Introduction

Phenylalanine hydroxylase deficiency (PAHD), one of the genetic disorders that can lead to hyperphenylalaninemia (HPA), is a phenotypically heterogeneous disorder of phenylalanine (Phe) metabolism. Recessive mutations in PAH result in high Phe levels in the blood and brain. More than 1,000 PAH mutations have been identified to date, recorded as PAHvdb in the Locus-Specific Database (http://www.biopku.org/pah). Undoubtedly, many more variants remain to be detected in different geographic populations. The metabolic phenotypes of PAHD range from mild hyperphenylalaninemia (MHP), which does not require treatment, to classic phenylketonuria (cPKU, OMIM 261600), which requires significant intervention. This broad phenotypic range is due to the different PAH genotypes conferring different degrees of reduced enzyme activity. PAHD has an incidence of 1/10,000 in the Europeans population1, 1/11,614 in the Chinese population2, 1/143,000 in the Japanese population, and a much higher incidence of 1/2,600 in the Turkish population3.

Accumulated evidence has demonstrated that the phenotypic and mutational spectrum of PAHD varies among different ethnic and geographic populations. Overall, the two most frequent variants listed in PAHvdb are p.R408W (19.2%) and c.1066-11G > A (6.8%). Variants p.R261Q (15.7% of alleles) and p.A403V (11.6% of alleles) are most common in PKU patients from southern Italy4, while the most prevalent variant in the Spanish population is p.I65T5. In the Chinese population, the highly p.R241C (36%) mutation was detected in Taiwan patients as a founder effect on PAHD6, while p.R243Q (17.53%) is common in PKU patients from both northern and southern Mainland China7. Additionally, mild PKU (mPKU) and MHP are present with higher incidence in the Chinese population compared with cPKU, which is more prevalent in Eastern Europe8,9. PAH analysis and phenotype assessment are critical for effective clinical diagnosis and treatment of PAHD, but the parameters of these remain incompletely characterized for both PKU and MHP Chinese patients.

In the present study, we summarized the PAHD mutational and phenotypic spectrum by analyzing PAH variants and genotypes in both PKU and MHP patients using samples from neonatal screening for HPA collected between 1999 and 2016 in Zhejiang Province. Seven novel variants of PAH were recorded. Heterogeneous genotypes with specific allele frequencies were observed for both PKU and MHP patients.

Materials and Methods

Subjects and phenotypic classification

Unrelated samples (n = 420) were collected via the neonatal screening program at the newborn screening center of the Children’s Hospital, Zhejiang University School of Medicine, Zhejiang Province, between 1999 and 2016. The metabolic phenotypes of all subjects were classified solely on the maximum pretreatment plasma Phe levels without dietary therapy as either cPKU (cPKU, ≥1200 μmol/L), mPKU (mPKU, 360–1200 μmol/L), and MHP (120–360 μmol/L) according to our national consensus10. Patients with BH4 cofactor deficiency were excluded. Informed consent, blood samples and clinical evaluations were obtained from all participants and families, under protocol approved by the Institutional Review Board of the Children’s Hospital of Zhejiang University School of Medicine.

PAH variant analysis

For the samples collected between 2010 and 2016, two hundred and nine patients consented to taking a gene test. DNA was extracted from peripheral blood using DNA Blood Mini Kit (250) (QIAamp) according to the manufacturer’s protocol. Variants detected by sequencing servicer agency (Biosan, Zhejiang) were further confirmed by Sanger sequencing with 13 pairs of primers (Supplementary Table S1) in an ABI 3700 automated DNA sequencer using the Big Dye Terminator Cycle sequencing reaction kit (Applied Biosystems)11. Sequences were aligned with the PAH transcript (NM_000277.2) to identify nucleotide variants. The nomenclature was checked with the Mutalyzer program suite (https://mutalyzer.nl/) and followed the HGVS-nomenclature (http://varnomen.hgvs.org/) for description12. After Sanger sequencing, novel missense variants were evaluated with several bioinformatics programs including PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/), SIFT (http://sift.jcvi.org/), Mutation Assessor (http://mutationassessor.org/), Provean (http://provean.jcvi.org/ index.php), Fathmm (http://fathmm.biocompute.org.uk/), and ANNOVAR (http://wannovar.wglab.org/)13,14. Variants predicted to be deleterious based on more than four of the seven parameters were considered putative pathogenic mutations. The annotation version used was Homo sapiens GRCh37.

Genotype analysis

Biallelic mutations were found in 183 cases out of the 209 PAHD patients. According to the PAHvdb database, the effects of variants were sorted as synonymous, missense, nonsense, frameshift and splicing site. Null PAH alleles are defined as those containing nonsense, deletion, frameshift that result in a truncated PAH protein, missense mutations with <3% enzyme activity in vitro compared with the native enzyme, and splice sites8,15. Therefore, genotypes were categorized into three defined types in this study: null/null, null/missense, and missense/missense.

Statistical analyses

Statistical analyses were performed with IBM SPSS Statistics 21. The chi-squared test was used to distinguish between the three genotypes and to analyze the null frequency between the cPKU and mPKU group and between the mPKU and MHP group. A p value of <0.01 was considered statistically significant.

Results

Clinical phenotype of PAHD patients

A total of 420 patients (206 males and 214 females) were diagnosed with PAHD with an incidence of 1 in 20,445 in Zhejiang Province as part of the NBS program between 1999 and 2016. As shown in Table 1, patient phenotypes were classified as cPKU (183 patients, 43.57%), mPKU (139 patients, 33.10%), or MHP (98 patients, 23.33%). The BIOPKU database (http://www.biopku.org) lists the distribution of these three phenotypes as 54.8%, 27.4%, and 17.8%, respectively16. The peak values for pretreatment plasma Phe level varied among the different phenotypes and individual patients with a mean level (µmol/L) of 1,755 for cPKU, 743 for mPKU, and 239 for MHP. No significant correlations were found between phenotype and gender (p = 0.89). During follow-up, blood Phe values of three MHP cases were increase to 389 (µmol/L) in one 3-year-old case, 421 (µmol/L) in one 4-year-old case and 367 (µmol/L) in one 6-year-old case respectively after a high protein diet, and then return to normal range by dietary management. The outcomes of mPKU and cPKU patients here with normal treated Phe levels had similar intelligence quotient (IQ) to control individuals throughout their childhood so far.

Table 1 Characteristics of PAHD patients enrolled in PAH genetic testing.
Full size table

PAH variant and allele distributions

Genetic testing was performed for 209 patients who consented to. The results indicated 392 different PAH alleles with a mean detection rate of 93.80% (Table 1). A total of 72 variants were identified (Table 2), including 46 missense mutations (63.89%), 8 nonsense mutations (11.11%), 10 splice site mutations (13.89%), 4 frameshift mutations (5.56%) resulting from 3 deletions and one duplication, and 4 synonymous variants (5.56%) including the complex one, c.1197A > T (p.V399V)17. Ten truncating mutations were produced by 8 nonsense and 2 frameshift mutations (c.598dupA (p.T200Nfs*6) and c.722delG (p.R241Pfs*100)). Variants were unevenly distributed across PAH (Fig. 1). There were 63 (87.5%) variants found in the exons, including 16 (22.22%) in exon 7, 8 (11.1%) in exon 11, 7 (9.7%) in exon 12, and 6 (8.3%) each in exons 3 and 6. None were observed in exon 13. It appeared that exons 3, 6, 7, 11, and 12 are hotspot regions for PAHD-associated variants and alleles, consisted with the previous evidence7.

Table 2 PAH variants and allele distribution in PKU and MHP patients from Zhejiang Province.
Full size table
Figure 1
figure 1

Seventy-two variants were identified across PAH for PKU and MHP patients from Zhejiang Province. Purple, novel variants identified in this study; Blue, variants identified in MHP patients only. Amino acid position at the start and end of each exon was labeled numerically.

Full size image

Among 72 variants, only 12 variants were identified in MHP patients, including p.R53H, p.F121L, p.P147P, p.R155H, p.R169S, p.R176Q, p.T186I, p.G247S, p.M276K, p.T380M, p.F392I, and c.1200-8G > A. Of these, p.R53H and p.F392I were common variants accounting for 19.3% and 8.0% of the alleles, with a predicted residual enzyme activity of 79% (http://www.biopku.org/pah/result-details-pah.asp?ID=668) and 98% (http://www.biopku.org/pah/result-details-pah.asp?ID=811), respectively. Among the 60 variants recorded for the PKU patients, seven variants with relatively high frequencies were p.R243Q, p.R241C, p.Y204C, p.R111*, c.442-1G > A, p.Y356*, and p.R408Q (26.0%, 14.8%, 9.9%, 6.3%, 6.3%, 3.6%, and 3.3%, respectively). These variants were present in approximately 70% of the alleles for PKU. Moreover, eleven variants—p.R243Q, p.R241C, p.Y204C, c.442-1G > A, p.R111*, p.R408Q, c.707-1G > A, p.G257V, p.R413P, p.R252Q, and p.S70del—were observed for all phenotypic groups (i.e., cPKU, mPKU, and MHP).

Seven novel variants were identified (Supplementary Fig. 1.), including two synonymous variants: c.285C > T (p.I95I) in a cPKU patient and c.441T > C (p.P147P) in an MHP patient. Besides the conservation analysis of amino acid, other five pathogenic nonsynonymous variants were evaluated with bioinformatic programs and predicted as putative functional variants (Supplementary Table S2). Of these, c.107C > A (p.S36*) and c.1238G > A (p.R413H) were observed for cPKU; c.764T > G (p.L255W) and c.904T > G (p.F302V) were observed for mPKU; and c.557C > T (p.T186I) was observed for MHP. According to the PAH protein domains, p.S36* variant localized in the regulatory domain, p.T186I, p.L255W and p.F302V in the catalytic domain, and the p.R413H in the oligomerization domain. The pathogenicity of novel variants should be proved by more PAHD cases and their functional analysis in vitro.

Genotypic distribution for PAHD

In total, biallelic mutations were genotyped in 183 individuals, and of which 85.8% (157/183) were compound heterozygous alleles, and 78.38% of these observed for cPKU, 90.28% for mPKU, and 91.89% for MHP patients. Monoallelic mutations were found and confirmed for the remaining 26 patients, of which 16 had the MHP phenotype. There were 112 different genotypic combinations among the PAHD patients including 8 homoallelic genotypes harbored by 26 individuals (Supplementary Table S2): p.[R111*];[R111*], c.[442-1G > A];[442-1G > A], p. [R241Pfs*100];[R241Pfs*100], and p.[Y356*];[Y356*] were carried by cPKU patients only; p.[R243Q];[R243Q] and p.[Y204C];[Y204C] were carried by both cPKU and mPKU patients, p.[R241C];[R241C] was harbored by both mPKU and MHP patients, and p.[R53H];[R53H] was observed for MHP patients only. The most abundant genotypes in the PAHD patients in this study were p.[R241C];[R243Q], p.[R243Q];[R243Q], p.[Y204C];[R243Q], p.[Y204C];[R241C], p.[R111*];[R241C], p.[R111*];[R243Q], [p.R243Q];[c.442-1G > A], p.[R241C];[R241C], p.[Y204C];[Y204C] and p.[R111*];[R261Q], with frequencies of 8.7%, 7.1%, 3.3%, 2.7%, 2.7%, 2.2%, 2.2%, 2.2%, 2.2%, and 1.6%, respectively (Table 3). Strikingly, none of these genotypes were observed to coexist in any of the three phenotypes. However, seven genotypes were concurrent in both cPKU and mPKU patients, and four genotypes associated with not only the mPKU phenotype but also the MHP phenotype (Supplementary Table S3).

Table 3 The most prevalent genotypes among the PAHD patients.
Full size table

We hypothesized that the genotypes and null alleles would exhibit significant differences in distribution among the three phenotypes. Thus, 29 null PAH alleles were sorted into three defined genotype classes (null/null, null/missense, and missense/missense) as described in previous reports (Supplementary Table S4)8,18. Of the cPKU patients, 77.03% carried at least one null allele, with 54.79% observed for mPKU patients. Furthermore, the null/null genotype was identified in 26 cPKU patients, but only in six mPKU patients (Table 4). The distribution of the three defined genotypes, null/null, null/missense, and missense/missense, were significantly different (p < 0.001) between cPKU and mPKU patients. The null allele frequency also differed remarkably between cPKU and mPKU (p < 0.001). No significant difference was found between mPKU and MHP patients in either the distribution of genotypes or the null allelic frequency (p = 0.21 and p = 0.57, respectively).

Table 4 The three classes of defined genotypes and null allele frequencies associated with the PAHD phenotypes.
Full size table

Discussion

Here, we carried out a retrospective study on samples obtained via neonatal screening for HPA since 1999 in Zhejiang Province of southeast China, illustrating the mutational and phenotypic spectrum of PAHD for a greater understanding of PAH gene variants and their association with PAHD phenotypes.

It is confirmed that the incidence of PAHD for south China is lower than that for Northern China alone (1/3,425–1/7,849)19,20. Compared with the data from the BIOPKU database, the comprising of the three PAHD phenotypes in the present study is different that more than 50% of PAHD cohort from Zhejiang Province comprises mPKU and MHP patients as reported in Denmark PAHD group20.

The identified variants included 63.89% missense mutations, 13.89% splice site mutations, and 11.11% nonsense mutations as previously observed21. Exons 3, 6, 7, 11, and 12 appeared to be hotspots for PAHD-associated variants and alleles, which was consistent with previous reports for Chinese and Asian populations7,22,23,24. However, phenotype frequency depends on the allele frequency of particular PAH variants. Among the PKU patients, 70% of the allele frequency was contributed by only 7 mutations. The most frequent allele was p.R243Q (22.19%), followed by p.R241C (13.27%), p.Y204C (8.93%), p.R111*(5.36%), c.442-1G > A (5.61%), p.Y356* (3.06%), and p.R408Q (2.81%), but excluded the p.V399V and p.R413P mutations which common in Northern China22. Notably, p.R241C seemed to be present with higher incidence in mPKU and MHP patients from the south of China than that described for other regions, which may be represent the expansion or migration of the certain populations7,11. The classic PKU mutation c.1222C > T (p.R408W), the most frequent allele in Eastern Europe, was carried by only one cPKU patient in our cohort who also had a 1,706 µmol/L maximum pretreatment Phe level. The p.R53H and p.F392I variants observed for MHP patients with a Phe value of <360 (µmol/L) were present in 27% of the alleles and 49% of the MHP patients (Table 2). While p.R53H is known to be harbored by Japanese PKU patients25, it is more common in the general Korean population with a frequency of 2.57%26,27. Since these two mutations are observed in healthy subjects and they retain >70% residual PAH activity, p.R53H and p.F392I are classified here as mutations associated with MHP. Another notable finding was that 30% of the MHP patients (16/53) with abnormal Phe values observed during newborn screening were carriers of a pathogenic allele. Genetic counseling should be provided to such patients prior to conception.

The genotypes in our PAHD cohort were highly heterogeneous as >60% of the patients presented with a unique genotype, and 85.8% were compound heterozygotes compared with 76% reported in the BIOPKU database16. The p.[R241C];[R243Q] (8.7%) genotype was the most prevalent, followed by p.[R243Q];[R243Q] (7.1%), p.[Y204C];[R243Q] (3.3%), p.[Y204C];[R241C] (2.7%), and p.[R111*];[R241C] (2.7%), which is in accordance with earlier findings that p.[R243Q];[R243Q], p.[Y204C];[R243Q], and p.[R241C];[R243Q] constitute the major genotypes in southern China8. Interestingly, two p.R241C homozygous patients showed MHP, which was also detected in Korean PAHD population before23. The distribution of null alleles and the genotypes null/null, null/missense, and missense/missense showed significant differences between cPKU and mPKU patients as well as between cPKU and MHP patients. However, no significant differences were observed for the distribution of either between mPKU and MHP patients. This indicates that other modifier factors may influence PAHD phenotype, especially for mPKU and MHP. It is therefore unsurprising that several genotypes are shared by both mPKU and MHP phenotypes.

Certainly, the specific PAH genotype is key for determining the metabolic phenotype. More than 80% (26/32) of the double-null genotypes correlated to cPKU phenotype in the present study. Interestingly, MHP and mPKU phenotypes have similar frequencies of null alleles in genotypes. This implies that the interaction between compound heterozygous alleles play a major role in phenotypic outcome28. Additionally, we confirmed that in MHP patients, mild mutations determine disease severity. Genotypes comprising a combination of p.R58H with one of the null alleles p.R111*, c.442-1G > A, p.Y204C, p.Y356*, or p.L255S, resulted in the MHP phenotype. A similar phenomenon was observed for the p.F392I/null genotype associated with the MHP phenotype, although further samples are needed to confirm this, and the effect on PAH structure and function needs to be considered for further insights. In conclusion, the data presented in this study will provide a valuable tool for improved genetic counseling and management of future patients of PAHD in China.