Research ArticleImmunology

Desynchronization of the molecular clock contributes to the heterogeneity of the inflammatory response

See allHide authors and affiliations

Science Signaling  05 Mar 2019:
Vol. 12, Issue 571, eaau1851
DOI: 10.1126/scisignal.aau1851

Desynchronized macrophage responses

Within a population of cells that are genetically and developmentally identical, individual cells do not necessarily respond identically to the same stimulus. Such differences may be due to cell-intrinsic variables, extracellular factors, or stochastic events. Allen et al. found that the proportion of macrophages within a population that produced the cytokine IL-12p40 in response to an inflammatory stimulus depended on the phase of the circadian clock (a cell-intrinsic factor) and the relative expression of clock genes. Although the central body clock resets the clock in peripheral tissues daily, the clocks of individual cells within a population gradually become desynchronized. These findings demonstrate that such differences in clock phase between individual macrophages contribute to the heterogeneity in their response to inflammatory stimuli.

Abstract

Heterogeneity in the behavior of genetically and developmentally equivalent cells is becoming increasingly appreciated. There are several sources of cellular heterogeneity, including both intrinsic and extrinsic noise. We found that some aspects of heterogeneity in the response of macrophages to bacterial lipopolysaccharide (LPS) were due to intercellular desynchronization of the molecular clock, a cell-intrinsic oscillator. We found that the ratio of the relative expression of two clock genes, Nfil3 and Dbp, expressed in opposite phases of the clock, determined the fraction of cells that produced the cytokine IL-12p40 in response to LPS. The clock can be entrained by various environmental stimuli, making it a mechanism by which population-level heterogeneity and the inflammatory response can be regulated.

INTRODUCTION

From prokaryotes to complex multicellular organisms, cells from seemingly homogeneous or even clonal populations have diverse responses to the same stimulus (1, 2). This functional heterogeneity plays critical roles in processes as different as the generation of population fitness in cyanobacteria, developmental patterning in Drosophila melanogaster, and even differentiation of the vertebrate hematopoietic system (14). Given the multitude and diversity of scenarios in which such heterogeneity has been observed, it is not unexpected that cells of the innate immune system, which includes macrophages and dendritic cells (DCs), also demonstrate highly heterogeneous behavior (57). Much of the described heterogeneity among innate immune cells is fixed through differentiation or influenced by environmental signals (810), leading to phenotypically distinct cellular populations. Here, we sought to identify heterogeneity in the innate immune response among a phenotypically homogeneous population.

Heterogeneity among developmentally and phenotypically equivalent cells is often attributable to noise. Noise can be characterized as either “intrinsic” or “extrinsic” (11). Intrinsic noise refers to the noise inherent in gene transcription and can be demonstrated as differential expression from two identical alleles within a genome. Extrinsic noise encompasses all other cell-intrinsic variables that lead to intercellular variability in behavior. Some examples of extrinsic noise include heterogeneity in ribosome number or the abundance of a pathway-specific transcription factor (11, 12). One specific type of extrinsic noise is heterogeneity due to the desynchronization of cell-intrinsic oscillators (12, 13). This type of noise is unique, because cell-to-cell variability can be modulated by the degree of oscillator synchronization between cells.

One well-recognized cell-intrinsic oscillator is the circadian clock, which is present in organisms from cyanobacteria to humans and regulates various aspects of physiology and behavior, from sleep-wake cycles to nutrient and xenobiotic metabolism (1416). In mammals, the clock is controlled by transcriptional-translational feedback loops. The core components of the circadian clock include the transcriptional activators brain and muscle ARNT-like (BMAL) and circadian locomotor output cycles protein kaput (CLOCK), which form a heterodimeric complex and drive the expression of genes encoding the transcriptional repressors period circadian protein homolog (PER) and cryptochrome (CRY). PER and CRY, in return, negatively regulate CLOCK-BMAL complexes by associating with them and preventing the histone acetylation that is required for gene transcription (15). In addition to the core loop, there are additional loops that interact with the core loop. One of these secondary loops involves transcriptional activators of the proline and acidic amino acid–rich basic leucine zipper (PAR-bZIP) family, including albumin D box–binding protein (DBP), thyrotroph embryonic factor (TEF), and hepatic leukemia factor (HLF), all of which function in opposition to the transcriptional repressor NFIL3 (nuclear factor, interleukin-3–regulated) (16, 17).

We found that cells of a phenotypically homogeneous population of mouse bone marrow (BM)–derived macrophages (BMDMs) exhibited a heterogeneous cytokine response to bacterial lipopolysaccharide (LPS). The response to LPS was not a fixed property of the cells. Instead, the response of an individual cell depended on the phase of the circadian oscillation. The phase of the circadian clock can be modulated by various environmental stimuli (18, 19); therefore, it is likely that both the phase of the clock and the degree to which populations are synchronized with respect to the phase of the clock are important mechanisms by which responses are controlled at the population level.

RESULTS

The macrophage response to LPS is heterogeneous and unstable

To initially characterize heterogeneity in the response of mouse BMDMs to LPS stimulation in vitro, we analyzed cytokine production by flow cytometry. Both of the cytokines tumor necrosis factor (TNF) and interleukin-12 p40 (IL-12p40) were produced in a bimodal manner, meaning that within the cell population, there were distinct groups of cells with respect to the amount of cytokine produced (Fig. 1, A and B). This was in contrast to inducible nitric oxide synthase (iNOS), which was produced in an analog fashion, referring to a continuous spectrum of iNOS production among the population (Fig. 1C). Whereas all BMDMs produced TNF when exposed to a large enough concentration of LPS, there always remained a fraction of BMDMs that did not produce IL-12p40 regardless of the concentration of LPS or the length of stimulation (Fig. 1B). To determine whether heterogeneity in the IL-12p40 response was due to macrophage preprogramming during differentiation, we tested the stability of IL-12p40 production. To do this, we stimulated BMDMs from mice carrying a fluorescent IL-12p40 reporter (yet40) (20) with LPS and sorted them by flow cytometry based on the abundance of yellow fluorescent protein (YFP). These sorted populations were then rested in culture for 4 days to enable the reversal of LPS tolerance and subsequently restimulated with LPS. LPS tolerance describes a macrophage state that develops after initial macrophage activation in the presence of persistent LPS. It is characterized by a dampened inflammatory response to LPS, preventing excessive inflammation, whereas other less destructive antimicrobial processes are continuously expressed (21). Upon restimulation, bimodal IL-12p40 production was reproduced in both the previously IL-12p40–negative and IL-12p40–positive populations, suggesting that heterogeneity in IL-12p40 production is largely controlled by nonstable cellular variables rather than differences in cell differentiation or identity (Fig. 1D).

Fig. 1 Heterogeneity in BMDM cytokine production is bimodal and unstable.

(A) BMDMs were stimulated with the indicated concentrations of LPS for 7 hours (top) or 3 hours (bottom) in the presence of brefeldin A (BFA) and then analyzed by intracellular flow cytometry for IL-12p40 (top) or TNF (bottom). Numbers within flow cytometry plots indicate the percentages of cytokine-positive cells. Count (y axis) refers to the number of cells with a given amount of cytokine measured by fluorescence (x axis). (B) Concentration-time response curves for BMDMs stimulated with LPS for the indicated times and then analyzed by intracellular flow cytometry for IL-12p40 or TNF. Graphs indicate the percentage of the total population of cells that were positive for the indicated cytokine at each time point. Data in (A) and (B) are representative of two independent experiments. (C) BMDMs were stimulated with 10-ng LPS/ml for 10 hours. BFA was added after 4 hours of stimulation. Cells were then analyzed by flow cytometry for iNOS and IL-12p40. Data are representative of three independent experiments. (D) Day 6 BMDMs from IL-12p40 (yet40) YFP reporter mice were stimulated with 10-ng LPS/ml for 6 hours and subsequently sorted by flow cytometry into YFP+ and YFP populations. Cells were returned to culture for 4 days and on day 10 were left unstimulated or were restimulated with 10-ng LPS/ml for 6 hours in the presence of BFA before being subjected to intracellular staining for IL-12p40. Data are representative of two independent experiments.

Molecular clock components control the macrophage response to LPS

Desynchronization of cell-intrinsic oscillators is one mechanism by which unstable heterogeneity can be generated within a population of cells (Fig. 2A) (13). This was an appealing model for control of the inflammatory response, because modulation of heterogeneity through cell-intrinsic oscillators would enable increased control relative to other noise-mediated mechanisms. The molecular clock is a cell-intrinsic oscillator present nearly ubiquitously in mammalian cells, and it is composed of a transcription-translation feedback loop with an about 24-hour period (Fig. 2B) (16). The clock protein NFIL3 inhibits Il12b transcription (22), and we therefore suspected that desynchronization of the clock could contribute to the observed heterogeneity in IL-12p40 production.

Fig. 2 Clock genes Nfil3 and Dbp oppositely control the fraction of LPS-induced BMDMs that are IL-12p40+.

(A) Intercellular heterogeneity can be generated through desynchronization of cell-intrinsic oscillators. (B) Schematic representation of the primary and secondary loops of the molecular clock. (C) Flow cytometry plots of IL-12p40 in BMDMs of the indicated genotype unstimulated or stimulated with 10-ng LPS/ml for 6 hours. Data are representative of two independent experiments. The numbers within the flow cytometry plots indicate fraction of cells positive for IL-12p40. (D) BMDMs retrovirally transduced with Nfil3, Dbp, or empty expression vectors containing an IRES-hCD2 to allow for identification of successfully transduced cells. BMDMs were either unstimulated or stimulated with 10-ng LPS/ml for 6 hours. Flow cytometry plots for IL-12p40 in transduced BMDMs are shown. Data are representative of three independent experiments. (E) Relationship between the amount of gene overexpression and the fraction of LPS-induced cells that were IL-12p40+. Data are representative of two independent experiments. (F) Reverse transcription–quantitative polymerase chain reaction (RT-qPCR) for Il12b primary transcripts in BMDMs stimulated with 10-ng LPS/ml for the indicated times. Error bars represent mean with SD. Graph is representative of two independent experiments. (G) ChIP for NFIL3 at the Il12b enhancer at time 0 (no stim), 2 hours, and 6 hours after BMDM stimulation with 10-ng LPS/ml. (H) ChIP for FLAG at the Il12b enhancer at the indicated times in Dbp-Flag retrovirally transduced BMDMs after stimulation with 10-ng LPS/ml. For (G) and (H), the percent input represents enrichment of Il12b enhancer DNA relative to DNA immunoprecipitated from an unrelated region of the genome (negative control). Data are representative of three independent experiments (fig. S3, A and B).

Consistent with NFIL3 being a negative regulator of Il12b expression, we observed that with the loss of Nfil3 through genetic deletion in Nfil3+/− and Nfil3−/− mice, the fraction of BMDMs producing IL-12p40 in response to LPS increased (Fig. 2C). This finding was also observed in BM-derived DCs (BMDCs), demonstrating a more generalizable contribution of NFIL3 to Il12b gene regulation (fig. S1). NFIL3 inhibition of Il12b transcription is thought to be mediated through binding to a D-box within the Il12b enhancer (22, 23). PAR-bZIP proteins share the ability to bind D-boxes and function as transcriptional activators. DBP is a member of the PAR-bZIP family and is produced in the opposite phase of the clock as NFIL3 (24). To test whether DBP functions in the opposite manner as NFIL3, we overexpressed either Nfil3 or Dbp in BMDMs using retroviral transduction of overexpression vectors (fig. S2, A to C). Whereas Nfil3 overexpression decreased the fraction of BMDMs that were IL-12p40+, Dbp overexpression increased this fraction. Additionally, within transduced BMDMs, the degree to which Dbp or Nfil3 was overexpressed correlated with the extent to which the IL-12p40+ fraction was increased or decreased, respectively (Fig. 2, D and E). Because NFIL3 and DBP are both capable of binding to D-boxes, we predicted that NFIL3 and DBP competitively bind at the Il12b enhancer. Consistent with this hypothesis, chromatin immunoprecipitation (ChIP) of Flag in BMDMs transduced with a retrovirus overexpressing Flag-tagged Dbp revealed enrichment of DBP at the Il12b enhancer at the time of peak Il12b transcription, whereas enrichment of NFIL3 was observed later at times of Il12b repression (Fig. 2, F to H, and fig. S3, A and B). This late enrichment of NFIL3 at the IL12b enhancer also correlated with LPS-induced Nfil3 expression and Dbp repression (fig. S3, C and D).

Given the opposing roles of DBP and NFIL3 in the regulation of Il12b, we suspected that there might be other genes that can be similarly regulated by these two factors. To investigate this, we performed RNA sequencing (RNA-seq) on BMDMs from wild-type and Nfil3−/− mice that were either unstimulated or stimulated with LPS for 4 or 24 hours. We noted that even before LPS stimulation Nfil3−/−, BMDMs had statistically significantly higher expression of the gene encoding the chemokine receptor CCR2, which is found predominately on inflammatory monocytes (25). After LPS stimulation, Nfil3−/− BMDMs had enrichment for a subset of genes involved in the proinflammatory antibacterial response and decreased expression of a subset of genes involved in inhibition of the LPS inflammatory response (2628) (Fig. 3A). Consistent with our hypothesis that DBP and NFIL3 can play opposing roles in the LPS inflammatory response, gene expression analysis of Dbp-overexpressing BMDMs revealed that for several genes, Dbp overexpression phenocopied the loss of Nfil3 (Fig. 3, B and C).

Fig. 3 Nfil3 and Dbp regulate BMDM inflammatory response to LPS.

(A) RNA-seq of wild-type (WT) and Nfil3 knockout (KO) BMDMs stimulated with 10-ng LPS/ml for 0 (no stim), 4, or 24 hours. Two independent samples for each condition were sequenced. Genes differentially expressed greater than twofold (Padj < 0.01) in WT or Nfil3 KO BMDMs during at least one of the assayed times are shown. Differential expression is plotted in log2. Blue indicates enriched in WT BMDMs; red indicates enriched in Nfil3 KO BMDMs. Il12b is shown but was only 1.8× enriched in Nfil3 KO at 4-hour LPS. IFN, interferon. (B) RT-qPCR for selected transcripts in WT and Nfil3 KO BMDMs under the indicated conditions (expressed as relative to WT for the indicated conditions). Error bars represent mean with SD from three technical replicates. Data are representative of three independent experiments. (C) RT-qPCR for selected transcripts in BMDMs either transduced with a Dbp-overexpressing or empty vector (EV) under the indicated conditions. Expressed as relative to empty vector for the indicated conditions. Error bars represent mean with SD from three technical replicates. Data are representative of two independent experiments. (D) Ccr2 expression in BMDMs sorted by CCR2 abundance (normalized to Ccr2 expression in CCR2-low cells). Data are representative of three independent experiments. (E) IL-12p40 production in BMDMs sorted by CCR2 abundance and immediately stimulated with LPS (10 ng/ml) with BFA for 6 hours. Data are representative of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by Student’s t test.

Our data suggest that CCR2 might serve as an indicator of the clock phase given that expression of Ccr2 is affected by the abundances of DBP and NFIL3. If that was the case, macrophages with greater CCR2 abundance should have a higher LPS-induced IL-12+ fraction than those with low CCR2 abundance. To test this hypothesis, macrophages were sorted on CCR2 abundance and immediately stimulated with LPS. As expected, CCR2-high macrophages had larger fraction of IL-12+ cells than did CCR2-low macrophages (Fig. 3, D and E). Of note, in all the CCR2 sorting experiments, the total fraction of IL-12+ BMDMs was low, which we suspect is due to the immediate stimulation of the cells after sorting without allowing time for the cells to rest.

The molecular clock controls heterogeneity in the macrophage response to LPS in vivo

We sought to determine whether the phase of the clock controls fractional heterogeneity in vivo. Consistent with a previous report (29), we confirmed that peritoneal macrophages exhibited circadian oscillation of clock gene expression (Fig. 4A and fig. S4, A to C). Given that the Nfil3/Dbp ratio was highest in macrophages harvested at zeitgeber time (ZT) 0, the time at which the lights came on in the animal facility, and lowest in macrophages harvested at ZT 12, the time at which the lights switched off, we predicted that the fraction of macrophages producing IL-12p40 in response to LPS would be smaller at ZT 0 relative to that at ZT 12. As predicted, peritoneal macrophages isolated and stimulated in vitro at ZT 0 or ZT 1 had a significantly smaller IL-12p40+ fraction than those isolated at ZT 12 (Fig. 4, B and C). We then examined the response to LPS in peritoneal macrophages from Bmalfl/flLysM-cre mice, in which the core clock gene Bmal is deleted specifically in macrophages. Expression analysis of peritoneal macrophages from ZT 1 and ZT 12 in these mice revealed that they had lost their circadian regulation of Nfil3 and Dbp. Additionally, loss of Bmal led to persistently high Nfil3 and low Dbp expression, mimicking ZT 0, throughout the ZT 0 to ZT 12 light phase of the circadian cycle (Fig. 4, D and E). When peritoneal macrophages from Bmalfl/fl-LysM-cre+ mice were stimulated with LPS, they demonstrated a loss of circadian variation in IL-12+ fraction. As predicted by the high Nfil3/Dbp ratio in these cells, they behaved as if they were at ZT 0 regardless of the time of stimulation (Fig. 4C).

Fig. 4 Resident peritoneal macrophage response to LPS is time of day dependent and controlled by the molecular clock.

(A) RT-qPCR analysis of clock genes Nfil3 and Dbp from peritoneal cells isolated from C57BL/6 mice (n = 3 per time point) at ZT 0, ZT 6, and ZT 12. ZT 0 indicates the time at which the lights turn on, and ZT 12 indicates the time at which lights turn off in the animal facility. The horizontal bar above each graph indicates the light/dark cycle. Error bars represent mean with SD from three technical replicates. (B) IL-12p40 production in peritoneal macrophages (CD11b+ F4/80+) isolated from C57BL/6 mice at ZT 0 or ZT 12 and stimulated in vitro under the indicated conditions for 4 hours in the presence of BFA. Numbers within boxes indicate fraction of macrophages that are IL-12+ as measured by flow cytometry. (C) Fraction of peritoneal macrophages from Bmal fl/fl;LysMcre (Bmal WT) and Bmal fl/fl;LysMcre+ (Bmal MKO) mice that were IL-12p40+ after stimulation with LPS at either ZT 1 or ZT 12. Cells were stimulated as described in (B). Each circle represents one mouse. Error bars represent mean with SD. (D) Dbp and (E) Nfil3 expression in Bmal fl/fl;LysMcre (Bmal WT) and Bmal fl/fl;LysMcre+ (Bmal MKO) mice at either ZT 1 or ZT 12. Error bars represent mean with SD. (F) Proposed mechanism of why clock phase may control IL-12 production only in a fraction of cells. Each oscillation represents the molecular clock in an individual cell with respect to a “threshold” for IL-12 production. All data are representative of at least three independent experiments. ***P < 0.001, **P < 0.01. n.s., not significant by Student’s t test.

DISCUSSION

We showed that the heterogeneity in IL-12p40 production by macrophages in response to LPS was not stable and that this instability was, at least in part, attributable to the phase of the molecular clock. We focused on the heterogeneity in IL-12 production because it is a key cytokine in the T helper 1 inflammatory response. Although NFIL3 was previously identified as a negative regulator of Il12b expression (22), our studies demonstrate the importance of NFIL3 as a clock protein in the regulation of IL-12 production. Additionally, we showed that in combination with NFIL3, the clock protein DBP modulated the inflammatory response. The molecular clock as a functional regulator of heterogeneity in the inflammatory response is logical for several reasons. First, other noise-mediated mechanisms of heterogeneity are inherently difficult to control. Oscillator-driven mechanisms have the benefit of being able to regulate a response through both the phase of oscillation and the degree to which populations are synchronized. Because the molecular clock can be entrained to various environmental stimuli (18, 19, 30), the population-level response can be tuned to environmental demands. This is an intuitively effective way to control the inflammatory response because it enables an adequate strength of response for pathogen clearance while minimizing immunopathology.

A remaining question is the degree to which the molecular clock is synchronized in vivo. It is generally considered that our cellular (peripheral) clocks in vivo are synchronized by daily entrainment signals from the central (master) clock in the hypothalamus. This viewpoint is, at least in part, supported by our data demonstrating the time-of-day difference in macrophage IL-12 production in addition to the circadian oscillation of clock gene expression at the population level. If macrophages are synchronized in vivo, there needs to be an explanation as to why we did not see 100% of peritoneal macrophages responding at ZT 12. One explanation for this would be that because of other noncircadian factors, individual cells have different activation barriers that need to be overcome to sustain Il12b transcription and that only cells fluctuating near the activation threshold will show circadian variation in response (Fig. 4F). However, it is also possible that not all macrophages are synchronized in vivo, as was shown to be the case among epidermal stem cells within individual hair follicles. This desynchronization is biologically important and ensures that not all stem cells have the same responsiveness to activating signals (31). With currently available tools, we were unfortunately unable to answer this question in our system.

An additional finding from this study is that the LPS-induced cytokines IL-12 and TNF exhibited bimodal production, as compared to iNOS, which was produced in an analog manner. This raises the interesting question of why some genes are expressed digitally, whereas others are analog. One difference between cytokines and iNOS is that cytokines are secreted, whereas iNOS functions intracellularly to produce reactive nitrogen species that can both directly and indirectly promote intracellular bacterial killing (32). These data raise the possibility that the cell-intrinsic function versus cell-extrinsic function of gene products determines whether genes are regulated in a digital or analog fashion. This hypothesis is supported by another study demonstrating bimodal production of the cytokine IL-6 in macrophages (5). Logically, all cells need to express genes that are required for their intrinsic functions, whereas production of secreted factors can be delegated to a fraction of the population. Additionally, the partitioning of responses may enable the segregation of incompatible processes. Given that several of the LPS-induced cytokines are produced digitally, heterogeneity in the population-level response is particularly important, because a homogeneous “all-or-none” inflammatory response risks either failure to respond to an infection or a hyperinflammatory response with resultant immunopathology.

In sum, our data demonstrate that the phase of the molecular clock is an important determinant in the macrophage response to LPS. Furthermore, this interaction between the clock and the environment is bidirectional, enabling changing functional demands to control cellular responses through manipulation of the clock phase. Although we examined the role of cell-intrinsic oscillators in the control of the macrophage inflammatory response, we suspect that these findings are generalizable as a mechanism by which many other cell decisions are made.

MATERIALS AND METHODS

Animals

All mice were housed in the Yale Animal Resources Center in a specific pathogen–free environment. All management and handing of animals were in accordance with approved guidelines by the Institutional Animal Care and Use Committee at Yale. Mice were kept on a strict 12-hour light/12-hour dark (LD) cycle, and for circadian experiments, mice were kept on the LD cycle for a minimum of 2 weeks before use. C57BL/6J mice were purchased from National Cancer Institute or the Jackson Laboratory for experiments. Yet40 (B6.129-Il12btm1Lky/J), Bmal flox (B6.129S4-Arntltm1Weit/J), and LysM-cre [B6.129P2-Lyz2tm1(cre)Ifo/J] mice were purchased from the Jackson Laboratory and maintained within our animal facility. Nfil3 KO mice were provided by P. B. Rothman at the University of Iowa (USA). Nfil3 genotyping primers are NFscreen28809anti (common 3′) CGATGTCCAGTGTCTTCCTTA, NFWT5’ (wild type 5′) GGAGATGGATGCCTCAGTTGGGGT, and PTEN Neo (KO 5′) ACGAGACTAGTGAGACGTGC. Cycling conditions are as follows: 94°C, 30 s; 58°C, 30 s; 72°C, 45 s.

BMDM differentiation and stimulation

BMDMs were generated by plating mouse BM cells in non–tissue culture (TC) dishes in macrophage growth medium (MGM) composed of 30% L929 supernatant/70% RPMI 1640 with l-glutamine (Corning). RPMI 1640 was supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Gibco), penicillin/streptomycin, sodium pyruvate, Hepes, and 2-mercaptoethanol. BM was prepared by crushing mouse femurs and tibias to release marrow, followed by ACK (ammonium-chloride-potassium) lysis and passage through a 70-μm cell strainer. BM was plated on day of isolation (day 0) at 7 × 106 cells/20 ml of MGM in a 15-cm non-TC dish. On day 4 of differentiation, cells were supplemented with 10 ml of MGM. Unless otherwise stated, on day 6, adherent cells were lifted with cold phosphate-buffered saline (PBS) containing 5 mM EDTA and were replated at 0.75 × 106 cells per well of 12-well non-TC dish or 1.5 × 106 cells per well of six-well non-TC dish in MGM for stimulation on day 7 of differentiation. All cell culture was done at 37°C with 5% CO2. LPS stimulation was performed with Ultrapure LPS, Escherichia coli 0111:B4 (Invivogen, catalog no. tlrl-3pelps). Concentrations and length of stimulations are indicated in the figure legends and/or the results section of the main text. If intracellular cytokine staining was to be performed, GolgiPlug (BD Biosciences, catalog no. 555029), a BFA-containing reagent, was added at the time of stimulation.

BMDC differentiation

To generate BMDCs, murine BM was cultured in DC growth medium, which consisted of Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS, sodium pyruvate, l-glutamine, Hepes, penicillin/streptomycin, β-mercaptoethanol, and granulocyte-macrophage colony-stimulating factor for 6 days (33). BM cells (1 × 106) were plated in 1 ml of DC growth medium, and 500 μl of medium was replaced every 2 days. BMDCs were gated on live, singlets CD11c+ MHC IIhi, CD11b.

Flow cytometry staining and analysis

BMDMs were collected for flow cytometry from non-TC dishes with cold PBS with 5 mM EDTA. All staining steps and washes were done in staining buffer (PBS with 2% FBS, 2.5 mM EDTA) unless otherwise indicated. Cells were Fc blocked with anti-CD16/CD32 (clone 93, eBioscience) at 1:200 for 15 min on ice before staining with labeled antibodies. Cells were stained with extracellular antibodies for 30 min. Antibodies include anti-CD11b–fluorescein isothiocyanate at 1:250 (clone M1/70, eBioscience, catalog no. 11-0112-82), anti-F4/80–allophycocyanin (APC) at 1:500 (clone BM8, eBioscience, catalog no. 17-4801-82), anti-hCD2 at 1:50 (clone RPA-2.10, eBioscience, catalog no. 12-0029-42), and anti-CCR2-APC at 1:10 (clone no. 475301, R&D Systems, catalog no. FAB5538A). After extracellular staining, cells were washed and fixed with 4% paraformaldehyde in PBS for 15 min on ice. For intracellular staining, cells were subsequently permeabilized with BD Perm/Wash buffer (catalog no. 554723), and intracellular staining was performed in BD Perm/Wash buffer for 45 min. Antibodies included anti–TNFα–phycoerythrin (PE) at 1:100 (clone MP6-XT22-PE, eBioscience, catalog no. 12-7321-81) and anti–IL-12/IL-23 p40 at 1:100 (clone C17.8, eBioscience, catalog no. 12-7123-82 or 50-7123-82). After intracellular staining, cells were washed with BD Perm/Wash buffer followed by staining buffer. Samples were run on a BD FACSCalibur or LSR II followed by analysis with FlowJo software.

Cell sorting

BMDMs were stimulated as described in the figure legends and main text and subsequently lifted with cold PBS containing 5 mM EDTA. Cells were spun at 1350 rpm at 4°C and washed once with sterile staining buffer (PBS with 2% FBS, 2.5 mM EDTA). If needed, extracellular staining was performed as described earlier. Cells were then filtered and sorted on a Beckman Coulter MoFlo or BD FACSAria in the Yale fluorescence-activated cell sorting (FACS) Facility. Peritoneal cells were collected as described under the section “Peritoneal cell isolation and stimulation.” They were then stained with anti-CD11b and anti-F4/80 as described under the section “Flow cytometry staining and analysis.” Stained cells were washed with sterile staining buffer and filtered (as described above), followed by sorting on a BD FACSAria in the Yale FACS Facility.

RNA isolation, reverse transcription, and RT-qPCR

BMDM RNA was collected with RNA Bee (Amsbio). Phase separation was performed with chloroform, followed by isopropanol precipitation of RNA and ethanol wash according to the manufacturer’s protocol. For peritoneal cells and sorted cells, RNA was isolated using the RNeasy Mini Kit including On-Column DNAse Digestion with the RNase-Free DNase Set (Qiagen). Purified RNA was resuspended in nuclease-free water and quantitated using a NanoDrop 8000 Spectrophotometer before reverse transcription (RT). Oligo(dT) (Sigma-Aldrich) or random hexamer (Sigma-Aldrich) primers were used for mRNA or primary transcript analysis, respectively. Primers were annealed to RNA by adding 1 μl of oligodT (0.5 μg/μl) or random hexamer to 10 μl (1 μg) of RNA, followed by 70°C for 3 min and 4°C for 5 min in a thermocycler. Subsequent RT was performed using SMART MMLV Reverse Transcriptase (Clontech) in a 20-μl total reaction volume according to the manufacturer’s protocol. Upon completion of RT, complementary DNA (cDNA) volume was brought to 100 μl with nuclease-free water. RT-qPCR was performed with PerfeCTa SYBR Green SuperMix, Low ROX (Quanta Biosciences) in a 10-μl reaction volume according to the manufacturer’s protocol. Bio-Rad CFX96 was used for thermocycling. Primers for RT-qPCR are listed in table S1.

Retroviral transduction

Human embryonic kidney (HEK) 293 T (293 T) cells were grown in DMEM with 10% FBS at 37°C, 5% CO2. 293 T cells (0.75 × 106/6-cm dish/3-ml medium) were transfected with 4 μg of retroviral vector (MIGR2.IRES-hCD2) and 4 μg of pCL-Eco packaging vector with Lipofectamine 2000 (Life Technologies) as a transfection reagent. Twenty-four hours after transfection, medium was removed and replaced with MGM. Twenty-four hours later, HEK 293 T viral supernatant was collected and passed through a 0.45-μm filter. Fresh MGM (3 ml) was added to 293 T cultures. Lipofectamine 2000 was added to viral supernatant at 1/250 and incubated at room temperature for 10 min. BM from one mouse was isolated as described under “BMDM differentiation and stimulation” and resuspended in 200 μl of MGM. Fifty microliters of BM was mixed with 3 ml of viral supernatant and spinfected by centrifuging at 2500 rpm, 90′ at 32°C in a six-well non-TC plate. After spinfection, 1.5 ml of fresh MGM was added to each well, and cells were incubated overnight at 37°C. The following day, fresh viral supernatant was filtered and incubated with Lipofectamine 2000 as described above. BM cells were collected, spun down, and resuspended in freshly prepared viral supernatant. Spinfection was repeated. After spinfection, BM was collected and washed once with MGM, and cells were counted. D1 BM cells were plated at ~7 × 106 cells per 15-cm petri dish in 20 ml of MGM. From this point, BMDMs were differentiated as described under “BMDM culture.”

Cloning

Murine Dbp was cloned from BMDM cDNA, and murine Nfil3 was cloned from pcDNA3.1-mNFIL-3, a gift from P. Murray (Addgene plasmid no. 34572). Inserts were subcloned into the multiple cloning site of the modified pMSCV.IRES-hCD2 retroviral overexpression vector, MIGR2 (provided by D. Stetson, University of Washington, Seattle, WA). Q5 Site-Directed Mutagenesis kit (New England Biolabs, catalog no. E0554) was used to add a FLAG sequence to the 5′ end of Dbp.

Chromatin immunoprecipitation

About 40 × 106 BMDMs were used per condition. After simulation with LPS, cells were directly fixed with 1% formaldehyde and incubated at room temperature for 10 min. Cross-linking was quenched by adding freshly prepared glycine in PBS to a final concentration of 0.125 M and let sit for 5 min at room temperature. Plates were washed twice with cold PBS and were scraped into a 50-ml conical tube using cell scrapers. Collected cells were spun and resuspended in 0.5 ml of SDS lysis buffer [1% SDS, 10 mM EDTA, and 50 mM tris (pH 8.0)] with 1× protease inhibitors (Sigma-Aldrich, catalog no. S8830). Cells were sonicated using a Virtis Virsonic 600 with a microtip and the following settings: total process time, 5 min; pulse on, 0.5 s; pulse off, 0.5 s; output, 1. Chromatin was stored overnight (O/N) at −80°C and thawed the following day on ice. Samples were centrifuged for 15 min, with a maximum speed at 4°C in a tabletop centrifuge, and supernatant was taken for quantification on a NanoDrop 8000. Chromatin (100 μg) was used per immunoprecipitation (IP) in a total volume of 300 μl in ChIP dilution buffer (16.7 mM tris-HCl, 167 mM NaCl, 1.2 mM EDTA, 0.01% SDS, and 1.1% Triton X-100) with protease inhibitors. After this, the appropriate antibodies were added to each IP, 10 μg of anti-NFIL3 (Santa Cruz Biotechnology, catalog no. sc-9550x), 8 μg of anti-FLAG (Sigma-Aldrich, catalog no. F7425), and samples were rotated overnight at 4°C. To pull down immune complexes, 50 μl of Protein G Agarose/Salmon Sperm DNA (EMD Millipore) was added per IP and rotated 2 hours at 4°C. Fifteen microliters of supernatant from no-antibody control samples was saved as 5% input. Samples were washed in the following sequence: 1× with 1 ml of low-salt wash buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM tris-HCl (pH 8.1), and 150 mM NaCl], 1× with 1 ml of high-salt wash buffer [0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM tris-HCl (pH 8.1), and 500 mM NaCl], 1× with LiCl1 wash buffer [0.25 M LiCl, 1% IGEPAL CA-630, 1% deoxycholic acid (sodium salt), 1 mM EDTA, 10 mM tris (pH 8.1)], and 2× with 1-ml tris-EDTA (TE) (pH 8.0). All washes were done with ice-cold buffer at 4°C with rotation, except TE washes that were done at room temperature. After washes, samples and inputs were resuspended in 300-μl SDS lysis buffer with 1-μl Proteinase K (20 μg/μl) and incubated for 2 hours at 55°C to digest proteins. Samples were reverse–cross-linked by incubating overnight at 65°C. Immunoprecipitated DNA was purified with the Qiagen PCR purification kit, and samples were eluted in 200 μl of water. RT-qPCR was performed as described in the section “RNA isolation, reverse transcription, and RT-qPCR.” ChIP primers were as follows: Il12b enhancer, TTCACCAGTGACTCCAGCAG (forward) and AGGACCATGGCTGGTACAAC (reverse); Il12b promoter, GGGGAGGGAGGAACTTCTTA (forward) and CTTTCTGATGGAAACCCAAAG (reverse); and negative control, AGCTGTGTAGGGACACATATTGAG (forward) and CACACAAACTCTTAGTCCAGTTCC (reverse).

RNA-seq and sample preparation

RNA for next-generation sequencing was isolated using the RNeasy Mini Kit (Qiagen), including QIAshredder (Qiagen) sample homogenization and On-Column DNAse Digestion with the RNase-Free DNase Set (Qiagen). RNA purity and quality were analyzed on an Agilent Bioanalyzer 2100 before library preparation. Libraries were prepared by the Yale Center for Genome Analysis and run on a HiSeq 2000 (eight samples per lane) at 1 × 75 base pairs. Raw data were analyzed as follows: The raw sequence reads of RNA-seq experiments from Illumina were trimmed off sequencing adaptors and low-quality regions by btrim (34). The trimmed reads were mapped to mouse genome (mm10) by TopHat2 (35). The gene definitions of mouse genome were based on University of California, Santa Cruz annotation and downloaded from iGenomes (http://support.illumina.com/sequencing/sequencing_software/igenome.html). After the counts were collected, the differential expression analysis was carried out by DESeq2 (36) using the default setting to normalize the read counts and calculate the adjusted P values.

Peritoneal cell isolation and stimulation

Mice were euthanized by CO2 asphyxiation. Abdominal skin was removed, and peritoneum was flushed with 8 ml of RPMI 1640 with 10% FBS and 1× penicillin/streptomycin (supplemented RPMI 1640) using a 20-gauge needle. Medium was prewarmed to 37°C if cells were to be used for culture; otherwise, medium was at 4°C. For subsequent RNA isolation, cells were spun at 1350 rpm, 5 min at 4°C, and cell pellet was resuspended in 350 μl of Buffer RLT supplemented with 2-mercaptoethanol as directed by RNeasy Mini Protocol (Qiagen). For subsequent cell culture, peritoneal cells were spun at 1350 rpm, 5 min at room temperature, and were resuspended in supplemented RPMI 1640 with GolgiPlug (BD Biosciences, catalog no. 555029). Cells were transferred to a 96-well round-bottom plate for subsequent stimulation. Cells were stimulated with Ultrapure LPS, E. coli 0111:B4 (Invivogen, catalog no. tlrl-3pelps) as described in the figure legends and main text.

Statistical analysis

When comparing two groups with three or more technical replicates per group, a two-tailed unpaired Student’s t test was performed to determine statistical significance. Error bars indicate SD. For experiments with only two samples for group, error bars represent mean with range.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/12/571/eaau1851/DC1

Fig. S1. IL-12p40 production by BMDCs depends on Nfil3.

Fig. S2. The surface abundance of hCD2 correlates with the extent of retroviral gene overexpression.

Fig. S3. NFIL3 and DBP dynamically interact with the Il12b enhancer after LPS stimulation.

Fig. S4. Peritoneal macrophages exhibit circadian oscillation of clock gene expression.

Table S1. RT-qPCR primers used in this study.

REFERENCES AND NOTES

Acknowledgments: We thank P. B. Rothman for the Nfil3−/− mice, P. Murray for the Nfil3 cDNA containing plasmid, and D. Stetson for the MIGR2 retroviral overexpression vector. We would also like to thank Y. Okabe for intellectual contributions to the work. Funding: This work was supported by the Howard Hughes Medical Institute, Else Kröner Fresenius Foundation, the Blavatnik Family Foundation, and the Trudeau Fellowship from Yale University. Author contributions: R.M. and N.C.A. conceived and designed the project. N.C.A., N.H.P., L.H., X.Z., and R.A.F. designed and performed the experiments. N.C.A., N.H.P., and Y.K. performed the computational and statistical analysis. N.C.A. wrote the manuscript, and all authors were involved in the interpretation, discussion, and preparation of the final manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article