Evaluation of a bone morphogenetic protein 6 variant as a cause of iron loading

Hereditary hemochromatosis (HH) is relatively common in individuals of northern European descent, with the majority of cases caused by mutations in the HFE (hemochromatosis) gene [1, 2]. Conversely, mutations in HFE are rare in non-European populations, and mutations in other genes involved in the regulation of iron homeostasis are more common, such as hemojuvelin (HFE2), hepcidin (HAMP), transferrin receptor 2 (TFR2), and ferroportin (SLC40A1), collectively termed non-HFE HH [3]. Potentially, all regulators of iron homeostasis are functional gene candidates for non-HFE HH [4]. BMP6 (bone morphogenetic protein 6) is a key regulator of iron homeostasis [5]. Bmp6 null mice have significant iron overload due to deficiency of the iron-regulatory hormone hepcidin [6, 7].

Under normal conditions, Bmp6 expression is upregulated by increased iron and increases the expression of hepcidin which in turn limits iron absorption and recycling [8]. A recent report associated heterozygosity for three variants p.P95S, p.L96P, and p.Q113E in the BMP6 pro-peptide region with nine clinical cases of unexplained atypical iron overload [9]. The functional link between BMP6 and iron overload remains to be explained, however, and further validation is necessary to better assess causality of these variants [10]. In addition, p.E112Q and p.R257H have also been associated with more severe iron overload in patients with HFE mutations [11]. Here, we report on three clinical cases of unexplained atypical iron overload carrying an amino acid insertion within the same region of the BMP6 pro-peptide.

Three cases are reported in this study; in each case, HFE genotyping was negative, and liver specialist assessment was undertaken. Case 1 was a 62-year-old male who over the past 7 years had raised serum ferritin (SF) ranging between 1261 and 1675 μg/L (normal range 30–300 μg/L), with transferrin saturation (TS) 40 to 53% (normal range 20–55%). His alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma-glutamyl transferase (GGT) levels all remained normal during this time. Liver biopsy revealed grade 4 stainable iron in hepatocytes, with accentuation of iron in the pericanalicular region (Fig. 1). No other pathology was noted. In case 2, a male aged 51 who did not consume any alcohol, presented with a SF of 1714 μg/L and TS of 34%. ALT, AST, and GGT were all normal although ultrasound revealed a diffusely echogenic liver suggesting fatty liver. Ferriscan identified a hepatic iron concentration of 44 μmol/g (normal range 3–33 μmol/g). In case 3, also male aged 62 presented with a SF of 1133 μg/L and TS of 90%. His ALT, AST, and GGT levels were mildly elevated. Liver biopsy demonstrated grade 1 iron with a hepatic iron content of 39 μmol/g and F1 fibrosis. Case 3’s pathophysiology is complex with positivity for α-thalassemia trait and hepatitis C virus (genotype 2).

Targeted next-generation sequencing revealed that all three patients were negative for mutations in the five currently clinically associated HH genes. Further analysis identified a tri-nucleotide duplication in BMP6, c.353-355dupAGC (NM_001718.4), resulting in the in-frame duplication of glutamine p.Q118dup (NP_001709.1), which was confirmed by Sanger sequencing. The nomenclature for this variant is noted in three different forms across various datasets: dbSNP rs771616962 refers to this variant as p.Q118_L119insQ, wANNOVAR, and the NHLBI Exome Sequencing Project (http://evs.gs.washington.edu/EVS/) (ESP) dataset as p.E112delinsEQ and Exome Aggregation Consortium database (ExAC) [12] as p.Gln115dup. While all of these references result in the same outcome, extension of the poly-Q stretch, the various designations may cause underestimation of its population frequency. This variant, or its pseudonyms, is not present in the 1000 Genomes database; however, it is present in the ESP (MAF = 0.016) and ExAC (MAF = 0.041).

To investigate possible functional effects of the p.Q118dup variant on BMP6, immunofluorescence microscopy studies were conducted using BMP6 wildtype and p.Q118dup proteins with a pro-peptide N-terminus HA tag and a mature protein C-terminus V5 tag. We used both CHO cells and liver endothelial (SK-HEP-1) cells as a model for examining the molecular and functional effects of the p.Q118dup variant. CHO cells (Fig. 2a) and SK-HEP-1 cells (Fig. 3a) stably expressing these constructs showed some peri-nuclear Golgi localization of p.Q118dup pro-peptide; however, this was very mild and not consistent within all cells. No difference in localization was observed for the mature V5-tagged protein in either the CHO or endothelial cell models.

BMP6 is an important signaling molecule required for iron homeostasis [6, 8]. In response to increase in body iron levels BMP6 levels increase, this results in an increase in BMP-SMAD signaling. The increased BMP_SMAD signaling then leads to an increase in hepcidin production. Recently, Daher and colleagues reported on three BMP6 pro-domain variants associated with iron overload and attributed the cause to reduced export of BMP6 [9]. We compared levels of the full-length and mature wildtype and mutant BMP6 protein in cell lysate and conditioned media (Fig. 2b, c). This revealed no difference in the capacity of cells to export p.Q118dup BMP6 protein compared to wildtype. Recent studies have suggested that the endothelial cells may be responsible for BMP6 production in response to iron [13]. Using an endothelial cell model, we show that there is no difference in the expression of the mature BMP6 protein for either the wildtype or p.Q118dup variant of BMP6 (Fig. 3b).

As population frequency can provide insight into the likelihood of a variant being benign, we investigated the frequency of p.Q118dup in the ExAC dataset (http://exac.broadinstitute.org/) which contains exome sequences from 60,706 unrelated individuals, and carrier rates were then calculated using the Hardy-Weinberg equation. This analysis reveals that in East Asians, p.Q118dup has a carrier rate of 1 in 3 (MAF = 0.191), with other ethnicities also showing lower though still significant carrier rates, confirming the experimental result that this variant is unlikely to affect gene function. Finally, we examined the classification of the p.Q118dup variant according to the “standards and guidelines for the interpretation of sequence variants” as set out by the American College of Medical Genetics and Genomics [14]. Based on the above data plus bioinformatic functional impact prediction using several tools available through ANNOVAR, it confirmed a categorization of the BMP6 pro-peptide p.Q118dup variant as benign.

The data from this study suggest no functional relationship between the p.Q118dup BMP6 pro-peptide variant and iron overload. Recognition of this variant as non-deleterious and not causative is notably important for this specific variant, as cursory literature and database analysis may easily lead to the alternate and incorrect assumption that this variant is likely causative of the disease.

Firstly, the alternative nomenclature of p.E112indelEQ (which while technically incorrect is the annotation returned by several common sequence alignment programs) places this variant at the same location as the p.Q113E variant recently reported by Daher et al. [9] to be deleterious and the cause of atypical iron overload in two members of a single family. Secondly, the absence of p.Q118dup from the two most commonly known variant data sets, dbSNP and 1000 Genomes, erroneously suggests that this variant is novel and thus is not carried in the healthy population. Finally, using CHO and hepatocyte endothelial cell models, we showed no difference in the localization or secretion between the wildtype and p.Q118dup BMP6 variant.

These data demonstrated no relationship between BMP6 variants and atypical iron overload and provided no functional evidence of causality for the identified variant. Furthermore, the analysis raises uncertainty about the population prevalence of BMP6 variants and their potential role in iron overload conditions. This has important clinical implications, for monitoring and treatment, and should lead to pause before suggesting genetic and family counseling based on BMP6 variation. Atypical, non-HFE-related iron overload remains an important question; however, despite prior literature and the appeal of its underlying biological plausibility, the genetic causes of some forms of HH appear elusive. Further research will be required to uncover the potential genetic causes of these atypical iron overload disorders.