Hyperphosphatemia increases inflammation to exacerbate anemia and skeletal muscle wasting independently of FGF23-FGFR4 signaling

Elevations in plasma phosphate concentrations (hyperphosphatemia) occur in chronic kidney disease (CKD), in certain genetic disorders, and following the intake of a phosphate-rich diet. Whether hyperphosphatemia and/or associated changes in metabolic regulators, including elevations of fibroblast growth factor 23 (FGF23) directly contribute to specific complications of CKD is uncertain. Here, we report that similar to patients with CKD, mice with adenine-induced CKD develop inflammation, anemia, and skeletal muscle wasting. These complications are also observed in mice fed high phosphate diet even without CKD. Ablation of pathologic FGF23-FGFR4 signaling did not protect mice on an increased phosphate diet or mice with adenine-induced CKD from these sequelae. However, low phosphate diet ameliorated anemia and skeletal muscle wasting in a genetic mouse model of CKD. Our mechanistic in vitro studies indicate that phosphate elevations induce inflammatory signaling and increase hepcidin expression in hepatocytes, a potential causative link between hyperphosphatemia, anemia, and skeletal muscle dysfunction. Our study suggests that high phosphate intake, as caused by the consumption of processed food, may have harmful effects irrespective of pre-existing kidney injury, supporting not only the clinical utility of treating hyperphosphatemia in CKD patients but also arguing for limiting phosphate intake in healthy individuals.


Introduction 57
Phosphate (Pi) is an essential mineral nutrient (Erem and Razzaque, 2018). Once absorbed and in 58 circulation, Pi is utilized by cells for various structures and functions. Pi metabolism is regulated 59 by a specific set of hormones to maintain physiological Pi concentrations. Fibroblast growth 60 factor 23 (FGF23) is the chief hormone maintaining body Pi balance by promoting renal Pi 61 excretion when Pi load is high (Fukumoto and Yamashita, 2007;Isakova et al., 2011). al., 2020) and induces its degradation, thereby restricting iron efflux into the circulation from 79 iron recycling macrophages, a process also known as reticuloendothelial system (RES) blockade, 80 and from duodenal enterocytes responsible for dietary iron absorption. Collectively, these events 81 reduce serum iron levels (hypoferremia), limiting the supply of iron for erythrocyte production 82 (Nemeth et al., 2004a). interventions to lower phosphate intake are challenging because they require long-term 99 behavioral changes made more difficult by the lack of disclosure of Pi content of foods and 100 beverages by the food industry (Gutiérrez and Wolf, 2010). 101 We assessed gene expression of inflammatory cytokines and acute phase proteins in both 146 genotypes in adenine-induced CKD. Unlike in healthy control mice, liver IL1 (Il1b), IL6 (Il6) 147 and serum amyloid A1 (Saa1) transcript levels were significantly and similarly elevated (Fig. 1c, 148 d) in both FGFR4 +/+ and FGFR4 -/mice. 149 To identify if FGF23-FGFR4 signaling contributes to functional iron deficiency, we both FGFR4 -/and FGFR4 +/+ mice following adenine (Fig. 1h) but less so in FGFR4 -/mice 168 when compared to FGFR4 +/+ mice. 169 To establish whether excess Pi and/or FGF23 contributes to hypoferremia in the absence 170 of CKD, we exposed FGFR4 +/+ and FGFR4 -/mice to a graded dietary Pi load for 12 weeks. 171 Serum FGF23 levels increased in both genotypes on 2% Pi and 3% Pi diet, in comparison to 172 mice on 0.7% Pi diet (Fig. 2a). Despite 2% Pi increasing serum FGF23, serum Pi levels 173 significantly increased only on 3% Pi, in comparison to mice fed 0.7% Pi (Fig. 2a). Notably, 174 these serum Pi levels are comparable to the elevated serum Pi levels observed in adenine-induced 175 CKD (Fig. 1b). No significant differences in serum calcium levels were observed between 176 genotypes (Supplemental Fig. 2a), despite elevated Pi levels (Fig. 2a) although not on 2% Pi diets (Fig. 2b, c). Liver injury was not detected, as no significant 188 elevations in hepatic alanine aminotransferase (Alt1) or aspartate aminotransferase (Ast1) mRNA 189 levels were found on 3% Pi (Supplemental Fig. 3a, b). These data support the notion that dietary 190 Pi overload induces inflammation, but not via FGF23-FGFR4 signaling. 191 To explain these effects of 3% Pi diet and determine if increased tissue Pi deposition is 192 associated with adverse outcomes, we analyzed the relationship between liver and serum Pi 193 levels in both FGFR4 +/+ and FGFR4 -/mice following a graded dietary Pi load. A positive 194 correlation was detected between hepatic and serum Pi levels in both genotypes, beginning with 195 2% Pi (Fig. 2d). These results show liver Pi deposits increase following elevations in dietary Pi 196

content. 197
Next, we tested if increased liver Pi accumulation affects correlations between liver Pi 198 and liver Hamp mRNA levels, as a high Pi diet induces Hamp expression (Nakao et al., 2015). A 199 positive correlation between liver Pi and liver Hamp mRNA levels were detected in both 200 FGFR4 +/+ and FGFR4 -/mice, again only with diets containing 2% or 3% Pi (Fig. 2e). As liver 201 injury was not detected (Supplemental Fig. 3a, b), these data indicate that elevations in liver 202 Hamp mRNA are independent of liver injury and may result from Pi-driven inflammation. 203 We next explored if increased dietary Pi loading led to changes in hematological 204 responses. Marked reductions in RBC, MCV, hemoglobin and serum iron levels were detected 205 on 3% Pi and were similar in both FGFR4 +/+ and FGFR4 -/mice (Fig. 2f). HCT% and MCH 206 were also significantly decreased (Supplemental Fig. 3c). Supporting these findings, spleen 207 tissue sections revealed both FGFR4 +/+ and FGFR4 -/mice show increased intracellular iron 208 deposits on 3% Pi (Fig. 2g). Thus, dietary Pi loading causes iron restriction and anemia even in 209 the absence of CKD. 210 211 Mouse models of hyperphosphatemia exhibit signs of skeletal muscle wasting which are 212 independent of FGF23-FGFR4 signaling. 213 As inflammation is a known contributor of muscle wasting (Raj et al., 2008;Schaap et al., 2006;214 Verzola et al., 2016), and since hyperphosphatemia and excess FGF23 are associated with 215 inflammation, we explored whether hyperphosphatemia contributes to skeletal muscle wasting, 216 and if pathologic FGF23-FGFR4 signaling is involved in these effects. We analyzed skeletal 217 muscle from FGFR4 +/+ and FGFR4 -/mice exposed to adenine-induced CKD (Fig. 1) or a 218 graded dietary Pi load (Fig. 2). Examination of skeletal muscle strength indicates that mice on 219 adenine or 3% Pi diet exhibit reduced grip strength in both FGFR4 +/+ and FGFR4 -/mice, in 220 comparison to respective control mice (Fig. 3a). A reduction in gastrocnemius mass was also 221 observed (Fig. 3b). Notably, gastrocnemius metallothionein-1 (Mt1) transcript levels were 222 significantly elevated in both genotypes following adenine and 3% Pi diets (Supplemental Fig.  223 4a, b), indicating that either condition fosters skeletal muscle abnormalities. 224 We next investigated if these muscle deficits resulted from inflammation inducing 225 myostatin and downstream atrophy-related gene programs, as both experimental models display 226 elevated levels of liver Il1b and Il6 (Fig. 1c, Fig. 2b). Compared to respective control mice, 227 gastrocnemius myostatin (Mstn) transcript levels were significantly elevated in both FGFR4 +/+ 228 and FGFR4 -/mice following adenine or 3% Pi diet (Fig. 3c). Additionally, both genotypes on 229 2% Pi showed an increased trend in Mstn mRNA levels (Fig. 3c). As these findings suggest 230 increased myofibrillar protein degradation, we further analyzed the expression of two specific 231 ubiquitin ligases of muscle-protein breakdown, muscle RING-finger protein 1 (Murf1) and 232 Atrogin-1. Compared with their respective control mice, gastrocnemius Murf1 and Atrogin1 233 transcript levels were significantly elevated in both FGFR4 +/+ and FGFR4 -/mice following 234 adenine and 3% Pi diets (Fig. 3d). 235 As elevated myostatin and increased myofibrillar protein degradation are features of 236 skeletal muscle wasting, we assessed if these results prompt a shift towards smaller myofibers. 237 Indeed, gastrocnemius tissue sections stained with H&E from FGFR4 +/+ and FGFR4 -/mice, on 238 either adenine or 3% Pi diet, showed smaller muscle fiber size compared with controls (Fig. 3e). 239 Taken together, these data suggest skeletal muscle wasting in adenine-induced CKD and 240 hyperphosphatemia does not require FGF23-FGFR4 signaling.  To test if hyperphosphatemia aggravates these pathologic complications, we exposed COL4A3 -/-246 mice to a low Pi diet treatment (0.2% Pi) for 6 weeks. In comparison to wild-type (COL4A3 +/+ ) 247 mice on normal diet (0.6% Pi), COL4A3 -/mice showed renal dysfunction by increased BUN and 248 serum creatinine levels (Fig. 4a). COL4A3 -/mice on low Pi diet displayed a reduction in both 249 parameters (Fig. 4a), along with reduced pathologic alterations in kidney morphology 250 (Supplemental Fig. 5a). As compared to wild-type mice, serum levels of FGF23 and Pi 251 significantly increased in COL4A3 -/mice on normal diet (Fig. 4b), but less so in COL4A3 -/mice 252 on low Pi diet (Fig. 4b). 253 To identify if a low Pi diet affects inflammation or the acute phase response in Alport 254 mice, we assessed gene expression of inflammatory cytokines and acute phase proteins. 255 Compared to wild-type mice on normal diet, liver Il1b, Il6 and Saa1 transcript levels were 256 significantly elevated in COL4A3 -/mice but much less elevated on 0.2% Pi (Fig. 4c, d). These 257 data support the notion that excess Pi in Alport mice aggravates inflammation. 258 As our results show that a low Pi diet decreases inflammation, we next explored its 259 impact on functional iron deficiency. Compared to wild-type mice on normal diet, liver Hamp 260 transcript levels were significantly elevated in COL4A3 -/mice with complete reversal by low Pi 261 diet (Fig. 4e). Assessing hematological responses, COL4A3 -/mice on normal diet were anemic, 262 with significant reductions in RBC count, MCV, hemoglobin and serum iron levels (Fig. 4f), as 263 well as in HCT%, MCH and TSAT% (Supplemental Fig. 5b). COL4A3 -/mice on normal diet 264 displayed profound intracellular iron sequestration in spleen (Fig. 4g), along with excessive 265 spleen and liver non-heme iron levels (Supplemental Fig. 5c, d). These effects were substantially 266 ameliorated by low Pi diet in COL4A3 -/mice, with improved hematologic parameters and 267 reduced iron deposits in spleen (Fig. 4f, g). Non-heme iron levels in spleen and liver were 268 reduced in COL4A3 -/mice by treatment, indicating increased iron mobilization and decreased 269 iron restriction (Supplemental Fig. 5c, d). These data indicate that dietary Pi restriction improves 270 hematological responses and alleviates hypoferremia. 271 As liver Pi accumulation is increased in adenine-induced CKD (Fig. 1h), we next 272 explored if low Pi diet treatment reduces pathologic liver Pi deposits in progressive CKD. 273 Compared to wild-type mice on normal diet, liver Pi levels were increased in COL4A3 -/mice 274 and were reduced on low Pi diet (Fig. 4h). Taken together, our data demonstrate that Pi (COL4A3 +/+ ) and Alport (COL4A3 -/-) mice subjected to either a normal diet (0.6% Pi) or a low Pi 284 diet treatment (0.2% Pi) (Fig. 3). Compared to wild-type mice on normal diet, COL4A3 -/mice 285 showed significant reduction in grip strength which was improved by low Pi diet (Fig. 5a). 286 Gastrocnemius mass was also reduced in COL4A3 -/mice, and treatment tended to improve 287 muscle weight (Fig. 5b). In particular, gastrocnemius Mt1 transcript levels were significantly 288 elevated in COL4A3 -/mice compared to wild-type mice on normal diet and were reduced by low 289 Pi diet (Supplemental Fig. 4c). These results suggest that Pi restriction as a dietary intervention 290 may improve skeletal muscle abnormalities in CKD. Furthermore, compared to wild-type mice 291 on normal chow, COL4A3 -/mice displayed significant elevations in gastrocnemius Mstn 292 transcript levels which was reduced by treatment (Fig. 5c). Additionally, COL4A3 -/mice on 293 normal diet showed increased gastrocnemius Murf1 and Atrogin1 transcript levels, which were 294 also reduced by low Pi diet (Fig. 5d). These data support the notion that dietary Pi restriction, as 295 a treatment in COL4A3 -/mice, reduces myostatin synthesis and subsequent atrophy-related gene 296

programs. 297
As with adenine-induced CKD mice, or mice fed high phosphate diet, gastrocnemius 298 tissue sections from COL4A3 -/mice showed smaller muscle fiber size compared to wild-type 299 controls (Fig. 5e). However, COL4A3 -/mice fed low Pi diet showed improved muscle fiber size 300 (Fig. 5e). Taken together, these data suggest that hyperphosphatemia affects skeletal muscle 301 wasting in Alport mice, possibly by exacerbating systemic inflammatory cytokine concentrations 302 and their catabolic effects on muscle. 303 304

Pi targets hepatocytes and increases expression of inflammatory cytokines and hepcidin 305
Having shown that inflammation, hypoferremia, and muscle wasting induced by high Pi are 306 independent of FGF23-FGFR4 signaling, we tested if Pi directly affects inflammatory cytokine 307 and hepcidin expression in mouse primary hepatocytes. 308 We first analyzed the expression profile of the three families of Na/Pi cotransporters 309 (type I -III). Quantitative polymerase chain reaction (qPCR) analysis detected high levels of Pit1 310 and Pit2, but not Npt1 and 4, or NaPi2a, 2b and 2c (Supplemental Fig. 6a). This analysis 311 indicates that type III Na/Pi cotransporters are the predominant Na/Pi family in primary 312 Immunoblot analysis of ERK1/2, STAT3 and NFB showed that Pi treatments increased 319 phosphorylated NFB levels without changing total NFB expression (Fig. 6a). Increased 320 concentrations of Na 2 SO 4 , a salt generating another anionic species, had no effect on phospho-321 NFB levels, indicating this response was specific to elevated Pi and not an unspecific response 322 to increased anions. Pi treatment did not affect pERK1/2 or pSTAT3, and FGF23 had no effect 323 on any of the pathways examined. 324 We next analyzed gene expression by qPCR of inflammatory cytokines and acute phase 325 proteins in isolated hepatocytes treated with increasing concentrations of Pi or Na 2 SO 4 with LPS 326 and IL6 treatments used as positive control. Elevations in Il1b, Il6 and Saa1 transcript levels 327 were noted not only following LPS and IL6 but also Pi treatments (Fig. 6b, c). Treatments with 328 Na 2 SO 4 had no effect on gene expression. As inflammation is a known mediator of hepcidin We report that hyperphosphatemia, either as a result of adenine-induced CKD or dietary 382 Pi excess, increases inflammation to exacerbate anemia and skeletal muscle wasting. These 383 complications are associated with increased liver Pi levels, which correlated with serum Pi 384 concentrations. Supplying a low Pi diet treatment to Alport mice, a genetic model of CKD, 385 results in beneficial outcomes that reduce functional iron deficiency and skeletal muscle wasting. 386 Furthermore, our mechanistic in vitro studies indicate that Pi elevations induce hepatic 387 production of IL6 and IL1 to increase hepcidin expression in hepatocytes, a potential causative 388 link between hyperphosphatemia, anemia, and skeletal muscle dysfunction. 389 Previously, we reported pathologic FGF23-FGFR4 signaling might contribute to excess 390 inflammatory mediators (Singh et al., 2016), and we now followed up on the FGF23 391 inflammatory role in clinically-relevant CKD models in vivo. Here, we examined wild-type 392 (FGFR4 +/+ ) and constitutive FGFR4 knockout (FGFR4 -/-) mice subjected to adenine diet or a 393 graded dietary Pi load. We found that on adenine, both FGFR4 +/+ and FGFR4 -/mice show 394 comparable macroscopic parameters (Table 1) and degrees of functional iron deficiency (Fig. 1). 395 These findings coincide with greater levels of liver Pi, which raises the possibility that 396 pathologic Pi deposits, in tissues apart from the vasculature, may contribute to additional We employed a graded Pi diet to study the effects of excess Pi on the liver (Fig. 2).  (Table 2). Also, no pathologic changes in kidney (Supplemental Fig. 2) or liver were detected 414 (Supplemental Fig. 3). However, mice on a 3% Pi diet exhibit increased liver inflammation and 415 Hamp expression, which corroborates previous observation that high dietary Pi influences 416 hepcidin production (Nakao et al., 2015). These results coincide with positive correlations 417 between liver Pi and liver Hamp mRNA expression, with onset of this correlation preceding 418 significant elevations in serum Pi. Despite these data suggesting that liver Pi influences liver 419 hepcidin production, our finding might indicate that increased extra-renal Pi accumulation 420 provides a reservoir for storing excess Pi until tissue accumulation achieves saturation, in which 421 case the serum Pi compartment then gradually rises, resulting in hyperphosphatemia. 422 Furthermore, our data suggest that prolonged exposure to Pi, if not maintained in adequate 423 quantities, might trigger pathologic outcomes, as mice on 3% Pi show a noticeable degree of 424 hypoferremia. None of the observed effects of the high Pi diet were mediated by FGFR4, as 425 FGFR4 -/mice were comparable to wild-type mice in all the parameters measured. However, this 426 work cannot exclude the potential of alternative FGFRs which might mediate the effects of 427 excess FGF23 towards functional iron deficiency following either adenine or 3% Pi diet, as a 428 recent report exhibits the utilization of a single intraperitoneal injection of FGF23 blocking 429 peptide was sufficient to rescue anemia (Agoro et al., 2018).  To assess whether reducing hyperphosphatemia can improve inflammation, anemia, and 449 skeletal muscle wasting, we exposed Alport mice, a genetic model of progressive CKD, to a low 450 Pi diet treatment. Indeed, despite severe elevations in serum FGF23, dietary Pi restriction limited 451 functional iron deficiency (Fig. 4). Our data also shows liver Pi levels were reduced in Alport 452 mice following low Pi diet treatment, in comparison to Alport mice on normal diet. These 453 findings provide strong evidence that hyperphosphatemia, specifically pathologic liver Pi 454 accumulation, rather than pathologic FGF23-FGFR4 signaling, might exacerbate inflammation 455 and hypoferremia. Skeletal muscle function and mass were also improved by a low Pi diet in 456 Alport mice, along with decreased expression of muscle myostatin and atrophy-related gene 457 programs, culminating in larger myofiber size. These findings suggest the contribution of 458 hyperphosphatemia to skeletal muscle wasting may result from an indirect mechanism that 459 regulates inflammatory cytokines and their pleotropic activities, such as increased liver-derived 460 IL1 and IL6, which might increase overall systemic levels that effectively target skeletal 461 muscle. Despite this postulate, further work will be needed to determine if high extracellular Pi 462 directly targets skeletal muscle cells to affect muscle function. Nonetheless, these data add to the 463 growing list of adverse outcomes of Pi toxicity such as gingivitis, accelerated aging, vascular 464 calcification and tumorigenesis (Erem and Razzaque, 2018). Furthermore, Alport mice on low Pi 465 diet treatment displayed a reduced degree of pathologic kidney function, alterations in kidney 466 morphology, and macroscopic parameters (Fig. 4, Supplemental Fig. 5, Table 3). Thus, we 467 cannot exclude that these beneficial outcomes in Alport mice observed on treatment may be a 468 repercussion of slightly improved kidney function, as a recent report demonstrates elevated Pi 469 concentrations directly affect proximal tubular function (Shiizaki et al., 2021). 470 Importantly, we identify a molecular mechanism that potentially links 471 hyperphosphatemia to anemia and skeletal muscle dysfunction. Utilizing mouse primary 472 hepatocytes, we demonstrate high extracellular Pi activates NFB signaling and leads to 473 subsequent inflammatory cytokine and hepcidin production (Fig. 6) neutralization assay of these targeted cytokines (Fig. 7). Based on these findings, we speculate Our study has some limitations. Although we confirm hyperphosphatemia affects specific 497 complications, our study does not specifically address the actions of certain aggregate 498 byproducts formed by increased extracellular Pi. Nonetheless, this principal emphasis on 499 elevations in plasma Pi concentrations will ultimately impact the formation of byproducts. 500 Notably, the identification of the specific Pi sensor which mediates our observed hepatic Pi 501 actions are not definite and remains to be defined with our ongoing studies. Moreover, our study 502 does not address the specific actions of hyperphosphatemia on bone metabolism. As bone is a 503 reservoir of extracellular Pi, potential alterations in bone health could relay a crosstalk between 504 bone, liver and/or skeletal muscle, which might contribute to our reported observations. 505 In summary, we investigated whether hyperphosphatemia and/or pathologic FGF23-506 FGFR4 signaling aggravates inflammation, anemia, and skeletal muscle wasting. We establish 507 hyperphosphatemia, as found in dietary Pi overload or in CKD, is a detrimental trigger which 508 activates hepatic NFB signaling to stimulate an inflammatory response, which in turn, 509 exacerbates hypoferremia and widespread complications such as skeletal muscle wasting.  are considered a genetic model of Alport syndrome and progressive CKD. When maintained on a 581 mixed Sv129/C57BL/6 background, Alport mice die at 10 weeks of age due to rapid renal injury. 582 Wild-type littermates placed on the customized 0.6% phosphate diet served as controls. At 10 583 weeks of age, mice were euthanized under 2.5% isoflurane anesthesia and samples were 584 hemocytometer after staining with trypan blue (25900Cl, Corning), seeded at a density of 653 2.5x10 5 cells/6-well or 1.0x10 5 cells/12-well on plates coated with 100 g/mL of collagen type 1 654 (354236, Corning) and allowed to adhere for 4 hours in a humidified 5% CO 2 incubator at 37°C. 655 After this attachment period, medium was exchanged with fresh warm William's E medium 656 solution supplemented with primary hepatocyte maintenance supplements (CM4000, Gibco) and 657 incubated overnight in a humidified 5% CO 2 incubator at 37°C. Next morning, media was 658 buffers to produce final desired concentrations and incubated for 24 hours in a humidified 5% 677 CO 2 incubator at 37°C. As described above, DMEM with 1x Penicillin/Streptomycin served as a 678 reference control (Ctrl) and sodium sulfate served as a negative control. 679 For experiments investigating the role of high extracellular phosphate, cells were seeded 680 on either 6-well or 12-well collagen-coated plates and pre-incubated for 1 hour with or without 681 the addition of PFA [1 mM] in a humidified 5% CO 2 incubator at 37°C. Cells were then either 682 treated for 30 minutes to assess NFB activation or treated for 24 hours to analyze expression 683 levels of specific target genes, and incubated accordingly in a humidified 5% CO 2 incubator at 684 37°C. Specific treatments were conducted with factors described above. DMEM with 1x 685

Penicillin/Streptomycin served as a reference control (Ctrl). 686
For experiments analyzing the participation of NFB signaling, cells were seeded on 12-687 well collagen-coated plates and pre-incubated for 1 hour with or without the addition of BAY 688 11-7082 [20 M] in a humidified 5% CO 2 incubator at 37°C. Cells were then treated and 689 incubated for 24 hours in a humidified 5% CO 2 incubator at 37°C to analyze expression levels of 690 specific target genes. Specific treatments were conducted with factors described above. DMEM 691 with 1x Penicillin/Streptomycin served as a reference control (Ctrl). Total protein lysates were 692 prepared from 30-minute treatments as described below. Total RNA was prepared from 24-hour 693 treatments as described below. All 24-hour treatments were supplemented with 0.70% FBS 694 (CM3000, Gibco). 695 Cytokine neutralization. Hepatocytes were seeded on 12-well collagen-coated plates, cultivated 696 and serum starved as described in supplemental methods. Primary mouse hepatocytes were 697 treated for 24 hours to analyze expression levels of specific target genes with either LPS [100 698 ng/mL] or appropriate amounts of sodium phosphate buffer [1 M; pH 7.4] to produce a final 699 desired phosphate concentration and incubated accordingly in a humidified 5% CO 2 incubator at 700 37°C. Treatments were performed with or without the addition of neutralizing antibodies against 701 IL6 [6 g/mL] and/or IL1 [6 g/mL] as indicated. Total RNA was prepared from treatments as 702 described below. All treatments were supplemented with 0.70% FBS (CM3000, Gibco). 703 RNA isolation and quantification. Total RNA was extracted from liver and cultured 704 hepatocytes using a RNeasy Plus Mini Kit (74136, Qiagen) and from gastrocnemius tissue using 705 a RNeasy Plus Universal Mini Kit (73404, Qiagen) following the manufactures' instructions. 706 Employing a two-step reaction method, 1 g of total RNA was reverse transcribed into cDNA 707 using iScript Reverse Transcription Supermix (1708840, Bio-Rad). Quantitative PCR was 708 performed with 100 ng of cDNA, SsoAdvanced Universal SYBR Green Supermix (172-5272, 709 Bio-Rad) and sequence specific primers (as indicated in Table 7). Samples were run in duplicate 710 on a CFX96 Touch Real-Time Detection Instrument (1855196, Bio-Rad). Amplification was 711 performed in forty cycles (95°C, 30 seconds; 98°C, 15 seconds; 60°C, 30 seconds; 65°C, 5 712 seconds). The generated amplicon was systematically double-checked by its melting curve. 713 Relative gene expression was normalized to expression levels of housekeeping genes 18S rRNA 714 (for in vitro studies) or Gapdh (for in vivo studies). Results were evaluated using the 2 -∆∆Ct 715 method and expressed as mean ± SEM. 716 Protein isolation and Immunoblotting. Total protein was extracted from cells which were 717 placed on ice and scraped from 6-well or 12-well plates, using a 300 L or 150 L volume of 718 RIPA lysis buffer (50 mM Tris-HCl pH 7.5, 200 mM NaCl, 1% Triton X-100, 0.25% 719 deoxycholic acid, 1 mM EDTA, 1 mM EGTA) respectively, with addition of protease inhibitor 720 (11836153001, Roche) and phosphatase inhibitors (P5726, P0044, Sigma-Aldrich). Cell lysates 721 were then incubated on ice for 30 minutes and cleared by centrifugation at 13,000 g for 30 722 minutes at 4°C. Supernatants were collected and protein was quantified using a Pierce BCA 723 Protein Assay Kit (23225, Thermo Fisher Scientific). 724 Following protein quantification, supernatants were appropriately aliquoted and 725

740
Next day, membranes were subjected to three wash periods for 5 minutes in 1x TBS with 741 0.5% Tween and then probed with horseradish peroxidase-conjugated goat anti-mouse or goat 742 anti-rabbit secondary antibodies at 1:2,500 (W4021, W4011, Promega) in 1x TBS with 5% 743 nonfat dry milk and 0.5% Tween at room temperature for 1 hour. Membranes were then 744 subjected to three wash periods for 10 minutes in 1x TBS with 0.5% Tween at room temperature. 745 Horseradish peroxidase activity was detected using enhanced chemiluminescence detection 746 solution (RPN2106, GE Healthcare) and imaged on an SRX-101A X-ray film developer. All 747 immunoblots were repeated with a minimum of three independent trials, with comparable results.

Competing Interests 799
The authors declare competing financial interests: CF has served as a consultant for Bayer and 800 Calico Labs, and he is the founder and currently the CSO of a startup biotech company (Alpha 801 Young LLC); OG has received honoraria and grant support from Akebia and Amgen, grant 802 support from GSK, honoraria from Ardelyx, Reata, and AstraZeneca, and serves on the Data 803    149.0 ± 3.8 151.5 ± 3.7 152.9 ± 3.2 145.6 ± 3.5 152.5 ± 2.5 149.6 ± 3.9