The Prostate Single-Cell Analysis of Circulating Tumor Cells Identifies Cumulative Expression Patterns of EMT-Related Genes in Metastatic Prostate Cancer Chun-Liang Chen, 1 Devalingam Mahalingam, 2 Pawel Osmulski, 1 Rohit R. Jadhav, 1 Chiou-Miin Wang, 1 Robin J. Leach, 3,4 Tien-Cheng Chang, 5 Steven D. Weitman, 6,7 Addanki Pratap Kumar, 4 LuZhe Sun, 3 Maria E. Gaczynska, 1 Ian M. Thompson, 4 and Tim Hui-Ming Huang 1 * 1 Department of Molecular Medicine,CancerTherapy and Research Center, University of Texas Health Science Center, San Antonio,Texas 2 Department of Medicine,CancerTherapy and Research Center,University of Texas Health Science Center, San Antonio,Texas 3 Department of Cellular and Structural Biology,CancerTherapy and Research Center, University of Texas Health Science Center, San Antonio,Texas 4 Department of Urology,CancerTherapy and Research Center,University of Texas Health Science Center, San Antonio,Texas 5 Department of Obstetrics and Gynecology,CancerTherapy and Research Center, University of Texas Health Science Center, San Antonio,Texas 6 Department of Pediatrics,CancerTherapy and Research Center,University of Texas Health Science Center, San Antonio,Texas 7 Department of Institute for Drug Development,CancerTherapy and Research Center, University of Texas Health Science Center, San Antonio,Texas BACKGROUND. Prostate tumors shed circulating tumor cells (CTCs) into the blood stream. Increased evidence shows that CTCs are often present in metastatic prostate cancer and can be alternative sources for disease profiling and prognostication. Here we postulate that CTCs expressing genes related to epithelial–mesenchymal transition (EMT) are strong predictors of metastatic prostate cancer. METHODS. A microfiltration system was used to trap CTCs from peripheral blood based on size selection of large epithelial-like cells without CD45 leukocyte marker. These cells individually retrieved with a micromanipulator device were assessed for cell membrane physical properties using atomic force microscopy. Additionally, 38 CTCs from eight prostate cancer patients were used to determine expression profiles of 84 EMT-related and reference genes using a microfluidics-based PCR system. RESULTS. Increased cell elasticity and membrane smoothness were found in CTCs compared to noncancerous cells, highlighting their potential invasiveness and mobility in the Grant sponsor: Integrative Cancer Biology Program; Grant number: U54CA113001; Grant sponsor: Early Detection Research Network; Grant number: U01CA086402; Grant sponsor: Cancer Center Support Grant; Grant number: P30CA054174. *Correspondence to: Tim Hui-Ming Huang, Department of Molecular Medicine/Institute of Biotechnology, University of Texas Health Science Center, 7703 Floyd Curl Drive, Mail Code 8257, STRF, San Antonio 78229-3900, TX. E-mail: huangt3@uthscsa.edu Received 21 August 2012; Accepted 2 November 2012 DOI 10.1002/pros.22625 Published online in Wiley Online Library (wileyonlinelibrary.com). ß 2012 Wiley Periodicals, Inc.
Transcript
TheProstate
Single-Cell Analysis of CirculatingTumorCells IdentifiesCumulative Expression Patterns of EMT-RelatedGenes
inMetastatic ProstateCancer
Chun-Liang Chen,1 Devalingam Mahalingam,2 Pawel Osmulski,1 Rohit R. Jadhav,1
Chiou-Miin Wang,1 Robin J. Leach,3,4 Tien-Cheng Chang,5 Steven D. Weitman,6,7
Addanki Pratap Kumar,4 LuZhe Sun,3 Maria E. Gaczynska,1 Ian M. Thompson,4 andTim Hui-Ming Huang1*
1DepartmentofMolecularMedicine,CancerTherapyandResearchCenter,Universityof TexasHealth ScienceCenter, San Antonio,Texas
7Departmentof Institute forDrugDevelopment,CancerTherapyandResearchCenter,Universityof TexasHealth ScienceCenter, San Antonio,Texas
BACKGROUND. Prostate tumors shed circulating tumor cells (CTCs) into the blood stream.Increased evidence shows that CTCs are often present in metastatic prostate cancer and canbe alternative sources for disease profiling and prognostication. Here we postulate that CTCsexpressing genes related to epithelial–mesenchymal transition (EMT) are strong predictors ofmetastatic prostate cancer.METHODS. A microfiltration system was used to trap CTCs from peripheral blood basedon size selection of large epithelial-like cells without CD45 leukocyte marker. These cellsindividually retrieved with a micromanipulator device were assessed for cell membranephysical properties using atomic force microscopy. Additionally, 38 CTCs from eight prostatecancer patients were used to determine expression profiles of 84 EMT-related and referencegenes using a microfluidics-based PCR system.RESULTS. Increased cell elasticity and membrane smoothness were found in CTCscompared to noncancerous cells, highlighting their potential invasiveness and mobility in the
Grant sponsor: Integrative Cancer Biology Program; Grant number: U54CA113001; Grant sponsor: Early Detection Research Network; Grantnumber: U01CA086402; Grant sponsor: Cancer Center Support Grant; Grant number: P30CA054174.
*Correspondence to: Tim Hui-Ming Huang, Department of Molecular Medicine/Institute of Biotechnology, University of Texas Health ScienceCenter, 7703 Floyd Curl Drive, Mail Code 8257, STRF, San Antonio 78229-3900, TX. E-mail: [email protected] 21 August 2012; Accepted 2 November 2012DOI 10.1002/pros.22625Published online in Wiley Online Library(wileyonlinelibrary.com).
! 2012WileyPeriodicals,Inc.
peripheral circulation. Despite heterogeneous expression patterns of individual CTCs, genesthat promote mesenchymal transitioning into a more malignant state, including IGF1, IGF2,EGFR, FOXP3, and TGFB3, were commonly observed in these cells. An additional subset ofEMT-related genes (e.g., PTPRN2, ALDH1, ESR2, and WNT5A) were expressed in CTCs ofcastration-resistant cancer, but less frequently in castration-sensitive cancer.CONCLUSIONS. The study suggests that an incremental expression of EMT-related genesin CTCs is associated with metastatic castration-resistant cancer. Although CTCs represent agroup of highly heterogeneous cells, their unique EMT-related gene signatures provide anew opportunity for personalized treatments with targeted inhibitors in advanced prostatecancer patients. Prostate # 2012 Wiley Periodicals, Inc.
During the formation and growth of a prostate tu-mor, malignantly transformed cells can be shed fromthe primary site and circulate in the bloodstream.These circulating tumor cells (CTCs) are found atvery low levels, one in a billion blood cells, and mostdie in the circulation [1,2]. Nonetheless, a proportionof these rare cells survive and can be further preprog-ramed by integrins and chemokines, enabling their at-tachment at distant sites [3,4]. After seeding to ametastatic location, CTCs adapt to survive in inhospi-table conditions, for example, low blood oxygen per-fusion or low pH for extended periods [5]. As CTCscan be obtained through routine phlebotomy, there issignificant interest in their use as a measure of diseaseprognosis and treatment response as well as for thepotential of treatment selection.
Despite the promise of CTC characterization forclinical use, detecting this rare cell population is tech-nically challenging. The FDA-approved CellSearch1
system has to date been considered the gold standardfor CTC detection in the clinical setting [6,7]. This sys-tem uses antibodies against the epithelial cell adhe-sion molecule (EpCAM), which positively select CTCsin a magnetic field [8]. Immunocytological analysiscan then be used to confirm if these enriched cells ex-press cytokeratins or intermediate filaments of epithe-lial cells, but not the common leukocyte antigen CD45[9]. Using EpCAM-based or equivalent approaches,studies have shown that the presence of high CTCcounts (!5 cells/7.5 ml of blood) is associated withshorter progression-free survival and lower overallsurvival in prostate cancer patients [9–11]. Further-more, in patients with castration-resistant prostatecancer lower CTC counts detected post-treatmentscan be a stronger prognostic indicator for survival[6,12].
While this EpCAM-based detection technology isuseful for detecting advanced prostate cancer pro-gression, CTCs are heterogeneous and display stemcell-like properties [13]. Emerging evidence suggeststhat a subset of CTCs may lack EpCAM or cytokeratin
expression and instead exhibit a feature of epithelial–mesenchymal transition (EMT) [14]. EMT is a gradualprocess, and gene markers specific for mesenchymaland stem-like cells can be detected in CTCs [15,16].CTCs once reaching a particular site acquire an ‘‘or-gan-mimetic phenotype’’ and may lose prostate epi-thelial hallmarks [17,18].
In this study, we developed an approach to enrichand process CTCs based on their unique differencesin sizes and deformability that are distinct from bloodand non-invasive cells. Single CTCs individually re-trieved using a micromanipulator system were sub-ject to atomic force microscopy (AFM) as well asmicrofluidics-based PCR analyses. The results showthat CTCs isolated from advanced prostate cancerpatients frequently lose the typical features of epithe-lial prostate cancer cells. This shift was accompaniedby expressing highly diverse patterns of EMT-relatedgenes in CTCs. Furthermore, incremental increases inthe expression of these genes in these circulating cellsare associated with metastatic castration-resistantprostate cancer.
The University of Texas Health Science Center atSan Antonio’s Institutional Review Board approvedthe study and consent was obtained prior to samplecollection. Patient blood samples ("10 ml) were col-lected in K2-EDTA tubes, which were inverted fivetimes and kept at 48C or on ice. The patient blood wassubjected to single CTC isolation. CTCs were first iso-lated from blood cells using ScreenCell1 CC filtrationkit (cat #CC 3LC-ha, ScreenCell, Paris, France) accord-ing to manufacturer’s protocol with modifications[19]. After blood filtration, the circular-filter was re-leased onto an uncoated sterile petri dish with thecell-retained side up. From this point on, the rest ofisolation process was carried out under an invertedEvos fl digital fluorescence microscope (cat #1253460,
2 Chenet al.
The Prostate
AMG, Bothell, WA). The filter was washed 2–3 timeswith 50 ml PBS. During the washes, the residual bloodcells were further carried through the filter using gen-tle pipetting or dragging the filter against the bottomof petri dish using sterile forceps. If blood cell clump-ing occurred that could interfere with single CTC iso-lation, clumps were dissociated by incubation with50 ml TrypLE Express (cat #12604-013, Invitrogen,Carlsbad, CA) for 10 min in a petri dish before PBSwashes. CTCs and residual blood cells retained on thefilter were stained with anti-CD45 conjugated withphycoerythrin (PE; BD, Maryland) for 15 min andsubjected to three PBS washes as described above.CTCs on the filter were incubated with 25 ml TrypLEExpress for 10 min and removed and placed onto anew petri dish for CD45-negative selection and singleCTC isolation using a Narishige micromanipulatorand Ferty Syringe Plus Microinjector (cat #MN-153and INJ-FS-PLUS, Origio MidAtlantic Devices, Mt.,Laurel, NJ). Single CD45-negative CTCs were isolatedindividually, ejected in 4.5 ml PBS with 0.5 ml lysisbuffer (cat #55827, Invitrogen) in a 0.2 ml PCR tubesand frozen on dry ice immediately and stored at#208C until microfluidics-based PCR analysis. SomeCTCs were pooled together in RPMI medium supple-mented with 10% FBS and ampicillin/streptomycinfor atomic force microscope analysis.
ProstateCellCulture
Prostate cancer cell lines, LNCaP-AD (androgen-dependent), LNCaP-AI (androgen-independent) wereroutinely maintained in the laboratory. PC-3, andDU145 and the cell line were obtained from ATCC.The cells were cultured in RPMI medium with 10%FBS.
Individual CTCs and prostate cells suspended in"50 ml PBS were loaded on a poly-L-Lys (300 kDa;0.01% in PBS) coated glass disc glued to a steel disc.The discs were mounted in the MultiMode Nano-scope IIIa microscope (Bruker) equipped with the Jtype scanner and the glass chamber for in-liquidwork. The SQube probes with a colloidal gold spherewith a diameter of between 1.5 and 3 mm as a tip, andnominal spring constant of 0.08 N/m were appliedfor elasticity testing and topography imaging. Probeswith spherical tips were used as they produce lessharsh indentation than sharp tips and are less likelyto cause physical damage or trigger molecular re-sponse. The surface of the glass disc was surveyed forthe presence of cells under a video camera used forprobe position control, and the probe was directed
above the selected cell. Subsequently, the height im-age of the cell for roughness analysis was collected ina contact mode followed by a cell indentation for theelasticity testing. A standard plane fit was executedon the height mode images with the Nanoscopesoftware version 5.12. Roughness and force plotswere analyzed with the SPIP v.5.11 software (ImageMetrology, Denmark).
Cell elasticity. To determine the Young modulus,we performed cellular indentation mapping with theforce AFM. The central area on a cell surface wasprobed to obtain the most consistent elasticity data.We collected a 3 $ 3 array of force curves (total ninedata points) covering area of 4 mm2, with at least fiveindentations for each point. Indentation depth was re-stricted to 400 nm. A constant pulling rate was main-tained throughout all the experiments. The applieddesign allowed for data collection in less than 2 minper cell minimizing the cell stress response inducedby the prolong instrumentation of the cell surface. Foreach evaluated point, the force versus indentationcurve was constructed based on the force–load plots.We then applied the Hertz model to calculate theYoung’s modulus using the force-indentation curves.The model describes the physical relationship be-tween the applied force and the cantilever indenta-tion. It assumes spherical shape of the end of a tipplaced on a flat surface. The model is valid when thesphere radius is substantially larger than indenta-tions. The elasticity for each cell was averaged, andnominal elasticity was tested against cleaned glassdisks.
Cell roughness. To assess a level of morphologicalcomplexity of cell membranes, we determined theirsurface roughness. A contact mode image of each cellwas collected using a scan size from 5 mm $ 5 mm to30 mm $ 30 mm with a matrix of 512 $ 512 pixels perscan at 1 Hz scan rate. We analyzed roughness valueswithin two to four square areas of a cell surface cover-ing from 1 to 25 mm2. When analyzing multiplepatient-derived samples, we used the same sphericalprobe for force plots and image collecting. As a mea-sure of a cell membrane roughness, we employedRoot Mean Squared (RMS) of height calculated fromheights of all image pixels included in the area of in-terest. Images of a glass surface surrounding the cellswere used as a blank.
Single-CellMicrofluidics-BasedRT-PCRAnalysis
Single-cell microfluidics-based RT-PCR analysiswas carried out using CellsDirectTM one-step qRT-PCRkit (cat #11753–100, Invitrogen) and a microfluidics
device, BioMark HD MX/HX system (cat #BMKHD-PKG-MH, Fluidigm, Inc., South San Francisco, CA)[20]. Single CTCs in PBS/lysis buffer were thawed,mixed well, and spun down before lysed at 758C for10 min. To reduce contamination, genomic DNA wasdegraded in an 18 ml reaction volume using DNase I(five units) with 1$ DNase I buffer at RT for 5 min.PCR primers of selected genes for expression profilingwere selected from the PrimerBank database. Theseprimers were divided into two panels to fit BioMark48 $ 48 chips.
Reverse transcription (RT), preamplification, andPCR amplification were carried out according to theprotocol of single-cell gene expression (cat #BMK-M-48.48, Fluidigm). Target genes were amplified usingBioMark HB MX/HX system with 1$ SsoFast Eva-Green supermix with low ROX (cat #PN172-5211, Bio-Rad, Hercules, CA) and 1$ DNA binding dye sampleloading reagent (cat #PN 100-3738, Fluidigm). In eachchip assay, universal RNA (200 pg) from human nor-mal tissues (cat #4234565, BioChain, Newark, CA)and no template control (NTC) served as positive andnegative controls.
DataAnalysis
Expression data of genes of interest were displayedin cycles of threshold (Cts) after analysis usingReal-Time PCR analysis software (Fluidigm). Rela-tive expression values of the genes was obtainedusing 2#DDCt method in that each gene expression isnormalized to a reference gene and then normalizedto lowest expressed genes that have Ct 40 as describedpreviously [21]. Although three housekeeping genes(ACTB, GAPDH, and UBB) were initially included asreference genes, we found Ubiquitin B (UBB) to be ahighly stable gene for microfluidics-based PCR analy-sis, as its reliability has previously been validated in ameta-analysis of over 1,000 clinical samples [22].However, expression levels of ACTB and GAPDHwere less stable and weaker among different CTCs,consistent with a previous finding for single-cellCTC analysis [23]. Therefore, we only selected cellsthat expressed UBB at a threshold of Ct % 30 afterpreamplification, assuming that CTCs expressingrobust expression of UBB are less likely to contain de-graded RNA. Log2 values of gene expression in eachCTC were summed up as cumulative gene expressionaccording to the groups of frequently expressed EMT-related genes (detected in !44% CTCs) and lessfrequently expressed EMT-related genes (present in<44% CTCs) and different oncogenic signaling path-ways for comparisons. Cumulative gene expressionsof CTCs from prostate cancer patients were analyzedusing one-way ANOVA and unpaired Student’s t-test
using Prism 6 (GraphPad Software, La Jolla, CA).A P-value of <0.05 is considered as statisticallysignificant.
For in silico analysis of EMT-related gene expres-sion in clinical samples, raw probe cel intensity (&.cel)files were obtained from Gene Expression Omnibus(GEO) series GSE6919. Expression data for samplesrepresenting Normal Prostate Tissue free of any path-ological alteration (n ¼ 18), Normal Prostate TissueAdjacent to Tumor (n ¼ 63), Primary Prostate Tumor(n ¼ 65), and Metastatic Prostate Tumor (n ¼ 25),generated using Affymetrix Human Genome U95Version 2 Array were used for this study. RMA (Ro-bust Multichip Average) expression measures werecalculated for probes in all the samples by RMA nor-malization and background correction using Biocon-ductor Affy package in R [24]. The expression wasthen collapsed to gene level by averaging the meas-ures for the probes representing a gene. These expres-sion data were further used to compare the Metastaticsamples with Normal samples by calculating the sig-nificance (using Student’s t-test along with Benjamini,Hochberg false discovery rate adjustment) and foldchange.
The separation of malignant cells from the primarysite via acquisition of invasive properties and trans-port into the bloodstream are initial steps of metasta-sis. To characterize CTCs, we used a microporousdevice to filter and select CD45-negative cells fromblood samples (Table I; see the schematic diagram inFig. 1). Larger than blood cells, these cells showedirregular fibroblastoid morphology, suggestive of epi-thelial to mesenchymal transition (Fig. 1, inserts).These CTCs were individually retrieved by a micro-manipulator and used to determine their surface to-pography and mechanical properties by AFM. TheAFM-based analysis utilizes interactions between aprobe (‘‘tip’’) and a cell. Raster scanning of the cellwith a probe results in the image of cell surface, suit-able for comparing general features of surface topog-raphy, here represented by membrane roughness,between individual single cells (Fig. 2A). On the otherhand, in the AFM force mode the probe indents a cellwith a controlled force load. As a result, the cantileverto which the probe is attached is deflected proportion-ally to the applied force [25]. Figure 2B shows an ex-ample of a force curve resulting from indentation of asingle CTC at one preselected site. A blue trace repre-sents a tip approach phase in which the tip is brought
4 Chenet al.
The Prostate
into a direct contact with a cell surface from approxi-mately 1,000–700 nm. Next, the cantilever is progres-sively deflected as the tip encounters stronger cellresistance. At a preset Z position, a tip stops and thenretracts (red trace) not exactly following the approachtrace.
Based on a plot describing dependence of the canti-lever deflection on indentation, the Young modulusconstituting a measure of individual cell elasticity wasderived (Fig. 2B) [26,27]. We determined the Youngmodulus of the cultured cells from the followingestablished lines: the immortalized BPH-1 prostatecells and three prostate cancer cell lines, LNCap-AD,LNCap-AI, and PC-3. Noncancerous BPH-1 cells werethe least elastic with the Young modulus about 3.7-kilopascal (kPa), whereas the highly metastatic PC-3cells were almost 30$ more elastic (0.13-kPa, Fig. 2C).Interestingly, androgen-independent LNCap-AD cellswere more elastic then androgen-dependent LNCap-Al (0.88-kPa vs. 1.2-kPa). We also measured elasticityof four CTCs isolated from blood of a patient withcastrate-resistant prostate cancer and bone metastasis(Fig. 2D). Young moduli of these CTCs ranged from0.23 to 1.1-kPa, and the obtained values were similarto that of PC-3 elasticity, but much lower than thosevalues calculated for BPH-1 cells.
To determine cell surface roughness, images of thesame cells were acquired immediately after elasticitydetermination, by scanning the cells in contact modewith the same spherical probe. We measured rough-ness with a root RMS parameter, which correspondsto a variance of pixel heights included in an area ofinterest [28]. The RMS is measured in nm and doesnot depend on area size in the range of 1–5 mm2
(Fig. 2E). Therefore, the higher RMS value reflectsa richer relief of a cell surface and its lower valuecorresponds to a smoother surface. RMS values foundin a single cell were quite diverse reaching from 22 to90 nm. On average the PC-3 cells showed a roughercell surface than CTCs, which appeared smoother.
Specifically, the average RMS for all the PC-3 cellswas 48.7 nm, whereas for CTCs was only 25.2 nmwith the difference statistically significant at P < 0.05(Fig. 2F).
The AFM analysis presented here indicates thatcell elasticity and smoothness can be considered use-ful parameters to distinguish between non-metastaticand metastatic cells. Differences in elasticity also re-flect a histological background of a cell. The smooth-ness, commonly used to characterize a surfaceproperty of a variety of materials, reflects cell mobili-ty, distribution of surface proteins, and loss of cell po-larity [29,30]. These results suggest that the highdeformity and high smoothness of CTC membranesurface can be the result of a morphological transi-tioning of these cells into mesenchymal-like cells formalignant invasion. Considering the changes in a cellmembrane accompanying EMT and propensity to ad-here, we expect that softer and smoother cells repre-sent the most aggressive metastatic cells possiblyindicating poor prognosis.
Lossof Epithelial ProstateCancerFeaturesinCTCs
Our microporous filtration-micromanipulator sys-tem was further used to isolate 308 CD45-negativeCTCs from blood samples of eight prostate cancerpatients (Table I). CTCs were not detectable in bloodsamples from two healthy individuals (data notshown). Sixty-two of these captured cells were sub-jected to single-cell microfluidics-based RT-PCR anal-ysis of a panel of 11 known prostate epithelialmarkers and one negative control gene (CD45;Table II). Of these, 38 cells showed robust expressionof UBB, and their expression data were subsequentlyused for normalization with the expression value ofthis housekeeping gene. Included in the analysis werethree prostate cancer cell lines—PC-3, DU145, andLNCap-AD, and universal RNA as a positive controland water as a negative control.
TABLE I. Clinical InformationofProstateCancerPatients
Patient
# Age
Gleason
score
PSA
(ng/ml)
CTC isolated
(# analyzed) Metastasis Status of treatment
CC01 72 9 331.47 84 (10) Bone Castration-resistant, chemo-resistant
As shown in Figure 3, the heat map displays aremarkable heterogeneity of gene expression in these38 CTCs analyzed. The majority (93%) of these cellsexpressed EpCAM, suggesting their epithelial origin.
However, only "20% of these CTCs showed detect-able PSA and PCA-3 that are known to encode com-mon prostate-specific antigens. Other prostate cancermarkers (e.g., PSAP and PSMA) and epithelial
Fig. 1. A schematic flowchartof CTC isolation and analyses.The details are described inMaterials andMethods Section. In anti-CD45 negativeselection, representative microscopic photos show a CTC (white arrow head) was negative for anti-CD45-PE staining, whereas a blood cell(white arrow) positive. In singleCTC selection, the leftpanel shows four representativeCTCs and therightpanel illustrates the single cell isolationusing a micromanipulator and an Evos flmicroscope.A: An Evos flmicroscope and a micromanipulator (inset).B: A pipette tip pointing to a cell(white arrow) selected using a micromanipulator.C:The single cell was aspirated into the pipette tip from the place it was previously located at(white arrow).D: The selected single cell was placed on a petri dish. A higher magnification of the single cell (black rectangle) is shown in theinset.
6 Chenet al.
The Prostate
markers (cytokeratins 5, 7, and 8) were also present in20% of these circulating cells. Seven cells wereEpCAM-negative, but expressed various prostate-re-lated gene markers. Although we cannot rule outtechnical limitations of detecting some prostate can-cer-related genes at the single-cell level, our initialresults suggest a dramatic shift of gene expression oc-curring in CTCs that escaped from their primary tu-mor sites [16]. When seeded in metastatic locations,these cells may recirculate back into the bloodstreamand progressively lose their epithelial prostate charac-teristics [5].
Because of the invasive nature of CTCs, we also de-termined expression profiles of 56 EMT-related genesin these prostate cancer patients that were categorizedinto castration-resistant (i.e., four patients resistant toboth castration therapy and docetaxel chemotherapy),one castration-resistant/immunotherapy-responsive(in regards to patient’s serum PSA response observedfollowing the Provenge immunotherapy), and castra-tion-sensitive (i.e., three patients obtaining PSA re-sponse following initiation of castration therapy)groups (see Table I). Despite high degrees of tran-scriptional heterogeneity, 18 of these EMT-relatedgenes were commonly expressed in 44–100% ofthese CTCs analyzed (Fig. 3). Furthermore, expressionlevels of these genes (e.g., PTPRN2, ALDH1, ESR2,and WNT5A) were significantly higher in CTCs ofcastration-resistant patients than those of castration-resistant/immunotherapy-responsive (P < 0.01) andcastration-sensitive (P < 0.001) patients (Fig. 4A). Theexpression of the remaining 24 EMT-related geneswas less frequent (<44%) in these CTCs by the micro-fluidics-based PCR system. When expressed, incre-mental numbers and high expression values of thesegenes were significantly found in circulating cellsisolated from castration-resistant patients (P < 0.05;Fig. 4B). When further categorizing EMT-related genesinto different oncogenic signaling pathways, we foundthat upregulation of these genes was significantlyassociated with Sonic Hedgehog (P < 0.005), WNT(P < 0.05), and TGF-b (P < 0.05), suggesting theirimportant roles in metastatic castration-resistance andimmunotherapy. In silico analysis using availableexpression microarray data of a published prostatecancer cohort confirmed frequent upregulation offourteen (e.g., ESR2, WNT5A, IGF1R, PTCH1, GSK3B,MMP3, PTPRC, and EGFR) of these candidate genesin metastatic sites of prostate cancer (Fig. 5) [31].
Genes encoding for the regulation and mainte-nance of stem-cell characteristics were detected in
Fig. 2. AFM probing of cell surface indicates that CTCs exhibitmechanical phenotyperesemblinghighlymetastatic culturedpros-tate cancer cells.A: A scheme illustrating the principle ofmeasur-ing cell elasticity. A cell (blue) bound to a mica surface (gray) isindentedbya tip (red triangle)mountedona flexible cantilever (redboard) proportionally to the cell elasticity.Deflection of the canti-lever (blue arrow) changes aposition of a laserbeamreflection thatmeasures force needed to indent the cell.The distance between atip end and the cell is representedby the Z position (thick verticalarrow) directlymeasured by a piezoelectric element of themicro-scope.B: An example of a force plot of individual CTC (cell #4).Blue arrowspoint atpositions of little humps atwhich the tip likelysensed a cytoskeleton discontinuity. Adhesion forces between thetip and the cellbent the cantilever in the opposite direction as indi-cated by the red arrow.C: Histogram comparing elasticity of fourprostate cell lines.The elasticity is presented as theYoungmodulus.The benign BPH-1 cells are the stiffest (showed the largest Youngmodulus), whereas androgen dependent LNCap are more elasticfollowed by the LNCap androgen independent, and by PC-3 cellsthat are highlymetastatic, and also the softest.Histograms repre-sentmeanvalueswith corresponding SD.D: Histogramcomparingelasticity of four individual CTCs. These cells were as soft as thecancerous cell linespresentedin thepanelC.HistogramsrepresentmeanvalueswithcorrespondingSD.E:Height topographyimage ofa single CTC (cell #4) recorded with a spherical tip in contactmode. Roughness (rms in nm) of the cell membrane calculatedfor three 2.5 mm$ 2.5 mm.The cell is flat since it is tightly boundto a glass plate with poly-L-Lys and also may represent a stronglymetastaticphenotype.
TABLE II. Genes (n ¼ 84)SelectedforSingle-CellMicrofluidics-BasedRT-PCRAnalysis
Gene names Biological functions
Stem cell markerPTPRN2 Protein tyrosine phosphatase, receptor type, N polypeptide 2ALDH1(A1) Aldehyde dehyfrogenase 1 family, member A1; involved in metabolismCD44 CD44 antigen; involved in cell–cell interaction, cell adhesion, and migrationPTEN Phosphatase and tensin homolog; tumor suppressor. Acts as a dual-specificity protein
carcinogenesisMYC V-myc avian myelocytomatosis viral oncogene homolog; activate the transcription of
growth-related genesATXN1 (SCA-1) Ataxin 1; chromatin-binding factor that repress Notch signalingGATA3 GATA binding protein 3; transcription factor contains two GATA-type zinc fingersTNFSF11 (RANKL) Tumor necrosis receptor (ligand) superfamily, member 11; ligand of cytokineTNFRSF11B Tumor necrosis receptor superfamily, member 11b; acts as decoy receptor in
osteroclastogenesisTACSTD2 Tumor-associated calcium signal transducer 2; A cell surface receptor that transduces
calcium signalsOther EMT-related genesCXCL13 Chemokine (C-X-C motif) ligand 13; chemotactic for B-lymphocytesESR2 (ESRb) Estrogen receptor 2 (ER beta); nuclear receptor transcription factorsASPA Aspartoacylase; catalyzes the conversion ofN-acetyl_L-aspartic acid to aspartate and acetateCDH2 Cadherin 2; cadherin, neuronal (N-cadherin); a calcium dependent cell–cell adhesion
glycoprotein; contribute to the sorting of heterogeneous cell typesCDH1 Cadherin 1; E-cadherin (epithelial); a calcium dependent cell–cell adhesion glycoprotein;
loss of funtion contribute to progression of several cancersCOL1A2 Collagen, type 1, alpha 2; a type-I fibril-forming collagenDAB2IP DAB2 interacting protein; functions as a Ras GTPase-activating protein.FN1 Fibronectin 1; involved in cell adhsion and migration processesVIM Vimentin; Class-III intermediate filaments found in non-epithelial cells, especially
mesenchymalITGB1 (CD29) Integrin, beta 1; membrane receptors involved in cell adhesion and recognition
CTCs, but appeared as a less frequent event ("10%).However, two additional stem-cell gene markers,PTPRN2 and ALDH1, were related to EMT and werefrequently expressed in CTCs of castration-resistantpatients.
DISCUSSION
The current EpCAM-based technologies are largelyrestricted to count increased numbers of CTCs knownto correlate with advanced prostate cancer [6,9,10].Using an innovative strategy by coupling a microfil-tration system with a micromanipulator device, wehave developed a new system to characterize physicalproperties and expression patterns of individualCTCs in advanced prostate cancer patients, as well as
in established prostate cancer cell lines. This noveltechnology has permitted us to make the unexpecteddiscovery that the majority of EpCAM-positive CTCsshow loss of epithelial characteristics. In spite of highPSA values detected in the blood of these patients,these cells may not express PSA and other frequentlydetectable markers in primary prostate tumors. Shed-ding from the primary sites, these cells become highlydeformed by increasing their membranous elasticityand smoothness. It is possible that aberrant expres-sion of EMT-related genes can completely or partiallyreplace prostate epithelial features, instead displayingmesenchymal and stem-like characteristics [18,32].Activation of TGF-b signaling leads to increasedactivities of transcription factors in the TWIST, ZEB,
Fig. 3. Heterogeneous expression profiles of EMT-related and other genes among CTCs. RNA from CTCs was subjected to microfluidics-based single-cell qRT-PCRanalysisusing a BioMarkHD system.Gene expression for eachgenewas obtained as describedinMaterials andMeth-ods Section anddisplayedin ablue-whitegradient.Gene symbols andgenegroupswere labeledon the top andCTCnumbers andpatientgroupson theright.EMT-relatedgenes are furtherdivided into twogroups: the frequentlyexpressedgroup and the less frequentlyexpressedgroup.
10 Chenet al.
The Prostate
and SNAIL gene families that repress epithelial celladhesion and induce other mesenchymal proteins[16]. Overexpression of WNT agonists, FZD7 andFZD4, results in increased expression of MMP genefamilies that promote metastatic dissemination [33].
Clinically relevant to this discussion is the differ-ence observed in the EMT-related gene profiles be-tween the patients with advanced castrate-sensitiveprostate cancer (i.e., responding to castration withPSA response) and patients who are castrate- and che-mo-resistant (i.e., progressed on both castration thera-py and docetaxel chemotherapy). Patients with newlydiagnosed advanced prostate cancer are almostalways treated with medical castration therapy, themajority of which will respond favorably to therapywith improvement in PSA response, defined as aPSA %4 ng/ml at 7 months after therapy, with thoseachieving a PSA %0.2 ng/ml having a much bettermedian overall survival of 75 months. About a thirdof men however fail to achieve a PSA %4 ng/ml,develop early castrate-resistant disease, and have amedian OS of just 13 months. Identifying this subsetof patients early in their course of castration therapybased on expression patterns of EMT-related genes inCTCs would have prognostic value.
A further interesting finding relates to theEMT-related gene profile for one patient with cas-trate-resistant disease treated with the Provenge im-munotherapy instead of docetaxel chemotherapy.Provenge is now being used to treat men withasymptomatic or advanced metastatic castrate-resis-tant prostate cancer. Despite improvement in medianoverall survival, most patients did not achieve PSAresponse to therapy [34]. This patient, however, hadan improvement of PSA response from 9.29 to6.4 ng/ml following immunotherapy. Interestingly,his EMT-related gene profile in CTCs most closelyresembles that of patients with castrate-sensitive dis-ease. One of the limitations of our study is small sam-ple size, however, it is possible that this type ofsingle-cell analysis may have a predictive role in asubset of patients with castrate-resistant disease whowould benefit from immunotherapy.
In this regard, we additionally conducted a micro-fluidics-based PCR analysis of 12 oncogenes forwhich targeted inhibitors are readily available in earlyphase clinical studies at our institution. The CTC anal-ysis on these patients will allow their clinicians toconsider targeted treatments, such as PIM kinaseinhibitors, mTOR inhibitors, G-202 (a PSMA targetingpro-drug), Axl and MUC-1 inhibitors, as therapeuticoptions for these men with castrate and chemo-resis-tant disease who have exhausted all FDA-approvedagents available to them. For example, three genes,PIM3, MTOR, and ACP5 were frequently foundin CTCs of both castration-resistant and -sensitivepatients (see Fig. 3). This finding suggests that meta-static potential of CTCs may depend on the oncogenicaddiction of related signal transduction. Consider-ation should be given to this type of assessment forpatients with advanced prostate who have failedhormone ablation and second-line therapies. It isnoteworthy to mention that recently Darshan et al.reported a significant correlation between cyto-plasmic sequestration of AR and clinical responseto chemotherapy using CTCs from patients [35]. Inaddition this correlation was observed in EpCAM-positive, PSMA positive, and CD45-negative CTCs.Further mutations in AR were also detected usingCTCs from CRPCA patients [36]. However we did notinclude AR in our panel of genes as we solely focusedour efforts on EMT processes not AR signaling. Al-though EMT has been demonstrated to play a criticalrole in tumorigenesis, whether AR plays a significantrole in EMT is relatively unexplored. Nevertheless re-cent reports show that androgen deprivation inducesEMT in both normal prostate and prostate cancer[37,38]. Given these emerging data showing relation-ship between EMT and AR, identification of changesin expression of AR in CTCs would be interesting and
Fig. 4. Elevated cumulative expression of EMT-related genes andsignaling pathways in CTCs from castration-resistant patients. Cu-mulative expression EMT-relatedgenes in eachCTC are displayed inbox plots among CR, CR-IS, and CS patients. A: Cumulative geneexpression of frequently expressed EMT-related genes.B: Cumula-tive gene expression of less frequently expression EMT-relatedgenes.C: Cumulative gene expressions of WNT, SHH, and TGF-bsignaling pathways.Datawere analyzedusing one-way ANOVA andunpaired Student’s t-test. A P-value of<0.05 is considered as statisti-cally significant. CR, castration-resistant; CR-IS, castration-resis-tant and immunotherapy sensitive; CS, castration-sensitive.
Fig. 5. Elevated expression of EMT-related genes and drug target genes in metastatic prostate cancer. In silico analysis of gene expressionrevealed that expression of nine EMT-related genes and five drug target genes are higher in clinicalmetastatic prostate tumors than normalprostate.Datawere analyzedusing Student’s t-test.N, normal; AN, normal tissue adjacent to tumor;T, tumor; andM,metastatic. &P < 0.05;&&P < 0.01;&&&P < 0.001.
12 Chenet al.
The Prostate
useful. Therefore future studies will be conducted byplacing these CTCs in short-term cell culture for test-ing specific inhibitors that target EMT-related signal-ing and exploring the role of AR signaling in EMT.
ACKNOWLEDGMENTS
The authors would like to thank the staff at theCore for Advanced Translational Technologies forproviding their expertise in single-cell microfluidics-based PCR. This work is supported by grantsU54CA113001 (Integrative Cancer Biology Program),U01CA086402 (Early Detection Research Network),and P30CA054174 (Cancer Center Support Grant)of the National Institutes of Health, the University ofTexas STARS award, and gifts from the Cancer Thera-py and Research Center Fund and the Voelcker Fund.
REFERENCES
1. Aguirre-Ghiso JA. Models, mechanisms and clinical evidencefor cancer dormancy. Nat Rev Cancer 2007;7:834–846.
2. Meng S, Tripathy D, Frenkel EP, Shete S, Naftalis EZ, Huth JF,Beitsch PD, Leitch M, Hoover S, Euhus D, Haley B, Morrison L,Fleming TP, Herlyn D, Terstappen LW, Fehm T, Tucker TF,Lane N, Wang J, Uhr JW. Circulating tumor cells in patientswith breast cancer dormancy. Clin Cancer Res 2004;10:8152–8162.
3. Dittmar T, Heyder C, Gloria-Maercker E, HatzmannW, ZankerKS. Adhesion molecules and chemokines: The navigation sys-tem for circulating tumor (stem) cells to metastasize in an or-gan-specific manner. Clin Exp Metastasis 2008;25:11–32.
4. Sun YX, Schneider A, Jung Y, Wang J, Dai J, Cook K, OsmanNI, Koh-Paige AJ, Shim H, Pienta KJ, Keller ET, McCauley LK,Taichman RS. Skeletal localization and neutralization of theSDF-1(CXCL12)/CXCR4 axis blocks prostate cancer metastasisand growth in osseous sites in vivo. J Bone Miner Res 2005;20:318–329.
5. Alix-Panabieres C, Schwarzenbach H, Pantel K. Circulating tu-mor cells and circulating tumor DNA. Annu Rev Med2012;63:199–215.
6. Danila DC, Heller G, Gignac GA, Gonzalez-Espinoza R, AnandA, Tanaka E, Lilja H, Schwartz L, Larson S, Fleisher M, ScherHI. Circulating tumor cell number and prognosis in progres-sive castration-resistant prostate cancer. Clin Cancer Res 2007;13:7053–7058.
7. Allard WJ, Matera J, Miller MC, Repollet M, Connelly MC, RaoC, Tibbe AG, Uhr JW, Terstappen LW. Tumor cells circulate inthe peripheral blood of all major carcinomas but not in healthysubjects or patients with nonmalignant diseases. Clin CancerRes 2004;10:6897–6904.
8. Helo P, Cronin AM, Danila DC, Wenske S, Gonzalez-EspinozaR, Anand A, Koscuiszka M, Vaananen RM, Pettersson K, ChunFK, Steuber T, Huland H, Guillonneau BD, Eastham JA, Scar-dino PT, Fleisher M, Scher HI, Lilja H. Circulating prostate tu-mor cells detected by reverse transcription-PCR in men withlocalized or castration-refractory prostate cancer: Concordancewith CellSearch assay and association with bone metastasesand with survival. Clin Chem 2009;55:765–773.
9. Scher HI, Jia X, de Bono JS, Fleisher M, Pienta KJ, Raghavan D,Heller G. Circulating tumour cells as prognostic markers inprogressive, castration-resistant prostate cancer: A reanalysisof IMMC38 trial data. Lancet Oncol 2009;10:233–239.
11. Goodman OB Jr, Fink LM, Symanowski JT, Wong B, GrobaskiB, Pomerantz D, Ma Y, Ward DC, Vogelzang NJ. Circulatingtumor cells in patients with castration-resistant prostate cancerbaseline values and correlation with prognostic factors. CancerEpidemiol Biomarkers Prev 2009;18:1904–1913.
12. Coumans FA, Doggen CJ, Attard G, de Bono JS, TerstappenLW. All circulating EpCAM þ CK þ CD45- objects predictoverall survival in castration-resistant prostate cancer. AnnOncol 2010;21:1851–1857.
13. Danila DC, Pantel K, Fleisher M, Scher HI. Circulating tumorscells as biomarkers: Progress toward biomarker qualification.Cancer J 2011;17:438–450.
14. Yu M, Stott S, Toner M, Maheswaran S, Haber DA. Circulatingtumor cells: Approaches to isolation and characterization.J Cell Biol 2011;192:373–382.
15. Taylor RA, Toivanen R, Risbridger GP. Stem cells in prostatecancer: Treating the root of the problem. Endocr Relat Cancer2010;17:R273–R285.
16. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymaltransition. J Clin Invest 2009;119:1420–1428.
17. Doyen J, Alix-Panabieres C, Hofman P, Parks SK, Chamorey E,Naman H, Hannoun-Levi JM. Circulating tumor cells in pros-tate cancer: A potential surrogate marker of survival. Crit RevOncol Hematol 2012;81:241–256.
18. Danila DC, Fleisher M, Scher HI. Circulating tumor cells as bio-markers in prostate cancer. Clin Cancer Res 2011;17:3903–3912.
19. Desitter I, Guerrouahen BS, Benali-Furet N, Wechsler J, JannePA, Kuang Y, Yanagita M, Wang L, Berkowitz JA, Distel RJ,Cayre YE. A new device for rapid isolation by size and charac-terization of rare circulating tumor cells. Anticancer Res 2011;31:427–441.
20. Citri A, Pang ZP, Sudhof TC, Wernig M, Malenka RC. Compre-hensive qPCR profiling of gene expression in single neuronalcells. Nat Protoc 2012;7:118–127.
21. Livak KJ, Schmittgen TD. Analysis of relative gene expressiondata using real-time quantitative PCR and the 2(-Delta DeltaC(T)) Method. Methods 2001;25:402–408.
22. Popovici V, Goldstein DR, Antonov J, Jaggi R, Delorenzi M,Wirapati P. Selecting control genes for RT-QPCR using publicmicroarray data. BMC Bioinformatics 2009;10:42.
23. Powell AA, Talasaz AH, Zhang H, Coram MA, Reddy A, DengG, Telli ML, Advani RH, Carlson RW, Mollick JA, Sheth S, Kur-ian AW, Ford JM, Stockdale FE, Quake SR, Pease RF, Mindri-nos MN, Bhanot G, Dairkee SH, Davis RW, Jeffrey SS. Singlecell profiling of circulating tumor cells: Transcriptional hetero-geneity and diversity from breast cancer cell lines. PLoS ONE2012;7:e33788.
24. Gautier L, Cope L, Bolstad BM, Irizarry RA. Affy–analysis ofAffymetrix GeneChip data at the probe level. Bioinformatics2004;20:307–315.
25. Cross SE, Jin YS, Rao J, Gimzewski JK. Nanomechanical analysisof cells from cancer patients. Nat Nanotechnol 2007;2:780–783.
26. Faria EC, Ma N, Gazi E, Gardner P, Brown M, Clarke NW,Snook RD. Measurement of elastic properties of prostate cancercells using AFM. Analyst 2008;133:1498–1500.
27. Lekka M, Pogoda K, Gostek J, Klymenko O, Prauzner-Bechcicki S, Wiltowska-Zuber J, Jaczewska J, Lekki J, StachuraZ. Cancer cell recognition—Mechanical phenotype. Micron2012; 43:1259–1266.
28. Girasole M, Pompeo G, Cricenti A, Congiu-Castellano A,Andreola F, Serafino A, Frazer BH, Boumis G, Amiconi G.Roughness of the plasma membrane as an independent mor-phological parameter to study RBCs: A quantitative atomicforce microscopy investigation. Biochim Biophys Acta 2007;1768:1268–1276.
29. Chen DT, Weeks ER, Crocker JC, Islam MF, Verma R, Gruber J,Levine AJ, Lubensky TC, Yodh AG. Rheological microscopy:Local mechanical properties from microrheology. Phys RevLett 2003;90:108301.
30. Wang J, Wan Z, Liu W, Li L, Ren L, Wang X, Sun P, Zhao H,Tu Q, Zhang Z, Song N, Zhang L. Atomic force microscopestudy of tumor cell membranes following treatment with anti-cancer drugs. Biosens Bioelectron 2009;25:721–727.
31. Yu YP, Landsittel D, Jing L, Nelson J, Ren B, Liu L, McDonaldC, Thomas R, Dhir R, Finkelstein S, Michalopoulos G, BecichM, Luo JH. Gene expression alterations in prostate cancerpredicting tumor aggression and preceding development ofmalignancy. J Clin Oncol 2004;22:2790–2799.
32. Nauseef JT, Henry MD. Epithelial-to-mesenchymal transitionin prostate cancer: Paradigm or puzzle? Nat Rev Urol 2011;8:428–439.
33. Gupta S, Iljin K, Sara H, Mpindi JP, Mirtti T, Vainio P, RantalaJ, Alanen K, Nees M, Kallioniemi O. FZD4 as a mediatorof ERG oncogene-induced WNT signaling and epithelial-to-mesenchymal transition in human prostate cancer cells. CancerRes 2010;70:6735–6745.
34. Kantoff PW, Higano CS, Shore ND, Berger ER, Small EJ, Pen-son DF, Redfern CH, Ferrari AC, Dreicer R, Sims RB, Xu Y,Frohlich MW, Schellhammer PF. Sipuleucel-T immunotherapyfor castration-resistant prostate cancer. N Engl J Med 2010;363:411–422.
35. Darshan MS, Loftus MS, Thadani-Mulero M, Levy BP, EscuinD, Zhou XK, Gjyrezi A, Chanel-Vos C, Shen R, Tagawa ST,Bander NH, Nanus DM, Giannakakou P. Taxane-inducedblockade to nuclear accumulation of the androgen receptorpredicts clinical responses in metastatic prostate cancer. CancerRes 2011;71:6019–6029.
36. Jiang Y, Palma JF, Agus DB, Wang Y, Gross ME. Detectionof androgen receptor mutations in circulating tumor cellsin castration-resistant prostate cancer. Clin Chem 2010;56:1492–1495.
37. Zhu ML, Kyprianou N. Role of androgens and the androgenreceptor in epithelial-mesenchymal transition and invasion ofprostate cancer cells. FASEB J 2010;24:769–777.
38. Sun Y, Wang BE, Leong KG, Yue P, Li L, Jhunjhunwala S, ChenD, Seo K, Modrusan Z, Gao WQ, Settleman J, Johnson L. An-drogen deprivation causes epithelial-mesenchymal transitionin the prostate: Implications for androgen-deprivation therapy.Cancer Res 2012;72:527–536.
