Proc. Nadl. Acad. Sci. USA
Vol. 89, pp. 4845-4849, June 1992
Tumor necrosis factor soluble receptors circulate during
experimental and clinical inflammation and can protect
against excessive tumor necrosis factor a in vitro and in vivo
(cytokine inhibitors/endotoxemia/septic shock/humans/primates)
KIMBERLY J. VAN ZEE*, TADAHIKO KOHNOt, EVA FISCHER*, CRAIG S. ROCK*, LYLE L. MOLDAWER*,
AND STEPHEN F. LOWRY*t
*The Laboratory of Surgical Metabolism, New York Hospital-Cornell University Medical Center, 525 East 68th Street, F-2016, New York, NY 10021; and
tSynergen, Inc., 1885 33rd Street, Boulder, CO 80301
Communicated by Igor Tamm, January 24, 1992 (receivedfor review November 11, 1991)
mediator of systemic responses to sepsis and infection, can be
injurious to the organism when present in excessive quantities.
Here we report that two types of naturally occurring soluble
TNF receptors (sTNFR-I and sTNFR-II) circulate in human
experimental endotoxemia and in critically ill patients and
demonstrate that they neutralize TNFa-induced cytotoxicity
and immunoreactivity in vitro. Utilizing immunoassays that
discriminate between total sTNFR-I and sTNFR-I not bound to
TNFa, we show that sTNFR-I-TNFa complexes may circulate
even in the absence ofdetectable free TNFa. To investigate the
therapeutic possibilities of sTNFR-I, recombinant protein was
administered to nonhuman primates with lethal bacteremia
and found to attenuate hemodynamic collapse and cytokine
induction. We conclude that soluble receptors for TNFa are
inducible in inflammation and circulate at levels sufficient to
block the in vitro cytotoxicity associated with TNFa levels
observed in nonlethal infection. Administration of sTNFR-I
can prevent the adverse pathologic sequelae caused by the
exaggerated TNFa production observed in lethal sepsis.
Tumor necrosis factor a (TNFa), a primary
Tumor necrosis factor a (TNFa) is widely appreciated as a
principal mediator ofsystemic responses to sepsis and injury.
Produced by inflammatory cells in response to diverse in-
fectious stimuli and tissue injury, TNFa induces a cascade of
endogenous mediators that direct host immunologic func-
tions (1). While TNFa may thus serve as an essential element
in host defense, the excessive tissue production ofTNFa can
mediate detrimental systemic effects by acutely precipitating
a syndrome similar to that of septic shock (2), and lesser
degrees of chronic TNFa production appear to induce ano-
rexia and cachexia (3, 4). Thus, pathologic conditions may
result from the excessive production and activity of TNFa.
Naturally occurring inhibitors ofTNFa activity have been
identified in human urine and serum and in cell-culture sys-
tems (5-10). The isolation and characterization of these in-
hibitors have revealed at least two distinct species, an '=30-
kDa protein with the NH2-terminal sequence Asp-Ser-Val-
Cys-Pro-Gln and an -40-kDa protein with the NH2-terminal
sequence Leu-Pro-Ala-Gln-Val-Ala (7-11). These two pro-
teins are the extracellular domains of the TNF receptors
(TNFRs) types I and II (TNFR-I and TNFR-II), respectively
(8, 9, 12-15), and apparently are shed from the cell surface in
response to many of the same inflammatory stimuli that are
known to induce TNFa production (16). The shedding ofsuch
receptors and resultant acute decrease in the number of
TNFRs on the cell surface may serve to transiently desensitize
cells, thereby providing a mechanism for inhibition ofTNFa
activity. This process may have additional significance in vivo,
as released soluble receptors may inhibitTNFa bioactivity by
binding to the molecule and preventing ligand binding to the
cellular TNFR. Hence, the appearance of such extracellular
soluble receptors may provide a regulatory mechanism for
modulation of excessive TNFa activity arising in response to
severe injury or infection.
MATERIALS AND METHODS
TNFa Immunoactiit. TNFa immune activity was deter-
mined by a sandwich ELISA utilizing a mouse anti-human
monoclonal antibody [provided by P. Tekamp-Olsen (Chi-
ron) and identified as 18.1.1] as the capture protein and a
rabbit anti-humanTNFa antiserum (rabbit no. A5293) (23). A
standard curve was generated with recombinant human
TNFa (human rTNFa; Synergen). The sensitivity of the
assay is 34-100 pg/ml.
TNFa Cytotoxicity. Cytotoxicity was assessed by using the
WEHI 164 clone 13 fibroblast bioassay (17). Human rTNFa
in normal human plasma was used as a standard. The
sensitivity of the assay is 15-30 pg/ml.
To confirm that the cytotoxicity observed was specifically
due to TNFa, the bioassay was repeated with neutralizing
antibodies against human rTNFa. Samples from endotox-
emic volunteers and bacteremic baboons that were found to
have cytotoxic activity were incubated for 30 min at room
temperature with and without antibodies raised against hu-
man rTNFa. These samples were reassayed, and those
incubated with the antibody were found to have no measur-
Other Cytokine Assays. Interleukin-lp (IL-1ip) levels were
assayed by ELISA as described (18). The sensitivity of this
assay is 30 pg/ml. A B.9 hybridoma proliferation assay was
used to determine IL-6 levels, with the number of units/ml
being defined as the reciprocal ofthe sample dilution required
to produce half-maximal proliferation (lower limit of detec-
tion, 2 B.9 units/ml) (19).
Recombinant Soluble TNFRs (rsTNFRs). Two fragments of
human TNFR-I and TNFR-II have been purified, their
cDNAs cloned, and the encoded products characterized (9,
20). These soluble receptors were expressed in Escherichia
coli and chromatographically purified to homogeneity. Both
forms are nonglycosylated and competitively inhibit 1251_
labeledTNFa binding to cell-surface receptors on U-937 cells
Abbreviations: TNF, tumor necrosis factor; rTNFa, recombinant
TNFa; TNFR, TNF receptor; sTNFR-I and sTNFR-II, soluble
TNFR types I and II; rsTNFR-I and rsTNFR-II, recombinant human
sTNFR types I and II; NYH-CUMC, New York Hospital-Cornell
University Medical Center; IL-1p and IL-6, interleukins 1,B and 6;
APACHE, acute physiology and chronic health evaluation.
tTo whom reprint requests should be addressed.
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked "advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Medical Sciences: Van Zee et al.
with the same binding capacity as naturally occurring inhib-
itors (9, 20).
sTNFR Assays. Concentrations of sTNFR-I and sTNFR-II
were determined by ELISA. To determine the total concen-
tration of sTNFR-I or sTNFR-II, an affinity-purified poly-
clonal antibody was utilized as the capture protein. To
determine the quantity of sTNFR-I not bound to TNF, an
ELISA with human rTNFa as the capture protein was used.
This latter ELISA takes advantage of the fact that sTNFR-
I-TNF complexes do not bind to TNFa-coated polyvinyl
chloride plates, whereas free, unbound sTNFR-I does.
Rabbit polyclonal antisera raised against human rsTNFR-I
(rabbit no. 8081) and rsTNFR-II (rabbit no. 2138) were purified
by affinity chromatography using Affi-Gel 10 (Bio-Rad) col-
umns coupled with rsTNFR-I or rsTNFR-II, respectively.
Aliquots of the affinity-purified antibodies were biotinylated
with NHS-LC-biotin (Pierce) according to the manufacturer's
specification and were used as the second antibody in the
ELISAs. The concentration ofsTNFR was calculated accord-
ing to a standard curve generated with human rsTNFR-I or
rsTNFR-II. The sensitivity of the assays is 0.2 ng/ml.
In Vitro Neutralization of Endogenous TNFa by Soluble
Receptors. rsTNFR-I or rsTNFR-II was added to plasma
obtained from volunteers 1.5 hr after administration of 20
units of endotoxin per kg of body weight (see "Human
Studies" below). Endogenous sTNFR-I and sTNFR-II levels
(determined by ELISA) were less than 5 ng/ml. Plasma
samples from different individuals were allowed to incubate
at room temperature for 30 min with increasing quantities of
soluble receptors (5-50 ,ug/ml) and then assayed in triplicate
for TNFa cytotoxicity by using the WEHI 164 clone 13
fibroblast bioassay. A standard curve in normal human
plasma was generated with human rTNFa and used to equate
cytotoxicity of the samples with concentrations of TNFa.
In Vitro Neutralization ofHuman rTNFa by Soluble Recep-
tors. rsTNFR-I and rsTNFR-II (5 ng/ml to 5 ,ug/ml) were
added to normal human plasma to which 1.5 ng of human
rTNFa had been added per ml. The mixtures were allowed to
incubate at room temperature for 30 min and then assayed for
TNFa cytotoxicity by WEHI bioassay and TNFa immuno-
reactivity by ELISA. All samples were assayed in triplicate
and compared with those of the control samples (normal
human plasma with 1.5 ng of added human rTNFa per ml).
Human Studies. Adult male volunteers were admitted to
the Clinical Research Center ofNew York Hospital-Cornell
University Medical Center(NYH-CUMC) after screening by
history, physical examination, and biochemical and hemato-
logical profiles to exclude those with preexisting disease.
Informed written consent was obtained under guidelines
approved by the Institutional Review Board ofNYH-CUMC.
At t = 0, 20 units of national reference endotoxin (E. coli
0113, lot EC-5; Bureau of Biologics, Food and Drug Admin-
istration, Bethesda, MD) per ml was administered i.v. Hep-
arinized arterial blood samples were obtained prior to endo-
toxin administration (baseline) and at intervals thereafter.
Afterobtaining informed consentfrom the patient orfamily
members, and following guidelines of the Institutional Re-
view Board of NYH-CUMC, we obtained blood samples
from 12 critically ill patients repeatedly during their clinical
course (total number of samples = 56). The patients had
sustained a major traumatic injury or undergone a major
surgical procedure, and each experienced episodes of hem-
orrhagic shock and/or sepsis. To quantify the severity of
illness in these patients, a revised acute physiology and
chronic health evaluation (APACHE II) score (21) was cal-
culated for each patient. The APACHE II scores in the 12
patients rangedfrom 11(5% predicted mortality) to 31 (70%o
predicted mortality) with a mean score of 19 ± 2 (mean +
SEM, 20%predicted mortality).
In Vivo Adinistration of sTNFR-I. The experimental pro-
tocol was approved by the Institutional Animal Care and Use
Committee at NYH-CUMC, and is similar to that described
(22). After an overnight fast, six male or female Papio anubis
baboons were anesthetized with ketamine (10 mg/kg, i.m.),
and anesthesia was maintained thereafter by administration of
sodium pentobarbital (5 mg/kg per hr, i.v.). All animals were
then administered 1011 live E. coli 086:B7 as an i.v. infusion
over 30 min. Animals were studied in pairs (1 treatment, 1
control) to negate any variation in the E. coli preparation.
Beginning simultaneously with the E. coli infusion, treatment
animals (n = 3) received a 3-hr primed continuous i.v. infusion
ofsTNFR-I (3 mg/kg priming dose, followed by 4.5 mg/kg per
hr of rsTNFR-I). Control animals received an isovolemic
infusion of human serum albumin (3 mg/kg priming dose
followed by 4.5 mg/kg per hr). All animals received a main-
tenance (3 ml/kg per hr) i.v. fluid (155 mM NaCl) infusion. In
addition, animals were resuscitated with 10 ml of fluid per kg
of weight i.v. every 15 min ifjudged to be hemodynamically
unstable by investigators blinded to treatment group (hemo-
dynamic instability was defined as meeting two or more ofthe
following criteria: mean arterial blood pressure < 70%6 of
baseline, heart rate > 130%o of baseline, pulmonary capillary
wedge pressure < 2 mmHg, and urine output < 1 ml/kg/hr).
Blood samples were obtained prior to E. coli administration
(baseline) and at intervals thereafter.
Statistics. All values are expressed as mean + SEM unless
Two of five healthy adult male volunteers had detectable
baseline levels of both sTNFRs (0.4 and 1.0 ng of sTNFR-I
per ml; 0.8 and 1.2 ng of sTNFR-II per ml) (Fig. 1). Plasma
samples from these five volunteers were also examined at
intervals following endotoxin administration. Like TNFa,
the appearance ofcirculating sTNFR-I and sTNFR-II species
was found to be monophasic, with peak levels occurring after
that of TNFa and detectable levels persisting several hours
beyond the period in which circulating TNFa was detectable
by ELISA (Fig. 2). Concentrations ofsTNFR-I and sTNFR-II
peaked at 3.7 ± 1.6 ng/ml and 1.4 ± 0.2 ng/ml, respectively, at
2 to 3 hr after the administration of endotoxin and remained
above baseline levels for up to 24 hr.
To evaluate further the presence of sTNFRs in human
disease, the plasma of 12 critically ill patients was assayed at
unteers, volunteers after endotoxin administration, and critically ill
patients. The maximal concentration of sTNFR-I and sTNFR-II for
each volunteer and each concentration measured in the critically ill
patients are shown. The median value-for each of the groups is
represented by a horizontal bar.
sTNFR-I and sTNFR-II concentrations in normal vol-
Proc. Natl. Acad. Sci. USA 89(1992)
Proc. Natl. Acad. Sci. USA 89 (1992)
healthy volunteers administered endotoxin [lipopolysaccharide
(LPS)]. Concentrations (mean ± SEM) of circulating TNFa (o),
sTNFR-I (e), and sTNFR-II (v) are shown at baseline and at intervals
after endotoxin administration (indicated by arrow).
Appearance of TNFa and sTNFRs in the circulation of
various points during their clinical course, resulting in 2-12
samples for each patient (mean + SD = 5 + 3). TNFa was
detected by ELISA in only37%ofthe 56 samples andbyWEHI
bioassay in only 20%. In contrast, sTNFR-I and sTNFR-II were
detected in 94% and 89%o of samples, respectively, and a
positive correlation between sTNFR-I and sTNFR-II levels
was evident (n = 56, R = 0.48, P < 0.001). There was no
statistically significant correlation between sTNFR levels and
APACHE II scores, although there was a trend toward higher
sTNFR-II levels in patients with higher APACHE II scores
(R = 0.26, P = 0.06) and higher sTNFR-I levels in nonsurvivors
(P=0.06; two-sided t test). The concentrations ofsTNFR-I and
sTNFR-II observed in these critically ill patients were similar to
the peak levels observed during experimental endotoxemia
To determine whether the concentrations of sTNFRs
found in the circulation are capable of neutralizing endoge-
nous TNFa, plasma samples from volunteers administered
endotoxin were incubated with rsTNFR-I and rsTNFR-II at
S ng/ml to 50;Lg/ml.The addition of S ng of rsTNFR-I per
ml reduced endogenous TNFa bioactivity (calculated to be
585 pg/ml by WEHI) by 66% (Table 1). In contrast, 500 ng of
rsTNFR-II per ml was required to neutralize >50% of
endogenous TNFa cytotoxicity. These findings show that the
addition of physiologic concentrations of rsTNFR-I to
plasma will significantly neutralize the cytotoxicity ofendog-
enous levels of TNFa and suggest that the higher levels of
sTNFR-I observed circulating in critically ill patients (5-10
ng/ml) are sufficient to at least partially attenuate the bio-
logical responses to TNFa production. The same concentra-
tion ofrsTNFR-II is inadequate to block TNFa cytotoxicity.
To evaluate the quantities of soluble receptors required to
neutralize lethal concentrations ofTNFa, plasma samples con-
taining 1.5 ng of human rTNFa per ml were coincubated with
5-5000 ng of rsTNFR-I and rsTNFR-II per ml. This concen-
tration ofhuman rTNFa is roughly 2-3 times the peak concen-
sTNFR-I and sTNFR-II
Neutralization of endogenous TNFa cytotoxicity by
Calculated TNFa, pg/ml
tration observed in human volunteers receiving a mild endot-
oxemia (23, 24) and is comparable to the maximum plasma
value seen in patients with lethal burn sepsis (25). A rsTNFR-I
concentration of50 ng/ml (a 30 molar excess ofsTNFR-I) was
necessary to achieve >50%o inhibition of in vitro TNFa cyto-
toxicity, and >80%o ofTNFa cytotoxicity was inhibited by the
addition of rsTNFR-I at 500 ng/ml (Fig. 3A). TNFa immuno-
activity, as determined by ELISA, was reduced in a similar
fashion by the presence ofrsTNFR-I (Fig. 3B). rsTNFR-II also
reduced TNFa cytotoxicity although to a lesser degree, with
500 ng/ml (a 300 molar excess) required to neutralize >50%o of
TNFa cytotoxicity. In contrast to rsTNFR-I, TNFa immuno-
activity was unaffected by all but the highest concentration of
rsTNFR-II (5000ng/ml). The inhibitionofcytotoxicity resulting
from the simultaneous presence of rsTNFR-I and rsTNFR-II
appears tobe due tothe additive effects ofthe individual soluble
receptors, rather than any synergistic influence on TNFa cy-
Because TNFa-sTNFR-I complexes exhibit reduced cy-
totoxic activity and immunoactivity with the present assays,
it is possible that some portion of TNFa may circulate in a
cryptic form in critically ill patients in whom TNFa concen-
trations could not be detected. To address this issue, plasma
samples from critically ill patients and endotoxemic volun-
teers were assayed for sTNFR-I by an immunoassay with
human rTNFa as the capture protein. This assay takes advan-
tage ofthe observation that free sTNFR-I in plasma will bind to
TNFa that has been coated onto polyvinyl chloride plates,
whereas TNFa-sTNFR-I complexes in plasma will not. The
quantity ofunbound sTNFR-I detected was only 42 ± 4% (n =
36) of the total quantity of binding protein detected when
antibodies against sTNFR-I were used, indicating that roughly
71 . 100|''l
Neutralization of TNFa cytotoxicity and immunoactivi-
ty by combinations of sTNFR-I and sTNFR-II. rsTNFR-I and
rsTNFR-II were added in the indicated concentrations to normal
human plasma to which 1.5 ng of human rTNFa per ml had been
added. The mixtures were allowed to incubate at room temperature
for 30 min and then were assayed for TNFa cytotoxicity by the
WEHI bioassay (A) and for TNFa immunoreactivity by ELISA (B).
All samples were assayed in triplicate and compared with those ofthe
control samples (normal human plasma with added TNFa at 1.5
ng/ml). All values are expressed as a percentage of the calculated
TNFa concentration found in the control samples and represent the
mean of five experiments.
Medical Sciences: Van Zee et al.
Medical Sciences: Van Zee et al.
ine induction in septic baboons
receiving sTNFR-I. Plasma con-
TNFa (A), IL-1f3 (B), and IL-6 (C)
are shown for the animals receiv-
ing treatment (E. coli + rsT-
NFR-I) (W) and the control ani-
mals (E. coli only (o). Cytokine
levels were determined by ELISA
(TNFa and IL-1,) and B.9 hybrid-
oma bioassay (IL-6), as described.
Attenuation of cytok-
± SEM) of
60% of the circulating sTNFR-I is bound to TNFa. In 20 of25
samples from critically ill patients in which TNFa was unde-
tectable by both WEHI bioassay and ELISA, TNFa-sTNFR-I
complexes were detectable, suggesting that TNFa may circu-
late in a large proportion of critically ill patients solely in such
acryptic form. Similarly, TNFa-sTNFR-I complexes werealso
detected after the disappearance of free TNFa from the circu-
lation of human volunteers administered endotoxin.
To directly evaluate the in vivo effects ofadministration of
sTNFRs in sepsis, Papio anubis baboons administered an
LD100 of live E. coli received a 3-hr primed continuous
infusion of rsTNFR-I, sufficient to provide a 300-fold molar
excess of rsTNFR-I over the expected maximum TNFa
concentration. The circulating concentration of rsTNFR-I
achieved during the primed continuous infusion was 18.1 +
0.9Ag/ml,with a calculated clearance of4.1 ± 0.2 ml/kg per
min. Treatment with rsTNFR-I completely prevented the
detection (by ELISA) of TNFa in the circulation of E. coli
septic shock baboons (Fig. 4A) and markedly delayed the
appearance ofTNFa bioactivity, as measured by the WEHI
assay (Fig. 5). Circulating levels of the TNFa-inducible
by administration of rsTNFR-I to septic baboons. Plasma TNF a
SEM) as determined by WEHI bioassay are
shown for the animals receiving treatment (E. coli + rsTNFR-I) (e)
and the control animals (E. coli only) (o).
Delay in plasma appearance ofTNFa cytotoxicity caused
cytokines IL-1f3 and IL-6 were also attenuated by adminis-
tration of rsTNFR-I (Fig. 4 B and C). rsTNFR-I treatment
significantly reduced the volume of resuscitation fluid re-
quired to maintain hemodynamic stability (163 ± 64 ml/kg in
controls vs. 13 ± 13 ml/kg in treatment animals) and the
maximum fall in mean arterial pressure, which declined 56 +
7% from baseline in controls but only 29 ± 6% in baboons
receiving rsTNFR-I (P < 0.01, paired t test). The improve-
ment in hemodynamic performance seen with rsTNFR-I
treatment was of similar magnitude to that which we had
previously observed with monoclonal antibodies to TNFa
(22). Although not statistically significant, the single mortal-
ity occurring within the 24-hr observation period occurred in
the control group not receiving sTNFR-I.
To investigate the appearance of sTNFRs in vivo, a model of
human infection was utilized wherein healthy adult volun-
teers were intravenously administered 20 units of national
reference endotoxin per kg of body weight. This model uses
a mild inflammatory stimulus that induces well-characterized
hormonal and cytokine responses. We and others have
previously demonstrated a monophasic circulating TNFa
response to endotoxin administration, with values peaking
within 1-2 hr and becoming undetectable within 4 hr (23, 24).
It has been reported that such endotoxin administration
also induces the appearance of plasma inhibitors of TNFa
(26), although the inhibitors had not been completely char-
acterized. Until now, sera from humans have been assayed
for TNF inhibitors by in vitro bioassays (10, 26). The recent
sequencing and cloning of the cDNAs encoding the 30-kDa
and 40-kDa extracellular portions of TNFR-I and TNFR-II,
respectively (9, 12, 13, 15), has made it possible to develop
an ELISA for each of these soluble receptors. Using these
ELISAs, we have demonstrated that circulating sTNFR-I
and sTNFR-II are inducible by endotoxemia; the appearance
is monophasic and occurs after that ofTNFa, and detectable
concentrations persist longer than those of TNFa.
The presence of sTNFRs in human disease was evaluated
by sampling the plasma of 12 critically ill patients. The
detection ofsTNFRs was more than twice as frequent as that
of TNFa, occurring in 90%o of the samples. Although the
presence ofcirculating TNFa levels has been correlated with
mortality from severe septic shock in some populations, the
frequency with which this protein is detected in most critical
illness and infection settings is similar to that observed here
(25, 27). The higher frequency of detection of the soluble
receptors in critically ill patients demonstrates that such
soluble receptors may circulate for prolonged periods in
response to continuing inflammation or injury.
The frequent and prolonged appearance of sTNFRs in the
plasma of humans after endotoxemia and in severe illness
suggests a potential physiologic role forthese proteins, perhaps
as a mechanism to protect the organism from the adverse
consequences of excessive TNFa activity. Because it was
unclear whether, at the concentrations reported here, soluble
receptors can bind sufficient quantities ofTNFa to mitigate its
pathologic consequences, studies were undertaken to explore
the capabilities ofthese human sTNFRs to inhibit endogenous
TNFa cytotoxicity in vitro. Endogenous TNFa in plasma
obtained from volunteers 90 min after receiving an endotoxin
challenge was found to be significantly neutralized by 5 ng of
rsTNFR-I or 500 ng of rsTNFR-II per ml. These experiments
establish that physiologic concentrations of rsTNFR-I will
significantly neutralize the cytotoxicity oflevelsofTNFafound
endogenously and suggest that at least the higher concentra-
tions of sTNFR-I detected in the circulation of critically ill
patients (5-10 ng/ml) are sufficient to partially attenuate TNFa
bioactivity, while these concentrations ofrsTNFR-II are insuf-
ficient to inhibit TNFa cytotoxicity.
Proc. Natl. Acad. Sci. USA 89(1992)
Proc. Natl. Acad. Sci. USA 89 (1992)
sTNFRs seen in endotoxemic volunteers or critically ill
patients are able to neutralize the higher concentrations of
TNFa that are observed in patients with overwhelming
bacterial infections. Using a concentration ofhuman rTNFa
associated with lethal sepsis, 50 ng ofrsTNFR-I or 500 ng of
rsTNFR-II per ml were required to neutralize TNFa cyto-
toxicity by at least 50%. This implies that the endogenous
levels of soluble receptors observed in critically ill patients
are inadequate to block the detrimental effects of high levels
ofTNFa. In conjunction with the observation that in human
experimental endotoxemia the peak appearance of both sol-
uble receptors occurred after that ofTNFa, and presumably
after the initiation of any detrimental TNFa-induced
changes, these in vitro studies suggested a therapeutic role
for administered sTNFRs. Our in vitro studies demonstrated
that plasma samples with 1.5 ng ofTNFa per ml required the
addition of 500 ng of rsTNFR-I (a 300-fold molar excess) per
ml to achieve >80% inhibition of TNFa cytotoxicity. Ex-
trapolating to the maximum concentration ofTNFa observed
in previous baboons administered E. coli, -30-80 ng/ml, one
would expect that 9,000-24,000 ng/ml would be necessary to
achieve a similar level of inhibition.
To investigate whether exogenous sTNFR-I sufficient to
achieve a 300-fold molar excess (over the expected maximum
circulating TNFa concentration) would ameliorate the
TNFa-mediated pathophysiologic consequences of septic
shock, an in vivo study was undertaken. Treatment with
human rsTNFR-I resulted in attenuation of TNFa-induced
cytokines and hemodynamic instability in Papio anubis ba-
boons with Gram-negative septic shock. This supports the
recent reports of others wherein a fusion product of the
TNFR and human IgG provided protection against endotoxic
shock in mice (28, 29).
Of interest is the fact that although both the early immuno-
activity and bioactivity due to TNFa seen in the control
animals (Figs. 4A and 5) were eradicated by administration of
rsTNFR-I to septic animals, a late (4-8 hr after E. coli
administration) appearance of TNFa-induced cytotoxicity in
the plasma occurred. Treatment of plasma samples with a
monoclonal antibody against TNFa in vitro eliminated this
cytotoxicity, thereby demonstrating that it is due to TNFa
rather than some other agent cytotoxic to WEHI cells. The
biological significance of the delayed appearance of TNFa
bioactivity is unclear. However, the absence ofTNFa immu-
noactivity in vivo (Fig. 4A) atthe latertime points suggests that
free TNFa is not present in the circulation. Rather, one
possible explanation is that TNFa-sTNFR-I complexes cir-
culate for extended periods and that such complexes are not
protective when incubated under in vitro conditions in a
WEHI bioassay but are protective in vivo, as demonstrated by
the improvement in hemodynamic parameters and attenuation
of the subsequent IL-1f3 and IL-6 responses.
Thus, we conclude that sTNFR-I and sTNFR-II appear in
the circulation of man during endotoxemia and circulate in
the plasma for a longer period oftime than does free TNFa.
In addition, most critically ill patients exhibit detectable
plasma levels of sTNFRs. Our findings suggest that, in the
absence of detectable TNFa bioactivity or immunoactivity,
TNFa circulates bound to sTNFR-I in a cryptic inactive
form. This observation is of unclear clinical significance.
However, the clinical significance of circulating sTNFRs is
potentially profound, as in vitro studies suggest that circu-
lating soluble receptor concentrations seen in critically ill
patients and in experimental endotoxemia are sufficient to
neutralize the bioactivity associated with TNFa concentra-
tions observed in mild inflammation, whereas such levels are
inadequate to neutralize the cytotoxicity associated with
TNFa levels seen in overwhelming or lethal sepsis. The in
vivo primate studies demonstrate that rsTNFR-I may have
it was unclear whether circulating levels of
use as a therapeutic intervention in situations where there is
insufficient or delayed production of endogenous sTNFRs.
We thank A. Hudson, J. Sjoberg, and S. Baker for their excellent
technical assistance and D. Bloedow for helpful discussions and
assistance with the sTNFR assays. This work was supported by
grants GM-34695, GM-40586, CA-52108, and RR-00047 awarded by
the National Institutes of Health and a clinical fellowship from the
American Cancer Society (to K.J.V.Z.).
Fong, Y., Tracey, K. J., Moldawer, L. L., Hesse, D. G., Manogue,
K. B., Kenney, J. S., Lee, A. T., Kuo, G. C., Allison, A. C.,
Lowry, S. F. & Cerami, A. (1989) J. Exp. Med. 170, 1627-1633.
Tracey, K. J., Beutler, B., Lowry, S. F., Merryweather, J., Wolpe,
S., Milsark, I. W., Hariri, R. J., Fahey, T. J., III, Zentella, A.,
Albert, J. D., Shires, G. T. & Cerami, A. (1986) Science 234,
Moldawer, L. L., Andersson, C., Svaninger, G., Gelin, J. & Lund-
holm, K. (1988) Am. J. Physiol. 254, G450-G456.
Tracey, K. J., Wei, H., Manogue, K. R., Fong, Y., Hesse, D. G.,
Nguyen, H. T., Kuo, G. C., Beutler, B., Cotran, R. S., Cerami, A.
& Lowry, S. F. (1988) J. Exp. Med. 167, 1211-1227.
Peetre, C., Thysell, H., Grubb, A. & Olsson, I. (1988) Eur. J.
Haematol. 41, 414-419.
Seckinger, P., Isaaz, S. & Dayer, J.-M. (1988) J. Exp. Med. 167,
Engelmann, H., Aderka, D., Rubinstein, M., Rotman, D. &
Wallach, D. (1989) J. Biol. Chem. 264, 11974-11980.
Engelmann, H., Novick, D. & Wallach, D. (1990) J. Biol. Chem.
Kohno, T., Brewer, M. T., Baker, S. L., Schwartz, P. E., King,
M. W., Hale, K. K., Squires, C. H., Thompson, R. C. & Vannice,
J. L. (1990) Proc. Nati. Acad. Sci. USA 87, 8331-8335.
Gatanaga, T., Hwang, C., Kohr, W., Cappuccini, F., Lucci, J. A.,
III, Jeffes, E. W. B., Lentz, R., Tomich, J., Yamamoto, R. S. &
Granger, G. A. (1990) Proc. Nati. Acad. Sci. USA 87, 8781-8784.
Smith, C. A., Davis, T., Anderson, D., Solam, L., Beckmann,
M. P., Jerzy, R., Dower, S. K., Cosman, D. & Goodwin, R. G.
(1990) Science 248, 1019-1023.
Loetscher, H., Pan, Y.-C. E., Lahm, H.-W., Gentz, R., Brockhaus,
M., Tabuchi, H. & Lesslauer, W. (1990) Cell 61, 351-359.
Schall, T. J., Lewis, M., Koller, K. J., Lee, A., Rice, G. C., Wong,
G. H. W., Gatanaga, T., Granger, G. A., Lentz, R., Raab, H.,
Kohr, W. J. & Goeddel, D. V. (1990) Cell 61, 361-370.
Seckinger, P., Zhang, J.-H., Hauptmann, B. & Dayer, J.-M. (1990)
Proc. Natl. Acad. Sci. USA 87, 5188-5192.
Gray, P. W., Barrett, K., Chantry, D., Turner, M. & Feldmann, M.
(1990) Proc. Natd. Acad. Sci. USA 87, 7380-7384.
Porteu, F. & Nathan, C. (1990) J. Exp. Med. 172, 599-607.
Espevik, T. & Nissen-Meyer, J. (1986) J. Immunol. Methods 95,
Kenney, J. S., Masada, M. P., Eugui, E. M., Delustro, B. M.,
Mulkins, M. A. & Allison, A. C. (1987) J. Immunol. 138, 4236-
Aarden, L. A., De Groot, E. R., Schaap, 0. L. & Lansdorp, P. M.
(1987) Eur. J. Immunol. 17, 1411-1416.
Hale, K. J., Smith, C. G., Vanderslice, R. W., Baker, S., Russell,
D. A., Rivera, R. I., Dripps, D. & Kohno, H. (1991) J. Cell.
Biochem., Suppl. 15F, 113 (abstr.).
Knaus, W. A., Draper, E. A., Wagner, D. P. & Zimmerman, J. E.
(1985) Crit. Care Med. 13, 818-829.
Tracey, K. J., Fong, Y., Hesse, D. G., Manogue, K. R., Lee, A. T.,
Kuo, G. C., Lowry, S. F. & Cerami, A. (1987) Nature (London) 330,
Fong, Y., Marano, M. A., Moldawer, L. L., Wei, H., Calvano,
S. E., Kenney, J. S., Allison, A. C., Cerami, A., Shires, G. T. &
Lowry, S. F. (1990) J. Clin. Invest. 85, 1896-1904.
Michie, H. R., Manogue, K. R., Spriggs, D. R., Revhaug, A.,
O'Dwyer, S., Dinarello, C. A., Cerami, A., Wolff, S. M. & Wil-
more, D. W. (1988) N. Engl. J. Med. 318, 1481-1486.
Marano, M. A., Fong, Y., Moldawer, L. L., Wei, H., Calvano,
S. E., Tracey, K. J., Barie, P. S., Manogue, K., Cerami, A., Shires,
G. T. & Lowry, S. F. (1990) Surg. Gynecol. Obstet. 170, 32-38.
Spinas, G. A., Bloesch, D., Kaufmann, M.-T., Keller, U. & Dayer,
J.-M. (1990) Am. J. Physiol. 259, R993-R997.
Waage, A., Halstensen, A. & Espevik, T. (1987) Lancet i, 355-357.
Lesslauer, W., Tabuchi, H., Gentz, R., Brockhaus, M., Schlaeger,
E. J., Grau, G., Piguet, P. F., Pointaire, P., Vassalli, P. & Loet-
scher, H. (1991) Eur. J. Immunol. 21, 2883-2886.
Ashkenazi, A., Marsters, S. A., Capon, D. J., Chamow, S. M.,
Figari, I. S., Pennica, D., Goeddel, D. V., Palladino, M. A. &
Smith, D. H. (1991) Proc. Natl. Acad. Sci. USA 88, 10535-10539.
Medical Sciences: Van Zee et al.