Molecular analyses of patients with hyperferritinemia and normal serum iron values reveal both L ferritin IRE and 3 new ferroportin (slc11A3) mutations

Ferritin is the major protein implicated in the constitution of iron stores. Multiple isoferritins are found in tissues, resulting from the incorporation of varying proportions of H and L ferritin subunits into the protein shell.¹  The relative abundance of these 2 subunits in ferritins from different tissues results from complex transcriptional and translational regulatory processes.²  Although ferritin is an intracellular protein, it is present in trace amounts in the serum. Serum ferritin is iron poor and mostly composed of partially glycosylated L subunits. Normal serum ferritin values are between 50 and 350 μg/L and vary within this range with age and sex. Elevated serum ferritin values are considered a good index of increased iron stores, although several clinical conditions can give rise to increased serum ferritin levels in the absence of high iron stores, including cancer, inflammation, and infection.³  The translation of both H and L ferritin mRNAs depends on the iron status of the cell and is regulated by the interaction between the iron-responsive element (IRE), present in the 5′ noncoding region of both H and L ferritin mRNAs, and cytoplasmic iron regulatory proteins (IRPs). Iron entry into cells induces a conformational change in IRP1 or degradation of IRP2, resulting in the initiation of translation of ferritin mRNAs and synthesis of ferritin subunits.⁴  This system allows a constant adaptation of ferritin levels to the iron status of the cell.

One notable exception to the correlation between serum ferritin levels and tissue iron stores is the hereditary hyperferritinemiacataract syndrome (HHCS). The hallmark of HHCS is the presence of both elevated serum ferritin levels and early onset bilateral cataracts, together with normal to low serum iron and transferrin saturation.⁵  Patients with HHCS are heterozygous for a mutation in the IRE of the L ferritin gene. The presence of L ferritin mRNA molecules bearing a mutated IRE leads to the uncontrolled production of L ferritin subunits, independent of the cell iron status. It has been shown by several researchers that mutated IREs have a markedly reduced binding affinity for both IRP1 and IRP2.^6-9  As a result, ferritin synthesis remains elevated, even under conditions of iron depletion. In the first 2 families described, point mutations were identified in 2 different nucleotides coding for the IRE loop, which resulted in a 40A>G⁶  and a 41G>C¹⁰  change at the heterozygous state (numbering starts at the cap site of the L ferritin mRNA). Several other mutations in the loop and stem of the IRE have subsequently been reported, as well as partial deletion of the IRE structure (reviewed in Beaumont and Girelli⁵ ). The identification of many different mutations in patients with HHCS clearly demonstrated that this syndrome is due to mutations in the L ferritin gene present on chromosome 19q13.3-qter.

With the increasing awareness of the high prevalence of genetic hemochromatosis in white populations, routine screening of biochemical markers of iron metabolism, including serum ferritin, is frequently performed. Indeed, the detection of high serum ferritin level is not uncommon in hospitalized patients, and when associated with otherwise normal markers of iron status, is frequently considered as “unexplained.” However, a new form of dominant hereditary hemochromatosis (type 4 hemochromatosis) has recently been described, in which elevated serum ferritin levels associated with normal or moderately elevated transferrin saturation are observed.¹¹  These individuals have heavy iron deposits in Küpffer cells, but no hepatic fibrosis.^12,13  This phenotype is distinct from that of recessive HFE-linked hemochromatosis (type 1), in which iron overload occurs primarily in hepatocytes, and, in the absence of treatment, often leads to liver fibrosis and cirrhosis.¹⁴  Heterozygous mutations in the ferroportin gene have been identified in patients with dominant hemochromatosis.^12,13,15,16  Ferroportin is a transmembrane Fe (II) transporter, mostly expressed at the basolateral membrane of mature enterocytes and in macrophages.^17,18  Ferroportin mutations were postulated to impair iron export from macrophages with subsequent iron accumulation. In some individuals, even a limited number of phlebotomies were found to rapidly reduce blood hemoglobin content to the point of frank anemia,¹³  whereas serum ferritin levels remain elevated. Hyperferritinemia with normal transferrin saturation and abnormal sensitivity to phlebotomy is reminiscent of the phenotype of the patients with HHCS.

Over the past 4 years, 52 DNA samples have been referred to us for L ferritin IRE sequencing, from patients with unexplained hyperferritinemia and normal serum iron and transferrin saturation. The presence of cataracts was not always documented. L ferritin exon 1 was sequenced in all the patients. When no IRE mutation was found, we genotyped HFE mutations and sequenced both the H ferritin and the ferroportin genes. Here, we report the results of these molecular studies, in which new ferroportin mutations have been identified.

Blood samples from patients were sent to us by physicians from French hospitals, with the exception of one case coming from Canada and one from Sweden. In most cases, hyperferritinemia was discovered on routine screening and was associated with normal levels of serum iron and transferrin saturation (information available for one third of the patients). Prior to sending the samples for molecular analysis, the physicians had ruled out common causes of serum ferritin elevation. One patient was known to be homozygous for the Cys282Tyr mutation in the HFE gene and was referred to us because of the associated early onset cataract. Approval for these studies was obtained from the Hôpital Xavier Bichat. Informed consent was provided according to the Declaration of Helsinki.

Genomic DNA from individuals belonging to families collected by the Centre Français du Polymorphisme Humain (CEPH) was used as control DNA.

Genomic DNA was extracted from peripheral blood collected on EDTA (ethylenediaminetetraacetic acid) with QIAamp DNA mini kit (Qiagen, Courtabeuf, France). Polymerase chain reaction (PCR) amplification for L ferritin exon 1 and for the 8 exons of the ferroportin gene was performed using Taq polymerase Platinum (Invitrogen, Cergy Pontoise, France). L ferritin exon 1 was amplified with a set of primers complementary to the promoter region for the forward primer (Fer L Prom, 5′-CCGGCGCACCATAAAAGAAGC-3′) and to intron 1 for the reverse primer (Fer L int1, 5′-TTACCCGACCGCACAAAGAAGG-3′). The deletion of a portion of an IRE was confirmed in one family by PCR amplification using Fer L prom as a forward primer and a reverse primer mapping to exon 1 (5′-CGGAGGTTGCAAGCGGAGAG6-3′). The normal DNA gave a 183-bp fragment, and the deleted DNA gave a 168-bp fragment.

Each exon of the ferroportin gene was amplified from patients and control samples with a set of primers complementary to flanking intron sequences. These data can be viewed on the Blood website; see the Supplemental Table link at the top of the online article.

Two mutations were confirmed by PCR using the same set of primers as for sequencing. The G>T change in exon 6 destroyed a HincII restriction site in the mutated allele, and the G>T change in exon 7 destroyed a MwoI site. The PCR products from exon 6 or from exon 7 were digested either by HincII or MwoI, respectively, and analyzed on a 2% agarose gel and detected with ethidium bromide.

HFE Cys282Tyr and His63Asp mutations were tested by PCR amplification and allele-specific molecular beacons as previously described.¹⁹  Hybridization of the probes was carried out at the end of the PCR reaction, and the emitted fluorescence was recorded using a Fluostar plate fluorometer (BMG, Offenburg, Germany) at 2 wavelengths: 480 to 520 nm for fluorescein and 520 to 590 nm for tetramethylrhodamine. The ratio of fluorescence emitted by the normal and the mutated probes allowed an accurate determination of the genotype.

Sequencing of the first exon of the L ferritin gene in 52 DNA samples identified 24 patients with a heterozygous mutation in the IRE. Most of these were point mutations (Figure 1A), with the notable exception of one Canadian individual with a 2 base deletion of nucleotides A38 and C39 in the IRE loop and one French woman with a 16 base deletion removing the second half of the IRE structure, from U42 to G57 (Figure 1B). The interstitial 38delA-39delC deletion is probably a de novo mutation. The mother did not carry the mutation, and the father, who was not available for molecular diagnosis, had no history of cataract. Furthermore, the proband was the only affected child of a kindred of 7 children. In the family with the large 16–base pair deletion, both the proband and her 2 sons carried the deletion.

Previously undescribed mutations in the upper ascending part of the IRE stem (U34>C) and in the descending part (47G>A) were also identified. The mutations found in the remaining 20 patients (Figure 1A) have already been described in patients with HHCS.⁵  Mutations of the bulge account for about 50% of the mutations, 8 mutations affecting G32 (with all 3 possible replacements), and 4 mutations affecting C33. Only 2 nucleotides of the loop were found to be mutated, 39C>U in 3 cases and 40A>G in 4 other cases. One 32G>C mutation occurred de novo, because the mutation was not present in either parent. Ferritin values and clinical findings of patients with IRE mutation are presented in Table 1.

In the remaining 28 DNA samples from patients with hyperferritinemia, no IRE mutation was found. Examination of the clinical findings for these individuals indicated differences from patients with HHCS. The patients without IRE mutations were older than patients with these mutations (59 ± 15 years and 39 ± 17 years, respectively) (Table 2). Cataracts were frequent in the 2 groups, but, whereas several of the patients with HHCS mutations had required surgery at a very young age, this surgery was not reported for patients without these mutations. A family history of cataracts was not reported for any of the patients without IRE mutations. One known hereditary hemochromatosis (patient 28) with a homozygous Cys282Tyr mutation in the HFE gene was referred to us because of a cataract requiring operation at the age of 40. Because hemochromatosis is not a known cause for cataract, we sequenced the L ferritin IRE, but no mutation was found.

Several hypotheses were explored to account for unexplained hyperferritinemia in the 28 patients without IRE mutations. First, the HFE genotype was assessed in these patients. Three individuals were heterozygous for the HFE Cys282Tyr mutation, 1 individual was homozygous for the His63Asp mutation, and 12 individuals were heterozygous for the His63Asp mutation. Thus, the prevalence of HFE mutations (57%) was higher than in the healthy population (25%).²⁰  There was no obvious correlation between the presence of HFE mutations and the serum ferritin levels.

We previously reported that inactivation of the H ferritin gene at the heterozygous state in recombinant mice is responsible for increased serum L ferritin levels in the absence of iron overload.²¹  This finding prompted us to sequence the H ferritin gene in patients with no L ferritin IRE mutations. The 4 exons and exon-intron boundaries were sequenced, but no mutation was found.

In dominant hemochromatosis with ferroportin mutations, the most striking finding in the patients is the apparent discrepancy between elevated serum ferritin levels and normal transferrin saturation. Although we did not have information concerning iron stores in our 28 patients without IRE mutations, we amplified and sequenced the 8 exons and exon-intron boundaries of the ferroportin gene. We identified 3 patients with ferroportin mutations (patients 38, 39, and 48 from Table 2), and 2 of these mutations were confirmed by restriction analysis of the amplification products from the involved region (Figure 2).

The clinical information available on these patients is summarized in Table 3. Patient 48, a 61-year-old man, had high serum ferritin values (4069 μg/L) and no evidence of cataract. He was found to carry an A>G change in exon 5 (position 774 of the cDNA), which results in the Asp157Gly mutation. Patient 38, a 63-year-old man with serum ferritin of 3018 μg/L had a heterozygous G>T change in exon 6 (position 850), leading to a Gln182His replacement. Bilateral cataracts were diagnosed at the age of 46 years. His daughter also had increased serum ferritin levels and was found to carry the same mutation (Figure 2). In addition, this patient was homozygous for the His63Asp mutation in the HFE gene. However, his serum iron indices were absolutely normal, and this HFE mutation probably does not contribute to the hyperferritinemia. Finally, patient 39, a 49-year-old woman, with serum ferritin ranging between 1700 and 2000 μg/L had a G>T change in exon 7 (position 1272), leading to a Gly323Val mutation. None of these 3 mutations were found in the genomic DNA of a panel of 80 healthy individuals from the CEPH.

The Asp157Gly and Gln182His mutations are located in a putative loop between 2 transmembrane domains, in a region where several mutations have been found in humans^12,13,15,16  and in zebra fish.²²  The Gly323Val mutation occurs within a transmembrane domain (Figure 3).

Mutations in the IRE of the L ferritin gene alter the negative feedback regulation that represses L ferritin synthesis in conditions of low iron stores and are responsible for HHCS, a disease characterized by cataract and hyperferritinemia.²³  Genetic analysis of 52 DNA samples referred to us for molecular diagnosis of HHCS, however, demonstrates that both L ferritin IRE mutations and ferroportin mutations are associated with otherwise unexplained hyperferritinemia.

Mutations in the IRE of the L-ferritin gene were identified in only 24 of 52 cases. In these patients, we found 2 deletions and 22 point mutations in the IRE region, including 2 previously undescribed mutations of the IRE stem. The distribution of the mutations along the IRE structure highlights the importance of the bulge in maintaining IRE function, and especially G32, where all 3 possible mutations were found, and of the C₃₆-G₄₇ pairing in the middle of the stem, because both C36>A and G47>A disrupt ferritin translational control. Our study also confirms the phenotypic variability that has been previously reported in this syndrome.²⁴  Indeed, IRE mutations were associated with serum ferritin levels ranging from 800 to 3000 μg/L. Similarly the age of onset of cataract varied from early childhood (5-7 years) to late adulthood. In one case, a nonsymptomatic cataract was diagnosed at 35 years of age, following slit lamp examination performed after identification of the IRE mutation. In addition, there was no genotype/phenotype correlation between unrelated patients with the same mutations. For example, the C39>T mutation was associated with a serum ferritin of 2800 μg/L and early onset cataract in one individual and a serum ferritin of 800 μg/L and asymptomatic cataract in another patient.

Hyperferritinemia with normal transferrin saturation is somewhat reminiscent of the phenotype of patients with dominant type 4 hemochromatosis, a disorder recently shown to result from mutations in the ferroportin gene. Screening of the 28 patients for whom no L-ferritin IRE mutations were detected permitted the identification of 3 individuals with previously undescribed ferroportin mutations. Patient 48 had the highest serum ferritin level of our series (4069 μg/L). Liver biopsy had not been performed in this case; therefore, we were not able to confirm the presence of iron overload in macrophages. Both patient 38 and his daughter had hyperferritinemia, and both carried the same heterozygous ferroportin mutation, supporting the causative relationship between the mutation and the increased serum ferritin levels. To date, the ferroportin mutations associated with hyperferritinemia have involved amino acids that are conserved between human, rodents, and zebra fish sequences (Figure 3). The predictions of ferroportin structure vary between different studies. Regardless of the model, however, the Asp157Gly and the Gln182His mutations reported here are located in a loop between 2 transmembrane domains, as are the 2 previously reported human mutations (Asn144His and Val162del), as well as the zebra fish Leu169Phe (equivalent to Leu170 in the human sequence) mutation responsible for hypochromic anemia (Figure 3). These results suggest that this loop is a functional domain likely to be essential for ferroportin function. Depending on the model, this loop is either extracellular^13,22  or intracellular,¹⁸  and, given the uncertainty as to its orientation, speculation on its putative function is premature. The Gly323Val reported here falls in a transmembrane domain and might change iron export activity, membrane targeting, and/or stability of the protein.

It is intriguing that none of those previously described ferroportin mutations were identified in our patients, especially Val162del, which has been identified repeatedly in dominant hemochromatosis.^12,13,25,26  This finding might suggest that the phenotype of the mutations we have identified is milder and might have remained undetected if the patient had not developed cataract or if the hyperferritinemia had not been discovered fortuitously in a routine examination. In previously described families, the mutations were found to be associated with the presence of excess iron deposits in macrophages. It would appear advisable to recommend liver biopsy for these patients to assess the extent of iron overload and the possible onset of cirrhosis, both of which are potential indications for phlebotomy.

In 25 of the 52 patients referred for possible HHCS, no definite cause of hyperferritinemia was found. In these individuals, we did uncover an apparent increase in the frequency of heterozygous His63Asp mutations in the HFE gene (12 of 28). The role of His63Asp HFE mutations in the onset of iron overload is not clearly established, but the only expected effect, if any, would be an increase in transferrin saturation, not an increase in serum ferritin levels.²⁷  No mutations in H ferritin mutation were identified in these individuals, although H ferritin haplo-insufficiency in mouse is responsible for hyperferritinemia in the absence of iron overload. Although serum iron indices were not often available, the patients in this study were referred to us on the assumption that other known causes of serum ferritin elevation had been ruled out, including inflammation and infection, liver diseases, and cancer. Thus, our findings suggest that additional causes for hyperferritinemia remain to be identified.

Cataract formation is a direct consequence of L ferritin IRE mutations in patients with HHCS, because intracellular ferritin accumulation occurs in all tissues, including in the lens. Light-diffracting crystals of L subunit-rich ferritins have been found in the lens of patients with HHCS^28,29  and are thought to be responsible for the pulverulent dustlike opacities observed on slit lamp examination.^30,31  In this syndrome, additional individuals with cataract are usually found in the pedigree. Nevertheless, several cases have been described in which the mutation³²  or the deletion (this study) was a de novo mutation, and a family history of cataract was absent. In the patients with no IRE mutation, the relationship between hyperferritinemia and cataract formation is less clear. The DNA samples evaluated by us were sent for molecular diagnosis of HHCS. Because of this strong bias for selection of patients with both hyperferritinemia and cataract, it is possible that the 2 conditions were unrelated in patients without IRE mutation. Classical HFE-related hemochromatosis does not lead to cataract. Increased ferritin synthesis only occurs in tissue in which iron accumulates, and the lens seems to be spared by iron overload. It is not known whether this situation is also the case for type 4 hemochromatosis resulting from ferroportin mutations. Ferroportin is most strongly expressed in enterocytes and macrophages, and its expression in the lens has never been examined. Similarly, we cannot exclude a possible cause-effect relationship between hyperferritinemia and cataract formation in the patients in whom no mutation was identified. Cataract formation was documented in 17 of 25 of these individuals. It is noteworthy that a number of these patients required surgery in their fourth and fifth decade, which differs from age-related cataract in which surgery is usually performed later in life.³³  This information raises the intriguing possibility that lens ferritin accumulation might be a factor contributing to age-related cataract in the general population.

This study demonstrates that both L ferritin IRE mutations and ferroportin mutations can be found in patients with otherwise unexplained hyperferritinemia. The presence of cataracts does not permit the unambiguous identification of patients with HHCS, although the existence of a family history of cataract was only encountered in these patients. The measurement of ferritin iron has been proposed as a test to discriminate between elevated serum ferritin levels because of iron overload or because of inflammation.³⁴  This test may also discriminate patients with HHCS and those with ferroportin mutations, because serum ferritin iron content is likely to be low in patients with HHCS and elevated in patients with ferroportin mutations, although this finding remains to be demonstrated.

We thank all the clinicians who have identified the patients and without whom this study could not have been possible, including Drs L. Bettan, D. Bonneau, A. Bourguignat, P. Brissot, D. Cattan, Y. Consigny, J. Denis, Y. Deugnier, H. de Verneuil, B. Dingeon, I. Durieu, S. Durupt, B. Eschemann, J. Feys, P. Jouk, H. Journel, A. Klisnick, R. Levrat, S. Levy, E. Lilièvre-Berna, C. Moraine, A. Najman, E. Passot, P. Rohrlich, B. Rueff, F. Ruyssen, M. Schree, M. Suzan, P. Thibault, C. Vanlemmens, B. Varet, J.P. Villeneuve (Canada), D. Vincent, and J. Wallvik (Sweden).