A zebrafish embryo screen utilizing gastrulation identifies the HTR2C inhibitor pizotifen as a suppressor of EMT-mediated metastasis

  1. Joji Nakayama  Is a corresponding author
  2. Lora Tan
  3. Yan Li
  4. Boon Cher Goh
  5. Shu Wang
  6. Hideki Makinoshima
  7. Zhiyuan Gong  Is a corresponding author
  1. Department of Biological Science, National University of Singapore, Singapore
  2. Cancer Science Institute of Singapore, National University of Singapore, Singapore
  3. Tsuruoka Metabolomics Laboratory, National Cancer Center, Japan
  4. Shonai Regional Industry Promotion Center, Japan
  5. Institute of Bioengineering and Nanotechnology, Singapore
  6. Division of Translational Research, Exploratory Oncology Research and Clinical Trial Center, National Cancer Center, Japan

Abstract

Metastasis is responsible for approximately 90% of cancer-associated mortality but few models exist that allow for rapid and effective screening of anti-metastasis drugs. Current mouse models of metastasis are too expensive and time consuming to use for rapid and high-throughput screening. Therefore, we created a unique screening concept utilizing conserved mechanisms between zebrafish gastrulation and cancer metastasis for identification of potential anti-metastatic drugs. We hypothesized that small chemicals that interrupt zebrafish gastrulation might also suppress metastatic progression of cancer cells and developed a phenotype-based chemical screen to test the hypothesis. The screen used epiboly, the first morphogenetic movement in gastrulation, as a marker and enabled 100 chemicals to be tested in 5 hr. The screen tested 1280 FDA-approved drugs and identified pizotifen, an antagonist for serotonin receptor 2C (HTR2C) as an epiboly-interrupting drug. Pharmacological and genetic inhibition of HTR2C suppressed metastatic progression in a mouse model. Blocking HTR2C with pizotifen restored epithelial properties to metastatic cells through inhibition of Wnt signaling. In contrast, HTR2C induced epithelial-to-mesenchymal transition through activation of Wnt signaling and promoted metastatic dissemination of human cancer cells in a zebrafish xenotransplantation model. Taken together, our concept offers a novel platform for discovery of anti-metastasis drugs.

Editor's evaluation

We are so impressed with this new and ambitious concept for chemical screening using zebrafish embryos to find a novel anti-metastasis drug, Pizotifen. We hope many researchers will use this screening system for anti-cancer drug discovery.

https://doi.org/10.7554/eLife.70151.sa0

Introduction

Metastasis, a leading contributor to the morbidity of cancer patients, occurs through multiple steps: invasion, intravasation, extravasation, colonization, and metastatic tumor formation (Nguyen et al., 2009; Welch and Hurst, 2019; Chaffer and Weinberg, 2011). The physical translocation of cancer cells is an initial step of metastasis and molecular mechanisms of it involve cell motility, the breakdown of local basement membrane, loss of cell polarity, acquisition of stem cell-like properties, and epithelial-to-mesenchymal transition (EMT) (Tsai and Yang, 2013; Lu and Kang, 2019). These cell-biological phenomena are also observed during vertebrate gastrulation in that evolutionarily conserved morphogenetic movements of epiboly, internalization, convergence, and extension progress (Solnica-Krezel, 2005). In zebrafish, the first morphogenetic movement, epiboly, is initiated at approximately 4 hr post fertilization (hpf) to move cells from the animal pole to eventually engulf the entire yolk cell by 10 hpf (Latimer and Jessen, 2010; Solnica-Krezel, 2006). The embryonic cell movements are governed by the molecular mechanisms that are partially shared in metastatic cell dissemination.

At least 50 common genes were shown to be involved in both metastasis and gastrulation progression: Knockdown of these genes in Xenopus or zebrafish induced gastrulation defects; conversely, overexpression of these genes conferred metastatic potential on cancer cells and knockdown of these genes suppressed metastasis (Yang and Weinberg, 2008; Dongre and Weinberg, 2019; Thiery et al., 2009; Nieto et al., 2016; Table 1). This evidence led us to hypothesize that small molecules that interrupt zebrafish gastrulation may suppress metastatic progression of human cancer cells.

Table 1
A list of the genes that are involved between gastrulation and metastasis progression.

A list of the 50 genes that play essential role in governing both metastasis and gastrulation progression. The gastrulation defects in Xenopus or zebrafish that are induced by knockdown of each of these genes were indicated. The molecular mechanism in metastasis that is inhibited by knockdown of each of the same genes was indicated.

GenesGastrulation defectsRefEffects in metastasisRef
BMPConvergence and extensionKondo, 2007EMTKatsuno et al., 2008
WNTConvergence and extensionTada and Smith, 2000Migration and invasionVincan and Barker, 2008
FGFConvergence and extensionYang et al., 2002InvisionNomura et al., 2008
EGFEpibolySong et al., 2013MigrationLu et al., 2001
PDGFConvergence and extensionDamm and Winklbauer, 2011EMTJechlinger et al., 2006
CXCL12Migration of endodermal cellsMizoguchi et al., 2008Migration and invasionShen et al., 2013
CXCR4Migration of endodermal cellsMizoguchi et al., 2008Migration and invasionShen et al., 2013
PIK3CAConvergence and extensionMontero et al., 2003Migration and invasionWander et al., 2013
YESEpibolyTsai et al., 2005MigrationBarraclough et al., 2007
FYNEpibolySharma et al., 2005Migration and invasionYadav and Denning, 2011
MAPK1EpibolyKrens et al., 2008MigrationRadtke et al., 2013
SHP2Convergence and extensionJopling et al., 2007MigrationWang et al., 2005
SNAI1Convergence and extensionIp and Gridley, 2002EMTBatlle et al., 2000
SNAI2Mesoderm and neural crest formationShi et al., 2011EMTMedici et al., 2008
TWIST1Mesoderm formationCastanon and Baylies, 2002EMTYang et al., 2004
TBXTConvergence and extensionTada and Smith, 2000EMTFernando et al., 2010
ZEB1EpibolyVannier et al., 2013EMTSpaderna et al., 2008
GSCMesodermal patterningSander et al., 2007EMTHartwell et al., 2006
FOXC2Unclear, defects in gastrulationWilm et al., 2004EMTMani et al., 2007
STAT3Convergence and extensionMiyagi et al., 2004MigrationAbdulghani et al., 2008
POU5F1EpibolyLachnit et al., 2008EMTDai et al., 2013
EZH2Unclear, defects in gastrulationO’Carroll et al., 2001InvasionRen et al., 2012
EHMT2Defects in neurogenesisLin et al., 2005Migration and invasionChen et al., 2010
BMI1Defects in skeleton formationvan der Lugt et al., 1994EMTGuo et al., 2011
RHOAConvergence and extensionZhu et al., 2006Migration and invasionYoshioka et al., 1999
CDC42Convergence and extensionChoi and Han, 2002Migration and invasionReymond et al., 2012
RAC1Convergence and extensionHabas et al., 2003Migration and invasionVega and Ridley, 2008
ROCK2Convergence and extensionMarlow et al., 2002Migration and invasionItoh et al., 1999
PAR1Convergence and extensionKusakabe and Nishida, 2004MigrationShi et al., 2004
PRKCIConvergence and extensionKusakabe and Nishida, 2004EMTGunaratne et al., 2013
CAP1Convergence and extensionSeifert et al., 2009MigrationYamazaki et al., 2009
EZREpibolyLink et al., 2006MigrationKhanna et al., 2004
EPCAMEpibolySlanchev et al., 2009Migration and invasionNi et al., 2012
ITGB1/ ITA5Mesodermal migrationSkalski et al., 1998Migration and invasionFelding-Habermann, 2003
FN1Convergence and extensionMarsden and DeSimone, 2003InvasionMalik et al., 2010
HAS2Dorsal migration of lateral cellsBakkers et al., 2004InvasionKim et al., 2004
MMP14Convergence and extensionCoyle et al., 2008InvasionPerentes et al., 2011
COX1EpibolyCha et al., 2006InvasionKundu and Fulton, 2002
PTGESConvergence and extensionSpeirs et al., 2010InvasionWang and Dubois, 2006
SLC39A6Anterior migrationYamashita et al., 2004EMTLue et al., 2011
GNA12 /13Convergence and extensionLin et al., 2005Migration and invasionYagi et al., 2011
OGTEpibolyWebster et al., 2009Migration and invasionLynch et al., 2012
CCN1Cell movementLatinkic et al., 2003Migration and invasionLin et al., 2012
TRPM7Convergence and extensionLiu et al., 2011MigrationMiddelbeek et al., 2012
MAPKAPK2EpibolyHolloway et al., 2009MigrationKumar et al., 2010
B4GALT1Convergence and extensionMachingo et al., 2006InvasionZhu et al., 2005
IER2Convergence and extensionHong et al., 2011MigrationNeeb et al., 2012
TIP1Convergence and extensionBesser et al., 2007Migration and invasionHan et al., 2012
PAK5Convergence and extensionFaure et al., 2005MigrationGong et al., 2009
MARCKSConvergence and extensionIioka et al., 2004Migration and invasionRombouts et al., 2013

Here, we report a unique screening concept based on the hypothesis. Pizotifen, an antagonist for HTR2C, was identified from the screen as a ‘hit’ that interrupted zebrafish gastrulation. A mouse model of metastasis confirmed pharmacological and genetic inhibition of HTR2C suppressed metastatic progression. Moreover, HTR2C induced EMT and promoted metastatic dissemination of non-metastatic cancer cells in a zebrafish xenotransplantation model. These results demonstrated that this concept could offer a novel high-throughput platform for discovery of anti-metastasis drugs and can be converted to a chemical genetic screening platform.

Results

Small molecules interrupting epiboly of zebrafish have a potential to suppress metastatic progression of human cancer cells

Before performing a screening assay, we validated a core of our concept through comparing the genes expressed in zebrafish gastrulation with the genes which expressed in EMT-mediated metastasis. Gene set enrichment analysis (GSEA) demonstrated that 50%-epiboly, shield, and 75%-epiboly stage of zebrafish embryos expressed the genes which promote EMT-mediated metastasis: EMT induction, TGF-β signaling, wnt/β-catenin signaling, Notch signaling (Figure 1—figure supplement 1).

We further conducted preliminary experiments to test the hypothesis. First, we examined whether hindering the molecular function of reported genes, whose knockdown induced gastrulation defects in zebrafish, might suppress cell motility and invasion of cancer cells. We chose protein arginine methyltransferase 1 (PRMT1) and cytochrome P450 family 11 (CYP11A1), both of whose knockdown induced gastrulation defects in zebrafish but whose involvement in metastatic progression is unclear (Tsai et al., 2011; Hsu et al., 2006). Elevated expression of PRMT1 and CYP11A1 was observed in highly metastatic human breast cancer cell lines and knockdown of these genes through RNA interference suppressed the motility and invasion of MDA-MB-231 cells without affecting their viability (Figure 1—figure supplement 2A-C).

Next, we conducted an inverse examination of whether chemicals which were reported to suppress metastatic dissemination of cancer cells could interrupt epiboly progression of zebrafish embryos. Niclosamide and vinpocetine are reported to suppress metastatic progression (Weinbach and Garbus, 1969; Sack et al., 2011; Huang et al., 2012; Szilágyi et al., 2005). Either niclosamide- or vinpocetine-treated zebrafish embryos showed complete arrest at very early stages or severe delay in epiboly progression, respectively (Figure 1—figure supplement 2D).

These results suggest that epiboly could serve as a marker for this screening assay and epiboly-interrupting drugs that are identified through this screening could have the potential to suppress metastatic progression of human cancer cells.

132 FDA-approved drugs induced delayed in epiboly of zebrafish embryos

We screened 1280 FDA, EMA, or other agencies-approved drugs (Prestwick, Inc) in our zebrafish assay. The screening showed that 0.9% (12/1280) of the drugs, including antimycin A and tolcapone, induced severe or complete arrest of embryonic cell movement when embryos were treated with 10 μM. 5.2% (66/1280) of the drugs, such as dicumarol, racecadotril, pizotifen, and S(-)eticlopride hydrochloride, induced either delayed epiboly or interrupted epiboly of the embryos. 93.3% (1194/1280) of drugs have no effect on epiboly progression of the embryos. 0.6% (8/1280) of drugs induced toxic lethality. Epiboly progression was affected more severely when embryos were treated with 50 μM; 1.7% (22/1280) of the drugs induced severe or complete arrest of it. 8.6% (110/1280) of the drugs induced either delayed epiboly or interrupt epiboly of the embryos. 4.3% (55/1280) of drugs induced a toxic lethality (Figure 1A and B, Table 2). Among the epiboly-interrupting drugs, several drugs have already been reported to inhibit metastasis-related molecular mechanisms: adrenosterone or zardaverine, which target HSD11β1 or PDE3 and -4, respectively, are reported to inhibit EMT (Nakayama et al., 2020; Kolosionek et al., 2009); racecadotril, which targets enkephalinase, is reported to confer metastatic potential on colon cancer cell (Sasaki et al., 2014); and disulfiram, which targets ALDH (aldehyde dehydrogenase), is reported to confer stem-like properties on metastatic cancer cells (Liu et al., 2013). This evidence suggests that epiboly-interrupting drugs have the potential for suppressing metastasis of human cancer cells.

Figure 1 with 2 supplements see all
A chemical screen for identification of epiboly-interrupting drugs.

(A) Cumulative results of the chemical screen in which each drug was used at either 10 µM (left) or 50 µM (right) concentrations. 1280 FDA, EMA, or other agencies-approved drugs were subjected to this screening. Positive ‘hit’ drugs were those that interrupted epiboly progression. (B) Representative samples of the embryos that were treated with indicated drugs.

Table 2
A list of the drugs that interfere with epiboly progression in zebrafish.

Related to Figure 1. A list of positive ‘hit’ drugs that interfered with epiboly progression. Gastrulation defects or status of each of the zebrafish embryos that were treated with either 10 or 50 μM concentrations are indicated.

Chemical nameChemical formulaEffect of 10 µMEffect of 50 µM
AcitretinC21H26O3DelayedDelayed
AdrenosteroneC19H24O3DelayedDelayed
AlbendazoleC12H15N3O2SSevere delayedSevere delayed
Alfadolone acetateC23H34O5DelayedDelayed
AlfaxaloneC21H32O3DelayedDelayed
AlprostadilC20H34O5DelayedDelayed
AltrenogestC21H26O2Slightly delayedDelayed
AmpiroxicamC20H21N3O7SNon-effectDelayed
Anethole-trithioneC10H8OS3DelayedDelayed
Antimycin AC28H40N2O9DelayedDelayed
AvobenzoneC20H22O3DelayedDelayed
BenzoxiquineC16H11NO2Non-effectDelayed
BosentanC27H29N5O6SDelayedDelayed
Butoconazole nitrateC19H18Cl3N3O3SDelayedToxic lethal
Camptothecine (S,+)C20H16N2O4Severe delayedSevere delayed
Carbenoxolone disodium saltC34H48Na2O7DelayedToxic lethal
CarmofurC11H16FN3O3Slightly delayedDelayed
CarprofenC15H12ClNO2Severe delayedToxic lethal
CefdinirC14H13N5O5S2DelayedDelayed
CelecoxibC17H14F3N3O2SDelayedDelayed
ChlorambucilC14H19Cl2NO2Slightly delayedDelayed
ChlorhexidineC22H30Cl2N10Non-effectToxic lethal
Ciclopirox ethanolamineC14H24N2O3DelayedSevere delayed
CinoxacinC12H10N2O5DelayedSevere delayed
ClofibrateC12H15ClO3Non-effectSevere delayed
ClopidogrelC16H16ClNO2SNon-effectDelayed
Clorgyline hydrochlorideC13H16Cl3NODelayedDelayed
ColchicineC22H25NO6Non-effectDelayed
Deptropine citrateC29H35NO8DelayedDelayed
Desipramine hydrochlorideC18H23ClN2DelayedDelayed
Diclofenac sodiumC14H10Cl2NNaO2DelayedSevere delayed
DicumarolC19H12O6DelayedSevere delayed
DiethylstilbestrolC18H20O2DelayedToxic lethal
Dimaprit dihydrochlorideC6H17Cl2N3SSlightly delayedDelayed
DisulfiramC10H20N2S4DelayedDelayed
Dopamine hydrochlorideC8H12ClNO2DelayedDelayed
Eburnamonine (-)C19H22N2ODelayedDelayed
Ethaverine hydrochlorideC24H30ClNO4DelayedDelayed
EthinylestradiolC20H24O2DelayedSevere delayed
Ethopropazine hydrochlorideC19H25ClN2SDelayedDelayed
EthoxyquinC14H19NONon-effectDelayed
ExemestaneC20H24O2Slightly delayedDelayed
EzetimibeC24H21F2NO3Slightly delayedDelayed
FenbendazoleC15H13N3O2SNon-effectDelayed
Fenoprofen calcium salt dihydrateC30H30CaO8Slightly delayedDelayed
FentiazacC17H12ClNO2SToxic lethalToxic lethal
FloxuridineC9H11FN2O5DelayedToxic lethal
Flunixin meglumineC21H28F3N3O7DelayedToxic lethal
FlutamideC11H11F3N2O3DelayedToxic lethal
Fluticasone propionateC25H31F3O5SNon-effectDelayed
FurosemideC12H11ClN2O5SDelayedDelayed
GatifloxacinC19H22FN3O4DelayedDelayed
GemcitabineC9H11F2N3O4DelayedDelayed
GemfibrozilC15H22O3DelayedToxic lethal
GestrinoneC21H24O2DelayedDelayed
HaloproginC9H4Cl3IODelayedToxic lethal
HexachloropheneC13H6Cl6O2DelayedSevere delayed
HexestrolC18H22O2Slightly delayedDelayed
IbudilastC14H18N2ONon-effectDelayed
Idazoxan hydrochlorideC11H13ClN2O2Slightly delayedDelayed
Idazoxan hydrochlorideC11H13ClN2O2Non-effectDelayed
IdebenoneC19H30O5Severe delayedToxic lethal
IndomethacinC19H16ClNO4Non-effectDelayed
IpriflavoneC18H16O3DelayedSevere delayed
IsotretinoinC20H28O2Non-effectSevere delayed
IsradipineC19H21N3O5Non-effectDelayed
LansoprazoleC16H14F3N3O2SSlightly delayedDelayed
LatanoprostC26H40O5Non-effectDelayed
LeflunomideC12H9F3N2O2DelayedSevere delayed
LetrozoleC17H11N5Non-effectDelayed
Lithocholic acidC24H40O3Non-effectDelayed
LodoxamideC11H6ClN3O6Non-effectDelayed
LofepramineC26H27ClN2ONon-effectDelayed
LoratadineC22H23ClN2O2DelayedDelayed
Loxapine succinateC22H24ClN3O5DelayedDelayed
MebendazoleC16H13N3O3Severe delayedSevere delayed
MebendazoleC22H26N2O2Non-effectDelayed
MeloxicamC14H13N3O4S2DelayedToxic lethal
MethiazoleC12H15N3O2SDelayedDelayed
MevastatinC23H34O5Non-effectDelayed
MK 801 hydrogen maleateC20H19NO4Slightly delayedDelayed
NabumetoneC15H16O2Non-effectSevere delayed
Naftopidil dihydrochlorideC24H30Cl2N2O3Slightly delayedDelayed
NandroloneC18H26O2DelayedDelayed
Naproxen sodium saltC14H13NaO3DelayedDelayed
NiclosamideC13H8Cl2N2O4DelayedDelayed
NifekalantC19H27N5O5DelayedDelayed
Niflumic acidC13H9F3N2O2DelayedDelayed
NimesulideC13H12N2O5SNon-effectDelayed
NisoldipineC20H24N2O6DelayedToxic lethal
NitazoxanideC12H9N3O5SSevere delayedSevere delayed
NorethindroneC20H26O2Non-effectDelayed
NorgestimateC23H31NO3Slightly delayedDelayed
OxfendazolC15H13N3O3SSlightly delayedDelayed
OxibendazolC12H15N3O3Severe delayedSevere delayed
OxymetholoneC21H32O3Slightly delayedDelayed
ParbendazoleC13H17N3O2Severe delayedSevere delayed
ParthenolideC15H20O3Non-effectDelayed
PenciclovirC10H15N5O3Non-effectDelayed
PentobarbitalC11H18N2O3Non-effectDelayed
Phenazopyridine hydrochlorideC11H12ClN5DelayedToxic lethal
PhenothiazineC12H9NSNon-effectDelayed
Phenoxybenzamine hydrochlorideC18H23Cl2NONon-effectDelayed
Pizotifen malateC23H27NO5SDelayedSevere delayed
Pramoxine hydrochlorideC17H28ClNO3Slightly delayedDelayed
Prilocaine hydrochlorideC13H21ClN2ONon-effectDelayed
PrimidoneC12H14N2O2Slightly delayedDelayed
RacecadotrilC21H23NO4SSlightly delayedDelayed
Riluzole hydrochlorideC8H6ClF3N2OSNon-effectDelayed
RitonavirC37H48N6O5S2Non-effectSevere delayed
S(-)Eticlopride hydrochlorideC17H26Cl2N2O3DelayedDelayed
SalmeterolC25H37NO4Non-effectDelayed
Streptomycin sulfateC42H84N14O36S3Non-effectDelayed
Sulconazole nitrateC18H16Cl3N3O3SDelayedDelayed
TegafurC8H9FN2O3DelayedDelayed
TelmisartanC33H30N4O2Severe delayedToxic lethal
TenatoprazoleC16H18N4O3SNon-effectDelayed
TerbinafineC21H25NNon-effectDelayed
ThimerosalC9H9HgNaO2SNon-effectDelayed
ThiorphanC12H15NO3SDelayedDelayed
TolcaponeC14H11NO5Severe delayedSevere delayed
TopotecanC23H23N3O5DelayedDelayed
Tracazolate hydrochlorideC16H25ClN4O2Severe delayedDelayed
TribenosideC29H34O6DelayedDelayed
TriclabendazoleC14H9Cl3N2OSDelayedDelayed
TriclosanC12H7Cl3O2DelayedSevere delayed
TrioxsalenC14H12O3DelayedDelayed
TroglitazoneC24H27NO5SSevere delayedToxic lethal
Valproic acidC8H16O2Non-effectDelayed
VoriconazoleC16H14F3N5ONon-effectDelayed
ZardaverineC12H10F2N2O3Slightly delayedDelayed
Zuclopenthixol dihydrochlorideC22H27Cl3N2OSDelayedDelayed

Identified drugs suppressed cell motility and invasion of human cancer cells

It has been reported that zebrafish have orthologues to 86% of 1318 human drug targets (Gunnarsson et al., 2008). However, it was not known whether the epiboly-interrupting drugs could suppress metastatic dissemination of human cancer cells. To test this, we subjected the 78 epiboly-interrupting drugs that showed a suppressor effect on epiboly progression at a 10 μM concentration to in vitro experiments using a human cancer cell line. The experiments examined whether the drugs could suppress cell motility and invasion of MDA-MB-231 cells through a Boyden chamber. Before conducting the experiment, we investigated whether these drugs might affect viability of MDA-MB-231 cells using an MTT assay. Out of the 78 drugs, 16 of them strongly affected cell viability at concentrations less than 1 μM and were not used in the cell motility experiments. The remaining 62 drugs were assayed in Boyden chamber motility experiments. Out of the 62 drugs, 20 of the drugs inhibited cell motility and invasion of MDA-MB-231 cells without affecting cell viability. Among the 20 drugs, hexachlorophene and nitazoxanide were removed since the primary targets of the drugs, D-lactate dehydrogenase and pyruvate ferredoxin oxidoreductase, are not expressed in mammalian cells. With the exception of ipriflavone, whose target is still unclear, the known primary targets of the remaining 17 drugs are reported to be expressed by mammalian cells (Figure 2A and Table 3).

Figure 2 with 2 supplements see all
Pizotifen, one of epiboly-interrupting drugs, suppressed metastatic dissemination of human cancer cells lines in vivo and vitro.

(A) Effect of the epiboly-interrupting drugs on cell motility and invasion of MBA-MB-231 cells. MBA-MB-231 cells were treated with vehicle or each of the epiboly-interrupting drugs and then subjected to Boyden chamber assays. Fetal bovine serum (1% v/v) was used as the chemoattractant in both assays. Each experiment was performed at least twice. (B) Western blot analysis of HTR2C levels (top) in a non-metastatic human cancer cell line, MCF7 (breast) and highly metastatic human cancer cell lines, MDA-MB-231 (breast), MDA-MB-435 (melanoma), PC9 (lung), MIA-PaCa2 (pancreas), PC3 (prostate), and SW620 (colon); GAPDH loading control is shown (bottom). (C) Effect of pizotifen on cell motility and invasion of MBA-MB-231, MDA-MB-435, and PC9 cells. Either vehicle or pizotifen treated the cells were subjected to Boyden chamber assays. Fetal bovine serum (1% v/v) was used as the chemoattractant in both assays. Each experiment was performed at least twice. (D) and (E) Representative images of dissemination of 231R, shLacZ 231R or shHTR2C 231R cells in zebrafish xenotransplantation model. The fish larvae that were inoculated with 231R cells were treated with either vehicle (top left) or the drug (lower left) (D). The fish larvae that were inoculated with either shLacZ 231R or shHTR2C 231R cells (lower left) (E). White arrows head indicate disseminated 231R cells. The images were shown in 4× magnification. Scale bar, 100 µm. The mean frequencies of the fish showing head, trunk, or end-tail dissemination were counted (graph on right). Each value is indicated as the mean ± SEM of two independent experiments. Statistical analysis was determined by Student’s t test.

Table 3
Primary targets of the identified drugs.
The identified drugsPrimary targets of the identified drugs
HexachloropheneD-Lactate dehydrogenase (D-LDH), not expressed in mammalian cells
TroglitazoneAgonist for peroxisome proliferator-activated receptor α and γ (PPARα and -γ)
Pizotifen malate5-Hydroxytryptamine receptor 2C (HTR2C)
SalmeterolAdrenergic receptor beta 2 (ADRB2)
NitazoxanidePyruvate ferredoxin oxidoreductase (PFOR), not expressed in mammalian cells
Valproic acidHistone deacetylases (HDACs)
DicumarolNAD(P)H dehydrogenase quinone 1 (NQO1)
Loxapine succinateDopamine receptor D2 and D4 (DRD2 and DRD4)
AdrenosteroneHydroxysteroid (11-beta) dehydrogenase 1 (HSD11β1)
Riluzole hydrochlorideGlutamate R andvoltage-dependent Na+ channel
Naftopidil dihydrochloride5-Hydroxytryptamine receptor 1A (HTR1A) andα1-adrenergic receptor (AR)
S(-)Eticlopride hydrochlorideDopamine receptor D2 (DRD2)
RacecadotrilMembrane metallo-endopeptidase (MME)
IpriflavoneUnknown
FlurbiprofenCyclooxygenase 1 and 2 (Cox1 and -2)
ZardaverinePhosphodiesterase III/IV (PDE3/4)
LeflunomideDihydroorotate dehydrogenase (DHODH)
OlmesartanAngiotensin II receptor alpha
DisulfiramAldehyde dehydrogenase (ALDH)Dopamine β-hydroxylase (DBH)
Zuclopenthixol dihydrochlorideDopamine receptors D1 and D2 (DRD1 and -2)

We confirmed that highly metastatic human cancer cell lines expressed target genes through western blotting analyses. Among the genes, serotonin receptor 2C (HTR2C), which is a primary target of pizotifen, was highly expressed in only metastatic cell lines (Figure 2B and Figure 2—figure supplement 2A). Clinical data also shows that that HTR2C expression is correlated with tumor stage of breast cancer patients and is higher in metastatic and Her2/neu-overexpressing tumors (Pai et al., 2009). Pizotifen suppressed cell motility and invasion of several highly metastatic human cancer cell lines in a dose-dependent manner (Figure 2C). Similarly, dopamine receptor D2 (DRD2), which is a primary target of S(-)eticlopride hydrochloride, was highly expressed in only metastatic cell lines, and the drug suppressed cell motility and invasion of these cells in a dose-dependent manner (Figure 2—figure supplement 2A-C).

These results indicate that a number of the epiboly-interrupting drugs also have suppressor effects on cell motility and invasion of highly metastatic human cancer cells.

Pizotifen suppressed metastatic dissemination of human cancer cells in a zebrafish xenotransplantation model

While a number of the epiboly-interrupting drugs suppressed cell motility and invasion of human cell lines in vitro, it was still unclear whether the drugs could suppress metastatic dissemination of cancer cells in vivo. Therefore, we examined whether the identified drugs could suppress metastatic dissemination of these human cancer cells in a zebrafish xenotransplantation model. Pizotifen was selected to test since HTR2C was overexpressed only in highly metastatic cell lines supporting the hypothesis that it could be a novel target for blocking metastatic dissemination of cancer cells (Figure 2B). Red fluorescent protein (RFP)-labelled MDA-MB-231 (231R) cells were injected into the duct of Cuvier of Tg (kdrl:eGFP) zebrafish at 2 dpf and then maintained in the presence of either vehicle or pizotifen. Twenty-four hours post injection, the numbers of fish showing metastatic dissemination of 231R cells were measured via fluorescence microscopy. In this model, the dissemination patterns were generally divided into three categories: (i) head dissemination, in which disseminated 231R cells exist in the vessel of the head part; (ii) trunk dissemination, in which the cells were observed in the vessel dilating from the trunk to the tail; (iii) end-tail dissemination, in which the cells were observed in the vessel of the end-tail part (Nakayama et al., 2020).

Three independent experiments revealed that the frequencies of fish in the drug-treated group showing head, trunk, or end-tail dissemination significantly decreased to 55.3% ± 7.5%, 28.5 ± 5.0%, or 43.5% ± 19.1% when compared with those in the vehicle-treated group; 95.8% ± 5.8%, 47.1 ± 7.7%, or 82.6% ± 12.7%. Conversely, the frequency of the fish in the drug-treated group not showing any dissemination significantly increased to 45.4% ± 0.5% when compared with those in the vehicle-treated group; 2.0% ± 2.9% (Figure 2D, Figure 2—figure supplement 2 and Table 4).

Table 4
Effects of pharmacological inhibition of HTR2C on metastatic dissemination of MDA-MB-231 cells in zebrafish xenografted models.

Related to Figure 2D. The numbers and frequencies of the fish showing the dissemination patterns in vehicle- or pizotifen-treated group were indicated. The fish showed both patterns of dissemination were redundantly counted in this analysis.

Experiment_#1Experiment_#2Experiment_#3Average of experiments
 Drug: VehicleCell: MDA-MB-231Non-dissemination0% n1 = 0/170% n2 = 0/126.66% n3 = 1/152.22% ± 3.84%
Head58.82% n1 = 10/1791.66% n2 = 11/126.66% n3 = 1/1572.38% ± 17.15%
Trunk52.94% n1 = 9/178.33% n2 = 1/1220% n3 = 2/1527.09% ± 23.13%
End-tail100% n1 = 17/17100% n2 = 12/1286.66% n3 = 13/1595.55% ± 7.69%
 Drug: PizotifenCell: MDA-MB-231Non-dissemination55% n1 = 11/2031.57% n2 = 6/1945.45 % n3 = 10/2244.01% ± 11.77%
Head5% n1 = 1/2031.57% n2 = 6/1918.18% n3 = 4/2218.25% ± 13.28%
Trunk5% n1 = 1/2010.52% n2 = 2/194.45% n3 = 1/226.69% ± 3.32%
End-tail45% n1 = 9/2057.89% n2 = 11/1950% n3 = 11/2250.96% ± 6.50%

Similar effects were observed in another xenograft experiments using an RFP-labelled human pancreatic cancer cell line, MIA-PaCa-2 (MP2R). In the drug-treated group, the frequencies of the fish showing head, trunk, or end-tail dissemination significantly decreased to 15.3% ± 6.7%, 6.2% ± 1.3%, or 41.1% ± 1.5%; conversely, the frequency of the fish not showing any dissemination significantly increased to 46.3% ± 8.9% when compared with those in the vehicle-treated group; 74.5% ± 11.1%, 18.9% ± 14.9%, 77.0% ± 9.0%, or 17.2% ± 0.7% (Figure 2—figure supplement 2A and Table 5).

Table 5
Effects of pharmacological inhibition of HTR2C on metastatic dissemination of Mia-PaCa2 cells in zebrafish xenografted models.

Related to Figure 4. The numbers and frequencies of the fish showing the dissemination patterns in vehicle- or pizotifen-treated group were indicated. The fish showed both patterns of dissemination were redundantly counted in this analysis.

Experiment_#1Experiment_#2Average of experiments
 Drug: VehicleCell: MIA-PaCa2Non-dissemination17.64% n1 = 3/1716.66% n2 = 2/1217.15% + 0.69%
Head82.35% n1 = 14/1766.66% n2 = 8/1274.50% + 11.09%
Trunk29.41% n1 = 5/178.33% n2 = 1/1218.87% + 14.90%
End-tail70.58% n1 = 12/1783.33% n2 = 10/1776.96% + 9.01
 Drug: PizotifenCell: MIA-PaCa2Non-dissemination40% n1 = 4/1052.63% n2 = 10/1946.31% + 8.93%
Head20% n1 = 2/1010.52% n2 = 2/1915.26% + 6.69%
Trunk10% n1 = 1/105.26% n2 = 1/197.63% + 3.34%
End-tail40% n1 = 4/1042.05% n2 = 8/1941.4% + 1.48%

To eliminate the possibility that the metastasis suppressing effects of pizotifen might result from off-target effects of the drug, we conducted validation experiments to determine whether knockdown of HTR2C would show the same effects. Sub-clones of 231R cells that expressed short hairpin RNA (shRNA) targeting either LacZ or HTR2C were injected into the fish at 2 dpf and the fish were maintained in the absence of drug. In the fish that were inoculated with shHTR2C 231R cells, the frequencies of the fish showing head, trunk, and end-tail dissemination significantly decreased to 6.7% ± 4.9%, 6.7% ± 0.7%, or 20.0% ± 16.5%; conversely, the frequency of the fish not showing any dissemination significantly increased to 80.0% ± 4.4% when compared with those that were inoculated with shLacZ 231R cells; 80.0% ± 27.1%, 20.0% ± 4.5%, 90.0% ± 7.7%, or 0% (Figure 2E and Table 6).

Table 6
Effects of genetic inhibition of HTR2C on metastatic dissemination of MDA-MB-231 cells in zebrafish xenografted models.

Related to Figure 2E. The numbers and frequencies of the fish showing the dissemination patterns in the zebrafish that were inoculated with either shLacZ or shHTR2C MDA-MB-231 cells were indicated. The fish showed both patterns of dissemination were redundantly counted in this analysis.

Experiment_#1Experiment_#2Average of experiments
 shLacZNon-dissemination0% n1 = 0/100% n2 = 0/100%
Head60% n1 = 6/10100% n2 = 10/1080%± 28.28%
Trunk30% n1 = 3/1010% n2 = 1/1020% ± 14.14%
End-tail80% n1 = 8/10100% n2 = 10/1090% ± 14.14
 shHTR2CNon-dissemination80% n1 = 12/1576.84% n2 = 14/1976.84 ± 4.46%
Head6.66% n1 = 1/1515.78% n2 = 3/1911.22% ± 6.45%
Trunk6.66% n1 = 1/155.26% n2 = 1/195.96% ± 0.99%
End-tail20% n1 = 3/1526.31% n2 = 5/1923.15% ± 4.46%

These results indicate that pharmacological and genetic inhibition of HTR2C suppressed metastatic dissemination of human cancer cells in vivo.

Pizotifen suppressed metastasis progression of a mouse model of metastasis

We examined the metastasis-suppressor effect of pizotifen in a mouse model of metastasis (Tao et al., 2008). Luciferase-expressing 4T1 murine mammary carcinoma cells were inoculated into the mammary fat pads (MFP) of female BALB/c mice. On day 2 post inoculation, the mice were randomly assigned to two groups and one group received once daily intraperitoneal injections of 10 mg/kg pizotifen while the other group received a vehicle injection. Bioluminescence imaging and tumor measurement revealed that the sizes of the primary tumors in pizotifen-treated mice were equal to those in the vehicle-treated mice on day 10 post inoculation. The primary tumors were resected after the analyses. Immunofluorescence (IF) staining also demonstrated that the percentage of Ki67-positive cells in the resected primary tumors of pizotifen-treated mice were the same as those of vehicle-treated mice (Figure 3A–C), additionally, both groups showed less than 1% cleaved caspase 3 positive cells (Figure 3—figure supplement 1). Therefore, no anti-tumor effect of pizotifen was observed on the primary tumor. After 70 days from inoculation, bioluminescence imaging detected light emitted in the lungs, livers, and lymph nodes of vehicle-treated mice but not those of pizotifen-treated mice (Figure 3C). Vehicle-treated mice formed 5–50 metastatic nodules per lung in all 10 mice analyzed; conversely, pizotifen-treated mice (n = 10) formed 0–5 nodules per lung in all 10 mice analyzed (Figure 3D). Histological analyses confirmed that metastatic lesions in the lungs were detected in all vehicle-treated mice; conversely, they were detected in only 2 of 10 pizotifen-treated mice and the rest of the mice showed metastatic colony formations around the bronchiole of the lung. In addition, 4 of 10 vehicle-treated mice exhibited metastasis in the liver and the rest showed metastatic colony formation around the portal tract of the liver. In contrast, none of 10 pizotifen-treated mice showed liver metastases and only half of the 10 mice showed metastatic colony formation around the portal tract (Figure 3E). These results indicate that pizotifen can suppress metastasis progression without affecting primary tumor growth.

Figure 3 with 1 supplement see all
Pizotifen suppressed metastatic progression in a mouse model of metastasis.

(A) Mean volumes (n = 10 per group) of 4T1 primary tumors formed in the mammary fat pad of either vehicle- or pizotifen-treated mice at day 10 post injection. (B) Ki67 expression level in 4T1 primary tumors formed in the mammary fat pad of either vehicle- or pizotifen-treated mice at day 10 post injection. The mean expression levels of Ki67 (n = 10 mice per group) were determined and were calculated as the mean ration of Ki67-positive cells to 4’,6-diamidino-2-phenylindole (DAPI) area. (C) Representative images of primary tumors on day 10 post injection (top panels) and metastatic burden on day 70 post injection (bottom panels) taken using an IVIS Imaging System. (D) Representative images of the lungs from either vehicle- (top) or pizotifen-treated mice (bottom) at 70 days post tumor inoculation. Number of metastatic nodules in the lung of either vehicle- or pizotifen-treated mice (right). (E) Representative hematoxylin and eosin (H&E) staining of the lung (top) and liver (bottom) from either vehicle- or pizotifen-treated mice. Black arrow heads indicate metastatic 4T1 cells. (F) The mean number of metastatic lesions in step sections of the lungs from the mice that were inoculated with 4T1-12B cells expressing short hairpin RNA (shRNA) targeting for either LacZ or HTR2C. (G) Representative H&E staining of the lung and liver from the mice that were inoculated with 4T1-12B cells expressing shRNA targeting for either LacZ or HTR2C. Black arrow heads indicate metastatic 4T1 cells. Each value is indicated as the mean ± SEM. Statistical analysis was determined by Student’s t test.

To eliminate the possibility that the metastasis suppressing effects of pizotifen might result from off-target effects, we conducted validation experiments to determine whether knockdown of HTR2C would show the same effects. The basic experimental process followed the experimental design described above except that sub-clones of 4T1 cells that expressed shRNA targeting either LacZ or HTR2C were injected into the MFP of female BALB/c mice and the mice were maintained without drug. Histological analyses revealed that all of the mice (n = 5) that were inoculated with 4T1 cells expressing shRNA targeting LacZ showed metastases in the lungs. The mean number of metastatic lesions in a lung was 26.4 ± 7.8. In contrast, only one of the mice (n = 5) were inoculated with 4T1 cells expressing shRNA targeting HTR2C showed metastases in the lungs and the rest of the mice showed metastatic colony formation around the bronchiole of the lung. The mean number of metastatic lesions in the lung significantly decreased to 10% of those of mice that were inoculated with 4T1 cells expressing shRNA targeting LacZ (Figure 3F–H).

Taken together, pharmacological and genetic inhibition of HTR2C showed an anti-metastatic effect in the 4T1 model system.

HTR2C promoted EMT-mediated metastatic dissemination of human cancer cells

Although pharmacological and genetic inhibition of HTR2C inhibited metastasis progression, a role for HTR2C on metastatic progression has not been reported. Therefore, we examined whether HTR2C could confer metastatic properties on poorly metastatic cells.

First, we established a stable sub-clone of MCF7 human breast cancer cells expressing either vector control or HTR2C. Vector control expressing MCF7 cells maintained highly organized cell-cell adhesion and cell polarity; however, HTR2C-expressing MCF7 cells led to loss of cell-cell contact and cell scattering. The cobblestone-like appearance of these cells was replaced by a spindle-like, fibroblastic morphology. Western blotting and IF analyses revealed that HTR2C-expressing MCF7 cells showed loss of E-cadherin and EpCAM, and elevated expressions of N-cadherin, vimentin, and an EMT-inducible transcriptional factor ZEB1. Similar effects were validated through another experiment using an immortal keratinocyte cell line, HaCaT cells, in that HTR2C-expressing HaCaT cells also showed loss of cell-cell contact and cell scattering with loss of epithelial markers and gain of mesenchymal markers (Figure 4A–C and Figure 4—figure supplement 1A). Therefore, both the morphological and molecular changes in the HTR2C-expressing MCF7 and HaCaT cells demonstrated that these cells had undergone an EMT.

Figure 4 with 1 supplement see all
HTR2C induced epithelial-to-mesenchymal transition (EMT)-mediated metastatic dissemination of human cancer cells.

(A) The morphologies of the MCF7 and HaCaT cells expressing either the control vector or HTR2C were revealed by phase contrast microscopy. (B) Immunofluorescence staining of E-cadherin, EpCAM, vimentin, and N-cadherin expressions in the MCF7 cells from A. (C) Expression of E-cadherin, EpCAM, vimentin, N-cadherin, and HTR2C was examined by western blotting in the MCF7 and HaCaT cells; GAPDH loading control is shown (bottom). (D) Effect of HTR2C on cell motility and invasion of MCF7 cells. MCF7 cells were subjected to Boyden chamber assays. Fetal bovine serum (1% v/v) was used as the chemoattractant in both assays. Each experiment was performed at least twice. (E) Representative images of dissemination patterns of MCF7 cells expressing either the control vector (top left) or HTR2C (lower left) in a zebrafish xenotransplantation model. White arrow heads indicate disseminated MCF7 cells. The mean frequencies of the fish showing head, trunk, or end-tail dissemination tabulated (right). Each value is indicated as the mean ± SEM of two independent experiments. Statistical analysis was determined by Student’s t test.

Next, we examined whether HTR2C-driven EMT could promote metastatic dissemination of human cancer cells. Boyden chamber assay revealed that HTR2C expressing MCF7 cells showed an increased cell motility and invasion compared with vector control-expressing MCF7 cells in vitro (Figure 4D). Moreover, we conducted in vivo examination of whether HTR2C expression could promote metastatic dissemination of human cancer cells in a zebrafish xenotransplantation model. RFP-labelled MCF7 cells expressing either vector control or HTR2C were injected into the duct of Cuvier of Tg (kdrl:eGFP) zebrafish at 2 dpf. Twenty-four hours post injection, the frequencies of the fish showing metastatic dissemination of the inoculated cells were measured using fluorescence microscopy. In the fish that were inoculated with HTR2C expressing MCF7 cells, the frequencies of the fish showing head, trunk, and end-tail dissemination significantly increased to 96.7% ± 4.7%, 68.8% ± 6.4%, or 89.5% ± 3.4%; conversely, the frequency of the fish not showing any dissemination decreased to 0% when compared with those in the fish that were inoculated with vector control expressing MCF7 cells; 33.1% ± 18.5%, 0%, 56.9% ± 4.4%, or 43% (Figure 4E, Figure 4—figure supplement 1B and Table 7).

Table 7
Effects of HTR2C overexpression on metastatic dissemination of MCF7 cells in zebrafish xenografted models.

Related to Figure 4E. The numbers and frequencies of the fish showing the dissemination patterns in the zebrafish that were inoculated with MCF7 cells expressing either vector control (VC) or HTR2C were indicated. The fish showed both patterns of dissemination were redundantly counted in this analysis.

Experiment_#1Experiment_#2Average of experiments
VCNon-dissemination46.15% n1 = 6/1340% n2 = 4/1043.07% ± 4.35%
Head46.15% n1 = 6/1320% n2 = 2/1033.07% ± 18.49%
Trunk0% n1 = 0/130% n2 = 0/100%
End-tail53.84% n1 = 7/1360% n2 = 6/1056.92% ± 4.35%
 HTR2CNon-dissemination0% n1 = 0/140% n2 = 0/150%
Head100% n1 = 14/1493.33% n2 = 14/1596.66% ± 4.71%
Trunk64.28% n1 = 9/1473.33% n2 = 11/1568.80% ± 6.39%
End-tail85.71% n1 = 12/1493.33% n2 = 14/1589.52% ± 5.38%

These results indicated that HTR2C promoted metastatic dissemination of cancer cells through induction of EMT, and suggest that the screen can easily be converted to a chemical genetic screening platform.

Pizotifen induced mesenchymal-to-epithelial transition through inhibition of Wnt signaling

Finally, we elucidated the mechanism of action of how pizotifen suppressed metastasis, especially metastatic dissemination of cancer cells. Our results showed that HTR2C induced EMT and that pharmacological and genetic inhibition of HTR2C suppressed metastatic dissemination of MDA-MB-231 cells that had already transitioned to mesenchymal-like traits via EMT. Therefore, we speculated that blocking HTR2C with pizotifen might inhibit the molecular mechanisms which follow EMT induction. We first investigated the expressions of epithelial and mesenchymal markers in pizotifen-treated MDA-MB-231 cells since the activation of an EMT program needs to be transient and reversible, and transition from a fully mesenchymal phenotype to a epithelial-mesenchymal hybrid state or a fully epithelial phenotype is associated with malignant phenotypes (Kröger et al., 2019). IF and FACS analyses revealed 20% of pizotifen-treated MDA-MB-231 cells restored E-cadherin expression. Also, western blotting analysis demonstrated that 4T1 primary tumors from pizotifen-treated mice has elevated E-cadherin expression compared with tumors from vehicle-treated mice (Figure 5A–C and Figure 5—figure supplement 1). However, mesenchymal markers did not change between vehicle and pizotifen-treated MDA-MB-231 cells (data not shown). We further analyzed E-cadherin-positive (E-cad+) cells in pizotifen-treated MDA-MB-231 cells. The E-cad+ cells showed elevated expressions of epithelial markers KRT14 and KRT19; and decreased expression of mesenchymal makers vimentin, MMP1, MMP3, and S100A4. Recent research reports that an EMT program needs to be transient and reversible and that a mesenchymal phenotype in cancer cells is achieved by constitutive ectopic expression of ZEB1. In accordance with the research, the E-cad+ cells and 4T1 primary tumors from pizotifen-treated mice had decreased ZEB1 expression compared with vehicle-treated cells and tumors from vehicle-treated mice (Figure 5D and Figure 5—figure supplement 2). In contrast, HTR2C-expressing MCF7 and HuMEC cells expressed ZEB1 but not vehicle control MCF7 and HuMEC cells (Figure 4C and Figure 5—figure supplement 3). HTR2C-expressing MCF7 cells expressed not only ZEB1 but also Twist1 and Snail. In contrast, pizotifen-treated MDA-MB-231 cells showed decreased expression of ZEB1 and Twist1 compared with that in vehicle-treated cells. Furthermore, in the primary tumors of pizotifen-treated mice, only ZEB1 expression was decreased compared with those of vehicle-treated mice. These results indicate that HTR2C-mediated signaling induced EMT through up-regulation of ZEB1 and blocking HTR2C with pizotifen induced mesenchymal-to-epithelial transition through down-regulation of ZEB1 (Figure 5—figure supplement 4).

Figure 5 with 5 supplements see all
Pizotifen restored mesenchymal-like traits of MDA-MB-231 cells into epithelial traits through blocking nuclear accumulation of β-catenin.

(A) Immunofluorescence (IF) staining of E-cadherin in either vehicle- or pizotifen-treated MDA-MB-231 cells. (B) Surface expression of E-cadherin in either vehicle (black)- or pizotifen (red)-treated MDA-MB-231 cells by FACS analysis. Non-stained controls are shown in gray. (C) Protein expressions levels of E-cadherin, ZEB1, and β-catenin in the cytoplasm and nucleus of 4T1 primary tumors from either vehicle- or pizotifen-treated mice are shown; Luciferase, histone H3, and β-tubulin are used as loading control for whole cell, nuclear, or cytoplasmic lysate, respectively. (D) Protein expression levels of epithelial and mesenchymal markers and ZEB1 in either vehicle- or pizotifen-treated MDA-MB-231 cells or E-cadherin-positive (E-cad+) cells in pizotifen-treated MDA-MB-231 cells are shown. (E) IF staining of β-catenin in the MCF7 cells expressing either vector control (top left, bottom left) or HTR2C (top right, bottom right). (F) Expressions of β-catenin in the cytoplasm and nucleus of MCF7 cells. (G) IF staining of β-catenin in either vehicle (top left, bottom left) or pizotifen-treated MDA-MB-231 cells (top right, bottom right). (H) Expressions of β-catenin in the cytoplasm and nucleus of MDA-MB-231 cells and the E-cad+ cells.

We further investigated the mechanism of action of how blocking HTR2C with pizotifen induced down-regulation of ZEB1. In embryogenesis, serotonin-mediated signaling is required for Wnt-dependent specification of the superficial mesoderm during gastrulation (Beyer et al., 2012). Wnt signaling plays critical role in inducing EMT. In cancer cells, overexpression of HTR1D is associated with Wnt signaling (Sui et al., 2015; Zhan et al., 2017). This evidence led to a hypothesis that HTR2C-mediated signaling might turn on transcriptional activity of β-catenin and that might induce up-regulation of EMT-TFs. IF analyses revealed β-catenin was accumulated in the nucleus of HTR2C-expressing MCF7 cells but it was located in the cytoplasm of vector control-expressing cells (Figure 5E). Nuclear accumulation of β-catenin in HTR2C-expressing MCF7 cells was confirmed by western blot (Figure 5F and Figure 5—figure supplement 2). In contrast, pizotifen-treated MDA-MB-231 cells showed β-catenin located in the cytoplasm of the cells. Vehicle-treated cells showed that β-catenin accumulated in the nucleus of the cells. (Figure 5G), and western blotting analysis confirmed that it was located in the cytoplasm of pizotifen-treated MDA-MB-231 cells (Figure 5H and Figure 5—figure supplement 5). Furthermore, immunohistochemistry and western blotting analyses showed that β-catenin accumulated in the nucleus, and phospho-GSKβ and ZEB1 expression were decreased in 4T1 primary tumors from pizotifen-treated mice compared with vehicle-treated mice (Figure 5C and Figure 5—figure supplement 1). These results indicated that HTR2C would regulate transcriptional activity of β-catenin and pizotifen could inhibit it.

Taken together, we conclude that blocking HTR2C with pizotifen restored epithelial properties to metastatic cells (MDA-MB-231 and 4T1 cells) through a decrease of transcriptional activity of β-catenin and that suppressed metastatic progression of the cells.

Discussion

Reducing or eliminating mortality associated with metastatic disease is a key goal of medical oncology, but few models exist that allow for rapid, effective screening of novel compounds that target the metastatic dissemination of cancer cells. Based on accumulated evidence that at least 50 genes play an essential role in governing both metastasis and gastrulation progression (Table 1), we hypothesized that small molecule inhibitors that interrupt gastrulation of zebrafish embryos might suppress metastatic progression of human cancer cells. We created a unique screening concept utilizing gastrulation of zebrafish embryos to test the hypothesis. Our results clearly confirmed our hypothesis: 25.6% (20/76 drugs) of epiboly-interrupting drugs could also suppress cell motility and invasion of highly metastatic human cell lines in vitro. In particular, pizotifen, which is an antagonist for serotonin receptor 2C and one of the epiboly-interrupting drugs, could suppress metastasis in a mouse model (Figure 3A–E). Thus, this screen could offer a novel platform for discovery of anti-metastasis drugs.

Among the 20 drugs which suppressed both epiboly progression and cell motility and invasion of MDA-MB-231 cells, hexachlorophene and troglitazone showed the strongest effect on suppressing cell motility and invasion of MDA-MB-231 cells. However, the drug could not suppress cell motility and invasion of other highly metastatic human cancer cell lines: MDA-MB-435 and PC3. With the exception of pizotifen and S(-)eticlopride hydrochloride, the remaining 18 drugs could not show the suppressor effect on more than three highly metastatic human cancer cell lines. These results indicate that the strength of interrupting effect of a drug on epiboly progression is not proportional to the strength of suppressing effect of the drug on metastasis.

We have provided the first evidence that HTR2C, which is a primary target of pizotifen, induced EMT and promoted metastatic dissemination of cancer cells (Figure 4A–E). Clinical data shows that HTR2C expression is correlated with tumor stage of breast cancer patients and is higher in metastatic and Her2/neu-overexpressing tumors (Pai et al., 2009). That would support our finding.

Pharmacological inhibition of DRD2 with S(-)eticlopride hydrochloride suppressed cell invasion and migration of multiple human cancer cell lines in vitro. However, overexpression of DRD2 could not induce EMT on MCF7 cells. Therefore, we stopped focusing on DRD2 and S(-)eticlopride hydrochloride.

There are at least two advantages to the screen described herein. One is that the screen can easily be converted to a chemical genetic screening platform. Indeed, our screen succeeded to identify HTR2C as an EMT inducer (Figure 4A–E). In this research, 1280 FDA approval drugs were screened, this is less than a few percent of all of druggable targets (approximately 100 targets) in the human proteome in the body. If chemical genetic screening using specific inhibitor libraries were conducted, more genes that contribute to metastasis and gastrulation could be identified. The second advantage is that the screen enables one researcher to test 100 drugs in 5 hr with using optical microscopy, drugs, and zebrafish embryos. That indicates this screen is not only highly efficient, low-cost, and low-labor but also enables researchers who do not have high-throughput screening instruments to conduct drug screening for anti-metastasis drugs.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Zebrafish)AB lineNational University of Singapore
Strain, strain background (Zebrafish)Tg (kdrl:eGFP) zebrafishProvided by Dr Stainier
Strain, strain background (Mus musculus)BALB/cCharles River Laboratories
Cell line (Homo sapiens)MDA-MB-231ATCCHTB-26
Cell line (Homo sapiens)MCF7ATCCHTB-22
Cell line (Homo sapiens)MDA-MB-435ATCCHTB-129
Cell line (Homo sapiens)MIA-PaCa2ATCCCRM-CRL-1420
Cell line (Homo sapiens)PC3ATCCCRL-3471
Cell line (Homo sapiens)SW620ATCCCCL-227
Cell line (Homo sapiens)PC9RIKEN BRCRCB0446
Cell line (Homo sapiens)HaCaTCLI300493
Cell line (BALB/c Mus)4T1-12BProvided from Dr Gary Sahagian
AntibodyPRMT1 (A33)(Rabbit polyclonal)Cell Signaling TechnologyCat#_2449WB (1:1000)
AntibodyCYP11A1 (D8F4F)(Rabbit polyclonal)Cell Signaling TechnologyCat#_14217WB (1:1000)
AntibodyE-cadherin (4A2)(Mouse monoclonal)Cell Signaling TechnologyCat#_14472WB (1:1000)IF (1:100)
AntibodyEpCAM (VU1D9)(Mouse monoclonal)Cell Signaling TechnologyCat#_2929WB (1:1000)IF (1:100)
AntibodyVimentin (D21H3)(Rabbit polyclonal)Cell Signaling TechnologyCat#_5741WB (1:1000)IF (1:100)
AntibodyN-cadherin (D4R1H)(Rabbit polyclonal)Cell Signaling TechnologyCat#_13116WB (1:1000)IF (1:100)
AntibodyZEB1 (D80D3)(Rabbit polyclonal)Cell Signaling TechnologyCat#_3396WB (1:1000)
AntibodyHistone H3 (D1H2)(Rabbit polyclonal)Cell Signaling TechnologyCat#_4499WB (1:1000)
Antibodyβ-Tubulin (9F3)(Rabbit polyclonal)Cell Signaling TechnologyCat#_2128WB (1:1000)
AntibodyGAPDH (14C10)(Rabbit polyclonal)Cell Signaling TechnologyCat#_2118WB (1:1000)
AntibodyHTR2C (ab133570)(Rabbit polyclonal)AbcamCat#_ab133570WB (1:1000)
AntibodyDRD2 (ab85367)(Rabbit polyclonal)AbcamCat#_ab85367WB (1:1000)
AntibodyPhospho-GSK3β (Ser9) (F-2)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-373800WB (1:1000)
AntibodyGSK3β (1F7)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-53931WB (1:1000)
AntibodyKRT18 (DC-10)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-6259WB (1:1000)
AntibodyKRT19 (A53-B/A2)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-6278WB (1:1000)
AntibodyMMP1 (3B6)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-21731WB (1:1000)
AntibodyMMP2 (8B4)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-13595WB (1:1000)
AntibodyS100A4 (A-7)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-377059WB (1:1000)
AntibodyLuciferase (C-12)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-74548WB (1:1000)
Antibodyki67 (ki-67)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-23900WB (1:1000)
Antibodyβ-Catenin (E-5)(Mouse monoclonal)Santa Cruz BiotechnologyCat#_sc-7963WB (1:1000)IF (1:100)
AntibodyFITC-conjugated E-cadherin antibody (67A4)BiolegendCat#_324104FACS (1:100)
AntibodyAnti-mouse anti-rabbit immunoglobulin G (IgG) antibodies conjugated to Alexa Fluor 488Life TechnologiesA-11029IF (1:100)
AntibodyAnti-goat anti-rabbit immunoglobulin G (IgG) antibodies conjugated to Alexa Fluor 488Life TechnologiesA-11034IF (1:100)
Recombinant DNA reagentpLVX-shRNA1ClontechCat#_ 632,177
Recombinant DNA reagentpCDH-CMV-MCS-EF1α-HygroSystem BiosciencesCat#_CD515B-1Gene expression vector
Recombinant DNA reagentpMDLg/pRREAddgeneAddgene Plasmid #12251 RRID:Addgene_12251Lentivirus packaging vector
Recombinant DNA reagentpRSV-revAddgeneAddgene Plasmid #12253 RRID:Addgene_12253Lentivirus packaging vector
Recombinant DNA reagentpMD2.GAddgeneAddgene Plasmid #12259 RRID:Addgene_12259Lentivirus packaging vector
Recombinant DNA reagentProviding pCMV-h5TH2C-VSVProvided from Dr Herrick
Chemical compound, drugFDA-approved chemical librariesPrestwick Chemical
Chemical compound, drugPizotifenSanta Cruz BiotechnologyCat#_sc-201143
Chemical compound, drugS(-)Eticlopride hydrochlorideSanta Cruz BiotechnologyCat#_E101
Software, algorithmGraphPad Prism7GraphPad Software IncRRID:SCR_002798Data analysis
Software, algorithmFlowJoBD BiosciencesRRID:SCR_008520FACS data analysis

Zebrafish embryo screening

Request a detailed protocol

Zebrafish embryos at two-cell stage were collected at 20 min after their fertilization. Each drug was added to a well of a 24-well plate containing approximately 20 zebrafish embryos per well in either 10 or 50 μM final concentration when the embryos reached the sphere stage. Chemical treatment was initiated at 4 hpf and approximately 20 embryos were treated with two different concentrations for each compound tested. The treatment was ended at 9 hpf when vehicle- (DMSO) treated embryos as control reach 80–90% completion of the epiboly stage. The compounds which induced delay (<50% epiboly) in epiboly were selected as hit compounds for in vitro testing using highly metastatic human cancer cell lines. The study protocol was approved by the Institutional Animal Care and Use Committee of the National University of Singapore (protocol number: R16-1068).

Reagents

FDA, EMA, and other agencies-approved chemical libraries were purchased from Prestwick Chemical (Illkirch, France). Pizotifen (sc-201143) and S(-)eticlopride hydrochloride (E101) were purchased from Santa Cruz (Dallas, TX) and Sigma-Aldrich (St Louis, MO).

Cell culture and cell viability assay

Request a detailed protocol

MCF7, MDA-MB-231, MDA-MB-435, MIA-PaCa2, PC3, SW620, PC9, and HaCaT cells were obtained from American Type Culture Collection (ATCC, Manassas, VA). Luciferase-expressing 4T1 (4T1-12B) cells were provided from Dr Gary Sahagian (Tufts University, Boston, MA). All culture methods followed the supplier’s instruction. Cell viability assay was performed as previously described (Nakayama et al., 2020). PCR-based mycoplasma testing on these cells was performed once in 4 months.

Plasmid

Request a detailed protocol

A DNA fragment coding for HTR2C was amplified by PCR with primers containing restriction enzyme recognition sequences. The HTR2C coding fragment was amplified from hsp70l:mCherry-T2A-CreERT2 plasmid (Huang et al., 2012).

Immunoblotting

Request a detailed protocol

Western blotting was performed as described previously (Nakayama et al., 2020). Raw data of images of western blotting analyses are uploaded as source data for western. Anti-PRMT1 (A33), anti-CYP11A1 (D8F4F), anti-E-cadherin (4A2), anti-EpCAM (VU1D9), anti-vimentin (D21H3), anti-N-cadherin (D4R1H), anti-ZEB1 (D80D3), anti-histone H3 (D1H2), anti-β-tubulin (9F3), and anti-GAPDH (14C10) antibodies were purchased from Cell Signaling Technology (Danvers, MA). Anti-HTR2C (ab133570) and anti-DRD2 (ab85367) antibodies were purchased form Abcam (Cambridge, UK). Anti-phospho-GSK3β (Ser9) (F-2), anti-GSK3β (1F7), anti-KRT18 (DC-10), anti-KRT19 (A53-B/A2), anti-MMP1 (3B6), anti-MMP2 (8B4), anti-S100A4 (A-7), anti-luciferase (C-12), anti-ki67 (ki-67), and anti-β-catenin (E-5) antibodies were purchased from Santa Cruz Biotechnology (Dallas, TX).

Flow cytometry

Request a detailed protocol

Cells were stained with FITC-conjugated E-cadherin antibody (67A4, Biolegend, San Diego, CA). Flow cytometry was performed as described (Nakayama et al., 2009) and analyzed with FlowJo software (TreeStar, Ashland, OR).

shRNA-mediated gene knockdown

Request a detailed protocol

The shRNA-expressing lentivirus vectors were constructed using pLVX-shRNA1 vector (632177, TAKARA Bio, Shiga, Japan). PRMT1-shRNA_#3-targeting sequence is GTGTTCCAGTATCTCTGATTA; PRMT1-shRNA_#4-targeting sequence is TTGACTCCTACGCACACTTTG. CYP11A1-shRNA_#4-targeting sequence is GCGATTCATTGATGCCATCTA; CYP11A1-shRNA_#4-targeting sequence is GAAATCCAACACCTCAGCGAT. Human HTR2C-shRNA-targeting sequence is TCATGCACCTCTGCGCTATAT. Mouse HTR2C-shRNA-targeting sequence is CTTCATACCGCTGACGATTAT. LacZ-shRNA-targeting sequence is CTACACAAATCAGCGATT.

Immunofluorescence

Request a detailed protocol

IF microscopy assay was performed as previously described (Nakayama et al., 2020). Goat anti-mouse and goat anti-rabbit immunoglobulin G (IgG) antibodies conjugated to Alexa Fluor 488 (A-11029 and A-11034, Life Technologies, Carlsbad, CA) and diluted at 1:100 were used. Nuclei were visualized by the addition of 2 μg/ml of 4’,6-diamidino-2-phenylindole (DAPI) (62248, Thermo Fisher, Waltham, MA) and photographed at 100× magnification by a fluorescent microscope BZ-X700 (KEYENCE, Osaka, Japan).

Boyden chamber cell motility and invasion assay

Request a detailed protocol

These assays were performed as previously described (Nakayama et al., 2020). In Boyden chamber assay, either 3 × 105 MDA-MB-231, 1 × 106 MDA-MB-435 or 5 × 105 PC9 cells were applied to each well in the upper chamber.

Zebrafish xenotransplantation model

Request a detailed protocol

Tg(kdrl:eGFP) zebrafish was provided by Dr Stainier (Max Planck Institute for Heart and Lung Research). Embryos that were derived from the line were maintained in E3 medium containing 200 μM 1-phenyl-2-thiourea (P7629, Sigma-Aldrich, St Louis, MO). Approximately 100–400 RFP-labelled MBA-MB-231 or MIA-PaCa2 cells were injected into the duct of Cuvier of the zebrafish at 2 dpf. The fish were randomly assigned to two groups. One group was maintained in the presence of pizotifen-containing E3 medium and the other group was maintained in vehicle-containing E3 medium.

Spontaneous metastasis mouse model

Request a detailed protocol

4T1-12B cells (2 × 104) were injected into the #4 MFP while the mice were anesthetized. To monitor tumor growth and metastases, mice were imaged biweekly by IVIS Imaging System (ParkinElmer, Waltham, MA). The primary tumor was resected 10 days after inoculation. D-Luciferin Potassium Salt (LUCK-100) was purchased from GoldBio (St Louis, MO). The study protocol (protocol number: BRC IACUC #110612) was approved by A*STAR (Agency for Science, Technology and Research, Singapore).

Gene set enrichment analysis

Request a detailed protocol

Gene expression profiles obtained from zebrafish embryos at either 50%-epiboly, shield, or 75%-epiboly stage were analyzed based on the hallmark gene sets derived from the Molecular Signatures Database (MSigDB) (Subramanian et al., 2005; Liberzon et al., 2015). The zebrafish transcriptomic data was sourced from White et al., 2017. Gene sets that were significantly enriched (FDR < 0.25) were presented with the normalized enrichment score (NES) and nominal p value. Source data files for analysis of either gene expression and enriched pathways are uploaded as GSEA Source data 1 and 2, respectively.

Histological analysis

Request a detailed protocol

All OCT-embedded primary tumors, lungs, and livers of mice from the spontaneous metastasis 4T1 model were sectioned on a cryostat. Eight micron sections were taken at 500 µm intervals through the entirety of the livers and lungs. Sections were subsequently stained with hematoxylin and eosin. Metastatic lesions were counted under a microscope in each section for both lungs and livers.

Statistics

Data were analyzed by Student’s t test; p < 0.05 was considered significant.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting files.

References

    1. Kundu N
    2. Fulton AM
    (2002)
    Selective cyclooxygenase (COX)-1 or COX-2 inhibitors control metastatic disease in a murine model of breast cancer
    Cancer Research 62:2343–2346.
    1. Shi X
    2. Gangadharan B
    3. Brass LF
    4. Ruf W
    5. Mueller BM
    (2004)
    Protease-activated receptors (PAR1 and PAR2) contribute to tumor cell motility and metastasis
    Molecular Cancer Research 2:395–402.
    1. Yoshioka K
    2. Nakamori S
    3. Itoh K
    (1999)
    Overexpression of small GTP-binding protein RhoA promotes invasion of tumor cells
    Cancer Research 59:2004–2010.

Decision letter

  1. Yasuhito Shimada
    Reviewing Editor; Mie University, Japan
  2. Didier YR Stainier
    Senior Editor; Max Planck Institute for Heart and Lung Research, Germany

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "A chemical screen based on an interruption of zebrafish gastrulation identifies the HTR2C inhibitor Pizotifen as a suppressor of EMT-mediated metastasis" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Didier Stainier as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Three reviewers basically agree that your paper should be published in eLife. And I also think it is a very unique and interesting paper suitable for eLife. However, two reviewers think that additional data are necessary to strengthen your conclusions. I would be very grateful if you could perform these experiments in order to publish your work with more impact.

Comprehensive analysis using sequencing data is needed to support the concept that human cancer metastasis mimic/recapitulate zebrafish gastrulation. In addition, Pizotifen's functional part needs additional experiments and more supportive data. Details are in the recommendations to the authors.

Reviewer #1 (Recommendations for the authors):

The authors need to provide a detailed catalog numbers of the reagents used in this study.

Reviewer #2 (Recommendations for the authors):

1. It is not clear how the authors narrowed down the list from 66 hits to 22 compounds and further why Pizotifen was selected for further study. Can the authors provide additional information or data to justify exclusion of hits. Based on the data the compounds that effect dopamine based signaling would have been a more obvious pathway to study (4/22 hits).

2. The selection of human cancer cell lines to validate the findings are not clear. Is HTR2C relevant only to the subset of cancer cell lines studied. Analyses of TCGA or other data sets to show relevance of HTR2C gene expression and survival and or metastasis would greatly strengthen the rationale for picking these cell lines. Is HTR2C relevant for metastasis in all cancers or only in breast cancer. Also, it appears in figure 2B that MCF non metastatic breast cancer cells express higher levels of HTR2C than metastatic skin (MDA-MB-435) and lung (PC9) cells.

3. Some of the data provided does not support the conclusions drawn. For example, In figure 3 C. The luciferase images clearly show that Pizotifen treated mice have smaller tumors while the data in 3 A and B do not support this data. The measurements of tumor volume appear to be taken too early to actually be reproducible, also the data for proliferation and apoptosis was not shown. The authors should provide more data to justify the conclusions drawn.

4. Figure 5 A, 5B, 5G. The effects shown in the texts are not clearly visible. Can the authors provide higher magnification (40X) images to show differences noted in the text.

5. In figure 5D or 5H the effect on Pizotifen on ZEB1 or β-catenin reduction is not observed. Does this suggest this compound is effecting EMT and metastasis by other mechanisms?

6. It is not clear if HTR2C selectively only inhibits ZEB1 mediated metastases or are its effects more general i.e. its expression also effects other EMT regulators for example SNAIL, SLUG and or TWIST expression. Showing the effect of Pizotifen/ HTR2C on other EMT regulators especially the ones above is recommended.

Reviewer #3 (Recommendations for the authors):

1. Line 142: Mentioned DRD2 but did not elaborate on it and explain why it wasn't chosen as a target. In addition, please provide the rationale, why authors forgo other drugs, including the two with the strongest effect, Hexachloroghene and Troglitazone, to focus on Pizotifen.

2. Line 307, please indicate: in which cancer cells is overexpression of HTR1D associated with Wnt-signaling that enables induction of EMT?

3. Figure 5E and 5G: the figures do not show convincing nuclear accumulation of ß-catenin. Please show zoom in insert figures.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "A chemical screen based on an interruption of zebrafish gastrulation identifies the HTR2C inhibitor Pizotifen as a suppressor of EMT-mediated metastasis" for further consideration by eLife. Your revised article has been reviewed by 3 peer reviewers and the evaluation has been overseen by Didier Stainier as the Senior Editor, and a Reviewing Editor.

We are impressed with Dr. Nakayama's manuscript with a new and ambitious concept for the chemical screening for cancer metastasis using zebrafish embryos. However, two reviewers have requested that the authors clarify the mechanisms related to EMT. I also agree with their opinions because EMT is one of the most important factors for cancer metastasis and is well-studied already. It is very much expected by other researchers and would strengthen your story.

Reviewer #1 (Recommendations for the authors):

EMT is just one of the numerous steps of both gastrulation and tumor metastasis, as they mentioned that the transcriptomic data for zebrafish embryo development at the epiboly/gastrulation stage are based on the whole embryos which include all other activities and are not specific to EMT. As far as they observe phenotypic changes of whole embryo and focus on the similarity between gastrulation and metastasis in the 1st screening, what they should analyze in RNA-seq data is "global similarity" but not EMT. While they mentioned that this point is not really an objective of this study, I still feel this is one of the most important point that can be resolved, although they eventually find a drug that affects EMT.

Insufficiency of mechanistic parts has not been addressed.

Overall, I didn't feel in the revised manuscript that the authors have addressed many of my concerns, and recommend to resubmit to the other journal.

Reviewer #2 (Recommendations for the authors):

The authors have addressed most of the major comments except for Major comment 6. The authors provide no data (only text below) showing expression of SNAIL, SLUG or TWIST in MCF7 cells with and without Pizotifen, or any data from the mouse xenograft experiments. Since the authors emphasize the role of ZEB1 showing that other EMT factors are not affected or not consistently affected would help. At a minimum the authors should add their statement below to the text.

6. It is not clear if HTR2C selectively only inhibits ZEB1 mediated metastases are its effects more general i.e. its expression also effects other EMT regulators for example SNAIL, SLUG and or TWIST expression. Showing the effect of Pizotifen/ HTR2C on other EMT regulators especially the ones above is recommended.

Author response: HTR2C-expressing MCF7 cells expressed not only ZEB1 but also Twist1 and SNAIL. In contrast, Pizotifen-treated MDA-MB-231 cells showed decreased expression of ZEB1 and Twist1 compared with that in vehicle-treated the cells. In the primary tumors of Pizotifen-treated mice, only ZEB1 expression was decreased compared with those of vehicle-treated mice."

https://doi.org/10.7554/eLife.70151.sa1

Author response

Reviewer #1 (Recommendations for the authors):

The authors need to provide a detailed catalog numbers of the reagents used in this study.

I added catalog numbers of the reagents used in this study.

Reviewer #2 (Recommendations for the authors):

1. It is not clear how the authors narrowed down the list from 66 hits to 22 compounds and further why Pizotifen was selected for further study. Can the authors provide additional information or data to justify exclusion of hits. Based on the data the compounds that effect dopamine based signaling would have been a more obvious pathway to study (4/22 hits).

Zebrafish embryo screen identified 78 drugs as epiboly-interrupting drugs. Sixteen of the 78 drugs strongly affected cell viability at concentrations less than 1µM and were not used in the cell motility experiments. The remaining 62 drugs were assayed in Boyden chamber motility experiments. Twenty of the 62 drugs enabled to inhibit cell motility and invasion of MDAMB-231 cells without effecting cell viability and proceeded to next examination. The 42 drugs were passed over.

Inhibiting Dopamine receptor D2 (DRD2) suppressed metastatic dissemination of human cancer cells in a zebrafish xenografted model. However, overexpression of DRD2 could not induce EMT on MCF7 cells. Therefore, we have stop focusing DRD2. We incorporated this point into discussion part, line 777 to 781.

2. The selection of human cancer cell lines to validate the findings are not clear. Is HTR2C relevant only to the subset of cancer cell lines studied. Analyses of TCGA or other data sets to show relevance of HTR2C gene expression and survival and or metastasis would greatly strengthen the rationale for picking these cell lines.

According to cancer cell line encyclopedia, HTR2C mRNA expression is observed in cell lines which are established from embryo carcinoma, ovarian cancer, brain cancer, neuroblastoma, and myeloma. Ectopic expression of HTR2C is observed in not only the human cancer cell lines used in this study but also other cancer cell lines.

The reason why we select the human cancer cell lines used in this study, is that they are commonly used in an experimental study of metastasis and show metastasis in high frequency when they are xenografted into mice.

Is HTR2C relevant for metastasis in all cancers or only in breast cancer.

Pai et al., Breast Cancer Research 2009 shows that HTR2C expression is correlated with tumor stage of breast cancer patients and is higher in metastatic and Her2/ neu-overexpressing tumors (Vaibhav P Pai et al., 2009). We incorporate this point to main text in line 270-272.

Also, it appears in figure 2B that MCF non metastatic breast cancer cells express higher levels of HTR2C than metastatic skin (MDA-MB-435) and lung (PC9) cells.

Serotonin binds to HTR2C and activates HTR2C-mediated signaling. Biochemically, tryptophan hydroxylase 1 (TPH1) catalyzes amino acid tryptophan and yields serotonin. Serotonin in in vitro condition is provided from either an autocrine production of serotonin in cancer cells or culture media. Figure 5C clearly showed nuclear accumulation of β-catenin and p-GSK-3β were not observed in MCF7 cells. The result suggest serotonin would not exist in culture media and an autocrine production of serotonin would activate HTR2C-mediated signaling. Pai et al., revealed that MDA-MB-231 cells express TPH1 but MCF7 cells not. According to cancer cell line encyclopedia, TPH1 mRNA expressions in MDAMB-435 and PC9 cells is approximately five and two-fold rather than that in MCF7 cells, respectively. Taken together, both of HTR2C expression and an autocrine production of serotonin by TPH1 would need to activate HTR2C-mediated signaling.

3. Some of the data provided does not support the conclusions drawn. For example, In figure 3 C. The luciferase images clearly show that Pizotifen treated mice have smaller tumors while the data in 3 A and B do not support this data. The measurements of tumor volume appear to be taken too early to actually be reproducible, also the data for proliferation and apoptosis was not shown. The authors should provide more data to justify the conclusions drawn.

4T1 tumors grow aggressively in the primary sites and the tumor sizes often exceed the limits that are allowed in most animal protocols. Therefore, in 4T1 metastasis model, the primary tumors are usually resected at day 10 post inoculation. Following the conventional protocol on the model, we resected primary tumors at day 10 post inoculation. The images of top left and top right of Figure 3C were taken before the resection. In the image of bottom left of Figure 3C, bioluminescence emissions are detected around the mammary fat pad of vehicle-treated mice but the emissions might arise from either the lymph node or unresected primary tumor.

We have added data of apoptosis status in primary tumor of either vehicle or Pizotifen-treated mice as Figure S4. Proliferation status in primary tumor of either vehicle or Pizotifen-treated mice are indicated in Figure 3B.

4. Figure 5 A, 5B, 5G. The effects shown in the texts are not clearly visible. Can the authors provide higher magnification (40X) images to show differences noted in the text.

We inserted higher magnification images in Figure 5E and 5G. By using three different methods: immunofluorescence images, FACS and western blotting, we demonstrate Pizotifen recovers E-cadherin expression on a part of MDA-MB-231 cells. Immunofluorescence images in Figure 5A point that a part of Pizotifen-treated MDA-MB-231 cells show Ecadherin expression. Higher magnification image in Figure 5A would have missed the point.

5. In figure 5D or 5H the effect on Pizotifen on ZEB1 or β-catenin reduction is not observed. Does this suggest this compound is effecting EMT and metastasis by other mechanisms?

Quantification analyses of western blotting bands in Figure 5H reveals that b-catenin accumulation in the nucleus of Pizotifen-treated MDA-MB-231 cells decreased compared with those of vehicle-treated ones. Similar analyses in Figure 5D shows ZEB1 expression of Pizotifen-treated MDA-MB-231 cells slightly decreased compared with those of vehicle-treated ones. ZEB1 expression was significantly decreased in E-cadherin positive fraction of Pizotifen-treated MDA-MB-231 cells compared with ZEB1 expression in vehicle-treated ones. Therefore, we conclude Pizotifen affects ZEB1 expression.

6. It is not clear if HTR2C selectively only inhibits ZEB1 mediated metastases or are its effects more general i.e. its expression also effects other EMT regulators for example SNAIL, SLUG and or TWIST expression. Showing the effect of Pizotifen/ HTR2C on other EMT regulators especially the ones above is recommended.

HTR2C-expressing MCF7 cells expressed not only ZEB1 but also Twist1 and SNAIL. In contrast, Pizotifen-treated MDA-MB-231 cells showed decreased expression of ZEB1 and Twist1 compared with that in vehicle-treated the cells. In the primary tumors of Pizotifen-treated mice, only ZEB1 expression was decreased compared with those of vehicle-treated mice.

Reviewer #3 (Recommendations for the authors):

1. Line 142: Mentioned DRD2 but did not elaborate on it and explain why it wasn't chosen as a target.

We incorporated this point to discussion part, line 777-781.

Inhibiting Dopamine receptor D2 (DRD2) suppressed cell invasion and migration of multiple human cancer cell lines in vitro However, overexpression of DRD2 could not induce EMT on MCF7 cells. In contrast, pharmacologic and genetic inhibition of HTR2C suppressed metastatic progression; conversely, overexpression of HTR2C induced EMT. Therefore, we focus on Pizotifen.

In addition, please provide the rationale, why authors forgo other drugs, including the two with the strongest effect, Hexachloroghene and Troglitazone, to focus on Pizotifen.

We incorporated this point to discussion part, line 763-771. We have two reasons why we did not focus on Hexachloroghene. One is that Hexachloroghene strongly inhibited cell invasion and migration of MDA-MB-231 cells but did not affect cell invasion and migration of other cells: MDA-MB-436 and PC9 cells.

Second is that commercial products including Hexachloroghene killed 15 babies in the

United States and 39 babies in France in 1972. Even if we could demonstrate

Hexachloroghene has suppressor effect on metastasis, an issue of the toxic effects might abolish the new discovery. Therefore, we stopped focusing Hexachloroghene.

We have one reason why we did not focus on Troglitazone. Troglitazone has potential high liver toxicity and leads to drug-induced hepatitis. Once troglitazone was approved by FDA, it was withdrawn from the British market in December 1997, from the United States market in 2000, and from the Japanese market soon afterwards. Even if we could demonstrate Troglitazone has suppressor effect on metastasis, an issue of the toxic effects might abolish the new discovery. Therefore, we stopped focusing Troglitazone.

2. Line 307, please indicate: in which cancer cells is overexpression of HTR1D associated with Wnt-signaling that enables induction of EMT?

This paragraph is changed into "Wnt-signaling plays critical role in inducing EMT. In cancer cells, overexpression of HTR1D is associated with Wnt-signaling".

3. Figure 5E and 5G: the figures do not show convincing nuclear accumulation of ß-catenin. Please show zoom in insert figures.

We inserted higher magnification images in Figure 5E and 5G.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #1 (Recommendations for the authors):

EMT is just one of the numerous steps of both gastrulation and tumor metastasis, as they mentioned that the transcriptomic data for zebrafish embryo development at the epiboly/gastrulation stage are based on the whole embryos which include all other activities and are not specific to EMT. As far as they observe phenotypic changes of whole embryo and focus on the similarity between gastrulation and metastasis in the 1st screening, what they should analyze in RNA-seq data is "global similarity" but not EMT. While they mentioned that this point is not really an objective of this study, I still feel this is one of the most important point that can be resolved, although they eventually find a drug that affects EMT.

We compared the genes expressed in zebrafish gastrulation with the gene related with EMT through performing gene set enrichment analysis (GSEA). The analysis revealed that 50% epiboly, shield and 75%-epiboly stage of zebrafish embryos expressed the genes which promote EMT-mediated metastasis: EMT induction, TGF-β signaling, wnt/β-catenin signaling and Notch signaling (Figure 1—figure supplemental 1). This data demonstrates that a part of genes are commonly expressed between zebrafish gastrulation and EMT-mediated metastasis of human cancer cells.

Reviewer #2 (Recommendations for the authors):

The authors have addressed most of the major comments except for Major comment 6. The authors provide no data (only text below) showing expression of SNAIL, SLUG or TWIST in MCF7 cells with and without Pizotifen, or any data from the mouse xenograft experiments. Since the authors emphasize the role of ZEB1 showing that other EMT factors are not affected or not consistently affected would help. At a minimum the authors should add their statement below to the text.

6. It is not clear if HTR2C selectively only inhibits ZEB1 mediated metastases are its effects more general i.e. its expression also effects other EMT regulators for example SNAIL, SLUG and or TWIST expression. Showing the effect of Pizotifen/ HTR2C on other EMT regulators especially the ones above is recommended.

Author response: HTR2C-expressing MCF7 cells expressed not only ZEB1 but also Twist1 and SNAIL. In contrast, Pizotifen-treated MDA-MB-231 cells showed decreased expression of ZEB1 and Twist1 compared with that in vehicle-treated the cells. In the primary tumors of Pizotifen-treated mice, only ZEB1 expression was decreased compared with those of vehicle-treated mice."

We added following statement at line 321-328 and graphic data as Figure 5—figure supplement 4. “HTR2C-expressing MCF7 cells expressed not only ZEB1 but also Twist1 and Snail. […] These results indicate that HTR2C-mediated signaling induced EMT through up-regulation of ZEB1 and blocking HTR2C with Pizotifen induced mesenchymal to epithelial transition through downregulation of ZEB1 (Figure 5—figure supplement 4).”

https://doi.org/10.7554/eLife.70151.sa2

Article and author information

Author details

  1. Joji Nakayama

    1. Department of Biological Science, National University of Singapore, Singapore, Singapore
    2. Cancer Science Institute of Singapore, National University of Singapore, Singapore, Singapore
    3. Tsuruoka Metabolomics Laboratory, National Cancer Center, Tsuruoka, Japan
    4. Shonai Regional Industry Promotion Center, Tsuruoka, Japan
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Supervision, Validation, Visualization, Writing - original draft, Writing - review and editing
    For correspondence
    zmetastasis@gmail.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1077-140X
  2. Lora Tan

    Department of Biological Science, National University of Singapore, Singapore, Singapore
    Contribution
    Formal analysis, Investigation, Validation, Visualization
    Competing interests
    No competing interests declared
  3. Yan Li

    Department of Biological Science, National University of Singapore, Singapore, Singapore
    Contribution
    Data curation, Investigation
    Competing interests
    No competing interests declared
  4. Boon Cher Goh

    Cancer Science Institute of Singapore, National University of Singapore, Singapore, Singapore
    Contribution
    Funding acquisition, Project administration, Resources
    Competing interests
    No competing interests declared
  5. Shu Wang

    1. Department of Biological Science, National University of Singapore, Singapore, Singapore
    2. Institute of Bioengineering and Nanotechnology, Singapore, Singapore
    Contribution
    Funding acquisition, Resources
    Competing interests
    No competing interests declared
  6. Hideki Makinoshima

    1. Tsuruoka Metabolomics Laboratory, National Cancer Center, Tsuruoka, Japan
    2. Division of Translational Research, Exploratory Oncology Research and Clinical Trial Center, National Cancer Center, Kashiwa, Japan
    Contribution
    Funding acquisition, Project administration, Resources
    Competing interests
    No competing interests declared
  7. Zhiyuan Gong

    Department of Biological Science, National University of Singapore, Singapore, Singapore
    Contribution
    Funding acquisition, Project administration, Resources, Supervision
    For correspondence
    dbsgzy@nus.edu.sg
    Competing interests
    No competing interests declared

Funding

National Medical Research Council (R-154000547511)

  • Zhiyuan Gong

Ministry of Education - Singapore (R-154000A23112)

  • Zhiyuan Gong

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We sincerely appreciate Dr Joshua Collins (NIH/NIDCR) and Dr Shimada (Mie University) for helping this research. We thank Dr Herrick (Albany Medical College) for providing pCMV-h5TH2C-VSV with us. This study was funded by grants from National Medical Research Council of Singapore (R-154000547511) and Ministry of Education of Singapore (R-154000A23112) to ZG.

Ethics

The study protocol using zebrafish was approved by the Institutional Animal Care and Use Committee of the National University of Singapore (protocol number: R16-1068). The study protocol using mice (protocol number: BRC IACUC #110612) was approved by A*STAR (Agency for Science, Technology and Research, Singapore).

Senior Editor

  1. Didier YR Stainier, Max Planck Institute for Heart and Lung Research, Germany

Reviewing Editor

  1. Yasuhito Shimada, Mie University, Japan

Publication history

  1. Preprint posted: March 5, 2021 (view preprint)
  2. Received: May 7, 2021
  3. Accepted: December 17, 2021
  4. Accepted Manuscript published: December 17, 2021 (version 1)
  5. Version of Record published: February 8, 2022 (version 2)
  6. Version of Record updated: February 9, 2022 (version 3)

Copyright

© 2021, Nakayama et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,000
    Page views
  • 105
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Joji Nakayama
  2. Lora Tan
  3. Yan Li
  4. Boon Cher Goh
  5. Shu Wang
  6. Hideki Makinoshima
  7. Zhiyuan Gong
(2021)
A zebrafish embryo screen utilizing gastrulation identifies the HTR2C inhibitor pizotifen as a suppressor of EMT-mediated metastasis
eLife 10:e70151.
https://doi.org/10.7554/eLife.70151
  1. Further reading

Further reading

    1. Developmental Biology
    2. Neuroscience
    Mariah L Hoye et al.
    Research Article

    Mutations in the RNA helicase, DDX3X, are a leading cause of Intellectual Disability and present as DDX3X syndrome, a neurodevelopmental disorder associated with cortical malformations and autism. Yet, the cellular and molecular mechanisms by which DDX3X controls cortical development are largely unknown. Here, using a mouse model of Ddx3x loss-of-function we demonstrate that DDX3X directs translational and cell cycle control of neural progenitors, which underlies precise corticogenesis. First, we show brain development is sensitive to Ddx3x dosage; complete Ddx3x loss from neural progenitors causes microcephaly in females, whereas hemizygous males and heterozygous females show reduced neurogenesis without marked microcephaly. In addition, Ddx3x loss is sexually dimorphic, as its paralog, Ddx3y, compensates for Ddx3x in the developing male neocortex. Using live imaging of progenitors, we show that DDX3X promotes neuronal generation by regulating both cell cycle duration and neurogenic divisions. Finally, we use ribosome profiling in vivo to discover the repertoire of translated transcripts in neural progenitors, including those which are DDX3X-dependent and essential for neurogenesis. Our study reveals invaluable new insights into the etiology of DDX3X syndrome, implicating dysregulated progenitor cell cycle dynamics and translation as pathogenic mechanisms.

    1. Developmental Biology
    2. Evolutionary Biology
    Mathi Thiruppathy et al.
    Short Report

    Whereas no known living vertebrate possesses gills derived from the jaw-forming mandibular arch, it has been proposed that the jaw arose through modifications of an ancestral mandibular gill. Here, we show that the zebrafish pseudobranch, which regulates blood pressure in the eye, develops from mandibular arch mesenchyme and first pouch epithelia and shares gene expression, enhancer utilization, and developmental gata3 dependence with the gills. Combined with work in chondrichthyans, our findings in a teleost fish point to the presence of a mandibular pseudobranch with serial homology to gills in the last common ancestor of jawed vertebrates, consistent with a gill origin of vertebrate jaws.