Inheritance in Erythropoietic Protoporphyria: A Common Wild-Type Ferrochelatase Allelic Variant With Low Expression Accounts for Clinical Manifestation

Five Caucasian EPP families (Fig 1) were selected on typical clinical and laboratory criteria: the index cases had a classical history of skin photosensitivity and decreased FECH activity in lymphocytes.13 Each parent was characterized either as a transmitter (ie, bearing the deleterious FECHmutation with related decreased FECH enzyme activity) or as a normal parent (ie, with normal FECH enzyme activity). Heterozygosity for at least 1 of the 2 exonic dimorphisms (798G/C and1520C/T) was a prerequisite for relative FECH mRNA quantitation. Otherwise, ribonuclease protection assay was used in selected cases for absolute quantitation.

(1) Thirty-nine unrelated French Caucasian nuclear EPP families (including the above-described 5 families [77 subjects]) were investigated. The patient group consisted of 39 patients with a documented clinical and laboratory history of EPP. The transmitter parent group consisted of 22 asymptomatic FECH gene carrier parents. The normal parent group consisted of 16 parents with normal FECH enzyme activity. (2) The control group consisted of 70 healthy unrelated French Caucasian subjects. The allelic distribution of the 5 polymorphisms satisfied Hardy-Weinberg equilibrium (χ²test, P > .5). Procedures involving human subjects were performed in accordance with the Helsinki declaration revised in 1983, and informed consent was obtained from all subjects before their inclusion in the study.

The specific FECH gene defect in EPP family 1 has been previously described.9 In families 2 through 5, the specific FECH gene mutations were characterized by direct sequencing of FECH cDNA and further confirmed at the genomic DNA level. These four mutations were different and not previously described ones (Fig 1). The only missense mutation, the T⁷⁹⁰→C transition in family 4 leading to a serine to proline substitution at position 263 (S263P), was expressed in a prokaryotic system (pGEX-2T vector; Pharmacia Biotech, Uppsala, Sweden) using site-directed mutagenesis (Transformer site-directed mutagenesis kit; Clontech Laboratories, Palo Alto, CA) and showed a residual FECH enzyme activity at 0.5% ± 0.02% (mean ± 2 SD) as compared with 100% for the normal cDNA.

Five intragenic polymorphisms distributed over the FECH gene were studied. DNA was prepared from lymphoblastoid cell line as previously described.12 −251A/G in the 5′ promoter region,IVS1−23C/T,14,15 close to the branch point of intron 1, a dinucleotide repeat in intron 2 calledIVS2μsat:An_{n: 1→10}, 798G/C in exon 7, and 1520C/T in the 3′ UTR. Genotyping for −251A/G, 1520C/T, and IVS2μsat:An was performed as previously described.12 16 TheIVS1−23C/T dimorphism was screened as follows. The intron 1/exon 1 junction was amplified by polymerase chain reaction (PCR) and then digested by the restriction enzyme Cac8Iaccording to the manufacturer’s instructions (New England Biolabs, Beverly, MA). The −23T allelic fragment is specifically cut by this enzyme. The 798G/C dimorphism in exon 7 was screened as follows. After PCR amplification, exon 7 is digested withNla III restriction enzyme (New England Biolabs) and the 798C allelic fragment is cut into two parts.

Total RNA isolated from a lymphoblastoid cell line17 was reverse transcribed as described.12 Two regions ofFECH cDNA were amplified spanning nucleotides +693 to +1273 and +1328 to +1568. Primer extension assay was performed as previously described.12,18 19 Two fluorescent primers were used: 798F, 5′-flTTTTCTGCTCACTCACTG-3′; and 1520F, 5′-flCCTTTAATAAATGAGTTAA, respectively, located close to polymorphic loci 798G/C (+693/+1273 amplimer) and1520C/T (+1328/+1568 amplimer). Extended primers were separated on a Alf DNA automated sequencer (Pharmacia Biotech) and peak areas were calculated using the Fragment Manager program (Pharmacia Biotech). Results are expressed as the peak area ratios (R) of the corresponding allelic FECH mRNAs: R₇₉₈ = (peak area ofFECH mRNA bearing a 798G allele)/(peak area ofFECH mRNA bearing a 798C allele); and R₁₅₂₀= (peak area of FECH mRNA bearing a 1520C allele)/(peak area of FECH mRNA bearing a 1520T allele).

In some cases, FECH mRNA was quantified by ribonuclease protection assay as previously described.12 In family 5, the specific gene mutation (IVS9⁻¹ g → a) leads to exon 10 skipping, allowing a FECH probe that include exon 10 to be used for quantifying the normal allele independently to the mutated one. Two RNA antisense probes were synthesized: aFECH probe containing the last 414 FECH cDNA nucleotides cohybridized with an L ferritin control probe containing the first 125 L ferritin cDNA nucleotides to correct experimental variability. Results are expressed as the ratio (R) of the normal FECH mRNA to the L ferritin mRNA.

Genotype and allele distribution was analyzed using the χ² test. Odds ratio and 95% confidence intervals (CI) were calculated.20 The high prevalence of theIVS2μsatA9 allele resulted in a high percentage of homozygotes. The data were then collapsed and analyzed in 2 × 2 format using χ² statistics. Linkage disequilibrium (D) studies were performed according to Thomson et al.21 

Five EPP families with dominant inheritance were studied (ie, partial decrease in FECH enzyme activity associated with a specificFECH mutation in one of the FECH alleles in the proband and in one parent; Fig 1). In these 5 families, 5 differentFECH mutations were found, and 4 were newly described (Fig 1). Allelic segregation patterns and relative RNA quantitation showed that clinical expression in EPP patients in these families always resulted from the coinheritance of a wild-type low-expressed allele with a mutant one. Haplotyping using 5 intragenic polymorphisms spread over the 45 kb of the FECH gene from the 5′ promoter region to the 3′ untranslated region (3′ UTR) (−251A/G,IVS1−23C/T, IVS2μsat:An, 798G/C, and1520C/T) showed that this low-expression variant was (1) systematically transmitted from the normal parent to the patient and (2) associated with two haplotypes sharing the same 5′ part [G-T-A9] (Fig 1). Not one of these sequence variations was by itself directly involved in the low expression, as attested by allelic segregation in family 1. The normal parent I₁2, although homozygous for the 5′ partial [G-T-A9] haplotype, transmitted a low-expressed allele to the proband (II₁2) but a normally expressed allele to another sibling (II₁1), an asymptomatic carrier of an FECH mutation (Fig 1). Tugores et al22 described many regulatory elements of interest in the promoter region of the FECH gene. Nevertheless, they do not appear to be involved in low-expression mechanism, because no sequence variation has been found in more than 1 kb of the proximal promoter. Recently, Scott et al23 reported a new region 2 kb upstream from the transcription start site that may contribute to a high level of erythroid specific expression of FECH gene by maintaining an active chromatine configuration. One might hypothesize that mutations in this region bearing erythroid-specific regulatory elements could be involved in the FECH gene low expression. Such a mechanism clearly needs to be evaluated in further studies onFECH gene expression. On the other hand, sequencing the coding regions as well as part of intron 1 and 3′ UTR to search for mutations that might decrease FECH mRNA steady-state level failed to detect any other sequence variations.

These observations based on a limited number of EPP families were extended by a case-control association study in 39 EPP nuclear families in which the distribution of the 5 FECH intragenic polymorphisms was analyzed. The subjects were divided into 3 groups: (1) the patient group, (2) the transmitter parent group (ie, asymptomatic parents bearing a specific FECH mutation), and (3) the normal parent group. The most striking finding was that the patient group and the normal parent group exhibited a similar increase in the frequency of a specific set of alleles (G, T, A9, T; Table 1), one of which has previously been reported to be overrepresented in EPP patients,14 whereas none of them was associated with the transmitter parent (FECHmutation carriers) group. It is thus surprising that such a strong association exists between FECH polymorphisms and overt EPP patients but not with asymptomatic carriers of FECH mutations. This is in agreement with the vast heterogeneity in FECHmutations (38 different ones reported to date6). It also assesses that the normal parents transmit to the overt EPP siblings a common but specific wild-type FECH allele. In accordance with their physical proximity, we found evidence for linkage disequilibrium between −251G and IVS1−23T,IVS2μsatA9, and 1520T polymorphisms (Table 2). This strongly indicates that the low-expressed allele is associated with a major [G-T-A9-T] haplotype.

The invariable presence in the 7 EPP patients (Fig 1) of a wild-type low-expressed allelic variant with the same 5′ partial [G-T-A9] haplotype implies that this allele should appear with a fairly high frequency in the general population. This was tested in a control group of 70 unrelated normal subjects of whom 39 were heterozygous for one of the exonic dimorphisms 798G/C or1520C/T and thus available for RNA quantitation. The bimodal distribution of the quantitation data allowed isolation of a subgroup of 9 subjects with FECH allelic mRNA ratios of 0.7 or less. Eight of the nine low-expressed alleles were associated with the same 5′ partial [G-T-A9] haplotype (Fig 2). Haplotypic variations for the 3′ polymorphisms indicate that recombination in this part of the gene does not affect expression, suggesting that the mutation(s) causing low expression lies in the 5′ part of an FECHgene with an ancestral [G-T-A9] haplotype. The frequency of the low-expressed FECH allele in the control population could be estimated from 6.5% (9/140, assuming that none of the 31 unexplored subjects had a low-expressed allelic variant) to 11.5% (9/78, assuming that this ratio reflects the real frequency of the allelic variant in the whole group). At such a high heterozygote frequency, about 1% of the population should be homozygote. Such a homozygote was suspected on genotypic analysis in EPP family 5 (subject III₅3; Fig 1). mRNA relative quantitations showed that the mother II₅2 had transmitted a low-expressed allele to this subject (Fig 1). A ribonuclease protection assay was used for absolute quantitation of the nonmutant FECH allele in this family (see Subjects, Materials, and Methods). Compared with the control subject (R = 8.3), FECHmRNA level was 75% lower in the symptomatic father II₅1 (R = 2.6; 1 deleterious allele and 1 low-expressed one) and 50% lower in the daughter III₅3 (R = 4.1); as expected, these data were strongly correlated with a similar decrease in FECH enzyme activities (Fig 1). Because no de novo FECH mutation could be found by sequencing (not shown), this gives evidence for the presence in subject III₅3 of two coinherited low-expressed alleles. This subject has no clinical or biochemical signs of EPP but would have been misdiagnosed as a gene carrier in a family study based solely on FECH enzyme measurement.

These data lead to a number of conclusions. First, that the low expression of a wild-type allelic variant trans to a mutatedFECH allele is generally required for clinical expression of EPP. It is likely that when FECH enzyme activity falls below a critical threshold, the accumulation of protoporphyrin will exceed hepatic clearance capacity and lead to clinically manifest EPP. In fact, asymptomatic EPP carriers often have normal or slightly elevated protoporphyrin levels in erythrocytes2 and higher FECH enzyme activity than overt EPP patients.10,11 Second, that the low-expressed alleles are inherited from the normal parent and most if not all of these alleles probably originate from a mutational event(s) that occurred in a common ancestral [G-T-A9] haplotype. However, the mechanism of low expression remains unknown. Third, that the frequency of the low-expressed allele in the general Caucasian population is of the order of 6.5% to 11.5%. Finally, our results have important implications for genetic counseling in EPP. From the frequency of low-expressed alleles bearing specific 5′ haplotypes (−251G IVS1−23T or−251A IVS1−23C) in the 39 individuals studied, it can be estimated that the absolute risk that a carrier of an FECH mutation will develop symptomatic EPP when bearing the [−251G IVS1−23T] haplotype trans to a mutant one is close to 60%. More importantly, the risk decreases to less than 2% if this haplotype is [−251A IVS1−23C], which, fortunately, is the most frequent one (75% of [−251A IVS1−23C] homozygotes in the control group). Indeed, in these estimations we assume that the penetrance of EPP symptoms in subjects having a genotype with a mutantFECH allele associated with a low-expressed allele is almost 100% at 15 years of age. As compared with the empirical risk of EPP symptoms proposed by Went and Klasen,24 our results provide a dramatic improvement in risk prediction for an unborn child in an EPP family based on parental haplotype study.

Although autosomal recessive inheritance has been documented in 2 cases,7,8 EPP has long been considered as an autosomal dominant disorder with incomplete penetrance. Molecular analysis supports this notion, because only a single mutation is identified in patients in one of the FECH alleles. However, EPP patients exhibit a lower FECH enzyme activity than the 50% expected in autosomal dominant disease with haploinsufficiency. This strongly suggests a more complex mode of inheritance. Went and Klasen24 proposed a triallelic system in which, in addition to the usual deleterious and normal FECH alleles, they advanced the presence of a third FECH allele that, in association to the deleterious one, leads to the EPP clinical outcome. This third allele was also expected to present a fairly high frequency in general population. This triallelic system accounted for the autosomal recessive appearance of the EPP family pedigrees and the dominant transmission of biochemical characters. We provide here strong molecular evidence that the third allele is a low-expressed wild-type allele insufficient by itself, or in the homozygous state, to cause overt EPP yet sufficiently frequent in the population to cause manifest disease through coinheritance with a rare FECH EPP mutation. This allele appears thus more as a triggering factor than a deleterious allele. We therefore prefer still to consider EPP as an autosomal dominant disorder than a recessive one (strictly speaking 2 deleterious alleles). However, the mendelian rules of recessive and dominant inheritance of genetic traits may appear somewhat meaningless in EPP. A situation similar to that described here has been documented in hereditary elliptocytosis (HE).25,26 Finally, such a mechanism in which mild variations of the wild-type gene expression level may account for variable penetrance or clinical expression in many dominantly inherited disorders such as Hirschsprung disease,27 campomelic dysplasia, or basal cell nevus syndrome.28 

Special thanks to Prof G.H. Elder (Department of Medical Biochemistry, University of Wales College of Medicine, Cardiff, UK) and Dr X. Schneider-Yin (Stadspital Triemli, Zurich, Switzerland) for helpful discussions and critical review of the manuscript.