Article Text

PDF

Mutations in hepatocyte nuclear factor-1β and their related phenotypes
  1. E L Edghill,
  2. C Bingham,
  3. S Ellard,
  4. A T Hattersley
  1. Institute of Biomedical and Clinical Science, Peninsula Medical School, Exeter, UK
  1. Correspondence to:
 Professor Andrew T Hattersley
 Diabetes and Vascular Medicine, Peninsula Medical School, Barrack Road, Exeter, EX2 5AX, UK; A.T.Hattersley{at}exeter.ac.uk

Abstract

Background: Hepatocyte nuclear factor-1 beta (HNF-1β) is a widely distributed transcription factor which plays a critical role in embryonic development of the kidney, pancreas, liver, and Mullerian duct. Thirty HNF-1β mutations have been reported in patients with renal cysts and other renal developmental disorders, young-onset diabetes, pancreatic atrophy, abnormal liver function tests, and genital tract abnormalities.

Methods: We sequenced the HNF-1β gene in 160 unrelated subjects with renal disease, 40% of whom had a personal/family history of diabetes.

Results: Twenty three different heterozygous HNF-1β mutations were identified in 23/160 subjects (14%), including 10 novel mutations (V61G, V110G, S148L, K156E, Q176X, R276Q, S281fsinsC, R295P, H324fsdelCA, Q470X). Seven (30%) cases were proven to be due to de novo mutations. Renal cysts were found in 19/23 (83%) patients (four with glomerulocystic kidney disease, GCKD) and diabetes in 11/23 (48%, while three other families had a family history of diabetes. Only 26% of families met diagnostic criteria for maturity-onset diabetes of the young (MODY) but 39% had renal cysts and diabetes (RCAD). We found no clear genotype/phenotype relationships.

Conclusion: We report the largest series to date of HNF-1β mutations and confirm HNF-1β mutations as an important cause of renal disease. Despite the original description of HNF-1β as a MODY gene, a personal/family history of diabetes is often absent and the most common clinical manifestation is renal cysts. Molecular genetic testing for HNF-1β mutations should be considered in patients with unexplained renal cysts (including GCKD), especially when associated with diabetes, early-onset gout, or uterine abnormalities.

  • FJHN, familial juvenile hyperuricaemic nephropathy
  • GCKD, glomerulocystic kidney disease
  • HNF-1β, hepatocyte nuclear factor-1 beta
  • MODY, maturity-onset diabetes of the young
  • RCAD, renal cysts and diabetes
  • genetics
  • HNF-1β
  • maturity-onset diabetes of the young
  • MODY
  • RCAD syndrome
  • renal cysts

Statistics from Altmetric.com

Hepatocyte nuclear factor-1β is a widely distributed transcription factor that is vital for embryonic survival.1,2 Early expression of hepatocyte nuclear factor-1 beta (HNF-1β) is seen in the kidney, liver, bile ducts, thymus, genital tract, pancreas, lung, and gut.3,4 It can act either as a homodimer or as a heterodimer with HNF-1α.5

The HNF-1β (TCF2) gene is located on chromosome 17q21.3. The first description of HNF-1β mutations associated with disease was in 1997.6 A total of 30 heterozygous mutations in the HNF-1β gene have been described including missense, nonsense, frameshift, insertion/deletions, and splice site mutations.7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28 The majority of these mutations are familial, but eight spontaneous mutations7–9,13,15,17,23 and one germline mosaic mutation16 have been described.

The first HNF-1β mutation identified was in the autosomal dominant, young onset, subtype of diabetes known as maturity-onset diabetes of the young (MODY). Horikawa and colleagues showed that 1/57 Japanese MODY subjects had a mutation in the HNF-1β gene.6 A potential role for HNF-1β in MODY was suggested after mutations in the gene encoding its heterodimer partner HNF-1α were found to be a common cause of MODY.29 Although HNF-1β gene mutations are not a common cause of MODY in the UK,30 diabetes is frequently associated with HNF-1β mutations. It is usually of early-onset (median age 20 years (range 15 days–61 years) and insulin treatment is common.31 Consistent with HNF-1β having a key role in the development of the pancreas,32 pancreatic atrophy and exocrine dysfunction are common15 and neonatal diabetes has been reported.16

Extra-pancreatic disease was rapidly recognised with the description of renal cystic disease19 and since then the involvement of multiple organs has been described. The pattern of clinical features follows the expression of HNF-1β which is primarily in the kidney, pancreas, liver, and genital tract.3,4 The most consistent feature has been renal disease (seen in all probands and 78/88 (89%) mutation carriers). The renal disease described is very heterogeneous but is always due to aberrant renal development and includes: renal cysts,6,7,8,9,10,11,12,13,14,15,17,18,19,21,22,23,24,25,26,28,33 familial hypoplastic glomerulocystic kidney disease (GCKD),8,23,27 renal malformations (for example, single and horseshoe kidney),9,21 and atypical familial hyperuricaemic nephropathy.10 Other clinical features include genital tract malformations,9,18,20,26 abnormal liver function tests,6,15,24,26 pancreatic atrophy and exocrine insufficiency,15 biliary manifestations,25 gout and hyperuricaemia.10,12

Understanding the pathogenic mechanisms of HNF-1β mutations was delayed as homozygous HNF-1β knockout mice were not viable and heterozygous HNF-1β knockout mice had no phenotype.1,2 However, organ specific mouse knockouts have been created for the liver,34 kidney,35 and pancreas,36 and all of these have organ specific phenotypes similar to those seen in patients with HNF-1β mutations. Functional studies have shown mutations with either loss of function,17,18 dominant negative actions,17,25,37 or gain of function.22

The majority of patients reported in the literature have been described in case reports or belong to small cohorts (n<3). The largest previous series to date is a study of 20 subjects with diabetes diagnosed before 40 years of age (who did not have obesity or pancreatic autoantibodies) who also had renal disease. Eight of these patients (40%) were found to have HNF-1β mutations.15 In this study, we report a series of 160 patients with unexplained renal disease (with or without diabetes) and describe the genetics, phenotypes, and genotype-phenotype relationships in the 23 subjects with mutations in the HNF-1β gene.

METHODS

Subjects

A total of 160 Caucasian subjects were recruited from the UK, Europe, and the USA. The inclusion criterion was renal disease without a defined aetiology. Renal disease was classified into five groups: (1) GCKD diagnosed on the basis of histological or radiological appearance; (2) atypical familial juvenile hyperuricaemic nephropathy (FJHN) (subjects with young-onset hyperuricaemia, gout, renal impairment, and also disorders of renal development including renal cysts); (3) unexplained renal cystic disease in patients who had renal cysts but did not meet the criteria for GCKD or atypical FJHN (subjects in groups 1 and 2 may have renal cysts but are not included in this category); (4) renal dysplasia in which the kidneys have severe disturbance of differentiation and complete or partial lack of normal nephrons; and (5) renal malformations which are defined as gross renal developmental abnormalities and include single kidney (n = 20), horseshoe kidney (n = 1), and hypoplastic kidney(s) (n = 8). Details regarding renal disease, the presence of diabetes in the proband or a family member, and other clinical features were obtained from the referring clinician or the patient’s hospital records. Summary clinical characteristics for all 160 subjects are shown in table 1.

Table 1

 Summary of clinical findings in the renal disease cohort

HNF-1β genetic analysis

Genomic DNA was extracted from peripheral lymphocytes using standard procedures. Exons 1–9 including intron-exon boundaries and the minimal promoter of HNF-1β were amplified by the polymerase chain reaction.6 Sequence specific primers for each exon10 were tagged with 5′ M13 tails to allow sequencing to be performed with a universal M13 primer.

Both strands were sequenced in forward and reverse directions using the Big Dye Terminator Cycler Sequencing Kit (Applied Biosystems, Warrington, UK) according to the manufacturer’s instructions. Reactions were analysed on an ABI 3100 Capillary DNA sequencer (Applied Biosystems). Sequences were compared to the published sequence (NM_000458) using Sequence Navigator Software (Applied Biosystems, Warrington, UK). Changes in the sequence were checked against published polymorphisms and mutations and for conservation across species.

Microsatellite analysis

Family relationships were confirmed using three microsatellites on chromosome 17: D17S800, D17S927, and D17S1718.7

RESULTS

Genetics

We identified 23 heterozygous HNF-1β gene mutations (shown in fig 1) in 23/160 subjects (14%). Ten of these subjects have previously been reported (H69fsdelAC, E101X, S151P, P159fsdelT, IVS2nt+1G>T, IVS2nt+2insT, Q243fsdelC, P328L329fsdelCCTCT, Y352fsinsA, and A373fsdel29).7,8,9,10,11,12,13,14 The 13 unreported subjects include 10 novel mutations (V61G, V110G, S148L, K156E, Q176X, R276Q, S281fsinsC, R295P, H324S325fsdelCA, and Q470X) and three patients with previously reported mutations (R181X, IVS2nt+1G>A, and IVS2nt+2delAAGT).15,21,33 The 10 novel mutations include missense, nonsense, and frameshift mutations, none of which were found in 100 normal chromosomes. All the missense mutations affect residues which are conserved in both rat and mouse. Details of the novel mutations are shown in table 2.

Table 2

 Molecular genetics characteristics of the 10 novel HNF-1β mutations

Figure 1

HNF-1β gene structure illustrating the 40 different mutations identified to date which span the first seven exons of the gene; those identified in this study (n = 20) are in italics. The exons and protein domains are indicated; the exons are relative to their sequence length. The HNF-1β protein contains 557 amino acids with three distinct domains: the dimerisation domain (1–32 amino acids), the DNA binding domain (90–311 amino acids), and the transactivation domain (312–557 amino acids).

In addition to pathogenic mutations we also found two rare variants (V25L (GTG>CTG) in 1/159 subjects, and IVS4-22C>T in 1/159 subjects), but it is uncertain whether or not these variants represent aetiological mutations. The V25L carrier inherited this HNF-1β variant from his father who is not known to be affected by renal dysfunction or diabetes but has not had a renal ultrasound. As it was not possible to follow up this family, we therefore report V25L as a rare variant. The intron 4 variant -22C>T does not affect the conserved splicing branch site, but we were not able to check co-segregation within this family.

To investigate co-segregation in the families of the 13 newly reported probands, we tested 27 family members (10 affected and 17 unaffected) from 11 pedigrees. This and previously reported family studies allowed us to classify 39% of mutations as (9/23) as familial, where family members were affected and had an HNF-1β mutation. In the unreported families, three sets of parental DNA did not carry a mutation; microsatellite analysis confirmed these three cases as spontaneous de novo mutations (fig 2), in addition to the four previously reported spontaneous mutations.7–9,13 In the remaining pedigrees there was no known family history, suggesting that these were isolated cases.

Figure 2

 Pedigrees of 11 families with unreported HNF-1β mutations. Individuals with renal cysts are identified by a symbol with a filled upper left quarter, those with diabetes with a filled lower left quarter, those with other forms of renal disease with a filled upper right quarter, and those with other non-renal, non-pancreatic abnormalities with a filled lower right quarter.

Clinical phenotypes

The clinical characteristics of the family members from previously unreported pedigrees with HNF-1β mutations are shown in table 3. The commonest renal manifestation in patients with HNF-1β mutations was renal cysts which were present in 19/23 (83%). Four of these subjects had a specific diagnosis of GCKD. Diabetes was present in 11/23 (48%) and three (13%) had an affected first degree relative but no diabetes themselves. Only 6/23 families (26%) met the minimal MODY criteria (families with at least two generations of diabetes with at least one subject diagnosed under the age of 25 years). The common manifestation of both renal cysts and diabetes (RCAD) was found in 9/23 (39%) of probands (table 4).

Table 3

 Clinical characteristics of the 13 previously unreported HNF-1β mutation positive probands and their relatives

Table 4

 Renal disease and extra-renal phenotypes of all 23 probands with an HNF-1β mutation

In addition to renal disease and diabetes, 2/23 (9%) probands had genital tract malformations (table 4). These probands had uterus didelphys and a single ovary (DUK448) and hemi-uterus (DUK1094), respectively. Another subject (DUK972) had unexplained primary infertility. It was also noted that in three families a first degree family member had a malignancy although in no cases was it possible to establish that these family members had an HNF1β mutation. The tumours arose in biliary tract, uterine tract, and bowel, all tissues with high expression of HNF-1β.

None of our mutation carriers had clinical liver disease, although on biochemical testing the serum levels of liver enzymes (alanine aminotransferase and γ glutamyl transpeptidase) exceeded the upper limit of normal in 5/9 tested. Alkaline phosphatase alone was tested in a further seven patients and was abnormal in four, but this may represent renal bone disease rather than hepatic dysfunction as other liver enzymes were not measured.

Genotype-phenotype correlation

This cohort included four subjects with mutations of the intron 2 splice donor site: IVS2nt+1G>T, IVS2nt+2insT, IVS2nt+1G>A, and IVS2nt+2delAAGT. This is the most frequent site of mutations in the HNF-1β gene with a total of seven unrelated families (19 mutation carriers) reported to have one of these mutations.10,12,15,21,33 These mutations are predicted to result in an abnormally spliced variant (either exon loss or intron retention).12 We examined the clinical characteristics of these patients in order to investigate the possibility of a genotype-phenotype relationship. This analysis was not consistent with there being a specific phenotype associated with this mutation site. All of the 19 subjects with an intron 2 splice donor site mutation had renal disease: 11 had renal cysts (including one with GCKD), seven had FJHN (including two with FJHN and renal cysts), and one had a horseshoe kidney. Twelve of the 19 had diabetes and two had genital tract abnormalities.

DISCUSSION

We report the results of screening a large series of subjects with unexplained renal disease for HNF-1β mutations. We have identified 23 different HNF-1β mutations, of which 10 are novel. These novel mutations co-segregate with renal disease. The mutated residues are conserved across species and were not found in 100 normal chromosomes.

A total of 40 HNF-1β gene mutations have now been identified in 46 families.6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28 The majority (37/40) of these mutations are confined to the first four exons. Exons 1–4 encode the dimerisation and DNA binding domains of HNF-1β and mutations in these domains are predicted to alter the ability of hetero/homodimer formation or to change the DNA targeting sequence. There are no missense mutations in the transactivation domain, but the functional significance of this is not known. In this report we describe the most 3′ mutation to date: Q470X in exon 7. The majority of mutations are private, but there is a known hotspot at the intron 2 splice donor site (7/46 families). The most commonly mutated region of the gene is exon 2 where 16/40 mutations occur. Functional studies of exon 2 mutations have shown that this domain is critical for transcriptional activity. Studies looking at the effect of the exon 2 HNF-1β mutation H153N on the transcriptional activity of HNF-1α have shown that it has a repressive effect.25 Other mutations in the homeodomain result in a severe reduction in target gene activation38 or do not bind DNA27; therefore mutations in this region will alter the levels of HNF-1β gene targets.

There are limitations to our study. Firstly, the cohort was centre rather than population based and patients were referred on the basis of their phenotype. Therefore, we cannot use the data to determine how common HNF-1β mutations are in the general population with undetected renal disease. Secondly, clinical assessment was not standardised as patients were referred from multiple centres and assessment of potential phenotype features varied. Clinicians varied in the extent to which they tested for diabetes and uterine abnormalities in asymptomatic patients especially if they were young. All clinical features reported were confirmed and are robust, but some features may have been present but not clinically detected especially as regards uterine abnormalities and abnormal liver function tests. However, the features present would reflect the likely clinical features of a case being considered for genetic testing.

The identification of 23 HNF-1β mutations in subjects with renal disease has allowed us to refine the clinical phenotype associated with HNF-1β mutations. The first report of an HNF-1β gene was in a family with MODY, but within our series only six families met minimal MODY criteria. Overall, only 11/23 probands had diabetes and only three probands without diabetes had a family member with diabetes. This is in contrast to the series described by Bellanne-Chantelot et al who only screened subjects with diabetes and renal disease.15 Our study emphasises that diabetes is not an essential requirement for the identification of an HNF-1β mutation. In addition, a family history of diabetes or renal disease may not be present as de novo mutations are common in 32% of the families where parental DNA is available. Therefore, HNF-1β mutation analysis should not be restricted to subjects with a family history of diabetes and/or renal disease.

Our study confirms that HNF-1β mutations are a fairly common cause of renal disease and that the presentation of renal disease includes a wide spectrum of renal phenotypes. Renal cysts are present in most HNF-1β mutation carriers including those with the histologically proven form of cystic kidney disease, GCKD. Renal malformations are the second most common form of renal disease associated with HNF-1β mutations and include single and horseshoe kidneys. In our study, 9/23 (40%) probands had RCAD. This is the most common phenotype; together with reports from the literature, 24/46 (52%) probands with HNF-1β mutations have RCAD.6,15,17–22,24–28

Genital tract abnormalities, in particular those affecting the Mullerian tract, have previously been reported in 8/23 index cases with an HNF-1β mutation. In our series, only 2/23 probands had a genital tract abnormality. The low detection rate of genital tract abnormalities could be due to incomplete clinical investigation. Therefore, systematic screening of subjects with Mullerian tract abnormalities is required to determine the significance of these findings.

From published data and our data presented here, it is clear that the phenotype of HNF-1β mutation carriers is variable; phenotypes vary between and within families, and there is also little evidence to correlate mutation type or position and the clinical characteristics. We previously reported a family where the mother had diabetes and mild renal dysfunction, but her 19 week fetus had grossly enlarged cystic kidneys incompatible with life.7 Yorifuji et al have reported a family where two siblings both have a S148W missense mutation but different phenotypes; one had neonatal diabetes with no renal disease, whereas his brother had severe renal disease but no diabetes.16 These examples illustrate how the same mutation can result in a varied phenotype, suggesting that other genetic or environmental factors can play a considerable role in the pattern and severity of the disease associated with an HNF-1β mutation.

It is apparent that extra-pancreatic features are more common than initially thought in subjects with an HNF-1β mutation. Hence this study and others are limited by incomplete extra-pancreatic clinical investigations. The majority of HNF-1β mutations associated with renal disease and genital tract abnormalities are located within the intron 2 splice donor site or exon 2 (8/10 mutations). Exon 2 encodes the pseudo-POU domain which is involved in DNA binding specificity and is essential for the development of a normal kidney and genital tract.

This study also highlights the fact that that HNF-1β mutations are only one of many causes of renal cysts, as 82% of patients with renal cysts did not have a mutation. Polycystic kidney disease (OMIM 173900) is genetically heterogeneous with multiple loci including chromosome 16p (polycystic kidney disease 1, PKD1), 4q (PKD2), and 6p (autosomal recessive polycystic kidney disease) where mutations have been reported in the genes encoding polycystin-1, polycystin-2, and fibrocystin, respectively. There is a third autosomal dominant form of PKD (PKD3) which is linked to chromosome 2q. Subjects with PKD present from birth to late adulthood. Therefore, some patients in our cohort are likely to harbour a mutation in one of these PKD genes or in a novel gene.

In conclusion, mutations in the HNF-1β gene are a fairly common cause of renal disease and mutation carriers have a variety of extra-renal phenotypes. We report the largest series to date of HNF-1β mutations which includes three de novo mutations. As de novo mutations represent up to a third of cases, a family history of renal disease or diabetes should not be considered a prerequisite for molecular genetic testing. Following this study, a large number of HNF-1β mutations have now been reported but there is still no clear evidence for a genotype-phenotype relationship. This study highlights the fact that although HNF-1β was initially described as a MODY gene, patients usually present with renal disease or RCAD rather than with MODY.

Acknowledgments

The authors thank all the families for their participation, and their referring doctors for supplying the clinical information and making this study possible. These include: Dr N Gordjani, Dr C J Burton, Dr N Albers, Dr A Nicholls, Dr H Leslie, Dr C M Taylor, Professor G Rizzoni, Dr Arikoski, Dr Nadar, Dr M McGraw, Dr S Rigden, Dr S Arslanian, Dr Kuan, Dr T Cole, Dr R Davidson, Professor A Woolf, Dr P Williams, Dr Van’t Hoff, Dr Rees, Dr R Jones, Dr Trompeter, Dr K Jones, and Dr Skinner. We thank Dr Ewan Pearson for clinical expertise and Dr Lorna Harries and Lisa Allen for help with molecular genetic analysis.

REFERENCES

View Abstract

Footnotes

  • Published Online First 8 June 2005

  • We thank the National Kidney Research Fund (grant TF13/2000), Exeter Kidney Unit Development Fund, Royal Devon and Exeter NHS Foundation Trust R&D Directorate, British Medical Association, and Wellcome Trust who all supported this work. ATH is a Wellcome Trust research leave fellow and CB Bingham was an NKRF clinical research fellow

  • Competing interests: none declared

  • Ethics approval was given for this study by the South West Multi-Centre Research Ethics Committee

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.