HRG-9 homologues regulate haem trafficking from haem-enriched compartments

Abstract

Haem is an iron-containing tetrapyrrole that is critical for a variety of cellular and physiological processes^1-3. Haem binding proteins are present in almost all cellular compartments, but the molecular mechanisms that regulate the transport and use of haem within the cell remain poorly understood^2,3. Here we show that haem-responsive gene 9 (HRG-9) (also known as transport and Golgi organization 2 (TANGO2)) is an evolutionarily conserved haem chaperone with a crucial role in trafficking haem out of haem storage or synthesis sites in eukaryotic cells. Loss of Caenorhabditis elegans hrg-9 and its paralogue hrg-10 results in the accumulation of haem in lysosome-related organelles, the haem storage site in worms. Similarly, deletion of the hrg-9 homologue TANGO2 in yeast and mammalian cells induces haem overload in mitochondria, the site of haem synthesis. We demonstrate that TANGO2 binds haem and transfers it from cellular membranes to apo-haemoproteins. Notably, homozygous tango2^-/- zebrafish larvae develop pleiotropic symptoms including encephalopathy, cardiac arrhythmia and myopathy, and die during early development. These defects partially resemble the symptoms of human TANGO2-related metabolic encephalopathy and arrhythmias, a hereditary disease caused by mutations in TANGO2^4-8. Thus, the identification of HRG-9 as an intracellular haem chaperone provides a biological basis for exploring the aetiology and treatment of TANGO2-related disorders.

Extended Data Fig. 1 |. Analysis of hrg-9 and hrg-10 expression in C. elegans.

a, Workflow of RNA sequencing experiment. Synchronized L1 worms were grown in mCeHR2 medium containing 2 μM, 20 μM, or 400 μM haem to the gravid stage, and the intestines were isolated for transcriptomics analysis using the SMART-Seq technology. b, The hrg-9p∷nls-gfp transcriptional reporter is expressed in the intestine of C. elegans at all developmental stages. nls, nuclear localization signal; i, intestine. Scale bars, 20 μm. c, The hrg10p∷nls-gfp transcriptional reporter is predominantly expressed in the intestine of C. elegans at all developmental stages. nls, nuclear localization signal; i, intestine. Scale bars, 20 μm. d, Knockdown of hrg-9 and hrg-10 by RNAi does not impair the intestinal secretion of YP170∷GFP in C. elegans. The known trafficking gene rab-10 is used as a control. i, intestine; o, oocyte; e, embryo. Scale bars, 20 μm. e, RNA-seq demonstrates that intestinal hrg-10 expression is not significantly affected by haem levels in C. elegans. n = 3 independent samples for each group. f, Quantitative RT-PCR shows that haem does not regulate hrg-10 expression at the organismal level in C. elegans. n = 3 independent experiments. g, Haem does not regulate the expression of hrg-10p∷nls-gfp transcriptional reporter. Scale bars, 20 μm. Data in e and f are presented as mean ± s.e.m. Statistical significance was determined by one-way ANOVA followed by Tukey’s test.

Extended Data Fig. 2 |. hrg-9 and hrg-10 regulate haem homeostasis in C. elegans.

a, Knockout of hrg-9 and hrg-10 in C. elegans. The sgRNA targeting sites and the positions of genotyping primers (arrows) are shown in the knockout strategies (top), and the genotyping results are shown at the bottom. b, Representative raw mass spectra of ¹⁵N-haem and ¹⁴N-haem mixture from C. elegans. The ¹⁴N-haem shows a mass to charge ratio (m/z) of 679.51 and the ¹⁵N-haem shows m/z of 680.51 and 681.56. c, The uptake of ¹⁵N-haem in hrg-9 and hrg-10 knockout worms is comparable to that of wild type worms. Worms were incubated with 4 μM ¹⁵N-haem for 3 h, and the total haem in worms were extracted for analysis by mass spectrometry. Wild type worms cultured with 200 μM haem were used as a control. n = 3 independent experiments. d,e, Representative images (d) and quantification (e) of zinc mesoporphyrin IX (ZnMP) staining in worms treated with control, mrp-5, hrg-9, or hrg-10 RNAi. The known haem transporter mrp-5 is used as a control. n = 30 worms examined over 3 independent experiments. Scale bars, 20 μm. f, The ZnMP accumulation phenotypes in hrg-9 and hrg-10 knockout worms are rescued by the expression of HRG-9∷GFP and HRG-10∷GFP, respectively. tg9, HRG-9∷GFP; tg10, HRG-10∷GFP. n = 30 worms examined over 3 independent experiments. g, Wild type, hrg-9, hrg-10, and hrg-9;hrg-10 adults were stained with ZnMP and LysoSensor Green. Scale bars, 5 μm. h, ZnMP colocalizes with PGP-2∷GFP, a marker of lysosome-related organelles, in control, hrg-9, and hrg-10 RNAi worms. Scale bars, 10 μm. Data in c,e,f are presented as mean ± s.e.m. P values were determined by one-way ANOVA followed by Tukey’s test.

Extended Data Fig. 3 |. TANGO2 proteins regulate haem homeostasis in yeast and mammalian cells.

a, Growth curve of wild type, hem1Δ, and tango2Δ yeast cells treated with or without 5-aminolevulinic acid (ALA). OD, optical density at 600 nm. n = 2 technical replicates. b, The tango2Δ yeast cells exhibit increased expression (middle) and activity (top) of Ctt1p. GAPDH is used as a control. c,d, Dynamics of haem trafficking to mitochondria (c) and the cytosol (d) in wild type and tango2Δ cells. Labile haem levels in mitochondria and the cytosol were monitored by measuring fractional saturation of the haem sensors specifically targeted to these two locations, respectively, following initiation of haem synthesis. e, Flag-tagged human TANGO2 colocalizes with the mitochondrial marker COX IV in HEK293 cells. Scale bars, 10 μm. f, Western analysis of TANGO2 in whole lysates as well as the mitochondrial and non-mitochondrial fractions of HEK293 cells. The mitochondrial protein COX IV and the cytosolic protein GAPDH were used as controls. g, Strategy to knock out TANGO2 in HEK293 cells and verification of the knockout by genotyping. The sgRNA targeting sites and the positions of genotyping primers are indicated. h, Immunoblotting analysis validating the knockout of TANGO2 in HEK293 cells. i, Total haem content of the wild type and ΔTANGO2 HEK293 cells after treatment with succinylacetone (SA) for 24 h. j, Strategy to knock out Tango2 in MEL cells and verification by PCR genotyping. The sgRNA targeting sites and the positions of genotyping primers are indicated. k, Immunoblotting analysis validating the knockout of Tango2 in MEL cells. l, o-dianisidine staining of haemoglobin in wild type and Tango2-knockout MEL cells. Scale bar, 20 μm. m,n, Basal oxygen consumption rates (OCR, m) and maximal OCR (n) of wild type and TANGO2-knockout HEK293 and MEL cells. o, Representative images of TMRE staining in wild type and TANGO2-knockout HEK293 and MEL cells. Scale bars, 20 μm. n = 3 independent experiments for c,d,i,m,n. Data in a,c,d,i,m,n are presented as mean ± s.e.m. P values were determined by one-way ANOVA followed by Tukey’s test.

Extended Data Fig. 4 |. Loss of tango2 leads to reduced locomotor activity of zebrafish larvae.

a, Strategy to knock out tango2 in zebrafish and verification of the knockout by PCR genotyping. The sgRNA targeting sites and the positions of PCR primers are indicated. b, Comparison of the sequences between the wild type tango2 and the mutated allele in zebrafish. The exon2 - intron2 junction is deleted in the mutant allele. c, Verification of intron 2 retention in the tango2^−/− zebrafish at the mRNA level by reverse transcription (RT)-PCR. d, The predicted protein product of Tango2 in tango2^−/− zebrafish. e, Whole-mount in situ hybridization showing maternal expression of zebrafish tango2. Arrowheads indicate the expression of tango2 in 1-cell and 1000-cell stage embryos. Arrows indicate the expression of gata1 (control) in the intermediate cell mass and tango2 in neuronal tissues at 24 h post-fertilization (hpf). Scale bars, 200 μm. f, o-dianisidine staining of wild type and tango2-knockout zebrafish embryos at 48 and 72 hpf. Scale bars, 200 μm. g, Total haem content in wild type and tango2-knockout zebrafish embryos at 48 and 72 hpf. n = 3 independent experiments. Data are presented as mean ± s.e.m. h, Whole-mount in situ hybridization showing normal expression of the erythroid markers gata1 and αe3-globin in tango2-knockout zebrafish embryos. Scale bars, 200 μm. i, Knockout of tango2 did not reduce erythrocytes in Tg(globin-LCR:EGFP) transgenic zebrafish embryos. n = 3 independent experiments. Data are presented as mean ± s.e.m. j, Distance travelled per minute by 12-dpf wild type and tango2-knockout zebrafish larvae during the 10-min dark / 10-min light cycles. Black bars indicate dark and white bars indicate light. n = 16 fish for each group. k, Representative swimming tracks of 12-dpf wild type and tango2-knockout zebrafish larvae over a 40-min period under constant light. Scale bars, 2.5 mm. Statistical significance was determined by one-way ANOVA followed by Tukey’s test.

Extended Data Fig. 5 |. Knockout of tango2 causes brain and heart defects in zebrafish larvae.

a, Representative micro-CT images of the brains of wild type and tango2-knockout zebrafish larvae at 12 dpf. Images on the left are side views of the 3D reconstructed zebrafish brains. Numbers in the images indicate the distance (μm) from the top of the brain, as shown in the 3D images. Scale bars, 100 μm. b, Quantitative RT-PCR analysis of c-fos in wild type and tango2-knockout zebrafish larvae at 12 dpf. Data are presented as mean ± s.e.m from 3 independent experiments. Each sample contains a pool of 30 zebrafish larvae. c, Heart rates of wild type and tango2-knockout zebrafish larvae at 12 dpf. n = 28 fish for each group. Data are presented as mean ± s.e.m. d, Measurement of systolic time interval and diastolic time interval in wild type and tango2-knockout zebrafish larvae at 7 dpf. n = 12 fish for each group. Data are presented as mean ± s.e.m. e, Detection of heart contractions in 7-dpf wild type and tango2-knockout zebrafish larvae using the Changing Pixel Intensity Algorithm. Peaks indicate individual heart contractions. P values were determined by one-way ANOVA followed by Tukey’s test.

Extended Data Fig. 6 |. TANGO2 transfers haem in vitro.

a, Quantification of ZnMP fluorescence in wild type, hrg-9 and hrg-10 double knockout worms, and the double knockout worms expressing TANGO2 genes from human (hTANGO2), yeast (ytango2), or zebrafish (ztango2). n = 30 worms examined over 3 independent experiments. Data are presented as mean ± s.e.m. P values were determined by one-way ANOVA followed by Tukey’s test. b,c, Haem transfer from mitochondria isolated from mouse liver to apo-myoglobin in the absence or presence of yeast Tango2p (b) and zebrafish Tango2 (c). d,e, Haem transfer assays for human TANGO2, BSA (d), and GFP (e). Mitochondria isolated from mouse liver were used as the haem donor and apo-myoglobin was used as the haem acceptor. f,g, Incorporation of 10 μM (f) and 1 μM (g) haem into apo-myoglobin in the absence or presence of TANGO2. h, Haem transfer from the mitochondria isolated from TANGO2-knockout HEK293 cells to apo-myoglobin in the absence or presence of TANGO2. i, Coomassie blue staining validating the protein digestion by trypsin in mitochondrial membranes isolated from mouse liver. j, Haem transfer from protein-free mitochondrial membranes to apo-myoglobin in the absence or presence of TANGO2. Mito, mitochondria.

Fig. 1 |. hrg-9 regulates haem homeostasis in C. elegans.

a, Genes with altered expression at low (2 μM) or high (400 μM) haem (versus 20 μM haem) from transcriptomics analysis of C. elegans intestine. Previously reported haem-responsive genes are highlighted in red and R186.1 (also known as hrg-9) is shown in blue. b, hrg-9 quantification from the RNA-seq experiment on intestines isolated from worms growing at different haem concentrations. n = 3 for each group. FPKM, fragments per kilobase of transcript per million mapped reads. c, RT–qPCR analyses of hrg-9 expression in worms growing at different haem concentrations. d, Schematic (top) and expression (middle and bottom) of the hrg-9p∷nls-gfp transgene in C. elegans. Scale bar, 20 μm. DIC, differential interference contrast microscopy; i, intestine. e, Haem regulates the expression of hrg-9p∷nls-gfp reporter. Scale bars, 20 μm. g, gut granules. f, Schematic (top) and expression (middle and bottom) of hrg-10p∷nls-gfp reporter in C. elegans. Scale bar, 20 μm. h, hypodermis. g, Subcellular localization of HRG-9∷GFP and HRG-10∷GFP in C. elegans intestinal cells. Scale bars, 10 μm. c, cytoplasm; n, nucleus; a, apical plasma membrane. h,i, Representative images (h) and quantification (i) of hrg-1p∷gfp expression in worms with indicated RNAi. Scale bars, 100 μm. AU, arbitrary units. j,k, Representative images (j) and quantification (k) of hrg-1p∷gfp expression in wild-type and hrg-9-, hrg-10- and hrg-9;hrg-10-knockout worms. Scale bars, 100 μm. l, Quantification of endogenous hrg-1 mRNA by RT–qPCR in worms with indicated genotypes. m, Representative images of worms of the indicated genotypes after exposure to 0.5 μM GaPP for 3 days. Scale bars, 1 mm. n, Survival of worms of the indicated genotypes after exposure to 1 μM GaPP for 4 days. n = 3 independent experiments in c,i,k,l,n. Data in b,c,i,k,l,n are mean ± s.e.m. One-way ANOVA followed by Tukey’s test.

Fig. 2 |. Loss of hrg-9 and hrg-10 results in haem accumulation in LROs.

a, Total haem content in the worms of indicated genotypes. hrg-4 RNAi was used as a control. n = 3 independent experiments. b, Haem uptake in the worms of indicated genotypes. Wild-type worms cultured with 200 μM haem were used as a control. n = 3 independent experiments. c,d, Representative images (c) and quantification (d) of ZnMP staining in the worms of indicated genotypes. n = 30 worms examined over 3 independent experiments. Scale bars, 20 μm. e,f, Representative images (e) and quantification (f) of ZnMP fluorescence in wild-type and hrg-9 and hrg-10 double-knockout worms treated with control or pgp-2 RNAi. n = 30 worms examined over 3 independent experiments. Scale bars, 20 μm. g,h, Quantification of ZnMP fluorescence during ZnMP pulse (g) and chase (h) analyses in wild-type and hrg-9 and hrg-10 double-knockout worms. n = 36 worms examined over 3 independent experiments. i,j, Representative images (i) and quantification (j) of ZnMP fluorescence in wild-type and hrg-9 and hrg-10 double-knockout worms treated with control or hrg-1 RNAi. n = 30 worms examined over 3 independent experiments. Scale bars, 20 μm. Data in a,b,d,f–h,j are mean ± s.e.m. One-way ANOVA followed by Tukey’s test.

Fig. 3 |. TANGO2 regulates mitochondrial haem homeostasis in yeast and mammalian cells.

a, Total haem content in wild-type, hem1Δ and tango2Δ yeast cells. b, The ratio of EGFP to mKATE2 fluorescence for cytosolic haem sensor expressed in wild-type, hem1Δ, and tango2Δ yeast cells. The ratio correlates inversely with labile haem in the cytosol. c, The EGFP/mKATE2 fluorescence ratio for mitochondrial haem sensor in wild-type and tango2Δ yeast cells treated with or without 0.5 mM succinylacetone (SA). The ratio correlates inversely with the labile haem in mitochondria. d, Hap1 reporter activity as assessed by fluorescence of the indicated p415–CYC1–EGFP-expressing yeast cells. e, Dynamics of haem trafficking to the nuclei of wild-type and tango2Δ yeast cells. Labile haem in nuclei was monitored by measuring the fractional saturation of the nuclear haem sensor following initiation of haem synthesis. f–h, Haem content in total lysates (f), non-mitochondrial fractions (g) and mitochondrial fractions (h) of wild-type and TANGO2-knockout HEK293 cells. i, Activity of ascorbate peroxidase expressed in the mitochondria (mito-APX) or cytoplasm (cyto-APX) of wild-type and ΔTANGO2 HEK293 cells. j–l, Haem content in total lysates (j), non-mitochondrial fractions (k) and mitochondrial fractions (l) of wild-type and Tango2-knockout MEL cells. All experiments were performed three times. Data are mean ± s.e.m. One-way ANOVA followed by Tukey’s test.

Fig. 4 |. TANGO2 deficiency leads to mitochondrial defects in mammalian cells and lethality in zebrafish.

a,b, Oxygen consumption rate (OCR) of wild-type and TANGO2-knockout HEK293 (a) and MEL (b) cells. c, Flow cytometric analysis of tetramethylrhodamine ethyl ester (TMRE) staining in wild-type and TANGO2-knockout HEK293 and MEL cells. MFI, mean fluorescence intensity. d, ATP concentration in wild-type and TANGO2-knockout HEK293 and MEL cells. e, Survival curve of wild-type (n = 124) and tango2-knockout (n = 114) zebrafish. f, Representative images of wild-type and tango2-knockout zebrafish larvae at 12 days post-fertilization (dpf). The tango2^−/− larvae are shown under both pre-morbid and morbid conditions. Black arrow indicates pericardial oedema and white arrows indicate the swim bladder. Scale bars, 0.2 mm. g, The distance travelled by 12-dpf wild-type and tango2-knockout zebrafish larvae over a 40-min period. n = 16 fish for each group. h, Representative micro-CT images of brains from wild-type and tango2-knockout zebrafish larvae at 12 dpf. Arrows indicate longitudinal fissures. Scale bars, 100 μm. i, The arrhythmicity index of wild-type and tango2-knockout zebrafish larvae at 7 dpf. n = 12 fish for each group. j, Haematoxylin and eosin-stained muscle sections from wild-type and tango2-knockout zebrafish larvae at 12 dpf. Arrows indicate muscle fibres. Scale bars, 20 μm. k,l, Images of birefringence (k) and phalloidin staining (l) of wild-type and tango2-knockout zebrafish larvae at 12 dpf. Arrows indicate muscle fibres. Scale bars are 50 μm (k) and 20 μm (l). a–d, n = 3 independent experiments. Data in a–d,g,i are mean ± s.e.m. One-way ANOVA followed by Tukey’s test.

Fig. 5 |. TANGO2 binds to and transfers haem in vitro.

a,b, Haem transfer from mitochondria (mito) isolated from HEK293 cells (a) or mouse liver (b) to apo-myoglobin in the absence or presence of TANGO2. c, Haem transfer from mitochondrial membranes isolated from mouse liver to apo-myoglobin in the absence or presence of TANGO2. d, Sequence alignment of the N-terminal regions of HRG-9 and TANGO2 proteins. e, Haem transfer assays for the full-length TANGO2 and TANGO2 lacking the N-terminal 13 (ΔN13) or 28 (ΔN28) amino acids. Mitochondria isolated from mouse liver were used as the haem donor and apo-myoglobin was used as the haem acceptor. f,g, Haem activity assay (top) and immunoblotting analysis (bottom) of human (f) and mouse (g) TANGO2–Flag before and after incubation with haem in the presence of sodium dithionite. Human and mouse TANGO2–Flag were expressed and purified from HEK293 cells and MEL cells, respectively. Arrowheads indicate human or mouse TANGO2–Flag. h,i, Haem activity assay (top) and Coomassie staining (bottom) of yeast Tango2p before and after incubation with haem in the absence (h) or presence (i) of sodium dithionite. Myoglobin and lysozyme were used as positive and negative controls, respectively. j, Subtractive absorption spectra (ΔABS) of yeast Tango2p with increasing concentrations of haem in the absence (ferric haem) or presence (ferrous haem) of sodium dithionite. Differential absorbance was monitored at 419 nm and 414 nm for binding assays with ferric and ferrous haem, respectively. k, Data from j was used to calculate the dissociation constant (K_d). l, Proposed model of the role of HRG-9 or HRG-10 (HRG-9/10) and TANGO2 in regulating intracellular haem homeostasis. In haem auxotrophs (worms) and haem-synthesizing eukaryotes (yeasts and vertebrates), HRG-9/10 and TANGO2 translocate haem out of the haem storage site (LRO) and haem synthesis site (mitochondrion), respectively. HRG-9/10 and TANGO2 may transfer haem from haem-enriched membranes. Mito, mitochondria.