p53 functional deficiency in human colon cancer cells promotes fibroblast-mediated angiogenesis and tumor growth
p53 functional deficiency in human colon cancer cells promotes fibroblast-mediated angiogenesis and tumor growth
Yoshito Hayashi 2
Masahiko Tsujii 2
Takahiro Kodama 2
Tomofumi Akasaka 2
Jumpei Kondo 2
Hayato Hikita 2
Takuya Inoue 2
Yoshiki Tsujii 2
Akira Maekawa 2
Shunsuke Yoshii 2
Shinichiro Shinzaki 2
Kenji Watabe 2
Yasuhiko Tomit 1 2
Masahiro Inoue 0 2
Tomohide Tatsumi 2
Hideki Iijima 2
Tetsuo Takehara 2
0 Department of Biochemistry, Osaka Medical Center for Cancer and Cardiovascular Diseases , Osaka 537-8511 , Japan
1 Department of Pathology
2 Department of Gastroenterology and Hepatology, Osaka University Graduate School of Medicine , Suita 565-0871 , Japan
Cancer-associated fibroblasts (CAFs) create a microenvironment that contributes to tumor growth; however, the mechanism by which fibroblasts are phenotypically altered to CAFs remains unclear. Loss or mutation of the tumor suppressor p53 plays a crucial role in cancer progression. Herein, we analyzed how the p53 status of cancer cells affects fibroblasts by investigating thine vivo and in vitro effects of loss of p53 function in cancer cells on phenotypic changes in fibroblasts and subsequent tumor progression in human colon cancer cell lines containing wild-type p53 and in cells with a p53 functional deficiency. The growth of p53-deficient tumors was significantly enhanced in the presence of fibroblasts compared with that of p53-wild-type tumors or p53-deficient tumors without fibroblasts. p53-deficient cancer cells produced reactive oxygen species, which activated fibroblasts to mediate angiogenesis by secreting vascular endothelial growth factor (VEGF) bothin vivo and in vitro. Activated fibroblasts significantly contributed to tumor progression. Deletion of fibroblast-derived VEGF or treatment witNh-acetylcysteine suppressed the growth of p53-deficient xenograft tumors. The growth effect of blocking VEGF secreted from cancer cells was equivalent regardless of p53 functional status. Human colon cancer tissues also showed a significant positive correlation between p53 cancer cell staining activated fibroblasts and microvessel density. These results reveal that fibroblasts were altered by exposure to p53-deficient epithelial cancer cells and contributed to tumor progression by promoting neovascularization. Thus, p53 acts as a modulator of the tumor microenvironment.
The development of novel anticancer therapies is generally survival. As such, the tumor microenvironment is thought to be
approached by elucidating the biological characteristics of -can a novel treatment target in cancer.
cer cells; however, recent studies have demonstrated the imp-or Stroma cells surrounding cancer cells create their own
tance of understanding the tumor microenvironment. Cross microenvironments by producing numerous growth factors and
talk between cancer cells and surrounding stromal cells, such ascytokines. Previously reported growth factors and cytokines
fibroblasts and blood cells, through various mediators including expressed on the stroma cells include basic fibroblast growth
hormones, cytokines and chemokines is crucial for cancer cell factor, hepatocyte growth factor, transforming growth factor,
St. Louis, MO) was used at concentrations of 100 µM in culture medium
and H2O2 (Nacalai Tesque, Kyoto, Japan) was used at 50 µM. Transwell
coculture was performed using cell culture inserts (BD Biosciences, San
Jose, CA). WI-38 or CCD-18Co cells were seeded into the lower chamber
and HCT116 cells were seeded 24h later, followed by 7h2 coculture. Cell
viability was analyzed using water-soluble tetrazolium assays with SF cell
counting reagent (Nacalai Tesque). Human umbilical vein endothelial cells
(HUVECs) were kindly provided by Dr Takashima (Osaka University, Japan),
cultured using the EGM-2 Bullet Kit (Lonza, Basel, Switzerland) and used
for experiments within five passages.
factor (VEGF) 1(–6). Cancer-associated fibroblasts (CAFs) within
the tumor stroma play an important role in the development
of a variety of tumors 7(). CAFs are characterized by expression
markers such as α-smooth muscle actin (αSMA), fibroblast-act-i
vated protein or Gremlin-18(). αSMA-positive fibroblast
agglomerations are commonly observed in malignant breast, prostate,
pancreatic, gastric and colorectal cancers. Accordingly, high Quantitative real-time RT–PCR
Materials and methods
Biomedical Innovation, Osaka, Japan). HCT116 TP53+/+ and TP53−/− cells
(henceforth referred to as HCT11+6/+ and HCT116−/−, respectively) were a
generous gift from Dr Vogelstein (John Hopkins University1)5(). All cell ROS analysis
lines were maintained according to supplier instruction and were last Intracellular oxidative stress was estimated both visually and qu-anti
authenticated on 23 February 2015. The fibroblast lines were used for tatively. Cultured cells were stained with 20 µM CellROX Deep Red-rea
experiments within 15 passages. N-acetylcysteine (NAC) (Sigma-Aldrich, gent (Invitrogen) and Hoechst 33324 for 3m0in and then visualized
using a confocal laser-scanning inverted microscope (Olympus, Tokyo, (Figure 1C; Supplementary Figure 1A, available atCarcinogenesis
Japan). For quantitative evaluation, dichlorodihydrofluorescein flu-ores Online). No significant differences in HCT11+6/+ and
HCT116-/cence was detected using an OxiSelect Intracellular ROS Assay kit (Cell tumor size were observed by 21 days post-implantation; ho
wBiolabs, San Diego, CA). Specifically, HCT11+6/+ and HCT116−/− cells (5× 103 ever, tumor volumes in mice coinjected with HCT116−/− and
cells) were seeded onto 96-well plates and cultured forh.72Intracellular WI-38 cells were significantly higher than those from HCT1+/1+6
ROS production was estimated according to manufacturer protocol. To
evaluate ROS secretion in the supernatant, HCT1s1h6control and HCT116sh and WI-38 coinjection. Subcutaneous injection of WI-38 cells
p53 cells were seeded on 96-well plates ×(5103) and cultured for 72h. alone did not result in tumor formation for at least 30 days.These
Dichlorodihydrofluorescein fluorescence in the supernatant was m-eas results demonstrated that tumors containing p53-deficient c-an
ured using an OxiSelect In Vitro ROS Assay kit (Cell Biolabs). cer cells progressed in the presence of fibroblasts.
To exclude the possibility of non-p53-dependent
phenoTube formation assay typic changes in the HCT116−/− cells compared with HCT116+/+
HUVECs were cultured on growth factor-reduced Matrigel (BD Biosciences). counterparts, we established cells in which p53 was suppressed
The supernatants from HCT116sh control or HCT116sh p53 cells cocultured with (HCT116sh p53) using shRNA. Consistent with the results obtained
WI-38 cells for 72h were harvested and used as conditioned medium. using HCT116+/+ and HCT116−/− cells, qRT–PCR and western blo-t
HUVECs (1.2 × 105) were seeded into 24-well plates and cultured in the c-on ting showed that p53 downstream molecules were suppressed in
ditioned medium for 16 h. The total length of the HUVEC tube network HCT116sh p53 compared with HCT116sh control cells F(igure 1D and E).
iinndeeaxchoffiienlvditwroaasnagniaolgyezneedsiuss(iwnginiRmOaOgeF avnisaulaylsissyssotfetmw;aMreitaanndiCuos,eTdoaksyoa,n Additionally, when HCT11s6h p53 or HCT116sh control and WI-38 cells
Japan). were coinjected, tumors containing HCT11sh6p53 and WI-38 cells
grew significantly larger than those containing HCTsh11co6ntrol
Clinical specimens and immunohistochemistry and WI-38 cells, although no differences were observed between
HCT116sh p53- and HCT116sh control-derived tumors in the absence
Paraffin-embedded colon cancer specimens from patients with ava-il
able clinical data were obtained from Osaka Medical Center for Cancer of WI-38 cells F(igure 1F; Supplementary Figure 1B, available at
and Cardiovascular Diseases (Osaka, Japan) in accordance with the Carcinogenesis Online).
Declaration of Helsinki (2008). Endoscopically resected colorectal tumors To exclude the possibility that these results were cell line
diagnosed pathologically as submucosal invasive cancers were analyzed specific, these analyses were repeated using CCD-18Co fibr-o
for p53 status in the invasive lesion by scoring p53 nuclear staining as blasts and RKO colon cancer cells with stable p53 shRNA exp-res
none, weak or strong as described2(0). In the same visual field,VEGF stain- sion (Supplementary Figure 2A and B, available atCarcinogenesis
ing of the stromal and cancer cell areas, respectively, were evaluated and Online). Xenograft tumors derived from HCT11s6h p53 and
CCDdivided into two groups, weak or strong. Tumor sections were depara-ffi
nized, rehydrated and labeled with antibodies as listedSiunpplementary
Table II, available atCarcinogenesis Online. The area with the highest sta-in
ing density was identified at low magnification, and individuαaSlMA- or
CD31-positive cells were then counted at 200× magnification. The number
of CD31-positive vessels identified by immunohistological staining was
independently evaluated by three blinded investigators.
18Co cells grew significantly larger than HCTs1h 1co6ntrol and
CCD-18Co cell tumors, although no differences were observed
between HCT116sh p53- and HCT116sh control-derived tumors in the
absence of CCD-18Co cells F(igure 1G). Similarly, xenograft
tumors from RKOsh p53 and WI-38 cells grew significantly larger
than RKOsh control and WI-38 cell tumorsS(upplementary Figure 2C,
available atCarcinogenesis Online). Thus, we hypothesized that
Study approval p53-deficient colon cancer cells interacted with fibroblasts to
Written informed consent was obtained from all patients. The analysis stimulate tumor progression.
of human tissues was approved by the Osaka Medical Center for Cancer
aAnndimCaalrpdriootvoacscoulslawreDreiseaappsreosvIendstbiytutthieonAanlimReavliCewarBeoaanrdd U(Nsoe.C1o4m02m0i5t5t2e0e1). Fibroblast characteristics are altered by coculture
of Osaka University Graduate School of Medicine (No. 24-098-030). with p53-deficient colon cancer cells
Next, we evaluatedin vitro cell proliferation to confirm our
Statistics in vivo results and found no significant differences in cell
All data are expressed as the means ± SD. Statistical analyses were-per proliferation between HCT11s6h control and HCT116sh p53 cells
formed using unpaired Student’st-tests. Human tissues were analyzed (Supplementary Figure 3A, available atCarcinogenesis Online) or
statistically by Kruskal–Wallis tests. Differences were considered st-atisti between HCT116sh control or HCT116sh p53 and WI-38 cell cocultures
cally significant whenP < 0.05.
(Supplementary Figure 3B, available atCarcinogenesis Online).
There was also no significant difference between these coc-ul
Results tures using the transwell systemFi(gure 2A). These suggested
that the growth of tumors containing p53-deficient or
wildFibroblasts facilitate p53-deficient tumor progression type cancer cells was not directly affected by the presence of
To determine the effects of fibroblasts on tumor growth and fibroblasts. Therefore, we theorized that other factors such as
their dependency on cancer cell p53 status, we characterized immunocompetent cells, blood cells, blood vessels, cytokines
wild-type or p53-deficient HCT116 colon cancer cells (HCT1+1/+6 or chemokines present within thein vivo tumor
microenvironand HCT116−/−, respectively) and the human fibroblast cell line ment were likely responsible for altered tumor growth.
WI-38. Quantitative reverse transcription–polymerase chain To test this hypothesis, we examined whether changes in the
reaction (qRT–PCR) and western blotting revealed that do-wn gene expression signature of fibroblasts occurred after cocu-ltur
stream targets of p53, such as p21 C(DKN1A) and Puma ( BBC3), ing with p53-deficient cancer cells, as observed previously in
were expressed at lower levels in HCT11−/6− than in HCT116+/+ activated CAFs within the tumor microenvironment. Several
cells F(igure 1A and B), indicating that p53 function is su-p genes are upregulated during the process of fibroblast activation
pressed in HCT116−/− cells. and the phenotypic shift to CAFs. Therefore, we investigated
To investigate the importance of fibroblasts for tumor p-ro CAF-related gene expression in fibroblasts cocultured with
gression, we performed xenograft experiments in BALB/c nude cancer cells. Notably, transwell coculture experiments revealed
mice with HCT116+/+ or HCT116−/− cells coinjected with WI-38 cells that some CAF-related gene expression was significantly
enhanced in WI-38 cells cocultured with HCT1−1/−6 (Figure 2B; increased in tumors containing HCT11s6h p53 and WI-38 cells
comSupplementary Figure 4A, available atCarcinogenesis Online) pared with that in tumors containing HCT11sh6control and WI-38
or HCT116sh p53 (Figure 2B; Supplementary Figure 4B, available cells F(igure 2G). Thus, to evaluate the contribution of VEGF
at Carcinogenesis Online) cells in comparison with HCT11+6/+ secreted from fibroblasts to the progression of p53-deficient
or HCT116sh control cells, respectively. These results suggest that tumors via neovascularization, we next analyzed microvessels
fibroblasts grown in the presence of p53-deficient cancer cells in the xenografts. As expected, the number of CD31-positive cells
were activated, unlike those cultured with p53-wild-type cancer was increased significantly in tumors formed from HCT11s6hp53
cells. Among the expression-altered genes, we focused on the and WI-38 xenografts compared with that observed in the other
increase inVEGFA, the main human VEGF isoform. In particular, groups (Figure 2H). These results suggest that p53-deficient c-an
transwell coculture experiments revealed thVaEtGF expression cer cells enhanced the secretion of tumor-promoting VEGF from
was also significantly enhanced in CCD-18Co cells that had been fibroblasts.
cocultured with HCT11s6h p53 in comparison with HCT116sh control
cells F(igure 2B). Fibroblast-dependent VEGF secretion plays a crucial
VEGF is a critical pro-angiogenic factor secreted by cancer role in p53-deficient tumor progression
cells, fibroblasts, pericytes and endothelial cells and correlatesOur results suggested that p53-deficient cancer cells induce
with poor prognosis in patients with colon cancer21(,22). An fibroblast activation to promote VEGF secretion, which sub-se
earlier study found that p53 regulates angiogenesis by sti-mu quently contributes to tumor progression. Paracrine VEGF is
lating the expression of microRNAs 2(3). Thus, we performed released from cancer cells, myeloid cells, fibroblasts or other
qRT–PCR and ELISA to analyze VEGF expression levels in stromal cells to increase vessel branching and render tumor
p53-deficient or p53-wild-type cells. No significant differences vessels abnormal 2(4). To investigate the significance of fibr-o
were observed between HCT116+/+ and HCT116−/− or HCT116sh control blast VEGF secretion, we next performed xenograft experiments
and HCT116sh p53 cells F(igure 2C); however, the supernatant of using fibroblasts whereVEGF expression was suppressed using
cocultured WI-38 or CCD-18Co and HCT11s6hp53 cells contained siRNA. Transfection of WI-38 cells with two different siRNAs
significantly more VEGF compared with that of WI-38 or CCD- againstVEGF resulted in a substantial decrease VinEGF
expres18Co and HCT116sh control cocultures, respectivelyFi(gure 2D). sion (Figure 3A). We inoculated HCT116sh control or HCT116sh p53
Together, these results suggest that fibroblasts cultured withcells and WI-3 8si control or WI-38si VEGF cells into BALB/c nude mice.
p53-deficient cancer cells secreted more VEGF than those c-ul Coimplantation of HCT116sh p53 and WI-38si VEGF cells resulted in
tured with cancer cells exhibiting wild-type p53 expression. dramatically reduced tumor size compared with tumors co-n
To test whether the VEGF produced by p53-deficient cancer taining VEGF-expressing fibroblasts and p53-deficient cancer
cells and fibroblast cocultures exhibited pro-angiogenic function, cells F(igure 3B). Conversely,VEGF suppression in fibroblasts did
we performed endothelial tube formation assays. Supernatants not influence the size of tumors derived from cells expressing
harvested from cocultured HCT116 and WI-38 cells were added wild-type p53.
to HUVECs. Notably, total endothelial cell tube length was To confirm the effects of cancer-cell-derived VEGF, we inh-ib
increased by cocultured HCT11s6hp53 and WI-38 cell supernatants itedVEGF expression in HCT116sh control and HCT116sh p53 cells using
compared with those of HCT116sh control and WI-38 cells, sugges-t siRNA (Figure 3C), and then HCT116sh control/si control, HCT116sh control/si
ing that VEGF secreted from p53-deficient, cancer-cell-activated VEGF, HCT116sh p53/si control or HCT116sh p53/si VEGF cells were implanted
fibroblasts promoted neovascularizationFig(ure 2E).
with WI-38 cells into the trunks of BALB/c nude mice. Although
xenografts containing fibroblasts harboriVnEgGF siRNA showed
Neovascularization is enhanced in tumors composed significantly decreased growth, suppressinVgEGF expression in
of fibroblasts and p53-deficient cancer cells cancer cells did not affect tumor growth regardless of p53 status
To ascertain whether p53-deficient cancer cells stimulated (Figure 3D). These results showed that blocking VEGF expression
fibroblast activation and neovascularization, we next analyzed only in cancer cells had equivalent effects regardless of cancer
fixed tissues from the xenograft experiments described above cell p53 functional status. Although cancer cells are key-sup
by immunofluorescence staining. Notably, the number of an-ti pliers of VEGF, fibroblasts are crucial for the neovascularization
human αSMA-positive cells was increased in xenograft tumors that promotes tumor progression in p53-deficient tumors
comderived from HCT116−/− and WI-38 cells compared with tumors pared with p53-wild-type tumors. These data clearly demo-n
derived from HCT116+/+ cells with or without WI-38 cells and strated that the VEGF secreted from fibroblasts, but not cancer
HCT116−/− cells without WI-38 cellsSu( pplementary Figure 5A, cells, was essential for the progression specifically of p53-d-efi
available atCarcinogenesis Online). These results suggested cient tumors.
tnhuamtbetrheofpraecsteinvacteeodf mpy5o3-fdiberfoicbileansttsc.aInncearddcietliolsn,intchreeansuemdbtehre ROS production is enhanced in p53-deficient cells,
of CD31-positive endothelial cells, indicative of microvessels, inducing VEGF production from fibroblasts and
was increased in xenografts containing HCT11−/6− and WI-38 subsequent tumor growth
cells compared with those containing other cell combinations Next, we focused on ROS to examine the mechanism by which
(Figure 2F). fibroblasts are activated by p53-deficient cells. Both antioxidant
Consistent with the experiments performed using HCT11+6/+ and pro-oxidant functions for p53 have been reported previously
or HCT116−/− cells, the number ofαSMA-positive cells increased (25). For example, downregulation of p53 elevates intracellular
in tumors containing HCT116sh p53 and WI-38 cells compared with ROS levels under non-stressed conditions25(). Staining with
xenografts containing HCT11s6hcontrol cells with or without WI-38 CellROX Deep Red Reagent revealed that ROS production was
cells or HCT116sh p53 cells without WI-38 cellsSu( pplementary higher in HCT116−/− and HCT116sh p53 cells than in HCT11+6/+ and
Figure 5B, available atCarcinogenesis Online). To evaluate the HCT116sh control cells, respectivelyF(igure 4A and B). These
obserimportance of VEGF in tumor growth, we analyzedVEGF mRNA vations were confirmed by estimating the oxidative status of
levels in the HCT116sh control plus WI-38 and HCT116sh p53 plus RKO cells: ROS production was increased in RKsOhp53 cells
comWI-38 xenografts. qRT–PCR revealed thaVtEGF expression was pared with RKOsh control cells S(upplementary Figure 6A, available
at Carcinogenesis Online). Quantitative analysis of dichlorodi-hy We then analyzed the expression of several antioxidant genes
drofluorescein staining confirmed that ROS production was -sig to better understand the mechanism by which p53 loss of function
nificantly increased in HCT11s6hp53 cells compared with HCT116sh increases ROS levels. Notably, expression oNfQO1, GPX1, SESN1
control cells F(igure 4C). Moreover, analysis of ROS levels in the c-el and SESN2 was decreased in p53-deficient cells compared with
lular supernatants revealed that extracellular ROS productioncontrols F(igure 4E and F). Consistent results were also obtained
was higher in HCT116−/− cells than in HCT11+6/+ cells F(igure 4D). with RKOsh p53 and RKOsh control cells S(upplementary Figure 6B,
available atCarcinogenesis Online). Thus, inhibiting p53 function in antioxidant genes. These results suggested that the downreg-ula
non-stressed or physiologically stressed cells resulted in ex-ces tion of antioxidant genes caused by the loss of functional p53 was
sive oxidative stress accompanied by the suppression of several related to both intracellular and extracellular ROS production.
To test the hypothesis that ROS production in cancer cells Discussion
upregulates VEGF in fibroblasts, we investigated the effect of
fHib2Oro2—blaasrtesp.rHe2Ose2nsttaimtiuvleaRtiOoSn csoigmnpiofinceanntt—lyoinncVrEeGaFseeVdxEpGrFesmsiRoNnAin iUnngdteursmtoarndprinoggrtehsseiornolmeaoyf ltehaed ttuomtohre midicernoteifnivcairtoinonmeonftndo-uverl
and protein expression in WI-38 or CCD-18Co cellsF(igure 5A). To anticancer therapies. Fibroblasts play important roles in tumor
investigate whether elevated ROS levels in p53-deficient cells-con progression, infiltration and metastasis3,2(6). Recently, str-o
tributed to tumor progression by increasing VEGF production in mal signatures were identified that predict colon cancer re-cur
fibroblasts, we treated cells with the antioxidant NAC. Notably, therence (27,28); however, fibroblast or CAF depletion was shown
increasedVEGF expression observed in fibroblasts cocultured with to reduce survival in patients with pancreatic cancer by ac-cel
p53-deficient cancer cells was abolished by NAC treatmenintvitro erating tumor growth2(9,30). Thus, it remains unclear how
(Figure 5B). Moreover, NAC treatment inhibited VEGF secretion in fibroblasts acquire malignant potential within the tumor mi-cro
fibroblast and colon cancer cell coculture supernataFnigtsur(e 5C). environment. Herein, we found that the characteristic fi-bro
In vivo, tumor growth induced by coinjection of fibroblasts blast activation to support tumor progression was triggered by
and p53-deficient cells was partially inhibited by NAC tr-eat genetic alterations of cancer cells.
ment (Figure 5D). Consistent with this finding, NAC treatment Angiogenesis plays a crucial role in tumor progression and
significantly reduced the number of CD31-positive microvessels metastasis. Accordingly, anti-VEGF or VEGF receptor inh-ibi
in tumors consisting of p53-deficient cancer cells and fibroblasts tion significantly prolongs the survival of patients with c-olo
(Figure 5E). Additionally, NAC supplementation prevented the rectal cancer3(1,32). We confirmed that VEGF was produced
increase inVEGF expression in tumors containing p53-deficient by HCT116, WI-38 and CCD-18Co cells. Importantly, our results
cancer cells and fibroblastFsig(ure 5F). These results suggested do not contest the importance of cancer-cell-derived VEGF
that the ROS generation induced by p53 loss of function in c-an because we did not investigate cancer cell VEGF expression in
cer cells substantially stimulated VEGF production in fibroblasts,the absence of fibroblasts. We demonstrated that blocking VEGF
which led to tumor progression. expression from fibroblasts affected p53-deficient cancer cells
but not p53-wild-type cells. VEGF secreted from fibroblasts is
Vessel density is correlated with p53 staining in (k3n3o,3w4n); thoowbeevimerp,otrhteanmtefcohratnuismmorofaVnEgGioFgienndeuscistiaonndfprroomgrfeibsrsoi-on
human colon cancer blasts in cancer tissue remains unclear. In addition, VEGF levels
Lastly, we assessed the correlation between angiogenesis and are reported to increase when colon cancer cells are cocultured
p53 status in human colorectal cancer tissues. Because the data with fibroblasts 3(5). p53 directly modulates angiogenesis by
described above demonstrated that p53 suppression in cancer regulating the synthesis of thrombospondin-1, a potent ang-io
cells was involved in tumor progression through stimulation of genesis inhibitor, and semaphorin-3F (36,37). p53 promotes
VEGF secretion from activated fibroblasts, we analyzed cancer VEGF transcription in cancer cells exposed to hypoxia38().
tissues at the submucosal invasion stage. Submucosal invasive Correspondingly, the p53 status of cancer cells regulates their
cancers that had been resected endoscopically were analyzed by angiogenic potential3(9). Here, we showed that cancer cell p53
immunohistochemistry to assess the relationship between p53 status could alter VEGF secretion in fibroblasts by modulating
staining, fibroblast activation status and blood vessel densityROS production. Moreover, accumulating evidence from clinical
(Figure 6A). We analyzed samples from 36 patients with colon trials has demonstrated the efficacy of anti-angiogenic agents
cancer who were treated through endoscopic submucosal d-is in patients with colorectal cancer but lacks a useful biomarker
section; 12 patients underwent surgery after endoscopic tr-eat for activity 3(1,32). Our results suggested that drugs targeting
ment. Histological analysis revealed two cases of lymph node VEGF, particularly that secreted from fibroblasts, might exert
metastasis out of six patients with positive nuclear p53 expre-s beneficial effects in patients with colorectal tumors containing
sion levels indicating the presence of mutated p53 and deficient mutated p53.
p53 function (12), whereas there were no lymph node metas- ROS play opposing roles in tumorigenesis. First, ROS
tases in six patients with negative or weak p53 nuclear sta-in exert cytotoxic effects by damaging DNA. Recent studies
ing (Supplementary Table III, available atCarcinogenesis Online). showed that antioxidant molecules reduce ROS produ-c
Exclusion of the muscularis mucosa by differential immu- tion and decrease cancer cell sensitivity to chemotherapy
nostaining for desmin andαSMA is necessary to appropriately (
). Conversely, ROS are important for cancer initiation and
evaluate the presence of myofibroblasts, an activated type of other crucial cellular biological processes such as stemness
fibroblasts: the muscularis mucosa is positive for botαhSMA and (
). Stress-induced p53 activation causes pro-apoptotic ROS
desmin expression, whereas activated fibroblasts lack desmin generation but can also regulate antioxidant gene expression
expression. Myofibroblasts were quantified by measuring the to maintain lower ROS levels under non-stressed conditions
thickness of theαSMA-positive and desmin-negative fibroblast (
). The present in vitro experiments revealed that ROS
cells. Two cases of lymph node metastasis were included in the production was enhanced significantly in p53-deficient cells
activated myofibroblast groupS(upplementary Table IV, available compared with that in p53-wild-type cells. p53 is a critical
at Carcinogenesis Online). Additionally, microvascular density regulator of intracellular ROS levels because it upregulates
was determined by counting CD31-positive cells. High levels of antioxidant genes to reduce ROS levels. Although ROS levels
p53 nuclear staining significantly correlated with myofibroblast were not measured in xenograft tumors, results from
NACagglomeration and vessel densityFi(gure 6B). Finally, VEGF stain- treated mice demonstrated that oxidative stress promoted
ing was performed. The degree of VEGF positivity in the stromal the cross talk between p53-deficient cells and fibroblasts,
area significantly correlated with vessel density, whereas therealthough NAC treatment did not completely prevent the a-ngi
was no significant relationship between VEGF staining in c-an ogenic effects of fibroblasts. Therefore, further studies are
cer cells and vessel densityFi(gure 6C). These results suggested required to elucidate the specific mechanisms through which
a relationship among p53 functional deficient colon cancer cells, cross talk between p53-deficient cancer cells and fibroblasts
myofibroblasts and angiogenesis in invasive colon cancer tissues. enhances VEGF production.
Although the origin and role of CAFs are unclear, they are Positive cell staining for p53 is generally caused by lack of
known to be recruited from bone marrow or local healthy tissue its degradation by MDM2 owing to mutation. Generally, p53
and are characterized by the expression oαfSMA (44). During mutation implies a gain or loss of function. Inactivating TP53
the transformation of fibroblasts into CAFs, cancer cells induce mutations are the most common genetic alteration in cancers
alterations in fibroblasts by secreting signals such as growthand inactivation of the p53 pathway occurs in most tumors. Our
factors, cytokines and chemokines in addition to ROS, thereby results demonstrated that p53 deficiency caused angiogenesis
promoting the conversion of fibroblasts into myofibroblasts. The through fibroblast activation and that there was positive c-orre
present study suggests that targeting fibroblasts in p53-deficient lation between p53 positive staining and vessel density.
colorectal cancer may provide a survival benefit in patients by In the present study, we primarily utilized WI-38 cells
inhibiting the conversion of fibroblasts into myofibroblasts. originating from the lung, as these cells are widely used for
comparisons with cancer cells and fibroblasts in many organs mutation: influence of tumor site, type of mutation, and adjuvant
including colon 4(5,46). Additionally, we also used CCD-18Co treatment. J. Clin. Oncol., 23, 7518–7528.
fibroblasts derived from human colon and confirmed the same 13. Hamelin, R. et al. (1994) Association of p53 mutations with short s-ur
responses. vival in colorectal cancer. Gastroenterology, 106, 42–48.
In conclusion, the current study suggests that loss of p53 14. Muller, P.A. et al. (2009) Mutant p53 drives invasion by promoting in-te
grin recycling. Cell, 139, 1327–1341.
function activated fibroblasts, contributing to malignant a-ngio 15. Bunz, F. et al. (1998) Requirement for p53 and p21 to sustain G2 arrest
genesis. These results indicate that molecular therapy targeting after DNA damage. Science, 282, 1497–1501.
the activation and interaction of cells constituting the tumor16. Levine, A.J. et al. (2009) The first 30 years of p53: growing ever more
microenvironment may be effective and that genetic mutations complex. Nat. Rev. Cancer, 9, 749–758.
in cancer cells could serve as biomarkers to identify patients 17. Lujambio, A. et al. (2013) Non-cell-autonomous tumor suppression by
likely to respond to such treatments. Further elucidation of the p53. Cell, 153, 449–460.
various cellular components of complex tumor environments 18. Schwitalla, S. et al. (2013) Loss of p53 in enterocytes generates an
will be useful for developing combination therapies for perso-n inflammatory microenvironment enabling invasion and lymph node
alized cancer treatment. metastasis of carcinogen-induced colorectal tumors. Cancer Cell, 23,
19. Hayashi, Y. et al. (2013) CagA mediates epigenetic regulation to atte-nu
Supplementary material ate let-7 expression in Helicobacter pylori-related carcinogenesis. Gut,
Supplementary Tables I–IV and Figures 1–6 can be found at 62, 1536–1546.
20. Jädersten, M. et al. (2011) TP53 mutations in low-risk myelodysplastic
http://carcin.oxfordjournals.org/ syndromes with del(5q) predict disease progression. J. Clin. Oncol., 29,
Funding 21. Tokunaga, T. et al. (1998) Vascular endothelial growth factor (VEGF)
mRNA isoform expression pattern is correlated with liver metastasis
The present study was supported by JSPS KAKENHI Grant and poor prognosis in colon cancer. Br. J. Cancer, 77, 998–1002.
Number JP15K19326 from the Ministry of Education, Culture, 22. Guo, X. et al. (2008) Stromal fibroblasts activated by tumor cells
Sports, Science and Techonology, Japan (to Y.H.). promote angiogenesis in mouse gastric cancer. J. Biol. Chem., 283,
Acknowledgements 23. Yamakuchi, M. et al. (2010) P53-induced microRNA-107 inhibits HIF-1
and tumor angiogenesis. Proc. Natl Acad. Sci. USA, 107, 6334–6339.
We sincerely thank Prof. Vogelstein (Johns Hopkins University) 24. Carmeliet, P. et al. (2011) Molecular mechanisms and clinical appli-ca
for providing HCT116+/+ and HCT116−/− cells and Prof. Takashima tions of angiogenesis. Nature, 473, 298–307.
(Osaka University) for providing HUVECs. 25. Sablina, A.A. et al. (2005) The antioxidant function of the p53 tumor
Conflict of Interest Statement: None declared. suppressor. Nat. Med., 11, 1306–1313.
26. Hwang, R.F. et al. (2008) Cancer-associated stromal fibroblasts promote
References pancreatic tumor progression. Cancer Res., 68, 918–926.
27. Calon, A. et al. (2015) Stromal gene expression defines poor-prognosis
1. Grugan, K.D. et al. (2010) Fibroblast-secreted hepatocyte growth factor subtypes in colorectal cancer. Nat. Genet., 47, 320–329.
plays a functional role in esophageal squamous cell carcinoma in-va 28. Isella, C. et al. (2015) Stromal contribution to the colorectal cance-r tran
sion. Proc. Natl Acad. Sci. USA, 107, 11026–11031. scriptome. Nat. Genet., 47, 312–319.
2. Hägglöf, C. et al. (2010) Stromal PDGFRbeta expression in prostate 29. Özdemir, B.C. et al. (2014) Depletion of carcinoma-associated fibr-o
tumors and non-malignant prostate tissue predicts prostate cancer blasts and fibrosis induces immunosuppression and accelerates pa-n
survival. PLoS One, 5, e10747. creas cancer with reduced survival. Cancer Cell, 25, 719–734.
3. Quante, M. et al. (2011) Bone marrow-derived myofibroblasts cont-rib 30. Rhim, A.D. et al. (2014) Stromal elements act to restrain, rather than
ute to the mesenchymal stem cell niche and promote tumor growth. support, pancreatic ductal adenocarcinoma. Cancer Cell, 25, 735–747.
Cancer Cell, 19, 257–272. 31. Kabbinavar, F.F. et al. (2005) Combined analysis of efficacy: the ad-di
4. Calon, A. et al. (2012) Dependency of colorectal cancer on a TGβ-F- tion of bevacizumab to fluorouracil/leucovorin improves survival
driven program in stromal cells for metastasis initiation. Cancer Cell, for patients with metastatic colorectal cancer. J. Clin. Oncol., 23,
22, 571–584. 3706–3712.
5. Orimo, A. et al. (2005) Stromal fibroblasts present in invasive human 32. Saltz, L.B. et al. (2008) Bevacizumab in combination with
oxaliplatinbreast carcinomas promote tumor growth and angiogenesis through based chemotherapy as first-line therapy in metastatic colorectal-can
elevated SDF-1/CXCL12 secretion. Cell, 121, 335–348. cer: a randomized phase III study. J. Clin. Oncol., 26, 2013–2019.
6. Ito, T.K. et al. (2007) The VEGF angiogenic switch of fibroblasts is re-gu 33. Fukumura, D. et al. (1998) Tumor induction of VEGF promoter activity in
lated by MMP-7 from cancer cells. Oncogene, 26, 7194–7203. stromal cells. Cell, 94, 715–725.
7. Mueller, M.M. et al. (2004) Friends or foes - bipolar effects of the tumour 34. O’Connell, J.T. et al. (2011) VEGF-A and Tenascin-C produced by
stroma in cancer. Nat. Rev. Cancer, 4, 839–849. S100A4+ stromal cells are important for metastatic colonization. Proc.
8. Sneddon, J.B. et al. (2006) Bone morphogenetic protein antagonist Natl Acad. Sci. USA, 108, 16002–16007.
gremlin 1 is widely expressed by cancer-associated stromal cells and 35. Koshida, Y. et al. (2006) Interaction between stromal fibroblasts and
can promote tumor cell proliferation. Proc. Natl Acad. Sci. USA, 103, colorectal cancer cells in the expression of vascular endothelial growth
14842–14847. factor. J. Surg. Res., 134, 270–277.
9. Herrera, M. et al. (2013) Cancer-associated fibroblast and M2 m- ac 36. Dameron, K.M. et al. (1994) Control of angiogenesis in fibroblasts by p53
rophage markers together predict outcome in colorectal cancer regulation of thrombospondin-1. Science, 265, 1582–1584.
patients. Cancer Sci., 104, 437–444. 37. Futamura, M. et al. (2007) Possible role of semaphorin 3F, a candidate
10. Vousden, K.H. et al. (2002) Live or let die: the cell’s response to p53. Nat. tumor suppressor gene at 3p21.3, in p53-regulated tumor angiogenesis
Rev. Cancer, 2, 594–604. suppression. Cancer Res., 67, 1451–1460.
11. Hu, W. et al. (2010) Glutaminase 2, a novel p53 target gene regulating 38. Farhang Ghahremani, M. et al. (2013) p53 promotes VEGF expression
energy metabolism and antioxidant function. Proc. Natl Acad. Sci. USA, and angiogenesis in the absence of an intact p21-Rb pathway. Cell
107, 7455–7460. Death Differ., 20, 888–897.
12. Russo, A. et al. (2005) The TP53 colorectal cancer international- col 39. Ravi, R. et al. (2000) Regulation of tumor angiogenesis by p53-induced
laborative study on the prognostic and predictive significance of p53 degradation of hypoxia-inducible factor 1alpha. Genes Dev., 14, 34–44.
40. Trachootham , D. et al. ( 2009 ) Targeting cancer cells by ROS-mediated me-ch 44 . Paunescu , V. et al. ( 2011 ) Tumour-associated fibroblasts and mesenchymal anisms: a radical therapeutic approach? Nat . Rev. Drug Discov., 8 , 579 - 591 . stem cells: more similarities than differences . J. Cell. Mol. Med ., 15 , 635 - 646 .
41. Suda , T. et al. ( 2011 ) Metabolic regulation of hematopoietic stem cells in 45 . Crowder, R.N. et al. ( 2016 ) Deubiquitinase inhibitor PR-619 sensitizes the hypoxic niche . Cell Stem Cell , 9 , 298 - 310 . normal human fibroblasts to TRAIL-mediated cell death . J. Biol. Chem .,
42. Polyak , K. et al. ( 1996 ) Genetic determinants of p53-induced apoptosis 291 , 5960 - 5970 . and growth arrest . Genes Dev., 10 , 1945 - 1952 . 46. Baba , M. et al. ( 2008 ) Blocking CD147 induces cell death in cancer cells
43. Bensaad , K. et al. ( 2006 ) TIGAR, a p53-inducible regulator of glycolysis through impairment of glycolytic energy metabolism . Biochem . Bi-o and apoptosis . Cell , 126 , 107 - 120 . phys. Res . Commun., 374 , 111 - 116 .