Increased levels of ERFE-encoding FAM132B in patients with congenital dyserythropoietic anemia type II

To the editor:

Recessive mutations in the SEC23B gene cause congenital dyserythropoietic anemia type II (CDAII),¹  a rare hereditary disorder hallmarked by ineffective erythropoiesis, iron overload, and reduced expression of hepatic hormone hepcidin.^2,3  Some erythroid regulators have been proposed as pathological suppressors of hepcidin expression, such as growth differentiation factor 15 (GDF15) in thalassemia, CDAI and II,^4-6  even if alone GDF15 seems unnecessary for physiological hepcidin suppression.⁷  The most recently described is the erythroblast-derived hormone erythroferrone (ERFE), a member of the tumor necrosis factor superfamily that specifically inhibits hepcidin production. ERFE-encoding FAM132B is an erythropoietin (EPO)-responsive gene in experimental models.⁸  However, the function of ERFE in humans remains to be investigated.

To evaluate ERFE expression in CDAII patients, we enrolled 37 SEC23B-related nontransfusion-dependent patients (supplemental Table 1, available on the Blood Web site). Ten new cases from 9 unrelated pedigrees were described (supplemental Table 2). All patients were young adults (19.1 ± 3.0 years at diagnosis), with high transferrin saturation (71.4% ± 5.5%). For case-control study, 29 age- and sex-matched healthy controls (HCs) were also included (supplemental Table 1) (see supplemental Methods).

We observed a statistically significant overexpression of the FAM132B gene either in peripheral blood leukocytes (PBLs) (Figure 1A) or in reticulocytes (supplemental Figure 1a) from CDAII patients compared with HCs. Indeed, a statistically significant correlation between PBLs and reticulocyte FAM132B expression from the same patients was observed (Spearman ρ = 0.78; P = .02) (supplemental Figure 1b). Similarly, a marked upregulation of Erfe in both erythroblasts and PBLs from Hbb^th3/+mice compared with wild-type (WT) ones was observed (supplemental Figure 1c). Of note, ∼0.03% of erythroblasts (CD71⁺/CD45⁻ cells) were found in PBLs from CDAII patients (data not shown).

Consistent with previous studies and with the loss of function of SEC23B mutations,⁹  we observed a reduced SEC23B expression in our patients compared with HCs (supplemental Figure 2a). Indeed, FAM132B and SEC23B gene expression exhibited an inverse correlation (supplemental Figure 2b). Interestingly, we did not observe in our patients the same trend of expression of the FAM132A paralog gene, which shares 69% of transcript identity with FAM132B (supplemental Figure 2c). This suggests a possible involvement of ERFE-encoding FAM132B in the pathogenesis of hemosiderosis in CDAII.

During the preparation of this study, 2 articles reporting measurement of human ERFE protein were published.^10,11  According to gene expression analysis, we found increased ERFE expression in CDAII patients compared with HCs (Figure 1B).

Of note, we observed increased ERFE expression also in β-thalassemia (BT)-intermedia patients, which exhibit iron overload similarly to CDAII patients (supplemental Table 1). Conversely, only a slight increase of ERFE expression was observed in patients with mild well-compensated anemia, such as hereditary spherocytosis (HS) (Figure 1C). These data suggested that the markedly increased ERFE expression observed in both CDAII and BT-intermedia patients is mainly due to ineffective erythropoiesis.

As recently reported, the ablation of Erfe in thalassemic mice induces a slight amelioration of ineffective erythropoiesis but does not improve the anemia.¹²  In order to explore the possible interplay between the levels of erythroferrone and the degree of anemia in CDAII patients, we divided our cohort into 2 subgroups: low and high FAM132B. Unlike thalassemic Erfe-haploinsufficient mice, we observed a statistically significant reduction in hemoglobin level in the high-FAM132B subset compared with the low-FAM132B one. Of note, we focused on nontransfusion-dependent patients, thus the expression level of FAM132B was not biased by transfusion regimen; however, ERFE expression did not correlate even with the occasional need for transfusions (Table 1). The higher expression of ERFE seems to reflect the increase of iron demand for Hb production as well as the expanding abnormal erythropoiesis, characterizing CDAII, as attested by the increased levels of soluble transferrin receptor and EPO in high-FAM132B patients. This in turn leads to reduced hepcidin and hepcidin-to-ferritin ratio in the high-FAM132B group compared with the low-FAM132B one, resulting in augmented iron delivery to the erythron. Indeed, an inverse correlation between FAM132B expression and hepcidin levels was observed (Spearman r, −0.49; P = .004). However, the iron balance data do not differ significantly between the 2 CDAII subgroups, even if a trend to increased transferrin saturation was observed in high-FAM132B patients (Table 1). Thus, ERFE upregulation may contribute to the inappropriate suppression of hepcidin, but does not solely account for the subsequent hemosiderosis observed in CDAII patients.

High expression of FAM132B in ex vivo EPO-differentiating human erythroblasts has been documented.⁸  Accordingly, we confirmed a progressive increase of FAM132B expression in healthy CD34⁺ progenitor cells EPO induced to erythroid differentiation. Particularly, we observed a significant increase of FAM132B in the transition from basophilic (9 days, 3.93 ± 1.32) to polychromatic (11 days, 14.30 ± 2.46; P = .02) erythroblasts (Figure 1D). Of note, the onset of multinuclearity in an in vitro model of CDAII involves polychromatic and orthochromatic precursors.¹³  Thus, the accumulation of highly expressing FAM132B polychromatic erythroblasts in CDAII marrow could account for the pathological overexpression observed in CDAII patients. Our observation is consistent with the demonstration that polychromatic erythroblasts are the main source of Erfe in BT mice. Indeed, similarly to CDAII erythropoiesis, BT is characterized by an accumulation of polychromatic erythroblasts that do not mature to the orthochromatic stage. Thus, the increased Erfe expression in erythroblasts from Hbb^Th3/+ mice compared with WT mice is mainly due to the relative abundance of polychromatic erythroblasts.¹² 

CDAII is still lacking a reliable animal model because SEC23B deficiency results in different phenotypes in humans and mice.^14,15  So far, the only reliable in vitro model for CDAII is the SEC23B-silenced K562 cells.¹  We established K562 stable clones silenced for the SEC23B gene by lentiviral shRNAmir targeting human SEC23B, which ensures high-efficiency silencing. When we induced to erythroid differentiation K562 stable clones silenced for SEC23B, we observed a higher expression of FAM132B at 5 days of erythroid differentiation in K562 SEC23B-silenced cells compared with not-silenced ones (Figure 1E). Conversely, SEC23B expression was lower in both K562 SEC23B stably silenced cell lines compared with K562 sh-CTR cells at 2 and 5 days of differentiation (supplemental Figure 2d). Thus, we confirmed the ex vivo data about inverse correlation between FAM132B and SEC23B expression observed in our patients.

Literature data lack information about human ERFE protein. In order to validate our data at the protein level, we first evaluated the good performance of the commercial anti-FAM132B antibody we used (supplemental Figure 3a-d). We also provided the first demonstration that ERFE protein is N-glycosylated; indeed, the higher band obtained by western blotting (WB) determination is the glycosylated isoform of the same protein, as demonstrated by the digestion pattern after peptide-N4-(acetyl-β-glucosaminyl)-asparagine amidase treatment (supplemental Figure 3e). According to gene expression data, we observed an increased level of secreted ERFE in the extracellular medium of K562 SEC23B-silenced cells compared with not-silenced ones at 5 days of erythroid differentiation, whereas the ERFE intracellular level was lower (Figure 1F).

Although the mechanisms of hemin-induced differentiation are quite different from EPO-induced ones, we can hypothesize that ERFE overexpression is related to the maturative arrest and the subsequent increased number of erythroid precursors.

This study provides the first analysis on ERFE expression in a human model of dyserythropoietic anemia with ineffective erythropoiesis such as CDAII. Our ex vivo and in vitro data indicate that ERFE overexpression in CDAII patients might be most likely related to both physiological and pathological mechanisms leading to hepcidin suppression in a condition of dyserythropoiesis. Indeed, we clearly demonstrated that in 2 different genetic conditions sharing common clinical findings and similar pathogenesis, such as CDAII and BT intermedia, FAM132B overexpression is related to abnormal erythropoiesis. Nevertheless, the absence of a clear correlation between ERFE levels and CDAII iron balance suggests that ERFE cannot be the only erythroid regulator of hepcidin suppression, at least in CDAII patients.