Serum Glutamate Levels Correlate with Gleason Score and ... › ... › 18 › 21 ›...

Human Cancer Biology

Serum Glutamate Levels Correlate with Gleason Score andGlutamate Blockade Decreases Proliferation, Migration, andInvasion and Induces Apoptosis in Prostate Cancer Cells

Shahriar Koochekpour1,2, Sunipa Majumdar1, Gissou Azabdaftari3, Kristopher Attwood2,4, Ray Scioneaux1,Dhatchayini Subramani1, Charles Manhardt1, Giovanni D. Lorusso5, Stacey S. Willard1, Hillary Thompson6,Mojgan Shourideh1, Katayoon Rezaei8, Oliver Sartor7, James L. Mohler2,10,11, and Robert L. Vessella9

AbstractPurpose:During glutaminolysis, glutamine is catabolized to glutamate and incorporated into citric acid

cycle and lipogenesis. Serum glutamate levels were measured in patients with primary prostate cancer or

metastatic castrate-resistant prostate cancer (mCRPCa) to establish clinical relevance. The effect of glutamate

deprivation or blockade by metabotropic glutamate receptor 1 (GRM1) antagonists was investigated on

prostate cancer cells’ growth, migration, and invasion to establish biologic relevance.

Experimental Design: Serum glutamate levels weremeasured in normalmen (n¼ 60) and patients with

primary prostate cancer (n ¼ 197) or mCRPCa (n ¼ 109). GRM1 expression in prostatic tissues was

examined using immunohistochemistry (IHC). Cell growth, migration, and invasion were determined

using cell cytotoxicity and modified Boyden chamber assays, respectively. Apoptosis was detected using

immunoblotting against cleaved caspases, PARP, and g-H2AX.

Results: Univariate and multivariate analyses showed significantly higher serum glutamate levels in

Gleason score� 8 than in the Gleason score� 7 and in African Americans than in the Caucasian Americans.

African Americans with mCRPCa had significantly higher serum glutamate levels than those with primary

prostate cancer orbenignprostate.However, inCaucasianAmericans, serumglutamate levelswere similar in

normal research subjects andpatientswithmCRPC. IHC showedweakor no expression ofGRM1 in luminal

acinar epithelial cells of normal or hyperplastic glands but high expression in primary ormetastatic prostate

cancer tissues. Glutamate deprivation or blockade decreased prostate cancer cells’ proliferation, migration,

and invasion and led to apoptotic cell death.

Conclusions: Glutamate expression is mechanistically associated with and may provide a biomarker of

prostate cancer aggressiveness. Clin Cancer Res; 18(21); 5888–901. �2012 AACR.

IntroductionProstate cancer is estimated to account for 29% of all

new cancers and is the second leading cause of cancer-

related death in men in the United States in 2012 (1).Age, African-American race, and family history are une-quivocal risk factors for prostate cancer (2). Focusing onearly detection of prostate cancer has the capacity todecrease prostate cancer mortality. Serum prostate-specificantigen (PSA) is used for prostate cancer early detection, butits use is controversial because of its sensitivity and speci-ficity is limited and it does not provide reliable prognosticinformation. These limitations have led investigators toconsider new approaches, using systems biology, to identifybiomarkers of prostate cancer aggressiveness. This goalcould be reached by defining important interacting cellularnetworks and/ormetabolic pathways that reflect tissue- anddisease-specific phenotypes.

Metabolomics has emerged recently as a promisingmeth-od for prostate cancer detection (3), one that may supple-ment, or even replace current PSA testing. Metabolomicsprofiling of prostate cancer offers the potential for earlydetection and discrimination of clinically aggressive tumorsfrom indolent and nonaggressive diseases. Metabolicchanges are affected by genetic and epigenetic changes and

Authors' Affiliations: Departments of 1Cancer Genetics, 2Urology, and3Pathology, Roswell Park Cancer Institute; 4Department of Biostatistics,University of Buffalo, Buffalo, New York; Departments of 5Pathology and6Biostatistics, School of Public Health, Louisiana State University HealthSciences Center; 7Department of Medicine and Urology, Tulane CancerCenter, Tulane University School of Medicine, New Orleans, Louisiana;8Department of Pathology, GeorgeWashington University, Washington, D.C.; 9Department of Urology, University ofWashington and Puget Sound VAHealth Care System, Seattle, Washington; 10Department of Surgery (Divi-sion of Urology) and 11Lineberger Comprehensive Cancer Center, Univer-sity of North Carolina at Chapel Hill, Chapel Hill, North Carolina

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

Corresponding Author: Shahriar Koochekpour, Departments of CancerGenetics and Urology, Center for Genetics and Pharmacology, RoswellPark Cancer Institute, Elm and Carlton Streets, Buffalo, NY 14263. Phone:716-845-3345; Fax: 716-845-1698; E-mail:[email protected]

doi: 10.1158/1078-0432.CCR-12-1308

�2012 American Association for Cancer Research.

ClinicalCancer

Research

Clin Cancer Res; 18(21) November 1, 20125888

on June 15, 2020. © 2012 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst October 16, 2012; DOI: 10.1158/1078-0432.CCR-12-1308

http://clincancerres.aacrjournals.org/

a long list of environmental factors, which include diurnalvariation, dietary habits, and lifestyle. The prostate is aunique target formetabolomics analysis both under normalconditions and upon neoplastic transformation. Extremelyhigh amounts of citrate are produced and secreted by theprostate that reflects the unique state of anabolic andcatabolic processes of secretory luminal cells. Citratemetab-olism differs in prostate cancer compared with benignprostatic hyperplasia (BPH; ref. 4).Glutamate is a nonessential amino acid and an excitatory

neurotransmitter that is involved in neuroembryonicgrowth and development by regulating proliferation, sur-vival, migration, and invasion of neuronal progenitors andimmature neurons (5). A prominent consequence of sig-naling in rapidly proliferating cancer cells is increasednutrient uptake. Glucose and glutamine are the 2 mostabundant nutrients in mammals. During glycolysis andglutaminolysis, glucose is converted to lactate and gluta-mine to glutamate, respectively (6). In addition to being aprecursor of other amino acids and nucleotides, glutamatemetabolism is closely linked to ureagenesis, glutathionesynthesis, tricarboxylic acid cycle (TCA), amino acid trans-ferase, g-aminobutyric acid synthesis, lipogenesis, and ATPproduction (6). More than 80 years ago, Otto Warburgshowed aerobic glycolysis in tumor cells by their high rateof wasteful glucose consumption, increased lactate produc-tion, and bioenergetics demand despite accessibility tooxygen (7). This shift from mitochondrial ATP productionwas later proved to be a hallmark of aggressive tumors andrapidly proliferating nontransformed cells, such as lympho-cytes, thymocytes, and erythrocytes (8). In the PC-3 prostate

cancer cell line, c-Myc oncogenic transcription increasesmitochondrial glutaminase expression that leads toenhanced glutaminolytic activity, catabolic conversion ofglutamine to glutamate, and glutamine addiction (9). Tay-lor and colleagues showed elevated levels of glutamate in 12prostate cancer compared with 16 BPH tissues andincreased levels of additional metabolites in the urea cycleusing high throughput quantitative humoral response pro-filing (10). Glutamate levels were higher in the majority ofpatientswithprostate cancer than inBPH tissues in a follow-up study by the same group (11).

Glutamate in secreted form is a phylogenetically con-served cell signalingmolecule in addition to its intracellularactivities inbioenergetics, biosynthetic pathways,maintain-ing amino acids andnucleotide pool, andmetabolism (12).The glutamatergic system comprises the glutamate, gluta-mate receptors (GluR), and glutamate transporters. TheGluRs are divided into 2 different categories, the ionotropicglutamate receptors (iGluR) and the metabotropic gluta-mate receptors (mGluR; ref. 13). The iGluRs are gated ionchannels with excitatory action (13). The mGluRs promotecellular signaling pathways via 7 transmembrane domainG-protein–coupled receptors (GPCR) and are subdividedinto 3 groups and 8 subtypes: group I (mGluR1 andmGluR5), II (mGluR2 and mGluR3), and III (mGluR4,mGluR6, mGluR7, and mGluR8) based on their pharma-cology, sequence homology, response to agonists, anddownstream signaling. mGluRs are expressed in a varietyof nonneuronal cell types, such as hepatocytes, melano-cytes, keratinocytes, myocardial cells, pancreatic cells, andembryonic stemcells (14, 15). Recently, several studies haveindicated the involvement of mGluRs in tumorigenesis.mGluRs are highly overexpressed in the nervous system,so initial discoveries of mGluRs in human tumors werereported in neuro-glial–derived tumors, such as gliomas,neuroblastoma, andmedulloblastoma (14, 15).Glioma cellssecreting high amounts of glutamate have higher growthrates than their isogenic parental cells (16). Glutamate antag-onists decreased proliferation of a variety of cancer cells,which include colon adenocarcinoma, melanoma, lung car-cinoma, thyroid carcinoma, breast carcinoma, astrocytoma,neuroblastoma, and rhabdomyosarcoma (17). The firstproof for the involvementofmGluRs innonneuronal tumor-igenesis was obtained by Chen and colleagues (18, 19). Atransgenic mouse line overexpressing mGluR1 (also knownas GRM1) in melanocytes was predisposed to spontaneousmelanoma development with 100% penetrance, short laten-cy, and highmetastatic potential (18, 19). In prostate cancer,mGluR1–5 mRNA expression was reported in PC-3 andLNCaP cells and mGluR6 and 8 were expressed only inLNCaP cells. Dihydrotestosterone (DHT) induced mGluR7expression and inhibited mGluR8 expression in LNCaPcells (20). The clinical significance or biologic relevance ofglutamate in prostate cancer has not been investigated.

In this study,we examined the associationbetween serumglutamate levels and prostate cancer aggressiveness toestablish clinical relevance. To determine the biologic rel-evance, the effect of glutamate deprivation or blockade was

Translational RelevanceAerobic glycolysis is a hallmark of proliferating nor-

mal and malignant cells. During glutaminolysis, gluta-mine is catabolized to glutamate and incorporated intothe citric acid cycle and lipogenesis. Glutamate plays asubstantial and overlapping role betweenmetabolic andoncogenic signaling pathways. In this study, we foundthat serum glutamate levels were increased in patientswith primary prostate cancer and directly correlatedwithGleason score and prostate cancer aggressiveness. Com-paredwithCaucasianAmericans, serumglutamate levelswere higher in African Americans with metastatic cas-trate-resistant prostate cancer (mCRPCa) than thosewith primary tumors. Immunohistochemical stainingshowed weak or no expression of metabotropic gluta-mate receptor 1 (GRM1) in normal prostate cells butoverexpression in primary and metastatic prostate can-cer. Glutamate deprivation or blockade with GRM1antagonists reduced prostate cancer cells’ growth, migra-tion, and invasion and proved apoptogenic. Glutamatemay provide a metabolic biomarker of prostate canceraggressiveness, and glutamine metabolism or signalingmay provide a target for therapeutic intervention.

Serum Levels and Biologic Activities of Glutamate in Prostate Cancer

www.aacrjournals.org Clin Cancer Res; 18(21) November 1, 2012 5889




investigated onprostate cancer cells’ growth,migration, andinvasion.

Materials and MethodsPatients and samples

Primary prostate cancer cases investigated in this studyrefer to newly diagnosed, untreated, and clinically localizedprostate cancer. Normal research subjects in this study aredefined as men with no evidence of prostate cancer. Serumsamples used to evaluate glutamate levels in normal indi-viduals andmen with primary tumor or metastatic castrate-resistant prostate cancer (mCRPCa), were obtained fromthe biospecimen core facilities at the Louisiana CancerResearch Consortium (LCRC) affiliated to Tulane MedicalSchool and School of Medicine, Louisiana State UniversityHealth Sciences Center (LSUHSC, New Orleans, LA) andthe tumor bank at the University of Washington MedicalCenter (UWMC, Washington) after informed consent andwithout any personal identifiers. Briefly, peripheral bloodsamples were collected by phlebotomy from consentingnormal adult males or patients with primary or mCRPCa inthe Urology Oncology clinics at the UWMC and LCRC-affiliated hospitals. All of the controls came from patientswho presented at the University of Washington’s annualfree PSA screening week. All patients were more than 40years of age, and all stated that they did not have a history ofprostate cancer. PSA values were below 4.0 ng/mL. Bloodsamples fromUWMCwere processedby centrifugation, andserum samples were stored in aliquots at �80�C within �4hours of draw. Overall, all normal serum samples, 39% ofprimary prostate cancer, and 69%ofmCRPCa samples wereobtained from theUWMC. The remainder of serumsampleswas processed similarly andobtained from thebiospecimencore facilities at LCRC. Serum samples were stored untilused in the research study; no freeze–thaw cycles wereincurred. We did not acquire fasting or postmeal status onany of the research subjects. Prostate cancer aggressivenesswas defined using Gleason score, clinical stage, and PSA atdiagnosis as: (i) high aggressive (Gleason sum� 8, or PSA >20 ng/mL, or Gleason sum¼ 7, and stage cT3–cT4); (ii) lowaggressive (Gleason sum < 7, stage cT1–cT2, and PSA <10ng/mL); or (3) intermediate aggressive (all other cases;ref. 21). The prostate cancer tissue specimens used forimmunohistochemistry (IHC) were obtained from Accu-max Array (Accurate Chemical & Scientific Corp.).

Ethical considerationsEthical approval was obtained from the Institutional

Review Boards at University of Washington Medical Center(Seattle, Washington), Tulane Cancer Center, Tulane Uni-versity School of Medicine (New Orleans, LA), LSUHSC,School of Medicine (New Orleans, LA), and Roswell ParkCancer Institute (Buffalo, NY).

Determination of serum glutamate and creatininelevels

Serum glutamate levels were measured using BioVision’sglutamate assay kit (#K629-100) that provides sensitive

detection method for glutamate in a variety of biologicsamples and fluids including serum, urine, cells, and tissueextracts. The amount of glutamate is quantitated by measur-ing optical density at l ¼ 450 nm in a microplate reader.Background reading was corrected by subtracting the valuefrom the 0 glutamate control from all samples readings.Serum creatinine concentration was measured using theBioVision creatinine assay kit (#K625-100) and the manu-facturer’s instructions. After plotting the independent gluta-mate and creatinine standard curves, glutamate and creati-nine concentrations of the serum samples were calculatedusing the formulaC¼Sa/Svnmol/mL,ormmol/L, inwhichSais the sample amount of unknown (in nmol) from standardcurve and Sv is the sample volume (mL) added to each well.

The minimum detectable amounts for serum glutamateor creatinine were 2 nmol/mL. The standard curve repro-ducibility was shown in 5 independent assays. The intra-assay variance was evaluated in 10 replicates for 3 serumsamples of known concentrations on 1 plate. The interassayprecision was tested in 3 independent assays for 3 controlsamples in triplicates. Stability of serum glutamate wasdetermined by incubating aliquots of serum samples ofknown glutamate concentrations at�20�C or�70�C for 3,6, 12, 18, 24, and 36 months.

Immunohistochemical analysisThe ABC detection system (Vector Laboratories Inc.) was

used for immunostaining as described previously (22).Affinity purified, polyclonal rabbit antihuman-GRM1 anti-body (at 1:250, Sigma)- was used as the primary antibody.The incubation time for primary antibody was overnightat 4�C. The slides were counterstainedwith hematoxylin andmounted for examination. The negative control slides weretreated in an identical manner except that the primary anti-bodies were omitted. Descriptive examinations of cell type–specific anti-GRM1 immunoreactivity in various tissue con-stituents, normal, hyperplastic, and neoplastic glands in thesections were completed by the study pathologists.

Cell Culture, reagents, and Western blot analysisAndrogen-independent (PC-3, DU145, MDA-PCa2b,

VCaP, and 22Rv1) and androgen-sensitive (LNCaP) pros-tate cancer cell lines were purchased from American TypeCulture Collection and maintained in their individuallyspecifiedmedium (23). E006AA cell line wasmaintained inDulbecco’s modified Eagle’s medium (DMEM) supplemen-ted with 10% FBS (23). Noncompetitive GRM1 antagonists(riluzole and BAY36-7620) were purchased from Tocris(24). Customized glutamine- and glutamate-free RPMI-1640 or DMEM, dialyzed FBS, and GlutaMAX-I (a dipep-tide, L-alanine-L-glutamine) were purchased from Life Tech-nologies. GlutaMAX was used as a substitute for glutaminein cell growth, which cannot be degraded or converted intoglutamate. Cells were cultured in complete media up to75% confluency, washed with cold PBS, and whole-celllysates were extracted as described previously (23). GRM1was detected using a rabbit polyclonal antibody againsthuman GRM1 (Cat # 19955-1-AP at 1:1,000; Proteintech

Koochekpour et al.

Clin Cancer Res; 18(21) November 1, 2012 Clinical Cancer Research5890




Group Inc.) using enhanced chemiluminescence (ECL)detection system (GE Healthcare Biosciences). The anti-GRM1 specificity was examined by probing the duplicatemembrane with primary antibody that was preincubatedwith GRM1-specific blocking peptide (Proteintech GroupInc.) at 5 mg/mL for 2 hours with rotation at room temper-ature. Detection of apoptosis markers was conducted usingimmunoblotting. Cells were cultured up to 50% confluencyin complete culture media in the presence or absence ofdimethyl sulfoxide (DMSO) or riluzole at 0, 10, 25, or 50mmol/L for 4 or 6 days andwhole-cell lysates were prepared.Clarified protein samples (35 mg) were subjected to SDS-PAGE under reducing conditions. Western blot analyseswere carried out using monoclonal antibodies againstcleaved caspases-9, -7, -3, and PARP provided in an Apo-ptosis Sampler Kit (Cell Signaling Technology) and phos-phorylated histone H2AXSer139 (g-H2AX). Anti-GAPDH (at1:5,000; Cell Signaling Technology) was used for loadingcontrols. Each experiment was carried out in duplicate, andthe assays were repeated 3 times.

Cell proliferation assaysProstate cancer cells’ growth was examined in the pres-

ence of GRM1 antagonists, riluzole, and BAY36-7620. Rilu-zole functions in part by blocking GRM1 signaling andpreventing intracellular glutamate release. BAY 36-7620 is aspecific noncompetitive antagonist of GRM1 that binds thetransmembrane domain of the receptor and stabilizes itsinactive transformation (24). Cells were seeded at 500 (PC-3, DU145, and LNCaP) per 200 mL/well and in 24 replicatesin 96-well plates in complete medium. After 3 days, culturemedium was refreshed and supplemented with DMSO,riluzole, or BAY36-7620 at 0, 1, 10, 25, or 50 mmol/L.Culture plates were incubated at 37�C for 2, 4, or 6 daysand the media were refreshed every 48 hours. Cell prolif-eration was measured using The CellTiter 96 AQueous OneSolution Cell Proliferation Assay (Promega).Prostate cancer cell growth under glutamine- and gluta-

mate-free condition was examined in an independent set of96-well plates and at cell densities described earlier. Cellswere incubated in customized glutamine- and glutamate-free culture media supplemented with 10% dialyzed FBSand 2 mmol/L GlutaMAX for 3, 5, or 7 days, and the mediawere refreshed every 48 hours. After the indicated incuba-tion times, cell proliferation was measured as describedearlier and compared with their parallel isogenic cell linescultured simultaneously but in their complete growthmedium.

Cell migration and invasion assaysCell migration and invasion assays were conducted using

8-mm transwell filters (Costar) with the modificationsdescribed previously (25). For the invasion assay, the uppercompartment was coated with 50 mg Matrigel (BD Bios-ciences) to form amatrix barrier. A suspension of cells (5�104 for PC-3 or 2 � 104 for DU145) in basal mediumcontaining 0.1%bovine serumalbumin (BSA)was added tothe upper compartment. The lower compartment was filled

with 400 mL basal medium containing 10% FBS as che-moattractant. Riluzole or BAY36-7620 was included atequimolar concentrations (0–50 mmol/L) to upper andlower compartments. After 48 hours for PC-3 or 24 hoursfor DU145, the nonmigratory cells on the upper surfacewere removed using a cotton swab, and the cells on thelower surface were fixed and stained with Diff-Quick solu-tion (Dade Behring). Migrated or invaded cells in transwellfilters were counted from 10 randomly selected fields in atleast 4 independent wells. The experiment was repeated 3times independently.

Statistical design and data analysisAll analyses were conducted using SAS 9.3 (SAS Institute)

with a nominal significance level of 0.05. The serum sam-ples were stratified by clinical status of the research subjects(normal, primary prostate cancer, ormCRPCa)with patientdemographics (e.g., age, race, etc.) reported asmean and SDfor continuous variables, and as frequencies and relativefrequencies for categorical variable (Supplementary TablesS1 and S2). Patients with primary prostate cancer weredivided into 3 subgroups of aggressiveness (low aggres-siveness, intermediate aggressiveness, and high aggres-siveness) using clinical criteria as described in "Materialsand Methods." The demographic variables were comparedbetween normal research subjects and men with primaryprostate cancer or mCRPCa using one-way ANOVA and anextension of Fisher exact test for continuous and categoricalvariables, respectively. Age, PSA, and Gleason scores werecategorized as: 40–49, 50–59, 60–69, 70–79; �4, 4.1–10,�10.1; and �6, 7, �8, respectively.

Serum glutamate levels were compared for variables ofinterests and among the research subjects using one-wayANOVA,withTukey-adjusted post hocpairwise comparisons.All serum samples measurements were log-transformedbefore statistical analysis. Simple linear regression modelswere used to assess the association between research sub-ject’s serum glutamate and clinical status (normal, primaryprostate cancer with low-, intermediate-, or high-aggres-siveness score, mCRPCa) while adjusting for demographiccovariates. Tukey-adjustedpairwise comparisonsweremadebetween clinical status of research subjects and their demo-graphic strata when appropriate. The covariate-adjustedmodels considered a single demographic covariate (withand without interactions) as well as multiple demographiccovariates, as this information is not applicable or availablefor normal research subjects and patients with mCRPCa.Only research subjects with primary prostate cancer wereconsidered when Gleason score and stage were included ascovariates. All model assumptions were verified graphicallyusing quantile–quantile (QQ) and residual plots.

ResultsSerum glutamate levels in normal and prostate cancerresearch subjects

The research subjects’ race, Gleason scores, PSA values,age, and clinical aggressiveness were described in Supple-mentary Table S1. Serum creatinine values were within the






expected range and did not differ significantly amongresearch subject subgroups. Serum PSA levels were foundto be significantly (P < 0.001) different among normalindividuals (0.556 � 0.03), patients with primary prostatecancer (10.38� 2.19), and patients with mCRPCa (323.74� 51.43). The average age was found to be significantlyhigher (P < 0.001) in research subjects with mCRPCa (70.2� 1.0) than in those with primary prostate cancer (62.0 �0.6) or normal research subjects (61.4 � 1.2). Serumglutamate levels were not significantly different amongdifferent age groups of normal research subjects (Supple-mentary Table S2, P ¼ 0.1735).

The entire study cohort was analyzed regardless of ageand race. When the normal research subjects (n¼ 60) wereused for comparison, serum glutamate levels significantlyincreased inmenwith primary prostate cancer (n¼197;P¼0.003) but returned to normal levels in men with mCRPCa(n ¼ 109; P ¼ 0.002; Fig. 1A, left). When the patients withprimary prostate cancer were divided into low, intermedi-ate, and high aggressiveness, glutamate levels were signif-icantly different as: (i) between normal research subjectsand intermediate (P < 0.001) or high aggressiveness (P <0.001); (ii) between low aggressiveness group and inter-mediate (P < 0.002) or high aggressiveness (P < 0.001)groups; (iii) between intermediate aggressiveness group andmCRPCa (P < 0.001), and (iv) between high aggressivenessand mCRPCa (P < 0.001; Fig. 1A, middle). Pairwise com-parisons indicated a significant association between gluta-mate levels andGleason score (P < 0.001). Glutamate levelsdiffered significantly between Gleason � 6 (n ¼ 81) andGleason 7 (n ¼ 72; P < 0.001) or Gleason � 8 (n ¼ 30; P <0.001). Glutamate levels were higher in Gleason � 8 thanthe Gleason 7 groups, but the difference did not reachstatistical significance (P < 0.064; Fig. 1A, right). Univariateanalyses, after adjusting for PSA or stage, revealed nosignificant association between serum glutamate levels andany of the 3 aggressiveness categories in the pool of researchsubjects with primary prostate cancer (Table 1).

The association between prostate cancer aggressivenessand serum glutamate were explored among primary pros-tate cancer research subjects while adjusting for potentialconfounders. Age, PSA, and stage did not seem to be con-founders in the single and2 covariate (confounder)models;they were nonsignificant and did not change the directionor significance of the relationship between serum glutamatelevels and prostate cancer aggressiveness. However, theinclusion of Gleason score seemed to overpower the rela-tionship between serum glutamate levels and prostate can-cer aggressiveness. The Gleason score seemed to drive thedifference between serum levels within the primary prostatecancer research subjects (Tables 1 and 2).

Serum glutamate levels in Caucasian Americans versusAfrican Americans

The adjusted analysis involving race indicated thatpatient clinical status (primary or mCRPCa; P < 0.001),race (P ¼ 0.029), and interaction (P ¼ 0.029) effects weresignificant (Table 1). The pairwise comparisons of race

within each clinical subgroup of research subjects indicatedthat a significant difference exists in patients with mCRPCa(P < 0.037). Supplementary Fig. S1 shows that this differ-ence is the result of the significant interaction term, whereserum glutamate decreases for Caucasians from the highaggressiveness to mCRPCa but increase for AfricanAmericans.

Among Caucasian Americans, pairwise comparisonsindicated that significant differences in serum glutamateexisted between research subjects with primary prostatecancer and mCRPCa (P < 0.030; Fig. 1B, left), betweennormal research subjects and those with prostate cancer ofhigh aggressiveness (P < 0.004), between low and highaggressiveness subgroups (P ¼ 0.001), and betweenmCRPCa andhigh aggressiveness subgroups (P<0.001; Fig.1B, middle). Among Caucasian American research subjectswith primary prostate cancer, a significant difference wasfoundbetween research subjectswithGleason score�6 andGleason score 7 (P < 0.002) and between Gleason score� 6and Gleason score � 8 (P < 0.001; Fig. 1B, right).

Among African Americans, pairwise comparisons indi-cated that serum glutamate levels differed between normalresearch subjects and primary prostate cancer (P < 0.026)and mCRPCa (P < 0.016; Fig. 1C, left). A significant differ-ence in serum glutamate was found in African Americansbetween the normal research subjects and intermediateaggressiveness subgroup (P < 0.014) and between normaland both high aggressiveness subgroup (P < 0.027) andpatients withmCRPCa (P < 0.038; Fig. 1C,middle). AmongAfrican-American research subjects with primary prostatecancer, significant differences in serum glutamate levelswere found between research subjects with Gleason score�6orGleason score�8 (P<0.02; Fig. 1C, right). The serumglutamate levels differed significantly between the racesonly in the research subjects with mCRPCa (P ¼ 0.037;Supplementary Fig. S1).

Immunohistochemical staining of metabotropicglutamate receptor 1 in benign andmalignant prostatetissues

Specificity of anti-GRM1 immunoreactivitywas shownbydeleting the primary antibody or replacing it with a non-immune serum (data not shown). Basal cells exhibitednuclear staining, whereas no staining was found in luminalacinar epithelial cells ofBPHornormal prostate (n¼10; Fig.2A). Endothelial cells, infiltrating macrophages, and stro-mal cells exhibited intense nuclear staining (Fig. 2A, mes-enchymal cells). Intense cytoplasmic and nuclear stainingwas observed in proliferating epithelial cells of glands withbasal cell hyperplasia (Supplementary Fig. S2).Moderate tointense cytoplasmic staining was noted in high-grade pros-tate intraepithelial neoplasia (HGPIN; Fig. 2A). To furtherdetermine the GRM1 expression during prostate cancerprogression and metastasis, we carried out additionalimmunohistochemical staining in primary (n ¼ 15) andmetastatic prostate cancer tissues (n ¼ 6). Moderate tointense cytoplasmic stainingwith perinuclear enhancementwas noted in prostate cancer with Gleason scores 6 and 7

Koochekpour et al.





0

50

100

150P = 0.003

P = 0.002

A

≤ 6 7 ≥ 80

50

100

150 P < 0.001

P < 0.001P < 0.064

Gleason score

Nor

mal

Low-A

ggr

Inte

rmed

iate

-

Aggr

Hig

h-Aggr

mCRPCa

0

50

100

150

P < 0.001

P < 0.001

P < 0.001

P < 0.001P < 0.001

p < 0.002

Primary Pca

Norm

al

Prim

ary

PCa

mCRPCa

0

50

100

150 P < 0.030

Seru

m g

luta

mate

(μm

ol/L)

Norm

al

Low-A

ggr

Hig

h-Aggr

mCRPC

a

0

50

100

150P < 0.004

P < 0.001P < 0.001

Primary PCa

≤ 6 7 ≥ 8

0

50

100

150

P < 0.002

P < 0.111P < 0.001

0

50

100

150

P < 0.026

P < 0.016

C

0

50

100

150

P < 0.014

P < 0.038

P < 0.027

≤ 6 7 ≥ 80

50

100

150P < 0.020

Seru

m g

luta

mate

(μm

ol/L)

Seru

m g

luta

mate

(μm

ol/L)

Seru

m g

luta

mate

(μm

ol/L)

Seru

m g

luta

mate

(μm

ol/L)

Seru

m g

luta

mate

(μm

ol/L)

Seru

m g

luta

mate

(μm

ol/L)

Seru

m g

luta

mate

(μm

ol/L)

Seru

m g

luta

mate

(μm

ol/L)

Norm

al

Low-A

ggr

Hig

h-Aggr

mCRPC

a

Primary PCa

Norm

al

Prim

ary

PCa

mCRPCa

Norm

al

Prim

ary

PCa

mCRPCa

Gleason score

Gleason score

Inte

rmed

iate

-

Aggr

Inte

rmed

iate

-

Aggr

B

Figure 1. Box and whiskers plot of serum glutamate levels in normal individuals and patients with primary or metastatic castrate-resistant prostatecancer (PCa) of Caucasian- or African-American origin. Serum glutamate levels were measured as described in "Materials and Methods." Thestudy cohorts were analyzed according to their clinical status as normal research subjects, primary PCa, or metastatic castrate-resistant PCa (left). Allpatients with primary PCa were divided according to their aggressiveness score based on Gleason and PSA (middle) or according to Gleasonscore (right). A, the entire study cohorts. B, the entire study cohorts of Caucasian Americans. C, the entire study cohorts of African Americans.Actual P values are indicated on individual graphs. The distribution of log-transformed serum measurements was compared among groups usingone-way ANOVA with Tukey pairwise comparisons. The box frame defines the 25 to 75 percentiles, the ends of error-bar, such as whiskers, depict theminimum and maximum values, and the line within the box marks the median value.






(Fig. 2A). Intense cytoplasmic staining with perinuclearenhancement was also observed in malignant acinar cellsof Gleason score 8 (4 þ 4), 9, and 10 (Fig. 2A) tumors.Prostate cancer cells invading into extraprostatic adiposeandmuscular tissues exhibited intense cytoplasmic staining(Fig 2A). Intense nuclear staining was observed at tumor–inflammatory interfaces in inflammatory cells, endothelialcells, and isolated macrophages (Fig 2A). Intense cyto-

plasmic staining was found in multilayer prostate cancercells infiltrating around nerves (Fig. 2A). Intense cyto-plasmic staining with perinuclear enhancement wasdetected in prostate cancer cells metastatic to bone and theabdominalwall (Fig. 2A andSupplementary Fig. S3).GRM1staining seemed to be distributed in both nuclear andcytoplasmic compartments. The nuclear staining of GRM1was heterogeneous.

Table 1. Single covariate adjusted analysis of serum glutamate levels in normal individuals, patients withprimary or metastatic castrate-resistant prostate cancer

Normal Glu

Primary prostate cancer (Aggressivenesssubgroup) Glu (mmol/L � SE)

MCRPCa Glu

Variable(mmol/L � SE)(n ¼ 60) Low (n ¼ 80)

Intermediate(n ¼ 68) High (n ¼ 49)

(mmol/L � SE)(n ¼ 109) P valuea

Gleason scoreb,c

�6 (n ¼ 94) N/Ae 51.7 � 1.8(n ¼ 80)

53.8 � 3.5(n ¼ 13)

86.0 � N/Ad

(n ¼ 1)N/Ae Aggressiveness

subgroupg: 0.2227 (n ¼ 73) N/Af (n ¼ 0) 65.8 � 2.8

(n ¼ 55)55.9 � 4.8(n ¼ 18)

Gleason: 0.007

�8 (n ¼ 30) N/Af (n ¼ 0) N/Af (n ¼ 0) 75.7 � 5.0(n ¼ 30)

PSAb (ng/mL)�4 (n ¼ 119) 49.3 � 1.4

(n ¼ 60)51.3 � 2.4(n ¼ 39)

68.0 � 6.4(n ¼ 11)

86.1 � 12.0(n ¼ 6)

60.9 � 11.5(n ¼ 3)

Clinical statush:<0.001

4.1–10 (n ¼ 100) N/Ae 52.6 � 2.5(n ¼ 41)

63.6 � 3.7(n ¼ 37)

57.5 � 5.6(n ¼ 16)

69.9 � 14.5(n ¼ 6)

PSA: 0.252

�10.1 (n ¼ 139) N/Ae N/Af (n ¼ 0) 61.8 � 4.0(n ¼ 20)

68.0 � 4.5(n ¼ 27)

50.9 � 1.8(n ¼ 92)

Int: 0.178

Stageb

T1 (n ¼ 47) N/Ae 57.2 � 3.7(n ¼ 26)

69.4 � 5.9(n ¼ 14)

73.9 � 10.5(n ¼ 7)

N/Ae Aggressivenesssubgroupg: <0.001

T2 (n ¼ 113) 49.5 � 1.9(n ¼ 54)

63.3 � 2.9(n ¼ 44)

64.9 � 7.1(n ¼ 15)

Stage: 0.179

T3 (n ¼ 37) N/Af (n ¼ 0) 56.1 � 6.2(n ¼ 10)

69.4 � 5.1(n ¼ 27)

Int: 0.591

RaceWhites (n ¼ 222) 50.3 � 1.7

(n ¼ 44)48.8 � 2.4(n ¼ 35)

58.5 � 3.0(n ¼ 25)

70.7 � 6.5(n ¼ 19)

50.6 � 1.7(n ¼ 99)

Clinical statush:<0.001 Race: 0.029

Blacks (n ¼ 144) 46.8 � 2.3(n ¼ 16)

54.4 � 2.5(n ¼ 45)

66.4 � 3.3(n ¼ 43)

67.4 � 4.7(n ¼ 30)

72.6 � 9.4(n ¼ 10)

Int: 0.029

Abbreviations: Glu, glutamate; Int, interaction term; Med, medium; N/A, not available or not applicable.aIn each scenario a linear regression model was used to evaluate the association between patient clinical status and log-transformedserum glutamate while adjusting for different patient characteristics and their interaction. All model assumptions were checkedgraphically using quartile–quartile and residual-plots. Data are presented as mean � SE.bOnly patients with primary prostate cancer were included in the analysis. PSA data was not available for all research subjects in themCRPCa subgroup.cInteraction term not considered given the frequency of sparse cell counts.dCannot compute the SE based on a single observation. PSA value for this research subject was 28.0 ng/mL.eThe variable of interest was not applicable or analyzed for the specified research subjects.fNo research subject was identified in the low-aggressiveness subgroup for the variable of interest.gComparison made among aggressiveness subgroups of patients with primary prostate cancer.hClinical status refers to comparison made among all normal research subjects, aggressiveness subgroups of patients with primaryprostate cancer, and patients with mCRPCa.

Koochekpour et al.





GRM1 expression in prostate cancer cell linesAn affinity purified anti-GRM1 antibody detected a

strong band at approximately 85 kDa (Fig. 2B) and 3 other

weakly expressing bands at approximately 40, 60, and 170kDa, which were close to the predicted bands based onGRM1 molecular weight described by the manufacturer.

Table 2. Multiple covariate adjusted analysis of serum glutamate levels in patients with primary andmetastatic castrate-resistant prostate cancer

Adjusted variable Variable of

Primary prostate cancer (AggressivenessSubtype) Glu (mmol/L � SE)

MCRPCa Glu

subtype interest Low Intermediate High (mmol/L � SE) P valuesa

Race Gleason scoreWhites �6 48.2 � 2.4 47.9 � 1.8 N/A N/A Aggressiveness

subgroupd: 0.216¼7 71.5 � N/Ab 63.4 � 3.7 56.2 � 5.2�7 N/A N/A 79.1 � 9.1 Race: 0.201

Blacks �6 54.4 � 2.5 63.3 � 7.2 86.0 � N/Ab Gleason: 0.008¼7 N/A 66.8 � 3.7 55.7 � 7.4�7 N/A N/A 73.4 � 5.9

PSA (ng/mL) Race�4.0 White 47.9 � 2.4 60.6 � 4.3 86.9 � 12.8 60.9 � 11.5 Clinical statusc: <0.001

Blacks 55.4 � 4.4 76.7 � 12.5 82.8 � 43.8 N/A PSA: 0.0944.1–10 White 49.7 � 4.2 62.3 � 6.2 55.3 � 6.4 60.9 � 14.0 Race: 0.034

Black 54.4 � 3.2 64.2 � 4.6 58.5 � 7.8 114.5 � N/Ab

�10.1 White N/A 54.3 � 3.6 64.8 � 4.1 49.1 � 1.7Black 42.2 � N/Ab 65.6 � 5.5 69.1 � 5.9 68.0 � 9.2

Age (y) Gleason score40–49 �6 58.3 � 5.6 59.9 � 0.8 N/A N/A Aggressiveness

subgroupd: 0.175¼7 N/A 71.9 � 15.4 48.1 � 0.1�7 N/A N/A 59.3 � N/Ab Age: 0.419

50–59 �6 51.0 � 3.0 46.7 � 7.5 86.0 � N/Ab Gleason: 0.005¼7 N/A 71.2 � 5.4 70.1 � 14.8�7 N/A N/A 79.5 � 10.5

60–69 �6 50.9 � 2.8 49.9 � 1.0 N/A¼7 N/A 64.1 � 3.7 52.6 � 4.1�7 N/A N/A 77.3 � 7.9

70þ �6 51.8 � 5.1 56.1 � 7.3 N/A¼7 71.5 � N/Ab 61.1 � 6.6 43.0 � 1.9�7 N/A N/A 71.9 � 9.7

PSA (ng/mL) Gleason score�4.0 �6 50.6 � 2.4 N/A N/A N/A Aggressiveness

subgroupd: 0.247¼7 71.5 � N/Ab 68.0 � 6.4 44.9 � N/Ab

�7 N/A N/A 90.6 � 12.4 PSA: 0.5024.1–10 �6 52.6 � 2.5 48.5 � 2.5 N/A Gleason: 0.020

¼7 N/A 66.3 � 4.1 57.2 � 4.4�7 N/A N/A 57.7 � 9.0

�10.1 �6 N/A 57.1 � 5.3 86.0 � N/Ab

¼7 42.2 � N/Ab 65.6 � 5.8 58.0 � 7.6�7 N/A N/A 76.3 � 3.8

Abbreviations: Glu, glutamate; Int, interaction term; Med, medium; N/A, not available or not applicable.aIn each scenario a linear regression model was used to evaluate the association between aggressiveness subgroups of researchsubjects with primary prostate cancer or clinical status of the research subject and log serum glutamate while adjusting for differentcombinations of patient characteristics. All model assumptions were checked graphically using quartile–quartile and residual-plots.bCannot compute the SE based on a single observation.cClinical status refers to comparison made among all normal research subjects, aggressiveness subgroups of patients with primaryprostate cancer, and patients with mCRPCa.dComparison made among aggressiveness subgroups of patients with primary prostate cancer.






Parallel immunoblotting with anti-GRM1 antibody preab-sorbed by its immunogenic peptide confirmed specificity(Fig. 2B).

GRM1 expression in androgen-sensitive LNCaP cells wasless than androgen-independent PC-3, DU145 and andro-gen-independent, and bone metastatic prostate cancer cell

lines (i.e., MDA-PCa 2b and VCaP). GRM1 expression inthe E006AA cell line obtained from an organ-confinedGleason 6 score prostate cancer from an African Americanwas less than bone-metastatic MDA-PCa2b cells (Fig. 2B).GRM1 expression in 22Rv1 cells was higher than LNCaPcells.

LNCaP

MDA-P

Ca2b

E006AA

DU-145

PC-322RV1

VCaP

GRM1

GAPDH

+ Blocking

peptide

B

BPH Mesenchymal cells Basal cell hyperplasia HGPIN

Gleason 6 Gleason 9Gleason 8Gleason 7

Gleason 10 Extraprostatic invasionTumor-inflammatory area Perineural invasion

Bone metastasis

A

v

mp

S

inf

T

Tm

T

F

n

Abdominal wall metastasis

Figure 2. Expression pattern of metabotropic GRM1 in normal and malignant prostate cancer tissues and cells. A, representative IHC results; magnification,�200. Benign prostate hyperplasia showing nuclear staining in basal cells and absence of staining in normal luminal acinar cells. Intense nuclear andcytoplasmic staining is noted in vascular endothelial cells (v), tissue-infiltrating macrophages (mp), and stromal cells (s). Intense cytoplasmic and nuclearstaining is seen in an area of basal cell hyperplasia. Moderate to intense cytoplasmic staining is noted in a representative image of HGPIN, withrelatively absence of staining inmost nuclei.Moderate to intense cytoplasmic stainingwith perinuclear enhancement is noted inGleason6 (3þ3) andGleason7 (3þ 4) tumor. Intense cytoplasmic stainingwith perinuclear enhancementwas observed inmalignant luminal acinar cells withGleason 8 (4þ 4), Gleason 9 (5þ 4), and a Gleason 10 tumor. Moderate cytoplasmic staining in a Gleason 3 þ 4 tumor with intense nuclear staining was observed in inflammatory(inf) and endothelial cells, and isolated macrophage. Intense cytoplasmic staining was observed in tumor cells invading into extraprostatic adipose (F) andmuscle (M) tissues. Perineural invasionwas noted by intense cytoplasmic staining ofmultilayer tumor cells circling around nerve cells (n). Neural structures arenegative for staining and the background stroma shows moderate focal nuclear staining. Intense cytoplasmic staining with perinuclear enhancementis noted in scattered bone-metastatic prostate cancer cells. Bone tissue appears as fragmented pieces. Intense cytoplasmic staining was observed inmalignant prostate cells metastasized into abdominal wall. B, GRM1 protein expression in prostate cancer cell lines.Western blot analysis was conducted onwhole-cell lysates by SDS-PAGE in the presence of reducing agent (top). The anti-GRM1 specificity was examined by probing a duplicate membranewith primary antibody preincubatedwith 5 mg/mL blocking peptide for 2 hours (middle). Glyceraldehyde-3 phosphate dehydrogenase (GAPDH) detectionwasused for loading control. LNCaP ¼ androgen-sensitive PCa cell line; PC-3, DU145, MDA-PCa2b, VCaP, and 22RV1 androgen-independent PCa cell lines;E006AA ¼ primary African American PCa cell line.

Koochekpour et al.





Prostate cancer cells growth in glutamate-free cultureconditionProliferation of PC-3 cells was decreased by 42%,

DU145 cells by 63%, and LNCaP cells by 68%, as com-pared with their isogenic control cell types after culturing 3,5, or 7 days in customized glutamine- and glutamate-freemedium supplemented with dialyzed FBS and GlutaMAX(Fig. 3A).

Inhibition of prostate cancer cell proliferation byGRM1 antagonistsTo evaluate the potential of intracellular glutamate levels

on prostate cancer cells growth, we used riluzole that func-

tions in part as the glutamate release inhibitor (14, 15).Inhibition of glutamate release in a cell type-, dose-, andtime-dependent manner decreased prostate cancer cellsgrowth (Fig. 3B). LNCaP cells were the most sensitiveamong the prostate cancer cell lines investigated. LNCaPcells growth was reduced by 54% (day 2) and by 85%(day 6) compared with their control cell type at the sameincubation period (Fig. 3B, right). Among all prostatecancer cells, DU145 cells were least sensitive to riluzole(Fig. 3B, middle). Proliferation of DU145 cells wasdecreased by 14% (day 2) and by 18% (day 6). However,PC-3 growth was decreased by 10% (day 2) and by 40% onday 6 (Fig. 3B, left).

3 5 70.0

0.5

1.0

1.5

2.0Glut (+)Glut (-)

A PC-3

Days

OD

(490/6

30 n

m)

3 5 70.0

0.5

1.0

1.5

2.0Glut (+)Glut (-)

DU145

Days

OD

(490/6

30 n

m)

3 5 70.0

0.1

0.2

0.3

0.4Glut (+)Glut (-)

LNCaP

Days

OD

(490/6

30 n

m)

0 10 25 500.0

0.4

0.8

1.22

4

B

6

Riluzole (μmol/L) Riluzole (μmol/L) Riluzole (μmol/L)

OD

(490/6

30 n

m)

0 10 25 500.6

0.8

1.0

1.2 2

4

6

OD

(490/6

30 n

m)

0 10 500.0

0.2

0.4

0.6

0.8 2

4

6

OD

(490/6

30 n

m)

25

phospho-H2AX

GAPDH

0 10 25 50Riluzole (μmol/L) :

LNCaP

0 10 25 50

PC-3 DU145

Cleaved

GAPDH

C 0 10 25 50

Caspase-9

Caspase-3

Caspase-7

Cleaved-PARP

GAPDH

Figure 3. Inhibition of prostate cancer cell growth by glutamate deprivation or GRM1-antagonist. A, effect of glutamate deprivation on prostate cancer (PCa)cells proliferation.Cellswere seededat 500 (PC-3,DU145, andLNCaP) per 200mL/well in 24 replicates in 96-well plates in their completemedium.After 3days,cells were incubated in their maintenance medium or customized glutamine- and glutamate-free culture media supplemented with 10% dialyzed FBSand2mmol/LGlutaMAX for 3, 5, or 7days. Themediawere refreshedevery 48hours.Cell proliferationwasmeasuredbyadding 20mLMTSsolution perwell for1 hour andmeasuring the absorbance at 490/630nm.Bars,mean�SEM,P<0.001, Student t testswere used to compare cell growthbetween glutamate-freeand complete medium at each day. One-way ANOVA test with Bonferroni corrections was used to compare values among different days for each cell line. B,effect of GRM1-antagonist on PCa cells growth. Cells were seeded as described earlier and were incubated in their complete medium in the presence orabsence of riluzole at 10, 25, or 50 mmol/L for 2, 4, or 6 days with media refreshment every 48 hours, followed by cell proliferation assay as describedearlier. Data represented the averageof 3 independent experiments�SEM.Statistical significance (P<0.0001) between the control and treatment groupswasevaluated by one-way ANOVA test with Bonferroni adjustment. C, induction of apoptosis by riluzole. Cells were cultured in their complete medium in thepresence or absence of riluzole as described earlier. Whole-cell lysate were extracted and immunoblotting was conducted on 30 mg protein usingantibodies against cleaved caspases-9, -7, and -3, PARP, and phosphoserine-139 H2AX. Anti-GAPDH antibody was used for control loading. Experimentwas repeated twice independently.






To show GRM1 responsiveness as a functional glutamatereceptor, independent studies were conducted usingBAY36-7620 as a noncompetitive GRM1 antagonist at dif-ferent concentrations and for a period of 2 to 6 days. As inthe riluzole study, cells were maintained in their completemedium containing nondialyzed FBS and L-glutamine (2mmol/L). BAY36-7620 showed similar growth inhibitoryeffect on the 3 prostate cancer cell lines and LNCaP cellswere found tobe themost sensitive cell line (SupplementaryFig. S4).

Glutamate blockade induces apoptotic cell deathThe expression of cleaved (active) forms of the initiator

caspase-9 and its active downstream effectors (caspases-7and -3) were determined in the cells grown in the presenceof riluzole for 4 or 6 days. The expression of cleaved caspase-9, -7, and -3 increased in a cell type–specific and dose-dependent manner in all cells investigated (Fig. 3C). Ingeneral, riluzole induced a cell type–specific increase in theexpression level of initiator and effector caspases.

The intensity of cleaved PARP expression increased in adose-dependent manner in all prostate cancer cell lines.H2AX, a member of H2A histone family is phosphorylatedon serine residue 139 (known as g-H2AX) after the activa-tion of the caspase cascade and as a result of apoptotic DNAfragmentation (26). Cell lysates showed a dose-dependentincrease in phospho-H2AXSer139 levels in LNCaP cells (Fig.3C). An increase in g-H2AX levels was also detected at 50mmol/L riluzole in PC-3 and DU145 cells. Accumulation ofintracellular glutamate levels secondary to riluzole treat-ment seemed to be apoptogenic.

Glutamate blockade decreases prostate cancer cellsmigration and invasion

Addition of riluzole or BAY36-7620 to both upper andlower compartments of transwell filters decreased cellmotility and invasion in a dose-dependent manner inPC-3 and DU145 cells (Fig. 4A and B). Riluzole reducedmigration of PC-3 cells by 53% and DU145 cells by 76%and invasion of PC-3 by 65% and DU145 by 82% (Fig. 4Aand B). BAY36-7620 decreased migration of PC-3 cells by67%,DU145 cells by 88%and invasionof PC-3by 83%andDU145 by 96% (Fig. 4A). Similar data was obtained using awound assay (Supplementary Fig. S5A–S5C).

DiscussionSerum glutamate levels positively correlated with Glea-

son score and prostate cancer aggressiveness. Immunohis-tochemical staining showed weak or no expression ofGRM1 in luminal acinar cells of normal or benign glandsbut high expression levels in primary and metastatic pros-tate cancer tissues. Glutamate deprivation or blockade withGRM1 antagonists decreased prostate cancer cells growth,migration, and invasion and led to apoptotic cell death.

Glycolysis and glutaminolysis are major contributingmetabolic pathways of tumorigenesis. Glutamine is themost abundant serum amino acid and a necessary meta-bolic precursor for other amino acids and nucleotides.

Glutamine deprivation decreased cell proliferation, DNAsynthesis, and protein synthesis rates in a cell type–depen-dentmanner in hepatocellular carcinoma and several breastand colon adenocarcinoma cells (27). Similarly, the impor-tance of glutamate signaling in cancer has been revealed byantiproliferative activity of glutamate antagonists in differ-ent types of human tumor cells, such as melanoma, colonadenocarcinoma, breast carcinoma, astrocytoma, and lungcarcinoma (17).

Humoral response profiling of serum samples by Taylorand colleagues revealed elevated levels of nitrogen metab-olism during prostate cancer progression (10). Concomi-tant validation of increased levels of the constituents ofnitrogen metabolism pathway was determined further byusing high-throughput quantitative profiling of relativelevels of metabolites in BPH (n ¼ 16) and organ-confinedprostate cancer (n ¼ 12) tissues. This study revealed sig-nificantly increased levels of glutamate in the prostatecancer tissues (P ¼ 0.0003) compared with BPH. Furtherexamination of metabolomic changes during prostate can-cer progression by the same group verified elevated levels ofglutamate in a majority of prostate cancer tissues as com-pared with benign glands (11).

DMSO

DU145

PC-3

Riluzole (25 µmol/L) Bay 36-7620 (25 µmol/L)

0

120

240

360

DM

SO 0

10

25

50 0

10

25

50

DM

SO 0

10

25

50 0

10

25

50

Riluzole(µmol/L)

BAY 36-7620

(µmol/L)

Riluzole(µmol/L)

BAY 36-7620

(µmol/L)

Ce

ll n

o./

fie

ld

Migration Invasion

PC-3 DU145

A

B

Figure 4. Effect of glutamate blockade on prostate cancer (PCa) cellsmigration and invasion. Cell migration and invasionwas conducted using8-mm transwell filters. For the invasion assay, the upper compartmentwas coatedwith 50mgMatrigel per filter. A suspension of cells (5�104 forPC-3 or 2 � 104 for DU145) in basal medium containing 0.1% BSA wasadded to the upper compartment. The lower compartment was filled with400 mL basal medium containing 10% FBS as chemoattractant. Riluzoleor BAY36-7620 was included at equimolar concentrations (0–50 mmol/L)to upper and lower compartments. After 48 hours for PC-3 or 24 hours forDU145,migrated or invaded cells in transwell filterswere counted from10randomly selected fields in at least 4 independentwells. Data representedthe average of 3 independent experiments � SEM. Statisticalsignificance (P < 0.05) between the control and treatment groups wasevaluated by one-way ANOVA test with Bonferroni adjustment.

Koochekpour et al.





The excess amount of glutamate in prostate cancer tissuesand patients’ serum may be explained by: (i) increased rateof glutaminolysis or glutamine addiction in proliferatingprostate cancer cells and (ii) carboxypeptidase function ofprostate-specific membrane antigen (PSMA) overexpressedby prostate cancer cells. While "glutamine addiction" servesas a nonspecific metabolic hallmark of malignant andproliferating cells, it is likely that the known enzymaticactivity of PSMA may have a key contributing role inglutamate production by prostate cancer cells. PSMA is atype II transmembrane glutamate carboxypeptidase, andglutamate is the major constant product of its enzymaticactivity in prostate cancer cells (28, 29). PSMA is over-expressed in prostate cancer and high PSMA expression isassociated with advanced tumor stage, poor prognosis, andhigh risk of biochemical recurrence (28, 30). PSMA-gener-ated glutamate signaling may serve as a nonandrogenicregulator of prostate cancer growth (31). Therefore, PSMAmay have critical functions in metabolomics profiling dur-ing prostate carcinogenesis and progression.Analysis of the glutamate receptor, GRM1, in prostate

cancer cell lines showed a cell type–dependent expression.GRM1 expression was higher in metastatic or androgen-independent prostate cancer cell lines (e.g., DU145, MDA-PCa2b, VCaP, and 22RV1) than in the primary (i.e.,E006AA) or androgen-sensitive LNCaP cell lines. GRM1overexpression might be an adoptive change hypersensitiz-ing prostate cancer cells to extracellularly available gluta-mate or could confer oncogenic activity due to its postre-ceptor signaling activation. Aberrant expression of GRM1was found sufficient for oncogenic transformation of mel-anocytes in transgenic mice (18, 19). Overexpression ofseveral members of mGluRs has been reported in othercancers (32–34). Activation of mGluR-triggered signalingpathways may occur in response to free circulating gluta-mate, glutamate released by tumor cells (autocrine loop), orstromal cells in the tumor microenvironment (paracrineloop). The data presented here also support an intracrineregulatory role for glutamate in prostate cancer cell growth,migration, and invasion.Intense nuclear staining was detected in basal cells of

normal glands and in endothelial cells, infiltrating macro-phages, and stromal cells at the tumor–inflammatory inter-face or tumor microenvironment. Moderate to intensecytoplasmic staining was detected in HGPIN, which isconsidered to be a precursor lesion to prostate cancer.Further examination of GRM1 expression in primary andmetastatic prostate cancer tissue sections showed moderateto intense cytoplasmic staining in malignant acinar cellswith perinuclear enhancement in Gleason score� 7 tumorsand intense but similar expression pattern in Gleason score� 8 tumors and in highly aggressive ductal adenocarcino-mas (Supplementary Fig. S3). In addition, locally advancedprostate cancer produced intense cytoplasmic staining inprostate cancer cells invading into extraprostatic adiposeand muscular tissues, and in multilayer tumor cells infil-trating aroundnerve cells. Intense cytoplasmic stainingwithperinuclear enhancement was detected in metastatic pros-

tate cancer cells to bone and the abdominal wall. Overall,the descriptive immunohistochemical analysis showsGRM1 overexpression in primary and metastatic prostatecancer tissues compared with noncancerous and benigntissues. Additional large-scale studies are required to assessthe predictive or prognostic significance of GRM1 expres-sion in prostate cancer.

Serum glutamate levels in mCRPCa were significantlyhigher in African-American patients than Caucasian-Amer-ican research subjects. Univariate analysis showed serumglutamate levels affected by race (Table 1). Two covariatemodels adjusting for 2 potential confounders revealed thatage, PSA, race, and stage are not significant confounders(Table 2). However, when Gleason score was added to themodel with race, age, and PSA, race had a dominant effectand relationship with serum glutamate levels in African-American men. Gleason score is the most reliable predictorof lethal prostate cancer. Both groups showed significantassociation between serum glutamate levels and Gleasonscore (�6 vs. �8). African Americans with mCRPCa werefound to have higher serum glutamate levels than in theprimary prostate cancer. These data might reflect the smallsample size formCRPCa (n¼10) inAfrican-American studycohort compared with the Caucasian Americans (n ¼ 99).

These data suggest that high serum glutamate levels mayfavor higher growth rates and aggressive behavior for pri-mary prostate cancer in both racial groups and for mCRPCain African Americans. While inter-racial differences formetabolic genes have been reported in several studies(35), large-scale studies are needed to determine the exactcontribution of glutamate either as a metabolic byproductor a contributing factor in prostate cancer–racial disparity.

First evidence for a link between glutamate receptormGluR1 (GRM1) and tumorigenesis was provided by anaccidental discovery in which spontaneous melanoma in atransgenic mouse line (TG-3) was discovered to be due tomultiple tandem insertions of the transgene into intron 3 ofGRM1 leading to aberrant expression of GRM1 in melano-cyte and the development of melanoma at a high rate andyoung age (18, 19). This was verified in a subsequenttransgenic mouse line, TG(GRM1)Epv (E), in which GRM1expression was regulated by a melanocyte-specific promot-er, dopachrome tautomerase (36).Different pharmacologicagents have been used as GRM1 (i.e., mGluR1) agonist orantagonists. Riluzole is an U.S. Food and Drug Adminis-tration (FDA)–approved GRM1-blocking agent and well-tolerated oral medicine for the treatment of amyotrophiclateral sclerosis that has been used to prevent glutamaterelease (37). LY367385 or BAY36-7620 is known as com-petitive or noncompetitive GRM1 antagonist, respectively(24, 38). Riluzole, BAY36-7620, or LY367385 prevent thegrowth promoting effect and activation of signaling path-ways by glutamate in a variety of cancer cell types includingmelanoma and breast carcinoma (38, 39).

Prostate cancer cell proliferation was significantlyreduced in a time-dependentmanner in glutamate-depletedmedium or in the presence of GRM1 antagonists (Fig. 3AandB and Supplementary Fig. S4). These data indicated that






prostate cancer cells’ growth depended onGRM1-signaling.Riluzole inhibition of glutamate release and GRM1-signal-ing in prostate cancer cells cultured in complete growthmedium not only decreased growth rate, but led to apo-ptotic cell death, shown by increased expression levels ofcleaved caspase-9, -7, and -3 and PARP. These results wereconfirmed by the demonstration that phospho-H2AXSer139

levels (known as g-H2AX) appeared early during apoptosis.Initiation of DNA double-strand breaks following caspaseand apoptotic endonuclease activation induces phosphor-ylation of H2AX histone at Ser139 (26).

The inhibitory effect of glutamate blockade on prostatecancer cell proliferation and migration may be at leastpartially mediated by the PI3K/Akt/mTOR pathway (6).The PI3K/Akt/mTOR function as the main regulator ofaerobic glycolysis in malignant cells. Activation of thePI3K/Akt/mTOR pathway in cancer cells triggers many ofthe metabolic activities that support cellular biosynthesis,such as: (i) increased surface expression of nutrient trans-porters leading to increased uptake of glucose and aminoacid, (ii) activated Akt increases glycolysis and lactate pro-duction and induces a "Warburg effect" in cancer or evenproliferating nontransformed cells (40), (iii) activation ofPI3K and Akt induce lipogenesis and stimulate lipogenicgenes expression in a variety of cell types (41), and (iv)mTOR serves as a major regulator of protein translation.

A cell type–specific response to glutamate blockade sug-gests differential expression levels for GRM1 in differentprostate cancer cells or their dependence on glutamatesignaling. Excess intracellular glutamate and/or inhibitionof GRM1 signaling by riluzole treatment may have delete-rious effects on prostate cancer cells growth, migration, andinvasion, which cannot be salvaged by an intact extracel-lularly activated GRM1 signaling. In addition, accessibilityto a certain amount of glutamate is essential for prostatecancer cells growth that may signify "glutamine addiction."However, more than this necessary amount may lead to"glutamate toxicity" and activation of various interactiveapoptotic signaling pathway leading to prostate cancer cell-killing. Intracellular glutamate as a metabolic intermediate,metabolic byproduct, or metabolic regulator of interactivemultiple signaling pathways may play a major role inmalignancy. Glutamate via a combination of intracrine,

autocrine, and paracrine pathways may contribute to pros-tate cancer growth and aggressiveness. Our data suggest thatglutamate together with other predictive or prognosticfactors may be a useful metabolicmarker for early detectionand clinical discrimination of aggressive tumors from non-aggressive tumors. Glutamate-initiated signaling pathwaysmay provide novel therapeutic opportunities for prostatecancer. Additional work is needed to ascertain the clinicaland therapeutic implications of glutamate dependence inprostate cancer.

Disclosure of Potential Conflict of InterestThe content is solely the responsibility of the authors and does not

necessarily represent the official views of the National Institutes of HealthorNationalCancer Institute.Nopotential conflicts of interest were disclosed.

Authors' ContributionsConception and design: S. Koochekpour, J.L. MohlerDevelopment of methodology: S. Koochekpour, H. ThompsonAcquisitionofdata (provided animals, acquired andmanagedpatients,provided facilities, etc.): S. Koochekpour, S. Majumdar, G. Azabdaftari, R.Scioneaux, D. Subramani, K. Rezaei, O. Sartor, R.L. VessellaAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): S. Koochekpour, G. Azabdaftari, K. Att-wood, O. Sartor, J.L. MohlerWriting, review, and/or revision of the manuscript: S. Koochekpour, G.Azabdaftari, O. Sartor, J.L. Mohler, R.L. VessellaAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): S. Koochekpour, S. Majumdar, G.Azabdaftari, R. Scioneaux, D. Subramani, C. Manhardt, S.S. Willard, M.Shourideh, K. RezaeiStudy supervision: S. KoochekpourPreliminary evaluation of histology slides: G.D. Lorusso

AcknowledgmentsThe authors thank the patients and their families and physicians for

contributing to the studies conducted in the authors’ laboratory. The authorsalso thank Ms. Rebecca Trautman for editorial assistance.

Grant SupportThis work was supported by R01MD005824 (to S. Koochekpour) and by

Roswell Park Cancer Institute and National Cancer Institute (NCI) grant#P30 CA016056. Additional support was provided by P01 CA85859 and theNorthwest Prostate Cancer SPORE (to R.L. Vessella) and P01 CA77739 (toJ.L. Mohler).

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received April 28, 2012; revised July 26, 2012; accepted August 24, 2012;published OnlineFirst October 16, 2012.

References1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer

J Clin 2012;62:10–29.2. Platz EA, Rimm EB, Willett WC, Kantoff PW, Giovannucci E. Racial

variation in prostate cancer incidence and in hormonal systemmarkersamongmale health professionals. JNatlCancer Inst 2000;92:2009–17.

3. Trock BJ. Application of metabolomics to prostate cancer. Urol Oncol2011;29:572–81.

4. Costello LC, Franklin RB.Concepts of citrate production and secretionby prostate. 1. Metabolic relationships. Prostate 1991;18:25–46.

5. Komuro H, Rakic P. Modulation of neuronal migration by NMDAreceptors. Science 1993;260:95–7.

6. DeBerardinis RJ, LumJJ, HatzivassiliouG, ThompsonCB. The biologyof cancer: metabolic reprogramming fuels cell growth and prolifera-tion. Cell Metab 2008;7:11–20.

7. Koppenal W, Bounds PL, Dang CV. Otto Warburg's contributions tocurrent concepts of cancer metabolism. Nat Rev 2011;11:325–37.

8. Newsholme EA, Crabtree B, Ardawi MS. Glutamine metabolism inlymphocytes: its biochemical, physiological and clinical importance. QJ Exp Physiol 1985;70:473–89.

9. Gao P, Tchernyshyov I, Chang TC, Lee YS, Kita K, Ochi T, et al. c-Mycsuppression of miR-23a/b enhances mitochondrial glutaminaseexpression and glutamine metabolism. Nature 2009;458:762–5.

10. Taylor BS, Pal M, Yu J, Laxman B, Kalyana-Sundaram S, Zhao R, et al.Humoral response profiling reveals pathways to prostate cancerprogression. Mol Cell Proteomics 2008;7:600–11.

11. Sreekumar A, Poisson LM, Rajendiran TM, Khan AP, CaoQ, Yu J, et al.Metabolomic profiles delineate potential role for sarcosine in prostatecancer progression. Nature 2009;457:910–4.

Koochekpour et al.





12. Young VR, Ajami AM. Glutamate: an amino acid of particular distinc-tion. J Nutr 2000;130(Suppl 4):892S–900S.

13. Monaghan DT, Bridges RJ, Cotman CW. The excitatory amino acidreceptors: their classes, pharmacology, and distinct properties in thefunction of the central nervous system. Annu Rev Pharmacol Toxicol1989;29:365–402.

14. ShinSS,NamkoongJ,Wall BA,GleasonR, LeeHJ,ChenS.Oncogenicactivities of metabotropic glutamate receptor 1 (Grm1) in melanocytetransformation. Pigment Cell melanoma Res 2008;21:368–78.

15. Shin SS, Martino JJ, Chen S. Metabotropic glutamate receptors(mGlus) and cellular transformation. Neuropharmacology 2008;55:396–402.

16. TakanoT, Lin JH, ArcuinoG,GaoQ, Yang J, NedergaardM.Glutamaterelease promotes growth of malignant gliomas. Nat Med 2001;7:1010–5.

17. Rzeski W, Turski L, Ikonomidou C. Glutamate antagonists limit tumorgrowth. Proc Natl Acad Sci U S A 2001;98:6372–7.

18. Chen S, Zhu H, Wetzel WJ, Philbert MA. Spontaneous melanocytosisin transgenic mice. J Invest Dermatol 1996;106:1145–51.

19. ZhuH, Reuhl K, Zhang X, Botha R, RyanK,Wei J, et al. Development ofheritable melanoma in transgenic mice. J Invest Dermatol 1998;110:247–52.

20. PissimissisN, PapageorgiouE, Lembessis P, ArmakolasA, KoutsilierisM. The glutamatergic system expression in human PC-3 and LNCaPprostate cancer cells. Anticancer Res 2009;29:371–7.

21. Schroeder JC, Bensen JT, Su LJ, Mishel M, Ivanova A, Smith GJ, et al.The North Carolina–Louisiana Prostate Cancer Project (PCaP): meth-ods anddesign of amultidisciplinary population-based cohort study ofracial differences in prostate cancer outcomes. Prostate 2006;66:1162–76.

22. Koochekpour S, Hu S, Vellasco-Gonzalez C, Bernardo R, AzabdaftariG, Zhu G, et al. Serum prosaposin levels are increased in patients withadvanced prostate cancer. Prostate 2012;72:253–69.

23. Koochekpour S, Zhuang YJ, Beroukhim R, Hsieh CL, Hofer MD, ZhauHE, et al. Amplification and overexpression of prosaposin in prostatecancer. Genes Chromosomes Cancer 2005;44:351–64.

24. Carroll FY, Stolle A, Beart PM, Voerste A, Brabet I, Mauler F, et al.BAY36-7620: a potent non-competitive mGlu1 receptor antagonistwith inverse agonist activity. Mol Pharmacol 2001;59:965–73.

25. Hu S, Delorme N, Liu Z, Liu T, Velasco-Gonzalez C, Garai J, et al.Prosaposin down-modulation decreases metastatic prostate cancercell adhesion, migration, and invasion. Mol Cancer 2010;9:30.

26. Rogakou EP, Nieves-Neira W, Boon C, Pommier Y, Bonner WM.Initiation of DNA fragmentation during apoptosis induces phosphor-ylation of H2AX histone at serine 139. J Biol Chem 2000;275:9390–5.

27. Wasa M, Bode BP, Abcouwer SF, Collins CL, Tanabe KK, Souba WW.Glutamine as a regulator of DNA and protein biosynthesis in humansolid tumor cell lines. Ann Surg 1996;224:189–97.

28. Pinto JT, Suffoletto BP, Berzin TM, Qiao CH, Lin S, Tong WP, et al.Prostate-specific membrane antigen: a novel folate hydrolase inhuman prostatic carcinoma cells. Clin Cancer Res 1996;2:1445–51.

29. Barinka C, Rinnov�a M, S�acha P, Rojas C, Majer P, Slusher BS, et al.Substrate specificity, inhibition and enzymological analysis of recom-binant human glutamate carboxypeptidase II. J Neurochem 2002;80:477–87.

30. Minner S,Wittmer C, GraefenM, SalomonG, Steuber T, Haese A, et al.High level PSMA expression is associatedwith early PSA recurrence insurgically treated prostate cancer. Prostate 2011;71:281–8.

31. Aleman M, Milbank A, Harsch KM, Detore N, Klein E, Heston W.Expression of metabotropic glutamate receptors and its potentialsignaling throughprostate specificmembrane antigen. ProcAmAssocCancer Res 2004;45:Abstract # 5188.

32. ChangHJ, YooBC, LimSB, JeongSY,KimWH,Park JG.Metabotropicglutamate receptor 4 expression in colorectal carcinoma and itsprognostic significance. Clin Cancer Res 2005;11:3288–95.

33. Kalariti N, Lembessis P, Papageorgiou E, Pissimissis N, Koutsilieris M.Regulation of the mGluR5, EAAT1 and GS expression by glucocorti-coids in MG-63 osteoblast-like osteosarcoma cells. J MusculoskeletNeuronal Interact 2007;7:113–8.

34. Joo EJ, Kang CI, Ha YE, Park SY, Kang SJ, Joung MK, et al. Clinicaloutcome of catheter salvage in neutropenic cancer patients withcatheter-related infection. Scand J Infect Dis 2011;43:258–63.

35. Ragin CC, Langevin S, Rubin S, Taioli E. Review of studies onmetabolic genes and cancer in populations of African descent. GenetMed 2010;12:12–8.

36. Pollock PM, Cohen-Solal K, Sood R, Namkoong J, Martino JJ,Koganti A, et al. Melanoma mouse model implicates metabotropicglutamate signaling in melanocytic neoplasia. Nat Genet 2003;34:108–12.

37. Namkoong J, Shin SS, Lee HJ, Marín YE, Wall BA, Goydos JS, et al.Metabotropic glutamate receptor 1 and glutamate signaling in humanmelanoma. Cancer Res 2007;67:2298–305.

38. Le MN, Chan JL, Rosenberg SA, Nabatian AS, Merrigan KT, Cohen-Solal KA, et al. The glutamate release inhibitor riluzole decreasesmigration, invasion, and proliferation of melanoma cells. J InvestDermatol 2010;130:2240–9.

39. Speyer CL, Smith JS, Banda M, Devries JA, Mekani T, Gorski DH.Metabotropic glutamate receptor-1: a potential therapeutic target forthe treatment of breast cancer. Breast Cancer Res Treat 2012;132:565–73.

40. Elstrom RL, Bauer DE, Buzzai M, Karnauskas R, Harris MH, Plas H,et al. Akt stimulates aerobic glycolysis in cancer cells. Cancer Res2004;643892–9.

41. Bauer DE, Hatzivassiliou G, Zhao F, Andreadis C, Thompson CB. ATPcitrate lyase is an important component of cell growth and transfor-mation. Oncogene 2005;24:6314–22.






2012;18:5888-5901. Published OnlineFirst October 16, 2012.Clin Cancer Res Shahriar Koochekpour, Sunipa Majumdar, Gissou Azabdaftari, et al. Invasion and Induces Apoptosis in Prostate Cancer CellsGlutamate Blockade Decreases Proliferation, Migration, and Serum Glutamate Levels Correlate with Gleason Score and

Updated version

10.1158/1078-0432.CCR-12-1308doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2012/10/18/1078-0432.CCR-12-1308.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/18/21/5888.full#ref-list-1

This article cites 39 articles, 9 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/18/21/5888.full#related-urls

This article has been cited by 7 HighWire-hosted articles. Access the articles at:

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

Subscriptions

Reprints and

[email protected]

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/18/21/5888To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-12-1308

http://clincancerres.aacrjournals.org/content/suppl/2012/10/18/1078-0432.CCR-12-1308.DC1

http://clincancerres.aacrjournals.org/content/18/21/5888.full#ref-list-1

http://clincancerres.aacrjournals.org/content/18/21/5888.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/18/21/5888


Serum Glutamate Levels Correlate with Gleason Score and ... › ... › 18 › 21 ›...

Documents

Transcript of Serum Glutamate Levels Correlate with Gleason Score and ... › ... › 18 › 21 ›...