Common noncoding UMOD gene variants induce salt-sensitive hypertension and kidney damage by increasing uromodulin expression

Abstract

Hypertension and chronic kidney disease (CKD) are complex traits representing major global health problems. Multiple genome-wide association studies have identified common variants in the promoter of the UMOD gene, which encodes uromodulin, the major protein secreted in normal urine, that cause independent susceptibility to CKD and hypertension. Despite compelling genetic evidence for the association between UMOD risk variants and disease susceptibility in the general population, the underlying biological mechanism is not understood. Here, we demonstrate that UMOD risk variants increased UMOD expression in vitro and in vivo. Uromodulin overexpression in transgenic mice led to salt-sensitive hypertension and to the presence of age-dependent renal lesions similar to those observed in elderly individuals homozygous for UMOD promoter risk variants. The link between uromodulin and hypertension is due to activation of the renal sodium cotransporter NKCC2. We demonstrated the relevance of this mechanism in humans by showing that pharmacological inhibition of NKCC2 was more effective in lowering blood pressure in hypertensive patients who are homozygous for UMOD promoter risk variants than in other hypertensive patients. Our findings link genetic susceptibility to hypertension and CKD to the level of uromodulin expression and uromodulin's effect on salt reabsorption in the kidney. These findings point to uromodulin as a therapeutic target for lowering blood pressure and preserving renal function.

Figure 1. Biological effect of UMOD SNP variants

(a) Schematic representation of human UMOD gene structure. The position of top SNPs in the UMOD promoter associated with hypertension and CKD is shown. All variants are within the same linkage disequilibrium block, as shown by the LD plot (r² values, data from HapMap CEU, release #28). Genotyped SNPs are shown in red. (b) UMOD expression levels (Real-Time qRT-PCR) on human nephrectomy samples homozygous for either the protective (n = 17) or risk (n = 27) UMOD variants. We normalized UMOD expression to NKCC2 to account for TAL content in the different renal samples. Distribution and mean values of expression levels are shown. ***P < 0.001 (Mann-Whitney test). (c) Urinary uromodulin concentrations in individuals of the SKIPOGH cohort by genotype at rs4293393 (n = 18 CC, 214 CT, 532 TT). Bars indicate average ± s.e.m. The P value reflects significant association of rs4293393 genotype with square-root-transformed daytime urinary uromodulin-to-creatinine ratio in a mixed linear model. (d) Partial alignment of human (Hs), gorilla (Gg), gibbon (Nl), macaque (Mm), cow (Bt) and dog (Cf) UMOD promoter sequences. The intensity of the blue color shading corresponds to nucleotide conservation. The rs4293393 SNP is predicted to be part of a glucocorticoid response element that is lost in the presence of the protective (C) allele. (e) Quantitative analysis of the relative effect of protective (C) and risk (T) alleles of SNP rs4293393 on the transcriptional activity of UMOD promoter in mTAL, highly differentiated mouse primary TAL cells retaining uromodulin expression, immortalized MKTAL (mouse kidney TAL) and HEK293 (human embryonic kidney) cells. A UMOD promoter fragment (3.7 kb) was cloned upstream of LUC (firefly luciferase) gene. We assessed the effect of variants at SNP rs4293393 on the risk haplotype by measuring luciferase activity. Graphs represent 4 independent experiments. Bars indicate average ± s.e.m. **P < 0.01; ***P < 0.001 (ANOVA followed by Bonferroni’s test).

Figure 2. Uromodulin overexpression leads to hypertension and renal damage

(a) Box and whiskers plot showing the systolic BP at baseline in control and Umod transgenic mice at different age (n = 6–10 per group). Bars represent min and max values. (b) Average value and distribution of heart weight in 16 month-old control and Tg^Umodwt/wt mice (left panel) and heart histology showing left ventricular hypertrophy in Tg^Umodwt/wt mice (right panel) (Hematoxylin-Eosin; 1×). (c) Average systolic BP in 14 month-old control and Tg^Umodwt/wt mice on standard (1% NaCl) or low-sodium (0.01% NaCl) diet (n = 7 per group). Error bars indicate ± s.e.m. (d) Representative renal histological images of 16 month-old control and Tg^Umodwt/wt mice (n = 5). Control tissue displays normal features. Kidneys from Tg^Umodwt/wt mice show numerous dilated tubules (upper panel), mostly filled by casts (upper and lower panel). (e) Representative renal tissue from elderly subjects homozygous for the protective variants of rs4293393 and rs12917707 (left panels) showing normal interstitial compartment (upper panel) and mild focal tubular damage with increased thickness of the tubular basement membrane (lower panel) (n = 9). Renal tissue from subjects homozygous for the risk variants (right panels) displays dilated tubules, with detachment of the tubular epithelium (upper panel) and presence of tubular casts (lower panel) (n = 15) (PAS, bar = 100 μm). The graphs below d and e panels show average quantification of the histological analysis (only classes reaching statistical significance are shown). Error bars indicate ± s.d. (f) Tubular casts in renal tissue from a Tg^Umodwt/wt mouse are present in TALs (uromodulin-positive) or more distal tubules (uromodulin-negative) and are formed mostly by uromodulin (arrowhead) or by PAS-positive uromodulin-negative material (black arrows). Bar = 100 μm. (g) Transcript level (Real-Time qRT-PCR) of renal damage markers and chemokines in kidneys from 16 month-old mice: Lcn2 (Lipocalin 2); Kim-1 (Kidney injury molecule-1); Ccl2 (C-C motif chemokine 2); Ccl5 (C-C motif chemokine 5) (n = 5). Bars indicate average ± s.e.m. (h) Representative immunoblot showing Lcn2 and Kim-1 protein levels in kidney lysates of 16 month-old control and Tg^Umodwt/wt mice (n = 13 per group). Beta actin is shown as a loading control. Bars indicate average ± s.d. *P < 0.05; **P < 0.01; ***P < 0.001 determined by unpaired t test (a (left), b, c, h), ANOVA followed by Bonferroni’s test (a (right), g) or Mann-Whitney test (d, e).

Figure 3. Increased activation of Nkcc2 co-transporter and Spak kinase in Tg^Umodwt/wt mice

(a) Representative immunoblot analysis showing levels of phosphorylated (Thr⁹⁶ and Thr¹⁰¹) and total Nkcc2 in kidney lysates from 16 month-old mice. Densitometric analysis (average ± s.d. of 4 independent experiments) shows increased p-Nkcc2 in Tg^Umodwt/wt mice (n = 10–13 per group). # unspecific signal. The increase of p-Nkcc2 is linearly correlated with Umod gene dosage (P < 0.01, ANOVA post test for linear trend). Calnexin was used as a loading control. (b) Transcript level of Nkcc2 (Real-Time qRT-PCR) on total kidney extracts from 16 month-old mice (n = 5). Bars indicate average ± s.e.m. (c) Change in sodium excretion (left panel) and systolic BP (right panel) in 16 month-old Tg^Umodwt/wt and control mice, 2 hours after treatment with furosemide (n = 8 Ctr, 6 Tg^Umodwt/wt (sodium excretion), 5 per group (systolic BP)). Bars indicate average ± s.e.m. (d) Representative immunoblot analysis showing levels of phosphorylated and total Nkcc2 and uromodulin in HEK293 cells stably expressing Nkcc2 and transfected with either wild type (Umod_wt) or soluble (Umod_sol) uromodulin. Beta actin was used as a loading control. Densitometric analysis (average ± s.d. of 3 independent experiments) is shown. (e) Quantification of Nkcc2 activity as assessed by dpH_i dt⁻¹. Bars indicate average ± s.e.m. (f) Representative immunoblot analysis showing levels of phosphorylated and total Spak and Osr1 in kidney lysates from 16 month-old control and Tg^Umodwt/wt mice (n = 4 per group). Bars indicate average ± s.d. *P < 0.05; **P < 0.01 determined by ANOVA followed by Bonferroni’s test (a, b, d, e) or unpaired t test (c, f).

Figure 4. Increased blood pressure and response to furosemide in hypertensive patients homozygous for UMOD risk variants

(a) Twenty-four hour ambulatory BP monitoring, average values for systolic (SBP) and diastolic (DBP) blood pressure are shown. DBP is significantly higher in hypertensive subjects homozygous for the UMOD risk genotype. (b, c) Response to furosemide treatment in hypertensive patients with different rs4293393 genotype. Changes in sodium excretion (b), and systolic (SBP) and diastolic (DBP) blood pressure (c) 4 hours after treatment are shown. The loop diuretic is more effective in lowering BP in hypertensive patients homozygous for rs4293393 risk variant, with a significant difference in the decrease of DBP and a similar trend for SBP (P = 0.06). This effect is associated with higher natriuretic effect. Bars indicate average ± s.e.m. (a, c) or ± s.d. (b) *P < 0.05 (ANOVA test).