Mechanism of action of a serum oncolytic protein, rabbit tumor necrosis factor

MISSING IMAGE

Material Information

Title:
Mechanism of action of a serum oncolytic protein, rabbit tumor necrosis factor
Physical Description:
ix, 137 leaves : ill. ; 29 cm.
Language:
English
Creator:
Ruff, Michael Roland, 1953-
Publication Date:

Subjects

Subjects / Keywords:
Neoplasms   ( mesh )
Neoplasms, Experimental   ( mesh )
Rabbits   ( mesh )
Immunology and Medical Microbiology thesis Ph.D   ( mesh )
Dissertations, Academic -- Immunology and Medical Microbiology -- UF   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1980.
Bibliography:
Bibliography: leaves 127-136.
Statement of Responsibility:
by Michael Roland Ruff.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000895927
oclc - 22838780
notis - AEK4535
System ID:
AA00009123:00001


This item is only available as the following downloads:


Full Text





















MECHANISM OF ACTION OF A SERUM ONCOLYTIC PROTEIN,
RABBIT TUMOR NECROSIS FACTOR









By

Michael Roland Ruff


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA
1980














ACKNOWLEDGEMENTS

This dissertation is fondly and lovingly dedicated to Ruth and

Bob Ruff who, as my first teachers, conveyed to me those most mar-

velouus of gifts, a sense of wonderment and curiosity and, through

example, showed me their method toward personal greatness. I feel

a special bond of affection and gratitude for another of my tutors,

Roland Muenzen who, in addition to teaching me to sail, taught me

much about individual strength and integrity.

My scientific mentor, Dr. Gifford, has come to be a very close

friend over this period of four years or so and it is now that I feel,

perhaps, I've earned the privilege of calling him George. Another

student of his said George had taught him that there was a right and

a wrong way to do science. I couldn't imagine a better or more accurate

tribute.

Special xxxooo's to my friends and comrades who in numerous ways

were also collaborators during this great adventure. After all, you

guys are the only ones who know what it's really like.

Cindy you have particularly been a friend and counselor and

probably know more about TNF than anyone but me, well I guess that says

something; and thanks for all the times you helped by listening and

caring.














TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS . . .. ii

LIST OF TABLES. . . ... .. iv

LIST OF FIGURES . . ... v

ABSTRACT. . . ... ......... vii

INTRODUCTION. . . ... . 1

Historical . . . 2

Definition of TNF. . . .. 6

Cells Responsible for TNF Production . 8

MATERIALS AND METHODS ...................... 20

Methods. . . ... . 20

Purification and Physicalchemical Characterization of TNF 23

RESULTS . . . 29

Assays for TNF . . ... 29

Criteria for Purification and Further Estimation of
Molecular Weight. . .... 42

Mechanism of Action and Cellular Effects . .. 61

Effects on Macromolecular Synthesis. . 91

Cytoskeletal Aspects of TNF Action . .. 102

TNF A Metalloprotein . . .. 107

DISCUSSION. . . ... . 113

PROSPECTUS. . . .. .. 124

REFERENCES. . . ... . 127

BIOGRAPHICAL SKETCH . . .. 137














LIST OF TABLES


Table Page

I. Macrophage Toxins ... . 10

II. Purification Summary . .... .45

III. Stability of TNF to Digestion by Various Enzymes 60

IV. Physio-Chemical Characteristics of TNF ... .62

V. The Effect of Plating Density on TNF Cell Killing
of Normal and Transformed Cells. . ... 96

VI. Sugars and Lectins Tested for Their Effect on
TNF Activity . ... ... 101

VII. Protease Inhibitors Tested for Their Effect on
TNF Cell Killing . . .. 108

VIII. Effect of Heavy Metal Ions on TNF Cell Killing .... 112

IX. Synthetic Protein Analogs Tested for Hydrolysis by TNF 121














LIST OF FIGURES


Page


Dose response curves showing the survival ratio of L-929
cells as a function of the dilution of TNF serum. .


2. The effect of actinomycin D on the dose
cell killing . .

3. Dose response curves for TNF killing by
methods . .

4. DEAE Sephadex chromatography .

5. Gel filtration on Sephacryl-200 .

6. SDS-polyacrylamide gel electrophoresis.

7. Non-denaturing gel electrophoresis. .

8. Glycerol gradient centrifugation .

9. Isoelectric focusing . .

10. Two-dimensional electrophoresis .


kinetics of TNF
. 35

several assay
S . 37

. 40

. 44

S . 47

S . 50

. 52

. 55

S . 57


11. pH stability of TNF . . .

12. The effect of TNF dose on the killing of various clones
of L-929 cells . . .

13. Growth curves for L-929 cells cultured in the presence
of various doses of TNF or normal rabbit serum. .

14. Time course of L-929 cell killing the same experiment
as Figure 13, determined by the cytotoxic release
of isotope from cells which had been pre-labeled
with H-thymidine . . .

15. Time course of TNF killing in the presence or absence
of actinomycin D or cycloheximide . .

16. Effect of TNF on the functioning of mRNA .

17. Effect of TNF on the release of acid insoluble counts
from cells pre-labeled with H-isoleucine .

18. Effect of TNF dose on the length of the lag period
which precedes the onset of cytolysis .


Figure

1.















19. Logarithmic transformation of the data presented in
Figure 18. . . ... .. .85

20. Effect of TNF concentration on the killing kinetics
of L-929 cells . . 87

21. L-929 targets treated with high concentrations of
TNF for various periods of time. . ... 90

22. Time kinetics of L-929 cell killing at different
temperatures was determined in the presence
of actinomycin D . . 93

23. Serum dependency of TNF cell killing was determined
by culturing cells in the presence of different
amounts of fetal calf serum. . ... 98

24. Inhibitions of microtubules vinblastine and
colchicine, and an inhibitor of microfilament
assembly, cytochalasin B, were evaluated for
their effect on TNF cell killing ... 105

25. Inhibition of TNF killing by the metal chelator
o-phenanthroline . . .. 110


Figure


Page














Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


MECHANISM OF ACTION OF A SERUM ONCOLYTIC PROTEIN,
RABBIT TUMOR NECROSIS FACTOR

By

Michael Roland Ruff

March 1980

Chairman: Dr. George E. Gifford
Major Department: Medical Sciences (Immunology and Medical
Microbiology)

A soluble oncolytic protein, tumor necrosis factor (TNF), can be

induced to appear in the serum of experimental animals which have been

infected with viable Mycobacterium bovis, strain BCG, and then challenged

by an intraveneous administration of bacterial lipopolysaccharide (LPS)

in two weeks. TNF is capable of selectively killing several transformed

cells in vitro but no toxic effects on normal cells are observed. TNF

also shows necrotizing activity against a variety of transplantable

tumors in vivo.

Rabbit TNF has been purified by a series of salt precipitation, gel

filtration, ion exchange chromatography and lectin affinity chromatography

to a single homogeneous species on SDS-polyacrylamide gel electrophoresis

(SDS-PAGE). TNF activity could be recovered from non-denaturing gel systems

and has been shown to be an a-globulin with an isoelectric point of 5.1.

The molecular weight was estimated to be 68,000 by SDS-PAGE, 55,000 by gel

filtration, and 52,000 by glycerol gradient centrifugation. TNF activity

was table over the pH range of 6-10 and was relatively heat stable, not

being inactivated at 700 C for 1 hour. TNF activity was pronase sensitive

vii











but relatively trypsin resistant. Neuraminidase and phospholipase C

treatment did not destroy TNF activity. Partially purified TNF was

still capable of eliciting hemorrhagic necrosis in susceptible tumors.

Crude TNF serum had an interferon titer of 3000 units while the partially

purified sample had a titer of less than 30 units.

Rabbit TNF was also examined for its effects on transformed and

normal cells in culture. Several assays for TNF activity were developed

and their sensitivities and precisions compared. TNF killing of L-929 cells

was delayed by 10 12 hours and thereafter showed concentration and time

dependent increases in killing. Cells treated with actinomycin D or cyclo-

heximide showed enhanced rates of killing as well as shorter lag periods.

Several cell types, both normal and transformed;were tested for their

sensitivity to TNF. Normal cells were not killed by TNF and this discrim-

ination was shown not to be due to differences in growth rate between normal

and transformed cells. A cell cycle dependent mechanism of cell killing

was also excluded.

TNF killing of L-929 cells shows less than single-hit kinetics and

this is argued to suggest an enzymatic mode of action. Inhibitors of various

protease classes were screened for their effect on TNF killing. None of

the serine esterase inhibitors blocked TNF action; however the metal chelator

o-phenanthroline did abolish TNF activity. Killing activity was restored

with copper, molybdenum, and cobalt salts.

Inhibitors of receptor-mediated endocytosis such as colchicine, vinblas-

tine, or cytochalasin B were capable of protecting target cells to killing

by TNF, suggesting that the protein may have an intracellular site of action.

Chloroquine, a lysomotropic agent, also protects cells against killing by TNF.

viii










These data are argued to suggest that TNF may be a metal

containing enzyme which has to gain entrance to the intracellular

millieu to express its killing activity. Lysozomal enzyme functioning

may be important for the ultimate expression of killing, perhaps through

a protein processing step.











INTRODUCTION


Many elements of host defense contribute to tumor cell killing and

the scavenging of emerging aberrant clones of cells is becoming recog-

nized as one of the most important functions of the immune system

(Thomas, 1959, Burnet, 1967, 1970). How malignant cell killing pro-

ceeds in vivo is still largely unknown but based on several model in

vitro systems appears to be multifactorial. Specifically immune T-cells

(Hellstrom et al., 1971; Brunner and Cerottini, 1971), natural killer

(NK) lymphocytes (Herberman et al., 1975) with or without B or T cell

markers, antibody dependent cytotoxic T-cells (Pollack et al., 1972),

macrophages with specific antibody (Bennett et al., 1963) activated

macrophages (Cleveland et al., 1974) and polymorphonuclear leukocytes

(Gale and Zighelboim, 1975) are all possible effector cells. Antibody

with complement (Hellstrom et al., 1968), lymphotoxin (Granger and

Kolb, 1968) produced by lymphocytes, tumor necrosis factor (Carswell et

al., 1975) produced by macrophages, as well as other soluble mediators

are also candidates for tumor cell killing because of their demonstrat-

ed oncolytic capabilities. Interferons, produced by T and B lympho-

cytes as well as macrophages during immune reactions, are known to have

cytostatic effects on cells in culture, suppress immune reactions,

activate macrophages, and have some anti-tumor effects in vivo (Gresser

and Bourali, 1970; Gifford and Tobey, 1977; Huang et al., 1971). The

study of the immune system in tumor regression is made more complex by

observations that tumor cells may produce factors which subvert normal

immune mechanisms (Mansfield et al., 1977; Huget et al 1977; Friedman

et al., 1976; Snyderman and Pike, 1976).

1









The purpose of this report is to describe one of the soluble

mediators, tumor necrosis factor (TNF), and how it might contribute to

tumor cell killing. Although Lewis' initial thoughts on a surveillance

mechanism as formalized by Burnet are essentially cellular in nature

the selective toxicity of TNF for tumor cells compared to normal cells

suggests that this substance could also play an integral role in host

defense against neoplasia.

Historical

Events Leading to the Discovery of TNF

The anti-tumor effects of bacterial cells and their products have

been known for over one hundred years and have been investigated by

Bruns (1868) and by Coley (1891) who published pioneering studies

describing the spontaneous regressions of certain human tumors during

bacterial infections. In particular, erysipelas (reviewed by Nauts et

al., 1953) and tuberculosis (reviewed by Pearl, 1929) occasionally

caused a modifying, curative, or even a preventive effect on tumors.

Coley, and others, subsequently developed bacteria-free filtrates from

cultures of streptococci (isolated from erysipelas) and other organ-

isms, particularly Serratia marcescens, to be used in the treatment of

human malignancy. These "Coley's Toxins" were employed for over forty

years with favorable results in a significant number of cases (reviewed

by Nauts et al., 1953).

The study of bacterial toxins in experimental cancer seems to

have received little attention until Gratia and Linz (1931) demon-

strated hemorrhagic necrosis of a transplanted liposarcoma in guinea

pigs using Escherichia coli culture filtrates. Soon afterwards

Schwartzman and Michailovsky (1936) obtained hemorrhagic necrosis upon










parenteral injection of filtrates from meningococcal cultures into mice

bearing sarcoma 180 and some of the tumors receded completely.

The parenteral administration of bacterial products to experi-

mental animals, or human patients, may produce a severe hemorrhagic

reaction within the tumor. The anti-tumor effect is rapid (about 4

hours) and confined to the core of the tumor which darkens and is

eventually sloughed off. When administered in sufficient amounts,

these toxins cause disseminated systemic effects leading to circulatory

collapse and death. The rapid hemorrhagic reaction is followed by a

more slowly progressing necrotizing reaction at the tumor site which

increases in intensity over the next forty-eight hours.

Since these filtrates were quite toxic to the recipients, Shear

and Andervont (1936) attempted to separate the hemorrhage producing

substance from the toxic component. Shear and Turner (1944) eventually

isolated a polysaccharidee," which contained some lipid and is now

known to be endotoxin, from Serratia marcescens culture filtrates which

could elicit hemorrhagic necrosis of both experimental and primary

subcutaneous tumors (Shear 1944). Endotoxin today is known to have a

large number of biological effects in the recipient including stimu-

lation of the reticuloendothelial system, activation of macrophages,

mitogenicity for B lymphocytes, enhancement of antibody synthesis,

induction of interferon synthesis, as well as fever, leukopenia fol-

lowed by leukocytosis and diverse vascular disturbances (Elin and

Wolff, 1976; Kass and Wolff, 1973; Nowotny, 1969). An important

feature of endotoxin induced tumor regression is that although the

necrotic reaction is a common consequence of toxin administration,

complete tumor regression is rare. Typically a ring of viable tumor

tissue survives to eventually grow and kill the host.









Subsequent injections of endotoxin are usually less effective since the

animal becomes refractory to the tumor inhibiting action as well as the

other biological effects.

A favored mechanism for endotoxin induced tumor destruction, based

on the work of Algire et al. (1952), was one of endotoxin induced

systemic hypotension leading to collapse of the tumor vasculature

followed by anorexia and tumor cell death. Tumor cells are not killed

in tissue culture when grown in the presence of endotoxin, further

suggesting that they are affected indirectly as a result of endotoxic

reactions in the host. Subsequent work by several research groups

suggests that additional elements, both humoral and cellular, are

involved in the hemorrhagic reaction as well as the subsequent tumor

regression.

Recently Berendt et al. (1978a) have proposed a model for endo-

toxin induced tumor regression whose central precept is the requirement

for sensitized T cells. Endotoxin induced regression of established,

syngeneic murine tumors depends on the generation of a state of con-

comitant tumor immunity (Berendt et al., 1978b). It was suggested that

within this context endotoxin might more accurately be viewed as an

immunotherapeutic agent. The enhanced phagocytic capacity of the

reticuloendothelial system during the growth of various transplanted

tumors has been establiehd (Biozzi et al., 1958, Old et al., 1959). In

addition, certain agents that are active in the reticuloendothelial

system have been shown to alter the development of experimental tumors

(Ribi et al., 1975; Halpern et al., 1974; Hibbs et al., 1972; Cleveland

et al., 1974; Alexander and Evans, 1971; Hibbs, 1974).

Thus, when tumor bearing animals are treated with such other well

known immunostimulatory agents such as Mycobacterium bovis, strain BCG









(recently reviewed by Baldwin and Pimm, 1978) or Corynebacterium

parvum, significant regression rates may be obtained (Ribi et al.,

1975; Halpern et al., 1974). One of the first of these studies by Old

et al (1960) used BCG infected mice, subsequently challenged with

Sarcoma 180, carcinoma 755, or Ehrlich ascites cells. Mice inocu-

lated with S-180 usually die (80-95%) in 2 to 5 weeks, the remaining

regress completely. If mice were prophylactically treated with BCG

significant protection was observed; up to 100% survival for those

animals who were infected with BCG for 25 or more days prior to tumor

challenge. The two other tumors did not show such dramatic results but

the tumors did grow slower, and the mice survived for longer periods

when preinfected with BCG.

Ribi and co-workers (1975) showed a synergistic effect of BCG and

endotoxin in guinea pigs bearing hepatocarcinomas. The combination of

live or cell wall extracts of BCG in conjunction with endotoxin was a

particularly efficacious combination, with cure rates of up to 90%.

This compared with regression rates of 66% for the BCG alone group.

The Discovery of TNF

O'1lalley et al. (1962) reported the production of a tumor

necrotizing factor in the serum of normal mice which had been chal-

lenged with endotoxin from Serratia marcescens. Activity was assayed

by interperitoneal injection of the serum into other mice bearing

sarcoma 37. The appearance of this activity in serum was rapid, short-

lived, and refractory to further stimulation by repeated doses of endo-

toxin. Carswell et al. (1975) have also demonstrated a necrotizing

factor in the serum of mice when challenged with endotoxin. However,

the mice had to be primed first with BCG or other activators of the re-

ticuloendothelial system for 1 to 3 weeks prior to endotoxin challenge








in order for the factor to be elicited. This factor, which they called

tumor necrosis factor caused hemorrhagic necrosis of Meth A sarcoma

when passively transferred to tumor bearing mice. TNF was shown to be

active in cultures of transformed cells. The difference in these two

reports, i.e. the need for a priming agent, is not clear at this time

but may be due to differences in strains of mice or endotoxin prepara-

tions. Butler et al. (1978) have shown that tumor necrosis can be

passively transferred by serum from normal animals injected with endo-

toxin. They obtained a higher titer of this serum factor when animals

infected with BCG were employed. The increase in their serum necrosis

factor in BCG-infected animals may be related to the findings of Suter

and Kirsanow (1961) that such animals become exquistely sensitive to

the effects of endotoxin. Their serum factor was not capable of

killing the tumor cells in vitro. However, they performed their in

vitro assays for only 24 hours and, as we will show later, the killing

effects of TNF on cells in culture are delayed and maximal killing

occurs between 24 and 48 hours. An important contribution of this

study was the finding that a non-toxic polysaccharide-rich product ob-

tained by acid hydrolysis of endotoxin was capable of eliciting the

serum factor which causes tumor regression.

Definition of TNF

TNF is defined as a substance found in serum of animals sensi-

tized to BCG (or certain other immunopotentiators) and challenged later

with endotoxin and which causes the necrosis of some tumors when

passively transferred to tumor bearing animals. TNF has other unique

characteristics such as lack of species specificity and the ability to

discriminate between normal and certain tumor cells in vitro. The in

vitro cell killing effects of TNF are assumed to be due to the same









factor as that which causes hemorrhagic necrosis in vivo. Whether this

assumption is justified will have to await final purification. It is

reasonable that TNF activity measured in vitro is one component of a

complex host-tumor interaction leading to hemorrhagic necrosis.

Priming and Elicitation of TNF in vivo

TNF was first reported to be found in the sera of mice sensitized

to BCG and challenged two to three weeks later with endotoxin (Carswell

et al., 1975). Rats and rabbits also produce TNF under similar condi-

tions. The activity was not found in the sera of mice or rabbits given

either BCG alone or endotoxin alone. The optimal time for collection

of serum for maximal TNF is two hours following endotoxin injection; it

is not found six hours later. BCG is not the only successful priming

agent; heat killed Corynebacterium parvum or Corynebacterium granulosum

or yeast cell walls (zymozan), which also produce hyperplasia in the

reticuloendothelial system, are as effective as BCG (Green et al.,

1977).

Agents other than endotoxin have been shown to elicit TNF in BCG

primed mice (Carswell et al., 1975). These include a mixed bacterial

vaccine consisting of heat-killed Streptococcus pyogenes and Serratia

marcescens. It is probable that the endotoxin from Serratia was re-

sponsible for the elicidation of TNF. However, Brucella abortus, which

also contains endotoxin, did not elicit TNF in BCG primed mice. This

may be due to the observation that Brucella abortus endotoxin is not

very toxic. Old tuberculin was also ineffective in inducing TNF.

The strain of mouse is important. Old (1976) has reported that BP

8 tumors grew equally well in two strains of C3H mice known to differ

in their response to endotoxin as measured by B cell mitogenesis and








lethality. Only the endotoxin sensitive strain showed tumor necrosis

when injected with endotoxin. However, both strains of mice showed

tumor necrosis when TNF was injected. TNF can be produced in BCG in-

fected mice by endotoxin in endotoxin sensitive mice (C3H/HeN) but not

in endotoxin insensitive mice (C3H/HeJ) (Mannel et al., 1979b). Fur-

thermore, these authors have shown that the ability to release TNF

could be transferred into lethally x-irradiated endotoxin nonresponders

by reconstitution with endotoxin responder bone marrow cells.

Some studies have shown that a cytotoxic factor can be produced by

cultured macrophages under certain conditions that do not require prior

exposure to BCG. However, these factors have not been tested for their

ability to cause tumor necrosis in vitro and may be different from TNF.

These will be discussed in the next section.

Cells Responsible for TNF Production

Implication of the Macrophage

In their original paper (Carswell et al., 1975), the authors

speculate that the cellular origin of TNF is probably the macrophage.

They point out that massive hyperplasia of macrophages occur in the

spleen of BCG-infected mice and that pyknosis and disruption of the

population occurs two hours after the administration of endotoxin.

This is further supported by numerous observations that mouse peri-

toneal macrophages can be activated by a variety of diverse stimuli to

become broadly cytotoxic for syngeneic, allogeneic, or xenogeneic tumor

cells but not for normal cells (Hibbs, 1974; Piessens et al., 1975).

These mechanisms include chronic infection with BCG or exposure to en-

dotoxin (Cleveland et al., 1974), double stranded RNA (Alexander and

Evans, 1971) as well as co-cultivation with sensitized lymphocytes and








antigen (Evans and Alexander, 1971) or by the addition of lymphokines

to peritoneal cells (Piessins et al., 1975).

Old (1976) points out that TNF could not be induced in athymic

nude mice (nu/nu) when primed with Corynebacterium parvum and subse-

quently injected with endotoxin suggesting participation of T lympho-

ocytes. However, more recently, Mannel et al. (1979b) have found that

serum from nude mice infected with BCG and treated with LPS contained

as much cytotoxicity as did their heterozygous litternates. Perhaps,

there is a difference in the ability of the two priming agents (BCG and

C. parvum) to prime for TNF production in athymic mice.

Many cytotoxic factors in the supernatant fluids of macrophage

cultures from human, guinea pigs, rats and mice have been described

over the past 12 years (Table I). Some of these factors are spontan-

eously produced in culture but others are induced by such substances as

endotoxin and purified protein derivative (PPD) from old tuberculin

either in normal or sensitized animals. However, in some of the

inducible systems, a background level of activity is found in super-

natant fluids without inducer. Weinberg et al. (1978) have reported

that many reagents are contaminated with endotoxin including calf

serum. Thus, it is possible that the "nonstimulated" cultures were

actually stimulated with this contaminant.

When tested on normal cells, the macrophage supernatants are some-

times cytotoxic indicating that they are dissimilar to tumor necrosis

factor and similar to lymphotoxins in this regard. For example, Kramer

and Granger (1972) employing macrophages from sarcoma sensitized mice

found a toxic factor released into the medium when exposed to alleno-

geneic target L cells. Medium from cultures of normal cells plus

target cell or normal cells alone had a low level of activity. The





















.I
4-H
1-'
CO


00
00 0


oU

CO


:3 C 0








(U 4.1


4-
1-1
CO







0
co

(U












CU
4-1
r.


P4


0
4J
-I 0

40 0
Q) --
4-i 41- 0
4-4 4J

41 o

0 HI p



C 00
Cd 00>-






0 0
H0 0 04 a

u 4 x -xC co c -H


0
v



i-l



oH
,--4


O





I-i H
O rX 0
0 0l
4- -w

H- .-l0 Zu >a


o
0






H w 0 4-
4 H
0 ( 0






4 p4( 0 -
CO 1 U H ,
a o r X

n U 1 E-4


H
0
0






O o
(U 0
4-1 Z
*O


U H

0Q 0


0 p
0 N
. 0
C/ H-


4-
-3 CO
bo00 U

0 4J
3 co

'.4 1 -4
4Ji CU 0
-4 H 4


-4 U C

U H C


0

1-4
M 0
0-l 0
pq ca

0


0
9--







P4
0








74







r-i
Eo








M ^


04






-H
.5

,-7-4
41


CO





(d
-l
iCl

U

0}

4-i








'.4-4 -


41




co
0


(U
OU


t0
O4- 0

OH 0
o


0 o

0

rl p4
-M P4
a)O
J3 **I Q


0
4J


0
CO
CO



Ho



0






0p
ou

rC
0 4J



0 -H
A4 :





















.1-


$4










C )
:ij
u'
N




Oi







H-1
CO'
t(U

(U
0 '-


4-I
0
Q)
co
0

<4
a)





0,

CH
ca M
ON





0 0


p H (
M z z


(U


co
u H
0 H

a 0

CO U

0
N 4C
*H H
H 0
0,
0 41
co a CD
0 H

U) 0 0)
0 0-
S ,.c U


M-
0
1-1-
41)
ca
0
0
0 u
0 U




(U (U
ca a


4 1.
'o o


)0 c 0
H-1 I, z
0 -T C 0 0
C) H- rI Z >


0



41
0
4.

.0
H






oD r p
H o-
i-i m c


o
0
I
4-1


) 0



0 C 0 0 **
E*r U

U r-1 T) 0 (U
04 C0 H CO
W U H E-4


"O 0


,.-1 C (
H 0 p0
0 (U oM



0 0 0




4 4- c oo
H H 00

l 0 0 0


- 4a) 0
(0 H H 0
0 0 0 00
0 co
z m H


4.)

*H


0
(U


C
i-
60
c ^


0

H
C

0
u
I

1--]
H

N


IC-




r.


a


Cd

0
0


0
*







0
CO-

r-(

U 5


0



0 H

U H
0 0
4 0
S/)
W E-


H -H


r- )
0 q 4J
:J

(H 0
) (U
0 o















I 03


0 0 -
0 C 1-1 0)
U


I H ; ) Q)




W 0 wCd
c0 0

0 a A n
o o 4 -


r1 Q) -i
0 N



0 -40




0 O 0
S01 0 0 --








0* 0a 0
O 40 ,
0 OO 0













o H 0 4 01
*a0
04 0



co a 0 a -H

r. 0 H w c
O r. (U 0C


0I z *p
4 CO CO 0
0 0 9- H1
M H 0 000




40 H0
a) rU







0O 0 CO COH
o .o 0 04



O 0 3 0 0 M 4.4
-H 0 pC r 1 -H O
41 } En 4J 0 -H r-l :j U




C C
g G O !3 3
f U t *













0 0H 0 4 0 C
S ( u H 0 4 4-ic p
U l C 4 0
H o 0 H. i C 4 H H n
4-' 0 CB 0 0 m- 0 H H 0




CO HH3
P 0 H C O Q 0






H, 1- 0 HC CO







13








00



0 0
-H
,--I



0 o
r-4







0 0 cO
So 0 oo















r. o ) < r-








0 H
wU 0 0 wU i











eI 0 C S H
0 0 "a H 0 -oo
























C d C 0 w I 0
co









S 0 .4 0 0 < 0








02
0 l









J 0 w )
a n a


o o 0o
4u 00 u
r9 u



























3 TA *H
00 u p



























0 0 (U 0
HCd H

















H u H 0q u q o
c u r-4 co u p


> a 0 Ca 0 H 4 c
0 0) C ( 0 I 441

4i 5-






u C




















4
0
4-1

00





co a

0 0 0
0d a


ra Q a) 0
HH 6f H
S>-' o g 4- ) 1


4r-i ) 0


H 4 0
S0 00
4-1 r 1





0H H OH w 0
2 i 0p 0 0 w 0













- H u o o 0)










0a0 4 -) r.
H o C O 4 I













N0 0 C C u





Q* 10 r **
S 0 0 0)
pq wa .Z









0 0 **









40 4 4 z b0 u
S0 0









0 H p- W4 4H









0 0 a4 0 : Un
Ot < )H




















co P (n > 4 p
S 0 a

4U u cH w 0 S

> a) m 1co M ca
3 IU 0 441 .1 (>
0 0 0 0 0 I U1


H U) tL) E- W p M .

o d U *4 U, U Cu

















































ca




4-I


01


r-1
( 0 )
C co


0
0


CO
r-I















0 o 40
0 I
U C
U (U
*s
a0 G '0




0- 0 p


Y 0
o

u I
0 L ii-
Z in in


"U
o




O 0


co o
0 U H 0


0 i-4 0
H 0 H H


M 0
0 U4
3 4

CU 4
co
o 0)
3 U

H CU
0 ,U
1


a,
0
H
4I
0
u
I
H
w1


co




I)
-H

0









4.
rl








E-i
co








10












,-H

*u







0o
O N
00















z
0










4.J
Cf
r-



o 4J
0 0
HP

C C









factor was also toxic to normal mouse embryo fibroblasts. Two frac-

tions of toxic activity were eluted from Sephadex which had approximate

molecular weights of 150,000 and 47,000. Antiserum prepared against

partially purified PHA-induced mouse lymphotoxin was capable of neutra-

lizing both the lymphotoxin and the macrophage cytotoxin indicating

that the factors may be similar. Heat stability (1000C, 15 minutes)

was another feature in common with this lymphokine. Table I lists the

inducers and gives some properties of many of these factors. In those

studies where molecular weights have been estimated by gel filtration

on Sephadex the molecular weights have usually been in the range of

40,000 to 50,000. Some have an additional cytotoxic component whose

elution characterizes a molecule of approximately 150,000 daltons.

Thus there are similarities with the factor described by Kramer and

Granger (1972) relative to molecular weight. One factor of consider-

ably lower molecular weight has been reported. Pincus (1967) has

described a factor produced by macrophage cultures from normal guinea

pigs when challenged with PPD, bovine serum albumin or other anti-

gens, or the mitogen PHA. Occasionally low levels of activity could be

found in non-stimulated cultures. In a later paper (Sintex and Pincus,

1970), the authors used thioglycollate stimulated peritoneal cells from

tuberculin-positive guinea pigs and induced the factor with PPD. Cyto-

toxic activity from these supernatants chromato- graphed on Sephadex

with an estimated molecular weight of 30,000- 40,000. It was found to

be heat stable and could be boiled for 1 hour without losing activity.

Upon further purification, the activity was dissociated from protein

and the activity now chromatographed at 1000d. It was inactivated by

acid phosphatase and phospholipase D and was resistant to pronase and

is probably a phospholipid.









Cytotoxic activity has been found in the culture fluid from 4 of 5

macrophage cell lines (Aksamit and Kim, 1979). The factor was active

on a fibrosarcoma cell line; activity on normal cells was not reported.

Using Sephadex G-200 filtration, the activity was in the 40,000 to

50,000 d range with a second peak in the void volume.

Evidence that products secreted by stimulated lymphocytes activate

macrophages to become tumoricidal has been published by several

laboratories (e.g., see Piessens et al., 1975). Leonard et al., (1978)

have purified a lymphokine called macrophage activation factor (MAF)

from spleen cells from BCG-infected mice which were incubated with PPD.

This factor makes macrophages nonspecifically cytotoxic for tumor

cells.

Since the discovery and characterization of TNF there has been

increasing concern whether the macrophage cytotoxins are related to

TNF. Matthews (1978) has shown that apparently non-activated adherent

mononuclear cells from normal rabbit blood were cytotoxic to those cell

lines in vitro which were also susceptible to TNF and were not cyto-

toxic for TNF resistant cells. The bulk of the activity was expressed

by monocyte-enriched fractions. Detectable amounts of TNF-like

activity were found after 3 hours in culture and maximal amounts by 7

hours. Other similarities of the monocyte cytotoxic factor and TNF

were studied: both were precipitated with 50% ammonium sulfate,

eluted on gel filtration in the range 30,000-50,000 d, and migrated in

the same position by polyacrylamide gel electrophoresis.

Mannel et al. (1979 a,b) have studied four cytotoxins from macro-

phages. These were cytotoxic factors found in the supernatants from

BCG activated macrophages incubated with endotoxin, a macrophage cell

line incubated with endotoxin, a cytotoxin from thioglycollate induced








peritoneal cells propagated in the presence of supernatant fluids from

mouse L-929 cells and then stimulation with endotoxin, and cultivation

of peritoneal cells in lymphokine medium (from Con A stimulated spleen

cells) followed again by induction with endotoxin. The molecular

weight of all cytotoxins as detected by gel filtration was approxi-

mately 55,000 to 60,000 and the activity of all four eluted from

DEAE-Sephacel under similar conditions. Thermostability was similar.

Antibody prepared in rabbits against partially purified TNF was able to

neutralize the cytotoxicity of the 4 cytotoxins as well as TNF. These

data suggest but do not prove that cytotoxins produced by macrophages

under different conditions may be similar and related to TNF.

Are macrophage cytotoxins equated to TNF?

Considerable reservations have to be made in assuming that factors

from cell cultures are equated to TNF. For example, Currie and Basham

(1975) have shown that a soluble supernatant factor can be found in the

supernatant of rat and mouse macrophages which lyses malignant but not

normal cells. Currie (1978) subsequently has shown that this factor

has arginase activity and that the cytotoxic effects could be inhibited

by adding excess arginine. Currie and Basham (1978) have recently

shown that malignant cells require a higher concentration of L-arginine

in the medium than the normal counterparts and that complete arginine

deprivation caused a more rapid cytolysis of the malignant cells. They

also suggested that preliminary studies indicated that macrophages

freshly isolated from tumors contain high levels of arginase. We know

that TNF killing is not influenced by adding excess arginine or any

other amino acid videe infra). The possibility of nutrient depletion

has not been shown for any other of the other factors except for the






19

lymphotoxin-like factor from macrophages reported by Kramer and Granger

(1972).












MATERIALS and METHODS

Methods

Preparation of TNF Containing Serum.

New Zealand white female rabbits (NZW) weighing from 2-3 kg. were

bled by cardiac puncture as a source of normal rabbit serum (NRS) for

control experiments. Rabbit tumor necrosis factor (TNF) was produced

by the procedure by Carswell et al. (1975). Viable Mycobacterium

bovis, BCG (Tice stain,3x 108 organisms) were injected into the

marginal ear vein. Fourteen days later, 100 pg of endotoxin from

Salmonella typhimurium, virulent strain 7, was injected into the ear

vein. The endotoxin was a gift from Dr. Joseph Shands, University of

Florida. Rabbits were sacrificed by cardiac puncture 1.5 hours later.

Blood samples were also taken prior to injection of endotoxin, and

endotoxin was also administered to non-BCG infected rabbits. The

endotoxin alone did not demonstrate any TNF activity.

Cells.

L-929 is a mouse transformed cell line originally derived from the

C3H strain of mouse. Cells were grown in Eagle's minimal essential

medium (MEH) supplemented with 10% bovine serum and 250 units of peni-

cillin/ml and 125 yg of streptomycin/ml. Primary fibroblast cultures

from mouse, chicken, and hamster embryos and rabbit kidney cultures

were prepared by minicing near term embryos or mature kidneys, trypsin-

sinization (0.1%) with 0.04% ethylenediamine tetraacetic acid (EDTA)

for 60 minutes, followed by washing with Gey's balanced salt solution

(BSS) and dispensing the cells in glass bottles containing fortified

MEM with 10% fetal bovine serum (FBS). Fortified MEM contained 2x con-

centrations of essential amino acids, non-essential amino acids,

20








vitamins, and glutamine. Sodium pyruvate was also added (Dion et al.,

1977). Flow 7000 are human embryonic foreskin fibroblasts obtained

from Flow laboratories. B16C3 is a mouse melanoma HeLa is a human

cervical carcinoma. AV3 is a human amnion cell.

Assays for TNF.

TNF was assayed by several methods in order to better evaluate the

sensitivity and precisions of the various techniques. These assays are

similar to others which have been developed to study the actions of

other cytotoxins, most notably lymphotoxin (Sawada and Osawa, 1977;

Henney, 1973; and Kramer and Granger, 1972). In the morphological

microassay, cultures of cells are established in 96 well flat-bottomed

trays (3040, Falcon Plastics, Oxnard, CA.) at 50,000 cells/0.2 ml in a

humidified atmosphere at 370 with 5% C02, unless noted otherwise.

Cells are typically added at 2 x concentration in a volume of 0.1 ml to

dilutions of TNF are control serum in media in a volume of 0.1 ml.

Forty-eight hours later, media is decanted, and residual cells are

stained with crystal violet and cells in representative microorganic

fields are counted. A modification of this basic assay was to pre-

establish cells in 0.1 ml for 15 hours, then TNF dilutions and actino-

mycin D at 1 pg/ml final concentration was added to a final volume of

0.2 ml. Plates were stained after 18 hours. TNF killing was also

determined by establishing cells in 25 cm2 flasks for 18 hours

followed by addition of TNF or control serum. Specific plating

conditions are indicated in the figure legends. Morphologically intact

cells were determined at various times by cell counting

For the cytotoxic release assay L-929 cells were grown for three

days in the presence of 1.0 p Ci/ml [3H] thymidine (6Ci/mmole,

Schwarz/Mann). Cells were washed three times with warm MEM prior to








replating in 25 cm2 flasks, as above. Cytolysis was measured at

various times by sampling aliquots of centrifuged supernatant fluids

for scintillation counting. The specific lysis was computed by the

formula:

Specific lysis = Experimental Background (CPM)
Control Background (CPM)


The 100% releasable CPM was determined by incubating target cells in

0.5% SDS for 30 minutes. Spontaneous release was 8-14% of maximal

release. The mean and standard deviation of triplicate samples were

determined.

In the "end label" method cells were established in microtiter

trays as for the morphological assay. At 44 hrs after TNF addition

the cultures were pulsed with 0.5 pCi [3H] thymidine (56 Ci/mmole,

Schwarz/Mann) for 4 hours. Cells were collected onto glass fiber

filter strips with an automated sample harvester and radioactivity

counted. The means and standard deviations of triplicate samples were

determined.

Cell Counting.

Cell counting was carried out under an inverted microscope at

200x. Numbers of morphologically intact cells were determined by use

of a micrometer mounted in the ocular (5 x 5 mm2). Cells were

counted at several points on the field until 300 cells or

nine grid squares were counted. Cell killing is expressed as a

survival ratio, S/So, where So = number of cells in control culture

and S = number of cells in experimental cultures.








Purification and Physical-chemical Characterization of TNF

Ammonium Sulfate Fractionation

Saturated ammonium sulfate (SAS) was added to crude TNF serum and

the pH adjusted to 7.0 with NH40H. Samples were left to precipitate

overnight at 4C and then centrifuged at 10,000 X G for 30 minutes in

the cold to collect pellets and supernatant fluids. Pellets were

reconstituted with 50 mM phosphate buffer, pH 7.5.

Ultragel AcA 34 Gel Filtration

Protein from the ammonium sulfate precipitation procedure was

loaded onto an AcA 34 (LKB, Stockholm) column, 5 X 60 cm. The column

was eluted with 50 mM phosphate buffer, 120 mM NaC1, pH 7.0. Flow rate

was 10 ml/hr and 12 ml fractions were collected. This procedure and

all successive purification steps were carried out at 4.

DEAE Sephadex Chromatography

Fractions with TNF activity from the AcA 34 column were pooled,

concentrated by pressure ultrafiltration and loaded onto a DEAE-

Sephadex column (Pharmacia, Piscataway, N.J.) 2.5 x 25 cm, previously

equilibrated with starting buffer; 50 mM phosphate, 120 mM NaC1, pH

7.0. The column was washed with starting buffer until the major

portion of unbound proteins washed through, approximately four column

volumes. A linear salt gradient of five column volumes, from 120 to

400 mM NaC1, was used to elute TNF activity. Six milliliter fractions

were collected at a flow rate of 10 ml/hr.

CM Sephadex Chromatography

A cation exchange chromatographic step was occasionally performed

subsequent to the DEAE column. This step was removed from the final

purification scheme since it did not result in enhancement of

overall purification. CM sephadex (10 ml packed) was










equilibrated in starting buffer of 20 mM acetate, pH 5.0. TNF dialysed

into starting buffer did not bind to this column under these conditions

and was therefore recovered in the wash fractions. The pH was adjusted

to neutrality by collecting fractions into tubes containing small

volumes of a saturated Tris-base buffer.

Con A Sepharose Chromatography

Fractions with TNF activity from the anion exchanger DEAE-Sephadex

were concentrated and dialysed against PBS. This material was loaded

onto a Con A Sepharose column with a 10 ml bed volume. Elution was

with PBS since TNF was not bound by Con A. The column was regenerated

by washing with a-D-methylmannoside, 20 mg/ml in PBS, followed by

starting buffer.

Sephacryl-200 Chromatography

The pooled concentrated TNF containing fraction from the Con A

Sepharose step was loaded in a volume of 2.0 ml onto a Sephacryl-200

column, 1.5 x 180 cm. Elution was with 50 mM Tris, 0.15 M NaCl, 10 mM

EDTA buffer, with 10% glycerol at pH 7.5. Fractions of 1.5 ml were

obtained at a flow rate of 8 ml/hr. The column was separately

calibrated with blue dextran, human IgG (160,000d), bovine serum

albumin (68,000d), and ovalbumin (45,000d) as molecular weight markers.

Fractions with TNF activity were pooled and concentrated as the final

purified material.

Polyacrylamide Gel Electrophoresis

Molecular size and homogeneity of protein preparations was

determined by electrophoresis on 10% polyacrylamide gels with sodium

dodecyl sulfate (SDS-PAGE) according to the method of Laemmli (1970).

Proteins with known molecular weights served as internal markers.

Partially purified TNF and marker proteins








were also electorphoresed on 15% polyacrylamide gels under nonde-

naturing conditions. Gels were frozen and then cut into 8mm slices.

Each slice was macerated in a dounce homogenizer and proteins eluted by

incubation of the gel fragments in 1.0 ml of PBS with antibiotics for

two days at 370 on a shaker table. Samples were clarified by centri-

fugation and supernatant fluids were assayed for TNF activity. This

procedure was also employed for recovery of TNF activity from the

isoelectric focusing gels. Visualization of proteins was done by

staining with Coomassie Blue R-250, or for those cases where 125-iodin-

ated proteins were used, by autoradiography on Dupont Cronex film.

lodination of Purified TNF

Purified TNF, 5 pg in a volume of 20 pl of 0.1 M borate buffer,

pH 8.5, was reacted with 10 Ul of dried Bolton-Hunter reagent (4000

Ci/mmol; New England Nuclear) for 1 hour as described by (Bolton and

Hunter, 1973) and product literature. Unreacted ester was then reacted

with 0.5 ml of 0.1 M Tris buffer, pH 8.5 for minutes at 0O C. Cyto-

chrome c, 100 pg was added as cold carrier proteins and samples were

precipitated twice with cold acetone to remove non-macromolecular label

and finally resuspended in 50 pl of PBS. A portion of the carrier

protein was labeled, presumably by unreacted Bolton-Hunter reagent.

Accurate values for the specific activity of labeling could therefore

not be determined. Specific activity was estimated to be 5-10 pCi/pg

TNF. Marker proteins, 10 pg each, were mixed and iodinated as a

single batch by the same protocol.

Isoelectric focusing

Iodinated TNF was subjected to isoelectric focusing by a

modification of the method of O'Farrel (1975) using N, n'-diallytar-

tardiamide (DATD) rather than Bis-acrylamide as the crosslinking









agent (Horst and Roberts, 1979; Horst et al., 1979) and deleting the

solubilizing agent, NP-40. Briefly, 4% DATD-acrylamide gels with 3% pH

3.5-10 and 2% pH 5-7 ampholites were prepared by poly- merization with

riboflavin-TEMED and persulfate. Gels were run in 0.06 N H2S04 in

the lower chamber and degassed 0.04M NaOH in the upper chamber by the

following protocol: 75v for 15 min., 150v for 30 min., 300v overnight,

and 750v for 1-2 hours. After focusing, gels were frozen, cut into 4

mm slices, and assayed for pH and TNF activity.

Glycerol gradient ultracentrifugation

Samples of purified TNF containing 125 ng of protein, or marker

proteins at 1 mg/ml, were loaded in a volume of 25 Pl onto 5.0 ml

glycerol gradients (20-40%) in Tris-NaCl-EDTA (Sephacryl-200 buffer).

Centrifugation was performed in an SW 50.1 rotor (Beckman, Palo Alto,

CA.) at 45,000 RPM for 56 hours. Six drop fractions (approximately

0.14 mls) were collected from the bottom of the gradients and assayed

for TNF activity. Bovine serum albumin, ovalbumin, and chymotrypsin-

ogen were used as molecular weight markers and were run in parallel

buckets. Protein determinations were by the method of Bradford

(1976).

Enzyme Digestions

Purified TNF was tested for sensitivity to hydrolysis by trypsin,

neuraminidase, pronase, and phospholipase C. Purified TNF, 25 pg, was

incubated with 25 pg of trypsin-TPCK (269 U/mg, Worthington, Free-

hold, N.J.) in 0.5 m Tris-HCl buffer, pH 8.1 with 10 mm CaC12 in a

final volume of 50 1l at 370 for 180 minutes. Extensive trypsin

digestion was done with 250 pg of trypsin incubated for 32 hours.

Purified TNF, 25 ug, was also incubated with 8 pg

neuramindase (47 U/mg, Cl. perfringens, type IX, Sigma, St. Louis, MO)








in 50 mM acetate buffer pH 5.0, in a final vol of 50 pl at 370 for 180

minutes. Pronase digestions (Streptomyces griseus protease, B grade,

nuclease free, 97,100 P.U.K./g, Calbiochem-Behring, La Jolla, CA) were

performed with 25 pg TNF and 10 pg enzyme in PBS, also in a reaction

volume of 50 pl for 180 minutes at 37. Phospholipase C digestion (B.

cereus, B grade, 145 U/mg Calbiochem-Behring, La Jolla, CA) were per-

formed with 25 Pg TNF and 5 Pg of enzyme in PBS in a reaction volume of

50 Pl for 180 minutes at 37. Control reaction mixtures for all

digests were performed in the absence of added enzyme.

pH Stability of TNF

TNF containing serum, 1.0 ml, was mixed with 10 ml aliquots of

various 50 mm salt buffers with pH ranging from 2.0 to 11.0. Samples

were kept at 8 for 48 hours then diluted 1:10 with PBS prior to assay.

The buffers employed were: citrate-phosphate (pH 2-6), phosphate (pH

7.0), and glycine-NaOH (pH 8-11).

Interferon Assays

Interferon was assayed on primary or secondary rabbit kidney cell

cultures by a VSV plaque reduction method (Blalock and Gifford, 1976)

except that target monolayers were established in 24 well plastic trays

(Linbro, Hamden, Conn). Cells were prepared as described previously

(Ruff and Gifford, 1980a).

Necrosis of Meth A Tumors

Meth A tumor cells were kindly provided by Dr. Saul Green, Mem-

orial Sloan-Kettering Cancer Center, N.Y. Tumors were carried as an

ascites in the peritoneum of Balb/c mice by weekly passage of 2-4 x

106 cells into naive recipients. For necrosis assays, 106 tumor

cells, washed and suspended in MEM, in a volume of 0.05 ml were

injected intradermally into mice to establish solid










tumors. Six to seven days later mice were injected with either normal

rabbit serum, serum TNF, or partially purified rabbit TNF. Necrosis

reactions were scored 24 hours later at 1+ (slight necrosis confined

to the center of the tumor), 2+ (appreciable necrosis, encompassing

at least 50% of the tumor), 3+ (severe necrosis, involving the entire

tumor mass). Animals were injected by tail vein in a volume of 0.3)

mls. Serum TNF produced in rabbits had a titer in our 18 hour L-929

cell killing assay of 30,000 units. Partially purified TNF was DEAE

purified material, lacking any interferon activity, adjusted to a titer

of 30,000 units. Partially purified TNF had an interferon titer less

than 30 units, the crude TNF serum had a titer of 3000 units/ml.











RESULTS

Assays for TNF

Tumor Necrosis Assay

The original assay for TNF used by Carswell et al. (1975) was a

visual observation of necrosis in a subcutaneous transplant of a BALB/c

sarcoma, Meth A. Grades of response (- to +++) are recorded. In the

maximum response (+++) the major part of the tumor mass is destroyed

leaving only a peripheral rim of apparently viable tumor tissue. The

response is best and most consistently produced on well established

transplants (7 days) and less effective on 6 day transplants. For some

reason there is virtually no effect on 5 day transplants. This

response is similar to that obtained when endotoxin is given to tumor

bearing animals. Several other assays have been described.

In Vitro Assays

Carswell et al. (1975) used cultures of Meth A, L-929, and normal

mouse embryo cells and counted the number of treated cells remaining as

compared to control cultures after a 48 hour exposure to dilutions of

serum containing TNF. These authors claimed that in a broad range of

tests there were no discrepancies between the TNF activity against Meth

A in vivo and their toxicity to L-929 cells in vitro. We have employed

this procedure with L-929 cells (Ostrove and Gifford, 1979). More

recently, faced with a large number of assays in order to monitor our

purification procedures, we have employed 96 well microtiter plates

(Falcon 3040) (Ruff and Gifford, 1980b). Briefly, 2-fold serial dilu-

tions of TNF samples are made directly in plates which have been seeded

on the previous day with 60,000 L-929 cells in each well. Alternative-

ly, dilutions of TNF are made and then the cells added. The plates are

incubated for 48 hours and the residual cells stained with crystal
29









violet. Visual inspection can reveal that dilution of the sample which

destroys approximately 50% of the cells. This (dilution to extinction

titer) is usually sufficient when relative titers are being determined.

For better quantification, we count the number of cells remaining in

several low-powered microscopic fields using a grid in the eyepiece.

This procedure is similar to that employed by Matthews and Watkins

(1978). We have recently developed an improved method of determining

cell numbers based on a photometric measurement. Trays of L-929 cells

are prepared as usual to establish dense monolayers. At the end of the

culture period trays are stained for a uniform period of time, usually

10-15 minutes, washed with water, and allowed to dry. A photometric

measuring device was constructed by cementing a selenium photocell

(276-115, Radio Shack, Ft. Worth, TX) to a short piece of copper tubing

(6 x 30 mm) whose outside diameter approximated the inside diameter of

the micro- titer plate. A simple series circuit can be constructed

using a pH meter, in the millivolt mode, to read the reference voltage

drop across the photocell as a function of light intensity. Alterna-

tively the circuit in Fig. 1 can be constructed utilizing any conven-

ient DC milliameter as recording device (276-103, PNP germanium

transistor; 271-154, 5K potentiometer, Radio Shack, Ft. Worth, TX).

Our pH meter (Corning, Model 125) has a 1 my sensitivity. Under these

conditions the resolution of the detector is 1000-1500 cells (i.e.

Imv = 1500 cells). The photocell has a linear response at least over

the range 5000- 60,000 cells. The method is somewhat less sensitive

than individual cell counting but enjoys advantages of speed and lack

of bias in cell counting (Fig. 1).

We have shown (Ostrove and Gifford, 1979) that actinomycin D

greatly enhances the cell killing of L cells by TNF. The dose response






























o1i I4-0
c.M I E ) 41
NI 0I0





o :3






-H 44 -H 0)
o)






M 0 0 >


41 1- a o

S0 4 0
I- O
0 0














0 4-1 0)
04 4 U
4O -4 0H 0 P4
;> 0 0






44
,- *rli U
o 4.1 0 C



a)




c! 0
a)cd


0 0 0 f




Q -1 > CO






f-4
a00

-4
fl~d
tn(d~ r































\0


>


O o
0 w
o o

w X:


I
I
O


00



(70JlN03)
(-NIN)


7
Sz
0

O-
0-
-.L
I-


NQ


_ __ I


S/S '31.V -IVAIAlns









curve for cell killing when actinomycin D and TNF are added simultan-

eously is shown in Figure 2 (Gifford et al., 1979). Actinomycin D

shifts the dose response line two to three logs greater in sensiti-

vity. Eifel et al. (1975) have also shown that actinomycin D enhances

the cytotoxicity of lymphotoxin and have developed a similar sensitive

assay for that factor. The photometric assay using cells with TNF and

actinomycin D added simultaneously to microtiter plates is now used

routinely in our laboratory. The procedure shortens the assay time

from 48 to 18 hours and increases the sensitivity so that- lower doses

can be measured. We have developed two other assays for TNF activity

which gives us some clues to how TNF activity may be expressed (Ruff

and Gifford, 1980a). One method determines the extent of inhibition of

DNA synthesis after 48 hours exposure to TNF. The "end label" tech-

nique consists of treating cultures of L-929 cells with TNF for 48

hours in microtiter plates. At the end of the incubation period, the

medium is decanted and replaced with medium containing 3H-thymidine

for 30 minutes and the cells then harvested with a multiple automated

sample harvester. Another type of assay for TNF activity is used in

which the cells are prelabelled with 3H-thymidine or carrier-free

32p for 48 to 72 hours prior to distribution to microtiter plates.

Mannel et al. (1979a) have also used a 3H-thymidine release assay.

Dilutions of TNF are then added and the cultures incubated for 48 hours

at 370C. The released label in this "cytotoxicity release" assay is

then measured and compared to control cultures. The last two assays

are depicted in Figure 3 which is a composite comparison of the four

methods. The "cytotoxic release" is the most sensitive followed by the

"end-label" assay. The "cell-killing" assay is the least sensitive of

the three assays, except when killing is measured in the presence of

Actinomycin D.


































Figure 2. The effect of actinomycin D, 1 pg/ml, on the dose
kinetics of TNF cell killing. Assays performed in
the absence of the inhibitor were for 48 hours,
while those with the inhibitor were read at 18 hours.

















F91

NCi
N~~


N


C;
E
o >


c E
o C
r-
o0


-lOlLNOO JO .N3303d 'S3BYYnN -1-133


z
0

O
-i

C-
z
I.-

0
0


0 0 0 0 0
000 D i N


0.J
t o






















r-


O i H .
c c
o 0 0
o rm



4 3 4-4o
c o

> *i a 0
4O 0 0 0

000

OCO M- 0 O
CO I0 3 CO
S0 i H 4-
w a 1




H C 0 U 0
H% 0 ( 3 C C
0 0 p a u 4





*H -4rl n0 0
a0 4 i-H u0 o
0 0 OH 4-4 a0 H
Ca OO H w a) 0)

-i H 0 0 ,c C
U0-4 $ W 0 PJ W 4
-l 0 H-l m m
0-1 0 -H Co 4H



>O l r 0P -l U P

p 0 r-1 4 C 0 4c 0
0 10 0 a. 4
o U o 4 Co
0 0 a 0 0


<33Q (3
0 ) *r
rO 3 *U

O C 3 C
QA 0 u3
































































- 0 0 0 0
-: b d o


(Wd3 l S/Soy OJUO SnS
(wd3) N1 s/


ID0 I NI
0 0 0 0

PIS!j adoO*JO!UJ JOd 5junoO 110i *ojDOAV
(IOJIUOO) .Os/S 'O04H Io AIAnS
(JN.L)


-10NLN03 -0 iN30tl3d 'sua3814N T133


i :


I I I I
So o o c


(wdo)punojipog9- s!'1 % 001
(--* ------ --- --*1 1 A1131XO01A3 **
(wd3) punl| 30g -( uid3)|10|UMu*!

E
9
o -
c



s s


1 -
0
Cu
C4


8




8


N



0


o 3J
o a


o




I-

0

g-


8


o

N






0
o.
tO


-


c~-c~,








Cells Susceptible to TNF Action

Using these assays, numerous transplanted tumors and trans-

formed cells have been shown to have a high degree of sensitivity to

TNF. These include: Sarcomas S-180 (CD1-Swiss)and BP8 (C3H);

leukemias EL4 (C57BL/6) and ASL1 (A strain), and mastocytoma P815

(DBA/2). TNF also showed activity against several transformed lines in

culture. L-929 cells were most sensitive, Meth A cells less sensitive,

and normal mouse embryo fibroblasts (MEF) were insensitive. Helson et

al. (1975) extended these observations and showed that human melanoma

cells whose growth had been inhibited by mouse TNF when related in the

absence of TNF were still inhibited and did not resume their normal

growth rate.

Not all transformed cells are affected by TNF. For example Old

(1976) reported that 21 human tumor cell lines were screened for their

sensitivity to TNF. TNF was cytostatic for 10, cytotoxic for 1, and

had no effect on 10 others. The reason for the variable effects of

TNF on transformed cells is presently unclear. TNF presumably recog-

nizes common feature of some transformed cells, but not all, which is

missing in normal cells.

Purification and Physiochemical Properties of TNF

Since TNF is induced by a technique which causes a general reti-

culoendothelial hypertrophy one might expect TNF serum to be a rich

source of lymphokines as well. Indeed this is the case. Interferon is

produced in large quantities (Sauter and Gifford, 1966) as well as a

factor which causes differentiation of B lymphocytes from complement

receptor minus (CR-) to complement receptor positive (CR+) Hoffman

et al., 1977). TNF serum also stimulated bone marrow progeniter cells




























UA(


*0 Z H
14 1 -N c)
Cd0- C'4-4





a4Ju 0


4 0 0 4- 0,




U 0 *2 H
0 tH 1
r-r4 | 0 4-4 0
4J (0 "0 4-1 ) Ca



S0 W 0 S U4
u m H >1 o
a) ca "d 41 41




p0 D; 0 -d H
z t Cd >


cd Z cn O
D& (U -H 0



b20 4M d 0 o
H 0 H




4 *H H
cd



o0 H f Cd
lo r 4- a) 4- 0
0 dH 0 4-
4J 4- 1-4 i 0.


0 M 3 0D


c p a 1 o

0 *H H 41




M~ 0 14-4 4 *1-i







a0


*H
i4l
cloccl















x- x 'oquW!l!W (U/ ) aouoDs!sa8


-e '082(0









to differentiate to form colonies of macrophages or granulocytes in

vitro (Shah et al., 1978; Butler et al., 1978). We realized that it

would be necessary to purify it, free from other cell inhibitory and

cytotoxicity activities.

Since we started with large volumes of rabbit serum we employed

two preparative procedures as the first steps in purification of TNF,

an ammonium sulfate precipitation and a gel filtration. Pilot studies

indicate that less than 0.5% of TNF activity was present in the 60%

saturated ammonium sulfate (SAS) supernatant fluid and that no appreci-

able increase in specific activity was provided by excluding a first

precipitation with 20% SAS. The proteins precipitated in 60% ammonium

sulfate were resuspended in 50 mM phosphate buffer, pH 7.0, and were

then loaded onto an Ultragel AcA 34 column. Three major protein peaks

were generated corresponding approximately to an excluded peak greater

than 300,000 d, included proteins of range 40,000 d to 300,000 d and

those less than 40,000 d. Fractions with TNF activity in the 40,000 to

300,000 d second peak were pooled. Gel filtration data for TNF is

presented for a later purification step. Further resolution of TNF was

then obtained by anion exchange chromatography on DEAE Sephadex (Fig

4). Unbound proteins were washed off the column with starting buffer

and a linear NaCl gradient was used to elute TNF activity. Significant

amounts of TNF began to elute in 50 mM phosphate buffer, pH 7.0, with a

salt concentration of 160 mM and continued until approximately 260 mM

NaC1. Shallower salt gradients over this range did not improve the

resolution of this purification step. Fractions with TNF activity were

then passed through a Con A affinity column. TNF did not bind to this

column but several contaminating proteins were removed by this pro-

cedure. Since there was an appreciable loss of activity with this








step we verified that TNF did not exist as heterogeneously glycosylated

species by performing pilot studies using DEAE Sephadex purified

proteins with 125I labeled purified TNF added. TNF activity, as

well as radioisotope counts, was only recovered in the wash through

fractions. Sequential elution with a-methyl-mannoside 20 mg/ml, 200

mM borate buffer, pH 6.9, or 3.0 1 KC1 did not release any activity or

counts from the column. TNF activity is stable in borate and KC1, at

least for the time required to make these determinations. The wash

through fraction from the lectin column was therefore concentrated and

loaded onto a Sephacryl-200 column for gel filtration. Glycerol was

added to the buffer system to minimize protein adsorptive losses. As

shown in Fig. 5, TNF activity was well resolved from the major

contaminating proteins and gel filtered with a molecular weight of

approximately 55,000 daltons. The specific activities resulting from

the various purification steps are shown in Table II.

Criteria for Purification and Further
Estimation of Molecular Weight

A portion of the TNF recovered from the Sephacryl-200 gel filtra-

tion column was iodinated and subjected to several analytical pro-

cedures. Labeled TNF was electrophoresed on 10% SDS containing poly-

acrylamide gels which were then dried and exposed for autoradiography.

The homogeneity of the preparation was demonstrated by allowing films

to overexpose in order to detect minor protein species (Fig. 6). We

were not able to accurately compute the specific activity of TNF

iodination since some cytochrome c, which was added as cold carrier,

was also inadvertently labeled by unreacted Bolton-Hunter reagent.

Approximately 230,000 total CPM of TNF plus carrier protein was loaded

on this gel and film was exposed for 14 days. The apparent molecular































u Q
* l E r l



H M3
B 13 ) a)
w 4-1 60 i-l
4-4 0
r-4 >*
ri- 3



0 a)C 0
,-4 0 ,- I 4






4I r-1 40

ao ,
*i *r
C4 1 $3






0) 0 d
:34a -




dU 1
Co 3 C a)


MO *C d



0) 4 a) P4

0 0 *H1
o -i a





4J 0 4-J

u n)

Su-r a )









r11











o-----o OIX IH913M Iv-lnO33710I
o -. o
C oQC cO I 0 CIM


v v


'7VAIAInS 1133


O 0
I--- -)


I '


o ---

CD---- y


d d o d d
0 o d C5


1 1 I a I I I


i i

















00
aI 0






4-4 0 0
H11 4. 1

-t,-I 0





OW O
P-1 1-1







4









rC
1 -H-i


























0 "-
uE-4
a ,'










41-i
03











P-4
P-I
o






0 ^
























In


o 0 0 0
-4 00 c14 0
H CO CMD CV0


o L 0o
o 0t CM 4
H CM


--T
r-4
0 0 0 0
* *


0 0 00 0


o cc o
r-f









o c00 o cn
Co C'4 C0





0 0 0 1
h g


I- c'
CN


0oo
o.
CO)


0 0 \0
CM C *


-- 0
e n co 0
< H 0 C0 H
0 CO 01 P 3


Sl l E-- J 0
< 3 0 C a g m
0) M ca P4 e 0
3 p 4- <
o O 0 I0
U o 0 Q 0


HtJ
H
>-N



W I



4-1
l-

P.


c00
H 0 c-'
r-H 0 C






o C 0
.-I 0 C\
,-1






CV)











o C0 o
co 'o



















S M 0
Cn 10 o









S 0
,c-
4 C; o
*


a)


S4-


S






0
,V cd

r 0




Va II
C 4
CO 0


0 4





































SDS-polyacrylamide gel electrophoresis Purified TNF and
marker proteins were labeled by 125-iodination as detailed
in Methods. Gels were dried and exposed for autoradiography
for 14 days in order to reveal minor components. Markers
were: phosphorylase b, bovine serum albumin, catalase,
ovalbumin, and cytochrome c. Cytochrome c was added to
purified TNF as carrier protein.


Figure 6.









..- Phos. b

-*- BSA
-- Catalase


-- OA








S.-- Cyt. c


TNF









weight of purified TNF under these conditions was 68,000 d (Fig. 6).

Purified TNF samples prepared for electrophoresis by boiling with SDS

as well as the reducing agent B-mercaptoethanol showed an identical

protein staining profile implying the absence of any reductive subunit

structure for the molecule. (Data not shown.)

The electrophoretic characteristics of TNF were also examined

under non-denaturing conditions. Partially purified TNF obtained from

a cation exchange chromatographic step, CM Sepharose, was used in this

analysis. As seen in Fig. 7 the TNF activity ran to the anodic side of

the BSA marker, in the a -globulin region of the gel. TNF activity was

only associated with the single Coomassie R-250 staining band and is

clearly resolved from the BSA marker. On two occasions 50-60% of the

TNF activity could be recovered from the gel. Proteins recovered from

the native gel were re-run on SDS-denaturing gels with various marker

proteins. The major staining material co-migrated with the BSA marker,

as it does in Fig. 6.

Since there was some discrepancy in the molecular weight evalua-

tions for TNF between the gel filtration and SDS-PAGE we wished to

ascertain the size characteristics of the molecule by a third method.

Glycerol gradients were employed and TNF and marker proteins were run

in parallel buckets. As can be seen in Fig. 8, TNF ran as a symmetri-

cal peak with a molecular weight of 52,000 d, in agreement with the gel

filtration data.

Isoelectric focusing

Purified 1251 labeled and cold carrier TNF was subjected to

isoelectric focusing. Replicate gels were either dried and prepared

for autoradiography or sliced and assayed for TNF activity. Peak TNF

activity was recovered in a fraction whose pH was 5.1 and which





























FZ4 Q)
z U
4-4 *-




w 00oo 4J
0 ,U (



S00) 0
o 0 4 O
'Ca -
- 4-






0 o *H
SC. 1 0)
4 0-c
0 4-1 -
Cd P 03








r 0 -H ("
0 m 4.1

a) > -i
4-1 0O Ca





0o 0 a)oa
a 0 4-a4
r41 4
H0 *H





oO
H UH '0


0 H 0
~a) a)

I CO4-
3 Ca H 0






a)










C,,


o
N 8



I Q
2 U 2

m






.-Ko
A ;.


"It

i


0


I I I
0 tf 0
0
Pai
o!doosoJoi!u jad slunoo sll|9 a6oJaJAV
0S/S 'O!D8 I|DAIAJnS
































4-4
TC
10 z
CO H *I-9
0 0
0 24- .!



E4- 0 C
Ea 41 :

co-l U: (
1r Cd H


4-1 1- 0 r.
*t-l4 0 P
H. E-4



* cn Ji 9

0 30

bO a) P <

p .H

4- 0 0
1 %4 o >

o0 to
0000



S0 h-

4>1 P-

r-4 o cZ

n H< 0 n

00 4










bO
60



tnr








(IO4IUOO)
(JNi)


oS/s


'OUIV8 1VAIAfns


C:
4)
0.
C:
oC
Q.


E
-C


0


<
CoO
m)-


(D
-N



C0

-COJ
.o


-~ z
0



LL
-0


-(C


-CN


I I I I I I I I
0X0 ),- w toqt

X 1H913M vfIn" 3~IIOlV


o---o t,_01









contained the single autoradiographic band (Fig. 9). Two dimensional

isoelectric focusing and SDS-PAGE were also performed (Fig. 10) as

described by Horst et al. (1979).

Enzyme digestions

The protein nature of TNF was demonstrated by its susceptibility

to digestion by pronase (Table II). TNF was however relatively trypsin

resistant, implying a low number of exposed basic amino acids. Extens-

ive trypsin digestion did destroy TNF activity. TNF was also insensi-

tive to neuraminidase digestion although there was a decrease in activ-

ity observed due to the pH 5.0 incubation. Phospholipase C also had no

effect on TNF activity.

Sensitivity to pH

TNF activity was stable over the pH range of 6-10 (Fig. 11). The

protein showed some loss of activity after exposure to pH 5 to 5.5 for

two days at 4C. At more acid pH the activity began to drop much more

drastically, and the protein was completely inactivated at pH 3. In

working with the protein we have found that activity will survive

relatively brief (hours) exposure to pH levels as low as 4. However a

3 hour exposure at 370, pH 5.0, resulted in a 60% loss of activity

(Table III).

Temperature stability

Purified TNF in PBS, pH 7.0, survived heating at 700 for 1 hour

with no loss in activity. Over 99.9% of the activity was however lost

after 15 minutes at 800.

Hemorrhagic necrosis by Rabbit TNF

The activity originally described by Carswell et al. (1975) as

tumor necrosis factor had the following distinguishing characteristics.

It was elicited in experimental animals following C. parvum or BCG



































C~4

'P4
10 z
a) E-4



o
T3 >,













sH -
u-











U
P4
0 co

















0J

r4
*H







o i






OH
1r4
tn u
F^ 4-


Q\

(U
M3>
) n
ilCo












(IOJlUOo) os 'OLiVI -IVAIAINs
(dNi) S
io o co o o
IO 0D ) 0 ) o






.O


.0




II,
00








o 0 0 0 0
N
Hd
































MOC
0 w
S0 -4 a) c 9
0 0 *i3
o0. r-4 > H
Cx C )a (U (U 0 rl)0
Z W 4-J 4J 4.J
E-4 P O O O U 0
10 0 Z 4J 3 (U 0.

4-4 I4-4 :
H P 0 p
p 0 C0 -r-4 -)


.041 0 4 )

*rlF o 0 *4

0) AU .O'U0 H
0 W w bo 0
0 a0 Ci ml


-H m -
400 U 4-
i ,-4 (3 C -I
0r-1 H 4 0 ( 4
rCa bO4J -
,-l 0( ,-4 ( 0 4J

Hca p -r 4 -

4OJ C u -r 0

4J (U CO CO (UQ
S) 0 d co >
0t C4- 3 *

1-1 0 3 9 r-1 0
s d o Ba 4o 0n




r-4
00
r-i













SDS-PAGE


oo Ln Ln


LIn























u\
0

-LL

4
LU
I-
LU
0








O








CO
=
4 o.































*00
>m
*H z
4-J I4
U 44 0 0 0

, M 'Z
CU 4-4 p r.


P 0 00



0 d0


4J CO 1 4-
C3 > 4-i4- .
S0 3 -
o 0
41i 41
i-l ^3
0 40 C
W0 4- m0) ri C
3 > 0




h cc t
M (1 tfl Q

o 0 C 0 *
E n 4 C4-


4J U >- l4J
44 CO co c
0 0 CD 4
0 w C14



< 00 0p .
4 |r4 0 a cU -4


























-0

-0)

- co


0.


- O


-CMj


I I I I I I
mI O I O -0
ni nil -- -'r C


(sp!un -IN ) 601












TABLE III

Stability of TNF to Digestion by Various Enzymes


Concen- Activity Remaininga
tration Units Percent
-3
Enzyme yg (x 10 ) of Control

Control (PBS) 116.0 100.0

Trypsin 25 123.0 106.0

Trypsin (32 hours) 250 1.8 2.0

Pronase 10 0.93 0.8

Phospholipase C 5 104.0 90.0

Control (pH 5 acetate buffer) 39.5b 100.0

Neuraminidase 8 42.7 121.0


apurified TNF, 25 pg, was incubated at 370 C with indicated enzymes
for three hours as detailed in Methods. Extensive trypsin digestions
were performed for 32 hours.

bNeuraminidase digestions were carried out at pH 5.0. TNF loses
activity under these conditions (Figure 11); therefore a separate
control was performed.








priming followed by endotoxin challenge, it was capable of selective

toxicity for transformed compared to normal cells, and was capable of

eliciting a necrotizing reaction within the transplanted solid tumor

Meth A when passively administered to tumor bearing animals.

We employ the same procedure as Carswell et al. to produce crude

TNF containing serum, and we have already reported on the discrimina-

tory killing of transformed cells by this substance (Ostrove and

Gifford, 1979; Gifford et al., 1980; Ruff and Gifford, 1980a; Ruff and

Gifford, 1980b). In order to evaluate whether rabbit TNF, and in

particular whether purified rabbit TNF was capable of eliciting

hemorrhagic necrosis, tumor bearing animals were tested with control

serum, crude TNF serum, and partially purified TNF. Purified TNF was

not available in sufficient amounts for this test. Control serum did

not elicit a necrotic reaction within the tumor mass. Crude and

partially purified TNF elicited a 1+ hemmorhagic reaction indicating

that the designation of TNF for this protein activity is correct. A

list of the physical-chemical properties of purified rabbit and

partially purified mouse TNF is included in Table IV.

Mechanisms of Action and Cellular Effects

The following discussion will describe our initial attempts to

gain an insight of the general kinetic parameters as well as the range

of effects which TNF has on cells. Many of the studies were performed

with crude TNF while we were awaiting final purification. They are

therefore subject to certain reservations, primarily a concern that

effects which we observe may have some contribution from contaminating

activities present in TNF containing serum. Where results have been

verified with purified preparations they will be so indicated.












TABLE IV

Physio-Chemical Characteristics of TNF


Rabbit


Molecular Weight
Gel Filtration



Glycerol Gradient

SDS PAGE


Non-denaturing Electro-
phoresis

pH Stability

Isoelectric Point

Temperature Stability


Denaturing Agents


Enzyme Sensitivities


52K
40-50Kb


150Ka
125Kc
60K high salt


50K

68Kb
60K"


a-globulin

5-10 stable

5.1


70 C, 1 hr. stable
80 C, 15 min. unstable

Acetone, 2-mercapto-
ethanol

Neuraminidase-stable
Trypsin-sensitive
Pronase-sensitive
Phospholipase c-stable


a-globulina


56 C stabled
70 C, 1 hr. unstable


Glycoprotein


Presumably not


Presumably yes,
sialic acid
galactosamine


aGreen et al. (1976)

bMatthews and Watkins (1978)

cMannel (1979a)

Carswell et al. (197 )


Mouse









Dose Kinetics

The shape of the dose response curve (Figs. 1-3 ) has some

features worth noting. The curve is approximately sigmoidal when the

abscissa is plotted on a logarithmic scale. A common interpretation is

that the decrease in slope at high concentrations of TNF is due to re-

dundant sites of injury, while the decreased slope at low concentra-

tions is, in general terms, due to insufficient events to complete cell

killing. The slope of the dose curve can be expressed as the fraction-

al change in cell numbers compared to the log decrease in dose, meas-

ured over the linear portion of the curve. Theoretical considerations

(Gifford and Koch, 1969) show that a slope of 0.705 would be expected

for single-hit kinetics, i.e. one molecule of TNF can kill a cell.

Since TNF dose curves consistently have slopes much less than 0.705, in

the range of 0.35 0.45, one interpretation is that one molecule of

TNF can kill more than one cell. This is compatible with an enzymatic

mode of action for TNF, although other interpretations are possible.

Poisson statistics are appropriately applied to homogeneous popu-

lations and in their theoretical considerations Gifford and Koch (1969)

point out that variation in sensitivities among members of a population

may lead to artificially reduced slope. For example, they show that if

two types of cells exist, differing only in their sensitivity to a par-

ticular agent such that they have identical probit regressions, but

displaced along the abscissa relative to one another, and both curves

have slopes of 0.705, then an equal mixture of the two populations

would generate an intermediate curve with slope of 0.56 (i.e., low

slope).

In order to evaluate possible heterogeneity of cell types in the

population of L-929 cells with respect to sensitivity to TNF killing we

prepared clones of L-929 cells. Figure 12 shows the dose relations for































44-4
0

co

0
r-q
0



CH



0
r-4






.1-
r1







0

0
ril











0
Pr


4-4
0
Q)













4-1
0



CO4







H
CO









u4-






00
1-





65







O
O


_..1

0
x 0
O



0O
i, -x o


X



0
j -0o
0



0



\ T* I

-0 --








.0
V0


O U0
0
0





(IOUO) n


(1JNuo) os 'OIlvId 1VAIAIlS









five clones which we surveyed. We readily isolated clonal lines with

varied sensitivity to TNF. Interestingly however all the clones have

dose regressions with slope identical to the parental population. Thus

the "low slope" of the dose curve we obtain in the bulk population is

not due to the summation of "high slopes" of individual population

members. The similarity in slope among clones and parental population

was an unexpected observation since in the hypothetical case where

there is a 1:1 mixture of two cells with varying sensitivites to a

toxic agent the net toxicity as a function of dose would not be

expected to generate parallel curves. For example, at low concentra-

tions of the toxic agent the contribution to numbers of dead cells by

the less sensitive population would be negligible compared to

the more sensitive cells. The total mortality will be almost identical

to that for the sensitive cell alone. With increasing cytotoxin

concentration the contribution from the less sensitive population

begins to be observed so that now the total mortality is greater than

for the sensitive cell alone. Finney (1964) discusses these conditions

and further points out that for the case where the dosage regression

lines for mixtures of cells and for the constituent cells are all

parallel a synergistic mechanism may be operative. That is to say a

more TNF resistant cell might tend to "protect" a more sensitive cell,

or the converse.








Time Kinetics

Onset and rate of cell killing

Growth curves for L-929 cells cultured in the continuous presence

of various amounts of TNF (Fig. 13) reveal a time and concentration

dependence for cell killing. The onset of cell death, measured by

counting intact cells microscopically, is delayed for approximately 10

hours, even in the presence of relatively high doses of TNF. When, in

the same experiment, sensitive target cells are isotopically pre-

labeled with 3H-thymidine, one finds that high doses of TNF are more

rapidly cytocidal for these cells, as indicated by the 63% release of

label at 12 hours (Fig 14). The membrane permeability change precedes

the more slowly progressing overt cell destruction since cell recovery

did not begin to drop substantially for an additional 12 hours. More

dilute amounts of TNF result in a dose and time dependent decrease in

isotope release and growth inhibition. Thus increasing amounts of TNF

result in cytostasis, followed by leakage of labeled macromolecules or

degradation products, followed finally by complete cell disruption.

These data also indicated that both proliferation inhibition as well as

cytolysis occur in TNF treated cultures.

The time kinetics of cell killing due to TNF was examined in more

detail by two methods. In a cinemagraphic study, time-lapse photog-

raphy has provided a more detailed analysis of the rate of cell

killing, as well as the apparent lag period (Fig. 15). In the case

where one deals with larger populations of cells, it is not possible to

exclude the possibility that a small fraction of cells may be dying at

times significantly earlier than 10-12 hours. The absence or presence

of a lag period, as well as its duration must be incorporated into any

mechanistic theory of TNF action. The time lapse photography provides































4-4
0







0
(U
C,



01
4













I rn
0








'-*4


a
cA)
01





1J 0


E-4

COn


a)

'I




w0


0'4
3 -









rn





















O
I
I
I



I o

w

II
OE












0 0 0
0 O O
o -- -











o O O
O N


031J 3d03SOIO011Y 83d SINnoo -1-30 39V813AV


.LL
z
I-
UL
0
z
0
I-
-J































CO
(U
u -



ao



*Ju

)0 41


6 0o

4 k 0
)O




0 r- 41


0*0 0
il O r-H




-4 C
44 -W C-
-l I

0 ,w =
u: p

a0
to


U'0
41 4 0C



H CO *-'


o S

S60 r-






!-4


60
-)












-LL
z
I-



0 CM (0
0
I-


NO


(VWldO)punoJ6)Dog -


s!sAI %001


(WldO) punoJ6iODQ (VdO)loDuatuu!jdx3


=A1IIIX0101AO %


c0 t
cD

0

LLJ
I





























4-4 P
0 co 0
00o
to o3

O



0
0H H


0O
a4-

C C





E0 -
0) L-







r *H 0
H H
O4


at 0










0 p


u0 0
l I ) *0





0 m




U o g .
0 *
*40 r .0
H 0
oc nis














O
X ZS
< 0
x >
L Z
LL

>0
0
O

O


" 0O

90.

--0-
-0 6o


o 0o Do 0 0
-d o o d


(70lJNOO)
U 1No-)


S/S 'OllVI -7VAIAfnS


0
It






0
m
z

CIJ









the capability to observe single cells for the duration of the

experiment. Just before the cells die they start to round up and

become very granular, a process that takes about 15 minutes (real

time). At the moment of membrane depolarization the cells seem to

"pop" and immediately become very refractile. This event was used to

mark cell death. A plot of survival ratio S/So vs time of individual

cell death generated a first order decay curve with a lag period of 10

hours, the time the first cell died, in agreement with Fig 13. The

slope of this curve corresponds to a rate of cell killing of 1.6

percent/hr.

Effects of dose on cell killing

The effect of different doses of TNF on the onset of cell death

and rate of cell death were most conveniently determined by staining

trays of L cells at various times and determining survival ratios

photometrically. We have already referred to the synergistic killing

effects of TNF in the presence of actinomycin D and cycloheximide,

inhibitors of RNA and protein synthesis, respectively. Since cells

which were treated with these inhibitors were blocked from dividing, it

was possible to determine the kinetic parameters of cell killing

independent of a background of cell division.

Actinomycin D, at 1 pg/ml, enhanced the rate of cell killing by 10

fold compared to its non-inhibitor treated control. Additionally, the

lag period was shortened from 10 hours to 4 hours (Fig. 15). Cyclo-

heximide, at 5 pg/ml, showed identical results. Since cycloheximide

blocked cells did not die any sooner or faster than the actinomycin D

blocked cells, this would imply that essential new translation for pro-

longed survival does not detectably lag behind new transcription.

Cells pretreated with actinomycin D for four hours prior to TNF

























U)
41 rI C (3)
S a0 ) co *) Ha
4-4 H-1 4 4J 0
0o n ()C 41
>, T4 a) -
* UH 0 ) .
4-i Q)H u o41 i-i


.d c
00 *Q .0 3 0
n Q4JQ



o a co
Srl 1H *
41 o (U a 1f a fi 4
U 4-1 0
4-4 0 4-4 U 4- J

S cd 0 < M 0
0 0 0 0 r
o -W


*0 0 l )0 p
0H 0 i-4 05-'5-3 0
41 H r-1 0 4-4
u u 0 w ) a
0) <) 0 a- ) ( (3
w 0 0 M 044
,a -D a 4 co 0
41 I (U r *
z 54- H C U)
en 4-4 p r. *
0 (co 4J 4

za u 0 4-1
00 *r 41 ) 4-1

0 m- a) >
u 03 o d 3 41i
W *H *-H a) u 0) r
4-4 p i. bO w :3
4- 0) 0 r 3 ) 0
W 4-) > CO -HOP V U








rz1
-'-











N-




r.CD

L-
I 0
CO
LL I


Sz















oJ
I I


u H



FO



/ I

I O





IIn o n







C o I In I


1OdllNO3 JO % 'b^dZ










addition showed identical kinetics to non-pretreated cells. One

possible mechanism of action that we considered was that TNF may have

accelerated the normal decay of mRNA or that it inhibited mRNA

function. Such a mechanism would help explain why both cycloheximide

and actinomycin D treatment of target cells resulted in a similar

accelerated kinetics of cell killing by TNF. TNF and actinomycin D

were added to L cell cultures and at various periods of time thereafter

replicate wells were washed with isoleucine-free medium and then

isoleucine-free medium with actinomycin D was added and the cultures

pulsed with 3H-isoleucine for 30 minutes. Incorporation of

3H-isoleucine into trichloroacetic acid insoluble material was then

determined. As shown in Fig. 16, the loss of mRNA activity as a

function of time was not significantly influenced by TNF. We did not

determine any incorporation studies beyond 4 1/2 hours since signifi-

cant cell killing begins at that time. Since pre-treatment with actin-

omycin D for 3 to 4 hours prior to TNF addition does not result in any

further enhancement of cell killing, it was possible that TNF might

inhibit a more stable mRNA species. Cultures were pre-incubated with

actinonycin D for 3 hours and then TNF added. As can be seen in the

figure, TNF has no effect on the ability of the stable forms of the

messenger to be translated.

We also determined the effect of addition of cyclohexamide and

TNF on the possible hydrolysis and/or release of prelabelled protein

at various periods of time after the addition of TNF. Cells were pre-

labelled for 2 hours with 3H-isoleucine, washed to remove unincor-

porated label and incubated for an addition 1 1/2 hours to chase the

label into protein. Following washing again, medium with cyclohexi-

mide, with or without TNF was then added. The amount of label in the








supernatant fluid (acid soluble and acid insoluble) was measured as a

function of time. The amount of label found in the acid insoluble

material is shown in Fig. 17. Significant increase in protein released

from TNF treated cells appeared between 5 and 6 hours after TNF treat-

ment. The enhancement of isotope release between 5 and 6 hours is con-

sistent with the accelerated killing kinetics by TNF in the presence of

cycloheximide or actinomycin D. Significantly, no differences in acid

soluble extracellular label appeared, indicating that widespread and

general protein hydrolysis was not the cause of TNF induced cytolysis.

Within the context of the proposed repair capacity of the target cells

(Ostrove and Gifford, 1979) these data would be consistent with the

following. At a very preliminary level the enhancement of cell killing

by TNF in the presence of inhibitors of protein synthesis would suggest

a protein target for TNF. A corrolary argument is that TNF does not

merely block protein synthesis but must inactivate or destroy a protein

dependent cellular function, perhaps structural. Because sensitivity

to cycloheximide and actinomycin D show similar kinetics of killing,

which is not enhanced by pre-treatment with inhibitors, the putative

"repair" mRNA species is likely not to be constitutively expressed but

rather induced by TNF treatment and co-ordinately translated.

Some additional features of cell killing are worth noting at this

time. The lag period which precedes cytolysis is dependent on TNF con-

centration, more dilute amounts of TNF extend the lag period. As noted

earlier there appears to be a minimal time of approximately 4-5 hours

in which no cell killing occurs. The dependence of the lag on TNF dose

shows higher than first order kinetics (Fig. 18). A linear transform-

ation of these data can be arrived at by plotting lag (hours) vs. log





























El
0)
0.00
p 4J .C
4- C O ) 4-J

t0 -H *H4a

30 0
o *H 0 0


HOO4





1 4 -

H 0cQ) :l

Sr-l & U)



00 Q.1 V) a) n
r-I = 4
Ol 4-1




SC


H *H 0 0


0 0 U)






04-4 0 4H 0
3 H 3H*r





00


*H





80











L0
Z c
I- 0 0 _




|0
o 0




0 0


I A
00 1 1




6 I





O L0 to
0 r- 10 C

S1130 InOJd O3SV3138 N13l0Odd 031138V1 IN30d3d































0


0
i-H








60


ca
o






4-1 0
00
0 *I







00



41
00
C C


Co 0






0 p
C 44
0*
a) )


cU






a C
*Hl
U h)
'41 )
0 H 0




FE) i- S\






82







0
-0












T

z
0
O .j
0 D


C -

O

0
0


0
0

0
S-
00

0 0 0


(*sjL ) BD-









(TNF dose) (Fig. 19). An interpretation would be that the onset of

cell death, and by extension the killing process itself, is not

dependent solely on the concentration of TNF but rather, on a

simplistic level, there must be an additional component in the system.

One speculation, consistent with the observation, would be that TNF has

to gain entrance to the intracellular compartment to express its

effect. The lag represents the minimum time period required for the

cell to achieve a steady state concentration of TNF incompatible with

continued viability. In a later section the inhibition of TNF action

by agents which disrupt microtubules and microfilaments will be argued

as also suggesting internalization of TNF, as opposed to action

strictly in the external environment, such as at the cell surface.

The rate at which cells die, however, seems to be independent of

TNF concentration over the 160 fold range examined (Fig. 20). Once the

threshold limits which commit a cell to destruction have been reached

further insult does not cause it to die any faster. Serum TNF may

therefore not be the immediate agent of cell death. The actual cyto-

toxic mediator may be "activated" TNF or cellular hydrolytic enzymes.

One alternative explanation would be that as far as the actual killing

event is concerned we are still working in a range of TNF excess so

that the system, even at lower doses of TNF, is effectively saturated.

These type of studies provide useful guidelines for further work but

they suffer from an important flaw in the sense that a study of

kinetics of cell killing does not directly provide information on the





























,--4
oO

C)

.i-4
0)









4.i

0
cu








4-J



0
0
0
-H

41
Ca










.IJ








0
-4












0
>-1





























































B I I


(sJ ) B6D


GO
















-
C

C LL.




O
or






*. 1

0






to<



























ON p
1C




oo>
I *rl


&- 0

o 0


E4-4 o





0o 44
400






4 -1 0



4-J a ;
41 *H






S-4
00 0









44 m0

E-4H Ur)
0 & o







44 U 0 0
0
Scumo
0 '0 4
4 o


















'-H


rX4
60















o4o0xo
000000
*-0000


loJLuo3 = os,/ 'OIV -IVAIAflnS
INI sX









kinetics of TNF action. Cell death must necessarily be the last of

perhaps many intervening events and must also be the converging point

for any substance which kills a cell. Further work to elucidate the

mechanisms of any cytotoxin must focus on the earliest detectable

alterations within the cell.

Killing by limited exposure to TNF

We wondered how long TNF had to be present on cells to exert its

killing effects. In the absence of any inhibitors, cells were treated

with pulses of TNF for increasing periods of time. TNF was removed by

washing and cultures were re-fed in TNF free medium. Survival ratios

were determined at the end of 48 hours. A period of exposure of 1 hour

with a high dose of TNF is sufficient to result in appreciable

mortality at the end of 48 hours (Fig. 21). Pulses with ten fold lower

concentrations of TNF serum also show appreciable killing after limited

exposure to the cytotoxin. Cells thus seem to be committed to die

within the first hour and the continuous presence of TNF is not

required.

Temperature dependence of TNF action

Matthews and Watkins (1978) reported that L-929 cells grown in the

continuous presence of TNF at 210 were not killed at the end of four

days by a dose of TNF which killed most cells in 2 days at 37.

Furthermore we have ascertained that such temperature inhibited cells,

at the end of four days, when shifted up to 370, promptly all die

within the next 24 hrs. Although control 370 cultures do show some

lethality at 24 hrs (fig. 13), the acceleration of killing kinetics

after pre-treatment for several days at 250 suggests that there may be

both temperature dependent and independent aspects of TNF action.



























o


a) Cn o0
,= 4-J 3L
4-j1
C *u

co 0











or-1
co




4.J
0 *0




Cor 1

> 0 .H



,4 *I i o
0 Wo
C.. 0 )-
o0 e a
CO
*l *H C

















*n H0
rM 0










0 0
F *rl

cu o






c00 }




4-) >t
C 0





CS 00
4-1 -





90














4-.
*r-


z cm

S(D




-)0 o
.-


CL







E






0 00




I I I /

0 d d 0

(IOJIUO3) Oo
(j.IJ L) 'O2IVI ;SVAIAlnS









A more extensive survey of the time kinetics of TNF killing at

various temperatures utilizing the more sensitive modification of the

basic microassay is shown in Fig. 22. Despite their greatly enhanced

susceptibility to TNF when cultured in the presence of actinomycin D,

L-929 cells are still refractory to the lytic effects of TNF at 250.

As the temperature is elevated the onset of cell killing is acceler-

ated, as well as the rate at which the cells die. Cells grown at 390

showed substantially more lethality than their 370 counterparts at the

end of 10 hrs. The temperature dependence of killing is consistent

with the idea that metabolically active cells are better targets for

TNF action. The temperature dependence is also consistent with the

enzymatic mode of action suggested by the slope of the dose curves.

Effects on Macromolecular Synthesis

In a simple sense we knew that ultimately TNF must be inhibitory

to all anabolic pathways. We did not know if cell death occurred

directly as a result of TNF mediated shut down of host synthesis or

whether inhibition was the repercussion of an earlier, yet to be iden-

tified, lesion. If the ability of cells to synthesize DNA or protein

is measured by incorporation of labeled precursors in the continuous

presence of TNF, shutdown of these processes is delayed by approxi-

mately 12 hours and proceeds at a rate comparable to loss of cell

recovery. Ostrove and Gifford (1979) described a very interesting

paradoxical increase in 311-uridine incorporation in TNF treated L929

cells. A sixfold stimulation of RNA synthesis was reported 24 hours

after TNF addition. Normal mouse embryo fibroblasts, or clones of L

cells selected for their relative resistance to TNF, did not show this

effect. They further reported the synergistic effect of actinomycin D




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ELGUZI4EW_EOO6XJ INGEST_TIME 2012-03-02T21:30:33Z PACKAGE AA00009123_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES