FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313506871> ?p ?o ?g. }

• W4313506871 abstract "Article Figures and data Abstract Editor's evaluation eLife digest Introduction Results Discussion Materials and methods Appendix 1 Data availability References Decision letter Author response Article and author information Metrics Abstract The tumor suppressor gene PTEN is the second most commonly deleted gene in cancer. Such deletions often include portions of the chromosome 10q23 locus beyond the bounds of PTEN itself, which frequently disrupts adjacent genes. Coincidental loss of PTEN-adjacent genes might impose vulnerabilities that could either affect patient outcome basally or be exploited therapeutically. Here, we describe how the loss of ATAD1, which is adjacent to and frequently co-deleted with PTEN, predisposes cancer cells to apoptosis triggered by proteasome dysfunction and correlates with improved survival in cancer patients. ATAD1 directly and specifically extracts the pro-apoptotic protein BIM from mitochondria to inactivate it. Cultured cells and mouse xenografts lacking ATAD1 are hypersensitive to clinically used proteasome inhibitors, which activate BIM and trigger apoptosis. This work furthers our understanding of mitochondrial protein homeostasis and could lead to new therapeutic options for the hundreds of thousands of cancer patients who have tumors with chromosome 10q23 deletion. Editor's evaluation The authors identify co-deletion of the mitochondrial AAA+ ATPase ATAD1 with the tumor suppressor PTEN as a factor modifying cancer prognosis, based on a new mechanism of increasing sensitivity to proteotoxic stress induced by proteasome inhibition. The authors also identify the mitochondrial E3 ubiquitin ligase MARCH5 as a gene whose deletion is synthetically lethal with ATAD1. These findings suggest that the use of proteasome-targeting agents may be useful in patients with tumors dually deleted for ATAD1 and PTEN. The study is based on convincing evidence and makes an innovative contribution to the understanding of the biology of tumors with 10q23 deletions. https://doi.org/10.7554/eLife.82860.sa0 Decision letter Reviews on Sciety eLife's review process eLife digest Cancer cells have often lost genetic sequences that control when and how cell division takes place. Deleting these genes, however, is not an exact art, and neighboring sequences regularly get removed in the process. For example, the loss of the tumor suppressor gene PTEN, the second most deleted gene in cancer, frequently involves the removal of the nearby ATAD1 gene. While hundreds of thousands of human tumors completely lack ATAD1, individuals born without a functional version of this gene do not survive past early childhood. How can tumor cells cope without ATAD1 – and could these coping strategies become the target for new therapies? Winter et al. aimed to answer these questions by examining a variety of cancer cells lacking ATAD1 in the laboratory. Under normal circumstances, the enzyme that this gene codes for sits at the surface of mitochondria, the cellular compartments essential for energy production. There, it extracts any faulty, defective proteins that may otherwise cause havoc and endanger mitochondrial health. Experiments revealed that without ATAD1, cancer cells started to rely more heavily on an alternative mechanism to remove harmful proteins: the process centers on MARCH5, an enzyme which tags molecules that require removal so the cell can recycle them. Drugs that block the pathway involving MARCH5 already exist, but they have so far been employed to treat other types of tumors. Winter et al. showed that using these compounds led to the death of cancerous ATAD1-deficient cells, including in human tumors grown in mice. Overall, this work demonstrates that cancer cells which have lost ATAD1 become more vulnerable to disruptions in the protein removal pathway mediated by MARCH5, including via already existing drugs. If confirmed by further translational work, these findings could have important clinical impact given how frequently PTEN and ATAD1 are lost together in cancer. Introduction The tumor suppressor gene PTEN is deleted in more than 33% of metastatic prostate tumors and nearly 10% of glioblastoma multiforme and melanoma (Worby and Dixon, 2014). These deletions are imprecise and typically include PTEN as well as neighboring genes at the 10q23 locus. Prior research has demonstrated that the ‘collateral’ deletion of neighboring genes along with tumor suppressor genes can generate context-dependent vulnerabilities specific to tumor cells (Muller et al., 2015; Kryukov et al., 2016; Mavrakis et al., 2016). In some cases, such vulnerabilities can be exploited in a way that is toxic to mutant cancer cells but not to genetically intact host cells, which represents a therapeutic opportunity. Whether genomic deletions involving PTEN generate targetable vulnerabilities through the loss of neighboring genes is unknown, but would have relevance to a significant proportion of cancer patients given the frequency at which these deletions occur. Only 40 kb upstream of PTEN is ATAD1, which encodes a AAA+ ATPase involved in protein homeostasis on the outer mitochondrial membrane (OMM) (Chen et al., 2014; Okreglak and Walter, 2014; Nakai et al., 1993; Wang and Walter, 2020; Zhang et al., 2011). ATAD1 hydrolyzes ATP to directly remove substrate proteins from the OMM (Wang and Walter, 2020; Wang et al., 2022). ATAD1 appears particularly suited to extract tail-anchored (TA) proteins, which harbor a C-terminal, single-pass transmembrane domain. In contexts unrelated to cancer, it has been shown that the absence of ATAD1 leads to the accumulation of TA proteins on the OMM and significant mitochondrial dysfunction (Chen et al., 2014). This housekeeping role of ATAD1 is important for cellular health, as evidenced by the findings that ATAD1 is essential for life in mammals and has been conserved over the 1 billion years of evolution separating yeast and humans (Zhang et al., 2011; Ahrens-Nicklas et al., 2017). Here, we describe how the collateral deletion of ATAD1 along with PTEN sensitizes cells to apoptosis induced by dysfunction of the ubiquitin proteasome system. Results ATAD1 and PTEN are co-deleted in many human cancers Because the PTEN and ATAD1 genes are adjacent on human Chr10q23.31 (Figure 1A; Poluri and Audet-Walsh, 2018), we assessed whether ATAD1 is co-deleted with PTEN using immunohistochemistry on prostate adenocarcinoma tumors (Chung et al., 2019). We analyzed tumors that were PTEN-null by targeted sequencing, along with PTEN-wild-type (WT) controls. ATAD1 protein was undetectable in 21 of the 37 PTEN-null tumors analyzed, but was present in all 15 PTEN-WT control tumors (Figure 1—figure supplement 1A-C). Analysis of genomic data from The Cancer Genome Atlas corroborated these protein-level findings, as the majority of tumors harboring deep deletions in PTEN also had deep deletions in ATAD1 (Figure 1B). Importantly, ATAD1 is almost never deleted in the absence of PTEN deletion (Figure 1B), nor does it feature recurrent inactivating point mutations (Figure 1—figure supplement 1D,E), which argues that ATAD1 is not a tumor suppressor. Therefore, we hypothesize that ATAD1 deletion is simply a ‘hitchhiker’ with the oncogenic driver deletion of PTEN. These deletions most frequently span the 0.5 Mb surrounding PTEN, leading to collateral deletion of KLLN and RNLS in addition to ATAD1 (Figure 1—figure supplement 2). Altogether, ATAD1 is deleted at a high frequency across many tumor types, including in more than 25% of prostate cancer, 11% of melanoma, 7% of glioblastoma, and 4% of gastric adenocarcinoma (Figure 1C). Given the established role of ATAD1 in mitochondrial protein homeostasis, we hypothesized that a hitchhiker deletion of ATAD1 might confer unique vulnerabilities on tumors (Muller et al., 2015; Kryukov et al., 2016; Mavrakis et al., 2016; Muller et al., 2012). Figure 1 with 4 supplements see all Download asset Open asset ATAD1 is co-deleted with PTEN in cancer and its loss confers synthetic lethal vulnerabilities. (A) Schematic of PTEN and ATAD1 loci. (B) Oncoprint plots from three TCGA studies of cancer. ATAD1 and PTEN alteration frequencies are shown, with blue bars indicating deep deletions. (C) Frequency of ATAD1 deep deletions across various cancer types; data from cBioPortal. (D) CRISPR screen design for wild-type (WT) and ATAD1∆ Jurkat cells. (E) Jurkat CRISPR screen results; each point represents one gene. CRISPR score (CS) values were calculated by taking the average log2 fold-change in relative abundance of all sgRNAs targeting a given gene over 14 population doublings. WT CS values are shown on the y-axis. The CS values per gene for each of the two ATAD1∆ clones were averaged and are plotted on the x-axis. The top 10 genes that were differentially essential between WT and ATAD1∆ are labeled in blue, with MARCH5 labeled in red. (F) CRISPR screen design for HGC27 cells (Chr10q23 deletion, ATAD1-null) comparing gene essentiality in ATAD1 complemented cells or empty vector (EV) (ATAD1-null) control. (G) HGC27 CRISPR screen results; CS values are as described for (E). The x-axis depicts CS for the ATAD1-null condition of EV-transduced cells, and the y-axis depicts CS for the ATAD1-complemented (+ATAD1) condition. Labels are as described for (E). Figure 1—source data 1 Source data used to make Figure 1. https://cdn.elifesciences.org/articles/82860/elife-82860-fig1-data1-v2.zip Download elife-82860-fig1-data1-v2.zip ATAD1 is synthetic lethal with MARCH5 We conducted genome-wide CRISPR knockout screens to identify genes that are selectively essential in ATAD1∆ cells. Such genes represent pathways whose inhibition could be selectively toxic to ATAD1-deficient tumor cells in a patient. We generated two ATAD1∆ clones in the PTEN-null Jurkat T-cell acute lymphoblastic lymphoma cell line using transient expression of Cas9 and one of two sgRNAs targeting distinct exons of ATAD1 (Figure 1—figure supplement 3A and B). Jurkat cells were chosen as an experimentally tractable system that has been validated in genetic screening and as a cell line with background PTEN deficiency. ATAD1 deletion did not affect basal proliferation rate (Figure 1—figure supplement 3C). We conducted parallel screens on the WT Jurkat parental cell line and each of the two ATAD1∆ clonal cell lines, the comparison of which enabled us to minimize idiosyncrasies inherent to clonal cell lines (Figure 1D). A CRISPR score (CS) is assigned to each gene, and represents the mean log2 fold-change in relative abundance of sgRNAs targeting that gene. CS values for the two ATAD1∆ clones were averaged and compared against those of the WT cells (Figure 1E). Genes that are selectively essential in the ATAD1∆ background represent ATAD1 synthetic lethal candidate genes. As expected, the differential CRISPR score (dCS) values for each ATAD1∆ clone vs. WT significantly correlated with each other (p=2.16 × 10–16; Figure 1—figure supplement 3D). The top 10 candidates for ATAD1 synthetic lethality include five genes that encode mitochondrial proteins (MARCH5, TAZ, MTCH1, TOP3A, DNM1L) and two components of the ubiquitin proteasome system (MARCH5, PSMC6), which are both processes with clear relevance to the known functions of ATAD1 (Calvo et al., 2016; Figure 1E). MARCH5 (also known as MITOL) was a particularly interesting hit, since it is an E3 ubiquitin ligase that promotes protein degradation on the OMM (Nakamura et al., 2006). Many properties of cell lines can affect their particular genetic dependencies, including tissue of origin, driver, and passenger mutations, and even the media in which they grow (Hart et al., 2015; Rossiter et al., 2021). Therefore, we conducted an additional genetic screen in a different cellular context to gain a broader perspective on how ATAD1 deficiency creates synthetic lethal vulnerabilities. Rather than screening another pair of ATAD1-WT and engineered knockout cells, we sought to use cells that naturally have the Chr10q23 deletion seen in human tumors. Given the lack of available prostate cancer cell lines that harbor Del(10q23), we used a gastric adenocarcinoma cell line, HGC27, which is ATAD1 and PTEN deficient. ATAD1 is deleted in 4.1% of gastric cancer, a disease that causes nearly 800,000 deaths worldwide each year (Ferlay et al., 2019), therefore, we estimate that 32,000 patients die every year from ATAD1-deficient gastric cancer. This ranks second only to prostate cancer in terms of the number of deaths attributable to ATAD1-deficient tumors (Figure 1—figure supplement 4). We made ATAD1-proficient and -deficient HGC27 lines by transducing with lentiviral ATAD1-FLAG or empty vector (EV), and subsequently conducted genome-wide CRISPR screens. In this case, genes that are selectively essential for EV cells (ATAD1-deficient) represent putative ATAD1 synthetic lethal candidate genes (Figure 1F). The top candidate for synthetic lethality with ATAD1 was MARCH5 (Figure 1G), a gene also identified in the Jurkat screen described above. Deletion of MARCH5 had a negligible effect on fitness in ATAD1-proficient HGC27 cells, but was an essential gene (by MAGeCK) in the EV cells (ATAD1-deficient). A second OMM-localized ubiquitin E3 ligase, MUL1, was also a top hit for synthetic lethality with ATAD1. In summary, our two complementary and unbiased genetic screens indicate that dysfunction of the ubiquitin proteasome system is preferentially lethal to cells lacking ATAD1. We were particularly intrigued by the interaction of ATAD1 and MARCH5 for two main reasons. First, the same synthetic lethal interaction between ATAD1 and MARCH5 emerged as the top hit of CRISPR screens using two vastly different cellular contexts. Second, MARCH5 encodes a ubiquitin E3 ligase that ubiquitinates OMM proteins to trigger their extraction by p97/VCP and subsequent degradation by the proteasome (Nakamura et al., 2006; Cherok et al., 2017). MARCH5/p97 and ATAD1 mediate two parallel pathways by which OMM proteins are removed from mitochondria. Hence, it was intuitive that ATAD1 and MARCH5 could be synthetic lethal, given that they both contribute to protein homeostasis on the OMM, and synthetic lethal interactions classically involve two redundant pathways. An imbalance of BCL2 family proteins underlies the synthetic lethality of ATAD1 and MARCH5 It has recently become clear that the key function of MARCH5 is to suppress apoptosis (Djajawi et al., 2020; Haschka et al., 2020; Subramanian et al., 2016; Arai et al., 2020). Apoptosis is regulated by OMM-localized BCL2 family proteins and requires the permeabilization of the OMM by BAX/BAK (Kale et al., 2018). Pro-survival proteins such as MCL1 bind to and inhibit BAX/BAK to prevent inappropriate cell death (Greaves et al., 2019). A variety of stressors activate BH3-only proteins (e.g. BIM), which trigger apoptosis by binding and inhibiting pro-survival proteins like MCL1 and in some cases by directly activating BAX/BAK (Letai, 2017). BH3-only proteins serve as sentinels for cellular stress and, upon activation, initiate mitochondrial outer membrane permeabilization (Bhatt et al., 2020; Llambi et al., 2011). MARCH5 acts as a ‘guardian’ of MCL1 through an incompletely understood mechanism that involves the degradation of the pro-apoptotic BH3-only proteins BIM and/or NOXA (Kale et al., 2018; Letai, 2017; Llambi et al., 2011; Czabotar et al., 2014; Lin et al., 2022). We hypothesized that ATAD1 antagonizes these OMM-localized pro-apoptotic factors in parallel to MARCH5, such that simultaneous loss of both ATAD1 and MARCH5 leads to a lethal accumulation of pro-apoptotic proteins on the OMM. Indeed, BIM structurally resembles known substrates of ATAD1/Msp1 in that it is TA, has an intrinsically disordered region N-terminal to the transmembrane domain, and has basic residues at the extreme C-terminus (Castanzo et al., 2020; Li et al., 2019). Consistent with this hypothesis, the abundance of BIMEL (the predominant isoform of BIM) was increased in ATAD1∆ cells (Figure 2A). BIMEL can also be inactivated by phosphorylation by cytosolic kinases such as ERK. Deletion of ATAD1 decreased BIMEL phosphorylation, as assessed by decreased mobility in SDS-PAGE (Figure 2B), and with phospho-specific antibodies for residues Ser69 and Ser77 (Figure 2—figure supplement 1). While ATAD1 activity promoted the inhibitory phosphorylation of BIM at Ser69 and Ser77, it did not affect BIM phosphorylation at Thr112, a phosphorylation site that potentiates the pro-death activity of BIM (Figure 2—figure supplement 1). These data suggested that ATAD1 might act on BIM to promote its degradation and inhibitory phosphorylation. Figure 2 with 9 supplements see all Download asset Open asset ATAD1/MARCH5 synthetic lethality is partially mediated by BIM, which is a novel ATAD1 substrate. (A) Western blot of Jurkat cell lines, with quantification of BIMEL levels normalized to alpha-tubulin; one sample t and Wilcoxon test. (B) Western blot of whole cell lysates from wild-type (WT) or ATAD1∆ Jurkat cells stably expressing sgNT or sgBIM with Cas9-T2A-GFP. Lysates were mock treated or treated with lambda phosphatase (λ PPase) and analyzed by PhosTag/SDS-PAGE. (C) Western blots of Jurkat cell lines stably expressing Cas9-T2A-GFP with sgNT or sgBIM, harvested 4 days after transduction with additional indicated sgRNAs. (D) Viability of Jurkat cells after deletion of MARCH5, using different genetic backgrounds. Viability at 4 days post-transduction was normalized to that of cells transduced with sgRNA targeting AAVS1. Data analyzed by two-way ANOVA with Tukey’s multiple comparisons. (E) Viability of Jurkat cells stably expressing GFP or Myc-tagged MCL1 after deletion of MARCH5 and normalized as in (D). (F) Viability of Jurkat cells transduced with tetracycline-inducible GFP or GFP-BIMEL fusion; t=48 hr, normalized to viability of cells without doxycycline. (G) Western blot of cell lines as described in (D), treated with doxycycline (Dox) for 24 hr. (H) Schematic of in vitro extraction assay; ‘Ni2+ lipos’ indicates the use of nickel chelating headgroups of lipids in the liposomes; the star symbolizes a GST tag on the soluble chaperones, calmodulin (CaM) and SGTA, which are included to catch extracted TA substrates. (I) Extraction assay using His-ATAD1 and 3xFLAG-BIML (lanes 1–8, 13–20) or the negative control yeast TA protein, 3xFLAG-Fis1p (lanes 9–12); E193Q indicates the use of a catalytically inactive mutant of ATAD1; in samples shown in lanes 5–8, Ni2+ chelating lipids were omitted; in samples shown in lanes 17–20, ATP was omitted; ‘I’=Input, ‘FT’=flow-through, ‘W’=final wash, ‘E’=elution. Eluted fractions represent TA proteins extracted by ATAD1 and bound by GST-tagged chaperones; compare elution ‘E’ to input ‘I’. (J) Extraction assay as described in (H) but comparing different BH3-only proteins, BIM, BIK, and PUMA. (K) Quantification of assays as shown in (I), n=6 independent experiments. Figure 2—source data 1 Source data and uncropped blots used to make Figure 2. https://cdn.elifesciences.org/articles/82860/elife-82860-fig2-data1-v2.zip Download elife-82860-fig2-data1-v2.zip Since MARCH5 regulates the BIM/NOXA/MCL1 axis, we assessed abundance of these proteins in the context of single and double deletion of ATAD1 and MARCH5. Deletion of ATAD1 increased BIM levels, while deletion of MARCH5 increased NOXA levels (Figure 2C). Accordingly, deletion of both MARCH5 and ATAD1 increased the abundance of both NOXA and BIM, which work together to antagonize the pro-survival protein MCL1 (Figure 2C). We hypothesized that synergistic antagonism of MCL1 explained, at least in part, the synthetic lethality of ATAD1 and MARCH5. We tested whether BIM was required for the synthetic lethal interaction of ATAD1 and MARCH5. Deletion of MARCH5 caused a modest decrease in viability in WT Jurkat cells (Figure 2D). This effect was similar when we deleted MARCH5 in polyclonal BIM (BCL2L11) knockout cells (Figure 2D; generated by stably expressing Cas9 and sgRNA targeting BCL2L11). ATAD1∆ cells, however, were hypersensitive to MARCH5 deletion, as expected from our CRISPR screen results. Interestingly, BIM knockout partially rescued the hypersensitivity of ATAD1∆ cells to MARCH5 deletion (Figure 2D). These results validate the major finding from our CRISPR screens and demonstrate that BIM is partially responsible for the synthetic lethality of ATAD1 and MARCH5. If MCL1 antagonism underlies ATAD1/MARCH5 synthetic lethality, then we reasoned that overexpression of MCL1 might rescue this phenotype. Indeed, we found that the synthetic lethal interaction of ATAD1 and MARCH5 was suppressed in cells stably overexpressing MCL1 (Figure 2E). Again, we observed increased BIMEL and NOXA in ATAD1∆ cells, and deletion of MARCH5 led to a further increase in NOXA levels (Figure 2—figure supplement 2A). These effects were limited in cells overexpressing MCL1, which may reflect regulation of NOXA degradation by MCL1 binding. As a positive control for testing gene essentiality, we transduced cells with an sgRNA targeting the pan-essential gene PCNA, which is necessary for DNA replication (Girish and Sheltzer, 2020). Deletion of PCNA was similarly toxic to WT and ATAD1∆ cells, and was not rescued by BIM deletion nor by MCL1 overexpression (Figure 2—figure supplement 2B and C). These data indicate that ATAD1 protects cells from MARCH5 deletion specifically, rather than making them generally more resistant to perturbation of important genes. Altogether, our CRISPR screens identified a synthetic lethal interaction between ATAD1 and MARCH5, both of which enact protein extraction from the OMM. Loss of ATAD1 and MARCH5 antagonizes the anti-apoptotic function of MCL1 via increased abundance and activity of BIM and NOXA, which partially explains the synthetic lethality. We next asked if BIM was sufficient to trigger apoptosis preferentially in ATAD1∆ cells. We generated Jurkat cell lines expressing a tetracycline-inducible GFP-BIMEL fusion protein. The dose-dependent increase in GFP-BIM expression was equivalent in WT and ATAD1∆ cells (anti-GFP blot, Figure 2F). Maximal expression of GFP-BIMEL using 500 ng/mL doxycycline killed cells regardless of the presence or absence of ATAD1 (Figure 2G; Figure 2—figure supplement 3A). However, ATAD1∆ cells were hypersensitive to intermediate expression of ectopic BIMEL, as measured by cell viability assays (Figure 2G), cleaved PARP immunoblots (Figure 2F), and live cell imaging using the Incucyte platform (Figure 2—figure supplement 3A). Thus, endogenous ATAD1 protects against BIM but ATAD1 can be overwhelmed with sufficiently high levels of BIM. These results demonstrate that BIM is sufficient to induce apoptosis preferentially in cells lacking ATAD1. We further studied how ATAD1 affected apoptotic priming (the propensity of a cell to undergo intrinsic apoptosis) using BH3 profiling. H4 glioma cells (Del(10q23); ATAD1-null; Figure 2—figure supplement 3B) were highly sensitive to BIM peptide, which is rescued by re-expression of ATAD1WT but not the catalytically dead ATAD1E193Q mutant (Figure 2—figure supplement 3C). ATAD1 therefore appears to suppress overall apoptotic priming, at least in this context, as measured by sensitivity to BIM BH3 peptide. ATAD1 directly and specifically extracts BIM from membranes We hypothesized that BIM might be a direct substrate of the ATAD1 dislocase, which could explain how ATAD1 suppresses BIM-induced apoptosis. Consistent with BIM being an ATAD1 substrate, GFP-BIMEL co-immunoprecipitated with FLAG-tagged ATAD1 in H4 cells (Figure 2—figure supplement 3D). Inversely, ATAD1-FLAG co-immunoprecipitated with endogenous BIM (Figure 2—figure supplement 3E). Thus, reciprocal co-immunoprecipitation argues that ATAD1 and BIM physically interact in cells. We further tested whether ATAD1 can directly extract BIM from a membrane using an in vitro system with purified components (Wohlever et al., 2017). We used BIML because it is more soluble than BIMEL but shares the key structural features that would likely mediate ATAD1 recognition, including the tail-anchor and juxtamembrane regions (Ley et al., 2005; Liu et al., 2019; Chi et al., 2020). We were unable to purify active, full-length ATAD1. Instead, we swapped the N-terminal transmembrane domain with a His6 tag, which anchored His6-ATAD1 to liposomes doped with phospholipids containing nickel-chelated headgroups (‘Ni Lipos’, Figure 2H). In this extraction assay, TA proteins that are extracted from liposomes by ATAD1 bind soluble GST-tagged chaperones (SGTA and calmodulin), which are purified on a glutathione column and detected by immunoblotting (Figure 2H; Wohlever et al., 2017). We validated the Ni-His anchoring strategy using full-length or truncated yeast Msp1 (the yeast homolog of ATAD1) and positive and negative control substrates (Figure 2—figure supplement 4A–C). His6-ATAD1 directly and efficiently extracted 3xFLAG-BIML from liposomes in this assay (Figure 2I; compare lanes with elution ‘E’ to input ‘I’; lanes 13–16). As expected, this activity was ATP-dependent (Figure 2I, lanes 17–20) and was abolished when we used the catalytically inactive mutant, ATAD1E193Q (Figure 2I, lanes 1–4). Omission of Ni-chelating lipids from the liposomes (‘Mito Lipos’, which cannot anchor His6-ATAD1; Figure 2I, lanes 5–8) prevented ATAD1 from extracting BIM, demonstrating that ATAD1 requires membrane anchoring for its dislocase activity. Importantly, ATAD1 did not extract yeast Fis1, consistent with previous reports that Fis1 is not an Msp1 substrate (Li et al., 2019; Figure 2I, lanes 9–12). Many BH3-only proteins share key structural features, including a tail-anchor, so we next asked whether ATAD1 could extract other members of this protein family (Wilfling et al., 2012). In addition to BIM, we tested BIK, PUMA, and NOXA, since these have been proposed to mediate apoptosis triggered by proteotoxic stress. NOXA did not incorporate into proteoliposomes, which precluded it from our assay, and is consistent with a report that it lacks a transmembrane domain (Andreu-Fernández et al., 2016). ATAD1 extracted BIM in an ATP-dependent manner, as expected, but failed to extract BIK or PUMA under the same conditions (Figure 2J and K). While it is not clear how ATAD1 distinguishes between these substrates, BIM differs from BIK and PUMA in that it has a positively charged C-terminus and an intrinsically disordered region N-terminal to the transmembrane domain, which are important features for substrate recognition by Msp1, the yeast homolog of ATAD1 (Figure 2—figure supplement 5; Castanzo et al., 2020; Li et al., 2019). Taken together, we demonstrate that ATAD1 and BIM physically and genetically interact in cells, and that ATAD1 can directly extract BIM – but not other, related, BH3-only proteins – from membranes in a reconstituted system. These data raise the question of what happens to BIM after it has been extracted by ATAD1. We transduced SW1088 cells, which are a Del(10q23) cell line suitable for imaging, with either EV or ATAD1-FLAG. These cells were then transduced with TetON(GFP-BIMEL∆BH3), in which four point mutations in the BH3 domain of BIM neutralize its pro-apoptotic activity to permit live cell imaging. We assessed GFP-BIMEL∆BH3 localization using live cell confocal microscopy in the presence and absence of ATAD1, using MitoTracker Red to label mitochondria. ATAD1 altered the localization of BIM under basal conditions, generating GFP-positive puncta that did not colocalize with mitochondria (Figure 2—figure supplement 6A). Since BIM is regulated by proteasomal degradation, we additionally treated cells with bortezomib, a proteasome inhibitor. Treatment with bortezomib exacerbated this phenotype and resulted in larger, brighter GFP-positive puncta only in ATAD1 expressing cells (Figure 2—figure supplement 6B and C). We saw the same phenomenon when we genetically labeled mitochondria with mCherry, ruling out the possibility that these GFP-BIM puncta are merely depolarized mitochondria that cannot accumulate Mitotracker dye (Figure 2—figure supplement 7). GFP-positive puncta also did not colocalize with lysosomes or peroxisomes (Figure 2—figure supplements 8 and 9). In addition to localization, BIM is regulated by inhibitory phosphorylation, which we had previously observed to be affected by ATAD1 status. Bortezomib treatment led to the accumulation of phosphorylated BIMEL in SW1088 cells expressing ATAD1, while BIMEL accumulated in an unphosphorylated state in SW1088 cells transduced with EV (Figure 3—figure supplement 1A). We next used the PC3 prostate cancer cell line, which is PTEN-null and has a partial deletion of ATAD1 (Figure 3—figure supplement 1B). In parallel to our findings with SW1088 cells, BIM accumulated in a phosphorylated state after treatment with bortezomib in PC3 cells with ATAD1 present (sgNT), but this phosphorylation was abrogated in PC3 cells with ATAD1 deleted (Figure 3—figure supplement 1C and D). Thus, in the context of proteasome inhibition, ATAD1 shifts the localization of BIM from mitochondria to cytoplasmic puncta and promotes inhibitory phosphorylation of BIM. Proteasome inhibition is preferentially toxic to ATAD1-deficient cells We next sought to pharmacologically exploit ATAD1 deficiency in relevant cancer models, drawing on our discovery of synthetic lethality with MARCH5. Since MARCH5 is a ubiquitin E3 ligase, we hypothesized that disrupting the ubiquitin proteasome system downstream of MARCH5 might also be preferentially toxic to cells lacking ATAD1. Consistent with this idea, proteasome inhibitors are known to increase BIM and NOXA abundance, thereby triggering apoptosis (Baou et al., 2010; Meller et al., 200" @default.
• W4313506871 created "2023-01-06" @default.
• W4313506871 date "2022-10-05" @default.
• W4313506871 modified "2023-09-28" @default.
• W4313506871 title "Decision letter: Collateral deletion of the mitochondrial AAA+ ATPase ATAD1 sensitizes cancer cells to proteasome dysfunction" @default.
• W4313506871 doi "https://doi.org/10.7554/elife.82860.sa1" @default.
• W4313506871 hasPublicationYear "2022" @default.
• W4313506871 type Work @default.
• W4313506871 citedByCount "0" @default.
• W4313506871 crossrefType "peer-review" @default.
• W4313506871 hasBestOaLocation W43135068711 @default.
• W4313506871 hasConcept C121608353 @default.
• W4313506871 hasConcept C15744967 @default.
• W4313506871 hasConcept C181199279 @default.
• W4313506871 hasConcept C185592680 @default.
• W4313506871 hasConcept C23265538 @default.
• W4313506871 hasConcept C27740335 @default.
• W4313506871 hasConcept C2993632694 @default.
• W4313506871 hasConcept C502942594 @default.
• W4313506871 hasConcept C54355233 @default.
• W4313506871 hasConcept C55493867 @default.
• W4313506871 hasConcept C73484699 @default.
• W4313506871 hasConcept C86803240 @default.
• W4313506871 hasConcept C95444343 @default.
• W4313506871 hasConceptScore W4313506871C121608353 @default.
• W4313506871 hasConceptScore W4313506871C15744967 @default.
• W4313506871 hasConceptScore W4313506871C181199279 @default.
• W4313506871 hasConceptScore W4313506871C185592680 @default.
• W4313506871 hasConceptScore W4313506871C23265538 @default.
• W4313506871 hasConceptScore W4313506871C27740335 @default.
• W4313506871 hasConceptScore W4313506871C2993632694 @default.
• W4313506871 hasConceptScore W4313506871C502942594 @default.
• W4313506871 hasConceptScore W4313506871C54355233 @default.
• W4313506871 hasConceptScore W4313506871C55493867 @default.
• W4313506871 hasConceptScore W4313506871C73484699 @default.
• W4313506871 hasConceptScore W4313506871C86803240 @default.
• W4313506871 hasConceptScore W4313506871C95444343 @default.
• W4313506871 hasLocation W43135068711 @default.
• W4313506871 hasOpenAccess W4313506871 @default.
• W4313506871 hasPrimaryLocation W43135068711 @default.
• W4313506871 hasRelatedWork W1907775554 @default.
• W4313506871 hasRelatedWork W1979573002 @default.
• W4313506871 hasRelatedWork W2005997790 @default.
• W4313506871 hasRelatedWork W2036842083 @default.
• W4313506871 hasRelatedWork W2079283433 @default.
• W4313506871 hasRelatedWork W2149451293 @default.
• W4313506871 hasRelatedWork W2262989315 @default.
• W4313506871 hasRelatedWork W2949168305 @default.
• W4313506871 hasRelatedWork W3147681563 @default.
• W4313506871 hasRelatedWork W4210585717 @default.
• W4313506871 isParatext "false" @default.
• W4313506871 isRetracted "false" @default.
• W4313506871 workType "peer-review" @default.