Tumor necrosis factor-a induction by reovirus serotype 3 - CiteSeerX

5 downloads 0 Views 885KB Size Report
reovirus, when used in combination with. 1,3-bis(chloroethyl)-1- nitrosourea. (BCNU) chemotherapy, mediates the rejec- tion of murine ascites tumors. Surviving.
Tumor

necrosis Anthony

factor-a

L. Farone,

Department

induction

Patricia

of Micro&iology,

G.M.

Miami

O’Brien,

University,

and

Oxford,

Abstract: We have reported previously that reovirus, when used in combination with 1,3-bis(chloroethyl)-1nitrosourea (BCNU) chemotherapy, mediates the rejection of murine ascites tumors. Surviving anhnals reject a challenge with the same, but not a different, tumor, which suggests that tumor-specific immunity is induced by the treatment regimen. The present study was designed to characterize the interaction between reovirus and niurine peritoneal macrophages, both in vitro and in vivo, to determine whether such a relationship may play a role in immune modulation resulting in tumor rejection. The results demonstrated that reovirus can efficiently infect peritoneal macrophages in vitro and stimulate the secretion of tumor necrosis factor-alpha (TNF-a). In vivo administration of reovirus, however, did not produce high levels of infection in peritoneal exudate cells, even though the cells were stimulated to express detectable levels of membrane TNF-a. These results suggested that infection is not necessary for TNF-a expression and this hypothesis was supported by the observation that this expression was also stimulated in vitro by UV-inactivated reovirus. These findings suggest that one mechanism for immune stimulation by reovirus may be through the induction of TNF-a.-Farone, #{163}L.,O’Brien, P.C.M., Cox, D.C. Tumor necrosis factor-a (TNF-a) induction by reovirus serotype 3.J. Leukoc. Biol. 53: 133-137; 1993. Key

Words:

tumor

peritoneal

necrosis

factor-a

Macrophages tion

are

of the

come the infection

important

interest to the

it may

not

the

highly

and

reactive

function

nature

as a typical

bevirus

of

host

the

during

attachment and entry response leading to the resulting in the recruitcells important in viral

inflammatory

mediator

stimulated

by

virus infection is the cytokine, tumor necrosis factoralpha (TNF-a) (1-3, 5, 7). TNF-a, which is predominantly secreted by macrophages, is also expressed on the cell surface

(8).

tributed

study ment fected

to

sourea therapy mented

TNF-a;

biological those

are its antitumor and stimulation, cells

We have UV-inactivated the

Numerous

of

effects particular

activity and

and

cytolytic

have

been

interest immune effects

to cell on

atthis

recruitvirus-in-

(9-13).

previously reovirus

chemotherapy

reported serotype agent

that either infectious 3 in combination

or with

1 ,3-bis(2-chloroethyl)-1-nitro-

(BCNU) is a potent of murine tumors genome

of

causes

by two a variety

human

immune stimulator in the (14). Reovirus has a segdouble-stranded RNA (dsRNA) en-

Journal

protein coats. of mammalian

disease

Even

though types,

cell

reovirus it rarely

(15).

In this study, we have demonstrated that murine peritoneal macrophages can be productively infected by reovirus serotype 3 and that reovirus treatment stimulates the production of TNF-a both in vitro and in vivo. These results suggest that some of the immunomodulatory effects mediated by reovirus may occur through TNF-a.

MATERIALS

AND METHODS

Mice Female lan

CD2F1

(Balb/c

Laboratories,

x DBA/2)

mice,

Indianapolis,

IN),

age

were

7-9 wk used

in

(Harall

ex-

periments.

Macrophage

Collection

Thioglycollate-stimulated previously

the

cells

dish

macrophages

described

were

(Falcon

(16).

allowed

For

to

Labware,

were

the

adhere

to

Oxnard,

collected

infectious

CA)

center

a polystyrene for

90

mm

as assay,

petri at 37#{176}Cin

complete medium [RPMI 1640, Sigma Chemical Co., St. Louis, MO; 10% heat-inactivated fetal bovine serum (FBS; Hyclone Laboratories, Inc., Logan, Utah); 100 penicillin Co.,

and

and Louis,

St.

cells with

were

a cell

100 .Lg/ml MO]. After

removed

scraper,

and washed

streptomycin, incubation,

the

adherent cells with complete

Sigma non-

were colmedium,

counted.

Virus was

Reovirus

virus

3

C. Cox

capsidated can infect

lected

regula-

have recently targets for

serotype

45056

U/ml Chemical

#{149} membrane

induction

(1-5) and as potential

virus infection. In particular, may stimulate an inflammatory induction of immunomodulators, ment of immune effector One

Donald Ohio

adherent

in

response

focus of (6). Due

clearance.

#{149} Mac-2

#{149}reovirus

immune

macrophage,

ha ge

macrap

by reovirus

previously ing to the

purified

UV-inactivation Virus ticles

infectivity

7.2)

0.9

mM

infected

mouse

L929

cells

as

accord-

of Reovirus was

suspended

(D-PBS;

0.49

from

(17). Virus was then titrated of Shaw and Cox (18).

described procedure

inactivated

by

in Dulbecco’s mM

MgC12,

CaC12, 0.14

at a concentration

2.68 mM

irradiating

phosphate mM NaCl,

KCI,

1.47

8.03

mM

of 5 x 1012

virus

buffered

particles/mI

mM

par-

saline K1-12PO4,

Na2HPO4,

with

pH

a ger-

Abbreviations: BCNU, I ,3-bis(2.chloroethyl)-1-nitrosourea; BRMA?, Biological Response Modifiers Program; D-PBS, Dulbecco’s phosphate buffered saline; dsRNA, double-stranded RNA; FBS, fetal bovine serum; MOl, multiplicity of infection; pfu, plaque forming units; TNF.ct, tumor necrosis factor-alpha. Reprint requests: Anthony L. Farone, Harvard School of Public Health, Respiratory Biology Program. Department of Environmental Health, Bldg. 1, Rm. 307, 665 Huntington Ave., Boston, MA 20115. Received

June

of Leukocyte

5, 1992;

Biology

accepted

September

Volume

4, 1992.

53,

February

1993

133

micidal

lamp,

l.LW/cm2) was

(254

for

determined

Virus

nm)

mm.

15

at The

using

the

a distance

of

effectiveness

reovirus

15 of

titration

cm

(1250

Infectious

protocol.

vivo,

tion

virus

as

was

injected

described

above,

4 h before

above.

and

the

Antibody

Cells

macrophage

were

percentage

of

washed

collectwice,

infected

cells

plated

was

deter-

mined.

Culture

Macrophages

were

for TNF-a plated

at

medium in 24-well culture for 90 mm at 37#{176}C.After cells were removed with medium containing 1 lipopolysaccharide (LPS; MO), reovirus at a MOI equivalent to a MOI of was added to the adherent 4 h at 37#{176}Cand cell-free lected.

Release

3 x 106

cells/ml

in

complete

plates and allowed to adhere incubation, the nonadherent D-PBS, and 1 ml of complete .ig/ml Escherichia coli 0111:84 Sigma Chemical Co., St. Louis, of 10, UV-inactivated reovirus 10, or complete medium alone cells. Cells were stimulated for culture supernatants were col-

sample

(U)-neutralized

Detection

Peritoneal exudate mice injected with thioglycollate alone. to 1 x I 02 cells/mI azide

(w/v),

gelatin suspensions

Statistical

alone.

After

glutaraldehyde 1 h and stained

134

Journal

incubation,

Chemical equivalent

noninfectious the cells

Biology

St.

Louis, of

St. Louis, MO) for blue (Fisher Scien-

Volume

53,

St.

anti-murine CA) and San Mac-2

Louis,

MO;

Detroit, ml

were

MI]. then

TNF-cx antibody streptavidin-phycoerythrin Jose, CA) followed

monoclonal

0.1%

Cell stained (Pharby

antibody

60

tissue

%

Mac-2,

in

then

of Mac-2

% (%

subtracting binding

the the

percent isotype

used

in

with

mTNF-a

cells

Mac-2,

the

of cells controls.

following

+

%

nonadjusted

equation:

=

mTNF-a

mTNF-a

exhibiting The

x

100%.

Mac-2)

Analysis

LPS,

reovirus, or medium were fixed with 2.5%

(Sigma Chemical Co., with 0.05% methylene

of Leukocyte

Co., amounts

Co.,

Laboratories,

culture supernatant (M3/38.2.8.HL.2 hybridoma; a generous gift of Dr. Meena Subramanyam, Tufts University, Boston, MA) and 1 p.g biotin-conjugated goat affinity purified F(ab’)2 anti-rat IgG, whole molecule (Cappel Research Products, Durham, NC), followed by 1 p.g fluorescein-conjugated streptavidin (Pierce Chemical Co., Rockford, IL). Isotype controls were used to set markers on the double negative population. Data from the cell populations were acquired and analyzed with a FACScan flow cytometer (Becton-Dickinson). The percentage of the total Mac-2 cell population that was expressing mTNF-a was calculated by first determining the percentage of cells expressing Mac-2 alone, or, both Mac-2 and

j.tg/ml

(Sigma

In vivo

106 cells/0.1

anti-mouse

were

D

Chemical

Difco

with 1 .tg biotinylated Mingen, San Diego, (SAPE; Becton-Dickinson,

values

received

TNF-a

containing

rat

(U))

cells were collected as above from thioglycollate followed by reovirus or Cells were washed and resuspended in azide D-PBS [A-PBS; 0.1% sodium

Sigma

(w/v),

Culture supernatants were assayed in triplicate for TNF-a activity immediately after collection using the L929 cell cytotoxicity method (19). Briefly, 5 x 10 L929 cells/well were incubated at 37#{176}Cwith culture supernatants or recombinant TNF-a [Biological Response Modifiers Program (BRMP) standard] serially diluted in MEM supplemented with 5% FBS, penicillin, and streptomycin in 96-well microtiter plates for 18 h in the presence of 1 Actinomycin

activity

(U)

of Membrane

by

MO). Control wells infectious reovirus,

sample

sample

TNF-a specific

TN F-a Assay

of TNF-a

% Neutralization=

p.1 of

Macrophage

neutralization

TNF-u cytotoxic activity was specifically neutralized with anti-TNF-a antibody (Genzyme Corporation, Boston, MA) by incubating culture supernatants with antibody dilutions for 1 h at 37#{176}C before bioassay. Percent neutralization was determined as follows:

Assay

The percentage of cells infected with reovirus in vitro and in vivo was determined as described by Duncan et al. (17). For macrophage infection in vitro, the amount of reovirus required for the appropriate multiplicity of infection (MOI) was added to thioglycollate-elicited macrophages in 0.5 ml of RPMI 11640 medium without serum and incubated at 37#{176}C for 1 h. After incubation, the cells were washed two times with 5 ml of complete medium. The cell suspensions were then serially diluted and 2 ml of complete medium containing 10k, 10, or 102 cells added, in duplicate, to confluent monolayers of L929 cells in 6 well plates (Falcon) and allowed to incubate for 5 h at 37#{176}C before removal of medium and addition of agar overlays as described for the reovirus titration. Plaques were counted and the percentage of infected cells was determined. For determination of percent infection

as

Pittsburgh, PA) for 30 mm. The dye was with 0.33 N HCI and the absorbance (570 nm) was measured with a Mini-Reader II EUSA reader (Dynatech Laboratories Inc., Alexandria, VA). The activity of the culture supernatants was then determined from the standard curve.

previously with thioglycollate were injected plaque forming units (pfu) of reovirus in 0.2 or DPBS alone 4 h before peritoneal cell col-

Center

Co.,

released

Injection

Mice injected i.p. with l0 ml of D-PBS lection.

in

tific

inactivation

February

The levels mTNF-a

of TNF-a results were

variance

using

(Graphpad

1993

the

Software,

detected analyzed InStat

San

in

the L929 by one-way

statistical

Diego,

CA).

analysis

bioassay analysis program

and of

RESULTS

TABLE

Infectious Reovirus

Center Assay In Vitro

of Macrophages

Treated

with

Treatment of macrophages in vitro with reovirus resulted in a productive infection; however, the efficiency of infection was MO! dependent. Data from three independent infectious center assays (Table 1) demonstrate that 96 ± 4% of the macrophages were infected by a MO! of 100. A lower efficiency of infection of 37 ± 6% was observed when the cells were treated with a MOl of 10.

2.

Stimulation

of TNF-cz

by Reovirus

Production TNF-ct

Treatmenta

(U/ml)’

LPS

1280±316

Reovirus UV-reovirus

1840 2386

Medium

3 ± 63

control

aThioglycollateelicited appropriate

± 289 ± 233

macrophages

stimulator

or

complete

were

(medium

assay of the cell-free supernatants bUnits (U) of activity were

for TNF-ct determined

standard

are means

curve.

Data

presented

cultured

medium

activity. from ± SD of

for

4

h

control)

with

before

rTNF-u (BRMP) two independent

experiments.

Infectious Reovirus

Center In Vivo

Assay

of Macrophages

Treated

with neutralizing

antibody,

suggesting

that

the

activity

was

due

to TNF-a.

Administration of 10 pfu of reovirus i.p. produced very low levels of infectivity in the peritoneal macrophages. Results combined from two independent experiments demonstrated that 0.4 ± 0.05% of the macrophages were infected in animals injected with reovirus 4 h before cell collection.

Stimulation

of TNF-cz Production

by Reovirus

In Vitro

To determine if macrophages were stimulated to produce TNF-a by reovirus treatment, TNF-ct activity in macrophage culture supernatants treated with reovirus or UVinactivated reovirus was compared to supernatants from cells treated with 1 p.g/ml LPS in a L929 cytotoxicity assay. UV-inactivated virus was used to assess whether infectious virus was necessary for induction of TNF-a. Results combined from two independent experiments (Table 2) indicate that infectious and noninfectious reovirus are both capable of inducing TNF-a at levels greater than those stimulated by LPS. None of the agents produced any detectable cytotoxic effects when added directly to the L929 bioassay. Noninfectious virus produced the highest amount of TNF-a activity followed by infectious virus and LPS. Both of the values for virustreated cells were significantly different from each other as well as from the LPS-stimulated activity.

In Vivo Induction of Membrane Macrophages by Reovirus

TN F-a on Peritoneal

To determine if reovirus stimulated the expression of TNF-a on the surface of thioglycollate-elicited peritoneal exudate cells in vivo, peritoneal macrophages were elicited with thioglycollate 4 days before i.p. injection of reovirus or D-PBS. Four hours after reovirus administration, peritoneal exudate cells were collected and stained with fluorescent-labeled anti-TNF-a and anti-Mac-2 antibodies. Results combined from two independent experiments demonstrated significant differences in cells treated with reovirus in which 27.5 ± 2.3% of the total Mac2* cell population expressed mTNF-a compared with 8.9 ± 1.3% expressed on the untreated cells.

100 90 80 70

Neutralization Antibody

of TNF-a

Activity

by Anti-TNF-a 0

60

0

Supernatants containing TNF-a activity from reovirus, UV-inactivated reovirus, or LPS-treated cells were incubated with anti-murine TNF-a antibody to determine if the cytotoxic activity exhibited in the bioassay was due to TNF-a. Results in Fig. 1 demonstrate that the cytotoxic activity in reovirus, UV-inactivated reovirus, and LPS stimulated supernatants can be completely abrogated by

N

50

0 5

40

z

30 20 10

TABLE

1.

Percent

of Macrophages

Infection

Reovirus

Reovirus

Treatment

after

Anti-TNF Percent

(MOI)a

100

96±4

bData

serotype presented

macrophages

3 at multiplicities are

means

±

were

of infection SD

of three

incubated

1. Neutralization TNF.a-containing lated with 1 tg/ml Fig.

for

1 h

of 10 or 100 phi/cell.

independent

(Dilution)

Infectionb

37±6

amioglycollate.elicited

iO_3

10_2

10’

10

reovirus

0-

with

In Vitro

experiments.

with

of TNF-a

supernatants LPS

(0),

activity from reovirus

by anti-murine TNF-a peritoneal macrophages (#{149}), or UV-inactivated

(V) were incubated with dilutions of anti-TNF-a 37#{176}Cbefore assaying for L929 cytotoxicity. The sentative of experiments that have been repeated

Farone

d

at.

TNF-a

induction

antibody. stimureovirus

antibody for one h at above data are repreat least three times.

by

reovirus

135

DISCUSSION Macrophages

are

ing

inflammatory

a potent

multifunctional

cells

capable

response

upon

of

initiat-

activation

by

various agents (6). One important mediator of this response produced by activated macrophages is TNF-cz (20, 21). TNF-a is a cytokine that appears to mediate a variety of immunomodulatory and physiological effects, as well as being directly cytotoxic to some tumor cells (912, 22). We have previously demonstrated that reovirus/BCNU treatment affects significant therapy of murine ascites tumors. Resistance of the surviving animals to challenge with the tumor from which they were

cured,

but

not

therapy regimen The purpose of

a different

tumor,

causes tumor-specific this study was to

suggests

that

this

immunity (14). to characterize

begin

during

tumor

Initial

using

of

the

infectious

of reovirus infection Results suggested

could

established

ent

a

with

MO!

could

and

essentially

of bind

the

100. and

These productively

virus/cell

center

efficiency in vitro. be

assays

interaction

was

to determine

the

136

approximately untreated

Journal

was

from

the

supported

Ohio

by

Board

a

Research

Challenge

of Regents.

REFERENCES

control

of Leukocyte

1. Aderka,

entire

cell

data

population

demonstrate infect

infected

that

macrophages,

a threefold animals.

Biology

increase in This result

Volume

53,

February

Tumor

A.,

Horch,

infection (TNF-a)

at

H., Toker,

necrosis

factor

L., Hahn,

induction

T., Wallach,

by Sendai

D.

virus.

j

136, 2938-2942. J.-H., Sprenger, H., Hinder,

2. Gong,

reovirus possibly

mTNF-cz suggests

D., Holtmann,

(1986) Immunol.

of peritoneal macrophages that a productive infection that infectivity was MOl depend-

resulting in the production of cytokines. The focus of this study was to determine if reovirus could induce TNF-a production by macrophages. Reovirus-treated macrophage supernatants were analyzed in a L929 cytotoxicity assay. Secretion of TNF-a by these cells was detectable without the use of LPS as a costimulator to trigger the release of the molecule, as reported for other TNF-a-inducing agents (2). Reovirus alone stimulated a higher level of TNF-a activity than 1 p.g/ml LPS. Macrophages treated with UV-inactivated reovirus produced levels of TNF-a that were not only higher than LPS, but also significantly greater than infectious virus, suggesting that infection was not necessary for stimulation of the TNF-a response. This is of particular interest because we have recently found that UV-inactivated reovirus is as effective in tumor therapy as infectious virus (unpublished results). To begin to characterize the in vivo response to reovirus, infectious center assays were conducted on macrophages collected from mice 4 h after i.p. administration of virus. Results from these studies suggest that reovirus does not efficiently infect peritoneal cells. This is consistent with earlier studies demonstrating low levels of replication during reovirus therapy (14). Therefore, whereas TNF-a has been shown to mediate the selective cytolysis of virus-infected cells (13), only low levels of infectivity were exhibited by tumor cells during BCNU/ reovirus therapy, suggesting that TNF-a may not function directly during tumor rejection but indirectly as an immune stimulator (14). The reasons for this low efficiency of infection are not clear. However, when the reovirustreated peritoneal cells were assayed for membranebound TNF-a expression, the Mac-2 cell population exhibited over the

work

Award

therapy.

investigation

conducted

ACKNOWLEDGMENT This

this immune response by analyzing the interaction between reovirus and peritoneal macrophages that could contribute to the stimulation of specific immune effector cells

that TNF-a expression was induced by reovirus in vivo, but that infection was not required for induction. This is supported by the results observed in vitro with noninfectious virus. Future experiments will attempt to determine the extent of membrane TNF-a expression in peritoneal cells from mice bearing ascites tumors treated with BCNU and reovirus. The presence of TNF-a may suggest a possible immunomodulatory role in the immune-mediated rejection of tumors stimulated by the BCNU/reovirus treatment regimen.

S.,

Nain,

H.,

of macrophages: gene expression

TNF-a

J.

release.

3. Molina,

J.-M.,

Groopman,

Immunol.

Scadden,

J.E.

F., Bender, A., Schmidt, (1991) Influenza A virUs enhanced tumor necrosis factor-a and lipopolysaccharide-triggered 147, 3507-35 13. D.T., Byrn, R., Dinarello, CA.,

Gemsa,

(1989)

and interleukin immunodeficiency 4. Roberts, N.J.,Jr., a

lb

Production by monocytic virus. j Gun.

D.

of tumor necrosis cells infected with Invest. 84, 733-737.

factor human

Prill, A.H., Mann, T.N. (1986) Interleukin 1 inhibitor production by human macrophages exposed to influenza virus or respiratory syncytial virus.j Exp. Med. 163, 511-519. 5. Goldfield, A.E., Maniatis, T. (1989) Coordinate viral induction of tumor necrosis factor a and interferon b in human B cells and monocytes. Proc. Nail. Acad. Sci. USA 86, 14901 or

interleukin

1494.

6. Adams, D.O., Hamilton, T.A. (1984) The cell biology of macrophage activation. Annu. Rev. Immunol. 2, 283-318. 7. Lorence, R.M., Rood, R.A., Kelley, K.W. (1988) Newcastle disease virus as an antineoplastic agent: induction of tumor necrosis factor a and augmentation of its cytotoxicity. JNCI 80,

8.

1305-1311.

Kriegler,

M.,

A novel

Perez,

form

C., Defray,

of TNF/cachectin

K, Albert,

is a cell

I., Lu, S.D. (1988) surface cytotoxic

transmembrane protein: ramifications for the complex physiology of TNF. Cell 53, 45-53. 9. Asher, A., Mule,J.J., Reichert, C.M., Shiloni, E., Rosenberg, S.A. (1987) Studies of the anti-tumor efficacy of systemically administered recombinant tumor necrosis factor against several murine tumors in vivo. j Immunol. 138, 963-974. 10. Asher, A.L., Mule,J.J., Rosenberg, S.A. (1987) Recombinant

human murine 1341

tumor sarcoma

necrosis factor mediates in vivo via Lyt2p cells.

regression Federation

of Proc.

a 46,

(abstr.).

11. Carswell, E.A., Old, L.J., Kassel, R.C., Green, S., Fiore, N., Williamson, B.D. (1975) An endotoxin-induced serum factor that causes necrosis of tumors. Proc. Nail. A cad. Sri. USA 72, 3666-3670. 12. Sugarman, B.J., Aggarwal, B.B., Haas, P.E., Figari, I.S., Palladino, M.A., Shepard, H.M. (1985) Recombinant human tumor necrosis factor-a: effects on proliferation of normal and transformed cells in vitro. Science 230, 943-945. 13. Wong, G.H.W., Kamb, A., Goeddel, D.V. (1992) Antiviral

1993

Properties

14.

of TNF.

and

TheirEmerging

381,

Raven,

Bryson,

Cox,

D.C.

(1988)

Characteristics

mediated chemoimmunotherapy mia. Cancer Immunol. Immunother. 15. Weiner, H.L., Drayna, D., Averill, Molecular

Molecules pp. 371-

19.

York.

New

J.S.,

In Tumor Necrosis Factors: The Role in Medicine (Beutler, B., ed)

basis

of

reovirus

pathogenesis:

role

production by mouse pentoneal cellular cyclic AMP. Immunology

reovirus-

of

of murine L1210 26, 132-138. D.R., Fields, B.N. of

leuke(1977) the si

20. Adams,

21.

18. Shaw,

22.

J.E.,

synthesis

Cox,

D.C.

by high

(1973)

multiplicities

Early

inhibition

of infectious

of

and

cellular

UV-in-

D.O.,

Marino,

P.A.

macrophages: 64,

the

role

of

719-724.

(1984)

Activation

nuclear phagocytes for destruction of tumor for study of macrophage development. Topics in Hematology/Oncology. pp. 69-136,

gene. Proc. Nail. Acad. Set. USA 74, 5744-5748. 16. Gallily, R., Feldman, M. (1967) The role of macrophages in the induction of antibody in x-irradiated animals. Immunology 12, 197-206. 17. Duncan, M.R., Stanish, S.M., Cox, D.C. (1978) Differential sensitivity of normal and transformed human cells to reovirus infection. j Virol. 28, 444-449. DNA

activated virus. J. Virol. 12, 704-710. Katakami, Y., Nakao, Y., Koizumi, N., Katakami, N., Ogawa, R., Fujita, T. (1988) Regulation of tumour necrosis factor

of

mono-

cells as a model In Contemporary

Plenum,

New

York. Decker, T., Cell-associated mechanism

Lohmann-Matthes, M.L., tumor necrosis factor for activated macrophages.

Gifford,

(TNF)

J.

G.

(1987)

as

a killing Immunol. 138,

957-962. Old,

L.J.

(1985)

Tumor

necrosis

factor

(TNF).

Science

230,

630-632.

Farone

et at.

TNF-a

induction

by reovirus

137