• Nenhum resultado encontrado

Resultados MT1-MMP 123

N/A
N/A
Protected

Academic year: 2021

Share "Resultados MT1-MMP 123"

Copied!
49
0
0

Texto

(1)

               

(2)

124                   

Figura  37:  Co‐localização  de  decorim  e  MT1‐MMP  em  cortes  de  tecidos  comprometidos e não comprometidos por câncer de próstata. 

Amostras  de  tecido  de  regiões  comprometidas  ou  não  comprometidas  por  adenocarcinoma de próstata foram obtidas após a excisão da próstata de 4 pacientes  (ACM,  JSM,  LCE  e  CVO).  Os  tecidos  foram  fixados  com  paraformaldeído  4%  em  PBS,  desidratados com sacarose 30% em PBS e emblocados em meio de condicionamento  (TissueTek). Cortes de 4 m foram coletados em laminas silanizadas e analizados por  imunofluorescência  com  os  anticorpos  anti‐decorim  revelado  com  um  secundário  conjugado  com  Alexa  Fluor  594  (vermelho),  e  anti‐MT1‐MMP  revelado  com  um  secundário  conjugado  com  Alexa  Fluor  488  (verde).  O  núcleo  celular  foi  corado  por  DAPI  (azul)  e  a  co‐localização  dos  anticorpos  fornece  coloração  amarela.  Barra  de  escala: 20 m.                   

(3)

  Em 2 pacientes (JSM e CVO) houve aumento significante na expressão de MMP‐ 1 no estroma adjacente ao tecido tumoral nos tecidos comprometidos.  Curiosamente,  esse aumento segue o aumento de decorim, sendo as 2 moléculas co‐localizadas nas  células epiteliais do lúmen glandular (Figura 37).  MT1‐MMP é localizada e expressa em  células  que  circundam  as  glândulas  cancerosas.  Esta  enzima  está  provavelmente  envolvida na invasão do tumor, relacionando‐se a um estado mais avançado do tumor,  nos 2 tecidos analisados.  Assim, como o aumento na expressão de  decorim segue o  aumento  na  expressão  da  MMP‐1,  o  decorin  poderia  indicar  tumores  em  estado  avançado. 

Os  2  pacientes  (JSM  e  CVO)  que  apresentaram  um  aumento  significante  de  MMP‐1 e decorim também apresentaram aumento significante na expressão de TGFβ,  observado  tanto  por  imunofluorescência  quanto  por  qPCR  (Figura  38).  Estes  dados  corroboram  com  a  possibilidade  destes  2  pacientes  apresentarem  estágios  mais  avançados de câncer de próstata. O aumento da expressão de TGFβ já foi descrito em  câncer  de  próstata.  Os  demais  pacientes  apresentaram  aumento  de  TGFβ,  menos  expressivo:  o  tecido  comprometido  apresentou  um  aumento  de  expressão  1‐2  vezes  maior  quando  comparado  ao  tecido  não  comprometido  do  mesmo  paciente  (ACM  e  LCE).  Nos  2  pacientes  (JSM  e  CVO)  que  apresentaram  uma  grande  aumento  na  expressão  decorim,  MT1‐MMP  e  TGFβ  no  tecido  comprometido,  foi  observado  também  uma  maior  expressão  de  TGFβ  em  seus  controles  não  comprometidos,  quando comparado aos tecidos não comprometido dos demais pacientes (Figura 38).   Assim,  pode‐se  inferir  que,  num  estado  mais  avançado  de  câncer  de  próstata,  existe  um aumento significante de TGF‐β por todo o tecido prostático. 

  A expressão de biglicam não apresentou um padrão de aumento ou diminuição  entre  os  tumores,  quando  comparado  ao  tecido  não  comprometido.    Em  apenas  um  dos pacientes, que aparentemente apresenta câncer de próstata mais agressivo (JSM),  observou‐se  aumento  significante  na  expressão  de  biglicam  localizado  nas  células  epiteliais do lúmen glandular (Figura 38). O biglicam apresentou‐se co‐localizado com o  TGFβ  nas  células  epiteliais  do  lúmen  do  tecido.    Sabe‐se  que  os  membros  da  família 

(4)

126  dos SRLPs, como por exemplo, o biglicam, se ligam ao TGFβ, podendo apresentá‐lo ao  seu  receptor.    Este  mesmo  paciente  apresentou  um  maior  aumento  relativo  de  expressão de fibromodulina quando comparado aos demais pacientes (Figura 26). 

(5)
(6)

128                    Figura 38: Co‐localização de TGF‐ e biglicam em cortes de tecidos comprometidos e  não comprometidos por câncer de próstata. 

Amostras  de  tecido  de  regiões  comprometidas  ou  não  comprometidas  por  adenocarcinoma de próstata foram obtidas após a excisão da próstata de 4 pacientes  (ACM,  JSM,  LCE  e  CVO).  Os  tecidos  foram  fixados  com  paraformaldeído  4%  em  PBS,  desidratados com sacarose 30% em PBS e emblocados em meio de condicionamento  (TissueTek). Cortes de 4 m foram coletados em laminas silanizadas e analizados por  imunofluorescência  com  os  anticorpos  anti‐TGF‐  revelado  com  um  secundário  conjugado  com  Alexa  Fluor  594  (vermelho),  e  anti‐biglicam  revelado  com  um  secundário  conjugado  com  Alexa  Fluor  488  (verde).  O  núcleo  celular  foi  corado  por  DAPI  (azul)  e  a  co‐localização  dos  anticorpos  fornece  coloração  amarela.  Barra  de  escala: 20 m.                 

(7)

 

Lumicam,  decorim,  fibromodulina,  TGF‐β  e  MMP‐1  foram  detectados  no  citoplasma  de  células  circundantes  do  lúmen  das  glândulas  secretoras  de  tecido  prostático  normal.  Entretanto,  não  foi  possível  estabelecer  um  padrão  comum  de  expressão  destas  moléculas  pelas  células  secretoras  do  lúmen  das  glândulas  nos  tecidos  comprometidos  ou  não  comprometidos  por  adenocarcinoma  de  próstata.  A  fibromodulina, o decorim e o TGFβ foram imunodetectados no citoplasma das células  secretoras  do  lúmen  em  todos  os  tecidos  prostáticos  analisados,  existindo  uma  variabilidade  na  quantidade  expressa  entre  as  amostras  comprometidas  e  não  comprometidas.  O  lumican  foi  imunodetectado  no  citoplasma  das  células  secretoras  do  lúmen  do  tecido  prostático  normal,  ou  seja,  não  comprometido,  no  entanto  também foi encontrado no tecido comprometido de um paciente (JSM).  

As  linhagens  celulares  de  câncer  de  próstata,  PC3  e  DU  145  não  expressam  lumicam  e  expressam  baixas  quantidades  de  fibromodulina.    Comparando  os  dados  obtidos  dos  tecidos  prostáticos  com  aqueles  obtidos  das  linhagens  estabelecidas  de  câncer  de  próstata,  é  possível  que  o  aumento  detectado  por  qPCR  obtido  do  tecido  prostático seja na realidade reflexo de aumento na expressão de lumicam, sintetizada  por outros tipos celulares e  não  pelas  células tumorais. A exemplo do que ocorre no  câncer de mama, lumicam é secretado por fibroblastos adjacentes ao tumor, mas não  pelas  células  tumorais  (Leygue,  2000;  Leygue,  1998).  Isso  correlacionaria  muito  bem  com a localização estromal do lumicam, obtido por imunohistoquímica, principalmente  circundando  as  células  epiteliais  prostáticas,  sugerindo  certo  grau  de  fibrose,  composto provavelmente de colágeno e lumicam, e sintetizados pelos fibroblastos. 

A  reação  desmoplástica  foi  muito  bem  caracterizada  no  câncer  de  pâncreas  (Köninger, 2004). Consiste de uma forte reação fibrótica ao redor do tecido tumoral, e  supostamente,  uma  reação  do  hospedeiro  ao  crescimento  tumoral.    Os  miofibroblastos são a principal fonte para a produção excessiva deste tecido fibrótico,  que  consiste  principalmente  de  colágeno  tipos  I  e  IV,  além  de  uma  variedade  de  proteínas da MEC. 

(8)

130  Os  SLRPs  tem  sido  identificados  como  moduladores  da  fibrose  e  crescimento  tumoral e os nossos dados sugerem também para o lumicam um papel na reação do  hospedeiro à invasão tumoral.  

 

 

(9)

4.4 Contato célula-célula entre células de câncer coloretal e fibroblastos induz acúmulo de matriz extracelular afetando a invasão da célula tumoral

(10)

132  RESUMO

Miofibroblastos são fibroblastos ativados que se caracterizam pela expressão de actina de célula muscular lisa (actina α) e são encontrados circundando tumores possuindo papel crucial na reação desmoplástica. Esses miofibroblastos sintetizam colágenos e outras proteínas constituintes da matriz extracelular (MEC), como proteoglicanos. O proteoglicanos pequenos ricos em leucina (SRLPs), como o lumicam, decorim, biglicam e fibromodulina foram identificados circundando tumores de pancreáticos e de mama. A presença de SLRPs em tecidos fibrosos circundando células de câncer, e sua capacidade de suprimir sua proliferação é atualmente discutida. Essa acumulação de matriz extracelular circundando tumores afeta diretamente o crescimento e invasão das células de câncer, tornando a compreensão dos processos envolvidos na formação do tecido fibrótico de vital importância. Foi realizado o estudo das interações entre células de câncer coloretal (Caco-2 e HCT116) e fibroblastos pelo sistema de co-cultura de células. Foi demonstrado que o contato direto célula-célula entre fibroblastos e células de câncer coloretal é necessário para a indução da expressão de componentes de matriz (colágeno I, colágeno III, colágeno IV, colágeno V, biglicam e fibromodulina) pelos fibroblastos. Ainda, a exposição de fibroblastos a fatores solúveis produzidos por células de câncer coloretal diminuem a expressão de componentes de MEC. Os miofibroblastos circulantes também expressam proteases, incluindo as metaloproteinases de matriz (MMPs). A produção desses componentes resulta no remodelamento da MEC, podendo estimular o crescimento e migração das células de câncer. Ainda, os miofibroblastos induzem a expressão de MMPs quando expostos a fibroblastos. A importância do contado direto célula-célula entre as células de câncer coloretal e fibroblastos no acúmulo de MEC que circunda o tumor foi demonstrado.

(11)

Colorectal cancer desmoplastic reaction affects tumor cell invasion

Vivien J. Coulson-Thomas*, Yvette M. Coulson-Thomas*, Tarsis F. Gesteira*, Ana M. Mader, JaquesWaisberg, Maria A. Pinhal*#, Andreas Friedl§, Helena B.

Nader*, LenyToma*

*Department of Biochemistry, Universidade Federal de São Paulo, São Paulo, SP,

Brazil; †Department of Pathology, Faculdade de medicina ABC, Santo Andre, SP,

Brazil; ‡Department of Gastrosurgery, Faculdade de Medicina ABC, Santo Andre, SP,

Brazil; #Department of Biochemistry, Faculdade de Medicina ABC, Santo Andre, SP,

Brazil; §Department of Pathology and Laboratory Medicine, University of

Wisconsin-Madison, Madison WI, 53792, USA

Abbreviations: (sm), smooth muscle; (ECM), extracellular matrix; (SRLP), small leucine-rich proteoglycans, stromal reaction.

Key words: myofibroblasts, stromal reaction, proteoglycans, collagen, extracellular matrix, 2-D and 3D-cultures

Running Title: Desmoplasia affects colorectal tumor cell invasion

Address all correspondence to: Dr.Vivien J. Coulson-Thomas, Departamento de Bioquímica, Escola Paulista de Medicina, Universidade Federal de São Paulo, Rua 3 de Maio, 100 - CEP 04044-020, São Paulo, SP, Brazil. Tel: +55(11)5579-3175; FAX: +55(11)5573-6407; E-Mail: coulson.thomas@unifesp.br

(12)

134  Abstract

During cancer cell growth many tumors exhibit various grades of desmoplasia, unorganized production of fibrous or connective tissue, composed mainly of collagen fibers and myofibroblasts. The accumulation of extracellular matrix (ECM) surrounding tumors directly affects cancer cell proliferation, migration and spread, therefore the study of desmoplasia is of vital importance. Myofibroblasts synthesize an amalgam of products including collagens and other ECM proteins, such as proteoglycans and are activated during a desmoplastic reaction. Small leucine rich proteoglycans have been characterized surrounding breast and pancreatic tumors and have the ability to suppress cell proliferation. In this study we have analyzed desmoplasia co-cultivating colorectal cancer cells (Caco-2 and HCT116) and myofibroblasts using various co-culture systems. Our findings demonstrate that direct cell-cell contact between myofibroblasts and colorectal cancer cells evokes an up-regulation of the expression of ECM components (collagen I, collagen III, collagen IV, collagen V, biglycan and fibromodulin) by myofibroblasts. The ECM accumulation produced when myofibroblasts are co-cultivated with colorectal cancer cells appears unorganized and in bundles. This ECM accumulation slowed the migration and invasion of the colorectal tumor cells in both monolayer and 3-D co-culture systems. The participation of the ECM components analyzed in this study in desmoplasia is also demonstrated in vivo in human colorectal carcinoma tissue, validating our in vitro system.

(13)

Introduction

During cancer cell growth many tumors exhibit various grades of desmoplasia, unorganized production of fibrous or connective tissue, composed mainly of collagen fibers and fibroblasts, which affects cancer cell proliferation, migration and spread [1]. Desmoplastic reactions have been associated with invasive adenocarcinomas of breast and ovaries, gastrointestinal tract, and in the lung excluding small cell lung cancers, and also frequently described in squamous cell carcinomas and bilio-pancreatic carcinomas, although the prognostic significance of desmoplasia is still debated [1].The fibrotic deposition is produced mainly by stromal cells, principally fibroblasts [1, 2]. Myofibroblasts, activated fibroblasts characterized by the expression of smooth muscle (sm) α-actin, are found surrounding tumors and play an intricate role in this desmoplastic reaction. These myofibroblasts synthesize collagens, such as, collagens I, III and IV and proteoglycans which constitute the bulk of the desmoplastic reaction [3-6]. Myofibroblasts in the reactive stroma surrounding human colon adenomas and carcinomas synthesize extracellular matrix (ECM) components such as fibronectin and tenascin [7]. Fibronectin, tenascin, and chondroitin sulfate proteoglycans have been documented as the major components of colorectal adenoma and carcinoma stroma, whereas laminin, collagen IV, and heparan sulfate proteoglycans are down-regulated [8]. The expression of small leucine-rich proteoglycans (SRLPs), such as lumican, decorin, biglycan and fibromodulin, has been demonstrated surrounding breast, pancreatic and colon tumors [9-13]. Decorin and fibromodulin bind collagen types I and II [14, 15] and decorin has been shown to be essential for proper collagen fibrillogenesis [16, 17]. Heterologous expression of decorin has been shown to suppress tumor formation in vivo and retard the growth of colon cancer cells in vitro [18] through

(14)

136  modulating growth factor activity and ECM assembly. In accordance, decorin-deficient mice spontaneously develop intestinal tumors. Decorin expression is down-regulated in fibroblast-like cells within or immediately adjacent to human breast tumors [10]. Lumican expression, on the other hand, is up-regulated in fibroblast-like cells within the stroma surrounding human breast tumors compared to adjacent normal stroma [9] and has been associated with poor patient prognosis in advanced colorectal cancer [13]. Biglycan is over-expressed in the ECM of pancreatic tumors compared to normal pancreatic tissue and has been shown to inhibit the growth of pancreatic tumor cells [19].

Carcinoma cells have the capacity to induce normal fibroblasts to the reactive myofibroblast phenotype [20, 21]. In vitro studies have shown that transforming growth factor beta 1 (TGFβ1) stimulates the phenotypic switching of fibroblasts to myofibroblasts and regulates the expression of ECM components [22]. Therefore, myofibroblasts play a key role in cancer progression.

Emphasis in cancer research is given to the study of oncogenic events in the tumor cells, however, increasing evidence sheds light on the important role stromal cells play in tumor development. In vitro studies have shown that colon carcinoma cells produce soluble factors that stimulate the synthesis and secretion of proteoglycans by normal colon fibroblasts [23]. Many studies have focused on features of the stromal-cancer interactions leading to desmoplasia in pancreatic cancer; however, events involved in colorectal cancer desmoplastic reaction remain to be elucidated. Moreover, the physiological implications of this desmoplastic reaction remain to be fully established. In this study we established a desmoplastic reaction in 2-D co-culture systems between

(15)

colorectal cancer cell lines and fibroblasts using bi-layer and O-ring co-cultures. We examined components of the ECM, such as collagens (collagen types I, III, IV and V) and SLRPs (decorin, biglycan, lumican and fibromodulin) in order to characterize the desmoplastic reaction. The presence of these ECM components in the in vitro desmoplastic reaction is also demonstrated in vivo in colorectal carcinoma, thereby, validating our systems. Thereafter, we established and characterized a desmoplastic reaction using 3-D co-culture systems. We further elucidated whether the accumulation of ECM restricts or induces the migration of colorectal cancer cells through the surrounding stroma in both 2- and 3-D co-culture systems.

(16)

138  Materials and Methods

Cell culture

Cancer cell lines isolated from primary colorectal tumors, Caco-2 and HCT 166, were purchased from ATCC (HTB-37 and CCL-247, respectively) and maintained in DMEM medium (GibcoBRL, Life Technologies Inc., Grand Island, NY)/10% fetal bovine serum (FBS; Cultilab, Campinas, Brazil), containing 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin (all from Invitrogen, Carlsbad, CA), or RPMI medium (GibcoBRL)/10% FBS, containing 2 mM L-glutamine, 100 units/ml penicillin and 100 g/ml streptomycin, respectively, at 37C in a 5% CO2 humidified environment. Human primary cultures of fibroblasts isolated from amniotic fluid (WPF5) were kindly donated by Prof. Walter Pinto Jr. and used between passages 3 and 6. Fibroblasts often transform to a myofibroblastic-like state when placed in culture, which was characterized by the presence and organization of sm actin (results not shown). The cells were maintained in DMEM medium/10% FBS, containing 2mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin, at 37C in a 5% CO2 humidified environment. All experiments were previously approved by the University Ethics Committee (0038-08) according to national and international guidelines. Informed consent was obtained from all women prior to collection of the amniotic fluid as recommended by national guidelines and the Declaration of Helsinki.

Heterotypic co-culture bilayer system

Fibroblasts or colorectal cancer cells were removed from P100 culture dishes (Corning) with trypsin/EDTA (GibcoBRL), washed in EBSS, and then seeded at a density of 3 x 106 cells per well (6 well polystyrene culture dishes, Corning, Corning

(17)

Incorporated, Corning, NY) in DMEM or RPMI containing 10% FBS, L-glutamine/penicillin/streptomycin. The cells were maintained in culture for 24 h at 37C in a 5% CO2 humidified environment, and then either fibroblasts or colorectal cancer cells were seeded on top, at a cell density of 6 x 106 cells per well and maintained in culture for a further 24 h at 37C in a 5% CO2 humidified environment, and then total RNA extracted (see below). Experimental groups consisted of seeding fibroblasts underneath and colorectal cancer cells on top at different ratios (CACO2/WPF5, HCT116/WPF5) and vice versa (WPF5/CACO2, WPF5/HCT116). Control groups consisted of seeding the same cell type underneath and on top (WPF5/WPF5, CACO2/CACO2, HCT116/HCT116).

Bi-compartmental co-culture system

Culture plate inserts pore size 0.4 µm (30 mm, Millicell®-PCF, Millipore Corp., Bedford, MA) were placed one per well (6 well polystyrene culture dishes) in RPMI or DMEM containing 10% FBS, L-glutamine/penicillin/streptomycin, and left for 20 min in a sterile atmosphere (flow hood). Fibroblasts or colorectal cancer cells were removed from P100 culture dishes with trypsin/EDTA (GibcoBRL), washed in EBSS, and then seeded at a density of 3 x 106 cells per well in the bottom compartment of the insert co-culture system. The cells were maintained in co-culture for 24 h at 37C in a 5% CO2 humidified environment. Fibroblasts or colorectal cancer cells were then seeded in the top compartment of the insert co-culture system at a density of 3 x 106 cells and incubated for a further 24 h (37C, 5% CO2). Total RNA was extracted from the bottom compartment (see below). Experimental groups consisted of seeding fibroblasts in the bottom compartment and colorectal cancer cells in the top compartment

(18)

140  (CACO2/WPF5, HCT116/ WPF5) and vice versa (WPF5/CACO2, WPF5/HCT116). Control groups consisted of seeding the same cell type in the top and bottom compartments of the insert co-culture system (WPF5/WPF5, CACO2/CACO2, HCT116/HCT116).

RNA extraction and real-time reverse transcription-PCR analysis

Total RNA was isolated from fibroblasts and colorectal cancer cells after a co-culture of 24 h using Trizol® Reagent (Invitrogen, Carlsbad, CA). The concentration and purity of the RNA in each sample was determined using a spectrophotometer at 260 and 280 nm. First strand cDNA was reverse transcribed using 1 g of total RNA and the

kit Improm IITM Reverse Transcriptase (Promega, Madison, WI), according to the manufacturer’s protocol. Real time PCR amplification was performed on 2l of the cDNA (1:5) with specific primers for collagen types I, III, IV and V, fibromodulin, lumican, decorin, biglycan and the kit SYBR Green Master Mix (Applied Biosystems, Foster City, CA) in a 7500 Real-Time PCR System (Applied Biosystems, Warrington, UK), using the activation cycle of 95C for 10 min, 40 cycles of 95C for 15 sec (denaturation stage), 61C for 1 min (annealing and elongation stage) and 72C for 30 sec. The specificity of the amplified products was analyzed through dissociation curves generated by the equipment yielding single peaks. Negative controls were used in parallel to confirm the absence of any form of contamination in the reaction. Analysis of the data was carried out using the 2-∆∆Ct method [24] using the 7500 Real-Time PCR System’s software. The primer combination used for collagen I was forward:

5´AGTGTGGCCCAGAAGAACTGGTACAT3´ and reverse: 3´TCGAACTGGAATCCATCGGTCATGCT5´; for collagen III, forward:

(19)

5’TGCATACATGGATCAGGCCAGTGGAA3’ and reverse: 3’TTCGTGCAACCATCCTCCAGAACTGT5’; for collagen IV, forward:

5’TGGAGGAGTTTAGAAGTGCGCCAT3’ and reverse: 3’AGGCTTCTTGAACATCTCGCTCCT5’; for collagen V, forward:

5’TGCTCCAGGGATTCCTTCAAGGTT3’ and reverse: 3’ATAGGAGAGCAGTTTCCCACGCTT5’; for fibromodulin, forward:

5’TTCCCTCCCGCATGAAGTATGTGT3’ and reverse: 3’TATCACTGGTGATCTGGTTGCCGT5’; for lumican, forward:

5’TGGCATTGATTGGTGGTACCAGTG3’ and reverse: 3’TGGGTAGCTTTCAGGGCAGTTACA5’; for decorin, forward:

5’GCTGGACCGTTTCAACAGAGA3’ and reverse: 3’GGGGAAGATCCTTTGGCACT5’; for biglycan, forward:

5’TTGGACAACAACAAGTTGGCCAGG3’ and reverse: 3’TGAAGAGGCTGATGCCGTTGTAGT5’. Expression values were normalized to

the housekeeping gene ribosomal protein S29 (RPS29) using the primer combination forward: 5’CCTGGAGGAGAAGAGHAAAGAGA3’ and reverse: 3’TTGAGGACCTCTGTGTATTTGTCAA5’.

Heterotypic co-culture system using O-rings

Sterile glass cloning rings (O-rings) were placed on top of 12 mm diameter glass coverslips in 24 well polystyrene culture dishes. Fibroblasts and colorectal cancer cells were removed from P100 culture dishes with trypsin/EDTA (GibcoBR) and washed in EBSS. Fibroblasts were then seeded on the outside of the O-ring at a density of 1.5 x 104 cells per well in DMEM containing 10% FBS, L-glutamine/penicillin/streptomycin,

(20)

142  whereas colorectal cancer cells were seeded inside the O-ring at a cell density of 0.5 x 104 cells per well in RPMI containing 10% FBS, L-glutamine/ penicillin/streptomycin. The cells were maintained in culture for 48 h (37C, 5% CO2). The O-rings were then removed and the cells maintained in culture until the cells spread out into the cell-free area left by the O-ring (approximately 4 days). The cells were sequentially fixed using 4% paraformaldehyde and 1:1 methanol/acetone. The cells were stained for collagen types I, III, IV and V, fibromodulin, lumican, decorin, biglycan and -actin (see below).

Invasion assay of colorectal cancer cells through 3-D fibroblast-produced matrix

Fibroblasts were removed from P100 culture dishes (Corning) with trypsin/EDTA, washed in EBSS, and then seeded at a density of 3 x 106 cells per well (6 well polystyrene culture dishes, Corning) in DMEM or RPMI containing 10% FBS, L-glutamine/penicillin/streptomycin. The cells were maintained in culture at 37C in a 5%

CO2 humidified environment until confluent, and then either fibroblasts or colorectal cancer cells (in order to induce a desmoplastic reaction) were seeded on top, at a cell density of 1 x 105 cells per well and maintained in culture for an additional 6 days at 37C in a 5% CO2 humidified environment. After confluence the cultures were treated with 25 g/mL of ascorbic acid every other day in order to induce the production of the 3-D matrix.

Colorectal tumor cells were transfected with GFP plasmid using FuGENE®HD Transfection Reagent (Roche Applied Science, Mannheim, Germany) and maintained in culture for 24 hours and subsequently seeded into the 3-D stromagenic system. The GFP positive cells were left to migrate for 2 hours and observed using scanning confocal inverted microscope (Zeiss LSM510 Zeiss, Germany). Experimental groups

(21)

consisted of seeding colorectal tumor cells into a 3-D stromagenic system composed of fibroblasts seeded underneath colorectal cancer cells (WPF5 + CACO2, WPF5 + HCT). Control groups consisted of seeding colorectal tumor cells into a 3-D stromagenic system of solely fibroblasts (WPF5). Following the migration assay the cells were fixed with 4% paraformaldehyde and submitted to immunocytochemistry in order to verify co-localization of GFP with sm- actin and vimentin.

Immunocytochemistry and microscopy

Cells were washed 3 times in PBS (15 min each wash), and the coverslips then incubated in blocking solution (5% FBS) at room temperature for 1 h. Coverslips were then incubated with primary antibodies overnight at 4C. Primary antibodies used were: mouse monoclonal anti-collagen I (Calbiochem I-8H5, San Diego, CA, and Sigma COL1, Sigma Chemical Co., St. Louis, MI), goat anti-collagen III (Santa Cruz, Santa Cruz, CA), goat anti-collagen IV (against α1 type IV, Santa Cruz, Santa Cruz, CA), rabbit anti-collagen V (Chemicon AB763P, Chemicon International, Temecula, CA), rabbit anti-fibromodulin (Santa Cruz H-50), monoclonal mouse anti-lumican (produced at our laboratory), rabbit anti-lumican (Santa Cruz H-90), mouse anti-decorin (Seikagaku 6-B6, Seikagaku Corporation, Tokyo, Japan), rabbit anti-biglycan (Santa Cruz H-150), mouse anti-fibronectin (BD transduction laboratories), goat anti-perlecan (Santa Cruz L-20) and finally, mouse anti-smooth muscle actinconjugated with Cy3 (clone 1A4, Sigma-Aldrich, St. Louis, MI). Afterwards, the coverslips were washed 3 times in PBS and then incubated for 1 h at room temperature with appropriate fluorescent secondary antibodies conjugated to Alexa Fluor® 488 or Alexa Fluor® 594 (Molecular Probes/Invitrogen, Eugene, OR). In order to investigate the possible

(22)

co-144  localization of compounds, a second incubation with primary and secondary antibodies was performed. After incubation with the antibodies, the coverslips were washed 3 times in PBS, mounted on glass slides in Fluoromount G (2:1 in PBS, Electron Microscopy Sciences, Hatfield, PA) and sealed with nail polish. Negative control immunostaining was performed with omission of each primary antibody and did not yield specific immunostaining (not shown). Coverslips were examined using scanning confocal inverted microscope (Zeiss LSM510, Zeiss, Germany) and fluorescence quantified using LSM image browser.

Immunohistochemistry

Human colorectal adenocarcinoma tissues were obtained from 5 patients at the Hospital do Servidor Público Estadual, São Paulo, Brazil (Table 1). Normal tissues were obtained from the same patients at a distance of at least 10 cm from the tumors and histologically stated as non-neoplastic by a pathologist. Tissues were fixed for 12 h in 2 % buffered paraformaldehyde and further processed for paraffin embedding. Tissue sections 2 µm thick were cut, deparaffinized and rehydrated in PBS. Endogenous peroxidase was blocked with 3 % hydrogen peroxide in PBS for 30 min followed by blocking of unspecific protein binding sites with 5 % FBS in PBS for 30 min. Sections were then incubated with collagen I, collagen IV, collagen V, anti-biglycan and anti-fibromodulin for 16 h at 4oC in blocking solution. Sections were washed and incubated with biotin-conjugated anti-rabbit/mouse/goat antibody (Biotinilated Link Universal, DakoCytomation, Dako North America Inc., Vila RealCarpinteria, CA) for 1 h at 18oC followed by streptavidin-horseradish peroxidase, streptavidin-HRP conjugate (DakoCytomation, Dako North America Inc.). Bound HRP

(23)

was detected using diaminobenzidine-tetrahydrochloride, liquid DAB + substrate Chromogen System (Dako North America Inc.) for 15 seconds and the reaction stopped with double distilled water. The slides were counterstained with hematoxylin, mounted in Permount™ Mounting Medium (Electron Microscopy Sciences) and observed under a Nikon Eclipse E800 microscope. Negative controls were performed by omitting the primary antibody and did not yield specific signals. Tissue samples were obtained after patients consent as recommended by national guidelines and the Declaration of Helsinki and the study of human tissues was approved by the ethics committee of all institutions involved, UNIFESP (1796/08), Faculdade de Medicina do ABC (025/2008) and Hospital do ServidorPúblico (021/08).

Statistics

All values are presented as means ± standard deviation of the mean. The difference between 2 groups was compared by unpaired Mann-Whitney test. A level of p< 0.05 was considered as significant. Statistical analysis was performed with the GraphPad Prism version 5 software package (GraphPad Software, San Diego, CA).

(24)

146  Results

Gene expression of ECM components after heterotypic bilayer co-culture of colorectal cancer cells and fibroblasts

In order to elicit the steps leading to the production of fibrotic tissue surrounding many tumors, we cultured cancer cells isolated from primary colorectal carcinoma (cell lines Caco-2 and HCT 116) in direct contact with fibroblasts. We studied the expression of collagens (collagen types I, III, IV and V) and SLRPs (decorin,lumican, fibromodulin and biglycan). An increase in the expression of all the collagens analyzed as well as biglycan and fibromodulin was observed in this type of co-culture system (Figure 1). Two experimental conditions were studied: colorectal cancer cells seeded directly on top of previously seeded fibroblasts, and fibroblasts seeded directly on top of previously seeded colorectal cancer cells. Collagen I expression increased at least 2-fold and 3-fold when Caco-2 cells and HCT 116 cells were seeded on a monolayer of fibroblasts, respectively, but expression levels were not altered when fibroblasts were seeded on top of either colorectal cancer cell line (Figure 1A). Collagen III expression increased at least 5-fold in bilayer co-cultures of colorectal cancer cells plated on a monolayer of fibroblasts, independently of the colorectal cancer cell line, and approximately 2-fold in cultures where fibroblasts were cultured on a monolayer of Caco-2 cells, but no significant increases were observed when fibroblasts were cultured on a monolayer of HCT 116 cells (Figure 1B). Collagens IV and V were the ECM components with the greatest overall increase in expression when fibroblasts were cultured in cell-cell contact with colorectal cancer cells; the expression of collagen types IV and V increased over 10-fold in bilayer co-cultures of fibroblasts and colorectal cancer cells when the colorectal cancer cell was seeded on top, and approximately

(25)

5-fold when the fibroblasts were seeded on top (Figure 1C and D). Biglycan expression increased slightly and doubled when Caco-2 and HCT 116 cells were seeded on top of fibroblasts, respectively, but decreased when fibroblasts were cultured on top of colorectal cancer cells (Figure 1F). Fibromodulin expression increased at least 3-fold and almost 2-fold when Caco-2 cells and HCT 116 cells were cultured on a monolayer of fibroblasts, respectively, whereas significant increases were not observed when fibroblasts were seeded on either of the colorectal tumor cell lines (Figure 1H). The increases observed were always greater when colorectal cancer cells were seeded on top of fibroblasts compared to when fibroblasts were seeded on top of colorectal cancer cells. Slight decreases in the expression of decorin and lumican were observed in bilayer co-cultures of fibroblasts and colorectal cancer cells however these differences were not considered significant (Figure 1E and G).

With the exception of fibromodulin for both cell lines and collagen IV for Caco-2 cells, the colorectal cancer cell lines expressed no or insignificant levels of the ECM components studied here, and therefore, the expression levels observed accounted primarily for expression levels in the fibroblasts (Figure 1).

Gene expression of ECM components after bi-compartmental (transwell) co-culture of colorectal cancer cells and fibroblasts

Fibroblasts and colorectal cancer cells were cultured in a bi-compartmental (transwell) co-culture system allowing cross-talk through soluble factors between the two cell types. This allowed us to evaluate the role of soluble factors in the up-regulation of the expression of ECM components observed in the bilayer co-cultures where there is both cross-talk through soluble factors and cell-cell contact at play.

(26)

148  Distinct results were obtained when direct cell-cell contact was eliminated from the cross-talk between colorectal cancer cells and fibroblasts. When the fibroblasts were exposed solely to the soluble factors produced by the colorectal cancer cell lines Caco-2 and HCT 116, there was a general decrease in the expression of ECM components (Figure 2).

When fibroblasts were exposed to soluble factors produced by Caco-2 cells, significant reductions in the expression of collagen types I, IV and V (Figure 2), and the SLRPs analyzed (decorin, biglycan, lumican and fibromodulin; Figure 2) were observed. When fibroblasts were exposed to soluble factors produced by HCT 116 cells, significant reductions in the expression levels of collagen types I and IV (Figure 2A and C), and all four SLRPs (decorin, biglycan, lumican and fibromodulin) (Figure 2E-H) were observed. No significant increases in the expression levels of ECM components were observed, only decreases, when fibroblasts were exposed solely to the soluble factors produced by the colorectal cancer cells. Therefore, we may hypothesize that cell-cell contact or proximity between the colorectal cancer cell-cells and fibroblasts is necessary to evoke an increase in the expression of fibrotic components.

As observed for the control bilayer co-cultures (Caco-2 cells seeded on Caco-2 cells, HCT 116 cells seeded on HCT 116 cells, and fibroblasts seeded on fibroblasts), Caco-2 and HCT 116 cells expressed low levels of fibromodulin in the bi-compartmental co-culture system (Figure 2H). Caco-2 also expressed low levels of collagen IV and biglycan (Figure 2 C and F). Therefore, both colorectal tumor cell lines expressed no or insignificant levels of the other ECM components analyzed, hence, the expression levels observed accounted primarily for expression by the fibroblasts (Figure 2). Since RNA

(27)

was only extracted from the cell type in the bottom compartment of the bi-compartmental co-culture system, the decreases in expression levels observed account for expression in the cell line in this compartment.

Expression and localization of ECM components in heterotypic co-cultures of colorectal cancer cells and fibroblasts using the O-ring system

The O-ring co-culture system mimics the tumor microenvironment in vivo, composed of the tumor and surrounding stroma which consists of both ECM and stromal cells such as fibroblasts. Stromal cells are seeded around the cancer cells and cell-cell contact is observed at the fibroblast/cancer cell interface. Besides cell-cell contact, a gradient of soluble factors produced by the cancer cells is established, where fibroblasts in closer proximity to the cancer cells will be exposed to higher concentrations of these factors and fibroblasts located distantly will be exposed to lower concentrations. Cells cultured using the O-ring co-culture system were analyzed by immunocytochemistry. The results obtained corroborated the findings from the bilayer and transwell co-culture systems. An accumulation of ECM components was observed in the zone where colorectal cancer cells and fibroblasts came into direct cell-cell contact (Figure 3). An increase in collagen I and collagen V immunolabeling was observed solely in fibroblasts located in the region of contact with the colorectal cancer cells (Figure 3A-C). The excess collagen deposited appeared to be unorganized and in bundles (Figure 3A). An increase in collagen IV immunolabeling was also observed in the region of fibroblast/colorectal cancer cell interface (Figure 3G). Fibromodulin immunolabeling increased only in fibroblasts located at the fibroblast/colorectal cancer cell interface (Figure 3D-E). An increase in biglycan immunolabeling was observed in

(28)

150  fibroblasts located in the region of contact with the colorectal cancer cells but not in fibroblasts located at a distance from the colorectal tumor cells and solely exposed to the soluble factors produced by the colorectal cancer cells (Figure 3F).

Interestingly, a sharp fibroblast/cancer cell interface was established when the fibroblasts were co-cultivated with colorectal tumor cells (Figure 3H). Therefore, the increases in ECM components and accumulation of fibrotic deposits in the area of intersection between the fibroblasts and the colorectal cancer cells possibly inhibited the migration of the tumor cells through the fibroblast area of this system. In contrast, we have previously shown that metastatic prostate tumor cells induce a decrease in the ECM produced by fibroblasts, and the prostate cancer cells migrate through the fibroblast area when co-cultivated using the O-ring system [25].

The fibroblasts were characterized as myofibroblasts after immunolocalization of organized sm -actin in the fibroblasts in all experimental models (results not shown).

Invasion of colorectal tumor cells through a 3-D stroma

In order to verify whether the desmoplasia surrounding colorectal tumors has a limiting effect on colorectal tumor cell invasion we used a 3-D stromagenic system (Amatangelo et al., 2005). 3-D stroma cultures with a cellular component of solely fibroblasts were compared to stroma cultures composed of both fibroblasts and colorectal cancer cells, the latter presenting characteristics of desmoplasia (Figure 4). The 3-D stroma composed of solely fibroblasts presented an organized 3-D fibronectin and collagen I matrix, whereas when colorectal cancer cells were co-cultivated with fibroblasts, fibronectin and collagen I matrix appeared as clumps with thickened and unorganized fibronectin fibers (Figure 4B). Moreover, an increase and loss of

(29)

organization of versican and perlecan were observed when colorectal cancer cells were co-cultivated with fibroblasts using the 3D culture system. GFP-expressing colorectal tumor cells were seeded onto this 3-D stroma and left to invade for 2 hours and the system observed using a confocal microscope (Figures 5 and 6). Interestingly, the colorectal tumor cells plated onto the desmoplastic stroma were not able to penetrate the stroma, whereas, the colorectal tumor cells seeded onto the control stroma were able to penetrate the stroma. Moreover, the colorectal tumor cells seeded onto the desmoplastic stroma presented large flattened cytoplasms, typical of adhered cells, whereas those seeded onto the control stroma presented small irregular cell shapes typical of migrating cells. The system was subsequently submitted to immunocytochemistry and vimentin staining was present in GFP-positive cells in both 3-D systems (results not shown). However, organized strong -actin filaments were solely detected in GFP-positive colorectal cancer cells invading through the fibroblast 3-D matrix (results not shown).

In vivo localization of ECM components in human colorectal tumors compared to non-neoplastic tissues

Co-cultures of colorectal tumor cells with fibroblasts clearly led to a desmoplastic reaction characterized by an increase in the expression and deposition of ECM components. In vitro, an increase in the expression and deposition of collagens I, III, IV and V, as well as, biglycan and fibromodulin were observed. In order to elicit whether these events mimic those which take place in vivo the expression and localization of ECM components were analyzed by examining the immunolocalization of collagens I, IV and V, decorin and fibromodulin in colorectal carcinoma compared to normal tissues. DAB staining revealed an impressive increase in ECM accumulation

(30)

152  surrounding colorectal adenocarcinomas when compared to non-neoplastic tissues. Collagens I and IV (Figure 7), collagen V (Figure 8), biglycan and fibromodulin (Figure 9) immunolabeling increased solely in the stroma surrounding epithelial cells in tumor samples (B, D, F, H) compared to non-neoplastic samples (A, C, E, G). Strong positive immunolabeling was observed for collagens IV and V in tumor samples when compared to normal tissues. An increase in collagen IV, biglycan and fibromodulin immunolabeling was also detected in the basal membrane of neoplastic glands compared to non-neoplastic tissues. Therefore, an increase in the expression of the same ECM components immediately surrounding the colorectal cancer cells was also observed in vivo characterizing a desmoplastic reaction and reflecting the in vitro results thereby validating the relevance of our in vitro system.

Responsiveness of ECM genes to direct cell-cell contact and soluble factors

Increases in the expression levels of all of the collagen types analyzed were observed as a result of direct cell-cell contact when fibroblasts were co-cultured with colorectal cancer cells (Table 2). The expression levels of collagen types III and IV in fibroblasts were not significantly affected by soluble factors produced by colorectal cancer cells (Table 2). Decorin and lumican expression levels possibly decreased in the fibroblasts after direct cell-cell contact with colorectal cancer cells, however, a significant decrease in expression levels was observed when fibroblasts were exposed to soluble factors produced by colorectal cancer cells (Table 2). Based on the differences in gene expression observed between the two cell co-culture systems, bilayer and bi-compartmental (Figures 1 and 2), used in this study, we may organize the genes

(31)

analyzed into two groups, those responsive to direct cell-cell contact and those responsive to soluble factors (Table 2).

(32)

154  Discussion

ECM accumulation surrounding tumors directly affects the growth and invasion of cancer cells. This process is thought to be a host defense reaction (desmoplastic reaction) aiming to confine the developing tumor [26]. However, desmoplasia has been associated with tumor progression and poor prognosis of colorectal carcinoma, squamous cell carcinoma, breast cancer and scirrhous gastric cancer [27-30]. On the other hand, desmoplastic reaction has been suggested to lead to tumor regression [31]. Different tumors present varying grades of fibrosis and the bulk of this fibrotic tissue is composed of collagen fibers and fibroblasts. In view of the controversial involvement of desmoplasia in tumor progression, understanding the effect this ECM accumulation has on cancer cell migration and invasiveness is of vital importance.

The presence of SLRPs in fibrous tissue surrounding cancer cells and their ability to suppress proliferation has been well characterized [9, 10, 32, 33]. SLRP binding to collagen has been shown to increase collagen fibril stability [34, 35] as well as, protect collagen fibrils from proteolytic cleavage by various collagenases [36]. SLRPs are thought to regulate the assembly of collagen matrix in connective tissues via their bi-functional character where the protein moiety binds collagen fibrils at strategic loci and the highly charged hydrophilic glycosaminoglycans regulate interfibrillar spacing [37-40]. In co-cultures of colorectal cancer cells and fibroblasts, we observed increases in the expression of the ECM components collagen types I, III, IV and V, biglycan and fibromodulin, and possible decreases in decorin and lumican, as a result of direct cell-cell interactions. This would alter SLRP regulated collagen matrix assembly leading to disorganization of the ECM. The loss of collagen organization and control of fiber

(33)

thickness is evident in the 3D culture model, where the collagen and fibronectin fibers appear thickened and unorganized when fibroblasts where cultivated in the presence of colorectal tumor cells. SLRP and collagen up-regulation by fibroblasts could reflect a “host” response to the tumor cells, known as desmoplastic reaction. This up-regulation of ECM components was clear in both the bilayer culture system, the O-ring co-culture system and the 3D co-co-culture system, when cultivating fibroblasts and the colorectal cancer cell lines Caco-2 and HCT 116. Furthermore, our in vitro desmoplastic reaction was validated through immunohistochemistry of human colorectal carcinoma and non-neoplastic tissues. An impressive increase in collagens IV and V was observed in the desmoplastic reaction of colorectal adenocarcinoma. Collagen I, biglycan and fibromodulin were also confirmed as constituents of the fibrotic tissue surrounding colorectal tumors.

Wang and colleagues (2006) analyzed differential gene expression in co-cultures of prostate cancer cells and bone marrow stromal cells comparing the insert system and contact system, and identified three sets of genes: those that were modified only by soluble factors, those by both soluble factors and physical contact, and those only by physical contact [41]. These authors observed a unique set of genes that were regulated only by physical contact, including the genes encoding collagens III and IV, MMP-9, biglycan, and TGF1. We also observed regulation of expression levels of collagen types I, III, IV and V, biglycan and fibromodulin, as a result of direct cell-cell interactions between fibroblasts and colorectal tumor cells. However, the expression levels of collagen types I and V, biglycan and fibromodulin were also influenced by soluble factors. Interestingly,these ECM components increased in fibroblasts in response to cell-cell contact with colorectal cancer cells and decreased in fibroblasts in

(34)

156  response to soluble factors produced by colorectal cancer cells, but in the O-ring co-culture system where both cell-cell contact and soluble factors were at play, we observed an increase in these ECM components solely at the fibroblast-colorectal cancer cell interface.

Invasion of the fibrotic tissue surrounding tumors is necessary for cancer cell migration and eventual metastasis. Therefore, establishing whether the desmoplastic reaction surrounding many tumors restricts or supports the invasion of tumor cells is of vital importance; moreover, desmoplasia has been suggested as a prognostic marker and for staging of various cancers. Our data shows that colorectal tumor cells invade control stroma (produced by and consisting of normal fibroblasts) at a faster rate than desmoplastic stroma; therefore, supporting the idea that desmoplasia restricts colorectal tumor progression. Desmoplastic reaction in colorectal carcinoma has been associated with favorable prognosis [42, 43]. Moreover, basement membrane deposition has also been associated with favorable prognosis and tumor cell differentiation in colon cancer [44-46]. Cell migration is controlled by the ECM either through matrix bound adhesive molecules or by cells migrating along organized matrix fibers (contact guidance) [47]. Therefore, the unorganized matrix observed in the desmoplastic stroma, due to the up-regulation of ECM components and altered collagen matrix assembly, could have played a role in limiting the migration of the colorectal tumor cells in the invasion assay.

This study demonstrates that colorectal cancer cells Caco-2 and HCT-116 induce a desmoplastic reaction when co-cultivated in direct cell-cell contact with myofibroblasts. Fibronectin, collagens I and V were detected in this desmoplastic reaction as

(35)

up-regulated and unorganized, distributed as bundles of thickened fibers with no apparent orientation. Our in vitro desmoplasia model was characterized by the expression of collagens I, III, IV and V, as well as, fibromodulin and biglycan. These components were detected in in vivo desmoplasia presenting colorectal tissues, thereby validating our in vitro model. Interestingly, the presence of the desmoplastic reaction in in vitro stroma inhibited both the migration and invasion of colorectal tumor cells. Therefore, our data supports the idea that desmoplasia restricts colorectal tumor progression.

(36)

158  Acknowledgements

This work was supported byFundação de Amparo à Pesquisa do Estado de São Paulo, FAPESP (2007/59801-1), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq). We would like to thank Prof. Peter Reinach for his kind support throughout this study. We also acknowledge Caroline Z. Romera and Elizabeth N. Kanashiro (INFAR/UNIFESP, Brazil) for their technical assistance.

(37)

References

[1]  Kunz‐Schughart  LA,  and  Knuechel  R  (2002).  Tumor‐associated  fibroblasts  (part  I):  Active stromal participants in tumor development and progression? Histol Histopathol  17, 599‐621. 

[2]  Gress  TM,  Muller‐Pillasch  F,  Lerch  MM,  Friess  H,  Buchler  M,  and  Adler  G  (1995).  Expression and in‐situ localization of genes coding for extracellular matrix proteins and  extracellular matrix degrading proteases in pancreatic cancer. Int J Cancer 62, 407‐413.  [3]  Faouzi  S,  Le  Bail  B,  Neaud  V,  Boussarie  L,  Saric  J,  Bioulac‐Sage  P,  Balabaud  C,  and  Rosenbaum  J  (1999).  Myofibroblasts  are  responsible  for  collagen  synthesis  in  the  stroma of human hepatocellular carcinoma: an in vivo and in vitro study. J Hepatol 30,  275‐284. 

[4]  Shi  Y,  Niculescu  R,  Wang  D,  Ormont  M,  Magno  M,  San  Antonio  JD,  Williams  KJ,  and  Zalewski A (2000). Myofibroblast involvement in glycosaminoglycan synthesis and lipid  retention during coronary repair. J Vasc Res 37, 399‐407. 

[5]  Gressner  AM  (1994).  Activation  of  proteoglycan  synthesis  in  injured  liver‐‐a  brief  review of molecular and cellular aspects. Eur J Clin Chem Clin Biochem 32, 225‐237.  [6]  Apte MV, Park S, Phillips PA, Santucci N, Goldstein D, Kumar RK, Ramm GA, Buchler M, 

Friess H, McCarroll JA, et al. (2004). Desmoplastic reaction in pancreatic cancer: role of  pancreatic stellate cells. Pancreas 29, 179‐187. 

[7]  Hanamura N, Yoshida T, Matsumoto E, Kawarada Y, and Sakakura T (1997). Expression  of  fibronectin  and  tenascin‐C  mRNA  by  myofibroblasts,  vascular  cells  and  epithelial  cells in human colon adenomas and carcinomas. Int J Cancer 73, 10‐15. 

[8]  Hauptmann  S,  Zardi  L,  Siri  A,  Carnemolla  B,  Borsi  L,  Castellucci  M,  Klosterhalfen  B,  Hartung P,  Weis J, Stocker  G,  et al.  (1995).  Extracellular  matrix proteins  in  colorectal  carcinomas. Expression of tenascin and fibronectin isoforms. Lab Invest 73, 172‐182.  [9]  Leygue E, Snell L, Dotzlaw H, Hole K, Hiller‐Hitchcock T, Roughley PJ, Watson PH, and  Murphy LC (1998). Expression of lumican in human breast carcinoma. Cancer Res 58,  1348‐1352.  [10]  Leygue E, Snell L, Dotzlaw H, Troup S, Hiller‐Hitchcock T, Murphy LC, Roughley PJ, and  Watson PH (2000). Lumican and decorin are differentially expressed in human breast  carcinoma. J Pathol 192, 313‐320.  [11]  Ishiwata T, Cho K, Kawahara K, Yamamoto T, Fujiwara Y, Uchida E, Tajiri T, and Naito Z  (2007).  Role  of  lumican  in  cancer  cells  and  adjacent  stromal  tissues  in  human  pancreatic cancer. Oncol Rep 18, 537‐543. 

[12]  Augoff  K,  Rabczynski  J,  Tabola  R,  Czapla  L,  Ratajczak  K,  and  Grabowski  K  (2008).  Immunohistochemical  study  of  decorin  expression  in  polyps  and  carcinomas  of  the  colon. Med Sci Monit 14, CR530‐535. 

[13]  Seya  T,  Tanaka  N,  Shinji  S,  Yokoi  K,  Koizumi  M,  Teranishi  N,  Yamashita  K,  Tajiri  T,  Ishiwata T, and Naito Z (2006). Lumican expression in advanced colorectal cancer with  nodal metastasis correlates with poor prognosis. Oncol Rep 16, 1225‐1230.  [14]  Viola M, Bartolini B, Sonaggere M, Giudici C, Tenni R, and Tira ME (2007). Fibromodulin  interactions with type I and II collagens. Connect Tissue Res 48, 141‐148.  [15]  Svensson L, Heinegard D, and Oldberg A (1995). Decorin‐binding sites for collagen type  I are mainly located in leucine‐rich repeats 4‐5. J Biol Chem 270, 20712‐20716. 

[16]  Sampaio  Lde  O,  Bayliss  MT,  Hardingham  TE,  and  Muir  H  (1988).  Dermatan  sulphate  proteoglycan from human articular cartilage. Variation in its content with age and its 

(38)

160  structural  comparison  with  a  small  chondroitin  sulphate  proteoglycan  from  pig  laryngeal cartilage. Biochem J 254, 757‐764. 

[17]  Danielson  KG,  Baribault  H,  Holmes  DF,  Graham  H,  Kadler  KE,  and  Iozzo  RV  (1997).  Targeted disruption of decorin leads to abnormal collagen fibril morphology and skin  fragility. J Cell Biol 136, 729‐743. 

[18]  Santra  M,  Skorski  T,  Calabretta  B,  Lattime  EC,  and  Iozzo  RV  (1995).  De  novo  decorin  gene  expression  suppresses  the  malignant  phenotype  in  human  colon  cancer  cells. 

Proc Natl Acad Sci U S A 92, 7016‐7020. 

[19]  Weber CK, Sommer G, Michl P, Fensterer H, Weimer M, Gansauge F, Leder G, Adler G,  and Gress TM (2001). Biglycan is overexpressed in pancreatic cancer and induces G1‐ arrest in pancreatic cancer cell lines. Gastroenterology 121, 657‐667. 

[20]  Ronnov‐Jessen L, and Petersen OW (1993). Induction of alpha‐smooth muscle actin by  transforming  growth  factor‐beta  1  in  quiescent  human  breast  gland  fibroblasts.  Implications for myofibroblast generation in breast neoplasia. Lab Invest 68, 696‐707.  [21]  Ronnov‐Jessen L, Petersen OW, Koteliansky VE, and Bissell MJ (1995). The origin of the 

myofibroblasts  in  breast  cancer.  Recapitulation  of  tumor  environment  in  culture  unravels  diversity  and  implicates  converted  fibroblasts  and  recruited  smooth  muscle  cells. J Clin Invest 95, 859‐873. 

[22]  Qian SW, Burmester JK, Tsang ML, Weatherbee JA, Hinck AP, Ohlsen DJ, Sporn MB, and  Roberts  AB  (1996).  Binding  affinity  of  transforming  growth  factor‐beta  for  its  type  II  receptor  is  determined  by  the  C‐terminal  region  of  the  molecule.  J  Biol  Chem  271,  30656‐30662. 

[23]  Iozzo RV (1985). Neoplastic modulation of extracellular matrix. Colon carcinoma cells  release  polypeptides  that  alter  proteoglycan  metabolism  in  colon  fibroblasts.  J  Biol 

Chem 260, 7464‐7473. 

[24]  Livak  KJ,  and  Schmittgen  TD  (2001).  Analysis  of  relative  gene  expression  data  using  real‐time quantitative PCR and the 2(‐Delta Delta C(T)) Method. Methods 25, 402‐408.  [25]  Coulson‐Thomas VJ, Gesteira TF, Coulson‐Thomas YM, Vicente CM, Tersariol IL, Nader 

HB,  and  Toma  L  (2010).  Fibroblast  and  prostate  tumor  cell  cross‐talk:  Fibroblast  differentiation, TGF‐beta, and extracellular matrix down‐regulation. Exp Cell Res.  [26]  Angeli  F,  Koumakis  G,  Chen  MC,  Kumar  S,  and  Delinassios  JG  (2009).  Role  of  stromal 

fibroblasts in cancer: promoting or impeding? Tumour Biol 30, 109‐120. 

[27]  Halvorsen  TB,  and  Seim  E  (1989).  Association  between  invasiveness,  inflammatory  reaction, desmoplasia and survival in colorectal cancer. J Clin Pathol 42, 162‐166.  [28]  Breuninger  H,  Schaumburg‐Lever  G,  Holzschuh  J,  and  Horny  HP  (1997).  Desmoplastic 

squamous cell carcinoma of skin and vermilion surface: a highly malignant subtype of  skin cancer. Cancer 79, 915‐919. 

[29]  Hasebe  T,  Mukai  K,  Tsuda  H,  and  Ochiai  A  (2000).  New  prognostic  histological  parameter of invasive ductal carcinoma of the breast: clinicopathological significance  of fibrotic focus. Pathol Int 50, 263‐272. 

[30]  Yashiro  M,  Chung  YS,  Nishimura  S,  Inoue  T,  and  Sowa  M  (1996).  Fibrosis  in  the  peritoneum  induced  by  scirrhous  gastric  cancer  cells  may  act  as  "soil"  for  peritoneal  dissemination. Cancer 77, 1668‐1675. 

[31]  Martin  MS,  Caignard  A,  Hammann  A,  Pelletier  H,  and  Martin  F  (1987).  An  immunohistological  study  of  cells  infiltrating  progressive  and  regressive  tumors  induced by two variant subpopulations of a rat colon cancer cell line. Int J Cancer 40,  87‐93. 

[32]  Adany  R,  and  Iozzo  RV  (1990).  Altered  methylation  of  versican  proteoglycan  gene  in  human colon carcinoma. Biochem Biophys Res Commun 171, 1402‐1413. 

(39)

[33]  Reed CC, Gauldie J, and Iozzo RV (2002). Suppression of tumorigenicity by adenovirus‐ mediated gene transfer of decorin. Oncogene 21, 3688‐3695. 

[34]  Keene DR, San Antonio JD, Mayne R, McQuillan DJ, Sarris G, Santoro SA, and Iozzo RV  (2000). Decorin binds near the C terminus of type I collagen. J Biol Chem 275, 21801‐ 21804. 

[35]  Neame  PJ,  Kay  CJ,  McQuillan  DJ,  Beales  MP,  and  Hassell  JR  (2000).  Independent  modulation of collagen fibrillogenesis by decorin and lumican. Cell Mol Life Sci 57, 859‐ 863. 

[36]  Geng  Y,  McQuillan  D,  and  Roughley  PJ  (2006).  SLRP  interaction  can  protect  collagen  fibrils from cleavage by collagenases. Matrix Biol 25, 484‐491. 

[37]  Linsenmayer  TF,  Fitch  JM,  and  Birk  DE  (1990).  Heterotypic  collagen  fibrils  and  stabilizing collagens. Controlling elements in corneal morphogenesis? Ann N Y Acad Sci  580, 143‐160. 

[38]  Weber  IT,  Harrison  RW,  and  Iozzo  RV  (1996).  Model  structure  of  decorin  and  implications for collagen fibrillogenesis. J Biol Chem 271, 31767‐31770. 

[39]  Scott  JE  (1996).  Proteodermatan  and  proteokeratan  sulfate  (decorin,  lumican/fibromodulin)  proteins  are  horseshoe  shaped.  Implications  for  their  interactions with collagen. Biochemistry 35, 8795‐8799. 

[40]  Orgel JP, Eid A, Antipova O, Bella J, and Scott JE (2009). Decorin core protein (decoron)  shape  complements  collagen  fibril  surface  structure  and  mediates  its  binding.  PLoS 

One 4, e7028. 

[41]  Wang J, Levenson AS, and Satcher RL, Jr. (2006). Identification of a unique set of genes  altered during cell‐cell contact in an in vitro model of prostate cancer bone metastasis. 

Int J Mol Med 17, 849‐856. 

[42]  Ueno  H,  Jones  AM,  Wilkinson  KH,  Jass  JR,  and  Talbot  IC  (2004).  Histological  categorisation of fibrotic cancer stroma in advanced rectal cancer. Gut 53, 581‐586.  [43]  Caporale A, Amore Bonapasta S, Scarpini M, Ciardi A, Vestri A, Ruperto M, and Giuliani 

A  (2010).  Quantitative  investigation  of  desmoplasia  as  a  prognostic  indicator  in  colorectal cancer. J Invest Surg 23, 105‐109. 

[44]  Forster  SJ,  Talbot  IC,  and  Critchley  DR  (1984).  Laminin  and  fibronectin  in  rectal  adenocarcinoma: relationship to tumour grade, stage and metastasis. Br J Cancer 50,  51‐61. 

[45]  Havenith MG, Arends JW, Simon R, Volovics A, Wiggers T, and Bosman FT (1988). Type  IV  collagen  immunoreactivity  in  colorectal  cancer.  Prognostic  value  of  basement  membrane deposition. Cancer 62, 2207‐2211.  [46]  Bosman FT, de Bruine A, Flohil C, van der Wurff A, ten Kate J, and Dinjens WW (1993).  Epithelial‐stromal interactions in colon cancer. Int J Dev Biol 37, 203‐211.  [47]  Painter KJ (2009). Modelling cell migration strategies in the extracellular matrix. J Math  Biol 58, 511‐543.   

(40)

162  Figure legends

Figure 1. Gene expression of collagens and SLRPs after heterotypic bilayer co-culture of colorectal cancer cells and fibroblasts. The colorectal cancer cell lines Caco-2 and HCT 116 were seeded on top of previously seeded fibroblasts (CACO2/WPF5 and HCT116/WPF5, respectively) and fibroblasts were seeded on top of the previously seeded colorectal cancer cell lines (WPF5/CACO2 and WPF5/HCT116, respectively). After 24 hours, RNA was extracted for analysis by real time PCR of collagen I (A), collagen III (B), collagen IV (C), collagen V (D) decorin (E), biglycan (F), lumican (G) and fibromodulin (H). Controls were performed seeding the same cell type on top and underneath (WPF5/WPF5, CACO2/CACO2 and HCT116/HCT116). Gene expression was normalized against ribosomal protein S29 (RPS29).

Figure 2. Gene expression of collagens and SLRPs after bi-compartmental (transwell) co-culture of colorectal cancer cells and fibroblasts. Fibroblasts (WPF5) and prostate cancer cells (CACO2 and HCT116) were grown both in transwell inserts (0.4 μm membrane pores) and in microplate wells to study the effect of soluble factors in the cross-talk between the cell lines. RNA extracted from the cells in the bottom compartment was analyzed through quantitative RT–PCR. The experimental groups comprised seeding the colorectal tumor cell lines in the top compartment and the fibroblasts in the bottom compartment (tCACO2/WPF5 and tHCT116/WPF5) as well as the fibroblasts in the top compartment and the cancer cell lines in the bottom compartment (tWPF5/CACO2 and tWPF5/HCT116). The control groups entailed seeding the same cells in both the top and bottomcompartments (tWPF5/WPF5;

(41)

tCACO2/CACO2; tHCT116/HCT116). The expression of collagen I (A), collagen III (B), collagen IV (C), collagen V (D), decorin (E), biglycan (F), lumican (G) and fibromodulin (H) was analyzed through real time PCR. Gene expression was normalized against ribosomal protein S29 (RPS29). *p < 0.05.

Figure 3. Expression and localization of collagen types I, IV and V, fibromodulin and biglycan after co-culture of colorectal cancer cells and fibroblasts using the O-ring system. Experimental groups entailed seeding the colorectal cancer cell lines Caco-2 and HCT 116 inside the O-ring and fibroblasts around the O-ring (WPF5 & CACO2 and WPF5 & HCT116, respectively). Control groups consisted of seeding the same cell type inside and around the O-ring (WPF5, CACO2, HCT116). (A) Collagen I (red) and collagen V (green) were immunolocalized in cells cultured using the O-ring system. (B) Collagen I (red) and fibromodulin (green) were immunolocalized in cells cultured using the O-ring system.Fluorescent labeling of collagen I (C) collagen V (D) and fibromodulin (E) was quantified using the LSM image browser (*p<0.05). Biglycan (F) and collagen IV (G) were immunolocalized in cells cultured using the O-ring system where the colorectal cancer cell line Caco-2 (Ca) was seeded inside the O-ring and fibroblasts (Fi) around the O-ring. (H) Phase contrast image of the O-ring system where the colorectal cancer cell line Caco-2 was seeded inside the O-ring and fibroblasts around the O-ring. Asterisks indicate the region of fibroblast/colorectal cancer cell interface. Nuclei are DAPI stained (blue). Scale bar: 20 m (A and B) and 100 m (F and G). (*p<0.05).

(42)

164  Figure 4. 3-D stromagenic model. 3-D co-cultures were performed in which fibroblasts (WPF5) or colorectal cancer cells (Caco-2) were seeded on top of previously plated confluent fibroblasts (WPF5 and WPF5 + CACO2, respectively). The cultures were treated with ascorbic acid every other day to produce a 3-D stromagenic system with a cellular component of solely fibroblasts or a 3-D stromagenic system composed of both fibroblasts and colorectal cancer cells (with desmoplastic reaction). (A) Phase contrast images; (B) Fluorescent labeling of fibronectin, collagen I, perlecan and versican. Nucleus stained with DAPI (blue). Scale bar: 20m.

Figure 5. Invasion assay of colorectal cancer cells (Caco-2) through a 3-D fibroblast-produced matrix.Caco-2 cells expressing GFP (green) were seeded onto 3-D control stroma (WPF5) or 3-D desmoplastic stroma (WPF5 + CACO2) and confocal images obtained 2 hours later. A series of z-axis optical sections were collected, acquired at 1m intervals using the 40X objective. Maximum projection reconstructed confocal images of Caco-2 cells (green) migrating into a 3-D control stroma (WPF5) or a 3-D desmoplastic stroma (WPF5 + CACO2) were acquired using the x axis (A) and z axis (B). Arrows indicate direction of migration.

Figure 6. Invasion assay of colorectal cancer cells (HCT 116) through a 3-D fibroblast-produced matrix. HCT 116 cells expressing GFP (green) were seeded onto 3-D control stroma (WPF5) or 3-3-D desmoplastic stroma (WPF5 + HCT 116) and confocal images obtained 2 hours later. A series of z-axis optical sections were collected, acquired at 1m intervals using the 40X objective. Maximum projection reconstructed confocal images of HCT 116 cells (green) migrating into a 3-D control stroma (WPF5)

(43)

or a 3-D desmoplastic stroma (WPF5 + HCT 116) were acquired using the x axis (A) and z axis (B). Arrows indicate direction of migration.

Figure7. Localization of collagen types I and IV in colorectal carcinoma and non-neoplastic tissues. Immunoreactivity of collagen I (A and B) and collagen IV (C and D) in colorectal tissues shows an increase in collagen deposition surrounding adenocarcinoma (B and D) when compared to normal tissues (A and C). Increased expression of collagens I and IV were observed in basal membrane of neoplastic glands when compared to non-neoplastic tissue. Increased collagen IV immunoreactivity is also observed in the carcinoma cell cytoplasm when compared to non-neoplastic tissues. Images were obtained using a 40X objective. Scale bar: 20m.

Figure 8. Localization of collagen type V in colorectal carcinoma and non-neoplastic tissue. Collagen V was immunolocalized in colorectal tissues using DAB and counter stained with hematoxylin. Collagen V increases in the extracellular matrix of colorectal carcinoma (B,D,F) when compared to non-neoplastic tissues (A, C, E). A and B were obtained with a 20X objective, C, D, E and F were obtained using a 40X objective. Sample shown in E was revealed omittingthe primary antibody. Scale bar: 20m

Figure 9. Localization of biglycan and fibromodulin in colorectal carcinoma and non-neoplastic tissue. Biglycan and fibromodulin were immunolabeled in colorectal tissues using DAB and counter stained with hematoxylin. Biglycan (A,B) and fibromodulin (C,D,E,F) increase in the basal membrane of neoplastic glands (B,D,F) when compared to normal tissues (A,C,E). Increased fibromodulin immunoreactivity is also observed in

(44)

166  the carcinoma cell cytoplasm when compared to non-neoplastic tissues.C and D images were obtained using a 20X objective and A, B, E and F images with a 40X objective. Scale bar: 20m.

Supplemental Figure 1. Invasion assay of colorectal cancer cells (CACO2) through a 3-D fibroblast-produced matrix. Caco-2 cells expressing GFP (green) were seeded onto 3-D control stroma (WPF5) or 3-D desmoplastic stroma (WPF5 + CACO2) and confocal images obtained 2 hours later. A series of z-axis optical sections were collected, acquired at 1m intervals using the 40X objective of migrating into a 3-D control stroma (WPF5) or a 3-D desmoplastic stroma (WPF5 + CACO2).

Supplemental Figure 2. Invasion assay of colorectal cancer cells (HCT 116) through a 3-D fibroblast-produced matrix. HCT 116 cells expressing GFP (green) were seeded onto 3-D control stroma (WPF5) or 3-D desmoplastic stroma (WPF5 + HCT 116) and confocal images obtained 2 hours later. A series of z-axis optical sections were collected, acquired at 1m intervals using the 40X objective of migrating into a 3-D control stroma (WPF5) or a 3-D desmoplastic stroma (WPF5 + HCT 116).

(45)

Table 1. Clinical details of patients with colorectal cancer

Age Gender Location Size Differentiation

65 M Rectum 9 cm Moderate

56 M Rectum 4 cm Moderate

51 M Rectum 2 cm Well

57 M Rectum 6 cm Moderate

(46)

168  Table 2.Gene responsiveness to direct cell-cell contact, soluble factors and both cell-cell contact and soluble factors

Direct cell-cell contact Soluble factors Direct cell-cell contact and

soluble factors Collagen III Collagen IV Lumican Decorin Collagen I Collagen V Biglycan Fibromodulin  

(47)
(48)

Figurre 2.

(49)

Referências

Documentos relacionados

In the present research, the increase in cell viability and total protein production occurred when the MDPC-23 cells were subjected to a single irradiation with red LED at 20

However, compared to the control group, a 15 % increase in cell viability for cardiac fibroblasts was observed for cells in contact with the Pt/PU sample, Figure 6a. Furthermore, an

We found that gastric cancer cell migration, invasion, proliferation and apoptosis were restored in the AGS cell line with forced miR-338 expression and NRP1 restoration

In the TSPY1 knockdown experiments, the cell proliferation were reduced obviously in the MHCC97H cells silenced the expression of TSPY1 with shRNA compared to the mock group

TLR mRNA fold increase in corneal fibroblasts treated with cell wall components from Staphylococcus

Increased proliferation of human synovial mesen- chymal stem cells with autologous human serum: compari- sons with bone marrow mesenchymal stem cells and with fetal bovine serum.

In order to evaluate the percentage of cancer stem cells in the total HepG2 cell population following exposure to low-dose cisplatin, cells were labeled with the cancer stem

To assess the cell adhesion on different surfaces, HUVECs were seeded on different substrates. After cells were cultured for 4 h, cells were harvested and the cell number was