Activation of Cardiac Fibroblast Growth Factor Receptor 4 Causes Left Ventricular Hypertrophy

Abstract

Chronic kidney disease (CKD) is a worldwide public health threat that increases risk of death due to cardiovascular complications, including left ventricular hypertrophy (LVH). Novel therapeutic targets are needed to design treatments to alleviate the cardiovascular burden of CKD. Previously, we demonstrated that circulating concentrations of fibroblast growth factor (FGF) 23 rise progressively in CKD and induce LVH through an unknown FGF receptor (FGFR)-dependent mechanism. Here, we report that FGF23 exclusively activates FGFR4 on cardiac myocytes to stimulate phospholipase Cγ/calcineurin/nuclear factor of activated T cell signaling. A specific FGFR4-blocking antibody inhibits FGF23-induced hypertrophy of isolated cardiac myocytes and attenuates LVH in rats with CKD. Mice lacking FGFR4 do not develop LVH in response to elevated FGF23, whereas knockin mice carrying an FGFR4 gain-of-function mutation spontaneously develop LVH. Thus, FGF23 promotes LVH by activating FGFR4, thereby establishing FGFR4 as a pharmacological target for reducing cardiovascular risk in CKD.

Figure 1. FGF23 activates FGFR4 and PLCγ signaling in HEK293 cells

(A) FGF23 treatment increases levels of phosphorylated PLCγ in HEK293 cells within 30 minutes, as revealed by immunoprecipitation of endogenous PLCγ and Western blot analysis of eluates with anti-phospho-PLCγ. FGF23 does not increase phosphorylation of FRS2α or ERK1/2, as shown by Western blot analyses of total protein extracts. In contrast, FGF23 induces phosphorylation of FRS2α and ERK1/2, but not PLCγ, in cells that overexpress FLAG-tagged α-klotho (αKL). HEK293 cells that overexpress β-klotho (βKL) respond to FGF23 similar to untransfected cells. FGF2 induces phosphorylation of FRS2α and ERK1/2, regardless of presence of α- or β-klotho. Overexpression of FLAG-tagged α- or β-klotho does not induce phosphorylation of FRS2α, ERK1/2 or PLCγ in the absence of FGF treatment. GAPDH serves as loading control. (B) The four different FGFR isoforms (c splice variants of FGFR1-3) were overexpressed as V5-tagged fusion proteins in HEK293 cells for 2 days, serum starved overnight and treated with FGF23 for 30 minutes followed by anti-V5 immunoprecipitation from total protein extracts. Endogenous PLCγ co-precipitates only with FGFR4. In the absence of FGF23 treatment or if the FGFR4-Y751F mutant form is overexpressed, PLCγ cannot be detected in the eluate. Immunoprecipitations with control antibody (Ctrl-IgG) do not result in purification of V5-FGFR4 or PLCγ. See also Figure S1.

Figure 2. FGFR4 is expressed in cardiac myocytes

(A) Immunoprecipitation of FGFR4 from total heart tissue (H) and isolated ventricular myocytes of newborn wild-type and FGFR4 knock-out (FGFR4^−/−) mice (NMVM) followed by immunoblotting for FGFR4 reveals the expected 110 kDa signal of FGFR4 in wild-type but not FGFR4^−/− mice. The 50 kDa signal is the IgG heavy chain of anti-FGFR4. (B) Immunocytochemical analysis shows a striated localization pattern of FGFR4 (red) in wild-type NMVM that is absent in FGFR4^−/− cells. In wild-type NMVM, FGFR4 is present only in cardiac myocytes but not other co-purified cells, as evident by FGFR4 labeling of α-actinin expressing cells (green). DAPI (blue) identifies nuclei (original magnification, ×40; scale bar: 50 µm). (C) Representative dot blots of 3 independent flow cytometric analyses of single cell suspensions from adult wild-type mouse hearts. Labeling without primary antibodies is used as negative control (−). Numbers refer to percentages of quadrant analysis. FGFR4 is present in cardiac troponin-T (TNNT2)-positive myocytes, whereas TNNT2-negative non-myocytes show no FGFR4 labeling. (D) Immunoperoxidase labeling of fresh human cardiac tissue with anti-FGFR4 shows a positive signal that is absent if only secondary antibody is used (Ctrl). As positive control, human liver tissue shows high levels of FGFR4 expression. (E) Immunohistochemical analysis of human cardiac tissue with anti-FGFR4 (green) shows labeling of the myocardium, as visualized by differential interference contrast (DIC), and of cardiac myocytes, as identified by co-labeling with anti-α-actinin (red). Fluorochrome-conjugated secondary antibody only was used as negative control (Ctrl). DAPI (blue) shows nuclei. As positive control, anti-FGFR4 shows strong immunolabeling of hepatocytes in human liver tissue. See also Figure S2.

Figure 3. FGF23 activates FGFR4 in cultured cardiac myocytes

(A) NRVM were co-treated with FGF23 or FGF2 and an FGFR4 blocking antibody (anti-FGFR4) for 30 minutes followed by anti-PLCγ immunoprecipitation from total protein extracts. The amount of co-precipitated FGFR4 is increased in cells that were treated with FGF23 compared to FGF2-treated or untreated cells. In the presence of anti-FGFR4, levels of FGFR4 in anti-PLCγ eluates are reduced. Immunoprecipitation with control antibody (Ctrl-IgG) does not result in purification of PLCγ or FGFR4. (B) FGF23 treatment of NFAT-luciferase reporter NRVM for 2 hours significantly increases luciferase activity compared to vehicle- and FGF2-treated cells. The effect is blocked in cells that were co-treated with cyclosporine A (CsA) or anti-FGFR4, but not with the FGFR1-3 inhibitor, AZD4547. Values represent fold change in relative light units ± SEM compared with vehicle treated cells; *P < 0.0001. (C) NRVM were treated with FGF23 or FGF2 for 30 minutes followed by immunoblot analysis of total protein extracts. Only FGF2 causes an increase in levels of phospho-FRS2α and phospho-ERK. This effect is not reduced in the presence of anti-FGFR4. GAPDH serves as loading control. See also Figure S3 and S4.

Figure 4. FGF23-mediated hypertrophic growth of isolated cardiac myocytes requires FGFR4

(A) Immunofluorescence confocal images of isolated NRVM that were co-treated with FGF2 or FGF23 and anti-FGFR4 for 48 hours. Myocytes are labeled with anti-α-actinin (red), and DAPI (blue) identifies nuclei (original magnification, ×63; scale bar: 50 µm). NRVM treated with FGF23 or FGF2 appear larger than PBS-treated control cells. Anti-FGFR4 blocks the effect of FGF23 but not FGF2. (B) Compared with PBS-treated control cells, 48 hours of treatment with FGF23 or FGF2 significantly increases cross-sectional area of isolated NRVM (mean ± SEM). Co-treatment with the pan-FGFR inhibitor, PD173074, prevents any increase in area regardless of the FGF. Inhibition of FGFR1-3 by AZD4547 prevents FGF2- but not FGF23-induced hypertrophy. An FGFR4-specific blocking antibody (anti-FGFR4) prevents FGF23- but not FGF2-induced hypertrophy (150 cells per condition; n = 3 independent isolations of NRVM; *P < 0.0001 compared with vehicle). (C) Quantitative PCR analysis of NRVM 48 hours after co-treatment with FGF2 or FGF23 with PD173074 or anti-FGFR4. ANP and BNP expression are significantly elevated in FGF2 and FGF23-treated cells. PD173074 blocks the effects of both FGF2 and FGF23, but anti-FGFR4 blocks only the effects of FGF23 (n = 3 independent isolations of NRVM; *P < 0.05 compared with vehicle). (D) Immunofluorescence confocal images of NMVM isolated from wild-type or FGFR4^−/− mice that were treated with FGF2, FGF23 or AngII. Myocytes are labeled with anti-α-actinin (red), and DAPI (blue) identifies nuclei (original magnification, ×63; scale bar: 50 µm). Wild-type NMVM treated with FGF23, FGF2 or AngII appear larger than PBS-treated control cells. FGFR4^−/− NMVM increase in size in response to FGF2 and AngII, but are protected from the effect of FGF23. (E) FGF23, FGF2 and AngII induce significant increases in cross-sectional area of wild-type NMVM, but only FGF2 and AngII increase the size of NMVM isolated from FGFR4^−/− mice (150 cells per condition; n = 3 independent isolations of NMVM; *P < 0.01 compared with vehicle). (F) NMVM were treated with FGF23, FGF2 or AngII for 48 hours followed by immunoblot analysis with anti-β-MHC and anti-GAPDH, and signal quantification by densitometry. FGF23 induces increased β-MHC expression in wild-type but not FGFR4^−/− NMVM, whereas FGF2 and AngII increase β-MHC in NMVM of both genotypes (values represent fold change ± SEM compared with vehicle-treated cells; n = 4 independent isolations of NMVM; *P < 0.05). (G) Representative immunoblots for the quantification presented in (f). GAPDH shows equal protein loading.

Figure 5. Diet-induced elevation of serum FGF23 does not result in LVH in FGFR4^−/− mice

Compared to normal chow (Ctrl), high phosphate diet (P_i) significantly increases serum phosphate (A) and FGF23 (B) levels in wild-type (WT) and constitutive FGFR4 knock-out (FGFR4^−/−) mice, but does not affect blood urea nitrogen (BUN) levels (C). (D) Representative gross pathology of mid-chamber (MC) sections of the heart (hematoxylin and eosin stain; original magnification, ×5; scale bar: 2 mm), and wheat germ agglutinin (WGA)-stained sections (original magnification, ×63; scale bar: 50 µm) demonstrate LVH, and Picrosirius Red stainings (original magnification, ×10; scale bar: 100 µm) show myocardial fibrosis in wild-type mice on high phosphate diet. FGFR4^−/− mice on high phosphate or wild-type and FGFR4^−/− mice on normal diet do not develop LVH or fibrosis. Wild-type mice on high phosphate diet develop significant increases in left ventricular (LV) wall thickness (E), cross-sectional area of individual cardiac myocytes (F), and myocardial deposition of collagen fibers (G), that do not occur in FGFR4^−/− mice on high phosphate diet or wild-type and FGFR4^−/− mice on normal diet. All values are mean ± SEM (n = 4–11 mice per group; n = 100 cells per group for WGA analysis; *P < 0.001 compared with mice form the same genotype on normal chow). See also Figure S5.

Figure 6. Pharmacological inhibition of FGFR4 attenuates LVH in a rat model of CKD

Compared with sham nephrectomy, 5/6 nephrectomy (Nx) in rats results in increased ratio of heart weight to tibia length (A) and increased left ventricular mass (LVM) by echocardiography (B) at day 14 post-surgery. Each of these effects is attenuated by administering an FGFR4-specific blocking antibody (anti-FGFR4) (*P < 0.05 compared with Sham). (C) Representative gross pathology sections (hematoxylin and eosin stain; original magnification, ×2.5; scale bar: 2 mm), wheat germ agglutinin (WGA)-stained sections (original magnification, ×63; scale bar: 50 µm), and Picrosirius Red stainings (original magnification, ×10; scale bar: 100 µm) from the left ventricular mid-chamber (MC) at day 14 after 5/6 nephrectomy. Compared with vehicle, anti-FGFR4 attenuates the effects of 5/6 nephrectomy to increase left ventricular (LV) wall thickness (by gross pathology, D), cross-sectional area of individual cardiac myocytes (E), and myocardial fibrosis (F). All values are mean ± SEM (n = 6–14 rats per group; *P < 0.001 compared with Sham; #P < 0.01 compared with 5/6 nephrectomy treated with vehicle). Quantitative PCR analysis of cardiac tissue shows that expression levels of hypertrophic markers (ANP, BNP and β-MHC; G) and of some fibrotic markers (TIMP metallopeptidase inhibitor 1, Timp1; collagen type 1 alpha 1, Col1a1; collagen type 3 alpha 1, Col3a1; H) are significantly elevated in 5/6 nephrectomy rats that were treated with vehicle but not in anti-FGFR4 treated rats. (n = 6 rats per group; *P < 0.05 compared with Sham). Echocardiography shows no significant differences in ejection fraction among the three groups (I), but anti-FGFR4 significantly improves diastolic function (E/E’) in 5/6 nephrectomy rats compared to vehicle-treatment (J) (n = 6–8 rats per group; *P < 0.05 compared with vehicle-injected 5/6 nephrectomy). See also Figure S5 and S6.

Figure 7. Knock-in mice with an FGFR4 gain-of-function mutation spontaneously develop LVH

(A) Six-month old homozygous knock-in mice carrying the Arg385 substitution in FGFR4 (FGFR4Arg/Arg385) manifest significant increases in the ratio of heart weight to tibia length compared with wild-type littermates (FGFR4Gly/Gly385) (mean ± SEM; n = 9–12 mice per group; *P < 0.01). (B) Representative gross pathology of mid-chamber (MC) sections of the heart (hematoxylin and eosin stain; original magnification, ×5; scale bar: 2 mm), and wheat germ agglutinin (WGA)-stained sections (original magnification, ×63; scale bar: 50 µm) demonstrate LVH in FGFR4Arg/Arg385 mice but lack of myocardial fibrosis by Picrosirius Red staining (original magnification, ×100; scale bar: 100 µm). Compared with wild-type, FGFR4Arg/Arg385 mice develop significant increases in left ventricular (LV) wall thickness (mean ± SEM; n = 9–12 mice per group; *P < 0.01; C) and cross-sectional area of individual cardiac myocytes (mean ± SEM; n = 100 cells per group; *P < 0.0001; D), but no difference in collagen deposition (mean ± SEM; n = 9–12 mice per group; E). (F) Quantitative PCR analysis of total hearts from mice reveals significantly elevated BNP expression in 6-month old FGFR4Arg/Arg385 mice compared with wild-type (n = 4–6 mice per group; *P < 0.01). (G) Quantitative PCR analysis of cardiac tissue shows that expression levels of some fibrotic markers (TIMP metallopeptidase inhibitor 1, Timp1; collagen type 1 alpha 2, Col1a2; collagen type 5 alpha 1, Col5a1) are significantly elevated in FGFR4Arg/Arg385 mice compared to wild-type mice (n = 6 mice per group; *P < 0.01). Echocardiography shows that compared to wild-type mice, relative wall thickness (H), ejection fraction (I) and diastolic dysfunction (E/E’; J) are increased in FGFR4Arg/Arg385 mice (n = 5–9 mice per group; *P < 0.05). (K) FGFR4Arg/Arg385 mice demonstrate increased serum FGF23 levels compared with wild-type littermates (mean ± SEM; n = 9–12 mice per group; *P < 0.05). See also Figure S5 and S7.