0019-9567/80/04-0204/08$02.00/0
Generation and Characterization of a Lipopolysaccharide- Induced and Serum-Derived Cytotoxic Factor for Tumor Cells
DANIELA N.
MANNEL,It*
MONTES.MELTZER,'
ANDSTEPHAN E.MERGENHAGEN'
LaboratoryofMicrobiologyandImmunology, National InstituteofDental
Research,'
andLaboratoryofImmunobiology,
National CancerInstitute,'
National Institutesof
Health, Bethesda,Maryland
20205Serum from Mycobacterium bovis BCG-infected mice treated with lipopoly- saccharide
wascytotoxic
to tumorcells in vitro. Serum-induced cytotoxicity
wasestimated by measuring release of [3H]thymidine into culture supernatants
ofprelabeled
tumortargetcells. Serum from BCG-infected mice
nottreated with lipopolysaccharide
orfrom uninfected mice treated with lipopolysaccharide
wasinactive. Moreover, although
serumcytotoxic activity
wasevident with 10
syn-geneic
orallogeneic
tumorcell lines, little
or noeffect
wasobserved with normal embryonic fibroblast
targetcells. Maximal titers of
serumcytotoxic activity
weredetected
14days after BCG infection and 2 h after LPS
treatment.Serum
ofBCG-infected, T-cell-deficient nude mice developed
strongcytotoxic activity after LPS
treatment;however, lipopolysaccharide-insensitive C3H/HeJ mice could produce this cytotoxic activity only after adoptive transfer with lipopolysaccha- ride-responsive C3H/HeN bone
marrow.Physicochemical characterization of the
serumcytotoxic activity revealed
aheat-stable (56°C,
30min) entity with
amolecular weight of about 60,000 and
anisoelectric point
atpH
4.8.Biological and physicochemical characteristics of this
serumcytotoxic activity
asdefined by
anin vitro
assay wereverysimilar
tocharacteristics of
tumornecrosis factor and
suggestthat this molecule
maybe
amajor effector mechanism for the antitumor actions of lipopolysaccharide.
The antitumor effects of bacterial lipopolysac- charides (LPSs) have been the subject of
re-search for almost
acentury(6, 23). Transplant- able syngeneic
tumorsgrowing intradermally
orsubcutaneously
canbe completely cured with
asingle
butsometimes lethal injection of LPS
at adistant site (20). This effect
wasquite dra- matic. Within
6 to 8 hof the LPS injection, erythema developed
atthe
tumorsite. By
24h, extensive hemorrhagic
necrosis wasevident, and by
48 to 72h mostof the tumor masshad beensloughed
(2).Recently, Carswell
etal. discovered that
the tumornecrotic
action ofLPS
was not adirect effect
ofLPS
on tumorcells
but rather wasmediated by
afactor
present in sera of LPS- treated animals(5).
A similar factor could be isolated from sera ofLPS-treated mice and rab- bitswithout tumors butpreviously injected with Mycobacteriumbovis
strainBCG
(10, 14). The tumor necrosisactivity
ofserum
fromLPS-
treated animalscould be separated from many ofthe toxiceffects of LPS itself (10). Moreover,although
tumor cells ofdifferent
originswere inhibited
orkilled by thistumor necrosis factor invitro,
normalembryonic fibroblasts were un- affected under the same conditions (5, 11, 19).tPresent address:LaboratoryofImmunobiology,National CancerInstitute,Bethesda,MD 20205.
Basedon these earlierobservations, wehave
developed
asensitive and quantitative in
vitro assayfor cytotoxic factors in the sera ofLPS-treated animals.
In aprevious communication
weusedthis assaytodiscriminate thecytotoxic factor fromlymphocyte-activating factor which appearsin BCG-infected mice after LPS injec-
tion(13).
In this report, conditions foroptimal release of cytotoxic factors in sera ofBCG-in-
fected mice and physicochemical characteriza- tion ofthe active factorarepresented.
MATERIALS AND METHODS Mice.Female C3H/HeNmice, 6 to12 weeks of age, wereobtained fromthe Division of Research Services, National Institutes of Health. C3H/HeJ mice were purchased fromtheJackson Laboratory, Bar Harbor, Maine.
LPS.Escherichia coli K235 LPS was prepared by thephenol-waterextraction method ofMcIntireet al.
(16)orby the butanol extraction procedure described by Morrison (17). SalmonellaMinnesota Re mutant R595 LPS isolated by phenol-chloroform-petroleum ether method (8) was a generous gift of E. T.Rietschel, MPI, Freiburg, WestGermany. Polysaccharide of S.
Minnesota (Freemantype[7])waskindly given by C.
Bona, NIH, Bethesda, Md. The phenol-water-ex- tractedLPSof E. coli K235 was used for most of the studies.
Preparation ofcytotoxic serum. Mice were in- 204
LPS-INDUCED 205 fected intravenously (i.v.) with 2 x10"colony-forming
units (CFU) of living Mycobacterium bovis strain BCG (Phipps substrain TMC no. 1029, Trudeau Mycobac- terialCollection, SaranacLake, N.Y.), and then were injectedi.v. 14 days later with
10,jg
of LPS. Two hours after LPS injection, the animals wereexsanguinated, andthe serum was prepared (BCG-LPS serum). Con- trol serum was obtained from LPS-injected normal mice in a similar manner. All sera were stored at -20'C until use.Cytotoxicity assay. Tumor cells (mouse L 929, ATCC)at 4x104cells per 16-mm culture well (Costar 24,Cambridge, Mass.) were labeled in 0.5 ml of Eagle minimal essential medium (EMEM) with 25 mM HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesul- fonic acid) buffer, 5% heat-inactivated fetal calf serum (FCS), and0.5
1LCi
of[3H]TdR ([methyl-3H]thymidine, specific activity, 1.9 Ci/mmol; Schwarz/Mann, Or- angeburg,N.Y.)for18 to24h.Tumor cell monolayers were washed twice after labeling and incubated in dilutionsofcytotoxicserumin1ml ofDulbecco MEM containing 10%FCSand 501g
of gentamicin/ml. La- beledtumorcell monolayerslysedwith0.5%sodium dodecyl sulfate in waterwereused to estimate total incorporated countsperminute (cpm).For some ex- periments, the identical procedurewascarriedoutin 0.4-mm flat-bottom culture wells (Costar 96, Cam- bridge, Ma.) with 4X 103cells per well in 0.2 ml of medium. Tumor cytotoxicity was estimated by meas- uring the release of incorporated [3H]TdRfrom the labeled tumorcells in duplicate cultures and expressed as meancpm ±standard deviationor asspecificre- lease according to the formula (experimental cpm- cpmof control)/(cpm of sodium dodecyl sulfate-cpm of control) x 100. Cells were counted in a Coulter Counter (model ZBI, Coulter Electronics, Inc., Hi- aleah, Fla.) inquadruplicate after treating adherent cells with 0.5% trypsin-0.2% ethylenediaminetetraace- tic acid (EDTA) solution and suspending them in Isotone II(CoulterElectronics) containing 10% FCS.Chromatographic procedures.
Samples
weresubjectedtogelfiltrationon aSephacrylS-200 column (2.6by90cm)equilibratedinpH7.5buffercontaining 0.05Mtris(hydroxymethyl)aminomethane (Tris),0.1 M sodium chloride, or 1.6 M sodium chloride. All chromatographywasperformedat4°Cat aflowrate of approximately 12 ml/h. Up to 10 ml of either unfractionated serum ortheammonium sulfate pre- cipitate, dialyzed 4timesagainst pH 7.5Tris buffer adjustedto afinal volume of6ml in thesame
buffer,
wasappliedtothe column.Fractions ofapproximately 3mlwerecollected, and portionswere sterilizedby Millipore filtration before they were tested in the cytotoxicityassay.Upto10ml ofunfractionatedse- rum wasappliedon acolumn(1 by30cm)of
diethyl-
aminoethyl (DEAE)-Sephacel (Pharmacia) equili- brated in0.05 MTris buffer(pH 7.1).The columnwaswashed with4columnvolumes ofstartingbuffer be- foreinitiation ofalinear0to400mM sodium chloride gradient(totalgradientvolume280ml).Theflowrate was approximately 12 ml/h. Fractions (2 ml) were collected, andportionswerepreparedfor thecytotox- icity assay. The sodium chloridecontentof thecolumn fractionswasdetermined withanOsmometer
(model
3L, Advanced Instruments, Inc., NewtonHighlands,
Ma.).A 0.5-ml amount of unfractionated BCG-LPS serum was applied to a 5-mm-thick bed of Sephadex G75 Superfine with 5% Pharmalyte (Pharmacia) (pH 4 to 6.5) and focused for 18 h ataconstant power of 1 W on a flatbed apparatus (LKB 2117 Multiphor). The fractions were eluted with 6 ml of phosphate-buffered saline. After we measuredthe pH of individual frac- tions, we dialyzed the fractions against phosphate- buffered saline for 24 h, sterilized them by filtration, and assayed them for tumor cytotoxicity.
Induction of tumors and in vitro culture of tumor cells. Fibrosarcomas were induced intrader- mally in mice with 1 mg of 3-methylcholanthrene in 0.2ml of trioctanoin (tricaprylin, Sigma Chemical Co., St. Louis, Mo.). For in vitro culture, solid tumor pieces wereminced and a single cell suspension was obtained by enzymatic digestion (3). The cells were cultured in EMEM containing 10% FCS and 50,g of gentamicin perml.
Generation of chimeric mice. Bonemarrowcells were obtained by gently grinding and then rinsing femura andtibiae of donor mice in cold RPMI medium (GIBCOLaboratories, Grand Island, N.Y.). Bone frag- ments wereremoved by being allowed to settle for 5 min; the cells were washed and then suspended in medium to a concentration of
10"
cells/ml. Cell viabil- ity exceeded 90% as determined by exclusion of trypan blue dye. Mice received850roentgensof X-irradiation andwerereconstitutedwithin 6 hbytail vein injection of 107 bonemarrowcells. Three daysbefore and for2 weeks after reconstitution, the mice were given drink- ing water containing1geach ofampicillin (Totacillin- N, Beecham Laboratories, Bristol, Tenn.), carbenicil- lin (Pyophen, Beecham Laboratories), and cephalo- thin (Keflin, Eli Lilly & Co., Indianapolis, Ind.) per liter. Four to 6 weeks after reconstitution, the mice were used in the experiments.RESULTS
In vitro detection of cytotoxic factors in
serumof BCG-infected mice treated with LPS. Serum of BCG-infected mice treated with LPS (BCG-LPS serum) contained factors
cyto-toxic
for tumorcells
invitro (Table 1). Cytotox- icity
wasdetected with each of
10different
mu-rine fibrosarcomas after
treatment with1/100
dilution of BCG-LPS serum. Thesamedilutionof
serum had little or no effect onsyngeneic embryo fibroblasts.
It isimportant
to notethatfibrosarcomas from lipid A-unresponsive C3H/
HeJ mice
were assusceptible
tothe
activeserum as fibrosarcomas fromnormally responsive C3H/HeN
mice. That therewas nodifference in susceptibility between tumors ofLPS-respon-
sive and-unresponsive
strains suggests that a directeffect
ofLPS
upon target cells wasun-likely. Among
the tumorsassayed,
theC3H/
HeN fibrosarcoma L929 was themostsensitive target cell andwastherefore usedin all further studies. Serum from BCG-infected mice not
treated
i.v.with LPS,
orfrom normal micewith or without LPS treatment, had no effect on eithernormal
or tumorcells. LPSin mediumup 28,to aconcentrationof 100
tug/ml
was alsowithout effect (datanotshown).
Release of radiolabel from
[3H]TdR-prela-
beled L 929 cells into culturesupernatants
at48 hshowedaninverse correlation with the number of intact cells estimatedby
direct cell counts (Fig. 1).Significant cytotoxicity
to L 929cellsbyBCG-LPS/sera
could be detected with aslittle as a 1/1,000 dilution.Cytotoxic activity,
how- ever, even at the higher serumconcentration,
was dependent on continuous presence of the cytotoxic factor(Fig. 2). Target
cells incubated with a 1/100dilution of active serum for upto8TABLE 1. Tumorcytotoxicactivity ofBCG-LPS serum and normalmouseseruma
BCG-LPS
Normal
Targetcell
serumh
mouseserum
C3H/HeN fibrosarcoma
L929 55 ± 2 0±1
TumorA 19±1 1± 0
TumorB 19±0 0±1
Tumor C 13± 1 0±0
Tumor D 12 ±1 0 ± 0
C3H/HeJfibrosarcoma
Tumor A 20±2 0±1
Tumor B 17± 1 1±0
Tumor C 12 ± 1 2±1
Tumor D 10±1 0± 1
TumorE 10±0 1±0
Embryo fibroblasts 3 ±0 0 ± 0 a Each value represents percent
specific
label re- lease±standard deviation.'BCG-LPS
serumand normalmouse serum were diluted 1/100.C-
x
0 0
0 D
X.CU
15 >
0
10 m,
cn x
S
1
10-5 10-4 10-3 10-2 SERUM DILUTION
FIG. 1. Tumorcytotoxicity doseresponseofBCG- LPSserum and normalmouse serum. [3H]TdR-la- beled or unlabeled L 929 cells were incubated in differentdilutionsofBCG-LPSserum( )ornor-
malmouse serum(- )for48h. Tumorcytotox-
icitywasestimatedby[3H]TdRrelease(total incor- poratedcpm=45x 103)(0) and by countingintact cellsonaCoulter Counter(A). Morethan 95% of the recoveredcellswereviableasdeterminedbytrypan
bluedyeexclusion.
h and then
washed, released little
or no radio-label
at48 h.Even
a24-h pulse
withcytotoxic factor
wasless efficient than continuous
pres-ence
throughout
theassay.Optimal conditions for the in vivo
pro-duction
ofcytotoxic factors
in serum ofBCG-infected
micetreated
withLPS. Mice
weretreatedi.v. with
LPS
atvarioustimes
afterBCG infection (Fig. 3). Tumor cytotoxic factors
were
detected in the
serum ofBCG-LPS mice by
1 weekbut not 3days
afterBCG infection.
Maximal production
wasevident
at2 weeks andthen progressively declined
toabout
controllev- elsby
8weeks.The time
coursefor
appearance ofcytotoxic factors in
serum ofBCG-LPS mice is shown in
Fig.
4.Cytotoxic activity
wasdetected
30min
10_
lo /
it 8 /
0 0
0
23.a
5 15 25 35 45
HOURS OFCULTURE
FIG. 2. Tumor cytotoxicity after exposuretoLPS- BCGserum ornormal mouse serumforvarioustimes.
[3H]TdR-labeled
L929 cells wereincubatedwith a 1/100 dilution of BCG-LPSserum ( ) ornormal mouseserum(----) for0 to 48h.The supernatant wasreplaced by freshmediumwithout mouse serum, andcytotoxicitywasestimated by[3H]TdR
release at48h(totalincorporated cpm=40x103).
o^1
0'-o-b x
0-
I.-
__- _4._ _ _.4- _ _ _ _ -._
I I ~~~I I
1 2 4 8
WEEKS OF BCG INFECTION
FIG. 3. Kineticsof the appearance of cytotoxicse- rumactivity during BCG infection. Mice were infected i.v. with 2 x106 CFU ofBCGandinjectedi.v.with10 pgofLPSafter3daysandafter1, 2, 4,and 8weeks.
Twohoursafter LPS injection, the animals were bled andserum wasprepared.TumorcytotoxicityofBCG- LPS serum ( ) and BCG serum ( - - - ) was estimated by[3H]TdRrelease of labeled L 929cells at48 h(total incorporated cpm=60 x 103).
INFECT. IMMUN.
LPS-INDUCED TUMORICIDAL SERUM FACTOR 207 after LPS injection, reachedmaximal levels by
1 to 2 h, and then progressively declined and
wasabsent by 6 h. A second injection ofLPS6 h after the first injection failed toinducereap- pearanceoftumorcytotoxic activity.
Cytotoxic activity inserumof
BCG-LPS
micewasdependentupon asufficientdose of phenol- extracted LPS. Maximal activityoccurredafter 1
,ug
of LPS (Fig.5).Cytotoxicactivity could be detected inBCG-injected
mice aftertreatmentwith
polysaccharide-deficient
LPS of S.minne- sota R 595 (Re mutant), suggesting that the active principle of LPS was lipid A (Table 2).This is further confirmed by the fact that the polysaccharide partof LPS (Freemantype[7])
or asecond antigenic stimulus suchaspurified proteinderivative (PPD) wasunable toinduce
(vtotoxic
serum activity in theBCG-infected
°x_ 20 _
2 4 6 8
HOURSAFTERLPS CHALLENGE FIG. 4. Kineticsoftheappearanceof cytotoxicse- rum activity after LPS injection of BCG-infected mice. Mice were infected i.v. with 2 x 106 CFUof BCG andinjectedi.v. with IOAg ofLPS2 weeks later.
After10and 30minutes andafter 1, 2, 4, 6, and 8h, animalswerebled andserumwasprepared. Tumor cytotoxicitywith 1/100dilutionoftheserawasesti- matedby[3H]TdRreleaseoflabeled L 929 cells at 48 h(total incorporatedcpm=50xi03).
mice. In addition, lipid
A-unresponsiveC3H/
HeJ mice failed
toproduce serum cytotoxic fac- torsafter the BCG-LPS
treatment.C3H/HeJ mice treated with BCG and butanol-extracted LPS also failed
toproduce serum cytotoxic ac- tivity. Butanol-extracted LPS is a strong B cellmitogen
evenfor the lipidA-unresponsiveC3H/
HeJ mouse (9).
C3H/HeJ
mice could produce cytotoxic fac- torsonly after
adoptive transfer of lipid A-re- sponsiveC3H/HeN
bonemarrow (Table 3). Au-tologous
transfer of bone marrow cells into ir- radiatedC3H/HeJ
or C3H/HeN recipient micedid
notalter their
responsiveness to LPS. How- ever,irradiated C3H/HeN
recipients treated10
'CM
0 0
0
.001 .01 .1 1 10 100
LPS(pg)
FIG. 5. LPS dose responsefor induction of cyto- toxicserumfactor. Micewere infectedi.v. with2 x 106 CFUofBCG and injected i.v. with0.001 to 100 pgof LPS2weeks later.After2h, animalswerebled andserum wasprepared. Tumorcytotoxicitywitha 1/100dilutionoftheserawasestimatedby[3H]TdR releaseoflabeled L929cellsat48h(total incorpo- ratedcpm=35x103).
TABLE 2. Tumorcytotoxicactivity inserumofuntreatedorBCG-infectedmicechallengedwithdifferent stimuli
C3H/HeN C3H/HeJ
Challenge'
Untreated BCGinfected Untreated BCGinfected
Ph-LPS 1,500±150 28,300±50 2,050±50 1,850±100
Bu-LPS 1,550±50 26,450±900 1,600± 100 1,550± 150
R 595-LPS 1,500±50 25,700+ 850 1,450 ±50 1,850±300
PS 2,050±150 1,950±200 1,400±500 1,500±50
BCG 1,850±150 1,600±50 1,650±50 1,950± 100
PPD 1,650±300 1,700+ 50 1,600±50 1,550±100
aEach value represents cpm±standard deviation.
bC3H/HeNorC3H/HeJmice infected with2x
106
CFU of BCG for2weekswerechallenged intraperitoneally
with 50ugofphenol-extractedE.coliK 235LPS(Ph-LPS),50Agof butanol-extracted E. coli K 235LPS (Bu- LPS), 10pgofphenol-extractedS.minnesotaR 595LPS,50jgofpolysaccharidefrom S.minnesota(PS), 2 x 106 CFUofBCG,or100
jg
ofpurified proteinderivative (PPD).After 2h,animalswerebled andsera were prepared.Tumorcytotoxicitywasestimatedby[3H]TdR
release of labeled L 929 cells after 48 h of culture in a1/100 dilutionof the sera(totalincorporatedcpm=50X103).VOL. 28, 1980
208
with
nonresponsive C3H/HeJ bone
marrowfailed
toproduce cytotoxic
serumactivity after LPS challenge. Although the capacity of mice
torespond
tolipid
A wasessential for
tumorcyto-toxic activity in the
serum,there
was noappar- entneed for
Tcell participation. Serum of BCG- injected mice homozygous for the
nugene(nude mice) developed
strong tumorcytotoxic activity after intravenous LPS
treatment(Table 4). Al- though production of
tumorcytotoxic activity in the
serumof LPS-treated nude mice is
atvari-
ancewith
anearlier observation by Old (18),
a recent reportby Berendt
etal. showed
strongLPS-induced tumor-necrotizing effects in T cell- deficient mice (4).
Biochemical characterization of the
tu- morcytotoxic activity from
serum.Tumor cytotoxic activity
was asfollows.
(i)Cytotoxic activity
wasstable
at570C for
morethan
1h, and
nosignificant loss in activity
wasobserved when samples
werelyophilized
orstored in
so-lution
at40C for several weeks.
TABLE 3. Demonstrationoftumorcytotoxicfactor in serumafter LPStreatmentofBCG-infectedmice No.of mice withcytotoxic
Donor Recipient' serumfactor
per no.of micetested
C3H/HeN
C3H/HeN.
13/14C3H/HeJ C3H/HeN. 0/18
C3H/HeJ C3H/HeJ. 0/15
C3H/HeN C3H/HeJ. 6/8
aRecipient animals received 850 roentgens of X- irradiation before reconstitution with107bonemarrow cells. Six weeks afterreconstitution, animalswerein- fected i.v. with 2x106CFU ofBCG, and2weekslater theywereinjected i.v.with 10
,jg
ofLPS. After2h, micewerebled,and tumorcytotoxicity was estimated by[3H]TdRrelease of labeled L 929 cells after 48 h of culture ina1/100dilution of the sera.TABLE 4. Tumor
cytotoxic
activity of serafrom BCG-infectedC3H/HeNand nude micechallengedwith LPS'
Challenge(Mg C3H/HeNb C3H/HeN nu/nu ofLPS)
0 3,700± 100 2,900±350
0.5 11,200±550 15,700± 50 5 23,800±850 29,800±950 50 29,800±700 28,500+950 nEach value represents cpm±standard
deviation:
bC3H/HeN orC3H/HeN nu/nu mice infected
i.v.
with 2 x 106CFU ofBCG for 2 weeks were injected i.v.with0,0.5, 5, and50,ug of LPS and bled 2 h later.
Tumor cytotoxic activity was estimated by[3H]TdR releaseoflabeled L 929 cells after 48 h of culture in a 1/100 dilution of the sera (total incorporated cpm = 56x 103).
(ii)
BCG-LPS and control sera wereapplied
toSephacryl
S-200 columns and eluted with Tris buffercontaining
0.1 M sodium chloride. The elutionprofile
showed onemajor peak; cytotoxic activity
was found at an apparent molecularweight
of150,000.
Control serumfractions hadlittle
or nocytotoxic activity.
Most of thecyto-
toxicactivity
inserum wasfound in theprecip- itate ofa40to60% saturated ammoniumsulfate solution. The ammonium sulfate precipitatewaschromatographed
asdescribed above(Fig. 6A).
Again,
majorcytotoxicactivitywasfound intheimmunoglobulin
G range; however, a second peak withanapparentmolecularweight ofabout 55,000to60,000wasalso detected. FractionsofLPS-induced
control serum were inactive. Re-chromatography
ofthesameammoniumsulfate-x 0
0
E
0
I-
18 16 14 12 10 8 6 I4 2
8 7 6 5 4 3 2
(A
Cox
U.
A
I
I I I
160 200 240 20 320 360 400
ELUTION VOLUME(ml)
FIG. 6. Sephacryl S-200 chromatography of am- monium sulfate-precipitated cytotoxic activity from BCG-LPS serum. Materialprecipitated with 40 to 60%ammonium sulfatesaturation from 2 ml of BCG- LPS serum was subjected to Sephacryl S-200 col- umns. Thecolumns were 2.6 by 90 cm, and theflow ratewasapproximately 12 ml/h. The columns were equilibrated and eluted with 0.05 M Tris buffer (pH 7.5) containing0.1Msodium chloride (A) or 1.6 M sodium chloride (B). Tumor cytotoxicity of a 1/50 dilution ofthe3-ml sterile filtered fractions was es- timated by[3HJTdRrelease of labeled L 929 cells at 48h.Totalincorporated cpm=35x102 (A), 30x102
(B).
INFECT. IMMUN.
I
LPS-INDUCED TUMORICIDAL SERUM FACTOR
209 precipitated material
on anidentical column but
with
abuffer of higher sodium chloride
content (1.6M) shifted the elution of cytotoxic activity
to asingle peak
at55,000 to60,000(Fig. 6B). In
aprevious study,
wereported these dissociating conditions
in moredetail, and the
effect hasbeen used
to separatelymphocyte-activating factor from cytotoxic activity (13).
(iii) Tumor cytotoxic serum or
control
serum wasapplied
todiethylaminoethyl-Sephacel col-
umnsand washed with Tris buffer (pH 7.1). A gradient
wasstarted from
0 to 0.4M sodium chloride. Cytotoxic activity in BCG-LPS
serumeluted from the column
ata0.17 to0.2M sodium chloride concentration (Fig. 7).
Nocytotoxic
ac-tivity
wasdetected in LPS-induced control
se- rumfractions. Pooled fractions containing cyto-
toxicfactor
werethen applied
toaSephacryl S-
200column after extensive dialysis and lyophi-
lizationand chromatographed in
Trisbuffer
con-taining
0.1Msodium chloride. Cytotoxic activity
I
8X-
x, 6
-4 x 0C-
X 4 0 02
°a: 2
30 40 50 60
FRACTION NUMBER
z 0.2
Il
FIG. 7. DEAE-SephacelchromatographyofBCG- LPSserum.A2-mlamountofunfractionatedserum wasappliedon a column (1.5 by30 cm) ofDEAE- Sephacelequilibratedin0.5MTrisbuffer (pH 7.5).
The column waswashed with 4column volumesof starting buffer before initiation ofa 0 to 400 mM sodiumchloridegradient(totalgradientvolumewas 280 ml). Theflow rate wasapproximately 12ml/h.
Fractions (2 mleach) werecollected and sterilefil- tered. Tumor cytotoxicity ofa 1/10 dilution ofthe fractions was estimatedby[3HJTdR release ofla- beledL 929cellsat48h(totalincorporatedcpm=50 X103). Themolarityofthefractionswasdetermined withanosmometer.
iF
0x
6
eluted
as asingle
peak at an apparent molecularweight of
55,000 to60,000.(iv) A
0.5-ml
amount ofunfractionated BCG- LPS
serum wassubjected to flatbed electrofo-cusing in
apH gradient from 4 to 6.5. Cytotoxic activitywas recovered from the fraction eluates in asingle
peak together with serum albumin atabout pH
4.8(Fig.
8);yield
was lessthan 5% ofthat applied, explanable in
part by theinstabilityof activity below
pH 6(data
notshown).
DISCUSSION
Discovery
ofsoluble
mediators in sera ofBCG-infected mice treated
with microgramquantities of LPS that reproduced
tumor ne-crotic actions of milligram quantities
ofLPS
innoninfected animals
was animportant
advancein the analysis of the antitumor
properties ofbacterial endotoxin
(5).Characterization of these soluble
mediators(tumor necrosis factors) promised the
opportu-nity
of separating, for thefirst
time, the thera-peutic action of LPS from the toxic and often lethal side effects. Fulfillment of
this promise,however,
wasimpeded by
adifficult
and impre-cise in vivo
assay.Serum-induced necrosis of
anintradermal
tu- mortransplant required large volumes of active
serum(0.5 ml
permouse), and the
extentof the
tumornecrosis endpoint
wasdependent
uponlocation, size, and vascularization of the
tumor(18).
We have described
inthis
report a sensitive andquantitative in vitro
assay todetect
cyto-toxic mediators in
seraof LPS-treated, BCG-
15 2D a
FRACTKONNWUIER
12 1.0 o0 8
I .6
OA
02
FIG. 8. Electrofocusingof BCG-LPSserum.A 0.5- ml amount ofunfractionated BCG-LPS serum was
focusedonaflatbed ofUltrodex with5%Pharmalyte (pH4 to6.5)for 18h. Fractions wereremoved and eluted with 6 ml ofphosphate-buffered saline and dialyzed,and thetumorcytotoxicity ofa1/20 dilution of thefractionswasestimatedby
['H]TdR
releaseof
labeledL 929 cellsat48h(totalincorporated
cpm=35x102).
I I
VOL. 28, 1980
MANNEL, MELTZER,
AND MERGENHAGEN infectedmice;cytotoxic activity
in serum canbereproducibly
detectedthrough
a1/1,000
dilu- tion, and the cytotoxicendpoint
is nowdefined by release of [3H]TdR from prelabeled target cells instead ofsubjective grading
for tumor necrosis.Properties of cytotoxic
factors inserumby
the invitro
assay were very similartothose of tumornecrosis factor
asdefinedby
the in vivoassay.(i) The
time
courseforoptimal production
of tumor necrosis factor and forcytotoxic
serumactivity (intervals between BCG
infection and LPS treatmentand between
LPStreatmentand serumcollection)
wasidentical(5).(ii) Production of
tumor necrosisfactor andcytotoxic
serumactivity
were bothdependent
uponnormal LPS responsiveness;
LPS-insensi-tive C3H/HeJ mice
orX-irradiated
LPS-sensi- tiveC3H/HeN chimeric
micereconstituted
withC3H/HeJ bone
marrowfailed
toproduce
eitheractivity
after LPSchallenge (13a).
(iii) The in vivo
tumornecrotic action
of LPS wasevident
inmicehomozygous
forthe nugene(data
notshown);
serum fromLPS-treated, BCG-infected nude mice
wasalso cytotoxic
to tumortarget cells
invitro.(iv) Physicochemical characterization
of mouse tumornecrosis factor reveals
aheat-sta-ble activity with
anapparent molecularweight
of 150,000(10).
A recent report of a similar tumornecrotic factor of rabbit origin, however, described activity
with an apparent molecularweight of
about 50,000(14).
Thisconflict
in apparentmolecular weight
ofmouseand rabbit tumor necrosisfactor(s) could be explained by species differences. Our results, however,
suggest analternative explanation;
mousetumornecro- sis factor may exist as an aggregate in serum.Cytotoxic activity
of activeserum,
aspreviously
describedby Carswell
etal. (10), eluted from Sephacryl
S-200 in the125,000 to 150,000molec-
ularweight region.
Ammoniumsulfate precipi-
tation of this same serumfollowed by gel filtra-
tion inhigh
ionicstrength buffer
of theprecipi-
tate(conditions favoring separation
of aggre-gates) led
toelution of
cytotoxicactivity
at 55,000 to60,000 daltons.Several
investigators
havesuggested that the cellular source of tumornecrosis
factor may be themacrophage.
Indeed, soluble cytotoxic fac- tors have been isolated frommacrophage
cul- turesafter avariety
of invitro treatments (1,
12, 21,24). Matthewsdescribed
asoluble cytotoxic factor from rabbitmonocytes
whichresembled
the rabbitserum-derived
tumor necrosis factor (15).Preliminary
results from ourlaboratory confirm
and extend thisobservation.
Macro- phages fromBCG-infected
micetreated in vitrowith
LPS release solublecytotoxic
factors within 2 h oftreatment.Physicochemical characteris-
tics ofthismacrophage-derived cytotoxin
were very similar to those of the serum factor de- scribed in this report.Cytotoxic
factorswerenot detected in fluids ofLPS-treated
control mac-rophage
cultures. These results suggest that theLPS-induced
release of thiscytotoxin
mayde-pend
uponthe level ofmacrophage activation.Recent observations by Russel
and co-workers indicate thatmacrophages
withingrowing
tu- mors are eitheractivated
or can be activated with very small quantities of LPS (22). The tumornecroticaction of LPS could certainly be mediatedby
soluble factors released by theseintratumor-activated macrophages.
LITERATURE CITED
1. Aksamit,R.R.,and K. J.Kim. 1979.Macrophagecell linesproduceacytotoxin. J. Immunol. 122:1785-1790 2. Andervont, H.B. 1936. The reaction of mice and of variousmousetumors totheinjectionofbacterialprod- ucts.Am.J. Cancer27:77-81
3. Bartlett,G.L., B.Zbar,and H.J. Rapp. 1972.Sup- pressionofmurine tumorgrowth byimmunereaction tothe Bacillus Calmette-Guerinstrain ofMycobacte- riumbovis. J. Natl. Cancer Inst. 48:245-257
4. Berendt,M.J., R. J. North, and D. P. Kirstein. 1978.
Theimmunologicalbasis ofendotoxin-induced tumor regression.J.Exp. Med. 148:1550-1559
5. Carswell,E.A.,L.J. Old, R.L.Kassel,S.Green,N.
Fiore, and B. Williamson. 1975. An endotoxin-in- duced serum factor that causesnecrosisof tumors. Proc.
Natl. Acad. Sci. U.S.A.72:3666-3670
6. Coley, W. B. 1894.Treatmentofinoperable malignant tumorswith toxins oferysipelas and Bacillus prodigio- sus.Am. J.Med. Sci.108:50-66
7. Freeman, G. G.1942.The preparation andproperties of an specificpolysaccharide from Bact. typhosum Ty2.
Biochem. J.36:340-356
8. Galanos, C., 0. Luderitz, and0. Westphal. 1969. A newmethodfor theextractionof Rlipopolysaccharides.
Eur. J.Biochem. 9:246-249
9. Goodman,M.G.,D.E.Parks, and W.0.Weigle. 1978.
Immunologic responsiveness ofthe C3H/HeJmouse:
differential abilityofbutanol-extracted lipopolysaccha- ride(LPS)toevoke LPS-mediated effects.J.Exp.Med.
147:800-812
10. Green, S.,A.Dobrjansky,E. A.Carswell,R.L.Kas- sel, L. J.Old,N.Fiore, andM. K.Schwartz. 1976.
Partial purification ofaserum factorthat causes necro- sisof tumors.Proc. Natl. Acad. Sci. U.S.A., 73:381-385 11. Helson, L., S. Green,E.Carswell, andL. J.Old.1975.
Effect oftumor necrosis factoroncultured humanmel- anomacells.Nature (London),258:731-732
12. Kramer, J.J.,andG.A.Granger.1972. Theinvitro inductionandreleaseof acell toxinbyimmuneC57B1/
6mouseperitonealmacrophages. Cell. Immunol.3:88- 100
13. Mannel,D.W., J. J.Farrar,andS. E.Mergenhagen.
1980.Separationofaserum-derived tumoreidalfactor from ahelperfactor forplaque-forming cells.J.Immu- nol.124:1106-1110.
13a.Mannel,D. N., D.L.Rosenstreich,andS.E.Mergen- hagen. 1979. The mechanism of lipopolysaccharide inducedtumornecrosis:requirementforLPS-sensitive lymphoreticular cells. Infect. Immun. 24:573-576 14. Matthews, N., andJ. F.Watkins. 1978.Tumour-necro-
sis factorfromthe rabbit.I.Modeofaction,specificity andphysicochemical properties.Br.J.Cancer38:302- 309
15. Matthews, N. 1978. Tumour-necrosis factor from the rabbit. II.Productionbymonocytes.Br.J.Cancer 38:
310-315
16. McIntire,F. C., H.W.Sievert,G. H. Barlow, R. A.
Finley,and A. Y. Lee. 1967. Chemical, physical, and biologicalproperties ofalipopolysaccharidefrom Esch- erichiacoli K-235.Biochemistry6:2363-2372 17. Morrison,D.C.,S.J.Betz,and D. M.Jacobs.1976.
Isolation ofalipidAboundpolypeptideresponsiblefor
"LPS-mediated" mitogenesisofC3H/HeJ spleen cells.
J.Exp. Med.144:840-846
18. Old,L.J.1976.Tumor necrosis factor. Clin. Bull. 6:118- 120
19. Ostrove,J.M.,and G. E.Gifford. 1979. Stimulation of RNAsynthesisinL-929 cellsbyrabbittumornecrosis
factor. Proc. Soc.Exp. Biol. Med. 160:354-358 20. Parr, I., E.Wheeler, and P. Alexander.1973.Similar-
ities of theanti-tumour actions ofendotoxin, lipid A and double-stranded RNA. Br. J. Cancer 27:370-389 21. Reed, W. P., and Z. J. Lucas.1975.Cytotoxic activityof lymphocytes. V. Role of solubletoxin inmacrophage inhibited cultures oftumorcells.J. Immunol. 115:395- 404
22. Russel, S. W., W.F. Doe, and A. T. McIntosh.1977.
Functional characterization ofa stable, noncytolytic stageofmacrophageactivationintumors.J. Exp.Med.
146:1511-1646
23. Shear, M. J. 1944. Chemical treatmentoftumors. IX.
Reactions of micewithprimarysubcutaneoustumors toinjectionofahemorrhage-producing bacterial poly- saccharide.J. Natl. Cancer Inst.4:461-476
24. Trivers, G.,D.Braungart,and E. J. Leonard. 1976.
Mouselymphotoxin.J. Immunol.117:130-135