Introduction

Phenylketonuria (PKU; MIM#261600) is a genetic disorder resulting from the deficiency of phenylalanine hydroxylase (PAH). To date, more than 500 PAH mutations have been reported (online PAH database, http://www.pahdb.mcgill.ca/). These mutations vary in their impact on PAH enzymatic activity that lead to a range of clinical phenotypes from severe PKU to mild hyperphenylalaninemia (MHP). The spectrum of PAH mutations has been studied worldwide and a broad correlation of clinical phenotypes with mutational genotypes has been found.1 Moreover, various types of polymorphic makers in the PAH locus have been reported including single-nucleotide polymorphism (SNP), variable number of tandem repeats (VNTRs) motif and a short tandem repeats (STRs) polymorphism (online PAH database, http://www.pahdb.mcgill.ca/). A prevalent PAH mutation in an area is generally shown to be the founding mutation that strongly associates with a particular polymorphic haplotype,2 which has been used to understand origins and patterns of population migrations.3, 4, 5 Moreover, the number of different PAH mutations identified have been shown to reflect the ethnic complexity in a population.6

Present-day Taiwanese residents came from multiple origins.7, 8 In addition to Austronesian aboriginals who settled in Taiwan dating back to many thousands of years ago, there were also descendants from southeast coast of China in the last few centuries, and the more recent immigrants from mainland China after the Second World War. Other ethnic groups, such as Spanish, Dutch and Japanese, have resided for short periods of time in the last 400 years. Chien et al.9 have reported PAH mutations in 25 PAH-deficient families in Taiwan. They identified 18 different PAH mutations amongst which the R241C mutation accounted for 32% of the mutant chromosomes, and was proposed to be a founder mutation in the Taiwanese population.

To improve the representation of PAH mutations in Taiwan, we expanded the sample size to 71 unrelated PKU families in this study. In addition to the mutation spectrum, polymorphic haplotypes associated with prevalent PAH mutations were examined.

Materials and methods

Patients

Seventy-one PAH-deficient families in Taiwan were included. Majority (63 families) of the subjects were identified through neonatal screening. The remaining subjects were from schools of special education (4 families) and from outpatient clinics (4 families). BH4 deficiency has been ruled out in these patients. No consanguinity was known among PKU families. Informed consent for genetic analysis was obtained according to the regulation of Taipei Veterans General Hospital.

Mutation analysis

Genomic DNA was extracted from leukocytes by a described method.10 The PAH genomic sequences including promoter, 13 exons and their flanking exon–intron junctions were PCR amplified, and the sequence was determined using the Big Dye Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA, USA). Moreover, differential PCR was performed to detect mutation −4173_−407del, a large deletion occurred in the PAH enhancer.11 When sequence alteration was identified, paternal or maternal origin was determined whenever the parental DNA was available.

Analysis of VNTR, STR, SNP and RFLP haplotype

Structure of VNTR motif was analyzed by a PCR-based method.12 The number of repeats was deduced through size determination of the PCR products by agarose gel electrophoresis using PCR products of known repeat number as a reference. To determine VNTR cassette sequences, the PCR product was sequenced directly if the sample was homozygous for the VNTR genotypes; if the samples were of heterozygote types, allelic PCR products were first separated and purified by agarose gel electrophoresis and gel band purification kit. The STR number variation was determined by PCR-based microsatellite genotyping with the fluorescent dye FAM attached to the 5′-end of the reverse primer.13, 14 PCR products were analyzed using the ABI PRISM 377 automated DNA sequencer (Applied Biosystems). STR genotypes were defined and edited using the Applied Biosystems GeneScan with PCR products of known repeat number as references. To analyze diallelic restriction fragment length polymorphism (RFLP), in addition to the Southern blot method,10 the BglII, PvuII(a), PvuII(b), MspI, XmnI and EcoRI restriction enzyme sites of DNA samples were analyzed by the PCR-based method using primers described.3 The PCR products were sequenced directly to determine the status of enzyme sites except EcoRI recognition sequence, which was determined by EcoRI digestion. Moreover, during the PAH mutation screening, intragenic SNPs including four in introns, IVS2+19T/C, IVS3−22T/C, IVS4+47C/T and IVS10+97A/G, and three in exons, c.696G/A (Q232Q), c.735A/G (V245V) and c.1155G/C (L385L), were recorded accordingly.

PAH activity analysis

An expression plasmid carrying the PAH cDNA (complementary DNA) sequence under the transcriptional control of SV40 promoter was used for PAH activity assay. Mutation was introduced into the PAH sequence by the QuikChange site-directed mutagenesis (Stratagene, La Jolla, CA, USA). All constructs were confirmed by DNA sequencing. The expression plasmids were individually transfected into COS-1 cells by the calcium phosphate method.15 PAH activity was assayed in crude extracts of the transfected cells by measuring the conversion of L-[14C]phenylalanine to L-[14C]tyrosine using either natural cofactor BH4 (6R-5,6,7,8-tetrahydrobiopterin, Schircks, Jona, Switzerland) or synthetic cofactor 6MPH4 (6-methyl-5,6,7,8-tetrahydropterin, Sigma, St Louis, MO, USA).16 Residual PAH enzymatic activity of a particular allele was determined from at least three sets of lysates of transfected COS-1 cells using independently prepared plasmid DNAs.

Analysis of RNA splicing efficiency

To study the effects of IVS7+5G>A base substitution on RNA splicing, modified cDNA expression plasmid carrying the 1058-bp PAH intron 7 was constructed. To this end, PAH genomic DNA fragment comprising intron 6 to exon 8 was generated by PCR amplification using the forward primer, 5′-CTCCTAGTGCCTCTGACTCA-3′, located in intron 6, and reverse primer, 5′-TCCTTACCTGGGAAAACTGGG-3′, located at the junction of exon 8 and intron 8. Subsequently, the DNA fragment containing intron 7 and the flanking exonic sequences was obtained by co-digesting the PCR product with restriction enzymes BamHI and AflIII. The fragment obtained was used to replace the corresponding BamHI/AflIII fragment on PAH cDNA to generate a modified expression plasmid carrying PAH cDNA and intron 7 of the PAH gene. Northern blot analysis of PAH RNA was performed on RNA from such plasmid-transfected COS-1 cells.

Phenotype determination

Patients were assigned to three phenotype categories mainly according to the plasma phenylalanine (Phe) levels before clinical treatment: classical PKU (Phe⩾1200 μmol l−1), mild PKU (Phe=600–1199 μmol l−1) and MHP (Phe=240–599 μmol l−1).6 Data on dietary phenylalanine tolerance were used as references when available.1

Results and discussion

Identification of PAH mutations

The PAH DNA sequences including enhancer, promoter, 13 exons and flanking exon–intron junctions were determined by sequencing of PCR-amplified genomic DNAs from probands of 71 unrelated PKU families. Mutations were identified on both alleles in 68 families and on only a single allele in 3 families. Thus, mutations were determined in 139 of 142 PAH mutant chromosomes, a detection rate of 98%. The mutations represent 34 different types, including 20 missense mutations, 4 nonsense mutations, 4 deletion/insertion within structural gene, 1 deletion in enhancer region and 5 affecting splicing (Table 1). The most prevalent mutations in Taiwan are R241C, R408Q and Ex6-96A>G accounting for 23.2%, 12.0% and 9.2% of the 142 mutant chromosomes, respectively. The 17 mutations were found only once.

Table 1 PAH mutations identified in Taiwan and relative frequencies
Full size table

Functional analysis of novel PAH mutations

Three previously unreported alterations, V190G, Q267R and F392I, were found in this study in the Taiwanese population. To access functional significance associated with such changes, residual PAH activity from COS-1 cells transfected with the respective PAH cDNA expression plasmid was examined (Figure 1). The results showed that PAH activity expressed from PAH sequences harboring V190G or Q267R had lower than 2% of the wild-type enzymatic activity level, indicating that these alterations could result in a severe phenotype. Indeed, patients harboring the V190G mutation in functionally hemizygous state (V190G/E228X) displayed classical PKU phenotype (Table 2, genotype 6). Likewise, patient carrying compound heterozygous mutation of Q267R/G257V manifested classical PKU phenotype (Table 2, genotype 20). On the other hand, PAH expressed from the F392I allele showed about half the wild-type activity when assayed with the synthetic cofactor 6MPH4 (Figure 1a, right panel) whereas reaching the wild-type level when assayed with natural cofactor BH4 (Figure 1a, left panel).

Figure 1
figure 1

Effects of PAH mutations on PAH enzymatic activities. PAH cDNA expression plasmid carrying a defined substitution was transfected to COS-1 cells. Constant amounts of crude extract from transfected cells were aliquoted for determination of PAH enzymatic activities using BH4 (a, left panel, and b) or 6MPH4 (a, right panel) as a cofactor. The substrate L-[14C] phenylalanine (Phe) and the product L-[14C] tyrosine (Tyr) were separated on a TLC plate and visualized using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA, USA). PAH enzymatic activities were calculated as cpm of L-[14C] Tyrosine/(cpm of L-[14C] Phenylalanine+cpm of L-[14C] Tyrosine), where the mutant enzymatic activity was expressed as a percentage of that of the wild type (WT). Mutations are indicated at the top. cDNA, complementary DNA; PAH, phenylalanine hydroxylase.

Full size image
Table 2 Genotype–phenotype association
Full size table

In addition to being a coding sequence, exon has been shown to harbor cis-regulatory elements important for exon recognition during pre-mRNA splicing.17 In fact, we have shown that about one-quarter of naturally occurring mutations on the human PAH exon 9 affected RNA processing.16 Likewise, we found that the silent mutation V399V (c.1197A>T) affects exon-splicing enhancer element, resulting in exon 11 skipping during pre-mRNA splicing.18 To assess whether alteration of F392I (c.1174T>A) affects PAH exon 11 recognition, a study similar to that of V399V (c.1197A>T) was carried out. No significant effects of such an alteration on pre-mRNA splicing were found (data not shown). Clinical manifestation of patients carrying the F392I allele was MHP (Table 2, genotype 40, F392I/R408Q and genotype 41, F392I/?). As the associated PAH allele in these two patients was either a mild R408Q mutation or one of unknown identity, it was difficult to assign phenotype to F392I conclusively. It is noteworthy that no other mutations could be identified in the coding region of the PAH gene in these two patients, and none of the F392I (c.1174T>A) substitution was found among the 108 control alleles examined (data not shown), suggesting that the substitution may not be a neutral polymorphism but a mild PAH mutation.

Further functional analysis of selected PAH mutations

To study the relationship between genotype and clinical phenotype of PAH mutations, knowledge of the residual activity of a mutant gene was important. In this aspect, nucleotide alteration predicted to completely abolish PAH activity, that is, nonsense and frame-shift mutations, and splice-site mutations at both the 5′- and 3′-end of an intron are counted as null. Also considered null in gene activity is the −4173_−407del deletion in the PAH enhancer region that has been shown to severely impair PAH hepatic transcriptional activity.11

To improve on accurate assignment of mutational phenotypes, several mutations were chosen for analysis of in vitro activities. First, residual enzyme activities of a set of missense mutations in transfected COS-1 cells were determined (Figure 1b) and their relative activities (RA), that is, percent of in vitro activity relative to that of wild-type control, were determined (Table 2). It is to be noted that residual enzyme activities of R261Q, P314T, R408Q and R408W were not assayed in this study but were taken from published reports.16, 19, 20, 21 We found that in vitro residual enzyme activities associated with mutations can vary considerably to that of published reports. For example, PAH protein carrying the R241C substitution showed 25% of residual activity in the COS cell expression system in a report by Okano et al.22 however, we obtained a result of 49% for this mutant in our study. This may, in some extent, be due to the mutant protein manifesting variable residual activities dependent on assay conditions, such as substrate concentration, or the type of cofactors used. In this regard, the residual activity of PAH carrying F392I was shown to vary considerably depending on whether cofactor BH4 or 6MPH4 was used (Figure 1a). Therefore, the natural state of some mutant enzymes may not be able to be accurately reproduced in vitro.

Likewise, we examined how severe the IVS7+5G>A base substitution that alters intron 7 sequence outside GT–AG dinucleotides affected pre-mRNA splicing. COS-1 cells were transfected with a modified PAH cDNA expression plasmid carrying the intron 7 sequence (Figure 2a). Northern blot analysis of PAH RNA showed that when a construct carrying the G–A substitution at the fifth nucleotide of intron 7 (IVS7+5G>A), pre-mRNA splicing was completely abolished; only the 3.5-kb primary transcript was detected (Figure 2b, lane 1). In contrast, properly spliced mRNA of 2.4 kb, in addition to the 3.5-kb primary transcript, could be detected if a construct carrying the wild-type intron sequence was used (Figure 2b, lane 2), suggesting that the IVS7+5G>A substitution severely impairs PAH pre-mRNA splicing and hence the PAH enzymatic activity (Figure 2c, lane 1). Thus, the IVS7+5G>A mutation is classified as null. It is noted that the IVS7+5G>A base substitution does not involve the invariant GT dinucleotides at the splice-donor site, and yet such a substitution severely affects pre-mRNA splicing. This may be because the splice-donor sequence of intron 7 is ‘weak’ (CC/GTGAGT, where the slash ‘/’ indicates the exon–intron border) in that it matches only six bases of the eight-base consensus sequence (AG/GTRAGT, where R=purine). In such a configuration, a base change in the consensus sequence may lead to near-complete abolishment of pre-mRNA splicing.23

Figure 2
figure 2

Effects of the PAH IVS7+5G>A mutation on PAH activity. (a) Schematic representation of a PAH expression plasmid under transcriptional control of SV40 promoter (SV) and polyadenylation signal (pA). The bent arrow indicates transcriptional start site. The PAH sequence includes exons 1 to 13 and intron 7 (Int 7); the nucleotide sequences flanking the 5′ splice junction of intron 7 are shown with lowercase letters representing the intron sequence. The IVS7+5G>A mutation carrying nt 5 mutation in intron 7 is highlighted. (b) Northern blot analysis. Total RNA isolated from COS-1 cells transfected with PAH expression plasmid with intron 7 carrying the IVS7 nt 5 mutation (IVS+5G>A) (lane 1), intron 7 carrying the wild-type (WT) sequence (lane 2), untransfected COS-1 control (lane 3) or wild-type PAH cDNA (lane 4) were hybridized with P32-labeled PAH cDNA probe (upper panel) and subsequently to GAPDH cDNA probe (lower panel) after stripping. The unspliced and spliced PAH mRNAs are indicated. (c) PAH enzymatic activities assay. Constant amounts of crude extract from COS-1 transfected cells were aliquoted for determination of PAH enzymatic activities using 6MPH4 as a cofactor. The substrate L-[14C] phenylalanine (Phe) and the product L-[14C] tyrosine (Tyr) were separated on a TLC plate and visualized using a PhosphorImager. The samples in each lane are as in lanes 1–3 in (b). cDNA, complementary DNA; PAH, phenylalanine hydroxylase.

Full size image

Correlation of mutation genotype to clinical phenotype

The 34 different PAH mutations identified are variously combined into 50 genotypes in the 71 PKU families studied (Table 2). Homozygous mutations were found in 10 families, which gave an observed frequency of homoallelic PKU genotypes of 14% (Table 2, genotypes 1–4). On the basis of the pretreated plasma phenylalanine levels also taken as references to the phenylalanine tolerance data when available, three clinical phenotypes, that is, classical PKU, mild PKU and MHP, were assigned.1, 6 Among the 71 PKU families, 22 were assigned to classical PKU, 26 to mild PKU and 19 to MHP, whereas, 4 families were not assigned due to lack of sufficient clinical data (Table 2). In our PKU cohort, patients possessing the same mutation genotype were found to fall mostly into a single clinical category; the exception was genotype E286K/R241C, which was assigned to the categories of mild PKU and MHP (Table 2, genotype 26). Nevertheless, more cases are needed to firmly establish such a correlation, especially when many genotypes occurred in a single case.

It is noted that a number of mutations, although showed substantial in vitro activities, were found to manifest severe clinical phenotypes. As an example, PAH carrying the R243Q substitution showed 13% residual enzyme activity in our study or 10% in the study of Okano et al.22 Yet, this mutation appears to be associated with a severe phenotype as deduced by its homozygous state (Table 2, genotype 3) or functionally hemizygous state (Table 2, genotypes 10 and 30). Likewise, substantial enzymatic activities were shown for R297H and R261Q, and yet patients carrying this allele displayed severe clinical phenotype (Table 2, genotype 31, R297H/R413P and genotype 35, R261Q/?). It is noteworthy that patients carrying R261Q or R297H were identified at teenage when symptoms of mental retardation were apparent. More studies such as measurement of their phenylalanine tolerance level are needed to re-evaluate phenotype assignments. In particular, the R297H mutation in functional hemizygous state in a PKU cohort of Poland was found to be associated with the MHP phenotype.24 One would envision that if a mutant protein is displaying kinetic instability, the protein can be stabilized in vitro if substrates or cofactors are exceeding physiological concentrations. Therefore, to make a better correlation between residual enzyme activities and clinical severity, detailed biochemical characterization such as determinations of specific activity, substrate activation, enzyme kinetics and stability of the mutant PAH proteins may be required.25, 26

On the basis of the phenotype of the patient and data of in vitro expression study, phenotype of PAH mutation was assigned (Table 1). Among the 34 mutations, 23 were assigned to classical PKU, 2 to mild PKU, 5 to MHP and 4, namely, S70del, R261Q, R297H and V388M, remained to be assigned. It is noted that patients displayed classical PKU phenotype when carrying the S70del and R261Q mutations in their functionally hemizygous state, or R261Q in a homozygous state according to the study of Song et al.27 Therefore, both S70del and R261Q are likely associated with severe PKU phenotype. As for the V388M mutation, the mutant protein showed 18% of the normal enzyme activity in our work (Figure 1b), in agreement with 15% of a mutant in the study of Okano et al.,22 which is known to be a kinetic variant.28 PKU patients in Japan and Korea carrying V388M in their functionally hemizygous state showed classical PKU.29, 30, 31 Thus, the V388M mutation is most likely associated with severe phenotype. In line with this, mild PKU phenotype was observed in one patient in our PKU cohort carrying compound heterozygous mutation of V388M/R241C (Table 2, genotype 45). Furthermore, when phenotype assignments of PAH mutation were compared among Oriental PKU cohorts,22, 27, 29, 30, 31 a good agreement was found among the available mutations that were studied, except for that of R241C. The R241C mutation was found to be associated with mild PKU phenotype in our cohort as well as in that from Korea and Japan.29, 30, 31 However, in China as studied by Song et al.,27 R241C was found to be associated with mild PKU in three patients carrying the compound heterozygote mutation of R241C/R243Q, or with MHP in one patient with the compound heterozygote mutation of R241C/V399V. It is conceivable that clinical phenotype of R241C may vary depending on residual activity of the associated allele when present in a near-functional hemizygous state. It is noted that none of the patients from this study and that of Korea and Japan carried R241C/V399V genotype. In summary, among the PAH mutations identified in this study, classical PKU type occupied the highest proportion (23 out of 31 mutations, 74%), but more than 60% of the PAH-deficient families (45 out of 67 families) belonged to milder forms probably due to the fact that the most common R241C and R408Q mutations belong to mild forms of PAH deficiency.

On the other hand, Guldberg et al.1 have established a system for genotype-based prediction of metabolic phenotype using PKU cohort of Europe. They assigned each of the 105 PAH mutations an ‘assigned value’ (AV) to one of four phenotypic classes, that is, mutation-causing severe PKU (AV=1), moderate PKU (AV=2), mild PKU (AV=4) and MHP (AV=8), based on phenotypes of PAH mutations in association with functionally hemizygous state. These authors further demonstrated that in the majority of cases studied, phenotypic effects of two mutant PAH alleles could be determined by the sum of their AVs: classical PKU, AV=2; moderate PKU, AV=3; moderate/mild PKU, AV=4; mild PKU, AV=5–6; mild PKU/MHP, AV=8; and MHP, AV=9–16. To examine whether such a system is valid in our PKU cohort, phenotype assignment of the PAH mutations by Guldberg et al.,1 if available, is included in Table 2. The results showed that observed clinical phenotype matched the predicted phenotype in patients, except those carrying the EX6-96A>G and R241C alleles (Table 2). EX6-96A>G was assigned as moderate PKU by Guldberg et al.,1 yet patients homozygous (Table 2, genotype 1) or functionally hemizygous for EX-96A>G (Table 2, genotypes 18 and 19) manifested classical PKU phenotype in our cohort as in patients from China, Korea and Japan.27, 29, 30, 31 Likewise, R241C, the most frequent mutations in Taiwan, has been assigned by Guldberg et al.1 as MHP. However, the majority of our patients in or near functionally hemizygous state of R241C (Table 2, genotypes 12, 13, 24, 25 and 42) belonged to the mild PKU category. Mutations EX6-96A>G and R241C are prevalent among oriental populations,22, 27, 29, 30, 31, 32 yet are rare in European populations.1 Assignment of EX6-96A>G and R241C by Guldberg et al.1 was done from only one allele in a PKU patient, conceivably leading to discrepancy due to a small number of patients being analyzed. It is also possible that the R241C mutation may manifest MHP or mild PKU phenotype dependent on the associated allele as shown in the study of Song et al.27 Thus, in the majority of PKU mutations under study, a good agreement of phenotype assignment was found not only in Oriental PKU cohort but also in that of Caucasians, indicating that disease severity is primarily determined by PAH genotype as have also been suggested by other studies.1, 22

Genetic diversity of the common PAH mutations in Taiwan

In this study, mutations R241C, R408Q and EX6-96A>G were found to be most common in Taiwan, accounting 23.2%, 12.0% and 9.2% of the mutant chromosomes, respectively (Table 1). To study whether their prevalence was due to enrichment through the founder effect or the result of the recurrent mutation, the associated haplotypes of these mutant alleles were determined by analysis of RFLP, VNTR, STR and seven intragenic SNP markers (Table 3).

Table 3 Genetic diversity of the PAH common mutations in Taiwan
Full size table

The R241C mutation involving a CpG dinucleotide could be a mutational hotspot. Among the 33 chromosomes harboring R241C, all were found to be associated with RFLP haplotype 4 (Table 3), a major haplotype in both wild type and mutant alleles in Taiwan.10 This haplotype was found to be associated with three-repeat VNTR (VNTR 3) with a cassette sequence of a2-b2-c1, designated as haplotype 4.3.33 Variations on the associated STR number, including 234, 238 and 242, were observed for R241C with unimodal distribution centered on the 238 allele (Table 3). In contrast, a greater STR allelic diversity was found in 49 wild-type RFLP haplotype 4.3 alleles examined, where the STR variation exhibited a broad unimodel distribution with the 238 and 242 alleles at comparable frequencies (35% vs 49%) (Table 3). The study by Chien et al.9 have found that R241C is the most common PAH mutation in Taiwan suggesting a strong founder effect. In our studies of RFLP haplotype, VNTR configuration and SNPs pattern, we showed that all R241C alleles (n=33) fell into the same PAH chromosomal background to support that they were identical by descent (Table 3). However, variations on STR allele distribution were observed (Table 3). The STR loci evolve rapidly and have been shown to have high mutation rates.34 The association of the R241C mutation with chromosomes of various STR numbers suggests that R241C has occurred rather early in population history. Nonetheless, one cannot rule out that R241C has occurred independently in this chromosomal background, especially when haplotype 4 is the most common haplotype among Chinese.10 The R241C allele is also found in China, Korea and Japan with reported frequencies of 3.7%, 5.7% and 10.1%, respectively.30, 31, 32 It would be of interest to study their associated haplotypes to explore whether they originated from the same founder.

R408Q that accounted for 12% of mutant alleles in Taiwan has been reported to be in linkage disequilibrium with PAH RFLP haplotype 4.35 Interestingly, in our PKU cohort, 2 among a total of 17 R408Q alleles studied were found to be associated with VNTR of five repeats (VNTR 5) of the cassette sequence (a1)3-b5-c1 (Table 3). The RFLP haplotype of this allele was deduced to be either haplotype 11 or 34 depending on the status of the EcoRV recognition sequence, which cannot be analyzed by the current PCR-based method. Genotype of intragenic SNP markers including four in introns IVS2+19T/C, IVS3−22T/C, IVS4+47C/T and IVS10+97A/G, and three in exons c.696G/A (Q232Q), c.735A/G (V245V) and c.1155G/C (L385L), were determined by sequencing the PCR-amplified genomic DNA. Substantial SNP variations between R408Q alleles associated with VNTR 3 and VNTR 5 were found (Table 3). Apparently, the R408Q mutation that appeared on two chromosomal backgrounds is most likely due to the recurrent mutation. In contrast, the EX6-96A>G mutation that was associated with the RFLP 4.3 haplotype appeared in a single SNP haplotype with STR of 242, and is likely derived from a founder mutation (Table 3). It is noteworthy that although the major R408Q allele was found to associate with chromosome of the RFLP haplotype 4.3 similar to that of R241C and EX6-96A>G, R408Q SNP patterns at IVS3−22 and IVS4+47 are different (Table 3). R408Q is the only mutation associated with the minor subtype of RFLP haplotype 4.3.

Ethnic heterogeneity in the Taiwanese population

Among the 34 different PAH mutations identified, nine mutations, R241C, R408Q, EX6-96A>G, E286K, R413P, R243Q, −4173_−407del, R111X and V399V, accounting for 70% of all mutant alleles, showed a relative frequency of 3% or more and are regarded as prevalent. Amongst them, R241C, EX6-96A>G, R413P, R243Q and R111X are also considered as prevalent in Orientals, whereas, V399V reaches a relative frequency of 3.2% in Chinese Han population.32 The remaining mutations, R408Q (12.0%), E286K (5.6%) and −4173_−407del (3.5%), accounting for 21% of all mutant alleles in Taiwan, are either very rare or undetected among PKU cohorts of Chinese, Koreans and Japanese,22, 27, 29, 30, 31, 32 suggesting local founder effects. Therefore, the PAH mutation spectrum in Taiwan shows remarkable differences despite displaying various links to other Asian regions.

On the other hand, the R408W mutation, the most common European PKU mutation, is rare among PKU patients of Chinese, Koreans and Japanese ethnic groups,22, 27, 29, 30, 31, 32 yet R408W accounts for 2.8% of mutant alleles (n=4) in Taiwan. Particularly, this allele was found to be associated with haplotype 44 (n=2), 41 (n=1) and 4 (n=1).35, 36 The high recurrent rates of R408Q (c.1223G>A) and R408W (c.1222C>T) have been attributed to CpG dinucleotide methylation.37 The association of the R408Q and R408W alleles with various haplotypes supports the complexity of general population in Taiwan. On the other hand, homozygosity value at the PAH locus in a given population (determined by j=Σxi2, where xi is the frequency of the ith allele) has been used as an indicator of population heterogeneity.6 The homozygosity value was calculated to be 0.092 in Taiwanese, which is comparable to the value of 0.089 in Chinese.32 Such a low value supports the high degree of ethnic heterogeneity in the Taiwanese population. In this aspect, Lin et al.38 based on a HLA study, showed that the main ethnical constituents of the population in Taiwan, namely Minnan and Hakka, are mainly the descendants from tribal groups of the southeast coast of China with extensive intermarriage to the aboriginal people of Taiwan. Therefore, in addition to the shared ancestry between Chinese Han and Taiwanese populations, our finding supports the notion of a unique ancestral origin of the residents of Taiwan.38

Concluding remarks

Taiwan, a small island, has been at the crossroads of population movements.7, 8 Researches based on studying genetic variations in human gut bacteria Helicobacter pylori, and on linguistic evidences, indicate that most Pacific populations originated from Taiwan ∼5000 years ago.39, 40 However, by examining genetic signature on mitochondrial DNA, Soares et al.41 found that Polynesian people may have spread to the Pacific via New Guinea rather than via Taiwan as previously thought. In this work, we showed that the Taiwanese population has a unique spectrum of PAH mutations. It would be of interest to further study the PAH mutation spectrum in Pacific populations to clarify their origins and migration pattern in relation to Taiwan.