Phosphate Restriction Prevents Metabolic Acidosis and Curbs Rise in FGF23 and Mortality in Murine Folic Acid–Induced AKI

Visual Abstract

Abstract

Significance Statement

Patients with AKI suffer a staggering mortality rate of approximately 30%. Fibroblast growth factor 23 (FGF23) and phosphate (P_i) rise rapidly after the onset of AKI and have both been independently associated with ensuing morbidity and mortality. This study demonstrates that dietary P_i restriction markedly diminished the early rise in plasma FGF23 and prevented the rise in plasma P_i, parathyroid hormone, and calcitriol in mice with folic acid–induced AKI (FA-AKI). Furthermore, the study provides evidence for P_i-sensitive osseous Fgf23 mRNA expression and reveals that P_i restriction mitigated calciprotein particles (CPPs) formation, inflammation, acidosis, cardiac electrical disturbances, and mortality in mice with FA-AKI. These findings suggest that P_i restriction may have a prophylactic potential in patients at risk for AKI.

Background

In AKI, plasma FGF23 and P_i rise rapidly and are independently associated with disease severity and outcome.

Methods

The effects of normal (NP) and low (LP) dietary P_i were investigated in mice with FA-AKI after 3, 24, and 48 hours and 14 days.

Results

After 24 hours of AKI, the LP diet curbed the rise in plasma FGF23 and prevented that of parathyroid hormone and calcitriol as well as of osseous but not splenic or thymic Fgf23 mRNA expression. The absence of Pth prevented the rise in calcitriol and reduced the elevation of FGF23 in FA-AKI with the NP diet. Furthermore, the LP diet attenuated the rise in renal and plasma IL-6 and mitigated the decline in renal α-Klotho. After 48 hours, the LP diet further dampened renal IL-6 expression and resulted in lower urinary neutrophil gelatinase-associated lipocalin. In addition, the LP diet prevented the increased formation of CPPs. Fourteen days after AKI induction, the LP diet group maintained less elevated plasma FGF23 levels and had greater survival than the NP diet group. This was associated with prevention of metabolic acidosis, hypocalcemia, hyperkalemia, and cardiac electrical disturbances.

Conclusions

This study reveals P_i-sensitive FGF23 expression in the bone but not in the thymus or spleen in FA-AKI and demonstrates that P_i restriction mitigates CPP formation, inflammation, acidosis, and mortality in this model. These results suggest that dietary P_i restriction could have prophylactic potential in patients at risk for AKI.

Introduction

Phosphate (P_i) is an essential mineral, but its potential toxicity underscores the need of a tight homeostatic regulation.¹ P_i homeostasis is mostly orchestrated by a triad of directly interdependent hormones. Fibroblast growth factor 23 (FGF23) and parathyroid hormone (PTH) both suppress P_i reabsorption in the kidney, whereas calcitriol enhances P_i absorption in the small intestine.^2–4 FGF23 and PTH are negative and positive regulators of each other and calcitriol, respectively, where the latter is mediated via the transcriptional regulation of the renal calcitriol anabolic and catabolic enzymes encoded by Cyp27b1 and Cyp24a1.^5–10 In addition, α-Klotho serves as the obligate coreceptor for FGF23 while also acting as a renoprotective agent and early negative biomarker of kidney disease.^11,12

In AKI, defined as an abrupt decline in renal function precipitated by a prerenal, intrarenal, or postrenal insult or dysfunction, both serum P_i and FGF23 rise rapidly and are independently associated with disease severity and outcome.^13–17 Similarly, in CKD, serum levels of P_i and FGF23 are independently associated with increased risk of mortality and a heightened inflammatory state.^18–26 In fact, congruent associations are consistently observed in cardiovascular and all-cause mortality for both P_i and FGF23.^27–29

Several in vivo and in vitro studies have shown that P_i directly worsens inflammation. Treatment of hemodialysis patients with a noncalcium P_i binder was associated with a decrease in C-reactive protein, IL-6, and IL-8 and an increase in the anti-inflammatory IL-10.^30,31 Restricting dietary P_i in Col4a3^−/− mice, a CKD model, blunted the rise in BUN and hepatic Il1b and Il6 upregulation, while high medium P_i upregulated Il1b and Il6 in primary hepatocytes.³² Dietary P_i dose-dependently increased serum and tissue TNF as well as oxidative stress markers in rats with adenine-induced CKD, while high medium P_i directly upregulated TNF in human vascular smooth muscle cells.³³ Similarly, in the 5/6 nephrectomy rat model, intestinal P_i binding reduced renal injury, inflammatory cytokines, macrophage infiltration, and fibrosis.³⁴ Concordant results were also produced in macrophage, myoblast, and aortic smooth muscle cell lines.^35–37 However, whether the proinflammatory effects of dietary P_i are relevant in AKI and the concomitant rise of FGF23 constitutes a phosphaturic counter response serving to curb the inflammatory sequelae of hyperphosphatemia remain unexplored.

The aim of this study was to investigate the local and systemic effects of restricting dietary P_i on FGF23 metabolism, mineral homeostasis, and inflammation in mice with folic acid–induced AKI (FA-AKI).

Methods

Animals

Male C57BL6/JRj mice were obtained from Janvier Labs and allowed to acclimate for 1 week in a conventional animal facility at the Laboratory Animal Services Center, University of Zurich, with a 12 hours:12 hours light–dark cycle and an ambient temperature of 22°C±2°C. The mice had ad libitum access to water and food of either the normal phosphate (NP) (0.6% w/w phosphorus as calcium or sodium salts, 1% w/w calcium, 1000 IU/kg cholecalciferol, and 16 mg/kg folic acid (FA); S9151-E712, ssniff Spezialdiäten GmbH) or low phosphate (LP) (<0.1% w/w phosphorus as calcium or sodium salts, 1% w/w calcium, 1000 IU/kg cholecalciferol, and 16 mg/kg FA; S9151-E710, ssniff Spezialdiäten GmbH) diets (Supplemental Table 1). Except for phosphorus content, the NP and LP diets were of identical composition. Respective diets were kept constant for 4 days before experimental manipulation up until sacrifice. Pth^−/− mice and wild type littermates were bred (C57BL/6J background) at an Optimal Hygiene Conditions facility at the Laboratory Animal Services Center, where they were fed standard chow (3436, Kliba-Nafag, Kaiseraugst) and maintained with 1% w/v calcium d-gluconate monohydrate water (G4625, Sigma).^38,39 The genotyping protocol is found in the Supplemental Method. Four days before experimentation, the food was switched to the NP diet and plain water provided up until sacrifice.

All procedures conducted throughout this work were in accordance with the Swiss animal welfare laws and guidelines for animal care and approved by the Zurich Veterinary Office under the license number 169/2019.

Experimental Design

FA-AKI was induced by way of a single intraperitoneal injection of 250 mg/kg FA (F7876, Sigma) thoroughly dissolved at a concentration of 25 µg/µl in 150 mM NaHCO₃ at pH 7.4 or vehicle, as previously described.^40,41 In 4 separate experiments, mice were euthanized 3, 24, 48 hours, or 14 days after AKI induction by exsanguination and organ removal under isoflurane anesthesia. For the 48-hour time point, the electrocardiogram (ECG) was monitored before euthanasia. For the 14-day time point, mice were also weighed daily to monitor morbidity and injected subcutaneously with 1 ml of sterile 0.9% saline (395183, B. Braun) divided equally between the ventral inguinal and scruff regions starting 2 days after the FA injection to curb dehydration ensuing from the AKI-induced polyuria. Mice were euthanized as per humane end point: losing more than 15% of the initial bodyweight and/or animal health scoring exceeding 1. Spot urine was collected before euthanasia in all mice. Under continuous isoflurane anesthesia, further urine was collected by bladder puncture and blood, by cardiac puncture using heparinized syringes (heparin sodium, 50 μl, B. Braun). Plasma was separated by centrifugation at 6000 g and 4°C for 10 minutes. For the 48-hour time point, blood was collected via caudal vena cava puncture with a balanced heparin RAPIDLyte syringe (00925045, Siemens Healthcare), and plasma was separated by centrifugation at 3000 g and 4°C for 15 minutes. Organs were harvested, and bone marrow was isolated by cutting the epiphysis of tibias and femurs, placing the bones in a bottom perforated 0.5 ml tube placed in a 1.5 ml Eppendorf tube which was centrifuged for 15 seconds with a mini benchtop centrifuge. All samples were snap frozen in liquid nitrogen. Decapsulated kidneys were weighed and normalized to body weight. For histologic analysis, 2-mm transverse renal sections were excised and fixed according to the protocol described in the section “Histologic analysis and quantification.”

ECG Recordings

ECG was recorded for 10 minutes by external telemetry with the DSI (Data Sciences International) transmitter ETA-F10. Mice were anesthetized with inhalant anesthesia (1.5%–2% isoflurane, Piramal Critical Care) and placed on a warming pad. The temperature was monitored with a rectal probe (37.0°C±0.5°C) (PhysioSuite, Kent Scientific). The two electrodes were placed subcutaneously with the positive electrode positioned in the left caudal rib region 1 cm left of the xiphoid process and the negative electrode atop the right pectoral muscle, while the transmitter body was kept externally. ECG data were recorded using the software Ponemah (v6, DSI).

Plasma, Urine, and Blood Biochemical Assays

Using a UniCel SYNCHRON DxC 800 Synchron Clinical System (Beckman Coulter) at the Zurich Integrative Rodent Facility, University of Zurich, the following parameters were measured while adhering to the manufacturer's protocols: plasma P_i (PHOSm method), creatinine (enzymatic method), urea (enzymatic conductivity rate method) as well as urinary P_i, creatinine (Jaffé rate method), and NH₄⁺ (glutamate dehydrogenase method). Urinary osmolality was measured using the Osmomat 3000 (Gonotec). Blood was analyzed with the epoc blood gas analysis system (Siemens Healthcare) to measure blood pH, CO₂ partial pressure (pCO₂), HCO₃⁻, Cl⁻, K⁺, ionized Ca²⁺, lactate, and Na⁺.

Immunoassays

All commercial immunoassays were conducted in accordance with the manufacturers' protocols. Plasma hormones and urinary neutrophil gelatinase-associated lipocalin (NGAL) were measured using commercial ELISA kits: intact FGF23 (iFGF23) (60-6800, Quidel), C-terminal FGF23 (cFGF23) (60-6300, Quidel), PTH (1–84) (60-2305, Quidel), and NGAL (443707, BioLegend). Plasma calcitriol was measured using a radioimmunoassay (AA-54F1, Immunodiagnostic Systems). Plasma and renal cytokines were measured using a bead-based multiplex flow cytometric assay with a predefined panel (740446, BioLegend) on the FACS Canto II high-throughput sampler (BD Biosciences) at the Cytometry Facility, University of Zurich. The kidney homogenization protocol for the cytokine measurement is detailed in the section “Renal cytokine measurement.”

Renal Cytokine Measurement

Transversally cut half kidneys were homogenized with ceramic beads using the Precellys 24 (Bertin Instruments) in a buffer containing 25 mM Tris, 150 mM NaCl, 1 mM EDTA (EDS, Sigma), 1 mM EGTA (E4378, Sigma), 1% IGEPAL CA-630 (56741, Sigma), and 1 cOmplete protease inhibitor cocktail pill per 10 ml (4693132001, Merck) at pH 7.4. Tissue homogenate was then centrifuged for 15 minutes at 16,000 g and 4°C, and the supernatant was centrifuged again before measuring protein concentration using a colorimetric assay (5000112, Bio-Rad) and loading the samples at a concentration of 20 mg/ml in the cytokine multiplex assay.

RNA Extraction, Reverse Transcription, and Semiquantitative Real-Time PCR

Total RNA from tissues was extracted by first homogenizing the tissue using ceramic beads on the Precellys 24 (Bertin Instruments) either in TRIzol reagent (15596026, Invitrogen) followed by chloroform extraction or with NucleoSpin RNA lysis buffer (Macherey-Nagel) for kidney tissue. Total RNA was purified by NucleoSpin RNA Mini kit (740955, Macherey-Nagel) in accordance with the manufacturer's protocol. RNA was quantified using the NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific), which was then reverse transcribed on a thermocycler (SensoQuestGmbH) using the TaqMan Reverse Transcription Reagent Kit (Applied Biosystems): 7.5 ng/μl RNA template, 5.5 mM MgCl₂, 5 ng/μl random hexamers (79236, Qiagen), 500 μM each deoxynucleotide triphosphate, 0.4 U/ml ribonuclease inhibitor, 1.25 U/μl multiscribe reverse transcription, and ribonuclease-free water. The PCR thermal profile consisted of 10 minutes at 25°C, 30 minutes at 48°C, and 5 minutes at 95°C.

Semiquantitative real-time PCR (qPCR) was run on either 96-well or 384-well plates (4346906 and 4309849, Applied Biosystems, respectively). For probe-based qPCR, the KAPA Probe Fast qPCR kit was used (KK4715, Roche): a 20 µl reaction consisted of 0.1 µM fluorescent probe, 1 µM each of forward and reverse primers, 10 µl Taqman mastermix, 0.4 µl Rox Low passive dye, and 3 µl cDNA template. The PCR thermal profile comprised 20 seconds at 95°C, followed by 40 cycles of 1 second at 95°C and 20 seconds at 60°C. For dye-based reactions, the PowerUp SYBR Green Master Mix was used (A25742, Applied Biosystems): a 20 µl reaction consisted of 1 µM each of forward and reverse primers, 10 µl SYBR Green Master Mix, and 3 µl cDNA template. The PCR thermal profile comprised 2 minutes at 50°C and 2 minutes at 95°C, followed by 40 cycles of 1 second at 95°C and 30 seconds at 60°C. A melt curve was also conducted for dye-based reactions to confirm specificity. All qPCR reactions were run on the QuantStudio 6 Pro Real-Time PCR System (Thermo Fisher Scientific). Primer/probe sequences are listed in Supplemental Table 2.

Gene expression was quantified using the Livak method, where 2^−ΔCT was first calculated for each sample by normalizing to a reference gene, before dividing by the average of all 2^−ΔCT values in the NP-vehicle group: for every experiment and tissue, a triplex reaction of reference genes was run (Hprt, Gusb, and Tbp), and the gene that does not significantly vary with treatment or the geometric mean (of 2^−ΔCT values) of a combination thereof is used to normalize the results before the fold change calculation.⁴²

Western Blot Analysis

Transversally cut half kidneys were homogenized with ceramic beads using the Precellys 24 (Bertin Instruments) in RIPA buffer containing 20 mM Tris-HCl, 150 mM NaCl, 1% IGEPAL CA-630 (56741, Sigma), 0.5% Na deoxycholate (30970, Fluka Analytical), 1 mM EDTA (EDS, Sigma), 0.1% SDS (L3771, Sigma), and one cOmplete protease inhibitor cocktail pill per 10 ml (4693132001, Merck) at pH 7.4. Tissue homogenate was centrifuged for 20 minutes at 380 g and 4°C, and the supernatant was centrifuged again before measuring protein concentration using a colorimetric assay (5000112, Bio-Rad). Protein extracts were mixed with Laemmli buffer containing a final composition 62.4 mM Tris (pH 6.8), 2% w/v SDS (L3771, Sigma), 75 mM dithiothreitol (646563, Sigma), 10% v/v glycerol (G5516, Sigma), and 5% w/v bromophenol blue (114391, Sigma) and heated at 60°C for 5 minutes. The samples, along with the molecular weight marker (928-40000, LI-COR Biosciences), were run on 8%–15% polyacrylamide gels containing acryl-bisacrylamide mix (1610156, Bio-Rad), 375 mM Tris, 0.1% w/v SDS (L3771, Sigma), 0.1% w/v ammonium persulfate (09913, Sigma), and N,N,N′,N′ -Tetramethylethylenediamine (T9281, Sigma). The proteins were transferred onto a polyvinylidene fluoride membrane (IPVH00010, Merck) in a wet transfer system at 340 mA for 90 minutes.

The membrane was dried at 37°C for 10 minutes, rehydrated in 100% methanol for 30 seconds, rinsed in Tris-buffered saline (TBS) for 5 minutes, rinsed in distilled deionized water, stained with Revert 700 Total Protein Stain (926-11021, LI-COR Biosciences) for 5 minutes, and finally washed twice for 30 seconds in a solution containing 6.7% v/v glacial acetic acid and 30% v/v methanol, followed by a rinse in distilled deionized water. The total protein content was scanned with the Odyssey CLx Imaging System (LI-COR Biosciences) in the 700 nm channel. The membrane was blocked with either Casein blocking buffer (B6429, Sigma) or 5% bovine serum albumin in TBS (P06-1391,100, PAN Biotech) and incubated overnight with primary antibody at 4°C in the respective blocking buffer along with 0.1% v/v Tween 20 (93773, Sigma). After washing with 0.1% Tween 20 in TBS, the membrane was incubated for 1 hour at room temperature with secondary antibody in the respective blocking buffer containing 0.01% SDS w/v and 0.1% v/v Tween 20. Finally, the membrane was washed again and developed with Immobilon Chemiluminescent horseradish peroxidase Substrate (WBKLS0500, Merck) in accordance with the manufacturer's protocol and imaged with the LAS-4000 (Fujifilm). The total protein stain and band intensity were quantified using Image Studio Lite v5.2 (LI-COR Biosciences). All antibodies used are presented in Supplemental Table 3.

Plasma Calciprotein Monomer and Calciprotein Particle Measurement

Calciprotein monomer (CPM) and calciprotein particle (CPP) measurements were performed according to our published methods, with minor modifications as follows.^43,44 All solutions were twice 0.22 µm filtered before use. The monomeric form of bovine fetuin-A (F3004, Sigma) was purified by size exclusion chromatography, as described previously.⁴⁵ Frozen plasma samples were thawed in a water bath at 37°C for 60 seconds with gentle agitation. For CPP analysis, 5 µl portions were mixed with 40 µl HEPES-buffered DMEM (50 mM HEPES, no phenol red, pH 7.45) and 5 µl staining solution containing Alexa Fluor 647-conjugated risedronate (5 µM, BV500101, BioVinc), FITC-conjugated lactadherin (1.5 µg/ml, bovine lactadherin-FITC, Hematologic Technologies Inc.), and mFluor violet 450-labeled bovine fetuin-A (2 µg/ml, prepared in-house using the ReadiLink Rapid mFluor Violet 450 Labeling Kit, 1100, AAT Bioquest). After 120 minutes of incubation in the dark with gentle mixing, samples were diluted to 500 µl in HEPES-buffered DMEM. Measurements were acquired in triplicate using a calibrated Apogee A50/Micro flow cytometer equipped with 50 mW 405, 488, and 638 nm lasers (control parameter settings: sheath pressure of 150 mbar and four flush cycles). Flow rate (3 µl/min) and measurement times (120 seconds or until data storage buffer was full—5×10⁶ events) were held constant for all samples. Fluorescence threshold triggering was used to detect fetuin-A–stained particles, and a negative gating strategy was used to exclude membrane-delimited particles which stain positive for FITC-conjugated lactadherin. CPPs were gated as fetuin-A–positive lactadherin-negative events. Amorphous Ca-P_i containing primary CPP (CPP-I) and crystalline hydroxyapatite containing secondary CPP (CPP-II) populations were gated on the basis of their differential affinity for risedronate: risedronateLO and risedronateHI, respectively. For CPM analysis, 25 µl portions were centrifuged at 30,000 g and 4°C for 2 hours to sediment CPP and the resulting supernatants were incubated with gentle agitation for a further 24 hours at 37°C. Five microliter portions were then mixed with 45 µl of HEPES-buffered DMEM (100 mM HEPES, no phenol red, pH 7.8) containing Alexa Fluor 647-conjugated risedronate (5 µM). After light-protected incubation for 1 hour at room temperature, unbound dye was removed by gel filtration (Micro Bio-Spin Columns with Bio-Gel P-30, Bio-Rad, pre-equilibrated with buffer). Of the resulting flow through, 40 µl were mixed with 10 µl of 0.5 M EDTA solution (adjusted to pH 8.0) and 10 µl SDS solution (10% in water). Samples were assayed in triplicate, and fluorescence was measured using a multimode plate reader equipped with appropriate filter sets (Synergy HTX, BioTek).

Data Analysis and Statistics

All data were analyzed using GraphPad Prism 9, except for the multivariate ANOVA, which was conducted on IBM SPSS 14. However, raw data from the cytokine cytometric assay were first fitted using an asymmetric 5-parameter logistic model and interpolated using the manufacturer's cloud-based Data Analysis Software Suite version 2021.07.01 (BioLegend). All other immunoassays were fitted using a four-parameter logistic model, while data from the protein quantification colorimetric assays were fitted with a polynomial regression model. ECG data were analyzed using the software Ponemah (v6, DSI); QT intervals were corrected using the Mitchell method, as previously described.^46,47 For the generation of the modified Gamblegram, ion valence was used as an equivalence factor. Given the pH-dependent valence of P_i, its average valence was computed using Equation 1 (derived from the Henderson–Hasselbalch equation), where pK_a=6.8.⁴⁸

$${\left[P_i\right]}_{Eq}=\frac{\left[P_i\right]+2\left[P_i\right]\left({10}^{pH-{pK}_a}\right)}{1+\left({10}^{pH-{pK}_a}\right)}$$ (1)

Statistical analysis of all experiments was conducted using a two-way ANOVA, a three-way ANOVA (mixed-effects model with repeated measures), or t test as appropriate with α=0.05. Tukey's test was used for post hoc analysis to correct for multiple comparisons. Cytokine whole-panel data were analyzed using multivariate ANOVA. Outliers were excluded based on the robust regression and outlier removal method with a maximum false discovery rate Q=0.5%.⁴⁹ For censored cytokine cytometric data imputations, samples falling below the limit of detection (software calculated: predicted blank concentration+3 SD values) of a particular analyte were substituted with limit of detection/√2. All data are presented as mean±SD.

All data were also tested for heteroscedasticity using the nonparametric Spearman's rank correlation test conducted on predicted Y values and absolute residuals. Given the robustness of ANOVA to the violation of the homoscedasticity assumption under the condition of sample size homogeneity across experimental groups, a Spearman's ρ (R_s) threshold of ±0.7 was a priori dictated to indicate heteroscedasticity when significant (α=0.05).⁵⁰ The threshold was chosen based on the previously suggested correlation coefficient strength categorization.⁵¹ The Anderson–Darling, D'Agostino–Pearson omnibus, Shapiro–Wilk, and Kolmogorov–Smirnov tests were collectively used to assess whether the residuals were Gaussian. In cases of heteroscedasticity or non-normality, the data were first logarithmically (log₁₀) transformed before conducting the parametric statistical tests. Kaplan–Meier curve comparison was conducted using the Mantel–Cox log-rank test for the FA-treated groups.

Results

P_i Restriction Diminished the AKI-Induced Rise in Plasma iFGF23 and cFGF23

Twenty-four hours after AKI induction, kidney weight, plasma urea and creatinine, urinary NGAL, and renal expression of Havcr1 (kidney injury molecule 1) were significantly elevated and body weight decreased similarly in both FA groups confirming the induction of AKI (Figure 1, A–D and Supplemental Figure 1, A and B). The efficacy of P_i restriction was confirmed by a significantly lower urinary P_i per creatinine ratio in the LP compared with NP groups independent of FA treatment (Figure 1E). Plasma P_i was elevated after FA treatment in the NP group, as previously reported; however, it was unchanged in the LP group (Figure 1F).^41,52 Conversely, plasma total Ca²⁺ was increased in the LP-FA group but unchanged in the NP-FA group (Supplemental Figure 1C).

Figure 1.

Kidney injury markers and P_i 24 hours after FA-AKI. WT male C57BL6/JRj mice were fed either a normal (NP, 0.6% P_i) or low P_i (LP, <0.1% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 24 hours. (A and B) Plasma urea and creatinine levels. (C) Urinary NGAL normalized to creatinine (logarithmically transformed data, native units: ng/mg) and (D) renal Havcr1 mRNA (KIM-1) expression fold change. (E and F) Urinary P_i normalized to creatinine and plasma P_i. Data are expressed as mean±SD (n=5–7 for NP-vehicle; n=7 for LP-vehicle; n=8 for NP-FA; n=6–8 for LP-FA). **P < 0.01, ***P < 0.001, and ****P < 0.0001 between indicated groups by two-way ANOVA with Tukey's post hoc. FA-AKI, folic acid–induced AKI; KIM-1, kidney injury molecule 1; LP, low phosphate; NGAL, neutrophil gelatinase-associated lipocalin; NP, normal phosphate; P_i, phosphate; WT, wild type.

Consistent with previous studies, both plasma iFGF23 and cFGF23 were increased in FA-AKI.^41,53 However, in the LP-FA group, the rise in iFGF23 and cFGF23 was strongly attenuated relative to the NP group, and the baseline was lower (Figure 2, A and B). The difference in iFGF23 between the diet groups after FA treatment was also sustained when fold changes were computed (Supplemental Figure 1D). On FA treatment, the i/cFGF23 ratio was increased in the NP but not the LP group (Figure 2C). As previously shown, AKI upregulated Fgf23 mRNA expression in bone, thymus, and spleen, but only osseous upregulation was abrogated in the LP-FA group (Figure 2, D–F).⁵³ Fgf23 mRNA expression remained unchanged on FA treatment in the bone marrow and liver, but a significant FA- treatment group effect was detected in the heart and ectopic expression was observed in the kidney (Supplemental Figure 1, E–H).

Figure 2.

Plasma and tissue mRNA expression of FGF23 24 hours after FA-AKI. WT male C57BL6/JRj mice were fed either a normal (NP, 0.6% P_i) or low P_i (LP, <0.1% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 24 hours. (A–C) Plasma iFGF23 and total cFGF23 (logarithmically transformed data, native units: pg/ml) and their ratio (i/cFGF23). (D–F) Osseous, thymic, and splenic Fgf23 mRNA expression fold change. Data are expressed as mean±SD (n=7 for NP-vehicle; n=7 for LP-vehicle; n=8 for NP-FA; n=7–8 for LP-FA). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 between indicated groups by two-way ANOVA with Tukey's post hoc. cFGF23, C-terminal fibroblast growth factor 23; FA, folic acid; FGF23, fibroblast growth factor 23; iFGF23, intact FGF23.

P_i Restriction Abrogated the AKI-Induced PTH-Mediated Elevation in Plasma Calcitriol

Twenty-four hours after FA-AKI induction, plasma PTH increased in the NP-FA group, in line with previous studies, which was abrogated in the LP-FA group (Figure 3A).^41,54,55 Similarly, plasma calcitriol was increased in the NP-FA group compared with the LP-FA group, which was paralleled by higher renal Cyp27b1 mRNA expression. Renal Cyp24a1 mRNA was downregulated in the NP-FA but not the LP-FA group (Figure 3, B–D).

Figure 3.

Plasma PTH and calcitriol 24 hours after FA-AKI. WT male C57BL6/JRj mice were fed either a normal (NP, 0.6% P_i) or low P_i (LP, <0.1% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 24 hours. (A and B) Plasma PTH (1–84) and calcitriol. (C and D) Renal mRNA expression fold change of Cyp27b1 and Cyp24a1. Data are expressed as mean±SD (n=7 for NP-vehicle; n=7 for LP-vehicle; n=7–8 for NP-FA; n=8 for LP-FA). *P < 0.05, ***P < 0.001, and ****P < 0.0001 between indicated groups by two-way ANOVA with Tukey's post hoc. ND, non-detectable; PTH, parathyroid hormone.

The AKI-induced increase in plasma calcitriol was further investigated in Pth^−/− mice. Pth^−/− mice exhibited no genotype-dependent differences in kidney injury markers, plasma P_i, or plasma total Ca²⁺; however, urinary P_i and Ca²⁺ displayed a significant genotype group effect (Supplemental Figure 2, A–G). The AKI-induced increase in plasma calcitriol and renal Cyp27b1 mRNA was completely absent in Pth^−/− mice (Figure 4, A and B). AKI downregulated renal Cyp24a1 mRNA expression genotype independently (Figure 4C). The rise in plasma iFGF23 and its fold change from baseline were diminished in Pth^−/− mice with FA-AKI, while for plasma cFGF23, there was a significant genotype group effect (Figure 4D and Supplemental Figure 2, H and I). However, no difference in the AKI-induced upregulation of osseous or splenic Fgf23 mRNA was observed between genotypes, but thymic Fgf23 mRNA upregulation was further augmented in Pth^−/− mice (Supplemental Figure 2, J–L). No genotype-dependent differences in plasma or renal IL-6 (Supplemental Figure 2, M and N) or other inflammatory cytokines (data not shown) were detected.

Figure 4.

Plasma calcitriol and FGF23 24 hours after FA-AKI in PTH knockout mice. WT (Pth^+/+) and knockout (Pth^−/−) male mice were fed a normal-P_i (NP, 0.6% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 24 hours. (A) Plasma calcitriol. (B and C) Renal mRNA expression fold change of Cyp27b1 and Cyp24a1. (D) Plasma iFGF23 (logarithmically transformed data, native units: pg/ml). Data are expressed as mean±SD (n=10 for NP-vehicle; n=9 for LP-vehicle; n=11 for NP-FA; n=12 for LP-FA). ***P < 0.001 and ****P < 0.0001 between indicated groups by two-way ANOVA with Tukey's post hoc.

P_i Restriction Ameliorated the AKI-Induced Renal α-Klotho Depletion and Rise in Plasma IL-6

FA-AKI depleted α-Klotho protein in kidney homogenates after 24 hours, but the LP diet increased α-Klotho in the LP-vehicle group, as previously reported, and reduced its depletion in FA-AKI (Figure 5, A and B).⁵⁶

Figure 5.

Renal α-Klotho and inflammatory status after FA-AKI. WT male C57BL6/JRj mice were fed either a normal (NP, 0.6% P_i) or low P_i (LP, <0.1% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 24 or 3 hours. (A and B) Renal α-Klotho Western blot (normalized to total protein) and quantification 24 hours after FA-AKI. (C–F) Plasma and renal IL-6 and renal and hepatic Il6 mRNA expression fold change 24 hours after FA-AKI. (G) A heatmap of median fold change of a panel of inflammatory cytokines 3 hours after FA-AKI; trend for diet group effect (P = 0.077). For the 24-hour experiments, data are expressed as mean±SD (n=7 for NP-vehicle; n=7 for LP-vehicle; n=7–8 for NP-FA; n=7–8 for LP-FA). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 between indicated groups by two-way ANOVA with Tukey's post hoc. For the 3-hour experiment (heatmap), data are expressed as median fold change from the NP-vehicle group (per cytokine: n=6–7 for NP-vehicle; n=5–7 for LP-vehicle; n=7–8 for NP-FA; n=6–8 for LP-FA) and were analyzed using MANOVA. GM-CSF, granulocyte macrophage colony–stimulating factor; MANOVA, multivariate ANOVA; MCP-1, monocyte chemotactic protein 1.

Next, a panel of inflammatory cytokines was measured in the plasma and renal homogenates of mice 24 hours after FA-AKI. The AKI-instigated increase in IL-6 was mitigated with the LP diet in both plasma and renal tissue (Figure 5, C and D). Furthermore, renal Il6 mRNA was less upregulated in the LP-FA group, and hepatic Il6 mRNA expression was unchanged (Figure 5, E and F). Other measured cytokines did not differ across the diet groups (data not shown). To capture the early inflammatory response, the panel of inflammatory cytokines was measured again in the plasma of mice 3 hours after FA-AKI induction. The LP groups showed a trend toward a group effect for P_i restriction suppressing inflammatory cytokines (P = 0.077) (Figure 5G).

As α-Klotho has been previously suggested to exert its renoprotective effects via autophagic activation, renal LC3 protein levels were quantified 24 hours after FA-AKI.⁵⁷ AKI induced an increase of LC3-I irrespective of the diet group as well as a modest increase of the lipidated LC3-II in the NP-FA but not the LP-FA group (Supplemental Figure 3, A–C). In addition, given the established significance of ferroptosis in the pathogenesis of FA-AKI, renal ferroptotic markers were assessed.^58–61 With FA treatment, Gpx4 mRNA in the NP group was lower than in the LP group; Ptgs2, Chac1, and Slc7a11 transcripts were upregulated comparably in both diet groups; and Aifm2 mRNA was marginally upregulated in the LP group (Supplemental Figure 3, D–H). Furthermore, AKI caused a diet-independent increase in thiobarbituric acid-reactive substances, a surrogate for malondialdehyde (Supplemental Methods, Supplemental Figure 3I).

P_i Restriction Reduced the AKI-Induced Rise in Plasma iFGF23 and cFGF23 and Improved the 14-Day Survival

Next, longer-term implications of P_i restriction were explored 14 days after FA-AKI induction. The 14-day survival as per death or humane end point euthanasia was significantly ameliorated in the LP group (Figure 6A). Kidney weight remained increased in the survivors of the NP-FA group but was normalized in the LP-FA group, although plasma urea and creatinine remained comparably elevated after 14 days (Figure 6, B and C and Supplemental Figure 4A). Urinary NGAL was significantly higher in the NP-FA than in the LP-FA group on day 2 and thereafter remained similarly elevated (Figure 6D). Urine osmolality remained reduced throughout the course of the experiment in both FA groups (Supplemental Figure 4B).

Figure 6.

Survival, kidney injury markers, and plasma P_i and FGF23 14 days after FA-AKI. WT male C57BL6/JRj mice were fed either a normal (NP, 0.6% P_i) or low P_i (LP, <0.1% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 14 days. (A) Kaplan–Meier survival curve over 14 days as per death or humane end point euthanasia. (B and C) Plasma urea and creatinine levels. (D) Urinary NGAL normalized to creatinine at various time points after FA-AKI (logarithmically transformed data, native units: ng/mg). (E and F) Plasma P_i and iFGF23 (logarithmically transformed data, native units: pg/ml). Data are expressed as mean±SD (n=12 for NP-vehicle; n=11 for LP-vehicle; n=6 for NP-FA; n=12 for LP-FA). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 between indicated groups by two-way ANOVA with Tukey's post hoc; NGAL data were analyzed by three-way ANOVA with Tukey's post hoc; survival data are expressed as percent survival and 95% confidence interval and analyzed using the Mantel–Cox log-rank test of the FA groups.

Despite the return of plasma P_i to baseline 14 days after FA-AKI, plasma iFGF23 and cFGF23 continued to be increased in both diet groups albeit to a lower extent in the LP group; however, iFGF23 fold change from baseline did not differ between the diet groups (Figure 6, E and F and Supplemental Figure 4, C and D). Consistent with the 24-hour experiment, plasma i/cFGF23 ratio was reduced in both LP groups (Supplemental Figure 4E). There was no difference in gross renal morphology between the diet groups as evidenced by hematoxylin and eosin staining (Supplemental Methods, Supplemental Figure 5A). Furthermore, FA treatment caused diet-independent extensive renal fibrosis as evidenced by picrosirius red staining and COL1A1 protein levels (Supplemental Figure 5, B–E). Renal α-Klotho expression was diet-independently decreased in FA-treated mice 14 days after FA-AKI induction (Supplemental Figure 5F).

P_i Restriction Ameliorated the AKI-Induced Renal Dysfunction and Blood Ion Dysregulation and Prevented Metabolic Acidosis

Given the onset of mortality after 48 hours of FA-AKI induction, renal, cardiac, and systemic status were studied at that time point. Plasma urea, creatinine, urinary NGAL, and P_i and kidney weight were significantly more elevated in the NP-FA group than the LP-FA group (Figure 7, A–D and Supplemental Figure 6A). FA-AKI resulted in a decrease of blood pH in the NP group but not the LP group, while blood HCO₃⁻ and Cl⁻ were decreased and increased, respectively, in the NP compared with the LP groups irrespective of FA treatment (Figure 7E and Supplemental Figure 6, B and C). No differences were detected in blood lactate (Supplemental Figure 6D). Blood K⁺ was increased, and blood ionized Ca²⁺ and Na⁺ were decreased in the NP-FA compared with the LP-FA group, while plasma Mg²⁺ was increased diet-independently on FA-AKI (Supplemental Figure 6, E–H). In the FA groups, blood ionized Ca²⁺ was strongly negatively correlated with plasma P_i (Supplemental Figure 6I).

Figure 7.

Kidney injury markers, acid–base status, and blood ions 48 hours after FA-AKI. WT male C57BL6/JRj mice were fed either a normal (NP, 0.6% P_i) or low P_i (LP, <0.1% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 48 hours. (A and B) Plasma urea and creatinine levels. (C) Urinary NGAL normalized to creatinine (logarithmically transformed data, native units: ng/mg). (D) Plasma P_i. (E) Blood pH. (F and G) Urinary pH and NH₄⁺ (normalized to creatinine). (H) Modified Gamblegram of ions measured in blood with the indicated unmeasured anions (UA) reflecting the difference between total anions and cations. Ion valence was used as an equivalence factor; pH-dependent P_i average valence was calculated using the Henderson–Hasselbalch equation. Data are expressed as mean±SD (n=6–7 for NP-vehicle; n=6–7 for LP-vehicle; n=7–9 for NP-FA; n=9–10 for LP-FA). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 between indicated groups by two-way ANOVA with Tukey's post hoc.

Urinary pH was lower in the NP compared with LP groups and decreased on FA treatment in both groups; urinary NH₄⁺ was higher in the NP-vehicle group compared with the LP-vehicle group but comparable on FA treatment (Figure 7, F and G). The modified Gamblegram revealed increased P_i and decreased HCO₃⁻ contributions in the NP-FA group (Figure 7H). ECG monitoring on anesthetized mice uncovered a higher heart rate and a prolonged QTcm interval in the NP-FA group compared with the LP-FA group (Figure 8, A and B). Of note, the mouse with the most severe changes in blood pH, K⁺, Na⁺, and ionized Ca²⁺ exhibited a multitude of obvious ECG abnormalities: QRS broadening, second-degree atrioventricular block, and P-wave splitting (Supplemental Figure 7).

Figure 8.

Cardiac parameters from anesthetized ECG monitoring, CPP, and renal inflammatory cytokine panel 48 hours after FA-AKI. WT male C57BL6/JRj mice were fed either a normal (NP, 0.6% P_i) or low P_i (LP, <0.1% P_i) diet for 4 days before being treated with FA or NaHCO₃ (vehicle) and sacrificed after 48 hours. (A) Heart rate over a 1-minute interval. (B) Corrected QT interval (Mitchell's method, QTcm). (C–E) Plasma CPM, CPP-I, and CPP-II. (F) Proportion of CPP-II per total CPP (CPP-I+CPP-II). Data are expressed as mean±SD (n=7 for NP-vehicle; n=6–7 for LP-vehicle; n=7–9 for NP-FA; n=9–10 for LP-FA). *P < 0.05, **P < 0.01, and ***P < 0.001 between indicated groups by two-way ANOVA with Tukey's post hoc. (G) A heatmap of median fold change of a panel of inflammatory cytokines in kidney tissue; diet group effect (P = 0.009). For the heatmap, data are expressed as median fold change from the NP-vehicle group (per cytokine: n=7 for NP-vehicle; n=7 for LP-vehicle; n=9 for NP-FA; n=10 for LP-FA) and were analyzed using MANOVA. CPM, calciprotein monomer; CPP, calciprotein particle; CPP-I, primary CPP; CPP-II, secondary CPP; ECG, electrocardiogram; MCP-1, monocyte chemotactic protein 1.

P_i Restriction Prevented the AKI-Induced Increase of Amorphous and Crystalline CPPs and Attenuated the AKI-Induced Increase of Renal IL-6 Persistently

Next, the mineral buffering system in mice 48 hours after FA-AKI was studied by microparticle flow cytometry. In plasma, Ca-P_i complexes and fetuin-A form CPM which, under supersaturating conditions, spontaneously coalesce to form CPP-I that can subsequently transform into proinflammatory/procalcific crystalline CPP-II (Supplemental Figure 8A).⁶² Plasma CPM in mice fed the NP were increased compared with the LP diet independently of FA treatment (Figure 8C). Plasma CPP-I and CPP-II were significantly higher in the NP-FA group compared with the LP-FA group, whereas CPP-II increased in the NP-FA group compared with the NP-vehicle group (Figure 8, D and E). Similarly, the proportion of CPP-II per total CPP was increased on FA-AKI in the NP group but not in the LP group (Figure 8F). The staining and gating strategy for CPP detection is illustrated in Supplemental Figure 8, B and C.

The inflammatory status was revisited by measuring a panel of inflammatory cytokines in mouse renal homogenates 48 hours after FA-AKI. There was a significant effect for dietary P_i (P = 0.009), which largely depended on the persistently mitigated increase in renal IL-6 in the LP-FA group (Figure 8G). Il6 mRNA expression remained elevated in the FA groups but substantially less than at the 24-hour time point, and it was not significantly different between the diets (Supplemental Figure 9A). Other measured cytokines did not differ across the diet groups (Figure 8G). Renal α-Klotho protein remained depleted in FA-treated mice; however, the LP diet did not significantly reduce its depletion at this time point (Supplemental Figure 9, B and C).

In addition, renal ferroptosis markers were measured at the 48-hour time point. With FA treatment, Gpx4 and Aifm2 mRNA were decreased, whereas Ptgs2, Chac1, and Slc7a11 mRNA were increased (Supplemental Figure 9, D–H). Chac1 and Slc7a11 transcripts were more upregulated in the LP-FA group compared with the NP-FA group. Furthermore, FA-AKI caused a persistent diet-independent increase in thiobarbituric acid-reactive substances, a surrogate for malondialdehyde (Supplemental Figure 9J).

Discussion

This study sought to unravel the effect of dietary P_i restriction in mice on renal impairment, inflammation, and hormonal dysregulation precipitated by FA-AKI. Here, we show that P_i restriction in mice with FA-AKI (1) markedly diminished the early rise in plasma FGF23 and abrogated that of PTH and calcitriol; (2) blunted the early increase in systemic and renal IL-6; (3) prevented the increased formation of CPP-I and CPP-II; and (4) remarkably alleviated 14-day mortality likely by precluding cardiac electrical disturbance secondary to metabolic acidosis, hyperkalemia, and hypocalcemia.

Increased plasma FGF23 is a common feature of renal and extrarenal inflammatory diseases.^17,63–67 We showed that P_i restriction strongly dampened the rise in plasma P_i on AKI and substantially reduced the basal and AKI-induced levels of both plasma iFGF23 and cFGF23, indicating that the rise in plasma P_i contributed a sizeable component of the FGF23 increase. Consistent with hemodialysis patients having a higher proportion of iFGF23 compared with healthy controls,⁶⁸ P_i restriction decreased the basal i/cFGF23 ratio and persistently prevented its increase after FA treatment likely due to the absence of hyperphosphatemia.

Christov et al. reported that P_i restriction reduced plasma cFGF23 in mice with FA-AKI; however, in contrast to our findings, the fold change remained the same. Using genetically null mice, they also found that the increase in cFGF23 was independent of PTH and vitamin D receptor.⁴¹ The discordant results may be explained by the lower P_i content in the control diets, phosphate-enriched and/or calcium-enriched diets for Pth-null and Vdr-null mice, lower renal injury, and the bleeding of mice before inducing AKI to assess baseline parameters.

In FA-AKI, P_i restriction also prevented the rise in plasma PTH, as previously shown in CKD models, and additionally abrogated the rise in calcitriol.^69,70 Although a recent meta-analysis revealed that patients with AKI typically have lower calcitriol levels compared with healthy controls, another study reported significantly higher plasma calcitriol and P_i in nonsurvivors compared with survivors in critically ill patients with AKI.^71,72 In our AKI model, the rise in calcitriol was driven by PTH as the increase in renal Cyp27b1 mRNA expression and the concomitant increase in calcitriol was fully abrogated in Pth^−/− mice.

In Pth^−/− mice, the AKI-induced increase in plasma iFGF23 was reduced by 40% without affecting plasma P_i, suggesting independent effects of PTH/calcitriol and plasma P_i, all of which are very potent stimulators of iFGF23. However, although P_i restriction prevented the AKI-induced hyperphosphatemia and concomitant increase in PTH and calcitriol, and blunted the rise in plasma FGF23, FGF23 was still increased above baseline, suggesting that the increase is multifactorial. For instance, one of these factors may be glycerol-3-phosphate which is increased upon renal injury due to a switch from gluconeogenesis to glycolysis as well as upon intravenous and high dietary P_i treatment and has been shown to stimulate FGF23 production in bone and bone marrow stromal cells.^73,74

The upregulation of Fgf23 in osseous and extraosseous organs has been repeatedly reported in multiple disease models.^53,75–79 We demonstrated that plasma P_i and PTH are involved in the elevation of plasma FGF23 in FA-AKI, but while P_i restriction attenuated the osseous upregulation of Fgf23 mRNA expression, Pth knockout did not and instead further enhanced thymic upregulation. Neither P_i restriction nor PTH ablation affected splenic Fgf23 mRNA upregulation. This signifies that bone remains the major source of circulating FGF23 in FA-AKI, but extraosseous sources may cumulatively contribute to a lesser extent. Previous studies have demonstrated that splenic but not renal FGF23 contributes to elevated circulating iFGF23 in inflammation and uremia.^77,78 Hepatic FGF23 production was shown to be increased via IL-6/estrogen-related receptor gamma pathway in mice with FA-AKI.⁵⁵ This is in contrast with the data shown herein, as no hepatic Fgf23 upregulation was detected. Moreover, glycolysis-derived renal glycerol-3-phosphate has been reported to increase Fgf23 expression in bone and bone marrow.⁷⁴ We did not find any changes in Fgf23 mRNA expression in bone marrow, which may be due to the lack of discrimination between the stromal and hematopoietic cell compartments.

Inflammatory stimuli and the consequent functional iron deficiency, as inducers of FGF23, have been shown to concurrently promote FGF23 cleavage, thus offsetting rises in transcriptional drive and maintaining bioactivity. This balance between synthesis and cleavage is thought to break down in the context of other modulators of FGF23 cleavage, such as in CKD.^80–82 We found that in mice with FA-AKI, P_i restriction curbed the rise in IL-6 levels after 24 and 48 hours and exhibited a trend toward a dampened inflammatory state after 3 hours. IL-6, TNF, and IL-1β are well-documented positive regulators of FGF23.^75,83–86 Therefore, it is conceivable that plasma P_i contributed to the increase in plasma iFGF23 partially via PTH and possibly also via its effect on IL-6, where the latter's effect is amplified with the effect of P_i on the i/cFGF23 ratio.

P_i is a proinflammatory instigator, stimulating cytokine production^{30,32–37,87} possibly through enhanced CPP formation and/or suppression of α-Klotho. While CPM/CPPs help to buffer excess P_i, prolonged net formation and subsequent transformation to crystalline CPP-II is deleterious.^43,88–90 In CKD, plasma P_i is positively correlated with crystalline CPP formation,⁹¹ while in vitro, lower serum pH promotes the transformation to CPP-II (unpublished observation). Indeed, hyperphosphatemia, high IL-6, and acidosis were accompanied by increased levels of CPP-I and CPP-II in our model, suggesting that the NP diet further enhanced the AKI-induced proinflammatory environment. In line with previous reports, the acute rise in plasma P_i caused by FA-AKI had a small effect on CPM levels, whereas dietary P_i strongly affected the CPM levels.⁴³ However, unlike CPPs, CPMs are readily cleared by healthy kidneys and thought to be innocuous in the inflammatory milieu.⁸⁸

P_i-induced reduction in α-Klotho, which is well-documented and reproduced in this work, may foster a permissive environment for an enhanced inflammatory response as α-Klotho has a well-characterized anti-inflammatory role.^56,92 This may have also contributed to the observed diet-dependent difference in inflammatory status. However, while several studies have suggested that FGF23 may promote inflammation by upregulating TNF and IL-6, our results do not lend credence to this mechanism, at least at a systemic scale: the absence of Pth reduced the rise in plasma FGF23 in FA-AKI without a change in proinflammatory cytokines.^85,93–95

We showed that P_i restriction improved 14-day survival after FA-AKI despite comparable kidney impairment, renal fibrosis, and urine osmolality. The disparate mortality between the diet groups may have incorporated a survival bias. Indeed, mice from the 48-hour experiment—marking the onset of mortality—clearly showed that P_i restriction curtailed FA-induced renal injury by abrogating the development of a non-anion gap metabolic acidosis due to higher dietary P_i acid load. The aggravating effect of acid loads on coexisting renal dysfunction is a well-established phenomenon: For instance, dietary protein has been shown to decrease GFR in 5/6 nephrectomized rats via metabolic acidosis, CKD patients with metabolic acidosis have an increased risk to develop AKI, and ammonium chloride ingestion in rats exacerbates renal damage and mortality in the ischemia/reperfusion injury model.^96–98

Moreover, hyperkalemia and hypocalcemia may have affected cardiac function. Hyperkalemia developed due to worse renal function and was conceivably escalated by a vicious cycle of mutual-augmentation between hyperkalemia and acidosis, as hyperkalemia has been shown to induce and be induced by acidosis.^99–102 Hypocalcemia developed likely due to the increased complexation of Ca and P_i and led to a prolonged QTcm interval, a hallmark of hypocalcemia that increases the risk of arrhythmia and sudden cardiac death.^103–107 In fact, each of acidosis, hyperkalemia, and hypocalcemia are implicated in adverse cardiac effects that may precipitate in arrhythmias and death.^108–110 Moreover, the symptomatic constellation of severe acute hyperphosphatemia, hypocalcemia, hyperkalemia, and metabolic acidosis is reminiscent of the clinical manifestation of tumor lysis syndrome, in which 25% of patients suffer from cardiac arrhythmia.^111,112 In pediatric patients with tumor lysis syndrome, the administration of P_i binders lowered plasma P_i suggesting a dietary contribution to the hyperphosphatemia.¹¹³

In summary, we showed that dietary P_i was pivotal in the elevation of plasma FGF23 and that it was necessary for the elevation of plasma PTH which in turn was necessary for that of calcitriol. The rise in plasma FGF23 was multifactorial, with its complexity compounded by the interdependence of its stimuli effecting upregulation differentially in osseous and extraosseous organs as well as possibly affecting its cleavage. We also demonstrated that P_i restriction substantially mitigated mortality likely by preventing the development of metabolic acidosis, hypocalcemia, and hyperkalemia, which likely cumulatively contributed to cardiac electrical disturbances. While P_i restriction showcased a notable curbing of IL-6 levels under AKI conditions and basal differences in the renoprotective α-Klotho and Gpx4, which may have added to the improved survival, the predominant mechanism likely hinges on preventing the acidotic and, consequential, hyperkalemic burdens imposed by hyperphosphatemia as well as the concomitant hypocalcemia. These results warrant further investigation into the mechanistic role of acidosis in AKI severity and outcome in general and the prophylactic potential of P_i restriction in patients at risk of AKI specifically and highlight the implication of dietary P_i in inflammation.

Supplementary Material

jasn-35-261-s001.pdf ^{(7.3MB, pdf)}

Disclosures

A.K. Hamid reports Employer: F. Hoffmann-La Roche AG (internship), Numab Therapeutics AG (spouse), and University of Zurich; Research Funding: F. Hoffmann-La Roche AG (internship); and Patents or Royalties: Philochem AG (spouse, granted patent). M.L. Muscalu reports Employer: Roche. E.R. Smith reports honoraria from CSL Vifor and holds stock in Calciscon AG. E.R. Smith also reports Consultancy: Calciscon AG; Research Funding: CSL Vifor and Sanofi; and Honoraria: CSL Vifor. C.A. Wagner reports honoraria from Advicenne, Chugai, Kyowa Kirin, and Medice; and Other Interests or Relationships: Bayer AG. All remaining authors have nothing to disclose.

Funding

This study was supported by grants from the Swiss National Center for Competence in Research NCCR Kidney Control of Homeostasis to C.A. Wagner and D. Egli-Spichtig, the Jubiläumsstiftung Swisslife (1155) to D. Egli-Spichtig, Olga Mayenfisch foundation to D. Egli-Spichtig, and Gottfried und Julia Bangerter Rhyner foundation to D. Egli-Spichtig. E.R. Smith is supported by the Clinical Investigator grant from the Viertel Charitable Foundation.

Author Contributions

Conceptualization: Daniela Egli-Spichtig, Ahmad Kamal Hamid, Carsten Alexander Wagner.

Formal analysis: Daniela Egli-Spichtig, Ahmad Kamal Hamid.

Funding acquisition: Daniela Egli-Spichtig, Edward R. Smith, Carsten Alexander Wagner.

Investigation: Charlotte Calvet, Daniela Egli-Spichtig, Ahmad Kamal Hamid, Timothy D. Hewitson, Maria Lavinia Muscalu, Eva Maria Pastor Arroyo, Udo Schnitzbauer, Edward R. Smith.

Methodology: Charlotte Calvet, Daniela Egli-Spichtig, Ahmad Kamal Hamid, Timothy D. Hewitson, Edward R. Smith.

Project administration: Daniela Egli-Spichtig.

Supervision: Daniela Egli-Spichtig, Carsten Alexander Wagner.

Validation: Ahmad Kamal Hamid.

Visualization: Daniela Egli-Spichtig, Ahmad Kamal Hamid.

Writing – original draft: Ahmad Kamal Hamid.

Writing – review & editing: Charlotte Calvet, Daniela Egli-Spichtig, Ahmad Kamal Hamid, Timothy D. Hewitson, Edward R. Smith, Carsten Alexander Wagner.

Data Sharing Statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/JSN/E566.

Supplemental Table 1. Composition of the normal phosphate and low phosphate diet.

Supplemental Table 2. Primer and probe sequences used for genotyping or qPCR.

Supplemental Table 3. Antibodies/blocking buffers used for Western blot.

Supplemental Methods

Supplemental Figure 1. Basic parameters and fibroblast growth factor 23 (FGF23) tissue expression 24 hours after folic acid–induced AKI (FA-AKI).

Supplemental Figure 2. Basic parameters, fibroblast growth factor 23 (FGF23), and IL-6 24 hours after folic acid–induced AKI (FA-AKI) in parathyroid hormone (PTH) knockout mice.

Supplemental Figure 3. Renal autophagic and ferroptotic markers 24 hours after folic acid–induced AKI (FA-AKI).

Supplemental Figure 4. Basic parameters and fibroblast growth factor 23 (FGF23) 14 days after folic acid–induced AKI (FA-AKI).

Supplemental Figure 5. Renal fibrosis and α-Klotho expression 14 days after folic acid–induced AKI (FA-AKI).

Supplemental Figure 6. Basic parameters 48 hours after folic acid–induced AKI (FA-AKI).

Supplemental Figure 7. Anesthetized ECG monitoring of mouse with severe acidosis, hyperkalemia, hyponatremia, and hypocalcemia 48 hours after folic acid–induced AKI (FA-AKI).

Supplemental Figure 8. Calciprotein particle (CPP) formation and measurement in mouse plasma by microparticle flow cytometry.

Supplemental Figure 9. Renal Il6 mRNA and ferroptotic markers 48 hours after folic acid–induced AKI (FA-AKI).