was carried out using 0.2% Triton X-100 (Sigma- Confocal images and Z stacks were acquired and analyzed by confocal microscopy (Zeiss LSM710). Repositioning of scale bars and image layouts were prepared using Adobe Photoshop (Adobe Systems). All images in a group were treated equally.

BMDM cytokine and activation molecules analysis
Cytokines were measured from cell culture media using commercial ELISA kits from BD Biosciences according to manufacturer's instructions. For cell surface marker analysis, 1 × 10 6 BMDMs were incubated with LPS/IFNγ for 6 h and then processed for flow cytometry analysis.

2.13
Measurement of oxygen consumption rates and extracellular acidification rates

ATP assay
Monocytes from CSF1R-cre Rictor KO and CSF1R-creER Rictor WT mice were isolated and plated at 1 × 10 6 per well in triplicates. They were treated with either media only or 50 ng/ml CCL2 for 2 h. The cells were then lysed in 50 μl somatic cell ATP releasing reagent (Sigma; cat# FLSAR) and cellular ATP measured using the ATP determination kit (Thermofisher; cat# A22066) according to the manufacturer's instructions. The experiment was repeated with 3 biologic replicates.

RESULTS
Loss of Rictor expression results in proinflammatory differentiation of macrophages in vitro, but reduced monocyte recruitment in vivo.
In this study, we aimed to investigate the effects of Rictor deletion on macrophage functions in vitro. It has been previously shown that mice lacking Rictor in both macrophages and neutrophils under the control of the LysMcre promoter become hypersensitive to TLR4 stimulation. 19 Rictor-LysMcre KO BMDMs produced significantly higher levels of inflammatory cytokines following exposure to LPS compared with their WT counterparts. 19 To prevent the deletion of all myeloid cells, we utilized the tamoxifen-inducible Cre recombinase under the control of CSF 1 receptor promoter (CSF1R-icre), specific for the monocyte and macrophage lineage 23 (Figures S1(A) and S1(B)). When crossed with Rictor f/f or Raptor f/f mice, rictor and raptor were deleted only in monocyte/macrophage lineage ( Figure S1(C) and S1(D)). The lines were named CSF1R-icreER Rictor KO and CSF1R-icreER Raptor KO , respectively, whereas the littermates used as controls were CSF1R-icreER Rictor WT or CSF1R-icreER Raptor WT .
To assess the contribution of mTORC2 to monocyte/macrophage proinflammatory differentiation, BMDMs were exposed to LPS and In line with the data described in Figure 1, zymosan-treated BMDMs displayed higher mRNA levels of the proinflammatory mediator IL-1β in the absence of mTORC2 activity Figure S2(C)).
To formally rule out potential effects of Rictor deficiency on resident macrophages, we further investigated migration of monocytes in thioglycollate-induced peritonitis, collecting migrating leukocytes as a single cell suspension from the peritoneal cavity. As shown in Rictor WT cells (Figures 3(A) and 3(B)). We also analyzed the CSF1R-creER Raptor KO BMDMs in in vitro migration assays and did not observe any difference between the Raptor-deficient and WT groups ( Figure 3(C)).
These observations were further confirmed using the chemokine CCL2 to drive monocyte migration. As shown in Figure S3 27 We conclude that the mTORC2 complex is a key regulator of cytoskeleton reorganization during monocyte/macrophage migration.
Most Rictor-mediated functions are dependent on mTORC2 ability to phosphorylate the kinase Akt at Serine 473. 28 We therefore assessed key signaling mediators downstream of Akt, which could underlie the functional phenotype of Rictor-deficient macrophages.
BMDMs from CSF1R-icreER Rictor KO and CSF1R-icreER Rictor WT mice were stimulated in vitro with LPS/IFNγ at the indicated time points and lysates were analyzed by Western blot. As expected, phosphorylation of AKT at S473 was completely abolished in CSF1R-icreER Rictor KO BMDMs (Figure 4(A)).

F I G U R E 4 CSF1R-icreER Rictor
Phosphorylation of AKT at S473 remained intact in Raptor-deficient macrophages, while ribosomal protein S6 phosphorylation at S235-236 was significantly reduced ( Figure S4). In these cells, greater IĸB phosphorylation at S32 confirmed the previously described activation of the NFĸB-mediated pathway and higher cytokine expression in CSF1R-icreER Raptor KO BMDMs. 25 While efficient mTORC1 activity is indispensable for NF-ĸB activity in BMDMs, Rictor deficiency in CSF1R-icreER Rictor KO BMDMs did not alter the phosphorylation levels of MAPK p38 (Figure 4(B)) The transcription factor c-Myc, a major regulator of cell metabolism and survival, 30 is required for the differentiation of alternatively acti- Statistical test: unpaired t-test, *p < 0.05, *p < 0.01, ***p < 0.001 Expression of the transcription factor IRF4 has been recently shown to be dependent on mTORC2 signals during differentiation of alternatively activated macrophages, 34 where it promotes the glycolytic pathway. However, as shown in Figure 4(F), expression of IRF4 was not affected by Rictor deficiency in classically activated macrophages. Transcription of the integrin beta chain cd18 gene was also preserved in Rictor-deficient LPS/IFNγ-activated macrophages ( Figure 4(G)). Since c-Myc is an upstream regulator, this ruled out any adhesion-dependent defects in Rictor-deficient macrophage motility.
Rictor is required for coupling of promigratory and promaturation stimuli with metabolic responses in monocytes.
We recently reported that the PI3K-mTORC2-Akt pathway sustains migration of regulatory T cells (Tregs) via induction of the hexokinase isoenzyme glucokinase. 3 Having observed that LPS/IFNγ-stimulated CSF1R-icreER Rictor KO BMDMs have lower transcription of aldolasea and pfk while also displaying impaired cytoskeletal recombination and migratory ability, we set out to investigate the contribution of the mTORC2-Akt pathway in reprogramming metabolism of migrating monocytes. Exposure of WT BMDMs to the chemokine CCL2 led to an increase in phosphorylation of Akt at S473 (Figure 5(A)), which, in contrast, remained at basal levels in CSF1R-icreER Rictor KO monocytes.
In addition, CCL2-induced Akt phosphorylation in WT monocytes was accompanied by enhanced mitochondrial activation and ROS production ( Figures 5(B) and 5(C)).
Migration of monocytes pretreated either with vehicle, 2DG or etomoxir toward a CCL2 gradient was subsequently analyzed. 2DG is a glucose analogue that competitively inhibits glycolysis while etomoxir inhibits fatty acid oxidation and mitochondrial respiration (oxidative phosphorylation, oxphos). Both molecules significantly reduced the migration of monocytes in response to CCL2 (Figure 5(D)), suggesting that monocytes require glycolysis and oxidative phosphorylation (oxphos) to sustain the energetic demands of migration.
We next compared ATP levels in these cells and observed that, after stimulation with CCL2, CSF1R-icreER Rictor WT monocytes produced significantly more ATP than CSF1R-icreER Rictor KO  The observations made in vivo were confirmed by in vitro migration assays in which migration of 2DG pretreated monocytes was reduced, while FCCP enhanced CCL2-induced migration ( Figure 5(L)).
Further, monocytes were pretreated with oligomycin and then analyzed for their migratory activity in response to CCL2. Oligomycin, an ATP synthase inhibitor, selectively inhibits oxidative phosphorylation and so is indicative of energy release through aerobic glycolysis.
As shown in Figure 5(l), migration of oligomycin-treated monocytes to CCL2 was only partly reduced compared with migration of 2DG treated monocytes, suggesting that glycolysis can partly compensate for oxphos during monocyte migration, but not vice-versa.
We further sought to confirm the role of Akt activation in the events described above. Prior to exposure to CCL2, monocytes were treated with vehicle, 2DG and an Akt-selective inhibitor. Pharmacologic inhibition of Akt was also carried out in monocytes during treatment with FCCP and Oligomycin. As shown in Figure 6(A), inhibition of Akt phosphorylation reduced monocyte migration to levels similar to those elicited in the presence of 2DG. In addition, preventing Akt activation abrogated the enhancing effect of FCCP on monocyte migration and further increased the inhibitory effect of Oligomycin ( Figure S5(B)).
In line with these results, qPCR analysis of monocytes exposed to CCL2 revealed a significant induction of several enzymes of the glycolytic pathway, TCA cycle and lipid metabolism, which were abrogated by inhibition of Akt signaling. Specifically, exposure to CCL2 significantly increased transcription of key enzymes of the glycolytic pathway, including pkm2 and enolase-2 as well as c-Myc, an upstream transcription factor, and cpt1a, a major fatty acid oxidation enzyme ( Figures 6(B)-6(H)).
In line with our previous observation that the defective migration of Rictor-deficient monocytes is not due to a defect in expression of adhesion molecules, CD11a and CD18 expression was also unaffected by pharmacologic Akt inhibition (Figures 6(I) and 6(J)).
It has previously been shown that chemokines and integrins can induce monocyte maturation into macrophages. 38 We have shown that Collectively, these data suggest that Akt activation by mTORC2 is required to maintain the energy supply required for monocyte migration and maturation.

DISCUSSION
The showed that loss of mTORC2 is accompanied with a more proinflammatory macrophage signature that induces colitis and tumor growth in mice.
Despite the proinflammatory phenotype in vitro, monocyte/macrophage specific Rictor deficiency results in an antiinflammatory phenotype in vivo. Our data indicate that this phenotype is at least in part due to defective migration and maturation   45 ACLY activity has been linked to increased glucose consumption and higher NO and ROS production in macrophages 45,46 and has been linked to efficient macrophage differentiation and polarization through increased acetylation of genes in alternatively F I G U R E 7 Chemokine-mediated monocyte to macrophage differentiation is regulated by AKT-dependent metabolic reprograming. Bone marrow derived cells were stimulated with media only or CCL2 in the presence of the indicated inhibitors. The percentage of CD11b hi F480+ macrophages were analyzed by flow cytometry and shown as the mean percentage of total cells ± SD. Statistical test: unpaired t-test, **p < 0.01 activated macrophages. 47 Further studies will be needed to challenge this hypothesis.
Reduced oxphos as a consequence of Rictor deficiency did not affect monocyte migration as much as the impairment of the glycolytic flux. While the full functional consequences of mTORC2 regulation of macrophage oxphos remain to be fully investigated, the partial effect on migration can be explained by the failure of upstream glycolysis to engage oxphos, while ATP can still be produced by aerobic glycolysis independently of mitochondrial respiration.
Despite being functionally proinflammatory, Rictor-deficient, classically activated macrophages also displayed maturation defects illustrated by lower CD64, CD86, and F4/80 molecule expression. In addition, CCL2-induced maturation of monocytes in nonpolarizing conditions was substantially reduced following Akt and glycolysis inhibition, suggesting that mTORC2-induced, Akt-mediated regulation of metabolic responses is required in the early events of macrophagemediated immunity, namely monocyte migration and maturation. In a recent study, Rictor deficiency in TLR4-activated dendritic cells (DCs) led to enhanced IL-6 and IL-23 expression but also lower B7-H1 and CD40 expression. 48 Rictor deficiency in alternatively activated macrophages leads to a proinflammatory phenotype in vivo, suggesting that monocytes undergoing M2-like differentiation do not depend on Rictor to migrate into tissues. It is possible that these cells rely upon other signaling pathways to meet the energy demands of migration, while silencing of Rictor signaling is necessary for the development of M1-like function, independently of the environmental cues. Subsequently, continued mTORC2 activity in alternatively activated macrophages may promote the acquisition of an anti-inflammatory phenotype. In alternatively activated macrophages, mTORC2 has been shown to operate in parallel with the IL-4Ra-Stat6 pathway to facilitate increased gly-colysis during M2 activation via the induction of the transcription factor IRF4. 34 Our investigation shows that IRF4 transcription was not reduced in Rictor-deficient, classically activated macrophages suggesting that Rictor induces different transcriptional programs to increase glycolysis in M1 and M2-like macrophages, macrophages, which could explain the lack of effect of Rictor deficiency on migration and maturation of alternatively activated monocytes.
Besides revealing a distinct transcriptional programme downstream of Rictor in classically and alternatively differentiated macrophages, our study is relevant to the recent interest in the development of mTORC2 selective inhibitors in cancer therapy. 49 Based in part on the proinflammatory effects on monocyte differentiation, 50,17 it has been proposed that selective inhibitors of mTORC1 and 2 may be effective anti-cancer drugs. 51 Our data suggest that this might be a flawed approach, as the increased inflammatory phenotype elicited by loss of mTORC2 activity may be outweighed by the inability of monocytes to reach the tumor tissue and differentiate.

ACKNOWLEDGMENTS
This study was funded by the British Heart Foundation

DISCLOSURES
The authors declare no conflict of interests.