Article Association Between Genotype and Phenotype in Uromodulin

Article

Association between Genotype and Phenotype in Uromodulin-Associated Kidney Disease

Jonathan L. Moskowitz,* Sian E. Piret,† Karl Lhotta,‡§ Thomas M. Kitzler,| Adam P. Tashman,* Erin Velez,* Rajesh V. Thakker,† and Peter Kotanko*

Summary Background and objectives Uromodulin-associated kidney disease (UAKD) is an autosomal dominant disease caused by uromodulin (UMOD) gene mutations. This study explored genotype-phenotype correlations by ex- *Renal Research Institute, New York, amining the relationship between the type of UMOD mutation and the age at onset of ESRD. New York; †Academic Endocrine Unit, Design, setting, participants & measurements Extensive bibliographic research was used to ascertain patient-level Radcliffe Department of Medicine, Oxford data of all patients with UAKD published up to October 2011. Data included sex; ages at onset of hyperuricemia, fi Centre for Diabetes, gout, and ESRD; and UMOD genotype. Kaplan-Meier analysis and Cox proportional hazards models tted with Endocrinology and shared gamma frailty terms to adjust for within-family correlations were used to model time to event. Metabolism, Churchill Hospital, University of Oxford, Oxford, Results Thirty-one peer-reviewed publications reporting on 202 patients from 74 families with 59 different United Kingdom; UMOD mutations were included. Median ages at onset of hyperuricemia, gout, and ESRD were 24, 40, and 56 ‡Department of years, respectively. Men developed gout and ESRD significantly earlier than did women (age at ESRD was 50 Nephrology and years for men and 60 for women; P=0.04, shared frailty model). Median ages at ESRD development were lowest Dialysis and §Vorarlberg Institute with Cys77Tyr (37.5 years) and highest with Gln316Pro (65.5 years) UMOD mutations. Onset of ESRD was fi for Vascular signi cantly earlier with UMOD mutations located within the epidermal growth factor domains 2 and 3 (range, Investigation and 45–52 years; P,0.01 and 0.04, respectively) compared with the cysteine-rich domains (range, 60–65 years; by Treatment, Academic shared frailty model). Teaching Hospital Feldkirch, Feldkirch, | fi Austria; and Division Conclusions The UMOD genotype is related to the clinical phenotype of UAKD. This nding may assist in of Medical Genetics, counseling of patients. McGill University Clin J Am Soc Nephrol 8: 1349–1357, 2013. doi: 10.2215/CJN.11151012 Health Centre, Montreal, Quebec, Canada fi Introduction UMOD is rst synthesized as an 84-kD precursor, Correspondence: Dr. Uromodulin (UMOD) is encoded by the UMOD gene then slowly converted into the mature 95-kD 640– Peter Kotanko, Renal on chromosome 16p12 and is the most abundant pro- amino acid glycoprotein. UMOD mutations present Research Institute, tein found in human urine (1). UMOD is postulated in patients with UAKD cause a delay in the matura- 207 East 94th Street, to affect urine concentration by maintaining tubular Suite 303, New York, tion rate and protein export to the plasma membrane NY 10128. Email: water impermeability and regulating salt transport because of a longer retention time in the endoplasmic [email protected] (2,3). More recent research suggests it can also acti- reticulum (7–12). It is possible that mutations that vate dendritic cells to express proinflammatory cyto- differentially affect the rate of UMOD maturation kines and thereby possibly promote inflammation in and expression on the cellular membrane can influ- the renal interstitium (4). Uromodulin-associated kid- ence the clinical course of the disease. Most UMOD ney disease (UAKD) refers to autosomal dominant mutations cluster in exons 4 and 5 and cause changes interstitial kidney disease. Mutations in the UMOD to cysteine residues, which leads to a misfolding of gene were found to be associated with clinical man- the UMOD molecule (1). ifestations, such as hyperuricemia, gout, and CKD (5). UMOD contains three epidermal growth factor Different diagnoses were given to patients with these (EGF) domains; a cysteine-rich region, which conditions, which were also found to have mutations includes a domain of eight cysteines (D8C); and a in the UMOD gene (e.g., medullary cystic kidney dis- zona pellucida domain (12) (Supplemental Figure 1 ease type 2, familial juvenile hyperuricemic nephrop- and Supplemental Table 1). Exons 4 and 5 encode the athy, hereditary nephropathy with hyperuricemia first EGF domain to the end of the cysteine-rich re- and gout, uromodulin storage disease, and glomeru- gion. Mutations in each region have been found with locystic kidney disease). However, it has been pro- varying levels of clinical phenotype severity. Muta- posed that a clinical distinction among these terms tions that reduce the calcium-binding affinity in the is not warranted because they appear to represent a calcium-binding EGF (cbEGF) domain may impair spectrum of one and the same disease (6). global protein structure (7). Bound calcium stabilizes www.cjasn.org Vol 8 August, 2013 Copyright © 2013 by the American Society of Nephrology 1349 1350 Clinical Journal of the American Society of Nephrology

the cbEGF-like domain conformation, restricting interdo- defined as first diagnosis of serum uric acid level .1SD main flexibility (13), and may protect against proteolytic greater than normal values for age and sex (16). Age at degradation (14). gout onset was defined as the patient’sageatthefirst Families with UMOD mutations have been identified in episode of gouty arthritis. the United States, Austria, Spain, France, Portugal, the Czech Republic, the United Kingdom, Belgium, Germany, Statistical Analyses Switzerland, Latvia, Morocco, Japan, Turkey, and South The primary aim of the study was to explore the relation- Korea. However, there has not been a worldwide epide- ship between UMOD mutations and time to ESRD. Second- miologic study on UAKD. This analysis strives to ascertain ary aims were to analyze the relationships between sex and all UAKD cases published through October 2011 and to age at the onset of hyperuricemia and gout. To that end, relate UMOD genotype to clinical presentation. median age at the onset of hyperuricemia, gout, and ESRD were computed. In separate analyses, patients were stratified by sex, presence or absence of hyperuricemia or gout, Materials and Methods UMOD mutation, affected uromodulin domain, and cell Search Strategy surface expression pattern. Extensive electronic and manual bibliographic research The six UMOD mutations with the most affected indi- was performed to ascertain patient-level data of all cases of viduals were examined to determine whether having a dif- UAKD published through October 2011. Meticulous care ferent genotype could affect the outcome. In an additional was exercised to exclude duplicate reports of patients. We analysis, UMOD mutations were aggregated according to reviewed citations in English from PubMed, the D. Samuel their place within one of the five following functional do- Gottesman Library from the Albert Einstein College of mains: CysRich-1, CysRich-2, D8C, EGF2, and EGF3. Only Medicine in New York, and Google Scholar. The search functional domains with mutations affecting more than 14 strategy included keywords and synonyms for the follow- patients were considered. Mutations were further classi- ing terms: familial juvenile hyperuricemic nephropathy, fied into groups in which substitutions affected a cysteine uromodulin, Tamm-Horsfall protein, medullary cystic residue, a nonpolar residue, and a noncysteine polar resi- kidney disease, mutation, uric acid, gout, and hyperurice- due. Cysteine substitutions were then compared with non- mia. Full-text articles were retrieved for further indepen- cysteine substitutions. dent manual evaluation by two authors (J.M. and E.V.). For the modeling process, statistical adjustments for Supplemental searches using the names of authors of members of the same family were made. This was done by relevant articles were performed. multiplicatively adjusting the hazard rate of each family All retrieved published full-text case studies on individ- member by the same factor (the frailty) according to the uals and families with UAKD were included. When the model mentioned below. For example, it may be the case same family was investigated in more than one study, the that all members of family A are frailer than members of most recent study with the most thorough data were family B, adjusting for other covariates. In this case, the included. If the most recent publication had incomplete family A frailty will be greater than the family B frailty. clinical data, the data were combined with a previous study The gamma shared frailty model—which is an extension and adjusted for time. In some instances, authors were of the Cox model—was used to estimate hazard ratios fi personally contacted for data veri cation. In addition, one (17,18). If the frailty effect was not statistically significant hitherto unpublished UAKD family was included. A search at 0.05, it was excluded and a conventional Cox propor- for UMOD mutations in the 1000 Genomes database tional hazard model was used to fittothedata.TheR showed that UAKD-associated mutations were not detec- package survival was used to fit the models and perform fi ted in that population, con rming the rare, pathogenic na- hypothesis testing (19). ture of the UMOD mutations studied here. Allele Additionally, in models with significant frailty effect, the frequency variation across different populations is there- shared frailty model was used to estimate median time to fore an unlikely confounding factor. event and the 95% confidence interval (CI) around the median. The median can be computed for any given frailty, Data Ascertainment and here a value of unity was used (this assumes that the Articles were scanned for information on all patients individual has average frailty). For case frailty = 1, the reported. All clinical data were recorded in a database, formula for the median under the shared frailty model is including pedigree position, sex, race, country of residence, the same as the formula for the median under the Cox type of UMOD mutation, ESRD status, age at hyperurice- model, but the latter must use the parameter estimates from mia, gout, ESRD, death, and age at last follow-up. Not the former. every article had information on every category; reports A formula for computing CIs around the median based that lacked information on the UMOD mutation or ESRD on a shared frailty model is not known, and so a bootstrap status were excluded. UMOD cell surface expression data approach was used. This involved building 500 resamples were taken from respective reports (8–10,12,15) and calcu- drawn with replacement from the original dataset. How- lated as percentage of wild-type expression; a cutoff of ever, care was taken with building these resamples; when a 50% was used to separate UMOD mutations into two subject was queried and added to the database, further groups (high versus low expression). ESRD was defined information about family members was requested to be as the initiation of dialysis therapy, renal transplantation, provided as well. A simple bootstrap that would randomly or death from renal failure in the absence of renal replace- sample patients from the database would lose this family ment therapy initiation. Age at hyperuricemia onset was structure, and so a block bootstrap was performed. In lnJA o eho :14–37 uut 2013 August, 1349–1357, 8: Nephrol Soc Am J Clin

Table 1. Characteristics of patients with UMOD mutations

Effect on Demographics Morbidity (Mean Age at Onset), n (yr) Nucleotide Patients Peptide Domain Exon Reference Change (n) Sequence Country Men/Women (n/n) ESRD Gout Hyperuricemia c.C96G p.Cys32Trp EGF1 4 1 NA NA 1 (NA) 1 (NA) NA 12 c.G149C p.Cys50Ser EGF1 4 2 IT 2/0 1 (29) 0 1 (11) 27 c.T156G p.Cys52Trp EGF1 4 2 NA 0/2 NA 2 (NA) NA 28 c.G172T p.Gly58Cys EGF1 4 2 NA NA 1 (NA) 0 NA 29 c.A176C p.Asp59Ala EGF1 4 1 BE NA 1 (59) 0 NA 16 c.G206A p.Cys69Tyr EGF2 4 1 NA NA 1 (NA) 0 NA 29 c.T229G p.Cys77Gly EGF2 4 2 US 2/0 1 (45) NA 2 (26) 37 rs121917768 c.G307T p.Gly103Cys EGF2 4 8 NA NA NA NA NA 5,6 rs28934584 c.G317A p.Cys106Tyr EGF2 4 1 NA NA 0 0 NA 29 rs115048026 c.T334C p.Cys112Arg EGF3 4 1 BE NA 1 (64) 1 (30) NA 16 c.T376C p.Cys126Arg EGF3 4 2 IT NA 1 (52) 2 (25.5) NA 8,16,31 rs121917769 5 AT 3/2 4 (44.75) NA NA 3 NA 1/2 1 (36) 1 (31) 3 (22) c.A383G p.Asn128Ser EGF3 4 NA ES NA NA NA NA 6,32,33 rs12191770 c.T403A p.Cys135Ser EGF3 4 7 NA 6/1 NA 7 (NA) NA 28 UMOD c.G443A p.Cys148Tyr EGF3 4 16 NA NA NA NA NA 5,6

rs28934582 1351 al. et Moskowitz Association, Phenotype and Genotype c.C708T p.Pro236Leu D8C 4 5 NA 1/4 2 (NA) 3 (15) 3 (NA) 8 c.T444G p.Cys148Trp EGF3 4 3 NA NA 1 (39) 1 (NA) 3 (NA) 7 c.G449C p.Cys150Ser Cys-Rich1 4 16 NA NA 0 (NA) 10 (NA) 16 (NA) 7 c.G509A p.Cys170Tyr Cys-Rich1 4 3 FR NA 0 0 NA 16 c.T520C p.Cys174Arg Cys-Rich1 4 3 US 1/2 2 (60) 2 (41.5) 3 (36) —a c.C553A p.Arg185Ser Cys-Rich1 4 13 BE 6/6b 7 (48.3) 7 (20.6) 9 (NA) 16,34 c.C553G p.Arg185Gly Cys-Rich1 4 1 NA NA 0 1 (NA) 1 (14) 12 c.563_661del p.Glu188-Leu221del D8C 4 3 FR NA 1 (38) 1 (14) NA 16 c.G584T p.Cys195Phe Cys-Rich1 3 NR 2/1 NA 2 (NA) 1 (NA) 28 c.G586T p.Asp196Tyr Cys-Rich1 4 5 AT 3/2 1 (36) 1 (14) 3 (9.3) 1,31 c.G605C p.Trp202Ser D8C 4 4 NA 3/1 NA 4 (NA) NA 28 c.C610G p.Arg204Gly D8C 4 1 MA NA 0 1 (38) NA 16 c.T649C p.Cys217Arg D8C 4 1 NA NA NA 0 0 5,6 rs28934583 c.T649G p.Cys217Gly D8C 4 1 BE NA 1 (36) 1 (21) NA 16 c.G665C p.Arg222Pro D8C 4 7 FR NA 1 (47) 4 (26.5) NA 16 c.C674A p.Thr225Lys D8C 4 3 NA 2/1 2 (31.5) 1 (18) 3 (24.7) 35 32Ciia ora fteAeia oit fNephrology of Society American the of Journal Clinical 1352

Table 1. (Continued)

Effect on Demographics Morbidity (Mean Age at Onset), n (yr) Nucleotide Patients Peptide Domain Exon Reference Change (n) Sequence Country Men/Women (n/n) ESRD Gout Hyperuricemia

c.C674T p.Thr225Met D8C 4 8 FR NA 4 (52) 1 (15) NA 16 1 NA 1/0 1 (38) 1 (25) NA 2 c.T667C p.Cys223Arg D8C 4 1 NA NA 0 (NA) 1 (NA) 1 (45) 12 2 AT 0/2 1 (53) 2 (NA) NA 31 c.G668A p.Cys223Tyr D8C 4 2 NA 0/2 1 (49) 1 (13) 1 (13) 36 c.T686G p.Met229Arg D8C 4 10 NA 5/5 4 (52.5) 7 (20.6) 6 (18.7) 8 c.T688C p.Trp230Arg D8C 4 1 NA NA 0 1 (36) 1 (36) 29 c.C707G p.Pro236Arg D8C 4 NA IT NA NA NA NA 10 c.G743C p.Cys248Ser D8C 4 3 NA NA 1 (NA) 2 (NA) NA 29 c.C744G p.Cys248Trp D8C 4 1 IT 0/1 1 (18) 0 1 (9) 37 c.G764A p.Cys255Tyr D8C 4 14 ES 9/5 4 (59.2) 5 (31.2) 10 (25) 38 rs121917771 c.G808T p.Gly270Cys D8C 4 2 NA 0/2 0 1 (31) 1 (26) 39 c.A821G p.Tyr274Cys D8C 4 1 NA NA 0 0 NA 29 c.T844C p.Cys282Arg D8C 4 1 FR NA 1 (25) 1 (14) NA 16 c.G899A p.Cys300Tyr Cys-Rich2 5 2 NA 1/1 0 1 (19) 2 (16.3) 40 c.T891G p.Cys297Trp Cys-Rich2 4 4 NA 1/3 1 (39) 1 (17) 4 (7.8) 41 c.A920C p.Lys307Thr Cys-Rich2 5 2 PT 1/1 1 (49) 2 (21) 2 (21) 42 c.T943C p.Cys315Arg Cys-Rich2 4 3 NA NA 0 0 3 (NA) 7 rs121917773 c.A947C p.Gln316Pro Cys-Rich2 5 9 ES 6/3 2 (65.5) 3 (41.3) 6 (41.2) 43 1 ES 1/0 0 (NA) 1 (31) 1 (15) 43 c.T1039G p.Cys347Gly ZP 6 11 NA 4/7 NA NA 2 (NA) 9 c.C744G p.Cys248Trp D8C 3 2 TR 1/1 0 (NA) 0 (NA) 1 (10) 37 c.C744G p.Cys248Trp D8C 3 1 DE 1/0 1 (6) 0 1 (3) 37 c.C744G p.Cys248Trp D8C 3 1 CH 0/1 1 (52) 0 1 (46) 37 c.C744G p.Cys248Trp D8C 3 3 NA 2/1 1 (45) 0 3 (24.7) 35,37 c.C1382A p.Ala461Glu ZP 7 5 KP 5/0 0 5 (15.4) 1 (16) 44 c.G1462C p.Gly488Arg ZP 8 1 NA NA 0 1 (NA) 1 (25) 12 NA p.Cys32Tyr EGF1 4 1 NA 0/1 0 1 (26) 1 (26) 8 NA p.Cys77Tyr EGF2 4 11 AT 5/6 7 (36.6) 10 (29.4) 7 (25.4) 30,31 NA p.Asp196Asn Cys-Rich1 4 1 NA NA 0 NA NA 29 1 NA NA 0 0 1 (16) 12 NA p.Cys217Trp D8C 4 1 NA NA 1 (NA) NA NA 12 4 AT 1/3 4 (47) NA NA 31 Clin J Am Soc Nephrol 8: 1349–1357, August, 2013 UMOD Genotype and Phenotype Association, Moskowitz et al. 1353

particular, sampling was done at the family level, and the sample was constructed by adding random families until the size of the resampled dataset matched the size of the original dataset. The rule used was that so long as the

Reference number of patients in the resampled set was less than the number in the original set, another family was added. Next, the median time to event was computed for each resampled

(yr) dataset according to the procedure discussed above, and n 2.5th and 97.5th quantiles of the distribution were determined. When the frailty effect was not signiﬁcant, the median time to event and 95% CI were estimated from Kaplan- Meier curves. rst diagnosis with hyperuricemia, gout, ﬁ

Results We ascertained 202 patients from 74 UAKD families. Overall, 59 different UMOD mutations were found in patients from 25 countries (Table 1). Sixty-eight of these 202 patients (33%) had ESRD. The presence or absence of hyperuricemia was reported in 166 patients, of whom 151 had hyperuricemia (91%). Gout was present in 86 of 179 ) ESRD Gout Hyperuricemia patients with recorded gout status (48%). In the whole co-

n/n hort, the median ages at onset of hyperuricemia, gout, and ESRD were 24.0 (95% CI, 19.3 to 28.7), 40.0 (95% CI, 32.1 to 47.9), and 56.0 (95% CI, 52.0 to 60.0) years, respectively.

Relationship between Sex, Hyperuricemia, Gout, and ESRD Sex was not associated with age at diagnosis of hyperuricemia (Table 2). However, men developed gout at an Demographics Morbidity (Mean Age at Onset), earlier age (P=0.02; hazard ratio, 2.28 by the shared frailty model). Men experienced ESRD at a younger age than did women (50 years for men and 60 years for women; P=0.04 Country Men/Women ( by the shared frailty model). Neither hyperuricemia nor gout was associated with age at ESRD onset (Table 3). ) 18 NA NA NA NA 0 0 (NA) 2 (NA) 0 8 (NA) 1 (33) 7 29 n ( Relationship between UMOD Mutations and ESRD

Patients Thirty-two of the 59 (54%) UMOD mutations affected a cysteine residue. Because of the small number of affected patients in most of the mutations, only the relationship

ects cDNA numbering with 1 = A of the ATG translation initiation codon, followed by its amino acid change. If the cDNA between the six most prevalent UMOD mutations (i.e., ﬂ with the most affected individuals) and ESRD was examined. UMOD mutations were related to age at ESRD onset Domain Exon Table 2. Sex as predictor of hyperuricemia, gout, and ESRD

Hazard Ratio Factor (95% Conﬁdence P Value Interval)

Sex versus Peptide Effect on

Sequence hyperuricemia Male 1.37 (0.78 to 2.44) 0.28 Sex versus gout p.Val273Phep.Cys300Gly D8C Cys-Rich2 5 4 4 3 ES NAMale 3/0 1/32.28 (1.14 2 2 (62.5) (58.5) to 4.57) 3 2 (30) (NA) 2 (15) 3 (30)0.02 43 8 Sex versus ESRD Male 2.09 (1.05 to 4.16) 0.04

Cox proportional hazards analysis of the association of sex on

Change time to hyperuricemia, gout, and ESRD ﬁtted with a shared Nucleotide Gender was not reported in all family members. Meisels I, Charytan C. Personal communication. Summary of all uromodulin mutations found with their corresponding family, functional domain, exon, demographics, morbidity, and mean age at a b and renal failure. For eachsequence mutation, of a nucleotide reported mutation numbering was re basedBE, on Belgium; another US, template, United it was States; converted AT, toTurkey; Austria; the DE, Cys-rich, 1 = cysteine-rich German; A domain; of CH, ES, ATG Switzerland; notation Spain; in KP, D8C: the domain Korea. table. of EGF, eight epidermal cysteines; growth factor FR, like France; domain; MA, NA, Morocco; not PT, available; Portugal; IT, Italy; ZP, zona pellucida domain; TR, NA rs121917774 NA rs121917772 NA p.Cys317Tyr Cys-Rich2 5frailty model. 5 NA 2/3 2 (56) 3 (22.7) 4 (12.5) 8 Table 1. (Continued) 1354 Clinical Journal of the American Society of Nephrology

maturation/export rate of the UMOD correlates with clin- Table 3. Association of hyperuricemia and gout with ESRD ical phenotypes. When we compared UMOD mutations associated with low cell surface expression (,50%) to those Hazard Ratio with high expression (.50%), we found no association with Endpoint (95% Confidence P Value Interval) the age at onset of ESRD. Median age at ESRD was 59 years for patients with mutations with low in vitro UMOD ex- Hyperuricemia pression and 56 years for those with high expression. versus ESRD The type of amino acid (polar, nonpolar, and cysteine) Hyperuricemia 2.06 (0.30 to 14.01) 0.46 affected by a UMOD mutation was also analyzed. The type Gout versus ESRD of amino acid affected was not a significant factor in de- Gout 1.45 (0.55 to 3.82) 0.45 termining onset of ESRD by the shared frailty model (Sup- plemental Table 3, A and B). Cox proportional hazards analysis with shared frailty term: analysis of the association of having hyperuricemia and gout with time to ESRD. Within-family correlations are significant for both tests (P,0.01). Hyperuricemia and gout are not sta- Discussion tistically associated with time to ESRD. Our analysis indicates that the type of UMOD mutation affects the clinical phenotype of patients with UAKD, in particular the age at onset of ESRD. Patients with the p. Cys77Tyr UMOD mutation experienced ESRD almost 30 , (P 0.001; log-rank test) (Table 4 and Figure 1). Patients years earlier than patients with p.Gln316Pro. To the best of with the p.Cys77Tyr mutation had the youngest mean our knowledge, this report compiles results from the most manifestation age of ESRD (37.4 years), and patients comprehensive and geographically diverse database of pa- with the p.Gln316Pro mutation were at the other extreme tients with UAKD. Recognizing that within-family corre- (65.5 years) (Table 5). lations may bias any analysis of genotype-phenotype relationships, we used Cox proportional hazards models fi Relationship between Outcome and Uromodulin Domain, tted with shared gamma frailty terms to adjust for Surface Expression, and Altered Amino Acid of UMOD within-family correlations whenever analyses encom- Mutation passed multiple families lumped together in a single cate- The effects of mutations in different locations within the gory. Our results are similar to those of a recent study that UMOD protein domains were investigated (Supplemental Ta- found 37 UMOD mutations in 109 patients from 45 fami- ble1).OnsetofESRDwassignificantly affected by the amino lies, in which median time to ESRD was determined to be acid sequence alterations in certain UMOD domains (Table 6). 54.0 years (20). This study by Bollée et al. was actually Mutations in the first cysteine-rich region (Cys-Rich 1) had published after we closed our database; therefore, patients the longest median time to ESRD at 65.0 years, whereas pa- from this study were not included in our analysis. fi tients with mutations affecting the second calcium- Sex was signi cantly associated with onset of ESRD, binding EGF domain (cbEGF2) experienced ESRD onset at with men developing ESRD at a younger age. Although the youngest age of 45.0 years (Table 7). Mutations affecting men and women developed hyperuricemia at approxi- EGF domains 2 or 3 were associated with earlier development mately the same age, men experienced gout at a younger of ESRD compared with mutations affecting the other UMOD age. Gout is generally rare in young women (21). However, domains (Table 7). women with UMOD mutations still often have gout even Previous reports on UMOD in vitro cell surface expression though they develop it later than men. Gout is more prev- with regard to UMOD mutation allowed us to investigate alent in men in the general population, possibly because of this aspect in our study. In vitro cell surface expression com- lower fractional uric acid excretion (FEUA) in men. In pa- pared with wild type was assessed for 15 UMOD muta- tients with UAKD, FEUA is reduced to approximately tions (Supplemental Table 2) to determine whether the equal levels in men and women (9), suggesting that hyperuricemia due to low FEUA is not the only reason for the earlier occurrence of gout in men than women. The significant association of different UMOD domain UMOD Table 4. mutation as predictor of ESRD mutations with age at onset of ESRD indicates that the location of the mutation can affect the progression of renal Hazard Patients/ Ratio (95% P disease. The median age at ESRD development was lowest Mutation Families Confidence Value in patients with mutations in the EGF2 and EGF3 domains. (n/n) Interval) Time to renal failure was longest in patients with UMOD mutations located in the domain of eight cysteines (D8C) p.Cys255Tyr 14/1 0.97 (0.24 to 3.91) 0.97 and the cysteine-rich regions (Cys-Rich 1 and Cys-Rich 2). p.Cys317Tyr 13/1 0.77 (0.14 to 4.29) 0.76 Our analysis indicates that individuals with UAKD p.Cys77Tyr 11/1 9.03 (2.05 to 39.83) 0.004 often develop hyperuricemia in their early twenties. Men p.Gln316Pro 10/2 0.27 (0.05 to 1.35) 0.11 develop gout in their early thirties and women, in their p.Met229Arg 10/1 1.33 (0.23 to 7.79) 0.75 early forties. Finally, ESRD ensues in the mid-fifties. Of note, development of hyperuricemia and gout does not Reference was p.Arg185Ser. Cox proportional hazards analysis of the effect of six different UMOD mutations on time to ESRD. predict the age at which a patient will reach ESRD. It is clinically important to appreciate that ESRD can still Clin J Am Soc Nephrol 8: 1349–1357, August, 2013 UMOD Genotype and Phenotype Association, Moskowitz et al. 1355

Figure 1. | Kaplan-Meier survival curves stratified by the six UMOD mutations with the greatest number of affected individuals comparing the time to ESRD. Differences in time to ESRD were significant. The difference between median time to ESRD in p.Cys77Tyr (C77Y) and p.Met229Arg (M229R) mutations was almost 30 years. MMA = median manifestation age; N = number of individuals with known symptom status. develop without the previous occurrence of hyperuricemia CKD and are associated with a common polymorphism or gout. in the UMOD region (22). The adjusted odds ratio of The level of in vitro cell surface expression of mutated CKD was found to be 1.72 per 1-SD higher uromodulin UMOD was not associated with the median age at ESRD. level. Intracellular accumulation of UMOD in cell models This is contrary to what we expected because cellular re- has been associated with cell apoptosis (23). Evidence tention and reduced secretion of UMOD is thought to exert for a proteotoxic effect of mutant UMOD expression harmful effects. The findings open the intriguing possibil- has not been observed with use of different kidney cell ity that greater secretion of mutated UMOD may actually lines under basal or stressed conditions (15). Further- be more damaging than retention of the mutated protein. more, the lack of correlation between presence of hyperuri- This notion is in line with the observation that elevated cemia or gout and time to ESRD suggests that different urinary uromodulin concentrations precede the onset of mechanisms may be responsible for the development of ESRD and development of hyperuricemia or gout in these patients. Table 5. UMOD mutation and time to ESRD (Kaplan-Meier In one study, not only were the p.Cys77Tyr and p. analysis) Asn128Ser mutants secreted less than the wild type in transfected LLC-PK1 cells, but that secretion was also Estimated Mean reduced at the apical membrane and increased at the Age at ESRD Patients/ basolateral membrane (15). UMOD secreted into the renal Mutation Manifestation Families (n/n) fi interstitium may stimulate immune cells, leading to in- (95% Con dence flammation and progression of tubulo-interstitial damage Interval) (yr) (15). This finding thereby suggests that the direction of Arg185Ser 13/1 49.0 (37.2 to 60.9) UMOD secretion and the degree of UMOD retention/ Cys255Tyr 14/1 54.1 (47.3 to 61) secretion may be relevant in the development of clinical Cys317Tyr 13/1 56.7 (53.6 to 59.7) phenotypes. Cys77Tyr 11/1 37.4 (32.7 to 42) It is unclear whether allopurinol therapy can slow Gln316Pro 10/2 65.5 (62.6 to 68.4) progression of renal disease. Some reports indicate that Met229Arg 10/1 53.7 (48.4 to 58.9) allopurinol can slow down progression of kidney disease (24,25), whereas others state that it will treat only hyper- Results of the Kaplan-Meier analysis of the six most prevalent uricemia and gout, without any effect on renal diagnosis UMOD mutations. Mean age at manifestation of ESRD was (26). Because drug therapy has been reported in only a few calculated. The SEM and 95% confidence interval were also patients, our data cannot answer the central question of calculated for each mutation. Time to ESRD was significantly different between the mutations (P,0.001). whether allopurinol helps prevent or delay ESRD. Further studies on that subject are needed. 1356 Clinical Journal of the American Society of Nephrology

with high or low in vitro UMOD cell membrane expres- Table 6. Affected UMOD domain as predictor of ESRD sions were not associated with time to ESRD. This retro- spective study could not explore the relevance of factors Patients/ Hazard Ratio P related to environment and lifestyle on clinical outcomes; Domain Families (95% Conﬁdence Value (n/n) Interval) this topic will require prospective studies.

Cys-Rich2 39/10 1.62 (0.42 to 6.32) 0.49 Acknowledgments D8C 75/24 4.37 (1.38 to 13.87) 0.01 We would like to thank Dr. Martin Farrall from The Wellcome EGF2 23/4 7.74 (1.95 to 30.72) 0.004 Trust Centre for Human Genetics, University of Oxford, Head- EGF3 42/5 4.44 (1.11 to 17.81) 0.04 ington, Oxford, United Kingdom, for his help and advice on pedigree-based association studies. Furthermore, we would like to Reference group was Cys-Rich1. Cox proportional hazards thank Drs. Ira Meisels and Chaim Charytan from New York, New analysis of the effect of UMOD domain on time to ESRD ﬁtted York, for bringing a UAKD family with a hitherto unknown UMOD with a shared frailty model. Time to ESRD is signiﬁcantly as- mutation to our attention and for providing clinical details. sociated with UMOD domains. Patients with mutations in the S.E.P. and R.V.T are grateful to Kidney Research UK (KRUK); D8C, EGF2, and EGF3 domains are at a much greater risk for Wellcome Trust (UK); Medical Research Council (MRC) UK; and developing ESRD at an earlier age compared with those who European Union Framework 6 (EU-FP6) - EUREGENE, for funding have mutations in the Cys-Rich1 domain (hazard ratios, 4.37, support. 7.74, and 4.44, respectively). Cys-Rich2, cysteine-rich domain; D8C, domain of eight cysteines; EGF, epidermal growth factor. Disclosures None.

References Table 7. Affected UMOD domain and time to ESRD (shared 1. Lhotta K, Gehringer A, Jennings P, Kronenberg F, Brezinka C, frailty model) Andersone I, Strazdins V: Familial juvenile hyperuricemic nephropathy: Report on a new mutation and a pregnancy. Clin Nephrol Median Age at ESRD 71: 80–83, 2009 UMOD Patients/ 2. Merieau E, Al Najjar A, Halimi JM, Sacque´peé M, Nivet H, Manifestation Functional Families Lebranchu Y, Bu¨chler M: Disappearance of tophi in familial ju- (95% Confidence Domain (n/n) venile hyperuricemic nephropathy after kidney transplantation. Interval) (yr) Am J Transplant 7: 2634–2636, 2007 3. Bleyer AJ, Zivna´ M, Kmoch S: Uromodulin-associated kidney CysRich1 46/8 65 (47 to NA) disease. Nephron Clin Pract 118: c31–c36, 2011 CysRich2 39/10 60 (49 to NA) 4. Darisipudi MN, Thomasova D, Mulay SR, Brech D, Noessner E, D8C 75/24 52 (36 to 63) Liapis H, Anders HJ: Uromodulin triggers IL-1b-dependent in- EGF2 23/4 45 (35 to NA) nate immunity via the NLRP3 inflammasome. J Am Soc Nephrol EGF3 42/5 52 (38 to NA) 23: 1783–1789, 2012 5. Hart TC, Gorry MC, Hart PS, Woodard AS, Shihabi Z, Sandhu J, Shirts B, Xu L, Zhu H, Barmada MM, Bleyer AJ: Mutations of the The median manifestation age of ESRD was calculated ac- UMOD gene are responsible for medullary cystic kidney disease cording to the shared frailty model. The 95% confidence interval 2 and familial juvenile hyperuricaemic nephropathy. JMed was also calculated for each functional domain whenever pos- Genet 39: 882–892, 2002 sible. Cys-Rich2, cysteine-rich domain; NA, not available; D8C, 6. Scolari F, Caridi G, Rampoldi L, Tardanico R, Izzi C, Pirulli D, domain of eight cysteines; EGF, epidermal growth factor. Amoroso A, Casari G, Ghiggeri GM: Uromodulin storage dis- eases: Clinical aspects and mechanisms. Am J Kidney Dis 44: 987–999, 2004 7. Rampoldi L, Caridi G, Santon D, Boaretto F, Bernascone I, Obvious limitations of our study are underreporting of Lamorte G, Tardanico R, Dagnino M, Colussi G, Scolari F, cases and lack of complete data because many reports did Ghiggeri GM, Amoroso A, Casari G: Allelism of MCKD, FJHN not include all the key clinical data from each patient. and GCKD caused by impairment of uromodulin export dy- Hum Mol Genet Complete data could not be included for every patient namics. 12: 3369–3384, 2003 8. Vylet’al P, Kublova´ M, Kalbaćova´ M, Hodanova´ K, Baresova´ V, because many patients died before any genetic analysis Stibůrkova´ B, Sikora J, Hůlkova´ H, Zivny´J,MajewskiJ,SimmondsA, could be done; for others, the information was difficult to Fryns JP, Venkat-Raman G, Elleder M, Kmoch S: Alterations of uro- obtain from the relatives. Because the main research modulin biology: A common denominator of the genetically het- Kidney Int question was the relationship between UMOD genotype erogeneous FJHN/MCKD syndrome. 70: 1155–1169, 2006 9. Tinschert S, Ruf N, Bernascone I, Sacherer K, Lamorte G, and ESRD, we limited our analysis to patients with deter- Neumayer HH, Nu¨rnberg P, Luft FC, Rampoldi L: Functional mined or inferred UMOD genotype, known ESRD status, consequences of a novel uromodulin mutation in a family with and age at ESRD onset. Despite these limitations, our da- familial juvenile hyperuricaemic nephropathy. Nephrol Dial tabase, which is available to researchers and clinicians Transplant 19: 3150–3154, 2004 upon request, may prove to be a valuable resource for 10. Bernascone I, Vavassori S, Di Pentima A, Santambrogio S, Lamorte G, Amoroso A, Scolari F, Ghiggeri GM, Casari G, future research in UMOD-related disorders. Polishchuk R, Rampoldi L: Defective intracellular trafficking of In conclusion, certain UMOD genotypes are associated uromodulin mutant isoforms. Traffic 7: 1567–1579, 2006 with an earlier onset of ESRD. UMOD mutations affecting 11. Bachmann S, Mutig K, Bates J, Welker P,Geist B, Gross V,Luft FC, different protein domains vary in their clinical correlates, Alenina N, Bader M, Thiele BJ, Prasadan K, Raffi HS, Kumar S: Renal effects of Tamm-Horsfall protein (uromodulin) deficiency with mutations affecting EGF domains being associated in mice. Am J Physiol Renal Physiol 288: F559–F567, 2005 with the most severe clinical phenotypes. Hyperuricemia 12. Williams SE, Reed AA, Galvanovskis J, Antignac C, Goodship T, and gout did not predict the advent of ESRD. Mutations Karet FE, Kotanko P, Lhotta K, Morinie`re V, Williams P, Wong W, Clin J Am Soc Nephrol 8: 1349–1357, August, 2013 UMOD Genotype and Phenotype Association, Moskowitz et al. 1357

Rorsman P, Thakker RV: Uromodulin mutations causing familial mutations result in decreased ciliary uromodulin expression. juvenile hyperuricaemic nephropathy lead to protein maturation Hum Mol Genet 19: 1985–1997, 2010 defects and retention in the endoplasmic reticulum. Hum Mol 30. Kotanko P, Gebetsroither E, Skrabal F: Familial juvenile hyper- Genet 18: 2963–2974, 2009 uricaemic nephropathy in a Caucasian family associated with 13. Werner JM, Knott V, Handford PA, Campbell ID, Downing AK: inborn malformations. Nephrol Dial Transplant 17: 1333–1335, Backbone dynamics of a cbEGF domain pair in the presence of 2002 calcium. J Mol Biol 296: 1065–1078, 2000 31. Lhotta K, Piret SE, Kramar R, Thakker RV, Sunder-Plassmann G, 14. Reinhardt DP, Ono RN, Sakai LY: Calcium stabilizes fibrillin-1 Kotanko P: Epidemiology of uromodulin-associated kidney dis- against proteolytic degradation. J Biol Chem 272: 1231–1236, ease: Results from a nation-wide survey. Nephron Extra 2: 147– 1997 158, 2012 15. Jennings P, Aydin S, Kotanko P, Lechner J, Lhotta K, Williams S, 32. Puig JG, Prior C, Martıńez-Ara J, Torres RJ: Familial nephropathy Thakker RV, Pfaller W: Membrane targeting and secretion of associated with hyperuricemia in Spain: our experience with 3 mutant uromodulin in familial juvenile hyperuricemic ne- families harbouring a UMOD mutation. Nucleosides Nucleo- phropathy. J Am Soc Nephrol 18: 264–273, 2007 tides Nucleic Acids 25: 1295–1300, 2006 16. Dahan K, Devuyst O, Smaers M, Vertommen D, Loute G, Poux 33. Turner JJ, Stacey JM, Harding B, Kotanko P, Lhotta K, Puig JG, JM, Viron B, Jacquot C, Gagnadoux MF, Chauveau D, Bu¨chler M, Roberts I, Torres RJ, Thakker RV: UROMODULIN mutations Cochat P, Cosyns JP, Mougenot B, Rider MH, Antignac C, cause familial juvenile hyperuricemic nephropathy. J Clin En- Verellen-Dumoulin C, Pirson Y: A cluster of mutations in the docrinol Metab 88: 1398–1401, 2003 UMOD gene causes familial juvenile hyperuricemic nephropa- 34. Dahan K, Fuchshuber A, Adamis S, Smaers M, Kroiss S, Loute G, thy with abnormal expression of uromodulin. J Am Soc Nephrol Cosyns JP, Hildebrandt F, Verellen-Dumoulin C, Pirson Y: Fa- 14: 2883–2893, 2003 milial juvenile hyperuricemic nephropathy and autosomal 17. Clayton DG: A model for association in bivariate life tables and dominant medullary cystic kidney disease type 2: Two facets of its application in epidemiological studies of familial tendency in the same disease? J Am Soc Nephrol 12: 2348–2357, 2001 chronic disease incidence. Biometrika 65: 141–151, 1978 35. Wolf MT, Mucha BE, Attanasio M, Zalewski I, Karle SM, 18. Hougaard P: Analysis of multivariate survival data, New York, Neumann HP, Rahman N, Bader B, Baldamus CA, Otto E, Springer, 2000 Witzgall R, Fuchshuber A, Hildebrandt F: Mutations of the Uro- 19. Therneau T: Survival: Survival analysis, including penalized modulin gene in MCKD type 2 patients cluster in exon 4, which likelihood. R package, version 236-10. Available at: http://www. encodes three EGF-like domains. Kidney Int 64: 1580–1587, r-project.org. Accessed January 2012 2003 20. Bolleé G, Dahan K, Flamant M, Morinie`re V, Pawtowski A, 36. Bleyer AJ, Trachtman H, Sandhu J, Gorry MC, Hart TC: Renal Heidet L, Lacombe D, Devuyst O, Pirson Y, Antignac C, manifestations of a mutation in the uromodulin (Tamm Horsfall Knebelmann B: Phenotype and outcome in hereditary tubu- protein) gene. Am J Kidney Dis 42: E20–E26, 2003 lointerstitial nephritis secondary to UMOD mutations. Clin J Am 37. Wolf MT, Beck BB, Zaucke F, Kunze A, Misselwitz J, Ruley J, Soc Nephrol 6: 2429–2438, 2011 Ronda T, Fischer A, Eifinger F, Licht C, Otto E, Hoppe B, 21. Bingham C, Ellard S, van’t Hoff WG, Simmonds HA, Marinaki Hildebrandt F: The Uromodulin C744G mutation causes MCKD2 AM, Badman MK, Winocour PH, Stride A, Lockwood CR, and FJHN in children and adults and may be due to a possible Nicholls AJ, Owen KR, Spyer G, Pearson ER, Hattersley AT: founder effect. Kidney Int 71: 574–581, 2007 Atypical familial juvenile hyperuricemic nephropathy associated 38. Rezende-Lima W, Parreira KS, Garcıá-Gonza´lez M, Riveira E, with a hepatocyte nuclear factor-1beta gene mutation. Kidney Int Banet JF, Lens XM: Homozygosity for uromodulin disorders: 63: 1645–1651, 2003 FJHN and MCKD-type 2. Kidney Int 66: 558–563, 2004 22. Ko¨ttgen A, Hwang SJ, Larson MG, Van Eyk JE, Fu Q, Benjamin EJ, 39. Nasr SH, Lucia JP, Galgano SJ, Markowitz GS, D’Agati VD: Uro- Dehghan A, Glazer NL, Kao WH, Harris TB, Gudnason V,Shlipak modulin storage disease. Kidney Int 73: 971–976, 2008 MG, Yang Q, Coresh J, Levy D, Fox CS: Uromodulin levels as- 40. Bleyer AJ, Hart TC, Willingham MC, Iskandar SS, Gorry MC, sociate with a common UMOD variant and risk for incident CKD. Trachtman H: Clinico-pathologic findings in medullary cystic J Am Soc Nephrol 21: 337–344, 2010 kidney disease type 2. Pediatr Nephrol 20: 824–827, 2005 23. Rampoldi L, Scolari F, Amoroso A, Ghiggeri G, Devuyst O: The 41. Scha¨ffer P, Gombos E, Meichelbeck K, Kiss A, Hart PS, Bleyer rediscovery of uromodulin (Tamm-Horsfall protein): From tubu- AJ: Childhood course of renal insufficiency in a family with a lointerstitial nephropathy to chronic kidney disease. Kidney Int uromodulin gene mutation. Pediatr Nephrol 25: 1355–1360, 80: 338–347, 2011 2010 24. Fairbanks LD, Cameron JS, Venkat-Raman G, Rigden SP, Rees L, 42. Calado J, Gaspar A, Clemente C, Rueff J: A novel heterozygous Van’T Hoff W, Mansell M, Pattison J, Goldsmith DJ, Simmonds missense mutation in the UMOD gene responsible for Familial HA: Early treatment with allopurinol in familial juvenile hyer- Juvenile Hyperuricemic Nephropathy. BMC Med Genet 6: 5, puricaemic nephropathy (FJHN) ameliorates the long-term pro- 2005 gression of renal disease. QJM 95: 597–607, 2002 43. Lens XM, Banet JF, Outeda P, Barrio-Lucıá V: A novel pattern of 25. Cameron JS, Moro F, Simmonds HA: Gout, uric acid and purine mutation in uromodulin disorders: autosomal dominant medul- metabolism in paediatric nephrology. Pediatr Nephrol 7: 105– lary cystic kidney disease type 2, familial juvenile hyperuricemic 118, 1993 nephropathy, and autosomal dominant glomerulocystic kidney 26. Puig JG, Miranda ME, Mateos FA, Picazo ML, Jimeńez ML, Calvin disease. Am J Kidney Dis 46: 52–57, 2005 TS, Gil AA: Hereditary nephropathy associated with hyperuri- 44. Lee DH, Kim JK, Oh SE, Noh JW, Lee YK: A case of familial ju- cemia and gout. Arch Intern Med 153: 357–365, 1993 venile hyperuricemic nephropathy with novel uromodulin gene 27. Benetti E, Caridi G, Vella MD, Rampoldi L, Ghiggeri GM, Artifoni mutation, a novel heterozygous missense mutation in Korea. L, Murer L: Immature renal structures associated with a novel J Korean Med Sci 25: 1680–1682, 2010 UMOD sequence variant. Am J Kidney Dis 53: 327–331, 2009 28. Kudo E, Kamatani N, Tezuka O, Taniguchi A, Yamanaka H, Yabe Received: October 29, 2012 Accepted: April 4, 2013 S, Osabe D, Shinohara S, Nomura K, Segawa M, Miyamoto T, Moritani M, Kunika K, Itakura M: Familial juvenile hyperuricemic nephropathy: Detection of mutations in the uromodulin gene in Published online ahead of print. Publication date available at www. five Japanese families. Kidney Int 65: 1589–1597, 2004 cjasn.org. 29. Zaucke F, Boehnlein JM, Steffens S, Polishchuk RS, Rampoldi L, Fischer A, Pasch A, Boehm CW, Baasner A, Attanasio M, Hoppe This article contains supplemental material online at http://cjasn. B, Hopfer H, Beck BB, Sayer JA, Hildebrandt F, Wolf MT: Uro- asnjournals.org/lookup/suppl/doi:10.2215/CJN.11151012/-/ modulin is expressed in renal primary cilia and UMOD DCSupplemental.