Regulation of Intracellular Triiodothyronine Is Essential for Optimal Macrophage Function

Abstract

Innate immune cells, including macrophages, have recently been identified as target cells for thyroid hormone. We hypothesized that optimal intracellular concentrations of the active thyroid hormone triiodothyronine (T3) are essential for proinflammatory macrophage function. T3 is generated intracellularly by type 2 deiodinase (D2) and acts via the nuclear thyroid hormone receptor (TR). In zebrafish embryos, D2 knockdown increased mortality during pneumococcal meningitis. Primary murine D2 knockout macrophages exhibited impaired phagocytosis and partially reduced cytokine response to stimulation with bacterial endotoxin. These effects are presumably due to reduced intracellular T3 availability. Knockdown of the main TR in macrophages, TRα, impaired polarization into proinflammatory macrophages and amplified polarization into immunomodulatory macrophages. Intracellular T3 availability and action appear to play a crucial role in macrophage function. Our data suggest that low intracellular T3 action has an anti-inflammatory effect, possibly due to an effect on macrophage polarization mediated via the TRα. This study provides important insights into the link between the endocrine and innate immune system.

Reduced intracellular T3 concentrations and action in macrophages, through modulation of type 2 deiodinase or thyroid hormone receptor α, impair the proinflammatory macrophage response.

Thyroid hormones (THs) are essential for growth, development, and energy metabolism (1). Innate immune cells have recently been identified as TH target cells [for review, see van der Spek et al. (2)]. Macrophages are important innate immune cells that play essential roles in tissue homeostasis and immunity. Macrophage dysfunction has been linked to a large number of pathophysiological conditions, including cancer, diabetes, inflammatory bowel disease, and atherosclerosis (3). Triiodothyronine (T3), the active form of TH, is important for adequate macrophage function (4).

TH is produced by the thyroid gland mainly as the prohormone thyroxine (T4). Once taken up by the cell, T4 needs to be converted into T3, the active hormone, to exert its action. To this end, macrophages contain type 2 deiodinase (D2). This enzyme belongs to the deiodinases, a family of enzymes that activate or inactivate the different molecular forms of intracellular TH. D2 is the TH-activating deiodinase that converts T4 to T3 (5). T3 then binds the nuclear TH receptor (TR) to regulate gene transcription. TR expression is differentially regulated with different cell types expressing different isoforms (6). The predominant isoform in macrophages is TRα (4). In addition to its classic transcriptional effects, T3 has nongenomic effects (7). Macrophages that lack D2, and thus that presumably have lower intracellular T3 levels, exhibit an impaired proinflammatory cytokine response to stimulation with bacterial endotoxin [lipopolysaccharide (LPS)] and impaired phagocytosis (4). Furthermore, murine macrophages derived from TRα knockout (TRαKO) mice, which cannot mediate T3-dependent gene transcription, also display functional abnormalities, including impaired cholesterol efflux and increased proinflammatory cytokines at baseline (8, 9) but reduced proinflammatory cytokine response to LPS (4).

Macrophages are phagocytic cells that are capable of a wide range of functions in vivo. In response to stimuli from their microenvironment, macrophages can shift between a proinflammatory and an anti-inflammatory phenotype, a process known as “polarization,” making them key players in the regulation of the local immune response (10). Proinflammatory classically activated macrophages, or M1-like macrophages, are important for microbial killing and for the recruitment and activation of other immune cells (3). In contrast, anti-inflammatory alternatively activated macrophages, or M2-like macrophages, encompass a heterogeneous spectrum of phenotypes that play a role in tissue homeostasis, remodeling, and repair (3).

We hypothesized that optimal intracellular T3 concentrations are essential for proinflammatory macrophage function. To test this hypothesis, we determined the effects of reduced intracellular T3 generation due to D2 knockout (D2KO) in vivo in a zebrafish embryo model of pneumococcal meningitis. In addition, we assessed the effect of reduced intracellular T3 on ex vivo proinflammatory macrophage function using bone marrow–derived macrophages (BMDMs) from D2KO mice. Finally, to determine the role of intracellular T3 in macrophage polarization, we analyzed the effects of modulation of intracellular TH metabolism on macrophage polarization in a macrophage cell line.

Methods

Animal care and procedures

All animal experiments were conducted in compliance with relevant institutional, national, and international guidelines and regulations. Experimental protocols were approved by the Maine Medical Center Research Institute Institutional Animal Care and Use Committee or the Erasmus University Medical Center Animal Ethics Committee. All zebrafish protocols adhered to the international guidelines specified by the European Council Directive 86/609/EEC.

Mice

D2KO mice were kindly provided by Dr. V. A. Galton (Department of Physiology and Neurobiology, The Geisel School of Medicine at Dartmouth, Lebanon, NH) (11). Wild-type (WT) and D2KO mice in a C57BL/6 genetic background were generated from WT and D2KO parents, respectively. Both female and male WT and D2KO mice were used. Adult mice (3 to 5 months old) were euthanized using CO₂ asphyxiation. Bone marrow was isolated immediately and processed for flow cytometry staining or macrophage culture. Animals were housed at the Maine Medical Center Research Institute animal facility with 12 hour light/12 hour dark cycles and ad libitum access to water and regular chow.

Adult female C57BL/6 mice (Harlan) aged 2 to 4 months were euthanized using CO₂ asphyxiation. Bone marrow was isolated immediately and processed for macrophage culture. Animals were housed at the Erasmus Medical Center animal facility under specific pathogen-free conditions with 12 hour light/12 hour dark cycles and ad libitum access to water and regular chow.

Zebrafish

Husbandry and embryo care

Transparent adult Tg(mitfa^w2/w2; roy^a9/a9) casper zebrafish were maintained at 26°C in 5-L aerated tanks with a 14/10 h light/dark cycle at the VU University Medical Center, Amsterdam, Netherlands. Embryos were collected within the first hour of fertilization, before the one-cell stage of embryonic development. Zebrafish handling and embryo care were performed as described previously (12). Zebrafish were maintained according to standard protocols (zfin.org), and experiments were conducted in compliance with institutional and national animal welfare guidelines and regulations.

Morpholino injections

To transiently block the translation of D2, zebrafish embryos were injected in the one- to four-cell stage with antisense oligonucleotide morpholino (MO). A scrambled control MO (SCMO) sequence was used as a control (13). All MOs were purchased from Gene Tools. Specificity of the D2 morpholino (D2MO) has been previously demonstrated by rescue using T3 supplementation (14). Methods for microinjection, MO sequences, and concentrations have been previously described (13, 14). Briefly, casper zebrafish embryos were injected with 2 nL of 0.4 mM D2MO or SCMO dissolved in sterile 0.5% [weight-to-volume ratio (w/v)] phenol red solution (Sigma-Aldrich) at the one- to four-cell stage as described previously (13). MO sequences were derived from Bagci et al. (13) and are as follows: D2MO, 5′-TCCACACTAAGCAAGCCCATTTCGC-3′; scrambled control MO, 5′-CCTCTTACCTCAGTTACAATTTATA-3′ (Gene Tools). Embryos were raised at 28°C in E3 medium (5.0 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl·2H₂O, 0.33 mM MgCl₂·6H₂O) supplemented with 0.3 mg/L methylene blue. A rescue experiment was performed in which 50 nM T3 was directly added to E3 medium after D2MO injection of embryos, and after 24 hours the chorion was mechanically removed. All procedures were performed according to local animal welfare regulations.

Infection of zebrafish embryos

Meningitis was induced in zebrafish embryos as previously described (12). Briefly, D2MO, SCMO, and D2MO + T3 zebrafish embryos were injected in the hindbrain ventricle at 2 days after fertilization with 500 colony-forming units (CFU) of wild-type Streptococcus pneumoniae serotype 2 D39 strain to induce meningitis. S. pneumoniae serotype 2 D39 cells were grown overnight on Columbia agar plates supplemented with 5% defibrinated sheep blood at 37°C in a humidified atmosphere with 5% CO₂. Before injection, bacteria were collected from an overnight culture and suspended in Todd Hewitt broth supplemented with 0.5% yeast extract (Difco; BD Biosciences) and grown to mid-log phase at 37°C. Bacteria were harvested by centrifugation (6000 rpm, 10 minutes), washed with sterile phosphate-buffered saline (PBS), and suspended in sterile 0.5% (w/v) phenol red solution (Sigma-Aldrich) to aid visualization of the injection process. The number of CFU per injection was determined by quantitative plating of the injection volume.

Survival of the embryos was quantified by determining live and dead embryos at fixed time points between 0 and 96 hours after infection. The experiment was performed in triplicate with a total of 40 embryos in each group.

Macrophage isolation, culture, and stimulation

D2KO mice

For D2KO and WT BMDMs, bone marrow was flushed from freshly isolated femurs and tibias with cold, sterile PBS without calcium and magnesium, centrifuged, and thoroughly resuspended in Dulbecco’s modified Eagle medium (DMEM)/F12 medium (Gibco) containing 10% [volume-to-volume ratio (v/v)] fetal calf serum (FCS), 10 mM l-glutamine, 100 IU/mL penicillin, and 100 µg/mL streptomycin (hereafter indicated as DMEM/F12-10) supplemented with 20% (v/v) L929 conditioned medium [a macrophage colony stimulating factor (M-CSF)–producing cell line]. The same batch of L929-conditioned medium was used for all experiments. Cells were counted and plated in nontissue-cultured 150 × 15 mm petri dishes at ∼5 × 10⁶ cells per petri dish in 30 mL medium. Cells were cultured at 37°C with 5% CO₂. On day 3, an additional 15 mL of medium containing 20% L929-conditioned medium was added to the culture dishes. Cells were harvested by scraping on days 6 and 7, washed, and replated in DMEM/F12-10 without L929-conditioned medium in six-well plates at 1 × 10⁶ cells/well for LPS stimulation or in 24-well plates at 1 × 10⁵ cells/well for phagocytosis. The average yield from a 150-mm petri dish was 6 to 7 × 10⁶ BMDMs. BMDM purity was assessed using flow cytometry staining for F4/80 and was always >95%. Cells were rested for 16 to 24 hours prior to stimulation.

D2KO and WT BMDMs were stimulated with 100 ng/mL bacterial endotoxin or LPS (Escherichia coli O55 B5) for up to 16 hours. Unstimulated control samples were incubated in parallel. Samples were incubated in triplicate, and experiments were repeated independently four times. In total, BMDMs from eight mice per genotype were used.

C57BL/6 mice

For C57BL/6 BMDMs, bone marrow was harvested by flushing femurs and tibias with cold, sterile RPMI 1640 (Gibco) containing 2 mM l-glutamine, 10% FCS, 100 IU/mL penicillin, and 100 µg/mL streptomycin (RPMI1640-10). Bone marrow cells were filtered through a 40-µm cell strainer to obtain single-cell suspensions. Bone marrow cells were cultured for 7 days in six-well plates in RPMI1640-10 medium supplemented with 20% LADMAC-conditioned medium (an M-CSF producing cell line) and 10 ng/mL recombinant murine M-CSF (rmMCSF; Prospec) at a density of 5 × 10⁵ cells/mL. An equal volume of RPMI1640-10 medium with recombinant mouse M-CSF (10 ng/mL) was added after 3 days. After 6 days, the culture medium was replaced by fresh RPMI1640-10 medium containing recombinant mouse M-CSF (10 ng/mL) and potential stimuli.

C57BL/6 BMDMs were stimulated with either LPS (50 ng/mL; E. coli O55 B5) and 50 ng/mL interferon (IFN)-γ (Biosource) or with 10 ng/mL interleukin (IL)-4 (Biosource) for 24 hours. Unstimulated cells were included as controls. BMDMs from five mice were used.

RAW264.7 cell line and small interfering RNA knockdown

The murine macrophage cell line RAW264.7 was kindly provided by the Tytgat Institute, Academic Medical Center, Amsterdam, Netherlands. Mycoplasma contamination status was checked regularly using polymerase chain reaction (PCR) and was always negative. RAW264.7 cells were cultured in RPMI 1640 with l-glutamine, 10 U/mL of penicillin and streptomycin, and 10% (v/v)] FCS (all from Lonza). RNA knockdown of Dio2 (D2) and Thra (TRα) was performed by introducing small interfering RNA (siRNA) using electroporation as described in detail previously (4). All siRNAs were purchased from Invitrogen. Dio2 siRNAs were designed using Dharmacon software (4). Thra and control siRNAs were predesigned by Invitrogen. After electroporation, cells were plated at 5 × 10⁴ cells/mL and rested for 24 hours before stimulation.

Two different siRNAs were used for Dio2 knockdown and three different siRNAs for Thra knockdown, with similar results. Scrambled siRNAs with matching GC content were used as controls (low GC and medium GC). siRNA introduction resulted in 70% knockdown of Dio2 and 69% knockdown of Thra, which was determined using quantitative PCR. siRNA sequences are listed in Table 1. Primer sequences are listed in Table 2.

Table 1.

siRNA Sequences

siRNA Name	siRNA Target Gene	Control siRNA	Sequence (5′–3′)	Catalog No. (Invitrogen)
Dio 2-3	Dio 2	LOGC	CCUUCAGCUAUAACCUACAAGAAGU
Dio 2-6	Dio 2	LOGC	GAGAAGAAUUUCAGCAAGAGAUGAA
Dio 2-9	Dio 2	LOGC	GGACAAUAAUGCCAACGUAGCUUAC
Thra-1	Thra	LOGC	GACCUAGAGGCCUUCAGCGAGUUUA	MSS211754
Thra-2	Thra	LOGC	GCAUGUCAGGGUAUAUCCCUAGUUA	MSS211755
Thra-3	Thra	MEGC	GGCCAUGGACUUGGUUCUAGAUGAU	MSS211756
LOGC	Control			12935200
MEGC	Control			12935300

Table 2.

Primer Sequences for Quantitative PCR

Gene	Protein	Forward (5′–3′)	Reverse (5′–3′)	Source
Csf2	GM-CSF	TGAACCTCCTGGATGACATG	GTGTTTCACAGTCCGTTTCC	Bouaboula et al. (15)
Il1b	IL-1β	TTGACGGACCCCAAAAGATG	AGAAGGTGCTCATGTCCTCA	Bouaboula et al. (15)
Tnf	TNF-α	TCTCATCAGTTCTATGGCCC	GGGAGTAGACAAGGTACAAC	Bouaboula et al. (15)
Il6	IL-6	GTTCTCTGGGAAATCGTGGA	TGTACTCCAGGTAGCTATGG	Bouaboula et al. (15)
Il10	IL-10	ATGCAGGACTTTAAGGGTTACTTG	TAGACACCTTGGTCTTGGAGCTTA	Bouaboula et al. (15)
Nos2	iNOS	ACATCGACCCGTCCACAGTAT	CAGAGGGGTAGGCTTGTCTC	Harvard primer bank (no. 146134510c2)
Arg1	Arg 1	CAGCACTGAGGAAAGCTGGT	CAGACCGTGGGTTCTTCACA	Newly designed
Dio1	Deiodinase 1	GAGCAGCCAGCTCTACGCGG	TGGGGAGCCTTCCTGCTGGT	van Zeijl et al. (16)
Dio2	Deiodinase 2	GCTTCCTCCTAGATGCCTACAA	CCGAGGCATAATTGTTACCTG	Kwakkel et al. (17)
Dio3	Deiodinase 3	CCAACTCTAGCAGTTCCGCA	GCCTCCCTGGTACATGATGG	Newly designed
Slc16a2	MCT8	GTGCTCTTGGTGTGCATTGG	GGGACACCCGCAAAGTAGAA	Bloise et al. (18)
Slc16a10	MCT10	TGATTCCCCTGTGCAGCGCC	CCACGTCGTAGGTGCCCAGC	Kwakkel et al. (4)
Thra1	TRα1	CATCTTTGAACTGGGCAAGT	CTGAGGCTTTAGACTTCCTGATC	Bakker (19)
Thra2	TRα2	CATCTTTGAACTGGGCAAGT	GACCCTGAACAACATGCATT	Bakker (19)
Thrb1	TRβ1	CACCTGGATCCTGACGATGT	ACAGGTGATGCAGCGATAGT	Boelen et al. (20)
Hprt	HPRT	GCAGTACAGCCCCAAAATGG	AACAAAGTCTGGCCTGTATCCAA	Sweet et al. (21)
Ppib	Cyclophillin B	GAGACTTCACCAGGGG	CTGTCTGTCTTGGTGCTCTCC	de Vries et al. (22)
Eef1a1	EF1α1	AGTCGCCTTGGACGTTCTT	ATTTGTAGATCAGGTGGCCG	Kwakkel et al. (4)
Rplp0	RPL0	GGCCCTGCACTCTCGCTTTC	TGCCAGGACGCGCTTGT	Bloise et al. (18)

Abbreviations: GM-CSF, granulocyte macrophage colony–stimulating factor; iNOS, inducible nitric oxide; TNF, tumor necrosis factor.

RAW264.7 cells treated with siRNA were stimulated for 24 hours with 10 ng/mL LPS (Sigma-Aldrich) and 100 IU/mL IFN-γ (Peprotech) to generate proinflammatory M1-like macrophages or with 20 ng/mL IL-4 (Peprotech) to generate anti-inflammatory M2-like macrophages (23). Unstimulated cells were included as controls. Samples were incubated in quadruplicate, and experiments were repeated independently.

Flow cytometric analysis of whole bone marrow

Freshly isolated whole bone marrow was stained using a panel of fluorescently labeled antibodies. All samples were incubated with mouse FC block (BD Biosciences) prior to staining, and relevant isotype control antibodies were used to control for background staining. D2KO and WT bone marrow cell populations were quantified by analysis of fluorescence on a MACSQuant flow cytometer (Miltenyi Biotec), and data were analyzed using FlowJo software (v.10; FlowJo). Antibodies are listed in the Table 3. Flow cytometry gating was performed as described previously (24). Briefly, cells that expressed markers for B cells (25), natural killer cells (26), and T cells (27) were excluded from analysis. This was determined using antibodies against CD19 [Research Resource Identifier (RRID): AB_657663], CD335 (RRID: AB_2573441), or CD3e (RRID: AB_469571), respectively. Antibodies against CD117 (RRID: AB_469430), CD11b (RRID: AB_396772), Ly-6G (RRID: AB_1727563), and Ly-6C (RRID: AB_10805389) were used to further differentiate between cells of the myeloid lineage. The following populations were identified: early hematopoietic blast cells CD117⁺/CD11b⁻; neutrophil precusors CD117⁺/CD11b⁺; monocyte precursors Ly-6G⁻/Ly-6C⁺/CD11b^lo; monocyte Ly-6G⁻/Ly-6C⁺/CD11b^hi; neutrophil Ly-6G⁺/Ly-6C⁻/CD11b^hi.

Table 3.

Flow Cytometry Antibodies

Protein Target	Fluorescent Conjugate	Species Raised in; Monoclonal or Polyclonal	Clone	Catalog No.	Manufacturer
Ly-6G	PerCP-Cy5.5	Rat IgG2A; monoclonal	1A8	560602	BD Biosciences
Ly-6C	PE	Rat IgG2c; monoclonal	HK1.4	12-5932	eBioscience
CD117 (cKit)	APC	Rat IgG2b; monoclonal	2B8	17-1171	eBioscience
CD11b	APC-Cy7	Rat IgG2b; monoclonal	M1/70	557657	BD Biosciences
CD19	PE-Cy7	Rat IgG2A; monoclonal	1D3	25-0193	eBioscience
CD335	PE-Cy7	Rat IgG2A; monoclonal	29A1.4	25-3351	eBioscience
CD3e	PE-Cy7	Armenian hamster; monoclonal	145-2C11	25-0031	eBioscience
F4/80	PE	Rat IgG2A; monoclonal	BM8	12-4801	eBioscience
CD16/CD32 (FC block)		Rat IgG2A; monoclonal	93	14-0161	eBioscience

Abbreviations: APC, allophycocyanin; Cy7, cyanine 7; IgG, immunoglobulin G; PE, phycoerythrin.

Phagocytosis assay

Zymosan particles fluorescently labeled with Alexa Fluor 488 (Life Technologies) were opsonized using zymosan bioparticle opsonizing reagent (Life Technologies) at 37°C for 1 hour. Opsonized zymosan was added to BMDMs at a multiplicity of infection of 5 and incubated at 37°C 5% CO₂ for 4 hours, after which cells were washed in warm PBS four times and harvested in cold PBS with 0.5% bovine serum albumin by scraping. Samples were run on a MACSQuant flow cytometer (Miltenyi Biotec), and data were analyzed using FlowJo software (v.10). Samples were run in triplicate. The experiment was repeated independently three times, and a total of five to six mice per genotype were used.

RNA isolation and quantitative PCR

RNA from D2KO and corresponding WT BMDMs and RAW264.7 cells was isolated using the High Pure RNA isolation kit (Roche). RNA from polarized C57BL/6 BMDMs was isolated using Qiazol (Qiagen) according to the manufacturer’s instructions.

cDNA was synthesized with equal RNA input using AMV Reverse transcription enzyme with oligo d(T) primers (Roche). A cDNA synthesis reaction without reverse transcription was included as a control for genomic DNA contamination. Quantitative PCR was carried out using the Lightcycler 480 (Roche) and SensiFAST SYBR No-ROX (Bioline) and analyzed using LinReg software (linregpcr.nl) (28). The mean of the efficiency was calculated for each assay, and samples that deviated >0.05 of the efficiency mean value were excluded from the analysis (0% to 5%). Primer sequences are listed in Table 2 and include previously published and newly designed primers (4, 15–22). mRNA expression values were normalized using the geometric mean of three reference genes in accordance with the Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines (29) and corrected for plate average. Relative expression values are shown.

Cytokine measurements

Cytokines were measured in supernatant of stimulated D2KO and WT BMDM using the Cytometric Bead Array Mouse Inflammation kit (BD Biosciences) according to the manufacturer’s instructions. Samples were run in triplicate on a FACS Calibur flow cytometer (BD Biosciences). Data were analyzed using FlowJo software (version 10).

Statistical analysis

Statistical significance was tested using one-way or two-way analysis of variance followed by a post hoc test (Tukey or Sidak) or by Student two-tailed t test. Zebrafish survival data were analyzed using the log rank (Mantel-Cox) test. P values <0.05 were considered statistically significant. All tests were performed using Graphpad Prism 7.

Results

A lack of D2 impairs survival of zebrafish embryos during pneumococcal meningitis

To determine the role of D2 in proinflammatory macrophage function in vivo, we used a combination of two established models in zebrafish embryos. Successful modulation of intracellular T3 levels by inhibition of deiodinase enzymes using MO technology has been described by Walpita et al. (14). We established a D2 knockdown in zebrafish embryos using this technique. We combined this with a recently developed model for bacterial meningitis in zebrafish embryos (12) to determine the effects of changes in intracellular TH availability on macrophage function during bacterial infection in vivo. Zebrafish embryos pretreated with D2MOs or scrambled control MOs were injected with live S. pneumoniae directly into the hindbrain ventricle, resulting in bacterial meningitis. Knockdown of D2 resulted in a significant increase in mortality in zebrafish embryos exposed to pneumococcal meningitis compared with controls (Fig. 1A). This effect was rescued by the addition of T3 (Fig. 1B).

Figure 1.

D2 knockdown impairs zebrafish survival during bacterial meningitis. (A) Kaplan-Meier survival curve of zebrafish embryos pretreated with D2MO or SCMO during bacterial meningitis induced by injection of 500 CFU of S. pneumoniae into the hindbrain ventricle. Data represent 40 embryos per group. P value for log rank (Mantel-Cox) test is indicated. (B) In a separate experiment using the same model, the effect of T3 addition to the D2MO group during bacterial meningitis was assessed. Data represent 40 embryos per group. The P value for log rank (Mantel-Cox) test is indicated.

D2KO mice have unchanged bone marrow hematopoietic cell populations

To determine the effects of D2 on macrophage development, we assessed whether a lack of D2 resulted in changes in hematopoietic bone marrow populations because this could potentially affect hematopoietic cell function in vivo and ex vivo. Flow cytometric analysis of freshly isolated whole bone marrow from D2KO and WT mice demonstrated that there was no difference in the relative abundance of early hematopoietic blast cells, neutrophil precursors, monocyte precursors, differentiated neutrophils, and monocytes between the two genotypes (Fig. 2A). Thus, D2 is not essential for hematopoietic cell differentiation and development in vivo.

Figure 2.

Proinflammatory macrophage function in D2KO mice. (A) Flow cytometry of whole bone marrow was used to determine the relative amount of early hematopoietic blast cells (HSC), neutrophils/polymorphonuclear leukocyte precursor cells (PMN prec), monocyte precursor cells (mono prec), differentiated neutrophils (PMN), and monocytes (mono). All data are represented as mean ± standard error of the mean (SEM). (B and C) BMDM phagocytosis was quantified using flow cytometry after incubation of BMDMs with fluorescently labeled (Alexa Fluor 488) zymosan (zym) for 4 hours. BMDMs were gated into high, mid, low, and nonfluorescent populations as shown in (B). Median fluorescent intensity (MFI) of the “Macrophage” gate is shown. The percentage of low, mid, and high fluorescent cells in the macrophage gate is depicted. All data are represented as mean ± SEM. *P < 0.05. (D) Quantitative real-time PCR relative mRNA expression data from D2KO (open circles with dashed line) and WT (black circles with solid line) BMDMs after LPS stimulation. Data represent six animals per time point per genotype. Values are normalized to the geometric mean of mRNA expression for three reference genes (HPRT, EF1α1, and cyclophillin-b). All data are presented as mean ± SEM. (E) Cytokine concentrations measured in the supernatant of LPS stimulated D2KO (open circles with dashed line) and WT (black circles with solid line) BMDMs. Control values are from unstimulated samples at time (t) = 4 hours because t = 0 is directly after medium change, meaning there would not be any endogenously produced cytokines in the supernatant. Data represent six animals per time point per genotype. All data are presented as mean ± SEM. n.s., not significant; TNF, tumor necrosis factor.

D2KO macrophages display impaired phagocytosis and partially reduced cytokine response to LPS stimulation

Proinflammatory macrophage function was assessed in unpolarized BMDMs derived from D2KO and WT mice. Phagocytosis, a hallmark of macrophage function, was quantified using fluorescently labeled yeast particles (zymosan). D2KO BMDMs contained fewer cells with a high degree of fluorescence, a measure for the amount of particles ingested, and significantly more BMDMs with a low degree of fluorescence, indicating impaired phagocytosis compared with WT BMDMs (Fig. 2C).

Another essential element of proinflammatory macrophage function is the ability to produce and secrete proinflammatory cytokines. We assessed whether cytokine production was altered in D2KO BMDMs after stimulation with LPS. Transcriptional induction of the proinflammatory cytokine and hematopoietic growth factor granulocyte macrophage colony–stimulating factor (GM-CSF) was reduced in LPS-treated D2KO BMDMs compared with WT cells (Fig. 2D). Expression of the proinflammatory cytokines IL1-β, tumor necrosis factor (TNF) α, and IL-6 was unchanged in LPS-treated D2KO BMDMs compared with WT cells (Fig. 2D). In addition, no differences were observed in the secretion of the proinflammatory monocyte chemotactic protein (MCP-1/CCL2), TNF, and IL-6 into the supernatant of LPS-stimulated cells (Fig. 2E).

D2 knockdown does not affect polarization in a macrophage cell line

Because D2 appears to affect proinflammatory macrophage function, we assessed the effect of D2 knockdown on macrophage polarization using a murine macrophage cell line. RAW264.7 cells were transfected with an siRNA against Dio2 or a scrambled control siRNA. Transfection with an siRNA against Dio2 resulted in an average knockdown efficiency of 70% (Fig. 3A). Cells were then stimulated with LPS and interferon-γ or IL-4, polarizing the cells toward an M1 or M2 phenotype, respectively. An unstimulated control (M0) was included.

Figure 3.

Effect of D2 knockdown on polarization in a macrophage cell line. (A–C) Quantitative real-time PCR relative mRNA expression data for RAW264.7 cells transfected with an siRNA against D2 (black bars) or a control siRNA (gray bars) after polarization into M1 cells (LPS+IFN-γ), M2 cells (IL4), or unstimulated controls (M0). Values are normalized to the geometric mean of mRNA expression for three reference genes (HPRT, EF1α1, and RPL0). Data represent average value for four technical replicates from two independent experiments (n = 2). The experiment was repeated with two different siRNAs in total with similar results. Data from the siRNA Dio2-9 are shown (see Supplemental Fig. 1 for results of the additional two D2 siRNAs). All data are presented as mean ± standard error of the mean. P values for analysis of variance are indicated. Post hoc (Sidak) analysis P values: ***P < 0.001. iNOS, inducible nitric oxide; n.s., not significant.

Macrophage polarization results in the upregulation of specific genes, depending on the polarized subset (30). Activation of RAW264.7 cells with LPS and IFN-γ into M1 macrophages indeed resulted in a significant induction of the classic M1 genes Nos2 (inducible nitric oxide) and Il1b (IL-1β) (Fig. 3B) (30). However, D2 knockdown did not affect the transcriptional induction of Nos2 or Il1b. In the M2 subset, expression of the well-characterized M2 markers Arg1 (arginase 1) and Il10 (IL-10) was strongly induced (Fig. 3C) (30). The induction of these genes was unaffected by D2 knockdown (Fig. 3C). The experiment was repeated with two additional siRNAs for D2 and showed similar results (Supplemental Fig. 1).

TRα knockdown in a macrophage cell line reduces expression of M1 markers and increases expression of M2 marker arginase-1 after polarization

Because the knockdown of D2 in RAW264.7 cells was not complete, we determined the effect of TRα knockdown on macrophage polarization. Residual D2 expression means that there is still an intracellular source of T3 production. By knocking down TRα, the genomic effects of T3 are inhibited, regardless of the intracellular T3 concentrations.

RAW264.7 cells were transfected with an siRNA against Thra or a scrambled control siRNA. Transfection with a Thra siRNA resulted in a knockdown efficiency of 69% for TRα1, the active splice variant of TRα, and a knockdown efficiency of 64% for TRα2, the splice variant of the receptor that does not bind T3 (Fig. 4A) (6). Transfected cells were polarized into M1-like cells using LPS and IFN-γ [M(LPS+IFN-γ)] or M2-like cells using IL-4 [M(IL-4)]. An unstimulated control was included (M0).

Figure 4.

Effect of TRα knockdown on polarization in a macrophage cell line. (A–C) Quantitative real-time PCR relative mRNA expression data for RAW264.7 cells transfected with an siRNA against TRα (black bars) or a control siRNA (gray bars) after polarization into M1 cells (LPS+IFN-γ), M2 cells (IL4), or unstimulated controls (M0). Values are normalized to the geometric mean of mRNA expression for three reference genes (HPRT, EF1α1, and RPL0). Data represent average values for four technical replicates from two independent experiments (n = 2). The experiment was repeated with three different siRNAs in total with similar results. Data from the siRNA Thra-2 are shown (see Supplemental Fig. 2 for results from the siRNAs Thra-1 and Thra-3). All data are presented as mean ± standard error of the mean. P values for analysis of variance are indicated. Post hoc (Sidak) analysis P values: *P < 0.05; **P < 0.01; ***P < 0.001. iNOS, inducible nitric oxide; n.s., not significant.

M(LPS+IFN-γ) polarization resulted in a strong upregulation of Nos2 and Il1b in control cells. The transcriptional induction of these genes was significantly reduced in TRα knockdown M(LPS+IFN-γ) macrophages. In contrast, during M(IL-4) polarization, TRα knockdown macrophages exhibited increased induction of the M2 marker Arg1 compared with control cells (Fig. 4B). The experiment was repeated with two additional siRNAs against Thra (Supplemental Fig. 2), and an increased induction of Arg1 expression was consistently observed. One of the siRNAs against Thra also resulted in an increased expression of the M2 marker Il10 (Supplemental Fig. 2). These results indicate that a lack of intracellular T3 signaling via the TRα impairs proinflammatory M1 polarization and enhances immunomodulatory M2 polarization in macrophages. Polarization of primary murine BMDMs derived from C57BL/6 mice did not affect D2 or TRα1 expression (Supplemental Fig. 3). In M1 cells, D1 expression increased, whereas in M2 cells thyroid hormone receptor β1 (TRβ1) expression increased and TH plasma membrane transporter MCT10 decreased (Supplemental Fig. 3).

Discussion

Macrophages play essential roles in the innate immune response and tissue homeostasis. Macrophage dysfunction has been implicated in a wide range of diseases, including cancer, atherosclerosis, and various autoimmune diseases. Further insight into macrophage function can therefore have important clinical implications for a large variety of illnesses. This study indicates an important role for the regulation of intracellular T3 availability and action in macrophage function.

In zebrafish embryos, a lack of D2 impairs survival during pneumococcal meningitis in vivo. This effect is not observed when embryos that lack D2 are incubated with T3. This suggests that the regulation of intracellular T3 concentrations by D2 plays a critical role in the innate immune response to bacterial infection. D2 has previously been shown to affect macrophage function (4). In this zebrafish meningitis model, macrophages appear to be important for bacterial clearance in the later stages of inflammation (12). The only other study on the effect of D2 knockdown during inflammation in vivo found that D2 knockdown in mice exacerbated both LPS- and ventilator-induced lung injury due to increased inflammation (31). This appears contradictory to our results; however, Ma et al. (31) achieved D2 knockdown by local intratracheal administration of Dio2 siRNA, making these results difficult to interpret because there is no whole-body D2 knockdown and infiltrating macrophages could still express D2. Our results suggest that a lack of D2 activity impairs the innate immune response, which is likely due to impaired macrophage function.

To study the effects of a reduction in intracellular T3 concentrations on macrophage function in more detail, we assessed proinflammatory macrophage responses in primary bone marrow–derived D2KO macrophages. These cells were found to have reduced transcriptional induction of GM-CSF and impaired phagocytosis, which are both in accordance with earlier data in a D2 knockdown macrophage cell line (4). GM-CSF is a hematopoietic growth factor and inflammatory cytokine with a wide range of functions in different cell types (32). In macrophages, GM-CSF has a proinflammatory effect (32) and induces polarization into an M1-like phenotype (33). D2KO BMDMs also exhibit impaired phagocytosis. These data show a modulatory effect of D2 on proinflammatory macrophage function, suggesting that a reduction in intracellular T3 levels impairs the proinflammatory macrophage response.

The effects of T3 on macrophage function could be due to changes in macrophage polarization. To analyze this, we studied the effect of modulation of intracellular TH metabolism on M1 and M2 polarization in a macrophage cell line. siRNA-mediated knockdown of D2 had no effect on macrophage polarization, possibly because the knockdown was not complete and because residual D2 expression in these cells might have been sufficient for intracellular T3 production. We therefore analyzed polarization in macrophages in which the main TH receptor in this cell type, TRα (4), had been knocked down, thereby impairing the genomic action of T3. TRα knockdown macrophages exhibited a defect in polarization and favored a more anti-inflammatory phenotype. These data again support the view that reduced intracellular T3 action has an anti-inflammatory effect in macrophages.

Primary murine macrophages did not exhibit changes in D2 or TRα1 expression after polarization. We did observe an increase in D1 expression in M1 cells. However, T4 to T3 conversion by D2 is much more efficient than by D1 (Km ≈5 nM vs Km 1 to 5 µM, respectively) (34, 35). Thus, given the similar expression levels of D1 and D2, it is highly unlikely that D1 plays a substantial role in the intracellular generation of T3 in macrophages. In addition, TRβ1 expression increased in M2-polarized macrophages. Although others have reported this effect (36), TRβ1 was expressed at 100- to 200-fold lower levels than TRα1. Therefore, the significance of TRβ1 in this cell type is uncertain because TRα is the dominant TR isoform expressed.

Our data about the effects of T3 on macrophage polarization are in accordance with those from the only other available study, which found that incubation with T3 primed macrophages toward an M1-like phenotype and reduced the expression of M2 markers in M2-polarized cells (36). Our results are also consistent with the impaired function observed in macrophages obtained from TRαKO mice in an atherosclerosis model (8). These studies and our present work suggest that T3 signaling increases proinflammatory effects in macrophages. This is in contrast with a study by Furuya et al. (9) in a model of kidney obstruction in mice transplanted with TRαKO bone marrow. Macrophages isolated from these mice at the site of inflammation produced more proinflammatory cytokines, including IL-1β, than macrophages derived from mice transplanted with WT bone marrow (9). This study suggests an anti-inflammatory effect of T3 signaling via the TRα, which is in contrast to our observations. Furuya et al. (9) studied irradiated mice transplanted with TRαKO bone marrow, but it is unclear what the interplay is between WT tissue resident macrophages and transplanted TRαKO macrophages in the injured kidney. These contrasting results warrant additional studies in different in vivo models of inflammation to determine the exact role of TRα in macrophage function.

Our study suggests that in vivo macrophage function could be impaired in systemic TH deficiency, but to our knowledge there are no studies available on the effects of hypothyroidism on macrophage function. However, a number of in vitro studies have demonstrated a proinflammatory effect of hyperthyroid conditions on macrophage function. Incubation with TH was found to increase macrophage phagocytosis (37–39), macrophage reactive oxygen species production (39, 40), O₂⁻ and glucose consumption (41), and bacterial killing in vitro (39).

Observations in humans suggest a complex interplay between TH status and immune function. A study of patients with newly diagnosed Hashimoto disease found elevated levels of proinflammatory cytokines that improved after treatment with l-thyroxine (42). Higher circulating levels of proinflammatory cytokines correlated with lower serum T3 concentrations in subjects from the Leiden-85–plus cohort that were not on thyroid medication (43). The same study found higher production of proinflammatory cytokines in whole blood samples stimulated with LPS to be associated with higher circulating T3 concentrations, indicating that there is a discrepancy between the effects of thyroid status on baseline cytokine levels and induced cytokine production (43). In contrast, a positive correlation between serum TH concentrations and markers for inflammation was described in a cohort of healthy 55- to 70-year-old subjects (44). These observations illustrate the complex nature of the interaction between the endocrine and immune systems at the whole organism level.

The presence of the common nucleotide polymorphism Thr92Ala in the human D2 gene, which results in decreased conversion of T4 to T3 (45), was associated with increased protection from severe sepsis (31). The tissue injury observed in severe sepsis is thought to be in part due to an overactivation of the proinflammatory host response (46). This role for TH metabolism in macrophage function could have important clinical consequences for patients with altered thyroid status.

In summary, our data suggest that low intracellular T3 action has an anti-inflammatory effect by influencing macrophage polarization. We conclude that intracellular T3 availability and action, which are controlled by D2 and TRα, respectively, play a crucial role in macrophage function. This study provides critical insights into the role of TH in macrophage function and further elucidates the clinically relevant link between the endocrine and innate immune systems.

Glossary

Abbreviations:

BMDM

bone marrow–derived macrophage

CFU

colony-forming unit

D2

type 2 deiodinase

D2KO

type 2 deiodinase knockout

D2MO

type 2 deiodinase morpholino

DMEM

Dulbecco’s modified Eagle medium

FCS

fetal calf serum

GM-CSF

granulocyte macrophage colony–stimulating factor

IFN

interferon

IL

interleukin

M-CSF

macrophage colony stimulating factor

MO

morpholino

PBS

phosphate-buffered saline

PCR

polymerase chain reaction

RRID

Research Resource Identifier

SCMO

scrambled control morpholino

siRNA

small interfering RNA

T3

triiodothyronine

T4

thyroxine

TH

thyroid hormone

TNF

tumor necrosis factor

TR

thyroid hormone receptor

TRαKO

thyroid hormone receptor α knockout

TRβ1

thyroid hormone receptor β1

v/v

volume-to-volume ratio

WT

wild-type

w/v

weight-to-volume ratio