Translational Cancer Mechanisms and Therapy

Multi-omics Integration Analysis RobustlyPredicts High-Grade Patient Survival andIdentifies CPT1B Effect on Fatty Acid Metabolismin Bladder CancerVenkatrao Vantaku1, Jianrong Dong1, Chandrashekar R. Ambati2, Dimuthu Perera2,Sri Ramya Donepudi2, Chandra Sekhar Amara1, Vasanta Putluri2, Shiva Shankar Ravi1,MatthewJ. Robertson2, DanthasingheWadugeBadrajee Piyarathna1,MarianaVillanueva2,Friedrich-Carl von Rundstedt3, Balasubramanyam Karanam4, Leomar Y. Ballester5,Martha K. Terris6, Roni J. Bollag6, Seth P. Lerner7, Andrea B. Apolo8, Hugo Villanueva2,MinJae Lee9, Andrew G. Sikora10,Yair Lotan11, Arun Sreekumar1,12, Cristian Coarfa1,2,13, andNagireddy Putluri1

Abstract

Purpose: The perturbation of metabolic pathways inhigh-grade bladder cancer has not been investigated.We aimed to identify a metabolic signature in high-gradebladder cancer by integrating unbiased metabolomics,lipidomics, and transcriptomics to predict patient survivaland to discover novel therapeutic targets.

Experimental Design: We performed high-resolution liq-uid chromatography mass spectrometry (LC-MS) and bioin-formatic analysis to determine the global metabolome andlipidome inhigh-gradebladder cancer.We further investigatedthe effects of impaired metabolic pathways using in vitro andin vivo models.

Results: We identified 519 differential metabolites and 19lipids that were differentially expressed between low-gradeand high-grade bladder cancer using the NIST MS metabolo-mics compendium and lipidblast MS/MS libraries, respective-ly. Pathway analysis revealed a unique set of biochemical

pathways that are highly deregulated in high-grade bladdercancer. Integromics analysis identified a molecular gene sig-nature associated with poor patient survival in bladder cancer.Low expression of CPT1B in high-grade tumors was associatedwith low FAO and low acyl carnitine levels in high-gradebladder cancer, which were confirmed using tissue microar-rays. Ectopic expression of the CPT1B in high-grade bladdercancer cells led to reduced EMT in in vitro, and reduced cellproliferation, EMT, and metastasis in vivo.

Conclusions: Our study demonstrates a novel approachfor the integration of metabolomics, lipidomics, and tran-scriptomics data, and identifies a common gene signatureassociated with poor survival in patients with bladder cancer.Our data also suggest that impairment of FAO due to down-regulation ofCPT1Bplays an important role in the progressiontoward high-grade bladder cancer and provide potential tar-gets for therapeutic intervention.

IntroductionBladder cancer causes significant morbidity and mortality

worldwide. In the United States, there were about 79,000 casesand 17,000 deaths due to bladder cancer in 2017 (1). In the pastfew decades advances in cancer genomics, transcriptomics, pro-

teomics, and metabolomics resulted in the discovery of potentialbiomarkers for cancer (2, 3). Unfortunately, most of the biomar-kers have failed to demonstrate superior performance character-istics compared with existing clinical tests. Recent advancementsin omics technologies have ushered in a new era of targeted cancer

1Department of Molecular and Cell Biology, Baylor College of Medicine, Houston,Texas. 2Dan L. Duncan Cancer Center, Advanced Technology Core, Alkek Centerfor Molecular Discovery, Baylor College of Medicine, Houston, Texas. 3Depart-ment of Urology, Jena University Hospital, Friedrich-Schiller-University, Jena,Germany. 4Department of Biology and Center for Cancer Research, TuskegeeUniversity, Tuskegee, Alabama. 5Pathology & Laboratory Medicine, Neurosur-gery, University of Texas Health Science Center, Houston, Texas. 6AugustaUniversity, Augusta, Georgia. 7Scott Department of Urology, Baylor College ofMedicine, Houston, Texas. 8Center for Cancer Research, National Cancer Insti-tute, Bethesda, Maryland. 9Division of Clinical and Translational Sciences,Department of Internal Medicine, McGovern Medical School at The Universityof Texas Health Science Center, Houston, Texas. 10Department of Otolaryngol-ogy-Head &Neck Surgery, Baylor College of Medicine, Houston, Texas. 11Depart-ment of Urology, University of Texas Southwestern, Dallas, Texas. 12Verna andMarrs McLean Department of Biochemistry and Molecular Biology, Baylor

College of Medicine, Houston, Texas. 13Center for Precision EnvironmentalHealth, Baylor College of Medicine, Houston, Texas.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

V. Vantaku, J. Dong, C.R. Ambati, andD. Perera contributed equally to this article.

Primary Corresponding Author: Nagireddy Putluri, Department of MolecularandCellular Biology, Baylor College ofMedicine, Houston, TX 77030. Phone: 713-798-3139; E-mail: putluri@bcm.edu

Correspondence for Bioinformatics: Cristian Coarfa, E-mail: coarfa@bcm.edu

Clin Cancer Res 2019;25:3689–701

doi: 10.1158/1078-0432.CCR-18-1515

�2019 American Association for Cancer Research.

ClinicalCancerResearch

www.aacrjournals.org 3689

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

therapies by improving our understanding of molecular carcino-genesis. Omics studies have identified therapeutic targets andresulted in the successful development of new drugs to treatvarious solid tumors like breast cancer and lung cancer (4, 5).For bladder cancer, prognostication and treatment still dependmainly on pathologic and clinical characteristics (6). Patientsdiagnosed with high-grade invasive disease are difficult to treateffectively and have a relatively low life expectancy despite avail-able multimodal therapies. Therefore, novel prognostic and ther-apeutic targets against bladder cancer are needed.

Tumor progression relies on the reprogramming of cellularmetabolism (7). Our previous metabolomic and lipidomic stud-ies highlighted the significance of altered xenobiotic and fatty acidmetabolism in bladder cancer development (8–10). However, westill know little about the altered metabolic pathways during thetransition from low-grade to high-grade bladder cancer. Theintegration of omics facilitates the study of interactions amongall classes of biomolecules that can occur in cell, which in turn,determines the cellular physiology and behavior. Treatment forhigh-grade muscle-invasive bladder cancer (MIBC) has notadvanced beyond cisplatin-based combination chemotherapy inthe past three decades and very fewdrugs for the disease have beenapproved recently (11). The identification of markers to predictpoor outcomes among patients with bladder cancer will improveour ability to identify patients who might benefit from adjuvanttherapy. Furthermore, a greater understanding of the metabolicmolecular mechanisms of bladder cancer progression and theidentification of novel therapeutic targets will improve outcomesfor patients with high-grade bladder cancer.

We used a robust mass spectrometry (MS) platform to conductglobal unbiased metabolomic and lipidomic analyses to identifycritical alterations that may contribute to bladder cancer progres-sion. We mapped the altered metabolites and lipids to corre-sponding genes using the Human Metabolome Database(HMDB) and then a pathway analysis to discover the ways inwhich metabolic pathways are perturbed in high-grade bladdercancer. We integrated the relevant genes with publicly available

transcriptomic data and generated an integrated gene signature topredict the survival of patients with bladder cancer. Furthermore,we found that carnitine palmitoyl transferase 1B (CPT1B) is signif-icantly downregulated inhigh-grade bladder cancer, which in turnleads to low fatty acid b-oxidation (FAO). Ectopic expression ofCPT1B in high-grade bladder cancer cells increased FAO, which inturn reduced epithelial–mesenchymal transition (EMT). Finally,overexpression of CPT1B in high-grade bladder cancer cellsreduced cell proliferation and liver metastasis in a chick chorio-allantoic membrane (CAM) in vivo model. Our results clearlyindicate that CPT1B plays an important role in bladder cancertumor progression by affecting fatty acid metabolism. CPT1Bmight therefore have prognostic value and provide a target fortherapeutic intervention in high-grade bladder cancer.

Materials and MethodsReagents and internal standards

High-performance liquid chromatography (HPLC)-gradeammonium acetate, acetonitrile, methanol, chloroform, andwater were procured from Burdick & Jackson. MS grade formicacid, standards and internal standards, N-acetyl aspartic acid-d3,tryptophan-15N2, sarcosine-d3, glutamic acid-d5, thymine-d4,gibberellic acid, trans-zeatin, jasmonic acid, anthranilic acid-15N,and testosterone-d3 were purchased from Sigma-Aldrich. Mixtureof LPC 17:0/0:0, PG 17:0/17:0, PE 17:0/17:0, PC 17:0/17:0,TAG 17:0/17:0/17:0, SM 18:1/17:0, MAG 17:0, DAG 16:0/18:1,CE 17:0, ceramide 18:1/17:0, PA 17:0, PI 17:0/20:4, and PS 17:0/17:0. were obtained from Avanti polar lipids (SupplementaryTable S1).

Cell linesRT4, T24, UMUC3, J82, and TCCSUP were procured from

ATCC and maintained as per ATCC instructions. All cell lineswere verified using short tandem repeat DNA fingerprinting at theMD Anderson Cancer Center (Houston, TX) and were tested forMycoplasma contamination every 3 months. All the bladdercancer cell lines were procured from the ATCC less than 2 yearsbefore starting the experiments.

Sample preparation for metabolomics and lipidomicsAll the bladder cancer specimens were procured by a prior

written informed consent under Institututional Review Board–approved protocols. Metabolites were extracted from bladdercancer tissues, and mouse liver pool was used as a quality controland followed the extraction procedure described previously(8, 12–14). Briefly, 10 mg of tissue was used for the metabolicextraction. The extraction step starts with addition of 750 mLice-cold methanol: water (4:1) containing 20 mL spiked internalstandards (ISTD). After homogenization, ice-cold chloroformandwater were added in a 3:1 ratio for a final proportion of 4:3:2methanol:chloroform:water. The organic and aqueous layerswerecollected, dried, and resuspended in methanol: water (1:1). Theextract was deproteinized using a 3-kDa molecular filter and thefiltrate was dried under vacuum. The dried extracts were resus-pended in 100 mL of injection solvent composed of 1:1methanol:water and subjected to LC-MS.

Lipids were extracted using a modified Bligh–Dyer method(15). The extraction was carried out using 2:2:2 ratio of water:methanol:dichloromethane at room temparature after ISTDs intotissues andquality control pool (12). After homogenization of the

Translational Relevance

Cancer metabolism varies depending on the tumor grade.Currently, there is an absence of multi-omics data integrationto predict bladder cancer survival. To fill in this gap, weperformed unbiased metabolomics and lipidomics analysesof matched bladder cancer tissues and integrated them withbladder cancer transcriptomics analyses to generate an inte-grated gene signature that was associated with patient survivalin multiple bladder cancer cohorts. Our results revealedaltered metabolic differences between high-grade bladdercancer and low-grade bladder cancer and suggest that impairedfatty acid b-oxidation (FAO) due to the downregulation ofCPT1B plays a crucial role in the progression of low-gradeto high-grade bladder cancer. CPT1B overexpression inhigh-grade bladder cancer cell lines reduced cell proliferation,epithelial–mesenchymal transition, and metastasis to theliver in vivo by increasing FAO. Our metabolic-centeredmulti-omics based integrative analysis provides a system-levelperspective on bladder cancer and potential targets for noveltherapeutics against high-grade bladder cancer.

Vantaku et al.

Clin Cancer Res; 25(12) June 15, 2019 Clinical Cancer Research3690

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

samples, the organic layer was collected and completely driedunder vacuum. Before MS analysis, the dried extract was resus-pended in 100 mL of buffer containing 10 mmol/L NH4Ac andsubjected to LC/MS. The lipidome was separated using reverse-phase (RP) chromatography. To monitor the lipid extractionprocess, we used a standard pool of tissue samples from aliquotsof the same samples.

Data acquisition for metabolomics and lipidomics analysisFor metabolomics, 5 mL of the metabolite extract was injected

into a 3.5-mm particle 4.6 � 150 mm X bridge amide column,which heats to 60�C. Formic acid (0.1%) in water was used assolvent-A and acetonitrile as solvent-B in positive ionization and20mmol/L ammoniumacetate in 95%water; 5%acetonitrile and100% acetonitrile as solvent-B in negative ionization mode. Forlipidomics, 10 mL of the lipid extract was injected into a 1.8 mmparticle 50 � 2.1 mm column (Acquity HSS UPLC T3), whichheats to 55�C. Acetonitrile: water (40:60, v/v) with 10 mmol/Lammonium acetate was solvent-A and acetonitrile: water: iso-propanol (10:5:85 v/v) with 10 mmol/L ammonium acetate wassolvent-B. Chromatographic elution used a linear gradient over a20-minute total runtime, with 60% solvent-A and 40% solvent-Bgradient in the first 10 minutes. Then, the gradient was ramped ina linear fashion to 100% solvent-B, which was maintained for 7minutes. After that, the system was switched back to 60%solvent-B and 40% solvent-A for 3 minutes. The flow rate usedfor these experiments was 0.4mL/minute. The data acquisition ofeach sample was performed in both positive and negative ioni-zationmodesusing a TripleTOF5600 equippedwith aTurboVTMion source. The instrument performed one TOF MS survey scan(150 ms) and 15 MS/MS scans with a total duty cycle time of2.4 milliseconds (ms). The mass range in both modes was 50 to1,200 m/z. We controlled the acquisition in both MS andMS/MSspectra by data-dependent acquisition function of the Analyst TFsoftware (AB Sciex). Rolling collision energy spread was setwhereby the software calculated the collision energy value to beapplied as a functionofm/z.Mass accuracywasmaintained by theuse of an automated calibrant delivery system interfaced to thesecond inlet of the Duo Spray source.

Metabolomics and lipidomics data processingThe raw data (.wiff) of metabolomics and lipidomics from AB

Sciex are converted to ".mgf" data format using proteoWizardsoftware (16). We used the NIST MS Pep to search the convertedfiles against NIST14 libraries (17, 18) for metabolomics andLipidBlast libraries for lipidomics. Them/zwidthwas determinedby themass accuracy of ISTDs andwas set 0.001 for positivemodeand 0.005 for a negative mode with an overall mass error of lessthan 2 parts per million. The minimummatch factor used was setto 400 score. The MS/MS identification results from all the fileswere combined using an in-house software tool to create a libraryfor quantification. All raw data files were searched against thislibrary of identified metabolites with mass and retention timeusing Multiquant 1.1.0.26 (AB Sciex, USA). We used MS/MS dataas an intermediary step to help with identification, but quanti-fication was done using MS1 data only. Relative abundance ofpeak spectra was used for the analyses. Themetabolites identifiedin both positive and negative ion modes were initially analyzedseparately for their relationship with the outcome to ensurepersistent results. Identified metabolites were quantified by nor-malizing against their respective ISTDs. Quality control samples

were used to monitor the overall quality of the metaboliteextraction and MS analysis.

Statistical analysis for unbiased metabolomics and lipidomicsAfter data acquisition, themissing values formetabolites/lipids

were imputed using the k-nearest neighbors method. Then thedata were log2 transformed followed by normalization usingthe median interquantile range normalization. The compound-by-compound t-test was applied, followed by the Benjamin–Hochberg procedure for FDR correction accounting for multiplecomparisons. Significance was achieved for FDR < 0.25.

Quantification of overall b-oxidation activity by using theBiocrates AbsoluteIDQ Kit p180

Absolute concentrations of metabolites were quantified byAbsoluteIDQ p180 Kit (Biocrates Life Science AG) using QTRAP6500 LC/MS/MS system (AB Sciex) equippedwith an electrosprayionization (ESI) source, an Agilent G1367B autosampler and theAnalyst 1.51 software (AB Sciex). Ten microliters of the homog-enized tissue sample, quality control samples, blank, and cali-bration standards were added to the appropriate wells. The platewas dried and the samples were derivatized with phenylisothio-cyanate and dried again. The samples were eluted with 5 mmol/Lammonium acetate in methanol and then were diluted withrunning solvent provided in kit for flow injection analysis. Twentymicroliters of the samplewasdirectly injected into theMSat aflowof 30mL/minute. Concentrationswere calculated and evaluated inthe Analyst/MetIQ software by comparing measured analytes in adefined extracted ion count section to those of specific labeledinternal standards or nonlabeled, nonphysiologic standards(semiquantitative) providedbyBiocrates.Measurement of overallb-oxidation activity was derived by calculating the ratio of short-chain acylcarnitines to free carnitines.

Pathway analysisEnriched pathways were determined by using the hypergeo-

metric distribution against the Kyoto encyclopedia of genes andgenomes (KEGG) pathway database, as compiled by the MSigDBgene set compendium; significance was achieved for P < 0.05.Pathway networks were visualized using the Cytoscape scientificvisualization platform.

Integration of unbiased metabolomics, unbiased lipidomics,and transcriptomics

Metabolites and lipids were mapped into associated genes/proteins using theHMDB ID.Next,weobtained the transcriptomefingerprint between high and low grade bladder cancerusing the Kim cohort as described below. We generated anintegrated metabolomics/lipidomics/transcriptomics gene signa-ture, by intersecting the all three gene sets.

Gene expression analysisTo identify the gene signature of high versus low grade bladder

cancer, we downloaded the publicly available data from the Kimcohort. Gene significance was assessed by using the Student t test;significance was achieved for P < 0.05 and fold change eitherabove 1.25 or below 0.8. To identify genes associated with lowversus high CPT1B gene expression, we used the gene expressiondata from five independent public cohorts: The Cancer GenomicAtlas (TCGA), Kim (GSE13507), Sjodahl (GSE32894), Lindgren(GSE32548), and Riester (GSE31684). For each cohort, we first

Multi-omics: Identification of CPT1B Role in Bladder Cancer

www.aacrjournals.org Clin Cancer Res; 25(12) June 15, 2019 3691

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

stratified the patients by CPT1B, into the top 50% and bottom50%. Next, we generated a rank file for each expressed gene by thelog2 fold change of lowCPT1B samples over highCPT1B samples.We ran Gene Set Enrichment Analysis (GSEA) against theHALLMARK pathway collection, as compiled by MolecularSignature Database (MSigDB). Significance was achieved forFDR-adjusted Q-value < 0.25.

Analysis of prognostic power of the integrated gene signature inbladder cancer

Significant association with survival of an integrated bladdercancer metabolomics/lipidomics/transcriptomic gene signaturewas evaluated in four gene expression datasets of bladder cancercohorts fromTCGA, Kim, Sjodahl, and Lindgren forwhich clinicaloutcomes have been reported. For each gene in the signature andfor each specimen, we computed the Z-score for its expressionwithin the cohort, as described previously (9) and computed thesum Z-score for each specimen. Specifically, the Z-scores of genesrepressed in the integrated signature were subtracted from theZ-scores of genes induced in the integrated signature, resulting in acorresponding signature activity score for each specimen. In eachcohort, the specimens were ranked according to their integratedsignature activity score, and association with survival was evalu-ated via the log-rank test between the bottom 50% of the patientsand the top 50%of the patients. Survival significancewas assessedby employing the package survival in the R statistical system.

Univariate and multivariate survival analysisSurvival analysis was performed using the survival library in the

R statistical system. Univariate analysis used 27 gene signature orthe CPT1B with survival, after stratifying patients in top andbottom 50%. For multivariate models, we tested the gene signa-ture and CPT1B, while controlling for other demographic andclinical factors, specifically age, sex, and tumor stage. HRs derivedfrom Cox proportional hazards models are expressed with 95%confidence intervals and P values. For all analyses and eachvariable, significant association with survival was achieved forP < 0.05.

CAM assayThe CAM was assessed as described previously (19) on embry-

onic day 7; the eggs were inoculated with 5� 105 UMUC3 vectorcontrol and CPT1B-overexpressing cells per egg. In vivo lumines-cence imaging was conducted on embryonic days 10, 13, and 15(3, 5, and 7 days after seeding the tumor cells) by using in vivoimaging system (IVIS). Exposure time was 1 second, 5 minutesafter addition of 100 mL of 15 mg/mL D-luciferin, and region ofinterest was quantitated by the Lumina software. Seventh day afterinoculation, the eggswere euthanized as per theAVMAguidelines.The CAM tumors were fixed in 10% formalin and embedded inparaffin for IHC. IHCwasperformed forCPT1B, E-cad,N-cad, andvimentin to check the expression levels. IHC-stained slides werescanned and analyzed by Aperio CS2 system (Leica Biosystems)and number of positive cells/total cells were plotted.

To analyze metastasis of the tumor into the chick embryo, weharvested embryo visceral organs (i.e., lung, liver, bone, andbrain) on day 15 to 18 postfertilization. To achieve this, theeggshell opening was widened to allow excision of the CAMtumor. After the tumor excision, the embryo was humanelyeuthanized by decapitation with a scissors and the body incisedto dissect visceral organs including lung, liver, bone, and brain.

Once dissected, the tissues were collected, flash-frozen, and storedat�80�Cuntil further analysis. Quantitative assessment of tumormetastasis into visceral organs was achieved by quantification ofhuman DNA from the metastasized cells in the chicken DNA byperforming RT-PCR for human-specific Alu sequence.

To quantify the amount of human DNA from tumor cells thathad metastasized to the chick embryo tissues, a standard curvewas generated by amplification of genomic DNA isolated from aserial dilution of bladder cancer cells mixed with individualchick visceral tissue DNA (modified from published methods;refs. 20, 21). To detect the human Alu sequences in the chickenDNAbackground, humanAlu-detecting primers and aprobewereused (Supplementary Table S2). The standard curvewas generatedusing serial dilutions (0.1 ng–1 mg) of UMUC3 DNA mixedwith constant amount of chicken DNA. The RT-PCR was per-formed in triplicate for each standard as well as the experimentalsamples and analyzed the DNA content in control and CPT1Boverexpression groups.

ResultsIdentification of altered metabolites and lipids in low-gradeand high-grade bladder cancer

We characterized the global metabolome and lipidome oflow-grade (n¼ 5) and high-grade (n¼ 20) bladder cancer tissuesusing high-resolution LC-MS and clinical information for thepatients is shown in Supplementary Table S3. A cocktail of 10internal standards along with a repetitive analysis of pooled liverextracts served as controls to ascertain the reproducibilityand robustness of the profiling platform for metabolomics(Supplementary Figs. S1A and S2A) and lipidomics (Supplemen-tary Figs. S1B and S2B). An extensiveNISTdatabase and lipid blastsearch analysis of the LC-MS data identified a total of approxi-mately 2,000 metabolites and 786 lipids in positive and negativeionization modes (Supplementary Fig. S3). An overview of thestudy is shown in Fig. 1. A total of 519 metabolites (14 differentclasses)were significantly altered (FDR<0.25) between low-gradeand high-grade bladder cancer (Fig. 2A; Supplementary Table S4).These include significantly altered levels of metabolites inmultiple pathways (Fig. 2D). Likewise, a total of 19 lipids (7different classes) were significantly altered (FDR < 0.25; Fig. 2B;Supplementary Table S5), that were involved in many pathways(Fig. 2E). Using well-annotated, publicly available Kim transcrip-tomics data we found 3036 genes that had significantly altered(P < 0.05) between low-grade and high-grade bladder cancer(Fig. 2C). Like the altered lipids andmetabolites the altered geneswere involved in multiple pathways (Fig. 2F) and 51 commonpathways were implicated in all three platforms (Fig. 2G).

Integration of metabolomics, lipidomics, and transcriptomicsdata identified a gene signature with prognostic value inbladder cancer

We mapped the panels of altered metabolites, to their corre-sponding genes using HMDB (10, 22), resulting in the identifi-cation of 2,311 genes. A similar analysis of altered lipids led to theidentification of 173 genes. Analysis of transcriptomics publicdata from the Kim cohort resulted in the identification of 3,036genes that were differentially expressed between low-grade andhigh-grade bladder cancer. Integrationof these three platforms ledto 27 intersecting genes between low-grade and high-grade blad-der cancer (Fig. 3A).

Vantaku et al.

Clin Cancer Res; 25(12) June 15, 2019 Clinical Cancer Research3692

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

We next examined the expression of these 27 genes using Kimtranscriptomics data. Eleven of these genes were upregulatedwhile the other 16 were downregulated in bladder cancer(Fig. 3B). We examined the association between the 27 genesand survival using publicly available bladder cancer cohorts[Kim(23), Lindgren (24), andSjodahl (25)].Weused the log-ranktest to determine patient outcomes. In all three cohorts, highexpression levels of the 11 genes that were upregulated in bladdercancer was associated with poor survival. Similarly, low expres-sion levels of the genes that were downregulated in bladder cancerwas associated with poor survival in all three cohorts (Fig. 3C).Interestingly, 11upregulated gene signature thatwere upregulatedin bladder cancer in the Kim cohort were associated withpoor survival in Lindgren and TCGA bladder cancer cohorts(Supplementary Fig. S4).

We analyzed univariate and multivariate Cox proportionalhazards regression models using the Kim, Lindgren, Sjodahl,TCGA datasets to assess the associations between the 27 genesignature and the time-to-death. Using multivariate models, wetested the 27 gene signature (top 50% vs. bottom 50%; Supple-mentary Table S6A–S6D) while controlling for sex, age, andtumor stage. Univariate Cox proportional hazards regressionanalysis showed that the gene signature was significantly associ-ated with risk of death, with an HR above 1 in the Kim, Lindgren,

Sjodahl and TCGA cohorts. Despite that, a multivariate analysisshowed that although the HR was greater than 1 in the Kim,Lindgren, and TCGA, the 27 gene signature was not associatedwith survival independently of the other clinical variables con-sidered. The tumor stage was associated with survival in all fourcohorts, while age was associated with survival in Kim, Lindgren,and TCGA cohorts (Supplementary Table S6A—S6D).

To obtain additional insights into the 27 genes, we performedsurvival analysis based on the individual genes in the samepublicly available cohorts. We found that six upregulatedgenes (PLA2G4C LIPG, PIGS, PIGU, ATP8B2 and ATP8B4) hada significant clinical association with survival (SupplementaryFig. S5). High expression of PLA2G4C andATP8B2was associatedwith poor survival in multiple cohorts (Kim, Lindgren, Sjodahl,and TCGA; ref. 26; Supplementary Fig. S5). Eight of the down-regulated genes (CPT1B, PIGB, PLA2G4A, PIGV, INPP5D, PLCD3,PIGZ, and PLA2G10) were associated with poor survival (Sup-plementary Fig. S6).

Remarkably, low expression of CPT1B was associated withpoor survival in TCGA (Fig. 4A) Lindgren, and Sjodahl cohorts(Supplementary Fig. S6), whereas in the Kim cohort there was noassociationbetweenCPT1B and survival.However, the expressionof CPT1B was significantly lower in high grade bladder cancertumors in all three cohorts (Supplementary Fig. S7). We further

Figure 1.

Overview of the strategy used to profile the global metabolome and lipidome and integrated themwith the transcriptome to characterize low-grade and high-grade bladder cancer.

Multi-omics: Identification of CPT1B Role in Bladder Cancer

www.aacrjournals.org Clin Cancer Res; 25(12) June 15, 2019 3693

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

tested the univariate andmultivariatemodels forCPT1B (top50%vs. bottom 50%) by considering for sex, age, and tumor stageusing Kim, Lindgren, Sjodahl, and TCGA datasets. Univariate Coxproportional hazards regression analysis showed that CPT1B hadan HR below 1 in TCGA, Lindgren, and Sjodahl cohorts and wassignificantly associated with survival only in the TCGA cohort(Supplementary Table S7A–S7D). In a multivariate Cox propor-tional hazards regression model adjusted for sex, age, and tumorstage showed thatCPT1Bhad anHRbelow1 andwas significantly

associated with survival in the TCGA cohort (SupplementaryTable S7A–S7D). Pathway analysis also indicated that fatty acidmetabolism was significantly increased in high-grade bladdercancer. CPT1B is involved in fatty acid transport and acts as arate-limiting enzyme for FAO, thereby altering the fate of thecancer cells (27). Recent studies showed that cancer metastaticpotential is a consequence of an alteration in lipid metabolicpathways (28–30), which led to our hypothesis that CPT1B playsa crucial role in bladder cancer progression.

Figure 2.

Identification of altered metabolites, lipids, and transcripts and their respective pathways in high-grade bladder cancer.A, Heatmap of hierarchical clustering of519 differentially expressedmetabolites from different classes of biomolecules detected across low-grade and high-grade bladder cancer samples (FDR < 0.25).Each class of metabolites is represented in a different color. B, Heatmap of hierarchical clustering of 19 differentially expressed lipids from various classesdetected across low-grade and high-grade bladder cancer samples (FDR < 0.25). Each class of lipids is represented in a different color. C, Heatmap of hierarchicalclustering of 3,036 differentially expressed genes from the Kim dataset (GSE13507) across low-grade and high-grade bladder cancer samples (P < 0.05).Columns represent individual tissue samples; rows refer to distinct metabolites, lipids, and genes. Shades of yellow indicate different levels of increase inexpression; shades of blue indicate different levels of decrease in expression relative to the median levels. D–F, Pathway enrichment analysis of differentiallyexpressed metabolites, lipids, and genes shows the significantly altered pathways in high-grade bladder cancer (P < 0.05). The node size indicates the number ofgenes involved in the pathway (deregulated pathways common to the metabolomics, lipidomics, and transcriptomics data sets are highlighted in red). G, Venndiagram of the deregulated metabolic pathways common to the metabolomics, lipidomics, and transcriptomics datasets.

Vantaku et al.

Clin Cancer Res; 25(12) June 15, 2019 Clinical Cancer Research3694

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

Figure 3.

Metabolomic-lipidomic–transcriptomic integration strategies to identify potential prognostic gene signature for bladder cancer. A, The integration of asignificant genes derived frommetabolomics, lipidomics, and transcriptomics data comparing low-grade and high-grade bladder cancer resulted in a set of 27common genes. B, Eleven upregulated and 16 downregulated genes in high-grade bladder cancer in the Kim cohort (GSE13507) with their fold change. C,Survival analysis shows poor survival associated with the upregulated gene signature [Kim, log rank P¼ 0.03359; Lindgren (GSE32548), log-rank P¼ 0.030314;Sjodahl (GSE32894), log-rank P¼ 0.00018) and downregulated gene signature (Kim, log-rank P¼ 0.03144; Lindgren, log-rank P¼ 0.07708; and Sjodahl,log-rank P¼ 0.00456)].

Multi-omics: Identification of CPT1B Role in Bladder Cancer

www.aacrjournals.org Clin Cancer Res; 25(12) June 15, 2019 3695

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

Suppression of CPT1B impairs the FAO in high-grade bladdercancer

To better understand the role of CPT1B in high-grade bladdercancer, we examined the expression of CPT1B in low-grade andhigh-grade bladder cancer. CPT1B expression was significantlylower in high-grade bladder cancer than in low-grade bladdercancer (Fig. 4B) in an independent validation cohort (Supple-mentary Table S8). Histochemical analysis of a tissue micro array(TMA; Fig. 4C) confirmed the findings of the transcriptomiccohorts analyses (Supplementary Fig. S7). It is known smokingis associated with high-grade bladder cancer (31). In comparisonwith nonsmokers, smokers with bladder cancer had reducedexpression of CPT1B (Fig. 4D). Likewise, the levels of CPT1Bwere also reduced in high-grade compared with low-grade blad-

der cancer cell lnes (Fig. 4E). Consistent with the CPT1B expres-sion, high-grade bladder cancer cell lines displayed low FAOactivity compared to low-grade bladder cancer cell lines(Fig. 4F). We measured the levels of acylcarnitines, which areindicators of FAO, and found that the levels of palmitoyl andoctanoyl carnitines were significantly lower in high-grade bladdercancer patients (Fig. 4G). Taken together, our results demonstrat-ed the co-occurrence of CPT1B downregulation and low FAOlevels in high-grade bladder cancer.

Hallmark cancer pathways are associated with low CPT1Bexpression

To obtain additional insights into the effects of CPT1B onmetabolism in high-grade bladder cancer, an in silico analysis was

Figure 4.

Suppression of CPT1B is associatedwith poor survival, low b-oxidation,and low levels of acylcarnitines inhigh-grade bladder cancer. A, Lowexpression of CPT1B was associatedwith poor survival in TCGA cohort(log-rank P¼ 0.00123). B, Low CPT1BmRNA expression demonstrates thelow CPT1B expression in high-gradebladder cancer (n¼ 6 and n¼ 12low-grade and high-grade tissuesrespectively; P < 0.0004). C, TMAanalysis of bladder cancer patienttissues (n¼ 9 and n¼ 27, low-gradeand high-grade, respectively) showsa trend for lower intensity in thehigh-grade tissues (P < 0.02). IHCanalysis shows the low CPT1B proteinlevels in high-grade bladder cancer;on a scoring scale of 1–4, one is lowintensity and 4 is the highest intensityof the CPT1B expression. D,mRNAexpression analysis shows the lowCPT1B expression in smokers withbladder cancer (n¼ 8 and n¼ 12nonsmokers and smokers,respectively; P < 0.007). E,mRNAexpression analysis shows a gradualdecrease of CPT1B levels withincreased grades of the bladdercancer in cell lines (P < 0.0001). For allthe mRNAmeasurements, CPT1Blevels were normalized to a GAPDHinternal control. F, LC-MS–basedanalysis using Biocrates AbsoluteIDQp180 Kit shows the low b-oxidationactivity in high-grade bladder cancercell lines (P < 0.02). G, LC-MS–basedanalysis reveals low levels ofpalmitoyl (P < 0.0003) and octanoyl(P < 0.004) carnitines in patients withhigh-grade bladder cancer.

Vantaku et al.

Clin Cancer Res; 25(12) June 15, 2019 Clinical Cancer Research3696

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

performed on data divided into CPT1B high versus CPT1B lowsignature using public gene expression data sets. For each tran-scriptomics cohort that profiled the CPT1B gene (TCGA, Kim,Sjodahl, and Lindgren), we stratified the patients based onCPT1Bgene expression. Next, we divided the patients based on themedian CPT1B value into top 50% and bottom 50%. We thencomputed a gene signature using a parametric t-test and a min-imumdifference in expressionof 1.25 foldbetween thepatients inthe top 50% of CPT1B expression and bottom 50% of CPT1Bexpression. For each cohort, the signature consisted of the genesthat were differentially expressed by at least 1.25-fold between thepatients with high and low CPT1B expression, respectively. Wereferred the obtained signatures as "CPT1B high vs. CPT1B low"signatures. We used GSEA method to determine the enrichedpathways of the CPT1B high versus CPT1B low signatures ineach cohort. A total of 17 pathways were uniformly enriched(FDR-adjusted Q-value < 0.25) in all five datasets (Fig. 5A). Thesepathways include inflammation/immune system, metabolism,and oncogenic pathways (Fig. 5A). Notably, E2F targets and theEMTpathwaywere enrichedwith a strictQ-value <0.0001, furthersupporting the biological relevance of CPT1B in bladder cancerprogression (Fig. 5B).

Effects of CPT1B overexpression in high-grade invasive bladdercancer cell lines

To determine the effects of CPT1B, in high-grade bladdercancer, we ectopically expressed the CPT1B in high-gradeUMUC3bladder cancer cells (Fig. 6A). The CPT1B overexpressed cellsdisplayed increased FAO activity (Fig. 6B), with increased levelsof acyl carnitines and reduced levels of fatty acids compared withcontrol cells (Fig. 6C). EMT is one of the major hallmarks of themetastasis in high-grade cancer. The CPT1B overexpressed cellsdisplayed decreased levels of the EMTmarkers vimentin and snailcompared with vector control (Fig. 6D).

CPT1B overexpression in in vivo reduces tumor growth, EMT,and liver metastasis in CAM model

Given that CPT1B loss and FAO disruption led to significantlyincreased activity in pathways associated with EMT and invasion(Fig. 5), we hypothesized that CPT1B overexpression leads to areduction in the metastatic capabilities of cancer cells. To func-tionally test that hypothesis, we performed in vivo CAM xenograftassay (32) using CPT1B overexpressing cells. Compared to vectorcontrol UMUC3 cells, CPT1B overexpression cells demonstratedreduced tumor growth in the xenograft assays (Fig. 6E and F). IHCconfirmed that the CPT1B overexpression CAM tumors had highE-cad levels, but low N-cad and vimentin levels (Fig. 6G). Wetested the metastatic potential of the CPT1B overexpression cellsby Alu PCR. There was decreased metastasis to liver of the chickembryo from CPT1B overexpression tumors than controls(Fig. 6H). Overall, our results confirmed the role of CPT1B intumor growth andmetastasis in high-grade bladder cancer. Takentogether, our results suggested that CPT1B downregulationimpairs FAO and increases the tumor growth and metastasis inbladder cancer (Fig. 6I).

DiscussionTreatments for high-grade muscle-invasive and metastatic

bladder cancer have not advanced beyond cisplatin-basedcombination chemotherapy (33). The median survival time of

patients with high-grade MIBC is only 14 to 15 months withcisplatin-based chemotherapy (15), highlighting the need fornovel therapeutic approaches. Genomic integrative studiesshowed that genes involved in the PI3-kinase/AKT/mTOR,RTK/RAS, and CDKN2A/CDK4/CCND1 pathways are associatedwith high-grade bladder cancer (26). We used integratedapproach combining unbiased metabolomics, lipidomics, andtranscriptomics to investigate the progression of high-grade blad-der cancer and to delineate opportunities for therapeutic andprognostic intervention.

Alteration of metabolism is one of the hallmarkers for cancerprogression (34). Metabolic alterations enable cancer cells tomeet the increased energy demands for proliferation and survivalin the nutrient-deprived tumor microenvironment (35). Ourmetabolomics analysis identified 519 differentially metabolitesthat were differentially produced in high-grade bladder cancer,including glycolysis, TCA cycle, PPP pathway, polyamines, bileacids, carnitine, prostaglandins, and nucleotides indicating thatmetabolism is a highly dynamic process in bladder cancer. Unbi-ased lipidomics analysis identified 19 differentially producedlipids in high-grade bladder cancer, including triglycerides, phos-phatidylcholines, phosphatidylethanolamines, andphosphatidylinositols. The higher levels of triglycerides and low levels ofphosphatidylcholines andphosphatidylethanolamines are in linewith previous studies (10).Weused transcriptomics data from theKim cohort (26) to identify genes that were differentiallyexpressed in high-grade bladder cancer. We mapped the alteredmetabolites and lipids to their corresponding genes and then toenriched pathways. The results from metabolomics, lipidomics,and transcriptomics suggest that there were common deregulatedmetabolic pathways implicated the arachidonic acid, phosphati-dylinositol, and glycerophospholipid metabolism. These path-ways are important in various cancers (36, 37) (38, 39) andwill befocal points in future bladder cancer research. Clinical trials ofdrugs targeting these metabolic alterations in bladder cancerpatients are warranted.

Our integrative multi-omics analysis identified 27 commongenes that were implicated by metabolomics, lipidomics, andtranscriptomics data. Each of these genes has a profound impacton the cellular metabolism, and particularly lipid metabolismhighlighting the importance of fatty acids in cell proliferation andtumorigenesis (40). The upregulated genes (PLTP, PLA2G4C,LIPG, PIGS, PIGT, PIGU, PLCXD1, ATP8B2, PIGY, PTDSS1,ATP8B4) and downregulated genes (PIGB, PIGG, PIGN, PLCE1,PLA2G4A, CPTIB,PIGV, ATP8B1, CEL, PLA2G4F, PLCG2,PLA2G4B, INPP5D, PLCD3, PIGZ, PLA2G10) provided a genesignature that was associated with poor survival in Lindgren, Kim,and Sjodhal cohorts. Further characterization and validation ofthis gene signature may provide a prognostic marker for high-grade bladder cancer. Some of these genes in the signature havebeen reported to be involved in themetastasis of other cancers; forexample, PLTP, and LIPG upregulation in breast cancer (41, 42),ATP8B1 in colorectal cancer (43), and PLCG2 in chronic lym-phocytic leukemia (44). So far, the role of these genes in high-grade bladder cancer remains unclear. Further characterizationand functional validation of the genes could provide insights intothe molecular events that are related to progression of bladdercancer from low-grade to high-grade.

Our integrated analysis of these omics datasets revealed thepreviously unappreciated importance of CPT1B in high-gradebladder cancer. CPT1B resides at the outer mitochondrial

Multi-omics: Identification of CPT1B Role in Bladder Cancer

www.aacrjournals.org Clin Cancer Res; 25(12) June 15, 2019 3697

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

membrane and regulates intracellular lipid metabolism bytransporting long-chain fatty acids into the mitochondria forb-oxidation (45, 46). Overexpression of fatty acid synthase (47),leads to greater accumulation of fatty acids and low expression ofCPT1B in high-grade bladder cancer. That, in turn, supports therequirement of actively proliferating cells for membrane structur-al and signaling lipids. FAO and fatty acid synthesis are mutuallyantagonistic process (48). The low abundance of palmitoyl andoctanoyl carnitine and low FAO activity observed in patients withhigh-grade bladder cancer are consistent with low CPT1B expres-sion observed in patient with high-grade bladder cancer and celllines. GSEA of genes associated with lowCPT1B expression in fivedistinct public bladder cancer cohorts revealed alterations ofmany immunological, metabolic, and oncogenic pathways thatare known hallmarks of cancer. Low CPT1B expression wasassociated with high levels of EMT which regulates metastasis incancer by conferring an invasive phenotype (49), is the salient

feature of high-grade bladder cancer. CPT1B over expressionreduced the cell proliferation and EMT markers supporting therelevance of FAO in cellular transformation and bladder cancerprogression towards high grade disease. Furthermore, CPT1B overexpression in in vivo significantly reduced tumor growth, andmetastasis, suggesting that a strategy to increase the CPT1B levelsin high grade bladder cancer might help to inhibit tumor growthand metastasis.

In conclusion, for the first time, we identified critical alterationsof metabolite and lipid profiles that contribute to bladder cancerprogression by levaraging high resolution LC-MS based unbiasedmetabolomics, and lipidomics in combinationwith bioinformat-ics analysis. Our integrated study of low-grade versus high-gradebladder cancer provided numerous novel insights into diseasebiology and delineated pathways that provide potential oppor-tunities for therapeutic intervention. The low levels of FAO anddownregulation of CPT1B in high-grade bladder cancer indicate

0 2 4 6 8 10 12

XENOBIOTIC_METABOLISM

TNFA_SIGNALLING_VIA_NFKB

IL6_JAK_STAT3_SIGNALLING

IL2_STAT5_SIGNALLING

INFLAMMATORY_RESPONSE

PI3K_AKT_MTOR_SIGNALLING

REACTIVE_OXYGEN_SPECIES_PATHWAY

GLYCOLYSIS

MTORC1_SIGNALLING

MITOTIC_SPINDLE

ANGIOGENESIS

APOPTOSIS

KRAS_SIGNALLING

MYC_TARGETS_V1

G2M_CHECKPOINT

EPITHELIAL_MESENCHYMAL_TRANSITION

E2F_TARGETS

TCGA -BLCA Lindgren (GSE32548) KIM (GSE 13507) Sjodahl (GSE32894)

Enric

hmen

tsc

ore

SjodahlGSE32894

KimGSE13507

LindgrenGSE32548

TCGABLCA

CPT1B Low

CPT1B High

CPT1B Low

CPT1B High

CPT1B Low

CPT1B High

CPT1B Low

CPT1B High

NES = 9.29NES = 5.93 NES = 7.37 NES = 9.06

Q < 0.0001Q < 0.0001Q < 0.0001 Q < 0.0001

EMT

Enric

hmen

tsc

ore

NES = 7.77

Q < 0.0001

NES = 8.27

Q < 0.0001

NES = 10.12

Q < 0.0001

NES = 4.98

Q < 0.0001E2

FTa

rget

s

B

Oncogenic pathways

Metabolism

Inflammation/Immune system

A

Log2 fold change between CPT1B low (bottom 50%) and CPT1B high (top 50%)

Figure 5.

Low expression of CPT1B is associatedwith key oncogenic, metabolic pathwaysin high-grade bladder cancer. A, Gene setenrichment analysis (GSEA) revealedthat genes associated with CPT1Bdownregulation are significantly enrichedin oncogenic, metabolic, and immunepathways in the Kim, Lindgren, Sjodahl,and TCGA cohorts. B, CPT1Bdownregulation was highly associatedwith enrichment of E2F targets and EMTsignatures in all cohorts. The values on they-axis for each graph representenrichment scores (corresponding to themagnitude of the enrichment for eachanalysis). For each graph, the normalizedenrichment score (NES, computed via theGSEA analysis) and the significance of theenrichment (q¼ FDR, also computed viathe GSEA analysis). The NES scoresranged from 4 to 11 (all q < 0.0001).

Vantaku et al.

Clin Cancer Res; 25(12) June 15, 2019 Clinical Cancer Research3698

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

mR

NA

exp

ressio

n

0.3

0.2

0.1

0.0

P < 0.05

CPT1B

Vec CPT1B OE

A B

DVec CPT1B OE

0.14

0.16

0.18

0.20

0.22

β-Oxidation activity

Rel

ativ

ea b

unda

nce

Heptadecanoic acidDocosapentaenoic acidDocosatetraenoic acidOctaedecAnoic acidEicosapentaenoic acidEicosatrienoic acidOleic acidDocosahexaenoic acidEicosapentaenoic acidEicosatetraenoic acidPalmitoelic acidHexadecanoic acidOctadecadienoic acidStearic acid

−2 0 2

C

Acyl

car

nitin

esFa

tty a

cids

Stearoyl carnitinePalmitoylcarnitineMyristoyl carnitines

Vector control

F

CPT1B

Actin

Vec OE

Vec OE

P < 0.05

Rel

ativ

e ab

unda

nce 1.5

1.0

0.5

0.0

Vec CPT1B OE

P < 0.05

P < 0.05

5

4

3

2

1

0

mR

NA

expre

ssio

n

VimSnail

Actin

Vec OE

Rel

ativ

e ab

unda

nce 0.8

0.6

0.4

0.2

0.0Vec OEVec OE

Vector control CPT1B OE

CPT1B

N-cad

E-cad

E

G

Vector control CPT1B OE

Vim

Vim

Actin

CPT1B OE

CPT1B OE

0.0

0.5

1.0

1.5

Rel

ativ

eab

unda

nce

0.000

0.001

0.002

0.003

0.004

0.005

mR

NA

expr

essi

on

Snail

Vec

Tota

lflu

x[p

/s]

2.0 x 108

1.5 x 108

1.0 x 108

5.0 x 107

0.00

0

2

4

6

8

CAM tissues after 7 days

Hum

ange

nom

icD

NA

(ng)

per

tota

lgen

omic

DN

A(u

g) Vector controlCPT1B OE

Scale: 50mm

Scale: 10mm

H

0.0

0.2

0.4

0.6

0.8

1.0

Rel

ativ

ea b

unda

nce

Vector control

CPT1 B OE

Lung Liver Bone BrainCPT1B E-cad N-cad Vim

I Low CPT1B

Cytosol Mitochondrial matrix

Cytosol Mitochondrial matrix

LG BLCA HG BLCA

Fatty acyl CoA + carntines

Acyl carnitines

Acyl carnitines

Acyl carnitines b-oxidation

Energy

Acyl carnitines b-oxidation

Energy

Fatty acyl CoA + carntines

Acyl carnitines

Acyl cartines

P < 0.02

CPT1B OE

Vector control

P < 0.05

P < 0.05

Fatty acids Carntines+ Acyl CoA

Fatty acids

Acyl CoA Carntines+

P < 0.04

P < 0.05

Figure 6.

Effects of CPT1B in high-grade bladder cancer cell lines. A, Adenovirus-mediated overexpression of CPT1B in the high-grade bladder cancer cell line UMUC3compared with a vector control, confirmed by qPCR andWestern blot analysis. B, CPT1B overexpression increases the b-oxidation in high-grade bladdercancer cells. C, Heatmap of increased acyl carnitine levels and reduced fatty acid levels in CPT1B-overexpressed cells compared with control cells (FDR-correctedP < 0.05). D, Expression of the EMTmarkers vimentin and snail in CPT1B overexpressed cells compared with that in vector control cells. E, CPT1B overexpressedcells formed significantly smaller tumors on chicken CAMs as measured by total photon flux of the region of interest (day 15; red circle; F).G, CPT1Boverexpression significantly increased the E-cad, reduced the N-cad and vimentin levels in tumors after 7 days of tumor growth on CAMs, as evidenced by IHCanalysis and quantification. H, CPT1B overexpression reduced the metastasis of UMUC3 cells into the chick embryo liver tissue as measured by RT-PCR of ahuman-specific Alu sequence in total genomic DNA (gDNA). I,Graphical representation of the role of CPT1B in bladder cancer progression. OE, overexpression.

Multi-omics: Identification of CPT1B Role in Bladder Cancer

www.aacrjournals.org Clin Cancer Res; 25(12) June 15, 2019 3699

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

that the impairment of b-oxidation plays an important role inbladder cancer tumor progression. These results may have prog-nostic value andprovide targets for therapeutic intervention in thefuture. Taken together, our results suggest that interventions thatincrease the CPT1B levels may have therapeutic value for high-grade bladder cancer patients that are notably dependent on FAOmetabolism.

Disclosure of Potential Conflicts of InterestA.G. Sikora is an employee of Ovodex, reports receiving commercial research

grants from Tessa Therapeutics, other commercial research support fromAdvaxis, and holds ownership interest (including patents) in patents relatedto CAM model. No potential conflicts of interest were disclosed by the otherauthors.

Authors' ContributionsConception and design: V. Vantaku, C.R. Ambati, S.R. Donepudi, V. Putluri,F.-C. von Rundstedt, H. Villanueva, C. Coarfa, N. PutluriDevelopment of methodology: V. Vantaku, J. Dong, C.R. Ambati, D. Perera,S.R. Donepudi, V. Putluri, B. Karanam, C. Coarfa, N. PutluriAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): C.R. Ambati, S.R. Donepudi, V. Putluri,M. Villanueva, F.-C. von Rundstedt, L.Y. Ballester, M.K. Terris, R.J. Bollag,S.P. Lerner, A.B. Apolo, H. Villanueva, A.G. Sikora, Y. LotanAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): V. Vantaku, J. Dong, D. Perera, C.S. Amara,M.J. Robertson, D.W.B. Piyarathna, M. Villanueva, M.K. Terris, H. Villanueva,M. Lee, A.G. Sikora, Y. Lotan, C. Coarfa, N. PutluriWriting, review, and/or revision of the manuscript: V. Vantaku, J. Dong,C.R. Ambati, D. Perera, S.R. Donepudi, C.S. Amara, F.-C. von Rundstedt,

L.Y. Ballester, S.P. Lerner, A.B. Apolo, H. Villanueva, A.G. Sikora, Y. Lotan,A. Sreekumar, C. Coarfa, N. PutluriAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): M. Villanueva, H. Villanueva, A. Sreekumar,C. CoarfaStudy supervision: Y. Lotan, C. Coarfa, N. PutluriOther (in vitro/in vivo experiments): S.S. Ravi

AcknowledgmentsThis research was fully supported by American Cancer Society (ACS) Award

127430-RSG-15-105-01-CNE (N. Putluri), NIH/NCI R01CA220297(N. Putluri), and NIH/NCI R01CA216426 (N. Putluri), partially supported bythe following grants: NIH/NCI U01 CA167234 (A.S.K), CPRIT RP170295(C.C.), as well as funds from Alkek Center for Molecular Discovery (A.Sreekumar). This project was also supported by the Agilent Technologies Centerof Excellence (COE) in Mass Spectrometry at Baylor College of Medicine,Metabolomics Core, Human Tissue Acquisition and Pathology at BaylorCollege of Medicine with funding from the NIH (P30 CA125123), CPRITProteomics and Metabolomics Core Facility (N. Putluri; RP170005), and DanL. Duncan Cancer Center. CAM assay was supported by the Patient-DerivedXenograft andAdvanced in vivoModelsCore Facility atBaylorCollegeofMedicinewith funding from theCancer Prevention and Research Institute of Texas (CPRIT)grant #170691. We would like to thank the team of the Georgia Cancer CenterBiorepository / BRAG-Onc for biospecimen collection and annotation.

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received May 15, 2018; revised November 9, 2018; accepted March 6, 2019;published first March 7, 2019.

References1. Siegel RL, Miller KD, Jemal A. Cancer Statistics, 2017. CA Cancer J Clin

2017;67:7–30.2. Lin LL, Hsia CR, Hsu CL, Huang HC, Juan HF. Integrating transcriptomics

and proteomics to show that tanshinone IIA suppresses cell growth byblocking glucose metabolism in gastric cancer cells. BMC Genomics 2015;16:41.

3. Koplev S, Lin K, Dohlman AB, Ma'ayan A. Integration of pan-cancertranscriptomics with RPPA proteomics reveals mechanisms of epithelial-mesenchymal transition. PLoS Comput Biol 2018;14:e1005911.

4. Ettinger DS, Wood DE, Akerley W, Bazhenova LA, Borghaei H, CamidgeDR, et al. Non-small cell lung cancer, version 6.2015. J Natl Compr CancNetw 2015;13:515–24.

5. Gradishar WJ, Anderson BO, Balassanian R, Blair SL, Burstein HJ, Cyr A,et al. Breast Cancer Version 2.2015. J Natl Compr Canc Netw 2015;13:448–75.

6. Li R, Choi W, Ferguson Rd JE, Metcalfe MJ, Kamat AM. New discoveries inthe molecular landscape of bladder cancer. F1000Res 2016;5:2875.

7. Pavlova NN, Thompson CB. The emerging hallmarks of cancer metabo-lism. Cell Metab 2016;23:27–47.

8. Putluri N, Shojaie A, Vasu VT, Vareed SK, Nalluri S, Putluri V, et al.Metabolomic profiling reveals potential markers and bioprocesses alteredin bladder cancer progression. Cancer Res 2011;71:7376–86.

9. von Rundstedt FC, Rajapakshe K,Ma J, Arnold JM, Gohlke J, Putluri V, et al.Integrative pathway analysis of metabolic signature in bladder cancer: alinkage to the cancer genome atlas project and prediction of survival. J Urol2016;195:1911–9.

10. Piyarathna DWB, Rajendiran TM, Putluri V, Vantaku V, Soni T,von Rundstedt FC, et al. Distinct lipidomic landscapes associated withclinical stages of urothelial cancer of the bladder. Eur Urol Focus 2018;4:907–15.

11. Park JC, Citrin DE, Agarwal PK, Apolo AB. Multimodal management ofmuscle-invasive bladder cancer. Curr Probl Cancer 2014;38:80–108.

12. Vantaku V, Donepudi SR, Ambati CR, Jin F, Putluri V, Nguyen K, et al.Expression of ganglioside GD2, reprogram the lipid metabolism and EMTphenotype in bladder cancer. Oncotarget 2017;8:95620–31.

13. Amara CS, Ambati CR, Vantaku V, PiyarathnaDWB, Donepudi SR, Ravi SS,et al. Serummetabolic profiling identified a distinct metabolic signature inbladder cancer smokers: a key metabolic enzymes associated with patientsurvival. Cancer Epidemiol Biomarkers Prev 2019.

14. Vantaku V,Donepudi SR, PiyarathnaDWB, Amara CS, Ambati CR, TangW,et al. Large-scale profiling of serum metabolites in African American andEuropean American patients with bladder cancer reveals metabolic path-ways associated with patient survival. Cancer 2019;125:921–32.

15. Bligh EG, Dyer WJ. A rapid method of total lipid extraction and purifica-tion. Can J Biochem Physiol 1959;37:911–7.

16. ChambersMC,MacleanB, Burke R, AmodeiD, RudermanDL,Neumann S,et al. A cross-platform toolkit for mass spectrometry and proteomics.Nat Biotechnol 2012;30:918–20.

17. Kind T,Meissen JK, YangD, Nocito F, Vaniya A, Cheng YS, et al. Qualitativeanalysis of algal secretions with multiple mass spectrometric platforms.J Chromatogr A 2012;1244:139–47.

18. Ejsing CS, Duchoslav E, Sampaio J, Simons K, Bonner R, Thiele C, et al.Automated identification and quantification of glycerophospholipidmolecular species by multiple precursor ion scanning. Anal Chem 2006;78:6202–14.

19. LiM, Pathak RR, Lopez-Rivera E, Friedman SL, Aguirre-Ghiso JA, SikoraAG.The in ovo chick chorioallantoic membrane (CAM) assay as an efficientxenograft model of hepatocellular carcinoma. J Vis Exp 2015;104: doi:10.3791/52411.

20. Quigley JP, Armstrong PB. Tumor cell intravasation alu-cidated: the chickembryo opens the window. Cell 1998;94:281–4.

21. Zijlstra A, Mellor R, Panzarella G, Aimes RT, Hooper JD, Marchenko ND,et al. A quantitative analysis of rate-limiting steps in the metastatic cascadeusing human-specific real-time polymerase chain reaction. Cancer Res2002;62:7083–92.

22. Wishart DS, Tzur D, Knox C, Eisner R Guo AC, Young N, et al. HMDB: thehuman metabolome database. Nucleic Acids Res 2007;35:D521–6.

23. Kim WJ, Kim EJ, Kim SK, Kim YJ, Ha YS, Jeong P, et al. Predictive value ofprogression-related gene classifier in primary non-muscle invasive bladdercancer. Mol Cancer 2010;9:3.

Vantaku et al.

Clin Cancer Res; 25(12) June 15, 2019 Clinical Cancer Research3700

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

24. Lindgren D, Sj€odahl G, Lauss M, Staaf J, Chebil G, L€ovgren K, et al.Integrated genomic and gene expression profiling identifies two majorgenomic circuits in urothelial carcinoma. PLoS One 2012;7:e38863.

25. Sjodahl G, Lauss M, L€ovgren K, Chebil G, Gudjonsson S, Veerla S, et al. Amolecular taxonomy for urothelial carcinoma. Clin Cancer Res 2012;18:3377–86.

26. The Cancer Genome Atlas Research Network. Comprehensive molecularcharacterization of urothelial bladder carcinoma. Nature 2014;507:315–22.

27. Xiong J. Fatty acid oxidation in cell fate determination. Trends Biochem Sci2018;43:854–7.

28. Pascual G, Avgustinova A, Mejetta S, Martín M, Castellanos A, Attolini CS,et al. Targeting metastasis-initiating cells through the fatty acid receptorCD36. Nature 2017;541:41–5.

29. Luo X, Cheng C, Tan Z, Li N, Tang M, Yang L, et al. Emerging roles of lipidmetabolism in cancer metastasis. Mol Cancer 2017;16:76.

30. Beloribi-Djefaflia S, Vasseur S,GuillaumondF. Lipidmetabolic reprogram-ming in cancer cells. Oncogenesis 2016;5:e189.

31. Jin F, Thaiparambil J, Donepudi SR, Vantaku V, Piyarathna DWB, Maity S,et al. Tobacco-specific carcinogens induce hypermethylation, DNAAdducts, and DNA damage in bladder cancer. Cancer Prev Res 2017;10:588–97.

32. Deryugina EI, Quigley JP. Chick embryo chorioallantoicmembranemodelsystems to study and visualize human tumor cell metastasis.Histochem Cell Biol 2008;130:1119–30.

33. Spiess PE, Agarwal N, Bangs R, Boorjian SA, Buyyounouski MK, Clark PE,et al. Bladder Cancer, Version 5.2017, NCCN clinical practice guidelines inoncology. J Natl Compr Canc Netw 2017;15:1240–67.

34. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell2011;144:646–74.

35. Cairns RA, Mak TW. An Alternative Sugar Fuels AML. Cancer Cell 2016;30:660–2.

36. Liu P, Cheng H, Roberts TM, Zhao JJ. Targeting the phosphoinositide3-kinase pathway in cancer. Nat Rev Drug Discov 2009;8:627–44.

37. Dolce V, Cappello AR, Lappano R, Maggiolini M. Glycerophospholipidsynthesis as a novel drug target against cancer. CurrMol Pharmacol 2011;4:167–75.

38. Yang P, Cartwright CA, Li J, Wen S, Prokhorova IN, Shureiqi I, et al.Arachidonic acid metabolism in human prostate cancer. Int J Oncol 2012;41:1495–503.

39. Chaudry AA, Wahle KW, McClinton S, Moffat LE. Arachidonic acidmetabolism in benign and malignant prostatic tissue in vitro: effectsof fatty acids and cyclooxygenase inhibitors. Int J Cancer 1994;57:176–80.

40. Fritz V, Fajas L. Metabolism and proliferation share common regulatorypathways in cancer cells. Oncogene 2010;29:4369–77.

41. Gallo LH, Ko J, DonoghueDJ. The importance of regulatory ubiquitinationin cancer and metastasis. Cell Cycle 2017;16:634–48.

42. Slebe F, Rojo F, Vinaixa M, García-Rocha M, Testoni G, Guiu M, et al. FoxAand LIPG endothelial lipase control the uptake of extracellular lipids forbreast cancer growth. Nat Commun 2016;7:11199.

43. Aziz MA, Periyasamy S, Yousef ZA, AlAbdulkarim I, Otaibi MA, Alfahed A,et al. Integrated exon level expression analysis of driver genes explain theirrole in colorectal cancer. PLoS One 2014;9:e110134.

44. Zhang SQ, Smith SM, Zhang SY, Lynn Wang Y. Mechanisms of ibrutinibresistance in chronic lymphocytic leukaemia and non-Hodgkin lympho-ma. Br J Haematol 2015;170:445–56.

45. Britton CH, Schultz RA, Zhang B, Esser V, Foster DW, McGarry JD. Humanliver mitochondrial carnitine palmitoyltransferase I: characterization of itscDNA and chromosomal localization and partial analysis of the gene.Proc Natl Acad Sci U S A 1995;92:1984–8.

46. Weis BC, Esser V, Foster DW, McGarry JD. Rat heart expresses two forms ofmitochondrial carnitine palmitoyltransferase I. The minor component isidentical to the liver enzyme. J Biol Chem 1994;269:18712–5.

47. Sugino T, Baba K, Hoshi N, Aikawa K, Yamaguchi O, Suzuki T. Over-expression of fatty acid synthase in human urinary bladder cancer andcombined expression of the synthase and Ki-67 as a predictor of prognosisof cancer patients. Med Mol Morphol 2011;44:146–50.

48. McGarry JD, Brown NF. The mitochondrial carnitine palmitoyltransferasesystem. From concept to molecular analysis. Eur J Biochem 1997;244:1–14.

49. Davis FM, Stewart TA, Thompson EW, Monteith GR. Targeting EMT incancer: opportunities for pharmacological intervention. Trends PharmacolSci 2014;35:479–88.

www.aacrjournals.org Clin Cancer Res; 25(12) June 15, 2019 3701

Multi-omics: Identification of CPT1B Role in Bladder Cancer

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515

2019;25:3689-3701. Published OnlineFirst March 7, 2019.Clin Cancer Res   Venkatrao Vantaku, Jianrong Dong, Chandrashekar R. Ambati, et al.   Metabolism in Bladder CancerPatient Survival and Identifies CPT1B Effect on Fatty Acid Multi-omics Integration Analysis Robustly Predicts High-Grade

  Updated version

  10.1158/1078-0432.CCR-18-1515doi:

Access the most recent version of this article at:

  Material

Supplementary

  http://clincancerres.aacrjournals.org/content/suppl/2019/03/07/1078-0432.CCR-18-1515.DC1

Access the most recent supplemental material at:

   

   

  Cited articles

  http://clincancerres.aacrjournals.org/content/25/12/3689.full#ref-list-1

This article cites 47 articles, 9 of which you can access for free at:

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.org

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/25/12/3689To request permission to re-use all or part of this article, use this link

on February 27, 2020. © 2019 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 7, 2019; DOI: 10.1158/1078-0432.CCR-18-1515