|Home | About | Journals | Submit | Contact Us | Français|
Monocytes serve as a central defense system against infection and injury but can also promote pathological inflammatory responses. Considering the evidence that monocytes exist in at least two subsets committed to divergent functions, we investigated whether distinct factors regulate the balance between monocyte subset responses in vivo. We identified a microRNA (miRNA), miR-146a, which is differentially regulated both in mouse (Ly-6Chi/Ly-6Clo) and human (CD14hi/CD14loCD16+) monocyte subsets. The single miRNA controlled the amplitude of the Ly-6Chi monocyte response during inflammatory challenge whereas it did not affect Ly-6Clo cells. miR-146a–mediated regulation was cell-intrinsic and depended on Relb, a member of the non-canonical NF-κB/Rel family, which we identified as a direct miR-146a target. These observations not only provide novel mechanistic insights into the molecular events that regulate responses mediated by committed monocyte precursor populations but also identify novel targets to manipulate Ly-6Chi monocyte responses while sparing Ly-6Clo monocyte activity.
Monocytes are the circulating precursors of several types of macrophages and dendritic cells (Geissmann et al., 2010). They confer protection of injured or infected tissue but also propagate chronic diseases (Auffray et al., 2009; Qian and Pollard, 2010; Shi and Pamer, 2011). At least two CD11b+ CD115+ monocyte populations exist in mice: (i) Ly-6Chi (Gr-1+ CCR2+ CX3CR1lo) cells respond to pro-inflammatory cues such as CCL2 (or MCP-1), migrate to inflamed sites and draining lymph nodes and can differentiate into antigen-presenting dendritic cells (Cheong et al., 2010) and orchestrate inflammatory functions (Swirski et al., 2007; Tacke et al., 2007); and (ii) Ly-6Clo (Gr-1− CCR2−CX3CR1hi) cells patrol the resting endothelium (Auffray et al., 2007), can be recruited to tissue after the onset of inflammation, and participate in granulation tissue formation (Nahrendorf et al., 2007). Ly-6Chi monocytes recirculate into the bone marrow where they can convert into Ly-6Clo monocytes (Varol et al., 2007). Monocyte heterogeneity is conserved at least in part in mice and humans: mouse Ly-6Chi monocytes share phenotypic and functional features with human CD14hi cells, whereas mouse Ly-6Clo monocytes resemble human CD14lo CD16+ cells (Cros et al., 2010).
Infection (Shi and Pamer, 2011), injury (Nahrendorf et al., 2007), atherosclerosis (Swirski et al., 2007; Tacke et al., 2007); cancer (Movahedi et al., 2010) and other pathophysiological conditions alter monocyte subset ratios. Changes of ratios can occur rapidly (e.g., hours after pathogenic infection), be long lasting (e.g. in chronic inflammatory disorders), and typically result in the selective amplification of pro-inflammatory Ly-6Chi cells. Human studies have underscored the relevance of studying monocyte subsets because an imbalance in their relative proportion is linked to several diseases (Ziegler-Heitbrock, 2007). The factors that regulate the balance between monocyte subset responses are largely unknown. The identification of such factors is potentially useful as it may offer new vantage points for tailoring immune responses to a desired phenotype.
MicroRNAs (miRNAs) regulate target genes at the post-transcriptional level and can control distinct functional properties in cell types that are closely related ontogenically. miRNAs are known to regulate the development and function of various immune cell types (O'connell et al., 2010) but have to date not been investigated in the context of monocyte heterogeneity. Here we compared the expression levels of 380 miRNAs in sorted monocyte subsets (Fig S1a) and defined significant genes as those with at least 2-fold differential expression and a p<0.05 (Student’s t-test). The approach identified 9 miRNAs, which were highly expressed either in Ly-6Chi (miR-20b,-135a,-424,-702) or in Ly-6Clo monocytes (miR-146a, -150, -155, -342, -29b) (Fig 1a).
Independent assays indicated ~2 orders of magnitude higher expression of miR-146a in Ly-6Clo monocytes when compared to hematopoietic stem cells (HSC), granulocyte/macrophage progenitors (GMP), macrophage/dendritic cell progenitors (MDP) and Ly-6Chi monocytes in steady-state (Fig 1b). Thus, monocytes express miR-146a only at a late maturation stage and selectively in the Ly-6Clo subset. Steady-state dendritic cell populations expressed miR-146a at intermediate levels (Fig 1b).
miRNAs and their respective target genes are often mutually exclusively expressed in a given tissue (Farh, 2005). In keeping with previous observations that miR-146a suppresses NF-κB-dependent inflammatory pathways (Taganov et al., 2006), we confirmed with two independent genome-wide profiling methods that splenic miR-146alo Ly-6Chi monocytes showed increased inflammatory signatures (Swirski et al., 2009) and expressed components of the NF-κB signaling cascade at higher levels than their miR-146ahi Ly-6Clo counterparts (Fig 1c, S1b). Also, splenic and blood miR-146alo Ly-6Chi monocytes stimulated with lipopolysaccharide (LPS) produced more TNFα, IL-6 and IL-1b inflammatory cytokines than miR-146ahi Ly-6Clo cells (Fig 1d and S1c).
The elevated miR-146a expression in Ly-6Clo cells and the inflammatory profile of Ly-6Chi cells reported above were likely not due to a premature activation artifact induced by the isolation procedure because IκBα protein levels were similar in both monocyte subsets ex vivo (Fig S1d) and NF-κB subunit p65 only became detectable in the nucleus of Ly-6Chi cells upon in vitro challenge (Fig 1e). The cause for constitutive (NF-κB-independent) miR-146a expression in Ly-6Clo cells will require additional investigation.
We addressed the regulation of miR-146a expression in monocyte subsets upon ex vivo challenge with either LPS, heat killed Listeria monocytogenes (HKLM) or TNFα. miR-146a was induced only in Ly-6Chi monocytes, in response to all stimuli, and reached levels matching those in Ly-6Clo cells (Fig S1e). In vivo LPS challenge studies confirmed the in vitro findings (Fig S1f). miR-146a expression in Ly-6Chi cells increased within 4 h after LPS challenge and reached levels equivalent to those found in Ly-6Clo cells after 16 h (Fig 1f). Thus, miR-146a expression is constitutive in Ly-6Clo monocytes and inducible in Ly-6Chi monocytes. LPS-stimulated Ly-6Chi monocytes were CD11c+ MHC IIhigh (Ly-6Chi) and thus distinct from Ly-6Clo monocytes (Fig S1g).
To investigate the role of miR-146a in monocytes in vivo, we generated mice in which miR-146a expression was either up- or down-regulated experimentally. To constitutively over-express miR-146a we reconstituted mice with HSC transduced to co-express EGFP and miR-146a (Fig S2a–b). miR-146a overexpression did not alter monocyte numbers or subset ratios in steady-state (Fig S2c); however, upon Listeria monocytogenes (Lm) infection (Shi and Pamer, 2011), it prevented the unfolding of a full-fledged TNFα–producing Ly-6Chi monocyte response (Fig 2a,b).
To suppress miR-146a expression in vivo we used two independent approaches. The first one involved systemic delivery of anti-miRNA locked nucleic acid (LNA) formulations (Fig S2d,e). LNA treatment did not alter monocyte subset ratios in steady-state (Fig S2f) but it increased the number of TNFα–producing Ly-6Chi monocytes at Lm infected sites (Fig 2c,d).
The second approach to suppress miR-146a expression used recently described mice with targeted deletion of the miR-146a gene (Boldin et al., 2011) (Fig S2g). miR-146a−/− mice contained both monocyte subsets thus Ly-6Chi→Ly-6Clo monocyte conversion should not require miR-146a. Also, miR-146a knockdown did neither alter the ratio (Fig 2e) nor the phenotype (Fig S2h) of monocyte subsets in 8 wk old mice. To compare miR-146a−/− and wild-type monocyte responses as they developed in the same environments, we reconstituted wild-type (CD45.1) mice with equal numbers of miR-146a−/− (CD45.2) and wild-type (EGFP+ CD45.2) cells (Fig S2i). The absence of miR-146a strongly amplified TNFα–producing Ly-6Chi peritoneal monocytes in response to LPS challenge (Fig 2f, g). Ly-6Chi monocytes mediate immune defense in early phase of Lm infection (Shi and Pamer, 2011). Accordingly, Lm-infected miR-146a−/− mice contained reduced numbers of viable Lm 24 h post infection when compared to Lm-infected wild-type mice (Fig 2h). Amplification of the Ly-6Chi monocyte response in absence of miR-146a was confirmed in a model of sterile peritonitis induced by thioglycollate (Fig 2i).
The experiments above involved indiscriminate alteration of miR-146a expression in all hematopoietic cells. We reasoned that injection of miR-146a−/− GMP into wild-type mice would permit to track miR-146a−/− monocytes in a wild-type environment because miR-146a is only upregulated upon progenitor cell maturation. Specifically, we co-administered equal numbers of miR-146a−/− (CD45.2 EGFP−) and wild-type (CD45.2 EGFP+) GMP into non-irradiated wild-type (CD45.1) mice, which were subsequently challenged with LPS i.p. (Fig S2j). Wild-type and miR-146a−/− hematopoietic progenitor cells show comparable clonogenic potential (Boldin et al., 2011; Fig S2k) and the transferred cells’ progeny contained monocytes and neutrophils, as expected. miR-146a−/− monocytes recruited to the peritoneal cavity outnumbered their wild-type counterparts (Fig 2j, k) and were Ly-6Chi (Fig 2l); in marked contrast, miR-146a−/− neutrophils—which do not upregulate miR-146a in vivo—mounted a response that was similar to their wild-type counterparts (Fig 2k). Thus miR-146a should regulate Ly-6Chi monocytes at least in part in a cell-intrinsic manner.
In contrast to previous descriptions for other cell types (Nahid et al., 2009; Boldin et al., 2011), including macrophages (Fig S3a), the absence of miR-146a did not detectably alter inflammatory cytokine production by Ly-6Chi and Ly-6Clo monocytes on a per-cell basis (Fig 3a,b). However, LPS challenge increased the percentage of miR-146a−/− Ly-6Chi monocytes undergoing cell division in bone marrow (Fig 3c, d) and to a lower extent in the spleen and peritoneal cavity (Fig 3d). The absence of miR-146a did not affect proliferation of Ly-6Clo monocytes; Fig S3b. Co-cultures of miR-146a−/− and wild-type cells also indicated a proliferative advantage for bone marrow miR-146a−/− Ly-6Chi monocytes (Fig S3c, d).
In addition, co-injection of bone marrow miR-146a−/− (EGFP− CD45.2) and control (EGFP+ CD45.2) Ly-6Chi monocytes into LPS-treated wild-type (CD45.1) mice showed higher accumulation of miR-146a−/− cells at the site of inflammation within only 6 h (Fig 3e). The chemokine CCL2 controls Ly-6Chi monocyte migration to inflamed sites (Shi and Pamer, 2011). Interestingly, miR-146a−/− blood Ly-6Chi—but not Ly-6Clo—monocytes expressed the cognate receptor CCR2 at higher levels than their wild-type counterparts (Fig 3f, S3e) and migrated more efficiently toward a CCL2 gradient in vitro (Fig 3g).
These observations indicate that miR-146a controls the expansion of Ly-6Chi monocytes during acute inflammatory conditions in part through elevated proliferation of Ly-6Chi monocytes—predominantly in the bone marrow—and increased trafficking to inflamed sites.
We aimed to find endogenous miR-146a target genes that contribute to altering the monocyte response. The screening approach, which compared the expression profiles of miR-146a–predicted target genes in Ly-6Chi and Ly-6Clo monocytes either at 2 h or 8 h after Lm challenge (Fig S4a and supplementary information), identified the transcription factor Relb (Fig 4a). Experimental evidence also indicates that Relb is a miR-146a target. First, ectopic miR-146a expression in resting Ly-6Chi monocytes in vivo reduced Relb transcript levels (Fig 4b). Second, NIH-3T3 cells transfected with a luciferase reporter plasmid expressing Relb 3’ UTR (ENSMUST00000049912) containing a potential miR-146a binding sequence showed reduced luciferase activity upon miR-146a overexpression. The phenotype was rescued by mutating the seed sequence (Fig 4c). Third, immunofluorescence microscopy with a validated anti-Relb Ab (Fig S4b) showed efficient nuclear translocation of Relb protein at 30 min after LPS challenge in both wild-type and miR-146a−/− Ly-6Chi monocytes; however at 6 h cytoplasmic Relb levels were recovered more prominently in the miR-146a−/− cells (Fig 4d). Fourth, flow cytometry analysis confirmed that Relb protein levels remained higher in miR-146a−/− Ly-6Chi monocytes upon LPS challenge (Fig 4e).
To investigate whether modulation of Relb affects the monocyte response, we generated both Relbhi EGFPhi CD45.1 HSC (which expressed Relb from a cDNA sequence that could not be regulated by miR-146a) and control Relbnorm EGFPhi CD45.2 HSC, which were adoptively transferred at a 1:1 ratio into LPS-treated CD45.1/2 recipient animals (Fig S4c). Relb overexpression did not alter HSC expansion (Fig S4d) but amplified the monocyte response in vivo (Fig 4f) and thus recapitulated the phenotype observed for miR-146a−/− Ly-6Chi monocytes.
We also injected LPS-treated CD45.1 mice either with miR-146a−/− shRelb EGFPhi HSC (which expressed a miR30-hairpin based shRNA to silence Relb to the levels found in challenged wild-type monocytes) or with miR-146a−/− EGFPhi HSC (which expressed a control EGFP vector) (Fig 4g, S4e). Relb silencing did not alter HSC expansion (Fig S4f) but decreased miR-146a−/− Ly-6Chi monocyte responses in vivo (Fig 4h). These data indicate that miR-146a can control Ly-6Chi monocyte fate in response to acute inflammatory challenge via Relb targeting.
The human Relb 3’UTR contains a binding site for the alternative processing isoform miR-146a-3p (miR-146a*) instead of the “canonical” miR-146a-5p isoform (miR-146a) (Transcript ENST00000221452, Fig S4g). miR-146a, and most notably miR-146a*, were detected at higher levels in human CD16+(CD14lo) monocytes than in their CD14+(CD16−) counterparts ex vivo (Fig 4i, S4h,i), and were selectively induced in CD14+ (CD16−) monocytes 6 h post LPS challenge (Fig 4j and S4j,k). miR-146a* was also detected in mouse monocytes (Fig S4l). These data are in line with previous findings that CD14loCD16+ monocytes resemble Ly6Clo cells and respond less well to LPS in comparison to CD14+CD16+ and CD14+CD16− monocytes, which resemble mouse Ly-6Chi monocytes (Cros et al., 2010). Furthermore, human CD14+ monocytes challenged with LPS decreased Relb mRNA levels (Fig 4k), although treatment with a LNA to suppress miR-146a* induction (Fig S4m) was sufficient to prevent Relb downregulation (Fig 4k, l).
This study provides functional evidence that miR-146a and Relb differentially regulate monocyte subsets. Following inflammatory challenge, modulation of miR-146a expression tunes the amplitude of the Ly-6Chi–but not the Ly-6Clo—monocyte response: premature miR-146a induction aborts Ly-6Chi cell amplification whereas lack of miR-146a induction leads to expansion and increased recruitment of these cells. miR-146a in monocytes targets Relb, which expression levels tune the amplitude of Ly-6Chi monocyte responses.
Recent work has identified miR-146a as a negative regulator of the canonical NF-κB inflammatory cascade by targeting Traf6 and Irak1/2 (O'connell et al., 2010) and as a tumor suppressor gene by decreasing transcription of NF-κB–targeted genes (Boldin et al., 2011; Zhao et al., 2011). The present study extends the role of miR-146a to the control of Relb, which is mostly implicated in the non-canonical NF-κB pathway (Vallabhapurapu and Karin, 2009). Relb has sizable effects on mononuclear phagocytes as it controls dendritic cell development in humans (Platzer et al., 2004) and mice (Burkly et al., 1995; Cejas et al., 2005; Wu et al., 1998), and the generation of monocyte-derived osteoclasts (Vaira et al., 2008). In accordance with the present study, the non-canonical NF-κB pathway activator CD40L also controls Ly-6Chi monocyte expansion (Lutgens et al., 2010). Of note, miR-146a can regulate proinflammatory gene expression by controlling RelB-dependent reversible chromatin remodeling (El Gazzar et al., 2011).
Ly-6Clo monocytes constitutively express miR-146a in accordance with their non-inflammatory properties (Nahrendorf et al., 2007; Auffray et al., 2009). Nevertheless, miR-146a−/− Ly-6Clo cells did not mount an inflammatory response that was notably higher than their wild-type counterparts. It is possible that miR-146a does not play a significant role in Ly-6Clo cells; yet, other regulatory mechanisms may keep Ly-6Clo cells in check in absence of miR-146a. The study of Ly-6Clo cells that bear defects in several candidate factors (e.g., miR-146a and other miRNAs) may serve to address this question. Either way, the present findings indicate that selective targeting of the miR-146a pathway should control Ly-6Chi monocyte responses while preserving Ly-6Clo cells.
Previous work has identified that miR-146a−/− macrophages produce higher levels of inflammatory cytokines than their wild-type counterparts (Boldin et al., 2011); however, we could not recapitulate these findings in miR-146a−/− monocytes. Challenged miR-146a−/− and wild-type Ly-6Chi monocytes may produce the same amount of cytokines on a per-cell basis because miR-146a up-regulation is induced after the initial burst of inflammatory cytokine production (4–24 h vs 0–8 h, respectively). Yet, miR-146a−/− Ly-6Chi monocytes will contribute more cytokine production at target sites not only because more of these cells are recruited but also because they can give rise locally to miR-146a−/−macrophages, which exhibit heightened inflammatory functions.
The findings presented here place miR-146a and Relb as key regulators of monocyte subset population dynamics. miR-146a and Relb preferentially control Ly-6Chi monocytes, which are cells that selectively expand in many chronic inflammatory conditions. Targeting of miR-146a or Relb may serve to suppress adverse inflammatory Ly-6Chi monocyte responses while sparing Ly-6Clo monocyte activity.
The studies used 6–12 wk old mice. The institutional subcommittee on research animal care at Massachusetts General Hospital approved the animal studies. Human blood was obtained from healthy volunteers and collected in heparinized collection tubes in accordance to a protocol approved by the Committee on microbiological safety at Harvard Medical School.
Cell staining and cell sorting was performed as described in supplemental methods.
Gene expression studies were performed in accordance to MIAME guidelines and are described in supplemental methods.
LPS from Escherichia coli (serotype O55:B5, Sigma) was given at 0.4 mg/kg in PBS daily i.p. for 4 d (or 7 d when indicated). Lm bacteria (strain EGDe, ATCC) were expanded in Brain Heart Broth (Fluka) and given i.v. at 3×103 CFU. Thioglycollate was given i.p. as a 4% solution in 1 ml RMPI.
5–6×104 isolated cells were plated in complete medium (RPMI, Cellgro Mediatech Inc.), 10% FCS (Stem Cell Technologies), 100 U/ml Pen/strep, 2 mM L-Glu (both Cellgro Mediatech Inc.) in round bottom 96-well plates. Stimulations included LPS (100 ng/ml, Sigma), rmTNFα (50 ng/ml, Peprotech) and HKLM (5×108/ml, heat killed Lm, Invivo Gen). Luminex cytokine assays (R&D Biosciences) were analyzed on a Luminex FlexMap 3D (Agilent) instrument.
Results were analyzed with Prism 4.0 (GraphPad). P-values were determined using Student’s t tests. A p-value <0.05 was taken as statistically significant and higher significance is indicated in the figure legends. All graphs show mean ± SEM.
The authors thank Mike Waring, Andrew Cosgrove and Adam Chicoine (Ragon Institute of MGH, MIT and Harvard) for cell sorting; Borja Saez (Harvard Medical School), Patrick Stern and David Feldser (MIT) for help with retroviral gene transfer. Charles Vanderburg and Anna Krichevsky (Harvard Medical School) for help with analytical RNA techniques; and Yoshiko Iwamoto and Joshua Dunham (MGH Center for Systems Biology) for help with immunofluorescence staining and imaging. Martin Etzrodt is part of the International PhD program ‘Cancer and Immunology’ at the University of Lausanne, Switzerland. This work was supported in part by National Institutes of Health grants NIH-R01 AI084880 and P30 DK043351 (to M.J. Pittet).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The microarray data newly generated in this study are available on the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/gds) under the accession number GSE32364.
Supplemental Information includes Supplemental Experimental Procedures and Figures.