Essential role for Nix in autophagic maturation of erythroid cells

Abstract

Erythroid cells undergo enucleation and the removal of organelles during terminal differentiation^1–3. Although autophagy has been suggested to mediate the elimination of organelles for erythroid maturation^2–6, the molecular mechanisms underlying this process remain undefined. Here we report a role for a Bcl-2 family member, Nix (also called Bnip3L)^7–9, in the regulation of erythroid maturation through mitochondrial autophagy. Nix^−/− mice developed anaemia with reduced mature erythrocytes and compensatory expansion of erythroid precursors. Erythrocytes in the peripheral blood of Nix^−/− mice exhibited mitochondrial retention and reduced lifespan in vivo. Although the clearance of ribosomes proceeded normally in the absence of Nix, the entry of mitochondria into autophagosomes for clearance was defective. Deficiency in Nix inhibited the loss of mitochondrial membrane potential (ΔΨ_m), and treatment with uncoupling chemicals or a BH3 mimetic induced the loss of ΔΨ_m and restored the sequestration of mitochondria into autophagosomes in Nix^−/− erythroid cells. These results suggest that Nix-dependent loss of ΔΨ_m is important for targeting the mitochondria into autophagosomes for clearance during erythroid maturation, and interference with this function impairs erythroid maturation and results in anaemia. Our study may also provide insights into molecular mechanisms underlying mitochondrial quality control involving mitochondrial autophagy.

Nix, a BH3-only member of the Bcl-2 family, is upregulated in erythroid cells undergoing terminal differentiation¹⁰. To determine the potential function for Nix in erythroid maturation, we generated Nix^−/− mice using embryonic stem (ES) cells with a gene trap insertion between exons 3 and 4 of Nix (Supplementary Fig. 2). We first examined red blood cells in the peripheral blood (RBCs), including reticulocytes and erythrocytes, in Nix^−/− mice. Although RBC counts were decreased (Supplementary Table 1), polychromasia and increased reticulocytes were observed in Nix^−/− mice (Fig. 1a and Supplementary Fig. 3a). We also examined RBCs for the expression of an erythroid cell marker, glycophorin-A-associated Ter119, and for transferrin receptor CD71, which is downregulated during terminal erythroid differentiation ^11,12. Although Ter119^lowCD71^high and Ter119⁺ CD71^high early erythroblasts¹³ were absent in the peripheral blood, a significant increase in Ter119⁺ CD71⁺ reticulocytes was observed in Nix^−/− mice (Fig. 1b). Electron microscopy also showed more irregularly shaped cells characteristic of reticulocytes in the peripheral blood of Nix^−/− mice (Supplementary Fig. 3c). In contrast, CD71⁻ mature erythrocytes were reduced in Nix^−/− mice (Fig. 1b). Together, these data suggest a potential defect in the development from CD71⁺ reticulocytes into mature CD71⁻ erythrocytes in Nix^−/− mice.

Figure 1. Reticulocytosis and retention of mitochondria in Nix^−/− RBCs.

a, Polychromasia (Wright–Giemsa stain, indicated by arrows) and increased reticulocytes (brilliant cresyl blue stain) revealed by staining blood smears from 6-week-old wild-type (WT) or Nix^−/− mice (scale bar, 20 µm). b, c, Ter119 (b) or Mitotracker deep red (c) versus CD71 staining of RBCs from 3- or 6-week-old mice. Cell counts were also calculated (mean ± s.e.m.; n = 3). *P <0.05; **P <0.01; ***P<0.001. d, e, Analyses of RBCs as in b by immunocytochemistry (d; scale bar, 10 µm) or transmission electron microscopy (e; scale bar, 0.5 µm).

Terminal differentiation of reticulocytes into erythrocytes involves the coordinated removal of organelles¹⁴. Because Nix localizes to the mitochondria¹⁰, we examined whether Nix^−/− reticulocytes might be defective in eliminating the mitochondria. Consistent with increased reticulocytes, more RBCs (28% at 3 weeks and 12% at 6 weeks of age) were positive for CD71 and mitochondrial staining in Nix^−/− mice than in wild-type controls (Fig. 1c). Notably, a significant portion of CD71⁻ RBCs in Nix^−/− mice retained mitochondria (33% and 24% of total RBCs in 3-week-old and 6-week-old mice, respectively), whereas mitochondria were virtually absent in wild-type CD71⁻ RBCs (Fig. 1c). Immunocytochemistry staining for COX IV, a catalytic subunit of cytochrome c oxidase localized to the inner mitochondrial membrane, also confirmed the presence of mitochondria in both CD71⁺ and CD71⁻ RBCs in Nix^−/− mice (Fig. 1d). Electron microscopy revealed that very few RBCs in wild-type mice contained mitochondria or detectable autophagosomes (Fig. 1e). Autophagosomal structures were abundant in Nix^−/− RBCs (Fig. 1e). However, all Nix^−/− RBCs containing autophagosomes also displayed multiple mitochondria outside of autophagosomes, while approximately 20% of autophagosomes contained one mitochondrion inside (Fig. 1e). This suggests a potential defect in the inclusion of mitochondria by autophagosomes for clearance in Nix^−/− RBCs.

The loading and unloading of oxygen by haemoglobin can induce oxidant stress in RBCs¹⁵. The mitochondrion is a major site for the production of reactive oxygen species (ROS) and can function as an apoptotic machinery¹⁶. The removal of mitochondria is therefore potentially beneficial for the survival of RBCs. We therefore examined whether Nix deficiency could affect RBC survival in vivo. Wild-type and Nix^−/− mice were injected with N-hydroxysuccinimido (NHS)-biotin. Cell clearance was determined by the loss of biotinylated RBCs¹⁷. We observed that biotinylated RBCs disappeared more rapidly in Nix^−/− mice than in wild-type controls (Fig. 2a, top panel), suggesting a faster clearance of RBCs in Nix^−/− mice. We also labelled RBCs with 5-chloromethylfluorescein diacetate (CMFDA)¹⁸. We then measured the loss of CMFDA-labelled wild-type or Nix^−/− RBCs after transfer into wild-type recipient mice. We observed accelerated clearance of transferred Nix^−/− RBCs compared to wild-type controls (Fig. 2a, bottom panel). These data suggest that Nix^−/− RBCs have a shorter lifespan in vivo.

Figure 2. Decreased survival of RBCs in Nix^−/− mice.

a, Quantification of NHS-biotin-labelled RBCs or transferred CMFDA-labelled RBCs (n = 3). b, Western blot analyses of caspases in RBCs after in vitro culture. Asterisks denote processed caspases. FL, full length. c, The relative luminescence units (RLU) of caspase activities in cultured RBCs and the suppression of annexin V staining in Nix^−/− RBCs after 12-h culture in the presence of qVD-oph or solvent control (DMSO) (n = 3). d, Mitotracker versus annexin V staining of RBCs after in vitro culture. e, Mean fluorescent intensity (MFI) of ROS staining in RBCs after in vitro culture, and caspase activities in Nix^−/− RBCs after 24 h culture with solvent control, ROS scavengers or qVD-oph (n = 3). BHT, butylated hydroxytoluene; CAT, catalase; TEM, tempol.

f, Haematoxylin and eosin (H&E) staining of spleen sections of 9-week-old wild-type and Nix^−/− mice. Arrows denote iron deposits within macrophage cytoplasm. Iron deposits were also stained with Prussian blue, followed by counterstain with nuclear-fast red. Scale bar, 20 µm. g, Real-time RT–PCR of erythropoietin (Epo) (n = 12). For all relevant panels, statistical significance of the data (mean ± s.e.m.) is: *P<0.05; **P<0.01; ***P<0.001.

Consistent with decreased survival of Nix^−/− RBCs in vivo, spontaneous caspase activation was observed in Nix^−/− but not wild-type RBCs after in vitro culture (Fig. 2b, c). Moreover, mitochondrion-containing RBCs from Nix^−/− mice displayed phosphatidylserine, a signal for phagocytosis¹⁹, before and after in vitro culture, as revealed by annexin V staining (Fig. 2d). A pan-caspase inhibitor, quinolyl-valyl-O-methylaspartyl-[2,6-difluorophenoxy]-methyl ketone (qVD-oph)²⁰, blocked caspase activation and suppressed annexin V staining on Nix^−/− RBCs (Fig. 2c, e), indicating that caspase activation contributed to phosphatidylserine display. Increased ROS was detected in Nix^−/− RBCs, and ROS scavengers inhibited caspase activation in Nix^−/− RBCs (Fig. 2e). In agreement with accelerated clearance of RBCs, histochemistry showed increased iron-laden macrophages in the spleens of Nix^−/− mice (Fig. 2f). Decreases in mature erythrocytes have been shown to induce compensatory erythropoiesis¹³. Consistently, reduced mature RBCs were correlated with increased erythropoietin mRNA and an expansion of Ter119⁺CD71⁺ erythroblasts in Nix^−/− mice (Fig. 2g and Supplementary Fig. 8). The increased erythropoietin may contribute to the expansion of Ter119⁺ CD71⁺ erythroblasts, thus ameliorating the anaemia in Nix^−/− mice. Together, our data suggest that mitochondrial retention in Nix^−/− RBCs results in increased caspase activation and phosphatidylserine display, leading to faster clearance of RBCs and compensatory erythropoiesis. Removing mitochondria by autophagy would limit mitochondrial disruption in the cytosol and prevent the induction of apoptosis signalling in RBCs.

To induce the generation of sufficient numbers of reticulocytes for analyses, we treated mice with phenylhydrazine (PHZ), which induces reticulocytosis²¹. Ter119⁺ CD71⁺ reticulocytes in the peripheral blood were then sorted either on day 3 or on day 6 after PHZ treatment. Notably, a significant portion of the freshly sorted wild-type reticulocytes (18% at day 3 and 37% at day 6 after PHZ treatment) had already lost their mitochondria (Fig. 3a, top row). In contrast, fewer Nix^−/− reticulocytes were negative for mitochondrial staining (Fig. 3a). This implies that Nix^−/− reticulocytes were undergoing less efficient elimination of the mitochondria in vivo. After culture for in vitro maturation³, Nix^−/− reticulocytes showed significant delays in the removal of the mitochondria compared to wild-type controls (Fig. 3a and Supplementary Fig. 4a). This suggests that Nix^−/− reticulocytes are defective in clearing mitochondria during maturation.

Figure 3. Defective clearance of mitochondria by autophagy in Nix^−/− reticulocytes.

a, Ter119⁺ CD71⁺ reticulocytes sorted at day 3 or 6 after PHZ treatment were cultured for in vitro maturation for 0 (top row), 2 (middle row) or 4 (bottom row) days, followed by Mitotracker deep red and Ter119 staining. b, CD71⁺ Ter119⁺ reticulocytes sorted at day 3 after PHZ treatment were cultured and stained with new methylene blue. Scale bar, 20 µm. c, Cells as in b were stained for LC3 and COX IV and analysed by deconvolution microscopy. Scale bar, 5 µm. The Pearson coefficiency for LC3 and COX IV co-localization at day 0 (mean ± s.e.m.) is: wild type, 0.81 ± 0.001; Nix^−/−, 0.44 ± 0.036 (n = 35, P = 0.0006). d, Cells sorted as in b were analysed by electron microscopy. Scale bar, 0.5 µm.

Inhibition of autophagy with 3-methyladenine, wortmannin or chloroquine^22,23 suppressed the removal of mitochondria but not ribosomes in reticulocytes (Supplementary Fig. 5), indicating that the clearance of mitochondria, but not ribosomes, is dependent on autophagy in RBCs. Abnormal autophagosome formation in Nix^−/− reticulocytes might lead to defective mitochondrial clearance. We therefore stained reticulocytes for the processed form of microtubule-associated protein light chain 3 (LC3) that is characteristic of autophagosome formation²⁴. Wild-type and Nix^−/− reticulocytes had similar levels of LC3 punctate staining at the initiation of culture (Fig. 3c; LC3 dots per cell at day 0: wild type, 19.7 ± 3.9; Nix^−/−, 21.3 ± 4.5; P=0.19), indicating that the formation of autophagosomes was normal in the absence of Nix. Notably, LC3 co-localized with COX IV in wild-type but not Nix^−/− reticulocytes (Fig. 3c). Moreover, wild-type but not Nix^−/− reticulocytes had lost most COX IV staining by day 4 of in vitro culture (Fig. 3c). Electron microscopy showed that autophagic vacuoles in wild-type reticulocytes contained engulfed mitochondria; however, most mitochondria remained clustered outside of the autophagic vacuoles in Nix^−/− reticulocytes (Fig. 3d and Supplementary Fig. 6). This suggests that Nix has a specific role in targeting the mitochondria into autophagosomes for clearance. In contrast to defective mitochondrial removal, Nix^−/− RBCs showed no retention of the nucleus and underwent the loss of reticular staining similar to wild-type cells (Fig 1a, and Fig 3b and Supplementary Fig. 3c). Therefore, Nix seems to be required for the clearance of mitochondria, but not the nucleus or ribosomes, during erythroid maturation.

Incubation of Nix with purified mitochondria induced the loss of ΔΨ_m (ref. ⁹). Wild-type reticulocytes showed continued decreases in ΔΨ_m that preceded mitochondrial removal during in vitro maturation, whereas Nix^−/− reticulocytes were defective in the loss of ΔΨ_m (Supplementary Fig. 7a). This supports a role for Nix in disrupting ΔΨ_m during erythroid maturation. Treatment with carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP), an uncoupling agent that dissipates the proton gradient across mitochondrial inner membrane to abolish ΔΨ_m (ref. ²⁵) (Supplementary Fig. 7c, d), promoted the clearance of mitochondria and restored the entry of the mitochondria into autophagic vacuoles in Nix^−/− reticulocytes (Fig. 4a, b and Supplementary Fig. 4b). A BH3 mimetic, ABT-737 (ref. ²⁶), which induced the loss of ΔΨ_m (Supplementary Fig. 7c), also restored the entry of mitochondria into autophagosomes and promoted mitochondrial removal in Nix^−/− reticulocytes (Fig. 4a, c and Supplementary Fig. 4b). ABT-737 has been shown to promote the initiation of autophagy by releasing beclin 1 from the suppression mediated by Bcl-2 or Bcl-xL²⁷. Because Nix is not required for autophagosome formation (Fig. 3c), ABT-737 potentially rescued mitochondrial autophagy in Nix^−/− reticulocytes by inducing the loss of ΔΨ_m. Although a BH3 mimetic can rescue the defects in Nix^−/− RBCs, deletion of another BH3-only molecule, Bim, did not affect erythroid differentiation (Supplementary Fig. 8). Therefore, Nix may be a specific sensor triggered by maturation signals in reticulocytes to promote mitochondrial autophagy.

Figure 4. Removal of mitochondria by autophagy in FCCP-treated or ABT-737-treated Nix^−/− reticulocytes.

a, Ter119⁺ CD71⁺ reticulocytes were sorted from the peripheral blood of wild-type and Nix^−/− mice at day 6 after PHZ treatment. The cells were cultured for in vitro maturation in the presence of 10 µM FCCP or 1 µM ABT-737, followed by staining with Mitotracker deep red and phycoerythrin-conjugated anti-Ter119. b, c, Reticulocytes were cultured with FCCP (b) or ABT-737 (c) for 24 h and analysed by transmission electron microscopy. Scale bar, 0.5 µm.

Our study identifies an unexpected function for Nix in mitochondrial autophagy during erythroid differentiation. Nix-dependent loss of ΔΨ_m may induce the display of molecules on the outer surface of mitochondria for recognition and sequestration by autophagosomes (Supplementary Fig. 1). Two yeast mitochondrial proteins, Uth1p (ref. ²⁸) and Aup1p (ref. ²⁹), have been shown to regulate mitochondrial autophagy in yeast. Here, we provide an example of specific regulation of mitochondrial autophagy in mammalian cells. Defective mitochondrial removal in Nix^−/− RBCs was associated with spontaneous caspase activation, increased phosphatidylserine display and accelerated clearance by macrophages. Defective autophagy-dependent maturation and increased RBC clearance may contribute to anaemia in Nix^−/− mice. In human K562 cells that were induced to undergo erythroid maturation, Nix-dependent removal of mitochondria was also observed (Supplementary Fig. 9). Defects in autophagic maturation of erythroid cells have been linked to anaemia in both humans and mice^4,14. Understanding the mechanisms for mitochondrial autophagy in erythroid cells should facilitate the development of novel therapeutic approaches for treating haematological disorders involving defective erythroid maturation. It may also shed light on the mechanisms underlying mitochondrial quality control by autophagy in the protection against ageing, cancer and neurodegenerative diseases³⁰.

METHODS SUMMARY

Mice

A mouse embryonic stem cell clone with a gene trap insertion targeting the Nix locus (The Sanger Institute Gene Trap Consortium) was used to generate Nix^−/− mice (Supplementary Fig. 2). Mice were backcrossed to the C57BL/6 background for more than five generations.

Analyses of autophagosomes

RBCs were incubated with mouse anti-COX IV (Invitrogen) and a rabbit antibody to processed LC3 (Abgent), followed with Alexa Fluro-conjugated secondary antibodies (Molecular Probes). The slides were examined by deconvolution microscopy (Applied Precision). RBCs were also fixed and embedded in Spurr’s low-viscosity resin, and the sections were prepared and stained with uranyl acetate and lead citrate for analyses by transmission electron microscopy.

Analyses of apoptosis signalling

Cell lysates were used for SDS–PAGE and western blot analysis by probing with antibodies to different caspases, followed by incubation with horseradish peroxidase-conjugated secondary antibodies and developed by the chemiluminescent method (Pierce). Caspase activities were also measured using Ac-DEVD-pNA caspase-Glo reagent (Promega). To measure the display of phosphatidylserine on the cell surface, RBCs labelled with Mitotracker deep red were incubated with FITC–annexin V and analysed by flow cytometry.

Full Methodand any associated references are available in the online version of the paper at http://www.nature.com/nature.

METHODS

Mice

A mouse embryonic stem cell clone with a gene trap insertion targeting the Nix locus (The Sanger Institute Gene Trap Consortium) was used to generate Nix^−/− mice (Supplementary Fig. 2). Mice were backcrossed to the C57BL/6 background for more than five generations.

Flow cytometry and immunocytochemistry

Erythroid cells from peripheral blood, bone marrow or spleen of wild-type and Nix^−/− mice were stained with FITC–anti-CD71 and phycoerythrin–anti-Ter119 (BD Bioscience). Phosphatidylserine display was measured by staining with FITC–annexin V (BD Biosciences). To stain mitochondria, wild-type and Nix^−/− RBCs were labelled with 200 nM Mitotracker deep red (Molecular Probes) at 37 °C for 30 min. The cells were then stained with FITC–anti-CD71 and phycoerythrin–anti-Ter119 and analysed by flow cytometry. For immunocytochemistry analyses, RBCs from wild-type and Nix^−/− mice were stained with FITC–anti-CD71 and applied to slides by cytospin. The cells were fixed, incubated with mouse anti-COX IV (Invitrogen) and Alexa Fluor-conjugated secondary antibody. For mitochondria and LC3 co-staining, wild-type and Nix^−/− RBCs were incubated with mouse anti-COX IV (Invitrogen) and a rabbit antibody to processed LC3 (Abgent), followed by staining with Alexa Fluor-conjugated secondary antibodies (Molecular Probes). The slides were examined using an AxioPlan2 fluorescent microscope (Zeiss) or a SoftWorx Image deconvolution microscope (Applied Precision). The Pearson coefficiency of correlation for LC3 and COX IV co-localization was determined by using the SoftWorx software (Applied Precision), with a value greater than 0.5 considered to have co-localization between the two signals.

Measurement of clearance of RBCs

Wild-type and Nix^−/− mice were injected intravenously with (150 mg kg⁻¹ body weight) NHS-biotin (Pierce), which covalently binds to free amino groups of cell surface proteins¹⁷. Cell clearance was determined by the loss of biotinylated RBCs¹⁷. RBCs were collected at indicated time points and the cells were incubated with phycoerythrin–Streptavidin (BD Bioscience). The percentage of labelled cells was analysed by flow cytometry. RBCs of wild-type or Nix^−/− mice were also labelled with 4mM CMFDA (Molecular Probes) which emits green fluorescence after cleavage by intracellular esterases¹⁸. The labelled cells were injected into wild-type recipient mice intravenously. Blood was collected at indicated time points to quantify CMFDA-labelled cells by flow cytometry.

Induction of reticulocytosis in mice

Mice were injected intraperitoneally on day −1 and day 0 with phenylhydrazine (PHZ, Sigma; 60 mg kg⁻¹ body weight). Ter119⁺CD71⁺ reticulocytes were then sorted on day 3 or day 6 after PHZ treatment using a FACSAria cell sorter (BD Bioscience).

Transmission electron microscopy

RBCs of wild-type and Nix^−/− mice were fixed and embedded in Spurr’s low viscosity resin similar to described protocols³¹. Sections were prepared and stained with uranyl acetate and lead citrate, followed by analyses using a Hitachi H-7500 transmission electron microscope.

Quantitative real-time RT–PCR

RNA was isolated from kidneys of 6-week-old wild-type and Nix^−/− mice using the MELT total nucleic acid isolation system (Ambion). Erythropoietin mRNA was quantified by real-time RT–PCR and normalized against 18S RNA as described³².

Analyses of apoptosis signalling

Cell lysates were used for SDS–PAGE and western blot analysis by probing with antibodies to Nix (Kamiya), caspase-6, caspase-9 (Cell Signaling), or actin (Santa Cruz Biotechnology), followed by incubation with horseradish peroxidase-conjugated secondary antibodies (Southern Biotechnology). The blots were developed by the chemiluminescent method (Pierce). To detect caspase activities, cells were added to 96-well plates with caspase-Glo reagent containing Ac-DEVD-pNA (Promega) and incubated at room temperature for 2 h. The relative luminescence unit (RLU) was measured in a luminometer (Labsystems). In some experiments, Nix^−/− RBCs were cultured with solvent control, 4,000 U ml⁻¹ catalase, 100 µM butylated hydroxytoluene, 5 µM tempol or 200 µM qVD-oph for 24 h. Caspase activities were measured as above. To measure the display of phosphatidylserine on the cell surface, RBCs with or without Mitotracker deep red labelling were incubated with FITC–annexin V (BD Biosciences) and analysed by flow cytometry. To measure ROS, wild-type and Nix^−/− RBCs were cultured in vitro and stained with 20 µM 2′,7′-dichlorofluorescin diacetate. Mean fluorescent intensity (MFI) was plotted.

Statistical analysis

Data were presented as the mean ± s.e.m. and P values were determined by two-tailed Student’s t-test using GraphPad Prism software version 4.0 for Macintosh.

Supplementary Material

01

Supplementary Information is linked to the online version of the paper at http://www.nature.com/nature.