Nuclear magnetic resonance spectroscopy to identify metabolite

Subscriber access provided by UNIV AUTONOMA DE COAHUILA UADEC

Article

Nuclear magnetic resonance spectroscopy to identify metabolite biomarkers of non-responsiveness to targeted therapy in glioblastoma tumor stem cells Ingvild Comfort Hvinden, Henriette Engen Berg, Daniel Sachse, Erlend Skaga, Frøydis Sved Skottvoll, Elsa Lundanes, Cecilie J. Sandberg, Einar O. Vik-Mo, Frode Rise, and Steven Ray Wilson J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.8b00801 • Publication Date (Web): 09 Apr 2019 Downloaded from http://pubs.acs.org on April 10, 2019

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 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 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.

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 24 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

Nuclear magnetic resonance spectroscopy to identify metabolite biomarkers of non-responsiveness to targeted therapy in glioblastoma tumor stem cells Ingvild Comfort Hvindena,b, Henriette Engen Berga, Daniel Sachsea, Erlend Skaga,c,d, Frøydis Sved Skottvolla, e, Elsa Lundanesa, Cecilie J. Sandbergc, Einar O. Vik-Mc, Frode Risea, Steven Ray Wilsona, e* a Department

of Chemistry, University of Oslo, Post Box 1033, Blindern, NO-0315 Oslo,

Norway b Department

of Chemistry, University of Oxford; Chemistry Research Laboratory, 12 Mansfield Rd, Oxford OX1 3TA, United Kingdom

c Vilhelm

Magnus Laboratory of Neurosurgical Research, Institute for Surgical Research and Department of Neurosurgery, Oslo University Hospital, 4950 Nydalen, NO-0424 Oslo, Norway

d Institute

of Clinical Medicine, Faculty of Medicine, University of Oslo, Post Box 1171, Blindern NO-0318 Oslo, Norway

e Hybrid

Technology Hub-Centre of Excellence, Institute of Basic Medical Sciences, Faculty of Medicine, University of Oslo, PO Box 1112 Blindern, 0317 Oslo, Norway *Corresponding author: [email protected], +47 97010953. https://orcid.org/0000-00029755-1188

ACS Paragon Plus Environment

1

Journal of Proteome Research 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 24

Abstract Glioblastoma is the most common and malignant brain tumor, and current therapies only confer modest survival benefits. A major obstacle is our ability to monitor treatment effect on tumors. Current imaging modalities are ambiguous and repeated biopsies are not encouraged. To scout for markers of treatment response, we have used NMR spectroscopy to study the effects of a survivin inhibitor on the metabolome of primary glioblastoma cancer stem cells. Applying high resolution NMR spectroscopy (1H resonance frequency: 800.03 MHz) to just 3 million cells per sample, we achieved sensitive and high resolving determinations of e.g. amino acids, nucleosides and constituents of the citric acid cycle. For control samples that were cultured, prepared and measured at varying dates, peak area relative standard deviations were 15-20%. Analyses of unfractionated lysates were performed for straightforward compound identification with e.g. COLMAR and HMDB databases. Principal component analysis (PCA) revealed that citrate levels were clearly up-regulated in non-responsive cells, while lactate levels substantially decreased following treatment, for both responsive and non-responsive cells. Hence, lactate and citrate may be potential markers of successful drug uptake and poor response to survivin inhibitors, respectively. Our metabolomics approach provided alternative biomarker candidates compared to spectrometry-based proteomics, underlining benefits of complementary methodologies. These initial findings make a foundation for exploring in vivo MR spectroscopy (MRS) of brain tumors, as citrate and lactate are MRS-visible. Taken together, NMR metabolomics is a tool for addressing glioblastoma.

Keywords: Cancer; Glioblastoma; NMR; Metabolomics; Proteomics


2

Page 3 of 24 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 Glioblastoma (GBM) is the most common form of malignant brain cancer. Survival after diagnosis is dismal, with a median survival of 15 months. Only 5% of diagnosed patients are alive 5 years after diagnosis 1. The diagnosis of GBM is based on histopathological characteristics, as well as a limited set of molecular biomarkers 2. The clinical stratification of tumors is limited, despite an increasing description of the widespread genetic, molecular, cellular and anatomical heterogeneity both within tumors and between patients

3 4 5 6

. This

complexity indicates the need for development of personalized treatment strategies. However, the scant diagnostic stratification of patients and intratumoral heterogeneity makes suitable choices of treatment difficult to make, even with advanced molecular technologies 7. A major obstacle to effective therapy development is the limited availability of methods for evaluation of tumor response to a chosen treatment. While repeated analyses of malignant cells are possible in selected malignancies (e.g. hematological and dermatological), the risks involved in repeated brain tumor biopsies makes serial and repeated analysis of brain tumors not feasible 8. New imaging technologies with magnetic resonance imaging (MRI) can give a range of supplementary information. Perfusion and diffusion series can display intratumoral heterogeneity and characteristics, but falls short on their ability to discern between tumor growth and treatment induced changes in the tumors

9 10.

Similarly, positron emission

tomography (PET) imaging has only given modest additions in evaluating tumor response for any given treatment

11.

In vivo magnetic resonance spectroscopy (MRS) is an MRI-based

method that provides tumor-specific data without the need for invasive procedures

12 13 14.

Previous reports have tried to identify possible metabolic fingerprints of tumor response 16,

15 14

looking at standard metabolite identifications or a combination of spectral changes.

However, to establish this methodology for treatment evaluation, better biomarkers for tumor


3


Page 4 of 24

response or resistance to treatment with chemotherapeutic agents must first be established under more controlled conditions. A natural choice for establishing biomarkers in vitro may be nuclear magnetic resonance spectroscopy (NMR), which is basically the same technique as MRS 17. NMR is well suited to perform untargeted metabolomics because it requires little a priori knowledge of the chemical structure of the analytes. We have chosen to use NMR for scouting for biomarkers of treatment response with cultured cells harvested from biopsy samples, with the hope to establish biomarkers that can be determined in vivo for subsequent patients, avoiding the need for biopsies. NMR is robust, reproducible and nondestructive, but is comparatively less sensitive than mass spectrometry (MS) 18. Despite issues with sensitivity, NMR can still be a powerful tool in bioanalytical science 19. It is possible to mitigate sensitivity issues 20, and here we have done so by using a high magnetic field (18.8 Tesla), a cryogenically cooled NMR probe, and non-uniform sampling (NUS). NUS can allow for notable gains in sensitivity and/or savings in acquisition time 21-22. In this initial study, we demonstrate the feasibility of our approach, using an 800.03 MHz NMR instrument for prediction of response to the survivin inhibitor YM155 using a variety of primary glioblastoma cells enriched with glioblastoma stem cells (GSCs). We also complement results obtained with 800 MHz NMR spectroscopy with proteomics and transcriptomics analysis.

Experimental Cell culture Glioblastoma biopsies were obtained from five informed and consenting patients undergoing surgery for GBM at Oslo University Hospital, Norway, approved by The Norwegian Regional Committee for Medical Research Ethics (REK 2017/167). Histopathological diagnostics were


4

Page 5 of 24 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


performed according to the WHO classification. Biopsies were kept in ice-cold Leibowitz-15 medium (L-15, Invitrogen) until isolation. Cell cultures were prepared as previously described 23.

In short, single cells were isolated mechanically and enzymatically with trypsin-EDTA

(Thermo Fisher, Waltham, MA), blocked by 2 mg/mL human albumin (Octapharma Pharmazeutika Produktionges, Lachen, Swizerland) and washed twice in L-15 (Lonza, Basel, Swizerland). Isolated cells were counted using Countess automated cell counter (Invitrogen) and cultured as floating spheres in non-treated 175cm2 cell culture flasks (Nunc; VWR) under serum-free conditions: Dulbecco’s Modified Eagle’s Medium (DMEM; Thermo Fisher), 10 ng/mL

bFGF,

20

ng/mL

EGF

(both

from

R&D

Systems,

Minneapolis,

MI),

penicillin/streptomycin 100 U/mL (Lonza), heparin 1 ng/mL (Leo Pharma, Ballerup, DK), HEPES 8 mM (Lonza) and 1:50 B27-supplement without retinoic acid (Thermo Fisher).

Cell viability assay Cells were plated at 5000 cells/well in a 96-well plate (Sarstedt, Numbrecht, Germany) under sphere conditions, cultured for 24 h before the addition of YM155 (Selleck Chemicals, Munich, Germany) in a 6-point dose-escalating pattern, and further incubated for 72 h. Viability was assessed using Cell Proliferation Kit II XTT (Roche, Basel, Switzerland) solution, which was incubated for 24 h before absorbance was measured with a PerkinElmer EnVision plate reader. The percentage cell survival is reported relative to the negative control (DMSO, Thermo Fisher) and corrected for background. Dose-response curves were fitted based on the basis of a fourparameter sigmoidal logistic fit function by maximal and minimal cell survival, slope and inflection point (EC50). In the curve fitting, the maximal cell survival was fixed to 100%, the minimal cell survival was allowed to float between 0% and 75% and the slope between 0 and 2.5. Statistical analyses of differences in the sensitivity to YM155 were performed using ordinary one-way ANOVA corrected for multiple comparisons using Tukey’s test. Doseresponse and statistical analyses were performed using GraphPad Prism 7.0.


5


Page 6 of 24

Cell treatment and sample preparations For each experiment, 3 million cells were plated at 300,000 cells/mL under sphere forming conditions. YM155 (final concentration 1250 nM) and negative control (0.5% DMSO) were added in two different flasks, respectively. Following incubation for 24 hours, the culture media was removed by centrifugation (300xg) and cells were washed two times in distilled phosphate buffered saline (pH 7.4) and snap frozen in liquid nitrogen) and stored at -80 0C before transport to NMR facility.

NMR instrumentation and equipment An AVIII800 NMR instrument (800 MHz) with a 5 mm Triple Resonance (TCI) cryoprobe and a temperature adjustable Sample Case was used, all from Bruker (Fallanden, Switzerland). See also supporting file S1, Disposables and small equipment. An AVI600 NMR instrument (600 MHz) with a 5 mm Triple Resonance (TCI) cryoprobe and a BACS sample changer, all from Bruker.

NMR solutions (buffering, internal standard) A 1.5 M potassium phosphate monobasic buffer in deuterated water (D2O from Merck, Kenilworth, NJ) at pH 7.4 with 0.1 % trimethylsilylpropanoic acid (TSP) as an internal standard was made by dissolving 20.4 g of KH2PO4 in 80 mL of D2O. The pH was adjusted to approximately 5 by dropwise adding a strong (~12 M) solution of KOH dissolved in D2O. Once desired pH was reached, 100 mg of TSP dissolved in 10.0 mL of D2O was added to the buffer, which was mixed vigorously with a vortex mixer. The pH was further adjusted to 7.4 by adding more ~12M KOH in D2O and the volume adjusted to 100 mL with D2O. Finally, approximately 20 mg of NaN3 was added to the buffer as a bactericide.


6

Page 7 of 24 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


Preparing cell pellets for NMR analysis Each pellet was stored in separate 15 mL centrifuge tubes. The tubes were placed on ice and 600 µL type I water was added. Rapidly, pellet and liquid were transferred to Eppendorf tubes and then placed in an ultrasonic bath for further preparation, applying a variant of procedure described in reference

24,

to reduce sample heating; in the bath, the cells were subjected to

ultrasonic treatment for 30 seconds on/30 seconds off 10 times, taking a total of 10 minutes. Afterwards the Eppendorf tubes were transferred to a centrifuge cooled to 4 °C and spun for 5 minutes at 12,000 rpm, to spin down cellular debris. Before transfer to NMR tubes, 540 µL of the supernatant was moved to a fresh Eppendorf tube and mixed with 60 µL NMR buffer. An additional 40 µL of supernatant was set aside in Thermo-Tubes and stored at -80 °C pending liquid chromatography–mass spectrometry (LC-MS) analysis. For control samples and treated cells with low sensitivity towards the drug, samples analyzed contained comparable amounts of protein (580-625 µg), measured with nanodrop instrumentation (Thermo) and BCA reagent (calibration with bovine serum albumin). Before NMR analysis, the samples were either stored at 4 °C in a refrigerator or at 7 °C in the Sample Case attached to the 800 MHz NMR instrument.

NMR experiments Acquisition of one dimensional (1D) nuclear magnetic resonance spectra: All 1D spectra used for further statistical analysis were acquired with the pulse program zgesgp (this program was used to increase sensitivity, although the suppression region is broader than with noesypresat). For each sample, pulses were calibrated prior to acquisition with the script pulsecal. Suppression of the water signal was obtained with excitation sculpting. Acquisition parameters are given in S1, Table S1. Acquisition of two dimensional (2D) nuclear magnetic resonance spectra: The J resolved experiments were acquired with the pulse program jresgpprqf and the TOCSY experiments


7


Page 8 of 24

were acquired with the pulse program dipsi2esgpph. General acquisition parameters for the J resolved experiment and TOCSY experiment are given in S1, Table S2. Both experiments were acquired with non-uniform sampling (NUS). NUS data could not be processed with conventional Fourier transform algorithms. Thus, once acquired, both the J resolved and the TOCSY data were processed with the Compressed Sensing algorithm. The J resolved spectra all had the size of the real spectrum increased to 16,384 and were calibrated by placing TSP at 0.00 ppm in all spectra.

Pre-processing The 1D spectra were phased and calibrated manually in TopSpin, before being transformed to ASCII files (.txt) with the command convbin2asc. Certain regions of the NMR spectra were removed before further analysis. This included all regions above 10.0 ppm and below -0.5 ppm, as well as the range 4.5 to 4.9 ppm (residual water peak), 1.16 to 1.21 ppm and 3.64 to 3.70 ppm (ethanol peaks, traces present due to lab sterilization procedures), and 2.64 to 2.70 ppm (DMSO peak, drug solvent). Normalization was either to the integral of the TSP peak or to the integral of the total spectrum (not including the TSP peak). Unit variance was used as the scaling method; each variable was divided by its standard deviation. The spectral baselines were corrected after normalization and removal of ethanol and DMSO signals using the BarkauskasXi-Rocke algorithm from the FTICRMS package 25 26 of the statistics software R. Further preprocessing, e.g. binning/bucketing and warping, was not carried out as the samples had satisfactory peak alignments and we wished to retain as high resolution as possible. Cell extracts are not prone to the same pH and ionic strength variations as e.g. urine, and the sample pH was controlled with the 150 mM phosphate buffer.


8

Page 9 of 24 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


PCA PCA was carried out with a NIPALS (Non-linear Iterative Partial Least Squares) algorithm 27, also using the statistics software R, and validated by considering R2 and Q2 values obtained by five-fold cross-validation 28. In each case, only the first two or three principal components (PC) were meaningful before Q2 values decreased. R2 values of the first two components represent the major share of explained variation and are presented in the figure margins of the score´s plots.

LC-MS instrumentation During preparation of capillary column frits, a GC-17A Gas Chromatograph oven from Shimadzu (Kyoto, Japan) was used. Column packing was carried out with an in-house pressure bomb system, described in

29.

The LC-MS analyses were done with an Agilent 1200 series

(G1376) capillary pump with an 1100 series degasser (G1379), both from Agilent (Santa Clara, CA, USA) for the metabolomics study, and an Easy-nLC100 pump (Thermo Fisher Scientific) for the proteomics study. The Q Exactive™ Hybrid Quadrupole-Orbitrap™ mass spectrometer equipped with a “Nanospray Flex” ion source was from Thermo Fischer Scientific. See S1, Disposables and small equipment, for disposables and small equipment, e.g. silica capillary, fittings etc.

LC-MS parameters for the targeted metabolomics study The fused silica capillary column (100 µm inner diameter (ID)/360 µm outer diameter (OD)/length 25 cm, whereof 12 cm with particles) was connected to the Agilent 1200 series pump via a four-gated valve with nuts, unions, ferrules and empty capillary (30 µm ID/360 µm OD/length 10 cm). The valve had an inner loop with a volume of 500 nL. The capillary column was connected to a stainless steel (SS) emitter (20 µm ID, 40 mm length) with Upchurch PEEK fittings, sleeves and PEEK Microtight® Connector Butt. Waste was led away from the valve


9


Page 10 of 24

via red PEEK tubing (127 µm ID/1.6 mm OD/length 5 cm). The mobile phase (MP) was led from the pump to the valve via a fused silica capillary (30 µm ID/360 µm OD/length 75 cm). The MP composition was 65/35 (v/v) of acetonitrile and buffer (30 mM ammonium formate, pH 4.5), and the ratio was achieved by mixing the solvents in the same MP flask. The flow rate was kept at 0.60 µL per minute at all times. The mass spectrometer was run in negative mode. Further details on MS parameters are given in S1, Table S3.

LC-MS parameters for the comprehensive proteomic study For the proteomic study, a column switching set up was employed. The pre-column (50 µm ID/360 µm OD/length 10 cm, whereof 3 cm with particles) was connected to the pump via an empty capillary (20 µm ID/360 µm OD/length ~40 cm) with nuts, union, and ferrules, and to the analytical column through a SS tee. An empty fused silica capillary (50 µm ID/360 µm OD/length ~50 cm) was connected to the waste valve from the tee. The analytical column was connected to a SS emitter (20 µm ID, 40 mm length) with PEEK fittings, sleeves and Microtight® Connector Butt. A 120 min linear gradient from 3 to 20 % MP B was employed with a flow rate of 150 nL/min and injection volume of 5 µL. Further details on the gradient is given in S1, Table S4. The mass spectrometer was run in positive mode with full MS (m/z = 350-1850) and data dependent tandem mass spectrometry (ddMS2), additional details on MS parameters are given in S1, Table S3. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD010180.

Microarray analysis Total RNA from the cell cultures was extracted using the RNeasy Micro Kit (Qiagen, Hilden, Germany). Quality control and labeling were performed as previously described 30. The RNA


10

Page 11 of 24 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


samples were run on a HumanHT-12 chip (Illumina, Thermo Fisher). Analysis and statistics were performed using J-Express (Molmine, Bergen, Norway).

Results and Discussion Primary cell cultures from glioblastoma Glioblastoma stem cell (GSC) cultures were established from biopsies of five patients undergoing first surgery for IDH-wildtype GBM tumors. Cells were established as free-floating tumorspheres (Figure 1A). We have previously reported the ability of such cultures to be longterm serially passaged, express stem cell markers, differentiate in lineage-specific patterns, and form invasive tumors upon xenografting to severe combined immunodeficient mice, while retaining individual tumor traits 30 23. We next explored whether tumor heterogeneity is present in the sensitivity to the survivin inhibitor YM155 in patient-derived GSC cultures. The drug was tested over a concentration range covering clinically achievable plasma concentrations. Initially, we performed repeated independent experiments (n=5) of sensitivity to YM155 in three GSC cultures (T1454, T1456, T1459), and found significant differences in the total drug sensitivity (area under the curve, AUC, p

Nuclear magnetic resonance spectroscopy to identify metabolite

Recommend Documents