Mutational and phenotypical spectrum of phenylalanine hydroxylase deficiency in Denmark

Clin Genet 2016: 90: 247 251 Printed in Singapore. All rights reserved Short Report 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd CLINICAL GENETICS doi: 10.1111/cge.12692 Mutational and phenotypical spectrum of phenylalanine hydroxylase deficiency in Denmark Bayat A., Yasmeen S., Lund A., Nielsen J.B., Møller L.B.. Mutational and phenotypical spectrum of phenylalanine hydroxylase deficiency in Denmark. Clin Genet 2016: 90: 247 251. John Wiley & Sons A/S. Published by John Wiley & Sons Ltd, 2015 We describe the genotypes of the complete cohort, from 1967 to 2014, of phenylketonuria (PKU) patients in Denmark, in total 376 patients. A total of 752 independent alleles were investigated. Mutations were identified on 744 PKU alleles (98.9%). In total, 82 different mutations were present in the cohort. The most frequent mutation c.1315+1g>a (IVS12+1G>A) was found on 25.80% of the 744 alleles. Other very frequent mutations were c.1222c>t (p.r408w) (16.93%) and c.1241a>g (p.y414c) (11.15%). Among the identified mutations, five mutations; c.532g>a (p.e178k), c.730c>t (p.p244s), c.925g>a (p.a309t), c.1228t>a (p.f410i), and c.1199+4a>g (IVS11+4A>G) have not been reported previously. The metabolic phenotypes of PKU are classified into four categories; classical PKU, moderate PKU, mild PKU and mild hyperphenylalaninemia. In this study, we assigned the phenotypic outcome of three of the five novel mutations and furthermore six not previously classified mutations to one of the four PKU categories. Conflictofinterest The authors have no conflict of interest to declare A. Bayat a,b,, S. Yasmeen a,, A. Lund c, J.B. Nielsen a and L.B. Møller d a Clinical Genetic Clinic, Kennedy Center, Copenhagen University Hospital, Glostrup, Denmark, b Department of Pediatrics Hvidovre Hospital Kettegård Alle 30, 2650 Hvidovre, Denmark, c Clinical Genetic Clinic, Centre for Inherited Metabolic Diseases, Copenhagen University Hospital, Copenhagen, Denmark, and d Department of Science, Systems and Models (NSM), Roskilde University, DK 4000 Roskilde, Denmark These authors are treated as shared first authors. Key words: classification genotype phenotype PAH PKU Corresponding author: Lisbeth Birk Møller, Clinical Genetic Clinic, Kennedy Center, Copenhagen University Hospital, Rigshospitalet, 2600 Glostrup, Denmark. e-mail: Lisbeth.Birk.Moeller@Regionh.dk Received 15 August 2015, revised and accepted for publication 3 November 2015 Phenylketonuria (PKU, OMIM 261600) is an inherited autosomal recessive metabolic disease. It is caused by mutations in the phenylalanine hydroxylase (PAH)gene and is the most common inborn error of amino acid metabolism in the Caucasian population, with an average incidence of 1/10,000 (1). PAH catalyzes the conversion of phenylalanine (Phe) to tyrosine and its deficiency results in accumulation of Phe in the blood. Left untreated condition leads to progressive mental retardation, seizures and cerebral palsy. As the introduction of screening of newborns and dietary Phe restriction, children born with PKU can expect to live a normal life (2). Since 1962, newborns in Denmark have been screened for PKU. The Clinical Genetic Clinic, Kennedy Center is the only national center for PKU, and we are not aware of any missed cases. In Denmark, members of immigrant families are screened if they show symptoms compatible with PKU. The metabolic phenotypes are based on daily tolerance of Phe in milligrams and classified into four categories: classical PKU, moderate PKU, mild PKU and mild hyperphenylalaninemia (MHP). 247

Bayat et al. More than 800 variant PAH alleles have been recorded and include a broad range of alterations, including missense mutations, small deletions and insertions, large deletions and splice-site mutations (3). A simple system for genotype-based prediction of metabolic phenotype has been suggested using the milder of two mutations to predict the phenotypic outcome (2). Thus, in functionally hemizygous patients, e.g. patients with one mutation known to abolish the PAH activity completely on one of the alleles, it has been possible to assign the phenotypic effect of the second mutation to one of the four classes (2, 4). In this study, we describe the mutations identified in the complete cohort of 376 PKU patients in Denmark. Five mutations were not reported previously. Furthermore, we assign nine not previously classified mutations, including three of the five novel mutations, to one of the four PKU categories. Materials and methods Predictive analysis Two online algorithms, sorting intolerant from tolerant (SIFT; http://sift.jcvi.org/) (5) and polymorphism phenotype (PolyPhen-2, http://genetics.bwh.harvard. edu/pph2/dokuwiki/start) (6), were used to predict the functional consequences of PKU missense mutations. A SIFT score less than 0.05 indicates a deleterious amino acid substitution. PolyPhen-2 categorizes the mutations into three groups: benign, possible or probably (7). A value of 1.000 indicates probably. To predict the functional consequences of a PAH splice-site mutation, the Human Splice Finder (HSF, http://www.umd.be/hsf/) software version 2.4 was used (8). The numbering of the affected nucleotides is based on the PAH transcript NM_000277.1, where the ATG start codon is defined as c.1. Patients We included 376 PKU patients from 362 unrelated Danish families, representing all PKU patients in Denmark from the period 1967 2014 referred to the National Center for PKU. The number of patients per family varied from one to three. All patients had PKU according to biochemical criteria (9). Because this is a retrospective study, approval from the Institutional Review Board was not needed. Phenotypic classification Some of the patients with unclassified mutations were assigned to one of the four categories based on daily Phe tolerance. Phe tolerance was determined at the age of 5 years and indicated the amount of daily Phe intake a patient could tolerate without an increase in the blood Phe concentration above the upper target range (400 μmol/l). Patients with classical PKU tolerate <20 mg/kg/day of Phe (250 350 mg/day); patients with moderate PKU tolerate 20 25 mg/kg/day (350 400 mg/day); patients with mild PKU tolerate 25 50 mg/kg/day (400 600 mg/day); and patients with MHP tolerate >50 mg/kg/day (600 mg/day). As we no longer determine Phe tolerance in all patients at age 5, data on Phe tolerance are not available in the younger patients. In cases where the patients were younger than 5 years of age and where a Phe tolerance had not been determined, the pretreatment serum Phe levels were used to discriminate between the four different PKU phenotypes: classic PKU had Phe pretreatment levels >1200 μmol/l, moderate PKU 900 1200 μmol/l, mild PKU 600 900 μmol/l and MHP <600 μmol/l. Screening for PAH mutations Before 2005, screening for PAH mutations was performed by denaturing gradient gel electrophoresis and subsequent sequencing as described previously (10). After 2005, identification of PAH mutations was performed by direct sequencing of polymerase chain reaction (PCR)-amplified exons and flanking intronic sequences. Screening for exon deletions or duplications was performed by multiplex ligation-dependent probe amplification assay using the PAH kit (SALSA P055) MRC-Holland (Amsterdam, The Netherlands) as described previously (11). Results Phenotype distributions, identified mutations and allele frequencies The 376 patients were screened for PAH mutations, and potential causative mutations were identified on 744 alleles. A minor fraction of these results have been published previously (10, 11). Two mutations were identified in 368 patients, and in 8 only 1 was found. In total, a mutation was detected on 98.9% of 752 Danish PKU alleles (744/752). A total of 82 different mutations were identified. The relative frequencies of all mutations identified in the Danish PKU population are given in Table S1, Supporting information. The most frequent mutation IVS12+1G>A was found on 25.80% of the alleles. Other very frequent mutations were p.r408w (16.93%) and p.y414c (11.15%) both located in exon 12. Mutations in exon 12 accounted for 58.43% of the mutated alleles. The mutation IVS10-11G>A accounted for 4.83% and p.r158q for 2.28% of the alleles. The remaining mutations were less frequent or rare (0.13 1.61%). Among the identified mutations, five mutations: p.e178k, p.p244s, p.a309t, p.f410i, and IVS11+4A>G have not previously been reported to the PAH mutation databases (3) or published according to the Human Gene Mutation Database (http://www.hgmd.cf.ac.uk/ac/index.php). Classification of the mutations Parental testing confirmed that the two mutations identified in the 368 patients all were present in trans. 248

PAH mutations in Denmark Table 1. Attempt to classify previously unclassified mutations including five novel mutations a Patient (ethnic origin) Allele 1: nucleotide change, predicted protein change (classification) Allele 2: unclassified mutation. Nucleotide change, predicted protein change SIFT score Polyphen HSF Phenotype observed Predicted classification of mutation 1 (Pakistan) c.293t>c, p.l98s c.293t>c, p. L98S 0.01: Deleterious 0.998: probably 2 (Iraq) c.997c>t, p.l333f (classic) c.329c>t, p.s110l 0.01: Deleterious 0.995: probably 3 (The Netherlands) c.1222c>t, p.r408w (classic) 4 (Unknown) c.116_118del, p.f39del (classic) 5a + 5b (Syria) c.782g>a, p.r261q (moderate) c.385g>t, p.d129y 0.00: Deleterious 1.000: probably Mild p.l98s: mild MHP b p.s110l: MHP Mild PKU c p.d129y: mild c.532g>a, p. E178K 0.27: Tolerated 0.029: benign Probably MHP b,d p.e178k: MHP c.545a>g, p. E182G 0.01: Deleterious 0.924: probably 6 (Denmark) c.473g>a, p.r158q (classic) c.730 C>T, p. P244S 0.00: Deleterious 1.000: probably 7 (Turkey) c.532g>a, p.e178g (MHP) c.733g>a, p.v245m 0.00 Deleterious 1.000: probably 8 (Denmark) c.1315+1g>a, c.922c>g, p. L308V 0.00 Deleterious 1.000: probably 9 (Denmark) c.1241a>g, p.y414c (mild) c.925g>a, p. A309T 0.00: Deleterious 1.000: probably 10 (Denmark) c.1315+1g>a, 11 (Denmark) c.1315+1g>a, MHP e p.e182g: MHP Classic to moderate PKU p.p244s: classic (moderate) MHP p.v245m: unclassified Mild PKU c p.l308v: mild Mild p.a309t: unclassified c.1199+4a>g (IVS11+4A>G) ΔCV = 8.7% Classic PKU c IVS11+4A>G: classic PKU c.1228t>a, p. F410I 0.00: Deleterious 1.000: probably 12 (Denmark) c.916a>g, p.i306v (MHP) c.1229t>c, p. F410S 0.00: Deleterious 1.000: probably Mild to moderate PKU p. F410I: moderate (mild) MHP c p. F410S: unclassified HSF, human splice finder; MHP, mild hyperphenylalaninemia; Phe, phenylalanine; PolyPhen, Polymorphism Phenotype; SIFT, Sorting intolerant from tolerant. a Novel mutations are in bold. b Phenotype based on Phe tolerance and pretreatment Phe level. c Phenotype based on pretreatment Phe level. d Lost to follow-up at the age of 3 months, but at that age, the patient was clinically MHP with no diet requirement. e Phenotype based on Phe tolerance. 249

Bayat et al. Both mutations in 355 patients were previously classified, based primarily on the phenotypic characteristics observed in functionally hemizygous patients (2, 4) (http://www.biopku.org/home/home.asp). Based on this classification, we found that 166 patients had severe, 22 moderate, and 94 mild PKU; 73 had MHP (Table S2). We found a good genotype phenotype correlation based on the patient s daily tolerance of Phe. The remaining 13 patients, with two identified mutations, had either one or two unclassified mutations. In total, 12 different mutations were unclassified including the 7 previously identified mutations: p.l98s, p.s110l, p.d129y, p.e182g, p.v245m, p.l308v and p.f410s and the 5 novel not previously described mutations: p.e178k, p.p244s, p.a309t, p.f410i, and IVS11+4A>G (Table1). Seven of the unclassified mutations were observed in trans combination with a previously classified, classic mutation, abolishing the PAH activity completely (patients 2, 3, 4, 6, 8, 10, and 11). Patient 1 was homozygous for a previously unclassified mutation. Based on the degree of Phe tolerance in combination with the pretreatment Phe level, we could classify the clinical phenotype in two of the patients as MHP (patients 2 and 4) (Table 1 and Table S2). Based on the pretreatment levels, we were able to safely classify two patients as mild (patients 3 and 8) and one patient as classic (patient 10) (Table 1 and Table S2). Patients 1, 6 and 11 were more difficult to classify. Although the pretreatment Phe level in patients 1 and 11 was low (300 and 404 μmol/l, respectively, indicating a MHP phenotype), Phe-restricted diet was necessary, indicating that the clinical phenotypes in the two patients were mild and not MHP. Currently, patient 11 has a daily Phe intake at 20 25 mg/kg/day with blood Phe levels at 210 250 μmol/l, indicating that the patient suffers from either mild or moderate PKU. Based on the pretreatment Phe level (1537 μmol/l), the clinical phenotype of patient 6 is classic, but the observed daily Phe intake and blood Phe levels indicated that the phenotype could be moderate. A Phe tolerance determination is however not available in patients 1 and 6 because of age above 5 and in patient 11 because of age below 5. As the mildest mutation determines the clinical phenotype (2), this observation leads to the classification of p.s110l in patient 2 and p.e178k in patient 4 as mutations leading to MHP, p.d129y in patient 3 and p.l308v in patient 8 as mutations leading to mild PKU, p.f410i in patient 11 as a mutation leading to moderate/mild PKU, p.p244s in patient 6 as a mutation leading to classic/moderate PKU and the novel mutation IVS11+4A>G in patient 10 as a classic PKU mutation. The mutation p.l98s was observed in homozygous form in patient 1 with mild PKU, indicating that p.l98s could be classified as a mild mutation. Two siblings (patient 5a and 5b) with MHP had the previously classified mutation p.r261q (moderate) on one allele and the mutation p.e182g on the other allele, indicating that p.e182g can be classified as a mutation leading to MHP. The classification of the last three mutations was not possible. Classification of p.v245m in patient 7 and p.f410s in patient 12, both with MHP phenotype, was not possible because they were in trans combination with the MHP mutations p.e178g and p.i306v, respectively. Finally, the phenotype of patient 9 was mild and as the novel mutation p.a309t was in trans combination with the mild mutation p.y414c, p.a309t could not be classified. In an alternative attempt to classify the remaining three unclassified amino acid substitutions, p.v245m, p.a309t, and p.f410s, we explore the correlation between in situ predicted effect, using the algorithms SIFT and PolyPhen, and real severity. The values for all mutations are shown in Table S1. Although no clear correlation between values for predicted effect and actual classification of the previously classified mutations values was present, predicted values as possibly, benign and tolerated were obtained only for mild and MHP mutations. Only for MHP missense mutations, a combined prediction, from both in situ prediction tools, as tolerated and benign, respectively, was obtained, but only in 4 out of 21 MHP mutations. Furthermore, the number of missense mutations with a SIFT value different from 000 or a PolyPhen value different from 1.000 decreased with increasing severity of the associated phenotype, for MHP 48%, for mild PKU 20% and for classic PKU 19%. Same tendency was observed when investigating only the amino acid substitutions, p.l98s, p.s110l, p.d129y, p.e182g, p. 244S, p.l308v, p.f410i, and p.e178k classified in this study. As the three unclassified amino acid substitution p.v245m, p.a309t, and p.f410s all had the SIFT value 0.00 combined with the PolyPhen value 1.000, classification was not possible. The predicted reduction in the ΔCV value, obtained from HSF, of the splice-site mutation IVS11+4A>G, is 8.7%. As an effect of 7%, for a mutation in position +4, is expected to affect the splicing, this is in agreement with the classification of this mutation as disease causing. The classification of all mutations identified in Danish PKU patients, including the classification performed as a result of this study, is shown in Table S1, and the genotype of all Danish PKU patients is shown in Table S2. Discussion We have presented genotype data and corresponding phenotype of 376 Danish PKU patients (Table S2). It has been possible to create a national database to produce a detailed picture of the mutational spectrum in the entire Danish PKU population. A total of 82 different mutations were identified and 42 were classified as classical PKU mutations, 6 as moderate, 11 as mild and 20 as MHP and 3 were unclassified mutations (Table S1). The mutations included 70% missense mutations, 14% splice-site mutations, 8% nonsense mutations and 8% small deletions or indels, and a single big deletion spanning exons 1 and 2. We attempted to classify 12 mutations not previously classified, including 5 novel mutations and we were able to classify 9 of the 12 mutations. Attempts to classify the remaining three 250

PAH mutations in Denmark mutations using the prediction algorithms SIFT and PolyPhen failed as an investigation of already classified mutations showed no clear association between the predicted values and the classification. Looking at the relative frequencies of the PKU alleles, the PKU mutational spectrum in Denmark is heterogeneous. The most common mutation is the splice-site mutation IVS12+1G>A, with 25.80% occurrence, although it was even more frequent with 37.3% occurrence in 1993 (10). The frequencies of p.y414c (11.15%) and p.r408w (16.93%) have not changed compared with 1993 (10). The changes in the Danish mutational profile probably reflect the changes in the ethnic composition of the community with more migration from other countries than previously. The mutation p.a300s occur with a frequency of 1.2% and p.l48s with a frequency of 0.53%, both were absent in 1993 (10). A total of 9 of 12 patients with the mutation p.a403v were from Mediterranean countries, and the mutations S349A and p.l48s were only identified in patients from Mediterranean countries (two and four patients, respectively). Recent migration has brought p.a403v, p.l48s, S349A and other Mediterranean mutations to northern Europe (12, 13). The data presented in this study have facilitated the establishment of a database of causative Danish PAH mutations and will be a valuable tool in genetic counseling and prognostic evaluation of future cases of PKU. Supporting Information Additional supporting information may be found in the online version of this article at the publisher s web-site. Acknowledgements We thank the dieticians from the Kennedy Center for assistance. References 1. Scriver CR. Whatever happened to PKU? Clin Biochem 1995: 28 (2): 137 144. 2. Guldberg P, Rey F, Zschocke J et al. A European multicenter study of phenylalanine hydroxylase deficiency: classification of 105 mutations and a general system for genotype-based prediction of metabolic phenotype. Am J Hum Genet 1998: 63 (1): 71 79. 3. Scriver CR, Waters PJ, Sarkissian C et al. PAHdb: a locus-specific knowledgebase. Hum Mutat 2000: 15 (1): 99 104. (http://www.pahdb.mcgill.ca.database or http://www.biopku.org/home/ home.asp). 4. Guttler F, Azen C, Guldberg P et al. Relationship among genotype, biochemical phenotype, and cognitive performance in females with phenylalanine hydroxylase deficiency: report from the maternal phenylketonuria collaborative study. Pediatrics 1999: 104 (2): 258 262. 5. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc 2009: 4 (7): 1073 1081. 6. Adzhubei IA, Schmidt S, Peshkin L et al. A method and server for predicting missense mutations. Nat Methods 2010: 7 (4): 248 249. 7. Sunyaev S, Lathe W III, Bork P. Integration of genome data and protein structures: prediction of protein folds, protein interactions and "molecular phenotypes" of single nucleotide polymorphisms. Curr Opin Struct Biol 2001: 11 (1): 125 130. 8. Desmet FO, Hamroun D, Lalande M, Collod-Beroud G, Claustres M, Beroud C. Human Splicing Finder: an online bioinformatics tool to predict splicing signals. Nucleic Acids Res 2009: 37 (9): e67. 9. Guttler F. Hyperphenylalaninemia: diagnosis and classification of the various types of phenylalanine hydroxylase deficiency in childhood. Acta Paediatr Scand Suppl 1980: 280: 1 80. 10. Guldberg P, Henriksen KF, Guttler F. Molecular analysis of phenylketonuria in Denmark: 99% of the mutations detected by denaturing gradient gel electrophoresis. Genomics 1993: 17 (1): 141 146. 11. Birk ML, Nygren AO, Scott P et al. Low proportion of whole exon deletions causing phenylketonuria in Denmark and Germany. Hum Mutat 2007: 28 (2): 207. 12. Dobrowolski F, Heintz C, Miller T et al. Molecular genetics and impact of residual in vitro phenylalanine hydroxylase activity on tetrahydrobiopterin responsiveness in Turkish PKU population. Mol Genet Metab 2011: 10: 116 121. 13. Zschocke J. Phenylketonuria mutations in Europe. Hum Mutat 2003: 21 (4): 345 356. 251