AMPKα1 Deletion Shortens Erythrocyte Life Span in Mice

AMP-activated protein kinase (AMPK) is an energy sensor essential for maintaining cellular energy homeostasis. Here, we report that AMPKα1 is the predominant isoform of AMPK in murine erythrocytes and mice globally deficient in AMPKα1 (AMPKα1−/−), but not in those lacking AMPKα2, and the mice had markedly enlarged spleens with dramatically increased proportions of Ter119-positive erythroid cells. Blood tests revealed significantly decreased erythrocyte and hemoglobin levels with increased reticulocyte counts and elevated plasma erythropoietin concentrations in AMPKα1−/− mice. The life span of erythrocytes from AMPKα1−/− mice was less than that in wild-type littermates, and the levels of reactive oxygen species and oxidized proteins were significantly increased in AMPKα1−/− erythrocytes. In keeping with the elevated oxidative stress, treatment of AMPKα1−/− mice with the antioxidant, tempol, resulted in decreased reticulocyte counts and improved erythrocyte survival. Furthermore, the expression of Foxo3 and reactive oxygen species scavenging enzymes was significantly decreased in erythroblasts from AMPKα1−/− mice. Collectively, these results establish an essential role for AMPKα1 in regulating oxidative stress and life span in erythrocytes.

Reactive oxygen species (ROS) 2 are ubiquitously generated in living cells, including erythrocytes, and overwhelming evidence indicates that they play key roles in essential cellular functions. Oxidative stress, defined as a pathological state characterized by increased ROS production or decreased ability to detoxify ROS, plays a causative role in tissue injury in many disease conditions, including cardiovascular diseases, neurological disorders, cancers, and aging (1,2). Mature erythrocytes, with their absence of protein synthesis and high oxygen-carrying capacity, are particularly susceptible to oxidative damage because they are rich in heme iron and oxygen, which can spontaneously generate H 2 O 2 and lipid peroxides (3). Under normal conditions, red blood cells (RBCs) have effective defense mechanisms against oxidative stress. When the production of ROS overwhelms erythrocyte antioxidant systems, or antioxidant systems become defective, excessive ROS cause oxidative damage to the cytoplasmic membrane and associated cytoskeleton in the mature red cell. The consequences of oxidative damage in erythrocytes are often manifested as decreased deformability and splenic sequestration. Uptake of abnormal red cells in the spleen results in a decreased life span of RBCs leading to anemia (4). Thus, disruption of this oxidant-antioxidant equilibrium can markedly shorten the life span of erythrocytes, as reported in mice deficient for peroxiredoxin I (5), peroxiredoxin II (6), SOD1 (7), or SOD2 (8). Defects in these enzymes, which are critical to the oxidative stress response, have also been implicated in human diseases that involve acute or chronic hemolysis (9). The enzymopathy most commonly responsible for hemolytic anemia is deficiency of glucose-6-phosphate dehydrogenase, which converts NADP to NADPH. NADPH is required for maintenance of GSH, a major component of the cellular ROS detoxification system. The absence of these antioxidant protections can result in moderate hemolysis (10); thus, a normal erythrocyte life span depends on an adequate antioxidant response (11).
AMP-activated protein kinase (AMPK) (12) is a serine/threonine kinase, conserved from yeast to humans, that regulates energy homeostasis and metabolic stress. Mammalian AMPK is a heterotrimer consisting of three subunits designated ␣, ␤, and ␥. The ␣ subunit contains a kinase domain and can exist as either an ␣1 or ␣2 isoform. AMPK is activated under conditions that elevate the AMP/ATP ratio, such as glucose deprivation, hypoxia, and muscle contraction (13). Recent evidence also suggests that AMPK may have a much wider range of functions. For instance, it is involved in the regulation of mitochondrial biogenesis (14) and angiogenesis (15). We previously demonstrated that pathologically relevant concentrations of peroxynitrite (ONOO Ϫ ), a potent oxidant formed by the reaction of nitric oxide and superoxide anions (O 2 . ), are capable of activating AMPK independently of changes in the AMP/ATP ratio (16) and, furthermore, that the activation of AMPK can inhibit intracellular O 2 . production in human umbilical vein endothelial cells (17). Overall, AMPK appears to be essential in maintaining redox homeostasis. Whether AMPK controls oxidative homeostasis in erythrocytes and is therefore critical for their survival is not clear. Here, we report that AMPK␣1 depletion triggers oxidative stress and reduces the life span of erythrocytes by lowering the expression of Foxo3-mediated antioxidant enzymes.

EXPERIMENTAL PROCEDURES
Reagents and Animals-Rabbit anti-AMPK␣ antibody was obtained from Cell Signaling Technology (Beverly, MA). Rabbit anti-AMPK␣1 and rabbit anti-AMPK␣2 antibodies were obtained from Bethyl Laboratories, Inc. (Montgomery, TX). Mouse anti-glyceraldehyde-3-phosphate dehydrogenase antibody was from Abcam Inc. (Cambridge, MA). Other chemicals and organic solvents of the highest available grade were obtained from Sigma. AMPK␣1 Ϫ/Ϫ mice and AMPK␣2 Ϫ/Ϫ mice were described elsewhere (18,19). Tempol (4-hydroxy-2,2,6,6-tetramethylpiperidinyloxy) was provided to the animals for the indicated times (see "Results") at a concentration of 1 mM in drinking water. Mice were handled in accordance with study protocols approved by the Institutional Animal Care and Use Committee of the University of Oklahoma Health Science Center.
Histological and Hematological Analyses-10 -12-Week-old male and female mice were weighed and sacrificed. Selected organs, including spleen, liver, kidney, heart, and femur, were removed and weighed to calculate an organ index (organ weight/body weight ϫ 100). Tissues processed for microscopic evaluation were fixed in 10% neutrally buffered formalin, embedded in paraffin, sectioned, mounted on slides, and stained with hematoxylin and eosin or Prussian blue. Bones were decalcified before fixing in formalin. Tissue sections were examined with an Olympus AX70 microscope with UPlanFI 20ϫ/0.50 or 40ϫ/0.85 lens, blood smears with oil immersion UPlanApO 100ϫ/1.30 lens, and an Qimaging Retiga Exi digital CCD camera (QImaging Co.) using QCapture software (version 2.98.0). Peripheral blood samples were harvested in EDTAcoated microtubes (Aktiengesellschaft Co., Germany) by retroorbital sinus bleeding and analyzed with a HEMAVET 950 Veterinary Hematology Analyzer (Drew Scientific Inc.) using the mouse species program. Blood smears were stained with Giemsa or new methylene blue and analyzed microscopically.
Flow Cytometry Analysis-Mice spleens were mechanically dissociated through a 70-m strainer and washed with cold phosphate-buffered saline containing 2% fetal calf serum. Splenocyte single cell suspensions were double-stained with antibodies against fluorescein isothiocyanate-conjugated CD71 (CD71-FITC) and phycoerythrin-conjugated erythroid antigen (Ter119-PE; BD Biosciences). Flow cytometry was performed using a FACSCalibur (BD Biosciences), and FACS data were analyzed using Summit version 4.3 software. Reticulocytes were assayed using a Retic-Count reticulocyte reagent system (BD Biosciences).
Erythrocyte Life Span Determination-In vivo biotinylation followed by FACS analysis were performed according to Suzuki and Dale (21) with modifications (22). Briefly, RBCs were labeled in vivo by intravenous injection of 30 mg/kg N-succinimidyl-6-[biotinamido] hexanoate (Thermo Fisher Scientific). Small samples (5 l) of peripheral blood were collected from tail vein 1 h after biotin labeling and stained using streptavidinphycoerythrin. Staining was followed by FACS analysis to ensure that at least 95% of red cells were labeled with biotin. Blood samples were analyzed at 7-day intervals to quantify biotin-labeled cells remaining in the circulation.
Erythropoietin (EPO) Enzyme-linked Immunosorbent Assay-EPO level in blood plasma was determined using the Quantikine mouse EPO enzyme-linked immunosorbent assay kit from R & D Systems, Inc. (Minneapolis, MN) according to the manufacturer's protocol.
Erythrocyte Reinfusion and Clearance Test-In vivo RBC tracking was performed as reported previously (23,24) with modifications. Briefly, RBCs of donor mice were biotin-labeled in vivo by tail vein injection of N-succinimidyl-6-[biotinamido] hexanoate, as described above. One hour after biotin infusion, blood was collected from donor mice and placed in tubes containing heparin as an anticoagulant. A small aliquot of labeled RBCs was incubated with streptavidin-phycoerythrin and analyzed by flow cytometry to ensure that at least 95% of the blood cells were labeled. Blood was washed and resuspended in sterile saline, and 100 l of RBC suspension was infused into each recipient mouse by tail vein injection. Initial postinfusion blood samples were obtained after 30 min and analyzed by flow cytometry. The typical initial percentage of biotin-labeled RBCs (recorded as day 0%) was greater than 5% in recipient mice. Seven days after blood reinfusion, the remaining labeled red cells were determined (recorded as day 7%). The red cell clearance rate was calculated as follows: clearance rate ϭ ((day 7% Ϫ day 0%)/day 0%) ϫ 100%.
Protein Oxidation Detection-The Oxyblot TM protein oxidation detection kit (Millipore, MA), based on a previously described method (25), was used as described by the manufacturer. Briefly, RBC lysates from wild-type mice and AMPK␣1 Ϫ/Ϫ mice were derivatized with dinitrophenylhydrazine, and derivatized samples (20 g of protein/sample) were separated by SDS-PAGE on 12% gels. Proteins were transferred to nitrocellulose membranes, and membranes were blocked and immunoblotted using OxyBlot kit methods and reagents. Bands were visualized with chemiluminescence (captured on film) and analyzed densitometrically. Membranes stained with Red Ponceau S (Sigma) were used as loading controls.
Intracellular ROS Measurement-Red cell intracellular ROS measurement was performed essentially as reported previously (8). RBCs were washed and resuspended in staining buffer (BD Biosciences) and loaded with 5 M 5-(and -6)-chloromethyl-2Ј,7Ј-dichlorodihydrofluorescein diacetate, acetyl ester (Invitrogen), in the dark for 30 min at 37°C in a 5% CO 2 atmosphere. Intracellular fluorescent products were measured immediately by flow cytometry.
Osmotic Fragility Assay-The osmotic fragility of erythrocytes in freshly collected blood from AMPK␣1 Ϫ/Ϫ and wildtype mice was measured with modifications (26). Briefly, RBCs were harvested into heparin-coated tubes and suspended in varying concentrations of NaCl, and samples were incubated at room temperature for 10 min and then centrifuged at 1500 ϫ g for 10 min to sediment unlysed cells and stroma. 200 l of supernatant was collected, and the hemoglobin concentration of each sample was measured by Synergy HT multidetection microplate reader (BioTek Instrument, Inc) at 540 nm, with appropriate controls. The lyses percentage of RBCs was calculated from the absorbance, and a fragility curve was generated by plotting varying salt concentrations versus hemolysis.
Fluorescence-activated Cell Sorting (FACS) and Quantitative Reverse Transcription (RT)-PCR-Bone marrow cell suspensions were stained with antibodies against CD71-FITC and Ter119-PE (BD Biosciences) before sorting on an Influx Cell Sorter. Total RNA in sorted cells was extracted using an RNeasy mini kit (Qiagen), and cDNA was synthesized from total RNA using an iScript cDNA synthesis kit (Bio-Rad), as described by the respective manufacturers. Real time quantitative RT-PCR was performed using iQ SYBR Green SuperMix and the MyiQ single color real time PCR detection system (Bio-Rad). Primer sequences are presented in the supplemental material. Quantitative RT-PCR results were analyzed according to the 2 Ϫ⌬⌬CT method (27) using expression of the ␤-actin gene as a housekeeping control.
Statistical Analysis-Two-tailed Student's t test was conducted, or two-way analysis of variance followed by Bonferroni post hoc analyses was used to determine statistical differences between various experimental and control groups. p value Ͻ 0.05 was considered statistically significant.

AMPK␣1 Is the Predominant Isoform in Murine Erythrocytes-We
found that AMPK␣1 is the predominant isoform of AMPK␣ in murine erythrocytes, as demonstrated by Western blots. Using an antibody specific for AMPK␣1, we confirmed the expression of AMPK␣1 in mice RBCs (Fig. 1A). Because no recognizable bands were seen on the blot with the same loading with an antibody specific for AMPK␣2, we proceeded to use an antibody that recognizes both AMPK␣1 and AMPK␣2 isoforms. As shown in Fig. 1A, AMPK␣ was largely absent in erythrocytes from AMPK␣1 Ϫ/Ϫ mice, whereas there was no significant difference of the expression of total AMPK␣ between AMPK␣2 Ϫ/Ϫ and wild-type erythrocytes.
Genetic Deletion of AMPK␣1, but Not AMPK␣2, in Mice Causes Splenomegaly and Moderate Anemia-Next, we determined if genetic deletion of AMPK␣1 or AMPK␣2 affected the pathology of organs, including hearts, kidneys, and spleens. AMPK␣1 Ϫ/Ϫ mice exhibited splenomegaly (Fig. 1, B and C), whereas body weights and the weights of other major organs were not significantly changed (supplemental Fig. 1); this phenotype was not observed in AMPK␣2 Ϫ/Ϫ mice. Hematological tests showed moderate anemia in AMPK␣1 Ϫ/Ϫ mice, as evidenced by low erythrocyte number, low hemoglobin content, and increased red cell distribution width compared with wild-type mice. Leukocyte and platelet counts were not significantly changed in AMPK␣1 Ϫ/Ϫ mice compared with wild-type littermates (Table 1). In AMPK␣2 Ϫ/Ϫ mice, there were no significant differences in erythrocyte or platelet parameters compared with wild-type mice (supplemental Table 1). Accordingly, subsequent experiments focused on AMPK␣1 Ϫ/Ϫ mice. Hematoxylin and eosin-stained sections of spleens from AMPK␣1 Ϫ/Ϫ mice revealed expanded red pulps and an increase in erythroid precursor cells (Fig. 1D), indicating a compensatory reaction to anemia. Prussian blue staining showed elevated iron deposits in spleen sections from AMPK␣1 Ϫ/Ϫ mice (Fig. 1E), indicating increased destruction of erythrocytes in the spleen, and Giemsa staining of blood smears indicated polychromasia, combined macrocytosis and microcytosis, and reticulocytosis in AMPK␣1 Ϫ/Ϫ mice (Fig. 1F). Collectively, these data indicate that a deficiency of AMPK␣1 caused moderate anemia, characterized by a significant decrease in red cell count, hematocrit, and hemoglobin in peripheral blood relative to wildtype littermate controls. The phenotypic differences between AMPK␣1 Ϫ/Ϫ mice and AMPK␣2 Ϫ/Ϫ mice likely stem from the distinctively different expression levels of these two isoforms in erythrocytes.
Erythropoiesis Is Increased in AMPK␣1 Ϫ/Ϫ Mice-The anemia observed in AMPK␣1 Ϫ/Ϫ mice could be due to defective red cell maturation (ineffective erythropoiesis), increased red cell destruction (hemolysis), or a combination of both processes. We first analyzed splenocytes by flow cytometry. Erythroid cells (Ter119 ϩ ) in the spleens of AMPK␣1 Ϫ/Ϫ mice were increased compared with those of wild-type mice (supplemental Fig. 2A), whereas the relative abundance of other cell types in the spleen was not increased (supplemental Fig. 2, B-E). A subpopulation of Ter119 high cells was distinguished based on their expression of the transferrin receptor (CD71), which decreases with erythroblast maturation. Combining Ter119 and CD71 expression distinguishes four subpopulations of erythroid cells, Ter119 med CD71 high , Ter119 high CD71 high , Ter119 high CD71 med , and Ter119 high -CD71 low , corresponding to increasingly mature erythroblasts (28). Using this approach, we found a dramatic expansion of the early erythroblast population (Ter119 high CD71 high ) and increased proerythroblast population (Ter119 medium CD71 high ) in AMPK␣1 Ϫ/Ϫ mice ( Fig. 2A). Further analysis of the Ter119 high erythroblasts was arbitrarily based on the forward scatter (FSC) parameter and CD71 expression level. Ter119 high erythroblasts can be subdivided into three groups as follows: Ery.A (Ter119 high CD71 high FSC high ) are basophilic; Ery.B (Ter119 high CD71 high FSC low ) are late basophilic and polychromatic; and Ery.C (Ter119 high CD71 low FSC low ) are orthochromatic erythroblasts and reticulocytes. Increasingly mature erythroblasts, according to the reports (29,30), indicated that the major elevated population was late basophilic and polychromatic cells (supplemental Fig. 4).
Next, using new methylene blue-stained blood smears (data not shown) and flow cytometry with thiazole orange staining (Fig. 3A), we determined the number of reticulocytes and found that the reticulocyte count in AMPK␣1 Ϫ/Ϫ mice was 3-4-fold higher than that in wild-type mice (Fig. 3B). The increased reticulocyte levels were correlated with elevated plasma levels of EPO (Fig. 3C). These data indicate that erythropoiesis was elevated in AMPK␣1 Ϫ/Ϫ mice, possibly reflecting a compensatory reaction to the moderate anemia observed in these mice.
Erythrocyte Life Span Is Shortened in AMPK␣1 Ϫ/Ϫ Mice-The normal life span of circulating RBCs, which is determined by their clearance from the peripheral circulation (predominantly by the spleen), has been reported to be ϳ120 days and 40 days in humans and mice, respectively (31). To determine more directly whether the observed anemia was due to decreased production of mature RBCs or increased destruction of RBC in the circulation, we measured erythrocyte life span using direct in vivo biotin labeling. To this end, wild-type and AMPK␣1 Ϫ/Ϫ mice were injected with N-succinimidyl-6-[biotinamido] hexanoate to label RBCs, and cell life span was determined by flow cytometric analysis of circulating, biotinylated RBCs. The time required for 50% of labeled erythrocytes to be lost from wildtype mice (half-life) was about 24 days, consistent with previous reports (5,22). However, the half-life of labeled erythrocytes decreased to about 14 days in AMPK␣1 Ϫ/Ϫ mice (Fig. 4A).
To further demonstrate the destruction of erythrocytes in AMPK␣1 Ϫ/Ϫ mice, we performed a blood cross-transfusion experiment. Wild-type mice received either biotin-labeled wild-type erythrocytes (WT-WT) or biotin-labeled AMPK␣1 Ϫ/Ϫ erythrocytes (KO-WT), and AMPK␣1 Ϫ/Ϫ mice received either biotin-labeled wild-type erythrocytes (WT-KO) or biotin-labeled AMPK␣1 Ϫ/Ϫ erythrocytes (KO-KO). The result demonstrated that the KO-WT group had a higher clearance rate of infused erythrocytes than did the WT-WT group, and similar results were observed when compared KO-KO with WT-KO group (Fig. 4B), suggesting accelerated clearance of erythrocytes in AMPK␣1 Ϫ/Ϫ mice. Consistent with these results, Prussian blue-stained spleen sections from AMPK␣1 Ϫ/Ϫ mice showed a marked increase in iron deposits.
Osmotic Fragility Is Decreased in Erythrocytes from AMPK␣1 Ϫ/Ϫ Mice-To uncover the possible cause of why erythrocytes from AMPK␣1 Ϫ/Ϫ mice had an increased clearance rate, we first determined the osmotic fragility of erythrocytes. As depicted in Fig. 5A, AMPK␣1 Ϫ/Ϫ -deficient erythrocytes were significantly more resistant to hypotonic lysis than were wild-type red cells (Fig. 5A), suggesting that they had a more rigid and less deformable cell membrane.

TABLE 1 Hematological parameters, analysis of mouse peripheral blood with a HEMAVET 950 veterinary hematology analyzer
The abbreviations used are as follows: HCT, hematocrit; MCV, mean corpuscle volume; MCH, mean corpuscle hemoglobin; MCHC, mean corpuscle hemoglobin concentration; RDW, red cell distribution width; WBC, white blood cell count; NE, neutrophil count; LY, lymphocyte count; MO, monocyte count; EO, eosinophil; BA, basophil count; PLT, platelet; MP, mean platelet volume; k, thousands; M/l, million cells per l of blood; k/l, thousand cells per l of blood. All data are expressed as means Ϯ S.D. (n ϭ 5 in each group).
Intracellular ROS Levels and Protein Oxidation Are Increased in AMPK␣1 Ϫ/Ϫ Erythrocytes-Next, we determined if increased levels of protein aggregates were due to increased oxidative stress in erythrocytes. Protein oxidation can be measured indirectly by Western blotting for carbonyls that result from the reaction of side chains of lysine, proline, threonine, or arginine with ROS (25,33). Intracellular ROS concentrations measured under base-line conditions and after challenging with exogenous H 2 O 2 (50 M) showed that ROS levels were elevated in erythrocytes from AMPK␣1 Ϫ/Ϫ mice compared with wild-type mice under both conditions (Fig. 6A). In agreement with the observed elevation in ROS concentrations, oxidized protein levels in erythrocytes from AMPK␣1 Ϫ/Ϫ mice were significantly increased (ϳ7fold) as measured by 2,4-dinitrophenylhydrazine-derivatized carbonyl immunochemical staining (Fig. 6, B and C).

Chronic Administration of Tempol Prolongs the Life Span of Erythrocytes in AMPK␣1
Ϫ/Ϫ Mice-To further evaluate the potential effect of ROS on shortened red cell survival, we fed tempol, a potent antioxidant, to wild-type and AMPK␣1 Ϫ/Ϫ mice. Tempol treatment had no effect on reticulocyte index or erythrocyte life span in wild-type mice. However, it significantly increased the life span of erythrocytes in AMPK␣1 Ϫ/Ϫ mice but decreased the reticulocyte index (Fig. 7, A and B). 4 months after tempol treatment, the spleen index was significantly decreased in treatment groups compared with no treatment AMPK␣1 Ϫ/Ϫ controls (Fig. 7C).

Expression of the Foxo3 Transcription Factor and Antioxidant Genes Is Decreased in Erythroblast Cells from AMPK␣1
Ϫ/Ϫ Mice-Finally, we investigated the molecular basis of increased oxidative stress in the erythrocytes from AMPK␣1 Ϫ/Ϫ mice by measuring the mRNA levels of several antioxidant genes known to play a key role in erythrocytes. Total RNA was extracted from sorted CD71 high Ter119 high erythroblast cells from wild-type and AMPK␣1 Ϫ/Ϫ mice and analyzed by quantitative real time RT-PCR. As shown in Fig.  7D, the expression of the Foxo3 transcription factor was significantly reduced in the erythroblast cells from AMPK␣1 Ϫ/Ϫ mice compared with that in wild-type mice; the mRNA levels of Sod2, catalase, and glutathione peroxidase (GPx-1) were also reduced.

DISCUSSION
In this study, we provide the first evidence to suggest that AMPK␣1, but not AMPK␣2, maintains the oxidant-antioxidant balance in erythrocytes by controlling Foxo3 transcription factor-mediated antioxidant enzyme expression. We further show that genetic deletion of AMPK␣1 causes oxidative stress, resulting in a shortened erythrocyte life span, RBC hemolysis, and anemia in vivo.
The most important findings of this study are that erythrocytes predominantly express AMPK␣1 and genetic deletion AMPK␣1 leads to anemia in vivo. The anemia could be caused by defective red cell maturation (ineffective erythropoiesis), increased red cell destruction (hemolysis), or a combination of both processes. Our data clearly indicate that the anemia observed in AMPK␣1 Ϫ/Ϫ mice is likely the result of increased erythrocyte removal by the spleen. This conclusion is supported by several lines of evidence. First, the rate at which labeled erythrocytes were eliminated from the circulation was markedly higher in AMPK␣1 Ϫ/Ϫ mice than in wildtype littermates (Fig. 4A). Consistent with this result, the Prus-  sian blue-stained sections of spleen showed marked increases in iron deposits in AMPK␣1 Ϫ/Ϫ mice. Second, reticulocytes were increased in AMPK␣1 Ϫ/Ϫ mice, and the increased reticulocyte levels were correlated with elevated levels of plasma EPO. In addition, Ter119 high CD71 high erythroblasts in the spleen were also dramatically expanded ( Fig. 2A), and a histological examination of bone sections of AMPK␣1 Ϫ/Ϫ showed no significant deficiency (supplemental Fig. 3). We noticed that there was a decrease in the late erythroblast population (Ter119 high CD71 low cells) in spleens from AMPK␣1 Ϫ/Ϫ mice compared with wild-type controls, whereas the early erythroblast population (Ter119 medium CD71 high and Ter119 high CD71 high ) was increased (Fig. 2, A and B). This result suggested that the maturation of the Ter119 high CD71 high stage to the Ter119 high CD71 low stage is the limiting step for anemia compensation in AMPK␣1 Ϫ/Ϫ mice and that the enhanced loss of abnormally fragile RBCs from the body exceeds the rate of recruitment of the late eryth-  roblast population in spleen. Furthermore, the increased reticulocyte count accompanied by anemia is highly suggestive of peripheral hemolysis in these animals. In line with our observations, Foller et al. (34) have reported genetic deletion in AMPK␣1 leads to anemia. Importantly, our results suggest that elevated intracellular ROS levels play important roles for the shortened life span of red cells, which was supported by our FACS data, protein oxidation immunoblot results, and PHZ treatment results; more importantly, our results showed that in vivo treatment of the mice with the antioxidant can significantly, even if not completely, reverse the phenotype. Taken together, these results suggest that the shortened erythrocyte life span in AMPK␣1 Ϫ/Ϫ mice is due to elevated destruction of circulating erythrocytes within the spleen due to factors intrinsic to the erythrocytes.
The specialized structure of the venous system of the red pulp gives the spleen a unique capacity to remove abnormal erythrocytes. RBCs must cross the narrow interendothelial slits in the walls of the venous sinuses in order to reenter the venous system, a process that requires erythrocytes to undergo remarkable deformation (35,36). Alterations in membrane characteristics can affect cell survival. Perhaps not surprisingly, abnormal structure and deformability of the erythrocyte membrane are observed in many clinical red cell disorders (37), and reduced erythrocyte deformability plays an important role in shortened erythrocyte survival in many types of hemolytic anemia (38). Consistent with this interpretation, we found that the erythrocytes from AMPK␣1 Ϫ/Ϫ mice had markedly decreased osmotic fragility, an indication of increased rigidity of erythrocytes.
The clearance of abnormal RBC occurs primarily in spleen by two methods. First, the rigid red cells would be trapped by specific venous sinus structures of red pulps. Second, the macrophage in the spleen would recognize the specific surface markers such as phosphatidylserine exposure or CD47 on the abnormal or foreign infused RBCs (39,40). Unlike the data shown in Fig. 4A, the KO-KO combination did not have higher clearance rate than that in WT-WT. This apparent conflict between A and B in Fig. 4 can be reconciled by abnormal macrophage function in spleen to recognize infused RBCs. Indeed, there is a significant decrease in uptake of fluorescence-labeled low density lipoproteins by isolated macrophages from AMPK␣1 knock-out mice compared with wild type controls. 3 Due to low metabolic rate in red cells, these cells are relatively not sensitive to nutrient depletion but are very sensitive to oxidative stress because of the high physiological hemoglobin level (9,41). The deformability of erythrocytes is also affected by oxidative stress (42)(43)(44). Excessive ROS in erythrocytes causes damage to the cytoplasmic membrane and associated cytoskeleton in the mature red cell, effects that manifest as decreased deformability of RBCs and splenic sequestration. In this study, 3 M. Zhang, unpublished data. A, reticulocyte index was measured after 5 weeks in all groups. NS, not significant. B, in vivo tempol therapy improves the life span of erythrocytes in AMPK␣1 Ϫ/Ϫ mice. Wild-type and AMPK␣1 Ϫ/Ϫ mice were provided normal drinking water or water containing tempol (1 mM) for 5 weeks, and the life span of in vivo biotinylated erythrocytes was measured. Tempol was administered continuously during the experimental period. C, selected antioxidant genes were monitored in wild-type and AMPK␣1 Ϫ/Ϫ erythroblast cells by quantitative real time RT-PCR. Gene expression was normalized to ␤-actin and is expressed as a percentage of the control values. Data were expressed as means Ϯ S.D. (*, p Ͻ 0.05; **, p Ͻ 0.01, n ϭ 4 in each group). D, spleen indexes after the mice were provided tempol (1 mM) in drinking water for 4 months, (**, p Ͻ 0.01, n ϭ 6 in each group). E, proposed mechanism illustrating that AMPK␣1 is essential for normal erythrocyte life span.
we found increased oxidative damage in erythrocytes from AMPK␣1 Ϫ/Ϫ mice, as evidenced by increased ROS levels, increased protein oxidation, decreased RBC survival, and decreased osmotic fragility. This phenotype is similar to that observed in mice deficient for the antioxidant enzymes, peroxiredoxin I (5), peroxiredoxin II (6), Sod1 (7), or Sod2 (8). That the antioxidant defenses in AMPK␣1 Ϫ/Ϫ mice are defective is also supported by PHZ tolerance experiments, which showed that AMPK␣1 Ϫ/Ϫ mice were hypersensitive to PHZ-induced oxidative stress compared with wild-type littermates. Most importantly, we found that chronic administration of the antioxidant, tempol, markedly prolonged the life span of erythrocytes and ameliorated anemia in AMPK␣1 Ϫ/Ϫ mice. Foller et al. (34) have reported that genetic deletion in AMPK␣1 leads to anemia.
The specialized structure of the venous system of the red pulp gives the spleen a unique capacity to remove abnormal erythrocytes. RBCs must cross the narrow interendothelial slits in walls of the venous sinuses in order to reenter the venous system, a process that requires erythrocytes to undergo remarkable deformation (35,36). Alterations in membrane characteristics can affect cell survival. Perhaps not surprisingly, abnormal structure and deformability of the erythrocyte membrane are observed in many clinical red cell disorders (37), and reduced erythrocyte deformability plays an important role in shortened erythrocyte survival in many types of hemolytic anemia (38). Consistent with this interpretation, we found that the erythrocytes from AMPK␣1 Ϫ/Ϫ mice had markedly decreased osmotic fragility, an indication of increased rigidity of erythrocytes.
The deformability of erythrocytes is also affected by oxidative stress (42)(43)(44). Excessive ROS in erythrocytes causes damage to the cytoplasmic membrane and associated cytoskeleton in the mature red cell, effects that manifest as decreased deformability of RBCs and splenic sequestration. In this study, we found increased oxidative damage in erythrocytes from AMPK␣1 Ϫ/Ϫ mice, as evidenced by increased ROS levels, increased protein oxidation, decreased RBC survival, and decreased osmotic fragility. This phenotype is akin to that observed in mice deficient for the antioxidant enzymes, peroxiredoxin I (5), peroxiredoxin II (6), Sod1 (7), or Sod2 (8). That the antioxidant defenses in AMPK␣1 Ϫ/Ϫ mice are defective is also supported by PHZ tolerance experiments, which showed that AMPK␣1 Ϫ/Ϫ mice were hypersensitive to PHZ-induced oxidative stress compared with wild-type littermates. Most importantly, we found that chronic administration of the antioxidant, tempol, markedly prolonged the life span of erythrocytes and ameliorated anemia in AMPK␣1 Ϫ/Ϫ mice. In supporting this concept, our published studies have demonstrated that AMPK␣2 deletion causes excessive oxidative stress and atherosclerosis in vivo (17,45). Taken together, our results strongly imply that AMPK␣1 deletion increases the removal of erythrocytes in vivo, likely via increased levels of intracellular ROS.
Another important finding of this study is that Foxo3 might be an important target of AMPK in erythrocytes, and the AMPK-Foxo3 axis is essential in maintaining redox homeostasis. An earlier study from Tothova and Gilliland (46) reported a significant elevation of ROS in the Foxo3-deletion hematopoietic stem cell population, as compared with wild-type controls, which is correlated with decreased expression of anti-oxidant genes, including catalase, MnSOD, GADD45, etc. In the study carried out by Marinkovic et al. (47), they found that animals deficient in Foxo3 have a shortened life span of erythrocytes together with decreased expressions of ROS-scavenging enzymes, including catalase, SOD1, SOD2, and glutathione peroxidase 1 in erythroblasts. Consistently, Foxo3-deficient mice died rapidly when exposed to erythroid oxidative stressinduced conditions, and antioxidant treatment can significantly improve the life span of Foxo3 Ϫ/Ϫ erythrocytes and corrected the reticulocyte index in these mice. Our observations are in line with earlier reports that Foxo3 is a downstream target of AMPK (48) and that AMPK activity, mediated at least partly by Foxo3 transcription factors, is required for the extension of life span in Caenorhabditis elegans by dietary restrictions (49). During the preparation of this paper, Colombo et al. (50) reported that AMPK␣1 regulates the status of antioxidants in human umbilical vein endothelial cells and that Foxo3 is involved in this process. How AMPK␣1 regulates Foxo3 transcription factors remains unknown and warrants further investigation.
In summary, our results demonstrate that AMPK␣1 is important in regulating the expression of genes involved in antioxidative defense in red blood cells; genetic deletion of AMPK␣1 resulted in oxidant-antioxidant imbalance in RBCs, and increased intracellular ROS levels caused elevated protein oxidation and increased osmotic fragility. The anemia phenotypes observed in AMPK␣1 Ϫ/Ϫ mice due to the recruitment of RBCs cannot compensate for the elimination of those fragile erythrocytes by the spleen (Fig. 7E). Our research suggested that AMPK activation might be a therapeutic target for treating erythrocyte disorders.