Alterations in Sphingolipid Composition and Mitochondrial Complex I Deciency Represent Synergistic Therapeutic Vulnerabilities Linked to Vincristine Resistance

Background: Modifications in sphingolipid (SL) metabolism and mitochondrial bioenergetics are key factors implicated in cancer cell response to chemotherapy, including chemotherapy resistance. Vinca alkaloids such as vincristine (VCR), widely used in cancer treatment, are no exception, as their beneficial actions are often supplanted by resistance. In the present work we utilized HL-60 human leukemia cells and a VCR-resistant counterpart, HL-60/VCR cells, to determine potential interplay between SL metabolism and mitochondrial bioenergetics supportive of VCR resistance. Methods: Cellular SL alterations were assessed via a combination of enzyme immunoblot analysis and lipidomics. Mitochondrial function was evaluated in intact cells, permeabilized cells and isolated mitochondria using a comprehensive diagnostic assay system designed to directly interrogate oxidative phosphorylation (OXPHOS) kinetics. Informed by SL and bioenergetic readouts, experiments were designed to assess the therapeutic efficacy of co-targeting SL metabolism and mitochondrial flux to combat VCR resistance. Results: Relative to wild-type cells, HL-60/VCR presented with global alterations in SL composition, typified by upregulated expression of sphingosine kinase (SPHK1), which catalyzes formation of sphingosine 1-phosphate (S1P), glucosylceramide synthase (GCS), which catalyzes formation of glucosylceramides (GC), and acid ceramidase, responsible for ceramide hydrolysis. In support of these changes, VCR resistance was also characterized by increases in S1P, several molecular species of ceramide and GC, and changes in sphingomyelin (SM) molecular species. With respect to mitochondria, despite increased basal respiration in HL-60/VCR cells, direct interrogation of the mitochondrial network revealed intrinsic respiratory complex insufficiency, largely localized to complex I (CI). Importantly, forced ceramide accumulation in wild-type cells phenocopied the respiratory insufficiency observed in HL-60/VCR, and co-targeting SL metabolism and CI induced synergistic cytotoxicity in HL-60/VCR.


Background
Chemotherapy resistance is a major element underlying treatment failure in cancer. Factors contributing to resistance include drug efflux driven by multidrug resistance (MDR) membrane transport proteins (e.g.,Pglycoprotein; P-gp), drug activation/inactivation missteps, DNA damage repair, cell death inhibition (i.e., resistance to apoptosis), mutation of the target to reduce inhibitor binding, and alterations in mitochondrial bioenergetics [1][2][3]. These current research paradigms by and large drive the study of drug resistance in cancer.
Vinca alkaloids, such as vincristine (VCR) and vinblastine (VBL), part of the World Health Organization's List of Essential Medicines, are no exception; their benefits are supplanted by upregulated expression of MDR proteins, dysfunctional apoptosis, and alterations in β-tubulin that thwart Vinca-associated microtubule targeting benefits.
Despite these limitations, Vinca alkaloids are used widely in treatment of both liquid and solid tumors, and VCR as well as VBL are frequent agents in combination regimens. VCR, the focus of the present study, is a component in CVP (cyclophosphamide, VCR, prednisone) used in follicular B-cell lymphoma, CHOP (cyclophosphamide, doxorubicin, VCR, prednisone) used in large B-cell lymphoma, and was recently demonstrated to exert antileukemic effects in CD44+ AML when administered as A6 peptide functionalized polymersomal vincristine sulfate [4].
In addition to the hallmark factors underlying chemotherapy resistance, the interplay of SL's and mitochondrial bioenergetics merits consideration. Many anticancer agents raise intracellular levels of ceramides, SL's that exhibit strong tumor-suppressing activities [5]. With respect to VCR, Olshefski and Ladisch [6] have shown that VCR exposure increases ceramide levels in leukemia cells, resulting in apoptotic responses that were intensified by blocking ceramide glycosylation. Interestingly, inhibitors of SL enzymes were recently found to interfere with energy transduction across the AML mitochondrial network, suggestive of a direct link between SL metabolism and mitochondrial bioenergetics [7]. Together, these findings raise the intriguing possibility that SL-mitochondria crosstalk may play a fundamental role in chemotherapy response and thus represent an opportunity for novel therapeutic interventions against drug resistance.
Herein, we utilized VCR-resistant human leukemia cells (HL-60/VCR) to comprehensively evaluate the interplay between SL metabolism and mitochondrial bioenergetics and explore the therapeutic potential of cotargeting these pathways in drug resistance. In HL-60/VCR cells as opposed to wild-type (wt) HL-60 cells, we detail striking alterations in SL composition, highlighted by specific increases in S1P, ceramides, as well GC, and changes in levels of specific SM molecular species. These changes were accompanied by intrinsic mitochondrial remodeling that allowed for maintenance of cellular respiratory kinetics, despite intrinsic limitations in respiratory CI. Interestingly, despite minimal cytotoxicity as single agents, the combination of various inhibitors of SL metabolism with CI inhibition induced synergistic cell death in HL-60/VCR.

Materials Immunoblotting
Immunoblotting was conducted as previously described [9,10] with modifications. SPHK1 (47 kDa, catalog # 12071), used at 1:1000 dilution, and β-actin (45 kDa, catalog # 3700), used at 1:3000 dilution, were from Cell Signaling. P-gp (180 kDa, catalog # NB60D-1036), used at 1:1000 dilution, was from Novus (Imgenex). GCS (40 kDa, catalog # PA5-42111), used at a 1:1000 dilution, was from Invitrogen. AC (13 kDa, catalog # 612302), used at a 1:1000 dilution, was from Labome (BD Bioscience). Briefly, cells were lysed in 1x RIPA buffer (Cell Signaling Technology, Danvers, MA, USA) containing phosphatase inhibitor cocktail 2, protease inhibitor, and PMSF (1.0 mM), according to the manufacturer's protocol (Sigma Chemical Company). The bicinchoninic acid (BCA) method was employed for determining protein concentrations. Lysates were centrifuged to remove debris and then heated in 1x NuPage LDS sample buffer at 95°C for 5 min (GCS lysate was heated at 55°C for 5 min). Equal amounts of total protein were separated by SDS-polyacrylamide gel electrophoresis and subsequently transferred onto PVDF or nitrocellulose membranes (Millipore, Billerica, MA, USA). The membranes were blocked with 5% non-fat milk in PBS for 1 hr at room temperature and incubated with primary antibodies according to the manufacturer's instructions, at 4°C overnight. The membranes were washed with PBST, followed by incubation with the corresponding HRP-conjugated secondary antibodies according to the manufacturer's instructions, for 1 hr at room temperature. The membranes were washed with PBST before exposure. Bound antibodies were measured using an enhanced chemiluminescence detection kit (Thermo Fisher Scientific) and surveyed using a ChemiDoc imaging system (Bio-Rad, Hercules, CA). The relative intensity of protein expression was calculated using ImageJ software (National Institutes of Health).

Quantitative reverse transcription PCR (RT-PCR)
An RNA extraction kit (Qiagen Pty., Ltd, Waltham, MA) was used to extract total RNA, which was then reverse-transcribed into cDNA using a ProtoScript II First Strand cDNA Synthesis Kit (New England Biolabs, Ipswich, MA). The synthesized cDNAs were amplified with SYBR Premix (Bimake, Houston, TX) using the ABI VII7 Real-Time PCR System (Invitrogen). The PCR cycling parameters were 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Relative mRNA levels were calculated using the comparative Ct method (Applied Biosystems instruction manual) and presented as ratios to their biological controls. The fold change in expression of each target mRNA relative to β-actin was calculated as 2 Δ(ΔCt) , where ΔCt = ΔCt β-acin − Ct genes. β-actin transcript levels were confirmed to correlate well with total RNA levels and therefore were used for normalization throughout the experiments [11,12] flux was stimulated with FCCP titration (0.5-3.0µM). Cytochrome C (10 µM) was added to check the integrity of the outer mitochondrial membrane and then respiration was inhibited with rotenone (Rot; 5 µM) and antimycin A (Ant A; 5 µM). Data were normalized to viable cell count and expressed as pmol/s/million cells.

Mitochondrial isolation
As described previously [13] and with some modifications, cells were centrifuged at 300 x g for 10 min followed by a PBS wash. Cell pellets were resuspended in Mitochondrial Isolation Buffer with BSA (100mM KCl, 50mM MOPS, 1mM EGTA, 5mM MgSO 4 , 0.2% BSA, pH 7.1) and homogenized using a borosilicate glass mortar and Teflon pestle. Homogenates were centrifuged at 800 x g for 10 min at 4°C. The supernatant was collected, and this process was repeated for a total of 3 times. The collected supernatant was centrifuged at 10,000 x g for 10 min at 4°C to pellet the mitochondrial fraction. The fraction was resuspended in Mitochondrial Isolation Buffer without BSA, transferred to a microcentrifuge tube and spun down at 10,000 x g. Final resuspension of the mitochondrial pellet was done in Mitochondria Isolation Buffer and protein concentration was calculated using the Pierce BCA assay. Respiration assays using isolated mitochondria were similar to that described for permeabilized cells.

Mitochondrial functional assessment (isolated mitochondria)
High-resolution O 2 consumption rate (JO 2 ) measurements were conducted using the Oroboros Oxygraph-2K (Oroboros Instruments, Innsbruck, Austria). The base assay buffer was identical to that used in permeabilized cells (105mM potassium-MES, 30mM KCl, 10mM KH 2 PO 4 , 5mM MgCl 2 , 1mM EGTA, 2.5g/L BSA and 5mM creatine monohydrate; pH=7.2). As previously described [13,14], a modified version of the CK clamp technique was used to determine steady-state JO 2 across a physiological ΔG ATP span. For complete details regarding the calculation of ΔG ATP at each titration point see [14]. To begin, mitochondria were added to respiration buffer, followed by the addition of respiratory substrates to stimulate State 4 (i.e., non- inhibit ATP synthesis and cyanide (CN, 10mM) was added to induce 100% reduction of the matrix NADH pool.
Relative fluorescence was converted to pmol of H 2 O 2 using an H 2 O 2 standard curve, as previously described [15].

Mitochondrial lysis and sample prep for label-free proteomics
Isolated mitochondria were lysed in urea lysis buffer (8M urea in 40mM Tris, 30mM NaCl, 1mM CaCl 2 , 1 x cOmplete ULTRA mini EDTA-free protease inhibitor tablet; pH=8.0), as described previously [13,16]. The samples were subjected to three freeze-thaw cycles, and sonication with a probe sonicator in three 5s bursts (Q Sonica #CL-188; amplitude of 30). Samples were centrifuged at 10,000 x g for 10min at 4°C. Protein concentration was determined by BCA. Equal amounts of protein were reduced with 5mM DTT at 37°C for 30min,

Data analysis for label-free proteomics
As described previously [16], with some modification, Proteome Discoverer 2.2 (PDv2.2) was used for raw data analysis, with default search parameters including oxidation (15.995 Da on M) as a variable modification and carbamidomethyl (57.021 Da on C) as a fixed modification. Data were searched against the Uniprot Homo Sapiens reference proteome (Proteome ID: UP000005640), as well as the Human Mito Carta 2.0 database [17].
PSMs were filtered to a 1% FDR and grouped to unique peptides while maintaining a 1% FDR at the peptide level. Peptides were grouped to proteins using the rules of strict parsimony and proteins were filtered to 1% FDR.
Peptide quantification was done using the MS1 precursor intensity. Imputation was performed via low abundance resampling. Using only high confidence master proteins, mitochondrial enrichment factor (MEF) was determined as previously described [13] by comparing mitochondrial protein abundance (i.e., proteins identified to be mitochondrial by cross-reference with the MitoCarta 2.0 database) to total protein abundance.

Statistical evaluation
All mass spectrometry samples were normalized to total protein abundance, and the protein tab in the PDv2.2 results was exported as a tab delimited .txt. file and analyzed. Protein abundance was converted to the Log 2 space. For pairwise comparisons, tissue mean, standard deviation, p-value (p; two-tailed Student's t-test, assuming equal variance), and adjusted p-value (Benjamini Hochberg FDR correction) were calculated [18,19].
Mitochondrial assay results are expressed as the mean ± SEM (error bars). Data were normalized to protein loaded per experiment and then corrected for the mitochondrial enrichment factor (*) calculated for that sample, or the group mean, with the final values expressed as pmol/s/mg protein*. For a detailed description of this method, see previous publication [13]. Throughout the paper, differences between groups were assessed by t-test, or one-way ANOVA, followed by Tukey's test where appropriate using GraphPad Prism 8 software

HL-60/VCR cell characteristics
The cell line used in the present work, HL-60/VCR, was developed by Safa and colleagues from HL-60/Vinc cells, originally developed by Melvin S. Center, Division of Biology, Kansas State University, Manhattan, KS, by step-wise selection to increasing concentrations of VCR [20]. To confirm the VCR resistance status in our model, we determined sensitivity to VCR in wt and in HL-60/VCR cells. As shown, HL-60/VCR cells were more resistant to VCR compared to the VCR-naïve, wt counterpart (Fig. 1A), with IC50 values of 10.6 (12.8 µM) and 0.001 µg/ml (1.2 nM), respectively, translating to a >10,000-fold difference in sensitivity. From a morphological perspective, wt and VCR-resistant cells were similar in size and appearance; however, HL-60/VCR cells tend to clump in suspension (photos not shown). Cell growth rate was however slower in drug resistant cells compared to wt, with HL-60/VCR growing at 57% the rate of wt cells (Fig. 1B). Lastly, and as expected, ABCB1 transcription (P-glycoprotein, P-gp) was markedly elevated (2000-fold) in HL-60/VCR cells along with a slight but significant increase (3-fold) in ABCC1 (multidrug resistance protein, MRP), and (25-fold) ABCG2 (breast cancer resistant protein, BCRP) compared to wt cells (Fig. 1C).

Alterations in SL enzyme expression accompany VCR resistance
Because of the key role that SL's play in cancer growth [21,22] and response to chemotherapy [23], we were interested in assessing the impact of VCR resistance on expression of several enzymes involved in SL metabolism. Firstly, levels of total SPHK1 were elevated in HL-60/VCR cells compared to the wt counterpart ( Fig. 1D-E). Regarding SPHK1, it is important to note that the product of this enzyme, S1P, is mitogenic, and thus favors neoplastic growth. GCS, which catalyzes glycosylation of ceramide to GC, and in effect dulls ceramide's tumor-suppressor activity by reducing ceramide levels, was overexpressed in VCR-resistant cells ( Fig. 1D-E). This change in glycosphingolipid metabolism has been shown to accompany MDR [24]. Acid ceramidase (AC) expression was also enhanced in HL-60/VCR compared to wt cells (Fig. 1D-E), as previously demonstrated in HL-60/VCR cells [25]. Lastly, although expected, HL-60/VCR cells exhibited markedly elevated levels of P-gp ( Fig. 1D-E).

Lipidomic analysis
Complementary with immunoblot data, HL-60/VCR cells presented with multi-fold increases in S1P and dh-S1P levels, compared to wt cells (Fig. 1F). Sphingosine levels were similar in both cell lines, whereas the levels of dihydrosphingosine (dhSph), also termed sphinganine, were diminished by approximately 75% in HL-60/VCR cells (Fig. 1F). The levels of several ceramide molecular species were increased in HL-60/VCR cells, including C14:0, C16:0, C24:1, as well as the C26 species, C26 and C26:1 (Fig. 1G). The increases in ceramide molecular species were mirrored by concomitant increases in corresponding GC species, as shown for example in C14:0, C16:0, C22:1, C24:1, and C26 molecular species (Fig. 1G). Several molecular species of SM were decreased in HL-60/VCR cells versus wt, as demonstrated in C18, C20, C22, and C24 (Fig. 1G). On the opposing side, C22:1 SM levels were elevated and a marked increase in C26:1 SM was clearly shown in VCRresistant cells (Fig. 1G). In summary, the data in Fig. 1 demonstrate widespread alterations in SL composition occur with VCR resistance, and interestingly, C22:1 ceramide appears to be sequestered in GC and SM.

Intact cellular respirometry in VCR resistance reveals increased mitochondrial content.
Mitochondria play a key role in cancer cell biology and are increasingly recognized as potential targets in cancer therapy [26]. Mitochondria are also important downstream targets of ceramide [27,28]. Considering the dynamic changes in SL profiles and enzyme expression patterns that accompanied VCR resistance, we sought to determine whether mitochondrial metabolism was impacted in the process. To begin, initial respirometry experiments were performed in intact cells under basal conditions, as well as in response to ATP synthase inhibition (i.e., oligomycin), chemical uncoupling (i.e., the addition of the protonophore FCCP), and electron transport inhibition (i.e., rotenone and antimycin A). Respiration under basal conditions was 2-fold higher in HL-60/VCR compared to wt cells ( Fig. 2A, basal abbreviated by 'Basal'). Interestingly, the ability of oligomycin to lower respiration was blunted in HL-60/VCR ( Fig. 2A, 'Oligo'). Although inhibition of JO2 by oligomycin is typically interpreted to reflect OXPHOS 'coupling efficiency', we hypothesized that high P-gp expression in HL-60/VCR could be driving oligomycin efflux. To test this, oligomycin-mediated inhibition of basal respiration was assessed in HL-60/VCR cells in the presence of the P-gp inhibitor verapamil. Under these conditions, oligomycin inhibited basal respiration similar to that seen in wt cells (Fig. 2B), confirming P-gp mediated efflux of oligomycin. Maximal uncoupled respiration was nearly 2-fold higher in HL-60/VCR and respiration remained higher in HL-60/VCR following the addition of both rotenone and antimycin A, suggesting increased non-mitochondrial oxidase activity in HL-60/VCR ( Fig. 2A, 'Rot/Ant'). Correcting for non-mitochondrial JO 2 , maximal uncoupled respiration was higher in HL-60/VCR (Fig. 2C), consistent with previous findings linking increased mitochondrial content to AML chemoresistance [2,29]. In our model system, HL-60/VCR were continuously cultured in the presence of 1.0µg/mL (1.2µM) VCR. To determine if the bioenergetic phenotype observed in VCR resistant cells was attributable to maintaining our stock cultures in VCR medium, we repeated our respirometry experiments in HL-60/VCR cultured in growth media devoid of VCR for seven days. Results revealed nearly identical results to that observed in cells continuously exposed to VCR ( Fig. 2A, 'VCR +VCR ' vs 'VCR -VCR ').

Comprehensive mitochondrial phenotyping in VCR resistance reveals functional limitations in CI
Increased basal respiration and higher mitochondrial mass are commonly observed bioenergetic phenotypes associated with drug-resistance [2,30]. Although such phenotypes are presumed to reflect an increased reliance on oxidative ATP production, how the mitochondrial network is intrinsically remodeled to support the drug resistant phenotype remains poorly understood. To begin to address this gap, we first sought to validate our intact cell respirometry data by directly measuring mitochondrial respiratory capacity in substrate replete permeabilized cells. Surprisingly, despite higher basal respiration (Fig. 2D 'Cells'), maximal respiratory capacity induced via FCCP titration was decreased, rather than increased, in HL-60/VCR cells (Fig. 2D 'FCCP'), suggesting intrinsic dysregulation across the VCR mitochondrial network. Similar results were observed in HL-60/VCR cultured for seven days in VCR-free culture media (Fig. 2D). Importantly, the addition of cytochrome C did not increase respiratory flux in any group (Fig. 2D), indicating that the respiratory limitations in VCR were not attributable to outer membrane permeabilization. To probe deeper, we isolated mitochondria from wt and VCR resistant HL-60 cells and comprehensively evaluated mitochondrial bioenergetic function. Using this workflow, we can model the impact of a physiological ATP free energy (i.e, ΔG ATP ) span on respiratory flux [14,31].
Experimental manipulation of ΔG ATP is made possible via a modified version of the creatine kinase (CK) energetic clamp [32]. This technique leverages the enzymatic activity of CK, which couples the interconversion of ATP and ADP to that of phosphocreatine (PCr) and free creatine (Cr) to titrate the extra-mitochondrial ATP/ADP ratio and allow direct interrogation of OXPHOS kinetics in the presence of various substrate/inhibitor combinations.
To begin, isolated mitochondria from wt and VCR resistant HL-60 cells, either continuously exposed to VCR in culture or following 7-day withdrawal, were energized with saturating carbon (pyruvate, malate, glutamate, octanoyl-carnitine, succinate; indicated by 'Multi') and respiration was stimulated with the CK clamp at minimal ATP free energy to mimic maximal OXPHOS demand. Subsequent titrations in PCr were then performed to gradually lower ATP resynthesis. Both maximal respiratory flux and OXPHOS conductance (i.e., the slope of the JO 2 vs ΔG ATP relationship) were decreased in HL-60/VCR (Fig. 3A-B). To identify the source of decreased OXPHOS conductance in HL-60/VCR, NADH/NAD + redox poise was measured in parallel in substrate-replete isolated mitochondria exposed to an identical ΔG ATP span. Results revealed NADH/NAD + hyper-reduction in HL-60/VCR mitochondria (Fig. 3C), indicating that OXPHOS impairments in HL-60/VCR are not due to a generalized impairment in dehydrogenase flux, but likely attributable to respiratory complex limitations (i.e., the electron transport system; ETS).
To investigate complex specific OXPHOS kinetics, mitochondria were energized with either a CI (Pyr/M) or CII (Succ/R) specific substrate combination and respiration was again stimulated with the CK clamp. At minimal ΔG ATP , respiration support by Pyr/M was nearly 2-fold lower in HL-60/VCR (Fig. 3D), despite no differences in the presence of S/Rot (Fig. 3E). These findings suggested that the respiratory impairments observed with Pyr/M were either caused by limitations in pyruvate uptake/metabolism or due to a generalized impairment in NADH-linked flux (i.e., CI limitations). To distinguish between these possible phenotypes, we performed an additional respirometry assay in which CI-supported flux was fueled with G/M to provide NADH that circumvents pyruvate metabolism (Fig. 3F). Under these conditions, G/M supported JO 2 was again reduced in HL-60/VCR (Fig. 3F), suggesting a specific bioenergetic limitation localized at CI. In support of this, a dihydroorotate (DHO) assay was performed to provide electrons directly to the ubiquinone pool, thus bypassing CI in a manner comparable to succinate. Respiration supported by DHO was higher in HL-60/VCR (Fig. 3G), further corroborating a CI-specific limitation in HL-60/VCR. Interestingly, even in the absence of carbon substrate, mitochondrial respiration was elevated in HL-60/VCR, suggesting the involvement of increased mitochondrial oxidase activity (Fig 3G, 'ADP'). To address this, we assessed mitochondrial H 2 O 2 emission in deenergized mitochondria, as well as in response to Pyr/M and the CK clamp ( Fig. 3H-I). Results revealed an over 10-fold increase in H 2 O 2 emission in HL-60/VCR under all conditions ( Fig. 3H-I). Interestingly, while the addition of the CK clamp decreased peroxide emission in wt mitochondria, presumably as a result of decreased redox pressure, H 2 O 2 emission was unaffected by ATP free energy in HL-60/VCR (Fig. 3H-I). Such results are consistent with elevated NAD(P)H oxidase activity in HL-60/VCR mitochondria and provide a potential explanation for the increased rotenone/antimycin insensitive respiration observed in intact cells ( Fig. 2A).
To identify the mechanism of CI deficiency in HL-60/VCR, we conducted a label-free quantitative proteomics screen using peptides prepared from the same isolated mitochondria samples used for functional characterization. In total we identified 385 differentially expressed mitochondrial proteins (adjusted P value < 0.01) comparing wt to HL-60/VCR (Supplementary Table 1). Focusing in on the respiratory complexes exclusively, the vast majority of CI subunits were specifically downregulated in HL-60/VCR (Fig. 4A).
Interestingly, the core subunits of succinate dehydrogenase (i.e., CII) were increased in HL-60/VCR, presumably to compensate for CI partial loss-of-function (Fig. 4A). To explore the biological consequences of CI deficiency in HL-60/VCR, we assessed cell viability in the absence and presence of increasing concentrations of the CI inhibitor, rotenone ( Fig. 4B-C). Consistent with partial CI loss-of-function, HL-60/VCR cells were found to be highly sensitized to rotenone-induced cytotoxicity (Fig 3C). Taken together, these findings reveal CI deficiency as a previously unidentified bioenergetic vulnerability that accompanies VCR resistance.

Forced ceramide accumulation in wt cells phenocopies respiratory complex insufficiency observed with VCR
resistance.
Pharmacological inhibition of AC, sufficient to increase cellular ceramide molecular species, was recently demonstrated to decrease mitochondrial respiratory competence in drug resistant AML [7]. Interestingly, respiratory limitations were independent of both carbon substrate and respiratory stimulus and were observed despite no evidence of mitochondrial outer membrane permeabilization. Such findings suggested the intriguing possibility that cellular ceramides may be capable of directly regulating mitochondrial ETS function. In the present 13 study, VCR resistance was associated with substantial increases in a variety of ceramide molecular species (Fig. 1G). Based on this, we hypothesized that CI deficiency may be consequential to ceramide accumulation in HL-60/VCR. To determine whether elevated ceramide levels seen in HL-60/VCR cells infringe on mitochondrial function, we grew HL-60 wild-type cells for extended periods (7-days) in medium containing either C6-ceramide (C6; 10µM) or ceramide delivered via nanoliposome (CNL; 10µM), both sources for the production of long-chain ceramides [33,34]. Following 7-days of exposure, mitochondrial OXPHOS kinetics in permeabilized cells was assessed. Despite no change respiration in the presence of Pyr/M in the absence of adenylate stimulation ( Fig.   5A; 'Pyr/M'), stimulation of near-maximal OXPHOS flux with the CK clamp revealed a nearly 2-fold decrease in respiratory flux in cells exposed to both C6 and CNL ( Fig. 5A; 'ΔGATP'). The addition of cytochrome C did not impact respiration, thus confirming functional integrity of the outer mitochondrial membrane across all groups ( Fig. 5A; 'Cyt C'). Relative to vehicle control, respiration remained lower in C6/CNL exposed cells upon addition of glutamate, as well as succinate, following rotenone ( Fig. 5A; 'Succ). Similar differences in respiration regardless of the carbon substrate strongly suggests that the respiratory impairments induced by C6/CNL are most likely attributable to ETS dysregulation. Such findings are highly reminiscent to those observed in HL-60/VCR and thus support a model whereby mitochondrial dysregulation, primarily localized to the ETS (e.g., CI), may be secondary to altered SL metabolism.
Remarkably, despite no change in viability for any of the single agents, 48hr exposure to both CNL and phenformin led to a dramatic (>80% of control) decrease in cell viability (Fig. 5B). To furth the therapeutic efficacy of targeting mitochondrial CI alongside SL metabolism, experiments were designed to determine the ability of metformin or phenformin to boost cytotoxicity of other agents that can act to increase ceramide levels (inhibitors of SL metabolism). Except for SK1-i, minimal cytotoxicity was observed in HL-60/VCR cells in the presence of the indicated doses of SACLAC, PDMP, metformin or phenformin alone (Fig. 5C-E). However, the combination of each SL enzyme inhibitor with either metformin or phenformin boosted cytotoxicity, relative to vehicle or single agent conditions (Fig. 5C-E). This relationship was particularly evident for the AC inhibitor SACLAC, where relative to the single agent conditions, the combination of SACLAC and phenformin led to a near 70% decrease in cell viability after only 24 hours (Fig. 5C). Together, these data raise the intriguing possibility that co-targeting of cellular sphingolipid metabolism and mitochondrial metabolism, specifically at CI, may represent a novel treatment strategy in VCR resistance.

Discussion
A number of studies have been conducted on the relationship of SL's and SL metabolism to chemotherapy resistance, and results have been captioned in excellent works, including reviews [21,42,43].
Gouaze-Andersson and Cabot [44] have reviewed the subject with specific regard to hematological malignancies, and recent work by Snider et al. discloses multiple actions of doxorubicin on SL metabolism in MCF-7 breast cancer cells [45], including increased levels of S1P and C16 ceramide. In agreement with recent observations in daunorubicin and cytarabine resistant AML [29], VCR resistance herein was accompanied by substantial alterations in SL composition. The novelty of the present findings stems from our focus on VCR resistance and its impact on both SL's and mitochondrial function. Moreover, given our recent findings linking SL metabolism to mitochondrial bioenergetics [7,29], we hypothesized that developing a detailed understanding of the mitochondrial perturbations that accompany the drug-resistant phenotype may reveal new opportunities for targeted therapies that regulate cancer growth. To this end, the present study leveraged a recently described mitochondrial phenotyping platform [46], specifically designed to comprehensively evaluate OXPHOS efficiency and capacity, relative to the underlying mitochondrial proteome. Using this platform, we made the exciting discovery that VCR resistance is accompanied by intrinsic mitochondrial remodeling, typified by specific limitations in respiratory CI. These changes were subsequently found to sensitize VCR resistant cells to combinatorial treatment directed at increasing cellular ceramides and further decreasing CI flux within the mitochondria, thus offering an exciting, potential treatment strategy for combatting VCR resistance in the clinic.
With respect to SL metabolism, VCR resistance was accompanied by upregulated expression of SPHK1, GCS, and AC, in addition to the concomitant and expected upregulation of P-gp, a predominant player in MDR [10]. Interestingly, the changes observed in enzyme expression were reflected in the results from lipidomic analyses. For example, increased expression of SPHK1 in HL-60/VCR cells was mirrored by elevated levels of S1P. Overexpression of AC, as noted in HL-60/VCR cells, can also contribute to elevated S1P, as AC liberates sphingosine that is converted to S1P by SPHK1. With regard to ceramide, it is possible that elevated ceramide levels in HL-60/VCR cells resulted from recycling of sphingosine via the salvage pathway of ceramide synthesis [47]. Interestingly, however, increases in cellular ceramides were countered in part by conversion to GC, a reflection of elevated levels of GCS. High GCS expression and elevated levels of GC have previously been linked to MDR [24,[48][49][50]. As ceramides induce apoptosis, our data suggest that VCR resistant cells keep levels of ceramide "in check" via the combined activities of GCS and AC. This is supported by work of Tan et al [51] and Morad et al [52] who showed that the employ of AC and GCS inhibitors, respectively, in HL-60/VCR cells, increased ceramide levels. Interestingly, even though HL-60/VCR cells contained high levels of S1P, their growth rate was considerably lower than HL-60 cells. This biology could be related to S1P receptor status in HL-60/VCR cells, although this has not been evaluated. We also detected reductions in C18, C20, C22, and C24 molecular species of SM in HL-60/VCR cells, indicative of high sphingomyelinase activity that would contribute to increased ceramide levels. However, only some minor increases in C18, C20, C22, and C24:0 species were detected in the ceramides or in GC. This discrepancy may be a product of altered SL flux.
gp and GCS in breast cancer and epidermoid carcinoma cells selected for resistance to natural product chemotherapy drugs [24]. Thus, the association of GCS and GC levels with drug resistance [53][54][55] has solid footing. Of further relevance, Shabbits and Mayer [56] have shown that ceramide sensitivity in breast cancer cells is modulated by P-gp, and using vincristine as a ceramide generator, Olshefski and Ladisch [6] have shown that GCS inhibition boosts vincristine efficacy.
In addition to changes in SL metabolism, VCR resistance in the present study was accompanied by extensive mitochondrial remodeling. In comparison to wt cells, HL-60/VCR presented with increases in basal respiration, as well as higher maximal uncoupled oxygen consumption, the latter suggestive of higher mitochondrial content. Increased reliance on mitochondrial metabolism is a commonly emerging theme in cancer drug resistance [2,3]. Although such findings are often interpreted to reflect an increased reliance on oxidative ATP synthesis, it is critical to point out that a large collection of these findings are based solely on measurements of intact cellular respirometry. As demonstrated herein, such methodologies are largely incapable of identifying nuanced bioenergetic inefficiencies across the mitochondrial network in cancer. As such, too much reliance on macro-level changes in cellular bioenergetics (i.e., binary assessments of mitochondrial versus glycolytic 'reliance') may in fact be masking a 'goldmine' of cancer-specific mitochondrial biology that if known could potentially be leveraged to develop highly efficacious targeted therapies with a favorable therapeutic index.
Results from experiments in Fig. 5 highlight the integration between SL's and mitochondrial function and suggest that this link can be exploited, therapeutically. For example, with ceramide levels already high in HL-60/VCR cells compared to wild-type (see Fig. 1G), the inclusion of either SACLAC or PDMP would be expected to drive ceramide to cytotoxic levels [29,36,57,58]. SK1-i could also backup ceramide levels while limiting S1-P's mitogenic effects. Interestingly, consequent to baseline disruptions in CI expression, the addition of CI inhibitors metformin or phenformin (see Fig. 4B-E) to the mix likely add "insult to injury", as amply demonstrated by the synergistic, cytotoxic outcome, most strikingly regarding the CNL-phenformin combination and the SACLAC-phenformin regimen. Moreover, it is noteworthy that we were able to mimic infringement of mitochondrial function typified in HL-60/VCR cells, by cell supplementation with either C6 or CNL.

Conclusions
In conclusion, increases in SPHK1, GCS, and AC expression that accompany VCR resistance serve to support cancer cell growth, as the coordinated action of these enzymes may limit the full cancer suppressor actions of ceramide while boosting mitogenicity and leukemic dissemination. Although levels of several ceramide molecular species increased with VCR resistance, a change perhaps favoring apoptosis, it is clear from the lipidomic profiles that VCR-resistant cells endeavor to keep ceramide levels in-check by increasing several GC and SM molecular species. Ceramides may also be of benefit in induction of autophagy for survival [59]. These factors, in conjunction with the intrinsic mitochondrial remodeling allied with VCR resistance, promote a resilient cancer phenotype, but also reveal(s) opportunities for design of novel, targeted therapeutics [21,[60][61][62]. Our

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryTable1.xlsx