Experimental model and subject details
Cell culture
RKO, H508, HepG2, Huh7, HEK293T (293T), A172, U118-MG, U87, MCF-7, MDA-MB-468, and PC3 cells were obtained from ATCC. DLD1 and HCT116 cells were a generous gift from Lukas Dow. 22RV1 and was a generous gift from Dawid Nowak. 22Rv1, PC3, and H508 cells were cultured in full RPMI (Corning, Corning, NY) supplemented with 10% fetal bovine serum (FBS) (Gemini, Sacramento, CA) and 1% penicillin/streptomycin (Life Technologies, Carlsbad, CA). All of the other cells were cultured in DMEM (Corning) supplemented with 10% FBS and 1% penicillin/streptomycin (Life Technologies). HepG2 cells were grown on collagen coated plates (2 ug/cm^2). Cell lines were STR fingerprinted and/or bought from ATCC directly. Cells were tested for mycoplasma (Lonza, Basel, Switzerland).
Sugarless RPMI (Life Technologies) and DMEM (Life Technologies) were used in many experiments. Glucose (Millipore-Sigma, Burlington, MA) and fructose (St. Louis, MO) powders were diluted to 1 M stock in water before filtration. This stock solution was diluted into sugarless media.
To generate the semi-trained PC3 line, the parental cells were cultured in RPMI (Life Technologies) containing 1 mM glucose 10 mM fructose and 5% dFBS (Life Technologies). Cells were passaged approximately once per week. After >20 passages, semi-trained cells were cultured in 10 mM fructose in order to generate trained PC3 cells.
Method Details
RNA extraction, RT-qPCR, and RNA-seq
Total RNA was isolated directly from plates using the RNeasy Mini Kit (Qiagen, Hilden, Germany). For qPCR, 1.25 µg RNA was reversed transcribed using SuperScript VILO Master Mix (Thermo Fisher, Waltham, MA). Resulting cDNA was diluted 1:10 and qPCR was performed with Fast SYBR Green Mastermix (Life Technologies). The relative expression of each gene was calculated by comparative ΔCt method after normalizing to endogenous controls (Raw dCt in Table S2, Primers in Table S3). A heatmap of the results was produced using the Qlucore Omics Explorer (Qlucore, Lund, Sweden).
RNA samples from PC3 and semi-trained PC3 were submitted to the Weill Cornell Medicine Genomics Core for paired-end RNA-seq on a NovaSeq 6000. Raw sequenced reads were aligned to the mouse reference GRCm38 using STAR (v2.4.1d, 2-pass mode) aligner. Aligned reads were quantified using Cufflinks (v2.2.1) to obtain fragments per kilobase per million (FPKM). Statistical analyses on the normalized expression values (FPKM) were performed using the Qlucore Omics Explorer (Qlucore, Lund, Sweden). Gene expression levels were log2 transformed before performing PCA and differential gene expression analysis.
Genomic DNA (gDNA) extraction and qPCR
500 uL genomic lysis buffer (20 mM Tris-HCl pH 7.5, 20 mM EDTA, 1% SDS, 400 ug/mL proteinase K) was used to lyse 500,000 cells. Proteinase K was heat inactivated at 95°C for 15 minutes and allowed to cool to room temperature. Protein was precipitated with 5 M NaCl, and sample was centrifuged at 13,000 x rpm at room temperature for 10 minutes. Supernatant was poured out and pellet was washed with 1 mL 70% ethanol. Samples were centrifuged for at 13,000 x rpm for 5 minutes and supernatant was drained. Pellets were resuspended in 10 mM Tris-HCl pH 8.0. To analyze SLC2A5 copy number, qPCR was performed on 40 ng of gDNA using Fast SYBR Green Mastermix. Primers were designed to be within the same exon for SLC2A5 and B2M and can be found in Table S3.
Cell line mutation and clinical data analysis
Cell line genomic data were downloaded from the Cancer Cell Line Encyclopedia (CCLE, Broad Institute) (29) or the COSMIC (Wellcome Sanger Institute) (30) databases and cross referenced with known oncogenic mutations from COSMIC tier 1 genes (Table S1). Full list of oncogenic mutations for each cell line can be found in Supplementary File 1. Mutation and clinical data for each cell line were cross referenced with Cellosaurus (Table S1) (31).
Cell confluence, viability, and the fructolytic index
Cells were plated at low confluency in a 6- or 12-well dish. After settling, cells received a PBS wash and were given 5% dFBS, 1% penicillin-streptomycin media containing no sugar, 10 mM glucose, or 10 mM fructose. Plates were loaded into IncuCyte ZOOM Live Cell Analysis System (Essen Bioscience, Ann Arbor, MI) for imaging. 16 frames per well were analyzed at each timepoint to determine confluency. Change in confluency per hour was measured by linear regression on Prism (Graphpad, San Diego, CA).
Independent cell proliferation experiments were used to produce the fructolytic index (n = 3). It was calculated by dividing the relative growth in fructose (growth rate in fructose – growth rate in the no-sugar control) by the relative growth in glucose (growth rate in glucose – growth rate in the no-sugar control). After 3-4 days in the Incucyte system, cells were fixed with ice cold 80% methanol before. Crystal violet reagent (Sigma-Aldrich) was added to each well, and the plates were placed on a rocker for 30 minutes. Cells were then rinsed with water and imaged with a scanner.
For the competition assay, phase contrast and fluorescent images from the Incucyte system were exported as TIFF files. A custom ImageJ (Bethesda, MD) program (https://github.com/sam-taylor/CompCount) was used to acquire cell count and size. A bandpass filter, automatic threshold, and watershed algorithm were employed to distinguish cells from background. Data from the individual images were compiled into groups using MATLAB (Natick, MA) statistical software.
To measure sensitivity to drugs, cells were plated at low confluency with several replicates in a 96-well white bottom plate. The next day, powdered 2-DG (Sigma-Aldrich) was reconstituted in 10 mM glucose or 10 mM fructose media to make 100 mM 2-DG stock, which was then serially diluted. Cells were washed with PBS and were given 5% dFBS, 1% penicillin-streptomycin media containing either 100 µL of 10 mM glucose or 10 mM fructose media containing serially diluted 2-DG. Cell viability was measured after 72 hours using Cell Titer Glo reagent according to manufacturer’s instructions (Promega, Fitchburg, WI). Plates were covered and rocking for 15 minutes before luminescence was measured.
Western Blots
Whole cell lysates were extracted with RIPA buffer (CST, Danvers, MA) containing protease and phosphatase inhibitor (Life Technologies) and quantified with BCA reagent (Thermo Fisher). Murine muscle, liver, and Khk-/- liver was obtained from our previous study (21). Equal amounts of protein were diluted in 4x LDS buffer (Life Technologies) before being run in 4-12% bis-tris gels (Invitrogen, Carlsbad, CA). Gels were transferred to PVDF membranes (Perkin-Elmer, Waltham, MA) and blocked for 1 hour in 5% BSA in Tris-buffered saline containing .1% Tween 20 (TBST). Membranes were probed while rocking at 4°C with the following antibodies and concentrations: GLUT1 (Millipore 07-1401) 1:1000, GLUT2 (Abcam, Cambridge, UK, ab192599) 1:1000, GLUT5 (Invitrogen, PA5-80023) 1:1000, KHK (Abcam) 1:1000, HK1 (CST 2024) 1:1000, HK2 (CST 2867) 1:1000, ALDOA (CST8060) 1:1000, ALDOB (Abcam ab152828) 1:1000, ALDOC (Proteintech, Rosemont, IL, 14884-1-AP) 1:1000, LDHA (CST) 1:1000, LDHB (Abcam) 1:1000, GAPDH (Proteintech 10494-1-AP) 1:5000, Pan-Actin (CST 4968) 1:1000, and V5-HRP (Life Technologies R96125) 1:5000. After incubation, cells were washed with TBST before appropriate HRP-conjugated secondary antibody was added for 1 hour. After 3 more TBST washes, membranes were exposed to Supersignal West Dura (Life Technologies) and imaged using a ChemiDoc MP Imaging System (BioRad, Hercules, CA).
Plasmids and cloning
The following plasmids were generously provided by researchers via Addgene: pSpCas9(BB)-2A-Puro (PX459) V2.0 (#62988) from Dr. Feng Zhang (Broad Institute) (Ran et al., 2013)m pDONR221-SLC2A5 (#132090) from the RESOLUTE Consortium and Giuliu Superti-Furga (Research Center for Molecular Medicine of the Austrian Academy of Sciences), and pLenti-U6-tdTomato-P2A-BlasR (Lrt2B) (#110854) from Dr. Lukas Dow (Weill Cornell Medicine) (32).
We selected sgRNA (Figure S3) for human KHK at the beginning of exon 5 using CRISPRdirect (33). Oligonucleotide pairs were annealed and cloned into PX459 using BbsI-HF (New England Biolabs, Ipswich, MA) followed by a ligation reaction (New England Biolabs). PDONR221-GLUT5 was cloned according to Gateway Technology (Invitrogen) into pLenti7.3_V5_DEST (Invitrogen) using LR Clonase (Invitrogen) in order to generate pLenti7.3_V5_SLC2A5. These plasmids were generated in Stbl3 bacteria (Life Technologies) and were purified using Qiagen miniprep or maxiprep kits (Qiagen).
Generating knockouts
We plated 200,000 cells/well in a 6-well dish. The following day, cells were transfected with 3 µL Lipofectamine 2000 (Life Technologies) and 3 ug plasmid containing sgRNA in Optimem (Life Technologies). The following day, media was changed. The day after media change, cells were selected with 2 ug/mL puromycin for 48 hours. 50 or 100 cells were then passaged into 10 cm dishes and were allowed to proliferate into visible colonies over 2 weeks. Single colonies with normal morphology were selected using cloning cylinders (Thermo Fisher). Knockouts were verified by western blot and sanger sequencing.
Transduction
2,000,000 293T cells were plated in a 10-cm dish. The next day, cells were transfected with 30 µL Lipofectamine 2000, 9 ug psPAX2, 1 ug VSV-G, and 9 ug of either Lrt2b, pLenti7.3-V5 EV, pLenti7.3-V5-SLC2A5. Media was changed the following day. Viral particles were harvested 48 and 72 hours after initial media change. The 2 harvests were combined and aliquoted for storage in -80 C.
To generate PC3-red, parental cells were given 50% Lrt2b virus and 50% media as well as 10 ug/mL polybrene. The next day, cells were given 50% virus and 50% media as well as 10 ug/mL polybrene. Media was changed after 24 hours. The day after media change, cells were grown in media containing 10 ug/mL blasticidin (Invivogen, San Diego, CA). Overexpression was verified by microscopy.
To overexpress GLUT5, non-fructolytic cell lines from several origins were plated at low confluence in 6-well dishes. The next day, cells were given 50% EV or SLC2A5 virus and 50% media as well as 10 ug/mL polybrene. Media was changed after 24 hours. Overexpression was verified by western blot.
Seahorse Assay
ECAR was measured with the Seahorse XFe96 Analyzer (Agilent, Santa Clara, CA), following manufacturer’s Glycolytic Stress Test protocol. Briefly, 5,000 cells were plated in each well of a 96-well Seahorse assay plate. That same day, the assay cartridge was hydrated and kept in a non-CO2 incubator at 37°C. After 12-24 hours, cells were washed with PBS before they were given reconstituted sugarless DMEM powder (Sigma-Aldrich) supplemented with 2 mM glutamine and 5 mM HEPES buffer. Cells were then incubated for 45 minutes at 37°C in a non-CO2 incubator. Compounds (final concentrations: Glucose 10 mM or fructose 10 mM, oligomycin 1 uM, and 2-DG 50 mM) were prepared, loaded into the flux pack, and put into the Seahorse XFe96 Analyzer. The plate containing cells were subsequently loaded into the machine. ECAR was analyzed using Seahorse Wave software.
Metabolite extraction, targeted analysis, and untargeted analysis
Metabolomics were carried out on cells to measure polar metabolites. 500,000 cells were plated in triplicate in 6-well dishes for each condition. The next day, cells were washed briefly with 37°C PBS before given media containing no glucose and 10 mM [U-13C]-fructose (Cambridge Isotope Laboratories, Tewksbury, MA). After 30 minutes incubation, cells were washed briefly with warm PBS and immediately harvested into 2 mL Eppendorf tubes using with ice cold 80% methanol (Yuan et al., 2012) and 0.02 M formic acid. Cells were vortexed and stored in -80C overnight. Samples were spun down at 13,000 x RPM for 10 minutes at 4°C. Supernatant was transferred to a new Eppendorf tube and was evaporated for LC/MS.
Quantitative metabolomics were performed on samples as previously described (21). Briefly, 5 µL of each filtered extract was injected through an Agilent ZORBAX Extend C18, 2.1 x 150 mm, 1.8 (Agilent) downstream of an Agilent ZORBAX SB-C8, 2.1 mm x 30 mm, 3.5 um guard column (Agilent) heated to 40°C in the Agilent 1290 Infinity LC system. Solvent A (97% water/ 3% methanol containing 5 mM tetrabutylammonium hydroxide (TBA) and 5.5 mM acetic acid) and Solvent B (methanol containing 5 mM TBA and 5.5 mM acetic acid) were infused at a 0.250 mL/min flow rate. The reverse phase gradient was as follows: 0-3.5 min, 0% B; 4-7.5 min, 30% B; 8-15 min, 35% B; 20-24 min, 99% B; followed by a 7-minute run at 0% B. Acquisition was performed on the Agilent 6230 TOF mass spectrometer (Agilent) using an Agilent Jet Stream electrospray ionization source (Agilent) operated at 4000 V Cap and 2000 V nozzle voltage in high resolution, negative mode. During acquisition, the sample nebulizer was set to 45 psig with sheath gas flow of 12L/min at 400°C. Drying gas was kept at 325°C at 8 L/min. The fragmentor was set to 125 V, with the skimmer set to 50 V and Octopole Vpp at 400 V. Samples were acquired in centroid mode for 1.5 spectra/s for m/z’s from 50-1100.
Data was analyzed by batch processing with Agilent MassHunter Profinder software (Agilent) for both targeted and untargeted analysis. For targeted analysis, we identified metabolites by both retention time and with authentic standards. We identified untargeted compounds using Profinder Batch Targeted Feature Extraction. Then, we processed hits through Agilent Mass Profiler Professional software for quality control.
Quantification and statistical analysis
Sample size was estimated based on prior data (21). Data is presented as ± standard error of the mean (SEM), calculated by Graphpad Prism 8. For total metabolites and GLUT5 rescue growth rates, unpaired two-tail t tests were done between control and experimental conditions. For RT-qPCR data and 13C metabolomics, two-way ANOVA was done with post-test comparisons made by Fisher’s LSD test. Statistical significance is indicated in figures using the following denotation: *P < 0.05, **P< 0.01, ***P < 0.001, and ****P < 0.0001. Sample number was noted in figure legends.
Software availability
An application to perform cell quantification from images was created by S.T and is available on https://github.com/sam-taylor/CompCount.