Primary metabolism and medium-chain fatty acid alterations precede

Subscriber access provided by MT SINAI SCH OF MED

Article

Primary metabolism and medium-chain fatty acid alterations precede long-chain fatty acid changes impacting neutral lipid metabolism in response to an anti-cancer lysophosphatidylcholine analogue in yeast Nicolas P Tambellini, Vanina Zaremberg, Saikumari Krishnaiah, Raymond Joseph Turner, and Aalim M. Weljie J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00430 • Publication Date (Web): 29 Aug 2017 Downloaded from http://pubs.acs.org on August 29, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Primary metabolism and medium-chain fatty acid alterations precede long-chain fatty acid changes impacting neutral lipid metabolism in response to an anti-cancer lysophosphatidylcholine analogue in yeast Nicolas P. Tambellini1,2, Vanina Zaremberg1,*, Saikumari Krishnaiah3, Raymond J. Turner1 and Aalim M. Weljie1,2,3,* From the 1Department of Biological Sciences, University of Calgary, Calgary, Alberta T2N 1N4, Canada, the 2Metabolomics Research Centre, University of Calgary, Calgary, Alberta T2N 1N4, Canada, and the 3 Department of Systems Pharmacology and Translational Therapeutics, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania 19104-5158, United States of America. *Co-corresponding authors, [email protected] +1-403-220-4298, [email protected] +1-215-9394945

ACS Paragon Plus Environment

1


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 30

Abstract The non-metabolizable lysophosphatidylcholine (LysoPC) analogue edelfosine is the prototype of a class of compounds being investigated for their potential as selective chemotherapeutic agents. Edelfosine targets membranes, disturbing cellular homeostasis. Is not clear at this point how membrane alterations are communicated between intracellular compartments leading to growth inhibition and eventual cell death. In the present study, a combined metabolomics/lipidomics approach for the unbiased identification of metabolic pathways altered in yeast treated with sub-lethal concentrations of the LysoPC analogue was employed. Mass spectrometry of polar metabolites, fatty acids and lipidomic profiling was used to study the effects of edelfosine on yeast metabolism. Amino acid and sugar metabolism, the Krebs cycle, and fatty acid profiles were most disrupted, with polar metabolites and short-medium chain fatty acid changes preceding long and very long-chain fatty acid variations. Initial increase in metabolites like trehalose, proline and gamma-amino butyric acid with a concomitant decrease in metabolites of the Krebs cycle, citrate and fumarate, are interpreted as a cellular attempt to offset oxidative stress in response to mitochondrial dysfunction induced by the treatment. Notably, alanine, inositol and myristoleic acid showed a steady increase during the period analyzed (2,4 and 6 hours after treatment). Of importance was the finding that edelfosine induced significant alterations in neutral glycerolipid metabolism resulting in a significant increase in the signalling lipid diacylglycerol.

Keywords Anticancer drug, Cell metabolism, Drug metabolism, Fatty acid metabolism, Lipid metabolism, Lipidomics, Mass spectrometry (MS), Metabolomics, Multivariate Statistical Analysis, Saccharomyces cerevisiae.


2

Page 3 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Introduction Selective targeting of cellular membranes provides a non-mutagenic alternative to more traditional chemotherapeutic strategies targeting DNA or the cytoskeleton. Synthetic ether-linked lysophosphatidylcholine (LysoPC) analogues have emerged as compounds that could be used as chemotherapeutic agents and are thought to act by targeting the cell membrane1-4. Many of these lipids have demonstrated selective antitumour activities in vitro and in vivo2,5 as well as antiparasitic activity6,7. The alkyl-lysophospholipid 1-O-octadecyl-2-O-methyl-rac-glycero-3-phosphocholine, commonly known as edelfosine, is a prototype of this group of lipids. Edelfosine is a LysoPC analog that has been shown to selectively induce apoptosis in tumour cells, and this has been linked to the ability of these cells to uptake the lipid drug4,8. Genetic studies performed in the model organism Saccharomyces cerevisiae identified a conserved flippase that facilitates internalization of edelfosine (9) 9. Localization studies using fluorescent edelfosine derivatives have shown that edelfosine accumulates in the endoplasmic reticulum (ER) once inside the cell10,11, leading to ER stress12 and inhibition of phosphatidylcholine synthesis through the Kennedy pathway13. In addition, a number of surface receptors, phospholipases and signalling pathways have been proposed as cellular targets of edelfosine14. These disparate effects of edelfosine can be reconciled by the fact that edelfosine alters membrane architecture and its biophysical properties, impacting on a wide range of membrane associated proteins. Unbiased genetic studies conducted in budding yeast support this view, where choline-containing lysolipid analogues induce biophysical modifications of the plasma membrane15 leading to alteration of pH homeostasis16. In addition, defects in vesicle trafficking pathways resulted in resistance to edelfosine, by allowing the plasma membrane architecture to be repaired11. A highly enriched cluster of genes related to mitochondria function was also identified in these studies, but the role of this organelle in modulating sensitivity to edelfosine is not yet clear11. Interestingly, a metabolic flux study using 13C-labeled glucose in cancer cells17, was able to identify progressive alterations of the TCA cycle induced by edelfosine, leading to ROS-induced apoptosis. Altogether these results point to a mechanism by which changes in cellular membranes by edelfosine lead to a shift in metabolism involving mitochondria, changes in the TCA cycle and oxidative stress. A gap remains however in the understanding of how cellular membranes and organelles communicate and what kind of checkpoints operate in order to trigger the appropriate response upon sensing alterations in membranes. The combined use of unbiased approaches from the omics fields was favored in this study to elucidate the temporal changes induced by edelfosine and to identify pathways and metabolites involved in the cellular response to the treatment. Metabolomics allows for the metabolic status to be monitored and gives a snapshot of the processes occurring in a cell. As metabolomics has emerged as a viable tool it has been used extensively in drug discovery and development in addition to many other fields18. Recent advances in metabolomics technologies and data processing have allowed for studies that encompass the global cellular metabolism in contrast to previous studies that targeted specific metabolite classes or metabolic pathways. This untargeted approach has recently been used with great success in our labs to study the effects of metal toxicity on bacteria19,20 and cancer hypoxia21. Metabolomic studies have been successfully used to uncover changes in the fatty acid and lipid profile of yeast cells during different


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 30

phases of growth22 or caused by defective lipid biosynthesis23, as well as the response to furfural, phenol, acetic acid24 and copper25. Here a combined metabolomics and lipidomics study is reported identifying comprehensive metabolic changes induced by edelfosine in the model organism Saccharomyces cerevisiae (budding yeast). Yeast is a powerful model for the identification of pathways involved in drug action26-29 and has been extensively used as an experimental system to investigate the mode of action of edelfosine and other medically relevant lysophosphatidylcholine analogues11,15,16,30-32. Yeast is susceptible to edelfosine at similar concentrations to those used to induce apoptosis in tumoral cells15. One additional advantage of using budding yeast for this study is that it possesses a relatively simple lipidome with the main fatty acid species consisting of 16:0, 16:1, 18:0 and 18:123. Yeast has been at the vanguard of the development of functional genomic, transcriptomic, proteomic and metabolomic approaches and currently represents the organism with the most comprehensive and well curated experimental dataset available to the research community33,34. New levels of understanding of the physiological roles of lipids and their remodeling as part of the cell response to drugs and environmental changes is emerging as consequence of the use of these systems biology endeavors. In this work, untargeted profiling of the polar metabolites, targeted fatty acid analysis and untargeted lipidomic profiling of yeast treated with edelfosine at concentrations that induce a cytostatic effect identifies perturbations induced by the treatment through comparison to untreated yeast.

A surprisingly wide-scale metabolic perturbation is identified that suggest a metabolic switch in central carbon metabolism and alterations of neutral lipid reservoirs as part of a cellular attempt to revert the lysolipid burden, which reduces proliferative ability. Integration of the identified metabolic pathways with the results of previous chemogenomic screens11,16 point to mitochondrial dysfunction and oxidative stress as critical mediators of the cellular response to edelfosine. The results open new hypotheses on how cells sense membrane alterations and how different cellular compartments communicate in order to trigger the appropriate response, with diacylglycerol arising as an important signalling lipid in this process.

Experimental Procedures Yeast and Edelfosine Growth Curves Yeast strain BY4741 (MATa his3 leu2 met15 ura3) was obtained from Euroscarf (Frankfurt, Germany) and grown in 50mL cultures in liquid synthetic medium containing 0.67% Yeast Nitrogen Base with ammonium sulphate (MP Biomedical, Solon OH, USA) 2% glucose and histidine, leucine, methionine and uracil to fulfill auxotrophies, as described previously35. Two biological replicates were started to generate technical replicates, and this was repeated twice. Each culture was grown to an initial A600 of 0.2 from a start point of approximately 0.05 and edelfosine was then added. Edelfosine was dissolved in anhydrous ethanol and added at the indicated concentrations while 0.2% (v/v) ethanol only was added


4

Page 5 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


to the control culture. A600 readings were taken every 90 minutes after edelfosine addition for the indicated times. A final reading was taken at 32 hours to determine if recovery of the culture had occurred. Two sets of triplicate readings were taken for each concentration of edelfosine and the control culture. The mean and standard error for each concentration were then graphed using GraphPad Prism 3.03 software (La Jolla, CA, USA).

Yeast Sample Growth and Sample Harvesting Wild type strain BY4741 was grown in 1.75 L cultures to log phase (A600 of 0.2). At this point the culture was split and 2 μg/ml edelfosine or vehicle (0.2% v/v anhydrous ethanol) was added to each flask (treated and control culture respectively). Cells (10 optical density units as 600 nm) were harvested in multiples of six at each time point of 0, 2, 4 and 6 hours after edelfosine addition from both the edelfosine treated and untreated culture. Each harvested sample was washed twice with water to remove all growth media, flash frozen in liquid nitrogen to prevent further growth and metabolite turnover and stored at −80 °C unLl extracLon.

Sample Extraction and Derivitization A modified version of a cell extraction method previously described by McCombie et al.36 and described in detail by Tambellini et al.35 was used. Briefly, yeast cell pellets were re-suspended in CHCl3/CH3OH (1:2, v/v) and cells disrupted with acid-washed glass beads at 4 °C using a BioSpec Multi-Bead Beater (Bartlesville, OK, USA). Next CHCl3 and H2O were added to facilitate phase separation and extraction of lipids and polar metabolites. Seventy-five microliters (75μl) aliquots from the organic phase of the edelfosine treated and untreated samples were then transferred to 1.5mL microcentrifuge tubes after extraction had occurred. Isolated organic fractions containing the lipids and fatty acids and aqueous fractions containing the polar metabolites were dried down in a fume hood or speed vacuum (Eppendorf 5415 C, Hamburg, Germany) and stored at -80°C. Aqueous samples were prepared for GC-MS analysis by derivatization with methoxyamine and MSTFA (N-methyl-N-(trimethylsilyl) trifluoroacetamide) as previously described19. Briefly 50 μL of 20 mg/mL solution of methoxylamine-hydrochloride in pyridine was added to each dry sample with shaking at 37°C for 2.5 h. After the addition of 50 μL of MSTFA, 45 min of additional shaking at 37 °C followed. Organic samples were prepared for GC-MS fatty acid methyl ester (FAME) analysis by derivitization with BF3/methanol as previously described36. Briefly the dried down organic fractions were dissolved in 750 μL of 1:1 (CHCl3/CH3OH) under sonication for 15 minutes. 125 μL of BF3/methanol was then added and the samples were incubated in glass vials at 80 °C for 90 min. After cooling, 300 μL of H2O and 600 μL of hexane were added to each sample and phase separation was allowed. The organic layer was isolated and evaporated to dryness overnight in a fume hood.


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 30

GC-MS Data Acquisition GC-MS acquisition was carried out as previously described35 using a Waters Micromass GCT Premier GCTOF-MS (Waters, Mississauga Ontario, Canada) coupled to a 7683 B Series Injector and Autosampler (Agilent Technologies, Mississauga Ontario, Canada). A DB-5MS 30 m × 0.25 mm column with a 0.25 μm filament size (Agilent Technologies, Mississauga Ontario, Canada) was used for polar metabolite analysis and a DB-23 60 m × 0.25 mm column with a 0.15 μm filament size (Agilent Technologies, Mississauga Ontario, Canada) was used for FAME (fatty acid methyl ester) analysis.

UPLC-QTOF-MS Based Data Acquisition for Untargeted Lipidomic Profiling: Four-hour time point edelfosine treated and untreated samples were thawed on ice. Lipids were extracted by a 2:1 (CH3OH /CHCl3) mixture (300μl), then vortexed and sonicated for 15mins. Chloroform and water were added (100μl each) and the samples vortexed. Organic and aqueous layers were separated by centrifugation for 7mins at 10,000g at 4°C. Internal lipid standards were added to the organic layer and the samples dried in a fume hood. All lipid standards PC (17:0/14:1), PC (17:0/20:4), PC (21:0/22:6), Lyso PC (13:0/0:0), Lyso PC (17:1/0:0), PE (17:0/14:1), D5-16:1 DG, D5-18:1 DG, D5-18:2 DG were purchased from Avanti Polar Lipids (Alabaster, AL, USA). The dried lipid samples were re-dissolved in a mixture of 60% solvent A (40% H2O, 60% ACN, 10mM Ammonium formate) and 40% solvent B (90% IPA, 10% ACN, 10mM Ammonium formate). The samples were centrifuged at 10000g for 5min to remove fine particulates. The supernatant was transferred to a glass vial for Ultraperformance liquid chromatography on qTOF Xevo G2S (Waters Corporation, Milford, MA, USA) for high-throughput LC-MS lipidomics. For each sample, 10μl was injected on to a reverse-phase HSS T3 1.8μm 2.1x150mm column using an Aquity H-class UPLC system (Waters Corporation, Milford, MA, USA). Each sample was resolved for 20min, at a flow rate of 0.3ml/min. The UPLC gradient consisted of 75% solvent A and 25% solvent B for 1 min, a ramp of 50% solvent A and 50% solvent B for 3 mins, followed by a ramp of 10% solvent A and 90% solvent B for 8 mins, and a final ramp to 6% solvent A and 94% solvent B for 8 mins. The column eluent was introduced directly into the mass spectrometer. Mass spectrometry was performed on a quadrupole time-of-flight instrument (Xevo G2-S, Waters, Milford, MA) operating in positive ion, sensitive mode with a capillary voltage of 3000V and sampling cone temperature of 40°C. The desolvation gas flow was set to 800 L/hr and the temperature was set to 450°C. The source temperature was set at 80°C. Accurate mass was maintained by introduction of a lock-spray interface of leucineenkephalin (556.2771m/z) at a concentration of 0.5ng/μl in 50% aqueous acetonitrile and a rate of 5μl/min. Data was acquired in centroid MSe mode from 50-1200m/z mass ranges for both MS and MSe modes. Low energy or fragmented data was collected without collision energy, while high energy or fragmented data collected by using collision energy ramp from 15-40eV. The entire set of triplicate sample injections was bracketed with a test mix of standard metabolites at the beginning and at the end of the run to evaluate instrument performance with respect to sensitivity and mass accuracy. The


6

Page 7 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


sample queue was randomized to remove bias. Lipid analysis and identifications were done using Progenesis QI software (Waters, Milford, MA).

GC-MS Data Processing and Multivariate Statistical Projection and Analysis Raw GC-MS data from polar metabolite analysis was imported to MetaboliteDetector37 for peak detection and identification using an untargeted approach. Peak detection and identification for FAME analysis was done with AMDIS/MetIdea using a targeted approach with a 37 FAME standard (Supelco, Bellefonte, PA, USA). After peak detection and identification the data were normalized using Excel 2010 (Microsoft, Redmond, WA, USA) first to the internal standard, D-25 Tridecanoic Acid, in the case of FAME profiling, followed by integral normalization for both the FAME and the polar metabolite profiling. After normalization and peak detection, univariate scaling and mean centering were applied before the model construction and validation steps and was all done using SIMCA-P (Version 13) (MKS Umetrics, Umeå, Sweden) a multivariate statistical analytical software. To avoid over fitting of the data, the autofit routine of SIMCAP was used during model construction and for seven-fold cross-validation of the OPLS-DA (orthogonal partial least squares-discriminant analysis) models and CV-ANOVA (cross validated-analysis of variance) p-values were used.

Data analysis and multivariate projection modelling Raw LC-MS data files were converted to mzXML format using masswolf. The data from 12 edelfosine treated and 10 untreated yeast samples was uploaded to XC-MS online38. Analysis and peak detection was then carried out using the HPLC/Waters TOF parameters built into the server with [M+H], [M+NH4], [M+Na], [M+H-H20], [M+K], [M+ACN+H] and [M+ACN+Na] adducts detected. Upon initial examination of the resulting data, it was observed that edelfosine treated samples had approximately 5fold higher total ion counts than untreated samples for an unknown reason even though the internal standard concentrations were verified to be statistically the same using a t-test. FIGURE S1 is a cloudplot demonstrating the extent of this disparity. As a result the data was exported out of XC-MS online and normalized using total integral normalization to allow for compositional comparison of treated and untreated samples using multivariate projection modelling and statistical analysis in SIMCA-P13 (Umetrics AB, Umeå Sweden). Edelfosine (523.73 g/mol) and presumed downstream metabolite adducts and isotope peaks were identified and removed before normalization as they could affect the normalization and modelling steps. These peaks were determined by a series of criteria including similar mass and retention times, as well as detection only in the treated samples through the first phase of XCMS peak detection, and fold changes > 10 between the treated and untreated samples following the peakfilling procedure. The complete list of 51 removed peaks is provided in TABLE S1.


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 30

After import into SIMCA-P Version 13, mean centering and pareto scaling were applied to the data followed by PCA and OPLS-DA modelling. Additionally an S-plot was constructed to identify significantly increased or decreased lipids as the use of VIP (variable influence on projection) scores for screening is not ideal with a high number of variables, in this case thousands. S-plots combine the modelled variance and modelled correlation from the OPLS-DA plot using the magnitude of each variable (p) and reliability of each variable (p(corr)) to screen out metabolite peaks that have low magnitudes of change or intensities39. Waters Progenesis software (Waters Corporation, Milford, MA, USA) was used to identify the lipid features found to be noticeably increased or decreased by edelfosine treatment. Metabolite Modelling and Pathway Analysis Polar metabolites and fatty acids identified to have VIP scores greater than one through OPLSDA modelling of edelfosine treated and untreated yeast cells were subjected to pathway analysis using the MetaboAnalyst 3.040 server. Pathway analysis was done with 17 polar metabolites and 8 fatty acids using the S. cerevisiae pathway library and hypergeometric test and relative-betweeness centrality algorithms for the over representation analysis and pathway topology analysis portions respectively.

Results Edelfosine has a rapid impact on the polar metabolome at cytostatic concentrations In order to minimize biological variation introduced through cell death, a cytostatic concentration of edelfosine was determined (FIGURE 1). Dose-dependent growth of wild type yeast exposed to edelfosine was recorded, and 2 μg/mL of edelfosine achieved the desired effect as evidenced by the halting of growth during the first six hours followed by culture recovery overnight. Higher concentrations did not recover, and as such were deemed overly cytotoxic for the purpose of this study. Based on the known effects of edelfosine on ER fuction and lipid interaction, the metabolome and fatty acid composition of the model organism S. cerevisiae was assessed using comparative metabolomics and lipidomics. Yeast treated with the sub-lethal concentration of edelfosine was compared to vehicle control yeast at 0, 2, 4 and 6 hours. Polar metabolites were extracted and analyzed using GC-MS, followed by orthogonal partial least squares-discriminant analysis (OPLS-DA) to elucidate global patterns of change (TABLE 1) as OPLS-DA is a pattern recognition approach. The OPLS-DA models for polar metabolites show a significant difference between the treated and untreated samples at 2, 4, and 6 hours after the addition of edelfosine as evidenced by the predictive ability (Q2 >0.4) and low CVANOVA (cross-validated analysis of variance) p-values < 0.01. Conversely, the low Q2 value at time 0 hours (Q2 = 0.005) demonstrates that there is no predictive value of the model and suggests little difference between the treated and untreated samples. As samples were obtained immediately after addition of edelfosine (time 0 hours) it would be expected that the drug would not instantaneously induce measurable changes in metabolism. The Q2 value of 0.464 for the 6 hour condition was relatively lower than other other; this time point falls just after two doubling cycles of growth for untreated yeast


8

Page 9 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


while lack of proliferation with maintenance of viability was observed for the drug treated cultures32 (Figure 1). As such, this is interpreted to mean that the polar metabolome undergoes a rapid change upon exposure to edelfosine between 2 and 4 hours compared to control, but then both conditions stabilize, possibly due to nutrient depletion. Alterations in fatty acids lag the polar metabolome Fatty acid profiling using FAME (fatty acid methyl ester) GC-MS analysis revealed significant separation between treated and untreated samples 4 and 6 hours after edelfosine addition as evidenced by the Q2 > 0.6 and significant CV-ANOVA p-value 0.4) obtained at 0 and 2 hours after treatment suggest there is little multivariate difference in the fatty acid profiles of treated and untreated samples until 4 hours post-treatment. The significant separation of untreated and edelfosine treated samples observed at the 4 and 6 hours time points for FAME analysis contrasted to that of polar metabolite profiling which showed significant separation already at the 2 hours time point. This suggests a temporal component to the metabolic reaction to treatment. Lipidomics analysis reveals extensive perturbations 4 hours post-treatment Examining the polar and fatty acid alterations holistically, it was determined that the 4 hour time point best represented the state of maximum overall difference between the control and treated situations. Based on the importance of this time point, it was reasoned that an in-depth lipidomics screen would reveal changes in intact lipids. The results noted here are compositional changes following normalization as noted in the methods section as edelfosine treatment resulted in systematically elevated ion counts (FIGURE S1). PCA modelling of the normalized data showed clear separation of edelfosine treated and untreated samples with R2 = 0.776 and Q2 = 0.544 values (FIGURE 2A). OPLS-DA modelling confirmed this separation of with strong model parameters of R2(X) = 0.722, R2(Y) = 0.963, Q2 = 0.874 and CV-ANOVA pvalue = 1.43 e-5 (Figure 2B). An S-plot was constructed to identify features that were noticeably increased or decreased by edelfosine treatment in yeast but also were of highest amplitude to be reliable indicators (FIGURE 2C). Thirty features that decreased and 71 that increased by edelfosine treatment were identified as having both high amplitude and significance, highlighted in red in Figure 2C. A total of 30 features were identified from these significant ones, 24 elevated and 6 reduced upon treatment as described in Table S2. These features represented a total of 26 different lipids from major lipid classes. The major lipid classes found to be affected by edelfosine consist of monoacylglycerols (MAG, n=3), diacylglycerols (DAG, n=8), triacylglycerols (TAG, n=5), phosphatidylcholines (PC, n=1), lysophosphatidylcholines (LPC, n=2), phosphatidylethanolamines (PE, n=1), lysophosphatidylethanolamines (LPE, n=2) and phosphatidylinositols (PI, n=4). Polar metabolites, fatty acids and lipids found to be altered by edelfosine treatment cluster into specific metabolic pathways Based on the multivariate profiling (Figure 2), a list of metabolites and fatty acids (FAs) significantly contributing to the separation between the edelfosine treated and untreated samples was identified (TABLE S3). This was based on OPLS-DA modelling of 2 and 4 hours time points of polar


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 30

metabolites and the 4 and 6 hours time points from the FAME profiling using a cutoff of VIP (variable influence on the projection) scores >1, with a higher score indicating a greater influence on the separation of the two sample groups. These time points displayed Q2 >0.5 and CV-ANOVA p-values < 0.01, indicating significant separation between the untreated and edelfosine treated sample groups. In total, 22 different polar metabolites from the 2 and 4 hours time points and 8 different fatty acids from the 4 and 6 hours time points were identified to have statistically significant changes in the treated samples when compared to the untreated samples. Model coefficients obtained from the OPLS-DA modelling were used to identify if the levels were increased or decreased in the edelfosine treated samples compared to the untreated samples (TABLE S3). Pathway analysis was then carried out using the 17 polar metabolites and 8 fatty acids identified to be significantly perturbed at least at one time point during the edelfosine treatment. Four metabolic pathways were found to be enriched with the criteria of having an impact score > 0.1 and raw p-values < 0.05 as determined by the MetaboAnalyst 3.0 software40 using the yeast specific metabolite background set (FIGURE 3). These pathways included: alanine, aspartate and glutamate metabolism; arginine and proline metabolism; glutathione metabolism; and TCA cycle metabolism. Temporal delineation of edelfosine impact on metabolism In order to better understand the specific temporal differences underlying the multivariate differences observed in the pattern recognition approach, specific differences between polar metabolites and fatty acid species in response to edelfosine was investigated (FIGURE 4). Medium chain C10:0, C12:0 as well as the long chain C14:0 and C14:1 fatty acids showed an initial response to edelfosine treatment at 2 hours that continued to strengthen through the 6 hours period, while longer chain C16:0, C20:0 and very long chain FAs C22:0 and C:24:0 did not change in response to edelfosine until the cells were exposed to edelfosine for 6 hours (FIGURE 4A). In contrast, polar metabolites changes in response to edelfosine treatment were already detected within 2 hours and most were maintained up to the 6 hours time point (FIGURE 4B). To further identify patterns associated with metabolic responses induced by edelfosine treatment, heatmaps were constructed for the polar metabolite (FIGURE 5A) and fatty acid (FIGURE 5B) data. Analysis of polar metabolite data showed that edelfosine treated samples clustered away from the zero time and from the untreated samples at all time points (FIGURE 5A). Similarly, clustering of the fatty acid profiling data clearly showed grouping of edelfosine treated samples at 2, 4 and 6 hour time points, with the 2 hour timepoint interestingly most similar to the 6 hour vehicle treated condition. An increase in C12:0, 14:0 and 14:1 fatty acids over time after edelfosine treatment was evident. Moreover, examination of a heatmap of p-values from rank ordered significance testing every metabolite and fatty acid at each time (FIGURE 5C) further supported the temporal partitioning of polar metabolites and medium chain fatty acid metabolism versus long and very long chain fatty acids. Analysis of the heatmap based on the lipid metabolite data (FIGURE 6) from ten replicates of control and treated samples indicated a major impact of edelfosine on the levels of neutral glycerolipids including monoglycerides (MGs), diacylglycerides (DGs) and triglycerides (TGs). Edelfosine also appears to stimulate remodelling of yeast glycerophospholipids at the expense of their corresponding lysolipid species. The most abundant yeast glycerophospholipids, phosphatidylcholine (PC) and


10

Page 11 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


phosphatidylinositol (PI), increased their degree of unsaturation. Interestingly, only one lipid was identified outside the glycerolipid class, corresponding to the sphingolipid intermediate inositolphosphorylceramide (Ins-P-Cer), which also increased. The major metabolic pathways and patterns as well as potential biological interpretations for the changes induced by edelfosine treatment are discussed below. Discussion Non-metabolizable LysoPC analogues like edelfosine target cellular membranes and disturb cellular homeostasis. Alterations in the metabolic behavior of the cell occur as a direct consequence of the perturbations experienced at the plasma membrane and organelles. Alongside, a cascade of signalling circuits is set in motion as part of a cellular program activated to restore homeostasis. Failure to do so may lead to cell death. In order to dissect the metabolic and signalling pathways affected by edelfosine, in this study a combined metabolomics and lipidomics approach was used for the unbiased identification of altered metabolism in yeast treated with the LysoPC analogue. Temporal metabolic profiles using mass spectrometry of polar metabolites and fatty acids (FAs) of yeast exposed to sublethal concentrations of edelfosine revealed that 17 polar metabolites and 8 FAs were perturbed over multiple timepoints (2, 4, and 6 hours post-exposure), identifying metabolic pathways not implicated in previous studies in addition to further supporting earlier findings obtained with tumoral cells17,41 and parasites42. From the analysis of aqueous metabolites, amino acid and glucose metabolism emerged as the pathways that were most impacted by treatment of yeast with sub-lethal concentrations of edelfosine (Figure 7). A steady increase in alanine levels was registered after exposure of cells to edelfosine for 2, 4 and 6 hours, reaching more than 20-fold change during this period. Alanine metabolism is positioned at the interphase of carbon and nitrogen metabolism and has been recently proposed as modulator of yeast life span through regulation of cytochrome c oxidase subunit 2 (COX2) expression (35) 43. Interestingly, a cluster of mutants defective in cytochrome c oxidase (cox6, cox16, cox17, cox19, cox23, mss51) was enriched in a recent chemogenomic screen for the identification of yeast deletion mutants resistant to edelfosine11. Considering the similarities between yeast and cancer cells with regard to cell energy metabolism44 and their response to edelfosine15 is interesting to note that alanine and several LysoPC species have been recently identified as plasma markers with potential diagnostic value for breast cancer45. Furthermore, alanine was the only aqueous metabolite (out of ~150) that emerged in a recent metabolomic study as a biomarker for distinguishing serous ovarian carcinoma and benign serous ovarian tumors46. In addition, alanine synthesis and glycolysis alterations have been linked to c-Mycinduced tumor formation47. Therefore, although is not clear at this point how edelfosine perturbs alanine levels, targeting of a pathway associated with cancer metabolism could be of significance to the antitumor activity displayed by this lipid drug. During the time course of 6 hours, spanning two cell divisions in control untreated cells, a rise in glucose 6-phosphate levels was also observed (Figure 4B). Glucose 6-phosphate represents a key metabolite in central carbon metabolism. It sits at the branching point between glycolysis/gluconeogenesis and the pentose phosphate pathway which provides reducing power for fatty acid synthesis and other anabolic


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 30

pathways and which is constantly adjusted to control oxidative stress and to maintain redox homeostasis. In addition, glucose 6-phosphate is a precursor for the synthesis of inositol and trehalose, which also increased in response to edelfosine in this study. While inositol registered a constant change during the entire time course analyzed, differences in trehalose, together with proline, were only observed at the 2 hours time point in the multivariate analysis (TABLE S3), but overall were generally elevated as well (Figure 3B). Trehalose is a well known oxidative stress protectant of yeast membranes and proteins48-50 while increased levels of proline and inositol were shown to confer tolerance to various stress conditions by reducing reactive oxygen species levels in yeast51,52. Another metabolite that increased with the treatment was gamma-aminobutyrate (GABA) which has also been linked to oxidative stress protection53 and thermotolerance54 in yeast. Therefore these metabolites could be part of an early response to oxidative stress induced by edelfosine which has been previously observed in yeast32 and tumoral cells55. In summary, the emerging map of pathways altered by edelfosine suggests a metabolic reprogramming that favors gluconeogenesis and glucose storage in the form of trehalose (Figure 7). In addition, pyruvate seems to be diverted towards alanine synthesis impacting also glutamate levels which significantly dropped by 6 hours of treatment. A slowdown of the TCA cycle accompanies these metabolic changes as reflected by the decrease in citrate and fumarate, possibly indicative of a deleterious effect of edelfosine on mitochondrial performance. It is interesting to note that a large enrichment of mitochondrial related genes was found in a recent yeast chemogenomic screen using cytotoxic concentrations of edelfosine11. Considering mitochondria are mostly responsible for the generation of reactive oxygen species, altogether our results on primary metabolism support the notion that edelfosine induces mitochondrial dysfunction, contributing to oxidative stress. An interesting and significant finding of our study is that eight fatty acids (FAs) were found to be different in treated and untreated samples with a delayed temporal profile as compared to the polar disruptions. The overall trend in FA perturbation points to an increase in C12-C14 FAs at the expense of C10:0 at early time points (2 and 4 hours) followed by a significant decrease in saturated long (C16:0, C20:0) and very long FAs (C22:0, C24:0) at the latest time point analyzed (6 hours). Of note is a steady increase of myristoleic acid (C14:1) across all time points. This suggests that edelfosine treatment induces acyl chain remodeling. The shortening of fatty acids and the increment in the degree of unsaturation resembles the adaptation described for yeast mutants defective in acyl-CoA-binding protein, Acb1p56,57. Furthermore, a reduction in inositol-mediated repression of UASINO containing genes has been described for acb1 mutants56. Inositol is the master regulator of glycerophospholipid biosynthesis in yeast58 making it of particular interest that myo-inositol was found to be increased in edelfosine treated compared to untreated yeast samples (TABLE S2). Insight on how edelfosine could be inducing shortening of acyl chains could be linked to the fact that yeast strains defective in genes involved in mitochondrial medium chain fatty acid synthesis (etr1, eht1, lip5) as well as in long and very long fatty acid elongation (elo2, elo3) were identified in a genetic screen for edelfosine hypersensitivity in yeast16 . It is therefore speculated that mitochondrial FA synthesis could be involved in providing medium long chain fatty acids (C12-14) for remodelling of glycerolipids. A recent study has described alterations of neutral glycerolipid metabolism in the mitochondrial FA synthesis mutant etr159. Similarly,


12

Page 13 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


a striking effect of edelfosine on the levels of MAG, DAG and TAG was detected in our untargeted lipidomic profiling herein. Diacylglycerol (DAG) is not only an intermediate in glycerolipid metabolism but is also a lipid second messenger with a complex strategic role in the regulation of biochemical networks60,61. Eleven out of 50 identified lipids perturbed by edelfosine treatment were DAG species including DAG 28:1, DAG 32:0, DAG 32:2, DAG 34:0, DAG 34:2 and DAG 36:0, which were all increased by treatment. Although an increase in DAG could be the result of the known inhibitory effect of edelfosine on the Kennedy pathway for synthesis of PC13,15, this explanation is not favored as these experiments were conducted in the absence of exogenous choline and did not show evidence of changes in PC levels other than an increase in PC 30:2 probably arising from acylation of the LysoPC pools which decreased alongside. Judging from the observed increases in several MAG (16:0, 18:1 and 24:1) and TAG (44:2, 46:2, 48:3, 50:3) species and a decrease in TAG (46:3) the increment in DAG may be the result of neutral lipid turnover and resynthesis. It is worth noting four lysolipid species with choline and ethanolamine head groups decreased while PC and PE species increased concomitantly. This reflects a cellular effort aimed at decreasing the presence of lysolipids through acylation, in response to the burden imposed by edelfosine. Neutral lipid turnover would result in an increase in fatty acid availability for such acylations. In addition, synthesis of several new PI species must be consuming PA and inositol while also increasing the synthesis of complex sphingolipid species like inositol-phospho-ceramide (IPC). Biosynthesis of IPC produces DAG, possibly contributing to the DAG increment observed. A similar effect on neutral lipid metabolism and specifically on DAG levels has been described for the treatment of tumour cell lines with the alkylphospholipid miltefosine41. DAG is a cone shaped lipid with a small area occupied by the polar head group compared to that of the apolar tails. Thus, the presence of DAG in membranes induces negative curvature and favours nonbilayer inverted phase transitions62. This is in great contrast with the positive curvature induced by inverted cone shaped lysolipids like edelfosine, so DAG could potentially correct the curvature stress introduced by edelfosine owing to its complementary shape63. Therefore, reducing the levels of LysoPC and LysoPE pools while increasing the levels of DAG could be part of a novel cellular strategy aimed at counteracting the stress inflicted by edelfosine. While our methodology was not able to conclusively identify cardiolipiin (CL) species, three accurate mass results were consistent with CLs and all three were significantly elevated with edelfosine treatment. Such changes would be consistent with mitochondrial stress and the proposed mechanism described above. A functional crosstalk between mitochondria and neutral lipid metabolism has been proposed, where oxidative damage due to mitochondrial dysfunction could be ameliorated by lipid droplets, with neutral lipids acting as a “sink” to terminate propagation of radicals64. Our experimental findings combined with those from previous genetic screens are consistent with this model. Furthermore, the transit of fatty acids through neutral lipid metabolism has been proposed as a means to rapidly regulate acyl-CoA pools and acyl chain remodeling of phospholipids in response to cellular needs65. A primary aim of this work was to identify metabolic changes triggered by edelfosine which could be used to advance our understanding of not only how this lipid induces cell death but also how cells


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 30

respond to a LysoPC burden. A possible link between DAG accumulation and mitochondrial metabolism emerging from this study, is intriguing. Further work will aim to determine the contribution of each of these pathways to the toxic properties of edelfosine and other LysoPC analogues, like the naturally occurring platelet activating factor (PAF). Recent work using yeast as model organism, has lead to the discovery that intraneuronal accumulation of PAF alone is sufficient to disrupt mitochondrial function and increase ROS in cultured human neurons, which could be relevant in development of Alzheimer’s disease30. Future research efforts will be directed at studying the molecular mechanisms by which edelfosine induces the metabolic changes observed, with focus on mitochondria and neutral lipid metabolism.

Associated Content Supporting Information FIGURE S1: Cloudplot of raw data from untargeted lipidomics analysis TABLE S1: Ions removed from the analysis presumed to be of drug or drug metabolite origin TABLE S2. Proposed lipids identified as altered by edelfosine treatment TABLE S3: Polar metabolites and fatty acids identified as significantly changed for Edelfosine treated samples compared to untreated samples

Acknowledgements: Funding for this work was provided in part by Discovery grants from the National Sciences and Engineering Research Council (NSERC) of Canada to AMW, RJT, and VZ, as well as a Discovery Accelerator Supplement to VZ and an American Cancer Society award to AMW (124268-IRG-78-002-35IRG).

Figure Legends: FIGURE 1: Yeast and edelfosine treatment growth curves. Wild type yeast strain BY4741 was to an initial OD600 (optical density at 600nm) of approximately 0.2 from a start point of approximately 0.05. Edelfosine was then added at 2, 4, 8 and 16 μg/mL. Two sets of triplicate readings were taken and the mean and standard error of the mean (SEM) for each concentration were calculated. FIGURE 2: PCA and OPLS-DA models of control (n=10) and edelfosine treated (n=12) yeast samples from lipidomic profiling. (A) PCA scores plot using cross-validated principal components (tcv) with R2X = 0.776 and Q2X = 0.544 values. (Blue= vehicle; Green=edelfosine+vehicle). PC, principal component. (B) OPLS-DA model using tcv and tocv variables from cross-validation with R2X = 0.722, R2Y = 0.963, Q2 = 0.874 and CV-ANOVA p = 1.43 x 10-5 values. LV, latent variable; OC, orthogonal component. C) S-plot of to identify lipids decreased or increased by edelfosine treatment. The magnitude of each


14

Page 15 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


variable (p) and reliability of each variable (p(corr)) are used to screen out metabolite peaks that have low magnitudes of change or intensities, with the most relevant markers highlighted in red. FIGURE 3: Schematic overview of pathways, polar metabolites, fatty acids and lipids identified to be affected by edelfosine in yeast through metabolomic and lipidomic profiling. Summary of overrepresented pathways across any timepoint; the p-value calculated from the enrichment analysis (Metaboanalyst 3.0) and the impact score is the pathway impact calculated from the pathway topology analysis. Only significantly perturbed pathways are shown, defined as having a p-value of less than 0.05 and an impact score of greater than 0.1. FIGURE 4 Examples illustrating the different temporal responses from 0 to 6 hours after edelfosine treatment observed for fatty acids and aqueous metabolites. A) Fold change of fatty acid response compared to initial conditions (time=0), with decanoic acid significantly elevated in the vehicle treated (Control, blue), C12 and C14 species elevated in the edelfosine treated (red), and only slight differences observed in other FAs at the 6 hour timepoint. Significance noted when p-values > 0.05 at non-zero timepoints (#, p

Primary metabolism and medium-chain fatty acid alterations precede

Recommend Documents