Dual regulation of SLC25A39 by AFG3L2 and iron controls mitochondrial glutathione homeostasis

Summary

Organelle transporters define metabolic compartmentalization and how this metabolite transport process can be modulated is poorly explored. Here, we discovered that human SLC25A39, a mitochondrial transporter critical for mitochondrial glutathione uptake, is a short-lived protein under dual regulation at the protein level. Co-immunoprecipitation mass spectrometry and CRISPR KO in mammalian cells identified that mitochondrial m-AAA protease AFG3L2 is responsible for degrading SLC25A39 through the matrix loop 1. SLC25A39 senses mitochondrial iron-sulfur cluster using four matrix cysteine residues and inhibits its degradation. SLC25A39 protein regulation is robust in developing and mature neurons. This dual transporter regulation, by protein quality control and metabolic sensing, allows modulating mitochondrial glutathione level in response to iron homeostasis, opening avenues for exploring regulation of metabolic compartmentalization. Neuronal SLC25A39 regulation connects mitochondrial protein quality control, glutathione and iron homeostasis, which were previously unrelated biochemical features in neurodegeneration.

Graphical Abstract

In Brief

Shi et al. identify a dual regulation of the putative mitochondrial glutathione transporter SLC25A39 by mitochondrial protease AFG3L2 and iron sensing, which enables an autoregulatory mechanism orchestrating glutathione and iron metabolism in mitochondria.

Introduction

Mitochondrial metabolite transporters^1–3 define metabolic compartmentalization between mitochondria and rest of the cell⁴, and transport dysregulation is associated with many pathological conditions such as neurodegenerative disorders, metabolic syndromes and cancers. Despite the recent discoveries of protein identities that facilitate metabolite transport, the mechanism by which transporter proteins can be regulated remain unexplored.

Given the critical role of mitochondrial redox metabolism and oxidative stress^5–8 in physiology, we focus on the regulation of mitochondrial transport for glutathione, the most abundant endogenous antioxidant. Glutathione⁹ critically supports redox signaling and biosynthesis of iron-sulfur (Fe-S) cluster cofactors inside the mitochondrial matrix^10–12. Because glutathione is exclusively synthesized in the cytosol, how glutathione enters the mitochondrial matrix remained a decades-old mystery until the recent identification that a previously uncharacterized mitochondrial metabolite transporter, SLC25A39 (A39), is critical for mitochondrial glutathione transport^13,14. A39 belongs to the SoLute Carrier 25 (SLC25) family, the largest transporter family that is responsible for translocating diverse metabolite ligands across the mitochondrial inner membrane^1–3. Despite the relatively small mitochondrial volume, mitochondrial glutathione accounts for 10–15% of the total cellular glutathione pool¹⁰. We therefore hypothesize that a direct regulation of A39 could have a major impact on glutathione compartmentalization and cellular metabolism.

Results

SLC25A39 protein regulation depends on GSH and the matrix cysteine residues.

To explore the regulatory mechanism of A39, we performed amino acid sequence analysis^13,15 and Alphafold structure prediction¹⁶ of A39. In addition to the typical structural fold of the SLC25 transporters, the threefold pseudo-symmetrical, six transmembrane α-helices surrounding the central transport cavity, the analysis revealed that A39 contains an extra mitochondrial matrix loop between transmembrane domain 1 and 2 with a low structural prediction confidence score (Fig. 1A). This loop 1 of amino acids 41–105 contains four matrix-exposed, thiol Cysteine (Cys) residues^17,18 (Fig. 1A), which exceeds Cys frequency in proteins at around 1.7%¹⁹. C-terminal FLAG-tagged A39 construct with all four cysteine residues mutated to alanine, A39^4CA-FLAG, as well as A39^WT-FLAG, can be expressed as a level similar to that of the endogenous A39 (Fig. 1B). A39^4CA-FLAG can fully restore mitochondrial GSH depletion in the A39 CRISPR KO cells similar to the wild type A39-FLAG (Fig. 1C), suggesting that the transporter activity is preserved in the A39^4CA mutant, while another cysteine mutant of the substrate binding site A39^C334S-FLAG can only rescue partially (Fig. 1C). However, A39^4CA -FLAG completely abolished A39 protein upregulation upon pharmacological inhibition of glutathione biosynthesis using buthionine sulfoximine (BSO) (Fig. 1B), a regulation specific to A39 protein at the post-transcriptional level (Fig. S1A–C) that was previously reported with an unknown mechanism¹⁴.

Figure 1. SLC25A39 (A39) is a short-lived protein, and the degradation is regulated by GSH and the matrix cysteine residues.

(A) AlphaFold prediction of A39 structure revealed a unique cysteine-rich matrix loop 1.

(B) Western blot of endogenous A39 and ectopically expressed FLAG-tagged A39^WT, substrate binding pocket mutant A39^C334S, and matrix loop 1 four cysteine mutant A39^4CA in the K562 cells upon GSH depletion by BSO treatment (1mM, 2d). GAPDH, loading control.

(C) Mitochondrial GSH level assayed by HA-MITO tag immunoisolation followed by MS-based metabolite profiling in the control and clonal A39 CRISPR KO K562 cells expressing the indicated A39 constructs. Statistical significance was calculated using two-tailed t test. Significance level was indicated as *** p < 0.001, ** p < 0.01. Data are expressed as mean ± SD.

(D) Scatter plot showing the correlation of protein abundance (log₁₀ nM) and mRNA abundance (Log₁₀ TPM) for all coding genes expressed in the HEK293 cells (OpenCell, CZI). Blue dots, SLC25 transporter proteins. A39 and A40 are highlighted in red.

(E-F) Western blot showing A39-FLAG (E) and A39^4CA-FLAG (F) protein level in the clonal A39 CRISPR KO K562 cells treated with CHX (cycloheximide, 150 μg/ml) for the indicated times, either under basal condition (−BSO) or with BSO (1mM and with 2 d prior treatment). GAPDH, loading control. A39 band intensity was quantified and modeled using the non-linear fitting with one phase decay (A39, −BSO; A39^4CA, −BSO and +BSO) and plateau followed by one phase decay (A39, +BSO).

See also Figure S1.

SLC25A39 is a short-lived protein and degradation is regulated.

Because ectopically over-expressed recombinant A39-FLAG protein driven by the CMV promoter can only be expressed as a level similar to that of the endogenous A39 (Fig. 1B), we suspected that A39 protein might be tightly maintained at a low level under basal condition and is then alleviated upon GSH depletion by BSO. Indeed, we first identified that human A39 protein level is extremely low in comparison to the A39 mRNA level, which stands out as a proteome-wide outlier of protein-mRNA correlation in the HEK293T cells (CZI Open Cell²⁰, Fig. 1D) — the majority of the mitochondrial SLC25 transporters exhibit correlated protein and mRNA ratios (Fig. 1D). Because all inner membrane SLC25 transporters are imported through a common TIM22 import mechanism, we suspected that A39 protein stability is regulated. We then performed cycloheximide (CHX) chase experiments and discovered that pre-synthesized A39 protein indeed exhibited a very short half-life < 2 h under basal condition (Fig. 1E). A39^4CA protein exhibited a similar short half-life (Fig. 1F). This half-life is drastically shorter than the majority of mitochondrial proteins^21,22 of approximately median half-life at 87 h²³. BSO significantly extended A39’s half-life (Fig. 1E), but failed to extend the short half-life of A39^4CA (Fig. 1F). This suggests that under basal condition, both pre-synthesized wild-type A39 and A39^4CA mutant were quickly degraded, and BSO treatment inhibited A39 degradation through a mechanism that depends on the matrix cysteines. These two regulatory mechanisms of A39, both fast degradation and BSO-mediated stabilization, are yet to be explored.

Mitochondrial m-AAA protease AFG3L2 is responsible for A39 degradation.

To characterize the quality control pathway that degrades A39, we combined co-immunoprecipitation mass spectrometry (MS) and CRISPR KO in cells to identify its regulatory machinery. We first selected K562 clonal cells ectopically expressing relatively high level of A39-FLAG and a mitochondrial tag (HA-MITO, 3xHA-eGFP-OMP25^24,25) from the clonal A39 CRISPR KO cells (Fig. 2A and S2A). We treated the cells with BSO for 2 days to increase the A39-FLAG protein level (Fig. S2A), rapidly immuno-isolated the mitochondria via HA-MITO using anti-HA magnetic beads, and then immunoprecipitated A39 binding partners using anti-FLAG resin in a mild 1% digitonin-containing lysates (Fig. 2A). Wild type K562 cell expressing HA-MITO tag was used as the control. Silver staining of the immunoprecipitants revealed an enrichment of A39-FLAG protein that migrated around 40 kD, confirmed by MS (Fig. 2A). One major band around 60 kD that is specific to A39-FLAG IP was identified as HSPD1/HSP60 by MS (Fig. 2A), an abundant mitochondrial matrix chaperonin that facilitates folding and refolding of matrix proteins²⁶. HSPD1 enrichment suggested that the matrix portion of A39 protein may be flexible and misfolded, however, because HSPD1 CRISPR KO did not affect A39 protein level under basal condition (Fig. 2C), we did not follow up here. We then performed MS analysis of the entire gel lane from the A39-FLAG immunoprecipitants and control to identify specific A39 binding partners (Table S1). We categorized the hits into pathways relevant to mitochondrial quality control and A39-related metabolite sensing mechanisms that include mitochondrial proteases, matrix chaperones, mitophagy-related, and iron-related proteins (Fig. 2B). We then performed CRISPR to knock out 11 representative protein hits from these pathways to investigate the impact on A39 protein level, both under basal condition and BSO treatment (Fig. 2C). Strikingly, CRISPR KO of only two proteins dramatically stabilized A39 under basal condition, AFG3L2 and ABCB7 (Fig. 2C, red), among which AFG3L2 KO upregulated basal A39 protein level to a much higher level and A39 level in the AFG3L2 KO was no longer sensitive to BSO treatment (Fig. 2C). Western blot (Fig. S2B) and qPCR (Fig. S2C) confirmed successful depletion of AFG3L2 and ABCB7 in the corresponding CRISPR KOs. And mRNA level of the endogenous SLC25A39 did not change (Fig. S2D).

Figure 2. Co-immunoprecipitation mass spectrometry and CRISPR KO identify the mitochondrial m-AAA protease AFG3L2 in A39 degradation.

(A) Left, flow chart showing the experimental procedure to identify putative A39-interacting regulatory proteins. Right, silver stained SDS-PAGE of anti-FLAG immunoprecipitants from the control and A39-FLAG mitochondrial lysates. Wild type K562 cells expressing 3HA-eGFP-OMP25 (HA-MITO tag) were used as control. A39-FLAG protein identified by MS is labeled; *, a specific binding partner HSPD1 identified by MS.

(B) Categorization of specific A39-interacting protein hits into relevant quality control pathways. Protein level intensity based on top three peptides was shown in the control and A39-FLAG co-immunoprecipitants. Bold, hits that were followed up by CRISPR KO.

(C) Western blot of A39 from CRISPR KO K562 cells using sgRNAs targeting indicated A39-interacting proteins, and related mitochondrial proteases. Cells were analyzed under basal condition and BSO (200 μM) treatment for either 1 d or 2 d. Two CRISPR KO lines that increased basal A39 level were highlighted in red. GAPDH, loading control.

(D-E) Western blot of endogenous A39, ectopically expressed FLAG-tagged A39^WT and A39^4CA in the AFG3L2 CRISPR KO cells (+ sgAFG3L2, D) and ABCB7 CRISPR KO cells (+sgABCB7, E).

(F) Western blot of endogenous A39 in HeLa cells upon AFG3L2 CRISPR KO and re-expressing AFG3L2–3xFLAG. GAPDH, loading control.

(G) Western blot of the endogenous A39 protein in the K562 cells treated with CHX (cycloheximide, 150 μg/ml) for the indicated times, with or without AFG3L2 CRISPR KO. GAPDH, loading control.

(H) A39 band intensity was quantified and modeled by the non-linear fitting with one phase decay.

See also Figure S2 and Table S1.

AFG3L2 is the subunit of the mitochondrial m-AAA proteases in the inner membrane with catalytic sites facing the matrix²⁷, and its mutations cause neurological disorders including dominant spinocerebellar ataxia (SCA28)²⁸ and Spastic ataxia 5, autosomal recessive (SPAX5)²⁹. AFG3L2 CRISPR KO increased the basal level of endogenous A39, A39-FLAG and A39^4CA-FLAG (Fig. 2D), suggesting its role in the quality control of A39 (and A39^4CA) under basal condition. ABCB7 is a mitochondrial inner membrane protein and a putative transporter for certain Fe-S cluster species³⁰. Because ABCB7 CRISPR KO only increased wild-type A39-FLAG but not A39^4CA-FLAG (Fig. 2E), we suspected that ABCB7 might function in the second branch of the regulation, the BSO-mediated A39 stabilization, that depends on the matrix cysteines.

The quality control regulation of A39 protein is specific to AFG3L2. Two hexametric m-AAA proteases exist in human, homo-oligomeric AFG3L2 complexes and hetero-oligomeric complexes of the homologous AFG3L2 and SPG7^31,32. Together with i-AAA protease active in the intermembrane space YME1L and other membrane-bound peptidases, they play a central role for inner membrane protein quality control. Notably, our co-immunoprecipitation MS did not recover SPG7 and YME1L, and CRISPR KO of SPG7 and YME1L did not affect A39 protein level (Fig. 2C), suggesting a specificity by AFG3L2.

CHX chase experiment identified that AFG3L2 CRISPR KO completely abolished A39 degradation beyond 8 hrs (Fig. 2G and H) -- a complete stabilization upon AFG3L2 loss is different from the moderate BSO-mediated stabilization (Fig. 1E). This suggests that A39 is a specific, short-lived substrate of AFG3L2, and other mitochondrial proteases cannot compensate and substitute for A39 proteolytic regulation. We validated these findings in HeLa cells, in which AFG3L2 CRISPR KO cells also upregulated A39 level that can be reduced by re-expressing AFG3L2–3xFLAG (Fig. 2F). We therefore concluded that AFG3L2 is necessary for degrading A39 under basal condition and this function cannot be substituted by other mitochondrial proteases.

A39 degradation is targeted through the matrix loop 1, which controls mitochondrial glutathione level.

We then tested whether the matrix loop 1 of A39 is responsible for A39 degradation by AFG3L2. We generated two A39 fusion constructs by swapping the long A39 loop 1 to the corresponding loop of either the homologous transporter SLC25A40 (A40), A39^A40L1 -FLAG, or to that of the dicarboxylate SLC25 transporter SLC25A11 (A11), A39^A11L1-FLAG (Fig 3A and B). As expected, due to the shorter sequences, A39^A11L1-FLAG and A39^A40L1-FLAG migrated at lower molecular weights than A39-FLAG (Fig. 3C). Importantly, A39^A11L1-FLAG and A39^A40L1-FLAG were expressed at a much higher level than A39-FLAG under basal condition (Fig. 3C), and AFG3L2 CRISPSR KO cannot further stabilize these two fusion proteins (Fig. 3C), suggesting they were resistant to degradation. A lack of AFG3L2-mediated degradation for the loop 1-swapping A39 fusion constructs strongly suggested that A39’s matrix loop 1 is responsible for AFG3L2 recognition and degradation.

Figure 3. A39 degradation is targeted through the matrix loop 1, which controls mitochondrial glutathione uptake and steady state level.

(A) Sequence alignment showing matrix loop 1 of A39, SLC25A40, and other representative SLC25 transporters SLC25A15, SLC25A1 and SLC25A11.

(B) Diagram showing Loop 1 swapping chimeric constructs.

(C) FLAG western blot for the indicated constructs in the clonal A39 CRISPR KO K562 cells, upon AFG3L2 CRISPR KO (+sgAFG3L2). GAPDH, loading control.

(D) Western blot of endogenous A39 in the whole cell lysates and mitochondrial lysates from the control and AFG3L2 CRISPR KO K562 cells. Mitochondrial protein VDAC, loading control.

(E) Mitochondrial GSH level assayed by HA-MITO tag immunoisolation followed by MS-based metabolite profiling in the control and AFG3L2 CRISPR KO K562 cells, normalized by total mitochondrial protein abundance.

(F) Western blot of A39 from BSO (1mM, 2 d)-treated clonal A39 CRISPR KO cells expressing either wild type A39 and A39^4CA-FLAG. VDAC, loading control.

(G) GSH uptake (1mM labeled GSH + 4mM unlabeled GSH, 15 min at room temperature) into HA-MITO immunoisolated mitochondria from BSO-treated clonal A39 CRISPR KO cells expressing either wild type A39 or A39^4CA-FLAG.

(H) Western blot of A39^WT and A39^A11L1-FLAG in the clonal A39 CRISPR KO cells.

(I) Mitochondrial GSH level assayed by HA-MITO tag immunoisolation followed by MS-based metabolite profiling in control, A39 CRISPR KO, and A39 CRISPR KO re-expressing A39^WT, or A39^A11L1-FLAG, normalized by NAD⁺ level.

(J) GSH uptake (5mM labeled GSH, 15 min at room temperature) into isolated mitochondria from A39 CRISPR KO cells expressing A39^WT and A39^A11L1-FLAG. All statistical significance was calculated using two-tailed t test. Significance level was indicated as *** p < 0.001, ** p < 0.01. Data are expressed as mean ± SD.

We then explored the functional consequence of A39 upregulation on mitochondrial glutathione level using three different experimental conditions. First, we performed mitochondrial metabolite profiling in the AFG3L2 CRISPR KO cells upon rapid mitochondrial immunoisolation using HA-MITO tag and discovered that the KO mitochondria lacking the protease had a higher A39 protein level as expected (Fig. 3D), and increased GSH level (Fig. 3E). Second, we immunoisolated mitochondria from BSO-treated A39 CRISPR KO cells expressing either A39-FLAG or A39^4CA-FLAG where only A39-FLAG can be upregulated but not A39^4CA-FLAG (Fig. 3F); and then performed in vitro mitochondria-based uptake assay using stable isotope labeled GSH [M+3] (GSH-[glycine-¹³C2, ¹⁵N]) for 15 min. The uptake assay revealed significantly higher GSH uptake in the A39-FLAG-expressing cells (Fig. 3G), despite A39^4CA-FLAG is a functional transporter (Fig. 1C). Third and specifically, we compared mitochondrial glutathione level in the A39 KO cells overexpressing A39-FLAG or A39^A11L1-FLAG, in which A39^A11L1-FLAG is expressed at a much higher level than A39-FLAG (Fig. 3H). Indeed, while A39-FLAG can rescue mitochondrial glutathione depletion in the A39 KO cells to the control level, A39^A11L1-FLAG further increased mitochondrial GSH higher than that of the control level (Fig. 3I). Labeled GSH mitochondrial uptake assay from these cells also confirmed a high mitochondrial glutathione uptake rate (Fig. 3J). Therefore, we concluded that a regulated A39 transporter protein level directly controls mitochondrial matrix glutathione level.

SLC25A39 degradation is inhibited by a coordinated sensing of mitochondrial Fe-S cluster by the matrix cysteines.

What is the mechanism that stabilizes A39 protein during BSO treatment? We first tested and ruled out the possibility that BSO affects AFG3L2 activity, because BSO treatment did not affect level of NDUFA9 (Fig. 4A), whose degradation depends on AFG3L2 in the mammalian cells³³. Because previous studies of A39 identified a coupling between A39 and mitochondrial iron homeostasis^13,14,34,35, we hypothesized a BSO-mediated, A39’s matrix-cysteine-dependent metabolic sensing mechanism. Consistent with this notion, our CRISPR KO assay (Fig. 2C) already observed that CRISPR KO of the putative mitochondrial Fe-S cluster transporter ABCB7 stabilized A39 protein in a similar manner that depends on A39’s matrix cysteines (Fig. 2E). Following up on this, we deployed three strategies to perturb iron homeostasis to investigate the impact on A39 protein level (Fig. 4B). First, depleting cellular iron using an iron chelator deferoxamine (DFO), which is sensed by increased cytosolic iron-sensing IRP2 level (Fig. S3A and S3B), diminished BSO-mediated A39 protein upregulation (Fig. 4C and S3B). Second, depleting mitochondrial iron by double CRISPR KO of the two mitochondrial iron transporters, SLC25A28 and SLC25A37 (Fig. S3C and S3D), completely abolished BSO-mediated A39 protein upregulation (Fig. 4D). Third, inhibiting Fe-S cluster biosynthesis chaperone by CRISPR KO of HSCB or GLRX5 (Fig. S3E and S3F) significantly dampened BSO-mediated A39 protein upregulation (Fig. 4E). Then, to investigate the contribution of individual matrix cysteines in A39, we generated four FLAG-tagged single cysteine mutants, A39^C74A, A39^C78A, A38^C88A and A39^C94A. While all four mutants can be expressed similar to the wild-type level under basal condition, all four mutants exhibited reduced stabilization upon BSO treatment (Fig. S3G), suggesting a coordinated regulation through all four cysteines.

Figure 4. A39 is stabilized through a coordinated sensing of mitochondrial iron-sulfur cluster by the matrix cysteines. A39 regulation occurs in neurons.

(A) Western blot of A39 showing BSO treatment (1mM, 2d) increased A39 in the control K562 cells but did not affect NDUFA9 level, while the basal level of A39 and NDUFA9 were increased in the AFG3L2 CRISPR KO K562 cells.

(B) Diagram of iron manipulation strategies.

(C) Western blot of A39 in K562 cells treated with either or a combination of BSO (1mM, 2d) and DFO (20 μM, 2d).

(D) Western blot of A39 with and without BSO (1mM, 2d) in the CRISPR KO K562 cells of either or both mitochondrial iron transporters SLC25A28 and SLC25A37.

(E) Western blot of A39 upon BSO (1mM) treatment for 1 and 2d in the CRISPR KO K562 cells of HSCB or GLRX5.

(F) Diagram of A39 cysteine labeling and mobility shift assay. Cells were treated with either or a combination of BSO (100 μM, 2d) and DFO (deferoxamine, 20 μM, 2d).

(G) Anti-FLAG western blot and mobility shift assay for A39-FLAG in the indicated conditions, without or with cysteine labeling.

(H) Diagram of neuronal differentiation from the iPSC line, KOLF2.1J cells expressing dox-inducible NGN2 in the safe locus.

(I) Western blot of A39, HSCB, AFG3L2, and Synaptophysin, a synaptic marker from the indicated days (H) during neuronal differentiation. GAPDH, loading control.

(J) Western blot of A39 in the mature iNeurons (differentiation Day 21) upon 1 mM BSO treatment for 3 and 7 d. GAPDH, loading control.

(K) Western blot of A39 and IRP2 in the mature iNeurons (differentiation Day 21), treated with either or a combination of BSO (1mM, 7d) and DFO (20 μM, 7d). GAPDH, loading control.

(L) A39 regulation model.

See also Figure S3.

To assay whether the four matrix cysteines might be directly involved in iron sensing, we applied reactive cysteine labeling reagent, iodoacetyl-PEG2-biotin (MW = 542 Da), to label cysteine residues of A39^WT-FLAG in the A39 and AFG3L2 double KO K562 cell mitochondrial lysates, and assayed iron-dependent labeling efficiency through A39-FLAG mobility shift (Fig. 4F). Because A39 protein contains three additional cysteine residues besides the four matrix cysteines, we observed a modest mobility shift of the labeled A39-FLAG from the BSO-treated condition, corresponding to a partial labeling (Fig. 4G, black asterisk). Importantly, cellular iron depletion by either DFO or BSO + DFO treatment allowed A39-FLAG to be maximally labeled and to migrate at a higher molecular weight (Fig. 4G, cyan asterisk), suggesting more cysteine residues of A39 became exposed upon iron depletion and reactive to labeling. The mobility shift reflects cysteine labeling, because A39-FLAG from mitochondrial lysates without labeling reaction migrated at the similar molecular weight (Fig. 4G, red asterisk) and the protein abundances were comparable due to AFG3L2 KO. We therefore concluded that A39’s four matrix cysteines are responsible for coordinated sensing of matrix iron homeostasis, protecting A39 from degradation.

A39 regulation in the neuronal cells.

Mutations in AFG3L2 are known to cause neurological disorders^28,29,36 and A39 is a candidate risk gene for late onset Parkinson’s disease³⁷. In addition, as glutathione, iron, and mitochondrial dysregulation have all been implicated in neurological disorders, we then explored if this A39 protein regulation mechanism also occurs in the neuronal cells. We chose the human induced pluripotent stem cell (iPSC) KOLF2.1J cell-derived glutamatergic neurons utilizing Tet-On doxycycline (Dox)-inducible NGN2 expression (Fig. 4H)^38–40. The neuronal differentiation was robust (>95%) based on neuronal morphology and they were considered mature iNeurons on differentiation Day 21 (Fig. S3H). Western blotting for synaptic marker Synaptophysin confirmed synaptic development (Fig. 4I). We observed a time-dependent reduction of A39 protein level during neuronal differentiation (Fig. 4I), which is accompanied by a decline of the Fe-S cluster machinery HSCB (Fig. 4I), suggesting a coordinated regulation of mitochondrial glutathione and Fe-S cluster during differentiation. The AFG3L2 protein level remained unchanged (Fig. 4I). A39 reduction in neurons is likely due to reduced iron sensing that activates degradation, because BSO treatment in the mature iNeurons dramatically upregulated A39 protein level (Fig. 4J and 4K), which can be dampened by DFO treatment (Fig. 4K). The occurrence of A39 regulation in the neuronal cells might suggest a regulated neuronal glutathione compartmentalization in the context of brain disorders.

Discussion

Here, we reported a dual regulatory mechanism acting on a mitochondrial transporter protein, by both protein quality control and metabolic sensing, which directly controls mitochondrial glutathione uptake in response to iron homeostasis (Fig. 4L), a finding independently reported by Liu et al.⁴¹. The proteolytic regulation of A39 is specific to the mitochondrial m-AAA protease AFG3L2, and cannot be substituted by other mitochondrial proteases that also act around the inner mitochondrial membrane^42,43. A direct sensing of mitochondrial iron homeostasis inhibits A39 degradation, further enabling a specific transporter regulation and coordinated metabolism of two essential and reactive redox metabolites, glutathione and iron. We further went on to reveal this A39 regulation in the developing and mature neurons. This unexpected link between mitochondrial glutathione and m-AAA protease AFG3L2 might lead to unexplored pathophysiological mechanisms under a diverse clinical spectrum of AFG3L2 mutation-associated neurodegenerative disorders that range from ataxia to parkinsonism^28,36. This dual regulatory mechanism of SLC25A39 transporter, by AFG3L2 and iron, further advances our understandings of metabolic regulation through protein quality control and metabolite sensing, opening new avenues for exploring regulation of metabolic compartmentalization in cellular metabolism.

Limitations of the Study

We reported a dual protein regulation of SLC25A39 by the mitochondrial protease AFG3L2 and iron sensing, however, the exact mechanisms how SLC25A39 is targeted for degradation and how iron sensing inhibits this degradation are yet to be explored. And whether SLC25A39 regulation by AFG3L2 might be involved in pathology of AFG3L2-mutation associated neurological disorders needs to be investigated.

STAR Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Hongying Shen ( hongying.shen@yale.edu)

Materials availability

All reagents generated in this study are available from the lead contact without restriction.

Data and code availability

• The raw data for the proteomics have been deposited with the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository with the dataset identifier PXD047327 and is publicly available as of the date of publication.

• This paper does not report original codes.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell cultures

K562 cells (ATCC, CCL-243), HeLa-M cells (a subclone of HeLa, ATCC, CCL-2 from P. De Camilli laboratory), and HEK293T cells (ATCC, CRL-3216) were maintained in DMEM (Thermo Fisher, 11995) supplemented with 10% FBS (Sigma, F2442) and 100 U /ml Pen/Strep (Thermo Fisher, 15140122).

For human induced pluripotent stem cell (hiPSC)-derived neuronal experiments, we used an established human KOLF 2.1J iPSCs (Jackson Laboratory, JIPSC1000)⁴⁰ engineered to harbor a doxycycline-inducible NGN2 transgene expressed at the AAVS1 safe harbor locus^38,39,44. Cell maintenance and differentiation is following the previous literature³⁹. For maintenance, the iPSCs were seeded at a density of 0.3 M per well in a 6 well Geltrex coated plate (Thermo Fisher Scientific, A1413302) supplemented with 10 μM ROCK inhibitor (Selleck Chemicals, S1049) in E8 Flex medium (Thermo Fisher Scientific, A2858501). The media was replaced with fresh E8 Flex Medium without ROCK inhibitor within 24 hours and the cells were passaged every 2–3 days at 70–80% confluency with StemPro Accutase (Thermo Fisher Scientific, A1110501). To generate excitatory neurons (iNeurons), hiPSCs were induced by DMEM/F12 (Thermo Fisher Scientific, 11320033) containing 2 mg/ml of doxycycline (Sigma, D9891) for 2 days. These cells were then incubated with StemPro Accutase for 5 min, quenched with DMEM/F12 medium, and counted. The cell suspension was centrifuged at 300 g at room temperature for 5 min and the cell pellet was gently resuspended in 2 ml iNeuron medium prepared with the following reagents: BrainPhys medium (StemCell Technologies, 5790), B27 supplement (Thermo Fisher Scientific, 17504044), NT-3 (Peprotech, 450–03), BDNF (Peprotech, 450–02), laminin (Thermo Fisher Scientific, 23017015), and 2 mg/ml doxycycline. The induced cells were then seeded at a density of 2 M per well on a PLO-borate (Millipore Sigma, P3655; Thermo Fisher Scientific, 28341) and Geltrex-coated 6-well plate. The following day, 1 ml of iNeuron medium was added to the medium and a half medium change was performed every 2–3 days until respective time points required for the time course assays (Day 7, Day 14, and Day 21).

All the cells were tested for mycoplasma quarterly using Universal Mycoplasma Detection kit (ATCC, 30–1012K).

METHOD DETAILS

Plasmids and cloning

For single KO cells, the following sgRNAs were synthesized and cloned into the lentiCRISPR v2 vector (Addgene, 52961), and the resulting plasmids were used to generate the knockout lines. sgRNA sequences used for targeted integration can be found in Table S2. For the DKO cells, spCas9 guides were synthesized and cloned into either lentiCRISPR v2 vector (puromycin selection) and pXPR_BRD051 (hygromycin selection) vector, and the resulting plasmids were used to generate the single knockout (puromycin selection) and the double knockout (a combination of puromycin and hygromycin selection) lines. For the study of SLC25A39, Lentiviral plasmids were used to ectopically express human SLC25A39^WT, SLC25A39^C334S, SLC25A39^4CA, SLC25A39^C74A, SLC25A39^C78A SLC25A39^C88A, SLC25A39^C94A, SLC25A39^A11L1 and SLC25A39^A40L1 proteins with a C-terminal FLAG tag following a flexible linker. A lentiviral plasmid for the human SLC25A39^WT with a C-terminal FLAG tag without the linker was used in the Fig. S2A. To study AFG3L2, the lentiviral vector was used to express AFG3L2 with a C-terminal 3xFLAG tag. For rapid mitochondrial isolation experiments, the following plasmid was used: pMXs-3xHA-EGFP-OMP25 (Addgene, 83356).

Lentivirus production and transduction for cell line generation

Lentivirus was produced using HEK 293T cells. Specifically, 0.5 × 10⁶ HEK293T cells were plated in a 6 cm dish one day before induction. A mixture of 0.6 μg pMD2.G (Addgene, 12259), 0.9 μg psPAX2 (Addgene, 12260), and 1 μg lentiviral plasmids was mixed with 125 μl serum-free DMEM and 5 μl of Lipofectamine P3000 reagent (Invitrogen, 100022057). The mixture was then combined with 2.5 μl of Lipofectamine 3000 reagent (Invitrogen, 100022050) and 125 μl serum-free DMEM, and incubated for 15–30 min at room temperature before adding dropwise to the cells. Two days after induction, the spent media was collected and filtered through a 0.45 μm sterile filter, and the resulting virus was stored at −80 °C until use. For infection, 0.5×10⁶ cells were plated in 24-well plate in DMEM containing 10% FBS, 10 μg/ml polybrene, and 250–300 μl of virus-containing media. Spin infection was performed at 2000 g for 60 min at 37 °C. After overnight incubation, the infected cells were transferred to a 10 cm dish containing the fresh media supplemented with the corresponding antibiotic.

To generate the DKOs and corresponding controls, cells were first infected with plasmids bearing a hygromycin selection cassette, followed by selection with 0.25 mg/ml hygromycin (Roche, 10843555001) for one week. After, the cells were infected again with plasmids bearing a puromycin selection cassette followed by 2 μg/ml puromycin (Gibco, A1113803) selection for two days. Two cutting controls were used for the growth fitness defects caused by double-strand DNA breaks during the Cas9 cutting. For SLC25A39 and AFG3L2 double KO expressing various FLAG-tagged SLC25 fusion proteins, sgRNA targeting SLC25A39 cloned in the pLentiCRISPR V2-blast was used followed by 10 μg/ml blasticidin (Alfa Aesar, J61883.FPL) selection for 6 days clonal selection.

qPCR for mRNA level

The mRNAs were isolated from 0.5 million cells using RNeasy Mini kit (Qiagen, 74104); the cDNA was prepared using SuperScript III First-Strand Synthesis SuperMix for qRT-PCR kit (Invitrogen, 11752050). The qPCR experiments were performed on the QuantStudio 6 Flex Real-Time PCR Systems using the recommended TaqMan probes (See Key Resources Table).

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit Anti-SLC25A39	Proteintech	Cat#14963-1-AP
Rabbit Anti-VDAC	Cell Signaling Technology	Cat#4661T
Mouse Anti-GAPDH	Invitrogen	Cat#39-8600
Mouse Anti-SLC25A40	Abcam	Cat#ab69075
Rabbit Anti-AFG3L2	Proteintech	Cat#14631-1-AP
Mouse Anti-FLAG	GenScript	Cat#A00187
Mouse Anti-NDUFA9	Invitrogen	Cat#459100
Rabbit Anti-HSCB	Proteintech	Cat#15132-1-AP
Rabbit Anti-Synaptophysin 1	Synaptic Systems	Cat#101002
Rabbit Anti-IRP2	Cell Signaling Technology	Cat#37135S
Rabbit Anti-SLC25A37(Mitoferrin1)	Proteintech	Cat#26469-1-AP
Rabbit Anti-GLRX5	Sigma	Cat#HPA042465
Rabbit-Anti-ABCB7	Proteintech	Cat#11158-1-AP
Goat anti-Rabbit IgG (H+L) Secondary Antibody, HRP	Invitrogen	Cat#PI31460
Goat anti-Mouse IgG (H+L) Secondary Antibody, HRP	Invitrogen	Cat#PI31430
Chemicals, peptides, and recombinant proteins
L-Butathione Sulfoximine	Sigma	Cat#B2515
Cycloheximide	Sigma	Cat#C4859
Deferoxamine mesylate salt	Sigma	Cat#D9533
GSH-[glycine-13C2, 15N]	Sigma	Cat#683620
EZ-link Iodoacetyl-PEG2-Biotin	Thermo Scientific	Cat#21334
DYKDDDDK Peptide	GenScript	Cat#RP10586
Hygromycin B	Roche	Cat#10843555001
Puromycin Dihydrochloride	Gibco	Cat#A1113803
Blasticidin S hydrochloride	Alfa Aesar	Cat#J61883.FPL
Digitonin (5%)	Invitrogen	Cat#BN2006
Doxycycline	Sigma	Cat#D9891
Critical commercial assays
Pierce BCA protein assay kit	Thermo Scientific	Cat#23225
Silver Stain Plus Kit	Bio-Rad	Cat#1610449
Superscript™ III First-Strand Synthesis SuperMix for qRT-PCR	Invitrogen	Cat#11752050
Lipofectamine 3000 Transfection kit	Invitrogen	Cat#L3000008
RNeasy Mini kit	Qiagen	Cat#74104
Universal Mycoplasma Detection kit	ATCC	Cat#30-1012K
Deposited data
The raw data for the proteomics have been deposited with the ProteomeXchange Consortium via the PRIDE partner repository.	This paper	PXD047327
Experimental models: Cell lines
K562 cells	ATCC	CCL-243
HeLa	Pietro De Camilli lab	ATCC-CCL-2
HEK293T	ATCC	CRL-3216
Human iPSC KOLF 2.1J line	Jackson Laboratories	JIPSC1000
Oligonucleotides
spCas9 sgRNA guide sequences to generate the CRISPR KO cells	This paper	Table S2
Recombinant DNA
lentiCRISPR v2 vector	Addgene	Cat#52961
lentiCRISPR v2 - hygro	Broad institute	pXPR_BRD051
lentiCRISPR v2 - Blast	Addgene	Cat#83480
pMXs-3xHA-EGFP-OMP25	Addgene	Cat#83356
psPAX2	Addgene	Cat#12260
pMD2.G	Addgene	Cat#12259
SLC25A39-FLAG (linker) - hygro resistance	This paper	N/A
SLC25A39FLAG (no linker) - hygro resistance	This paper	N/A
SLC25A394CA-FLAG - hygro resistance	This paper	N/A
SLC25A39C334S-FLAG - hygro resistance	This paper	N/A
SLC25A39A11L1-FLAG - hygro resistance	This paper	N/A
SLC25A39A40L1-FLAG - hygro resistance	This paper	N/A
SLC25A39C74A-FLAG - hygro resistance	This paper	N/A
SLC25A39C78A-FLAG - hygro resistance	This paper	N/A
SLC25A39C88A-FLAG - hygro resistance	This paper	N/A
SLC25A39C94A-FLAG - hygro resistance	This paper	N/A
AFG3L2-3xFLAG - hygro resistance	This paper	N/A
Software and algorithms
GraphPad Prism 10.1.1	GraphPad Software	N/A
Xcalibur 4.1	Thermo Scientific	N/A
StepOne Software version 2.1	Applied Biosystems	N/A
FIJI (ImageJ2)	https://imagej.net/software/fiji/	N/A
Proteome Discoverer version 2.5	Thermo Scientific	N/A
Other
human SLC25A28 TaqMan probe	Thermo Fisher Scientific	Cat#4331182/Hs00945863_m1
human SLC25A39 TaqMan probe	Thermo Fisher Scientific	Cat#4331182/Hs00924971_g1
human ABCB7 TaqMan probe	Thermo Fisher Scientific	Cat#4331182/Hs00188776_m1
human AFG3L2 TaqMan probe	Thermo Fisher Scientific	Cat#4351372/Hs01064997_m1

Immunoblotting

For the western blotting experiments, samples were lysed with RIPA lysis buffer (Thermo Scientific, J60645.AK) supplemented with 1% SDS and protease inhibitor cocktail (Roche, 11836170001). SDS-PAGE was performed using NuPAGE 4–12% Bis-Tris gel (Invitrogen, NP0335BOX) and the blot was visualized for labeling by chemiluminescent on a Bio-Rad ChemiDoc MP imaging system. The protein content was measured using Pierce BCA protein assay kit (Thermo Scientific, 23225) for sample normalization. The following antibodies and concentration were used: Anti-SLC25A39 (Proteintech, 14963–1-AP, 1:300); Anti-VDAC (Cell Signaling Technology, 4661T, 1:1000); Anti-GAPDH (Invitrogen, 39–8600, 1:1000); Anti-SLC25A40 (Abcam, ab69075, 1:1000); Anti-AFG3L2 (Proteintech, 14631–1-AP, 1:1000); Anti-FLAG (GenScript, A00187, 1:1000); Anti-NDUFA9 (Invitrogen, 459100, 1:1000); Anti-HSCB (Proteintech, 15132–1-AP, 1:1000); Anti-GLRX5 (Sigma, HPA042465, 1:500); Anti-Synaptophysin 1 (Synaptic Systems,101002, 1:1000); Anti-IRP2 (Cell Signaling Technology, 37135S, 1:500); Anti-SLC25A37(Mitoferrin1) (Proteintech, 26469–1-AP, 1:1000); Anti-ABCB7 (Proteintech, 11158–1-AP, 1:1000); Goat anti-Rabbit IgG (H+L) Secondary Antibody, HRP (Invitrogen, PI31460, 1:5000 ); and Goat anti-Mouse IgG (H+L) Secondary Antibody, HRP ((Invitrogen, PI31430, 1:5000 ).

Protein degradation assay

The cells were pretreated as described in figure legends. For each sample, 2 M cells were incubated with 150 μg/ml CHX (cycloheximide, Sigma, C4859) in a six-well plate, which was collected at the indicated times. For degradation assay in BSO, cells were pre-treated with 1 mM BSO for 2d, and incubated with CHX together with 1mM BSO for the indicated times. Western blot was used to quantify the A39 protein level. The intensity of the A39 protein was measured and analyzed using Fiji⁴⁵.

Co-immunoprecipitation from the mitochondrial lysates

BSO-treated clonal SLC25A39 KO K562 cells expressing SLC25A39-FLAG and 3xHA-EGFP-OMP25 was selected. 150 million cells were washed with PBS and collected at 1,000 g for 2 min at 4 °C, suspended in 3 ml KPBS buffer (136 mM KCl, 10 mM KH₂PO₄, pH 7.25) containing protease inhibitor cocktail to yield a homogenous single cell suspension, and aliquots were individually lysed by 20 gentle strokes using hand-held Dounce tissue grinder with tight-fitting pestle B (VMR, Kontes). Cell lysates were cleared by centrifugation at 1,000 g for 2 min at 4 °C, and transferred to new EP tubes containing 300 μl pre-washed anti-HA magnetic beads (Pierce, 88837). After an end-to-end rotator incubation for 5 min in the cold room, the isolated mitochondria were washed with KPBS buffer for 3 times, and lysed with 600 μl of TBS buffer (pH=7.5) containing 0.5% digitonin (w/v) in wet ice for 15 min. The lysate was transferred to a new 1.5 ml tube and spun at 10,000 g for 5 min at 4 °C. The supernatant was transferred to a new 1.5 ml EP tube containing 70 μl pre-washed anti-FLAG resin (GenScript, L00432). After an end-to-end rotator incubation for 40 min at room temperature, the resins were washed with 0.04% digitonin (w/v) TBS buffer three times. A39-FLAG protein was eluted from the FLAG resin with 60 μl of TBS buffer containing 300μg/ml FLAG peptide (GenScript, RP10586) after a 20-min incubation at room temperature with pipetting, and collected by centrifugation using spin cups (Thermo Scientific, 69702) at 2,000 g for 2 min at 4 °C.

Silver staining

The sliver staining is following the product protocol in a glass container. The gel was immediately fixed using 40% methanol/10% acetic acid (v/v) for 60 min. Then the gel was transferred to a solution containing 10% ethanol/5% acetic acid (v/v) for 30 min, immersed in a 1x oxidizer (Bio-Rad, 1610444) for 10 min, washed using deionized water several times (10 min each) until all the yellow color was removed from the gel, and stained by the silver reagent (Bio-Rad, 1610445) for 30 min. After that, the gel was quickly washed using deionized water to remove excess silver reagent. Lastly, the developer solution (Bio-Rad, 1610450) was used to develop the gel several times until the protein bands became clearly visible. To stop the staining process, 5% acetic acid (v/v) was used.

Sequence analysis and structural modeling

Primary amino acid sequences were aligned using MUSCLE (MUltiple Sequence Comparison by Log-Expectation). A structure model of human SLC25A39 was generated using the AlphaFold^16,46.

Mitochondrial isolation and uptake assay

A rapid mitochondria immunoisolation method was following the protocols^24,25. Approximately 30 million cell lines expressing mito IP constructs were washed with PBS, collected at 1,000 g for 2 min at 4 °C, suspended in 1 ml KPBS buffer (136 mM KCl, 10 mM KH₂PO₄, pH 7.25) to form homogenous single-cell suspension, and lysed by 20 gentle strokes in a 2 ml hand-held Dounce tissue grinder with tight-fitting pestle B (VMR, Kontes). Cell lysates were cleared by centrifugation at 1,000 g for 2 min at 4 °C, and transferred to a new 1.5 ml EP tube containing 80 μl pre-washed anti-HA magnetic beads (Pierce, 88837). After an end-to-end rotator incubation for 3.5 min in the cold room, the isolated mitochondria were washed with KPBS buffer for 3 times. For the metabolite profiling experiment, mitochondrial metabolites were extracted with 50 μl acetonitrile-methanol-water (27:9:1 v/v/v). For the western blotting experiment, mitochondrial proteins were lysed with 100 μl RIPA lysis buffer containing 1% SDS and protease inhibitor cocktail (Roche, 11836170001).

For the mitochondrial uptake assay, the isolated mitochondria on beads were incubated with 5 mM labeled GSH, GSH-[glycine-13C2, 15N] (Sigma, 683620), in 100 μl assay buffer (110 mM sucrose, 20 mM HEPES pH7.4, 10 mM KH₂PO₄, 3 mM MgCl₂, 1 mM EGTA, 0.1% BSA). The uptake assay was performed at room temperature by horizontal shaking at 170 rpm and manually mixed one time. At the indicated time point, the incubated mitochondrion was washed three times with KPBS buffer, and metabolites were extracted with 50 μl acetonitrile-methanol-water (27:9:1 v/v/v) for immediate LC-MS analysis.

Mass spectrometry for mitochondrial GSH level

LC/MS-based analyses were performed on a Q Exactive plus benchtop orbitrap mass spectrometer equipped with an Ion Max source and a HESI II probe, which was coupled to a Vanquish UHPLC system. The polar metabolites from mitochondria were analyzed on a SeQuant ZIC-pHILIC polymeric 5 μm, 150×2.1 mm column (EMD-Millipore 150460). Mobile phase A was consisted of 20 mM ammonium carbonate in water at pH 9.6 (adjusted with ammonium hydroxide), and mobile phase B consisted of 100% acetonitrile. The column temperature was held at 27 °C with the injection volume of 5 μl, and an autosampler temperature of 4 °C. The LC conditions were set as following: the flow rate was 0.15 ml/min with a gradient of 80% B at 0 min, 80% B at 0.5 min, 20% B at 20.5 min, 20% B at 21.3 min, 80% B at 21.5 min and till 29 min. The mass data were acquired in the polarity switching mode with full scan mode in a range of 70–1000 m/z, with the resolution at 70,000, the AGC target at 1e⁶, the maximum injection time at 80 ms, the sheath gas flow at 50 units, the auxiliary gas flow at 10 units, the sweep gas flow at 2 units, the spray voltage at 2.5 kV, the capillary temperature at 310°C, and the auxiliary gas heater temperature at 370°C. Xcalibur 4.1 (Thermo Fisher Scientific) was used to pick peaks and integrate peak intensity.

Mass spectrometry for SLC25A39-FLAG immunoprecipitants

The co-immunoprecitants were loaded in a SDS-PAGE gel and run for a short time to remove the elution buffer composition. Cut gel slices were further cut into small pieces, fixed with 1 ml of methanol-water-acetic acid (45:45:10 v/v/v) for 15 min on a tilt-table, and then washed three times with 1 ml water for 5 min. The bands were washed with 1 ml 100 mM NH₄HCO₃ (ammonium bicarbonate, ABC) 50% acetonitrile (ACN) for 20 min, then reduced by the addition of 175 μl 4.5 mM dithiothreitol (DTT) in 100 mM ABC and incubated at 37 °C for 20 min. The DTT solution was removed, and the samples were cooled to room temperature. The samples were alkylated by the addition of 175 μl 10 mM iodoacetamide (IAN) in 100 mM ABC and incubation at room temperature in the dark for 20 min. The IAN solution was removed, and the gels were washed with 1 ml 50% ACN/100 mM ABC twice for 10 min, with 1 ml 50% ACN/25 mM ABC twice for 10 min, then briefly dried by SpeedVac. Digestion was performed by adding 1 gel volume of 25 mM ABC containing 2.5 ng/μl of digestion grade trypsin (Promega, V5111) and incubating at 37 °C for 16 hours. The supernatants containing the tryptic peptides were transferred to new Eppendorf tubes. The residual peptides in the gel bands were extracted with 500 μl 80% ACN/0.1% trifluoroacetic acid (TFA) for 15 min and combined with the original digests, then dried in a SpeedVac. The peptides were dissolved in 30 μl MS loading buffer (2% acetonitrile, 0.2% trifluoroacetic acid), with 5 μl injected for LC-MS/MS analysis.

LC-MS/MS analysis was performed on a Thermo Scientific Q Exactive Plus equipped with a Waters nanoAcquity UPLC system utilizing a binary solvent system (A: 100% water, 0.1% formic acid; B: 100% acetonitrile, 0.1% formic acid). Trapping was performed at 5 μl/min, 99.5% Buffer A for 3 min using a Waters Symmetry^® C18 180μm x 20mm trap column. Peptides were separated using an ACQUITY UPLC PST (BEH) C18 nanoACQUITY Column (1.7 μm, 75 μm x 250 mm) maintained at 37 °C and eluted at 300 nl/min with the following gradient: 3% buffer B at initial conditions; 5% B at 1 minute; 25% B at 90 min; 50% B at 110 min; 90% B at 115 min; 90% B at 120 min; return to initial conditions at 125 min. MS was acquired in profile mode over the 300–1,700 m/z range using 1 microscan, 70,000 resolution, AGC target of 3E6, and a maximum injection time of 45 ms. Data dependent MS/MS were acquired in centroid mode on the top 20 precursors per MS scan using 1 microscan, 17,500 resolution, AGC target of 1E5, maximum injection time of 100 ms, and an isolation window of 1.7 m/z. Precursors were fragmented by HCD activation with a collision energy of 28%. MS/MS were collected on species with an intensity threshold of 1E4, charge states of 2–6, and peptide match preferred. Dynamic exclusion was set to 20 seconds.

Data was analyzed using Proteome Discoverer (version 2.5) (Thermo Scientific) software and searched in-house using the Mascot algorithm (version 2.8.0) (Matrix Science). The data was searched against the Swissprotein database with taxonomy restricted to Homo sapiens (20,405 sequences) along with a custom database containing the hSLC25A39-FLAG sequence. Search parameters used were trypsin digestion with up to 2 missed cleavages; peptide mass tolerance of 10 ppm; MS/MS fragment tolerance of 0.02 Da; fixed modification of carbamidomethyl cysteine, and variable modifications of methionine oxidation and deamidation on asparagine and glutamine. Normal and decoy database searches were performed, with the confidence level set to 95% (p<0.05). Scaffold v5.1.2 (Proteome Software Inc.) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability. Protein identifications were accepted if they could be established at greater than 99.0% probability (assigned by the Protein Prophet algorithm) and contained at least 2 identified peptides.

Cysteine labeling and mobility shift assay

Approximately 25 million clonal SLC25A39 and AFG3L2 double CRISPR KO K562 cells expressing SLC25A39-FLAG in each condition were washed with PBS, collected at 1,000 g for 2 min at 4 °C, suspended in 1.5 ml hypotonic buffer (100 mM sucrose, 1mM EGTA, 20 mM HEPES, pH 7.4) containing protease inhibitor cocktail to form homogenous single-cell suspension, and kept on wet ice for 10 min, and then lysed by 30 gentle strokes in the 2 ml hand-held Dounce tissue grinder with tight-fitting pestle B (VMR, Kontes). Cell lysates were cleared by centrifugation at 1,000 g for 10 min at 4 °C. The supernatant was transferred to a new 1.5 ml EP tube and spun at 10,000 g for 10 min at 4 °C. Afterward, the pellet was lysed with 150 μl of 50 mM Tris-HCl (pH 8.0) reaction buffer containing 1% digitonin (w/v) wet ice for 15 min. Mitochondrial lysates were cleared by centrifugation at 10,000 g for 5 min at 4 °C. For Cysteine labeling, 50 μl of mitochondrial lysates were reacted with the 50 μl of freshly prepared reaction buffer (4 mM EZ-link Iodoacetyl-PEG2-Biotin (Thermos scientific, 21334) in 50 mM Tris-HCl pH 8.0) and 1mM TCEP. The reaction was conducted in the dark at room temperature for 90 min, and quenched by an aliquot of 4x Laemmli Sample buffer (Bio-Rad, 1610747). To assay mobility of the labeled A39-FLAG, the samples were separated by SDS-PAGE using NuPAGE 4–12% Bis-Tris gel (Invitrogen, NP0335BOX) at 75 V for 6 hours, and anti-FLAG antibody was used to detect the both labeled and unlabeled A39-FLAG.

QUANTIFICATION AND STATISTICAL ANALYSIS

For metabolite profiling experiments, data were shown as mean ± s.d., n≥3 biologically independent samples. Statistical significance was calculated using two-tailed t-test. Significance level was indicated as *** p < 0.001, ** p < 0.01, * p < 0.05 and n.s. p > 0.05. All other important experiments were validated at least two times. Statistical analysis was performed using GraphPad Prism (8.4.3a), Microsoft Excel (Microsoft Office 365), or as reported by the relevant computational tools.

Supplementary Material

1

Table S1. A39-FLAG SDS-PAGE gel band mass spectrometry results, related to Figure 2.

Highlights.

• SLC25A39 is degraded by m-AAA protease AFG3L2 through the matrix loop 1

• SLC25A39 senses mito iron using the matrix Cys residues, inhibiting degradation

• Protein level of SLC25A39 controls mito glutathione uptake and homeostasis

• SLC25A39 regulation occurs in neurons