Identification of Two Distinct Carcinoma-AssociatedFibroblast Subtypes with Differential Tumor-PromotingAbilities in Oral Squamous Cell Carcinoma
Daniela Elena Costea1, Allison Hills4, Amani H. Osman1, Johanna Thurlow5, Gabriela Kalna6,Xiaohong Huang4, Claudia Pena Murillo4, Himalaya Parajuli1, Salwa Suliman1,2, Keerthi K. Kulasekara7,Anne Chr. Johannessen1,3, and Max Partridge4
AbstractHeterogeneity of carcinoma-associated fibroblasts (CAF) has long been recognized, but the functional
significance remains poorly understood. Here, we report the distinction of two CAF subtypes in oral squamouscell carcinoma (OSCC) that have differential tumor-promoting capability, one with a transcriptome andsecretome closer to normal fibroblasts (CAF-N) and the other with a more divergent expression pattern(CAF-D). Both subtypes supported higher tumor incidence in nonobese diabetic/severe combined immunode-ficient (NOD/SCID) Ilg2(null) mice and deeper invasion of malignant keratinocytes than normal or dysplasia-associated fibroblasts, but CAF-N was more efficient than CAF-D in enhancing tumor incidence. CAF-N includedmore intrinsically motile fibroblasts maintained by high autocrine production of hyaluronan. Inhibiting CAF-Nmigration by blocking hyaluronan synthesis or chain elongation impaired invasion of adjacent OSCC cells,pinpointing fibroblast motility as an essential mechanism in this process. In contrast, CAF-D harbored fewermotilefibroblasts but synthesized higher TGF-b1 levels. TGF-b1 did not stimulate CAF-Dmigration but enhancedinvasion and expression of epithelial–mesenchymal transition (EMT) markers in malignant keratinocytes.Inhibiting TGF-b1 in three-dimensional cultures containing CAF-D impaired keratinocyte invasion, suggestingTGF-b1–induced EMTmediates CAF-D–induced carcinoma cell invasion. TGF-b1–pretreated normal fibroblastsalso induced invasive properties in transformed oral keratinocytes, indicating that TGF-b1–synthesizingfibroblasts, as well as hyaluronan-synthesizing fibroblasts, are critical for carcinoma invasion. Taken together,these results discern two subtypes of CAF that promote OSCC cell invasion via different mechanisms. Cancer Res;73(13); 3888–901. �2013 AACR.
IntroductionInduction of invasion is the key property differentiating
malignant from benign lesions. Similar to other carcinomas,oral squamous cell carcinoma (OSCC) is a multistep process inwhich the altered epithelium undergoes malignant conversionover a period of time (1). As carcinoma evolve, changes in theepithelia result also in concomitant adaptations of the adja-cent normal stroma and one of itsmain cell type, thefibroblast,
and this was shown to be important for tumor progression (2,3). There is now convincing evidence that dysplasia-associatedfibroblasts (DAF) and carcinoma-associated fibroblasts (CAF)differ from those associated with normal epithelium (4, 5), andthat these adaptations have functional consequences fortumor progression and invasion (6, 7). Initially, these carcino-ma-promoting effects were attributed mainly to the subpop-ulation of myofibroblasts in the mixed CAF populations thathad higher ability to secrete stimulative paracrine factors(3, 8, 9).
However, CAF were shown to show phenotypic and geno-typic diversity (4, 10, 11), as well as complex changes in theirsecretory activity (12–14), but the functional significance ofthis heterogeneity has been poorly addressed so far. Onlyrecently, the essential role of this heterogeneity and the coop-eration between different subpopulations was shown for pros-tate tumorigenesis, with the differential TGF-b signalingin tumor stromal cells indicated to be pivotal (15). TGF-b1essential role in carcinogenesis, either as tumor suppressor atearly stages and/or as tumor promoter at late stages has beenlong recognized (16). One of the processes throughwhich TGF-b1 participates to carcinogenesis is the epithelial–mesenchy-mal transition (EMT) phenomenon, via increased carcinoma
Authors' Affiliations: 1Section for Pathology, TheGade Institute; 2Instituteof Clinical Dentistry, University of Bergen; 3Department of Pathology,Haukeland University Hospital, Bergen, Norway; 4Head and Neck Unit,Guy's and St. Thomas' Hospitals NHS Foundation Trust, London; 5Centerfor Stem Cell Biology, Department of Biomedical Science, University ofSheffield, Sheffield; 6Beatson Institute, Glasgow, United Kingdom; and7Rural Health School, La Trobe University, Bendigo, VIC 3552, Australia
Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).
Corresponding Author: Daniela Elena Costea, Section for Pathology, TheGade Institute, Haukeland University Hospital, 5021 Bergen, Norway.Phone: 47-5597-2564; Fax: 47-5597-3158; E-mail:[email protected]
cell motility, invasiveness, and ultimately metastasis, and thisphenomenon was shown to occur in OSCC as well (17).It was the aim of this study to investigate stromal hetero-
geneity in OSCC at both molecular and functional levels, andidentify the mechanisms by which different CAF subsetssupport oral carcinoma development and invasion. We showhere a differential global gene expression profile for normaloral fibroblasts (NOF), DAF, and CAF derived from OSCC, butthe novelty is the observation that unsupervised clusteringcould identify two subgroups of CAF. Detailed analysis of theirmigratory and secretory characteristics presented here indi-cates that the 2 subtypes of CAF are able to induce oralcarcinoma cell invasion by different mechanisms.
Materials and MethodsClinical samples, cell isolation, and characterizationOSCCs, lesions with histologic evidence of dysplasia, and
contra lateral normal oral biopsies were collected followingethical approval and written informed consent. CAF (n ¼ 18),DAF (n¼ 6), and NOF (n¼ 17) were isolated and forwarded formorphologic examination, 2-dimensional (2D) cultures, 3-dimensional (3D) constructs in collagen type I biomatrices,gene microarray, quantitative reverse transcription PCR (qRT-PCR), immunohistochemistry (IHC), and flow cytometry anal-ysis (7, 18). For flow cytometry, subconfluent cells (105) weresuspended in 100 mL PBS with 2% FBS and 1% HEPES (allSigma), and incubated with 20 mL of fluorochrome-conjugatedantibody fluorescein isothiocyanate (FITC)mouse anti-humanepithelial-specific antigen (ESA; Biomeda), phycoerythrin (PE)mouse anti-human CD31, CD45, CD146, CD140b (known asplatelet-derived growth factor receptor B—PDGFRB), FITC, orPE mouse immunoglobulin G 1 (IgG1)k—isotype control (allBD Pharmingen) for 15 minutes and analyzed using a MoFlocell sorter (Beckman Coulter). For IHC, fixed cells and tissueswere stained with lineage-specific antibodies recognizingvimentin, aSMA (smooth muscle actin), CD31, pancytokeratin(panCK), S100A4 (FSP; all DAKO; 1:50), and fibroblast-activat-ing protein (FAP; Affinity Bioreagents; 1:5), and vizualized withEnVision kit (DAKO), as previously described (19). For scan-ning electronmicroscopy (SEM), 2� 104 fibroblasts were fixedin 2% glutaraldehyde/0.1 mol/L phosphate buffer pH 7.2, for 2hours at 4�C, mounted on grids and viewed using a Jeol JSM-7400 field emission-SEM. For transmission electron micros-copy (TEM), cells were postfixed in 1% osmium tetroxide(Sigma) in PBS (30 minutes), dehydrated using graded etha-nols, embedded in epoxy resin, ultrathin sectioned, doublestained with uranyl acetate and lead citrate (Sigma). Speci-mens were examined using a TEM (JEOL 1230; Jeol Ltd.), andthe micrographs processed using an Arcus II scanner (Agfa-Gevaert N.V.). For detailed characterization, see Supplemen-tary Figs. S1 and S2.
Gene expression profilingCells (2� 106) at passage 2 were seeded in 3D collagen type I
(BD Biosciences) biomatrices in duplicates for 5 days. RNAwasextracted using RNA Stat (Biogenesis Ltd.) for analysis withU133 plus 2 arrays (Affymetrix Inc.). CEL file data were nor-malized using Robust Multichip Average expression summary
(RMAexpress; http://rmaexpress.bmbolstad.com/). CEL filesfor samples and normalized data matrix are available fromhttp://www.ncbi.nlm.nih.gov/geo/ (GSE38517). Rank productanalysis (20) was conducted to identify differentially expressedgenes betweenNOF (n¼ 5), DAF (n¼ 4), and CAF (n¼ 7). GeneOntology analysis was conducted for the differentiallyexpressed genes using DAVID 6.7 (21). Data were analyzedusing Spectral Clustering, an unsupervised algorithm (22), forall strainsmaintained in 3Dbiomatrices and 2Dmonolayers, orfor 3D alone only.
genic oral dysplasic keratinocytes (DOK cell line; ref. 23)obtained from European Collection of Cell Cultures weresuspended in 50 mL of growth factor–reduced matrigel (BDBiosciences), and inoculated alone (n¼ 6) or together with 105
fibroblasts of the strains NF5 (n ¼ 6), CAF1 (n ¼ 6), or CAF5(n ¼ 6) subcutaneously on the back of 12-week-old nonobesediabetic/severe combined immunodeficient (NOD/SCID)IL2rg(null) mice (The Jackson Laboratory). Tumor incidenceand development (volume) was assessed at every 3 days. Allmice were sacrificed 45 days after inoculation and tissues wereharvested for histologic assessment. The Norwegian AnimalResearch Authority approved all animal procedures.
Tissue engineering and evaluation of carcinoma cellinvasion.
Fibroblasts at passage 3 to 7 (split ratio 1:4) were embeddedin collagen type I biomatrix (BD Biosciences), and seeded ontop with malignant or transformed oral keratinocytes intriplicate, as previously described (7). The majority of experi-ments were carried out using Ca1 malignant oral keratinocytecell line (24) that showed minimal invasion when seeded ontoNOF-populated biomatrices. Nontumorigenic DOK cells andadditional stains of malignant oral keratinocytes UK1, H357,5PT, CaLH3 (25), and SCC25 (26) were used to validate thefindings. For some experiments, NOF (n ¼ 3) were pretreatedwith 10 ng/mL TGF-b1 (BD Biosciences) for 10 days, then usedto construct 3D biomatrices. 3D constructs were harvested,formalin-fixed, and paraffin-embedded. To measure depth ofinvasion, 3-mm sections were stained for panCK (DAKO; 1:50).Subsequently, each section was divided into fifths. The centraland the two outer fifths were excluded from measurements,depth of invasion being assessed in the remaining two fifthsonly. For this, a horizontal line was drawn (using the softwareOlympus DP.Soft 5.0) through the uppermost remnants of thecollagen gel to visualize the basement membrane zone; depthof invasion was determined every 100 mmalong this horizontalline as the vertical distance from this line to the limit ofinvading epithelial cells (19).
Secretory profileEighteen-hour serum-free conditioned media was collected
from cells maintained in 2D and 3D culture at similar passagesand analyzed for levels of growth factors, cytokines, matrixmetalloproteinases (MMP), and hyaluronan by ELISA withLuminex beads (R&D Systems, Inc.), the Widescreen Human
Subtypes of Oral Carcinoma-Associated Fibroblasts
www.aacrjournals.org Cancer Res; 73(13) July 1, 2013 3889
Cancer Panel 2 (Novagen,Millipore), the FlurokineMAPTGF-bMultiplex Kit (R&D Systems, Inc.), and hyaluronan ELISA(Corgenix). The results are presented with values normalizedfor 106 cells; data represent the mean � SD. Deposition inpatient tissue and 3D constructs of hyaluronan was visualizedby histochemical staining in deparaffinized, rehydrated tissuesections stained with a 5% acidic solution of Alcian blue(Sigma) for 45 minutes and then counterstained with nuclearfast red (Sigma).
Protein detectionProtein lysates were resolved by PAGE and the membranes
were probed with antibodies recognizing Smad2, pMLC (bothCell Signaling Technology Inc.; 1:1,000), Smad2/3 (R&D Sys-tems, Inc.; 1:1,000), Hyaluronan-mediated motility receptor(RHAMM) (gift from V. Assmann, Center for ExperimentalMedicine, Institute of Tumor Biology, University HospitalHamburg, Germany; 1:400), and glyceraldehyde-3-phosphatedehydrogenase (GAPDH; Abcam; 1:1,000) as previouslydescribed (7). For validation of microarray data, additionalWestern blot analysis and IHC of monolayers and tissuesections was conducted with antibodies recognizing ITGA6,ITGA5 (all Cell Signaling Technology Inc.), GAPDH, DUSP4(both Abcam), MFAP5, COL15A1 (all Sigma), PMEPA1, CADPS(both Abnova), TAGLN (Leica), INTA11 (gift from Prof. D.Gullberg, Department of Biomedicine, University of Bergen,Norway), RAB27B (gift from M.C. Seabra, Imperial CollegeLondon, UK), and CHI3L1 (gift from S. Werner, Institute ofMolecular Health Sciences, Zurich, Switzerland), all 1:1,000.
Quantitative reverse transcription PCRCells were lysed with RNA-Stat-60 (AMS Biotechnology
Europe Ltd.), total RNA was extracted following the manufac-turer's instructions, and cDNA synthesis was conducted usingHigh-Capacity cDNA Archive Kit system (Applied Biosystems).qRT-PCR was then conducted using inventoried TaqManassays with exon-spanning probes detecting MFAP5, CADPS,ASPN, TAGLN, TIAMI,MEST, EVI1, CDKNIC, EDIL3, SRGN, VIM,S100A4, TWST1, HMGA2, FAPa, PDGFRB, COL1A2, DDR, THY1,and ACTB. Comparative 2�DDCt method was used to quantifythe relative mRNA expression.
Inhibition of hyaluronanFor some experiments, 0.3 mmol/L 4-methylumbelliferone
(4-MU; Sigma) was added to collagen matrix and media for 24hours to block elongation of hyaluronan chains andmotility ofcells. Specific inhibition of hyaluronan synthesis was achievedby hyaluronan synthase HAS2 short hairpin RNA (shRNA)lentiviral particles transduction of CAFs 1 and 3. Fibroblastswere seeded at 105 cells per well in 6-well plates and after 24hours were infected with HAS2 shRNA or control shRNAlentiviral particles (Santa Cruz Biotechnology) at a multiplicityof infection (MOI) of 100 in presence of polybrene (5 mg/mL;Santa Cruz Biotechnology), then centrifuged at 32�C for 2hours at 2,300 rpm. After 48 hours, puromycin (2 mg/mL; SantaCruz Biotechnology) was added for selection of transducedcells, and after additional 5 days, the cells were split 1:4 inpuromycin-containing medium.
TGF-b1 inhibitionA potent and specific inhibitor of TGF-b1 superfamily type I
activin receptor-like kinase (ALK) receptors ALK4, ALK5, andALK7, the small molecule SB431542 (Sigma), was used at aconcentration of 10 mmol/L, and the antibody against TGF-b1(R&D Systems, Inc.) or the isotype control were used at a con-centrationof 10mg/mL. Inhibitorswereadded30minutes beforeTGF-b1 treatment in monolayer. In 3D constructs, these reag-ents were added each second day, with the change of medium.
Motility assaysCell migration assays were conducted as previously
described (27) with and without prior exposure of cells to 10ng/mL TGF-b1 for 18 hours. For cell invasion assay, Transwellwere coated with 6 mg/mL Matrigel (Invitrogen). In someexperiments 0.3 mmol/L 4-MU (Sigma) was added to collagenmatrix and media to block elongation of hyaluronan for 24hours prior the assessment. For time-lapse microscopy, 105
cellswere seeded on 1.7mg/mL collagen (Nutacon) for 24 hoursand then treated with CellTracker Green CMFDA (MolecularProbes, Invitrogen). One hour before imaging, medium waschanged to 1% FBS/1% HEPES. Images were taken every 10minutes for 18 hours with a Zeiss LSM 510 META confocalmicroscope to determine average track speed, straightness,length, and displacement using Open Source imaging, Fiji andImaris software (MeasurementPro and ImarisTrack; Bitplane).
Statistical analysisAll data are presented as mean� SD. Mann–Whitney U test
(SPSS version 18) was used for analysis of ELISA data, ANOVAwith a post hoc Bonferroni test for RhoA/Rac G-LISA data,paired Student t test for the comparison of invasion scores, andindependent Student t test for analysis of population doublingsdata. To examine the biologic significance of the genes over-expressed in CAFs, we examined whether they correlated withoutcome using our independent microarray database of 71head and neck squamous cell carcinoma (22), with disease-specific death as the primary endpoint. Cox univariate andmultivariate analysis was conducted using the R environment(http://www.r-project.org) and Survival bioconductor package(http://www.bioconductor.org).
ResultsCAFs show increased expression of TGF-b target genes
Rank product analysis showed that 335 genes were signif-icantly upregulated and 347 were downregulated [false dis-covery rate (FDR)� 0.01] when CAF and NOF were compared(Supplementary Table S1). These genes were linked to sub-strate adhesion, tissue remodeling, cell migration, secretion,growth regulation, and angiogenesis (Table 1). Gene Ontologyanalysis confirmed that the overexpressed genes clustered intomany functional groups (Supplementary Table S1). Notably, ofthe top 100 genes (the unique gene symbols), 52 were TGF-btargets (Fig. 1A; ref. 28). In addition, many transcripts of factorsthat bind to, or modulate the bioactivity and cellular responseto TGF-b1, including ASPN (29), BGN (30), DCN (31), PMEPAI(32), and CTHRC1 (33) were upregulated when CAF werecompared with NOF (Table 1). Other upregulated genes were
Costea et al.
Cancer Res; 73(13) July 1, 2013 Cancer Research3890
NOTE: The genes differentially expressed in CAF when compared with NOF (FDR � 0.01) were linked to substrate adhesion, tissueremodeling, cell migration, secretion, growth regulation, and angiogenesis. TGF-b1 targets are shown in italics. Fewer changes werefound when DAF and NOF were compared (127 genes significantly upregulated and 75 downregulated, FD rate of �0.01). The mostoverexpressed functional group in DAF when compared with NOF were transcription factors.
Subtypes of Oral Carcinoma-Associated Fibroblasts
www.aacrjournals.org Cancer Res; 73(13) July 1, 2013 3891
known modulators of cellular secretion, such as CADPS (34),RAB27B (35), and SRGN (8-, 9-, and 6-fold, respectively; ref. 36).Fewer changes were found when DAF and NOF were com-pared, with 127 genes significantly upregulated and 75 down-
regulated (FDR� 0.01). Interestingly, many of the upregulatedgenes in DAF were transcription factors (Table 1). The prog-nostic significance of the top-ranked 50 probe sets and TGF-b1targets identified here by rank product analysis as being
Figure 1. A, heatmap showing differentially expressed transcripts of TGF-b target genes between NOF and CAF. Expression values are log2, mean centered(red, higher expression; green, lower expression). B, Kaplan–Meier analysis of 71 head and neck SCC cases for PMEPA1 (probe set 222450_at). The blackline shows the quartile with the lowest PMEPA1 expression (0–0.25; 4 events/18 cases); the red line the other quartiles (0.25–1; 23 events/53 cases).C, Kaplan–Meier analysis of 71 HNSCC cases for BGN (201261_x_at). The black line shows the quartile with lowest BGN expression (0–0.25; 1 events/18cases); the red line the other quartiles (0.25–1; 26 events/53 cases). D, Kaplan–Meier analysis of 71 head and neck SCC cases for CADPS (1568603_at).The black line shows the quartiles with lowest CADPS expression (0–0.75; 15 events/54 cases); the red line shows the quartile with highest CADPSexpression (0.75–1; 12 events/17 cases). E, spectral clustering showing distinct grouping of fibroblasts grown in 2D and 3D, and for those maintained in 3Dseparate grouping of NOF, DAF, and CAF. CAF clustered into 2 subgroups: one with a transcriptome closer to NOF (CAF 1-4, termed CAF-N), the otherwith a more divergent expression profile (CAF 5-7, termed CAF-D). F, hierarchical clustering showing the tight grouping of the more homogenousgroup of CAF-N (closest to DAF and NOF) when compared with the more heterogenous and divergent group of CAF-D. G, heatmap showing differentiallyexpressed probe sets between CAF-N and CAF-D of genes enriching GO BP 5, GO:0016477�cell migration. Expression values are log2, mean centered(red, higher expression; green, lower expression). H, heatmap showing differentially expressed probe sets between CAF-D and CAF-N of genesenriching GO:0014032�neural crest cell development and GO BP 5, GO:0016477�cell migration. Expression values are log2, mean centered (red, higherexpression; green, lower expression).
Costea et al.
Cancer Res; 73(13) July 1, 2013 Cancer Research3892
upregulated in CAF versus NOF was investigated on ourindependent microarray database of 71 head and neck SCC.Univariate analysis showed that 20 of these probe sets weresignificantly associatedwith reduced disease-free survival (P�0.05). These genes included PMEPA1 (P ¼ 0.025), BGN (P ¼0.0058), and CADPS (P < 0.001), Kaplan–Meier analysis showedthe quartile of cases with the lowest expression of these geneshad the best outcome (Fig. 1B–D). Significant changes in geneexpression were validated by qRT-PCR, IHC, and Western blotanalysis with additional strains of fibroblasts to those profiledon Affymetrix arraysmaintained in 3D culture (SupplementaryFig. S3).
Gene expression profile identifies two subgroups of CAFderived from OSCCSpectral clustering showed a clear separation of fibro-
blasts grown in 2D cultures from those maintained in 3Dbiomatrices (Fig. 1E), and for the cells in 3D biomatrices,NOF and DAF grouped at different positions from CAF,reflecting their distinct gene expression signatures (Fig.1E). Further analysis using hierarchical clustering (Fig.1F), and various clustering arrangements of 3D culturesalone, showed that CAF 1–4 tightly grouped together, indi-cating that these strains were transcriptionally more similarto each other than the rest of the CAF strains (CAF 5–7). CAF1–4 often clustered closely to NOF, and for these reasonswere termed CAF-N; whereas the more transcriptionallydivergent CAF 5–7 were termed CAF-D. Of note, the NOFstrain (NF1) closest to CAF strains in some clusteringarrangements was from a juvenile, and the DAF strain(DAF3) that also clustered closest to CAF strains wasisolated from a patient who later developed OSCC. GeneOntology analysis revealed that differentially expressedgenes between CAF-N and CAF-D clustered in the functionalgroups of cell migration and cell surface signal transduction(Supplementary Table S1.11 and S1.12) Some of the genesupregulated in CAF-N versus CAF-D belonged to cell migra-tion (e.g., TBX1, KIT, ITGA4, CXCL12, and NR2F1) andthe angiogenesis functional group (Fig. 1G), whereas someof the genes upregulated in CAF-D versus CAF-N (e.g., PAX3,NRP2, and EDNRB) were associated with mesenchymal/neural crest development and amoeboidal cell movement(Fig. 1H).
The two CAF subgroups provide differential support fortumor formation and invasionDOK cells did not develop tumors when xenotransplanted
subcutaneously in NOD/SCID IL2rg(null) mice. Tumor inci-dence increased from 0 (0 of 6), when DOK were inoculatedalone, to 66.66% (8 of 12) by coinoculating DOK with CAF. Asignificantly higher tumor incidence was observed for DOKcoinoculated with CAF-N (83.33%, 5 of 6) when comparedwith DOK coinoculated with CAF-D (50%, 3 of 6; Fig. 2A). Inaddition, DOK/CAF-N tumors developed after a shorter lagtime (median of 7 days) than DOK/CAF-D tumors (median of14 days; Fig. 2A), and showed invasion into the musculaturelayer (Fig. 2B), in contrast to the well-circumscribed DOK/CAF-D tumors (Fig. 2C). Quantification of the fibroblast
support for carcinoma cell invasion using in vitro 3D con-structs showed that CAF-N were significantly more effectivein supporting deeper invasion of carcinoma cells than CAF-D (P ¼ 0.005; Fig. 2D), with a mix of small islands and singlecarcinoma cells invading the 3D biomatrix (Fig. 2E). Never-theless, both subtypes of CAF supported significantly deeperinvasion of oral carcinoma cells when compared with NOFembedded in 3D biomatrix (Ca1 cell line; Fig. 2D and E). Thiswas confirmed using other cell lines (DOK, SCC25, CaLH3,and UK1; Supplementary Fig. S4). An intermediate pattern ofinvasion was observed when DAF were incorporated into the3D biomatrix (Supplementary Fig. S4).
The two CAF subgroups show a differential secretoryprofile
To identify growth factors and multifunctional cytokinesthat could be linked to the key differences in CAF behavior,we investigated their secretory profile (Supplementary Fig.S5). Multiplex ELISA assay of conditioned medium fromfibroblasts maintained in 3D biomatrices showed that CAFsynthesized significantly more TGF-b1 (P < 0.001), interleu-kin (IL)-1a (P < 0.001), acidic fibroblast growth factor (aFGF;P¼ 0.008), TNF-a (P¼ 0.014), and keratinocyte growth factor(KGF) (P ¼ 0.04) than NOF, whereas more VEGF (P ¼ 0.003)and hepatocyte growth factor (HGF; P ¼ 0.007) was secretedby NOF (Supplementary Fig. S5B). CAF-N secreted signifi-cantly more KGF (P ¼ 0.019), HGF (P ¼ 0.002), aFGF (P ¼0.008), and MMP3 (P ¼ 0.016) than CAF-D (Fig. 4A), whereasCAF-D secreted 9 times more TGF-b1 (P < 0.001; Fig. 3A) andmore IL-1a (P ¼ 0.006) than CAF-N (Fig. 3A).
TGF-b1 increases invasion and expression of EMTmarkers in OSCC cells, whereas TGF-b1 inhibition bySB431542 impairsCAF-D–induced invasionofOSCCcellsin 3D constructs
Provided that CAF-D stimulated significantly deeper car-cinoma cell invasion when compared with NOF, the obser-vation that TGF-b1 was the most obvious change in theirsecretory profile determined us to assess the TGF-b1 effecton carcinoma cell migration and invasion. Transwell assaysshowed significant increase of oral carcinoma cell migrationand invasion after treatment with TGF-b1 for all cell linestested (P ¼ 0.012 and 0.002, respectively), varying from 2- to3-folds (Fig. 3B and C), Although the amplitude and kineticsof the response to TGF-b1 treatment varied between the celllines investigated, qRT-PCR showed an increased expressionof EMT-related markers after 5 hours of exogenous TGF-b1exposure (Fig. 3D). IHC showed significant increase invimentin expression as long as 5 days after TGF-b1 treat-ment (Fig. 3E–G). Significant decrease in cell invasion wasobserved when 3D biomatrices populated with CAF-D wereseeded with DOK and various other carcinoma cell lines ontop (n ¼ 3), and 24 hours later treated with SB431542 (Fig.3F–I). Significant decrease in invasion was also observed forCa1 and 5PT cell lines occurring after treatment with theanti-TGF-b1-Ab, but this effect was not statistically signif-icant (Fig. 3F–I). Inhibition of activation of TGF-b1 down-stream pathways after treatment with inhibitors 30 minutes
Subtypes of Oral Carcinoma-Associated Fibroblasts
www.aacrjournals.org Cancer Res; 73(13) July 1, 2013 3893
before TGF-b1 stimulation could be shown with SB431542but not anti-TGF-b1-Ab (Fig. 3J).
The two CAF subgroups have distinct motilityphenotypes
To determine other critical differences between the 2 CAFsubtypes that might be linked to their differential ability tosupport tumor development and invasion, we investigatedfibroblast motility. Transwell migratory experiments showedthat CAF-N contained a significantly higher percentage ofintrinsically migratory fibroblasts than CAF-D or NOF (Fig.
4A). The presence of this subpopulation of migratory fibro-blasts predominantly in CAF-N was also indicated byincreased expression of pSmad2 when CAF-N and NOF werecompared (Fig. 4B). The significantly higher levels of RhoA-GTPase and Rac1/2/3-GTPase in CAF-N when comparedwith NOF (P ¼ 0.0001; Supplementary Fig. S6) reflected alsothe higher proportion of migratory fibroblasts in CAF-Nwhen compared with NOF. Furthermore, time-lapse micros-copy showed that CAF-N moved faster and for longer dis-tances from the initial point (longer track displacement)than NOF (Fig. 4C and D).
D
Ca1/CAF3
CAF-N
Ca1/CAF1
Ca1/CAF3Ca1/CAF1
Ca1/CAF5
Ca1/CAF5
Ca1/CAF6
Ca1/CAF6
Ca1/NF4
Ca1/NF4
Ca1/NF7
Ca1/NF7
CAF-D
CAF-N
Depth
of in
vasio
n (
μm)
CAF1
200
180
160
140
120
100
80
60
40
20
0CAF3 CAF4 CAF5 CAF6 CAF7 NF1 NF4 NF6 NF7
CAF-D NOF
NOF
pan
CK
H&
E
B
C
E
DOK/CAF1 DOK/CAF5
-2
3
8
13
18
6
5
4
3
2
1
0
23
28
33
38
45 d35 d28 d21 d14 d7 d42 d35 d28 d21 d14 d
No
. tu
mo
rs
7 d
DOK ALONE
DOK/CAF1
DOK/CAF5
DOK/CAF1 DOK/CAF5Ap
anC
K
H
&E
F
Figure 2. A, incidence and lag timeof tumors formed by DOK cellscoinoculated subcutaneously withCAF-N (CAF1) or CAF-D (CAF5) inNOD/SCID/Ilg2(null) mice. B,microphotograph showing thetumor front of invasion into themuscular layer of a tumordeveloped after coinoculation ofDOK and CAF1; hematoxylinand eosin (H&E) staining. C,microphotograph showing a welldemarcated tumor developed aftercoinjection of DOK andCAF5 (H&Estaining; magnification, �200;scale bar, 200 mm). D, the depth ofinvasion of malignant oralkeratinocytes (Ca1) wassignificantly higher when CAF-Ns,as opposed to CAF-D and NOF,were embedded into the biomatrix.Significant differences (P ¼ 0.001)are marked by a star. E, top,representative H&E-stainedsections of 3D constructs withmalignant oral keratinocytes (Ca1)seeded onto CAF-N, CAF-D, orNOF. Bottom, panCK stainingof serial sections. Magnification,�200. Scale bar, 200 mm.
Costea et al.
Cancer Res; 73(13) July 1, 2013 Cancer Research3894
CAF-N subpopulation of intrinsically motile fibroblastsis dependent on hyaluronan and is essential forsupporting carcinoma cell invasionDifferential intrinsic migratory activity has been linked to
inherent differences in the production of hyaluronan; there-
fore hyaluronan was next investigated (37). Significantlyhigher levels of hyaluronan were found when comparing allavailable CAF strains (n¼ 9) with NOF (n¼ 3) maintained in3D biomatrices (P ¼ 0.008; Supplemental Fig. S4A), and thelevels of secreted hyaluronan were significantly higher for
Figure 3. A, significant differencesin levels of secretion of variousgrowth factors, cytokines, andMMPs determined for NOF,CAF-N, and CAF-Dmaintained in 3Dbiomatrices. CAF-N secreted higherlevels of KGF and HGF than CAF-D,whereas more TGF-b1 was secretedby CAF-D, P < 0.05. B, cell migrationin a Transwell assay for OSCC-derived cell lines and DOK cells.Cells treated with 1 ng/mL TGF-b1for 18 hours migrated significantlymore than the control group. C, cellinvasion in a Matrigel Transwellassay for the same cell lines.TGF-b1–treated cells invadedsignificantly more than the controlgroup. D, the expression of mRNAfor various EMTmarkers in Ca1 cellstreated with TGF-b1 relative tocontrol cells. E, expression ofvimentin as detected by IHC in Ca1cells. F, expression of vimentin asdetected by IHC in TGF-b1–treatedCa1 cells. G, the number of vimentin-positive cells after TGF-b1 treatmentwas significantly higher than incontrol cells. Statistical significantdifferences between the groupsP < 0.05 are highlighted by a star. H,representative hematoxylin andeosin–stained sections of 3Dconstructs with malignant oralkeratinocytes (CaLH3) seeded ontoCAF-D and 24 hours later treatedeitherwithSB431542 or anti-TGF-b1monoclonal antibody for 10 days(magnification, �200; scale bar, 200mm). I, quantification of the depth ofinvasion of malignant keratinocytesin 3D constructs with and withoutaddition of TGF-b1 inhibitors. J,Western blot analysis for pSMAD2and SMAD2/3 showing inhibition ofTGF-b1downstreampathway inCa1cells treated with with SB431542 butnot when treated with anti-TGF-b1monoclonal antibody.
1
10
100
1,000
10,000
100,000
300
250
200
150
100
50
0
2.5
2.0
1.5
1.0
0.5
0.0
80
70
60
50
40
30
20
10
0
% o
f positi
ve c
ells
200
180
160
140
120
100
80
60
40
20
0
TGF-β1aFGFMMP3HGF
Ctrl.
H357
EMT markers in TGF-β–treated Cal relativ to ctrl. Vimentin in Cal +/– TGF-β
mR
NA
rela
tive v
alu
e
(TG
F-β
tre
ate
d v
s.
ctr
l cells
)
N.a.N. 1 h 5 h 24 h
VimentinS100A4TwistHMGA2
72 h 120 h
Cal DOK H357 Cal DOK
TGF-β1 Ctrl.
TGF-β1
KGF
Mean m
igra
ting c
ells
/feild
Mean invadin
g c
ell/
feild
IL-1αng/m
L p
er
10
6 c
ells
NOFCAF-NCAF-D
A
CB
E
F
GCa1 ctrl.
Ctrl.
Ca1+TGF-β1
TGF-β1
D
HCaLH3/CAF5+SB 431542
CaLH3/CAF5
0.00
20.00
40.00
60.00
80.00
100.00
120.00
140.00
DOK4PTCaLH3Ca1
Dep
th o
f in
vasi
on
(μm
)
Ctrl.
SB 431542
Anti TGFB1 Ab
I
J
pSMAD2
SMAD2/3
Ctrl. TGF-β1 SB AB
CaLH3/CAF5+AB anti-TGFβ-1
Subtypes of Oral Carcinoma-Associated Fibroblasts
www.aacrjournals.org Cancer Res; 73(13) July 1, 2013 3895
CAF-N than CAF-D (P ¼ 0.002; Fig. 5A). HAS2 was alsoupregulated 3-fold in CAF-N when compared with CAF-D(Supplementary Table S1.4). Histochemical visualization ofhyaluronan by Alcian blue staining showed abundant hya-luronan in the tumor stroma from which CAF-N strains wereisolated (Supplementary Fig. S7D), and a fine, discontinuousdeposition of hyaluronan around cords of invasive cells in 3Dconstructs with Ca1 cells onto CAF3 (Supplementary Fig.S7E and S7F). IHC and Western blot analysis revealed thathyaluronan receptors RHAMM (Fig. 5B and C) and CD44(data not shown) were expressed by CAF. Blocking hyalur-onan chain elongation by addition of 4-MU resulted in asignificant decrease in the proportion of intrinsically migra-tory fibroblasts in CAF-N (Fig. 5D), identifying hyaluronan asa key factor for the maintenance of this CAF-N migratorysubpopulation. SEM of control and 4-MU–treated CAF-Nrevealed a reduction in the number of lamellipodia andfilopodia (Fig. 5E). The functional importance of hyaluro-nan-dependent subpopulation of intrinsically migratoryCAF-N for oral carcinoma cell invasion was proven by asignificant reduction in the depth of invasion of oral carci-noma cells when CAF-N pretreated with 4-MU were incor-porated into 3D constructs and seeded on top with Ca1 cells(P ¼ 0.001; Fig. 5F and G). 4-MU is described as a selectiveinhibitor of nonsulfated GlcUA-containing glycosaminogly-cans, and thus of hyaluronan production, but other effectson related glycosaminoglycans cannot be excluded. For thisreason, we also specifically targeted hyaluronan synthesiswith HAS2 shRNA lentiviral particles. qPCR showed a 84.50%reduction in HAS2 mRNA in HAS2 shRNA-treated CAF3compared with control shRNA-treated CAF3 (Supplementa-ry Fig. S8), and this downregulation induced a statistically
significant inhibition of invasion of carcinoma cells in 3Dconstructs (P ¼ 0.0067; Fig. 5H and I).
TGF-b1 activation of NOFs is sufficient to promoteinvasion of transformed oral keratinocytes
Upregulation of TGF-b1 target genes in CAF prompted us toestablish whether this had any functional relevance for CAFmotility and subsequent oral carcinoma cell invasion. Trans-well assays revealed that after exposure to 10 ng/mL TGF-b1, asignificantly higher proportion of migratory fibroblasts werepresent in CAF-N and NOF than in CAF-D (Fig. 6A), indicatingthat CAF-D contained only low numbers of fibroblasts thatmigrated in response to exogenous TGF-b1. This was alsoindicated by the increased expression of Smad2 in TGF-b1–treated CAF-N and NOF (Fig. 6B), whereas pSmad2 was almostundetectable in CAF-D either before or after exposure toexogenous TGF-b1 (data not shown). Broadly, similar percen-tages of na€�ve fibroblasts that migrated in response to exog-enous TGF-b1 were observed when CAF-N and NOF werecompared (Fig. 6A). These data show that CAF-N include, inaddition to the subpopulation of intrinsically motile fibro-blasts, another subset of na€�ve fibroblasts that are able tobecome motile after stimulation with exogenous TGF-b1, andthat these subtypes of motile fibroblasts are largely absent inCAF-D (Fig. 6C). The functional importance for the TGF-b1–migratory activation of fibroblasts for carcinoma cell invasionwas tested in 3D models by pretreating NOF with 10 ng/mLTGF-b1 for 10 days and embedding afterward these activatedfibroblasts into a 3D biomatrix seeded on top with DOKcells. Nonactivated, matched fibroblasts supported only aminimal DOK cell invasion (Fig. 6D), whereas TGF-b1–acti-vated fibroblasts supported significantly deeper invasion in 3D
B
pMLC
Smad2
pSmad2
GAPDH
μmμm/s
CAF1 CAF3 CAF4 NF1 NF2 NF4
NOF CAF-N
CAF-N NOF CAF-N NOF
CAF-N CAF-D NOF
A
D C
% M
igra
tory
ce
lls
Figure 4. A, Transwell migrationassay showing various fibroblaststrains with distinct migratorycharacteristics: CAF-N with thehighest proportion of intrinsicallymigratory cells and the relativelystationary CAF-D. B, Western blotanalysis showing increased levelsof pMLC and pSmad2 in CAF-Nwhen compared with NOF,reflecting the presence of a higherproportion of intrinsicallymigratoryfibroblast in CAF-N. C and D, trackspeed (C) and track displacement(D) of CAF-N and NOF quantifiedby time-lapse microscopy over 18hours: CAF-N migrated at higherspeed and showed greaterdisplacement (the shortest pathbetween the start and theendpoints of the total cellmovement track) than NOF.Statistical significant differencesbetween groups, P < 0.05, arehighlighted by a star.
Costea et al.
Cancer Res; 73(13) July 1, 2013 Cancer Research3896
biomatrices (Fig. 6E). The pattern of invasion induced by TGF-b1–activated NOF showed predominantly cords of cells andsingle cells invading vertically in the biomatrix (Fig. 6F), similarto the pattern of carcinoma cell invasion induced by CAF-N(Fig. 6G).
DiscussionTwo distinct CAF subtypes were identified in this study by
transcriptome and secretome analysis: CAF-N with a geneexpression pattern and secretory profile closer to NOF, andCAF-D with a divergent transcriptome and secreting very highlevels of TGF-b1. The CAF-N subtype included a high percent-age of intrinsically motile CAFs, and the number of migratorycells could be increased in response to TGF-b1, whereas the
CAF-D subtype included few motile cells and the motilityphenotype was largely unchanged in response to TGF-b1.
Functional effects of the distinct CAF subtypes were eluci-dated by implanting transformed, nontumorigenic oral kera-tinocytes together with CAF into NOD/SCID IL2rg(null) mice.CAF-N significantly increased tumor incidence and shortenedthe lag time for tumor formation as compared with CAF-D.CAF-N also supported deeper oral carcinoma cell invasionthan CAF-D in 3D models, with areas of noncohesive invasionas well as vertical proliferation of cords of carcinoma cells, apattern of invasion previously correlated with poor prognosis(37). The observation that CAF-N supported the best tumorformation and invasion indicates the pivotal role of CAFmigration for oral carcinoma cell invasion, in linewith previous
Figure 5. A, CAF secretedsignificantly more hyaluronan thanNOF (P ¼ 0.008) and that CAF-Nsynthesized significantly higherlevels than CAF-D (P ¼ 0.002). B,IHC and (C) Western blot analysisconfirmed that CAF and malignantoral keratinocytes expressedhyaluronan-receptor RHAMM. D,significantly fewer CAF-N migratedthroughTranswell after exposure to4-MU that prevents elongation ofhyaluronan chains, P < 0.001.E, CAF-N morphology as depictedby SEM showing larger, flattenedcells with fewer filopodia afterexposure to 4-MU when comparedwith untreated controls. Top,magnification, �1,500; scale bar,10 mm; bottom, magnification,�5,000; scale bar, 1 mm. F, panCKIHC showing reduced invasion ofCa1 carcinoma cells after CAF-Nwere exposed to 4-MU, confirmedby measurement of thedepth of invasion (G;magnification,�200; scale bar, 200 mm). H,representative hematoxylin andeosin–stained sections of 3Dconstructs with malignant oralkeratinocytes (Ca1 and CaLH3)seeded onto ctrl-shRNA- andHAS2shRNA-treated CAF3(magnification, �200; scale bar,200 mm). I, quantification of thedepth of invasion of malignantkeratinocytes in 3D constructs withctrl-shRNA- and HAS2shRNA-treated CAF3.
B
CAF-N NOF
CAF-N NOF
Control
4MU-treated
Control
4MU-treated
C
ED
F G
CAF1-M CAF1+4MU CAF3-M CAF3+4MU
RHAMMRHAMMSecretion of hyaluronan
0
5,000
10,000
15,000
20,000
25,000
30,000
35,000
40,000
45,000
50,000
CAF-DCAF-N
CAF1 CAF3 NF1 NF4
CAF1 CAF3 NF1
NOF%
Mig
rato
ry c
ells
50
45
40
35
30
25
20
15
10
5
0
60
50
40
30
20
10
0
ng
/mL
pe
r 1
06 c
ells
Ca1/CAF1 Ca1/4MU-treated CAF1
A
RHAMMGAPDH
CAF1 CAF3 CAF4 NF1 NF2 NF4
NOFCAF-N
H I
0.00
20.00
40.00
60.00
80.00
100.00
120.00
140.00
160.00
180.00
200.00
DOK4PTCaLH3Ca1
Dep
th o
f in
vasi
on
(μm
)D
epth
of
inva
sio
n (
μm)
shGFP-ctrl
shHAS2
CaL
H3
Ca1
CAF3 ctrl-shRNA CAF3 HAS2 shRNA
Subtypes of Oral Carcinoma-Associated Fibroblasts
www.aacrjournals.org Cancer Res; 73(13) July 1, 2013 3897
publications (38). However, this was previously linked to therequirement for expression of the integrins A3 and A5. Theseintegrins were not upregulated in the present study (Supple-mentary Table S1 and Supplementary Fig. S3), suggesting thatalternative integrins, including A6, A10, and A11 that were
upregulated in the present study (Supplementary Table S1 andSupplementary Fig. S3) and other reports (39, 40) may substi-tute for A3 and A5.
Analysis of data from migration assays identified at least3 different phenotypes of fibroblasts contributing to the
pSmad2
Smad2/3
GAPDH
+TGF-β1 +TGF-β1 +TGF-β1TGF-β1–treatedControl
90
80
70
60
50
40
30
20
10
0
% M
igra
tory
ce
lls
CAF3 CAF3 NF1 NF1 SF SF
CAF-N NOF Ctrl.
CAF-NCAF1 CAF3 CAF6 CAF7
CAF1 CAF3 CAF6 CAF7
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
NF1 NF4
Stationary/secretory
Migratory afterTGF-β1stimulation/naïve
Migratory/intrinsicallyactivated
CAF-D NOF
CAF-NNF1 NF4
CAF-D NOF
B A
C
DOK/NF16D
DOK/TGF-β1-pretreated ENF16
DOK/CAF1GTGF-β1-pretreated FNF16
Figure 6. A, Transwell migration assay showing CAF distinct response to TGF-b1: CAF-N showed increase of the subpopulation of migratory cells afterexposure to TGF-b1, whereas CAF-Dwere refractory to exposure to TGF-b1. B,Western blot analysis showing increased levels of pSmad2 after 1 hour TGF-b1 exposure of CAF-N and NOF. Skin fibroblasts (SF) were used as control for activation of TGF-b1 downstream pathway. C, the distribution ofdifferent cell phenotypes contributing to the heterogeneity of each fibroblast strain analyzed: (i) intrinsically motile cells; (ii) TGF-b1-responsive cells; and (iii)stationary/low-migratory and high-secretory cells. CAF-N and NOF contained significantly higher percentages of intrinsically and TGF-b1–inducedmotile cells when compared with CAF-D (P¼ 0.05). CAF-D contained a significantly high percentage of low-motile, highly secretory cells (P < 0.01). D and E,hematoxylin and eosin staining of representative 3D constructs with DOK cells seeded onto control NOF (D) or matched NOF pretreated with TGF-b1 (E),showing that TGF-b1–activatedNOF supported a deeper invasion (magnification,�200; scale bar, 200mm). F, closer view (magnification,�400; scale bar, 50mm) showing invasion of DOKwhen seeded onto NOF pretreatedwith TGF-b1, with small islands, single cells, and cords of cells growing deep into thematrix(noncohesive pattern of invasion). G, closer view (magnification, �400; scale bar, 50 mm) showing invasion of DOK when seeded onto CAF-N with anoncohesive, vertical pattern of invasion, similar to the invasion pattern seen when DOK were seeded onto NOF pretreated with TGF-b1.
Costea et al.
Cancer Res; 73(13) July 1, 2013 Cancer Research3898
heterogeneity of each fibroblast strain: (i) intrinsically motilecells; (ii) TGF-b1–responsive cells; and (iii) stationary/low-migratory cells (Fig. 6C). It has been previously shownthat migratory cells are not active synthesizers (41), there-fore it is most likely that the third subpopulation of sta-tionary/low migratory fibroblasts is responsible for synthesisof various growth factors, cytokines, MMPs, and extracellu-lar matrix (ECM) components including hyaluronan. Asdepicted in Fig. 6C, all fibroblast strains were heterogenous,but presence of these subtypes in various proportions, andtheir cooperative effects was found to be essential for thefunctional differences observed between NOF, CAF-N, andCAF-D. CAF-N were most efficient at maximizing the poten-tial contact with transformed keratinocytes due to the highproportion of intrinsically migratory fibroblasts that wereable to create a network of tracks used by oral transformedcells for invasion, as previously suggested (38). The main-tenance of this subpopulation and its effects on carcinomacell invasion was dependent on hyaluronan secretion. TheCAF-N strain that secreted the highest levels of hyaluronan(CAF3) supported the deepest invasion of malignant oralkeratinocytes, pinpointing the notion that cooperationbetween different subpopulations of CAF (e.g., motile cellsand hyaluronan-synthesizers) results in increased ability topromote carcinoma invasion. The broad similarity betweenCAF-N, which supported the highest rate of tumor formationand the deepest invasion, and NOF was unexpected but isconsistent with the notion that na€�ve fibroblasts undergoadaptations in response to signals in the milieu (2).The microarray analysis conducted in this study indicated
the pivotal role of TGF-b in the stromal adaptations that occuras oral carcinoma evolves, as we found that approximately 50%(52 of 100) of the top genes upregulated in CAF compared withNOF were TGF-b targets. The importance of TGF-b1 for CAFactivation was confirmed by demonstration of significantdeeper invasion occurring when TGF-b1–pretreated NOFwereembedded in 3D constructs and seeded on top with trans-formed oral keratinocytes, as compared with matched con-trols. Most likely, inflammatory cells that are present in vivo inthe tumor stroma provide TGF-b1 that initiates this process(42), but over time, the change in the transcriptome, withincreased expression of many TGF-b1 target genes, leads toemergence of the nonmotile (41), high TGF-b1–secretingfibroblast subtype (14). By correlating Transwell motility assaywith secretome analysis, we found this subset of high TGF-b1secretors to be present at significantly higher proportions inCAF-D strains. This finding might indicate that CAF-N andCAF-D are actually 2 different stages of CAF in the OSCCprogression, with CAF-N representing an earlier stromalchange than CAF-D. This is also suggested by the findingsthat CAF-N are closer to NOF and DAF in many aspects, andtheir impact on carcinoma cell invasion is more dramatic, as itmight be necessary for inducing invasion at early, but notlate stages of carcinogenesis. It has been previously shownthat increased HGF synthesis suppresses TGF-b1 secretion,and that changes in the balance between HGF and TGF-b1synthesis by fibroblasts might be decisive in pathogenesis ofother diseases (43). Such switch might also occur with OSCC
progression, as indicated by our results here. The high secre-tion of HGF might also maintain low the synthesis of TGF-b1in NOF; it lowers progressively in CAF-N, and switchescompletely to low HGF and high TGF-b1 secretion in CAF-D (Fig. 3A). The mechanism responsible for this shift iscurrently unknown, as is the mechanism switching fromTGF-b1–responsive to TGF-b1–refractory CAF. A possibleexplanation might be loss of TGFBR2 by CAF as previouslysuggested (44), or defects in Smad signaling as found formalignant keratinocytes (45). Increased expression of genesthat bind to, or ameliorate the response to TGF-b1, includingBGN and ASPN (31, 46), or PMEPA1 and CTHRC1 that interferewith phosphorylation of Smads 2/3 (32, 33, 47) may alsomaintain the distinct CAF subpopulations.
Nevertheless, secretion of high levels of TGF-b1 not onlyinduces NOF to evolve into CAF; it can also stimulate invasionof malignant keratinocytes directly through EMT (48). In linewith this, our results suggest that CAF-D, being high TGF-b1secretors, are able to stimulate EMT and oral cancer cellinvasion via TGF-b1 (Fig. 3). However, this high TGF-b1–secreting CAF phenotype (CAF-D) did not support deep inva-sion in our experiments. Migration of keratinocytes deep intostroma probably requires additional mechanisms such as asubpopulation of highly motile fibroblasts, and a subpopula-tion of cells that also secrete hyaluronan, MMPs, or growthfactors that support proliferation (KGF andHGF). In vivo, theseother subpopulations of cells might be continuously renewedfrom bone marrow or adjacent normal connective tissue, butwhether or not the fibroblast subtypes identified in this studyshare a common lineage is presently unknown. Highly motilefibroblasts are also known to be present in wounds (49), thus itwill be instructive to tease out differences between wound-associated fibroblasts and CAF to establish whether they arebroadly similar, or whether CAF have unique characteristics.Most likely, it is the persistence of the transcriptome changes infibroblasts in the developing tumor milieu that drives thealtered balance in the subpopulations identified in the presentstudy. Our experiments indicate that such changes are of adynamic nature, and the reduced life span of CAF (Supple-mentary Fig. S1)may indicate that in vivo, thefibroblasts with achanged transcriptome are eliminated and replaced continu-ously. Nevertheless, these adaptations support tumor forma-tion and have prognostic significance, and we provide here forthe first time evidence that CAF subtypes have functionalrelevance via different mechanisms to support tumor forma-tion and invasion.
Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.
Authors' ContributionsConception and design: D.E. Costea, A.H. Osman, J. Thurlow, H. Parajuli,S. Suliman, A.C. Johannessen, M. PartridgeDevelopment of methodology: D.E. Costea, A. Hills, A.H. Osman, J. Thurlow,H. Parajuli, S. Suliman, K.K. Keerthi, M. PartridgeAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): D.E. Costea, A.H. Osman, C.P. Murillo, H. Parajuli,S. Suliman, K.K. Keerthi, M. PartridgeAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): D.E. Costea, A.H. Osman, J. Thurlow, G. Kalna,X. Huang, C.P. Murillo, H. Parajuli, S. Suliman, M. Partridge
Subtypes of Oral Carcinoma-Associated Fibroblasts
www.aacrjournals.org Cancer Res; 73(13) July 1, 2013 3899
Writing, review, and/or revision of the manuscript: D.E. Costea, A.H.Osman, J. Thurlow, G. Kalna, H. Parajuli, S. Suliman, A.C. Johannessen,M. PartridgeAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): D.E. Costea, A. Hills, J. Thurlow,X. Huang, K.K. Keerthi, A.C. Johannessen, M. PartridgeStudy supervision: D.E. Costea, A.C. Johannessen, M. Partridge
AcknowledgmentsThe authors thank Dr. Hege A. Dale [Molecular Imaging Center, Functional
Genomics (FUGE) Program, The Research Council of Norway, University ofBergen, Bergen, Norway] for assistance withTEM, SEMand confocalmicroscopy,and quantification of time-lapse microscopy and to E.M. McCormack (Institutefor Internal Medicine, University of Bergen) and Prof. K. Mustafa (Institute
for Clinical Odontology, University of Bergen) for assistance with the animalexperiments.
Grant SupportThis study was cofunded by King's Medical Research Trust (to M. Partridge),
BergenMedical Research Foundation (to D.E. Costea; grant no. 20/2009) and TheNorwegian Cancer Research Association (to D.E. Costea; grant no. 515970/2011).
The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received November 5, 2012; revised March 15, 2013; accepted March 31, 2013;published OnlineFirst April 18, 2013.
References1. Lippman SM, Hong WK. Molecular markers of the risk of oral cancer.
N Engl J Med 2001;344:1323–6.2. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer 2006;
Stromal features are predictive of disease mortality in oral cancerpatients. J Pathol 2011;223:470–81.
4. Tchou J, Kossenkov AV, Chang L, Satija C, Herlyn M, Showe LC, et al.Human breast cancer associated fibroblasts exhibit subtype specificgene expression profiles. BMC Med Genomics 2012;5:39.
5. Lim KP, Cirillo N, Hassona Y, Wei W, Thurlow JK, Cheong SC, et al.Fibroblast gene expression profile reflects the stage of tumourprogression in oral squamous cell carcinoma. J Pathol 2011;223:459–69.
6. Orimo A, Gupta PB, Sgroi DC, Arenzana-Seisdedos F, Delaunay T,Naeem R, et al. Stromal fibroblasts present in invasive human breastcarcinomas promote tumor growth and angiogenesis through elevat-ed SDF-1/CXCL12 secretion. Cell 2005;121:335–48.
7. Gaballah K, Costea DE, Hills A, Gollin SM, Harrison P, Partridge M.Tissue engineering of oral dysplasia. J Pathol 2008;215:280–9.
8. De Wever O, Mareel M. Role of tissue stroma in cancer cell invasion.J Pathol 2003;200:429–47.
9. Sobral LM, Bufalino A, Lopes MA, Graner E, Salo T, Coletta RD.Myofibroblasts in the stroma of oral cancer promote tumorigenesisvia secretion of activin A. Oral Oncol 2011;47:840–6.
10. Sugimoto H, Mundel TM, Kieran MW, Kalluri R. Identification offibroblast heterogeneity in the tumor microenvironment. Cancer BiolTher 2006;5:1640–6.
11. BauerM, SuG, Casper C, HeR, RehrauerW, Friedl A. Heterogeneity ofgene expression in stromal fibroblasts of human breast carcinomasand normal breast. Oncogene 2010;29:1732–40.
12. Lewis MP, Lygoe KA, Nystrom ML, Anderson WP, Speight PM,Marshall JF, et al. Tumour-derived TGF-beta1 modulates myofibro-blast differentiation and promotes HGF/SF-dependent invasion ofsquamous carcinoma cells. Br J Cancer 2004;90:822–32.
13. Davalos AR, Coppe JP, Campisi J, Desprez PY. Senescent cells as asource of inflammatory factors for tumor progression. Cancer Metas-tasis Rev 2010;29:273–83.
14. Rosenthal E, McCrory A, Talbert M, Young G, Murphy-Ullrich J,Gladson C. Elevated expression of TGF-beta1 in head andneck cancer-associated fibroblasts. Mol Carcinog 2004;40:116–21.
15. Kiskowski MA, Jackson RS II, Banerjee J, Li X, Kang M, Iturregui JM,et al. Role for stromal heterogeneity in prostate tumorigenesis. CancerRes 2011;71:3459–70.
16. Prime SS, Davies M, Pring M, Paterson IC. The role of TGF-beta inepithelial malignancy and its relevance to the pathogenesis of oralcancer (part II). Crit Rev Oral Biol Med 2004;15:337–47.
17. Ramos DM, Dang D, Sadler S. The role of the integrin alpha v beta6 inregulating the epithelial to mesenchymal transition in oral cancer.Anticancer Res 2009;29:125–30.
18. Harper LJ, Costea DE, Gammon L, Fazil B, Biddle A, Mackenzie IC.Normal andmalignant epithelial cells with stem-like properties have an
extended G2 cell cycle phase that is associated with apoptotic resis-tance. BMC Cancer 2010;10:166.
19. Costea DE, Kulasekara K, Neppelberg E, Johannessen AC, VintermyrOK. Species-specific fibroblasts required for triggering invasivenessof partially transformed oral keratinocytes. Am J Pathol 2006;168:1889–97.
20. Breitling R, Armengaud P, Amtmann A, Herzyk P. Rank products: asimple, yet powerful, new method to detect differentially regulatedgenes in replicated microarray experiments. FEBS Lett 2004;573:83–92.
21. Dennis G Jr, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, et al.DAVID: database for annotation, visualization, and integrated discov-ery. Genome Biol 2003;4:P3.
22. Thurlow JK, Pena Murillo CL, Hunter KD, Buffa FM, Patiar S, Betts G,et al. Spectral clustering of microarray data elucidates the roles ofmicroenvironment remodeling and immune responses in survival ofhead and neck squamous cell carcinoma. J Clin Oncol 2010;28:2881–8.
23. Chang SE, Foster S, Betts D,MarnockWE. DOK, a cell line establishedfrom human dysplastic oral mucosa, shows a partially transformednon-malignant phenotype. Int J Cancer 1992;52:896–902.
24. Mackenzie IC. Growth of malignant oral epithelial stem cells afterseeding into organotypical cultures of normal mucosa. J Oral PatholMed 2004;33:71–8.
25. Harper LJ, Piper K, Common J, Fortune F, Mackenzie IC. Stem cellpatterns in cell lines derived from head and neck squamous cellcarcinoma. J Oral Pathol Med 2007;36:594–603.
26. Rheinwald JG, Beckett MA. Tumorigenic keratinocyte lines requiringanchorage and fibroblast support cultures from human squamous cellcarcinomas. Cancer Res 1981;41:1657–63.
27. Sapkota D, Costea DE, Blo M, Bruland O, Lorens JB, Vasstrand EN,et al. S100A14 inhibits proliferation of oral carcinoma derived cellsthrough G1-arrest. Oral Oncol 2012;48:219–25.
28. McBryan J,Martin FT. A database of TGF-B responsive genes. Dublin,Ireland: University College Dublin; 2007.
29. Nakajima M, Kizawa H, Saitoh M, Kou I, Miyazono K, Ikegawa S.Mechanisms for asporin function and regulation in articular cartilage.J Biol Chem 2007;282:32185–92.
30. Lopez-Casillas F, Payne HM, Andres JL, Massague J. Betaglycan canact as a dual modulator of TGF-beta access to signaling receptors:mapping of ligand binding and GAG attachment sites. J Cell Biol1994;124:557–68.
31. Hildebrand A, Romaris M, Rasmussen LM, Heinegard D, Twardzik DR,Border WA, et al. Interaction of the small interstitial proteoglycansbiglycan, decorin and fibromodulin with transforming growth factorbeta. Biochem J 1994;302(Pt 2):527–34.
32. Watanabe Y, Itoh S, Goto T, Ohnishi E, Inamitsu M, Itoh F, et al.TMEPAI, a transmembrane TGF-beta-inducible protein, sequestersSmadproteins fromactiveparticipation in TGF-beta signaling.MolCell2010;37:123–34.
33. Li J, Cao J, Li M, Yu Y, Yang Y, Xiao X, et al. Collagen triple helix repeatcontaining-1 inhibits transforming growth factor-b1-induced collagentype I expression in keloid. Br J Dermatol 2011;164:1030–6.
Costea et al.
Cancer Res; 73(13) July 1, 2013 Cancer Research3900
34. Cisternas FA, Vincent JB, Scherer SW, Ray PN. Cloning and charac-terization of human CADPS and CADPS2, newmembers of the Ca2þ-dependent activator for secretion protein family. Genomics 2003;81:279–91.
35. Ostrowski M, Carmo NB, Krumeich S, Fanget I, Raposo G, Savina A,et al. Rab27a and Rab27b control different steps of the exosomesecretion pathway. Nat Cell Biol 2010;12:19–30.
36. Kolset SO, Tveit H. Serglycin—structure and biology. Cell Mol Life Sci2008;65:1073–85.
37. Bryne M, Koppang HS, Lilleng R, Kjaerheim A. Malignancy grading ofthe deep invasive margins of oral squamous cell carcinomas has highprognostic value. J Pathol 1992;166:375–81.
38. Gaggioli C, Hooper S, Hidalgo-Carcedo C, Grosse R, Marshall JF,HarringtonK, et al. Fibroblast-led collective invasionof carcinomacellswith differing roles for RhoGTPases in leading and following cells. NatCell Biol 2007;9:1392–400.
39. Navab R, Strumpf D, Bandarchi B, Zhu CQ, Pintilie M, Ramnarine VR,et al. Prognostic gene-expression signature of carcinoma-associatedfibroblasts in non–small cell lung cancer. Proc Natl Acad Sci U S A2011;108:7160–5.
40. Zhu CQ, Popova SN, Brown ER, Barsyte-Lovejoy D, Navab R, Shih W,et al. Integrin alpha 11 regulates IGF2 expression in fibroblasts toenhance tumorigenicity of human non–small-cell lung cancer cells.Proc Natl Acad Sci U S A 2007;104:11754–9.
41. Masur SK, Dewal HS, Dinh TT, Erenburg I, Petridou S. Myofibroblastsdifferentiate from fibroblasts when plated at low density. Proc NatlAcad Sci U S A 1996;93:4219–23.
42. Bhowmick NA, Neilson EG, Moses HL. Stromal fibroblasts in cancerinitiation and progression. Nature 2004;432:332–7.
43. Mou S, Wang Q, Shi B, Gu L, Ni Z. Hepatocyte growth factor sup-presses transforming growth factor-beta-1 and type III collagen inhuman primary renal fibroblasts. Kaohsiung J Med Sci 2009;25:577–87.
44. MengW, Xia Q, Wu L, Chen S, He X, Zhang L, et al. Downregulation ofTGF-beta receptor types II and III in oral squamous cell carcinoma andoral carcinoma-associated fibroblasts. BMC Cancer 2011;11:88.
45. Bornstein S, White R, Malkoski S, Oka M, Han G, Cleaver T, et al.Smad4 loss in mice causes spontaneous head and neck cancer withincreased genomic instability and inflammation. J Clin Invest2009;119:3408–19.
46. Kizawa H, Kou I, Iida A, Sudo A, Miyamoto Y, Fukuda A, et al. Anaspartic acid repeat polymorphism in asporin inhibits chondrogenesisand increases susceptibility to osteoarthritis. Nat Genet 2005;37:138–44.
47. LeClair RJ, Durmus T,WangQ, Pyagay P, Terzic A, Lindner V.Cthrc1 isa novel inhibitor of transforming growth factor-beta signaling andneointimal lesion formation. Circ Res 2007;100:826–33.
48. Yu C, Liu Y, Huang D, Dai Y, Cai G, Sun J, et al. TGF-beta1 mediatesepithelial to mesenchymal transition via the TGF-beta/Smad pathwayin squamous cell carcinoma of the head and neck. Oncol Rep2011;25:1581–7.
49. Wang C, Tammi M, Guo H, Tammi R. Hyaluronan distribution in thenormal epithelium of esophagus, stomach, and colon and their can-cers. Am J Pathol 1996;148:1861–9.
Subtypes of Oral Carcinoma-Associated Fibroblasts
www.aacrjournals.org Cancer Res; 73(13) July 1, 2013 3901
2013;73:3888-3901. Published OnlineFirst April 18, 2013.Cancer Res Daniela Elena Costea, Allison Hills, Amani H. Osman, et al. Squamous Cell CarcinomaSubtypes with Differential Tumor-Promoting Abilities in Oral Identification of Two Distinct Carcinoma-Associated Fibroblast
Updated version
10.1158/0008-5472.CAN-12-4150doi:
Access the most recent version of this article at:
