To examine FGF23 inflammatory actions in vivo, we explored pathologic FGF23-FGFR4 signaling and its role in functional iron deficiency. We subjected wild-type (Fgfr4^+/+) and constitutive FGFR4 knockout (Fgfr4^−/−) mice to adenine diet to induce CKD. Following adenine diet for 14 weeks, Fgfr4^+/+ and Fgfr4^−/− mice displayed comparable renal dysfunction, as shown by elevations in blood urea nitrogen (BUN) and serum creatinine levels (Figure 1A). On adenine diet, serum FGF23 and Pi levels significantly increased in both genotypes (Figure 1B), as expected with marked kidney injury. No significant changes in serum calcium levels were observed between genotypes on adenine (Figure 1—figure supplement 1A), despite increased serum Pi levels (Figure 1B).

We assessed gene expression of inflammatory cytokines and acute phase proteins in both genotypes in adenine-induced CKD. Unlike in healthy control mice, liver IL1β (Il1b), IL6 (Il6), and serum amyloid A1 (Saa1) transcript levels were significantly and similarly elevated (Figure 1C, D) in both Fgfr4^+/+ and Fgfr4^−/− mice.

To identify if FGF23-FGFR4 signaling contributes to functional iron deficiency, we evaluated liver hepcidin (Hamp) mRNA and hematological responses. Compared with control mice, Hamp transcript levels were significantly and similarly elevated in both Fgfr4^+/+ and Fgfr4^−/− mice on adenine (Figure 1E). Complete blood count and serum analyses displayed significant reductions in red blood cell (RBC) count, mean corpuscular volume (MCV), hemoglobin, and serum iron levels on adenine diet (Figure 1F). Significant reductions in hematocrit percentage (HCT%), mean corpuscular hematocrit (MCH), and serum transferrin saturation percentage (TSAT%) were also observed (Figure 1—figure supplement 1B). Spleen tissue sections stained with Perls’ Prussian blue revealed profound intracellular iron sequestration, indicating severe RES blockade in both Fgfr4^+/+ and Fgfr4^−/− mice on adenine (Figure 1G, Figure 1—figure supplement 1C). Taken together, our data demonstrate that FGF23-FGFR4 signaling does not affect inflammation, the acute phase response or functional iron deficiency and anemia in adenine-induced CKD.

Excess Pi is a hallmark of CKD, but direct pathologic effects of Pi on tissues other than the vasculature are poorly understood (Komaba and Fukagawa, 2016; Scialla and Wolf, 2014). To examine if liver Pi deposition is increased in CKD, we analyzed liver Pi levels in adenine-induced mouse CKD model by colorimetric quantification. Hepatic Pi levels were elevated in both Fgfr4^−/− and Fgfr4^+/+ mice following adenine (Figure 1H) but less so in Fgfr4^−/− mice when compared to Fgfr4^+/+ mice.

To establish whether excess Pi and/or FGF23 contributes to hypoferremia in the absence of CKD, we exposed Fgfr4^+/+ and Fgfr4^−/− mice to a graded dietary Pi load for 12 weeks. Serum FGF23 levels increased in both genotypes on 2% Pi and 3% Pi diet, in comparison to mice on 0.7% Pi diet (Figure 2A). Despite 2% Pi increasing serum FGF23, serum Pi levels significantly increased only on 3% Pi, in comparison to mice fed 0.7% Pi (Figure 2A). Notably, these serum Pi levels are comparable to the elevated serum Pi levels observed in adenine-induced CKD (Figure 1B). No significant differences in serum calcium levels were observed between genotypes (Figure 2—figure supplement 1A), despite elevated Pi levels (Figure 2A). No pathologic changes were detected in kidneys regardless of genotype as BUN, serum creatinine, and kidney tissue sections stained with hematoxylin and eosin (H&E) appeared similar to those of mice on 0.7% Pi (Figure 2—figure supplement 1B–D). Interstitial fibrosis was not detected in kidneys as shown by Masson’s trichrome staining (Figure 2—figure supplement 1E). These data indicate elevations in serum levels of FGF23 and Pi in mice on 2% Pi or 3% Pi diet are consequences of an increasing dietary Pi load and not renal dysfunction.

As a high Pi diet has been reported to exacerbate inflammation and serum FGF23 levels (Sugihara et al., 2017; Takashi et al., 2019; Yamada et al., 2014), we evaluated gene expression of inflammatory cytokines and acute phase proteins. Compared to 0.7% Pi, liver Il1b, Il6, and Saa1 transcript levels significantly increased in both Fgfr4^+/+ and Fgfr4^−/− mice on 3% Pi, although not on 2% Pi diets (Figure 2B, C). Liver injury was not detected, as no significant elevations in hepatic alanine aminotransferase (Alt1) or aspartate aminotransferase (Ast1) mRNA levels were found on 3% Pi (Figure 2—figure supplement 2A, B). These data support the notion that dietary Pi overload induces inflammation, but not via FGF23-FGFR4 signaling.

To explain these effects of 3% Pi diet and determine if increased tissue Pi deposition is associated with adverse outcomes, we analyzed the relationship between liver and serum Pi levels in both Fgfr4^+/+ and Fgfr4^−/− mice following a graded dietary Pi load. A positive correlation was detected between hepatic and serum Pi levels in both genotypes, beginning with 2% Pi (Figure 2D). These results show liver Pi deposits increase following elevations in dietary Pi content.

Next, we tested if increased liver Pi accumulation affects correlations between liver Pi and liver Hamp mRNA levels, as a high Pi diet induces Hamp expression (Nakao et al., 2015). A positive correlation between liver Pi and liver Hamp mRNA levels were detected in both Fgfr4^+/+ and Fgfr4^−/− mice, again only with diets containing 2% or 3% Pi (Figure 2E). As liver injury was not detected (Figure 2—figure supplement 2A, B), these data indicate that elevations in liver Hamp mRNA are independent of liver injury and may result from Pi-driven inflammation.

We next explored if increased dietary Pi loading led to changes in hematological responses. Marked reductions in RBC, MCV, hemoglobin, and serum iron levels were detected on 3% Pi and were similar in both Fgfr4^+/+ and Fgfr4^−/− mice (Figure 2F). HCT% and MCH were also significantly decreased (Figure 2—figure supplement 2C). Supporting these findings, spleen tissue sections revealed that both Fgfr4^+/+ and Fgfr4^−/− mice showed increased intracellular iron deposits on 3% Pi (Figure 2G, Figure 2—figure supplement 1F). Thus, dietary Pi loading causes iron restriction and hypoferremia even in the absence of CKD.

As inflammation is a known contributor of muscle wasting (Raj et al., 2008; Schaap et al., 2006; Verzola et al., 2017), and since hyperphosphatemia and excess FGF23 are associated with inflammation, we explored whether hyperphosphatemia contributes to skeletal muscle wasting, and if pathologic FGF23-FGFR4 signaling is involved in these effects. We analyzed skeletal muscle from Fgfr4^+/+ and Fgfr4^−/− mice exposed to adenine-induced CKD (Figure 1) or a graded dietary Pi load (Figure 2). Examination of skeletal muscle strength indicates that mice on adenine or 3% Pi diet exhibit reduced grip strength in both Fgfr4^+/+ and Fgfr4^−/− mice, in comparison to respective control mice (Figure 3A). A reduction in gastrocnemius mass was also observed (Figure 3B). Notably, gastrocnemius metallothionein-1 (Mt1) transcript levels were significantly elevated in both genotypes following adenine and 3% Pi diets (Figure 3—figure supplement 1A, B), indicating that either condition fosters skeletal muscle abnormalities.

We next investigated if these muscle deficits resulted from inflammation inducing myostatin and downstream atrophy-related gene programs, as both experimental models display elevated levels of liver Il1b and Il6 (Figures 1C and 2B). Compared to respective control mice, gastrocnemius myostatin (Mstn) transcript levels were significantly elevated in both Fgfr4^+/+ and Fgfr4^−/− mice following adenine or 3% Pi diet (Figure 3C). Additionally, both genotypes on 2% Pi showed an increased trend in Mstn mRNA levels (Figure 3C). As these findings suggest increased myofibrillar protein degradation, we further analyzed the expression of two specific ubiquitin ligases of muscle-protein breakdown, muscle RING-finger protein 1 (Murf1) and Atrogin-1. Compared with their respective control mice, gastrocnemius Murf1 and Atrogin1 transcript levels were significantly elevated in both Fgfr4^+/+ and Fgfr4^−/− mice following adenine and 3% Pi diets (Figure 3D).

As elevated myostatin and increased myofibrillar protein degradation are features of skeletal muscle wasting, we assessed if these results prompt a shift toward smaller myofibers. Indeed, gastrocnemius tissue sections stained with H&E from Fgfr4^+/+ and Fgfr4^−/− mice, on either adenine or 3% Pi diet, showed smaller muscle fiber size compared with controls (Figure 3E). Taken together, these data suggest skeletal muscle wasting in adenine-induced CKD and hyperphosphatemia does not require FGF23-FGFR4 signaling.

Alport (Col4a3^−/−) mice are an established model of progressive CKD which develop hyperphosphatemia with severe inflammation, hypoferremia, and anemia (Francis et al., 2019). To test if hyperphosphatemia aggravates these pathologic complications, we exposed Col4a3^−/− mice to a low Pi diet treatment (0.2% Pi) for 6 weeks. In comparison to wild-type (Col4a3^+/+) mice on normal diet (0.6% Pi), Col4a3^−/− mice showed renal dysfunction by increased BUN and serum creatinine levels (Figure 4A). Col4a3^−/− mice on low Pi diet displayed a reduction in both parameters (Figure 4A), along with reduced pathologic alterations in kidney morphology (Figure 4—figure supplement 1A). As compared to wild-type mice, serum levels of FGF23 and Pi significantly increased in Col4a3^−/− mice on normal diet (Figure 4B), but less so in Col4a3^−/− mice on low Pi diet (Figure 4B).

To identify if a low Pi diet affects inflammation or the acute phase response in Alport mice, we assessed gene expression of inflammatory cytokines and acute phase proteins. Compared to wild-type mice on normal diet, liver Il1b, Il6, and Saa1 transcript levels were significantly elevated in Col4a3^−/− mice but much less elevated on 0.2% Pi (Figure 4C, D). These data support the notion that excess Pi in Alport mice aggravates inflammation.

As our results show that a low Pi diet decreases inflammation, we next explored its impact on functional iron deficiency. Compared to wild-type mice on normal diet, liver Hamp transcript levels were significantly elevated in Col4a3^−/− mice with complete reversal by low Pi diet (Figure 4E). Assessing hematological responses, Col4a3^−/− mice on normal diet were anemic, with significant reductions in RBC count, MCV, hemoglobin, and serum iron levels (Figure 4F), as well as in HCT%, MCH, and TSAT% (Figure 4—figure supplement 1B). Col4a3^−/− mice on normal diet displayed profound intracellular iron sequestration in spleen (Figure 4G, Figure 4—figure supplement 1E), along with excessive spleen and liver nonheme iron levels (Figure 4—figure supplement 1C, D). These effects were substantially ameliorated by low Pi diet in Col4a3^−/− mice, with improved hematologic parameters and reduced iron deposits in spleen (Figure 4F, G). Nonheme iron levels in spleen and liver were reduced in Col4a3^−/− mice by treatment, indicating increased iron mobilization and decreased iron restriction (Figure 4—figure supplement 1C, D). These data indicate that dietary Pi restriction improves hematological responses and alleviates hypoferremia.

As liver Pi accumulation is increased in adenine-induced CKD (Figure 1H), we next explored if low Pi diet treatment reduces pathologic liver Pi deposits in progressive CKD. Compared to wild-type mice on normal diet, liver Pi levels were increased in Col4a3^−/− mice and were reduced on low Pi diet (Figure 4H). Taken together, our data demonstrate that Pi restriction as a dietary intervention, in a genetic model of progressive CKD, reduces pathologic Pi accumulation in the liver and alleviates the severity of renal injury and functional iron deficiency.

To determine if reducing hyperphosphatemia limits skeletal muscle wasting, an important complication of CKD (Verzola et al., 2019), we analyzed skeletal muscle from wild-type (Col4a3^+/+) and Alport (Col4a3^−/−) mice subjected to either a normal diet (0.6% Pi) or a low Pi diet treatment (0.2% Pi) (Figure 3). Compared to wild-type mice on normal diet, Col4a3^−/− mice showed significant reduction in grip strength which was improved by low Pi diet (Figure 5A). Gastrocnemius mass was also reduced in Col4a3^−/− mice, and treatment tended to improve muscle weight (Figure 5B). In particular, gastrocnemius Mt1 transcript levels were significantly elevated in Col4a3^−/− mice compared to wild-type mice on normal diet and were reduced by low Pi diet (Figure 3—figure supplement 1C). These results suggest that Pi restriction as a dietary intervention may improve skeletal muscle abnormalities in CKD. Furthermore, compared to wild-type mice on normal chow, Col4a3^−/− mice displayed significant elevations in gastrocnemius Mstn transcript levels which was reduced by treatment (Figure 5C). Additionally, Col4a3^−/− mice on normal diet showed increased gastrocnemius Murf1 and Atrogin1 transcript levels, which were also reduced by low Pi diet (Figure 5D). These data support the notion that dietary Pi restriction, as a treatment in Col4a3^−/− mice, reduces myostatin synthesis and subsequent atrophy-related gene programs.

As with adenine-induced CKD mice, or mice fed high phosphate diet, gastrocnemius tissue sections from Col4a3^−/− mice showed smaller muscle fiber size compared to wild-type controls (Figure 5E). However, Col4a3^−/− mice fed low Pi diet showed improved muscle fiber size (Figure 5E). Taken together, these data suggest that hyperphosphatemia affects skeletal muscle wasting in Alport mice, possibly by exacerbating systemic inflammatory cytokine concentrations and their catabolic effects on muscle.

Having shown that inflammation, hypoferremia, and muscle wasting induced by high Pi are independent of FGF23-FGFR4 signaling, we tested if Pi directly affects inflammatory cytokine and hepcidin expression in mouse primary hepatocytes.

We first analyzed the expression profile of the three families of Na/Pi cotransporters (types I–III). Quantitative polymerase chain reaction (qPCR) analysis detected high levels of Slc20a1and Slc20a2 (encoding for PiT-1 and PiT-2), but not Slc17a1 and Slc17a3 (encoding for Npt-1 and Npt-4), or Slc34a1, Slc34a2 and Slc34a3 (encoding for NaPi-2a-c) (Figure 6—figure supplement 1A). This analysis indicates that type III Na/Pi cotransporters are the predominant Na/Pi family in primary hepatocyte cultures.

Based on studies demonstrating that high extracellular Pi activates signaling pathways such as MAPK and NFκB (Chavkin et al., 2015; Zhao et al., 2011), we assessed if MAPK, STAT3, and NFκB signaling are activated in cultured hepatocytes in response to treatments with FGF23 or graded concentrations of Pi. Treatment with TNFα or IL6 was used as a positive control for activation of these established networks regulating inflammatory gene expression. Immunoblot analysis of ERK1/2, STAT3, and NFκB showed that Pi treatments increased phosphorylated NFκB levels without changing total NFκB expression (Figure 6A, Figure 6—source data 1A). Increased concentrations of Na₂SO₄, a salt generating another anionic species, had no effect on phospho-NFκB levels, indicating this response was specific to elevated Pi and not an unspecific response to increased anions. Pi treatment did not affect pERK1/2 or pSTAT3, and FGF23 had no effect on any of the pathways examined.

We next analyzed gene expression by qPCR of inflammatory cytokines and acute phase proteins in isolated hepatocytes treated with increasing concentrations of Pi or Na₂SO₄ with LPS and IL6 treatments used as positive control. Elevations in Il1b, Il6, and Saa1 transcript levels were noted not only following LPS and IL6 but also Pi treatments (Figure 6B, C). Treatments with Na₂SO₄ had no effect on gene expression. As inflammation is a known mediator of hepcidin synthesis, we analyzed Hamp mRNA levels. LPS, IL6, and Pi treatments all elevated Hamp transcript levels when compared to control (Figure 6D). These data indicate that high extracellular Pi can act on hepatocytes to increase the synthesis of inflammatory cytokines and hepcidin.

Given that inflammation and NFκB signaling regulate PiT-1 expression (Koumakis et al., 2019) and primary hepatocytes express Slc20a1 and Slc20a2 (Figure 6—figure supplement 1A), we examined if Slc20a1 and/or Slc20a2 mRNA levels were altered following LPS, IL6, or Pi treatment. Expression analysis showed that Pi significantly increased Slc20a1 transcript levels in a dose-dependent manner but had no effect on Slc20a2 expression (Figure 6E, Figure 6—figure supplement 1B). To determine if this result is an action of hepatocytes sensing high extracellular Pi, we cotreated hepatocytes with phosphonoformic acid (PFA), a compound which is reported to inhibit chemisorption of calcium–Pi clusters, as aggregate formation is a byproduct of increased extracellular Pi (Villa-Bellosta et al., 2007; Villa-Bellosta and Sorribas, 2009). In the presence of PFA, Pi-induced Slc20a1 expression was reduced compared to vehicle-treated control cells (Figure 6F). Interestingly, PFA also altered the effects of LPS and IL6 on Slc20a1 expression. To confirm whether inhibiting high extracellular Pi and/or aggregate byproducts disrupts downstream actions of increased extracellular Pi, we cotreated hepatocytes with Pi and PFA, and observed that PFA interfered with Pi-induced effects on phospho-NFκB levels without changing total NFκB expression (Figure 6G, Figure 6—source data 1G).

Furthermore, when isolated hepatocytes were treated with either LPS, IL6, or Pi in the presence or absence of PFA, the significant elevations in Il1b, Il6, Saa1, and Hamp transcript levels following Pi treatments were reversed in the presence of PFA (Figure 6H–J). Interestingly, PFA slightly altered the effects of LPS on Il6 expression and IL6 on Hamp expression (Figure 6H–J). Taken together, our results indicate that in primary hepatocyte cultures type III Na/Pi cotransporters are predominant, and that increased extracellular Pi activates NFκB signaling, increases PiT-1 expression and induces inflammatory cytokine and hepcidin production.

Next, we explored if NFκB is a necessary mediator for inflammatory cytokine and hepcidin regulation by high extracellular Pi in vitro. We treated mouse primary hepatocytes with either LPS or graded Pi concentrations in the presence or absence of a selective NFκB pharmacologic inhibitor, BAY 11-7082 (Koumakis et al., 2019; Pierce et al., 1997). Significant elevations in Il1b, Il6, and Hamp mRNA levels were detected following LPS or Pi treatments, and these were attenuated by BAY 11-7082 (Figure 7A, B). Similar effects of BAY 11-7082 on Saa1, haptoglobin (Hp) and Slc20a1 mRNA expression were also observed, corroborating reports that NFκB directly regulates PiT-1 abundance (Figure 7—figure supplement 1A–C).

As inflammation directly regulates hepcidin, we explored if Pi-induced hepcidin expression is a response resulting from direct or indirect actions of NFκB. Testing indirect effects, we cotreated primary hepatocytes with either LPS or graded Pi concentrations in the presence or absence of anti-IL1β antibody, anti-IL6 antibody, or both neutralizing antibodies in combination. Expression analysis detected significant elevations in Hamp mRNA levels following LPS or Pi treatments and was blunted by addition of either antibody alone or in combination (Figure 7C). To ensure treatments generated endogenous IL1β and IL6 protein, we analyzed Saa1 and Hp mRNA expression, as both genes are regulated by IL1β and IL6 (Zhou et al., 2016). Compared to vehicle-treated control cells, Saa1 and Hp transcript levels were reduced in the presence of either antibody alone or in combination, following LPS or Pi treatment (Figure 7—figure supplement 1D, E). Collectively, these results show Pi-mediated hepcidin production in cultured hepatocytes is a result of NFκB amplifying PiT-1 expression, which in turn, might intensify high extracellular Pi to augment NFκB regulated inflammatory gene programs, prompting the induction of required cytokines IL1β and IL6 to mediate hepcidin production.

Recombinant proteins used are mouse FGF23 (2629-FG, R&D Systems), mouse TNFα (410-MT, R&D Systems), and mouse IL6 (406 ML, R&D Systems). This FGF23 peptide contains an arginine to glutamine amino acid substitution at position 179 which yields it resistant to furin protease-mediated degradation, thus prolonging its half-life. Lipopolysaccharide (LPS) from E. coli serotype 0111:B4 (tlrl-3pelps, Invivogen) was used as endotoxin. Sodium phosphate dibasic anhydrous (Na₂HPO₄) (BP332-500, Fisher Scientific) and sodium phosphate monobasic anhydrous (NaH₂PO₄) (BP329-1, Fisher Scientific) were used to prepare a 1 M stock sodium phosphate buffer solution containing 500 mM Na₂HPO₄ and 500 mM NaH₂PO₄ at an adjusted pH of 7.4. Sodium sulfate (Na₂SO₄) (239313, Sigma-Aldrich) was used to prepare a 1 M stock sodium sulfate buffer solution at an adjusted pH of 7.4. PFA (P6801, Sigma-Aldrich) and BAY 11-7082 (S2913, Selleckchem) were used as agents to elucidate underlying signal transduction mechanisms. Anti-IL6 (MP5-20F3, R&D Systems) and anti-IL1β (AF-401-NA, R&D Systems) were used as antibodies in a cell-based assay to neutralize the biological activity of targeted cytokines.

Animal studies were performed in the conformity with applicable laws and guidelines and were approved by the Animal Research Committee at the University of Alabama Birmingham School of Medicine (UAB). Studies were performed using male mice and were maintained on a NIH 31 rodent diet (Harlan Teklad) and fed ad libitum, unless otherwise indicated. Constitutive FGF Receptor four null (Fgfr4^−/−) mice (Weinstein et al., 1998) were maintained on a C57BL/6 background. Constitutive Col4a3^−/− null (Alport) mice (Cosgrove et al., 1996) were maintained on a mixed Sv129/C57BL/6 background. Both mouse models were housed in our UAB rodent facility, in a heterozygous breeding state.

For experiments exploring the contribution of pathologic FGF23-FGFR4 signaling to CKD-associated pathologies, 10- to 14-week-old Fgfr4^−/− mice and corresponding wild-type littermates were placed on a customized diet containing 0.2% adenine (TD.140290, Envigo) for 6 weeks, switched to a customized diet containing 0.15% adenine (TD.170304, Envigo) for 2 weeks and transitioned back to the customized 0.2% adenine diet for an additional 6 weeks. Wild-type littermates placed on a customized control diet (TD.170303, Envigo) served as controls. All experimental groups were permitted a 1-week dietary acclimation period, using the customized control diet. After dietary acclimation, mice were unbiasedly assigned to either the customized control diet or customized adenine diet. After the 14-week duration, mice were euthanized under 2.5% isoflurane anesthesia and samples were prepared as described below. This experimental timeline is in accordance with previous studies (Noonan et al., 2020; Taylor et al., 2019). Adenine is known to induce kidney tubule-interstitial damage and is considered a dietary model of CKD.

For experiments testing the contribution of pathologic FGF23-FGFR4 signaling to systemic effects of a graded dietary phosphate load, 10- to 14-week-old Fgfr4^−/− mice and corresponding wild-type littermates were unbiasedly assigned and fed a customized 0.7% phosphate diet (TD.180287, Envigo), a customized 2.0% phosphate diet (TD.08020, Envigo) or a customized 3.0% phosphate diet (TD.180286, Envigo) for 12 weeks. Wild-type littermates placed on the customized 0.7% phosphate diet served as controls. At end of the experimental period, mice were euthanized under 2.5% isoflurane anesthesia and samples were prepared as described below.

For experiments investigating the contribution of hyperphosphatemia to CKD-associated pathologies, 4-week-old Alport mice and corresponding wild-type littermates were unbiasedly assigned and fed a customized 0.6% phosphate diet (TD.200407, Envigo) or a customized 0.2% phosphate diet (TD.200406, Envigo) as treatment for 6 weeks. Constitutive Col4a3^−/− null mice are considered a genetic model of Alport syndrome and progressive CKD. When maintained on a mixed Sv129/C57BL/6 background, Alport mice die at 10 weeks of age due to rapid renal injury. Wild-type littermates placed on the customized 0.6% phosphate diet served as controls. At 10 weeks of age, mice were euthanized under 2.5% isoflurane anesthesia and samples were prepared as described below. A detailed description of diet compositions is indicated in Table 4; Table 5; Table 6. All experimental group numbers were predetermined on the basis of experience from previous publications. Investigators were not blinded to mouse genotypes.

Mouse blood was collected by cardiac puncture and transferred into microvette serum gel tubes (20.1344, Sarstedt). Samples were then centrifuged at 10,000 × g for 5 min at room temperature. Serum supernatants were harvested and stored at −80°C. Clinical chemistry analyses were performed by the Animal Histopathology & Laboratory Medicine Core at the University of North Carolina, which is supported in part by an NCI Center Core Support Grant (5P30CA016086-41) to the UNC Lineberger Comprehensive Cancer Center. Serum intact FGF23 was assessed using ELISA (60-6800, Quidel).

Muscle strength was assessed using a Chatillon DFE series digital force gauge (E-DFE-200, Chatillon) with a metal grid adaptor, provided by the Behavioral Assessment Core at UAB. Mice were allowed to grip the metal grid with fore- and hindlimbs and then gently pulled backwards by their tail, until mice could not grip the metal grid. Each mouse was given 10 trials, excluding the highest and lowest values. These eight trials were then averaged. These averaged values are used to represent the muscle grip strength of each individual mouse. Investigators were blinded to each experimental group.

Unless otherwise indicated, tissues were excised, weighed, and either immediately flash frozen in liquid nitrogen or fixed for histologic examination. Organ and gastrocnemius weights were measured with an OHAUS scout portable balance (SJX323N/E).

Spleen, kidney and gastrocnemius tissues were fixed in 10% formalin solution for 24 hr, transferred into 70% ethanol and subjected to paraffin embedding (IDEXX). Spleen, kidney, and gastrocnemius sections were cut and either stained with Perls’ Prussian blue, H&E, or Masson’s trichrome (IDEXX) and used for representative images. Images were captured on a Keyence BZ-X800 fluorescent microscope with a ×20 and ×40 objective lens.

To quantify liver phosphate concentrations in mouse tissue, liver samples were weighed, homogenized in protein precipitation solution (0.53 N HCl, 5.3% trichloroacetic acid [TCA]), boiled for 30 min at 95°C and cooled in room temperature water for 2 min. Samples were then centrifuged at 13,300 × g for 30 min at 4°C. Supernatants were collected and subjected to colorimetric phosphate quantifications (ab65622, Abcam) according to the manufactures’ instructions.

To quantify nonheme iron concentrations in mouse tissues, spleen, and liver samples were weighed, homogenized in protein precipitation solution (0.53 N HCl, 5.3% TCA), boiled for 30 min at 95°C and cooled in room temperature water for 2 min. Samples were then centrifuged at 13,300 × g for 10 min at room temperature. Supernatants were harvested and subjected to colorimetric iron quantifications (157-30, Sekisui Diagnostics) according to the manufactures’ instructions.

Mouse blood was collected by cardiac puncture, transferred into microvette EDTA tubes (20.1341, Sarstedt), inverted to prevent clotting and stored at 4°C prior to shipment. Complete blood counts were measured by the Animal Histopathology & Laboratory Medicine Core at the University of North Carolina. In addition, serum supernatants were analyzed for iron- and total iron-binding capacity (TIBC) concentrations. Transferrin saturation percentage (TSAT%) = (serum iron/TIBC) × 100.

Hepatocytes were isolated from 10- to 14-week-old male wild-type C57BL/6J mice, which were anesthetized and placed on a 37°C heated surface to maintain adequate body temperature. Ventral laparotomy from the pubis to the cranial border of the liver was performed and the abdominal wall was incised to both sides, caudal of the diaphragm, exposing the inferior vena cava (IVC). Following suprahepatic diaphragm incision and surgical silk (5/0) ligation of the thoracic IVC, the infrarenal IVC was cannulated using a 24-Gauage shielded catheter (381412, BD) attached to a perfusion line. A peristaltic pump was utilized to perfuse the liver with 30 ml of liver perfusion medium (17701-038, Gibco) followed by 30 ml of liver digest medium (17703-034, Gibco), both prewarmed in a 37°C water bath. The portal vein was incised to route consecutive retrograde perfusion through the liver at a rate of 3 ml/min until each solution was empty. Next, the digested liver was excised, transferred into a 10-cm dish containing hepatocyte wash medium (17704-024, Gibco) and was minced within a cell culture hood. The mixture was then filtered through a 70-μm nylon cell strainer (352350, Falcon) using a 20-ml plastic serological pipette into a 50-ml polypropylene conical tube (352098, Falcon) to remove debris. Cells were washed twice with chilled hepatocyte wash medium with centrifugation at 60 × g for 3 min at 4°C to allow a soft separation of parenchymal cells from nonparenchymal cells. To enrich the hepatocyte cell population, the cell pellet was resuspended and inverted four times in 20 ml of chilled 36% iso-osmotic percoll gradient solution (P1644, Sigma-Aldrich) (percoll gradient solution:William’s E medium solution [four parts:six parts]) and centrifuged at 200 × g for 7 min at 4°C. The enriched hepatocyte population was resuspended in 10 ml of chilled hepatocyte wash medium and subjected to two washes with centrifugation at 60 × g for 2 min at 4°C. The washed pellet was resuspended in 12 ml of warm William’s E medium (12551-032, Gibco) supplemented with primary hepatocyte thawing and plating supplements (CM3000, Gibco), counted in a hemocytometer after staining with trypan blue (25,900 Cl, Corning), seeded at a density of 2.5 × 10⁵ cells/6-well or 1.0 × 10⁵ cells/12-well on plates coated with 100 μg/ml of collagen type 1 (354236, Corning) and allowed to adhere for 4 hr in a humidified 5% CO₂ incubator at 37°C. After this attachment period, medium was exchanged with fresh warm William’s E medium solution supplemented with primary hepatocyte maintenance supplements (CM4000, Gibco) and incubated overnight in a humidified 5% CO₂ incubator at 37°C. Next morning, media was exchanged with fresh warm Dulbecco's Modified Eagle's Medium (DMEM) (26140079, Gibco) supplemented with 1× penicillin/streptomycin (15140122, Gibco) and incubated for 6 hr in a humidified 5% CO₂ incubator at 37°C. This 6 hr serum-starvation period allows cells to synchronize to an identical cell cycle arrest phase, thus eliminating the potential impact between contrasting cell cycles and a cells overall response to exogenous treatment, as serum contains various growth factors and cytokines which promote the activation of signal transduction pathways related to cell proliferation and survival.

Hepatocytes were isolated, cultivated, and serum starved as described in supplemental methods. For experiments investigating the activation of signal transduction mediators, cells were seeded on 6-well collagen-coated plates and treated with either TNFα (100 ng/ml), IL6 (50 ng/ml), FGF23 (25 ng/ml), or appropriate amounts of sodium phosphate (1 M; pH 7.4) and sodium sulfate (1 M; pH 7.4) buffers to produce final desired concentrations and incubated for 30 min in a humidified 5% CO₂ incubator at 37°C. DMEM supplemented with 1× penicillin/streptomycin, which contains ~1 mM phosphate, served as a reference control (Ctrl). Sodium sulfate served as a negative control in response to increased anions.

For experiments analyzing expression levels of specific target genes, cells were seeded on 12-well collagen-coated plates and treated with either LPS (100 ng/ml), IL6 (50 ng/ml), or appropriate amounts of sodium phosphate (1 M; pH 7.4) and sodium sulfate (1 M; pH 7.4) buffers to produce final desired concentrations and incubated for 24 hr in a humidified 5% CO₂ incubator at 37°C. As described above, DMEM with 1× penicillin/streptomycin served as a reference control (Ctrl) and sodium sulfate served as a negative control.

For experiments investigating the role of high extracellular phosphate, cells were seeded on either 6- or 12-well collagen-coated plates and preincubated for 1 hr with or without the addition of PFA (1 mM) in a humidified 5% CO₂ incubator at 37°C. Cells were then either treated for 30 min to assess NFκB activation or treated for 24 hr to analyze expression levels of specific target genes, and incubated accordingly in a humidified 5% CO₂ incubator at 37°C. Specific treatments were conducted with factors described above. DMEM with 1× penicillin/streptomycin served as a reference control (Ctrl).

For experiments analyzing the participation of NFκB signaling, cells were seeded on 12-well collagen-coated plates and preincubated for 1 hr with or without the addition of BAY 11-7082 (20 μM) in a humidified 5% CO₂ incubator at 37°C. Cells were then treated and incubated for 24 hr in a humidified 5% CO₂ incubator at 37°C to analyze expression levels of specific target genes. Specific treatments were conducted with factors described above. DMEM with 1× penicillin/streptomycin served as a reference control (Ctrl). Total protein lysates were prepared from 30-min treatments as described below. Total RNA was prepared from 24-hr treatments as described below. All 24-hr treatments were supplemented with 0.70% fetal bovine serum (FBS) (CM3000, Gibco).

Hepatocytes were seeded on 12-well collagen-coated plates, cultivated and serum starved as described in supplemental methods. Primary mouse hepatocytes were treated for 24 hr to analyze expression levels of specific target genes with either LPS (100 ng/ml) or appropriate amounts of sodium phosphate buffer (1 M; pH 7.4) to produce a final desired phosphate concentration and incubated accordingly in a humidified 5% CO₂ incubator at 37°C. Treatments were performed with or without the addition of neutralizing antibodies against IL6 (6 μg/ml) and/or IL1β (6 μg/ml) as indicated. Total RNA was prepared from treatments as described below. All treatments were supplemented with 0.70% FBS (CM3000, Gibco).

Total RNA was extracted from liver and cultured hepatocytes using a RNeasy Plus Mini Kit (74136, Qiagen) and from gastrocnemius tissue using a RNeasy Plus Universal Mini Kit (73404, Qiagen) following the manufactures’ instructions. Employing a two-step reaction method, 1 μg of total RNA was reverse transcribed into cDNA using iScript Reverse Transcription Supermix (1708840, BioRad). Quantitative PCR was performed with 100 ng of cDNA, SsoAdvanced Universal SYBR Green Supermix (172-5272, BioRad) and sequence specific primers (as indicated in Table 7). Samples were run in duplicate on a CFX96 Touch Real-Time Detection Instrument (1855196, BioRad). Amplification was performed in forty cycles (95°C, 30 s; 98°C, 15 s; 60°C, 30 s; 65°C, 5 s). The generated amplicon was systematically double checked by its melting curve. Relative gene expression was normalized to expression levels of housekeeping genes 18S rRNA (for in vitro studies) or Gapdh (for in vivo studies). Results were evaluated using the 2^−∆∆Ct method and expressed as mean ± standard error of the mean (SEM).

Total protein was extracted from cells which were placed on ice and scraped from 6- or 12-well plates, using a 300 or 150 μl volume of RIPA lysis buffer (50 mM Tris–HCl, pH 7.5, 200 mM NaCl, 1% Triton X-100, 0.25% deoxycholic acid, 1 mM EDTA, 1 mM ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid tetrasodium (EGTA)), respectively, with addition of protease inhibitor (11836153001, Roche) and phosphatase inhibitors (P5726, P0044, Sigma-Aldrich). Cell lysates were then incubated on ice for 30 min and cleared by centrifugation at 13,000 × g for 30 min at 4°C. Supernatants were collected and protein was quantified using a Pierce BCA Protein Assay Kit (23225, Thermo Fisher Scientific).

Following protein quantification, supernatants were appropriately aliquoted and suspended in volumes of Laemmli sample buffer (1610747, BioRad) with β-mercaptoethanol (1610710, BioRad) as reducing agent, denatured at 100°C for 5 min and stored at −80°C. Protein samples [20 μg total protein] were loaded onto 8% or 10% SDS polyacrylamide gels and separated by SDS–PAGE. Polyacrylamide gels were run in 1× Tris/Glycine/SDS buffer (1610732, BioRad) at 20 mA per gel and stopped when sample dyes reached the end of the gels. Proteins were electroblotted onto PVDF membranes (IPVH00010, Merck Millipore) via a semi-dry cassette (1703940, BioRad) in 1× Tris/Glycine Buffer (1610734, BioRad) with 20% methanol at 20 V for 1 hr. Membranes were then blocked in 5% nonfat dry milk with 0.1% Tween-20 diluted in 1× Tris buffered saline (TBS) pH 7.5 for 1 hr and probed with primary antibodies at 1:1000 against specific antigens overnight at 4°C. ERK1/2 (4695, Cell Signaling), STAT3 (4904, Cell Signaling), NFκB (8242, Cell Signaling), and β-actin (4970, Cell Signaling) primary antibodies were used in 1× TBS with 5% nonfat dry milk and 0.1% Tween-20. Phospho-ERK1/2 (9101, Cell Signaling), phospho-STAT3 (9145, Cell Signaling), and phospho-NFκB (3033, Cell Signaling) primary antibodies were used in 1× TBS with 5% BSA and 0.1% Tween-20.

Next day, membranes were subjected to three wash periods for 5 min in 1× TBS with 0.1% Tween and then probed with horseradish peroxidase-conjugated goat anti-mouse or goat anti-rabbit secondary antibodies at 1:2,500 (W4021, W4011, Promega) in 1× TBS with 5% nonfat dry milk and 0.1% Tween at room temperature for 1 hr. Membranes were then subjected to three wash periods for 10 min in 1× TBS with 0.1% Tween at room temperature. Horseradish peroxidase activity was detected using enhanced chemiluminescence detection solution (RPN2106, GE Healthcare) and imaged on an SRX-101A X-ray film developer. All immunoblots were repeated with a minimum of three independent trials, with comparable results.

Data organization, scientific graphing, and statistical significance of differences between experimental groups were performed by using GraphPad Prism (version 9.0.0). All results are expressed as mean ± SEM. Depending on number of experimental groups and factors analyzed, we performed a two-way analysis of variance (ANOVA) followed by a post hoc Tukey test (for studies affected by two factors) or in the form of a one-way ANOVA (for studies measuring variance in three groups or more). Correlation and slope analyses were examined by simple linear regression. Statistical significance was set at a p value of less than or equal to 0.05. Sample size was determined on the basis of sample availability, prior experimental studies performed in our laboratory and from prior literature. No formal randomization was used in any experiment. For in vivo experiments, animals were unbiasedly assigned into different experimental groups, regardless of genotype. Group allocation was not performed in a blinded manner. Whenever possible, investigators were blinded to experimental groups (e.g., analysis of all grip-strength measurements).

All animal protocols and experimental procedures for adenine diet in FGFR4^+/+ and Fgfr4^−/− mice, graded phosphate diets in Fgfr4^+/+ and Fgfr4^−/− mice, low phosphate diets in Col4a3^+/+ and Col4a3^−/− mice and primary hepatocyte isolations from wild-type C57BL/6J mice, were approved by the Institutional Animal Care and Use Committees (IACUC) at the University of Alabama Birmingham School of Medicine. All animals were maintained in a ventilated rodent-housing system with temperature-controlled environments (22–23°C) with a 12-hr light/dark cycle and allowed ad libitum access to food and water. All protocols adhered to the Guide for Care and Use of Laboratory Animals to minimize pain and suffering. No animals were excluded from analysis.