Epigenetic Regulation of Micrornas Controlling CLDN14 Expression As a Mechanism for Renal Calcium Handling

BASIC RESEARCH www.jasn.org

Epigenetic Regulation of MicroRNAs Controlling CLDN14 Expression as a Mechanism for Renal Calcium Handling

† † † Yongfeng Gong,* Nina Himmerkus, Allein Plain, Markus Bleich, and Jianghui Hou*

*Department of Internal Medicine, Renal Division, Washington University Medical School, St. Louis, Missouri; and †Section of Membrane Transport Physiology, Physiologisches Institut der Christian-Albrechts-Universität, Kiel, Germany

ABSTRACT The kidney has a major role in extracellular calcium homeostasis. Multiple genetic linkage and association studies identiﬁed three tight junction genes from the kidney—claudin-14, -16, and -19—as critical for calcium imbalance diseases. Despite the compelling biologic evidence that the claudin-14/16/19 proteins form a regulated paracellular pathway for calcium reabsorption, approaches to regulate this transport pathway are largely unavailable, hindering the development of therapies to correct calcium transport abnormalities. Here, we report that treatment with histone deacetylase (HDAC) inhibitors downregulates renal CLDN14 mRNA and dramatically reduces urinary calcium excretion in mice. Furthermore, treatment of mice with HDAC inhibitors stimulated the transcription of renal microRNA-9 (miR-9) and miR-374 genes, which have been shown to repress the expression of claudin-14, the negative regulator of the paracellular pathway. With renal clearance and tubule perfusion techniques, we showed that HDAC inhibitors transiently increase the paracellular cation conductance in the thick ascending limb. Genetic ablation of claudin-14 or the use of a loop diuretic in mice abrogated HDAC inhibitor-induced hypocalciuria. The genetic mutations in the calcium-sensing receptor from patients with autosomal dominant hypocalcemia (ADH) repressed the transcription of miR-9 and miR-374 genes, and treatment with an HDAC inhibitor rescued the phenotypes of cell and animal models of ADH. Furthermore, systemic treatment of mice with antagomiRs against these miRs relieved claudin-14 gene silencing and caused an ADH-like phenotype. Together, our ﬁndings provide proof of concept for a novel therapeutic principle on the basis of epigenetic regulation of renal miRs to treat hypercalciuric diseases.

J Am Soc Nephrol 26: 663–676, 2015. doi: 10.1681/ASN.2014020129

Regulation of extracellular calcium homeostasis re- Mg++ reabsorption and (2) generates a lumen- sides principally within the kidney. A monogenic positive diffusion potential that supplies the driving renal disorder, autosomal recessive familial hypo- force for Ca++ and Mg++ reabsorption. A genome- magnesemia with hypercalciuria and nephrocalcino- wide association study has identiﬁed a novel claudin— sis (FHHNC; Online Mendelian Inheritance in Man claudin-14—as a major risk gene of hypercalciuric [OMIM] #248250), is caused by mutations in the nephrolithiasis.8 Our recent publications have claudin genes: claudin-161 and claudin-19.2 The claudin-16 and claudin-19 genes are exclusively expressed in the thick ascending limb (TALH) of the Received February 3, 2014. Accepted June 13, 2014. nephron, where a major percentage of ﬁltered diva- Published online ahead of print. Publication date available at lent cations is reabsorbed paracellularly (30%–35% www.jasn.org. ++ – ++ 3 in vitro4,5 Ca and 50% 60% Mg ). A run of and Correspondence: Dr. Jianghui Hou, Washington University Re- in vivo6,7 studieshasshownthatclaudin-16and nal Division, 660 South Euclid Avenue, St. Louis, MO 63110. claudin-19 confer cation permeability to the tight Email: [email protected] ++ junction (TJ), which (1) allows paracellular Ca and Copyright © 2015 by the American Society of Nephrology

J Am Soc Nephrol 26: 663–676, 2015 ISSN : 1046-6673/2603-663 663 BASIC RESEARCH www.jasn.org revealed the pathogenic basis for claudin-14 in the kidney: (1) model. HDAC inhibitors accordingly were able to correct the claudin-14 interacts with claudin-16 and inhibits its perme- CaSR mutational effect on miRs. In vivo antagomir treatments ability9;(2) transgenic overexpression of claudin-14 in the ablated endogenous miR-9 and miR-374 and induced ADH-like mouse kidney phenocopies FHHNC10;(3) claudin-14 is natu- phenotypes in laboratory mice. In animals receiving the calci- rally silenced by two microRNA (miR) molecules—miR-9 and mimetic compound cinacalcet to mimic ADH conditions, miR-3749; and (4) the expression levels of both miRs are phys- HDAC inhibitors effectively ameliorated cinacalcet-induced hy- iologically regulated by the Ca++-sensing receptor (CaSR) by percalciuria and hypermagnesiuria. nuclear factor of activated T cells (NFAT) binding and nearby histone deacetylation.10 Such data have raised the tantalizing possibility that epigenetic regulation may represent a novel RESULTS mechanism to correct physiologic abnormalities, such as hypercalciuria, and prevent kidney stone formation. Here, with a An Epigenetic Program Regulates the miR–CLDN14 range of in vitro and in vivo approaches, we have shown that Axis in the Kidney histone deacetylase (HDAC) inhibitors reduce renal Ca++ ex- The epigenetic drugs acting to regulate chromatin remodeling cretion through repression of the claudin-14 gene expression. and gene transcription have gained considerable recognition in The promoter of the claudin-14 gene itself was not regulated the treatments of various types of cancers,19,20 neurologic dis- through histone deacetylation; instead, it was the two claudin- orders,21 HIV infections,22 cardiovascular and inflammatory 14–targeting miRs—miR-9 and miR-374—that were tran- diseases,23 and others. Tosystematically screen for the epigenetic scriptionally regulated by HDAC. effects on CLDN14 gene expression, we treated the primary Therenaltubulesare composedof heterogeneousepithelia:the cultures of mouse TALH cells (freshly isolated from the proximal tubule (PT), the loop of Henle, the distal convoluted C57BL/6 mice described before)9,10 in vitro for 16 hours with tubule (DCT), and the collecting duct. The predominant Ca++ the DNA methyltransferase inhibitor 5-Aza-29-deoxycytidine, reabsorption sites are in the PT, the TALH, and the DCT.11,12 the histone-lysine methyltransferase inhibitor BIX 01294,24 the Although the PTand TALH use the paracellular pathway to han- histone demethylase inhibitor tranylcypromine,25 and the dle Ca++,11 the DCT relies on a coordinated transcellular Ca++ HDAC inhibitors trichostatin A (TsA) and suberanilohydroxa- transport pathway based on the apically localized transient re- mic acid26 (SAHA; approved by the Food and Drug Adminis- ceptor potential channel V member 5 (TRPV5) and the basolat- tration as Vorinostat). 5-Aza-29-deoxycytidine, BIX 01294, and erally localized Na+/Ca++ exchanger 1 (NCX1).12 Because HDAC tranylcypromine were without significant effects on claudin-14 is a global regulator of gene expression, its renal effects could stem mRNA levels (Supplemental Figure 1). By contrast, SAHA sig- from different tubular segments. Here, with ex vivo tubule per- nificantly downregulated CLDN14 mRNA levels by 38% at 0.1 mM fusion approaches and using animal models defective for TALH (P,0.01, n=3 versus vehicle) (Figure 1A); the downregula- Ca++ transport (CLDN14 knockout [KO] or furosemide-treated tion persisted through 0.33–3.3 mM and reached the maximal mice), we have conclusively shown that the TALH is the primary level of 54% at 0.33 mM (Figure 1A). The CLDN14 expression target segment of HDAC inhibition. We show that the GFR re- was more sensitive to TsA treatments. At concentrations as low mains unaltered, whereas the DCT exhibits secondary adaptive as 0.01 mM, TsA significantly reduced CLDN14 mRNA levels attenuation of cellular Ca++ buffering and basolateral NCX1- by 64% (P,0.001, n=3 versus vehicle) (Supplemental Figure 2A); driven Ca++ extrusion on the basis of an unaltered luminal Ca++ the reduction was significant through 0.033–0.33 mM(Supple- influx capability mediated by TRPV5. mental Figure 2A). To examine the in vivo effects of HDAC in- The CaSR, a member of the G protein-coupled receptors, hibitors on CLDN14, wild-type C57BL/6 mice were treated with plays a critical role in extracellular Ca++ homeostasis.13 Its loss- SAHA or TsA over a range of doses and durations. The in vivo of-function mutations cause two recessive genetic disorders— effects of SAHA and TsA were particularly fast. A single dose of 2 FHH (OMIM #145980) and neonatal severe hyperparathyroidism SAHA at 25 mg/kg body wt 1 significantly reduced CLDN14 (OMIM #239200).14 Its gain-of-function mutations cause a mRNA levels in the kidney by 56% (P,0.01, n=5 versus vehicle) dominant genetic disorder—autosomal dominant hypocalce- (Figure 1B) at 2 hours; by 4 hours and through to 8 hours, the mia (ADH; OMIM #146200).15,16 A severe form of ADH has reduction of CLDN14 expression persisted in the kidney (Figure been characterized in patients with hypocalcemia, hypomagne- 1B). By 12 hours, the effects of SAHA completely dissipated semia, and urinary loss of Ca++ and Mg++, and it is accompanied (Figure 1B). We then varied the SAHA treatment doses from 5 by additional Bartter-like phenotypes.17,18 Despite the rapid ex- to 25 mg/kg and found significant downregulation of CLDN14 panding repertoire of known CaSR genetic mutations, it remains at a dosage as low as 5 mg/kg (P,0.05, n=5 versus vehicle) and elusive how these mutations affect CaSR downstream signaling through 10–25 mg/kg at 4 hours (Figure 1C). The CLDN14 targets. Here, we studied three common CaSR mutations found protein levels were measured in freshly isolated TALH cells in patients with ADH—E127A, F128L, and T151M—with pooled from SAHA or vehicle-treated mouse kidneys (n=5). 2 quantitative RNA analyses. We found that the transcription of SAHA treatment of 25 mg/kg body wt 1 profoundly reduced both miR-9 and miR-374 genes was consistently repressed by the CLDN14 protein levels by 60% when assayed at 4 hours CaSR gain-of-function mutations in an overexpressing cell (Figure 1D). TsA was more effective in suppressing CLDN14

664 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 663–676, 2015 www.jasn.org BASIC RESEARCH

Figure 1. HDAC inhibitor regulation of CLDN14 and miR gene expression in the kidney. (A) The dose response of HDAC inhibitor SAHA effects on CLDN14 mRNA levels in primary cultures of mouse TALH. (B) The time response of SAHA effects on CLDN14 mRNA levels in mouse kidneys. (C) The dose response of SAHA effects on CLDN14 mRNA levels in mouse kidneys. (D) CLDN14 proteins levels in SAHA- or vehicle-treated freshly isolated mouse TALH tubules. (E and F) The effects of SAHA on miR transcription measured for (E) pri-miR-9–1 and (F) pri-miR-374 levels. (G and H) The effects of SAHA on histone acetylation levels over the miR gene promoters measured for (G) miR-9–1 and (H) miR-374 genes with ChIP. Veh, vehicle. *P,0.05; **P,0.01; ***P,0.001.

2 2 expression. At 1 mg/kg body wt 1 and the 4-hour time point, 25 mg/kg body 1—by 2.40-fold for miR-9–1(P,0.05, n=5) TsA decreased CLDN14 mRNA levels in the kidney significantly (Figure 1G) and 2.45-fold for miR-374 (P,0.01, n=5) (Figure by 71% (P,0.001, n=5 versus vehicle) (Supplemental Figure 1H). The NFAT binding to the miR-9–1 or miR-374 promoter 2B). There are two mechanisms to explain the CLDN14 gene was not affected by SAHA (Supplemental Figure 4). Together, regulation by HDAC inhibitor—cis regulation on the level of these data have established a molecular basis for SAHA regula- CLDN14 promoter and trans regulation through miRs-based tion of CLDN14 in the kidney. silencing.9,10 Because the claudin-14 KO/Reporter mouse line was generated with an lacZ reporter gene replacing the last The Short-Term Effects of HDAC Inhibitors on Urinary exon of the claudin-14 gene (including the entire coding region Calcium and Magnesium Excretion and the 39 untranslated region),27 the expression levels of the Because CLDN14 is a negative regulator of CLDN169 and Ca++ lacZ gene serve as a faithful measurement for endogenous reabsorption10 in the kidney, curbing CLDN14 expression by CLDN14 promoter activity, regardless of any miR-based regu- SAHA may provide a novel route to lower urinary Ca++ ex- lation. The lacZ mRNA levels in the KO mouse kidneys cretion. To test this hypothesis, we treated wild-type C57BL/6 2 showed no significant change 4 hours after receiving 25 mg/kg mice with SAHA at 5 mg/kg body wt 1 and traced urinary Ca++ 2 body wt 1 SAHA treatment (Supplemental Figure 3). By con- and Mg++ levels (as ratios to creatinine) from 4 to 12 hours in trast, the CLDN14 targeting miRs miR-9 and miR-374 were both spot urine collections. Consistent with the downregulation upregulated by SAHA in the kidney. At the same dosage and time of CLDN14 gene expression at 4 hours, the urinary excretion point, SAHA significantly increased the transcriptional levels of of Ca++ and Mg++ was significantly reduced by 47% and 40%, the miR-9–1 gene by 1.49-fold (P,0.05, n=5 versus vehicle) respectively (P,0.01, n=5 versus vehicle) (Supplemental Figure 5). (Figure 1E) and the miR-374 gene by 1.85-fold (P,0.01, n=5 By 8 hours, hypocalciuria was still pronounced in SAHA-treated versus vehicle) (Figure 1F). To determine if SAHA caused direct animals (Supplemental Figure 5A), but urinary magnesium excre- histone acetylation on the miR promoters, we analyzed the pre- tion recovered to the vehicle level (Supplemental Figure 5B). The viously discovered promoter regions in miR-9–1 and miR-374 SAHA effects on urinary Ca++ and Mg++ levels were completely genes containing the NFAT binding site (AGGAAAAT) located lost after 12 hours (Supplemental Figure 5). To capture time- 1–2 kb upstream of the respective miR hairpin sequence,10 sensitive changes in renal tubular transport function, we where the histone acetylation was vividly regulated through performed 1-hour renal clearance measurements on age- and CaSR signaling. Chromatin immunoprecipitation (ChIP) assays sex-matched wild-type mice infused with FITC-inulin (Concise revealed that the same promoter region experienced significant Methods) starting at 4 hours after a single dose of 5 mg/kg body 21 ++ increases in histone H3 Lysine-9 (H3K9) and Lysine-14 wt SAHA. The plasma Ca level (PCa) (Figure 2A) was not (H3K14) acetylation levels 4 hours after receiving SAHA at significantly altered by SAHA at 4 hours, despite pronounced

J Am Soc Nephrol 26: 663–676, 2015 Epigenetic Regulation of Calcium Metabolism 665 BASIC RESEARCH www.jasn.org

++ reduction of both fractional Ca excretion level (FECa)(Figure Table 1. Plasma and urine electrolyte levels in wild-type ++ 2B, Table 1) and absolute Ca excretion level (ECa)(Figure2C, animals receiving SAHA treatments measured by renal Table 1) in SAHA-treated animals compared with vehicle-treated clearance ++ animals (P,0.01, n=6). The plasma Mg level (PMg)(Figure Group Vehicle SAHA Significance 2D) was evidently higher in the SAHA group than the vehicle Weight, g 21.7360.64 21.8961.05 NS group (SAHA: 2.6060.09 mg/dl versus vehicle: 2.2060.08 mg/dl; UV, ml/hzg 6.1460.84 5.6360.92 NS P,0.01, n=6) (Table 1) owing to significant decreases in frac- GFR, ml/hzg 0.5160.09 0.4760.07 NS ++ 6 6 tional and absolute Mg excretion levels (FEMg and EMg, respec- PCa, mg/dl 10.02 0.22 10.12 0.28 NS 6 6 , tively) (Figure 2, E and F) induced by SAHA. The GFR based on PMg, mg/dl 2.20 0.08 2.60 0.09 P 0.01 6 6 , FITC-inulin clearance (Table 1) was not significantly different FECa,% 0.56 0.07 0.23 0.06 P 0.01 m z 6 6 , between SAHA- and vehicle-treated animals, and there was no ECa, g/h g 0.27 0.03 0.10 0.02 P 0.01 6 6 , significant difference in urinary volume (UV) between the two FEMg, % 17.52 3.02 7.87 1.53 P 0.05 E , mg/hzg 1.7760.24 0.8760.11 P,0.01 groups (Table 1). Together, these data indicate a renal tubular- Mg Values are expressed as means6SEMs; n=6 male animals ages 10–12 weeks. specific effect of HDAC inhibitor on Ca++ and Mg++ excretion, which is independent of glomerular hemodynamics. 10.4060.10 mg/dl [Figure 3A]; PMg: SAHA group: 2.7160.12 The Long-Term Effects of HDAC Inhibitors on Urinary mg/dl versus control: 2.4360.05 mg/dl [Figure 3B]; P,0.05, Calcium and Magnesium Excretion n=9) (Table 2), compatible with the observation of significantly To reveal the renal transport function over a chronic phase, we reduced FECa (P,0.01, n=9) (Figure 3C, Table 2) and FEMg performed 24-hour urinalysis on age (8–10 weeks old) and sex (P,0.01, n=9) (Figure 3D, Table 2) in the SAHA group. The (male) matched wild-type mice receiving SAHA at 10 mg/kg ECa and EMg levels were also significantly lowered by SAHA 21 body wt per day. The PCa and PMg levels in SAHA-treated (P,0.05, n=9) (Table 2), despite unchanged GFR and UV levels animals were both significantly higher than in the vehicle (Table 2). The plasma levels of Na+,K+, and phosphate were not group (PCa: SAHA group: 11.3860.44 mg/dl versus control: affected by SAHA, and there was no significant change in urinary

Figure 2. Renal clearance analyses of plasma and urine electrolyte levels in wild-type mice treated with SAHA. (A–C) The effects of SAHA on (A) plasma Ca++ levels, (B) fractional excretion rates, and (C) absolute excretion rates of Ca++ in 1-hour urine collections. (D and E) The effects of SAHA on (D) plasma Mg++ levels, (E) fractional excretion rates, and (F) absolute excretion rates of Mg++ in 1-hour urine collections. Veh, vehicle. *P,0.05; **P,0.01.

666 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 663–676, 2015 www.jasn.org BASIC RESEARCH

the CLDN14 KO animals were refractory to- ward SAHA-induced hypercalcemia and hypermagnesemia (Figure 3, A and B, Table 2), mindful that the CLDN14 KO was already hypermagnesemic (Figure 3B) because of reduced baseline levels of urinary Mg++ excretion.10 The urinary Ca++ and Mg++ changes were both abolished in CLDN14 KO (Figure 3, C and D, Table 2), suggesting that SAHA primarily acted on the kidney where CLDN14 was exclusively localized. To elucidate the tubular origin of SAHA effects, we pretreated wild-type animals with furosemide 2 at 40 mg/kg body wt 1 per day. Furosemide, by inhibiting the NKCC2 transporter in the TALH, induces natriuresis and diuresis, which in turn, abolished the transepithelial voltage for paracellular Mg++ and Ca++ reabsorption.3,28–30 The furosemide treatment significantly increased UV, FECa,andFEMg (Figure 3, C and D, Table 2). In addition to the elevated baseline levels, furosemide Figure 3. Metabolic analyses of plasma and urine electrolyte levels in wild-type (WT) abolished the SAHA effects on plasma and CLDN14 KO mice treated with SAHA and/or furosemide. (A and B) The effects of (Figure 3, A and B) and urine (Figure 3, C SAHA on plasma levels of (A) Ca++ and (B) Mg++ in WT versus CLDN14 KO mice or and D) Ca++ and Mg++ levels, identifying mice pretreated with furosemide. (C and D) The effects of SAHA on fractional excretion the TALH as the major nephron segment ++ ++ rates for (C) Ca and (D) Mg in WT versus CLDN14 KO mice or mice pretreated for SAHA function. Together, these results fi with furosemide in 24-hour urine collections. Furo, furosemide; n.s., not signi cant; indicate that HDAC inhibitors act on , , veh, vehicle. *P 0.05; **P 0.01. CLDN14 to regulate Ca++ and Mg++ reabsorption in the TALH of the kidney. excretion of Na+,K+, and phosphate (Supplemental Figure 6). The Tubular Differences in HDAC Inhibitor Effects 2 TsA at 1 mg/kg body wt 1 per day exerted similar effects on renal To reveal the tubular specific contribution to observed HDAC Ca++ and Mg++ metabolism. We then asked whether SAHA inhibitor effects, we studied the Ca++ transport pathways in affected other Ca++-related hormonal or genetic pathways in freshly isolated ex vivo perfused TALH and DCTs from SAHA- 2 the kidney. Because chronic SAHA treatment induced mild hy- treated animals. At 4 hours after a single dose of 5 mg/kg body wt 1 percalcemia, the circulating parathyroid hormone (PTH), 1,25- SAHA or vehicle treatment, age- and sex-matched mice were (OH)2-vitD3, and fibroblast growth factor 23 (FGF23) levels euthanized for immediate kidney harvest followed by manual were measured to determine if hypercalcemia originated from tubule isolation. On average, more than five tubules were per- changes in these hormonal systems. None of these hormones fused for each animal. Five to six animals were studied in each changed during SAHA treatments (Supplemental Figure 7), and treatment group. The tubule perfusion protocol for TALH had there was no significant change in the expression levels of other been described before by us.6 The TALHs from SAHA- or vehicle- major Ca++/Mg++ transporters or regulators from the kidney— treated animals showed similar lumen-positive potential CLDN16, CLDN19, TRPV5, TRPM6, NKCC2, ROMK, TSC, (Vte) (Figure 4A) that was fully inhabitable by luminal furose- Klotho, and VDR (Supplemental Figure 8). Together, these re- mide (0.05 mM), suggesting that the transcellular pathway sults have revealed a Ca++/Mg++-specific effect for the HDAC mediated through NKCC2 was not affected by SAHA. The trans- inhibitor in the kidney, which is independent of systemic endo- epithelial resistance (Rte) in the presence of furosemide was sig- crine factors. nificantly reduced by SAHA (P,0.05, n=5–6) (Figure 4B), suggesting its role in stimulating paracellular conductance. To The Genetic and Cellular Origins of HDAC Inhibitor- determine the ion-specific permeabilities of the TJ, changes in Induced Hypocalciuria in the Kidney the junctional diffusion potential produced by a NaCl gradient To elucidate the genetic origin of SAHA effects, the CLDN14 of 145 versus 30 mM were recorded in the absence of active KO mice (previously backcrossed to C57BL/6 background)9,10 transport (inhibited by furosemide). The Goldman equation + 2 were treated with the same SAHA regimen as described else- allowed calculation of paracellular Na (PNa)andCl (PCl)per- where for wild-type animals. In contrast to the wild-type mice, meabilities from the diffusion potential. The PNa in the SAHA

J Am Soc Nephrol 26: 663–676, 2015 Epigenetic Regulation of Calcium Metabolism 667 BASIC RESEARCH www.jasn.org

Table 2. The 24-hour plasma and urine electrolyte levels in wild-type and claudin-14 KO animals under SAHA and furosemide treatments Group WT+Veh WT+SAHA CLDN14KO+Veh CLDN14KO+SAHA WT+Veh+Furo WT+SAHA+Furo Weight, g 19.1560.42 19.4160.56 18.6461.93 19.1860.56 19.1760.28 18.3060.43 UV, ml/24 hzg 44.9563.96 37.1164.54 39.8365.78 44.0362.07 137.11615.95a 110.5969.03 GFR, ml/24 hzg2.8260.36 2.0960.51 1.9760.32 2.4260.29 1.7760.44 1.6960.13 b PCa, mg/dl 10.4060.10 11.3860.44 10.7360.59 11.1360.34 10.6160.43 9.9560.41 b a PMg, mg/dl 2.4360.05 2.7160.12 2.7560.05 2.7660.05 2.3360.11 2.2060.06 c d FECa,% 0.6860.08 0.2760.08 0.6060.18 0.4560.07 1.8960.67 1.9760.38 b ECa, mg/24 hzg1.9660.33 0.7260.26 1.1160.20 1.2060.24 2.8460.78 3.3760.75 c a FEMg,% 14.0161.53 6.0661.71 10.6861.00 9.9060.83 32.8361.30 22.5066.40 c EMg, mg/24 hzg9.4161.18 3.1660.80 5.7561.01 6.5060.55 13.2463.13 8.3462.15 Values are expressed as means6SEMs; n=5–9 male animals ages 8–10 weeks. WT, wild type; Veh, vehicle; Furo, furosemide. aP,0.01 versus the WT group with Veh treatment. bP,0.05 versus Veh treatment in the same group. cP,0.01 versus Veh treatment in the same group. dP,0.05 versus the WT group with Veh treatment.

++ Figure 4. Characterization of major Ca transport pathways in perfused TALHs and DCTs. (A) The transepithelial voltage Vte,(B) + 2 transepithelial resistance Rte, (C) paracellular Na permeability PNa, and (D) paracellular Cl permeability PCl values in ex vivo-perfused TALH tubules from wild-type mice treated with SAHA or vehicle. (E and F) Normalized Fura-2 fluorescence ratio in DCTs from wild-type mice treated with SAHA or vehicle perfused with (E) luminal high Ca++ buffer (+3 mM) or (F) basolateral low Na+ buffer (30 mM). Veh, vehicle. *P,0.05. group was approximately 2-fold higher than that in the vehicle DCT epithelia, we adopted intracellular calcium imaging as a group (P,0.05; n=5–6) (Figure 4C); the PCl in the SAHA group tool to report transcellular calcium transport levels for perfused showed increasing trend but did not reach significance (Figure DCTs (Concise Methods). A ratiometric fluorescence dye (Fura-2) 4D). Together, these results are consistent with our previous re- was used to report intracellular calcium levels (Supplemental cording that CLDN14 primarily affected CLDN16 cation per- Figure 9) independent of dye concentration, imaging parame- meability through protein interaction.9 The TRPV5-mediated ters, or cell volume. The normalized ratio of fluorescence emis- transcellular pathway in the DCT is another major component sion at 340 versus 380 nm for perfused DCTs from SAHA- or of renal Ca++ transport machinery. Because endogenous TRPV5 vehicle-treated animal groups is shown in Figure 4E. A high or TRPM6 conductance has never been captured from native luminal Ca++ solution (3 mM) was used to elicit changes in

668 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 663–676, 2015 www.jasn.org BASIC RESEARCH intracellular Ca++ levels. The intracellular Ca++ changes mea- levels by 30% compared with wild-type CaSR (P,0.05, n=4) sured at 20 seconds after high Ca++ challenge were regarded as (Figure 5A). Pretreating the HEK293 cells with SAHA at TRPV5-mediated effect. Shown in Figure 4E, the initial peak in 0.33 mM completely abolished the CaSR mutational effect on intracellular Ca++ levels was similar in magnitude between miR gene transcription (Figure 5B). To mimic the ADH disease in SAHA and vehicle groups (P=0.92, n=5–6), which can be fully vivo, we pretreated age- (8–10 weeks old) and sex-matched inhibited by ruthenium red (a TRPV5 inhibitor) at 1 mM, sug- (male) wild-type mice (strain C57BL/6) with the calcimimetic 2 gesting that TRPV5 activity was not affected by SAHA. How- compound cinacalcet at 30 mg/kg body wt 1 per day and ex- ever, the intracellular Ca++ level in SAHA group became different amined the SAHA effect on miR gene expression (at 4 hours after 2 from that in the vehicle group after 50 seconds, and the difference receiving 25-mg/kg body wt 1 SAHA treatment). The transcript reached significance at 100 seconds (Figure 4E) (P=0.04, n=5–6). levels of pri-miR-9–1 and pri-miR-374 in isolated TALH tubules Unlike the vehicle-treated animals, the DCT cells from the SAHA were both profoundly increased by SAHA to 2.13-fold group were not able to remove the intracellular Ca++ surge in- (P,0.001, n=5) (Figure 5C) and 3.61-fold (P,0.01, n=5) (Fig- duced by the luminal Ca++ load (Figure 4E), suggesting defects in ure 5D) of the vehicle level, respectively, which led to a reciprocal the basolateral Ca++ efflux. The major Ca++ efflux mechanism decrease in renal CLDN14 mRNA levels (P,0.001, n=5) (Figure in the DCT is through the NCX1, which can be inhibited by 5E). Notably, the SAHA effect size on miRNA and CLDN14 was low extracellular Na+ levels. When perfused DCTs were bathed much higher in cinacalcet pretreated animals (Figure 5, C–E) with low Na+ solution (30 mM), intracellular Ca++ levels than not treated animals (Figure 1, C, E, and F). Cinacalcet sig- ++ exhibited a transient increase caused by inhibition of Ca nificantly reduced the baseline level of PCa (to 7.9960.27 mg/dl; efflux (Figure 4F). The intracellular Ca++ peak (measured at P,0.01, n=7) (Figure 5F, Table 3) compared with that in 100 seconds) was clearly lower in SAHA-treated animals than animals not treated with cinacalcet (PCa: 10.3660.22 mg/ in vehicle treatment animals (Figure 4F), suggesting that the dl); baseline PMg showed a clear decreasing trend (P=0.07, NCX1 activity was already blunted by SAHA. Because low extra- n=7) (Figure 5G, Table 3), consistent with our previous ob- cellular Na+ caused unexpected fluctuation in intracellular Ca++ servation.10 SAHA reversed the cinacalcet effect by increas- (Figure 4F), the statistical significance value (P) for comparing ing PCa and PMg (Figure 5, F and G) in cinacalcet pretreated SAHA and vehicle treatments was 0.08 (n=5–6). Using a more animals. PMg recovered to above the control level with no 31 specific NCX1 inhibitor, such as KB-R7943 mesylate, although cinacalcet treatment, but PCa was still below the normal not within the scope of this study, may clarify its role in medi- range. The cinacalcet-induced hypercalciuria (FECa)(Figure5H, ating SAHA effects. Together, these results indicate that SAHA Table 3) and hypermagnesiuria (FEMg) (Figure 5I, Table 3) were ++ reduces the transcellular transport capacity for Ca in the DCT both effectively reversed by SAHA. FECa was 1.8260.31% in ani- that likely results from reduced luminal Ca++ delivery because of mals receiving both cinacalcet and SAHA (only 2.60-fold higher increased reabsorption in the TALH. than animals receiving no treatment), whereas cinacalcet treatment alone increased FECa by 7.66-fold (Figure 5H, Table 3). HDAC Inhibitors Correct CaSR Mutational Effects in The SAHA effect on FEMg in cinacalcet pretreated animals was ADH Disease even more dramatic. Despite a 1.70-fold increase of baseline ADH is caused by activating mutations in the CaSR gene. FEMg by cinacalcet (Figure 5I), animals receiving both cinacalcet Knowing that miRs are an important second messenger of CaSR and SAHA had even lower FEMg levels (4.9461.21%) (Figure 5I, signaling,9 we asked whether CaSR mutations deregulated miR Table 3) compared with those receiving no treatment at all abundance as a potential causal mechanism and whether HDAC (15.7861.64%; P,0.01, n=7) (Figure 5I, Table 3). Together, these inhibitors corrected the CaSR mutational effects as a novel ther- results have revealed a novel pathogenic mechanism for ADH dis- apeutic approach for ADH disease. We generated three common ease based on miR deregulation and suggested a novel therapeutic ADH mutations in the human CaSR gene (E127A, F128L, and approach using an HDAC inhibitor to reverse miR abnormalities. T151M previously described by Pollak et al.15 and Pearce et al.16), transfected the wild-type CaSR gene and its mutants into a well miR Manipulation Provides Therapeutic Principles for established cell model for studying CaSR signaling (HEK293 HDAC Inhibitors cells that expressed no endogenous CaSR),17 and quantified Because the SAHA effect and the ADH disease mechanism the transcriptional levels of four miR genes—miR-9–1, miR- intersected at the miRs, we asked whether manipulation of miRs 9–2, miR-9–3 (all three loci transcribing mature miR-9 hairpin per se was sufficient to induce changes in renal calcium metab- sequence), and miR-374. Among them, the miR-9–1andmiR- olism. Since our previous discovery of the CLDN14 targeting 374 genes were transcriptionally suppressed by the CaSR miRs miR-9 and miR-374,9 attempts to knockdown miR expres- mutants. The cellular abundance of the primary miR-9-1 (pri- sion have been made with antagomirs32 based on the locked miR-9-1) transcript was significantly reduced to 53% of the nucleic acids (LNAs). We have screened for LNA sequences wild-type level by the F128L and T151M mutations (P,0.05, ranging from 8 to 23 nucleotides long33–35 using cultured n=4) (Figure 5A); the E127A mutation elicited a decreasing TALH cells and found the most effective sequences for miR-9 trend in miR-9–1 transcription (P=0.07, n=4) (Figure 5A). All and miR-374 (Concise Methods). In vitro in the primary TALH three mutations significantly reduced the pri-miR-374 transcript cultures, transfection with anti–miR-9 or anti–miR-374 but not

J Am Soc Nephrol 26: 663–676, 2015 Epigenetic Regulation of Calcium Metabolism 669 BASIC RESEARCH www.jasn.org

Figure 5. Activation of CaSR effects on miR and CLDN14 gene expressions and plasma and urine electrolyte levels in cell and animal models. (A and B) Pri-miR transcript levels in HEK293 cells transfected with CaSR mutations found from patients with ADH in the (A) absence or (B) presence of SAHA. (C–E) The effects of SAHA on (C) pri-miR-9–1, (D) pri-miR-374, or (E) CLDN14 transcript levels in the TALH of the mouse kidneys pretreated with cinacalcet. (F and G) The effects of SAHA on plasma levels of (F) Ca++ and (G) Mg++ in wild- type (WT) mice pretreated with cinacalcet. (H and I) The effects of SAHA on fractional excretion rates for (H) Ca++ and (I) Mg++ in WT mice pretreated with cinacalcet in 24-hour urine collections. Cina, cinacalcet; veh, vehicle. *P,0.05; **P,0.01; ***P,0.001.

Table 3. The 24-hour plasma and urine electrolyte levels in wild-type animals under SAHA and cinacalcet treatments Group WT+Veh WT+SAHA WT+Veh+Cina WT+SAHA+Cina Weight, g 20.2360.23 19.8960.41 20.9060.71 20.4860.33 UV, ml/24 hzg 41.2863.40 41.5165.03 40.9366.60 33.8464.11 GFR, ml/24 hzg2.5360.16 2.2360.44 2.5660.51 2.7860.40 a b a PCa, mg/dl 10.3660.22 11.2060.29 7.9960.27 9.5960.47 c d a PMg, mg/dl 2.3360.04 2.6960.10 2.1560.10 2.4960.07 a b c FECa,% 0.7060.09 0.3460.10 5.3660.23 1.8260.31 a b a ECa, mg/24 hzg1.7960.21 0.9560.29 10.6661.86 4.5260.63 c b c FEMg,% 15.7861.64 5.5661.96 26.8763.10 4.9461.21 c c EMg, mg/24 hzg9.5061.27 3.3161.28 14.8963.43 3.0160.72 Values are expressed as means6SEMs; n=7 male animals ages 8–10 weeks. WT, wild type; Veh, vehicle; Cina, cinacalcet. aP,0.05 versus Veh treatment in the same group. bP,0.01 versus the WT group with Veh treatment. cP,0.05 versus Veh treatment in the same group. dP=0.07 versus the WT group with Veh treatment. scrambled antagomir induced signiﬁcant increases in CLDN14 or the podocytes37 with systemic injection of antagomirs, we gene expression by 3.07-fold and 5.31-fold, respectively found that systemic injection of anti–miR-9 or anti–miR-374 (P,0.001, n=4) (Figure 6A). The in vivo knockdown seemed was not able to knockdown miR in renal tubules throughout 2 to be much more difﬁcult. Although several studies have reported thedosagerangeof1–25 mg/kg body wt 1 in mice. Instead, when effective ablation of glomerular miRs from the mesangial cells36 the LNA antagomirs were coinjected with the in vivo jetPEI

670 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 663–676, 2015 www.jasn.org BASIC RESEARCH

levels. In freshly isolated TALH cells pooled from antagomir-treated mouse kidneys (n=5), the CLDN14 protein level was 3.25-fold higher with anti–miR-374 than scrambled antagomir treatment (Figure 6C). The change in CLDN14 levels at the TJ was particularly striking. The thin inter- digitated TJ lines faintly stained for CLDN14 in control mouse kidneys (Figure 6D) were profoundly strengthened in anti– miR-374–treated mouse kidneys (Figure 6D, arrow), with occasional subapical stain- ing spotted in some TALH cells (Figure 6D, arrowhead). To reveal the functional effects of antagomir treatment in the kidney, we performed 24-hour urinalysis on age- (7–8 weeks old) and sex-matched (male) wild-type mice receiving anti–miR-374 injec- 2 tions at 2.5 mg/kg body wt 1 per day. The PCa was not changed by anti–miR-374, but the PMg was significantly lower in the anti– miR-374 group than the scrambled (P,0.05, n=6) (Figure 6E, Table 4). The fractional excretion rates for Ca++ and Mg++ were both significantly increased by anti– miR-374 in 24-hour urine collections (FECa:3.45-fold;FEMg: 2.36-fold versus scrambled [Figure 6, F and G]; P,0.01, n=6) (Table 4). Other renal parameters, such as GFR and UV, were not changed by anti–miR-374 (Table 4), indicating a direct tubular effect. Together, the anti– miR-374 knockdown animals phenocopied Figure 6. The renal effects of antagomir treatments. (A) CLDN14 mRNA levels in the CLDN14 transgenic overexpression ani- primary TALH cultures transfected with antagomirs against miR-9 or miR-374. (B) mals described by us before,10 establishing a CLDN14 mRNA levels in mouse kidneys treated with antagomirs against miR-9 or miR- therapeutic principle for using miRs to ma- 374. (C) CLDN14 protein levels in freshly isolated TALHs from mouse kidneys treated nipulate CLDN14 expression and urinary with anti–miR-374 versus scrambled antagomir. (D) Mouse kidney sagittal sections calcium excretion. immunostained for CLDN14 from anti–miR-374 versus scrambled treatments. Scale bar, 10 mm. (E) Plasma Mg++ and fractional excretion of (F) Ca++ and (G) Mg++ levels in anti–miR-374 versus scrambled animal groups. a-miR, anti-miR; scrbl, scrambled. DISCUSSION *P,0.05; **P,0.01; ***P,0.001. This work has identified a novel approach DeliveryReagent(basedonthepolyethyleniminecationic based on epigenetic drugs and miR manipulation to reduce polymers), a single dose of anti–miR-9 or anti–miR-374 at urinary calcium excretion. For many years, the only available 2 2.5 mg/kg body wt 1 significantly increased the CLDN14 therapy for hypercalciuria was through the use of diuretics, such mRNA level in the kidney 24 hours after injection by 1.48- as thiazide and amiloride. Although it is still debated, the fold and 2.02-fold, respectively, relative to the scrambled anta- physiologic mechanism of diuretic-induced hypocalciuria is gomir (P,0.01, n=5) (Figure 6B). Because anti–miR-374 was considered secondary to extracellular volume depletion and more effective than anti–miR-9 and miR-9 had known tumor- reduced GFR, which increase paracellular Ca++ reabsorption in igenic implication through targeting E-cadherin,38 we, thus, the PTs.39 Since the discovery of CLDN16 and CLDN19 from the focused on miR-374 to study its role in renal tubular functions. FHHNC disease, the possibility of directly manipulating loop We gave wild-type C57BL/6 mice anti–miR-374 injections at Ca++ reabsorption has emerged. The CLDN16- and CLDN19- 2 2.5 mg/kg body wt 1 per day and measured CLDN14 protein deficient animals that we have generated using small inter- abundance, localization, and plasma and urinary electrolytes fering RNAs (siRNAs)6,7 have proven a principle to induce

J Am Soc Nephrol 26: 663–676, 2015 Epigenetic Regulation of Calcium Metabolism 671 BASIC RESEARCH www.jasn.org

Table 4. Plasma and urine electrolyte levels in antagomir- we report using an extremely low dose of SAHA (at 15 mg/m2) injected animals to promote positive Ca++ homeostasis. Because of the low Group Scrambled Anti–miR-374 Significance dosage, AEs can be significantly reduced. The physiologic out- fi Weight, g 17.7560.42 17.8460.72 NS come of SAHA stems from its direct and speci c effect on UV, ml/24 hzg 34.8163.22 41.6567.67 NS augmenting the TJ permeability in the TALH. From a cell bi- GFR, ml/24 hzg3.3060.56 2.7560.47 NS ology stand, the SAHA effect proves that TJ function can be 6 6 PCa, mg/dl 10.43 0.23 10.50 0.22 NS modulated transiently and reversibly with pharmacologic re- 6 6 , PMg, mg/dl 2.68 0.20 2.10 0.09 P 0.05 agents, supporting a novel view of a drugable TJ. SAHA is a 6 6 , FECa,% 0.93 0.27 3.21 0.16 P 0.01 paninhibitor for class I HDAC, which includes HDAC1, -2, -3, m z 6 6 , ECa, g/24 h g2.670.36 9.11 1.38 P 0.01 -6, and -8.21 Additional refinement of this class of inhibitors with 6 6 , FEMg,% 8.86 2.23 20.94 1.53 P 0.01 improved substrate specificity will allow synthesis of drugs m z 6 6 EMg, g/24 h g7.611.98 11.92 2.25 NS against each specific HDAC member and significantly improve 6 – Values are expressed as means SEMs; n=6 male animals ages 7 8weeks. safety. It is not know which specific HDAC member mediates miR transcriptional regulation and promotes Ca++ reabsorption in the kidney. hypercalciuria in humans based on siRNA treatments. Nev- Since the discovery of CaSR, studies of its downstream ertheless, because there was nogain-of-functionmutationever signaling pathways have attracted immense research interest. found for CLDN16 or CLDN19, therapies relying on this path- The classic CaSR signaling pathway is mediated through the way to reduce Ca++ excretion have been difficult until our recent binding to the G proteins, which stimulate the production of discovery of the inhibitory subunit CLDN14.9,10 Because miR-9 diacylglycerol and inositol 1,4,5-trisphosphate andincrease the and miR-374 directly target CLDN14 in the kidney, approaches intracellular Ca++ levels.49 These second messengers (inositol acting to regulate these miRs are specifictothekidneyandavoid 1,4,5-trisphosphate and intracellular Ca++) have been broadly the unwanted effects from manipulating CLDN14 itself using used to assess the genetic mutations of CaSR in Xenopus oo- approaches, such as siRNA, that can also affect the inner ear, cytes and HEK293 cells.15–18 Our study has revealed a novel where CLDN14 is important for the generation of endocochlear second messenger based on miRs that is particularly useful for potential.40 The HDAC inhibitors acting farther upstream of assessing the activating mutations of CaSR found in ADH miRs will be particularly important. Although generally consid- disease. The common ADH mutations invariably repress the ered as a global regulator, pharmacologic inhibition of HDACs transcription of miR-9 and miR-374 genes, which can be re- alterstheexpressionofonly2%ofalltranscribedgenes.41 Our lieved by HDAC inhibitors. Ablation of endogenous miR-374 animal studies have strengthened this view that the HDAC in- with antagomirs generated ADH-like phenotypes in mice, em- hibitors only affect the miR–CLDN14 pathway in the kidney, phasizing the pathogenic role of miR in ADH development. with no promiscuous effect on other Ca++/Mg++ and Na+/K+ The HDAC inhibitors were particularly effective to ameliorate transporters in the kidney (Supplemental Figure 8). Because hypomagnesemia, hypercalciuria, and hypermagnesiuria in a HDAC inhibitor does not alter GFR, the glomerulotubular bal- pharmacologic ADH animal model. Hypocalcemia was only ance is well maintained. Functional recordings from perfused partially corrected, largely owing to the continuous suppression renal tubules further clarified the HDAC inhibitor effects by of PTH secretion in these animals. It is not known whether showing that increased Ca++ reabsorption came from the HDACinhibitorsactontheparathyroidglands.Anelegantin TALH but not the DCT. vitro study has shown direct binding of type 1 HDAC to the PTH The antitumor ability of SAHA was first shown in several in gene promoter and shown that HDAC inhibitors stimulate the vivo animal models of cancer using a dosage range of 50–100 PTH gene transcription in several cell models.50 In vivo, the 2 mg/kg body wt 1 (equivalent to 150–300 mg/m2 surface area HDAC inhibitor treatment had no overt effect on circulating 2 [SA] 1), including a xenograft mouse model of human pros- PTH levels (Supplemental Figure 7), despite mild hypercalcemia. tate cancer42 and a transgenic mouse model of leukemia.43 Initial toxicology studies in rodents showed that the maximal administered dose of 3000 mg/m2 produced no organ toxic- CONCISE METHODS ity.44 Based on these preclinical data, phase I and II trials have been carried out to test SAHA as a novel treatment for patients Reagents, Abs, Cell Lines, and Animals with refractory cutaneous T cell lymphoma at a dosage of The following Abs were used in this study: rabbit polyclonal anti- 300 mg/m2.44,45 The common adverse effects (AEs) include Tamm–Horsfall protein (Biomedical Technologies), rabbit polyclonal fatigue, diarrhea, nausea, thrombocytopenia, and hyperglyce- anti-CLDN14 (against RAPSVTSAAHSGYRLNDYV), rabbit poly- mia.46 The AE frequency increased with the dosage, but the clonal anti-CLDN16 (against SYSAPRTETAKMYAVDTRV), and rabbit pattern remained the same. No abnormality in serum or urine polyclonal anti-CLDN19 (against NSIPQPYRSGPSTAAREYV). Hu- Ca++ level has been reported for SAHA, although hypocalce- man HEK293 cells (Joan Brugge) were cultured in DMEM (Invitrogen) mia was found in patients receiving other types of HDAC in- supplemented with 10% FBS, penicillin/streptomycin, and 1 mM so- hibitors, such as Entinostat47 or Romidepsin.48 In this study, dium pyruvate. All mice were bred and maintained according to the

672 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 663–676, 2015 www.jasn.org BASIC RESEARCH

Washington University animal research requirements, and all proce- TALH Tubule Perfusion and Electrophysiologic dures were approved by the Institutional Animal Research and Care Recording Committee. Wild-type C57BL/6 mice were from Charles River Labora- The methods for perfusion and transepithelial measurements in tory. The CLDN14+/lacZ reporter/KO mice were from Tamar Ben-Yosef freshly isolated mouse TALH segments were performed as described and backcrossed to the C57BL/6 background for seven generations. previously.6 The TALH tubule was held and perfused by a concentric glass pipette system. The perfusion pipette was double-barreled, with Pharmacologic Manipulation in Experimental Animals an outer diameter of 10–12 mm. One barrel was used for perfusion, SAHA and TsA were dissolved in 5% (wt/vol) ethanol/0.9% saline fluid exchange, and voltage measurement. The second barrel was solution and fed to mice with gavage syringe. Furosemide was used for constant current injection (13 nA). The collection side con- dissolved in 5% (wt/vol) DMSO/0.9% saline solution and intraper- sisted of a glass pipette with an inner diameter of 45 mm. Cable itoneally (ip) injected to animals. All animals had free access to water equations as described52 were used appropriately to calculate trans- and were housed under a 12-hour light cycle. Blood samples were epithelial resistance Rte. Equivalent short circuit current Isc was calcu- taken by cardiac puncture rapidly after initiation of terminal lated from Rte and Vte according to Ohms law. The rates of perfusion anesthesia and centrifuged at 4°C for 10 minutes. Kidney were dis- were 10–20 nl/min. The bath was thermostated at 37°C. Continuous sected out and immediately frozen at 280°C. bath perfusion at 3–5 ml/min was obtained by gravity perfusion. Rela- tive paracellular permeabilities were calculated from the observed trans- Animal Metabolic Studies epithelial diffusion potentials in the presence of 0.1 mM furosemide Mice were housed individually in metabolic cages (Harvard Appa- according to the Goldman equation. ratus) with free access to water and food for 24 hours. Urine was ++ ++ collected under mineral oil. The plasma and urine Ca and Mg ++ levels were determined with an atomic flame absorption spectropho- DCT Perfusion and Intracellular Ca Imaging DCTs were identified by morphology and cut at the macula densa and tometer (PerkinElmer). The creatinine levels were measured with an approximately 300–500 mm downstream using a dissection micro- enzymatic method that was independent of plasma chromogens.51 scope (Leica MZ16) at 4°C. DCTs were loaded at 37°C for 30–45 The fractional excretion of electrolytes was calculated using the fol- minutes with 5 mM Fura-2-AM in the presence of 0.09% (vol/vol) lowing equation: FE =volume (V)3urine (U )/(GFR3P ), ion ion ion ion pluronic. DCTs were then transferred into the bath on a heated mi- where GFR was calculated according to the clearance rate of creati- croscope stage. The bath was thermostated at 37°C and continuously nine (GFR=V3U /P ). creatinine creatinine perfused at 8–10 ml/min with control solution (140 mmol/L NaCl, 0.4 mmol/L KH PO ,1.6mmol/LK HPO ,1mmol/LMgCl ,5mmol/L Renal Clearance Protocol 2 4 2 4 2 The method for performing renal clearance measurements in laboratory glucose, and 1.3 mmol/L Ca-gluconate at pH 7.4). Tubules were held – mice has been described by Hou et al.6 and Gong and Hou.10 Mice were and perfused (10 20 nl/min) by a concentric glass pipette system as 52 anesthetized by ip injection of Inactin (100 mg/kg; Sigma-Aldrich). The described. The perfusion pipette was double-barreled and allowed the jugular vein was catheterized for intravenous infusion of 0.9% saline at alternate perfusion of control solution, control solution plus an addi- 2 ml/min, with 1% FITC-inulin included in the infusate. After an equil- tional 3 mM Ca-gluconate (+3C), or control solution plus an additional m ibration period of 60 minutes, renal clearance measurements were car- 3 mM Ca-gluconate and 1 M ruthenium red (+3C+RuR). Only well fi ried out for a 60-minute period. Urine was collected under mineral oil, perfused DCTs with well de nable, and open lumens were used. Fluo- and a 30-ml blood sample was taken at hourly intervals. Urine and plasma rescence was monitored by a digital imaging system (Visitron Systems Ca++ and Mg++ concentrations were measured by atomic flame absorp- GmbH) and analyzed by MetaFluor software. Ratio images (340:380 nm) fl tion spectrophotometer (PerkinElmer). Urine and plasma FITC-inulin were obtained every 3 seconds and stored for of ine analysis. After a levels were measured in 100 mM Hepes buffer (pH 7.0) with fluorescence short equilibration period, luminal perfusion was switched to either spectrophotometer (BioTek). The fractional excretion of electrolytes was +3C or +3C+RuR for 3 minutes. The luminal calcium load was sub- sequently washed out, and DCTs were challenged with a basolateral calculated using the following equation: FEion=V3Uion/(GFR3Pion), where GFR was calculated according to the clearance rate of FITC-inulin exchange to iso-osmolar low NaCl (30 NaCl) solution (30 mmol/L NaCl, 0.4 mmol/L KH2PO4, 1.6 mmol/L K2HPO4, 1 mmol/L MgCl2, (GFR=V3Uinulin/Pinulin). 5 mmol/L glucose, 230 mmol/L mannitol, and 1.3 mmol/LCa-gluconate Establishing Primary Cultures of Mouse TALH Cells at pH 7.4) for another 3 minutes. We used an immunomagnetic separation method to isolate the TALH cells from the mouse kidney.9 Abs against the TALH cell-specificsurface Real-Time PCR Quantification antigen Tamm–Horsfall protein (a glycosylphosphatidylinositol [GPI]- Total RNA, including miR, was extracted using Trizol (Invitrogen). anchored protein that is exclusively expressed in the TALH and the early Cellular mRNA was reverse transcribed using the Superscript-III Kit part of the DCT) were coated onto the paramagnetic polystyrene beads (Invitrogen) followed by real-time PCR amplification using SYBR (Dynabeads M-280; Dynal), allowing immunoprecipitation of the Green PCR Master Mix (Bio-Rad) and the Eppendorf Realplex2S TALH cells from collagenase-digested mouse kidneys. The isolated cells System. The design of claudin-14 primers was described before by were plated in DMEM medium supplemented with 10% FBS, penicillin/ Gong et al.9 The design of pri-miR primers and ChIP primers was 2Δ streptomycin, and 1 mM sodium pyruvate for 16 hours followed by according to Gong and Hou.10 Results were expressed as 2 Ct values immediate pharmacologic treatment. with ΔCT=Ctgene2Ctb-actin.

J Am Soc Nephrol 26: 663–676, 2015 Epigenetic Regulation of Calcium Metabolism 673 BASIC RESEARCH www.jasn.org

ChIP DISCLOSURES Freshly isolated TALH cells or primary TALH cell cultures were None. crosslinked in 1% paraformaldehyde (pH 7.0) for 10 minutes at room temperature. The crosslinked cells were lysed following the instruc- tions of the MAGnify ChIP System (Invitrogen). Chromatin was sonicated into 500-bp fragments with the Bioruptor UCD-200 (high REFERENCES power, 15 cycles of 30 seconds on and 30 seconds off; Diagenode). Sheared chromatin was precipitated with Ab-bound Dynabeads 1. Simon DB, Lu Y, Choate KA, Velazquez H, Al-Sabban E, Praga M, Casari G, Bettinelli A, Colussi G, Rodriguez-Soriano J, McCredie D, Milford D, (Invitrogen). After reverse crosslink and DNA extraction, Ab-bound Sanjad S, Lifton RP: Paracellin-1, a renal tight junction protein required DNA was eluted and analyzed with real-time PCR. Two criteria for for paracellular Mg2+ resorption. Science 285: 103–106, 1999 normalization were applied: fold enrichment over anti-IgG back- 2. Konrad M, Schaller A, Seelow D, Pandey AV, Waldegger S, Lesslauer A, ground signal and input control over chromatin input signal. Vitzthum H, Suzuki Y, Luk JM, Becker C, Schlingmann KP, Schmid M, Rodriguez-Soriano J, Ariceta G, Cano F, Enriquez R, Juppner H, Bakkaloglu SA, Hediger MA, Gallati S, Neuhauss SC, Nurnberg P, Immunolabeling and Fluorescence Microscopy Weber S: Mutations in the tight-junction gene claudin 19 (CLDN19) are For viewing claudin localization in the kidney, fresh cryostat sections associated with renal magnesium wasting, renal failure, and severe (10 mm) were fixed with cold methanol at 220°C followed by block- ocular involvement. Am J Hum Genet 79: 949–957, 2006 ing with PBS containing 10% FBS and incubation with primary Ab 3. Greger R: Ion transport mechanisms in thick ascending limb of Henle’s (1:300) and FITC-labeled secondary Ab (1:200). After washing with loop of mammalian nephron. Physiol Rev 65: 760–797, 1985 PBS, slides were mounted with Mowiol (Calbiochem). Epifluores- 4. Hou J, Paul DL, Goodenough DA: Paracellin-1 and the modulation of ion selectivity of tight junctions. JCellSci118: 5109–5118, 2005 cence images were taken with a Nikon 80i Photomicroscope equipped 5. Hou J, Renigunta A, Konrad M, Gomes AS, Schneeberger EE, Paul DL, with a DS-Qi1Mc Digital Camera. All images were converted to TIFF Waldegger S, Goodenough DA: Claudin-16 and claudin-19 interact format and arranged using Photoshop CS4 (Adobe Systems, Inc.). and form a cation-selective tight junction complex. J Clin Invest 118: 619–628, 2008 6.HouJ,ShanQ,WangT,GomesAS,YanQ,PaulDL,BleichM, Antagomir Treatments Goodenough DA: Transgenic RNAi depletion of claudin-16 and the Antagomirs for miR-9, miR-374, and scrambled control miR were syn- renal handling of magnesium. JBiolChem282: 17114–17122, 2007 – thesized by Exiqon using LNAs using the following sequences: anti miR-9: 7. Hou J, Renigunta A, Gomes AS, Hou M, Paul DL, Waldegger S, A*T*+A*C*A*+G*C*T*+A*G*A*+T*A*A*+C*C*A*+A*A*G and anti– Goodenough DA: Claudin-16 and claudin-19 interaction is required for miR-374: A*C*+T*T*A*+G*C*A*+G*G*T*+T*G*T*+A*T*T*+A*T*A their assembly into tight junctions and for renal reabsorption of mag- – (DNA base: G, A, T, C; LNA base: +G, +A, +T, +C; phosphorothioated nesium. Proc Natl Acad Sci U S A 106: 15350 15355, 2009 8. Thorleifsson G, Holm H, Edvardsson V, Walters GB, Styrkarsdottir U, DNA base: G*, A*, T*, C*). In vitro, 30 pmol each antagomir was Gudbjartsson DF, Sulem P, Halldorsson BV, de Vegt F, d’Ancona FC, transfected to the primary cultures of TALH in 12-well culture dishes den Heijer M, Franzson L, Christiansen C, Alexandersen P, Rafnar T, using Lipofectamine 2000 (Invitrogen). In vivo, each antagomir was Kristjansson K, Sigurdsson G, Kiemeney LA, Bodvarsson M, Indridason mixed with the in vivo jetPEI Delivery Reagent (VWR) according to the OS, Palsson R, Kong A, Thorsteinsdottir U, Stefansson K: Sequence manufacturer’s guidelines followed by ip injection to animals at a dose variants in the CLDN14 gene associate with kidney stones and bone 2 mineral density. Nat Genet 41: 926–930, 2009 of 2.5 mg/kg body wt 1. 9. Gong Y, Renigunta V, Himmerkus N, Zhang J, Renigunta A, Bleich M, Hou J: Claudin-14 regulates renal Ca++ transport in response to CaSR signal- Statistical Analyses ling via a novel microRNA pathway. EMBO J 31: 1999–2012, 2012 The significance of differences between groups was tested by ANOVA 10. Gong Y, Hou J: Claudin-14 underlies Ca++-sensing receptor-mediated ++ (Statistica 6.0; Statsoft). When the all-effect F value was significant Ca metabolism via NFAT-microRNA-based mechanisms. JAmSoc – (P,0.05), post hoc analysis of differences between individual groups Nephrol 25: 745 760, 2014 – 11. Hou J, Rajagopal M, Yu AS: Claudins and the kidney. Annu Rev Physiol was made with the Newman Keuls test. Values were expressed as 75: 479–501, 2013 means6SEMs unless otherwise stated. 12. Dimke H, Hoenderop JG, Bindels RJ: Molecular basis of epithelial Ca2+ and Mg2+ transport: Insights from the TRP channel family. JPhysiol 589: 1535–1542, 2011 13. Brown EM, Gamba G, Riccardi D, Lombardi M, Butters R, Kifor O, Sun A, ACKNOWLEDGMENTS Hediger MA, Lytton J, Hebert SC: Cloning and characterization of an extracellular Ca(2+)-sensing receptor from bovine parathyroid. Nature – We thank Drs. Tamar Ben-Yosef (Technion Israel Institute of Tech- 366: 575 580, 1993 14. Pollak MR, Brown EM, Chou YH, Hebert SC, Marx SJ, Steinmann B, Levi nology), Feng Chen (Washington University), and Joan Brugge T, Seidman CE, Seidman JG: Mutations in the human Ca(2+)-sensing (Harvard Medical School) for kindly sharing reagents and cell and receptor gene cause familial hypocalciuric hypercalcemia and neonatal animal models. We thank Daniel Martin (Washington University severe hyperparathyroidism. Cell 75: 1297–1303, 1993 O’Brien Center) and Dr. Tong Wang (Yale University O’Brien Center) 15. Pollak MR, Brown EM, Estep HL, McLaine PN, Kifor O, Park J, Hebert for assistance in plasma and urine electrolyte analyses. SC, Seidman CE, Seidman JG: Autosomal dominant hypocalcaemia caused by a Ca(2+)-sensing receptor gene mutation. Nat Genet 8: 303– This work was supported by National Institutes of Health Grants 307, 1994 R01-DK084059 and P30-DK079333 and American Heart Association 16. Pearce SH, Williamson C, Kifor O, Bai M, Coulthard MG, Davies M, Grant 0930050N. Lewis-Barned N, McCredie D, Powell H, Kendall-Taylor P, Brown EM,

674 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 663–676, 2015 www.jasn.org BASIC RESEARCH

Thakker RV: A familial syndrome of hypocalcemia with hypercalciuria 34. Elmén J, Lindow M, Silahtaroglu A, Bak M, Christensen M, Lind- due to mutations in the calcium-sensing receptor. N Engl J Med 335: Thomsen A, Hedtjärn M, Hansen JB, Hansen HF, Straarup EM, 1115–1122, 1996 McCullagh K, Kearney P, Kauppinen S: Antagonism of microRNA-122 17. Vargas-Poussou R, Huang C, Hulin P, Houillier P, Jeunemaître X, in mice by systemically administered LNA-antimiR leads to up-regulation Paillard M, Planelles G, Déchaux M, Miller RT, Antignac C: Functional of a large set of predicted target mRNAs in the liver. Nucleic Acids Res 36: characterization of a calcium-sensing receptor mutation in severe au- 1153–1162, 2008 tosomal dominant hypocalcemia with a Bartter-like syndrome. JAm 35. Castoldi M, Vujic Spasic M, Altamura S, Elmén J, Lindow M, Kiss J, Soc Nephrol 13: 2259–2266, 2002 Stolte J, Sparla R, D’Alessandro LA, Klingmüller U, Fleming RE, 18. Kinoshita Y, Hori M, Taguchi M, Watanabe S, Fukumoto S: Functional Longerich T, Gröne HJ, Benes V, Kauppinen S, Hentze MW, activities of mutant calcium-sensing receptors determine clinical pre- Muckenthaler MU: The liver-specific microRNA miR-122 controls sys- sentations in patients with autosomal dominant hypocalcemia. JClin temic iron homeostasis in mice. JClinInvest121: 1386–1396, 2011 Endocrinol Metab 99: E363–E368, 2014 36. Kato M, Putta S, Wang M, Yuan H, Lanting L, Nair I, Gunn A, Nakagawa 19. Johnstone RW: Histone-deacetylase inhibitors: Novel drugs for the Y, Shimano H, Todorov I, Rossi JJ, Natarajan R: TGF-beta activates Akt treatment of cancer. Nat Rev Drug Discov 1: 287–299, 2002 kinase through a microRNA-dependent amplifying circuit targeting 20. Helin K, Dhanak D: Chromatin proteins and modifications as drug tar- PTEN. Nat Cell Biol 11: 881–889, 2009 gets. Nature 502: 480–488, 2013 37. Gebeshuber CA, Kornauth C, Dong L, Sierig R, Seibler J, Reiss M, 21. Pipalia NH, Cosner CC, Huang A, Chatterjee A, Bourbon P, Farley N, Tauber S, Bilban M, Wang S, Kain R, Böhmig GA, Moeller MJ, Gröne Helquist P, Wiest O, Maxfield FR: Histone deacetylase inhibitor treat- HJ, Englert C, Martinez J, Kerjaschki D: Focal segmental glomerulo- ment dramatically reduces cholesterol accumulation in Niemann-Pick sclerosis is induced by microRNA-193a and its downregulation of WT1. type C1 mutant human fibroblasts. Proc Natl Acad Sci U S A 108: 5620– Nat Med 19: 481–487, 2013 5625, 2011 38. Ma L, Young J, Prabhala H, Pan E, Mestdagh P, Muth D, Teruya- 22. Archin NM, Liberty AL, Kashuba AD, Choudhary SK, Kuruc JD, Crooks Feldstein J, Reinhardt F, Onder TT, Valastyan S, Westermann F, AM, Parker DC, Anderson EM, Kearney MF, Strain MC, Richman DD, Speleman F, Vandesompele J, Weinberg RA: miR-9, a MYC/MYCN- Hudgens MG, Bosch RJ, Coffin JM, Eron JJ, Hazuda DJ, Margolis DM: activated microRNA, regulates E-cadherin and cancer metastasis. Nat Administration of vorinostat disrupts HIV-1 latency in patients on anti- Cell Biol 12: 247–256, 2010 retroviral therapy. Nature 487: 482–485, 2012 39. Nijenhuis T, Vallon V, van der Kemp AW, Loffing J, Hoenderop JG, 23. Haberland M, Montgomery RL, Olson EN: The many roles of histone Bindels RJ: Enhanced passive Ca2+ reabsorption and reduced Mg2+ deacetylases in development and physiology: Implications for disease channel abundance explains thiazide-induced hypocalciuria and hy- and therapy. Nat Rev Genet 10: 32–42, 2009 pomagnesemia. JClinInvest115: 1651–1658, 2005 24. Coward WR, Watts K, Feghali-Bostwick CA, Jenkins G, Pang L: Re- 40. Wilcox ER, Burton QL, Naz S, Riazuddin S, Smith TN, Ploplis B, pression of IP-10 by interactions between histone deacetylation and Belyantseva I, Ben-Yosef T, Liburd NA, Morell RJ, Kachar B, Wu DK, hypermethylation in idiopathic pulmonary fibrosis. Mol Cell Biol 30: Griffith AJ, Riazuddin S, Friedman TB: Mutations in the gene encoding 2874–2886, 2010 tight junction claudin-14 cause autosomal recessive deafness DFNB29. 25. Lee MG, Wynder C, Schmidt DM, McCafferty DG, Shiekhattar R: His- Cell 104: 165–172, 2001 tone H3 lysine 4 demethylation is a target of nonselective anti- 41. Van Lint C, Emiliani S, Verdin E: The expression of a small fraction of depressive medications. Chem Biol 13: 563–567, 2006 cellular genes is changed in response to histone hyperacetylation. 26. Marks PA, Breslow R: Dimethyl sulfoxide to vorinostat: Development of Gene Expr 5: 245–253, 1996 this histone deacetylase inhibitor as an anticancer drug. Nat Biotechnol 42. Butler LM, Agus DB, Scher HI, Higgins B, Rose A, Cordon-Cardo C, 25: 84–90, 2007 Thaler HT, Rifkind RA, Marks PA, Richon VM: Suberoylanilide hy- 27. Ben-Yosef T, Belyantseva IA, Saunders TL, Hughes ED, Kawamoto K, droxamic acid, an inhibitor of histone deacetylase, suppresses the Van Itallie CM, Beyer LA, Halsey K, Gardner DJ, Wilcox ER, Rasmussen growth of prostate cancer cells in vitro and in vivo. Cancer Res 60: J, Anderson JM, Dolan DF, Forge A, Raphael Y, Camper SA, Friedman 5165 –5170, 2000 TB: Claudin 14 knockout mice, a model for autosomal recessive deaf- 43. Cohen LA, Amin S, Marks PA, Rifkind RA, Desai D, Richon VM: Che- ness DFNB29, are deaf due to cochlear hair cell degeneration. Hum moprevention of carcinogen-induced mammary tumorigenesis by the Mol Genet 12: 2049–2061, 2003 hybrid polar cytodifferentiation agent, suberanilohydroxamic acid 28. Ryan MP, Devane J, Ryan MF, Counihan TB: Effects of diuretics on the (SAHA). Anticancer Res 19: 4999–5005, 1999 renal handling of magnesium. Drugs 28[Suppl 1]: 167–181, 1984 44. Kelly WK, Richon VM, O’Connor O, Curley T, MacGregor-Curtelli B, 29. Ryan MP: Diuretics and potassium/magnesium depletion. Directions Tong W, Klang M, Schwartz L, Richardson S, Rosa E, Drobnjak M, for treatment. Am J Med 82: 38–47, 1987 Cordon-Cordo C, Chiao JH, Rifkind R, Marks PA, Scher H: Phase I 30. Breiderhoff T, Himmerkus N, Stuiver M, Mutig K, Will C, Meij IC, clinical trial of histone deacetylase inhibitor: Suberoylanilide hydroxa- Bachmann S, Bleich M, Willnow TE, Müller D: Deletion of claudin-10 mic acid administered intravenously. Clin Cancer Res 9[10 Pt 1]: 3578– (Cldn10) in the thick ascending limb impairs paracellular sodium per- 3588, 2003 meability and leads to hypermagnesemia and nephrocalcinosis. Proc 45. Duvic M, Talpur R, Ni X, Zhang C, Hazarika P, Kelly C, Chiao JH, Reilly Natl Acad Sci U S A 109: 14241–14246, 2012 JF, Ricker JL, Richon VM, Frankel SR: Phase 2 trial of oral vorinostat 31. Naro F, De Arcangelis V, Coletti D, Molinaro M, Zani B, Vassanelli S, (suberoylanilide hydroxamic acid, SAHA) for refractory cutaneous T-cell Reggiani C, Teti A, Adamo S: Increase in cytosolic Ca2+ induced by lymphoma (CTCL). Blood 109: 31–39, 2007 elevation of extracellular Ca2+ in skeletal myogenic cells. Am J Physiol 46. Fraczek J, Vanhaecke T, Rogiers V: Toxicological and metabolic con- Cell Physiol 284: C969–C976, 2003 siderations for histone deacetylase inhibitors. Expert Opin Drug Metab 32. Krützfeldt J, Rajewsky N, Braich R, Rajeev KG, Tuschl T, Manoharan M, Toxicol 9: 441–457, 2013 Stoffel M: Silencing of microRNAs in vivo with ‘antagomirs’. Nature 438: 47. Shah MH, Binkley P, Chan K, Xiao J, Arbogast D, Collamore M, Farra Y, 685–689, 2005 Young D, Grever M: Cardiotoxicity of histone deacetylase inhibitor 33. Obad S, dos Santos CO, Petri A, Heidenblad M, Broom O, Ruse C, Fu C, depsipeptide in patients with metastatic neuroendocrine tumors. Clin Lindow M, Stenvang J, Straarup EM, Hansen HF, Koch T, Pappin D, Cancer Res 12: 3997–4003, 2006 Hannon GJ, Kauppinen S: Silencing of microRNA families by seed- 48. Vansteenkiste J, Van Cutsem E, Dumez H, Chen C, Ricker JL, Randolph targeting tiny LNAs. Nat Genet 43: 371–378, 2011 SS, Schöffski P: Early phase II trial of oral vorinostat in relapsed or

J Am Soc Nephrol 26: 663–676, 2015 Epigenetic Regulation of Calcium Metabolism 675 BASIC RESEARCH www.jasn.org

refractory breast, colorectal, or non-small cell lung cancer. Invest New for the pathophysiology of FHHNC patients. Am J Physiol Renal Physiol Drugs 26: 483–488, 2008 295: F1641–F1647, 2008 49. Riccardi D, Brown EM: Physiology and pathophysiology of the calcium- 52. Greger R: Cation selectivity of the isolated perfused cortical thick as- sensing receptor in the kidney. Am J Physiol Renal Physiol 298: F485– cending limb of Henle’s loop of rabbit kidney. Pﬂugers Arch 390: 30– F499, 2010 37, 1981 50. Bhakat KK, Izumi T, Yang SH, Hazra TK, Mitra S: Role of acetylated human AP-endonuclease (APE1/Ref-1) in regulation of the parathyroid hormone gene. EMBO J 22: 6299–6309, 2003 51. Himmerkus N, Shan Q, Goerke B, Hou J, Goodenough DA, Bleich M: This article contains supplemental material online at http://jasn.asnjournals. Salt and acid-base metabolism in claudin-16 knockdown mice: Impact org/lookup/suppl/doi:10.1681/ASN.2014020129/-/DCSupplemental.

676 Journal of the American Society of Nephrology J Am Soc Nephrol 26: 663–676, 2015