https://orcid.org/0000-0003-0284-7961
Figures
Abstract
Kaposi’s sarcoma-associated herpesvirus (KSHV) causes several malignancies in people with HIV including Kaposi’s sarcoma and primary effusion lymphoma (PEL). We have previously shown that PEL cell lines require myeloid cell leukemia-1 (MCL1) to inhibit apoptosis. MCL1 is an oncogene that is amplified in cancers and causes resistance to chemotherapy regimens. MCL1 is thus an attractive target for drug development. The emerging clinical relevance and therapeutic potential of MCL1 motivated us to study the roles of this oncogene in PEL in depth. Using a systems biology approach, we uncovered an unexpected genetic interaction between MCL1 and MARCHF5 indicating that they function in the same pathway. MARCHF5 is an E3 ubiquitin ligase best known for regulating mitochondrial homeostasis and antiviral signaling, but not apoptosis. We thus investigated how MCL1 and MARCHF5 cooperate to promote PEL cell survival. CRISPR knockout (KO) of MARCHF5 in PEL cell lines resulted in a significant increase in apoptosis despite the presence of MCL1. The anti-apoptotic function of MARCHF5 was dependent on its E3 ligase and dimerization activities. Loss of MARCHF5 or inhibition of the 26S proteasome furthermore stabilized the MCL1 antagonist NOXA without affecting levels of MCL1. Interestingly, NOXA KO provides a fitness advantage to PEL cells suggesting that NOXA is the pro-apoptotic signal that necessitates the anti-apoptotic activities of MCL1 and MARCHF5. Endogenous reciprocal co-immunoprecipitation experiments show that MARCHF5 and NOXA are found in the same protein complex. Finally, NOXA KO rescued the loss of cell viability due to MARCHF5 KO. Our findings thus provide the mechanistic link that underlies the genetic interaction between MCL1 and MARCHF5. We propose that MARCHF5 induces the degradation of the MCL1 antagonist NOXA thus reinforcing the pro-survival role of MCL1 in these tumor cells. This newly appreciated interaction of the MCL1 and MARCHF5 oncogenes may be useful to improve the design of combination therapies for KSHV malignancies.
Author summary
Kaposi’s sarcoma-associated herpesvirus (KSHV) is a tumor virus that causes the cancers Kaposi’s sarcoma and primary effusion lymphoma (PEL). We previously identified that PEL tumor cells require the oncogene MCL1 to protect cells from apoptosis. In this study, we uncovered that this function is reinforced by another oncogene, MARCHF5. We found that MARCHF5 is similarly required to protect from apoptosis by causing the turnover of the MCL1 antagonist NOXA. By alleviating the inhibitory effect of NOXA on MCL1, we propose that the oncogenes MCL1 and MARCHF5 cooperate in the same pathway to promote tumor cell survival. Understanding this functional reinforcement may be useful in the future to improve the design of combination therapies for KSHV cancers.
Citation: Viswanathan P, Bersonda JR, Gill J, Navarro A, Farrar AC, Caldarera T, et al. (2026) The mitochondrial ubiquitin ligase MARCHF5 cooperates with MCL1 to inhibit apoptosis in KSHV-transformed primary effusion lymphoma cell lines. PLoS Pathog 22(10): e1014665. https://doi.org/10.1371/journal.ppat.1014665
Editor: Michael Lagunoff, University of Washington, UNITED STATES OF AMERICA
Received: May 18, 2026; Accepted: September 23, 2026; Published: October 1, 2026
Copyright: © 2026 Viswanathan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting information files.
Funding: This work is supported by a Transition Career Development Award K22 CA241355 (PI: M.M.) from the National Cancer Institute of the National Institutes of Health (NIH). Additional support was provided by the Center for Microbial Pathogenesis and Host Inflammatory Responses program grant P30 GM145393 (PI: Dr. Mark Smeltzer) from the National Institute of General Medical Sciences (NIGMS) of the NIH and the Winthrop P. Rockefeller Cancer Institute at UAMS (PI: M.M.). J.G., A.N., and A.F. were supported by the Undergraduate Summer Research Fellowship from the Arkansas IDeA Network of Biomedical Research Excellence (INBRE) funded by P20 GM103429 (PI: Dr. Lawrence Cornett) from NIH/NIGMS. The authors declare no conflicts of interest. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Kaposi’s sarcoma-associated herpesvirus (KSHV) is a human gammaherpesvirus that establishes life-long latency primarily in B cells. While the majority of infections remain asymptomatic, KSHV-positive people with HIV (PWH) can develop several hematologic diseases, including the lymphoproliferative disorder multicentric Castleman disease, and the rare but aggressive primary effusion B cell lymphoma (PEL) [1,2]. PEL classically presents as effusions in body cavities, including the pleural, pericardial, and peritoneal spaces, although extra-cavitary cases have been reported [3,4]. PEL comprises 4–8% of non-Hodgkin lymphoma cases in PWH [5,6]. Currently, there is no standard therapy for PEL. Patients receive similar chemotherapy regimens as with aggressive non-Hodgkin lymphomas (cyclophosphamide, hydroxydaunomycin/doxorubicin, oncovine/vincristine, and prednisolone or CHOP) and anti-retroviral therapy (ART) [1]. Nevertheless, the median overall survival time is 6–22 months [3,5–10].
While mutations in primary PEL tumors have been reported, the majority of PEL cases do not have genetic alterations in the tumor suppressor genes TP53 or RB1, nor do they have MYC translocations frequently observed in other non-viral B cell lymphomas [2–4,11–14]. Instead, the major driver of PEL is oncogenic “addiction” or constitutive dependency on host genes that are transcriptionally or post-transcriptionally dysregulated by KSHV genes. In our previous genome-wide CRISPR-Cas9 knockout (KO) screens, we identified 210 PEL-specific oncogenic dependencies (PSODs) or addictions that are preferentially required by PEL cell lines but not by non-PEL cancer cell lines [15]. These PSODs function in viral persistence strategies that overlap with hallmarks of cancer: metabolic rewiring, transcriptional reprogramming, cell cycle progression, sustaining proliferative signaling, immune evasion, and inhibition of cell death. Because of their essential roles in tumor cell survival, PSODs are attractive therapeutic candidates to treat PEL.
At the top of this list was the anti-apoptotic protein myeloid cell leukemia-1 (MCL1), which ranked as the second most essential PSOD [15,16]. MCL1 is an oncogene that has copy number amplification in solid tumors and hematologic malignancies, including non-Hodgkin lymphoma [17]. Overexpression of MCL1 in different cancers is associated with disease severity and poor prognosis, underscoring the pro-survival role of this oncoprotein [18,19]. With the development of the first-in-class MCL1 inhibitor S63845 [20], new and more potent MCL1 inhibitors are now being evaluated in the clinic [21]. This growing attention to MCL1 as a chemotherapeutic target to treat different cancers initiated our interest in the role of this oncogene in PEL.
MCL1 is a member of the B cell lymphoma-2 (BCL2) protein family that functions to prevent intrinsic apoptosis. BCL2 proteins are anchored to the mitochondrial outer membrane (MOM) to sequester monomers of the cell death effectors BAX or BAK1 [22]. The binding of BCL2 proteins to BAX/BAK1 prevents the monomers from assembling into a pore complex and initiating the apoptosis cascade. During intrinsic apoptosis, pro-apoptotic BCL2 homology region 3 (BH3)-only proteins are upregulated and bind with high affinity to the BCL2 proteins to displace BAX/BAK1. The released BAX/BAK1 molecules oligomerize to form mitochondrial pores and trigger MOM permeabilization (MOMP) [23]. While some BH3-only proteins can bind to all BCL2 family members (e.g., BIM, truncated BID, and PUMA), others are more selective. For instance, NOXA specifically neutralizes MCL1, while HRK has high specificity to BCL-XL.
We previously showed that PEL primary tumors and cell lines highly express MCL1 [15,24]. PEL cell lines are only addicted to MCL1 and do not depend on the other pro-survival BCL2 genes [15]. Consistent with this finding, we and others demonstrated that inhibitors against MCL1, but not other BCL2 proteins, are highly effective in killing PEL cell lines [24–26]. Moreover, the KSHV latency-associated nuclear antigen (LANA) sequesters the MCL1 E3 ubiquitin ligase FWB7, thereby stabilizing the MCL1 protein and promoting tumor cell survival [26]. Together, these observations underscore the role of the oncogenic addiction to MCL1 in this viral lymphoma.
Given the importance of MCL1 in PEL, we sought to uncover how MCL1 is regulated. In this study, we used an unbiased systems biology approach to identify that another high-scoring PSOD, MARCHF5, cooperates with MCL1 function. We show that this mitochondrial E3 ubiquitin ligase targets the MCL1 antagonist NOXA for degradation and consequently reinforces MCL1 function in protecting tumor cells from apoptosis.
Results
MCL1 and MARCHF5 are co-essential oncogene addictions in PEL cells
To gain unbiased insights into functions of MCL1, we took a systems biology approach to identify the genetic interactions of MCL1. We reanalyzed publicly available data from the Cancer Dependency Map (DepMap) [27]. DepMap integrates data from CRISPR screens from >1000 human cancer cell lines that identify genes essential for cell fitness to create “co-essentiality” maps. Genes that function in the same pathway have positive interactions because their individual knockout have similar phenotypic effects (co-essential genes or synergists) regardless of context. On the other hand, genes that have opposing functions have antagonistic interactions because the knockout of one negates the phenotype caused by the knockout of the other (antagonists). Because co-essentiality mapping only measures phenotypes and is thus agnostic to prior knowledge of gene function, this approach can lead to the discovery of new gene functions and pathway relationships [27–30].
We examined the co-essentiality maps of MCL1 from DepMap. Consistent with its known role in preventing intrinsic apoptosis, MCL1 shows antagonistic interactions with several pro-apoptotic effectors, including PMAIP1 (encodes for NOXA), BAX, APAF1, and BAK1 (Fig 1A, S1 Table). Similarly, MCL1 exhibits co-essentiality with other members of the BCL2 family, BCL2L2 and BCL2 (ranks 3 and 11, respectively). However, the top co-essential gene with MCL1 was the mitochondrial E3 ubiquitin ligase MARCHF5, which had a correlation score that is twice as strong as compared to BCL2L2 or BCL2. In addition, the E2 ubiquitin ligase UBE2J2 ranked second, just before the BCL2 genes. Conversely, MCL1 was the top co-essential gene of MARCHF5, followed by UBE2J2 and another E2 ubiquitin ligase UBE2K (Fig 1B, S1 Table). This relationship is surprising because MARCHF5 is more established as a regulator of mitochondrial homeostasis and antiviral signaling [31–33]. Recent reports, however, have implicated MARCHF5 in the degradation of MCL1 itself or the MCL1-NOXA complex that could result in the inhibition of apoptosis [34,35], facilitation of apoptosis [36] or can also have no effect [37] depending on the context. Regardless, these findings underscore the functional link between MCL1 and this mitochondrial ubiquitin ligase complex.
A-B. Top co- and anti-dependencies of A. MCL1 or B. MARCHF5 from the Cancer Dependency Map 23Q4. The top 100 genes are listed in S1 Table. C. Median fold changes of gRNAs targeting MCL1 or MARCHF5 in our gene essentiality CRISPR screens in PEL cell lines [15]. D. Western blots of MARCHF5 in a panel of PEL cell lines that are infected with KSHV only or co-infected by KSHV and EBV.
To investigate if MCL1 and MARCHF5 exhibit positive co-essentiality in PEL cell lines, we reexamined our previous CRISPR screens that identify PSODs [15,16]. MCL1 and MARCHF5 ranked as the second and third strongest PSOD for PEL cell lines in an updated reanalysis of our previous screens [16]. While these indicate that MCL1 and MARCHF5 are dependency factors in PEL, these ranks, by themselves, do not inform potential interactions between these genes as in DepMap.
We therefore calculated the Pearson correlation coefficient of the consolidated depletion scores for guide RNAs (gRNAs) targeting MCL1 or MARCHF5 similar to DepMap. These strongly correlated across the individual cell lines, even more consistent than the computed correlation coefficient from DepMap (Pearson r = 0.83, Fig 1C). In addition, MARCHF5 protein was expressed at similar levels in a wider set of cell lines (Fig 1D). Together, co-essentiality analyses of CRISPR screens suggests that the two mitochondrial PSODs MCL1 and MARCHF5 may cooperate to promote fitness and survival of PEL cell lines.
MARCHF5 is required for the survival of PEL cell lines
To validate if PEL cell lines are addicted to MARCHF5 expression, we targeted MARCHF5 for CRISPR-Cas9 KO using two independent gRNAs in two representative PEL cell lines: BC-3 expressing a constitutive Cas9 [15] and BCBL-1 expressing a doxycycline (Dox)-inducible Cas9 (iCas9) [38]. In both cell lines, MARCHF5 KO resulted in lower numbers of total live cells compared to the negative gRNA, which targets the safe harbor locus AAVS1 (Fig 2A–C). CRISPR screens from DepMap independently confirm that MARCHF5 is a dependency gene in PEL cell lines but not in uninfected non-cancerous immortalized B cells, prostate and retinal pigment epithelial cells (Fig 2D).
A. Cumulative live cell counts of MARCHF5 Cas9 knockout (KO) cells. To start the growth curves, BC-3 cells were seeded at 3 x 105 cells/mL 3-4 days after transduction with lentiviruses expressing both Cas9 and the targeting gRNA against AAVS1 or MARCHF5. BCBL-1 iCas9 cells expressing the gRNAs were similarly seeded in triplicates and Cas9 expression was induced by Doxycycline (Dox) treatment (0.5 μg/mL). Every three days, total live cells were counted by trypan blue exclusion assay and split back to 3 x 105 cells/mL to allow the cells to continue proliferating. Cumulative live cell counts are expressed relative to the negative control gRNA AAVS1. B. Representative Western blots of MARCHF5 from cells harvested at Day 6 from similar experiments from panel A. C. Quantification of MARCHF5 Western blots normalized to Revert Total Protein stain signal from MARCHF5 KO cells from experiments collected for this study. For panels A and C, statistical differences were calculated using two-way ANOVA with Tukey’s multiple comparison post-hoc test. Adjusted p values reflect post-hoc comparisons between gAAVS1 to either MARCHF5 gRNA at the specific time point. Error bars, standard error of the mean. D. MARCHF5 CRISPR Scores (Gene Effect Score) from genome-wide CRISPR screens in non-cancerous control immortalized cell lines (5 prostate cell lines, 5 retinal pigment epithelial cells, and 2 B cell lines) as well as 3 PEL cell lines were downloaded from DepMap 26Q1. As reference points, a Gene Effect Score of 0 (gray line) denotes that a gene does not affect fitness of a specific cell line while a Gene Effect Score (red line) of 1 represents the median fitness score for pan-essential genes (“housekeeping”) defined by DepMap. The box plots and whiskers represent the median and lower and upper limits of each group.
The co-essentiality of MARCHF5 and MCL1 suggests that these mitochondrial proteins are functioning in the same pathway. To test whether MARCHF5 also prevents intrinsic apoptosis like MCL1, we measured the activities of the terminal caspases 3/7 using a luminescence-based Caspase-Glo 3/7 Assay. In both BC-3 Cas9 and BCBL-1 iCas9 cells, MARCHF5 KO significantly increases caspase 3/7 activities (Fig 3A). Similarly, levels of the early apoptotic marker Annexin V and dead cell stain 7-AAD increased in BCBL-1 MARCHF5 KO cells (Fig 3B). Taken together, these data confirm that MARCHF5 is a dependency factor in PEL cell lines required to inhibit apoptosis.
Apoptosis was measured using A. Caspase 3/7 activity using the Caspase-Glo 3/7 Luminescence Assay or B. levels of Annexin V-7/AAD using the FITC Annexin V Apoptosis Detection Kit with 7-AAD for flow cytometry analysis from BC-3 and BCBL-1 iCas9 MARCHF5 KO cells harvested at Day 6 of the growth curves from Fig 2. Statistical differences were calculated using one-way ANOVA with Tukey’s multiple comparison post-hoc test (n = 3-5). Adjusted p values reflect post-hoc comparisons between gAAVS1 to either MARCHF5 gRNA. Error bars, standard error of the mean.
Ubiquitin ligase activity and dimerization of MARCHF5 are required for the survival of PEL cell lines
MARCHF5 is an E3 ubiquitin ligase anchored to the mitochondrial outer membrane [31,32]. It is known to regulate mitochondrial structure and dynamics by ubiquitinating proteins that regulate either mitochondrial fission or fusion. Furthermore, MARCHF5 protein levels are tightly regulated by auto-ubiquitination and homodimerization through GxxxG motifs to maintain proper mitochondrial homeostasis [39].
To determine if dimerization and/or the ubiquitin ligase activity is essential for MARCHF5 function in PEL cells, we transduced BCBL-1 iCas9 cells expressing the MARCHF5 gRNAs with lentiviruses expressing cDNAs encoding for mCherry (negative control), a positive control 3XFLAG-tagged codon-optimized wild-type MARCHF5 (coMARCHF5 WT), a ligase mutant (coMARCHF5 H43W), or a dimerization mutant (coMARCHF5 4GL) [39,40]. Our codon optimization strategy disrupts the targeting sites for the MARCHF5 gRNAs, thus restricting editing to only the endogenous alleles.
Surprisingly, these cells did not appear to over-express the coMARCHF5–3XFLAG cDNAs by Western blot (S2A Fig). We hypothesized that over-expressing MARCHF5 increases its auto-ubiquitination [39] thus reducing its steady state levels. To determine if over-expression of MARCHF5 increases its proteasome-mediated turnover, we treated the BCBL-1 gAAVS1 cells expressing the MARCHF5 cDNA constructs with MG132. S2B Fig shows that this increased MARCHF5 protein expressed from all the cDNAs consistent with an autoregulatory feedback. We thus repeated the Western blots on Dox-induced BCBL-1 iCas9 cells (9 days) treated with MG132 for 8 h prior to harvesting. Fig 4A–B show that Dox-induced editing reduced endogenous MARCHF5 protein but not the 3X-FLAG-tagged cDNA in the presence of MG132.
A. Western blots of endogenous and 3XFLAG-tagged MARCHF5. BCBL-1 iCas9 cells expressing gRNAs targeting MARCHF5 or AAVS1 were transduced with lentiviruses constitutively expressing cDNAs for mCherry, codon-optimized MARCHF5 (coMARCHF5) wildtype, H43W ligase mutant, or 4GL dimerization mutant and treated with Doxycycline (Dox) (0.5 μg/mL) for 9 days. To visualize the coMARCHF5 expressed from the lentivirus, cells were treated with 2.5 μM MG132 for 8 hour prior to harvesting for Western blotting. B. Quantification of Western blots in panel A normalized to total protein stain and gAAVS1. C. Cumulative live cell counts (relative to control gAAVS1) of BCBL-1 iCas9 cells in panel A after the addition of Dox (no MG132). Cells were seeded at 3 x 105 cells/mL and Cas9 expression was induced by Dox treatment (0.5 μg/mL). Every three days, total live cells were counted by trypan blue exclusion assay and split back to 3 x 105 cells/mL to allow the cells to continue proliferating. Cumulative live cell counts are expressed relative to the negative control gRNA AAVS1. Statistical differences were calculated using two-way ANOVA with Tukey’s multiple comparison post-hoc test (n = 6-8). Adjusted p values reflect post-hoc comparisons between gAAVS1 to either MARCHF5 gRNA at the specific time point. Error bars, standard error of the mean.
Nevertheless, the expression of coMARCHF5 WT completely rescued the decrease of total live cells in MARCHF5 g2 or g3 cells in the absence of MG132 (Fig 4C). This result indicates that the defects on cell fitness are specific to MARCHF5 KO and not from off-target effects. Moreover, neither the H43W ligase mutant nor the 4GL dimerization mutant rescued MARCHF5 KO cells despite being expressed at similar levels with the ectopic WT copy. Although the cells expressing the coMARCHF5 WT have lower endogenous MARCHF5 levels (Fig 4A), this did not affect the proliferation of these cells compared to cells expressing mCherry or the coMARCHF5 mutants (S2C Fig). This is likely because the expression of the ectopic coMARCHF5 wt can compensate for the decrease in the endogenous protein. These data collectively show that both the ubiquitin ligase activity and dimerization of MARCHF5 are required for the survival of PEL cell lines.
Cell death mediated by MARCHF5 KO is not caused by MAVS activation or lytic replication
One of the described functions of MARCHF5 is to negatively modulate antiviral signaling [33]. Upon sensing viral infections, RIG-I-like receptors (RLRs) are activated and bind to the signaling adapter MAVS on the mitochondrial surface. MAVS consequently transduces RLR signaling by activating antiviral kinases. This eventually leads to the phosphorylation of the transcription factors IRF3 and IRF7 and the transcription of interferon-stimulated genes. MAVS-mediated signaling is kept in check through a negative feedback loop wherein MARCHF5 induces MAVS protein turnover. Because MAVS restricts KSHV lytic replication [41], it is possible that the cell death from loss of MARCHF5 results from altered MAVS signaling and, consequently, the latency-lytic balance.
To test if cell death from MARCHF5 KO is caused by dysregulated MAVS levels and signaling, we immunoblotted for total MAVS, phosphorylated IRF3 (p-IRF3), and total IRF3. As a control for p-IRF3, we ran protein lysates from 293T cells transfected with the pCDNA vector or pCDNA expressing STING which is well-established to activate IRF3 and promoters of interferon stimulated genes [42,43]. Total MAVS, p-IRF3, and IRF3 remained unchanged in BC-3 MARCHF5 KO cells while p-IRF3 dramatically increased in STING-transfected 293T cells (Fig 5A). Moreover, we did not detect upregulation of representative early and late lytic proteins (K-bZIP and ORF26, respectively) unlike our positive control lanes for reactivation which are BCBL-1 cells treated with 12-O-tetradecanoylphorbol-13-acetate (TPA) and sodium butyrate (NaB) for 2 days. Similar results were observed in BCBL-1 iCas9 cells (Fig 5B). These data indicate that MARCHF5 KO-induced apoptosis is not due to MAVS stabilization or aberrant lytic reactivation.
Western blots of proteins involved in KSHV lytic reactivation (K-bZIP) or activation of the IRF3 signaling pathway (MAVS, phospho-IRF3) in A. BC-3 and B. BCBL-1 iCas9 MARCHF5 KO cells harvested at Day 6 of the growth curves. BCBL-1 PEL cells treated with 20 ng/mL 12-O-tetradecanoylphorbol-13-acetate (TPA) and 1 mM sodium butyrate (NaB) for 2 days serve as a positive control for viral reactivation. 293T cells transfected with an expression construct for human STING for 2 days serve as a positive control for phospho-IRF3. Quantification of Western blots are shown on the right normalized to total protein stain and gAAVS1. Statistical differences were calculated using two-way ANOVA with Tukey’s multiple comparison post-hoc test. Adjusted p values reflect post-hoc comparisons between gAAVS1 to either MARCHF5 gRNA. Error bars, standard error of the mean.
NOXA restricts the steady-state survival of PEL cell lines
Thus far, our current and previous work [15,24] has shown that PEL cells exhibit oncogenic co-dependencies to MARCHF5 and MCL1 to prevent intrinsic apoptosis, independent of inhibition of MAVS signaling or lytic reactivation. However, it remains unclear how they are mechanistically connected. MARCHF5 has been reported to ubiquitinate MCL1 itself to induce apoptosis [36] or target the MCL1-NOXA complex to inhibit cell death [34,35]. Because both MARCHF5 and MCL1 are pro-survival dependency factors in PEL cell lines, it is unlikely that MARCHF5 ubiquitinates MCL1 for degradation as this will induce apoptosis. We hypothesize that MARCHF5 prevents apoptosis by instead targeting the MCL1 antagonist NOXA for degradation (Fig 6A), consistent with previous reports [34,35]. Consistent with this, co-dependency interactions from DepMap reveal that both MCL1 and MARCHF5 are strongly anti-correlated with PMAIP1 (NOXA) which supports our hypothesis that MARCHF5 ubiquitinates an MCL1 antagonist for degradation (Fig 1A-B, S1 Table).
A. Model to link the oncogenic functions of MARCHF5 and MCL1. We propose that in PEL cell lines, MARCHF5 cooperates with MCL1 function to induce the proteasomal turnover of the MCL1 antagonist NOXA. B. mRNA expression of BH3-only proteins in PEL cell lines from Cell Model Passports. C. Western blots of BH3-only proteins that were targeted for Cas9 KO in BC-3 cells. D. Experimental flow chart of the Fitness Competition Assay. BC-3 Cas9 cells were transduced with a lentivirus encoding the control gAAVS1 and an iRFP670 fluorescent protein or a second lentivirus encoding the targeting gRNA and an eGFP reporter. Equal amounts of control iRFP670 cells were mixed with targeted eGFP cells at the beginning of the growth curve. Fitness indices as reflected by the ratio of eGFP to iRFP670 (eGFP:iRFP670) were measured over 15 days by flow cytometry. E. Fitness indices for each BH3-only KO competition relative to day 0. Statistical differences were calculated using two-way ANOVA with Dunnett’s multiple comparison post-hoc test (n = 3). Adjusted p values reflect post-hoc comparisons between gSAFE4501 to BH3-only protein gRNA. Error bars, standard error of the mean.
To confirm whether NOXA is the apoptotic trigger in PEL cells that necessitates the anti-apoptotic activities of MCL1 and MARCHF5, we measured the relative fitness of BC-3 cells with wildtype or NOXA KO. We hypothesized that Cas9 KO of the apoptotic trigger would provide a fitness advantage to these tumor cells. Although MCL1 is specifically neutralized by NOXA, other promiscuous pro-apoptotic BH3-only proteins can also bind to MCL1 (Fig 6B). To get an unbiased picture of the apoptotic trigger(s) in PEL cells, we extended our analysis to other related BH3 only proteins regardless of their specificity to MCL1. We included the top four BH3-only genes that had measurable basal RNA expression in different PEL cell lines: BCL2L11 (BIM), BID, BAD, and PMAIP1 (NOXA) (Fig 6B). We targeted each of these genes for KO in BC-3 Cas9 cells using two independent test gRNAs expressed from a lentivirus encoding an eGFP fluorescence reporter (Fig 6C). As a negative control, we used the SAFE4501 gRNA that targets a non-functional, non-genic region [44]. In parallel, we derived another negative control cell line expressing the gAAVS1 gRNA from a lentivirus encoding an iRFP670 fluorescence reporter.
Given the extensive adaptation of cancer cell lines during tumor development in patients and prolonged culture in vitro, we anticipated that the fitness advantage from single-gene KOs would be modest due to redundant or compensatory pathways that optimize cellular fitness. We performed a Fitness Competition Assay (FCA) instead of the standard cell counting method we used in Fig 2. We chose this approach because the FCA provides greater sensitivity and is less susceptible to variability introduced during counting and cell splitting of multi-day growth curves. In the FCA, eGFP+ cells expressing the test gRNA are mixed at a 1:1 equal ratio with the negative control iRFP670 + cells expressing gAAVS1 (Fig 6D). Relative fitness of the KO cells can be monitored using the ratio of eGFP to iRFP670 as a readout by flow cytometry.
Of the eight test KO cell lines, NOXA KO cells consistently had a significant fitness advantage for both gRNAs over the negative control cells gSAFE4501 and gAAVS1 (Fig 6E). While both BAD KO and BIM KO cell lines had statistically significant changes, these effects are minor and can only be seen in one gRNA. These experiments suggest that NOXA is the major constitutive apoptotic signal that restricts the steady state survival of PEL cell lines.
MARCHF5 is in the same complex as the MCL1 antagonist NOXA
To test if NOXA is constitutively regulated at the protein level to promote the survival of PEL cells, we treated PEL cell lines with the proteasome inhibitor MG132. MG132 treatment dramatically stabilized NOXA between 2-fold (BC-3 and JSC-1) to 4-fold (BCBL-1) (Fig 7A, B). In contrast, no changes in MARCHF5 levels were observed while MCL1 was marginally but consistently stabilized ~1.5-fold in all three cell lines. This is consistent with a previous report that MCL1 is regulated by the E3 ubiquitin ligase FWB7 in PEL cells [26] although this does not exclude that it can still be ubiquitinated by MARCHF5. This indicates that the NOXA and, to some extent MCL1, undergo rapid turnover by the proteasome at steady-state in all three PEL cell lines.
A. Western blots of PEL cell lines were treated with DMSO or 1.25 μM MG132 for 4 hours. B. Quantification of Western blots in panel A normalized to total protein stain and DMSO-treated cells. Statistical differences were calculated using paired t tests between untreated and MG132 treated in each cell line. Error bars, standard error of the mean.
To demonstrate that MARCHF5 facilitates NOXA turnover, we immunoblotted for NOXA in MARCHF5 KO cells. MARCHF5 KO cells had higher NOXA similar to MG132 treatment alone while MCL1 levels remained statistically unchanged (Fig 8A–B). In contrast, qRT-PCR did not reveal any major increases in NOXA mRNA for either cell line (Fig 8C). Moreover, treatment of MARCHF5 KO cells with MG132 did not further increase total NOXA protein (Fig 8D–E) suggesting that this E3 ligase is involved in the proteasomal turnover but not the transcriptional regulation of NOXA.
A. Western blot for MCL1 and NOXA in wildtype or MARCHF5 KO BC-3 or BCBL-1 iCas9 cells after 6 days of editing. B. Quantification of Western blot in panel A normalized to total protein stain and gAAVS1. C. qRT-PCR of NOXA mRNA in wildtype or MARCHF5 KO BC-3 or BCBL-1 iCas9 cells after 6 days of editing. D. BCBL-1 iCas9 gAAVS1 or MARCHF5 g3 were first Dox treated for 9 days to induce editing. Cells were then treated with either DMSO of 1.25 μM MG132 for 4 h before harvesting for Western blotting. E. Quantification of Western blots in panel D normalized to total protein stain and gAAVS1. F. Day 9 Dox-treated BCBL-1 iCas9 gAAVS1 or MARCHF5 g3 cells were treated with DMSO, 0.1 μM, or 0.25 μM of the MCL1 inhibitor S63845 for 2 days. Cell viability was measured with CellTiter-Glo 2.0 kit and expressed relative to gAAVS1. G. Reciprocal co-immunoprecipitation (co-IP) of the endogenous MARCHF5-NOXA complex in BC-3 cells. Statistical differences were calculated using two-way ANOVA with Dunnett’s multiple comparison post-hoc test or unpaired t test as noted. Adjusted p values reflect post-hoc comparisons between each gRNA to gAAVS1 unless specifically noted. Error bars, standard error of the mean.
We next asked whether loss of MARCHF5 makes PEL cells more dependent on MCL1 since MARCHF5 KO should increase NOXA protein. In line with this hypothesis, MARCHF5 KO cells showed a modest but significant increase in sensitivity to the MCL1 inhibitor S63845 (Fig 8F, MCL1i).
To determine if MARCHF5 directly interacts with NOXA, we performed co-immunoprecipitations (co-IP). IP of endogenous MARCHF5 complexes in BC-3 cells pulled down NOXA while reciprocal co-IP of NOXA complexes conversely pulled down MARCHF5 (Fig 8G). These biochemical experiments confirm that MARCHF5 and NOXA form complexes in PEL cells.
NOXA KO rescues the loss of viability and fitness in MARCHF5 KO cells
Finally, to demonstrate that MARCH5 and NOXA are functionally connected, we targeted both genes for Cas9 KO. We initially attempted this in BC-3 Cas9 cells so we can measure the fitness effect of MARCHF5-NOXA double KO (DKO) in the FCA. However, we were unsuccessful in performing sequential editing in these cells. We therefore used another PEL cell line BC-1 Cas9 which we previously established and validated for efficient editing [15,38].
We first transduced the BC-1 Cas9 cells with puromycin-resistant eGFP lentiviruses expressing the SAFE4501 or NOXA g1 gRNAs (Fig 9A). After confirming NOXA KO by Western blotting (Fig 9B), we again transduced the SAFE4501 or NOXA g1 cells with hygromycin-resistant lentiviruses expressing the gAAVS1, MARCH5 g2 or g3 gRNAs (no fluorescent reporter). After 7 days of selection with hygromycin and confirmation by Western blotting (Fig 9C), we split the cells into three groups: one to harvest for direct measurement with CellTiter-Glo 2.0, another for Caspase-Glo 3/7 Assay and a third to repeat the FCA with BC-1 Cas9 iRFP670 gAAVS1 cells. After two days, MARCHF5 KO in the NOXA KO background had increased cellular viability and fitness as measured by higher CellTiter-Glo 2.0 signals and eGFP:iRFP670 ratios, but also reduced caspase 3/7 activities when compared to the MARCHF5 KO in the negative control SAFE4501 gRNA cells (Fig 9D–F). We note that while we can fully rescue cellular viability and fitness as measured by CellTiter-Glo 2.0 and FCA (Fig 9D and F, SAFE4501-gAAVS1 vs. NOXA g1-MARCHF5 g2), we only saw partial rescue for both readouts in cells with the more efficient MARCHF5 g3 gRNA (by Western blot). Moreover, while NOXA KO significantly reduced the caspase 3/7 activities in both MARCHF5 g2 and g3 cells (Fig 9E, from 5-fold to 1.5-fold for MARCHF5 g2, and 3-fold to 2-fold in MARCHF5 g3), the caspase 3/7 activities remain ~1.5-2X slightly above baseline (Fig 9E, SAFE4501-gAAVS1 vs. NOXA g1-MARCHF5 g2/3). Together, our data suggests that the major function of MARCHF5 is to induce the degradation of NOXA. In addition, the partial rescues in NOXA g1-MARCHF5 g3 cells indicates that MARCHF5 may have other essential functions in these viral tumor cells although these are likely minor relative to its roles in NOXA regulation.
A. Workflow of the characterization experiments of the NOXA-MARCHF5 double KO (DKO) BC-1 Cas9 cells. Cells were transduced with a lentivirus encoding the targeting gRNA (SAFE4501 or NOXA g1), an eGFP reporter, and puromycin resistance gene (puroR). After selection and confirmation of NOXA KO by Western blotting (panel B), these cells were transduced with a second lentivirus encoding another targeting gRNA (gAAVS1, MARCHF5 g2 or g3) and the hygromycin resistance gene (hygR). After 7 days of selection (200 µg/mL) and confirmation of NOXA KO by Western blotting (panel C), cell densities were adjusted to 4 x 105 cells/mL and seeded for CellTiter-Glo2.0 Assay, Caspase-Glo 3/7 Assay and FCA for 2 days. B. Western blot of NOXA in SAFE4501 or NOXA g1 cells. C. Western blot of NOXA in MARCHF5-NOXA DKO cells. D-F. DKO cells were characterized using D. CellTiter-Glo2.0 (viability), E. Caspase-Glo 3/7 Assay (caspase 3/7 activity), or F. FCA (fitness). Caspase-Glo3/7 luminescence signals were normalized to CellTiter-Glo2.0 signals to normalize for total viable cells. All data in panels D-F, were further normalized to the SAFE4501-gAAVS1 DKO cells to establish the baseline. Statistical differences were calculated using one-way ANOVA with Dunnett’s multiple comparison post-hoc test (n = 3). Error bars, standard error of the mean.
In summary, our data support the model that MARCHF5 cooperates with MCL1 by promoting the proteasomal degradation of the MCL1 antagonist NOXA.
Discussion
In this study, we sought to uncover the genetic interactions of the anti-apoptotic protein MCL1, a clinically relevant oncogene in multiple cancers and KSHV-driven PEL [15,17–21,24]. Co-essentiality mapping from DepMap and our own CRISPR screens in PEL cell lines unexpectedly identified that MCL1 exhibits the strongest genetic interaction with the mitochondrial-resident E3 ubiquitin ligase MARCHF5, despite MARCHF5 not being known for being a canonical regulator of apoptosis. Strikingly, MCL1 and MARCHF5 are ranked back-to-back as the second and third strongest PSODs in PEL cell lines, underscoring this co-dependency. We find that the amount of MCL1 protein is insufficient to protect from cell death because single CRISPR KO of MARCHF5 also results in apoptosis. Mechanistically, we find that MARCHF5 induces the turnover of NOXA protein thereby reinforcing the pro-survival role of MCL1 in these tumor cells.
The dependence on both MCL1 and MARCHF5 oncogenes underscores that PEL cell lines constantly require NOXA to be kept in check to protect from cell death. The link between MARCHF5 and MCL1 in apoptosis is now being appreciated but is context specific. In prostate cancer, their roles appear to be antagonistic. MCL1 is amplified in ~10% of cases while MARCHF5 is deleted in up to ~5% of cases with a higher frequency in metastatic prostate cancers suggesting that they function as an oncogene or tumor suppressor, respectively [36]. Loss of MARCHF5 in pancreatic cancer cell lines leads to a decrease in dependence in MCL1 consistent with the negative regulation of MCL1 by MARCHF5. This is in sharp contrast to our data in PEL where MARCHF5 KO cells become more dependent to MCL1 and thus sensitive to MCL1 inhibitors. This mirrors other reports that show a synthetic lethal effect between MCL1 and MARCHF5 consistent with a cooperative function between these proteins [34,45]. In this anti-apoptotic role, MARCHF5 has been shown to facilitate the turnover of the MCL1-NOXA complex [35,36]. However, we do not see this in PEL cells. MARCHF5 KO results in pronounced stabilization of NOXA but not MCL1 in both BC-3 and BCBL-1 cells (Fig 8A–B). Interestingly, MG132 treatment of BC-3 and JSC-1 leads to a small but statistically significant stabilization of the MCL1 protein (Fig 7A–B). This modest stabilization of MCL1 in MG132-treated PEL cells is likely mediated by other E3 ubiquitin ligases because we do not see similar increases of MCL1 protein in MARCHF5 KO cells. Our data thus point to a different mechanism by which MARCHF5 regulates NOXA separately from the MCL1-NOXA complex.
It remains unclear why NOXA is maintained at steady-state levels that necessitates the addictions to both MCL1 and NOXA. It is worth noting that most PEL cell lines have functional p53 and low but detectable activated p53 [13,46]. It is possible that latent infection and/or replicative stress induced by proliferating tumor cells causes this low-level activation of p53. Thus, p53-responsive genes like NOXA need to be controlled by other mechanisms such as MARCHF5-mediated degradation of NOXA and the LANA-mediated stabilization of MCL1 [26]. Interestingly, KSHV encodes for a lytic viral homolog of the BCL2 family (vBCL2) [47,48]. Biochemical studies demonstrate that vBCL2 most resembles MCL1 in sequence and function [49]. Thus, it appears that a major strategy of KSHV to protect infected cells from cell death is to reinforce MCL1 activity or to bring in an MCL1-like protein.
Our study does not rule out that MARCHF5 may have other essential functions in PEL cells. It is possible that the addiction to MARCHF5 also involves the regulation of mitochondrial dynamics to meet the altered energy demands of these actively proliferating tumor cells. Moreover, MARCHF5 has recently been linked to mediating both peroxisome biogenesis and turnover [50,51]. Given that peroxisome proteins are induced by the latency genes and are required for survival by KSHV-infected endothelial cells [52], it is also possible that the dependency to MARCHF5 facilitates peroxisomal metabolism. However, these functions are difficult to observe in PEL due to the confounding effects of apoptosis in subcellular structures. Future studies on the role of MARCHF5 in mitochondrial or peroxisome homeostasis in PEL cell lines should therefore be done in the absence of the apoptotic machinery such as using BAX/BAK1 double knockout cells.
Finally, the co-essential relationship of MCL1 and MARCHF5 may be exploited for combination therapy to treat PEL. While MCL1 inhibitors or proteasome inhibitors effectively kill PEL cell lines in vitro and in vivo [15,24,26,53], their use as monotherapies has limitations in the clinic. Two recent Phase I trials using MCL1 inhibitors show that these treatments result in elevated cardiac troponin and cardiovascular risk in patients [54,55]. In addition, although bortezomib is being used to treat multiple myeloma patients, it also induces lytic reactivation in PEL cell lines [53,56] which may pose complications in PWH. By targeting the MCL1-MARCHF5 axis using combination therapy with MCL1 inhibitors and bortezomib, it is possible that a synergistic effect can be achieved at lower doses. More effective treatments will likely prevent the development of resistance and induce apoptosis faster before the virus can complete the lytic cycle. In addition, the lower doses may prevent dose-dependent cardiotoxicity associated with MCL1 inhibitors. Thus, factoring co-essentiality interactions of oncogenes in rational therapy design for KSHV malignancies may improve therapeutic strategies for these cancers.
Materials and methods
Cells and treatments
All cloning was performed in E. coli Stbl3 cells (Invitrogen, Carlsbad, CA). PEL cell lines were maintained in RPMI-1640 Medium with L-glutamine and sodium bicarbonate (Millipore-Sigma, St. Louis) supplemented with 10% or 20% Serum Plus II Medium Supplement (Millipore-Sigma), 10 μg/mL gentamicin (Gibco, Waltham, MA), and 0.05 mM β-mercaptoethanol (Millipore-Sigma). HEK293T cells were cultured in DMEM with L-glutamine and 4.5 g/L glucose (Millipore-Sigma) supplemented with 10% Serum Plus II Medium Supplement and 10 μg/mL gentamicin. To induce editing in BCBL-1 iCas9, cells were treated with 0.5 μg/mL doxycycline until the end of the growth curve. To induce lytic reactivation, cells were treated with 20 ng/mL TPA and 1 mM sodium butyrate for 2 d. To inhibit proteasome activity, cells were treated with 1.25 μM MG132 for 4 h or 2.5 μM MG132 for 8 h.
Primers and synthetic genes
All primers were purchased from Integrated DNA Technologies (Coralville, IA). Synthetic genes were purchased from Twist Biosciences (South San Francisco, CA). Sequences are found in S2 Table.
gRNAs
Primer sequences and cloning strategies for specific gRNA constructs are found in S2 Table. gRNAs for the constitutive Cas9 backbone were cloned using annealed primers into BsmBI-linearized lentiGuide-Puro (a gift from Feng Zhang; Addgene catalog no. 52963) or pLenti SpBsmBI sgRNA Hygro (a gift from Rene Maehr; Addgene catalog no. 62205). gRNAs for iCas9 were cloned into pLX-sgRNA (a gift from Eric Lander and David Sabatini; Addgene catalog no. 50662) as before [38]. To clone gRNAs targeting the BH3-only proteins, we modified the pRDA122_mKate backbone (a gift from the Genetic Perturbation Platform at the Broad Institute). First, the eGFP cDNA was PCR amplified from pLCE [57] using primers 229 and 230. Then, the eGFP PCR product or codon-optimized iRFP670 synthetic gene was cloned into a AgeI/BamHI-linearized pRDA122_mKate. The resulting gRNA vectors pRDA122_eGFP and pRDA122_iRFP670 now express eGFP and iRFP670, respectively, instead of mKate. The negative control gRNA SAFE4501 [44] was cloned into pRDA122_iRFP670. The negative control gAAVS1 and the rest of the targeting gRNAs were cloned into pRDA122_eGFP. Control gAAVS1 in pLenti SpBsmBI sgRNA Hygro was previously cloned [58].
cDNA cloning
All cDNAs used were cloned by Gibson assembly into pLC-coMCL1-IRES-hyg [24] using the NheI/BamHI restriction sites. pLC-mCherry-IRES-hyg is a gift from Eva Gottwein and was previously published [24]. coMARCHF5–3XFLAG cDNA was synthesized by Twist Biosciences. Site-directed mutagenesis were performed to construct the coMARCHF5–3XFLAG mutants using the pLC-coMARCHF5–3XFLAG WT-IRES-hyg as a template (S2 Table). pCDNA3.1 human STING plasmid is a gift from Dr. Megan Orzalli.
Lentiviruses and lentiviral infection
5 x 106 HEK293T cells were plated in 10 cm dishes. The next day, the medium was replaced with DMEM without any supplements. 5 μg of transfer plasmid, 2.44 μg of pMD2.G (a gift from Didier Trono; Addgene catalog no. 12259) and 3.74 μg of pSPAX2 (a gift from Didier Trono; Addgene catalog no. 12260) were transfected with 39.13 μL of 1 mg/mL polyethylenimine (Sigma-Aldrich, 25kDa Mw branched, pH 7.4). After 4–6 h, media was replaced with complete RPMI-1640 with 10% Serum Plus II Medium Supplement. Cells were grown in the cell culture incubator for three days and viruses were harvested from the supernatant by filtering through a 0.45 μm polyethersulfone membrane.
Cells were either directly infected with undiluted lentiviral supernatants or spinfected with concentrated lentivirus at 1600 g for 45 min. After 24 h, cell culture media was replaced with fresh media with antibiotic selection (1 μg/mL puromycin or 100–250 μg/mL hygromycin).
Growth curves
3 x 105 cells were seeded in 1 mL in 12-well plates. Cells were counted by trypan blue exclusion using the Countess II Automated Cell Counter (Thermo Fisher) every 3 d. Cell densities were adjusted back to 3 x 105 cells/mL to ensure that the cells did not undergo cell cycle arrest.
Competition assay
We first generated our single BH3-only protein KO in BC-3 or BC-1 Cas9 cells using a lentivirus expressing the targeting gRNA or a negative control gRNA SAFE4501. This lentivirus expresses the fluorescent reporter eGFP in its backbone. In parallel, we also engineered our control competitor BC-3 or BC-1 Cas9 cell line by transducing it with a second lentivirus expressing gAAVS1 and the fluorescent reporter iRFP670 instead of eGFP. After puromycin selection and confirmation of KO by Western blots, each eGFP + KO cell line was mixed at 1:1 with iRFP670 + cells in 12-well plates. To establish an accurate baseline of eGFP:iRFP670 ratios, cell pools were analyzed by flow cytometry at this initial time point. Cells were then grown similar to our Growth Curve Assays and eGFP:iRFP670 ratios were continually monitored every 3 d for 15 d.
Caspase 3/7 assay
Twenty microliters of resuspended cells were dispensed into 96-well half-area microplates (Greiner Bio-One, Monroe, NC). An equal volume of reconstituted Caspase-Glo 3/7 Reagent (Promega, Madison, WI) was added to each sample and mixed by repeated pipetting. Samples were incubated in the dark for 30 min. Luminescence was read in a Synergy H1 microplate reader (Biotek, Winooski, VT) with an integration time of 2 s.
Annexin V/7-AAD staining
7 x 105 cells were pelleted and washed twice in 0.5 mL 2% BSA in PBS. The cell pellets were resuspended in 100 μL Annexin V binding buffer (Biolegend, San Diego, CA) with 5 μL Annexin V-APC and/or 5 μL 7-AAD. Cells were stained in the dark for 15 min at 4℃. After incubation, an additional 350 μL of Annexin V binding buffer was added. Samples were analyzed on the BD LSR Fortessa.
Western blots
Cells were lysed on ice for 10 min with radioimmunoprecipitation assay (RIPA) buffer (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% IGEPAL CA-630, 1% sodium deoxycholate) supplemented with protease inhibitor cocktail III (Millipore) and Simple Stop 2 Phosphatase Inhibitor Cocktail (GoldBio, St. Louis, MO). Lysates were clarified by centrifuging at 12,000 g for 10 min at 4℃. Total protein concentrations from clarified lysates were determined using the Pierce BCA Protein Assay Kit (Thermo Scientific). Equal amounts of total protein were loaded and separated in bis-tris acrylamide gels in MES-SDS Running Buffer (Invitrogen) at 100 V. The proteins were transferred to a 0.2 µm nitrocellulose membrane in Towbin buffer (25 mM Tris, 192 mM glycine, 20% methanol, pH 8.3) for 1 h and 45 min at a constant current of 400 mA. Membranes were airdried, stained with Revert 700 Total Protein Stain (LICOR), and imaged using Odyssey Fc. Western blots were blocked with Intercept (TBS) blocking buffer (LICOR) diluted 1:2 in tris-buffered saline (TBS; 20 mM Tris, 150 mM NaCl, pH 7.6) or 5% BSA in TBS for phosphorylated proteins for 1 h. The membranes were incubated with the primary antibody in TBS supplemented with 0.1% Tween-20 (TBST) overnight at 4℃ (See S2 Table). The next day, the membranes are washed 5x with TBST for 10 min each before incubating with IR800CW secondary antibody (LICOR) for 1 h. Final washes in TBST were done before imaging on the Odyssey M (LICOR). Uncropped Western blots are in S1 Fig.
Reciprocal co-immunoprecipitation (co-IP)
For each co-IP reaction, 2 x 107 BC-3 cells were washed thrice with phosphate-buffered saline (PBS, 137 mM NaCl, 2.7 mM KCl, 1.47 mM KH2PO4, 8.1 mM Na2HPO4, pH 7.4) then lysed with 600 μL IP buffer (25 mM HEPES, 100 mM NaCl, 10 mM CaCl2, 5 mM MgCl2) freshly supplemented with 1% digitonin and protease inhibitor cocktail III. To 400 μL of clarified lysate, IP antibodies were added and incubated on a rotisserie overnight at 4℃: 3.5 μL anti-NOXA antibody (0.55 mg/mL), 4.5 μL anti-MARCHF5 antibody (0.21 mg/mL) or 0.77 μL rabbit IgG (2.5 mg/mL). The remaining lysates were stored at -80℃ as input. The following day, 8 μL of Protein A/G magnetic beads (Pierce) were added to each IP and reincubated on a rotisserie for 4.5 h at 4℃. The beads were washed twice with 500 μL IP buffer with 0.1% digitonin and resuspended in 45 μL 1.5X LDS loading buffer (40 mM Tris-Cl, 53 mM Tris Base, 0.75% lithium dodecyl sulfate, 3.75% glycerol, 0.2 mM EDTA, 3.75% β-mercaptoethanol, pH 8.5). For controls, 27 μL of input was mixed with 9 μL 4X LDS loading dye. All samples were denatured at 90℃ for 7 min. 36 μL of input and 20 μL of each IP were run in each lane of a 12% bis-tris acrylamide gel in MES-SDS Running Buffer at 80V. Western blots for MARCHF5 or NOXA were performed as above.
RNA isolation
Cells were lysed in 750 μL TRIzol Reagent (Invitrogen) and phase separated with 150 μL chloroform by centrifugation at 12,000 g for 15 min at 4℃. The aqueous phase was transferred to a new tube and mixed with 375 μL 100% isopropanol to precipitate the RNA. The samples were incubated for 10 minutes at room temperature before centrifuging for 10 min at 12,000 g, 4°C. The RNA pellets were washed with 0.75 mL of 75% ethanol before resuspending in pre-warmed nuclease-free water (>60℃).
DNase treatment
6 μg total RNA was treated with 2U TURBO DNase (Invitrogen) in 1X TURBO DNase Buffer for 30 minutes at 37℃. Reactions were stopped by precipitation of the DNase with the addition of DNase Inactivation Reagent and incubating for 5 min at room temperature. The RNA-containing supernatant was separated and transferred to a new tube after centrifugation at 10,000 g for 3 min.
cDNA synthesis
cDNA synthesis was done using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosciences). Briefly, 1 μg DNase-treated RNA was first combined with 2 μL 10X RT Random Hexamers and incubated at 65°C for 5 min then flash cooled on ice. Each of the RNA-hexamer mixes were mixed in a total of 20 μL reaction volume with 1X RT Buffer, 4 mM dNTPs, 40 U RNase Inhibitor, and either 50U MultiScribe Reverse Transcriptase or nuclease-free water (for no RT control). All samples were incubated at 10 minutes at 25°C, 2 hours at 37°C, and 5 minutes at 85°C in a thermocycler.
qPCR
The cDNA or RT- control (0.5 μL) was combined with 2X LUNA Universal qPCR Master Mix (New England Biolabs) and 0.5 μM each of primers (797 and 798 for NOXA; 743 and 744 for B2M control) in a 20 μL reaction (See S2 Table). Reactions were run in the QuantStudio 6 Pro (Applied Biosystems) under the following conditions: 95°C for 2 min, 40 cycles of 95°C for 15 s, 60°C for 15 s, 72°C for 1 min (data collection step), followed by the default Melt Curve protocol. Ct values and Melt Curve analyses were calculated by QuantStudio 6 Pro using default settings. Fold changes were calculated using 2-(ΔΔCt) method with the B2M as the reference gene.
MCL1i treatments
BCBL-1 iCas9 gAAVS1 or MARCHF5 g3 cells were first treated with Dox for 9 days to induce editing. Cells were then seeded in triplicates in 96-well U-bottom plates at 5000 cells/well and were treated with DMSO only, 0.1 μM, or 0.25 μM of the MCL1 inhibitor S63845 (MCL1i) in a total volume of 150 μL. After 2 days, 20 μL of cells were harvested and mixed with 20 μL of CellTiter-Glo 2.0 (Promega) to measure relative cellular viabilities. Luminescence was read as above.
Supporting information
S1 Table. The Top 100 Codependencies of MCL1 and MARCHF5 downloaded from the Cancer Dependency Map 23Q4.
https://doi.org/10.1371/journal.ppat.1014665.s001
(XLSX)
S2 Table. List of sequences of primers, synthetic genes, and antibodies used in this study.
Also included are the strategies for cloning for the expression or gRNA vectors that are used in the experiments.
https://doi.org/10.1371/journal.ppat.1014665.s002
(XLSX)
S1 Fig. Uncropped Western blots from this study.
https://doi.org/10.1371/journal.ppat.1014665.s003
(PDF)
S2 Fig. MARCHF5 protein levels are autoregulated.
A. Western blots of endogenous and 3XFLAG-tagged coMARCHF5 without MG132 treatment from the same growth curves in Fig 4C. BCBL-1 iCas9 cells expressing gRNAs targeting MARCHF5 or AAVS1 were transduced with lentiviruses constitutively expressing cDNAs for mCherry, codon-optimized MARCHF5 (coMARCHF5) wildtype, H43W ligase mutant, or 4GL dimerization mutant and treated with Doxycycline (Dox) (0.5 μg/mL) for 9 days and harvested for Western blotting. B. Unedited BCBL-1 iCas9 gAAVS1 cells expressing mCherry and coMARCHF5–3XFLAG constructs (no Dox) were left untreated or treated with 2.5 μM MG132 for 8 hours prior to harvesting for Western blotting of endogenous and 3XFLAG-tagged coMARCHF5. C. Cumulative live cell counts from the same growth curves of edited BCBL-1 iCas9 gAAVS1 cells in Fig 4C were replotted relative to the matched mCherry cells to compare the impact of the cDNAs on cellular viability. Statistical differences were calculated using two-way ANOVA with Tukey’s multiple comparison post-hoc test (n = 6–8). Adjusted p values reflect post-hoc comparisons between mCherry cells to the other coMARCHF5 constructs at the specific time point. Error bars, standard error of the mean.
https://doi.org/10.1371/journal.ppat.1014665.s004
(TIF)
References
- 1. Cesarman E, Chadburn A, Rubinstein PG. KSHV/HHV8-mediated hematologic diseases. Blood. 2022;139(7):1013–25. pmid:34479367
- 2. Cesarman E, Chang Y, Moore PS, Said JW, Knowles DM. Kaposi’s sarcoma-associated herpesvirus-like DNA sequences in AIDS-related body-cavity-based lymphomas. N Engl J Med. 1995;332(18):1186–91. pmid:7700311
- 3. Chadburn A, Hyjek E, Mathew S, Cesarman E, Said J, Knowles DM. KSHV-positive solid lymphomas represent an extra-cavitary variant of primary effusion lymphoma. Am J Surg Pathol. 2004;28(11):1401–16. pmid:15489644
- 4. Nador RG, Cesarman E, Chadburn A, Dawson DB, Ansari MQ, Sald J, et al. Primary effusion lymphoma: A distinct clinicopathologic entity associated with the Kaposi’s sarcoma-associated herpes virus. Blood. 1996;88(2):645–56. pmid:8695812
- 5. Simonelli C, Spina M, Cinelli R, Talamini R, Tedeschi R, Gloghini A, et al. Clinical features and outcome of primary effusion lymphoma in HIV-infected patients: A single-institution study. J Clin Oncol. 2003;21(21):3948–54. pmid:14581418
- 6. Guillet S, Gérard L, Meignin V, Agbalika F, Cuccini W, Denis B, et al. Classic and extracavitary primary effusion lymphoma in 51 HIV-infected patients from a single institution. Am J Hematol. 2016;91(2):233–7. pmid:26799611
- 7. Castillo JJ, Shum H, Lahijani M, Winer ES, Butera JN. Prognosis in primary effusion lymphoma is associated with the number of body cavities involved. Leuk Lymphoma. 2012;53(12):2378–82. pmid:22591071
- 8. Lurain K, Polizzotto MN, Aleman K, Bhutani M, Wyvill KM, Gonçalves PH, et al. Viral, immunologic, and clinical features of primary effusion lymphoma. Blood. 2019;133(16):1753–61. pmid:30782610
- 9. Boulanger E, Gérard L, Gabarre J, Molina JM, Rapp C, Abino JF, et al. Prognostic factors and outcome of human herpesvirus 8-associated primary effusion lymphoma in patients with AIDS. J Clin Oncol. 2005;23(19):4372–80. pmid:15994147
- 10. Ramos JC, Sparano JA, Chadburn A, Reid EG, Ambinder RF, Siegel ER, et al. Impact of Myc in HIV-associated non-Hodgkin lymphomas treated with EPOCH and outcomes with vorinostat (AMC-075 trial). Blood. 2020;136(11):1284–97. pmid:32430507
- 11. Gaidano G, Capello D, Cilia AM, Gloghini A, Perin T, Quattrone S, et al. Genetic characterization of HHV-8/KSHV-positive primary effusion lymphoma reveals frequent mutations of BCL6: Implications for disease pathogenesis and histogenesis. Genes Chromosomes Cancer. 1999;24(1):16–23. pmid:9892104
- 12. Gaidano G, Capello D, Fassone L, Gloghini A, Cilia AM, Ariatti C, et al. Molecular characterization of HHV-8 positive primary effusion lymphoma reveals pathogenetic and histogenetic features of the disease. J Clin Virol. 2000;16(3):215–24. pmid:10738140
- 13. Petre CE, Sin SH, Dittmer DP. Functional p53 signaling in Kaposi’s sarcoma-associated herpesvirus lymphomas: Implications for therapy. J Virol. 2007;81(4):1912–22. pmid:17121789
- 14. Yang DM, Chen WG, Xiong J, Sherrod CJ, Henry DH, Dittmer DP. Interleukin 1 receptor-associated kinase 1 (IRAK1) mutation is a common, essential driver for Kaposi sarcoma herpesvirus lymphoma. Proc Natl Acad Sci USA. 2014;111(44):E4762–8. pmid:25341731
- 15. Manzano M, Patil A, Waldrop A, Dave SS, Behdad A, Gottwein E. Gene essentiality landscape and druggable oncogenic dependencies in herpesviral primary effusion lymphoma. Nat Commun. 2018;9(1):3263. pmid:30111820
- 16. Kuehnle N, Gottwein E. Druggable host gene dependencies in primary effusion lymphoma. Curr Opin Virol. 2022;56:101270. Epub 20220928. pmid:36182745; PMCID: PMC10043043
- 17. Beroukhim R, Mermel CH, Porter D, Wei G, Raychaudhuri S, Donovan J, et al. The landscape of somatic copy-number alteration across human cancers. Nature. 2010;463(7283):899–905. pmid:20164920
- 18. Cho-Vega JH, Rassidakis GZ, Admirand JH, Oyarzo M, Ramalingam P, Paraguya A, et al. MCL-1 expression in B-cell non-Hodgkin’s lymphomas. Hum Pathol. 2004;35(9):1095–100. pmid:15343511
- 19. Kuramoto K, Sakai A, Shigemasa K, Takimoto Y, Asaoku H, Tsujimoto T, et al. High expression of MCL1 gene related to vascular endothelial growth factor is associated with poor outcome in non-Hodgkin’s lymphoma. Br J Haematol. 2002;116(1):158–61. pmid:11841410
- 20. Kotschy A, Szlavik Z, Murray J, Davidson J, Maragno AL, Le Toumelin-Braizat G, et al. The MCL1 inhibitor S63845 is tolerable and effective in diverse cancer models. Nature. 2016;538(7626):477–82. pmid:27760111
- 21. Roberts AW, Wei AH, Huang DCS. BCL2 and MCL1 inhibitors for hematologic malignancies. Blood. 2021;138(13):1120–36. pmid:34320168
- 22. Newton K, Strasser A, Kayagaki N, Dixit VM. Cell death. Cell. 2024;187(2):235–56. pmid:38242081
- 23. Moldoveanu T, Czabotar PE. BAX, BAK, and BOK: A coming of age for the BCL-2 family effector proteins. Cold Spring Harb Perspect Biol. 2020;12(4):a036319. pmid:31570337
- 24. Dunham D, Viswanathan P, Gill J, Manzano M. Expression ratios of the antiapoptotic BCL2 family members dictate the selective addiction of Kaposi’s sarcoma-associated herpesvirus-transformed primary effusion lymphoma cell lines to MCL1. J Virol. 2022;96(23):e0136022. pmid:36416587
- 25. Quentmeier H, Geffers R, Hauer V, Nagel S, Pommerenke C, Uphoff CC, et al. Inhibition of MCL1 induces apoptosis in anaplastic large cell lymphoma and in primary effusion lymphoma. Sci Rep. 2022;12(1):1085. pmid:35058488
- 26. Kim YJ, Kim Y, Kumar A, Kim CW, Toth Z, Cho NH, et al. Kaposi’s sarcoma-associated herpesvirus latency-associated nuclear antigen dysregulates expression of MCL-1 by targeting FBW7. PLoS Pathog. 2021;17(1):e1009179. pmid:33471866
- 27. Tsherniak A, Vazquez F, Montgomery PG, Weir BA, Kryukov G, Cowley GS, et al. Defining a cancer dependency map. Cell. 2017;170(3):564-576.e16. pmid:28753430
- 28. Wang T, Yu H, Hughes NW, Liu B, Kendirli A, Klein K, et al. Gene essentiality profiling reveals gene networks and synthetic lethal interactions with oncogenic ras. Cell. 2017;168(5):890-903.e15. pmid:28162770
- 29. Pan J, Meyers RM, Michel BC, Mashtalir N, Sizemore AE, Wells JN, et al. Interrogation of mammalian protein complex structure, function, and membership using genome-scale fitness screens. Cell Syst. 2018;6(5):555-68 e7. pmid:29778836
- 30. Wainberg M, Kamber RA, Balsubramani A, Meyers RM, Sinnott-Armstrong N, Hornburg D, et al. A genome-wide atlas of co-essential modules assigns function to uncharacterized genes. Nat Genet. 2021;53(5):638–49. pmid:33859415
- 31. Yonashiro R, Ishido S, Kyo S, Fukuda T, Goto E, Matsuki Y, et al. A novel mitochondrial ubiquitin ligase plays a critical role in mitochondrial dynamics. EMBO J. 2006;25(15):3618–26. pmid:16874301
- 32. Nakamura N, Kimura Y, Tokuda M, Honda S, Hirose S. MARCH-V is a novel mitofusin 2- and Drp1-binding protein able to change mitochondrial morphology. EMBO Rep. 2006;7(10):1019–22. pmid:16936636
- 33. Yoo YS, Park YY, Kim JH, Cho H, Kim SH, Lee HS, et al. The mitochondrial ubiquitin ligase MARCH5 resolves MAVS aggregates during antiviral signalling. Nat Commun. 2015;6:7910. pmid:26246171
- 34. Subramanian A, Andronache A, Li YC, Wade M. Inhibition of MARCH5 ubiquitin ligase abrogates MCL1-dependent resistance to BH3 mimetics via NOXA. Oncotarget. 2016;7(13):15986–6002. pmid:26910119
- 35. Djajawi TM, Liu L, Gong JN, Huang AS, Luo MJ, Xu Z, et al. MARCH5 requires MTCH2 to coordinate proteasomal turnover of the MCL1:NOXA complex. Cell Death Differ. 2020;27(8):2484–99. pmid:32094511
- 36. Arai S, Varkaris A, Nouri M, Chen S, Xie L, Balk SP. MARCH5 mediates NOXA-dependent MCL1 degradation driven by kinase inhibitors and integrated stress response activation. Elife. 2020;9:e54954. pmid:32484436
- 37. Cherok E, Xu S, Li S, Das S, Meltzer WA, Zalzman M, et al. Novel regulatory roles of Mff and Drp1 in E3 ubiquitin ligase MARCH5-dependent degradation of MiD49 and Mcl1 and control of mitochondrial dynamics. Mol Biol Cell. 2017;28(3):396–410. pmid:27932492
- 38. Manzano M, Günther T, Ju H, Nicholas J, Bartom ET, Grundhoff A, et al. Kaposi’s sarcoma-associated herpesvirus drives a super-enhancer-mediated survival gene expression program in primary effusion lymphoma. mBio. 2020;11(4):e01457-20. pmid:32843547
- 39. Kim SH, Park YY, Yoo YS, Cho H. Self-clearance mechanism of mitochondrial E3 ligase MARCH5 contributes to mitochondria quality control. FEBS J. 2016;283(2):294–304. pmid:26476016
- 40. Karbowski M, Neutzner A, Youle RJ. The mitochondrial E3 ubiquitin ligase MARCH5 is required for Drp1 dependent mitochondrial division. J Cell Biol. 2007;178(1):71–84. pmid:17606867
- 41. West JA, Wicks M, Gregory SM, Chugh P, Jacobs SR, Zhang Z, et al. An important role for mitochondrial antiviral signaling protein in the Kaposi’s sarcoma-associated herpesvirus life cycle. J Virol. 2014;88(10):5778–87. pmid:24623417
- 42. Ishikawa H, Barber GN. STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. Nature. 2008;455(7213):674–8. pmid:18724357
- 43. Zhong B, Yang Y, Li S, Wang YY, Li Y, Diao F, et al. The adaptor protein MITA links virus-sensing receptors to IRF3 transcription factor activation. Immunity. 2008;29(4):538–50. pmid:18818105
- 44. Morgens DW, Wainberg M, Boyle EA, Ursu O, Araya CL, Tsui CK, et al. Genome-scale measurement of off-target activity using Cas9 toxicity in high-throughput screens. Nat Commun. 2017;8:15178. pmid:28474669
- 45. Wang Y, Poon RYC. MARCH5 regulates mitotic apoptosis through MCL1-dependent and independent mechanisms. Cell Death Differ. 2023;30(3):753–65. pmid:36329234
- 46. Sarek G, Kurki S, Enbäck J, Iotzova G, Haas J, Laakkonen P, et al. Reactivation of the p53 pathway as a treatment modality for KSHV-induced lymphomas. J Clin Invest. 2007;117(4):1019–28. pmid:17364023
- 47. Sarid R, Sato T, Bohenzky RA, Russo JJ, Chang Y. Kaposi’s sarcoma-associated herpesvirus encodes a functional bcl-2 homologue. Nat Med. 1997;3(3):293–8. pmid:9055856
- 48. Suraweera CD, Hinds MG, Kvansakul M. Structural insight into KsBcl-2 mediated apoptosis inhibition by Kaposi sarcoma associated herpes virus. Viruses. 2022;14(4):738. pmid:35458468
- 49. Flanagan AM, Letai A. BH3 domains define selective inhibitory interactions with BHRF-1 and KSHV BCL-2. Cell Death Differ. 2008;15(3):580–8. pmid:18084238
- 50. Zheng J, Chen X, Liu Q, Zhong G, Zhuang M. Ubiquitin ligase MARCH5 localizes to peroxisomes to regulate pexophagy. J Cell Biol. 2022;221(1):e202103156. pmid:34747980
- 51. Verhoeven N, Oshima Y, Cartier E, Neutzner A, Boyman L, Karbowski M. Outer mitochondrial membrane E3 Ub ligase MARCH5 controls mitochondrial steps in peroxisome biogenesis. bioRxiv. 2023. pmid:37693581
- 52. Sychev ZE, Hu A, DiMaio TA, Gitter A, Camp ND, Noble WS, et al. Integrated systems biology analysis of KSHV latent infection reveals viral induction and reliance on peroxisome mediated lipid metabolism. PLoS Pathog. 2017;13(3):e1006256. pmid:28257516
- 53. Sarosiek KA, Cavallin LE, Bhatt S, Toomey NL, Natkunam Y, Blasini W, et al. Efficacy of bortezomib in a direct xenograft model of primary effusion lymphoma. Proc Natl Acad Sci U S A. 2010;107(29):13069–74. pmid:20615981; PMCID: PMC2919898
- 54. Desai P, Lonial S, Cashen A, Kamdar M, Flinn I, O’Brien S, et al. A phase 1 first-in-human study of the MCL-1 inhibitor AZD5991 in patients with relapsed/refractory hematologic malignancies. Clin Cancer Res. 2024;30(21):4844–55. pmid:39167622
- 55. Yuda J, Will C, Phillips DC, Abraham L, Alvey C, Avigdor A, et al. Selective MCL-1 inhibitor ABBV-467 is efficacious in tumor models but is associated with cardiac troponin increases in patients. Commun Med (Lond). 2023;3(1):154. pmid:37880389
- 56. Brown HJ, McBride WH, Zack JA, Sun R. Prostratin and bortezomib are novel inducers of latent Kaposi’s sarcoma-associated herpesvirus. Antivir Ther. 2005;10(6):745–51. pmid:16218174
- 57. Zhang J, Jima DD, Jacobs C, Fischer R, Gottwein E, Huang G, et al. Patterns of microRNA expression characterize stages of human B-cell differentiation. Blood. 2009;113(19):4586–94. pmid:19202128
- 58. Kuehnle N, Osborne SM, Liang Z, Manzano M, Gottwein E. CRISPR screens identify novel regulators of cFLIP dependency and ligand-independent, TRAIL-R1-mediated cell death. Cell Death Differ. 2023;30(5):1221–34. pmid:36801923