IL-12 enhances the natural killer cell cytokine response to Ab-coated tumor cells.

Abstract

The anti-tumor activity of recombinant mAb's directed against tumor cell growth receptors has generally been considered to result from direct antiproliferative effects, the induction of apoptosis, or possibly Ab-dependent cellular cytotoxicity mediated against tumor targets. However, it remains unclear to what degree these mechanisms actually aid in the clearance of Ab-coated tumor cells in vivo. We show here that NK cells secrete a distinct profile of potent immunostimulatory cytokines in response to dual stimulation with Ab-coated tumor cells and IL-12. This response could not be duplicated by costimulation with other ILs and was significantly enhanced in the presence of monocytes. Cytokine production was dependent upon synergistic signals mediated by the activating receptor for the Fc portion of IgG (FcgammaRIII) and the IL-12 receptor expressed on NK cells. Coadministration of Ab-coated tumor cells and IL-12 to BALB/c mice resulted in enhanced circulating levels of NK cell-derived cytokines with the capacity to augment anti-tumor immunity. These findings suggest that, in addition to mediating cellular cytotoxicity and apoptosis, the anti-tumor activity of mAb's might also result from activation of a potent cytokine secretion program within immune effectors capable of recognizing mAb-coated targets.

Full text

Page 1

Introduction
HER2/neu is a proto-oncogene that encodes a cell-sur-
face transmembrane receptor with intracellular tyro-
sine kinase activity (1). Upon ligand binding, the HER2
receptor dimerizes with other members of the EGF-R
family and initiates intracellular signals that result in
proliferation. The HER2/neu gene is amplified in
approximately 25% of human adenocarcinomas (2).
Overexpression of this receptor leads to ligand-inde-
pendent activation of the HER2 receptor kinase and
aberrant epithelial cell proliferation. Breast cancer
patients with this alteration frequently exhibit a worse
histological grade, decreased relapse-free and overall
survival, and altered sensitivity to standard chemother-
apeutic regimens (1, 3). A murine mAb directed against
the extracellular domain of the HER2 receptor
(mu4D5) was developed in the hope that tumors
exhibiting HER2-dependent growth might respond
with reduced proliferation (4). Indeed, this Ab does
inhibit the growth of HER2-positive tumor cells in
vitro and mediates regression of established tumors in
a variety of animal models (5). Therapy with a recom-
binant, humanized form of this Ab (trastuzumab, trade
name Herceptin) has shown efficacy in clinical trials,
both as a single agent and when administered with
chemotherapeutic drugs (6, 7). However, only 25–30%
of patients with HER2-overexpressing malignancies
will respond to these treatments. The ability to effec-
tively combine Herceptin with other anti-tumor agents
might be enhanced with a more complete understand-
ing of its mechanism of action.
The binding of Herceptin to the HER2 receptor may
lead to alterations in intracellular signaling and cell
cycle kinetics within the target cell (8). In addition,
Herceptin may also stimulate important immune
responses such as Ab-dependent cellular cytotoxicity
(ADCC) and complement activation that may lead to
direct tumor lysis (9). Indeed, it has been shown in a
murine tumor xenograft model that the anti-tumor
effects of Herceptin are dependent upon the presence
of host immune cells that express receptors for the Fc
portion of IgG (FcR) (10). Important ADCC-mediating
effector cells that express FcR include mono-
cytes/macrophages, resting and activated granulocytes,
and NK cells. Although most immune cells coexpress
both activating and inhibitory FcR’s, NK cells are
unique in that they constitutively express only the acti-
vating, low-affinity receptor FcγRIII (11). They also
contain abundant cytolytic granules, prominently
express cellular adhesion molecules, and constitutive-
ly display multiple cytokine receptors. Indeed, NK cell
cytotoxicity against HER2-expressing tumor cells is
The Journal of Clinical Investigation|October 2002| Volume 110|Number 7
983
IL-12 enhances the natural killer cell cytokine response
to Ab-coated tumor cells
Robin Parihar,1,2Julie Dierksheide,3Yan Hu,2and William E. Carson1,2,3
1Department of Molecular Virology, Immunology, and Medical Genetics,
2The Arthur G. James Comprehensive Cancer Center, and
3Department of Surgery, The Ohio State University College of Medicine, Columbus, Ohio, USA
The anti-tumor activity of recombinant mAb’s directed against tumor cell growth receptors has gen-
erally been considered to result from direct antiproliferative effects, the induction of apoptosis, or
possibly Ab-dependent cellular cytotoxicity mediated against tumor targets. However, it remains
unclear to what degree these mechanisms actually aid in the clearance of Ab-coated tumor cells in
vivo. We show here that NKcells secrete a distinct profile of potent immunostimulatory cytokines in
response to dual stimulation with Ab-coated tumor cells and IL-12. This response could not be dupli-
cated by costimulation with other ILs and was significantly enhanced in the presence of monocytes.
Cytokine production was dependent upon synergistic signals mediated by the activating receptor for
the Fc portion of IgG (FcγRIII) and the IL-12 receptor expressed on NK cells. Coadministration of
Ab-coated tumor cells and IL-12 to BALB/c mice resulted in enhanced circulating levels of NK
cell–derived cytokines with the capacity to augment anti-tumor immunity. These findings suggest
that, in addition to mediating cellular cytotoxicity and apoptosis, the anti-tumor activity of mAb’s
might also result from activation of a potent cytokine secretion program within immune effectors
capable of recognizing mAb-coated targets.
J. Clin. Invest. 110:983–992 (2002). doi:10.1172/JCI200215950.
Received for publication May 17, 2002, and accepted in revised form
August 30, 2002.
Address correspondence to: William E. Carson, N924 Doan
Hall, 410 West 10th Avenue, The Ohio State University College of
Medicine, Columbus, Ohio 43210, USA. Phone: (614) 293-6306;
Fax: (614) 688-4366; E-mail: carson-1@medctr.osu.edu.
Conflict of interest: No conflict of interest has been declared.
Nonstandard abbreviations used: Ab-dependent cellular
cytotoxicity (ADCC); macrophage inflammatory protein-1α
(MIP-1α); antigen presenting cell (APC); IL-12 receptor (IL-12R);
recombinant human (rhu); recombinant murine (rmu);
phycoerythrin (PE); protein tyrosine kinase (PTK).

Page 2

markedly enhanced following treatment with IL-2 or
IL-12 (12, 13). NK cells are also sources of potent
immunostimulatory cytokines such as TNF-α,
GM-CSF, macrophage inflammatory protein-1α(MIP-
1α), and IFN-γ(11). Thus, while the anti-tumor activi-
ty of Herceptin has largely been attributed to direct
antiproliferative and proapoptotic effects, we theorized
that the clearance of Ab-coated tumor cells might be
enhanced by the coadministration of immunologic
adjuvants with the capacity to stimulate NK cell cyto-
toxicity and cytokine production.
IL-12 is an antigen presenting cell–derived (APC-
derived) cytokine that stimulates T cells and NK cells to
secrete IFN-γ and augments the proliferation and
cytolytic activity of these cells (14). In addition to its crit-
ical role in the regulation of early inflammatory respons-
es and promotion of the Th1-type repertoire, preclinical
studies have suggested that IL-12 is an effective anti-can-
cer agent against various experimental malignancies.
Work in numerous murine models has demonstrated
that the anti-tumor effects of IL-12 are primarily medi-
ated via the induction of IFN-γ secretion by T and NK
cells (15, 16). Of note, IL-12–induced IFN-γ production
has been shown to prevent mammary carcinogenesis in
HER2/neutransgenic mice when IL-12 was administered
alone or in combination with tumor cell– or dendritic
cell–based vaccines (17, 18). Neutralization of IFN-γ in
this and other models strongly inhibits the anti-tumor
activity of IL-12. Similarly, depletion of NK cells within
hosts given exogenous IL-12 has been shown to attenu-
ate the anti-tumor effect, suggesting a vital role for NK
cells in the anti-tumor activity of IL-12 (19).
In the current report, we have examined the ability of
IL-12 to enhance NK cell–mediated cytotoxicity and
promote immunostimulatory cytokine secretion. We
demonstrate that costimulation of pure NK cells or
whole PBMCs with Herceptin-coated human breast
cancer cells and IL-12 results in a unique cytokine secre-
tion profile highlighted by abundant production of
IFN-γ. Furthermore, we provide evidence that cytokine
production in response to Herceptin-coated tumor cells
and IL-12 is mediated by synergistic signals provided by
FcγRIII and the IL-12 receptor (IL-12R), and that this
synergistic effect could not be duplicated by IL-2, IL-10,
IL-15, or IL-18. These findings provide a strong ration-
ale for the concurrent administration of Herceptin and
IL-12 for the treatment of HER2-overexpressing malig-
nancies and lend insight into the potential role of NK
cells in the elimination of Ab-coated tumor targets.
Methods
Cytokines and Ab’s. Recombinant human IL-12 (rhuIL-
12) and murine IL-12 (rmuIL-12) were kindly provided
by Genetics Institute Inc. (Cambridge, Massachusetts,
USA). rhuIL-2 with a specific activity of 1.53 × 107
U/mg was obtained from Hoffman-LaRoche Inc.
(Nutley, New Jersey, USA). rhuIL-15, rhuIL-18, and
rhuIL-10 were purchased from PeproTech Inc. (Rocky
Hill, New Jersey, USA). Cytokines were resuspended in
1× PBS plus 0.1% BSA. The humanized anti-HER2
mAb Herceptin was kindly provided by Genentech Inc.
(San Francisco, California, USA) (20). Purified F(ab′)2
fragments of Herceptin were generated by pepsin diges-
tion as previously described (21). Polyclonal huIgG was
purchased from Sigma-Aldrich (St. Louis, Missouri,
USA). For preliminary costimulation experiments, 96-
well flat-bottom plates were coated with huIgG by
incubation with 100 µg/ml huIgG in 100 µl of PBS
overnight at 4°C. Wells were then washed twice with
200 µl fresh PBS and once with warm 10% HAB medi-
um consisting of heat-inactivated pooled human AB
serum (C-six Diagnostics Inc., Germantown, Wiscon-
sin, USA), 100 U/ml penicillin, 100 µg/ml streptomy-
ocin, and 0.25 µg/ml amphotericin, prior to use.
Cell lines. The SKBR3 (HER2-overexpressing) and
MDA-468 (HER2-negative) human breast adenocarci-
noma lines were obtained from American Type Culture
Collection (Manassas, Virginia, USA). The MCF-
7HER2/neu(HER2-overexpressing) human breast adeno-
carcinoma was obtained from Jose Baselga (Hospital
Universitari Vall D’Hebron, Barcelona, Spain). The
MCF-7HER2/neucell line was created by transfection of
parental MCF-7 cells with an expression cassette for
human HER2/neu and culture in standard medium
containing neomycin (22). A murine colon carcinoma
line overexpressing human HER2/neu, CT-26HER2/neu,
and the parental line, CT-26, were kind gifts from P.T.P.
Kaumaya (Ohio State University, Columbus, Ohio,
USA) and were used for in vivo costimulation studies
(23). Cancer cell lines were propagated in RPMI-1640
medium supplemented with 10% heat-inactivated FBS,
300 µg/ml L-glutamine, 100 U/ml penicillin, 100 µg/ml
streptomycin, 0.25 µg/ml amphotericin B, and 0.06
mg/ml anti-PPLO agent (all from Life Technologies
Inc., Rockville, Maryland, USA).
Mice. Female BALB/c mice between the ages of 5 and
7 weeks were purchased from The Jackson Laboratory
(Bar Harbor, Maine, USA). STAT4–/–breeder mice on a
B6 × 129 strain background were a kind gift from
James Ihle (St. Jude Children’s Research Hospital,
Memphis, Tennessee, USA), and females were used in
experiments at 5–7 weeks of age (24). Age-matched
wild-type B6×129 mice (The Jackson Laboratory) were
used as controls. Mice were maintained in the animal
facility at the Ohio State University Comprehensive
Cancer Center (CCC) with free access to food and
water. All protocols were approved by the Ohio State
University CCC Animal Care and Use Committee, and
mice were treated in accordance with the institutional
guidelines for animal care.
Isolation of human NK cells. PBMCs were isolated from
fresh leukopacks (American Red Cross, Columbus,
Ohio, USA) using Ficoll-Hypaque (Sigma-Aldrich) den-
sity gradient centrifugation. PBMCs were washed in
RPMI-1640 (Life Technologies Inc.) and adhered
overnight to plastic plates to eliminate the monocyte
population. Nonadherent peripheral blood lympho-
cytes were further depleted of T cells, B cells, and mono-
984
The Journal of Clinical Investigation| October 2002|Volume 110|Number 7

Page 3

cytes using an Ab cocktail against epitopes of CD3,
CD4, and HLA-DR, respectively, in conjunction with
goat anti-mouse immunomagnetic beads, as previous-
ly described (25). In some experiments, NK cells were
isolated directly from peripheral blood leukopacks by
30-minute incubation with RossetteSep cocktail (Stem
Cell Technologies, Vancouver, British Columbia, Cana-
da), followed by Ficoll-Hypaque density gradient cen-
trifugation. Depleted peripheral blood lymphocytes
were further stained with CD56-phycoerythrin (CD56-
PE) (Beckman Coulter, Fullerton, California, USA),
washed, and sorted for a population containing both
CD56dimand CD56bright(herein referred to as CD56+)
NK cells on a FACStar Plus cell sorter (Becton, Dickin-
son and Co., Franklin Lakes, New Jersey, USA). Sorted
CD56+NK cells were greater than 96% pure by FACStar
analysis. For our in vitro assay, human PBMCs and NK
cells were cultured in 10% HAB medium.
In vitro coculture assay. Wells of a 96-well flat-bottom
culture plate were coated with either the HER2-overex-
pressing cell line SKBR3, the HER2-negative cell line
MDA-468, or a breast cancer cell line engineered to over-
express HER2/neu, MCF-7HER2/neu. Tumor cells (5 ×104)
were grown to confluence overnight at 37°C. The cul-
ture supernatant was aspirated the following day, and
wells were treated with 100 µg/ml Herceptin for 1 hour
at 37°C and then washed twice with warm medium to
remove any unbound Ab. Washing with warm medium
had no appreciable effect on cell viability. Unfraction-
ated PBMCs or purified CD56+NK cells isolated from
normal donors were added at 2 ×105cells/well in 200 µl
medium in the presence of 10 ng/ml rhuIL-12. Control
wells contained tumor cells plus PBMCs or NK cells
supplemented with medium alone, Herceptin alone,
Herceptin F(ab′)2fragment, or rhuIL-12 alone. Culture
supernatants were harvested at the indicated timepoints
and analyzed for levels of various cytokines (IFN-γ,
TNF-α, GM-CSF, and MIP-1α) by ELISA using com-
mercially available mAb pairs (Endogen Inc., Woburn,
Massachusetts, USA). Based on the manufacturer’s
guidelines, a standard sandwich ELISA for each human
cytokine was developed, and cytokine concentrations
were determined by the use of standard curve regression
analysis as previously described (15). The detection limit
for all ELISAs was 10–30 pg/ml. All results are shown as
the mean of triplicate wells ±SEM.
Real-time PCR. Cells were harvested from culture and
lysed with 500 µl RNA lysis buffer. Total cellular RNA
was isolated using RNeasy Mini Kits (Qiagen Inc.,
Valencia, California, USA), and cDNA was generated
with random hexamer primers and MMLV-RT accord-
ing to manufacturer recommendations (Life Tech-
nologies Inc.). Using the generated cDNA as template,
real-time PCR for human IFN-γ transcript was per-
formed as multiplex reactions with primer and probe
sets specific for the cytokine transcript and an 18S
rRNA internal control (PE Applied Biosystems, Foster
City, California, USA). Reactions were performed
using an ABI Prism 7700 sequence detector, and data
were analyzed with Sequence Detector (version
1.6)software (PE Applied Biosystems). Data were ana-
lyzed according to the comparative CTmethod using
18S internal control transcript levels to normalize dif-
ferences in sample loading and preparation. Results
are semiquantitative and represent the fold difference
in transcript levels in a particular sample as compared
with levels from unstimulated cells.
Administration of Herceptin plus IL-12 to mice. CT-26 and
CT-26HER2/neucells were harvested from culture via mild
trypsinization, washed once with sterile PBS, and incu-
bated at 4°C in PBS plus 10% FBS at 1 × 107cells/ml
with either Herceptin (1 mg/ml) or normal huIgG (1
mg/ml) for 45 minutes. Cells were then washed twice in
sterile PBS, and 4 ×106cells in PBS were given intraperi-
toneally to each mouse concurrently with a separate
intraperitoneal injection of 1 µg muIL-12 in PBS. Each
group consisted of ten wild-type BALB/c mice, and con-
trol groups received injections of IgG-treated tumor
alone, IgG-treated tumor plus IL-12, or Herceptin-coat-
ed tumor alone. Serum was harvested from each mouse
at 6, 12, and 24 hours and analyzed for cytokine levels
by ELISA. Mice were immediately euthanized, and
splenocytes were harvested for use in our in vitro cocul-
ture assay using CT-26 and CT-26HER2/neucells as the
HER2–and HER2+tumor cell lines, respectively. Culture
supernatants were harvested after 24 hours and ana-
lyzed for levels of muIFN-γ by ELISA (R&D Systems
Inc., Minneapolis, Minnesota, USA).
Statistics. Statistical analysis of ELISA cytokine levels
was performed using the paired Student t test, with
P <0.05 considered significant.
Results
NK cells costimulated with immobilized IgG and IL-12 pro-
duce large quantities of IFN-γ. Our laboratory has previ-
ously demonstrated that NK cells can mediate ADCC
against Ab-coated tumor targets and that the lytic
activity of the NK cell compartment is markedly
enhanced by activation with IL-2 or IL-15 (12). Our
recent observations show that ADCC mediated by
breast cancer patient PBMCs (26) against HER2-over-
expressing cell lines is also markedly enhanced follow-
ing culture of effectors with IL-12, and the observed
lytic activity is comparable to that obtained with IL-2
(data not shown). We were next interested in determin-
ing the ability of IL-12 to enhance NK cell cytokine pro-
duction in the presence of immobilized huIgG, as the
NK cell response to this stimulus has not been previ-
ously characterized. Purified NK cells cultured for 72
hours in the presence of both immobilized huIgG and
rhuIL-12 produced large amounts of IFN-γ, while NK
cells stimulated with immobilized IgG alone or IL-12
alone produced negligible amounts of this cytokine
(Figure 1a). Further experiments revealed a clear IL-12
dose–dependent effect on secretion of IFN-γ.
Detectable NK cytokine secretion could be achieved
upon costimulation with only 0.1 ng/ml IL-12. How-
ever, optimal production of IFN-γ required higher
The Journal of Clinical Investigation|October 2002| Volume 110| Number 7
985

Page 4

levels of IL-12 (Figure 1b). While levels of IFN-γ were
detected after 6 hours in culture, peak accumulation
did not occur until 72 hours after stimulation (Figure
1c). In order to confirm the specific effects of IL-12,
other cytokines were examined for their ability to act as
costimuli for NK cell cytokine production. Interest-
ingly, markedly lower levels of IFN-γ were obtained
when NK cells were stimulated with immobilized IgG
in combination with IL-2, IL-10, IL-15, IL-18 (Figure
2), or IFN-α (not shown). Thus, under the conditions
of this assay, IL-12 appears to be the most effective cos-
timulus for NK cell secretion of IFN-γ.
Herceptin and IL-12 provide a potent costimulus for NK
cell cytokine production. Culture of human NK cells in
the presence of IL-12 and immobilized anti-tumor
mAb’s such as Herceptin or rituximab (a humanized
mAb recognizing the CD20 B cell antigen) (27) also
resulted in significant production of IFN-γ over the
course of 72 hours (data not shown). In order to
extend these observations in the context of combina-
tion therapy using Herceptin plus cytokine, we pro-
ceeded to develop an in vitro assay that might better
reflect the tumor microenvironment.
HER2-overexpressing (SKBR3 and MCF-7HER2/neu) and
HER2-negative (MDA-468) cell lines were grown to con-
fluence in flat-bottom wells and coated with Herceptin.
Each well was then supplemented with IL-12 and puri-
fied human NK cells. Herceptin-coated tumor cells
served as a potent stimulus for NK cell cytokine pro-
duction, provided that IL-12 was present in the culture
medium (Figure 3a). In contrast, there was minimal pro-
duction of IFN-γ in wells containing HER2-negative
cells, regardless of the presence of IL-12. Similar results
were obtained for a second pair of HER2-positive and
-negative cell lines (SKBR3 and HT-1080, respectively).
A time-course study revealed that NK cell production of
IFN-γbegan within 12 hours of exposure to Herceptin-
coated HER2-positive cells (Figure 3b). Interestingly, a
small amount of IFN-γ was secreted in response to the
HT-1080 cell line. Flow cytometric analysis revealed low-
level surface overexpression of the HER2 antigen on this
cell line (not shown), suggesting that combination ther-
apy with Herceptin and IL-12 might be effective in
patients whose tumors express modest levels of HER2.
This represents a significant finding, as Herceptin treat-
ments to date have shown efficacy only in patients
whose tumors have high-level expression of the HER2
antigen (8). Analysis by real-time RT-PCR revealed a 35-
fold increase in cytokine transcript in NK cells costim-
ulated with Herceptin-coated tumor cells and IL-12
(Figure 3c). In contrast, IFN-γtranscript induction ver-
986
The Journal of Clinical Investigation| October 2002| Volume 110| Number 7
Figure 1
NK cells costimulated with immobilized IgG and IL-12 secrete high lev-
els of IFN-γ. (a) Human NK cells were cultured on wells precoated with
huIgG. rhuIL-12 was added at a concentration of 10 ng/ml. Control
wells consisted of NK cells cultured with medium alone (Medium),
immobilized huIgG alone (IgG), or rhuIL-12 alone (IL-12). Culture
supernatants were harvested after 72 hours and analyzed for IFN-γ
content by ELISA. (b) Human NK cells were cultured on immobilized
huIgG in varying concentrations of rhuIL-12 (0.01–20 ng/ml), and
supernatants were harvested after 72 hours. (c) In time-course exper-
iments, NK cells were cultured with 10 ng/ml IL-12 on immobilized
huIgG. Supernatants were harvested at varying times (6–72 hours).
*P <0.001 versus medium, IgG, and IL-12.
Figure 2
IL-12 is the most effective costimulus with immobilized IgG for IFN-γ
production by NK cells. Purified NK cells were cultured on immobi-
lized huIgG in the presence of various human cytokines (IL-2, IL-10,
IL-15, IL-18, and IL-12) at 10 ng/ml. Standard controls were also
included. As a further control, some wells were coated with human
F(ab′)2fragment prior to the addition of NK cells (Fab). Culture
supernatants were harvested after 72 hours and analyzed for IFN-γ
content by ELISA. *P < 0.005 versus all conditions shown.

Page 5

sus an HER2-negative cell line was less than twofold for
all conditions (not shown).
Synergistic cytokine secretion induced by IgG and IL-12 is
mediated through distinct pathways. In order to elucidate
the mechanism by which Herceptin and IL-12 act in
synergy to induce cytokine secretion by NK cells, both
IL-12 and IgG signaling pathways were examined.
STAT4 is an important intracellular signaling molecule
and transcription factor responsible for mediating
IL-12–induced IFN-γ secretion by T cells and NK cells
(24). Abrogation of STAT4-mediated signaling has
been shown to dramatically affect the transcription of
IL-12–stimulated genes. Splenocytes from wild-type
mice and STAT4-deficient (STAT4–/–) mice were plated
onto wells precoated with immobilized murine IgG
and costimulated with murine IL-12. Whereas
wild-type splenocytes displayed synergistic IFN-γ pro-
duction in response to immobilized IgG and IL-12,
cells from STAT4–/–mice were unable to secrete IFN-γ
in response to stimulation with either IL-12 alone or
with IL-12 plus immobilized IgG (Figure 4a). Impor-
tantly, cytokine secretion in response to immobilized
IgG alone, while low, was not affected in STAT4–/–cells.
To further clarify the contribution of IgG-induced
signaling to cytokine secretion, specific inhibitors of
intracellular signaling were employed in our in vitro cos-
timulation assay. Freshly isolated human PBMCs were
plated with rhuIL-12 on wells precoated with huIgG in
the presence of piceatannol, a specific inhibitor of Syk
and ZAP-70 kinases (28). Inhibition of IgG-mediated
signaling by piceatannol dramatically reduced IFN-γ
secretion by costimulated PBMCs (Figure 4b). Taken
together, these results suggest that cytokine secretion
by NK cells in response to immobilized Herceptin and
IL-12 is critically dependent upon these two distinct sig-
nals relayed from the cell surface to the nucleus.
PBMCs secrete greater amounts of cytokines than do purified
NK cells in response to Herceptin-coated tumor cells and IL-12.
We next proceeded to determine whether there were
quantitative differences in cytokine secretion between
pure NK cells and whole PBMCs in response to costimu-
lation with Herceptin-coated tumor cells and IL-12. The
use of PBMCs in our coculture assay might provide a
more accurate picture of the cytokine levels to be found
within the tumor microenvironment following systemic
administration of Herceptin and IL-12. Surprisingly,
2 × 105PBMCs (consisting of approximately 12% NK
cells) consistently secreted greater amounts of IFN-γ in
response to Herceptin-coated HER2-overexpressing cells
in the presence of IL-12 than did 2 ×105purified NK cells
(Figure 5a). Large quantities of TNF-α, GM-CSF, and
MIP-1αwere also secreted by PBMCs in response to cos-
timulation (Figure5, b–d). These findings suggested that
either a compartment other than NK cells was directly
responsible for cytokine secretion in response to Her-
ceptin and IL-12 costimulation, or the monocyte com-
partment might act to enhance the NK cell production
of cytokines, or both. To help clarify these issues, cytokine
production by specific immune effectors was examined.
Both NK and T cells secrete IFN-γin response to Herceptin-
coated tumor cells and IL-12. Having observed consistent-
ly higher levels of cytokine secretion by a whole PBMC
population than by pure NK cells in our coculture
model, we next examined which cellular compartment
could be responsible for the additional cytokine secre-
tion. We used flow cytometry to analyze intracellular
The Journal of Clinical Investigation| October 2002| Volume 110| Number 7
987
Figure 3
NK cells secrete large quantities of IFN-γ in response to Herceptin-
coated human breast cancer cells and IL-12. (a) The MDA-468 and
SKBR3 cell lines were cultured with purified NK cells in our in vitro
coculture assay. Control wells contained tumor cells and NK cells
supplemented with medium alone, Herceptin alone (Her), or rhuIL-
12 alone. As a further control, some wells were treated with Her-
ceptin F(ab′)2fragment prior to the addition of NK cells. Culture
supernatants were harvested after 72 hours and analyzed for IFN-γ
content by ELISA. *P < 0.01 versus medium, F(ab′)2, Herceptin,
IL-12, Her + IL-12 (MDA-468), and F(ab′)2 + IL-12 (MDA-468 and
SKBR3). (b) The SKBR3 (HER2 high-level expression) and HT-1080
(HER2 low-level expression) human breast cancer cell lines were used
in our in vitro coculture assay. Supernatants were harvested at the
indicated times (12–72 hours). **P <0.05 versus medium, IgG, Her-
ceptin, IL-12 (HT-1080 and SKBR3), and Her + IL-12 (HT-1080).
(c) NK cells cultured in our in vitro coculture assay were harvested
after 72 hours and processed for real-time PCR analysis of the indi-
cated transcript (see Methods). Results are given as fold increase
in cytokine transcript over baseline versus the SKBR3 cell line.
***P < 0.05 versus medium, Her, and IL-12.