IL-1β drives production of FGF-23 at the onset of chronic kidney disease in mice

Abstract

FGF-23 has arisen as an early biomarker of renal dysfunction, but at the onset of chronic kidney disease (CKD) data suggest that FGF-23 may be produced independently of the parathyroid hormone (PTH), 1,25(OH)₂-Vitamin D₃ signaling axis. Iron status is inversely correlated to the level of circulating FGF-23, and improvement in iron bioavailability within patients correlates with a decrease in FGF-23. Alternately, recent evidence also supports a regulatory role of inflammatory cytokines in the modulation of FGF-23 expression. To determine the identity of the signal from the kidney inducing upregulation of osteocytic FGF-23 at the onset of CKD we utilized a mouse model of congenital CKD that fails to properly mature the glomerular capillary tuft. We profiled the sequential presentation of indicators of renal dysfunction, phosphate imbalance, and iron bioavailability and transport to identify the events that initiate osteocytic production of FGF-23 during the onset of CKD. We report here that elevations in circulating intact-FGF-23 coincide with the earliest indicators of renal dysfunction (P14), and precede changes in serum phosphate or iron homeostasis. Serum PTH was also not changed within the first month. Instead, production of the inflammatory protein IL-1β from the kidney and systemic elevation of it in the circulation matched the induction of FGF-23. IL-1β’s ability to induce FGF-23 was confirmed on bone chips in culture, and within mice in vivo. Furthermore, neutralizing antibody to IL-1β blocked FGF-23 expression in both our congenital model of CKD and a second nephrotoxic serum-mediated model. We conclude that early CKD resembles a situation of primary FGF-23 excess mediated by inflammation. These findings do not preclude that altered mineral availability or anemia can later modulate FGF-23 levels, but find that in early CKD they are not the driving stimulus for the initial upregulation of FGF-23.

Graphical Abstract

To determine the identity of the renal signal inducing osteocytic FGF-23 at the onset of CKD we utilized a mouse CKD model and profiled presentation of indicators of renal dysfunction, phosphate imbalance, inflammation, and iron bioavailability. Production of IL-1β from the kidney and systemic elevation of it in the circulation matched the induction of FGF-23. This was confirmed in culture, and in vivo, as well as with neutralizing antibody in two CKD models.

INTRODUCTION

During late Chronic Kidney Disease (CKD) phosphate homeostasis is disrupted.⁽¹⁾ Late CKD is considered a state of secondary FGF-23 excess as phosphate elevation occurs along with disruption of the PTH, Vitamin D₃, and calcium signaling axis. FGF-23 is produced by osteocytes, and counteracts hyperphosphatemia by increasing urinary phosphate wasting, decreasing phosphate absorption in the intestine, and decreasing the production of active vitamin D₃ (1,25(OH)₂D₃).^(2–6) Interestingly, however, FGF-23 levels are elevated during early CKD before any perturbations of phosphate homeostasis or elevations in PTH are present,⁽⁷⁾ and may possibly be regulated initially by decreased iron levels.⁽⁸⁾ In CKD, FGF-23 is regulated independently of phosphate and instead correlates inversely to iron status.⁽⁹⁾ Consistent with this relationship, FGF-23 is predictive of worse clinical outcomes independent of serum phosphate,⁽¹⁰⁾ and rats with 5/6 nephrectomy maintain high FGF-23 while on phosphate-free diets.⁽¹⁰⁾ In experimental animals dietary iron restriction induces FGF-23 and results in hypophosphatemia.⁽⁸⁾ Also, treatment of CKD patients with ferric citrate to correct anemia of CKD demonstrated decreased FGF-23 levels coincident with improvements in iron status.⁽¹¹⁾ These data suggest that iron status directly regulates FGF-23 and phosphate.

Anemia of chronic kidney disease (CKD) was first recognized over 170 years ago⁽¹²⁾, and it is multifactorial in its development. Damaged kidneys decrease erythropoietin (Epo) production from the stromal population. Epo circulates to the bone marrow to stimulate production of new erythrocytes.^(13–15) Erythrocytes also have a shortened lifespan in CKD patients.^(16,17) Free iron is transported through the blood bound to transferrin. Patients with glomerulopathies and CKD as early as stage 1 have marked elevations in urinary loss of transferrin, which leads to lower circulating transferrin levels and a decreased iron binding capacity in the circulation.^(18,19) Iron is taken up from the intestine, and released from macrophage stores, via the iron transporter ferroportin. The activity of ferroportin is regulated by hepcidin, a hepatic-produced peptide, that upon binding to ferroportin causes it to be internalized and degraded.⁽²⁰⁾ CKD is associated with reduced renal clearance of hepcidin and a resulting increase in plasma levels.^(21–24) Chronic inflammation leads to hepcidin elevation as well.^(21–24)

Anemia is not the only non-phosphate-related physiological regulator of FGF-23. Associations have been made between several inflammatory diseases and elevated FGF-23 expression,^(25–29) and several inflammatory signals including IL-6 and IL-1β, and TNF-α have all been demonstrated to increase circulating FGF-23.^(30–32) Direct transcriptional regulation of FGF-23 production by inflammatory proteins appears to be partially dependent upon activation of NFκB and utilizes a regulatory enhancer 16 Kb upstream of the Fgf23 locus.^(32,33) These inflammatory mediators also appear to be able to regulate the cleavage of FGF-23 to its inactive form by dampening the efficacy of furin-like proteases increasing circulating intact FGF-23 (iFGF-23) post-translationally.⁽³⁰⁾

Because CKD is a condition of high inflammation, anemia, and altered PTH/phosphate/1,25(OH)₂D₃ status we utilized a model of low nephron endowment CKD to determine which of these regulators of FGF-23 was initially inducing FGF-23 at the beginning of CKD. We provide the first examination of all three of these potential regulators of FGF-23 in a single CKD model, and have identified that during the onset of CKD renal production of the inflammatory cytokine IL1-β reaches the osteoblasts and osteocytes and first stimulates FGF-23 production.

MATERIALS AND METHODS

CKD was modeled using a mouse with of low nephron endowment CKD. Foxd1-cre mice (Foxd1^{tm1(GFP/cre)Amc}) originally generated by Andrew McMahon and purchased from JAX (stock #012463)⁽³⁴⁾ were mated with animals from R. Grosschedl where flox sites flank the exons encoding the DNA-binding domain of Ebf1 (Ebf1^fl/fl)⁽³⁵⁾ hereafter called Foxd1-Ebf1KO Mice were always compared to age-matched wild-type littermate controls, and were genotyped using the Kapa hot start genotyping kit (KK7352). Animals were group housed in pathogen-free conditions on corn-cob bedding with 12 hour light/dark cycles. For pups less than 21 days old they were raised with dams and sires. All experiments were conducted on mixed sex, littermate-matched animals. Control C57Bl/6 mice used for the collection of bone chips and induction of nephritis were purchased from JAX (stock # 000664). Bone chips were derived from both male and female mice. All mice older than 10 days old were sacrificed by CO₂ asphyxiation. Pups younger than 10 days old were euthanized by hypothermia on ice and then also decapitated. Yale Medical School IACUC approved the purchase and use of all animals (protocol #10543 and 10538).

To induce anti-GBM nephritis 150 μL of nephrotoxic sera raised in sheep immunized against rat GBM and purchased from Peprotech (#PTX-001S) was injected intraperitoneally (IP) into C57Bl/6 mice. Mice were allowed to recover in their home cage until the following day.

Ischemia reperfusion injury (IRI) was performed on 6-month-old mice as described previously.⁽³⁶⁾ Unilateral Ischemic Reperfusion Injury. 8 week old female mice were subjected to anesthesia by intraperitoneal injection with ketamine (100 mg/kg) and xylazine (10 mg/kg) on a 37°C warming pad. The abdomen was opened and warm renal ischemia was induced using a nontraumatic microaneurysm clip (B-2V; FST vascular clamps) on the left renal pedicle for 35 minutes, leaving the right kidney intact. During surgery, the mice were intraperitoneally injected with normal PBS and buprenorphine to avoid dehydration and postoperative pain, respectively. After recovering for 24 hours mice were administered either isotype control or neutralizing antibody to IL-1β (3 μg/mouse IP) once daily for 2 days, and then were sacrificed on the third day post-surgery.

Cisplatin-induced AKI was initiated by dosing three-month-old male C57Bl/6 mice with 20 mg/kg cisplatin (Sigma #232120) intraperitoneally. Mice recovered in their home cages until sacrifice 30 hours after dosing.

Bone chips were generated by dissecting tibiae and femora free of adjacent soft tissues. The periosteal layer was scraped off with a scalpel, epiphyses were removed, and the marrow cavity flushed with PBS. The remaining bone fractions were minced into a fine-grained texture, washed three times with PBS, and then transferred into 24-well plates for culture in vitro overnight in αMEM + 10% FBS + 1% antibiotic/antimycotic. The following day cells were stimulated as indicated. FGF-23 release into the media was measured by first concentrating the culture media 10-fold with 10 kDa cutoff spin columns (Millipore #MRCPRT010).

All urine measurements were made from spot urines at the ages indicated. Urine dipstick analysis was completed with Roche Chemstrip 10 uranalysis strips. Blood counts were acquired using heparinized blood samples on a Hemavet 9500 veterinary hematology analyzer that provides accurate CBC counts from whole blood of mice. Serum and plasma were collected from the vena cava. Serum was clotted at room temperature for 30 min. Serum and plasma were collected after separation by centrifugation at 10,000 x g for 10 min. Serum and urine were assessed for iFGF-23 with the Immunotopics/Quidel ELISA (#60-6800), Furin with Enzium Ensens® Furin Activity Detection kit (#032014-05-96), PTH with Immunotopics mouse PTH (1-84) ELISA (#60-2305), Epo with the R&D Quantikine ELISA (#MEP00B), transferrin with Bethyl Labs (#E99-129). IL-6 with Millipore ELISA (#EZMIL6), TGF-β with Affynmetrix ELISA (#88-8350), CRP ELISA was from R&D (#MCRP00), TNF-α measured by BioLegend ELISA (#430907), IL-1β with R&D Quantikine ELISA (#MLB00C), and IL-1α with R&D Quantikine ELISA (#MLA00). Total hepcidin was assessed using the Intrinsic LifeSciences murine complete ELISA kit (#HMC-001). Serum 1,25(OH)₂D₃ was measured by Tecan ELISA (#MG59051). Serum phosphorus was measured on samples collected after a 6 hour fast, and using Phosphorous Liqui-UV (#0380-125) from StanBio. Serum iron and total iron binding capacity were determined with commercial kits (Biovision, K392). Transferrin saturation was calculated as serum iron/total iron binding capacity.

Nonheme iron concentrations of liver (the major site of body iron stores) spleen (the major site of iron recycling of RBC), and kidney (due to its relevance to this model) were determined using batho-phenanthroline detection.⁽³⁷⁾

Purified recombinant IL-1β was purchased from Novus (#NBP2-35109). Neutralizing antibody against mouse IL-1β was purchased from R&D (#MAB4012). Neutralizing antibody was administered by IP injection and diluted to the indicated dose in sterile PBS. Mice recovered in their home cages until sacrifice. Neutralizing anti-TNF-α was from BioLegend (# 506331), and anti-NGAL was from R&D (#MAB1857).

RNA was collected using the Qiagen miRNeasy kit (#74104) and reverse transcribed using BioRad iScript cDNA synthesis kit (#1708891). Real time PCR was completed on an ABI One Step machine with Kapa Universal Sybr fast reagents.

All charts are expressed as boxplot with box height representing first and third quartile values, and the median shown by the bar across the middle. Plungers indicate the min and max values for the dataset. We evaluated differences between groups through student’s two tailed t-test assuming equal variance. Groups with p values less than 0.05 have the exact p values displayed on the graphs with brackets to highlight the groups being directly compared.

RESULTS

FGF-23 is upregulated in advance of any perturbations to the PTH/Phosphate/1,25(OH)₂D₃ Axis

To probe the relationship between anemia, FGF-23, phosphate homeostasis and inflammation during the early stages of CKD, we employed mice with targeted deletion of EBF1. EBF1 is a transcription factor, and was recently discovered by us to play an important role in renal development.⁽³⁸⁾ The loss of EBF1, via either complete knockout or specific deletion within the kidney stromal progenitors (utilizing Foxd1-cre with Floxed-Ebf1 mice, Foxd1-cre:Ebf1^fl/fl), results in poor development of the outer cortex and hypoplastic peripheral glomeruli during the final stages of kidney maturation.⁽³⁸⁾ In these mice renal development is abrogated between postnatal day 10 and 14 (P10-14 (Figure 1A). Albuminuria develops in the Foxd1-cre:Ebf1^fl/fl animals by P21. This is a highly penetrant model of low nephron endowment- CKD and because of the fulminant progression and temporal consistency of its onset it is an ideal model to investigate the early stages of CKD. A detailed description of the pathology of the Foxd1-cre:Ebf1^fl/fl animals has been published elsewhere⁽³⁹⁾ but a brief description of the timing of renal decline of these animals is provided in Figure 1A.

Figure 1: Abrogation of the PTH/Phosphate/Vitamin D₃ axis does not coincide with upregulation of iFGF-23 in early CKD.

(A) A brief summary of the onset of symptoms in the low nephron endowment CKD model we used (Foxd1-cre:Ebf1^fl/fl mice). (B-C) Measurement of the induction of iFGF-23 in the circulation of Cre⁺ control littermates (Ctrl), floxed control animals (Ebf1^fl/fl) and the CKD model (Foxd1-cre:Ebf1^fl/fl). RNA expression from bones was also measured. n=5-7 animals/genotype (D) Furin activity in the serum of Foxd1-cre:Ebf1^fl/fl mice at the ages indicated. n = 4 samples/genotype (E) Serum phosphorous at the onset of CKD. n=5-7 animals/genotype (F) Serum PTH as CKD progresses in Foxd1-cre:Ebf1^fl/fl mice. n=5 animals/genotype (G-H) Renal mRNA production of Vitamin D₃ activating and inactivating hydroxylation enzymes. n=5-7 animals/genotype (I) Circulating active Vitamin D₃ in early CKD. n=5-7 animals/genotype.

We first began by investigating when FGF-23 was upregulated in Foxd1-cre:Ebf1^fl/fl animals. Intact FGF-23 (iFGF-23) was elevated at P14 as soon as kidney development appeared abrogated (Figure 1B). This coincided with an increase in FGF-23 transcripts in bones. (Supplemental Figure 1 details the expression of the Foxd1-cre within the skeleton). There are several sites through the body that can induce the expression of FGF-23 apart from osteocytes. We examined heart, thalamus, hippocampus, thymus, spleen, and kidney tissue from P14 Foxd1-cre:Ebf1^fl/fl mice and could not detect an increase in extra-skeletal FGF-23 transcripts at P14 (Supplemental Figure 2). FGF-23 turnover through furin-mediated cleavage was also examined and not found to be inhibited at this early stage (Figure 1D).

Although FGF-23 was elevated at P14 in Foxd1-cre:Ebf1^fl/fl mice, neither hyperphosphatemia nor hyperparathyroidism were measurable at this age (Figure 1E–F). Animals do become slightly hypophosphatemic, in the face of this elevated iFGF-23, which is a condition that presents in the patient population as GFR declines through the early stages of CKD.⁽⁷⁾ Although renal expression of the 25-hydroxylase (Cyp27b1) decreases by P21, and the inactivating 24-hydroxylase (Cyp24a1) is increased at P14 and beyond, active vitamin D3 is not systemically altered in these mice at P14 (Figure 1G–I). This demonstrates that the FGF-23 profile in this model of early CKD more closely resembles a pathology of primary FGF-23 elevation and not a secondary PTH/phosphate/1,25(OH)₂D₃ mediated pathology.

Anemia is apparent 7-10 days after the induction of FGF-23 in early CKD

To determine if anemia was stimulating FGF-23 production in early CKD, we characterized the appearance of iron deficiency. When mice were 24 days old Foxd1-cre:Ebf1^fl/fl animals are anemic by hematocrit (Figure 2A). Hemoglobin content trended down, but wasn’t significantly changed, while erythrocyte numbers (RBC) similarly trended down at P21 but were different between groups by P24 (Figure 2B–C). This decrease in the total erythrocytes was reflected by a matched decrease in mean corpuscular volume (MCV) but there was no change in the hemoglobin content at a per cell basis (Mean corpuscular hemoglobin (MCH)) (Figure 2D–E). A main signal for the stimulation of erythropoiesis is the production of Epo from the kidney. We measured circulating Epo and found it was elevated in serum of Foxd1-cre:Ebf1^fl/fl animals at P24 but not before (Figure 2F). This correlates with the observable drop in erythrocytes and is 10 days after FGF-23 was induced. Anemia can be accompanied by extramedullary hematopoiesis manifested as enlargement of the spleen. Splenomegaly was also observed, but not prior to P28 (Figure 2G). While the mice do develop hematuria, it only is apparent once the mice are 5 weeks old (detected by dipstick, data not shown), and therefore this may be exacerbating the anemia, but is not an initiating event. These data reveal that anemia develops in the Foxd1-cre:Ebf1^fl/fl animals between 3 and 4 weeks of age.

Figure 2: Anemia develops between 3 and 4 weeks of age.

(A-E) Whole blood parameters were measured at either P21 or P24 in Foxd1-cre:Ebf1^fl/fl mice and controls. (HCT = hematocrit; RBC = red blood cell number; MCH = mean corpuscular hemoglobin; MCV = mean corpuscular volume) n=5-6 animals/genotype (F) Serum Epo was measured at the ages indicated. n=4 animals/genotype (G) Spleen weight expressed as a percentage of total body weight. n=5-7 animals/genotype.

Distinction needs to be made with any anemia model as to whether the phenomena is arising from an absolute iron deficiency (when total body stores are low) or a functional iron deficiency (total stores aren’t low, but there is some defect that prevents adequate utilization of iron for erythropoiesis). The tissue iron levels in spleen, liver, and kidney were examined at P21 and found to be unchanged Foxd1-cre:Ebf1^fl/fl mice (Figure 3A). Next, serum iron, was examined and found to be normal at P14, trending down at P21, and significantly decreased by P28 (Figure 3B). Free iron is carried through the serum by transferrin, and can be lost to the urine when the integrity of the glomerular basement membrane (GBM) is compromised during CKD. Coincident with the decrease in serum iron transferrin begins to be lost in the urine at P21 (Figure 3C–D). The percentage of transferrin bound to iron, was not significantly different between genotypes at either P14 or P21 (Figure 3E). Hepcidin regulates the release of iron from tissue stores and absorption of dietary iron from the intestines. Hepcidin mRNA production (Hamp) from the liver trended up at P14 (Figure 3F), and was significantly elevated by P21. Circulating hepcidin was elevated at P14 (Figure 3G), but this was not because it was actively being retained by the kidney (Figure 3H). Taken together these data reveal an absolute iron deficiency developing after the onset of CKD these mice. They are anemic by P24 and are attempting to compensate for an iron imbalance as early as P21, but all parameters of anemia development are normal until well after the induction of iFGF-23.

Figure 3: Anemia is not detectable until a week after the induction of FGF-23.

(A) Non-heme iron in tissues normalized per gram of tissue weight at P21. n=6-9 animals/genotype (B) Serum iron in the circulation. (C-D) Leakage of transferrin out of the circulation. (C) and into the urine (D) as CKD progresses. n=5-7 animals/genotype (E) TIBC and serum iron content were used to calculate percent transferrin bound in the circulation n=4 animals/genotype (F-H) Liver production of hepcidin transcripts. (F), circulating hepcidin in the serum (G) and the leakage of hepcidin into the urine (H) were measured. n=5-7 animals/genotype.

IL-1β production from the kidney correlates with FGF-23 expression

Having demonstrated that neither altered PTH/Phosphate/1,25(OH)₂D₃ nor anemia correlate with circulating iFGF-23 in early CKD, we next investigated for the presence of several inflammatory proteins that are known to be upregulated during CKD and have been demonstrated to regulate FGF-23 production. While production of Il6 transcripts were induced over 10 fold in the Foxd1-cre:Ebf1^fl/fl kidneys at P14, this never translated into a measurable accumulation of protein in the circulation at these early times (Figure 4A,B). Similarly, C-reactive protein was undetectable in the serum at P14 (Figure 4C). When TGF-β was profiled, both the RNA and detectable circulating levels were present at P21 (Figure 4D–E). TNF-α was elevated at the transcript level at P14 and P21, but it was not detectable in the circulation until P21 (Figure 4F–G). IL-1β, however, had an increase in both transcript production from the kidney and circulating protein at P14 (Figure 4H–I). IL-1α, a related interleukin that utilizes the same IL-1R1 receptor as IL-1β, was not upregulated until P21 and was not detectable at appreciable levels in the serum (Figure 3J–K). Of all the cytokines examined, only IL-1β had increased production within the injured kidney that then translated into an appreciable increase in circulating cytokine level that could reach the osteocytes to induce FGF-23 production. Furthermore, IL-1β is also known to stimulate the production of hepcidin from hepatocytes,⁽⁴⁰⁾ and may be why hepcidicn production increases before anemia of any form is detectable in our animals. It remains unclear why IL-6, which is also regulated by IL-1β, would not be increased in the circulation, but it is possible that although the mRNA is being stimulated this is not being translated into functional protein at a similar rate, or is being produced locally within the kidney, but not at a sufficient level to be detectable in the serum at this early stage of CKD. A summary chart for all the regulators of FGF-23 that were examined is provided in Figure 5.

Figure 4: Upregulation of IL-1β is apparent at the onset of CKD in Foxd1-cre:Ebf1^fl/fl mice.

Production of mRNA transcripts for the indicated cytokines (A, D, F, H, and J n=6-11 mice/treatment group) as well as the circulating levels of active cytokine (B, C, E, G, I, and K, n=5-6 mice/treatment group) are shown for Foxd1-cre:Ebf1^fl/fl mice at the ages indicated.

Figure 5: Development of CKD and onset of potential inducers of FGF-23 in Foxd1-cre:Ebf1^fl/fl mice.

Indicators of CKD progression are shown in light blue, disruptions of the PTH/phosphate/Vitamin D₃ axis in mauve, anemia progression in dark blue, and inflammatory cytokines in olive.

IL-1β controls osteocytic FGF-23 production in early CKD

Having correlated the expression of circulating IL-1β to induction of FGF-23 we next examined whether it controls production of iFGF-23 from osteocytes. Bone chips from normal C57Bl/6 mice were isolated and treated in vitro with doses of IL-1β equivalent to what we observed in vivo at P14 (0–50 pg/mL). Both transcripts of Fgf23 increased in the cultured cells, and iFGF-23 protein accumulated in the culture media (Figure 6A,B). We next injected C57Bl/6 mice with IL-1β and saw a transient increase in Fgf23 transcripts and iFGF-23 protein over 18 hours (Figure 6C,D). This demonstrates that FGF-23 is regulated by IL-1β both in vivo and in vitro. To prove that FGF-23 is regulated by IL-1β during the initial stages of CKD onset, we next used neutralizing antibody to IL-1β to optimize the concentration of antibody needed to block the upregulation of Fgf23 transcripts using in vitro bone chips (Figure 6E). We next exposed those bone chips to serum from mice with CKD (Foxd1-cre:Ebf1^fl/fl animals at P14, “EBF1-CKO serum”). In the presence of anti-IL-1β neutralizing antibody (αIL-1β) blocks upregulation of Fgf23 transcripts from the bone chips (Figure 6F). This was specific to the IL-1β-targeted antibody and not accomplished by blocking other inflammatory mediators including anti-TNFα⁽⁴¹⁾ (α-TNFα) nor anti-NGAL⁽⁴²⁾ (α-NGAL) antibodies (efficacy of these doses is shown in Supplemental Figure 3). Upregulation of FGF-23 by IL-1β during CKD was further supported by suppression of iFGF-23 within Foxd1-cre:Ebf1^fl/fl animals that were administered IL-1β neutralizing antibody from P11-P14 (Figure 6G).

Figure 6: IL-1β modulates FGF-23 expression from osteocytes during renal damage.

(A-B) Induction of Fgf23 transcripts (A) and iFGF-23 in culture media (B) from bone chips. n=4 wells/condition (C-D) Mice injected with IL-1β (100 ng/mouse) had long bones assayed for Fgf23 message production (C) and circulating iFGF-23 (D). n=6 mice/treatment group (E) Bone chips were treated with purified IL-1β in the presence of increasing concentrations of neutralizing antibody to IL-1β (0.1, 0.25, 0.5, and 1.0 μg/mL). n=4 wells/condition (F) RNA expression of Fgf23 was examined after serum from mice with early CKD was added to bone chips in culture. Neutralizing antibodies to either IL-1β, NGAL, or TNFα were included alone or in combination. n=4 well/condition. (G) Serum iFGF-23 was assayed after Foxd1-cre:Ebf1^fl/fl mice were administered daily IP injection of IL-1β neutralizing antibody from P11-14. n=6 mice/treatment group (H) Cultured bone chips were exposed to serum from either Foxd1-cre:Ebf1^fl/fl mice, nephritic serum collected 24 hours after administration of nephrotoxic α-GBM solution, or nephritic serum where the mice were also treated with IL-1β neutralizing antibody in vivo. n=4 wells/condition (I) Circulating iFGF-23 was measured from mice induced with nephrotoxic serum alone or with co-administration of IL-1β neutralizing antibody. n=6 mice/condition (J) Skeletal production of FGF-23 message was measured from mice whom underwent IRI alone or with administration of IL-1β neutralizing antibody. n=5 mice/condition (K) Circulating iFGF-23 was measured from mice induced with IRI alone or with administration of IL-1β neutralizing antibody. n=5 mice/condition (M-N) Cultured bone chips were exposed to serum from serum collected 72 hours after administration of IRI, and then treated in vitro with IL-1β neutralizing antibody. Production of FGF23 mRNA (M) and intact protein (N) were measured. n=4 wells/condition.

There are many mechanisms through which CKD can be precipitated, yet upregulation of iFGF-23 occurs in all of them. To confirm that IL-1β was not specific to just the congenital nephropathy associated with EBF1 deficiency, we examined mice treated with anti-GBM targeted sera to initiate glomerulonephritis (α-GBM). First, bone chips were treated in vitro with serum collected from either Foxd1-cre:Ebf1^fl/fl mice at P14, or serum from C57Bl/6 animals 18 hours after injection with α-GBM . Combinatorial suppression of IL-1β and TNF-α was required to significantly reduce Fgf23 message stimulated by nephritis serum in vitro, while neither demonstrated a significant reduction independently (Figure 6H). We next collected serum from mice that were administered α-IL-1β antibody at the same time that they were given anti-GBM, and found that inhibiting IL-1β activation in vivo diminished the cytokine activation cascade and also greatly diminished the ability of this nephritic serum to induce Fgf23 (Figure 6I, Supplemental Figure 4). As expected, addition of the Il-1β neutralizing antibody had no effect on IL-1β expression. Other inflammatory cytokines downstream of IL-1β signaling, TNF-α and IL-6 (Supplemental Figure 4), were also decreased or trended down, indicating that the effect IL-1β inhibition may have other beneficial consequences related to the role of IL-1β in the inflammatory cascade.

To further validate the role of IL-1β in the regulation of FGF-23 at the onset of kidney injury, we used two models of acute kidney injury: used ischemia reperfusion (IRI) and cisplatin toxicity. Three days after IRI injury both production of Fgf23 transcripts within the bones and circulating iFGF23 were reduced in animals given the neutralizing antibody against IL-1β (Figure 6J–K). Furthermore, incubating bone chips in vitro with serum from mice who had undergone IRI caused an upregulation of FGF-23 production that was neutralized by the addition of IL1β inhibitor (Figure 6L–M). Similar results were seen following cisplatin injury (Supplemental Figure 5). Together these data demonstrate that IL-1β controls the expression of FGF-23 during the initial stages of CKD in four disparate models of kidney injury.

DISCUSSION

We report evidence supporting a regulatory role of IL-1β on the production of circulating iFGF-23 from direct stimulation of the osteocytes during the onset of CKD. Through detailed investigation of the appearance of markers of anemia, disrupted PTH/phosphate/1,25(OH)₂D₃ signaling and inflammatory markers we identified systemic IL-1β as the primary driver of FGF-23 production. This is the first examination of all of these potential regulators of FGF-23 in a single CKD model, and highlights the important role for inflammatory signals in the initiation of FGF-23 expression during kidney injury. Development of anemia, the presence of other inflammatory markers at later stages, and eventual disruption of the phosphate axis can all further exacerbate FGF-23 expression, but it is IL-1β that initiates FGF-23 expression at the onset of CKD. The relative contributions of each of the other pathways to the total FGF-23 produced will need to be teased apart as a focus of future investigations.

It has been repeatedly shown that higher FGF-23 correlates to more progressive CKD and worse outcomes for patients.^(43–46) It also has been generally accepted that CKD is a condition of secondary FGF-23 excess. This is based mostly on observations from patients, the majority of which are not under observant care for their CKD until they are fairly advanced to stage 3, 4 or 5, and already have lost significant GFR. FGF-23 expression is enhanced though elevation of serum phosphate, secondary hyperparathyroidism and disrupted production of 1,25(OH)₂D₃, but effective treatment of the patient population requires a detailed understanding of the progression of the pathology underlying organ dysfunction at all stages, and prior to our investigations no one directly investigated FGF-23 stimulation during the initial stages of CKD onset. By using these mouse models we could time the onset of renal decline precisely and reproducibly. Our analysis of the perturbations to the phosphate and FGF-23 signaling pathway present in EBF1-deficient mice revealed a model that is reminiscent of a primary FGF-23 elevation, with early CKD resembling a pathology of primary FGF-23 excess ⁽⁴⁷⁾.

This is significant because our identification of IL-1β as a primary driver of FGF-23 in early CKD works cooperatively with its known role as also a potent stimulator of systemic inflammation. Stimulation of IL-1β signaling within the kidney itself is also already known to exacerbate the progression of kidney injury through activation of the NLRP3 inflammasome. Potentially, targeting IL-1β should have a multifactorial effect in improving CKD-related outcomes. It would not only reduce FGF-23 production, but also minimize the inflammatory response at renal and extra-renal locations. A clinical trial published this year using the IL-1β neutralizing antibody Canakinumab showed some promise to this approach.⁽⁴⁸⁾ These authors demonstrated that in a study of patients who had a previous myocardial infarction with CKD, use of Canakinumab reduced the risk of major adverse cardiovascular events, but also did not demonstrate either a substantive clinical benefit or harm with regard to CKD progression. They did not, however, look at circulating FGF-23. Inhibition of IL-1β is approved for patients with gout, and although the studies reported to date are small, promising improvement in kidney function observed in patients with gout and diabetic nephropathy with blockade of IL-1β.⁽⁴⁹⁾

The cardiovascular benefit of IL-1β neutralization is especially interesting with respect to FGF-23 regulation. Recent animal trials utilizing a neutralizing antibody against FGF-23 to minimize CKD consequences was effective at reducing secondary hyperparathyroidism, and normalizing serum 1,25(OH)₂D_3, but also resulted in aggravated disruption of serum phosphate and increased cardiovascular event-mediated death.⁽⁵⁰⁾ This suggests that although FGF-23 levels do correlate to a worse clinical outcome in CKD,⁽¹⁰⁾ targeting it directly would not be feasible based on the presence of increased cardiac incidents. Although the first is related to atherosclerosis and the second increases the incidence of atherocalcinosis, theoretically, using Canakinumab might both improve adverse heart events and decrease FGF-23. Furthermore, several case reports of patients with Hyperphosphatemic Familial Tumoral Calcinosis, who have a dysregulated FGF-23 and extra-skeletal calcification complications as a result, found a benefit for two patients treated with Canakinumab, Anakinra, or other IL-1β antagonists including resolution of their calcinosis cutis,^(51,52) This suggests the possibility that inhibiting IL-1β in this disease of broken FGF-23 signaling may be beneficial to limiting the unresolved signaling loop that creates this pathology. The parallels of this genetic condition to the unregulated mineral balance in late CKD are promising for a potential benefit of targeted IL-1β blockade in CKD patients.