ww.sciencedirect.com

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5

Available online at w

ScienceDirect

journal homepage: www.JournalofSurgicalResearch.com

Association for Academic Surgery

The glypican 3 hepatocellular carcinoma markerregulates human hepatic stellate cells viaHedgehog signaling

Paolo Magistri, MD,a,b Stephanie Y. Leonard, BS,a Chih-Min Tang, PhD,a

Jonathan C. Chan, BS,a Tracy E. Lee,a and Jason K. Sicklick, MDa,*aDivision of Surgical Oncology, Department of Surgery, Moores UCSD Cancer Center, University of California,

San Diego, Californiab Faculty of Medicine and Psychology, Azienda Ospedaliera Sant’Andrea, SapienzaeUniversita di Roma, Rome, Italy

a r t i c l e i n f o

Article history:

Received 27 June 2013

Received in revised form

9 December 2013

Accepted 13 December 2013

Available online 17 December 2013

Keywords:

Epithelial-mesenchymal

interactions

GPC3

Hepatoma

HCC

Hedgehog-interacting protein

HSC

Tumorestroma interactions

This article was presented at the seventh* Corresponding author. Division of Surgical

Sciences Drive, MC 0987 La Jolla, CA 92093-0E-mail address: jsicklick@ucsd.edu (J.K. S

0022-4804/$ e see front matter ª 2014 Elsevhttp://dx.doi.org/10.1016/j.jss.2013.12.010

a b s t r a c t

Background: Hepatocellular carcinoma (HCC) frequently represents two diseases as it often

arises in the setting of cirrhosis caused by the proliferation and activation of hepatic

stellate cells (HSCs). Previously, we identified that Hedgehog (Hh) signaling regulates HSC

viability and fibrinogenesis, as well as HCC tumorigenesis. Although it is increasingly

recognized that HSCs and HCCs communicate via paracrine signaling, Hh’s role in this

process is just emerging. We hypothesized that a secreted HCC tumor marker and Hh

mediator, glypican 3 (GPC3), may regulate HSC.

Methods: Using three human HCC lines (Hep3B, PLC/PRF/5 and SK-Hep-1) and one Hh-

responsive human HSC line (LX-2), we developed two in vitro models of HCC-to-HSC

paracrine signaling using a Transwell coculture system and HCC-conditioned media. We

then evaluated the effects of these models, as well as GPC3, on HSC viability and gene

expression.

Results: Using our coculture and conditioned media models, we demonstrate that the three

HCC lines decrease HSC viability. Furthermore, we demonstrate that recombinant GPC3

dose-dependently decreases the LX-2 viability while inhibiting the expression of Hh target

genes that regulate HSC viability. Finally, GPC3’s inhibitory effects on cell viability and Hh

target gene expression are partially abrogated by heparin, a competitor for GPC3 binding.

Conclusions: For the first time, we show that GPC3, an HCC biomarker and Hh mediator,

regulates human HSC viability by regulating Hh signaling. This expands on existing data

suggesting a role for tumorestroma interactions in the liver and suggests that GPC3 plays a

role in this process.

ª 2014 Elsevier Inc. All rights reserved.

Annual Academic Surgical Congress, February 14e16, 2012, Las Vegas, Nevada.Oncology, Moores UCSD Cancer Center, University of California, San Diego, 3855 Health987. Tel.: þ1 858 822 6173; fax: þ1 858 228 5153.icklick).ier Inc. All rights reserved.

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5378

1. Introduction in decreased HSC viability. Similarly, exogenous GPC3 treat-

Hepatocellular carcinoma (HCC) is one of the most common

cancers worldwide and the third leading cause of cancer-

related death [1]. Furthermore, patients with cirrhosis have

an approximately 20% risk of developing HCC over 5 y. In the

United States, HCC and cirrhosis often occur secondary to

hepatitis C virus [2]. Because of the latency of hepatitis C virus

and its increasing spread over the last 40 y, the incidences of

HCC and cirrhosis are expected to rise [3]. Alcoholic liver dis-

ease and nonealcoholic fatty liver disease serve as additional

risk factors for both diseases [4,5]. In addition, we know that

epithelialemesenchymal interactions are important in ma-

lignancies [6]. Emerging data in pancreatic adenocarcinomas

suggest that tumor cells send signals to the stroma to provide

a more favorable environment for tumor growth and pro-

gression [7]. As a result, these paracrine interactions may

serve as targets for treating cancers. During the evolution of

liver fibrosis and cirrhosis, hepatic stellate cells (HSCs) are the

major profibrogenic cells in the liver. Emerging data suggest

that tumorestroma, or HCCeHSC, paracrine interactions

occur in the liver [8e12]. Thus, the interplay between HCCs

and peritumoral HSCs deserves further investigation.

There is a growing body of knowledge demonstrating the

role of Hedgehog (Hh) signaling within liver cell populations

including HSC, progenitors, and endothelial cells [13e16]. We

first described evidence of Hh pathway activity in HSCs and

demonstrated that this pathway plays an important role in

maintaining their viability and activation [13,17]. It was later

shown that these stromal cells have Hh-mediated paracrine

interactions with hepatic epithelial cells [18]. We, and others,

have also defined a role for Hh signaling in the genesis of HCC

and shown that dysregulation of Hh signaling lies at several

checkpoints within the pathway [19,20].

Glypican 3 (GPC3), an Hhmediator [21], is expressed at high

levels in up to 72% of human HCCs [22] and in Hh-responsive

HCC lines (Hep3B, Huh7, and PLC/PRF/5) [23]. GPC3 is one of

six proteoglycans (PGs) that acts as an Hh pathway inhibitor

during embryogenesis [24]. GPC3 binds with Hh ligands on the

cell surface. In turn, this binding leads to internalization and

degradation of the GPC3eHh ligand complex [25]. Thus, the

amount of Hh ligand available for activating its receptor,

patched (Ptc), is reduced at the cell surface. This leads to a

reduction in Hh signaling and expression of Hh target genes,

including Ptc. As a result, loss-of-function mutations in GPC3

are the cause of the X-linked Simpson-Golabi-Behmel syn-

drome [26], which is characterized, in part, by hepatomegaly.

Conversely, high expression of GPC3 leads to suppression of

hepatocyte proliferation [27] and liver regeneration [28].

Because GPC3 is not detectable in normal hepatocytes, benign

liver diseases or peritumoral stroma (i.e., HSC) [29], it is used as

a serum tumormarker and prognostic factor for HCC. Patients

withHCCs that express high levels of GPC3 have a significantly

worse 5-y survival rate than patients with HCCs that do not

express GPC3 [30]. Given the previous data, we hypothesized

that this secreted HCC tumor marker, and Hh mediator, may

modulate HCC (tumor)-to-HSC (stroma) interactions.

Herein, we show evidence that coculture of an activated

humanHSC line with three GPC3-expressing HCC lines results

ment of HSCs decreases cell viability in a dose-dependent

fashion. We also show that exogenous GPC3 treatment of

HSCs decreases the messenger RNA (mRNA) expression of Hh

target genes including patched 1 (Ptc1) and Hh-interacting

protein (HHIP). Because GPC3 is classified as a heparan sul-

fate PGs (HSPGs) that display heparin-inhibitable binding, the

effects of GPC3 on HSC viability and gene expression can be

partially abrogated by HSC treatment with unfractionated

heparin. Taken together, this suggests that GPC3 plays an

unreported role in regulating hepatic tumor-to-stroma in-

teractions that is partially mediated by Hh signaling.

2. Materials and methods

2.1. Reagents

Transwell plates were purchased from Corning (Lowell, MA).

Recombinant human GPC3 and HHIP were purchased from

R&D Systems (Minneapolis, MN). Heparin sodium salt derived

from porcine intestinal mucosa was purchased from Sigma-

Aldrich (St. Louis, MO). Antibodies against human GPC3 and

b-actinwere purchased fromAbbiotec (SanDiego, CA) andCell

Signaling (Danvers,MA), respectively.Horseradishperoxidase-

conjugated goat anti-rabbit immunoglobulin Gwas purchased

from Thermo Scientific (Waltham, MA). Cell culture media

were purchased from GIBCO/BRL (Carlsbad, CA). Cell viability

wasmeasuredwith the Cell Counting Kit-8 (DojindoMolecular

Technologies, Rockville,MD)andfindingswereconfirmedwith

an MTT (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium

bromide) assay using Thiazolyl Blue Tetrazolium Bromide

(Sigma-Aldrich). Detergent compatible protein assay was

purchased from Bio-Rad (Hercules, CA). An enhanced chem-

iluminescence system for Western blot detection was pur-

chased from Pierce (Rockford, IL).

2.2. Cell culture

The activated human LX-2 HSC line was kindly provided by Dr

S.L. Friedman (Mount Sinai School of Medicine, New York, NY)

[31] and maintained in 2% serum-supplemented Dulbecco’s

modified Eagle’s medium -to-Ham’s F-12 (1:1, GIBCO/BRL),

10 mM hydroxyethyl piperazineethanesulfonic acid, penicillin

100 IU/mL and streptomycin 100 mg/mL (GIBCO/BRL). This is a

well-validated human HSC line that has been used as a model

systemforstudyingHSCbiology inmore than120peer-reviewed

manuscripts. As opposed to primary HSC isolates, this cell line

unequivocally eliminates other contaminating types of cells,

such as macrophages, endothelial cells, and vascular smooth

muscle cells. Hep3B, SK-Hep1, and PLC/PRF/5 cell lines were

purchased from American Type Culture Collection (Manassas,

VA) and maintained in 10% serum-supplemented Dulbecco’s

modified Eagle’s medium (GIBCO/BRL), 2 mM glutamine, peni-

cillin 100 IU/mL, and streptomycin 100 mg/mL (GIBCO/BRL).

During experimental manipulations as described in the

following, the respective media conditions for HSC and HCC

were adjusted to control for differences in serum conditions

between the sets of lines.

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5 379

2.3. Cell viability and apoptosis assays

Cell viability was measured with the Cell Counting Kit-8

(Dojindo) [13]. Briefly, HSCs were cultured for up to 72 h, and

then incubated with tetrazolium reagent and absorbance was

measured. Experiments were replicated and findings

confirmed with MTT reagent according to manufacturer’s in-

structions (Sigma-Aldrich). Results are presented as the

percent of control. Apoptotic activity was assayed in parallel

cultures using the Apo-ONE Homogenous Caspase 3/7

Apoptosis Assay (Promega, Madison, WI) according to the

manufacturer’s instructions.

2.4. Transwell coculture model

Similar to a previously described model [32], HSC and HCC

lines were cultured in a Transwell insert coculture system for

72 h, using 3-mm pore size polyester inserts (Corning). Briefly,

LX-2 cells were plated at a density of 500,000 cells/well in six-

well plates, and cultured overnight as monocultures. Insert

chambers containing HCC cells (500,000 cells/polyester insert)

were then transferred into the coculture systems and incu-

bated with the LX-2 cells for an additional 72 h. Thus, cocul-

ture experiments were performed with HSC in the bottom

well, and HCC in the top well (1:1 cell ratio), using 2% serum-

supplemented LX-2 media. For analyses of HCC lines, the

well positions and media were reversed.

2.5. Conditioned media model

AfterHCC lineculture for 72h, 2%,or 10%serum-supplemented

conditioned media were collected from culture flasks and

centrifugedat1500 rpmfor10mintopellet cellulardebris.HSCs

were cultured in LX-2 media for 24 h and then treated for 72 h

with HCC-conditioned medium. Results were compared with

cells treated with corresponding control medium (i.e., 2% or

10%serum-supplementedmediumwithoutHCCconditioning).

2.6. Two-step real-time RT-PCR

Two-step real-time RT-PCR (reverse transcription polymerase

chain reaction) was performed to compare the expression of

target genes. Total RNA was extracted from cells using Trizol

Reagent (Invitrogen/Life Technologies, Carlsbad, CA). Reverse

transcription to complementary DNA (cDNA) templates was

performed using iScript cDNA Synthesis kit (Bio-Rad). For

quantitative RT-PCR, human GPC3, HHIP, and b-actin (house-

keeping gene) primers were designed using Genbank se-

quences. Additional oligonucleotide primers were

synthesized by ValueGene (San Diego, CA). The sense (S) and

antisense (AS) primer sequences are as follows: GPC3: (S)

TGAAAGTGGAGACTGCGGTGATGA, (AS) TCCCGAGGTTGT

GAAAGGTGCTTA; Ptc1: (S) CCACCAGACGCTGTTTAGTCA,

(AS) CGATGGAGTCCTTGCCTACAA; HHIP: (S) CCCACACTTCA

ACAGCACCA, (AS) GCACATCTGCCTGGATCGT; Gli1: (S)

TGCAGTAAAGCCTTCAGCAATG, (AS) TTTTCGCAGCGAGC

TAGGAT; a-SMA: (S) CTTTTCCATGTCGTCCCAGT, (AS) GTGA

CGAAGCACAGAGCAAA; Col1a1: (S) TGTGAGGCCACGCAT

GAG, (AS) CAGATCACGTCATCGCACAA; b-actin: (S) AATGTGG

CCGAGGACTTTGATTGC, (AS) AGGATGGCAAGGGACTTCCTG

TAA. Quantitative RT-PCR was performed by using 1e2 mL of

the first-strand cDNA product. The real-time reaction was

amplified (N ¼ 3e9) using SsoFast EvaGreen supermix (Bio-

Rad), the previously mentioned primers, and the CFX96 Real-

Time PCR Detection System (Bio-Rad). Target gene levels in

cells are presented as a ratio to levels of b-actin, according to

the DCt method, or as a ratio to levels detected in the corre-

sponding control cells, according to the DDCt method [19].

These fold changes were determined using point and interval

estimates.

2.7. Western analysis

Western analyses of cell extracts were performed according to

conventional methods. Cells were lysed with radio-

immunoprecipitation assay buffer (Cell Signaling), which

included Halt Protease and Phosphatase Inhibitor Cocktail

(Pierce). Protein was quantified using a detergent compatible

protein assay. An amount of 40 mg protein samples were

subjected to 4%e12% Nu PAGE BIS-Tris gel (Life Technologies),

and transferred using the Invitrogen I-Blot 7 min transfer

system onto a polyvinylidene difluoride membrane. After

blocking themembranewith 5% (wt/vol) nonfat milk and 0.1%

(vol/vol) Tween-20 in phosphate-buffered saline for 30 min at

room temperature, the membrane was incubated with anti-

GPC3 antibody (1:500, Abbiotec) or anti-b-actin antibody

(1:1000, Cell Signaling) overnight at 4�C. Themembranes were

then washed three times with phosphate-buffered saline/

Tween-20 (0.1% [vol/vol]). Horseradish peroxidase-conjugated

goat anti-rabbit immunoglobulin G (1:40,000, Thermo Scien-

tific) was used as a secondary antibody for both primary an-

tibodies. Proteinwas detected using the SuperSignalWest Pico

Chemiluminescent Substrate (Pierce). Quantitative densi-

tometrywas performed using the Bio-Rad Gel Doc and Bio-Rad

Quantity One software.

2.8. Statistical analysis

Results are expressed as mean � standard error mean or

standard deviation as appropriate. Comparisons between

groups were performed using the Student t-test (Stata 9.0,

StataCorp, College Station, TX). Statistical significance was

accepted at the 5% level and statistical trends were accepted

at the 10% level.

3. Results

3.1. Three human HCC lines express GPC3

Previous work by Nakatsuraet al. [33] demonstrated that both

the Hep3B and PLC/PRF/5 lines express GPC3 mRNA and pro-

tein by real-time RT-PCR and Western analyses, respectively.

We confirmed these findings (Fig. 1), as well as determined

that the SK-Hep1 line also expresses GPC3mRNA (Fig. 1A) and

protein (Fig. 1B). To recapitulate the tumorestroma in-

teractions, and in particular, the effects of HCC on HSC, we

developed a Transwell coculture system using the activated

human HSC line, LX-2, and three different HCC cell lines

(Hep3B, SK-Hep-1, and PLC/PRF/5). Using our coculture assay,

Fig. 1 e HCC cell lines (SK-Hep1, Hep3B, and PLC/PRF/5) express GPC3. (A) Quantitative real-time RT-PCR of Gpc3 mRNA was

performed to compare expression with the b-actin housekeeping gene using the DCt method. (B) Western blot analyses

showing the protein expression of GPC3 in the HCC lines before (odd lanes) and after (even lanes) Transwell coculture with

activated human LX-2 HSC. b-actin serves as the control. (C) Densitometric analyses of Western blot. Volume units of GPC3

protein are adjusted for volume units of b-actin. Data are normalized to SK-Hep1 levels and reported as percent control.

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5380

we studied the LX-2 cells in the bottomwell and each HCC line

in the top well. These lines express GPC3 protein after cocul-

ture with LX-2 (Fig. 1B, even lanes). Despite having the lowest

mRNA expression, SK-Hep1 cells have the highest GPC3 pro-

tein expression. Conversely, the PLC/PRF/5 cells have the

lowest GPC3 protein expression despite having the highest

mRNA expression. Protein densitometry analysis confirmed

these findings (Fig. 1C).

3.2. Human HCC cell lines modulate the viability andapoptosis of a human LX-2 HSC line

Given that HCC rarely occurs in the absence of hepatic fibrosis

and/or cirrhosis, we aimed to determine if HCC cells affect the

viability of the LX-2 HSC line. Coculture with all three HCC

lines (top well) resulted in decreased HSC viability after 72 h

(Fig. 2A) as compared with LX-2 cultured without HCC cells in

the top well. Coculture with SK-Hep1, Hep3B, and PLC/PRF/5

cells resulted in 26.4% (P < 0.05), 20.9% (P < 0.05), and 9.3%

(P ¼ NS) decreases in viability, respectively. These reductions

paralleled the relative expression pattern of GPC3 protein in

the respective HCC lines (Fig. 1B and C). For instance, SK-Hep1

cells express the most GPC3 protein and coculture with LX-2

cells decreases LX-2 viability more than the other lines.

To validate the previously mentioned findings, we used a

second model of paracrine tumorestroma interactions. We

used conditioned media from confluent HCC lines for

culturing HSC. Figure 2B demonstrates that 72 h culture in the

presence of either 2% or 10% serum-supplementedHCCmedia

reduces HSC viability in two of three lines as compared with

culture in the presence of unconditioned HCC media. Culture

with 2% serum-supplemented SK-Hep1, Hep3B, and PLC/PRF/5

conditioned media resulted in 24.4% (P < 0.01), 5.9% (P ¼ 0.17),

and 20.5% (P < 0.01) decreases in viability, respectively.

Concomitantly, culture with 10% serum-supplemented SK-

Hep1, Hep3B, and PLC/PRF/5 conditioned media resulted in

19.6% (P< 0.001), 0.6% (P¼ 0.79), and 15.1% (P< 0.01) decreases

in viability, respectively. These findings generally parallel

those in Figure 2A while also controlling for the potential ef-

fects of serum supplementation. However, these effects on

cell viability appear to be independent of cellular apoptosis as

measured by caspase activity because we only observed

Fig. 2 e HCC-to-HSC interactions regulate the viability of

activated HSC. (A) HSC viability after 72 h coculture with

PLC/PRF/5, Hep3B, and SK-Hep1 versus LX-2 cultured

without cells in the top well. *P < 0.05. (B) HSC viability

after 72 h culture with 2% or 10% serum-supplemented

conditioned media from PLC/PRF/5, Hep3B, and SK-Hep1

compared with culture in the presence of unconditioned

HCC media. **P < 0.01 and yP < 0.001.

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5 381

0.24%e2.6% changes as compared with controls (data not

shown). Together, these data suggest that tumor-to-stroma

(HCC-to-HSC) paracrine signaling may occur in vitro and these

interactions regulate HSC viability.

Fig. 3 e GPC3 dose-dependently regulates the viability and

Hh target gene levels in activated HSC. (A) Dose-response

curves of LX-2 viability after treatment (0.0098e10 mg/mL)

with recombinant GPC3 (black square) and the positive

control Hh ligand antagonist, HHIP (red triangle). (B)

Quantitative real-time RT-PCR analysis of Ptc1, HHIP, and

Gli1 mRNA was performed in LX-2 cells without and with

GPC3 treatment (5 mg/mL) for 72 h. *P < 0.05 versus control.

(Color version of figure is available online.)

3.3. GPC3 modulates the viability of HSC in vitro

GPC3 is a HCC tumor marker and Hh pathway modulator that

is known to regulate the viability of parenchymal cells in the

liver [27,28]. However, its role in viability of HSC, an Hh-

dependent nonparenchymal cell [13], is unknown. Given

effects of HCC onHSC,we sought to determine if these could be

augmented with supplemental GPC3. We added recombinant

GPC3 (5 mg/mL) to HCC-conditioned media. In data not shown,

culture with GPC3-supplemented Hep3B and PLC/PRF/5

conditionedmedia (10% serum) resulted in an additional 19.2%

(P ¼ NS) and 4.4% (P ¼ NS) decrease in viability, respectively, as

compared with cells cultured with unsupplemented HCC-

conditioned media. In contrast, GPC3 supplementation of SK-

Hep1 conditioned media resulted in a 1.6% (P ¼ NS) increase

in viability. Thus, GPC3 did not have a significant additive or

synergistic effect in HSC exposed to HCC-conditioned media.

We then treated the LX-2 line with increasing concentra-

tions of GPC3 (0.0098e10 mg/mL). We determined that GPC3

reduces the viability of HSC in a dose-dependent manner

(Fig. 3A). We noted up to a 27.0% reduction in the viability of

cells treated with 5 mg/mL of GPC3 (P ¼ 0.005). Thus, this dose-

response curve spans the doses from no observable effect

(0.0098 mg/mL) versus control (P¼NS) to amaximal effect (5 mg/

mL) versus control (P ¼ 0.005). For the 10 mg/mL treatment,

there is no additional effect on viability as compared with the

5 mg/mL treatment (P¼NS). For the treatments<0.3125 mg/mL,

there is a plateau effect and no differences in viability as

compared with one another or control (P ¼ NS). Thus, the

changes seen in the dose response curve occur between 0.3125

and 5 mg/mL. These data suggest that GPC3, a protein secreted

by HCC, can partially modulate the viability of activated HSC,

which reside within the stroma of fibrotic and cirrhotic livers.

To confirm our findings that GPC3, an endogenous inhibitor of

Hh signaling, can reduce HSC viability by binding to Hh

ligands, we performed the same experiments using HHIP. This

membrane glycoprotein negatively regulates Hh signaling by

competitively binding to human Hh ligands with an affinity

similar to the Hh receptor, Ptc. [34e36]. Thus, its mechanism

of action is similar to that of GPC3. Moreover, this Hh inhibitor

is downregulated in liver cancers [34,36] and during HSC

activation [14,17]. As expected, treating LX-2 cells with

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5382

exogenous HHIP (0.3125e10 mg/mL) decreases HSC viability in

a dose-dependent manner (Fig. 3A). This also demonstrates

that active Hh signaling at the HSC surface is Hh ligand

dependent.

3.4. GPC3 regulates Hh target gene expression

When Hh signaling is activated in Hh-responsive cells, target

gene transcription occurs. Interestingly, several members of

the canonical Hh signaling pathway are also downstream

gene targets. Ptc1 and HHIP provide negative feedback

because of their inhibitor actions, whereas the Gli1 tran-

scription factor provides positive feedback. Alternations in

their relative mRNA levels indicate regulation of the Hh

pathway. We treated LX-2 cells with recombinant GPC3 at a

dose (10 mg/mL) which decreases cell viability. Real-time RT-

PCR analysis after 72 h treatment demonstrated that GPC3

treatment decreases Ptc1 and HHIP mRNA levels, although

these were not statistically significant (Fig. 3B). However, Gli1

mRNA expression increased by 9.02-fold (P ¼ 0.012) as

compared with untreated cells. Despite these changes in Hh

signaling, we did not observe a statistically significant effect

on the expression of HSC activation biomarkers including

a-SMA or Col1a1 (data not shown). These results suggest that

GPC3 regulates the viability, but not the further activation, of

HSC in vitro and that this is partially regulated by modulating

Hh signaling.

3.5. Heparin partially abrogates the inhibitory effects ofGPC3 on HSC viability and gene transcription

GPC3 is an HSPG. Members of this family display heparin-

inhibitable binding because heparin, a heparan sulfate gly-

cosaminoglycans, can covalently bind to HSPGs. As a result,

we hypothesized that heparin could competitively bind with

GPC3 to form GPC3-heparin complexes in lieu of GPC3-Hh

ligand complexes. In turn, we hypothesized that GPC3

effects on HSC viability and gene transcription could be

abrogated by heparin. To test this hypothesis, we treated LX-2

HSC with GPC3 (10 mg/mL) and heparin (1:2 serial dilutions

from 100e3.125 mg/mL). GPC3 alone decreased cell viability by

34.2% (P ¼ 0.04). Comparison of all GPC3- and heparin-treated

cells with GPC3-only treated cells demonstrated an abrogation

of GPC3 effects. We observed a 2.5%e22.4% increase in cell

viability across all groups (Fig. 4A). The effects were statisti-

cally significant (P < 0.05) at heparin doses of 25 mg/mL or

greater. Meanwhile, treatment of LX-2 with heparin alone

(100 mg/mL) results in a 2.0% change in HSC viability (P ¼ NS).

At the 25 and 50 mg/mL doses, we observed 18%e23% increases

in HSC viability. However, these were not statistically signifi-

cant because of variability in the replicates (N ¼ 4). At two

lower doses of heparin (6.25 and 12.5 mg/mL), we do see a

statistically significant increase in HSC viability (P ¼ 0.04 and

P ¼ 0.03, respectively). However, only at the 6.25 mg/mL dose

does this translate into either a partial rescue of GPC3’s effects

or an independent increase in HSC viability (P ¼ 0.13 versus

control, Fig. 4A). At all other doses (25e100 mg/mL) that result

in rescue from GPC3’s effects, there was not a statistically

significant effect from heparin alone on LX-2 viability (Fig. 4A).

On its own, the higher doses of heparin that result in rescue

from GPC3’s effects had no statistically significant effect on

the relative cell viability of the LX-2 HSC line. Our results

demonstrate that heparin is capable of a partial dose-

dependent rescue of GPC3’s effects on HSC viability. This

appears to be independent of heparin’s ability to regulate HSC

viability at lower doses. Thus, we find that the viability of

untreated HSC is similar to that of HSC treated with GPC3 and

higher doses of heparin. Taken together, this further demon-

strates GPC3’s regulation of HSC viability.

3.6. Heparin rescues Hh target genes expression inGPC3-treated HSC

To determine if heparin could rescue the transcriptional ef-

fects of GPC-treatment on HSC, we treated LX-2 cells with

recombinant GPC3 (10 mg/mL) and heparin (10 mg/mL). Inverse

to our earlier findings with GPC3 alone (Fig. 3B), quantitative

real-time RT-PCR analysis after 72 h treatment demonstrated

that heparin treatment increases Ptc1 and HHIP mRNA levels

by 4.92- (P ¼ 0.05) and 1904-fold (P ¼ 0.074), respectively, as

compared with GPC3-treated cells (Fig. 4B). Interestingly, Gli1

mRNA expression decreased by 2.0-fold (P ¼ NS) as compared

with GPC3-treated cells. Thus, we observed a heparin-induced

rescue of Hh target gene expression in GPC3-treated HSC.

4. Discussion

For the first time, we demonstrate that the HCCmarker, GPC3,

regulates human HSCs via paracrine signaling. In turn, this

modulates Hh signaling in the HSC. Because GPC3 binds with

Hh ligands (i.e., Sonic hedgehog and Indian hedgehog) on the

cell surface, there is a resultant reduction in Hh signaling, an

HSC viability factor [13]. Thus, these tumorestroma in-

teractions via GPC3 regulate the viability of activated human

HSC that are implicated in hepatic fibrosis and cirrhosis.

Furthermore, these effects correlate with the relative expres-

sion levels of GPC3 in HCC lines. Using a coculture model and

HCC-conditioned media, we replicated these findings.

Furthermore, we show that recombinant GPC3 dose depen-

dently decreases HSC viability and GPC3 can decrease Hh

signaling in HSC. Finally, because GPC3 is a HSPG, we deter-

mined that the inhibitory effects of GPC3 are partially abro-

gated by the treatment with unfractionated heparin. Taken

together, these results suggest that hepatic tumor-to-stroma

interactions are partially regulated by the HCC biomarker and

prognostic factor, GPC3 [33].

Our present findings complement and expand on earlier

studies suggesting that hepatic GPC3mRNA and protein levels

are inverse to the proliferative activities of hepatic paren-

chymal cells [27,28]. We now demonstrate that GPC3

decreases the viability of activated human HSC, a non-

parenchymal cell. Such intriguing findings are in alignment

with the paradoxical nature of GPC3. Despite GPC3 upregula-

tion in HCCs, earlier work by Michalopoulos’ group suggested

that GPC3 plays an overall growth inhibitory role in liver

regeneration and hepatocyte proliferation [28,37]. We now

add HSC to the list of hepatic cells that are negatively regu-

lated by GPC3. Mechanistically, we show that this is partially

regulated through the Hh signaling pathway and its target

Fig. 4 e Heparin partially abrogates the anti-HSC effects of GPC3. (A) Cells were treated with increasing doses of

unfractionated heparin (3.125e100 mg/mL) with or without a constant dose of GPC3 (10 mg/mL). Results are compared with

untreated control cells (0/0) and cells treated with GPC3, but not heparin (10/0). Absolute changes in normalized (percentage

control) LX-2 cell viability and P values are presented in the table. (B) Quantitative real-time RT-PCR analysis of Ptc1 and

HHIP mRNA was performed in GPC3-treated (10 mg/mL) LX-2 cells without and with heparin treatment (10 mg/mL) for 72 h.

*P < 0.05 versus control.

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5 383

genes, including HHIP. Given that HCCs are also Hh regulated

and secrete GPC3, this may explain the earlier finding that

HHIP mRNA expression is increased in peritumoral stroma

when compared with human HCCs [20,34,36]. It remains to be

determined whether this is through autocrine regulation of

the pathway.

The current findings also add to the growing body of evi-

dence suggesting that tumorestroma interactions are present

and active in the liver. We demonstrate a previously unknown

role for GPC3 in this process. Although we show that HSC

coculture with HCC lines (i.e., SK-Hep1 and Hep3B) with higher

GPC3 expression can decreaseHSC viability, others have shown

that coculture with HCC lines (i.e., Huh7) can stimulate cellular

proliferation and migration via PI3K/AKT activation. Moreover,

they demonstrated that Huh7 can increase the expression of

proangiogenic genes while decreasing the expression of profi-

brinogenic genes [11]. Together, our findings and the earlier

findings demonstrate the heterogeneity of HCC lines and their

variable effects on HSC in vitro.

HCC’s tumor-to-stroma interactions are crucial for modi-

fying this cancer’s natural history. Ju et al. [38] demonstrated

that a higher density of activated peritumoral HSCs correlates

with more advanced clinicopathologic features including

increased HCC size, vascular invasion, lack of tumor encap-

sulation, and elevated alpha-fetoprotein levels. Work by

Chanet al. [12] showed that malignant hepatocytes secrete Hh

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5384

ligands, which stimulate glycolysis in neighboring myofibro-

blasts. This results in the release of lactate, which cancer

cells uses as an energy source. Others [10,11] have also

demonstrated that HCC-HSC cross talk generates a permissive

pro-angiogenic microenvironment that promotes HCC pro-

gression. Although our findings may seem counterintuitive,

we can postulate that aggressive HCC cells that secrete more

GPC3 will kill HSC via inhibition of Hh signaling to possibly

allow for a niche that permits tumor outgrowth and/or

angiogenesis. In line with this hypothesis, GPC3 is known to

be highly expressed in HCCs, and its expression pattern differs

according to the degree of cell differentiation. [39] In fact,

membranous GPC3 expression in HCCs correlates with a

worse prognosis because after resection, GPC3-positive HCC

patients have lower disease-free survival rate than GPC3-

negative HCC patients. [40] Mechanistically, transfection of

HepG2, MHCC-97H, and Huh7 cell lines with anti-GPC3 short

hairpin RNA significantly reduced GPC3 mRNA and proteins

levels. [41] In turn, this reduced proliferative, migratory, and

invasive capacities, as well as significantly increased

apoptosis. Moreover, this was associated with increased Gli1

mRNA expression and decreased vascular endothelial growth

factor protein in HepG2 cells. Thus, GPC3 appears to

contribute to HCC migration, invasion, angiogenesis, and

apoptosis, possibly through its interactions with the Hh

signaling pathway. But, in opposition to these findings, Zit-

terman et al. [42] noted an antiangiogenic effect of GPC3 in

Huh7 and HepG2-derived tumors, but not in Huh6 xenografts.

Thus, the exact role of GPC3 in these processes remains

debatable.

Interestingly, our newly identified Hh-dependent tumore

stroma interactions are quite distinct from those in pancreatic

ductal adenocarcinoma, where tumor secretion of Hh ligands

drives stromal proliferation and promotes desmoplasia [7].

Teleologically, pancreatic ductal adenocarcinoma arise first

and secondarily leads to peritumoral fibrosis, which causes

chemoresistance. Conversely, chronic liver injury leads to

hepatic fibrosis and cirrhosis, which create a permissive

environment for HCC development. However, because the

HCCs arise in the milieu of fibrotic liver, their growth and

angiogenesis may be limited without serially depleting HSCs

or decreasing their expression of profibrinogenic genes [11]. In

line with this hypothesis, we show that the tumorestroma

interactions lead to decreased HHIP expression, which others

have shown can promote angiogenesis in HCC [36]. While

assigning a specific role to each protein mediator in this

complicated milieu remains difficult, a better comprehension

of tumorestroma interactions may have important conse-

quences for the treatment of HCC and cirrhosis. For instance,

the finding that heparin increases human HSC viability at low

doses is probably multifactorial and based on the fact that

HSCs express other cell surface HSPGs, including syndecan-1

and syndecan-3, as well as the matrix HSPG, perlecan. [43]

These are cofactors in cell-matrix adhesion processes, in

cellecell recognition systems, and in receptor-growth factor

interactions with HSC growth factors such as epidermal

growth factor, vascular endothelial growth factor fibroblast

growth factor-1/2, hepatocyte growth factor, and interleukin

8. [44] As a result, on its own, heparin appears to play a role in

regulating HSC viability at lower doses, possibly by

modulating HSPGs and the functional levels of the afore-

mentioned growth factors.

Clinically, GPC3 may represent not only a reliable tumor

marker, but also a prognostic factor and a therapeutic target.

GC33 (Chugai Pharmaceutical, Japan), a recombinant

humanized anti-GPC3 antibody, is now being studied as a

potential therapeutic agent for patients with advanced HCC

[45,46]. Preclinical pharmacologic assessments have shown

that rather than directly targeting GPC3, GC33 elicits antibody-

dependent cellular cytotoxicity mediated by natural killer

cells and macrophages targeted against GPC3-expressing

human HCC cells. [47] Moreover, our results demonstrate

that the regulation of the Hh pathway may be a useful tool to

control HCC progression, influencing both tumorestroma

cross talk and tumor growth. In this context, we provide

additional insight into potential consequences of such

approaches in patients with HCC and cirrhosis.

Acknowledgment

We thank Professor Stefania Uccini and Dr Michael Peterson

for their expertise and discussions. This work was supported

by the American Cancer Society’s Institutional Research Grant

No. 70-002 (JKS) and the SapienzaeUniversita di Roma student

research fellowship (PM).

Author contributions: PM, SYL, CMT, and JKS were partic-

ipated in conception and design; PM, SYL, CMT, and JKS were

responsible for analyses and interpretation; PM, SYL, CMT,

JCC, and TEL were responsible for data collection; PM, SYL,

CMT, and JKS were responsible for drafting the article; PM,

SYL, CMT, and JKS were responsible for the critical revision of

article; PM and JKS were responsible for funding.

r e f e r e n c e s

[1] El-Serag HB, Rudolph KL. Hepatocellular carcinoma:epidemiology andmolecular carcinogenesis. Gastroenterology2007;132:2557.

[2] El-Serag HB. Hepatocellular carcinoma: recent trends in theUnited States. Gastroenterology 2004;127(5 Suppl 1):S27.

[3] El-Serag HB. Epidemiology of hepatocellular carcinoma inUSA. Hepatol Res 2007;37(Suppl 2):S88.

[4] Ascha MS, Hanouneh IA, Lopez R, Tamimi TA, Feldstein AF,Zein NN. The incidence and risk factors of hepatocellularcarcinoma in patients with nonalcoholic steatohepatitis.Hepatology 2010;51:1972.

[5] Clark JM, Diehl AM. Nonalcoholic fatty liver disease: anunderrecognized cause of cryptogenic cirrhosis. JAMA 2003;289:3000.

[6] Sicklick JK. Correcting the misnomers of epithelial-mesenchymal relations. J Surg Res; 2012.

[7] Olive KP, Jacobetz MA, Davidson CJ, et al. Inhibition ofHedgehog signaling enhances delivery of chemotherapy in amouse model of pancreatic cancer. Science 2009;324:1457.

[8] Mazzocca A, Fransvea E, Dituri F, Lupo L, Antonaci S,Giannelli G. Down-regulation of connective tissue growthfactor by inhibition of transforming growth factor betablocks the tumor-stroma cross-talk and tumor progressionin hepatocellular carcinoma. Hepatology 2010;51:523.

j o u r n a l o f s u r g i c a l r e s e a r c h 1 8 7 ( 2 0 1 4 ) 3 7 7e3 8 5 385

[9] Amann T, Bataille F, Spruss T, et al. Activated hepatic stellatecells promote tumorigenicity of hepatocellular carcinoma.Cancer Sci 2009;100:646.

[10] Coulouarn C, Corlu A, Glaise D, Guenon I, Thorgeirsson SS,Clement B. Hepatocyte-stellate cell cross-talk in the liverengenders a permissive inflammatory microenvironmentthat drives progression in hepatocellular carcinoma. CancerRes 2012;72:2533.

[11] Sancho-Bru P, Juez E, Moreno M, et al. Hepatocarcinoma cellsstimulate the growth, migration and expression of pro-angiogenic genes in human hepatic stellate cells. Liver Int2010;30:31.

[12] Chan IS, Guy CD, Chen Y, et al. Paracrine Hedgehog signalingdrives metabolic changes in hepatocellular carcinoma.Cancer Res 2012;72:6344.

[13] Sicklick JK, Li YX, Choi SS, et al. Role for hedgehog signalingin hepatic stellate cell activation and viability. Lab Invest2005;85:1368.

[14] Choi SS, Omenetti A, Witek RP, et al. Hedgehog pathwayactivation and epithelial-to-mesenchymal transitions duringmyofibroblastic transformation of rat hepatic cells in cultureand cirrhosis. Am J Physiol Gastrointest Liver Physiol 2009;297:G1093.

[15] Sicklick JK, Li YX, Melhem A, et al. Hedgehog signalingmaintains resident hepatic progenitors throughout life. Am JPhysiol Gastrointest Liver Physiol 2006;290:G859.

[16] Witek RP, Yang L, Liu R, et al. Liver cell-derivedmicroparticles activate hedgehog signaling and alter geneexpression in hepatic endothelial cells. Gastroenterology2009;136:320.

[17] Yang L, Wang Y, Mao H, et al. Sonic hedgehog is anautocrine viability factor for myofibroblastic hepaticstellate cells. J Hepatol 2008;48:98.

[18] Omenetti A, Syn WK, Jung Y, et al. Repair-related activationof hedgehog signaling promotes cholangiocyte chemokineproduction. Hepatology 2009;50:518.

[19] Sicklick JK, Li YX, Jayaraman A, et al. Dysregulation of theHedgehog pathway in human hepatocarcinogenesis.Carcinogenesis 2006;27:748.

[20] Patil MA, Zhang J, Ho C, Cheung ST, Fan ST, Chen X.Hedgehog signaling in human hepatocellular carcinoma.Cancer Biol Ther 2006;5:111.

[21] Gallet A, Staccini-Lavenant L, Therond PP. Cellular traffickingof the glypican Dally-like is required for full-strengthHedgehog signaling and wingless transcytosis. Dev Cell 2008;14:712.

[22] Capurro M, Wanless IR, Sherman M, et al. Glypican-3: a novelserum and histochemical marker for hepatocellularcarcinoma. Gastroenterology 2003;125:89.

[23] Midorikawa Y, Ishikawa S, Iwanari H, et al. Glypican-3,overexpressed in hepatocellular carcinoma, modulates FGF2and BMP-7 signaling. Int J Cancer 2003;103:455.

[24] Capurro MI, Xu P, Shi W, Li F, Jia A, Filmus J, et al. Glypican-3inhibitsHedgehog signalingduringdevelopment bycompetingwith patched for Hedgehog binding. Dev Cell 2008;14:700.

[25] Yan D, Lin X. Opposing roles for glypicans in Hedgehogsignalling. Nat Cell Biol 2008;10:761.

[26] Pilia G, Hughes-Benzie RM, MacKenzie A, et al. Mutations inGPC3, a glypican gene, cause the Simpson-Golabi-Behmelovergrowth syndrome. Nat Genet 1996;12:241.

[27] Lin CW, Mars WM, Paranjpe S, et al. Hepatocyte proliferationand hepatomegaly induced by phenobarbital and 1,4-bis [2-(3,5-dichloropyridy-loxy)] benzene is suppressed inhepatocyte-targeted glypican 3 transgenic mice. Hepatology2011;54:620.

[28] Liu B, Bell AW, Paranjpe S, et al. Suppression of liverregeneration and hepatocyte proliferation in hepatocyte-targeted glypican 3 transgenic mice. Hepatology 2010;52:1060.

[29] Wang XY, Degos F, Dubois S, et al. Glypican-3 expression inhepatocellular tumors: diagnostic value for preneoplasticlesions and hepatocellular carcinomas. Hum Pathol 2006;37:1435.

[30] Shirakawa H, Suzuki H, Shimomura M, et al. Glypican-3expression is correlated with poor prognosis inhepatocellular carcinoma. Cancer Sci 2009;100:1403.

[31] Xu L, Hui AY, Albanis E, et al. Human hepatic stellate celllines, LX-1 and LX-2: new tools for analysis of hepaticfibrosis. Gut 2005;54:142.

[32] Omenetti A, Yang L, Li YX, et al. Hedgehog-mediatedmesenchymal-epithelial interactions modulate hepaticresponse to bile duct ligation. Lab Invest 2007;87:499.

[33] Nakatsura T, Yoshitake Y, Senju S, et al. Glypican-3,overexpressed specifically in human hepatocellularcarcinoma, is a novel tumor marker. Biochem Biophys ResCommun 2003;306:16.

[34] Tada M, Kanai F, Tanaka Y, et al. Down-regulation ofhedgehog-interacting protein through genetic and epigeneticalterations in human hepatocellular carcinoma. Clin CancerRes 2008;14:3768.

[35] Omenetti A, Popov Y, Jung Y, et al. The hedgehog pathwayregulates remodelling responses to biliary obstruction inrats. Gut 2008;57:1275.

[36] Olsen CL, Hsu PP, Glienke J, Rubanyi GM, Brooks AR.Hedgehog-interacting protein is highly expressed inendothelial cells but down-regulated during angiogenesisand in several human tumors. BMC Cancer 2004;4:43.

[37] Liu B, Paranjpe S, BowenWC, et al. Investigation of the role ofglypican 3 in liver regeneration and hepatocyte proliferation.Am J Pathol 2009;175:717.

[38] Ju MJ, Qiu SJ, Fan J, et al. Peritumoral activated hepaticstellate cells predict poor clinical outcome in hepatocellularcarcinoma after curative resection. Am J Clin Pathol 2009;131:498.

[39] Suzuki M, Sugimoto K, Tanaka J, et al. Up-regulation ofglypican-3 in human hepatocellular carcinoma. AnticancerRes 2010;30:5055.

[40] Yorita K, Takahashi N, Takai H, et al. Prognostic significanceof circumferential cell surface immunoreactivity of glypican-3 in hepatocellular carcinoma. Liver Int 2011;31:120.

[41] Yu DD, Yao M, Chen J, et al. [Short hairpin RNA mediatedglypican-3 silencing inhibits hepatoma cell invasiveness anddisrupts molecular pathways of angiogenesis]. ZhonghuaGan Zang Bing Za Zhi 2013;21:452.

[42] Zittermann SI, Capurro MI, Shi W, Filmus J. Soluble glypican3 inhibits the growth of hepatocellular carcinoma in vitroand in vivo. Int J Cancer 2010;126:1291.

[43] Roskams T, Rosenbaum J, De Vos R, David G, Desmet V, et al.Heparan sulfate proteoglycan expression in chroniccholestatic human liver diseases. Hepatology 1996;24:524.

[44] Friedman SL. Hepatic stellate cells: protean, multifunctional,and enigmatic cells of the liver. Physiol Rev 2008;88:125.

[45] Ho M. Advances in liver cancer antibody therapies: a focuson glypican-3 and mesothelin. BioDrugs 2011;25:275.

[46] Motomura Y, Ikuta Y, Kuronuma T, et al. HLA-A2 and -A24-restricted glypican-3-derived peptide vaccine induces specificCTLs: preclinical study using mice. Int J Oncol 2008;32:985.

[47] Ishiguro T, Sugimoto M, Kinoshita Y, et al. Anti-glypican 3antibody as a potential antitumor agent for human livercancer. Cancer Res 2008;68:9832.