FGF23 signaling impairs neutrophil recruitment and host defense during CKD

Abstract

Chronic kidney disease (CKD) has been associated with impaired host response and increased susceptibility to infections. Leukocyte recruitment during inflammation must be tightly regulated to protect the host against pathogens. FGF23 levels are increased in blood during CKD, and levels of this hormone have been associated with a variety of adverse effects in CKD patients. Here, we have shown that CKD impairs leukocyte recruitment into inflamed tissue and host defense in mice and humans. FGF23 neutralization during CKD in murine models restored leukocyte recruitment and host defense. Intravital microscopy of animals with chronic kidney failure showed that FGF23 inhibits chemokine-activated leukocyte arrest on the endothelium, and downregulation of FGF receptor 2 (FGFR2) on PMNs rescued host defense in these mice. In vitro, FGF23 inhibited PMN adhesion, arrest under flow, and transendothelial migration. Mechanistically, FGF23 binding to FGFR2 counteracted selectin- and chemokine-triggered β2 integrin activation on PMNs by activating protein kinase A (PKA) and inhibiting activation of the small GTPase Rap1. Moreover, knockdown of PKA abolished the inhibitory effect of FGF23 on integrin activation. Together, our data reveal that FGF23 acts directly on PMNs and dampens host defense by direct interference with chemokine signaling and integrin activation.

Figure 1. Neutrophil recruitment and host defense during pneumonia is decreased in mice with chronic kidney failure.

CKD was achieved in mice by 5/6-nephrectomy, and pneumonia was induced by E. coli instillation after 10 days. (A–D) Twenty-four hours after inducing pneumonia, neutrophils were counted in the BAL (A), and CFUs were analyzed in the BAL (B), lung tissue (C), and spleen (D) of sham-operated and CKD mice (n = 4). (E–H) Pneumonia was induce in sham mice and CKD mice, and sham mice after pretreatment with FGF23 and neutrophils were counted in the BAL (F), lung tissue (G), and spleen (H) (n = 4). (I–P) Pneumonia was also induced in CKD mice after injection of a neutralizing FGF23 antibody (3 mg/kg body weight) or isotype control (I–L) and mice transplanted with control or FGFR2-KD BM (M–P) (n = 4). (Q and R) Survival curves of sham-operated animals and CKD animals (Q) and sham-operated animals and CKD animals treated with isotype or FGF23 neutralizing Ab (R). *P < 0.05, ANOVA plus Bonferroni testing in E–L or 2-tailed t test in A–D and M–P.

Figure 2. CKD impairs neutrophil slow rolling and arrest.

Sham-operated mice and CKD mice, after injection of vehicle or PD173074, were intrascrotally injected with TNF-α. (A–C) After 2 hours, the rolling velocity (A), the number of adherent cells (B), and the number of emigrated cells (C) were analyzed by intravital microscopy of postcapillary venules of the cremaster muscle (n = 4). Adoptive transfer was performed by labeling BM-derived cells from sham-operated mice, CKD mice, or sham-operated mice pretreated with FGF23 with green or red cell trackers. The labeled cells were injected into WT recipient mice 30 minutes after intrascrotal TNF-α injection. (D–F) After 1.5 hours, the rolling velocity (D), the percentage of adherent fluorescent cells (E), and the percentage of transmigrated fluorescent cells (F) were analyzed by intravital microscopy of postcapillary venules of the cremaster muscle (n = 4). For autoperfused flow chamber assays, a catheter was placed in a common aortic artery and connected to coated glass capillaries by PE-50 tubing. (G and H) Leukocyte rolling velocities on E-selectin and E-selectin/ICAM-1 (G) or P-selectin and P-selectin/ICAM-1 (H) were analyzed by video microscopy (n = 4).*P < 0.05, ANOVA plus Bonferroni testing in A–F or 2-tailed t test in G and H.

Figure 3. CKD impairs neutrophil activation and recruitment by affecting β₂ integrin–dependent functions.

(A) Chemokine-induced arrest was investigated in vivo by injecting CXCL1 through an intravascular catheter and recording leukocyte arrest in postcapillary cremasteric venules by intravital microscopy (n = 4–5). (B) Chemokine-induced arrest was investigated in vitro by connecting glass capillaries coated with P-selectin/ICAM-1 or P-selectin/ICAM-1/CXCL1 to intraarterial catheters inserted in sham-operated mice or CKD mice (n = 4). FOV, field of view.(C–E) Murine WT PMNs were pretreated with control or FGF23 (1–100 ng/ml), and static adhesion on ICAM-1 coated surface (C) (n = 4), arrest under flow conditions in glass capillaries coated with P-selectin/ICAM-1 or P-selectin/ICAM-1/CXCL1 (D) (n = 9–12), and ICAM-1 binding (E) were analyzed (n = 3). (F) Superoxide production from untreated and FGF23-treated WT PMNs (n = 3). (G) Phagocytosis of pHrodo E. coli particles were analyzed by flow cytometry (n = 6). Control group (only cells) lacked coincubation with pHrodo E. coli particles. (H) Transendothelial migration of WT PMNs pretreated with control or FGF23 (n = 6). *P < 0.05, ANOVA plus Bonferroni testing in B–H or 2-tailed t test in A.

Figure 4. FGF23 inhibits integrin activation by signaling through FGFR2.

(A) Flow chambers were coated with E-selectin or E-selectin/Kim127 and perfused with HL60 cells pretreated with FGF23 suspended in human plasma, and the number of adherent cells per field of view (FOV) was analyzed by video microscopy (n = 4). (B) Flow chambers were coated with P-selectin/IL-8 or P-selectin/IL-8/mAb24 and perfused with HL60 cells with FGF23, and the number of adherent cells per FOV was analyzed (n = 4). (C) Flow chambers were coated with E-selectin or E-selectin/Kim127 and perfused with HL60 FGFR2-KD cells pretreated with or without FGF23 (100 ng/ml), and the number of adherent cells per FOV was analyzed by video microscopy (n = 4). (D) Flow chambers were coated with P-selectin/IL-8 or P-selectin/IL-8/mAb24 and perfused with HL60 FGFR2-KD cells pretreated with or without FGF23 (100 ng/ml), and the number of adherent cells per FOV was analyzed by video microscopy (n = 4). (E) LFA-1 clustering on the cell membrane of HL60 cells was analyzed by confocal microscopy (n = 5–8). (F) Exemplary micrographs. Scale bars: 5 μm. *P < 0.05, ANOVA plus Bonferroni testing in A–E.

Figure 5. FGF23 deactivates integrins by inhibitory signaling through PKA.

PKA substrate serine phosphorylation in isolated human neutrophils pretreated with or without FGF23 and stimulated with IL-8. (A) Western blot from four experiments. (B) Quantification by densitometry. (C) PKA substrate serine phosphorylation in control and FGFR2-KD HL60 cells pretreated with or without FGF23 and stimulated with IL-8 (Western blot from four experiments). NS, nonsilencing. (D) Control and PKA- KD HL60 cells were stimulated with IL-8 (100 ng/ml) at 37°C for 5 minutes, and binding of the reporter antibody mAb24 was assessed by flow cytometry (n = 6). (E) PKA substrate phosphorylation in murine PMNs after FGF23 pretreatment (100 ng/ml) was analyzed at different time points after washing away FGF23 (Western blot from four experiments). (F) Rap1 activation in human neutrophils with or without FGF23 pretreatment and IL-8 stimulation (Western blot from four experiments). *P < 0.05, ANOVA plus Bonferroni testing in B and D.

Figure 6. Selectin-mediated slow rolling and chemokine-induced arrest is impaired in human patients with CKD.

(A and B) Heparinised whole blood samples were obtained from healthy volunteers and CKD stage 3–5 patients and perfused through coated flow chambers, and the rolling velocity on E-selectin and E-selectin/ICAM-1 (A) or P-selectin and P-selectin/ICAM-1 (B) was analyzed by video microscopy (n = 8–13). (C and D) Whole blood samples from healthy volunteers were incubated with FGF23, and the rolling velocity in flow chambers with E-selectin and E-selectin/ICAM-1 (C) or P-selectin and P-selectin/ICAM-1 (D) was analyzed (n = 4). (E) To investigate chemokine-induced arrest, whole blood samples from healthy volunteers and CKD stage 3–5 patients were perfused through flow chambers coated with P-selectin/ICAM-1 or P-selectin/ICAM-1/IL-8 (n = 8–13). (F) Whole blood samples from healthy volunteers were incubated with FGF23, and the chemokine-induced arrest in flow chambers coated with P-selectin/ICAM-1 or P-selectin/ICAM-1/IL-8 was analyzed (n = 4). *P < 0.05, ANOVA plus Bonferroni testing in A, B, and E or 2-tailed t test in C, D, and F.