Heme Excess Impairs the Maturation of Proerythroblasts From Patients with Diamond Blackfan Anemia (DBA) and 5q- Myelodysplastic Syndrome (MDS)

Abstract 514

Free heme is toxic so heme and globin synthesis must be tightly coordinated. In early erythroid differentiation, CD71 (transferrin receptor) is upregulated, allowing iron influx and the initiation of heme synthesis (via an IRE in the 5' UTR of ALAS2 RNA). Heme then upregulates globin transcription and translation. Eventually heme and globin production synchronize to ensure efficient and robust hemoglobin assembly. Mice lacking FLVCR, the cytoplasmic heme exporter, have a severe macrocytic anemia and cell apoptosis and maturation arrest at the proerythroblast stage. This led us to hypothesize that early erythroid precursors must export heme to prevent heme toxicities (Science 319:825, 2008).

To test this hypothesis using primary human marrow cells, we adapted the culture system of Giarratana, Blood 118:5071, 2011. At 10 days, >95% of cells are committed to erythropoiesis (CD71^pos, lin^neg) and are at both early (predominantly) and later stages of differentiation, allowing sorting of maturing subsets (by CD36 and GlyA) for analyses of gene expression and heme content. We anticipated that: 1) Heme synthesis would intensify before globin synthesis; 2) Erythroid cells might lack heme oxygenases (HMOX1, HMOX2) which degrade heme and thus be especially vulnerable to heme toxicity; and 3) FLVCR would be high in early cells then decrease as cells mature. As expected, ALAS2 expression initiates in stage I (CD36^negGlyA^neg; 30±6×10³ copies/5ngRNA, n=5) and there is 4.0±0.4 ng heme/10⁴ cells. Both ALAS2 RNA and total heme content increase in stage II (CD36^posGlyA^neg) and are high and stable in stages III (CD36^posGlyA^pos) and IV (CD36^negGlyA^neg)(65–73×10⁴ and 16.4–16.7). In contrast, alpha and beta globin RNA are first upregulated in stage II (17 and 13×10⁵ copies/5ngRNA) then increase 10–16 fold more in stages III-IV. Also as expected, HMOX1 RNA is barely detectable in stages I-III and present at only 30±4×10² copies/5ng RNA in stage IV; and HMOX2 RNA is low (22–30×10² copies/5ng RNA) in stages I and II, then decreases further, indicating minimal heme oxygenase activity occurs during erythroid differentiation. And as predicted, FLVCR RNA is high in stage I, highest in stage II, and declines in stages III and IV.

We next explored abnormal erythropoiesis, including times when heme and globin synthesis may be discordant. DBA is a congenital macrocytic anemia. RPS19 is mutated in 25% of cases causing failed ribosome assembly; an additional ∼30% result from defects in 13 other ribosomal proteins. We hypothesized that in DBA, globin translation would initiate slowly, leading to an increased vulnerability of proerythroblasts to heme toxicity, as heme export via FLVCR may be insufficient to compensate for the added free heme load. Data from DBA patients support this. After 10 days of marrow culture, total cell numbers increase only 6–11 fold (n=3) compared to 21–28 fold for concurrent control studies (p<0.01). Heme synthesis initiates normally in DBA, as ALAS2 expression in stage I (20±1×10³ copies/5ng RNA) is normal. However, the total heme content of stage I cells (a stage where globin is not yet available) is significantly higher in DBA patients (6.4±0.2 ng/10⁴cells) than in controls (p<0.05) and by flow cytometry, differentiation appears blocked between stages I and II. Those relatively few DBA cells that reach stage II expand in number in stages III-IV, and interestingly this subset of cells that can mature has significantly less ALAS2 RNA (26–32×10⁴ copies/5ng RNA stages III- IV, p<0.05) and more FLVCR RNA (p<0.10), thereby assuring a total heme content that is normal. There is no difference in HMOX1 or 2. An erythroid differentiation pattern similar to DBA was also observed in studies of two patients with 5q- MDS (acquired haploinsufficiency of RPS14), suggesting a shared pathophysiology of erythroid marrow failure. In both, escaping cells had significantly lower ALAS2 and ∼10-fold higher FLVCR RNA (p values <0.01). MDS patients with refractory anemia with ringed sideroblasts (RARS, n=1) and refractory cytopenia with multilineage dysplasia (n=1) were studied as well and did not show a DBA-like pattern; indeed, as anticipated, the heme content of cells from the RARS patient was significantly lower than in normal persons, p values<0.01. Together our results show that proerythroblasts are especially vulnerable to heme toxicity and argue that the macrocytic anemia of DBA, and perhaps 5q- MDS, results from excess free heme.