Counter-regulatory paracrine actions of FGF-23 and 1,25(OH)₂D in macrophages

Abstract

Mechanisms underlying the association between fibroblastic growth factor 23 (FGF-23) and inflammation are uncertain. We found that FGF-23 was markedly up-regulated in LPS/INF-γ-induced proinflammatory M1 macrophages and Hyp mouse-derived peritoneal macrophages, but not in IL-4-induced M2 anti-inflammatory macrophages. NF-κB and JAK/STAT1 pathways mediated the increased transcription of FGF-23 in response to M1 polarization. FGF-23 stimulated TNF-α, but not IL-6, expression in M0 macrophages and suppressed Arginase-1 expression in M2 macrophages through FGFR-mediated mechanisms. 1,25(OH)₂D stimulated Arginase-1 expression and inhibited FGF-23 stimulation of TNF-α. FGF-23 has proinflammatory paracrine functions and counter-regulatory actions to 1,25(OH)₂D on innate immune responses.

Fibroblastic growth factor 23 (FGF-23) is principally produced by osteoblasts/osteocytes in bone and acts as a circulating hormone to regulate phosphate transport and vitamin D metabolism in the kidney [1]. FGF-23 and vitamin D participate in a systemic counter-regulatory endocrine loop, where 1,25(OH)₂D stimulates FGF-23 production in bone and FGF-23 suppresses 1,25(OH)₂D production and increases its degradation by the kidney. Pathologically, excess FGF-23 causes acquired and hereditary hypophosphatemic disorders, such as X-linked hypophosphatemic rickets (XLH) [2], whereas FGF-23 deficiency leads to tumoral calcinosis caused by hyperphosphatemia and elevated 1,25 (OH)₂D levels [3]. FGF-23 differs structurally from locally acting paracrine FGFs by the presence of a novel C-terminus that binds to FGFR/α-Klotho binary complexes that constitute the obligate FGF-23 receptor in target tissues [4].

α-Klotho gene has two transcripts that encode a long Type I transmembrane [TM] protein containing KL1 and KL2 domains and a short secreted protein containing only a single KL domain [5]. The ~ 130 kDa transmembrane protein is an obligate coreceptor for binding of FGF-23 to FGFRs [4,6]. Ectodomain shedding by ADAM10 and ADAM17 generates a circulating α-Klotho isoform that lacks the TM domain [7]. The short secreted ~ 60 kDa isoform (hereafter referred to as s-KL) is generated by the alternative spliced transcript. Similar sized soluble KL1 and KL2 fragments are also generated by additional post-translational cleavage of the shed isoform [5,8]. The ~ 60 kDA s-KL gene product emerged during evolution before FGF-23 and likely has FGF-23 independent functions, including antiaging, anti-inflammatory, and anti-fibrotic effects due to actions of secreted forms of KL to inhibit Wnt, IGF-1, and TGF-β signaling [9–13].

Secondary increases in FGF-23 are observed in chronic kidney disease (CKD) in an attempt to restore phosphate homeostasis. In patients with CKD, elevated FGF-23 levels are also associated with increased inflammation, cardiovascular mortality, and progression of kidney disease, through mechanisms that are not fully elucidated [14–21]. In contrast, treatment of CKD patients with vitamin D analogs is associated with reduced mortality, whereas vitamin D deficiency is linked to increased all-cause mortality and adverse cardiovascular and renal outcomes in CKD [22]. The salutary effects of vitamin D may be due to its role to modulate innate and adaptive immune responses through its action to modulate the anti-inflammatory and antifibrotic effects of macrophages [23]. Indeed, macrophages can be polarized into either proinflammatory M1 and anti-inflammatory M2 phenotypes in tissues [24,25]. Cyp27b1, the enzyme that produces intracrine 1,25(OH)₂D, is also present in macrophages and acts locally to promote anti-inflammatory effects [26]. 1,25(OH)₂D is also a potent stimulator of FGF-23 expression [27]. If FGF-23 has adverse effects in CKD, the paradox of why stimulation of FGF-23 production by active vitamin D analogs does not negate the positive effects of vitamin D remains unexplained. As 1,25(OH)₂D and FGF-23 participate in a counter-regulatory endocrine network controlling systemic mineral metabolism, we asked if these two factors also have paracrine counter-regulatory effects on proinflammatory and anti-inflammatory macrophage responses might explain these apparent contradictory effects.

In the current study, we explored the relationship between FGF-23 expression and function in classically and alternatively activated RAW264.7 macrophages [28]. We found a marked increase in FGF-23 in classically activated M1 macrophages and proinflammatory effects of FGF-23 on M0 and M2 macrophages in vitro. 1,25(OH)₂D blocked these proinflammatory effects of FGF-23, thereby creating a counter-regulatory paracrine network regulating immune responses.

Methods

Cell culture

RAW264.7 macrophages cells were obtained from American Type Culture Collection. Murine peritoneal macrophages were also isolated from wild-type and Hyp mice, using 3% Brewer thioglycollate medium [29] as described previously [30]. All animal research was conducted according to guidelines provided by the National Institutes of Health and the Institute of Laboratory Animal Resources, National Research Council. The University of Tennessee Health Science Center’s Animal Care and Use Committee approved all animal studies (Protocol number: 12-168.0). RAW264.7 macrophages (3–5 × 10⁴ cells) were cultured in 6-cm diameter tissue culture plates in Dulbecco’s modified Eagle’s medium/F12 (Life technologies, Grand Island, NY, USA) with 10% fetal calf serum at 37 °C in the presence of 5% CO₂ in a humidified incubator. RAW264.7 macrophages were induced to polarize into either classically activated M1 macrophages or alternative activated M2 macrophages by culturing cells for 24 h in the presence of lipopolysaccharide (LPS) (10 ng·mL⁻¹) and interferon gamma (IFN-γ) (10 ng·mL⁻¹), or IL-4 (50 ng·mL⁻¹), respectively. Peritoneal macrophages were isolated from wild-type and Hyp mice (n = 8) and cultured in the medium containing 10% FBS for 3 days (~ 80% confluent) before being used in the studies.

Promoter analysis

To assess FGF-23 promoter activity, we transiently transfected RAW264.7 macrophages with a FGF-23 promoter/firefly luciferase reporter construct (p600Fgf23-Luc). p600Fgf23-Luc consists of a 600 bp 5′-flanking region DNA from −600 to −1 relative to the translation start site ATG that was subcloned into a pGL3-Basic vector (Promega, Madison, WI, USA) between KpnI and Hind III restriction sites to create [27]. p600Fgf23-Luc plasmid DNA was introduced into RAW264.7 cells using cationic liposomes (LipofectAMINE2000; Life technologies). Transfections (0.25 μg of FGF-23 promoter plasmid DNAs) were carried out for 16–18 h, and then the cells were washed twice with phosphate-buffered saline (PBS). To standardize the transfection efficiency, 0.1 μg of pRL-CMV vector (pRL Renilla reniformisluciferase control reporter vector; Promega) was cotransfected in all experiments. Cells were harvested 48 h after transfection and lyzed in 50 μL of reporter lysis buffer (Promega). A luciferase assay (20 μL of cell lyzed) was performed using a dual luciferase assay kit (Promega), and activity was measured with an Optocomp 1 luminometer (MGM Instruments, Inc., Hamden, CT, USA). Promoter activity (mean ± SD of triplicate samples in relative fold changes) is represented by reporter activity normalized to pRL-CMV control.

To assess the effect of various experimental manipulations on FGF-23 promoter activity, transfected cells were incubated in fresh medium containing 10% fetal calf serum in the presence of indicated drug treatment as described in the Results. The following drugs and concentrations were used: LPS (10 ng·mL⁻¹; Sigma-Aldrich, St Louis, MO, USA), IFN-γ (10 ng·mL⁻¹, R&D), IL-4 (50 ng·mL⁻¹; Sigma-Aldrich), Bay 11-7082 (10 μM; Sigma-Aldrich), Flu-baratine (50 μM; Selleckchem), 1,25(OH)₂D (10 nM; Sigma-Aldrich, Houston, TX, USA), recombinant human FGF-23 (rFGF-23, 20 ng·mL⁻¹; Sigma-Aldrich), FGFR1 inhibitor PD173074 (10 nM; Sigma-Aldrich), ERK1/2 inhibitor U0126 (10 μM; Cell signaling Technology, Danvers, MA, USA). Macrophages were exposed to the above agents or vehicle controls for 24 h before cell harvesting.

For studies that examined the effects of FGF-23 on different polarized macrophages, we administered rFGF23 to RAW264.7 macrophages simultaneously with either vehicle, LPS/IFN-γ or IL-4 to maintain M0 and induce M1 or M2 polarization, respectively. For inhibitor studies, we pretreated with inhibitors 2 h before administration of rFGF-23, or LSP + IFN-γ, respectively. All experiments were performed in the presence of 10% FBS.

FGF-23 functional assays

For FGF-23-mediated activation of FGFR/α-Klotho complexes, the HEK-293 cells were transiently transfected with α-Klotho membrane or secreted form along with Elk1-Gal4 luciferase reporter system and Renilla luciferase-null as internal control plasmid by electroporation using Cell Line Nucleofector Kit R according to the manufacturer’s protocol (Amaxa, Inc., Gaithersburg, MD, USA). About 48 h after transfection, the transfected cells were treated with 1 nM FGF-23 from different macrophages for 6 h. The cells were then lyzed in 1× passive reporter lysis buffer (Promega), and luciferase activities were measured using an Autolumat Luminometer (Wallac-Berthold, Gaithersberg, MD, USA) and Promega Dual-Luciferase^® Reporter Assay System. Data represent results of at least three separate experiments.

Measurement of cytokines and FGF-23 in the cell culture medium

RAW264.7 macrophages (3–5 × 10⁴ per well) were seeded in six-well plate and cultured in medium containing 10% fetal calf serum. When cells reached 80% confluence, either rFGF-23 (20 ng·mL⁻¹), or LPS (10 ng·mL⁻¹) + IFN-γ (10 ng·mL⁻¹) were added to the culture medium. After 24-h incubation with rFGF-23 or LPS + IFN-γ, the culture media were collected and centrifuged for 10 min at 1000 g to remove any particulate material. Samples were immediately assayed or stored at −80 °C. IL-2, IL-4, IL-5, IL-6, IL-10, IL-12, IL-13, IL-17A, IL-23, INF-γ, TNF-α, and TGF-β were measured in the samples by ELISA following manufacturer’s instructions (Multi-Analyte ELISArray Handbook; Qiagen, Valencia, CA, USA). Full-length FGF-23 protein levels were also measured in media derived from RAW264.7 macrophages using an FGF-23 ELISA kit (Kainos Laboratories, Tokyo, Japan) as previously described [27].

RT-PCR

Total RNA was extracted from RAW264.7 macrophages or mouse peritoneal macrophages using an RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. cDNAs were synthesized by using a iScript™ cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA). RT-PCR was performed to examine the FGF-23 mRNA, FGFR1, Klotho, TNF-α, IL-6, IL-10, Arg-1 in the RAW264.7 macrophages using a pair of primers as following. FGF23: 5′-CAACTGGGGAAGCCTGACC-3′ (forward), 5′-CCTTCGAGTCATGGCTCCTG-3′ (reverse). α-Klotho: 5′-AGCGATAGTTACAACAAC-3′ (forward), 5′-GCATTCTCTGATATTATA-GTC-3′ (reverse), s-Klotho: 5′-TTGCTGGGTTCCCTTTGTGAGGAA-3′ (forward), 5′-AACCACTGAGCCAGACTCCA-ACAT-3′ (reverse). IL-6: 5′-AGCCAGAGTCCTT-CAG-AGAGA-TAC-3′ (forward), 5′-AATTGGATGGTCTTGGTCCTTAGC-3′ (reverse), TNF-α: 5′-GCTCTTCTGTCTACTGAA-CTTCGG-3′ (forward), 5′-ATGATCTGAGTGTGAGGGTC-TGG-3′ (reverse), Arg-1: 5′-ATGCTCACACTGACATCAACACTC-3′ (forward), 5′- CTCTTCCATCACCTTGCCAATCC-3′ (reverse), IL-10: Cyp24a1: 5′-TGGGAAGATGATG-GTGACCC-3′ (forward), 5′-ACTGTTCCTTTGGGTAGCGT-3′ (reverse), Cyp27b1: 5′-ACACTTCGCACAGTTTACG-3′ (forward), 5′-TTAGCAATCCGCAAGCAC-3′ (reverse). Relative expression values were evaluated with the $2^{-\mathrm{\Delta }\mathrm{\Delta }{C}_t}$ method using GAPDH as housekeeping gene (forward primer: 5′-CACCACCAACTGCTTAGCC-3′, and reverse primer: 5′-TGGCATGGACTGTGGTCA-3′).

Western blot analysis

RAW264.7 macrophages or mouse peritoneal macrophages from wild-type and Hyp mice were cultured in 6-cm dishes as described above. Macrophages cytoplasmic protein was isolated by using an M-PER Cytoplasmic Extraction kit (Thermo Scientific, Rockford, IL, USA). Samples were quantified and stored at −80 °C until use. For electrophoresis, samples were prepared by mixing 3× SDS loading buffer (Cell Signaling) with 1× DTT. About 50 μg of protein were loaded onto NuPAGE 4–12% Bis-Tris Gel (Invitrogen, Carlsbad, CA, USA). Proteins were separated at 150 V for 60 min and transferred to nitrocellulose membrane (Invitrogen). Membranes were blocked with Superblock blocking buffer in TBST (Thermo Scientific) for 30 min and then incubated with primary antibody (FGF-23, ERK1/2, pERK1/2, 1 : 1000; Cell Signaling Technology. Klotho antibody 1 : 1000, a gift for Gwen King at UAB) with gentle agitation overnight at 4 °C. After three washes with TBST (15 min once and 2 × 5 min), membrane was incubated with secondary antibody in Superblock blocking buffer at room temperature for 1 h. Membrane was then washed four times (15 min and 3 × 5 min) and subjected to ECL (Thermo Scientific) and analyzed with the FOTO/Analyst Luminary/FX imaging workstation (FOTODYNE INCORPORATED, Hartland, WI, USA). Western blot using GAPDH antibody (Santa Cruz Biotechnology, Dallas, TX, USA) was used as internal control of protein loadings.

Statistics

We evaluated differences between two groups by unpaired t-test and multiple groups by one-way ANOVA followed by Tukey’s post-test for pairwise comparisons, as appropriate. All values are expressed as means ± SD. All statistical tests are two-sided with an alpha of 0.05 as the significance threshold. All computations were performed using GRAPHPAD PRISM5 (GraphPad Software Inc. La Jolla, CA, USA).

Results

Macrophage polarization results in proinflammatory and anti-inflammatory phenotypes

RAW264.7 macrophages were induced to polarize into either classically activated M1 macrophages (Fig. 1A–F) or alternative activated M2 macrophages (Fig. 1G–L) by culturing cells for 24 h in the presence of LPS (10 ng·mL⁻¹) and IFN-γ (10 ng·mL⁻¹), or IL-4 (50 ng·mL⁻¹), respectively. IL-6 and TNF-α expression were used as markers of proinflammatory M1 macrophages, and IL-10 and Arginase 1 (Arg-1) were used as markers of M2 macrophages. Treatment with LPS and IFN-γ increased the proinflammatory M1 markers IL-6 and TNF-α messenger RNA levels as compared to controls (P < 0.05 vs vehicle-treated control) (Fig. 1A,B), but had little effect on IL-10 and Arg-1 expression (Fig. 1C,D). In contrast, treatment of M0 macrophages with IL-4 induced polarization to alternatively activated M2 phenotype, as evidenced by increased expression of Arg-1 and IL-10 (P < 0.05 vs vehicle-treated control) (Fig. 1I,J), but not IL-6 or TNF-α (Fig. 1G,H).

Fig. 1.

Expression of M1 or M2 markers in LPS/IFN-γ, or IL-4-polarized RAW264.7 macrophages. RAW264.7 macrophages were polarized to M1 macrophages, or M2 macrophages in the presence of LPS (10 ng·mL⁻¹) with IFN-γ (10 ng·mL⁻¹), or IL-4 (50 ng·mL⁻¹) for 24 h, respectively. Transcriptions of IL-6 (A) and TNF-α (B) M1 macrophages markers were determined by real-time PCR. Transcription of Arg-1 (C), IL-10 (D), Cyp27b1 (E), and Cyp24a1 (F) in M1 macrophages. Transcription of IL-6 (G) and TNF-α (H) in IL-4-polarized M2 macrophages. Up-regulation of Arg-1 (I), IL-10 (J), Cyp27b1 (K), and down-regulation of Cyp24a1 (L) expression in IL-4-polarized M2 macrophages. Data are expressed as the mean ± SD (n = 3). Unpaired t-test (*P < 0.05 vs Veh control).

Macrophage polarization also affected enzymes involved in vitamin D metabolism. We found that expression of Cyp27b1 [31], which converts 25(OH)D to 1,25(OH)₂D, was increased 7-fold in anti-inflammatory M2 macrophages compared to that of M0 macrophages (P < 0.05) (Fig. 1K), and was increased 2-fold in proinflammatory M1 macrophages compared to M0 macrophages (P < 0.05) (Fig. 1E). Cyp24a1, which initiates the degradation of both 25(OH) and 1,25 (OH)₂D [31], was decreased in both M1 and M2 macrophages (P < 0.05 vs vehicle-treated controls, respectively) (Fig. 1F,L). The ratio of vitamin D regulatory enzymes favoring the local production of 1,25 (OH)₂D was greater in M1 than M2 macrophages.

M1 proinflammatory macrophages express FGF-23

Ensuing studies examined if polarization of macrophages affected FGF-23 expression. We found a dramatic induction of FGF-23 by LPS and IFN-γ in M1-polarized RAW264.7 macrophages. LPS and IFN-γ stimulated FGF-23 message levels by 80-fold as compared to vehicle-treated control (Fig. 2A). In contrast, an IL-4 induction of M2 activation resulted in a 5-fold stimulation of FGF-23 message expression (Fig. 2A). To determine if increased FGF-23 message expression was mediated by increased gene transcription, we transiently transfected a FGF-23 promoter/luciferase construct into RAW264.7 macrophages and treated with either LPS/IFN-γ_, or IL-4. LPS and IFN-γ stimulated FGF-23 promoter activity in M1 macrophages by nearly 30-fold vs vehicle control (Fig. 2B), whereas IL-4 stimulated FGF-23 promoter activity approximately 3-fold vs vehicle control (Fig 2B). LPS and IFN-γ resulted in increased FGF-23 protein expression in total cell lysates (Fig. 2C) and conditioned media (CM) (Fig. 2D) in M1 macrophages. Consistent with the smaller increment in FGF-23 message and promoter activity in M2 macrophages, only a faint band corresponding to the FGF-23 protein was found in M2-polarized macrophages by western blot analysis (Fig. 2A–C). In M1 macrophages there was also evidence for both full-length FGF-23 and a lower molecular weight FGF-23 fragments, consistent with degradation of FGF-23. The presence of intact FGF-23 was confirmed by measurable concentrations of intact FGF-23 in CM collected for 24 h, as assessed by ELISA assay that detects only full-length FGF-23. FGF-23 concentrations, were significantly elevated in classically activated M1 macrophages compared to controls and M2 macrophages, but the differences were much less than observed using the other detection methods, raising the possibility of FGF-23 degradation (Fig. 2E).

Fig. 2.

Up-regulation of FGF-23 expression by LPS/IFN-γ through activation of NF-κB and STAT1 pathways in M1 RAW264.7 macrophages. (A) Effect of LPS + IFN-γ or IL-4 on FGF-23 transcription in RAW264.7 macrophages. (B) Up-regulation of FGF-23 promoter activity in M1 macrophages. (C) Expression of FGF-23 protein was increased in M1 macrophages. GAPDH was used as protein loading controls. (D) Western blot analysis of conditional medium FGF-23 protein. Supernatants from LPS + IFN-γ-polarized RAW264.7 macrophages were collected and concentrated using Amicon Ultra-4 Centrifugal Filters (10 K) (Millipore, Temecula, CA, USA). (E) Condition medium intact FGF-23 levels. (F) Effect of NF-κB inhibitor Bay11-7082 (Bay11) or STAT1 inhibitor fludarabine (Flu) on LPS + IFN-γ induced FGF23 promoter activity, respectively. (G–I) Effects of 1,25(OH)₂D on FGF-23 message expression in macrophage subpopulations. 1,25(OH)₂D had no effect on FGF-23 gene transcription in M0 macrophages (G), reduced FGF-23 expression in M1 macrophages (H), and slightly decreased FGF-23 gene transcription in M2 macrophages (I). Data are expressed as the mean ± SD (n = 3). One-way ANOVA (P < 0.05) with Tukey’s post-test for multiple comparisons. Values not sharing the same superscript letter are significantly different at P < 0.05.

NF-_κB and STAT signaling pathways regulate FGF-23 gene transcription in macrophages

Next, we examined the mechanism for LPS/IFN-γ stimulation of FGF-23 gene transcription. LPS binds to toll like receptor and activates NF-_κB pathway, while IFN-γ binds to the IFN-γ receptor (IFNGR) and induces JAK/STAT1 signaling [32]. We identified both NF-_κB and STATs sites in the proximal FGF-23 promoter region. To explore the mechanisms whereby macrophage polarization stimulates FGF-23 gene transcription, we tested the effects of inhibitors of LPS and IFN-γ downstream signaling on FGF-23 promoter activity. In this regard, we treated the RAW264.7 macrophages transfected with the FGF-23 promoter/luciferase construct with LPS and IFN-γ in the presence and absence of Bay11-7082 (Bay11), a NF-_КB inhibitor, and/or fludarabine (Flu), a STAT1 inhibitor. We found that inhibition of NF-_КB pathway by Bay11 completely blocked the effect of LPS/IFN-γ-stimulated FGF-23 promoter activity and inhibited by 50% in Flu treated M1 macrophages (Fig. 2F). These findings suggest that both NF-_КB and JAK/STAT1 pathways regulate FGF-23 gene transcription in M1 macrophages. Collectively, these data demonstrate that classically activated M1 macrophages produce and release FGF-23.

We also tested effect of 1,25(OH)₂D treatment on FGF-23 message expression in M0, M1, and M2 macrophages. We found that addition of 1,25(OH)₂D had no effect on FGF-23 message expression in unstimulated M0 macrophages (Fig. 2G), but significantly decreased FGF-23 expression in both LPS and IFN-γ activated M1 and in IL-4 stimulated M2 macrophages (Fig. 2H,I).

FGF-23 regulates TNF-α in RAW264.7 macrophages

To assess if FGF-23 produced by macrophages acts as a paracrine factor, we first examined the expression of FGFR1 expression and response to exogenously administered FGF-23 in unpolarized RAW264.7 macrophages. As shown in Fig. 3A, FGFR1 was expressed in RAW264.7 macrophages. rFGF-23 activated ERK1/2 in RAW264.7 M0 macrophages in a time-dependent manner (Fig. 3B), consistent with the presence of a functional FGFR/α-Klotho signaling complex (see below). Next, we screened rFGF-23 for the ability to stimulate 11 different cytokines. rFGF-23 only stimulated TNF-α protein levels in RAW264.7 macrophages (Fig. 3C). This contrasts to the effects of LPS and IFN-γ to stimulate secretion of both IL-6 and TNF-α in classically activated M1 macrophages (Fig. 3D). PD173074, a specific FGFR1 inhibitor, blocked rFGF-23 stimulation of ERK activity in RAW264.7 M0 macrophages (Fig. 3E). Inhibition of FGFR1 signaling by PD173074 also blocked rFGF-23-induced TNF-α gene expression and TNF-α protein secretion by RAW264.7 M0 macrophages, respectively (Fig. 3F,G). FGF-23-induced expression of TNF-α in RAW264.7 M0 macrophages was also blocked by U0126, a specific inhibitor of ERK1/2 activation (Fig. 3H,I).

Fig. 3.

FGF-23 stimulates TNF-α expression via activation of FGFR1 signaling in RAW264.7 macrophages. RAW264.7 macrophages were cultured in the medium containing 10% FBS in the presence of rFGF-23 (20 ng·mL⁻¹) or LPS (10 ng·mL⁻¹) + IFN-γ (10 ng·mL⁻¹) for 24 h. Media from vehicle-treated cells as controls. Culture media were collected for measuring of different cytokines as indicated in C and D. For inhibition of FGFR signaling study, RAW264.7 macrophages were pretreated with indicated inhibitors for 2 h, and then cells were continually cultured in the presence of rFGF-23 for another 24 h before harvesting medium for measurement of TNF-α. Cells were collected for RNA isolation and RT-PCR. (A) Western blot analysis of FGFR1 expression in RAW264.7 macrophages. (B) FGF-23-induced ERK1/2 activation in RAW264.7 macrophages in a time-dependent manner. (C) Medium levels of indicated cytokines in rFGF-23-treated RAW264.7 macrophages. Unpaired t-test (*P < 0.05 vs Veh control). (D) Medium levels of indicated cytokines in M1-polarized RAW264.7 macrophages. Unpaired t-test (*P < 0.05 vs Veh control) (E) PD173074, a FGFR1 inhibitor abolished FGF23-induced ERK1/2 activation. (F, G) PD173074 blocked FGF-23-induced TNF-α gene transcription and TNF-α secretion. (H, I) U0126, an ERK1/2 inhibitor blocked FGF23-induced TNF-α gene expression and TNF-α secretion. Data are expressed as the mean ± SD (n = 3). One-way ANOVA (P < 0.05) with Tukey’s post-test for multiple comparisons. Values not sharing the same superscript letter are significantly different at P < 0.05.

1,25(OH)₂D modulates FGF-23 responses in macrophages

As 1,25(OH)₂D is known to modulate macrophage functions, we examined the interaction between 1,25 (OH)₂D and FGF-23 in M0 macrophages. TNF-α message expression in M0 macrophages was increased (3.5-fold) by treatment with rFGF-23 as compared to vehicle-treated M0 macrophages. In contrast, administration of rFGF-23 to RAW264.7 M0 macrophages in the presence of 1,25(OH)₂D blocked the stimulation of TNF-α mRNA expression (Fig. 4A). In contrast, rFGF-23 had no effect on Arg-1 mRNA expression, and did not inhibit 1,25(OH)₂D-induced Arg-1 mRNA expression in M0 macrophages (Fig. 4B).

Fig. 4.

Effects of FGF-23 and 1,25(OH)₂D on TNF-α and Arg-1 expression in different macrophage subpopulations. (A, B) Effects of FGF-23 and 1,25(OH)₂D on TNF-α gene transcription (A) and Arg-1 expression (B) in unstimulated RAW264.7 macrophages. FGF-23 stimulated TNF-α expression and 1,25(OH)₂D blocked this effect. 1,25(OH)₂D enhanced Arg-1 gene expression. (C, D) Effect of FGF-23 and/or 1,25 (OH)₂D on M1-polarized RAW264.7 macrophages. M1 macrophages were resistant to FGF-23 and 1,25(OH)₂D modulation of TNF-α gene expression (C). In M2 macrophages, 1,25(OH)₂D also enhanced Arg-1 gene transcription, while rFGF-23 inhibited IL-4-induced Arg-1 gene transcription. 1,25(OH)₂D counteracted the suppressive actions of FGF-23 on Arg-1 gene transcription in IL-4-polarized M2 RAW264.7 macrophages (D). Data are expressed as the mean ± SD (n = 3). One-way ANOVA (P < 0.05) with Tukey’s post-test for multiple comparisons. Values not sharing the same superscript letter are significantly different at P < 0.05.

The responses to FGF-23 and 1,25(OH)₂D differed in macrophages induced to polarize into M1 and M2 macrophages. In proinflammatory M1 macrophages induced by LPS and IFN-γ, neither the addition of FGF-23, 1,25(OH)₂D, nor combination of FGF-23 and 1,25(OH)₂D added at the time of the polarization stimulus affected TNF-α transcription (Fig. 4C). During M2 macrophages induction by IL-4, 1,25 (OH)₂D further increased Arg-1 mRNA expression in combination with IL-4, and FGF-23 inhibited IL-4-induced Arg-1 mRNA expression (Fig. 4D). Concomitant treatment with 1,25(OH)₂D counteracted the effects of FGF-23 to suppress IL-4-mediated Arg-1 mRNA expression (Fig. 4D) as evidenced by a net stimulation of Arg-1 expression in IL-4 activated M2 macrophages by combined treatment with 1,25(OH)₂D and FGF-23.

Intracrine metabolism of vitamin D also plays an important role in innate and adaptive immune responses [33]; therefore, we examined the effects of FGF-23 and 1,25(OH)₂D on Cyp27b1 and Cyp24a1 mRNA expression in polarized macrophages (Fig. 5A–F). Recombinant FGF-23 (rFGF-23) significantly down-regulated Cyp24a1 and up-regulated Cyp27b1 expression in M0 macrophages. (Fig. 5A,B). FGF-23 alone had no effect on Cyp24a1 in either M1 or M2 macrophages (Fig. 5C,E), and respectively decreased and increased Cyp27b1 mRNA expression in M1 and M2 macrophages (Fig. 5D,F). Interestingly, 1,25(OH)₂D treatment by itself had no effect on the expression of both Cyp24a1 and Cyp27b1 in M0 macrophages (Fig. 5A,D), M1 (Fig. 5C,D), or M2 macrophages (Fig. 5E,F). The combination of 1,25(OH)₂D and FGF-23, however, markedly up-regulated Cyp24a1 mRNA expression in M0, M1, and M2 macrophages (Fig. 5A,C,E). 1,25(OH)₂D antagonized the effect of FGF-23 to increase Cyp27b1 expression in M0 (Fig. 5B) but had no effect on FGF-23 effects on M1 or M2 macrophages (Fig. 5D,F).

Fig. 5.

Regulation of Cyp24a1 and Cyp27b1 by FGF-23 in RAW264.7 macrophages. (A) FGF-23 down-regulated Cyp24a1 in M0 macrophages. Addition of 1,25(OH)₂D reversed FGF-23 effect and up-regulated Cyp24a1 by 3-fold. (B) FGF-23 up-regulated Cyp27b1 by 8-fold and 1,25(OH)₂D blocked FGF-23 effect in M0 macrophages. 1,25(OH)₂D alone did not affect Cyp27b1 gene transcription. (C) 1,25 (OH)₂D or FGF-23 alone did not affect Cyp24a1 transcription in M1 macrophages, but combination of 1,25(OH)₂D with FGF-23 increased Cyp24a1 transcription by 4-fold in M1 macrophages. (D) FGF-23 decreased Cyp27b1 transcription in M1 macrophages. Addition of 1,25 (OH)₂D did not reverse FGF-23 effect in M1 macrophages. (E) 1,25(OH)₂D or FGF-23 alone did not affect Cyp24a1 transcription in M2 macrophages, but combination of 1,25(OH)₂D with FGF-23 increased Cyp24a1 transcription by 3-fold in M2 macrophages. (F) FGF-23 increased Cyp27b1 transcription by 2-fold in M2 macrophages. Addition of 1,25(OH)₂D did not reverse FGF-23 effect in M2 macrophages. Data are expressed as the mean ± SD (n = 3). One-way ANOVA (P < 0.05) with Tukey’s post-test for multiple comparisons. Values not sharing the same superscript letter are significantly different at P < 0.05.

α-Klotho isoform expression in macrophages

Finally, we explored the expression of α-Klotho isoforms in macrophages. We found that M0 macrophages expressed low levels of the full-length α-KL message, as determined by real-time PCR. α-KL message was up-regulated 10-fold in M1-polarized macrophages (Fig. 6A). In contrast, α-KL message expression in M2 macrophages was not significantly different from M0 controls (Fig. 6A). Expression of the secreted Klotho (s-KL) transcript was also increased ~ 9-fold in M1 macrophages (Fig. 6B), whereas M2 polarization of macrophages had little impact on s-KL expression (Fig. 6B). The threshold cycle for detecting α-KL (C_t = 35.67) and s-KL (C_t = 30.49) message in macrophages was consistent with the higher expression of the s-KL. The threshold cycle for detecting both isoforms of KL were less in kidney (α-KL C_t = 28.76) and s-Kl C_t = 21.2), indicating that macrophages express relatively less KL compared to kidney. We failed to detect the full-length 130-kDa α-Klotho protein in any macrophage subtype type, in spite of observing functional response to rFGF-23 in both M0 and M2 macrophages (Fig. 6C). Interestingly, only a 60-kDa band (s-KL) was identified in RAW264.7 macrophages by western blot analysis using an anti-KL antibody (Fig. 6C), consistent in size with s-KL, which can be generated by alternative splicing or by ectodomain shedding of α-KL. Expression of s-KL protein was up-regulated by M1 polarization induced by LPS/IFN-γ (Fig. 6C), but expression was similar in M0 and M2 macrophages.

Fig. 6.

M1-polarized macrophages expresses Klotho. (A) Up-regulation of α-Klotho in M1 macrophages. (B) Up-regulation of s-Klotho in M1 macrophages. (C) Western blot analysis of Klotho expression in during M1 polarization by LPS and IFN-γ in RAW264.7 macrophages. (D, E) Different ability of α-Klotho (designated as the transmembrane mKL containing both the K1 and K2 domains) and s-Klotho (containing only the K1 domain) to function as a coreceptor for FGFRs. Cotransfection of mKL imparts FGF-23 stimulation of Elk-GAL reporter gene activity (D). In contrast, cotransfection of s-KL does not impart FGF-23 signaling (E). Data are expressed as the mean ± SD (n = 3). One-way ANOVA (P < 0.05) with Tukey’s post-test for multiple comparisons. Values not sharing the same superscript letter are significantly different at P < 0.05.

Because of the discordance between α-KL and s-KL expression and functional responses to FGF-23, we tested the ability of the 130 kDa α-KL and 60 kDa s-KL to act as coreceptors for FGF-23 activation of FGFRs. For these studies, we transfected either α-KL and s-KL cDNA constructs into HEK-293 cells and tested the effects of FGF-23 to activate FGFR signaling. We found that transfection of the α-KL construct imparted FGF-23 activation of FGFR signaling in HEK-293 cells, whereas transfection of s-KL did not (Fig. 6D,E). These findings confirmed the obligatory role of α-KL containing the KL1 and KL2 domains in FGF-23 activation of FGFR signaling [34] and the inability of s-KL (KL1) to support FGF-23 activation of FGFRs.

Confirmation of FGF-23 expression and function in peritoneal-derived macrophages

To extend the observations in RAW264.7 macrophages to an ex vivo model, we isolated peritoneal macrophages from wild-type and Hyp mice, a mouse homolog of XLH that has elevated FGF-23 expression [2]. We found that expression of FGF-23 mRNA is significantly higher in the peritoneal macrophages from Hyp mice compared to wild-type macrophages (Fig. 7A). Western blot analysis confirmed the increased expression of the full-length FGF-23 protein and degradation fragments in Hyp-derived macrophages compared to wild-type controls (Fig. 7B). Expression of both α-KL (Fig. 7C) and s-KL (Fig. 7D) transcripts were also increased to a greater extent in Hyp macrophages compared to wild-type controls. In addition, treatment with rFGF-23 (20 ng·mL⁻¹) significantly increased TNF-α mRNA in both wild-type and Hyp- derived peritoneal macrophages, compared to vehicle-treated controls (Fig. 7E). Wild-type macrophages, similar to RAW264.7, exhibited an increase in TNF-α expression in response to LPS + IFN-γ treatment. Peritoneal macrophages from Hyp mice had higher expression of TNF-α expression at baseline and exhibited a greater response to LPS + IFN-γ-induced TNF-α expression compared to peritoneal macrophages from wild-type mice (Fig. 7F).

Fig. 7.

FGF-23, Klotho and TNF-α expression in peritoneal macrophages isolated from wild-type and Hyp mice. Peritoneal macrophages were treated with rFGF-23 (20 ng·mL⁻¹), LPS (10 ng·mL⁻¹) + IFN-γ (10 ng·mL⁻¹), or IL-4 (50 ng·mL⁻¹) for 24 h before harvesting for protein or RNA isolation, respectively. (A) Real-time PCR analysis of FGF-23 mRNA. Unpaired t-test (*P < 0.05 vs wild-type control), (B) Western blot analysis of FGF-23 protein expression, (C) Real-time PCR analysis of α-Klotho mRNA. Unpaired t-test (*P < 0.05 vs wild-type control), (D) Real-time PCR analysis of s-KL mRNA. Unpaired t-test (*P < 0.05 vs wild-type control), (E) Effect of rFGF-23 on TNF-α gene expression, (F) Effect of M1 polarization by LPS and IFN-γ on TNF-α gene expression. Data are expressed as the mean ± SD (n = 8 separate microphage isolations). One-way ANOVA (P < 0.05) with Tukey’s post-test for multiple comparisons. Values not sharing the same superscript letter are significantly different at P < 0.05.

Discussion

This study shows for the first time that FGF-23 and 1,25(OH)₂D function together in a paracrine manner to differentially regulate proinflammatory and anti-inflammatory macrophage functions in vitro. First, we found that FGF-23 is produced by macrophages and is markedly increased by polarization to M1 proinflammatory macrophages. Second, we found that macrophages not only produce FGF-23 but also express FGFR1 [35] and α-KL transcripts, thereby producing both the ligand and receptor complexes necessary for paracrine FGF-23 signaling in local inflammatory milieu. Indeed, we discovered that recombinant FGF-23 stimulated TNF-α expression in RAW264.7 M0 macrophages and suppressed Arg-1 expression in M2 macrophages through FGFR-dependent signaling pathways. Peritoneal macrophages also respond to rFGF-23 by increases TNF-α; and peritoneal macrophages derived from Hyp mice, which have increased endogenous production of FGF-23, exhibited higher basal TNF-α expression and an exaggerated TNF-α response to LPS/IFN-γ-induced M1 polarization. Third, our studies revealed that Cyp27b1, the enzyme responsible for 1,25(OH)₂D production, was increased in IL-4-induced M2 macrophages, consistent with their anti-inflammatory phenotype. In addition, using the ratio of Cyp24a1 to Cyp27b1 as an index of net intracrine 1,25(OH)₂D production, our data suggest that FGF-23 and 1,25(OH)₂D would act in concert to reduce the local production 1,25(OH)₂D by macrophages, consistent with the actions of FGF-23 to inhibit 1,25(OH)2D synthesis in human monocytes [36]. Although we did not measure 1,25 (OH)₂D production in macrophages, others have shown that FGF-23 inhibits the intracrine synthesis of 1,25(OH)₂D in human monocytes [36]. Finally, we found that 1,25(OH)₂D counteracted the effects of FGF-23 to stimulate TNF-α expression in M0 macrophages and to suppress Arg-1 expression in M2 macrophages. 1,25(OH)₂D also attenuated FGF-23 message expression in M1 and M2 macrophages. Thus, locally produced FGF-23 and 1,25(OH)₂D appear to have paracrine counter-regulatory effects on macrophage gene expression, analogous to their counter-regulatory roles as circulating factors controlling systemic mineral homeostasis (Fig. 8) [3,37].

Fig. 8.

Schematic showing the proposed paracrine actions of FGF-23 in macrophages. LPS/INF-γ activation of M1 proinflammatory macrophages results in the increased production and secretion of FGF-23. Release of FGF-23 from M1 macrophages acts as paracrine factor to stimulate TNF-α in M0 macrophages and to suppress Arginase-1 in M2 macrophages. In contrast, IL-4 induction of alternatively activated M2 macrophages does not up-regulate FGF-23 expression, but instead promotes anti-inflammatory responses though stimulation of Arginase-1 and Cyp27b1, enzymes responsible for degradation of L-arginine and intracrine production of 1,25(OH)₂D, respectively. In turn, 1,25(OH)₂D, which has anti-inflammatory and antifibrotic effects, inhibits FGF-23 production in M1 macrophages. The s-KL isoform, which has anti-inflammatory effects, is paradoxically increased in M1-activated macrophages. Thus, alterations in the local release of FGF-23 and s-KL during macrophage polarization may modulate innate immune responses.

Our studies also raise several questions that will need to be clarified in future studies. For instance, the effects of 1,25(OH)₂D to suppress FGF-23 is opposite to effects of this vitamin D analog to stimulate FGF-23 expression in bone-derived osteoblasts [27]. The mechanism whereby 1,25(OH)₂D regulates FGF-23 gene transcription is not clear; but indirect effects have been proposed and tissue-specific modulation of FGF-23 may occur, as suggested by the ability of IL-6 treatment to blunt FGF-23 up-regulation in osteoblasts [38,39]. In addition, 1,25(OH)₂D alone is a potent inducer of Cyp24a1 expression in the kidney proximal tubules and in dendritic cells [40], but 1,25 (OH)₂D stimulated CYP24a1 in macrophages only in the presence of FGF-23.

In addition, the role of FGF-23 catabolism and mechanism whereby FGF-23 activates macrophages is uncertain. We observed FGF-23 degradative fragments in M1 macrophages, consistent with expression of furin-like proprotein convertases in these cells that cleave FGF-23 [41,42]. Cleavage of FGF-23 may represent another point for regulating the activity of this factor in the inflammatory milieu. We failed to detect full-length α-Klotho protein by western blot analysis in macrophages, although the ability of FGF-23 to activate signaling in macrophages is ex post facto evidence for a functional FGFR1/α-Klotho membrane complex necessary for FGF-23 signaling. A similar discordance between α-Klotho message and protein expression and functional responses to FGF-23 is observed in the renal proximal tubule [43,44], but recent studies also suggest that FGF-23 can signal through FGFRs independent of α-KL under some circumstances [17].

Finally, the functional role of s-KL and smaller molecular weight KL isoforms, which are highly expressed in M1 macrophages, need to be defined. We found that s-KL was not capable of mediating FGF-23 bioactivity, as previously reported [44]. s-KL may have anti-inflammatory and antifibrotic effects independent of the actions of FGF-23 [45]. KL is known to inhibit TGF-β signaling and may inhibit the profibrotic effect of TGF-β released by M2 macrophage subpopulations [11]. Interestingly, both α-KL and TNF-α undergo ectodomain shedding by ADAM-17 [46], which raises the possibility that these two factors are coordinately regulated and that paracrine release of s-KL modulates inflammatory responses.

The discovery of paracrine functions of FGF-23 in macrophages brings another perspective to the clinical associations between FGF-23 and adverse outcomes. Our findings provide a molecular basis for the association between increased circulating FGF-23 and inflammation [16,35,47,48]. Indeed, infiltration of M1 macrophages promotes inflammation in the kidney, while anti-inflammatory signals favor macrophage polarization toward anti-inflammatory or profibrotic ‘M2’ phenotypes [49–52]. FGF-23 produced by M1 macrophages may modulate proinflammatory functions in vitro and participate in the innate immune response to kidney injury. Recent studies showing that LPS stimulates FGF-23 production in activated dendritic cells and macrophages in vivo [35], suggest our findings are clinically relevant.

The contradictory findings that vitamin D can promote proinflammatory M1 over the M2 phenotype under some conditions may also be explained by our data showing concordant effects of FGF-23 and 1,25 (OH)₂D on vitamin D catabolism and opposing effects on proinflammatory TNF-α responses in macrophages [53]. Differential effects on macrophages may also account for the ability of 1,25(OH)₂D improve survival while increasing FGF-23 [54].

FGF-23 regulation of proinflammatory macrophage functions provides another explanation for the association between elevated FGF-23 and adverse cardiovascular and renal outcomes. At present FGF-23-associated adverse outcomes have been attributed to either direct effects of FGF-23 on the myocardium [17,55] or to effects of FGF-23 on the kidney that result in hypertension and secondary effects on the myocardium [43]. As cardiovascular disease and renal failure progression in CKD are associated with macrophage tissue infiltration, it is possible that the local production of FGF-23 by macrophages may contribute to cardiac and renal fibrosis. Consistent with this possibility, α-KL has been detected in atherosclerotic arteries [56] and experimentally induced macrophage infiltration of the heart results in increased FGF-23 expression in myocardium [47,57].

In conclusion, although FGF-23 evolved from paracrine FGFs to become a hormone, our findings suggest that FGF-23 can also act as a paracrine factor. In this schema (Fig. 8), FGF-23 is released by proinflammatory M1 macrophages and acts locally to increase TNF-α production in M0, and to decrease intracrine production of 1,25(OH)₂D. In addition, paracrine FGF-23 inhibits M2 anti-inflammatory functions, and possibly targets other cells in inflamed and injured tissues to promote fibrosis. The net effect of FGF-23, predicted from these in vitro and ex vivo macrophage culture models, is to accelerate proinflammatory responses. In contrast, 1,25(OH)₂D produced locally by M2 macrophages, and other mononuclear cells, counteracts the proinflammatory effects of FGF-23, as well as stimulates anti-inflammatory and antifibrotic responses. The release of s-KL may further modulate M2 responses. The balance between the opposing effects of FGF-23 and 1,25 (OH)₂D may determine the overall net inflammatory and fibrotic response. Further studies to test this new schema in vivo should advance our understanding of the molecular mechanisms underlying the maladaptive effects of excess FGF-23.

Abbreviations

Arg-1

arginase 1

CKD

chronic kidney disease

CM

conditioned media

FGF-23

fibroblastic growth factor 23

Flu

fludarabine

IFNGR

IFN-γ receptor

IFN-γ

interferon gamma

LPS

lipopolysaccharide

PBS

phosphate-buffered saline

TM

transmembrane