Iron-induced NCOA4 condensation regulates ferritin fate and iron homeostasis

Abstract

Iron is not only essential but also a toxic trace element. Under iron repletion, ferritin maintains cellular iron homeostasis by storing iron to avoid iron toxicity. Under iron depletion, the ferritin-specific autophagy adaptor NCOA4 delivers ferritin to lysosomes via macroautophagy to enable cells to use stored iron. Here, we show that NCOA4 also plays crucial roles in the regulation of ferritin fate under iron repletion. NCOA4 forms insoluble condensates via multivalent interactions generated by the binding of iron to its intrinsically disordered region. This sequesters NCOA4 away from ferritin and allows ferritin accumulation in the early phase of iron repletion. Under prolonged iron repletion, NCOA4 condensates can deliver ferritin to lysosomes via a TAX1BP1-dependent non-canonical autophagy pathway, thereby preventing relative iron deficiency due to excessive iron storage and reduced iron uptake. Together, these observations suggest that the NCOA4-ferritin axis modulates intracellular iron homeostasis in accordance with cellular iron availability.

Keywords: NCOA4; autophagy; ferritin; iron metabolism; phase separation.

Figure EV1. NCOA4 forms insoluble condensates under iron‐replete conditions

1. A

   Schematic diagram of ferritin reporter.

2. B

   Schematic representation of the CRISPR screening procedure.

3. C

   RFP/GFP ratio of MEFs expressing ferritin reporter in the CRISPR screen. Representative results from one of two screens are shown.

4. D–G

   (D) HepG2, (E) U2OS, (F) Hepa1–6, and (G) NCOA4 KO MEFs reconstituted with hNCOA4‐myc or mNCOA4‐GFP were treated or not treated with FAC (indicated as F) or Dfo (indicated as D) for 12 h. Cells were lysed with SDS‐containing sample buffer or Triton buffer and analyzed with immunoblotting with indicated antibodies.

5. H

   Wild type MEFs and NCOA4 KO MEFs were treated with 10 μg/ml FAC for 12 h and then immunostained with anti‐NCOA4 and anti‐LAMP1 antibodies. Scale bar, 10 μm.

6. I

   NCOA4 KO MEFs expressing myc‐hNCOA4 were treated with 10 μg/ml FAC for 12 h and then immunostained with anti‐myc and organelle‐specific antibodies. Scale bar, 10 μm.

Source data are available online for this figure.

Figure 1. NCOA4 forms detergent‐insoluble condensates in iron‐replete cells

1. Results of CRISPR screening analyzed by the MAGeCK algorithm.

2. Wild type (WT) and NCOA4 KO MEFs were pretreated with 25 μg/ml ferric ammonium citrate (FAC) for 16 h and then chased with 20 μg/ml cycloheximide (CHX) for the indicated times. Soluble ferritin expression levels were determined by immunoblotting with the indicated antibodies.

3. Wild type MEFs untreated or treated with 10 μg/ml FAC for 12 h were lysed with SDS‐containing sample buffer or Triton buffer, and analyzed by immunoblotting with the indicated antibodies.

4. Schematic representation of fractionation procedures.

5. Subcellular fractions of MEFs treated with 20 μM Dfo for 12 h, 10 μg/ml FAC for 12 h, or 100 μM Deferoxamine (Dfo) for 6 h after treatment with 10 μg/ml FAC for 12 h were prepared by differential centrifugation as described in (D). Each fraction was subjected to SDS–PAGE and analyzed by immunoblotting with the indicated antibodies.

6. Wild type MEFs were untreated, treated with 10 μg/ml FAC for 12 h or treated with 10 μg/ml FAC, 10 μM E64d, and 10 μg/ml pepstatin A (pep A) for 12 h. The cells were immunostained with anti‐NCOA4 and anti‐LAMP1. Scale bar, 10 μm. Quantitative data are shown as the means ± SD of three biological replicates. At least 30 cells were quantified in each replicate. P = 0.033 (Welch two sample t‐test).

7. NCOA4 puncta were photobleached in MEFs stably expressing NCOA4‐GFP after treatment with 10 μg/ml FAC for 12 h, and then fluorescent recovery was monitored. p62 puncta were photobleached in MEFs stably expressing GFP‐human p62, and then fluorescent recovery was monitored. Representative images are shown. Time 0 indicates the start of recovery after photobleaching. Scale bars, 2 μm.

8. Quantitative data of fluorescence recovery in (G) shown as means ± SD. 26 dots (NCOA4) or 25 dots (p62) were quantified from three biological replicates.

9. Time‐lapse imaging of MEFs stably expressing NCOA4‐GFP. Representative fusion and fission images are shown. Red arrowheads indicate the fission event. Scale bars, 5 μm.

Source data are available online for this figure.

Figure 2. Multivalent interactions are required for NCOA4 condensation under iron‐replete conditions

1. Schematic diagram of NCOA4 variants used in this study.

2. NCOA4 KO MEFs reconstituted with NCOA4 variants were treated or not treated with 10 μg/ml FAC for 12 h and then fixed for imaging. Cells were immunostained with an anti‐myc antibody. Representative images are shown. Scale bar, 10 μm.

3. Number of puncta per cell in (B) is shown as the means ± SEM of at least 40 cells in each condition from two biological replicates.

4. NCOA4 KO MEFs reconstituted with NCOA4 variants were treated or not treated with 10 μg/ml FAC for 12 h, fractionated, and analyzed by immunoblotting with the indicated antibodies.

5. DISOPRED3 disorder score of human NCOA4.

6. Wild type MEFs stably expressing NCOA4 IDR (aa 167–334) fused with tandem FKBP were treated with 10 μg/ml FAC or 20 μM Dfo for 12 h and then cultured with 0.1 nM FKBP ligand (AP20187) for 6 h. The cells were fixed for imaging and immunostained with anti‐myc to analyze the puncta of the fusion protein. Scale bar, 10 μm.

7. Number of puncta in (F) are shown as means ± SEM of at least 35 cells in each condition from two biological replicates. P < 0.0001 (Dfo + ligand vs F.AC + ligand) and P = 0.048 (FAC vs. FAC + ligand; Kruskal–Wallis ANOVA with Dunn's multiple comparison test). P values were adjusted using the Bonferroni method.

8. Coomassie‐stained SDS–PAGE gel of purified proteins.

9. The amount of co‐purifying iron in purified proteins in (H) was measured by inductively coupled plasma mass spectrometry (ICP–MS). Data are shown as means ± SD of three biological replicates. P = 0.032 (Welch two sample t‐test).

Source data are available online for this figure.

Figure EV2. Multivalent interactions are required for NCOA4 condensation under iron repletion

1. Image of purified GST and GST‐NCOA4 IDR.

2. Schematic diagram of NCOA4 variants used in (C–F) and Fig 6.

3. NCOA4 KO MEFs reconstituted with NCOA4 variants were treated or not treated with 10 μg/ml FAC for 12 h, fractionated, and analyzed by immunoblotting with the indicated antibodies.

4. NCOA4 KO MEFs reconstituted with NCOA4 variants were treated with 10 μg/ml FAC for 12 h and then immunostained with anti‐myc antibody and DAPI. Scale bar, 10 μm.

5. Quantitative data of (D) are shown as means ± SEM of at least 90 cells in each condition from two biological replicates.

6. NCOA4 puncta were photobleached in MEFs stably expressing NCOA4 ΔIDR‐GFP or ΔIDR + ΔC‐GFP after treatment with 10 μg/ml FAC for 12 h, and then fluorescent recovery was monitored. Representative images are shown. Time 0 indicates the start of recovery after photobleaching. Scale bars, 2 μm. Quantitative data are shown as means ± SD. 11 dots (ΔIDR) or 10 dots (ΔIDR + ΔC) were quantified from two biological replicates.

Source data are available online for this figure.

Figure EV3. Iron directly induces NCOA4 condensation

1. A

   Coomassie‐stained SDS–PAGE gel of purified human NCOA4.

2. B

   Purified NCOA4 proteins were incubated with FeSO₄ for 1 h under anaerobic conditions and fractionated. Samples were subjected to SDS–PAGE and stained with Coomassie Brilliant Blue.

3. C

   Quantitative results in (B) are shown as means ± SEM of three biological replicates.

4. D–F

   Purified NCOA4 proteins were incubated with 50 μM FeSO4 and (D) 500 μM diethylenetriamine pentaacetic acid (DTPA), (E) 100 μM Dfo, or (F) 100 μM bathophenanthroline disulfonic acid (BPS) for 1 h at 37°C under aerobic conditions. Samples were fractionated and subjected to SDS–PAGE; gels were stained with Coomassie Brilliant Blue.

5. G

   FlAsH‐NCOA4 and FeSO4 were incubated in each concentration for 1 h at 37°C under aerobic conditions and then were observed by confocal fluorescence microscopy. Scale bar, 10 μm.

6. H

   Wild type or TAX1BP1 KO MEFs (clone sg1) in which FTH1 was knocked out by lentiCRISPR were treated or not treated with 10 μg/ml FAC for 12 h. Cell lysates were fractionated and analyzed by immunoblotting with the indicated antibodies.

7. I

   MEFs in which FTH1 was knocked out by lentiCRISPR were treated with 10 μg/ml FAC for 12 h. The cells were immunostained with anti‐NCOA4. Scale bar, 10 μm.

8. J

   Additional electron micrograph of NCOA4 KO MEFs reconstituted with myc‐hNCOA4. Cryosections were labeled with anti‐myc antibodies. Scale bar, 200 nm.

Source data are available online for this figure.

Figure 3. Iron induces NCOA4 condensation

1. A

   Purified NCOA4 proteins were incubated with FeSO₄ for 1 h under aerobic conditions and fractionated. Samples were subjected to SDS–PAGE and stained with Coomassie Brilliant Blue.

2. B

   Quantitative results in (A) are shown as means ± SEM of three biological replicates.

3. C

   Purified FlAsH‐NCOA4 (20 nM) and 10 μM FeSO₄ were incubated at 37°C under aerobic conditions and observed at the indicated times. Scale bar, 10 μm.

4. D

   Phase diagram of the formation of FlAsH‐NCOA4 condensates at the indicated protein and FeSO₄ concentrations.

5. E

   Purified FlAsH‐NCOA4 (20 nM) and the indicated metals were incubated at 37°C under anaerobic conditions for 1 h and observed. Scale bar, 10 μm.

6. F

   Electron micrograph of NCOA4 KO MEFs reconstituted with myc‐hNCOA4. Cryosections were labeled with anti‐myc antibody. Scale bar, 200 nm.

7. G

   Quantification of the diameter of NCOA4 condensates. The red line shows the mean. The dot plot represents the diameter of 36 condensates from two biological replicates.

8. H, I

   Representative energy‐dispersive X‐ray spectra of (H) cytosol and (I) condensate. Black arrowheads indicate Fe (Kα) peaks.

9. J

   Quantification of iron count in cytosol and condensates. Data are shown as the means ± SEM of at least 10 cytosols or condensates from two biological replicates. P < 0.0001 (Welch two sample t‐test).

Source data are available online for this figure.

Figure 4. NCOA4 condensates fine‐tune ferritin levels via two distinct mechanisms

1. Mouse embryonic fibroblasts were treated with 10 μg/ml FAC for the indicated times and harvested. Cell lysates were fractionated and analyzed by SDS–PAGE with the indicated antibodies. Asterisks indicate a non‐specific band. TfR1, transferrin receptor 1.

2. Mouse embryonic fibroblasts were cultured for 12 h in a medium containing 10 μg/ml FAC supplemented with DMSO or 10 μM E64d and 10 μg/ml pepstatin A (pepA). Samples were analyzed by immunoblotting with the indicated antibodies. The black arrowhead indicates an intact ferritin band, and the white arrowhead indicates a partial degradation band of ferritin.

3. NCOA4 KO MEFs reconstituted with hNCOA4‐myc were cultured with 10 μg/ml FAC for the indicated times and fixed for imaging. Cells were immunostained with ferritin and myc antibodies. Scale bar, 10 μm.

4. Myc (NCOA4) puncta were extracted, and average signal intensities of myc and ferritin in the puncta in (C) were plotted.

5. Mouse embryonic fibroblasts expressing hNCOA4‐TurboID were treated with 10 μg/ml FAC for 6 h or 24 h and then cultured with DMSO or 50 μM biotin for 30 min. Cells were lysed with Triton buffer, and lysates were pulled down by streptavidin beads. Inputs and pulldown samples were analyzed with immunoblotting by indicated antibodies.

6. Wild type MEFs pretreated with 10 μg/ml FAC for 6 h or 24 h were chased with 20 μg/ml CHX and analyzed by immunoblotting with the indicated antibodies.

7. NCOA4 KO MEFs reconstituted with myc‐hNCOA4 were pretreated with 10 μg/ml FAC and then cultured with 20 μg/ml CHX or 200 nM bafilomycin A1 (BafA). Cells were fixed for imaging and immunostained using myc and LAMP1 antibodies. Scale bar, 10 μm.

Source data are available online for this figure.

Figure EV4. NCOA4 condensates fine‐tune ferritin levels to maintain cellular iron homeostasis

1. Quantitative results in Fig 4A are shown as the mean intensities of FTH1, FTL, TfR1, and FBXL5 bands. FTH1, n = 3; FTL, n = 7, and others, n = 5 biological replicates.

2. Wild type MEFs or TAX1BP1 KO MEFs were incubated with 10 μg/ml FAC for the indicated times and stained with calcein‐AM. The fluorescence of calcein‐AM was measured by FACS. Data are shown as means ± SEM of three biological replicates.

Figure 5. TAX1BP1 is essential for the clearance of NCOA4 condensates and ferritin via recruitment of FIP200

1. A

   Wild type, ATG7 KO, TAX1BP1 KO, and FIP200KO MEFs were treated or not treated with 10 μg/ml FAC for 12 h. Cell lysates were fractionated and analyzed by immunoblotting with the indicated antibodies.

2. B

   Schematic diagram of TAX1BP1 variants used in this study. NB, NCOA4 binding.

3. C

   HEK293T cells were transfected with GFP‐NCOA4 and the indicated FLAG‐His‐TAX1BP1 variants. Soluble lysates extracted from transfected cells were immunoprecipitated with anti‐FLAG antibodies. Inputs and immunoprecipitated samples were analyzed by immunoblotting with the indicated antibodies.

4. D–F

   Wild type MEFs and TAX1BP1 KO MEFs reconstituted with the indicated TAX1BP1 variants were untreated or treated with 10 μg/ml FAC for 12 h. (D, F) Cell lysates were fractionated and analyzed by immunoblotting with the indicated antibodies. (E) Cells were fixed for imaging and immunostained to analyze ferritin localization. Scale bar, 10 μm.

Source data are available online for this figure.

Figure EV5. TAX1BP1 is essential for clearance of insoluble NCOA4 condensates and ferritin under iron repletion

1. A

   Wild type, ATG7 KO, TAX1BP1 KO, and FIP200KO MEFs were pretreated with 10 μg/ml FAC for 12 h and cultured with or without 20 μg/ml CHX for 1 h. Cell lysates were fractionated and analyzed by immunoblotting with the indicated antibodies.

2. B

   Wild type and ATG7 KO MEFs pretreated with 25 μg/ml FAC for 15 h were cultured with or without 20 μg/ml CHX at the indicated times. Soluble lysates were analyzed by immunoblotting with the indicated antibodies.

3. C, D

   HEK293T cells were transfected with GFP‐NCOA4 and the indicated FLAG‐His‐TAX1BP1 variants. Soluble lysates extracted from transfected cells were immunoprecipitated with anti‐FLAG antibodies. Inputs and immunoprecipitated samples were analyzed by immunoblotting with the indicated antibodies.

Source data are available online for this figure.

Figure 6. The properties of NCOA4 affect the TAX1BP1 dependence of ferritin degradation and interaction with ferritin

1. A

   Mouse embryonic fibroblasts were pretreated with 10 μg/ml FAC for 12 h and then cultured with 100 μM Dfo for the indicated time. Cells were lysed with SDS‐containing sample buffer or Triton buffer. Lysates were analyzed by immunoblotting with the indicated antibodies.

2. B

   Mouse embryonic fibroblasts were treated with 20 μM Dfo for 12 h, 10 μg/ml FAC for 12 h, or 100 μM Dfo for 6 h after treatment with 10 μg/ml FAC for 12 h. Triton‐soluble lysates were subjected to glycerol gradient ultracentrifugation analysis.

3. C

   Wild type and TAX1BP1 KO MEFs were pretreated with 25 μg/ml FAC for 16 h and then cultured with 100 μM Dfo. Lysates were fractionated and then analyzed by immunoblotting with the indicated antibodies.

4. D

   Soluble FTL bands in (C) were quantified by densitometry. Data are shown as the means ± SD of three biological replicates.

5. E

   TAX1BP1 KO MEFs (sg1 clone) stably expressing FLAG‐OsTIR1 (F74G) and FLAG‐mAID‐mTAX1BP1 were pretreated with 10 μg/ml FAC for 12 h, cultured with DMSO or 1 μM 5‐Ph‐IAA for 2 h, and then treated with 100 μM Dfo. Soluble lysates were analyzed by immunoblotting with the indicated antibodies. 5‐Ph‐IAA, a derivative of Auxin; mAID, mini auxin‐inducible degron.

6. F

   TAX1BP1 KO MEFs (clone sg1) in which ATG7 was knocked down were pretreated with 10 μg/ml FAC for 12 h and then cultured with 100 μM Dfo. Soluble lysates were analyzed by immunoblotting with the indicated antibodies.

7. G

   TAX1BP1 KO MEFs (clone sg1) stably expressing myc‐hNCOA4 pretreated with 25 μg/ml FAC for 16 h were cultured with 100 μM Dfo, and soluble lysates were analyzed by immunoblotting with the indicated antibodies.

8. H, I

   NCOA4 KO MEFs reconstituted with NCOA4 variants were pretreated with 10 μg/ml FAC for 12 h and then chased with (H) 20 μg/ml CHX or (I) 100 μM Dfo for the indicated times. Soluble lysates were analyzed by immunoblotting with the indicated antibodies.

9. J

   Mouse embryonic fibroblasts expressing hNCOA4‐TurboID were treated with 10 μg/ml FAC for 3 h (indicated as F) or 100 μM DFO for 3 h after treatment with 10 μg/ml FAC for 12 h (indicated as D), and then cultured with DMSO or 50 μM biotin for 30 min. Triton‐soluble cell lysates were pulled down by streptavidin beads. Inputs and pulldown samples were analyzed by immunoblotting with the indicated antibodies. The black and white arrowheads indicate hNCOA4‐TurboID and endogenous NCOA4, respectively.

Source data are available online for this figure.

Figure 7. Schematic summary of the proposed model

Model of the regulation of ferritin fate by NCOA4 in cellular iron homeostasis. Under iron depletion, ferritin is degraded by the NCOA4 dependent macroautophagy pathway. In the early phase of iron treatment, NCOA4 forms condensates that sequester ferritin for ferritin accumulation. During prolonged iron treatment, NCOA4 condensates bind ferritin, forming a complex that is trafficked to the lysosome in a TAX1BP1‐dependent manner to avoid iron deficiency.