eIF4E-Binding Proteins 1 and 2 Limit Macrophage Anti-Inflammatory Responses through Translational Repression of IL-10 and Cyclooxygenase-2

In eukaryotes, translational control (i.e., regulation of the efficiency of mRNA translation) mostly occurs at the rate-limiting initiation step during which the ribosome is recruited to the mRNA (1). This process is facilitated by the eukaryotic translation initiation factor 4F (eIF4F), a heterotrimeric complex consisting of eIF4E, the mRNA 5′-m⁷G-cap–binding subunit; eIF4A, an RNA helicase; and eIF4G, a scaffolding protein. Assembly of eIF4F is blocked by a reversible association between eIF4E and eIF4E-binding proteins (4E-BPs), a family of repressor proteins that in mammals comprises 4E-BP1, 4E-BP2, and 4E-BP3 (2). The mechanistic target of rapamycin (mTOR) complex 1 (mTORC1) phosphorylates 4E-BP1 and 4E-BP2 (4E-BP1/2), promotes their dissociation from eIF4E, and thereby enables the assembly of a functional eIF4F complex. Conversely, in conditions of mTORC1 inhibition, hypophosphorylated 4E-BP1/2 and inducible 4E-BP3 bind with high affinity to eIF4E, prevent the formation of the eIF4F complex, and thereby inhibit initiation of translation (3, 4).

Translational control enables cells to rapidly adjust their proteomes in response to stress without the requirement of de novo mRNA synthesis (5). The ability to quickly modulate gene expression is a key feature of the immune system; therefore, several innate immune regulators are under translational control. For example, analysis of the translatome (i.e., the transcriptome-wide pools of efficiently translated mRNA) in mouse embryonic fibroblasts (MEFs) from 4E-BP1/2 double-knockout (DKO) mice (Eif4ebp1^−/−/Eif4ebp2^−/−) identified Irf7 as an mRNA translated in a 4E-BP1/2–sensitive fashion (6). Moreover, similar studies on MEFs from mice mutated at the residue where eIF4E is phosphorylated (i.e., eIF4E S209A knockin) demonstrated that translational efficiency of IκBα, the inhibitor of NF-κB, is controlled by the MAPK-interacting kinase (MNK)–eIF4E axis (7, 8). The transcription factors IFN regulatory factor (IRF) 7 and NF-κB promote the activation of Ifnα and Ifnβ genes (9, 10). Accordingly, 4E-BP1/2 DKO and eIF4E knockin MEFs and mice are resistant to viral infections owing to enhanced type I IFN responses (6–8). These findings support the notion that eIF4E-dependent translational control constitutes an important regulatory mechanism of innate immune responses. Yet, these studies used MEFs and therefore may not reflect the entire transcript repertoire under translational control in immune cells.

Macrophages are sentinels of the innate immune system that alter their phenotype, ranging from inflammatory to regulatory and anti-inflammatory depending on the environmental cues (11). Selective changes in translational efficiency direct macrophage differentiation (12) and activation (13–17). This includes the response to cytokines and TLR ligands, in which regulation of mTORC1 and MNK signaling modulates translational efficiency of immune-related transcripts (13, 14, 17). Moreover, previous studies conducted in MEFs demonstrated that 4E-BP1/2 play a crucial role in translational control of antiviral innate immunity (6, 18). Consistently, 4E-BP1/2 are involved in macrophage resistance to infection by a protozoan parasite through type I IFN– and NO-mediated mechanisms (19). These findings indicate that 4E-BP1/2 regulate macrophage proinflammatory and microbicidal functions via selective changes in translational efficiencies; however, the identities of such transcripts remain unknown. Notably, the impact of 4E-BP1/2–dependent translational control in macrophage anti-inflammatory responses is yet to be investigated. In this study, we show that translational efficiency of mRNAs encoding anti-inflammatory mediators, IL-10 and cyclooxygenase-2 (COX-2), is regulated through 4E-BP1/2, which thereby modulate the anti-inflammatory phenotype of macrophages.

LPS (Escherichia coli serotype 0111:B4) and cycloheximide were purchased from Sigma-Aldrich. NS-398 was provided by Cayman Chemical. Rat mAb against mouse IL-10 (no. MAB417) and rat IgG1 isotype control (no. MAB005) were obtained from R&D Systems. BP-1-102 was purchased from Selleck Chemicals. DMEM, FBS, HBSS, 0.05% EDTA-Trypsin, penicillin, and streptomycin were provided by Wisent.

Hind legs from Eif4ebp1^−/−/Eif4ebp2^−/− C57BL/6 mice (6, 19) and their wild-type (WT) C57BL/6 littermates, originally purchased from The Jackson Laboratory, were kindly provided by Dr. N. Sonenberg (McGill University, Montreal, QC, Canada). All procedures were in compliance with the Canadian Council on Animal Care guidelines and approved by the Comité Institutionnel de Protection des Animaux of the Institut National de la Recherche Scientifique (no. 1611-10). Bone marrow precursor cells were extracted from the femurs and tibias for differentiation into bone marrow–derived macrophages (BMDM). Briefly, marrow was flushed from femurs and tibias maintained in HBSS (100 U/ml penicillin, 100 μg/ml streptomycin, 4.2 mM sodium bicarbonate, 20 mM HEPES) at 4°C. Precursor cells were resuspended in BMDM culture medium (DMEM, 10% heat-inactivated FBS, 2 mM l-glutamate, 1 mM sodium pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin) supplemented with 15% L929 fibroblast-conditioned culture medium (LCCM). Cells were seeded in 10-cm-diameter tissue culture–treated dishes and incubated overnight at 37°C and 5% CO₂. The following day, nonadherent cells were collected, resuspended in BMDM culture medium supplemented with 15% LCCM, and plated in 10-cm-diameter nontreated petri dishes (approximately five dishes per mouse). LCCM was added every 2 d (∼1.5 ml/dish), and differentiated BMDM were collected at 8 d after marrow extraction. Differentiation of precursor cells into macrophages was routinely assessed by monitoring for CD11b and F4/80 coexpression by flow cytometry using allophycocyanin anti-mouse/human CD11b Ab no. 101211 and PE anti-mouse F4/80 Ab no. 123109 (BioLegend), as previously described (20).

Samples were processed for polysome profiling and RNA fractionation as previously described (19). BMDM were seeded in 15-cm-diameter culture dishes (3 × 10⁷ cells per plate) in DMEM containing 100 U/ml penicillin, 100 μg/ml streptomycin, and supplemented with 10% FBS and 1% LCCM. A total of 9 × 10⁷ cells per genotype were used to generate each polysome profile. Cells were treated with 100 μg/ml cycloheximide for 5 min and were washed three times with cold PBS containing 100 μg/ml cycloheximide. Cells were centrifuged at 200 × g for 10 min at 4°C and lysed in hypotonic lysis buffer containing 5 mM Tris-HCl (pH 7.5), 2.5 mM MgCl₂, 1.5 mM KCl, 100 μg/ml cycloheximide, 2 mM DTT, 0.5% Triton X-100, 0.5% sodium deoxycholate, and 200 U RNasin (Promega). Lysates were cleared by centrifugation (20,000 × g, 2 min at 4°C). A 50-μl sample was collected (10% of the lysate) to isolate cytoplasmic RNA using TRIzol (Invitrogen). Five to fifty percent (w/v) sucrose density gradients (20 mM HEPES-KOH [pH 7.6], 100 mM KCl, 5 mM MgCl₂) were generated using a Gradient Master 108 (Biocomp Instruments). Samples were loaded onto the sucrose gradients and subjected to ultracentrifugation at 221,830.9 × g (SW 41 Rotor; Beckman Coulter) for 2 h at 4°C. Sucrose gradients were fractionated (35 s for each fraction = 750 μl/fraction) by displacement by 60% sucrose/0.01% bromphenol blue. The OD at 254 nm was continuously recorded using a BR-188 Density Gradient Fractionation System (Brandel). Fractions were flash frozen immediately after fractionation and stored at −80°C. RNA from each fraction was isolated using TRIzol and purified using the RNeasy Kit (Qiagen). Fractions with mRNA associated with more than three ribosomes were pooled (heavy polysome-associated mRNA).

RNA samples from three independent biological replicas were prepared for NanoString nCounter assays and analyzed as previously described (21). In addition to heavy polysome-associated mRNA samples described above, a parallel sample was collected from the lysates loaded onto the sucrose gradient (total cytoplasmic mRNA), and RNA was isolated using TRIzol and purified using the RNeasy Kit. RNA quality was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies). Next, 150 ng RNA was used as input for the NanoString nCounter assays using the nCounter Mouse Immunology Panel (NanoString Technologies). Data were generated as previously described (22). For NanoString data analysis, the obtained counts were log₂-transformed. Per-sample normalization was performed using geometric means from three housekeeping genes (Rpl19, Eef1g, and Gapdh). Differential translation (false discovery rate [FDR] <0.25) was identified using Analysis of Translational Activity (anota) algorithm (23, 24), which corrects changes in polysome-associated mRNA for changes in cytoplasmic total mRNA, applies variance shrinkage (the random variance model) and adjusts the p values for multiple testing using Benjamini and Hochberg’s FDR method. The translational activity in 4E-BP1/2 DKO cells (i.e., the intercepts from analysis of partial variance) were obtained for those mRNAs that are translationally upregulated and compared with WT control cells to obtain relative changes in translational efficiency.

5′ untranslated regions (UTRs) of transcripts from the top 11 gene hits were retrieved from the mm10 genome build using the University of California, Santa Cruz Table Browser (https://genome.ucsc.edu). Minimum free energy (MFE) and secondary structures were obtained from the foldUtr5 table, which contains MFE structures computed using RNAfold (25). Secondary structures were plotted using Visualization Applet for RNA, known as VARNA (26).

Pools of efficiently translated mRNAs (i.e., mRNAs associated with more than three ribosomes) and total cytoplasmic RNA were isolated using TRIzol (Invitrogen). RNA (1 μg) was reverse transcribed with Superscript III Reverse Transcriptase and Oligo(dT) (both from Invitrogen). Quantitative PCR was performed with PowerUp SYBR Green Master Mix (Applied Biosystems) according to the manufacturers' instructions using a QuantStudio 3 Real-Time PCR System (Applied Biosciences). Analysis was carried out by relative quantification using the comparative cycle threshold method (ΔΔCt) (27). Experiments were performed in independent biological replicates (n = 3), whereby each sample was analyzed in a technical triplicate. Relative mRNA expression was normalized to Gapdh and Rpl19. Primers were designed using National Center for Biotechnology Information Primer Basic Local Alignment Search Tool (http://www.ncbi.nlm.nih.gov/tools/primer-blast/). The list of primers is provided in Supplemental Table I.

BMDM were seeded in six-well plates (2 × 10⁶ cells per well). After stimulation, cells were scraped in cold PBS (pH 7.4), collected by centrifugation, and lysed in cold radioimmunoprecipitation assay buffer containing 25 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, supplemented with phosphatase and EDTA-free protease inhibitor mixtures (Roche). Cell debris was removed by centrifugation at 20,000 × g for 15 min at 4°C, and total protein content was determined using the Pierce BCA Protein Assay Kit (ThermoFisher Scientific). Whole-cell protein extracts were subjected to SDS-PAGE, and the separated proteins were transferred onto a polyvinylidene difluoride membrane (Bio-Rad Laboratories). Membranes were blocked for 1 h in 5% skim milk TBST (0.1% Tween 20) and incubated with specific primary Abs overnight at 4°C. Proteins were then detected with IgG HRP-linked Abs by chemiluminescence using Clarity Western ECL Substrate (Bio-Rad Laboratories). Abs detecting phospho–4E-BP1 (T37/46) (no. 2855), phospho–4E-BP1 (T70) (no. 9455), phospho–4E-BP1 (S65) (no. 9451), 4E-BP1 (no. 9452), 4E-BP2 (no. 2845), COX-2 (no. 4842), C/EBPβ (no. 3087), eIF4G (no. 2498), phospho-STAT3 (Y705) (no. 9145), STAT3 (no. 9139), and β-actin (no. 3700) were obtained from Cell Signaling Technology. The Ab detecting secreted IL-1R antagonist (sIL-1Ra) (no. MAB4801) was purchased from R&D Systems. The anti–NF IL-3–regulated (NFIL-3) Ab (no. 685402) was provided by BioLegend. The following secondary HRP-conjugated Abs were used in this study: anti-rabbit IgG (no. A0545) and anti-mouse IgG (no. A4416) from Sigma-Aldrich and anti-rat IgG (no. HAF005) from R&D Systems.

BMDM were plated in 10-cm-diameter plates (1.5 × 10⁷ cells per plate). Cell treatment was followed by lysis in cold Buffer A (lysis buffer; 50 mM MOPS [pH 7.4], 100 mM NaCl, 2 mM EDTA, 2 mM EGTA, 1% IGEPAL CA-630, 1% sodium deoxycholate, 7 mM 2-ME) supplemented with phosphatase and EDTA-free protease inhibitor mixtures. Samples were incubated for 15 min on ice and regularly mixed gently, and the crude lysates were cleared by centrifugation. About 0.5 mg of proteins of each sample were mixed with 50% slurry of 2'/3′-EDA-m⁷GTP immobilized on agarose beads (no. AC-142S; Jena Bioscience) and diluted up to 1 ml with Buffer B (wash buffer; 50 mM MOPS [pH 7.4], 100 mM NaCl, 0.5 mM EDTA, 0.5 mM EGTA, 7 mM 2-ME, 0.1 mM GTP) supplemented with phosphatase and EDTA-free protease inhibitor mixtures. Samples were mixed for 1 h at 4°C with end-over-end mixing, and beads were pelleted by centrifugation. The supernatants (i.e., flow through) were kept, whereas the beads were washed in Buffer B and finally resuspended in Laemmli loading buffer for further analysis by Western blotting.

Cells were seeded in 96-well plates (2 × 10⁵ cells per well) and stimulated with 1–10 ng/ml E. coli LPS for 6, 12, or 24 h. Cell culture supernatant samples were collected, and the concentrations of secreted IL-10 and PGE₂ were assessed by ELISA using the Mouse IL-10 ELISA MAX Deluxe kit (BioLegend) and the PGE₂ ELISA Kit (Cayman Chemical) according to the manufacturers' instructions.

BMDM were seeded in 6-cm-diameter nontreated plates (5 × 10⁶ cells per plate) and were stimulated with 10 ng/ml LPS for 4, 6, or 10 h. After cell fixation and staining, samples were acquired on the ImageStreamX Mark II Imaging Cytometer (Amnis), as described previously (28). Briefly, BMDM were collected by trypsinization (0.05% EDTA-Trypsin) and fixed in 1.5% paraformaldehyde. Cells were permeabilized in ice-cold methanol and were washed twice with staining buffer (1% BSA-PBS). Supernatants were discarded, and cells were incubated with an anti-CD16/32 Ab (no. 101302; BioLegend) for 15 min on ice. Cell staining was performed by a 25-min incubation on ice using the following fluorescent reagents and Abs: DAPI (Sigma-Aldrich), A488-coupled anti–Y705-STAT3 (no. 557814; BD Biosciences), A488-coupled isotype control (no. 558055; BD Biosciences), A647-coupled anti–β-actin (no. 8584; Cell Signaling Technology), and A647-coupled isotype control (no. 3452; Cell Signaling Technology). Next, cells were washed twice with staining buffer, and samples were acquired on the ImageStreamX Mark II Imaging Cytometer (Amnis). Bright-field and fluorescent images were collected at 40× magnification. Ten to thirty thousand cell singlets were gated from each sample. The analysis was performed using the IDEAS software (Amnis).

BMDM were seeded in 24-well plates (5 × 10⁵ cells per well) and treated with 10 ng/ml LPS with or without 2 μg/ml anti–IL-10 Ab and 50 μM NS-398 for 24 h before infection. Conditioned media were present throughout the entire experiment. Bacterial infection was carried out as previously described (29). Briefly, E. coli MG 1655 (nonpathogenic strain) was grown overnight at 37°C and then subcultured at a dilution of 1:4 in fresh Luria–Bertani broth without antibiotics to midlogarithmic phase (OD₆₀₀ = ∼0.1). BMDM were infected with 2.5 × 10⁶ bacteria (5:1 ratio) and centrifuged for 10 min at 500 × g to synchronize phagocytosis. BMDM were cultured in 1% FBS DMEM without antibiotics at 37°C for 2 h, and after three washes in PBS, they were treated with 100 μg/ml gentamicin for 1 h to eliminate extracellular bacteria. The infected cells were either lysed in 1% Triton X-100 to assess bacterial invasion (t = 0) or further incubated for a total of 6, 8, or 24 h in presence of 12 μg/ml gentamicin. When indicated, 10 ng/ml LPS with or without 2 μg/ml anti–IL-10 Ab and 50 μM NS-398 were added to cells 24 h before infection and were also present during the subsequent steps of infection. Surviving bacteria were determined as CFU by plating serial dilutions (1:10) of whole-cell lysates in Luria–Bertani agar.

nCounter data were analyzed using the anota R package to identify mRNAs under translational control between WT and 4E-BP1/2 DKO BMDM (23, 24). Statistical differences were calculated using two-way ANOVA embedded in the GraphPad Prism 7 software package. Results are presented as the mean ± SD of the mean. Differences were considered to be statistically significant when *p < 0.05, **p < 0.01, and ***p < 0.001.

mTORC1 orchestrates effective immune responses through a number of effectors, including the translational repressors 4E-BP1/2 (6, 18, 30). However, the impact of 4E-BP1/2 on selective translational control in macrophages and the effects on anti-inflammatory responses remain largely unexplored. To begin addressing this issue, cytosolic RNA from WT and 4E-BP1/2 DKO BMDM at steady-state was subjected to polysome profiling (19). Polysome profiling generated a pool of efficiently translated mRNA that was quantified in parallel with total cytosolic mRNA (input) using targeted nCounter assays (mouse immunology panel) (Fig. 1A). In this study, translational efficiency is defined as the proportion of the mRNA copies transcribed from a gene that are in heavy polysomes (in this case associated with more than three ribosomes) and hence are efficiently translated. Changes in polysome-associated mRNA levels can be due to altered translational efficiency or to changes in mRNA levels (e.g., via altered transcription or mRNA stability) (5). To identify mRNAs whose translation depends on 4E-BP1/2 in BMDM, we employed the anota algorithm, which specifically captures differences in translational efficiency of individual transcripts independent of changes in total mRNA levels (i.e., changes in the amount of mRNA copies associated with heavy polysomes after adjusting for changes in total cytosolic mRNA) (24). We thereby identified 11 mRNAs more efficiently translated in 4E-BP1/2 DKO as compared with WT BMDM (Supplemental Table II). Among the most regulated transcripts were Il-10 and Cox-2/Ptgs2, which encode proteins involved in macrophage anti-inflammatory responses. The graphical representation of anota analysis illustrates that Il-10 mRNA was more abundant in the pool of polysome-associated mRNA in 4E-BP1/2 DKO than in WT BMDM despite having similar total cytosolic mRNA levels (2.48-fold change in translational efficiency) (Fig. 1B). Similarly, translational efficiency of Cox-2 mRNA was upregulated in the absence of 4E-BP1/2 (4.20-fold change) (Fig. 1C). In Fig. 1B and 1C, each biological replicate is represented by an X, and the lines correspond to regressions used by anota to adjust changes in polysome-associated mRNA levels (y-axis) for changes in total cytosolic mRNA levels (x-axis). A difference in intercepts of the regression lines on the y-axis (i.e., when total cytosolic mRNA is set to zero) indicates changes in translational efficiency (when there is no change in translational efficiency, there is no difference in intercept). In summary, this analysis demonstrates that Il-10 and Cox-2 mRNAs are under the control of the translational repressors 4E-BP1/2 DKO in BMDM at steady-state.

Select translational control through the mTORC1–4E-BP1/2–eIF4E axis is associated with distinct features in the 5′ UTRs of mRNAs (5, 31, 32). To assess whether the identified immune-related transcripts under translational control contain such features, we conducted 5′ UTR analysis. Eight of these mRNAs (Il-1a, Il-1b, Il-10, Il-12b, Ccl5, Ccl12, Cd40, and Cxcl10) harbor relatively short 5′ UTRs (between 50 and 90 nt) with MFE ranging from −5 to −30 kcal/M (Fig. 1D, Supplemental Table III). In addition, we identified four mRNAs (Cox-2, Ifit2, Il-1b, and Mx-1) that contain longer 5′ UTRs (up to ∼1300 nt) with MFE as low as −487.3 kcal/M. In particular, several 5′ UTR sequences are annotated for Il-1b and Mx1 (three and five sequences, respectively), which differ in length and structure. Interestingly, the 5′ UTR of Il-10 is relatively short (67 nt and MFE = −12.3 kcal/M), whereas that of Cox-2 is longer and more structured (193 nt and MFE = −48.92 kcal/M) (Fig. 1E), indicating that these may be regulated by distinct mechanisms. Thus, selective 4E-BP1/2–dependent translational control of immune-related mRNAs in macrophages could at least partially be linked to the length and structure of their 5′ UTRs, as previously reported for transcripts that are highly sensitive to eIF4E levels and/or availability (5, 31, 32).

After having identified Il-10 and Cox-2 as targets of translational control downstream of 4E-BP1/2, we set out to elucidate the biological impact of such regulation in macrophage anti-inflammatory responses. The bacterial endotoxin LPS, a TLR4 ligand, regulates a large number of immune cell functions by controlling gene expression at the levels of transcription, mRNA stability, and mRNA translation (33). Activation of mTORC1 signaling is required for IL-10 production in LPS-stimulated macrophages (34). However, it remains unclear whether this increase is, at least in part, dependent on regulation of Il-10 translational efficiency via the mTORC1–4E-BP1/2 axis. To address this, WT and 4E-BP1/2 DKO BMDM were stimulated with 10 ng/ml E. coli LPS for 4 h, and cytoplasmic RNA was fractionated by polysome profiling (Fig. 2A). Quantitative RT-PCR analyses for Il-10 were conducted in total cytosolic and heavy polysome–associated RNA isolated from control (unstimulated) and LPS-treated cells. Consistent with selective modulation of Il-10 via changes in translational efficiency, the amount of heavy polysome–associated mRNA encoding Il-10 was higher in 4E-BP1/2 DKO than in WT control cells (2.8-fold increase) (Fig. 2B), whereas total cytosolic Il-10 mRNA, although slightly higher (1.5-fold change) did not fully explain this difference. Notably, we also observed greater accumulation of Il-10 mRNA associated with heavy polysomes in 4E-BP1/2 DKO than in WT cells after LPS treatment (7.6-fold change), which was only partly explained by changes in total cytosolic mRNA levels (3.6-fold change) and therefore is consistent with activated translation of Il-10 mRNA in 4E-BP1/2 DKO macrophages as compared with WT counterparts following LPS exposure (Fig. 2B). To further assess differences in translational efficiency of Il-10 between WT and 4E-BP1/2 DKO cells, we monitored Il-10 mRNA distribution in subpolysomal, light polysome, and heavy polysome fractions. Note that subpolysomal fractions contain mRNAs that are not efficiently translated (i.e., free mRNAs or associated with one ribosome). In light polysome fractions, mRNAs are associated with one to three ribosomes and therefore are not translated as efficiently as those found in heavy polysome fractions (mRNAs associated with more than three ribosomes). In control WT cells, 60.1% of Il-10 mRNA was found in subpolysomal fractions, and the remaining 39.9% was equally distributed in the light and heavy polysome fractions (19.9 and 20%, respectively) (Fig. 2C, left top panel). In contrast, only 29.5% of Il-10 mRNA was present in subpolysomal fractions of 4E-BP1/2 DKO BMDM (30.6% less in DKO than WT), which resulted in greater Il-10 mRNA amount in light and heavy polysome fractions (37.6 and 32.9%, respectively). Further supporting that LPS promotes Il-10 translation by inactivating 4E-BP1/2, we observed a significant shift in the distribution of Il-10 mRNA from subpolysomal to light polysome fractions of LPS-treated WT cells as compared with control (∼25.7% increase in Il-10 accumulation in light polysomes) (Fig. 2C, left versus right top panels, WT). Conversely, no apparent differences were detected between control and stimulated 4E-BP1/2 DKO BMDM (Fig. 2C, left versus right top panels, DKO). Distribution of Gapdh across polysome profiles indicated that, in contrast to Il-10, this mRNA is efficiently translated in both WT and 4E-BP1/2–deficient cells (Fig. 2C, bottom panels). These observations are in line with the notion that 4E-BP1/2 control translational efficiency of select transcripts. In keeping with greater Il-10 mRNA translational efficiency, we detected a significant upregulation of IL-10 protein secretion by 4E-BP1/2 DKO BMDM stimulated with LPS as compared with WT cells (3078 pg/ml versus 1762 pg/ml) (Fig. 2D). Consistent with the involvement of the mTORC1–4E-BP1/2 axis in the regulation of IL-10 production, active-site mTOR inhibitor PP242 reduced IL-10 induction by LPS in WT BMDM (35% reduction) but had no effect in 4E-BP1/2 DKO counterparts (Supplemental Fig. 1A). In support of the role of 4E-BP1/2 in modulating translation in response to LPS, a substantial increase in 4E-BP1/2 phosphorylation at T37/46, T70, and S65 (Fig. 2E) and a dramatic reduction in the interaction of 4E-BP1/2 with eIF4E (Fig. 2F) were detected in LPS-stimulated as compared with untreated BMDM. Accordingly, the amount of eIF4G bound to eIF4E augmented upon LPS exposure. Thus, LPS leads to 4E-BP1/2 inactivation and thereby promotes eIF4F complex formation in macrophages. Note that similar total cytosolic and heavy polysome–associated Tlr4 mRNA levels were found in WT and 4E-BP1/2 DKO BMDM (Supplemental Fig. 1B), ruling out the possibility that enhanced responses to LPS in 4E-BP1/2–deficient cells were caused by higher Tlr4 transcription and/or translation. Overall, these data provide evidence that in addition to activating Il-10 transcription, LPS promotes Il-10 mRNA translation in macrophages by dampening the activity of the inhibitory proteins 4E-BP1/2.

IL-10 induces expression of anti-inflammatory genes via the signal transducer and activator of transcription 3 (STAT3) (35, 36). Therefore, in the absence of 4E-BP1/2, enhanced translational efficiency of Il-10 and elevated IL-10 secretion may promote STAT3 activity. To test this hypothesis, WT and 4E-BP1/2 DKO BMDM were stimulated with LPS for various time periods, and the phosphorylation status of STAT3 was assessed by Western blotting. As previously reported (37), LPS treatment induced STAT3 phosphorylation at Y705 (Y705-STAT3) in WT BMDM. Notably, this response was markedly amplified in the absence of 4E-BP1/2 (Fig. 3A, top panel), reaching maximal differences between 4 and 6 h poststimulation (∼3-fold change DKO/WT) (Fig. 3A, bottom panel). In parallel, culture supernatants were collected from the same cells to quantify the amount of IL-10 by ELISA. As predicted, greater secretion of IL-10 in 4E-BP1/2 DKO cells followed similar kinetics to that of Y705-STAT3 phosphorylation (Supplemental Fig. 1C). Accordingly, when the activity of IL-10 was blocked using a neutralizing Ab (Supplemental Fig. 1D), Y705-STAT3 phosphorylation was abrogated in both WT and 4E-BP1/2 DKO cells (Fig. 3B). This set of experiments indicates that higher endogenous production of IL-10 is responsible for the increase in Y705-STAT3 phosphorylation in LPS-stimulated 4E-BP1/2 DKO as compared with WT BMDM. Further supporting this notion, WT and 4E-BP1/2 DKO BMDM expressed the same levels of total cytosolic and heavy polysome–associated Il-10r1 and Il-10r2 mRNA (Supplemental Fig. 1E), confirming that upregulated IL-10 signaling in 4E-BP1/2 DKO over WT cells is not caused by differential transcription and/or translation of the IL-10R.

Phosphorylation of STAT3 at Y705 is a requirement for its dimerization and translocation to the nucleus (37). In agreement with our Western blot data, time course experiments analyzed by image flow cytometry revealed that STAT3 nuclear translocation is enhanced in 4E-BP1/2 DKO BMDM. Differential nuclear levels of STAT3 were observed as early as 4 h following LPS stimulation and were sustained up to 10 h (Fig. 3C, 3D). Nuclear STAT3 was found in 75.7% of Y705-STAT3⁺ 4E-BP1/2 DKO BMDM at 4 h posttreatment (Fig. 3D, right panel). By contrast, nuclear translocation of STAT3 was detected in only 45% of Y705-STAT3⁺ WT cells. Notably, nuclear levels of STAT3 remained higher in 4E-BP1/2 DKO BMDM for a longer period of time, as evidenced by the presence of nuclear STAT3 in 29% Y705-STAT3⁺ 4E-BP1/2 DKO versus 15.5% WT cells at 10 h poststimulation. Altogether, these data provide evidence that elevated STAT3 activity in 4E-BP1/2 DKO cells is triggered by translational derepression of Il-10, which enhances the secretion and the autocrine effect of IL-10.

Activation of the transcription factor STAT3 is required for anti-inflammatory responses induced by IL-10 (38). Chromatin immunoprecipitation sequencing for STAT3 identified a large repertoire of anti-inflammatory factors that are controlled by STAT3 in IL-10–stimulated macrophages (36). Our data showing elevated STAT3 nuclear translocation in 4E-BP1/2 DKO BMDM prompted us to investigate whether the expression of IL-10–STAT3–responsive genes was altered in the absence of 4E-BP1/2. We focused on two genes that are transcriptionally controlled by LPS via IL-10–STAT3 signaling, Nfil3 and sIl1ra. NFIL-3 is a key component of a negative feedback loop that suppresses proinflammatory responses in myeloid cells by inhibiting Il-12b transcription (39). sIL-1Ra is a naturally occurring inhibitor of the proinflammatory action of IL-1 because it binds to the IL-1R with high affinity but lacks IL-1–like activity (40). In keeping with previous reports (37, 39), LPS upregulated Nfil3 and sIl1ra mRNA expression in WT BMDM. This effect was substantially enhanced in 4E-BP1/2 DKO over WT cells, as evidenced by a 3.2- and a 1.8-fold change in Nfil3 and sIl1ra mRNA levels, respectively (Fig. 4A, 4B). Accordingly, Western blot analyses revealed that NFIL-3 and sIL-1Ra protein levels are higher in 4E-BP1/2–deficient cells treated with LPS than in WT counterparts (∼4- and ∼2.5-fold change DKO/WT, respectively) (Fig. 4C). BP-1-102, a small-molecule inhibitor of STAT3 activation (41), repressed Y705-STAT3 phosphorylation in LPS-stimulated BMDM in a dose-dependent manner (Supplemental Fig. 1F). Consistent with Nfil3 transcription being regulated by STAT3, BP-1-102 reduced Nfil3 mRNA expression in WT and 4E-BP1/2 DKO BMDM (51 and 65%, respectively). Similarly, a decrease in sIl1ra mRNA levels was detected in BP-1-102–treated cells (34% in WT and 39% in DKO) (Fig. 4A). Specific blockade of IL-10 activity with a neutralizing Ab reduced the expression of Nfil3 and sIl1ra to the same extent in WT and 4E-BP1/2 DKO BMDM (∼40% decrease in Nfil3 and ∼30% in sIl1ra) (Fig. 4B). Note that induction of Nfil3 and sIl1ra was significantly downregulated but not completely abrogated by blocking either STAT3 or IL-10 activity in WT and 4E-BP1/2 DKO cells (Fig. 4A, 4B). Interestingly, augmented expression of Nfil3 and sIl1ra in 4E-BP1/2 DKO BMDM was markedly reduced by BP-1-102 or anti–IL-10 Ab but remained higher than in WT cells (Fig. 4A, 4B). Collectively, these results indicate that transcriptional activation of the anti-inflammatory genes encoding NFIL-3 and sIL-1Ra is reduced by the translational repressors 4E-BP1/2, mainly through the control of IL-10–STAT3–dependent signaling.

COX-2 is a rate-limiting enzyme in the production of PGE₂, a lipid mediator involved in numerous physiological and pathological processes, including inflammation (42). We found that translation efficiency of Cox-2 mRNA is amplified in 4E-BP1/2 DKO as compared with WT BMDM at steady-state (Fig. 1C). Therefore, we set out to investigate the impact of Cox-2 translational control by 4E-BP1/2 in macrophage anti-inflammatory responses. To address this, we treated WT and 4E-BP1/2 DKO BMDM with LPS, isolated heavy polysome–associated mRNA and total cytosolic mRNA, and quantified Cox-2 mRNA levels by RT-qPCR. In keeping with anota analysis of translational efficiency of Cox-2 mRNA (Fig. 1C), there was a substantial increase in the amount of efficiently translated Cox-2 mRNA in 4E-BP1/2 DKO cells as compared with WT at steady-state (7.4-fold change) without any detectable change in total cytosolic mRNA (1.03-fold change) (Fig. 5A). The relative amount of Cox-2 mRNA associated with heavy polysomes was also augmented after LPS stimulation in 4E-BP1/2 DKO as compared with WT BMDM (3.10-fold change), without changes in total cytosolic mRNA levels (1.15-fold change) (Fig. 5A). Confirming and extending these data, mRNA fractionation and quantification across polysome profiles showed a 30.3% reduction in Cox-2 mRNA isolated from subpolysomal fractions of 4E-BP1/2 DKO BMDM as compared with WT cells, which resulted in higher accumulation of Cox-2 mRNA in light and heavy polysomal fractions (37.7 and 32% in DKO versus 20 and 19.5% in WT) (Fig. 5B, top panel). Notably, LPS treatment led to a remarkable shift in the distribution of Cox-2 mRNA from subpolysomal to light polysome fractions in WT BMDM, as indicated by an 18.8% reduction in subpolysomal Cox-2 mRNA concomitant with a 24.4% increase in light polysome fractions (Fig. 5B, left versus right top panels, WT). By contrast, no significant differences were detected in the distribution of Cox-2 mRNA across polysome profiles of LPS-treated versus control 4E-BP1/2 DKO BMDM (Fig. 5B, left versus right top panels, DKO). These results provide evidence that LPS stimulates Cox-2 mRNA translation through the inactivation of the repressors 4E-BP1/2. Accordingly, a more rapid kinetics and a greater induction of COX-2 protein expression were detected in 4E-BP1/2 DKO than in WT BMDM after LPS stimulation (∼13-fold change at 4 h and ∼6-fold change up to 8 h posttreatment) (Fig. 5C). Notably, PP242 blocked LPS-inducible COX-2 expression in WT BMDM, whereas it exerted only a mild effect in 4E-BP1/2 DKO counterparts (Supplemental Fig. 1G), a clear indicator that mTORC1-mediated 4E-BP1/2 inactivation contributes to the regulation of COX-2 production. In keeping with higher COX-2 levels, PGE₂ synthesis in response to LPS was amplified in absence of 4E-BP1/2, as evidenced by a 1.74-fold change in the accumulation of PGE₂ in cell culture supernatants of 4E-BP1/2 DKO BMDM as compared with WT (Fig. 5D). Collectively, these results support the notion that translational activity of Cox-2 mRNA and subsequent PGE₂ induction are controlled by the translational repressors 4E-BP1/2.

PGs are autocrine and paracrine lipid mediators that maintain local homeostasis. Endogenous PGE₂ production potentiates macrophage anti-inflammatory responses via activation of C/EBPβ signaling (42). Our data indicated that induction of PGE₂ synthesis in LPS-stimulated cells is augmented in the absence of 4E-BP1/2. Thus, we sought to determine whether enhanced PGE₂ secretion amplified its autocrine effect and potentiated C/EBPβ activity in 4E-BP1/2 DKO BMDM. RT-qPCR experiments showed that C/ebpb mRNA expression was substantially upregulated in 4E-BP1/2 DKO BMDM after LPS stimulation (1.56-fold change over WT) (Fig. 6A). Blockade of COX-2 activity with the selective NS-398 inhibitor (43) downregulated but did not abolish the expression of C/ebpb mRNA in WT and 4E-BP1/2 DKO BMDM (36 and 66%, respectively) (Fig. 6B). Thus, transcriptional activation of C/ebpb appears to require COX-2–PGE₂–dependent and –independent signals in macrophages. Remarkably, NS-389 reduced C/ebpb mRNA expression in 4E-BP1/2 DKO cells to WT levels (Fig. 6B). In keeping with the kinetics of C/ebpb accumulation, Western blot analyses revealed that C/EBPβ protein levels augmented more rapidly and to a greater extent in LPS-treated 4E-BP1/2 DKO than WT BMDM (∼2.5-fold change DKO/WT at 4–6 h) (Fig. 6C). As expected, C/EBPβ induction by LPS was reduced when cells were treated with NS-398 (Supplemental Fig. 1H). These data support the notion that enhanced PGE₂ secretion and autocrine action potentiates C/ebpb transcription and protein expression in 4E-BP1/2 DKO BMDM.

C/EBPβ is required for transcriptional activation of anti-inflammatory genes in macrophages (42, 44). We predicted that amplified PGE₂–C/EBPβ signaling would promote this cellular response in 4E-BP1/2 DKO BMDM. We focused on Arg1 and SerpinB2 because these genes are transcriptionally activated via C/EBPβ in LPS-stimulated macrophages and have central roles in local homeostasis (45, 46). Arginase-1 functions as an inhibitor of chronic inflammation in Th2-polarized immune responses (47), and SerpinB2 suppresses Th1 responses during inflammatory processes (48). At first, we stimulated WT and 4E-BP1/2 DKO BMDM with LPS over a 24-h period and monitored the expression of Arg1 and SerpinB2 by RT-qPCR. A dramatic increase in Arg1 mRNA levels was detected in 4E-BP1/2 DKO over WT cells at 12 h (4-fold change) and 24 h (10.5-fold change) posttreatment (Fig. 6D). Accumulation of SerpinB2 mRNA was also markedly amplified in absence of 4E-BP1/2. Significant differences were observed as early as 6 h after LPS stimulation (1.9-fold change DKO/WT), were maximal at 8 h (2.48-fold change), and remained detectable up to 24 h (Fig. 6E). Consistent with the requirement of COX-2 activity for Arg1 transcription, NS-398 prevented accumulation of Arg1 mRNA in WT and 4E-BP1/2 DKO BMDM (∼91% decrease) (Fig. 6F). Similarly, SerpinB2 mRNA levels were drastically downregulated in NS-398–treated WT and DKO cells (∼76% decrease) (Fig. 6G). Notably, greater expression of Arg1 and SerpinB2 mRNA in 4E-BP1/2 DKO BMDM was reduced to WT levels by NS-398 treatment (Fig. 6F, 6G). Collectively, this set of experiments provides evidence that in the absence of 4E-BP1/2, translational derepression of Cox-2 mRNA potentiates synthesis and autocrine action of endogenous PGE₂, which in turn promotes C/EBPβ-mediated transcriptional activation of anti-inflammatory genes.

Macrophage anti-inflammatory responses are essential for the resolution of inflammation and local tissue repair after elimination of invading pathogens; however, when dysregulated, host susceptibility to infection can emerge (11). IL-10–STAT3 and PGE₂–C/EBPβ signaling augment the expression of anti-inflammatory genes (36, 44, 45). In addition, IL-10 and PGE₂ exert a suppressive effect on proinflammatory and microbicidal mediator production (17, 49–51). Consistent with elevated levels of IL-10 and PGE₂ in 4E-BP1/2 DKO BMDM, transcriptional activation of several anti-inflammatory genes was enhanced in these cells. Conversely, LPS-inducible accumulation of proinflammatory transcripts Tnf, Il-6, and Nos2 declined more rapidly and to a greater extent in 4E-BP1/2 DKO than in WT BMDM (Fig. 7). Collectively, these data suggested that in the absence of 4E-BP1/2, the bactericidal capacity of LPS-stimulated macrophages could be diminished. To test this hypothesis, WT and 4E-BP1/2 DKO BMDM were treated or not with 10 ng/ml LPS for 24 h and subsequently infected with a nonpathogenic strain of E. coli (MG 1655). Unstimulated WT and 4E-BP1/2 DKO BMDM were able to control the infection. In stark contrast, bacteria survival increased in LPS-treated 4E-BP1/2 DKO over WT cells at 8 h postinfection (2.1-fold change) and remained augmented up to 24 h (3.7-fold change) (Fig. 8A). Bacterial numbers were equivalent in WT and 4E-BP1/2 DKO cells at 6 h postinfection, confirming that differential bacterial survival observed at later time points was not due to changes in phagocytic activity in response to TLR4 stimulation. Notably, simultaneous blockade of IL-10 and PGE₂ activity with a neutralizing anti–IL-10 Ab and a specific inhibitor of COX-2 restored bacteria killing by ∼82% in 4E-BP1/2 DKO BMDM, reaching similar levels to WT counterparts (Fig. 8B). By contrast, the same treatment had no significant effect in bactericidal activity of WT BMDM. These data support the notion that excess IL-10– and PGE₂-mediated anti-inflammatory responses hamper the bactericidal potential of LPS-stimulated 4E-BP1/2 DKO BMDM. In conclusion, our study uncovered a crucial role for 4E-BP1/2 in macrophage homeostasis by limiting the anti-inflammatory action of IL-10–STAT3 and PGE₂–C/EBPβ signaling (Fig. 9).

The mTORC1 downstream effectors 4E-BP1/2 play a crucial role in the regulation of proinflammatory mediators during viral and parasitic infections (6, 18, 19). Surprisingly, there are considerable gaps regarding the impact of 4E-BP1/2 in the control of anti-inflammatory responses. We identified the mTORC1–4E-BP1/2–eIF4E axis as a central regulator of macrophage homeostasis. In the absence of 4E-BP1/2, translational derepression of Il-10 and Cox-2 mRNAs triggered a transcriptional program that amplified anti-inflammatory gene expression and promoted bacteria survival in macrophages upon TLR4 stimulation. These data provide evidence that 4E-BP1/2–dependent translational control of select mRNAs contributes to modify gene expression networks that orchestrate macrophage responses.

Among 564 immune-related transcripts that were screened in this study, only 11 were identified as targets of translational control via 4E-BP1/2 in macrophages. These data are in line with numerous reports, including comparative analyses of the translatomes of WT and 4E-BP1/2 DKO mice and cells (6, 52), showing that 4E-BPs do not act as general translational repressors but rather target specific subsets of mRNAs (21, 53–59). Despite the fact that eIF4E is required for cap-dependent translation of all nuclear-encoded mRNAs, some of them are particularly sensitive to eIF4E levels and/or availability and therefore are referred to as “eIF4E-sensitive” (5, 31, 32). Accordingly, their translational efficiency is repressed by genetic deletion of eIF4E (32) or increased cellular 4E-BP-to-eIF4E ratios (59). Two major subsets of transcripts have been identified as being targets of select translational control via eIF4E-dependent mechanisms. The first one comprises mRNAs known as eIF4A-sensitive (5, 31, 32), which harbor long and highly structured 5′ UTRs and therefore depend on the RNA helicase activity of eIF4A to be efficiently translated (32, 60–62). eIF4E sensitivity of these mRNAs is thought to stem from eIF4E-dependent recruitment of eIF4A to the eIF4F complex and stimulation of eIF4A activity (5, 63). Consistent with central roles for 4E-BP1/2 as regulators of eIF4A-sensitive mRNAs, pharmacological inhibition of mTORC1 signaling represses translational efficiency of such transcripts in WT cells but not in 4E-BP1/2 DKO counterparts (21, 53). Accumulating evidence indicates that eIF4A-sensitive mRNAs encode proteins related to specific cellular processes, namely cell proliferation and survival (21, 32, 53) but also immune responses (6, 32). Indeed, Irf7, which encodes a central regulator of antiviral immunity (10), was identified as an eIF4A-sensitive mRNA (32). Accordingly, Irf7 mRNA harbors a long and highly structured 5′ UTR, and its translational efficiency is tightly controlled through 4E-BP1/2 (6). Note that Irf7 was not identified in our screening, which might be explained by differences in cell types employed (MEFs versus BMDM) and treatment (serum stimulated versus steady-state). Indeed, selective changes in mRNA translation via mTORC1-mediated inactivation of 4E-BP1/2 appear to be dependent on the nature of the stimulus or type of stress to which the cell is exposed (5, 31, 57). In support of the notion that 4E-BP1/2 limit translational efficiency of immune-related transcripts containing long and structured 5′ UTRs, our bioinformatic analysis indicated that Cox-2, Ifit2, Il-1b, and Mx-1 might fall into this category. Further investigation is required to determine whether selective translational control of these mRNAs via 4E-BP1/2 is dictated by their sensitivity to eIF4A activity.

A second subset of eIF4E-sensitive mRNAs is characterized by very short 5′ UTRs (i.e., 30 nt) that in some cases contain a translation initiator of short 5′ UTR element, also known as TISU element (32, 54, 56, 64). As such, their translational efficiency is less sensitive to eIF4A (32), and they are referred to as “eIF4A-insensitive” mRNAs (5, 31, 32). These transcripts encode proteins involved in mitochondrial-related functions (64), and their translational efficiency is limited by 4E-BP1/2 (54, 56). We identified several immune-related mRNAs (Il-1a, Il-1b, Il-10, Il-12b, Ccl5, Ccl12, Cd40, and Cxcl10) that harbor relatively short 5′ UTRs; however, none of them contains a translation initiator of short 5′ UTR element, and with the exception of Il-1a, their length exceeds 30 nt. Thus, they do not seem to meet the criteria to be considered as short 5′ UTR–containing eIF4E-sensitive transcripts. It is conceivable that these and other immune-related transcripts are subject to an alternative mechanism of 4E-BP1/2–dependent translational control, which might involve yet-to-be-discovered sequence and/or structural features in their 5′ UTRs. Alternatively, these mRNAs might be expressed in macrophages with 5′ UTRs differing from those reported in databases. Further investigation is needed to resolve these possibilities.

An additional explanation that could account for the small amount of selective translational control of immune-related transcripts through 4E-BP1/2 is that our comparative analysis was performed in macrophages at steady-state. Because we employed an immunology panel, it seems reasonable that changes in translational activity for many of these transcripts cannot be detected without cell activation. Further support for this hypothesis was obtained from our data showing that TLR4 stimulation was required to detect IL-10 and COX-2 at the protein level despite the fact that Il-10 and Cox-2 mRNAs were translated more efficiently in 4E-BP1/2 DKO than in WT BMDM at steady-state. These observations are in line with a study showing that even though ll-10 is transcribed in macrophages at basal level, IL-10 secretion can only be detected in stimulated cells. The authors presented experimental evidence indicating that macrophages are poised to secrete IL-10 and will do so if they receive appropriate signals (65). Thus, our data along with this previous report suggest that in the absence of 4E-BP1/2, select mRNAs might be primed to be more efficiently translated in response to triggers or cues, a phenomenon recently described in NK cells (66).

Consistent with the central role of IL-10 and COX-2 in anti-inflammatory responses (35, 36, 42, 44), their expression is tightly regulated through transcriptional and posttranscriptional mechanisms, including RNA stability and translational control via RNA-binding proteins, such as T cell intracellular Ag 1 (TIA-1) and tristetraprolin (reviewed in Refs. 67, 68). For example, the adenosine A2BR activates translation of Il-10 mRNA in macrophages by relieving the translational repressive effect of RNA-binding protein elements in its 3′ UTR (69). In regards to COX-2, different mechanisms of translational control have been reported. The translational silencer and RNA-binding protein TIA-1 represses Cox-2 mRNA translation (70). Moreover, COX-2 protein synthesis was shown to be dependent on mTORC1 signaling in neutrophils (71); however, the underlying mechanism remained unclear. Confirming and extending previous reports, we found that the inhibitors of cap-dependent translation, 4E-BP1/2, limit Il-10 and Cox-2 mRNA translation efficiency in macrophages. Because the activity of 4E-BP1/2 is altered by a number of pathogens (reviewed in Ref. 72), it is plausible that dysregulated translational control of Il-10 and Cox-2 may contribute to the pathogenesis of infections by skewing anti-inflammatory responses in macrophages.

Our results indicate that translational control of mRNAs encoding select immunomodulatory factors, such as IL-10 and COX-2, is required for fine tuning of macrophage responses to the bacterial toxin and TLR4 ligand LPS. In keeping with this reasoning, several reports showed that LPS-inducible expression of activators and suppressors of inflammation is, at least in part, controlled at the level of mRNA translation in macrophages. For instance, TLR4 stimulation with LPS activates mRNA translation of several proinflammatory mediators, including the transcription factor IRF-8 (14) and the TGF-activated kinase (15). In stark contrast, a previous study showed that LPS promotes translation of macrophage mRNAs encoding negative feedback regulators of the inflammatory response, such as inhibitors of NF-κB (e.g., IER3, NFKBID) and RNA-binding proteins that prevent the expression of cytokines at the posttranscriptional level (e.g., tristetraprolin) (16). These reports, along with our data, suggest that microbial components, such as LPS, trigger antagonistic translational control programs during infection (i.e., pro- and anti-inflammatory), which might contribute to pathogen clearance while helping to maintain macrophage homeostasis.

We found that 4E-BP1/2–dependent mTORC1 signaling is necessary to control Il-10 and Cox-2 mRNA translation and subsequent IL-10 and COX-2 production. These data are in agreement with previous studies that have linked translational control via eIF4E availability or activity to changes in translation of mRNAs encoding regulators of inflammation in macrophages. Indeed, IFN-γ enhances TLR2-stimulated M1 macrophage activation by suppressing mRNA translation of the transcriptional repressor HES-1 via MNK1/2 and mTORC1 inhibition (13). Similarly, IL-10 was shown to disrupt MNK signaling and thereby repress mRNA translation of the proinflammatory cytokine TNF (17). Conversely, LPS was found to activate the MNK pathway and induce protein synthesis of IRF-8. Notably, MNK-dependent regulation of IRF-8 promoted proinflammatory gene expression and M1 macrophage polarization (14). In view of these studies and our current findings, selective translational control through eIF4E-dependent mechanisms appears to regulate transcriptional programs that coordinate the onset and the resolution of inflammatory responses in macrophages.

Macrophages deficient in 4E-BP1/2 displayed a defect in their bactericidal capacity. We postulate that this phenotype is associated with translational derepression of Il-10 and Cox-2 and the amplified autocrine action of endogenous IL-10 and PGE₂ produced in response to LPS. Further supporting our model, the anti-inflammatory and immunosuppressive effects of IL-10 and PGE₂ are well documented and have been linked to their ability to inversely regulate anti- and proinflammatory gene expression (17, 36, 37, 42). Importantly, IL-10 and PGE₂ are negative regulators of LPS-mediated inflammatory responses (49, 51). However, we cannot rule out the possibility that in addition to IL-10 and COX-2, 4E-BP1/2 control other immunomodulatory factors that impact anti-inflammatory responses and bacterial survival in macrophages. In contrast to our observations, 4E-BP1/2 DKO MEFs were resistant to viral infections (6), and 4E-BP1/2 DKO peritoneal macrophages were less susceptible to a protozoan parasite (19). This discrepancy might be related to distinct translational programs triggered by specific stimuli or stressors in different cell types. Further characterization of the molecular mechanisms of 4E-BP1/2–dependent translational control during infections will shed light on this matter.

Collectively, this work provides evidence that the mTORC1–4E-BP1/2 axis orchestrates translational and thereby transcriptional programs that limit anti-inflammatory responses in macrophages. Notably, our data suggest that dysregulated activity of 4E-BP1/2 during pathological conditions, such as infections and cancer, might contribute to reprogram the translational and transcriptional landscape of macrophages and thereby favor disease progression. Targeted sequencing and transcriptome-wide analyses of the translatome of pathologic condition–associated macrophages will generate a more complete repertoire of the mRNAs that are translationally controlled through 4E-BP1/2–dependent mechanisms and will provide insight on the regulation of gene expression networks in health and disease.

We thank Dr. Nahum Sonenberg for providing the bone marrow of Eif4ebp1^−/−/Eif4ebp2^−/− mice, Annie Sylvestre and Annik Lafrance for invaluable technical assistance, Dr. Jennifer Raisch and Sebastien Houle for useful advice on bacterial infections, Dr. Medhi Jafarnejad for technical advice on m⁷GTP-agarose pull-down assays, and Jessie Tremblay for assistance with FACS experiments and data analysis.

The authors have no financial conflicts of interest.