Epstein-Barr Virus LMP1 Modulates the Malignant Potential of Gastric Carcinoma Cells Involving Apoptosis

AmericanJournal ofPathology, Vol. 152, No. 1,January 1998 Copyrngbt American SocietyforInvestigativePathology

Epstein-Barr

Virus LMP1 Modulates the Malignant

Potential of

Gastric Carcinoma Cells Involving

Apoptosis

Lai-Fa Sheu,* Ann

Chen,* Yu-Hui Wei,*

Kuo-Chieh Ho,t Jye-Young

Cheng,t

Ching-Liang

Meng,§

and Wei-Hwa Lee*

From theDepartmentsof Pathology,* Surgery,t and

Dentistry,s

Tn-Service GeneralHospital, National Defense Medical Center, and theDepartment of Botany,t National Taiwan University,

Taipei,

Taiwan, Republic of China

About 10% of

gastric carcinomas

including

lympho-epithelioma-like

carcinoma and

adenocarcinoma

are

associated with

Epstein-Barr virus

(EBV) infection.

In

EBV-associated gastric carcinomas, the

tumor

cells

express

Epstein-Barr nuclear

antigen

1

(EBNA-1) but

not

EBNA-2, -3A, -3B,

or

-3C, leader

protein,

or

latent

membrane

proteins

(LMPs) because of

gene

methyl-ation.

Only

a

few

exceptional

cases

have

LMP1

expres-sion in

tumorcells as

demonstrated

by

immunohis-tochemical

studies. To elucidate the

biological effects

of LMP1

and the

significance of its restricted

expres-sion in

EBV-associated gastric

carcinomas, the LMP1

genewas

transferred

into

EBV-negative gastric

carci-noma

celi

lines

(SCM1 and TMC1) and into

EBV-neg-ative

nasopharyngeal

carcinoma

(NPC)

cells

(HONE-1)

as a

controL The

biological effects of LMP1

in

gastric carcinoma cells

were

monitored in vitro

and

in

vivo.

These results showed that the

conse-quenceof

LMP1

expression is

a

growth enhancement

in

NPC

cells,

but

it is

a

growth suppression in gastric

carcinoma

cells.

The

LMPl-expressing gastric

carci-noma cells

had

a

reduced

growth

rate,

colony-form-ing

efficiency,

mean

colony size, and

tumorigenicity

and

a

lower

malignant

cytological grade. The reduced

growth

rate,

colony-forming efficiency, and

mean

colony size

were

partially reversible in vitro with

treatmentwith LMP1

antisense

oligonucleotide.

In

ad-dition, enhanced apoptosis

was

found in the

LMP1-expressing gastric carcinoma

cells.

This

suggests

that

LMP1 may

negatively modulate the

malignant poten-tial

of

gastric carcinoma

cellsviaan

enhancement of

apoptosis.

We

concluded that the restriction of

LMP1

expression in EBV-associated

gastric carcinomas

may

lead

toa

growth

advantage for

tumorcells

by

avoiding

LMP1

apoptotic effects and immunologically

medi-ated

elimination.

(Am

J

Pathol

1998,

152:63-74)

Epstein-Barr virus (EBV),

a

human herpesvirus,

can

cause

infectious mononucleosis and is

closely

associ-ated with endemic Burkitt's

lymphoma, nasopharyngeal

carcinoma (NPC), lymphoproliferative diseases in

immu-nocompromised

patients,1 nasal T-cell lymphomas,2'3

and

some

Hodgkin's

diseases.4 EBV

was

first shown to

be associated with gastric carcinomas

by Burke et al in

1990.5 This association

was

characterized

by the

pres-ence

of the EBV

genome

in

gastric carcinoma with

marked stromal

lymphocytic infiltration,

a

type

histologi-cally similar

to

nasopharyngeal lymphoepithelioma. EBV

involvement has also been demonstrated in

gastric

ade-nocarcinoma and adjacent

dysplastic epithelia.6 The

ev-idence of

carcinogenesis of EBV has been shown,

includ-ing

an

almost

uniform involvement of EBV in

tumor

cells

with

monoclonal proliferation and

an

elevation of

EBV-specific antibodies

in

patients.7'

Approximately

10%

of

gastric carcinomas

are

associated

with EBV infection.9

In

EBV-associated

gastric carcinomas, the

tumor

cells

ex-clusively

express

Epstein-Barr

nuclear

antigen

1

(EBNA-1) but

not

EBNA-2, -3A, -3B,

or

-3C,

leader

pro-tein,

or

latent

membrane

proteins

(LMPs) because

of

gene

methylation.7'10

Only

a

few

exceptional

cases ex-press

LMP1 in

tumor

cells

as

demonstrated

by

immuno-histochemical

studies.9'11

LMP1

is

an

integral membrane protein that consists of

a

24-amino-acid

amino-terminal

cytoplasmic domain,

six

membrane-spanning hydrophobic domains separated

by short

reverse

turns,

and

a

200-amino-acid

carboxyl-terminal

cytoplasmic domain.12'13 LMP1 aggregates

in

patches

on

the

plasma membrane14 and

canengage

the

tumor

necrosis factor

receptor

(TNFR)

family-associated

proteins for the TNFR

family.15'16 The interaction of LMP1

with

these TNFR

family-associated proteins

may prove

to

be

important

in

the

biological effects of LMP1. LMP1 has

properties of

a

classical

oncogene.

LMP1

can

transform

rodent

fibroblast cell lines

as

assayed by

tumorigenicity

in

nude mice

and foci formation.17,16

In

Burkitt's

lym-Supportedby grants from the National Science Council(NSC 85-2331-B-016-040), Tri-Service General Hospital (TSGH-C85-44 and TSGH-C86-46),and NationalDefense Medical Center(NDMC-85),Taipei,

Tai-wan, Republicof China.

Acceptedforpublication October 17, 1997.

Addressreprintrequests toDr.Lai-FaSheu, Department of Pathology, Tri-Service GeneralHospital, 8, Sec. 3,Ting-ChowRoad, Taipei,Taiwan,

Republic of China.

64

Sheu et al

AJPJanuary1998, Vol. 152, No. 1

phoma cells, LMP1 can induce villous

projections,

growth

in a

tight

clump, NF-KB activity, and

the

expression

of

activation markers

(CD23 and CD40), adhesion

mole-cules

(ICAM1, LFAl,

and

LFA3), and bcl-2

proto-onco-gene19'22

and can

inhibit p53-triggered apoptosis.23

EBV

recombinant genetic

analysis

indicates that

LMP1

is

essential

for primary B-lymphocyte transformation.24

In

squamous

epithelial

cells, the

consequences

of LMP1

expression

are

inhibition

of cellular

differentiation,25

mor-phological

transformation,26

and

induction of

epidermal

growth

factor

receptors.27 In

transgenic mice, LMP1 is

able

to

alter keratin

expression and induce hyperplasia of

the

skin.28 However,

LMP1 is

cytotoxic

in

a variety

of

cell

lines

and

induces

apoptosis

in

squamous epithelial cells

when

expressed

at a

high

level.2930 Collectively,

it

seems

that

the biological effects of LMP1

are

complicated.

Although

the relationship between EBV and gastric

carcinoma

has

been

demonstrated, the biological effects

of EBV infection and the

significance of restricted

LMP1

expression in

EBV-associated gastric carcinomas

are

still

unknown. To

our

knowledge, there has been

no

EBV-associated

gastric carcinoma cell line

or

EBV-immortal-ized

gastric

epithelial

cell

line

established.

Establishing

an

EBV-infected

gastric carcinoma

in

vitro

may

be

com-plicated

because of

the

absence of the EBV receptor

(CD21)

in

gastric epithelium

and

gastric carcinoma

cells.31

The evaluation

of

biological effects of

EBV

infec-tion

in

gastric carcinoma cells could

not

be

performed

in

our

study.

However, the expression of LMP1

gene in

EBV-negative gastric carcinoma cells by

gene

transfer is

one

of the

more

convenient methods

for the

initial

evalu-ation

of the

biological effects of EBV infection and the

significance

of restricted

LMP1

expression in

EBV-asso-ciated

gastric

carcinomas.

In

this

study,

the

LMP1 gene

was

introduced into

gastric carcinoma cells, and its

bio-logical

effects

were

monitored

in

vitro

and

in

vivo.

We

showed that LMP1 is able

to

negatively modulate

the

malignant potential of gastric carcinoma

cells

possibly

via

a

process

involving the enhancement of apoptosis.

Materials

and

Methods

Cells

and

Animals

Two

gastric

carcinoma

cell lines

(SCM1

and

TMC1)

were

used

for this

study. The SCM1

cell line

was

originally

derived

from

the

gastric

specimen

of

a

patient

with

poorly

differentiated tubular

adenocarcinoma

of

the stomach.

The

TMC1

cell line was derived

from

metastatic tumor

cells in the

lymph

node

of

a

patient

with

poorly

differen-tiated

adenocarcinoma of

the stomach. Both cell lines

were

EBV

negative, and

they

were

cultured

in

RPMI 1640

medium

containing

10%

fetal calf

serum

(FCS)

and

incu-bated

at

370C

in

an

incubator with 5%

CO2

and

water

saturation.

The

EBV-negative NPC cell

line

(HONE-1)32

was

used as a

comparative

study.

Pathogen-free

nude

(BALB/c

nu/nu)

mice

were

used for

tumorigenicity

as-says.

Construction of LMP1

Expression

Vector

The LMP1

gene was

excised from

the

EcoRI

D

fragment

of the B95.8 strain of EBV

by

Mlul

digestion.

This LMP1

gene

is a wild

type

without the

specific

mutations in exon

3

found

in

the NPC strain of EBV.3334 After

being

end-blunted

by

T4 DNA

polymerase,

the

LMP1

gene

was

cloned into the

pMAMneo expression

vector

(Clontech,

Palo

Alto, CA).

The

orientation of

the

LMP1 gene

relative

to

the MMTV

promoter

was

ascertained

by

restriction

mapping.

The

reconstructed plasmid

(pMAM-LMP1)

con-taining the LMP1

gene

and the

control

plasmid

(pMAM-neo)

were

used for cell transfection to establish the

LMP1-expressing cells

and

vector-transfected control

cells, respectively.

DNA

Transfection

DNA

transfection

was

performed

by the

lipofection

method according

to

the

manufacturer's

suggestion

(GIBCO

BRL,

Gaithersburg, MD).

In

brief, adherent cells

(2

x

105

cells)

in a 35-mm

tissue culture dish

were

transfected with

1

ml

of

a

DNA-liposome

complex.

The

DNA-liposome complex was prepared by mixing solution

A

(1

jig

of plasmid

DNA

diluted into

100

gl

of

Opti-MEM

I) and solution B (6 ,l of lipofectamine diluted

into 100

IlI

of

Opti-MEMI). After incubation of the DNA-liposome

complex

at

room temperature

for

15 to 45

minutes,

0.8 ml

of

Opti-MEM

medium

was

added and

mixed gently; then

the diluted

complex solution

was

overlaid

onto

rinsed

cells.

The cells

were

incubated

at

370C

in a

CO2

incuba-tor.

After

8

hours,

1

ml

of

growth medium containing

20%

FCS

was

added

without removing

the

transfection

mix-ture.

The

medium was

replaced with fresh

growth

me-dium 24

hours

after

the start

of

transfection. At

72

hours

after

transfection, cells

were

subcultured

at

a

1:10

dilu-tion

in

growth medium

containing G418

(400

,ug/ml).

G418-resistant

colonies were

selected.

Colonies

contain-ing

more

than

20

cells

were

counted

after

14

days.

Detection of LMP1

Expression by

Immunohistochemical

Stain

Detection

of

LMP1

expression in transfected cells and

their induced

tumors

in

nude mice

were

performed by

in

situ

immunostaining.

The cells were

grown

on

coverslips

to

approximately

70%

confluence,

washed twice with

ice-cold

phosphate-buffered saline

(PBS), and then fixed

in

methanol/acetone

(1/1,

v/v)

at

-200C for

30

minutes.

The

fixed cultured cells

were

rinsed

in

Tris-buffered

sa-line

(TBS;

145

mmol/L NaCI

and 20

mmol/L Tris, pH

7.6),

incubated

with 2%

H202

in

methanol for

10

minutes

to

destroy

endogenous peroxidase activity, blocked with

normal

serum

from

rabbits

(1:5

dilution)

for 5

minutes,

and then

incubated

with

mouse

monoclonal

antibody

S12

35

against

LMP1

for

2

hours.

After

three washes with

TBS,

the cells

were

exposed

to

rabbit anti-mouse

IgG

antibody conjugated

with

horseradish

peroxidase

(Santa

Cruz

Biotechnology,

Santa

Cruz, CA)

for

60

minutes and

washed with

TBS,

and the

positive

staining

was

devel-LMP1

Effects in Gastric Carcinomas

65

AJPJanuary 1998, Vol. 152, No. 1

oped

in

aminoethylcarbazole chromogen solution

(Dako,

Carpinteria, CA). The cells

were

lightly counterstained

with hematoxylin and

mounted for examination. The

for-malin-fixed and

paraffin-embedded

tissue sections of the

tumor masses

derived from the tumorigenicity assay in

nude mice were dewaxed,

pretreated by wet autoclaving

for

antigen

retrieval,36

rinsed in

TBS,

and stained by the

same procedures as the

culture cells except that the

secondary antibody was

replaced by

rabbit anti-mouse

IgG

antibody conjugated with alkaline phosphatase

(San-ta

Cruz

Biotechnology),

the

positive staining

was

devel-oped

in

naphthol phosphate-new fuchsin solution

(Dako),

and

the

light hematoxylin counterstain

was

omitted.

Western Immunoblotting

Proteins were extracted from

the

transfected

cells, and

tumor masses were derived

from the

tumorigenicity

as-say in nude mice

and the metastatic

NPC

in

neck

lymph

node. After washing with

PBS,

the cells and tumors were

lysed

in

RIPA

buffer

(150

mmol/L

NaCI,

50

mmol/L Tris,

pH

7.5, 5

mmol/L EDTA,

0.5% sodium

deoxycholate,

0.1%

sodium

dodecyl sulfate

(SDS), and

1

mmol/L

phe-nylmethylsulfonyl fluoride), homogenized by sonication,

and

centrifuged

at

15,000

x

g

for

10

minutes

at

40C.

The

extracted

proteins

in

the supernatant were

quantified

by

the

Lowry

assay

and then stored

in

aliquots

at

-70°C

before

use.

For

SDS-polyacrylamide gel electrophoresis,

extracted

protein

mixed with an

equal

volume

of

sample

buffer

(0.02% bromophenol blue,

4%

SDS,

4%

2-mercap-toethanol,

20%

glycerol,

and 50

mmol/L Tris)

was

boiled

for

8

minutes and

separated

on an

8.5%

polyacrylamide

gel. The separated proteins

were

electrotransferred

to

nitrocellulose membrane

(Amersham, Arlington Heights,

IL), probed

with mouse anti-LMP1

protein monoclonal

antibody

S12,

detected with

horseradish-peroxidase-conjugated rabbit

anti-mouse

IgG antibody (Serotech),

and

developed

in

enhanced

chemiluminescence

detec-tion reagents

(Amersham). For

the

LMP1-positive control

and

quantitative analysis, the B95.8 cells

were

included

in this assay. The

detection

of

bcl-2

expression in

LMP1-expressing gastric carcinoma cells and

vector-trans-fected cells

was

performed according

to

previously

de-scribed

methods.37

Growth Rate Assessment

Cells

were

plated

at a

density of

1

x

104

cells/ml

on

24-well

plates

in

RPMI 1640

medium containing

10%

FCS,

and this

growth medium was refreshed every other

day.

Each

day, triplicate

cultures were

trypsinized

indi-vidually, and the number of viable cells from each culture

was

determined

by

the

trypan blue exclusion method.

Colony-Forming Assay

The

vector-transfected

and

LMP1-expressing gastric

car-cinoma cells were

trypsinized,

and

viable cells

were

counted.

On

a

100-mm

culture

dish,

1000 cells were

seeded in RPMI 1640 medium

containing

10%

FCS,

and

this

growth

medium

was

refreshed

every 3

days.

After

2

weeks, the

culture cells were washed with

PBS, fixed

in a

10%

buffered

formalin,

and stained with a 1% solution

of

crystal violet to

determine

the

colony

number. This assay

was carried out in triplicate in three independent

exper-iments. For

the measurement of colony-forming

effi-ciency, the number

of colonies

per

dish, counted

under

low

magnification (x5),

was

divided

by the number of

viable cells

seeded and expressed as a percentage.

The

mean

colony

size

was

measured by counting

the

number

of cells

per

colony

at

x200

magnification in 60 randomly

selected colonies.

Tumorigenicity

Assay in Nude Mice

LMP1-expressing gastric

carcinoma cells

and

vector-transfected

cells

(2

x

106)

were

simultaneously

trans-planted

subcutaneously

at

different locations

on

the back

of each nude

mouse.

After

transplantation,

tumor

devel-opment was

monitored

daily. After

30

days, the

tumor

masses were

dissected,

weighed,

and cut into

three

pieces.

One piece

was

fixed

in

neutralized

10%

formalin

for

histopathological examination, and

the

others

were

frozen

in

liquid

nitrogen and preserved

at

-70°C for other

relevant studies.

Morphological

Study

The

morphological characteristics

of cultured

LMP1-ex-pressing

cells and vector-transfected cells

were

exam-ined in a

subconfluent

condition. The cellular

morphology

of each

tumor mass

derived

from the

tumorigenicity

as-say in

nude

mice was

evaluated

on

serial tissue sections

using hematoxylin and eosin

(H&E)

staining.

Detection of

Apoptotic

DNA

Fragmentation

The

DNA

fragmentation

assay was

performed

as

previ-ously described.38

In

brief,

the

LMP1-expressing gastric

carcinoma cells and

vector-transfected

cells were

cul-tured

to

approximately

70%

confluence

in

RPMI 1640

medium

containing

10%

FCS. The floating cells

were

collected

by

aspiration

of

the

medium,

and

the

adherent

cells were

collected

by trypsinization.

These

cells

were

washed twice

in

PBS containing

1

mmol/L

EDTA

and

0.1

mg/ml proteinase K, pelleted and

lysed

in 1

mmol/L

EDTA

containing

0.6%

SDS.

Sodium chloride

was

added

to 1

mol/L, and the solution

was

mixed by

gentle inversion

and

incubated

at

40C

overnight.

After

centrifugation

at

15,000

x

g for 60 minutes at

4°C,

low molecular weight

DNA was ethanol

precipitated from the supernatant. DNA

fragmentation

was

analyzed by subjecting

30

,ug

of

DNA

to

electrophoresis

on

a 1.5%

agarose

gel

using

Tris-acetate/EDTA

buffer, staining

with ethidium

bromide,

and

photography

under

ultraviolet illumination.

Antisense

Treatment

To

specify

the

biological effects of

LMP1,

the

cells

were

treated

with

LMP1 antisense

oligonucleotide for growth

66

Sheu

et

al

AJPJanuary 1998, Vol. 152, No. 1

Table 1. Comparison of the LMP1 Transfection in Gastric and Nasopharyngeal Carcinoma Cells

SCMi

TMC1

HONE1

Number of

Size of

Number of

Size of

Number of

Size

of

colonies

colonies

colonies

colonies

colonies

colonies

pMAMneo

Experiment

1

63

176

51

132

70

187

Experiment

2 71 182

53

124

75

199

Mean

67

179

52

128

73

193

pMAM-LMP1

Experiment

1

58

150

40

98

102

242

Experiment

2

50

142

35

107

90

213

Mean

54

146

38

103

96

228

The number of colonies and the size of colonies are expressed as the mean oftriplicate data of each experiment.

rate

assessment and

the

colony-forming assay.

LMP1

antisense

oligonucleotide (5' AG GTC GTG TTC CAT

CCT

CAG

GGC

3'), complementary

to a

23-bp

sequence

overlapping the

start

codon

of the

LMP1

mRMA,

was

synthesized using

an

Applied

Biosystem's 380B

DNA

synthesizer

(Applied Biosystems, Foster

City,

CA) with

a

phosphorothioate substitution

on

each

base.

Purity

was

assayed by

polyacrylamide gel electrophoresis after

two

cycles of ethanol precipitation. Sterile

aliquots of

1

mmol/L stock solution

were

stored at

-20°C and thawed

on

ice

immediately

before

use

in

each

experiment. The

LMP1-expressing

gastric

carcinoma

cells and

vector-transfected cells

were

cultured

in

growth medium

con-taining

3

,umol/L LMP1 antisense oligonucleotides. The

following

procedures

were

the

same

as

previously

de-scribed

for

growth

rate

assessment

and the

colony-form-ing assay.

Treatment of the control

oligonucleotide,

which

consisted of the

same

nucleotides

in

randomized

se-quence

(5' CT AGC AGT GCG AGG CGT TCT CTC 3'),

was

performed

in

the

control

study.

Results

Comparison

of LMP1

Transfection

Effects

in

Gastric Carcinoma

and

NPC Cells

Transfections,

under

the same

conditions,

were

per-formed

in

gastric carcinoma

cells

(SCM1

and

TMC1)

and

NPC cells

(HONE-1)

to

evaluate

the

differences of

LMP1

transfection effects in

gastric carcinoma and

NPC

cells.

Fourteen

days

after

transfection,

the

G418-resistant

col-onies

containing

more

than 20 cells

were

counted. The

mean

colony size

was

measured

by

counting

the number

of cells

per

colony

at

x200

magnification

in 60

randomly

selected

colonies. This

assay

was

carried

out

in

triplicate

in two

independent experiments.

These results are

sum-marized in

Table

1.

Using

this

approach

in

gastric

carci-noma

cells

(SCM1

and

TMC1),

we

found

consistently that

pMAM-LMP1 transfection of cells

that

express

LMP1

re-sulted in the

recovery

of

fewer

drug-resistant

cell colonies

and a smaller

mean

colony

size

than

when

the control

plasmid

(pMAMneo)

was

introduced

alone.

However,

the

NPC

cells

(HONE-1)

transfected

with

pMAM-LMP1

re-sulted

in

the

recovery of

more

drug-resistant

cell

colonies

and

a

larger

mean

colony

size than when the

pMAMneo

was

introduced

alone.

To determine

the

expression level

of

LMP1 in

transfected

gastric carcinoma

and

NPC

cells,

Western

analysis

was

performed

on

the

pooled

drug-resistant cells.

B95.8 cells and

a

specimen of

metastatic

NPC

with EBV association in neck

lymph node (the

re-sidual

lymphoid

tissue was removed under a

dissecting

microscope)

were

also

included

as

controls. As shown

in

Figure 1,

the LMP1

expression levels of

LMP1-trans-fected

gastric carcinoma and NPC cells

are

nearly

equal,

but

they

are

obviously

less than in

metastatic NPC

cells

and

B95.8 cells.

According

to

these

results,

the

conse-quence

of

the LMP1

expression

in

NPC cells is a

growth

enhancement,

but it is a

growth

suppression in

gastric

carcinoma

cells.

Establishment of LMP1

-Expressing

Gastric

Carcinoma Cells

To

further

characterize the biological

effects of

LMP1

in

gastric carcinoma cells,

two

monoclonal

drug-resistant

cell

lines (SCM-L14

and

-L22) with

persistent LMP1

ex-pression

were

randomly selected from

pMAM-LMP1-H

Pd P

U

>

I

XODA

I

V

s:X

V

U

~z;

z;

z

vzI

utI

Ez

:t

m

:

Figure 1.WesternblotassayforLMP1expressioninpMAM-LMP1-transfected gastriccarcinomaand NPCcells.Equalamountsofextracted protein (800

,tg)wereloadedineach lane except that50 .gofextracted protein of B95.8 cells and 150 ,ug of extracted protein ofmetastatic NPC wereloaded for quantitativeanalysis. NPC,metastaticNPC;SCM-V,vector-transfected SCM1 cells;SCM-LMP1,pMAM-LMP1-transfectedSCM1cells;TMC-V,

vector-trans-fected TMC1 cells; TMC-LMP1, pMAM-LMP1-transfected TMC1 cells; HONE-V, vector-transfected HONE-1cells;HONE-LMP1, pMAM-LMP1-trans-fectedHONE-1 cells. Thearrowindicates the positiveband of LMP1,

LMP1 Effects in Gastric Carcinomas

67

AJPJanuary1998, Vol. 152, No. 1

.S;

;.¢^:.:.jq,...,.:..

Afeitse

X$E''''X'.

.. .:

....

*.:w;...:

_...C;:,...

.

:r;,>..-.ir=w.N.gR.i-.,S1@U;<^,^---,.:iSg6;.;.:.w

,:t.:.;-t-:X}.:f.,.s...-....

'. -'

g REg@MiX i:n -zg2 R ^x

xa

Figure 2.ImmunostainingforLMP1 expression inLMP1-expressinggastric carcinomacells. a:Vector-transfected SCM1cells. b:LMP1-expressingSCM1cells (SCM-L22). c: Induced tumor derivedfromvector-transfectedSCM1 cells.d:Induced tumor derived fromLMP1-expressingSCM1cells(SCM-L22).e:

Vector-transfected TMC1 cells.f:LMPl-expressing TMC1cells(TMC-L1). g:Induced tumor derived fromvector-transfectedTMC1cells. h: Inducedtumorderived from

LMP1-expressingTMC1cells(TMC-L1). Anti-LMP1; magnification, X200.

transfected

SCM1

cells and

two

(TMC-L1 and -L4) from

pMAM-LMP1-transfected TMC1 cells. LMP1 expression

in

these selected

LMP1-expressing gastric carcinoma

cell lines and in the

tumors

induced

in

nude mice

was

demonstrated

by in situ immunostaining and Western blot

analysis. In culture, LMP1 expression

was

restricted

to

the

cytoplasm.

In

the

tumors

induced

in

nude

mice,

LMP1

expression

was

demonstrated

in

the

nuclei and

cyto-plasm (Figure 2). These selected LMP1-expressing

gas-tric

carcinoma cell lines all

expressed LMP1

at

approxi-mately 65 kd

as

demonstrated by Western blot analysis

(Figure 3). LMP1 expression

was

undetectable in

vector-transfected

SCM1

and TMC1 cells.

Thus, these selected

LMP1-expressing gastric carcinoma cell lines

were

con-firmed

to

constantly

express

LMP1 in

vitro and in vivo. In

addition, the LMP1 expression levels

in

these selected

LMPl-expressing gastric carcinoma cell lines

were

lower

than in B95.8 cells.

Reduced Growth Rate

of LMP1-Expressing

Gastric Carcinoma

Cells

in Vitro

The

growth

rates

of

LMP1-expressing gastric carcinoma

cells and

vector-transfected cells

were

evaluated

by daily

counting of viable cells. As shown in Figures

4

and

5, the

LMPl-expressing

SCM1 (SCM-L14 and -22) and TMC1

(TMC-Ll and -L4) cells showed

an

obviously slower

growth

rate

than

that of vector-transfected

SCM1

and

TMC1 cells. These

findings

suggest

that the

LMPl-ex-pressing gastric

carcinoma

cells have

a

reduced

growth

rate in

vitro.

Reduced

Colony-Forming Efficiency and

Colony

Size

of LMP1

-Expressing

Gastric

Carcinoma

Cells

in

Vitro

The results

of

the

colony-forming

assay are

summarized

in

Table

2.

The

colony-forming efficiency of

LMP1-ex-pressing

SCM1 cells

was

less

than that

of

vector-trans-fected

SCM1 cells. The

mean

colony

size

of

LMP1-ex-pressing

SCM1 cells

was

smaller than that of

vector-transfected

SCM1

cells.

These results indicate that

LMP1-expressing gastric carcinoma

cells

had

a

reduced

colony-forming efficiency and

mean

colony size

in

vitro.

Because

some

of

the

LMP1-expressing

and

vector-trans-fected

TMC1

cells grew in

suspension, the

colony-form-ing

assay could not be

tested

in

LMP1-expressing

TMC1

cells.

Reduced Tumorigenicity and Growth Rate of

LMP1-Expressing Gastric Carcinoma Cells

in

Vivo

A

subcutaneous

tumorigenicity

assay in nude mice

was

performed

to test

the

difference in

tumorigenicity of

LMP1-expressing

gastric

carcinoma cells and

vector-A

68

Sheu et al

AJPJanuary 1998, Vol. 152, No. 1

a

_ _ cn- cn

*

04

ri

v-I%

(

P2

rI

-

N - -

a

Z

I I I I I I I I @

e.

a

6

'd

d'

d'

eO

e

CO uz x m Co rq tn ux ux A

b

+

pqri i-

'4'

'4'

>

I

A

I A >

a

co I I I I I I I

u

u

r.

XtoS

S «

Figure 3. LMP1 expression in selectedLMP1-expressingSCM1 celllines (a) and TMC1 cell lines (b). Equal amountsofextractedprotein(250,ug)wereloaded ineach laneexcept that only 50 ,ug of extracted proteinof B95.8 cells was loaded for quantitativeanalysis. nm, inducedtumor mass innudemouse;AS,with

antisensetreatment;SCM-V,vector-transfected SCM1 cells; TMC-V,vector-transfected TMC1 cells. The arrow indicates the positive band of LMP1,approximately 65kd.

transfected cells. As shown

in

Table

3,

the

tumorigenesis

frequency of

SCM-L14

was 71%

and

of

SCM-L22

cells

was

62%,

which

was

significantly lower than that

of

vec-tor-transfected

SCM1 cells

(100%;

P

<

0.05).

The

growth

rates

of

SCM-L14 and -L22 cells, assessed by the weight

of the induced

tumor

masses, were

significantly slower

than that of

vector-transfected

SCM1 cells

(P

<

0.05).

Similar results

were

obtained from

LMP1-expressing

TMC1 cells (Table

4). The tumorigenesis

frequency of

TMC-L1 cells

was

13% and

of

TMC-L4 cells

was

0%,

which

was

significantly lower than that of

vector-trans-fected

TMC1 cells

(100%;

P

<

0.05).

The

growth

rate

of

LMP1-expressing TMC1 cells

was

also

significantly

slower than that

of

vector

control

TMC1 cells

in

the

animal

a

2 4 6 8 10 12 14

C

2 2 4 6 a 10 12 14 150 100 50 o0 0

b

I 2 4 6 8 10 12 l 50 50 0 x 1-'4-0 0 z 200 150 100 2 50-0 2 4 6 a 10 12 14 40 30 20 10 0 60 40 31 2C I14

a

I 2 4 6 8 10 12 50 40 30 20 10 o0 4 C 2 4 6 6 10 12 14 50 40 30 20 10 -b 2 4 6 a lo 12 1 4

d

0 2 4 6 8 10 12 14 Time(days)

Figure4.Growthcurves ofLMP1-expressing SCM1 cells and

vector-trans-fectedcells withorwithout LMP1 antisensetreatment. a:Growthcurvesof SCM-L14 (A), SCM-L22 (ED, and vector-transfected cells (0). b: Growth

curvesofvector-transfected cells(0)and with LMP1 antisensetreatment(0).

C:Growthcurvesof SCM-L14 cells(A)and with LMP1 antisensetreatment (A). d: Growth curves of SCM-L22 cells (0) and with LMP1 antisense

treatment(-).

Time(days)

Figure5. GrowthcurvesofLMPl-expressingTMC1cells and

vector-trans-fected cells withorwithoutLMP1antisense treatment. a: Growthcurvesof

TMC-Ll(AL),TMC-L4(E),and vector-transfectedcells(0).b: Growthcurves

of vector-transfected cells(0)and withLMP1antisense treatment(0). C:

Growthcurvesof TMC-Ll cells(A)and with LMP1 antisense treatment(A). d: Growthcurvesof TMC-L4 cells(E)and withLMP1antisense treatment (A). 150 100 0 x 4-0 z 50 200 150 100 50 o0 200 an so 0 An -L.

LMP1 Effects

in

Gastric Carcinomas

69

AJPJanuary

1998, Vol. 152, No. 1

Table

2. Colony-Formring Efficiency (CFE) of

LMP1-Expressing SCM1 Cells

Mean colony

size

CFE (%)

(cells/colony)

Cell lines

AS (-)

AS

(+)

AS (-)

AS

(+)

SCM-V 34

32

235

224

SCM-L14

12

21

163

194

SCM-L22

8

17

138

186

CFE, colony-fqrming efficiency; AS (-), without LMP1 antisense treatment; AS (+), with LMP1 antisense treatment. CFE = (colonies formed/cells inoculated) x 100.

test

(P

<

0.05). These findings

suggest that

LMP1-ex-pressing gastric carcinoma cells

have a

reduced

tumor-igenicity and growth

rate in vivo.

Morphological Alterations of LMP1 -Expressing

Gastri6

Carcinoma Cells

Morphological alterations of the

LMP1-expressing gastric

carcinoma cells

were

examined

in

vitro

and

in

vivo. In

subconfluent culture conditions, the

vector-transfected

SCM1

cells

predominantly displayed

more

spindle

shapes and

grew

in

a

less

organized

pattern

(Figure

6a).

However, the

LMP1-expressing SCM1

cells

showed

po-lygonal

or

ovoid shapes

and grew

in

a

cohesive

and well

organized

pattern

(Figure 6b). The vector-transfected

SCM1 cell

tumors

induced in nude mice showed

a

higher

malignant cytological grade, characterized by increased

Table 3.

Tumorigenesis

of

LMP1-Expressing SCM1 Cells

(SCM-L14

and

-L22)

and

Vector-Transfected

Cells

(SCM-V)

Tumor weight

(g)

SCM-V-

SCM-Li

4-

SCM-L22-derived

derived

derived

Mouse

tumors

tumors

tumors

1 0.144

0.028

.

0.013

2

0.063

0

0

3

0.124

0.042

0.033

4

0.250

0.100

0.030

5

0.190

0

0.005

6

0.287

0.113

0.032

7

0.246

0.103

0.066

8

0.272

0.085

0.037

9

0.246

0.278

0.180

10

0.260

0

0 11

0.182

0.069

0.030

12

0.140

0.145

0.005

13

0.057

0

0

14

0.233

0.055

0

15

0.116

0.048

0

16

0.333

0.078

0.063

17

0.398

0

0

18

0.128

0

0

19

1.165

0.214

0

20

1.681

0.432

0.227 21

1.889

0.788

0.292

Tumorigenesis was 100% for

SCM-V-derived

tumors, 71% for

SCM-L14-derived tumors, and 62% for

SCM-L22-derived

tumors. Differences

were considered significant at P < 0.05 by Fisher's exact test for

tumorigenicity and Mann-Whitney U test for growth rate (assessed by

weight).

Table

4. Tumorigenesis of LMPl-Expressing TMC1 Cells

(TMC-Ll and

-L4)

and

Vector-Transfected Cells

(TMC-V)

Tumor weight

(g)

TMC-V-

TMC-L1-

TMC-L4-derived

derived

derived

Mouse

tumors

tumors

tumors

1

0.804

0

0

2

0.653

0

0

3

0.596

0

0

4

0.761

0

0

5

0.556

0

0

6

0.314

0

0

7

0.643

0.190

0

8

0.285

0.025

0

9

0.507

0

0

10

0.869

0

0

11

0.690

0

0

12

0.707

0

0

13

0.655

0

0

14

0.341

0

0

15

0.716

0

0

Tumorigenesis was 100% forTMC-V-derivedtumors, 13%for

TMC-Li-derived

tumors, and 0% for TMC-L4-derived tumors. Differences were considered significant at P < 0.05 by Fisher's exact test for

tumorigenicityand Mann-Whitney U test for growth rate (assessed by weight).

cellular

polymorphism, conspicuous

eosinophilic

nucle-oli,

frequent mitosis, and relatively

scanty

cytoplasm

(Fig-ure

6c).

In contrast,

the

LMP1-expressing

SCMI1

cells

displayed

a

lower

malignant

cytological

grade,

charac-terized

by

monotonous tumor

cells with

polygonal

or

ovoid

nuclei,

inconspicuous

nucleoli,

occasional

mitosis,

and moderate

cytoplasm

(Figure 6d).

No

identifiable

dif-ferences between the mucin

production

of

vector-trans-fected

or

LMP1-expressing

SCM1 cells

were

found

by

serial tissue section

examination.

Similarly,

a

lower

grade

of

malignant

cytological

features

was

found

in

LMP1-expressing

TMC1 cells. The cultured vector-transfected

TMC1

cells had

a

tendency

to

grow

clumped together,

and

some

of them

formed clusters

(Figure

6e).

In

con-trast,

the

LMP1-expressing

TMC1 cells

mostly

grew

in

a

homogeneous monolayer attached

pattern

(Figure

6f).

The

vector-transfected

TMC1

cell

tumors

induced

in

nude mice showed

a

higher malignant

cytological

grade,

characterized

by cellular

polymorphism

with

giant

cell

formation, prominent eosinophilic nucleoli,

frequent

mito-sis, and

scanty

cytoplasm

(Figure

6g).

However, the

LMP1-expressing TMC1

cell

tumors

displayed

a

lower

malignant

cytological

grade, characterized by

monoto-nous

tumor

cells with

polygonal

or

ovoid

nuclei,

incon-spicuous

nucleoli,

occasional

mitosis,

and

increased

pro-duction of mucin

accumulated

in

the

intercellular

spaces

(Figure

6h).

Similar

morphological alterations

were

also

observed in the

pooled

LMP1-expressing SCMI1

and

TMC1

cells

(data

not

shown).

These

results

suggest

that

the

LMP1

-expressing

gastric

carcinoma cells

displayed

a

lower

malignant

cytological

grade

than that

of

70

Sheu

et

al

AJPJanuary 1998, Vol. 152, [Vo.1

Figure 6. Morphologicalalterations ofLMP1-expressinggastric carcinomacells.a:Vector-transfectedSCM1cells.b:LMP1-expressingSCM1cells(SCM-L22).C:

Tumorinducedbyvector-transfected SCM1 cells.d:TumorinducedbyLMP1-expressingSCM1 cells (SCM-L22). e:Vector-transfectedTMC1 cells.f:

LMP1-expressing TMC1cells(TMC-L1).g: Tumor inducedbyvector-transfected TMC1 cells. h: Tumor induced byLMP1-expressingTMC1cells (TMC-L1).Thearrows

indicatemucinaccumulatedintheintercellularspace. H&E stain(C,d, g,andh); magnification,X 200(a,b, d,e,f,andh)and X300(Candg).

Enhanced

Apoptosis of LMP1

-Expressing

Gastric Carcinoma Cells

Under routine

conditions,

more

floating

cells

appeared

in

the culture

of

LMPl-expressing

gastric carcinoma cells

than

in

the culture

of vector-transfected

cells.

To

charac-terize

this

phenomenon,

the low molecular

weight

DNA

isolated

from

the

total cell

population, including attaching

and

floating

cells, of the

LMPl-expressing

gastric

carci-noma

cells and

vector-transfected

cells were

analyzed

by

agarose

gel

electrophoresis.

DNA

fragmentation

in

the

ladder pattern was noted in

LMP1-expressing

SCM1

cells

(SCM-L14

and

-L22)

and

TMC1 cells

(TMC-Ll

and

-L4),

but it was not

clearly identified

in

vector-transfected cells

(Figure 7).

In serum

starvation

conditions, DNA

fragmen-tation

in

the

ladder

pattern

was

also

found

in

vector-transfected

SCM1

and

TMC1

cells

(data

not

shown).

These results suggest that

apoptosis

was

enhanced in

LMP1-expressing

gastric

carcinoma cells.

Lack of

Induced

bc1-2

Expression

in

LMP1-Expressing Gastric Carcinoma Cells

The

expression

of

bcl-2

in

LMPl-expressing

gastric

car-cinoma cells and

vector-transfected

cells

was

assayed

by

Western blot. No

LMPl-inducible

bcl-2

expression

in

LMPl-expressing

gastric

cells or

constitutive

bcl-2

ex-pression

in

vector-transfected

cells

was

detectable. The

expression of bcl-2

was

detected

only

in

MCF-7

cells

as

a

positive

control

(data

not

shown).

These results

suggest

that LMP1

cannot

induce

bcl-2 expression

in

gastric

car-cinoma

cells,

which are

compatible

with

the similar

find-ings

in

primary

B

cells, rodent

fibroblasts, and

immortal-ized

squamous

epithelial

cells.30,39

Specificity

of the

Biological

Effects of LMPl

by

LMP1 Antisense Treatment

After antisense

treatment

for

3

days,

the LMP1

expres-sion level

of

LMP1-expressing SCM1

cells

(SCM-L14

and

Figure7.DNAfragmentationofLMP1-expressinggastric carcinomacells.Lane

1 vector-transfectedSCM1cells; lane2,SCM-L14cells; lane 3,SCM-L22cells; lane4,vector-transfected TMCcells; lane 5, TMC-L1 cells;lane6,TMC-L4cells.

The molecularmarker, pCEP4 plasmid (Invitrogen)wasdigested byHinfI.

LMP1

Effects in

Gastric Carcinomas

71

AJPJanuary1998, Vol. 152,No. 1

-L22)

and

TMC1 cells (TMC-L1

and

-L4)

was

assayed by

Western blotting, which showed obvious inhibition of

LMP1

expression (Figure 3). After antisense treatment,

the

vector-transfected

SCM1

and TMC1 cells showed no

significant changes

in

growth rate,

but

the

LMP1-ex-pressing

SCM1

and TMC1

cells showed an obviously

increased

growth rate when compared with the relative

LMP1-expressing

gastric carcinoma cells without

anti-sense treatment

(Figures

4

and

5). In the colony-forming

assay

with

antisense

treatment,

the

vector-transfected

SCM1 cells showed

a

slightly

reduced

colony-forming

efficiency and

mean

colony

size, but the

LMP1

-express-ing

SCM1

cells

(SCM-L14

and

-L22)

showed an

in-creased

colony-forming efficiency and

mean

colony size

(Table 2).

The

cellular morphology

and

colony

shape of

LMP1-expressing

and vector-transfected

SCM1 cells

showed

no obvious

change after antisense

treatment.

The growth rate,

colony-forming efficiency,

and mean

colony

size

of

LMP1-expressing gastric carcinoma cells

was

partially

reversed by LMP1

antisense

treatment. For

treatment

with

the control

oligonucleotide, there

were no

significant

findings

except a

negligible toxic

effect

(data

not

shown).

Therefore, it is

conceivable that the

reduction

of

growth

rate,

colony-forming efficiency,

mean

colony

size,

and

tumorigenicity and the

morphological

alter-ations of

LMP1-expressing gastric cells

resulted

from

the

biological effects of

LMP1.

Discussion

In

this

study,

we

explored

the

biological

effects of the

LMP1 gene

(from the B95.8 strain of EBV without the

specific 30-bp deletion

in exon 3

found

in

the

NPC strain

of

EBV33,34)

in

gastric carcinoma

by

transfecting it

in

EBV-negative gastric carcinoma cells. The comparison of

LMP1

transfection effects between gastric carcinoma

and

NPC cells

were

performed,

and these results

sug-gest that

the

consequence

of

LMP1

expression

in

NPC

cells is

a

growth

enhancement,

but it

is

a

growth

sup-pression

in

gastric

carcinoma

cells. Multiple

monoclonal

LMP1-expressing

gastric

carcinoma cell lines

were

ran-domly

selected

for the additional characterization of

the

biological effects of

LMP1 in

gastric carcinoma cells. The

LMP1-expressing

gastric carcinoma cells had a reduced

growth

rate,

colony-forming efficiency,

mean

colony size,

and

tumorigenicity and

a

lower

malignant cytological

grade when compared with

vector-transfected

cells.

When

LMP1

expression

was

blocked with

LMP1

anti-sense

in

vitro, the reduced growth

rate,

colony-forming

efficiency,

and mean

colony size of

LMP1-expressing

gastric

carcinoma cells was

partially

reversed. In

addi-tion,

there was no

LMP1-inducible bcl-2

expression to

protect the

gastric carcinoma cells

from

apoptosis

in-duced

by

LMP1.

Our

findings

showed that LMP1 can

negatively

modulate the

malignant potential of

gastric

carcinoma cells

possibly

via the

enhancement of

apopto-sis. LMP1 is a

unique viral oncoprotein

on

account

of

its

oncogenic potential

in

rodent

fibroblast, B-lymphocytes,

and

epithelial

cells,40

but it

also exhibits

cytotoxicity

in a

variety of

cell

lines29

and induces apoptosis in squamous

epithelial cells

when

expressed

at

high levels.30

Collec-tively, these findings

suggest

that

LMP1

has paradoxical

biological effects:

an

oncogenic potential

associated

with

cytotoxicity. However,

our

findings

also suggest

that

gas-tric

carcinoma cells

are more

sensitive

to

the

toxic

effects

of LMP1 than NPC cells.

Oncogenes associated

with

toxicity

have also

been

demonstrated

in

v-abl,41

v-src,42

v-re/,43 and c-myc.44

For

example,

c-myc

is essential for cell

proliferation, but

it

also

can

induce cellular apoptosis. The activation of

ap-optosis

by

c-myc

is

dependent

on

regions that

are

also

essential for

transformation.44

Similarly,

mutant

analysis

has

shown that the

same

domains of

LMP1 are

respon-sible

for both the

transforming

function and the toxic

effects.29

The

co-expression

of

c-myc

and bcl-2 is

com-monly

present in human

follicular

lymphoma.45

In

addi-tion, the

apoptotic

effects of

c-myc can

be blocked

by

the

ectopic expression of bcl-2.46 Similarly, the coexistence

of activation of

oncogenes

and inactivation of the

p53

gene

has been

commonly observed

in

human

can-cers.47-49

It was

proposed

that

development

of

tumors

would involve the

deregulation of cellular proliferation

and

suppression of

apoptosis.46 In EBV-infected

normal

B

cells, the EBV

gene

products

that

initiate and maintain

B

cell growth and immortalization

in

vitro

include six

nu-clear

antigens, EBNAs 1, 2, 3a,

3b, 3c,

and

LP,

and two

latent membrane

proteins, LMPs

1

and 2.50

The

associ-ated

overexpression of bcl-2 also

was

found.51 The in vitro

studies have

suggested

that

LMP1 is

essential for

pri-mary B cell

transformation,24

which can

induce

bcl-2

expression

to

protect B

cells from

apoptosis20

and

coop-erate

with

EBNA2 for the induction

of

B

cell activation.52

On

the

other

hand, there

were no

similar

effects

demon-strated

in

the

EBV-infected normal

T

cells.

In

EBV-in-fected normal

human

thymocytes (CD21

'),

the EBV

ge-nome

is

in

linear

form, and its

gene

expression includes

EBNA1

from the

Fp promoter, EBNA2,

ZEBRA, RAZ,

and

gp350/220

but

no

LMPs

or

EBERs, which has

some

of the

same

specific

characteristics

of

lytic replication.53

How-ever,

outgrowth of immortalized T-cell clones from

EBV-infected thymocytes have

not

been

observed, which

sug-gests

that,

in contrast

with B

cells,

EBV

may

be

incapable

of

immortalizing T cells

in

vitro due

to

the lack

of

suppres-sion of

apoptosis.54

The EBV-encoded

LMP1 gene can

induce the

expres-sion

of

ICAM1,

LFA1 and

LFA3,19'25

which are the

essen-tial mediators for

conjugate formation between

T

lympho-cyte

and

target

cells.55

In

addition,

cytotoxic

T

lymphocyte

epitopes have also been identified

in the

LMP1

gene.56

Collectively,

it

was

predicted that the

ex-pression of LMP1 may itself be influenced

by biological

effects

in

host cells and the

patient's

immunosurveillance

function.

In

EBV-associated

neoplasms,

LMP1

expres-sion is

frequently

observed in

NPC,57

Hodgkin's

diseas-4,58 23 -eae ypo

es,

nasal T-cell

lymphomas,2'3

AIDS-related

lympho-mas, and

post-transplant

lymphomas,59

62 but it is

rarely

present in endemic Burkitt's

lymphoma63-65

or

gastric

carcinomas.7'9'11

A

characteristic

genotype

of

LMP1

with

a

30-bp

deletion and several

single-base

mutations

at

the

carboxyl-terminal

end

of

exon

3

were

initially identified

in