Molecular mechanisms of the defective hepcidin inhibition in TMPRSS6 mutations associated with iron-refractory iron deficiency anemia

Hepcidin, a circulating peptide synthesized by hepatocytes, is a major regulator of iron homeostasis.¹  When hepcidin is produced in excess, it limits the amount of ferroportin present at the plasma membrane of enterocytes and macrophages and induces iron-restricted erythropoiesis.²  Hemojuvelin (HJV) is a coreceptor for bone morphogenetic proteins³  that is required for hepcidin expression. Biallelic inactivation of the Tmprss6 gene in mice leads to excessive hepcidin production and consequently to iron-deficient erythropoiesis.^4,5  In humans, mutations of TMPRSS6 have been reported in patients with iron-refractory iron deficiency anemia (IRIDA).^6,–8 TMPRSS6 is highly expressed in the liver and encodes the membrane serine protease matriptase-2 (MT2), homologous to the ubiquitous matriptase-1.^9,10  MT2 represses hepcidin expression by cleaving membrane-bound HJV (m-HJV).¹¹  MT2 has a short N-terminal cytoplasmic tail, a single transmembrane domain, and a composite ectodomain with 2 C1r/C1s, urchin embryonic growth factor and bone morphogenetic protein 1 (CUB) domains, 3 class A low density lipoprotein receptor (LDLRA) domains, and a C-terminal serine protease domain.¹²  In this paper, we report a new case of IRIDA in a patient with compound heterozygosity for 2 mutations in the second LDLRA domain of the protein.⁶  The aim of the present work was to characterize the functional consequences of these and one other missense mutations outside the serine protease domain.

The proband originated from France. He was 10 months old when microcytic anemia was first diagnosed. Hemoglobin was 10 g/dL and mean corpuscular volume was 63 fL. Low serum iron (2.6 μM) and transferrin saturation (5%) were observed as well as low serum ferritin (4 μg/L). Anisocytosis and poikilocytosis were noticed. He was the third child of the family. A response to oral iron therapy was observed with a rise in ferritin (up to 180 μg/L) and hemoglobin (up to 11.8 g/dL) concentrations. After a few months, both ferritin and hemoglobin levels decreased despite ongoing treatment. He was given a course of intravenous iron (100 mg once a week for 8 weeks) when he was 7 years old that induced a rapid rise in ferritin and plasma iron. Hemoglobin increased up to 12.9 g/dL 3 months after the first iron infusion and remained normal for several months (Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article). However, the highest plasma iron concentration obtained at the end of the course of iron therapy was only 12 μM. Two other courses of intravenous iron produced similar results. Plasma hepcidin was measured on one occasion using an enzyme-linked immunosorbent assay¹³  (Intrinsic LifeSciences, La Jolla, CA) and found at 443 μg/L (normal range, 29-254 μg/L in men¹³ ) despite a hemoglobin concentration of 10.2 g/dL.

Blood samples of family members were obtained for genetic diagnosis after written informed consent of the parents in accordance with “Agence de Biomedecine” and the Declaration of Helsinki. Genomic DNA was extracted and TMPRSS6 gene was explored as previously described.⁷  The Service de Biochimie Hormonale et Génétique has received an approval for the practice of genetic diagnosis.

Expressing vectors encoding MT2 variants (G442R, D521N, and E522K) were obtained by mutagenesis of wild-type cDNA as described in the Supplemental data. MT2 devoid of the serine protease domain (ΔSP) and HJV^wt constructs were as described.¹¹  Functional studies on hepcidin promoter inhibition, MT2 processing, HJV cleavage, and MT2/HJV coimmunoprecipitation were performed as described.¹¹ 

We identified 2 missense mutations in exon 13 of the TMPRSS6 gene in a patient with IRIDA: a heterozygous c.1561G → A substitution leading to the replacement of an aspartic acid by an asparagine at position 521 of the protein (D521N) and a c.1564 G → A substitution leading to the replacement of a glutamic acid by a lysine at position 522 (E522K). Segregation of the mutations within the family is shown in Figure S2. These mutations are both predicted to modify the sequence of the second LDLRA domain of the protein (Figure 1A). The D521N mutation was previously reported in an IRIDA patient in combination with the G442R mutation in the CUB domain.⁶  Functional studies have only been performed on the R774C mutation, which affects the serine protease domain.¹¹  To characterize the effect of D521N, E522K, and G442R mutations on MT2 activity, HeLa cells were transfected with the corresponding cDNA constructs. By electron microscopy, we found that the proportion of protein expressed at the cell membrane was significantly reduced for D521N and E522K mutants, which partially remained in the Golgi apparatus, whereas G442R was normally targeted to the cell membrane (Figures 1B,C, S3). We have previously shown that activation of wild-type MT2 results in the release of shorter fragments in the culture medium, thought to reflect a catalytic cleavage.¹¹  Such fragments were undetectable in the supernatant of cells transfected with MT2^D521N and MT2^E522K cDNAs and only detectable as faint bands with MT2^G442R (Figure 1D), suggesting that the integrity of the CUB and LDLR domains is required for the activation of MT2.

The ability of MT2 mutants to cleave m-HJV was assessed in cotransfection experiments. HJV cleavage fragments were observed in the culture medium of cells cotransfected with normal MT2 and HJV, as previously described.¹¹  On expression of MT2^D521N and MT2^E522K, no cleavage fragments were observed, whereas expression of MT2^G442R generated only a small amount of HJV cleavage compared with normal MT2 (Figure 2A, CM). When the amount of HJV remaining at the plasma membrane was evaluated by treatment with phosphatidylinositol-phospholipase C (PI-PLC) that specifically cleaves glycosylphosphatidylinositol-anchored proteins, it appeared that MT2^D521N or MT2^E522K maintained a greater amount of m-HJV, compared with MT2^wt and MT2^G442R (Figure 2B PI-PLC). These results were consistent with those obtained with a binding assay measuring the proportion of m-HJV in the presence of escalating doses of normal and mutant MT2 constructs (Figure 2B).

To study the effect of MT2 mutants on hepcidin expression, Hep3B cells were transfected with a hepcidin promoter/firefly luciferase reporter construct.¹⁴  As expected, the presence of HJV enhanced the hepcidin promoter activity in this system, whereas the coexpression of normal MT2 with HJV almost completely prevented this activation. A significantly smaller stimulation of the hepcidin promoter was observed when G442R, D521N, or E522K variants were coexpressed with HJV, compared with the normal construct (Figure 2C).

Altogether, our results indicate that LDLRA mutants are fully, and the CUB mutant partially, defective in their ability to cleave m-HJV. However, all 3 variants are able to interact with HJV with similar efficiency, as assessed by coimmunoprecipitation studies (Figures 2D, S4). Both D521N and E522K substitutions replace highly conserved residues. From known 3-dimensional structures, both residues are predicted to bind Ca²⁺,¹⁵  and amino acid substitutions at homologous positions were shown to affect the folding of the protein.^16,17  Misfolding of the LDLR domain may explain a partial defect in the processing of the protein, resulting in lower expression of mutated MT2 at the plasma membrane and in a defective autocatalytic activation. A similar mechanism may explain the absence of TMPRSS3 activation in cases of autosomal recessive deafness resulting from a D103G mutation in the LDRA domain of the protein, at a position highly homologous to D521N of MT2. This mutation was shown to abolish the serine protease activity in a yeast expression system.¹⁸  Finally, the decrease in m-HJV cleavage observed with the G442R mutant and its inability to repress hepcidin expression underscore the importance of the CUB domain in the function of MT2.

The authors thank Carlos Lopez-Otin (University of Oviedo, Oviedo, Spain) for the MT2-FLAG–expressing vector, Paolo Arosio (University of Brescia, Brescia, Italy) for the gift of the anti-HJV antibody, Dr Serge Pissard (Hôpital Henri-Mondor, Creteil, France) for referring the patient to us and for globin gene analysis, and Roman Polishchuk from the Telethon Electron Microscopy Core Facility (TeEMCoF; Consorzio Mario Negri Sud, Santa Maria Imbaro, Italy) for the electron microscopy and morphometric analysis on MT2 mutants.

This work was supported by European Union Contract LSHM-CT-2006-037296, Italian Telethon Foundation Grant GGP08089 (C.C.), and by Agence Nationale de la Recherche–Groupement d' Intérêt Scientifique (ANR-GIS, France) Maladies rares, reference ANR-07-MRAR-008-01 (C.B., B.G.).

Contribution: L.S. performed research and contributed to the writing of the paper; F.G., C.O., and C.K. identified the mutations; A.P. and A.N. performed research; M.S. and F.T. followed up on the patient; C.B. contributed to the writing of the paper; C.C. contributed to the design of the experimental work and to the writing of the paper; and B.G. coordinated the work and contributed to the writing of the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Bernard Grandchamp, Service de Biochimie Hormonale et Génétique, Hôpital Bichat, 46 rue Henri Huchard, 75018, Paris, France; e-mail: bernard.grandchamp@inserm.fr.