FOXO3-mTOR metabolic cooperation in the regulation of ... metabolic... · FOXO3-mTOR metabolic cooperation in the regulation of erythroid cell maturation and homeostasis ... of metabolic

FOXO3-mTOR metabolic cooperation in the regulation oferythroid cell maturation and homeostasis

Xin Zhang,1 Gen�ıs Campreci�os,1 Pauline Rimmel�e,1 Raymond Liang,1,2 Safak Yalcin,1 Sathish Kumar Mungamuri,1

Jeffrey Barminko,1,3 Valentina D’Escamard,1 Margaret H. Baron,1,2,3,4,5,6 Carlo Brugnara,7 Dmitri Papatsenko,1,8

Stefano Rivella,9 and Saghi Ghaffari1,2,3,5,8*

Ineffective erythropoiesis is observed in many erythroid disorders including b-thalassemia and anemia ofchronic disease in which increased production of erythroblasts that fail to mature exacerbate the underlyinganemias. As loss of the transcription factor FOXO3 results in erythroblast abnormalities similar to the onesobserved in ineffective erythropoiesis, we investigated the underlying mechanisms of the defective Foxo32/2

erythroblast cell cycle and maturation. Here we show that loss of Foxo3 results in overactivation of theJAK2/AKT/mTOR signaling pathway in primary bone marrow erythroblasts partly mediated by redoxmodulation. We further show that hyperactivation of mTOR signaling interferes with cell cycle progression inFoxo3 mutant erythroblasts. Importantly, inhibition of mTOR signaling, in vivo or in vitro enhancessignificantly Foxo3 mutant erythroid cell maturation. Similarly, in vivo inhibition of mTOR remarkably improveserythroid cell maturation and anemia in a model of b-thalassemia. Finally we show that FOXO3 and mTORare likely part of a larger metabolic network in erythroblasts as together they control the expression of anarray of metabolic genes some of which are implicated in erythroid disorders. These combined findingsindicate that a metabolism-mediated regulatory network centered by FOXO3 and mTOR control thebalanced production and maturation of erythroid cells. They also highlight physiological interactionsbetween these proteins in regulating erythroblast energy. Our results indicate that alteration in the functionof this network might be implicated in the pathogenesis of ineffective erythropoiesis.Am. J. Hematol. 89:954–963, 2014. VC 2014 Wiley Periodicals, Inc.

� IntroductionErythroid cell maturation requires an exquisite coordination of cell proliferation and differentiation whose imbalance contributes to the patho-

genesis of erythroid disorders. This coordination is compromised in many diseases of erythroid cells characterized by anemia and ineffective eryth-ropoiesis including b-thalassemia, malaria, and anemia of chronic disease [1,2]. The importance of redox imbalance in altered erythropoiesis isevident as, depending on its degree of severity, results in subtle to highly defective RBC production [3–7]. The redox state is tightly coupled to cel-lular metabolism that in mature RBC is strictly limited to glycolysis [8,9]. Although RBC glycolysis has been extensively studied, less is knownabout the potential function of metabolic (glycolytic) pathways during erythroblast maturation [8–11].

The redox balance is sustained during erythroid cell maturation, at least in part, by FOXO3 transcriptional regulation of several anti-oxidantenzymes [5,12,13]. FOXO3 belongs to the FOXO Forkhead family of winged helix transcription factors. In addition to their antioxidant responseFOXO factors exert many fundamental biological functions including the regulation of cell cycle, apoptosis, DNA repair, and metabolism [14,15]. Thetranscriptional activity of FOXOs is negatively regulated by growth factor and cytokine receptor signaling via several protein kinases including AKT[15]. Conversely, FOXO factors are phosphorylated on distinct residues and activated in response to stress stimuli via a distinct set of protein kinases.

Additional Supporting Information may be found in the online version of this article.1Department of Developmental and Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York, New York, 10029; 2Developmental and Stem Cell Biol-ogy Multidisciplinary Training Area, Icahn School of Medicine at Mount Sinai, New York, New York, 10029; 3Division of Hematology and Medical Oncology, Depart-ment of Medicine, Icahn School of Medicine at Mount Sinai, New York, New York, 10029; 4Departments of Pediatrics Hematology-Oncology and Cell andDevelopmental Biology, Weill Cornell Medical College, New York, New York, 10021; 5Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NewYork, 10029; 6Department of Oncological Sciences, Icahn School of Medicine at Mount Sinai, New York, New York, 10029; 7Department of Lab Medicine, Children’sHospital, Boston, Massachusetts, 02115; 8Black Family Stem Cell Institute, Icahn School of Medicine at Mount Sinai, New York, New York, 10029; 9Departments ofPediatrics Hematology-Oncology and Cell and Developmental Biology, Weill Cornell Medical College, New York, New York, 10021

Conflict of interest: Nothing to report.X. Z. and G. C. contributed equally to this work.Xin Zhang is currently at the First Affiliated Hospital of Shantou University Medical College, Shantou, Guangdong, 515041, China*Correspondence to: Saghi Ghaffari; Department of Developmental and Regenerative Biology, Icahn School of Medicine at Mount Sinai, New York, NY10029. E-mail: [email protected] grant sponsor: Spanish Ministry of Education (to G.C.); Contract grant number: RO1 HL094283 (to J.B.).Contract grant sponsor: National Institutes of Health; Contract grant numbers: RO1 DK077174 to SG; RO1 HL116365 (SG, Co-PI).Contract grant sponsors: a Roche foundation award, a Black Family Stem Cell Institute award, a Myeloproliferative Neoplasm (MPN) Foundation award (toS.G.).Received for publication: 7 June 2014; Accepted: 11 June 2014Am. J. Hematol. 89:954–963, 2014.Published online: 25 June 2014 in Wiley Online Library (wileyonlinelibrary.com).DOI: 10.1002/ajh.23786

VC 2014 Wiley Periodicals, Inc.

954 American Journal of Hematology, Vol. 89, No. 10, October 2014 doi:10.1002/ajh.23786

RESEARCH ARTICLE AJHAJH

In addition to phosphorylation, FOXO are regulated by a number ofpost-translational modifications. FOXO factors in mammals are com-posed of four highly related members (FOXO1, FOXO3, FOXO4, andFOXO6). Among these FOXO3 is the major active FOXO during ery-throid cell maturation [5]. In response to erythropoietin (Epo), FOXO3is phosphorylated by AKT protein kinase [16–20] that by promoting itscytosolic localization represses FOXO3’s transcriptional activity.FOXO3’s expression, nuclear localization and transcriptional activityincreases with erythroblast maturation [5]. At the steady state, Foxo3 isrequired for erythroid cell formation [5] as loss of Foxo3 results inimpaired antioxidant response, cell cycle alterations associated withdelayed maturation of erythroblast precursors, as well as oxidativestress-mediated reduction of RBC lifespan [5]. These abnormalities leadto decreased RBC production. These combined abnormalities are highlyreminiscent of ineffective erythropoiesis in which FOXO3 may be aparticipant [21,22]. Nonetheless, the precise mechanism of cell cycleand maturation defects of Foxo3 mutant erythroblasts remains unclear.While the phenotype of Foxo3-deficient erythroid cells is relativelymild, these mice succumb to sudden death when exposed to exogenousoxidative challenge likely due to severe anemia [5], suggesting thatFoxo3 has a key function in stress erythropoiesis. Recent work in ourlaboratory and others’ indicate that in addition to the transcriptionalcontrol of antioxidant enzymes, Foxo3 is implicated in an array of met-abolic functions raising the possibility that Foxo3’s control of anti-oxidant response may be part of a broader metabolic program [23–29](and Campreci�os and Ghaffari, manuscript in preparation).

AKT provides an important signal for erythroid cell generation andmaturation [16–18,30–33]. In addition to Foxo3, AKT regulates mam-malian target of rapamycin (mTOR) kinase. mTOR signaling is one ofthe major regulators of cellular metabolism. mTOR is key to cellulargrowth (size) and proliferation, is highly sensitive to oxygen andnutrients including amino acids and glucose and has a central functionin protein synthesis [34,35]. mTOR protein kinase exists in two distinctcore complexes, mTOR complexes I and II (mTORC1 and mTORC2,respectively) which differ in their regulation and functions as well as intheir sensitivity to rapamycin. While the physiological function ofmTOR signaling during erythroid cell maturation remains unknown[36,37], conflicting results as to whether use of the mTOR inhibitors inpatients [35] is associated with anemias have been reported [38,39].

Here we investigated mechanisms underlying alterations of Foxo3-deficient erythroid cell cycling. Strikingly, we found that loss of Foxo3results in overactivation of the JAK2/AKT/mTOR signaling pathway inerythroblasts partly mediated by redox modulation. Activation ofmTOR leads to alterations of cycling and differentiation of immatureerythroblasts suggesting that activation of a feedback loop upstream ofFOXO3 compromises erythroid cell maturation. We further show usingin vitro and in vivo approaches that inhibition of mTOR signaling par-tially alleviates the abnormal maturation of Foxo3-deficient erythro-blasts leading to increased red blood cells (RBC) in the peripheralblood. Notably, we show that FOXO3 and mTOR together may bepart of a metabolic network during erythroid cell maturation. Thesefindings suggest that hyperactivation of mTOR signaling resulting fromloss of FOXO3 function contributes to the blockade of Foxo32/2

erythroblast maturation. In addition, they provide a platform to furtherdelve into redox and metabolic regulation of erythropoiesis.

� MethodsMice. Foxo31/2 mice (129xFBV/n) [40] were backcrossed >10 generations onto

C57Bl6 [41] and 10–12 week old C57Bl6 mice were used in all experiments. Proto-cols were approved by the Institutional Animal Care and Use Committee of MountSinai School of Medicine.

Cells. Bone marrow lineage negative cells were separated from mature cells usingthe EasySepTM mouse hematopoietic progenitor enrichment kit (StemCell Technolo-gies) and differentiated on fibronectin-coated plates for 18 h with IMDM1 15% FBS

supplemented with 2 U/ml Epo, IL-6 (10 ng/ml) and SCF (100 ng/ml) (modifiedfrom [32,33,42]). Cells were then starved in vitro for 2 hrs in IMDM supplementedwith 0.1% FCS and further stimulated with Epo (10 U/ml). In some experiments, cellswere differentiated in the presence of 100 mM NAC. Fetal liver cultures were per-formed using a modified protocol of [43]. Briefly, lineage negative cells were isolatedfrom E14.5 fetal livers and plated at <23 106 cells/ml with erythroid expansionmedium consisting of Stem Span SFEM (StemCell Technologies) supplemented with2 U/ml human recombinant Epo (Amgen), 100 ng/ml SCF (PreproTech), 40 ng/mlinsulin-like growth factor-1 (PreproTech), 1026 M dexamethasone (D2915; Sigma),0.4% cholesterol mix (Gibco), and 1% penicillin/streptomycin (Gibco). After 48 hrcells were washed with PBS and plated at a concentration <23 106 with either rama-pycin (20 nM; Enzo Life Sciences) or vehicle control with erythroid differentiationmedium consisting of IMDM supplemented with 2 U/ml Epo, 100 ng/ml SCF, 10%Serum replacement (Invitrogen), 5% platelet-derived serum, glutamine and 10%protein-free hybridoma media. After another 24 hrs, cells were collected and ery-throid maturation analyzed by flow cytometry.

Retroviral production and transduction of cells. Retroviral constructs and super-natant production were performed as previously described [32,33].

Colony-forming assays. For BFU-E and CFU-E analyses, 13 104 and 33 103 totalbone marrow cells were plated respectively in triplicates as previously described [32].

Flow cytometry. Bone marrow and fetal liver single cell suspensions were pre-pared and maintained in IMDM1 15% FBS, washed twice, preincubated with 10%rat serum and stained with CD71-FITC, CD44-APC, and TER119-PE or -FITCantibodies (BD Biosciences). Gating to distinguish erythroid populations accordingto their stage of maturation was performed as in [44]. Freshly isolated bone mar-row cells stained with CD44-APC and TER119-FITC, were fixed with fix/perme-abilization buffer (BD Biosciences) and incubated with 1:100 dilution of anti-pSer473 AKT and pSer235/236 S6 antibodies (Cell Signaling Technology, Cat #9271and #4858, respectively) followed by incubation with 1:1000 dilution of PE-conjugated secondary antibody (BD Biosciences) to measure intracellular AKT andS6 phosphorylation. Samples were washed and protein phosphorylation was ana-lyzed by flow cytometry. Data was analyzed by FlowJo software (Treestar).

Cell proliferation assay. Mice were injected intraperitoneally with 1 mg of BrdU.After 1 hr, bone marrow cells were isolated and stained with CD44-APC andTER119-PE antibodies (BD-Pharmingen, CA), then fixed and stained with anti-BrdU-FITC antibody (BD Biosciences) and 7-AAD for flow cytometric analysis ofcell proliferation following the manufacturer’s protocol. Similar results wereobtained when BrdU was injected 30 min before harvesting cells.

N-Acetyl-L-cysteine (NAC) treatment. Mice were injected intraperitoneally with100 mg/kg body weight of N-acetyl-L-cysteine (NAC; Sigma, MO) in phosphatebuffered saline solution (pH7.4) daily for 2 weeks.

Western blot analysis. Cells were starved in 0.1% serum for 2 hrs and then stimu-lated with Epo (10 U/ml). Lysates were prepared in 13 RIPA lysis buffer (20 mMSodium phosphate, 300 mM sodium chloride, 4 mM EDTA) containing 2% sodiumdeoxycholate, 2% NP-40, 0.2% SDS, 400 lM sodium orthovanadate, 0.2% b-mercaptoethanol, 2 mM PMSF, and 100 mM sodium fluoride. The buffer was alsomixed with protease cocktail inhibitors (Roche; Cat No: 11-697-498-001). The totalprotein was estimated using Bio-Rad Bradford’s Reagent (Cat #500-0006) followingmanufacturer’s instructions. Retrovirally transduced GFP-positive NIH3T3 cells wereFACS sorted and cell lysates were prepared in Laemmli sample buffer (Bio-Rad),resolved by SDS polyacrylamide gel electrophoresis (PAGE) and transferred on toPVDF membranes. Given the size and to enhance resolution, mTOR protein was runseparately. The following primary antibodies were used for western blotting: anti-pSer473 Akt (#4051), anti-Akt (#9272), anti-pThr389 p70 S6 Kinase (#9205), anti-p70 S6 kinase (#9202), anti-pSer2448 mTOR (#2971), anti-mTOR (#2972), anti-pTyr1007/1008 Jak2 (#3776), anti-TSC1 (# 6935), anti-p4EBP1 (Thr37/46) (# 2855),4EBP1 (# 9644), from Cell Signaling Technology, and antiglutamine synthase BD (#610517); all used at 1:1000 dilutions. Anti-Tubulin: Santa Cruz Biotechnology (#sc-8035) anti-Actin (#sc-1616) from Santa Cruz. Anti-JAK2 (#06–255; Upstate Biotech-nology; 1:500). Horseradish peroxidase (HRP)-conjugated secondary antibodies wereused at 1:5000 (Santa Cruz).

Rapamycin treatment. Mice received intraperitoneal administration of 4 mg/kgbody weight of rapamycin (Enzo Life Sciences, NY) in PBS1 5% Tween 801 5%PEG4001 4% Ethanol during five consecutive days/week for 2 weeks.

Hematological studies. Blood samples were obtained from the cava vein rightafter sacrificing the mice and collected in EDTA or Heparin. Complete bloodcounts (CBC) were measured with an Advia 120 analyzer.

Plasma Epo measurement. Plasma Epo concentrations were determined by theQuantikineVR ELISA kit for mouse erythropoietin from R&D Biosystems (Minneap-olis, MN) according to the manufacturer’s instructions.

RNA isolation and QRT–PCR. Was performed as previously described [41,45,46].Fluidigm—96.96 dynamic array IFC. For fluidigm dynamic array performance,

specific target amplification (STA) was performed according to the manufacturerÇsprotocol (PN 100-3488 B1). Briefly, cDNA was pre-amplified using the TaqManVR

PreAmp Master Mix (Applied Biosystems) for the 96 genes of interest. The amplifi-cation parameters were as follows: 95�C for 100, followed by 12 cycles at 95�C for1500 and 60�C for 40 . After STA, we performed exonuclease I treatment as recom-mended by the manufacturer. Briefly, Exonuclease I and Exonuclease I buffer (New

RESEARCH ARTICLE FOXO3/mTOR in Erythroid Maturation

doi:10.1002/ajh.23786 American Journal of Hematology, Vol. 89, No. 10, October 2014 955

England Biolabs) were added to the STA samples, and samples were then incubatedfor 300 at 37�C, followed by the enzyme inactivation at 80�C for 150 . Finally, toload the dynamic array IFC, samples were prepared with the SsoFast EvaGreenSupermix with Low ROX (Bio-Rad) and 203 DNA Binding Dye Sample LoadingReagent (Fluidigm). On the other hand, primers were diluted with Assay LoadingReagent (Fluidigm) and DNA Suspension Buffer (Teknova). After priming the963 96 chip in the IFC Controller MX, samples and primers were loaded into theirrespective inlets. The chip was then loaded by the IFC Controller MX (BioMarkTM

HD System). The chip was run following the GE 963 96 PCR1Melt v2.pcl proto-col in the Biomark using the Data Collection Software (Fluidigm). Results wereobtained with the Fluidigm Real-Time PCR Analysis software (Fluidigm) and fur-ther analyzed by the 22DDCt method. b actin was used as a loading control. Resultsshown as fold-change relative to Gate I wild type controls. Primer specific sequen-ces are listed in Supporting Information Table II.

Statistical analysis. Fluidigm data was normalized using standard 22DDCt methodusing actin readings as an internal standard in each series of PCR reactions. In experi-ments other than Fluidigm analysis of gene expression, the unpaired one- and two-tailed Student’s t test was used. A P< 0.05 was considered to be significant.

� ResultsFOXO3-mTOR control erythroblast cell cycling

To examine mechanisms underlying abnormalities of Foxo3 mutanterythroblast cycling and maturation we analyzed Epo-activated signal-ing pathways in primary mouse Foxo32/2 erythroid precursors. Pri-mary lineage negative (lacking mature erythroid cells) wild type (WT)and Foxo32/2 bone marrow cells cultured under an erythroid differen-tiation condition for 18 hrs (18 h) [33,42] produced �60% TER1191

cells (Supporting Information Fig. 1). These cells were serum starvedfor 2 h and stimulated with Epo before lysates were analyzed byimmunoblotting (Fig. 1). As anticipated JAK2 protein tyrosine kinasethat is necessary for Epo receptor (EpoR) signaling and its downstreameffector AKT, were rapidly phosphorylated in response to Epo [16,18](Fig. 1). In addition, mTOR protein kinase and its downstream targetribosomal S6 protein kinase 1 (S6K1) were phosphorylated in imma-ture mouse erythroblasts. Epo stimulation of primary Foxo3-null eryth-roblast precursors also resulted in increased JAK2 phosphorylation asin wild-type erythroblasts (Fig. 1). This was associated with enhancedphosphorylation of signaling proteins AKT, mTOR, and mTOR targetS6K1 in Foxo3-null erythroblast precursors at levels markedly higherthan the control. Up to 1 hr after Epo stimulation the AKT/mTOR/S6K1 remained highly phosphorylated in Foxo3-deficient erythroid pre-cursors with a distinct kinetic from that observed in wild type cells. Tofurther assess whether mTOR signaling was activated in Foxo3 mutanterythroblasts, we compared the phosphorylation of another down-stream target of mTOR, the eukaryotic initiation factor 4E (eIF4E)-binding protein 1 (4EBP1). mTORC1 phosphorylation and inhibitionof 4EBP1 has a more direct impact on mRNA translation [34]. Phos-phorylation of 4EBP1 in response to Epo stimulation was increased inprimary freshly isolated Foxo3 mutant erythroblasts (TER 1191) (Fig.1B, top panel) serum starved for 2 hrs as compared to control cells(quantification, Fig. 1B, bottom panel). Detection of pAKT, pS6K1 andp4EBP1 but not pJAK2 at time 0 even in the absence of Epo stimula-tion in Foxo3 mutant erythroblasts (Fig. 1) might indicate constitutiveactivation of mTOR signaling independent of JAK2. These results sug-gested that loss of FOXO3 may alter the signaling response to Epo inerythroblasts leading to prolonged activation of AKT/mTOR signaling(Fig. 1, see schematic on top).

As FOXO3 is a key regulator of oxidative stress, and ROS modulateprotein phosphorylation [14,47] we evaluated whether ROS are impli-cated in the alteration of EpoR signaling in Foxo3 mutant erythroblasts.In vitro treatment with ROS scavengers N-acetyl cysteine (NAC) for 2h had a noticeable effect on reducing phosphorylation of JAK2, AKT,mTOR, and S6K protein in primary Foxo3 mutant bone marrow eryth-roblasts (Fig. 2, compare lanes 8, 9 to 11, 12). These results suggest thatROS contribute to the enhanced phosphorylation of these proteins inFoxo3 mutant erythroblasts in response to Epo. NAC treatment also

reduced the levels of JAK2/AKT/mTOR/S6K1 phosphorylation in wild-type primary erythroid cells (Fig. 2, compare lanes 2, 3 to 5, 6).Although under these conditions, NAC reduced notably phosphoryl-ated AKT, mTOR and S6K in both wild type and Foxo3 mutant eryth-roblasts, the effect of NAC on pJAK2 was less pronounced (Fig. 2). Thephosphorylated form of ribosomal protein S6 (pS6) is a target of S6K1and a reliable indicator of mTORC1 activity [35]. In agreement withROS effects on mTORC1 activity, in vivo treatment with NAC reducedthe frequency of pS6 expressing cells and the levels of pS6 (SupportingInformation Fig. 2 and data not shown). These results are consistentwith the notion that redox state modulates mTOR signaling [48,49],and suggest that increased ROS mediate at least partially overactivationof mTOR signaling in immature Foxo32/2 erythroblasts in vitro (Fig. 1,Supporting Information Fig. 2).

We next investigated the function of mTOR signaling in normalerythropoiesis. We used RNA interference to inhibit the expression ofS6K1, a direct mTORC1 target [34,35,50] in primary erythroid pro-genitors (Fig. 3A). Inhibition of S6K1 in bone marrow cultures usingtwo distinct shRNA sequences resulted in significant reduction ofBFU-E and CFU-E-derived erythroid cell colony formation (Fig. 3B).The degree of inhibition of erythroid progenitor cell-colony formationwas consistent with the relative expression of S6K1 in response toRNA interference targeting in BFU-Es but not in CFU-Es (Fig. 3B)suggesting that while mTORC1 activation is required for both BFU-ES and CFU-Es, BFU-Es more than CFU-Es are highly sensitive to

Figure 1. Jak2-AKT–mTOR signaling pathway is overactivated in culturedFoxo32/2 bone marrow erythroid cells. (A) Schematic of EpoR-mediatedactivation of JAK2-AKT-mTOR on the left. Western blot analysis of phos-phorylation of signaling proteins. WT and Foxo32/2 lineage-negative bonemarrow cells were isolated and cultured under erythroid condition for 18hrs, serum- and cytokine starved for 2 hrs and stimulated with Epo (10 U/ml) for the indicated time points in vitro before preparing the whole cellextract. One representative of two experiments is shown. (B) Schematic ofactivation of feedback loop JAK2-AKT-mTOR in the absence of FOXO3 onthe left. TER 1191 WT and Foxo32/2 erythroblasts were freshly isolatedfrom mice serum- and cytokine starved for 2 hrs and stimulated with Epo(10 U/ml) for the indicated time points in vitro before preparing the wholecell extract to analyze of p4EBP1 (quantification of bands in the bottompanel). [Color figure can be viewed in the online issue, which is availableat wileyonlinelibrary.com.]

Zhang et al. RESEARCH ARTICLE


http://wileyonlinelibrary.com

levels of mTOR signaling. These results may reflect distinct sensitivityof these progenitors to EpoR signaling [51]. These combined resultssuggest that mTORC1 signaling is required for in vitro generation oferythroid progenitor cell-derived colonies.

As mTOR signaling is central to cell growth and proliferation, weasked whether inhibition of mTOR signaling by rapamycin that is aspecific inhibitor of mTORC1 [35] has any impact on erythroblastprecursor cell cycling. To address this, we used the thymidine analog5-bromo-2-deoxyuridine (BrdU) that is incorporated in dividing cellsin vivo. Mice were treated in vivo with rapamycin for 2 weeks andinjected with BrdU 30–60 min before harvesting the bone marrow.Erythroblast cell cycle distribution was analyzed at distinct stages of

maturation by flow cytometry examination of BrdU incorporationand the DNA marker 7-aminoactinomycin D (7-AAD). Maturingerythroblasts (proerythroblasts, basophilic erythroblasts, and poly-chromatophilic erythroblasts) expressing similar levels of transferrinreceptor CD71 [44] (and data not shown) were distinguished accord-ing to their size (forward scatter, FSC) and surface expression ofTER119 and CD44 [44] (Fig. 4A, FACS plot). The analysis yieldedimportant insights into primary bone marrow erythroblast cycling. Asignificant fraction (up to 75%) of wild type bone marrow erythro-blasts (Gates I to III proerythroblasts, basophilic, polychromatophilic)were in the S phase (Fig. 4B). Rapamycin treatment strongly blockederythroblast cell cycle progression at the G1/S transition phase inimmature erythroblasts (Gates I and II, proerythroblasts and baso-philic erythroblasts respectively) and reduced the fraction of proery-throblasts (Gate I) in the S phase. In addition, rapamycin treatmentstrongly reduced the fraction of basophilic erythroblasts (Gate II) thatwere in the G2/M transition without significant effects on cells atlater stages of maturation (Gates III and IV). In contrast to its effecton proerythroblasts, rapamycin treatment resulted in an increase,rather than a decrease in the fraction of basophilic erythroblasts(Gate II) that were in the S phase (Fig. 4B). These results indicatethat mTOR signaling is required for normal cell cycle progression ofimmature (Gates I and II) erythroblasts.

Immature Foxo3 mutant erythroblasts exhibit cell cycle defects asso-ciated with a failure to fully mature [5]. As a consequence, the rate ofFoxo3 mutant erythroblast maturation is decreased. We evaluatedwhether the overactivation of mTOR signaling in Foxo3 mutant eryth-roblasts contributes to their cell cycle alteration and defective matura-tion. BrdU analysis of Foxo3 mutant erythroblasts confirmed Foxo3mutant erythroblast cell cycle alterations [5] (Fig. 4A,B). Relative tocontrols, a reduced fraction of early Foxo3 mutant erythroblasts (GatesI) was in the S phase of cell cycle as previously observed [5]. In additionto a slight increase in the fraction of immature Foxo3 mutant precur-sors (Gates I and II) in G0/G1 and a slight increase in the S phase (GateII), a noticeable fraction of these cells was blocked at the G2/M transi-tion (Fig. 4B). Rapamycin treatment improved significantly the G2 toM transition of Foxo3 mutant erythroblasts (Fig. 4B, Gate I) suggestingthat the G2/M block is mediated in part by the overactivation of mTORsignaling in Foxo3 mutant erythroblasts. As observed in control cells,

Figure 3. RNA interference inhibition of mTOR signaling reduces bone marrow erythroid colony-formation capacity. A. Inhibition efficiency of short hairpinRNAs targeting S6 kinase was tested by transducing NIH3T3 cells with the indicated short hairpin RNAs and analyzing S6K1 expression by Western blot.B. Number of bone marrow BFU-E- and CFU-E-derived colonies formed by lineage negative cells transduced with short hairpin RNAs targeting S6 kinase orscrambled control are shown. Results shown are mean6SEM of triplicates; *P<0.05; **P<0.01; ***P<0.001; Student’s t test. [Color figure can be viewed inthe online issue, which is available at wileyonlinelibrary.com.]

Figure 2. Overactivation of Jak2-AKT–mTOR signaling pathway in culturedFoxo32/2 bone marrow erythroid cells is partly mediated by ROS. Westernblot analysis of phosphorylation of signaling proteins. WT and Foxo32/2

bone marrow lineage-negative cells were isolated and cultured under ery-throid condition for 18 hrs in the presence or absence of NAC (100 lM),serum- and cytokine starved for 2 hrs and stimulated with Epo (10 U/ml)for the indicated time points in vitro before preparing the whole cellextract.




the in vivo treatment with rapamycin enhanced the fraction of Foxo3mutant cells (Gate I) in G0/G1. The effects of rapamycin on wild typeand Foxo3 mutant cell cycle were highly similar at later stages of eryth-roblast maturation (Fig. 4B, lower panels).

As anticipated, the effect of rapamycin in early precursors (Gates Iand II) mediated the inhibition of mTORC1 signaling as shown byreduced phosphorylation of S6 at distinct stages of erythroblast matu-ration (Fig. 4C). In contrast, rapamycin treatment did not signifi-

cantly reduce the frequency of pAKT-expressing erythroblasts orlevels of pAKT (data not shown) suggesting that the AKT activatormTORC2 [35] might not be involved in these effects. Together theseresults suggested the effects of rapamycin in early erythroblasts (GatesI and II) are likely mediated by mTORC1.

Rapamycin treatment increases RBC production inFoxo32/2 mice in vivo

In vivo treatment with rapamycin increased significantly RBCnumbers and hemoglobin concentration in Foxo32/2 peripheralblood without modulating significantly the total number of erythroid(TER1191) cells (Table I; Fig. 5A,B). Importantly, rapamycin treat-ment tipped the balanced production of erythroid cells towards termi-nal maturation (Fig. 5C,D). A picture emerging from these findings isthat by increasing the fraction of cells in G1, rapamycin may blockcell cycle progression in immature erythroblasts (proerythroblasts,Gate I), reduce the fraction of cells in the S phase, and induce cellcycle exit in immature Foxo3 mutant erythroblasts. In agreementwith this interpretation, the ratio of mature to immature erythroblastsin the bone marrow also increased in response to rapamycin treat-ment (Fig. 5C,D).

The effect of rapamycin on maturation may be due to its intrinsiceffect in erythroblasts, or extrinsic due to its effect on bone marrowmicroenvironment [52,53]. To distinguish between these alternatives,we isolated E14.5 fetal livers that are site of definitive erythropoiesisand followed the ex vivo differentiation of erythroblasts in the presenceor absence of rapamycin. This approach enabled us to monitor pre-cisely stages of erythroblast maturation in response to treatment. Addi-tion of rapamycin to erythroblasts that were all at the proerythroblaststage reduced the frequency of immature (in Gate I) in favor of matureerythroblasts (in Gate III) similar to the effect seen in in vivo experi-ments (Supporting Information Fig. 3). These findings further supportthe notion that overactivation of mTOR signaling in immature Foxo3mutant erythroblasts reduces their rate of maturation.

Elevated erythropoietin (Epo) is associated with the state of ineffec-tive erythropoiesis and is likely to take part in abnormally enhancedsignaling in immature erythroblasts [54,55]. We argued that if theabnormal erythropoiesis in Foxo3 mutant mice is similar to ineffectiveerythropoiesis, then circulating Epo levels should be increased inFoxo32/2 mice. In addition, rapamycin treatment should result inreduced Epo levels in Foxo3 mutant peripheral blood. Consistent withan ineffective erythropoiesis phenotype, circulating Epo levels were sig-nificantly increased in Foxo3 mutant peripheral blood (SupportingInformation Fig. 4). In vivo treatment with rapamycin reduced theincreased levels of circulating Epo in Foxo3 mutant peripheral blood(Supporting Information Fig. 4) but did not significantly modulate thelevels of Epo in wild-type mice. These results further supported theconcept that the abnormal Foxo32/2 erythropoiesis has overlappingfeatures with ineffective erythropoiesis. They also raised the potential

Figure 4. In vivo rapamycin treatment inhibits cell cycle progression inbone marrow immature erythroblasts. A. Flow cytometry strategy to distin-guish four different bone marrow erythroid populations with increasingdegree of maturation (Gates I to IV) according to their TER119 and CD44cell surface expression and forward scatter (FSC) properties after fixationand permeabilization. Cell cycle analysis (B) and phosphorylated S6 (C)were analyzed by flow cytometry in each erythroid precursor populationfrom WT and Foxo32/2 mice treated with rapamycin (4 mg/kg*day) or con-trol vehicle for 2 weeks. In B each graph represents a distinct populationand numbers within each graph represent percentage of cells in G0/G1, S,and G2/M phases of cell cycle for each population. One representative oftwo independent experiments is shown (n53 mice in each group,mean6SEM; *P<0.05; **P<0.01; ***P<0.001 between vehicle- andrapamycin-treated; #P<0.05; ##P<0.01; ###P<0.001 between WT andFoxo32/2, Student’s t test). [Color figure can be viewed in the online issue,which is available at wileyonlinelibrary.com.]

TABLE I. Peripheral Blood Erythrocyte Parameters

WT Vehicle WT Rapamycin Foxo3-/- Vehicle Foxo3-/- Rapamycin Th3/1 Vehicle Th3/1 Rapamycin

RBC, 3106 10.460.3 10.960.2 9.060.1b 9.860.4a 7.560.2b 8.460.2a

HGB (g/l) 15.060.3 15.460.3a 14.860.1 16.060.7a 7.960.3b 9.260.3a

HCT 54.66 1.6 56.56 1.9 52.860.9 55.76 2.4 32.26 1.3b 36.46 1.3a

MCV (fL) 52.86 1.2 52.56 1.6 58.560.8b 56.860.9 42.960.8b 43.260.7MCH (pg) 14.660.2 14.960.2 16.460.2b 16.260.4 10.360.3b 10.560.2MCHC (g/l) 27.460.5 28.160.5 28.060.3 28.660.3 25.060.8b 25.260.7Retic (%) 3.660.3 3.860.5 6.660.5b 7.460.4 25.86 1.9b 28.262.2

n5 9 n5 8 n5 10 n59 n59 n5 10

Mice treated with rapamycin or vehicle control for two weeks.a P<0.05 between vehicle- and rapamycin-treatedb P<0.05 between WT and Foxo32/2 or Th3/1.




that rapamycin might have a beneficial effect on ineffectiveerythropoiesis.

Together, these results (Figs. 4 and 5, Supporting Information Fig.3, Table I) suggest that rapamycin enhances erythroid maturation byinducing cell cycle exit of immature erythroblasts resulting inincreased RBC production. These findings reflect the amplitude ofdynamic changes that are detected specifically in immature erythro-blasts (proerythroblasts) as compared to cells at later stages of matu-ration as previously reported [56,57].

FOXO3-mTOR regulate erythroblast metabolic geneexpression in vivo

ROS elimination is maintained in part by sustained generation ofreduced glutathione as a result of glucose metabolism and a functionalpentose phosphate pathway (Fig. 6A). Given the function of mTORprotein kinase in cellular metabolism [34,35] and its sensitivity to redoxmodulation [48], in particular in erythroblast precursors (Fig. 2, Sup-porting Information Fig. 2) we explored the possibility that the impacton erythroid cell production might be mediated by mTOR’s influenceon erythroblast metabolism. Using Fluidigm microfluidics technology thatenables monitoring the expression of 96 genes in 96 samples all at once(equivalent of 9216 real-time PCR), we interrogated the expression of anarray of metabolic genes (Fig. 6). Wild type and Foxo3 mutant mice weretreated for 2 weeks with rapamycin or control vehicle and RNA from

erythroblasts at distinct stages of maturation was isolated and subjected toFluidigm QRT-PCR analysis (Fig. 6). These experiments led to severalinteresting observations. First, in agreement with earlier observations[10,11], the transcript of the majority of metabolic genes surveyed in wild-type erythroblasts was at least as highly expressed in early immature eryth-roblasts (proerythroblasts, Gate I) as it was in maturing erythroblasts(orthochromatic erythroblast, Gate IV). Strikingly, we found the expres-sion of many metabolic genes specifically implicated in glucose metabo-lism was highly altered (mostly reduced) in Foxo3 mutant erythroblasts atdifferent stages of maturation, specifically in immature erythroblasts (Fig.6B). Expression of several of these genes in Foxo3 mutant erythroblastswas sensitive to rapamycin treatment (Fig. 6B). Specifically, in vivo treat-ment with rapamycin normalized and/or significantly improved the tran-script level of several genes including pyruvate kinase M2 (Pkm2), aldolaseA encoding fructose-bisphosphate aldolase (AldoA), Enolase 1 a (EnoA)(Eno1/EnoA), glyceraldehyde-3-phosphate dehydrogenase (Gapdh) andLactate Dehydrogenase (Ldh) B (LdhB) in Foxo3 mutant immature eryth-roblasts (Gates I and II cells, Fig. 6B). These results indicate that alteredexpression of many metabolic specifically glycolytic transcripts in Foxo3mutant immature erythroblasts is likely due to either decreased transcrip-tion (due to lack of FOXO3), or enhanced activation of mTOR signaling(or likely both) in Foxo3mutant erythroblasts.

In addition to the redox state, some of the direct targets of FOXO3[58,59] like Rictor, Tsc1, and glutamine synthetase (GS) [23] are impli-cated in the activation of mTOR. Rictor is part of the mTORC2 com-plex [60,61]. Rictor transcripts were slightly but significantly reduced inFoxo3 mutant erythroblast (Fig. 7A) supporting a potential decreasedassembly of mTORC2 in favor of mTORC1 complex in erythroblasts

Figure 5. In vivo rapamycin treatment increases erythroid cell maturation.Flow cytometric analysis of bone marrow erythroid cell distribution in wildtype and Foxo32/2 mice treated with rapamycin (4 mg/kg*day) or controlvehicle for 2 weeks. A. Schematic of flow cytometry analysis of five dis-tinct erythroid populations according to their TER119 and CD44 surfaceexpression and FSC properties. B. Percentage of TER1191 cells within thebone marrow. C. Distribution of TER1191 cells in each of the five gatesshown in A. D. Ratio of mature cells (Gates IV and V combined) relative tothe total TER1191 erythroid population. *P<0.05 between vehicle- andrapamycin-treated; #P<0.05 between WT and Foxo32/2 (n5 12, mean-6SEM; Student’s t test). Veh: vehicle, Rapa: rapamycin. [Color figure canbe viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 6. Alteration of transcripts of metabolic enzymes in Foxo32/2

immature erythroblasts. A. Schematic of glycolytic pathway and its interac-tion with tricarboxylic acid (TCA) cycle and pentose phosphate pathway,pyruvate kinase (PK). B. QRT-PCR expression analysis by Fluidigm micro-fluidics technology of metabolic genes in bone marrow Gates I–IV erythro-blasts from WT and Foxo32/2 mice treated with rapamycin (4 mg/kg*day)or control vehicle for 2 weeks. Quantification of target genes is relative tob actin. Results are mean6SEM of three cDNAs, each generated fromone mouse. *P<0.05 between vehicle- and rapamycin-treated; #P<0.05between WT and Foxo32/2. [Color figure can be viewed in the onlineissue, which is available at wileyonlinelibrary.com.]





[60,62]. While transcripts for Tsc1 that is an upstream negative regula-tor of mTOR signaling [63] were highly reduced in Foxo3 mutanterythroblasts (Fig. 7A), TSC1 protein was not significantly altered (Fig.7B, quantification in the right panel) suggesting that reduction of TSC1transcript expression was unlikely to mediate activation of mTOR inFoxo3 mutant erythroblasts. On the other hand the transcript for gluta-mine synthetase (GS) that is a negative regulator of mTOR [23] washighly and significantly reduced in Foxo3 mutant erythroblasts at allstages of maturation, including in early precursors (Gates I and II) (Fig.6B). GS protein expression was also significantly reduced in freshly iso-lated TER 1191 Foxo3 mutant erythroblasts (Fig. 7C). These findingsraise the possibility that reduced expression of GS might be implicatedin mTORC1 activation in Foxo3 mutant erythroblasts.

Increased proliferation of immature erythroblasts that are unable tofully mature, as seen in Foxo3 mutant erythroblasts (Figs. 4 and 5, Sup-porting Information Figs. 3,5) characterizes ineffective erythropoiesisobserved in many erythroid disorders including b-thalassemia [52,55].In b-thalassemia, the generation of erythroid cells is hampered by anumber of intricate mechanisms triggered partly by redox imbalance

that together lead to an exacerbated erythropoiesis that fails to producesufficient numbers of mature functional RBC [54,64]. Given that rapa-mycin treatment improved the ineffective erythropoiesis of Foxo3mutant mice, we evaluated the potential effect of rapamycin on b-thalassemic erythropoiesis. Strikingly, a 2-week in vivo treatment withrapamycin increased the number of RBC (*P< 0.05, Table I), hemato-crit (36.46 1.3% in rapamycin treated mice versus 32.26 1.3% in micetreated with control vehicle) and specifically hemoglobin by over 1 g/dlin the peripheral blood of a model of b-thalassemia (intermediaHbbth3/1, Th3/1) [55]. These findings suggest that activation of mTORsignaling in erythroblasts may contribute to the b-thalassemic pheno-type. The effects of rapamycin on bone marrow and spleen b-thalassemic erythroblast cell cycling was at best modest (data notshown) despite its effect on Foxo3 mutant erythroblast cycling (Fig. 4)suggesting additional mechanisms are involved. Nonetheless, consistentwith the notion that Foxo3 mutant and b-thalassemic erythropoiesismay share common features, there was a remarkable similarity betweenalteration of metabolic gene expression in b-thalassemic and Foxo3mutant erythroblasts (Fig. 6B). In particular, expression of Pkm2, Gpi1,

Figure 6. Continued



EnoA-Eno1, G6pd1, Pgk1, LdhA, LdhB, Idh1, and Idh2 was similarlyaltered in b-thalassemic and Foxo3 mutant erythroblasts. While rapa-mycin treatment increased expression of b major in b-thalassemic ery-throid cells (Fig. 8), the effect of rapamycin on metabolic genes wasrestricted to increasing the transcript expression of Pdk1 and Idh1 inearly and late stages of b-thalassemic erythroblast maturation respec-tively (Fig. 6B).

� DiscussionDefective proliferation of Foxo3 mutant erythroblasts that fail to

mature [5] is reminiscent of proliferation and maturation defects lead-ing to ineffective erythropoiesis [54]. Here we showed that defects ofFoxo3 mutant erythroblast maturation are mediated in part by hyperac-tivation of mTOR signaling In immature erythroblasts. These defectslead specifically to a G2/M block in Foxo3 mutant erythroblast cycling.Our results suggest that physiological cooperation of mTOR signalingwith FOXO3 is key to the control of cell cycle progression of immatureerythroblasts and their rate of maturation in vivo (see Model, Fig. 9).We have previously shown that FOXO3 is required for erythroblast cellcycling [5]. The defects in the G2/M progression of FOXO3 mutanterythroblasts isolated from C57Bl6 mice shown here are similar to whatwe observed in Foxo3 mutant hematopoietic stem cells [41]. However,we failed to detect the G2/M block in a heterogeneous population oferythroblasts isolated from mice on a mixed genetic background(FVB3129). Although our data does not implicate directly redox mod-ulations in mTOR activation in primary immature erythroblasts invivo, in contrast to our observations in primitive primary Foxo3 mutantmyeloid progenitors [45] overall our findings support a model in whicha feedback loop that is in part mediated by a redox switch amplified byloss of FOXO3 activates mTOR signaling in primary erythroblasts.Together these results indicate that the outcome of FOXO3/mTORinteractions may be cell context dependent [45]. Activation of mTORin primitive primary Foxo3 mutant myeloid progenitors [45] result inincreased cell cycle whereas in Foxo3 mutant immature erythroblastsmTOR activation leads to relative cell cycle delay (Fig. 4).

Erythroblasts have a remarkable capacity for proliferation and assuch are likely to be especially sensitive to metabolic perturbations[43]. FOXO3 may have a function in regulating key metabolic genes(Fig. 6) in erythroblasts in agreement with recent studies in neuralstem and progenitor cells [65]. Defective expression of metabolicgenes may contribute to the accumulation of ROS in Foxo3 mutanterythrocytes resulting in hemolytic anemia and ROS-mediated reduc-tion of RBC lifespan observed in Foxo32/2 mice [5]. It would beinteresting to investigate whether in addition to oxidative stress (Fig.2) [5], reduction of glutamine synthetase expression (Figs. 6B,7C)contributes to the overactivation of mTOR signaling of Foxo3 mutanterythroblasts in vitro and/or in vivo [23].

Figure 8. In vivo rapamycin treatment improves b globin expression in b-thalassemia erythroblasts. WT and Th3/1 mice treated with rapamycin (4 mg/kg*day) or control vehicle for 2 weeks. QRT-PCR expression analysis of globin genes in WT and Th3/1 bone marrow erythroblasts from Gates I to IV. QRT-PCR expression analysis was performed using Fluidigm microfluidics technology and quantification of target genes is relative to b actin. Results are mean-6SEM of three cDNAs, each generated from one mouse. Veh: vehicle, Rapa: Rapamycin. [Color figure can be viewed in the online issue, which is availableat wileyonlinelibrary.com.]

Figure 7. Glutamine synthetase protein is decreased in Foxo3 mutanterythroblasts. A. QRT-PCR expression analysis in freshly isolated erythro-blasts at distinct stages of maturation. B. Western blot and band quantifi-cation (right panel) of TSC1 in freshly isolated bone marrow TER 1191

erythroblasts from two distinct mice for each genotype. C. Western blotanalysis of glutamine syntethase in freshly isolated bone marrow TER 1191

erythroblasts from four distinct mice for each genotype. Band quantifica-tion in the right. Results shown are mean6SEM of triplicates; #P<0.05;***P<0.001; Student’s t test. [Color figure can be viewed in the onlineissue, which is available at wileyonlinelibrary.com.]





One of the most unexpected findings was the increased productionof mature RBC in response to rapamycin treatment in b-thalassemicmice. b-thalassemia arises as a consequence of mutations of b globingene and precipitation of excess unmatched a globin, resulting in anincrease in the pool of free iron that triggers enhanced redox reactionsand damage to RBC membrane. These changes ultimately lead to exces-sive compensatory proliferation of erythroid precursors that fail tomature [2,55,66,67]. Consistent with the therapeutic effect of JAK2 pro-tein tyrosine kinase inhibitors [55], our results implicate mTOR signal-ing in the ineffective b-thalassemic erythropoiesis. It is noteworthy thatrapamycin induces g-globin mRNA and fetal hemoglobin (HbF) pro-duction in cultured human erythroid progenitors from b-thalassemicpatients [68]. It will be important to investigate mechanisms wherebyrapamycin ameliorates b-thalassemic anemia and explore potentialeffects on protein translation, iron flux [69] and immune response [35].

One of the noticeable findings in these studies was ROS-mediatedphosphorylation of JAK2 in primary mouse erythroblasts. JAK2 pro-tein tyrosine kinase, an essential component of EpoR signaling isknown to generate ROS upon stimulation and be redox modulated, aproperty that is relevant to myeloproliferation [70]. However whetheroxidation activates or inhibits JAK2 has been debated [71–74]. Whilethe mechanism of JAK2 hyperphosphorylation in Foxo3 mutanterythroblasts is unclear, reduced expression of Lnk (SH2B3), a nega-tive regulator of JAK2 phosphorylation, in Foxo3 mutant erythro-blasts might be implicated (Supporting Information Fig. 6) [45].

Collectively, these studies support the notion that activation ofmTOR signaling as a result of loss of FOXO3 function [45] might beimplicated in the pathogenesis of ineffective erythropoiesis as seen inFoxo3 mutant mice. Future studies should elucidate whether and howmetabolic abnormalities associated with overactivation of mTOR sig-naling contribute to erythroblast cell cycle defects.

� Authors’ ContributionsXZ, designed experiments, performed experiments, analyzed data

and participated in writing the paper. GC, designed experiments, per-formed experiments, analyzed data and participated in writing thepaper. SY, PR, SKM, JB, RL designed experiments, performed experi-ments, analyzed data. VDE, performed experiments. MHB facilitatedthe set up of the in vitro assay. CB, analyzed data and participated indiscussions of the data and editing. SR, participated in discussions ofthe data and editing. DP, analyzed data. SG, designed experiments,analyzed data and wrote the paper.

� AcknowledgmentsWe thank Brigitte Izac for technical help, Dr. Jane Little (Case

Western Reserve University School of Medicine) for critical readingof the manuscript and the Flow Cytometry Shared Research Facilityat Icahn School of Medicine at Mount Sinai School.

� References1. McDevitt MA, Xie J, Gordeuk V, et al. The ane-

mia of malaria infection: Role of inflammatorycytokines. Curr Hematol Rep 2004;3:97–106.

2. Rivella S. Ineffective erythropoiesis and thalasse-mias. Curr Opin Hematol 2009;16:187–194.

3. Neumann CA, Krause DS, Carman CV, et al.Essential role for the peroxiredoxin Prdx1 inerythrocyte antioxidant defence and tumoursuppression. Nature 2003;424:561–565.

4. Friedman JS, Rebel VI, Derby R, et al. Absenceof mitochondrial superoxide dismutase resultsin a murine hemolytic anemia responsive totherapy with a catalytic antioxidant. J Exp Med2001;193:925–934.

5. Marinkovic D, Zhang X, Yalcin S, et al. Foxo3is required for the regulation of oxidative stressin erythropoiesis. J Clin Invest 2007;117:2133–2144.

6. Yu D, dos Santos CO, Zhao G, et al. miR-451protects against erythroid oxidant stress byrepressing 14-3-3zeta. Genes Dev 2010;24:1620–1633.

7. Kong Y, Zhou S, Kihm AJ, et al. Loss of alpha-hemoglobin-stabilizing protein impairs erythro-poiesis and exacerbates beta-thalassemia. J ClinInvest 2004;114:1457–1466.

8. Valentine WN. Metabolism of human erythro-cytes. Studies in health and disease. Arch InternMed 1975;135:1307–1313.

9. Bossi D, Giardina B. Red cell physiology. MolAspects Med 1996;17:117–128.

10. Nijhof W, Wierenga PK, Staal GE, et al.Changes in activities and isozyme patterns ofglycolytic enzymes during erythroid differentia-tion in vitro. Blood 1984;64:607–613.

11. Kim HD, Koury MJ, Lee SJ, et al. Metabolicadaptation during erythropoietin-mediated ter-

minal differentiation of mouse erythroid cells.Blood 1991;77:387–392.

12. Nemoto S, Finkel T. Redox regulation of fork-head proteins through a p66shc-dependent sig-naling pathway. Science 2002;295:2450–2452.

13. Kops GJ, Dansen TB, Polderman PE, et al.Forkhead transcription factor FOXO3a protectsquiescent cells from oxidative stress. Nature2002;419:316–321.

14. Zhang X, Rielland M, Yalcin S, et al. Regulationand function of FoxO transcription factors innormal and cancer stem cells: What have welearned? Curr Drug Targets 2011;12:1267–1283.

15. van den Berg MC, Burgering BM. Integratingopposing signals toward forkhead box O. Anti-oxid Redox Signal 2011;14:607–621.

16. Kashii Y, Uchida M, Kirito K, et al. A memberof Forkhead family transcription factor,FKHRL1, is one of the downstream moleculesof phosphatidylinositol 3-kinase-Akt activation

Figure 9. Proposed model for FOXO3/ROS/mTOR regulation of erythroblast maturation. During erythroid differentiation FOXO3 becomes nuclear and active, keepingROS levels under control through transcriptional regulation of antioxidant enzymes and certain metabolic enzymes. The absence of FOXO3 leads to increased ROS lev-els that further activate mTOR, which in turn influences glycolytic enzymes and erythroblast cell cycling leading to decrease in erythroblast maturation. Red arrowsindicate the main findings described in this article. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]



pathway in erythropoietin signal transduction.Blood 2000;96:941–949.

17. Uddin S, Kottegoda S, Stigger D, et al. Activa-tion of the Akt/FKHRL1 pathway mediates theantiapoptotic effects of erythropoietin in pri-mary human erythroid progenitors. BiochemBiophys Res Commun 2000;275:16–19.

18. Ghaffari S, Jagani Z, Kitidis C, et al. Cytokinesand BCR-ABL mediate suppression of TRAIL-induced apoptosis through inhibition of fork-head FOXO3a transcription factor. Proc NatlAcad Sci USA 2003;100:6523–6528.

19. Bakker WJ, Blazquez-Domingo M, Kolbus A,et al. FoxO3a regulates erythroid differentiationand induces BTG1, an activator of protein argi-nine methyl transferase 1. J Cell Biol 2004;164:175–184.

20. Bakker WJ, van Dijk TB, Parren-vanAmelsvoort M, et al. Differential regulation ofFoxo3a target genes in erythropoiesis. Mol CellBiol 2007;27:3839–3854.

21. Franco SS, De Falco L, Ghaffari S, et al. Resver-atrol accelerates erythroid maturation by activa-tion of FoxO3 and ameliorates anemia inbeta-thalassemic mice. Haematologica 2014;99:267–275.

22. Pourfarzad F, von Lindern M, Azarkeivan A,et al. Hydroxyurea responsiveness in beta-thalassemic patients is determined by the stressresponse adaptation of erythroid progenitorsand their differentiation propensity. Haemato-logica 2013;98:696–704.

23. van der Vos KE, Eliasson P, Proikas-Cezanne T,et al. Modulation of glutamine metabolism bythe PI(3)K-PKB-FOXO network regulatesautophagy. Nat Cell Biol 2012;14:829–837.

24. Rimmele P, d’Esamard V, Fatih K, et al. Meta-bolic cross talk between Foxo3 and mTOR isessential for hematopoietic stem cell function.Blood 2012;120:a856.

25. Warr MR, Binnewies M, Flach J, et al. FOXO3Adirects a protective autophagy program in hae-matopoietic stem cells. Nature 2013.

26. Sandri M, Sandri C, Gilbert A, et al. Foxo tran-scription factors induce the atrophy-relatedubiquitin ligase atrogin-1 and cause skeletalmuscle atrophy. Cell 2004;117:399–412.

27. Yeo H, Lyssiotis CA, Zhang Y, et al. FoxO3coordinates metabolic pathways to maintainredox balance in neural stem cells. EMBO J2013;32:2589–2602.

28. Camprecios G, Barminko J, Bernitz J, et al.Steady state differences in metabolic propertiesof bone marrow versus spleen erythroid cells.Blood 2013;122:943a.

29. Rimmele P, Bigarella C, Liang R, et al. Aging-like phenotype and defective lineage specifica-tion in SIRT1-deficient hematopoietic stem andprogenitor cells. Stem Cell Rep 2014 (in press)3:1–16.

30. Kadri Z, Maouche-Chretien L, Rooke HM, et al.Phosphatidylinositol 3-kinase/Akt induced byerythropoietin renders the erythroid differentia-tion factor GATA-1 competent for TIMP-1 genetransactivation. Mol Cell Biol 2005;25:7412–7422.

31. Hammerman PS, Fox CJ, Birnbaum MJ, et al.Pim and Akt oncogenes are independent regula-tors of hematopoietic cell growth and survival.Blood 2005;105:4477–4483.

32. Ghaffari S, Kitidis C, Zhao W, et al. AKT indu-ces erythroid-cell maturation of JAK2-deficientfetal liver progenitor cells and is required forEpo regulation of erythroid-cell differentiation.Blood 2006;107:1888–1891.

33. Zhao W, Kitidis C, Fleming MD, et al. Erythro-poietin stimulates phosphorylation and activa-tion of GATA-1 via the PI3-kinase/AKTsignaling pathway. Blood 2006;107:907–915.

34. Howell JJ, Manning BD. mTOR couples cellularnutrient sensing to organismal metabolic home-ostasis. Trends Endocrinol Metab 2011;22:94–102.

35. Zoncu R, Efeyan A, Sabatini DM. mTOR: Fromgrowth signal integration to cancer, diabetes andageing. Nat Rev Mol Cell Biol 2011;12:21–35.

36. Grech G, Blazquez-Domingo M, Kolbus A, et al.Igbp1 is part of a positive feedback loop in stemcell factor-dependent, selective mRNA transla-tion initiation inhibiting erythroid differentia-tion. Blood 2008;112:2750–2760.

37. Geslain R, Uddin S, Liu H, et al. Distinct func-tions of erythropoietin and stem cell factor arelinked to activation of mTOR kinase signalingpathway in human erythroid progenitors. Cyto-kine 2013;61:329–335.

38. Diekmann F, Rovira J, Diaz-Ricart M, et al. mTORinhibition and erythropoiesis: Microcytosis or anae-mia? Nephrol Dial Transplant 2012;27:537–541.

39. Fishbane S, Cohen DJ, Coyne DW, et al. Post-transplant anemia: The role of sirolimus. KidneyInt 2009;76:376–382.

40. Castrillon DH, Miao L, Kollipara R, et al. Sup-pression of ovarian follicle activation in mice bythe transcription factor Foxo3a. Science 2003;301:215–218.

41. Yalcin S, Zhang X, Luciano JP, et al. Foxo3 isessential for the regulation of ataxia telangiecta-sia mutated and oxidative stress-mediatedhomeostasis of hematopoietic stem cells. J BiolChem 2008;283:25692–25705.

42. Ghaffari S, Kitidis C, Fleming MD, et al. Eryth-ropoiesis in the absence of janus-kinase 2: BCR-ABL induces red cell formation in JAK2(2/2)hematopoietic progenitors. Blood 2001;98:2948–2957.

43. England SJ, McGrath KE, Frame JM, et al.Immature erythroblasts with extensive ex vivoself-renewal capacity emerge from the earlymammalian fetus. Blood 2011;117:2708–2717.

44. Chen K, Liu J, Heck S, et al. Resolving the dis-tinct stages in erythroid differentiation based ondynamic changes in membrane protein expres-sion during erythropoiesis. Proc Natl Acad SciUSA 2009;106:17413–17418.

45. Yalcin S, Marinkovic D, Mungamuri SK, et al.ROS-mediated amplification of AKT/mTOR sig-nalling pathway leads to myeloproliferative syn-drome in Foxo3(2/2) mice. EMBO J 2010;29:4118–4131.

46. Zhang X, Yalcin S, Lee DF, et al. FOXO1 is anessential regulator of pluripotency in humanembryonic stem cells. Nat Cell Biol 2011;13:1092–1099.

47. Liang R, Ghaffari S. Stem cells, redox signaling,and stem cell aging. Antioxid Redox Signal2014;20:1902–1916.

48. Sarbassov DD, Sabatini DM. Redox regulationof the nutrient-sensitive raptor-mTOR pathwayand complex. J Biol Chem 2005;280:39505–39509.

49. Yoshida S, Hong S, Suzuki T, et al. Redox regu-lates mammalian target of rapamycin complex 1(mTORC1) activity by modulating the TSC1/TSC2-Rheb GTPase pathway. J Biol Chem 2011;286:32651–32660.

50. Ma XM, Blenis J. Molecular mechanisms ofmTOR-mediated translational control. Nat RevMol Cell Biol 2009;10:307–318.

51. Gregory CJ. Erythropoietin sensitivity as a dif-ferentiation marker in the hemopoietic system:Studies of three erythropoietic colony responsesin culture. J Cell Physiol 1976;89:289–301.

52. Ramos P, Casu C, Gardenghi S, et al. Macro-phages support pathological erythropoiesis inpolycythemia vera and beta-thalassemia. NatMed 2013;19:437–445.

53. Hashimoto D, Chow A, Noizat C, et al. Tissue-resident macrophages self-maintain locallythroughout adult life with minimal contributionfrom circulating monocytes. Immunity 2013;38:792–804.

54. Rivella S. The role of ineffective erythropoiesisin non-transfusion-dependent thalassemia.Blood Rev 2012;26 Suppl 1:S12–S15.

55. Libani IV, Guy EC, Melchiori L, et al. Decreaseddifferentiation of erythroid cells exacerbates

ineffective erythropoiesis in beta-thalassemia.Blood 2008;112:875–885.

56. Peslak SA, Wenger J, Bemis JC, et al. EPO-mediated expansion of late-stage erythroid pro-genitors in the bone marrow initiates recoveryfrom sublethal radiation stress. Blood 2012;120:2501–2511.

57. Dev A, Fang J, Sathyanarayana P, et al. DuringEPO or anemia challenge, erythroid progenitorcells transit through a selectively expandableproerythroblast pool. Blood 2010;116:5334–5346.

58. Chen CC, Jeon SM, Bhaskar PT, et al. FoxOsinhibit mTORC1 and activate Akt by inducingthe expression of Sestrin3 and Rictor. Dev Cell2010;18:592–604.

59. Khatri S, Yepiskoposyan H, Gallo CA, et al.FOXO3a regulates glycolysis via transcriptionalcontrol of tumor suppressor TSC1. J Biol Chem2010;285:15960–15965.

60. Sarbassov DD, Ali SM, Kim DH, et al. Rictor, anovel binding partner of mTOR, defines arapamycin-insensitive and raptor-independentpathway that regulates the cytoskeleton. CurrBiol 2004;14:1296–1302.

61. Guertin DA, Stevens DM, Thoreen CC, et al.Ablation in mice of the mTORC componentsraptor, rictor, or mLST8 reveals that mTORC2is required for signaling to Akt-FOXO andPKCalpha, but not S6K1. Dev Cell 2006;11:859–871.

62. Sarbassov DD, Guertin DA, Ali SM, et al. Phos-phorylation and regulation of Akt/PKB by therictor-mTOR complex. Science 2005;307:1098–1101.

63. Efeyan A, Zoncu R, Sabatini DM. Amino acidsand mTORC1: From lysosomes to disease.Trends Mol Med 2012;18:524–533.

64. Thein SL. Genetic modifiers of beta-thalassemia.Haematologica 2005;90:649–660.

65. Yeo H, Lyssiotis CA, Zhang Y, et al. FoxO3 coor-dinates metabolic pathways to maintain redoxbalance in neural stem cells. EMBO J 2013;32:2589–2602.

66. Shinar E, Rachmilewitz EA. Oxidative denatura-tion of red blood cells in thalassemia. SeminHematol 1990;27:70–82.

67. Scott MD, Eaton JW. Thalassaemic erythrocytes:Cellular suicide arising from iron andglutathione-dependent oxidation reactions? Br JHaematol 1995;91:811–819.

68. Fibach E, Bianchi N, Borgatti M, et al. Effects ofrapamycin on accumulation of alpha-, beta- andgamma-globin mRNAs in erythroid precursorcells from beta-thalassaemia patients. Eur J Hae-matol 2006;77:437–441.

69. Bayeva M, Khechaduri A, Puig S, et al. mTORregulates cellular iron homeostasis through tris-tetraprolin. Cell Metab 2012;16:645–657.

70. Marty C, Lacout C, Droin N, et al. A role forreactive oxygen species in JAK2 V617F myelo-proliferative neoplasm progression. Leukemia2013;27:2187–2195.

71. Duhe RJ, Evans GA, Erwin RA, et al. Nitricoxide and thiol redox regulation of Janus kinaseactivity. Proc Natl Acad Sci USA 1998;95:126–131.

72. Maziere C, Conte MA, Maziere JC. Activationof JAK2 by the oxidative stress generated withoxidized low-density lipoprotein. Free RadicBiol Med 2001;31:1334–1340.

73. Nieborowska-Skorska M, Kopinski PK, Ray R,et al. Rac2-MRC-cIII-generated ROS causegenomic instability in chronic myeloid leukemiastem cells and primitive progenitors. Blood2012;119:4253–4263.

74. Walz C, Crowley BJ, Hudon HE, et al. Acti-vated JAK2 with the V617F point mutationpromotes G1/S-phase transition. J Biol Chem2006;281:18177–18183.



Documents

FOXO3-mTOR metabolic cooperation in the regulation of ... metabolic... · FOXO3-mTOR metabolic cooperation in the regulation of erythroid cell maturation and homeostasis ... of metabolic