Clin Exp Immunol 1999; 116:388–394

Studies of the mechanism of cytolysis by tumour-infiltrating lymphocytes

M. HISHII, J. T. KURNICK, T. RAMIREZ-MONTAGUT & F. PANDOLFI* Pathology Research Laboratory,

Massachusetts General Hospital, Boston, MA, USA, and *Chair of Semeiotica Medica, Catholic University, Rome, Italy

(Accepted for publication 8 January 1999)

SUMMARY
In order to determine the mechanism of tumour destruction by tumour-infiltrating lymphocytes (TIL),
we examined the ability of both CD4þ and CD8þ effector TIL, and TIL clones, to manifest granzyme-
mediated and Fas-mediated destruction of tumour targets. In many in vitro studies TIL have been shown
to manifest anti-tumour reactivity, yet many tumours escape immunological destruction. To investigate
the role of Fas expression and the concomitant sensitivity to the inducibility of apoptotic death, we
derived TIL from four melanomas and one glioma. The glioma, and all but one of the melanomas,
expressed Fas, but Fas-mediated apoptosis could only be detected if the targets were treated with
cyclohexamide. The melanomas and the glioma all expressed detectable cytoplasmic Bcl-2 protein,
known to exert anti-apoptotic activity. Lysis of tumours by CD8-enriched cultures and CD8þ clones was
Ca2þ-dependent and could not be modified by an anti-Fas MoAb. In CD4-enriched cultures or CD4þ
clones with cytotoxic potential against tumour cells, cytotoxicity was also Ca2þ-dependent. As Ca2þ-
dependent cytotoxicity is usually the result of secretion of perforin/granzyme-B, we investigated the
presence of perforin in cytotoxic CD4þ clones and demonstrated the presence of granular deposits of
this enzyme in some of the CD4þ clones. Although an anti-Fas MoAb did not block the lysis of
melanoma targets by CD4þ clones, the examination of Fas-dependent targets demonstrated that these
clones also had the potential to kill by the Fas/Fas ligand system. These data suggest that the
predominant mechanism in tumour killing by TIL appears to be perforin–granzyme-dependent, and that
the solid tumour cell lines we studied are less susceptible to Fas-mediated apoptosis. As non-apoptotic
pathways may enhance tumour immunogenicity, exploitation of the perforin–granzyme-dependent
cytotoxic T lymphocyte (CTL) pathways may be important for achieving successful anti-tumour
responses.

Keywords tumour-infiltrating lymphocytes cytotoxicity perforin Fas/CD95 FasL Bcl-2

apoptosis

INTRODUCTION particular lytic pathways, may heavily influence the outcome of

the immune response to tumours and our ability to modulate this
Despite numerous investigations, the role of tumour-infiltrating
response. In this study we evaluate the contribution of perforin and
lymphocytes (TIL) in controlling tumour growth remains unclear.
Fas-mediated lysis to the recognition of autologous human tumours
The ability of some TIL cultures to exert strong cytotoxic activity
by the T lymphocytes which infiltrate them.
against autologous tumour cells suggests that such cells should be
Although the literature is replete with articles using murine
able to suppress tumour growth, but the ability of some tumours to
models, including knock-out and transgenic animals, which
progress, even when infiltrated with tumour-specific TIL, suggests
describe the contributions of perforin and Fas-mediated lysis in
that the mechanism of killing may be impaired in vivo. The
the rejection of allografts [2], in the pathogenesis of viral [3,4] and
question of the mechanism of killing manifested in vivo may
autoimmune diseases [5], and in a few instances against leukaemic
have particular importance. Melcher et al. [1] have recently
malignant cells [6,7], the question of which lytic pathways are
demonstrated that apoptotic killing may result in reduced tumour
taken by TIL in the destruction of solid tumours has rarely been
immunogenicity compared with lytic mechanisms which induce
addressed [8]. Thus, although there is much data to suggest that
heat shock protein and produce inflammation. Thus, the question of
tumours may escape from immune recognition by a variety of
how tumours are lysed, and whether there is impairment of
pathways, there is no evaluation of the contribution of the Fas
Correspondence: Professor F. Pandolfi, Clinical Immunology, Viale versus the perforin and granzyme pathways in the lysis of solid
dell’ Universita’ 37, 00185 Rome, Italy. tumours by human TIL. In the light of tumour progression taking
E-mail: pandolfi@axrma.uniroma1.it place even in the presence of a lymphoid infiltrate with

388 q 1999 Blackwell Science

 Mechanism of cytotoxicity by TIL
tumour-specific cells, a better understanding of these mechanisms
could help explain how tumours escape from immune recognition.
Since the Fas–FasL mechanism appears to be more sensitive to
activation-induced regulation, we reasoned that down-regulation
of this mechanism could be involved in the down-regulation of
cytotoxicity. We therefore investigated Fas expression, and the
susceptibility to Fas-mediated apoptosis in several tumour cell
lines, including the tumour cells’ expression of the anti-apoptotic
protein Bcl-2 [8–11]. In addition, we studied the mechanisms of
killing involved in both CD8þ and CD4þ effector TIL mediating
anti-tumour cytotoxicity.
Several studies have correlated the presence of TIL with a
relatively better prognosis in melanomas and gliomas [12–15]. In
malignant melanoma, it has been shown that TIL are able to
recognize tumour-associated antigens, and in some cases to lyse
the malignant cells [16–30]. These findings provided a basis for the
use of in vitro cultured TIL in adoptive immunotherapy in an effort
to enhance the in vivo accumulation of tumour-lytic lymphocytes
[29,31]. However, there is considerable doubt concerning the
efficacy of TIL in vivo, even in cases in which it is possible to
demonstrate the accumulation in vivo of clones capable of exerting
strong anti-tumour cytotoxicity in vitro. For example, our observa-
tion of a predominant tumour-specific cytotoxic T lymphocyte
(CTL) cell clone in multiple metastatic lesions from the same
patient indicated that such cells were able to propagate, circulate,
home and accumulate within tumour deposits. However, these
tumour-specific CTL clones were found in a patient with progres-
sive melanoma, indicating that they were not sufficient to contain
tumour growth [32]. Thus, the discrepancy between a demon-
strated in vitro tumour lytic activity, and the in vivo tumour
progression in the presence of this lymphoid infiltrate, remains to
be clarified.
Several mechanisms of lymphocyte-mediated cytotoxicity
have been described, including specific cytotoxicity mediated by
CTL, as well as natural killer (NK) and lymphokine-associated
killer (LAK) activities, antibody-dependent cell-mediated cyto-
toxicity, and cytokine-mediated tumour destruction [8,33–35].
Specific cytotoxicity requires T cell receptor (TCR) recognition
of tumour-associated antigens presented in the context of MHC
molecules as well as an array of accessory molecules which
mediate both target binding and the delivery of regulatory
‘second signals’. Current models indicated that CTL-mediated
lysis usually occurs through the release of cytolytic proteins
(perforin, granzymes) accumulated in the granules abundantly
present in the cytoplasm of CD8þ cells. Cell lysis then occurs in
a Ca2þ-dependent fashion [36].
More recently, however, it has been shown that CTL can also
kill targets via a Ca2þ-independent mechanism. This event is
mediated by the engagement of a surface molecule (Fas or
CD95) capable of delivering a lethal signal. Cross-linking of Fas,
physiologically mediated by cells expressing FasL or artificially
mediated by the addition of anti-Fas MoAbs or soluble FasL, can
be shown to induce apoptotic death [37,38]. Cell death via Fas-
binding is independent of extracellular Ca2þ [39] and does not
require active macromolecular synthesis [40,41].
FasL is expressed by activated CD8þ and CD4þ cells, thus
providing a basis for the assumption that CD8þ cells can kill by
both Ca2þ-dependent and -independent mechanisms. In contrast,
among CD4þ cells, most Th1 cells lack perforins [42–44] and thus
presumably kill primarily through the Fas-based mechanism.
CD4þ cells of the Th2 subset have little killing potential
q 1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 116:388–394
389
[45–48]. A few studies have pointed out, however, that CD4þ
cells may also lyse targets via a Ca2þ-dependent mechanism
[49–51]. Importantly, it has been demonstrated that some tumours
may be resistant to Fas-mediated killing, by virtue of Bcl-2 expression,
but remain susceptible to perforin-mediated lysis [8,10,11]
In this study, we have assessed the ability of TIL to mediate
tumour destruction by both Ca2þ-dependent and -independent
mechanisms. Although our data demonstrate the ability of many
TIL to manifest Fas-mediated target destruction, in fact the tumour
cells from which they were derived were relatively insusceptible to
Ca2þ-independent lysis, indicating that such mechanisms are
unlikely to play a major role in tumour destruction in vivo.
However, both CD8þ TIL, and some CD4þ TIL, were able to
manifest strong Ca2þ-dependent lysis of tumour targets, and
perforin-containing granules were detected in these lytic cells,
indicating that this is probably the predominant tumour lytic
activity among TIL.
MATERIALS AND METHODS
Cell lines
Five pairs of autologous tumour cell lines and TIL were derived:
four melanoma cell lines (M-1, M-3, M-5 and Mel-EW), one
glioma cell line (Gli-PI) and the corresponding TIL. M-1, M-3 and
M-5 melanomas and TIL have been previously characterized
[52,53]. Mel-EW was derived from a metastatic melanoma patient
and Gli-PI from a glioma. Cultures supplemented with IL-2
demonstrated propagation of activated T cells in those specimens
containing lymphocytic infiltrates. These cultures were kept in IL-2
and restimulated with phytohaemagglutinin (PHA) and irradiated
feeder cells every 2–4 weeks to enhance proliferation, as pre-
viously described [54]. Other cell lines used included Jurkat
(derived from a T cell leukaemia), U-937 (derived from a promo-
nocytic leukaemia) and K562 (from an erythroleukaemia). These
cell lines were obtained from ATCC (Rockville, MD).
Cloning of TIL cultures
T cell clones were obtained by limiting dilution as previously
reported [27,55]. Limiting dilution of IL-2-propagated TIL was
performed within 4 weeks from original establishment of the
culture.
Antibodies
TIL (bulk cultures and clones) were studied with a panel of MoAbs
which included anti-CD3, CD4, CD8 (Ortho Diagnostic System,
Raritan, NJ). MoAbs were used to stain cells in an indirect
immunofluorescence test followed by analysis using a Becton
Dickinson FACScan (Palo Alto, CA). Fas expression was detected
by anti-Fas CH-11 MoAb obtained from UBI (Lake Placid, NY).
Fas expression was also studied after pre-incubation of tumour
cells with interferon-gamma (IFN-g) as previously described in
detail [53]. FasL expression was detected on TIL activated for 3 h
in culture with PHA, anti-CD3 (purified OKT3 from Ortho, used at
appropriate dilutions) or phorbol myristate acetate (PMA) and
Ca-ionophore (ionomycin, obtained from Sigma (St Louis, MO)
and used at 3 mg/ml), with a polyclonal rabbit anti-FasL (N-20)
serum (Santa Cruz Biotechnology, Santa Cruz, CA), in an indirect
immunofluorescence test with an anti-rabbit immunoglobulin FITC
serum.
Anti-Fas IgM CH-11 antibody was also used to test
susceptibility to Fas-mediated lysis by tumour cell lines at the
390 M. Hishii et al.
concentration of 0·5 mg/ml. Anti-Fas MoAb (IgG ZB4; Immuno- RESULTS
tech, Westbrook, ME) was used at a concentration of 1 mg/ml to
Fas and Bcl-2 expression on tumour cell line
inhibit Fas-mediated lysis.
We evaluated Fas expression on eight tumour cell lines. These
Perforin accumulations were detected in the cytoplasm of
included Jurkat (derived from a T cell leukaemia), U-937 (derived
cytocentrifuged smears by immunofluorescence with an anti-
from a promonocytic leukaemia), K562 (from erythroleukaemia),
perforin antibody (obtained from Kamiya Biomedical Co.,
four melanoma cell lines (M-1, M-3, M-5 and Mel-EW) and one
Thousand Oaks, CA), and used at a concentration of 150 mg/ml.
glioma cell line (Gli-PI). Our data confirmed Fas expression on the
The percentage of stained cells was calculated from a minimum of
Jurkat and U937 cell lines, while K562 was negative. In addition,
200 cells counted, with cells scored as positive if they contained
we observed Fas expression on three melanoma cell lines (M-3, M-
distinctly fluorescent granules.
5, Mel-EW) and the Gli-PI glioma cell line, while M-1 cells were
Cytoplasmic staining with Bcl-2 was assessed using anti-Bcl-2
negative. Pre-incubation of the tumour cells with IFN-g had only
(Novacastra Labs, Newcastle upon Tyne, UK) using cells which
insignificant effects on Fas expression in all the cell lines tested
were first fixed for 10 min in 1% paraformaldehyde, followed by
(not shown). All of the melanomas and the glioma cell lines
membrane permeabilization with 0·1% saponin for 10 min prior to
showed detectable cytoplasmic Bcl-2, although levels were gen-
addition of the anti-Bcl-2 antibody for 45 min. Following two
erally low.
washes, FITC-labelled goat anti-mouse IgG was added for 30 min.
All steps were performed at room temperature (228C). Assessment
Fas-mediated apoptosis in tumour cells lines
of staining was made by FACScan flow cytometry (Becton
We then tested the ability of an anti-Fas MoAb (clone CH-11) to
Dickinson).
induce apoptosis of target cells. Tumour cell lines, incubated with
the anti-Fas antibody CH-11, were evaluated for apoptosis using PI
Cytotoxicity
and flow cytometry and by MTT assays. As shown in Table 1, the
Cytotoxic activity was assessed by the lysis of 51Cr-labelled target
MTT assay (data shown are representative of four separate assays)
cells in a 4-h Cr-release assay, as previously reported [56]. Per cent
revealed that Fas ligation was able to induce cell death in Jurkat
specific lysis was calculated as: (experimental ¹spontaneous) ×
(80%) and, to a lesser extent, in U-937 (32%) cells. Of the other
100/(maximum ¹ spontaneous), where all experimental points
cell lines tested, only 19% of the M-3 cell line population was
were calculated as the average of triplicate cultures, and the
killed, while the other cells tested showed very little susceptibility
spontaneous and maximum release by the target cells was the
to cytotoxicity (# 12%). PI staining confirmed that the cell death
average of six wells each. Cytotoxicity was determined at serial
observed in these cultures was due to apoptosis (data not shown).
dilutions of effector cells at ratios of 50:1, 25:1, 12·5:1 and 6·25:1
As it has been shown that Fas-induced apoptosis in certain
to determine the level at which maximum lysis could be achieved
tumour cells may be inhibited by short lived cytoplasmic proteins
for each target/effector pair. As maximum release was achieved
[59], we also evaluated the ability of CH-11 to kill tumour targets
using 25 effectors per target, the data shown are for this ratio only
after pre-incubation with CHX (an inhibitor of protein synthesis).
in order to simplify data presentation. In all cases the ‘spontaneous
Data show that, under these conditions, Fas was able to trigger
release’ (background) was < 15% of the maximum, including
increased apoptotic death in some of the cells lines (Fig. 1). (Data
cultures in which antibodies and cyclohexamide were added.
shown are representative of three separate assays).
Target cells included autologous tumour cell lines, as well as the
series of tumour lines described above. Effector cells were TIL
Mechanisms of killing involved in TIL-mediated cytotoxicity of
maintained in IL-2 (100 U/ml) for between 8 and 20 days after the
tumours. Ca2þ-dependent killing by CD8þ and CD4þ cells
previous restimulation with PHA and irradiated feeder cells. Assays
We first investigated the mechanisms involved in cell-mediated
were performed at multiple dilutions resulting in effector:target
cytotoxicity by CD8þ TIL. We studied a previously described
ratios of 50, 25, 12 and 6:1. As maximal lysis was apparent at ratios
of 25:1, results are reported as percentage of specific lysis at this
ratio. Effects on cytotoxicity by a calcium chelant were measured
after adding 2 mM EGTA (Sigma) to the cultures.
Jurkat
Detection of Fas-induced cell death
After treatment of the tumour cell lines with anti-Fas (CH-11, at
1 mg/ml), apoptosis was evaluated by flow cytometry as previously
described [57] with minor modifications. Briefly, the cells were
centrifuged and resuspended in 0·6 ml of propidium iodide (PI) M-3
hypotonic fluorochrome solution (PI 50 mg/ml in 0·1% sodium
citrate plus 0·1% NP-40; Sigma). The tubes were incubated over-
night at 48C in the dark before the flow cytometric analysis. The
PI fluorescence was measured using an Ortho Absolute flow
M-5
cytometer.
Assessment of cell death was also evaluated by the reduction of
tetrazolium salt, MTT, by living cells to form a blue formazan 0 20 40 60 80 100
product as previously reported [58].
Fas-induced apoptosis was also evaluated after coincubation of Fig. 1. Mortality rate (expressed as percentage of dead cells in the MTT
tumour cells with CH11 and cyclohexamide (CHX; used at a test) of tumour cell lines incubated with anti-Fas MoAb (CH-11) with or
concentration of 1 mg/ml) as previously reported [59]. without cyclohexamide (CHX). A, CH-11; B, CH-11 þ CHX.
q 1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 116:388–394
 Mechanism of cytotoxicity by TIL 391
Table 1. Fas and Bcl-2 expression (as determined by flow cytometry), and Table 2. Killing (51Cr-release assay) of autologous tumour by CD8þ
mortality rate (expressed of percentage of dead cells in the MTT test) of tumour-infiltrating lymphocytes (TIL) (bulk culture and three clones)
tumour cell lines incubated with anti-Fas MoAb (CH-11)

Effector cells Percent killing þ EGTA þ anti-Fas (ZB-4)

Cell lines Fas expression Fas-induced mortality Bcl-2 expression
MU-bulk 34 6 6 13 6 4 32 6 6
Jurkat þ 80 6 6 NT* Clone MU-63 32 6 8 15 6 3 38 6 8
U937 – 32 6 3 NT* Clone MU-79 27 6 5 20 6 6 29 6 5
K562 – 261 NT* Clone MU-115 27 6 6 11 6 4 32 6 6
M-1 – 861 þ
M-3 þ 19 6 2 þ
M-5 þ 961 þ
Mel EW þ 12 6 1 þ
Therefore, we evaluated the presence of perforin in the cytoplasm
Gli PI þ 561 þ
of six CD4þ clones (three from the glioma and three from
melanoma patient Mel-EW), by indirect immunofluorescence. In
*These cell lines were not tested (NT) in our assays, but these cell lines three out of six CD4þ clones examined, we observed the presence
have been reported to have minimal, or absent Bcl-2 expression. of scattered perforin staining in the cytoplasm in a granular
distribution (Fig. 2). The positive clones were glioma clones PI-
A and PI-H, and melanoma clone EW-2E5. This clone showed
CD8þ TIL cell line, derived from a melanoma patient, with strong 43% killing of the autologous tumour cells, but killing was not
cytotoxic capacity against autologous tumour cells (M-1 cell line). significantly inhibited by either EGTA or ZB-4. In the three
Some clones derived from this bulk culture were also studied (MU- positive CD4þ clones, perforin staining was strong in only 15–
63, MU-79, MU-115). These CD8þ clones kill the autologous 50% of the cells (scored þ or stronger (out of þ þ þ þ) among 200
tumours through the recognition of the Melan A/MART-1 cells counted per sample). In contrast, in CD8þ cultures the
presented on HLA-A2 (not shown). As expected, killing by the number of strongly perforin-positive cells was > 80% (most
bulk culture and the three derived clones was dependent on þ þ þ or þ þ þ þ). In addition, Giemsa staining revealed the
the presence of Ca2þ, as shown by the inhibition of cytotoxicity presence of the characteristic azurophilic cytoplasmic granules in
in the presence of the Ca2þ chelant EGTA, whereas it could not be the CD8þ cells, while in the CD4þ clones azurophilic granules
blocked by the addition of the anti-Fas MoAb ZB-4 (Table 2). were not detected. Glioma clones PI-D and melanoma clones
(Data shown are representative of three experiments.) EW-2E10 and 3F8 did not show detectable perforin.
The mechanisms involved in killing by CD4þ cells were
investigated in TIL bulk cultures with a CD4þ phenotype and in Fas-mediated killing by CD4 TIL
CD4þ clones derived from both a melanoma and a glioma patient. To determine if the CD4þ effectors derived from tumour lesions
TIL derived from a glioma patient, for which an autologous tumour also had the potential to kill by Fas engagement, we first evaluated
cell line (Gli-PI) was available, were shown to be predominantly the ability of TIL cell lines to express FasL. Bulk cultures,
CD4þ (63%). This TIL cell line showed only marginal lytic
capacity against the autologous tumour (5%, 10% and 15% in
three different experiments). Three CD4þ clones (PI-A, PI-D and
PI-H) derived from the glioma TIL also showed little cytotoxicity
(not shown).
The expression of MHC class II antigens on glioma tumour
cells was undetectable by staining with an anti-DR antibody and
class II expression could not be induced by pretreatment of the
glioma cells with IFN-g (data not shown). As HLA class II
molecules were not expressed on tumour cells, this precluded
efficient recognition of the target by CD4 effectors. Therefore,
we cross-linked effectors and target cells with PHA before per-
forming the cytotoxic assay. Under these conditions, both the bulk
culture and the three clones showed killing of the autologous
tumour. In clones PI-A, -D and -H killing was 57%, 35% and
87%, respectively. Killing was Ca2þ-dependent, as shown by
> 90% inhibition observed with EGTA, while marginal (< 10%)
effects were seen with the addition of the ZB-4 anti-Fas antibody
(representative of three separate assays).

Expression of perforin on CD4þ cells

Ca2þ-dependent killing manifested by CD4þ cells in the experi-
ments with PHA-cross-linked targets suggested that Fas–FasL-
mediated killing of tumour cells may not have been involved in the
mechanism of lysis by these cells. Instead, the Ca2þ-dependent
killing observed is more consistent with perforin-mediated lysis.
Fig. 2. Expression of perforin by tumour-infiltrating lymphocytes (TIL) by
indirect immunofluorescence on cytocentrifuged preparations. CD4þ clone
PI-H (derived from a glioma lesion) showed bright cytoplasmic staining in
approximately half of the cells; at higher magnification (inset) the cyto-
plasmic stain shows a granular pattern.

q 1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 116:388–394

392
maintained in IL-2 for 6 or more days after restimulation with PHA
and feeder cells, showed little expression of FasL. However, when
we tested the same cell lines 3 h after stimulation with either PMA
and ionomycin, anti-CD3 or PHA (without feeder cells), we
observed increased FasL expression in all the tested cell lines,
suggesting transient expression after stimulation (Fig. 3).
To assess the ability of CD4þ TIL cells to kill via Fas–FasL
interactions, we developed a model using a Fas-sensitive target. As
shown in Table 1, among the tumour cell lines derived in our
laboratory from fresh tumours, only the M-3 targets were suscep-
tible to some Fas-mediated apoptosis (19%). We therefore tested
the ability of the melanoma-derived TIL CD4þ cell line M-3 and
some of the Mel-EW CD4þ clones to induce killing of the Fas-
sensitive target cell lines Jurkat and M-3. The K562 cell line was
used as a Fas-insensitive control. FasL expression on effector cells
was enhanced by pre-incubation with anti-CD3 antibodies for 3 h
before performing the cytotoxic assay (representative of duplicate
experiments). As shown in Fig. 4, the M3 (Fig. 4a) and Jurkat
targets (Fig. 4b) were lysed in the presence of anti-CD3 antibody
with or without anti-Fas or EDTA (P < 0·05 versus IL-2 alone). In
contrast, K562 cells (Fig. 4c) were not lysed significantly by the
EW clones, even in the presence of anti-CD3 antibody and added
anti-Fas and EGTA. The M-3 TIL bulk had anti-K562 activity even
in IL-2, which was not enhanced significantly by anti-CD3 or anti-
Fas antibody, indicating the Fas-independent lysis of this target.
Cells cultured in IL-2 alone (which expressed low levels of
FasL, see Fig. 3) showed no killing of the M-3 target (Fig. 4a).
With Jurkat (Fig. 4b) and K562 (Fig. 4c) targets, the M-3 TIL bulk
cultured in IL-2 alone showed some cytotoxic activity. After
activation with anti-CD3, M-3 cells and two EW clones (EW-
2E5 and EW-2E10) could lyse the Jurkat (Fig. 4b) and (to a lesser
degree), the M-3 targets (Fig. 4a). Killing was consistent with the
differential Fas sensitivity of Jurkat and M3 cell lines (Table 1).
Killing of Jurkat by M-3 bulk and EW clones was inhibited by
anti-Fas antibody, but EGTA induced little or no inhibition
(Fig. 4b). On the other hand, lysis of the M-3 target by M-3 bulk
and clone EW-2E5 appeared to be mediated by both Ca2þ- and
Fas-dependent mechanisms, since some inhibition could be
M. Hishii et al.
observed with both EGTA and anti-Fas (Fig. 4a). Lysis of this
target by clone 2E10 could not be blocked by either EGTA or anti-
Fas, thus precluding a delineation of the mechanism involved in
this interaction. Killing of the Fas-resistant K562 target by M-3
bulk could be inhibited only by EGTA, and not by anti-Fas,
confirming that killing of this target occurred in a Ca2þ-dependent,
Fas-independent fashion.
DISCUSSION
In this study we investigated the cytotoxic mechanisms involved in
tumour killing by TIL derived from melanoma or glioma patients.
Some TIL cultures or clones (expressing either the CD8þ or the
CD4þ phenotype) were able to exert cytotoxic activity against
autologous targets. The CD8þ TIL, and some of the CD4þ TIL
showed cytoplasmic perforin-containing granules and lysed their
targets in a Ca2þ-dependent fashion. In contrast, our tumour lines
were generally insensitive to Fas-mediated lysis, although several
of our fresh tumour-derived cell lines expressed Fas. As these
tumours also expressed cytoplasmic Bcl-2, it is possible that their
insensitivity to Fas-mediated destruction was due to the anti-
apoptotic activity of this protein.
It is not surprising that the cytotoxicity manifested by the
CD8þ TIL was shown to be Ca2þ-dependent. However, most of the
(a) Target: M3 tumour
Anti-CD3+anti-Fas
Clone EW-2E10 Anti-CD3+EGTA
Anti-CD3
IL-2 alone
Clone EW-2E5
M-3 TIL bulk
0 10 20 30 40 50 60
(b) Target: Jurkat
Clone EW-2E10

80 Clone EW-2E5

IL-2
PMA and M-3 TIL bulk
Percent FasL+ cells

60 ionophore
anti-CD3 0 10 20 30 40 50 60
PHA (c) Target: K-562
40
Clone EW-2E10
20
Clone EW-2E5

0
EW-TIL PI-TIL MU-TIL M-3 TIL bulk
(CD4+) (CD4+) (CD8+)
Fig. 3. FasL expression by tumour-infiltrating lymphocytes (TIL). Sample 0 10 20 30 40 50 60
1 is a CD4þ bulk population derived from a melanoma lesion (EW). Sample Fig. 4. Killing of the Fas-sensitive targets (Jurkat and M-3) and Fas-
2 is a CD4þ bulk population derived from a glioma lesion (PI). Sample 3 is resistant target (K-562) by CD4þ tumour-infiltrating lymphocyte (TIL)
a CD8þ bulk population derived from a melanoma lesion (MU). Cells in IL- effectors kept in IL-2 at least 6 days after restimulation (IL-2 alone), or
2 were stained 6 days after restimulation with phytohaemagglutinin (PHA) preactivated with anti-CD3 for 3 h. Anti-Fas (ZB-4) or EGTA were added
and feeder. For other conditions, cells were stained 3 h after restimulation to these CD3-stimulated cultures to determine the effect of these agents on
with phorbol myristate acetate (PMA)/ionophore, anti-CD3, or PHA. tumour lysis.
q 1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 116:388–394
 Mechanism of cytotoxicity by TIL
available literature argues that CD4þ cells, which are commonly
devoid of perforin-containing cytoplasmic granules, kill targets
through a Ca2þ-independent, Fas-mediated mechanism [42–48].
Our data demonstrate that killing by CD4þ bulk cultures and
clones (derived from both a melanoma and a glioma) could
occur in a Ca2þ-dependent fashion, and was not affected by anti-
Fas antibody. In fact, the demonstration of perforin-containing
granules in a proportion of these CD4þ clones adds a novel model
for TIL/tumour interaction, in which CD4þ cells kill in a Fas-
independent fashion.
Although our data suggest that CD4þ clones do have the
potential to kill by the Fas/FasL mechanisms, it appears that the
autologous tumour targets we tested are not susceptible to this lytic
activity. CD4þ TIL can easily express FasL upon activation and
can kill Fas-sensitive targets (such as the Jurkat cell line) by a
mechanism which can be inhibited by anti-Fas antibodies. How-
ever, this potential activity did not take place in the interactions of
TIL with the autologous tumours we investigated. This conclusion
was further supported by the fact that the melanoma and glioma
tumour cell lines we derived were also resistant to Fas-mediated
killing by the apoptosis-inducing antibody, CH-11. This antibody
had activity only when the target cells were treated with CHX. The
staining of these tumours with Bcl-2 antibody offers a likely
explanation for this lack of apoptosis induction [11], further
emphasizing the need for Ca2þ-dependent mechanisms of
tumour lysis to effect tumour destruction.
In conclusion, our data suggest that the Ca2þ-dependent,
perforin-associated mechanism appears to be the principle
cytotoxic pathway used by TIL-tumour pairs, although we
cannot exclude other lytic mechanisms, including cytokine-
induced tumour destruction [34], which could also play a role in
TIL–tumour interactions. However, it is noteworthy that when
Kuwano et al. blocked cytokine-induced killing, the ability to lyse
by a perforin–granzyme mechanism was retained [35]. The lack
of Fas-dependent lysis in our assays suggests that this is unlikely
to play a major role in tumour destruction, but does not preclude
another, perhaps counterproductive role for Fas. For example,
it has been observed that melanoma cell lines can express FasL,
and it has been proposed that they can destroy TIL by this mech-
anism, contributing to the immune privilege of tumours [60,61].
However, when we tested three available melanoma cell lines
(M-1, M-5, Mel-EW), all were FasL¹. Thus, the killing of TIL by
FasL-expressing tumour cells did not play a role in the tumour–
TIL interactions we studied, as TIL were not destroyed by contact
with tumour cells, and were able to deliver an efficient cytotoxic
signal when the TCR was engaged and perforin-mediated lysis
could occur. With the recent demonstration by Melcher [1] that
apoptotic killing of tumour cells may render them less immuno-
genic, it is ironic that the tumour killing we have observed was defi-
cient in Fas-mediated lysis, indicating that any killing which does
occur is more likely to be via conventional cell lysis. The fact that
we were able to isolate tumour-specific CTL from these tumours
perhaps reflects the fact that these tumours are indeed, at least
partially, immunogenic. However, the ultimate failure of these
CTL to control tumour growth remains a hurdle to be surmounted.
ACKNOWLEDGMENTS
This work was supported in part by NIH (grants HL-43793 and CA51345),
NATO (grant CRG940029), AIRC (Milan) and MURST-COFIN (Italy).
q 1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 116:388–394
393
REFERENCES
1 Melcher A, Todryk S, Hardwick N, Ford M, Jacobson M, Vile R. Tumor
immunogenicity is determined by the mechanism of cell death via
induction of heat shock protein expression. Nature Med 1998; 4:581–7.
2 Muller K, Mariani S, Matiba B, Kyewski B, Krammer P. Clonal
deletion of major histocompatibility complex class I-restricted
CD4þCD8þ thymocytes in vitro is independent of the CD95 (APO-
1/Fas) ligand. Eur J Immunol 1995; 25:2996–9.
3 Lowin B, Hahne M, Mattmann C, Tschopp J. Cytolytic T-cell
cytotoxicity is mediated through perforin and Fas lytic pathways.
Nature 1994; 370:650–2.
4 Bachmann M, Ohteki T, Faienza K, Zakarian A, Kagi D, Speiser D,
Ohashi P. Altered peptide ligands trigger perforin- rather than Fas-
dependent cell lysis. J Immunol 1997; 159:4165–70.
5 Fukuyama H, Adachi M, Suematsu S, Miwa K, Suda T, Yoshida N,
Nagata S. Transgenic expression of Fas in T cells blocks lymphopro-
liferation but not autoimmune disease in MRL-lpr mice. J Immunol
1998; 160:3805–11.
6 Dilloo D, Brown M, Roskrow M, Zhong W, Holladay M, Holden W,
Brenner M. CD40 ligand induces an antileukemia immune response in
vivo. Blood 1997; 90:1927–33.
7 Tsujimura K, Takahashi T, Iwase S, Matsudaira Y, Kaneko Y, Yagita
H, Obata Y. Two types of anti-TL (thymus leukemia) CTL clones with
distinct target specificities: differences in cytotoxic mechanisms and
accessory molecule requirements. J Immunol 1998; 160:5253–61.
8 Jaattela Benedict M, Tewari M, Shayman J, Dixit V. Bcl-x and Bcl-2
inhibit TNF and Fas-induced apoptosis and activation of phospholipase
A2 in breast carcinoma cells. Oncogene 1995; 10:2297–305.
9 Panayiotidis P, Ganeshaguru K, Foroni L, Hoffbrand A. Expression and
function of the Fas antigen in B chronic lymphocytic leukemia and
hairy cell leukemia. Leukemia 1995; 9:1227–32.
10 Lee R, Spielman J, Podack ER. Bcl-2 protects against Fas-based but not
perforin-based T cell-mediated cytolysis. Int Immunol 1996; 8:991–
1000.
11 Schroter M, Lowin B, Borner C, Tschopp J. Regulation of Fas (Apo-1/
CD95)- and perforin-mediated lytic pathways of primary cytotoxic T
lymphocytes by the protooncogene bcl-2. Eur J Immunol 1995;
25:3509–13.
12 Clark WH, Elder DE, Guerry D et al. Model predicting survival in stage
I melanoma based on tumor progression. J Natl Cancer Inst 1989;
81:1893–904.
13 Day C, Lew R, Mihm M et al. A multivariate analysis of prognostic
factors for melanoma patients with lesions >3·65 mm in thickness. Ann
Surg 1982; 195:44.
14 Mihm M, Clemente C, Cascinelli N. Tumor infiltrating lymphocytes in
lymph node melanoma metastases: a histopathologic prognostic
indicator and an expression of local immune response. Lab Invest
1996; 74:43–47.
15 Ridley A, Cavanagh JB. Lymphocytic infiltration in gliomas: evidence
of possible host resistance. Brain 1971; 94:117–24.
16 Mukherji B, Guha A, Chakraborty G, Sivanandham M, Nashed A,
Sporn J, Ergin M. Clonal analysis of cytotoxic and regulatory T cell
responses against human melanoma. J Exp Med 1989; 169:1961–76.
17 Itoh K, Platsoucas CD, Balch CM. Autologous tumor-specific cytotoxic
T lymphocytes in the infiltrate of human metastatic melanomas.
Activation by interleukin 2 and autologous tumor cells, and
involvement of the T cell receptor. J Exp Med 1988; 168:1419–41.
18 Sensi M, Salvi S, Castelli C et al. T cell receptor (TCR) structure of
autologous melanoma-reactive cytotoxic T lymphocyte (CTL) clones:
tumor-infiltrating lymphocytes overexpress in vivo the TCR beta chain
sequence used by an HLA-A2-restricted and melanocyte-lineage-
specific CTL clone. J Exp Med 1993; 178:1231–46.
19 Sensi M, Traversari C, Radrizzani M et al. Cytotoxic T-lymphocyte
clones from different patients display limited T-cell-receptor variable-
region gene usage in HLA-A2-restricted recognition of the melanoma
antigen Melan-A/MART-1. Proc Natl Acad Sci USA 1995; 92:5674–8.
394 M. Hishii et al.
20 Kurnick J, Kradin R, Blumberg R, Schneeberger E, Boyle L. Functional
characterization of T lymphocytes propagated from human lung
carcinomas. Clin Immunol Immunopathol 1986; 38:367–80.
21 Platsoucas C. Human autologous tumor-specific T cells in malignant
melanoma. Cancer Metastasis Rev 1991; 10:151–76.
22 Caignard A, Dietrich P, Morand V et al. Evidence for T-cell clonal
expansion in a patient with squamous cell carcinoma of the head and
neck. Cancer Res 1994; 54:1292–7.
23 Gervois N, Heuze F, Diez E, Jotereau F. Selective expansion of a
specific anti-tumor CD8þ cytotoxic T lymphocyte clone in the bulk
culture of tumor infiltrating lymphocytes from a melanoma patient:
cytotoxic activity and T cell receptor gene rearrangements. Eur J
Immunol 1990; 20:825–31.
24 Ioannides C, Freedman R. Selective usage of TCR V beta in tumor-
specific CTL lines isolated from ovarian tumor-associated lymphocytes.
Anticancer Res 1991; 11:1919–25.
25 Mackensen A, Ferradini L, Carcelain G, Triebel F, Faure F, Viel S,
Hercend T. Evidence for in situ amplification of cytotoxic T-lympho-
cytes with antitumor activity in a human regressive melanoma. Cancer
Res 1993; 53:3569–73.
26 Miescher S, Whiteside T, Moretta L, VonFliedner V. Clonal and
frequency analyses of tumor-infiltrating T lymphocytes from human
solid tumors. J Immunol 1987; 138:4004.
27 Pandolfi F, Boyle L, Trentin L, Oliva A, Kurnick J. T cell receptor gene
rearrangement and cytotoxic activities of clones isolated from tumor
infiltrating lymphocytes from melanoma patients. Clin Exp Immunol
1994; 95:141–7.
28 Peoples GE, Schoof DD, Andrews JV, Goedegebuure PS, Eberlein TJ.
T-cell recognition of ovarian cancer. Surg 1993; 114:227–34.
29 Rosenberg S, Packard B, Aebersold P. Use of tumor-infiltrating
lymphocytes and interleukin 2 in the immunotherapy of patients with
metastatic melanoma. N Eng J Med 1988; 25:1676–80.
30 Wong JT, Pinto CE, Gifford JD, Kurnick JT, Kradin RL. Charac-
terization of the CD4þ and CD8þ tumor infiltrating lymphocytes
propagated with bispecific monoclonal antibodies. J Immunol 1989;
143:3404–11.
31 Kradin R, Kurnick J, Lazarus D et al. Tumour-infiltrating lymphocytes
andinterleukin-2in treatment ofadvanced cancer. Lancet 1989;i:577–80.
32 Hishii M, Andrews D, Boyle L, Wong J, Pandolfi F, van den Elsen P,
Kurnick J. In vivo accumulation of the same anti-melanoma T cell clone
in two different metastatic sites. Proc Natl Acad Sci USA 1997;
94:1378–83.
33 Berke G. The binding and lysis of target cells by cytotoxic lymphocytes.
Annu Rev Immunol 1994; 12:735–73.
34 Hehner S, Hofmann T, Ratter F, Dumont A, Droge W, Schmitz M.
Tumor necrosis factor-alpha-induced cell killing and activation of
transcription factor NF-kappaB are uncoupled in L929 cells. J Biol
Chem 1998; 273:18117–21.
35 Kuwano K, Akashi A, Arai S. An anergic cytotoxic T lymphocyte clone
exhibits granule exocytosis-mediated cytotoxicity. Cell Immunol 1998;
185:114–22.
36 Henkart P. Lymphocyte-mediated cytotoxicity: two pathways and
multiple effector molecules. Immunity 1994; 1:343–6.
37 Suda T, Nagata S. Purification and characterization of the Fas-ligand
that induces apoptosis. J Exp Med 1994; 179:873–9.
38 Owen JJ, Jenkinson EJ. Apoptosis and T-cell repertoire selection in the
thymus. Ann NY Acad Sci 1992; 663:305–10.
39 Rouvier E, Luciani MF, Golstein P. Fas involvement in Ca(2þ)-
independent T cell-mediated cytotoxicity. J Exp Med 1993; 177:195–
200.
40 Itoh N, Yonehara S, Ishii A et al. The polypeptide encoded by the
cDNA for human cell surface antigen Fas can mediate apoptosis. Cell
1991; 66:233–43.
41 Stalder T, Hahn S, Erb P. Fas antigen is the major target molecule for
CD4þ T cell-mediated cytotoxicity. J Immunol 1994; 152:1127–33.
q 1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 116:388–394
42 Strack P, Martin C, Saito S, Dekruyff R, Ju S. Metabolic inhibitors
distinguish cytolytic activity of CD4 and CD8 clones. Eur J Immunol
1990; 20:179–83.
43 Lancki D, Hsieh C, Fitch F. Mechanisms of lysis by cytotoxic T
lymphocyte clones. J Immunol 1991; 146:3242–9.
44 Ozdemirli M,El-Khatib M, Bastiani L, Akdeniz H, Kuchroo V, Ju S. The
cytotoxic process of CD4 Th1 clones. J Immunol 1992; 149:1889–95.
45 Ju ST, Cui H, Panka DJ, Ettinger R, Marshak RA. Participation of target
Fas protein in apoptosis pathway induced by CD4þ Th1 and CD8þ
cytotoxic T cells. Proc Natl Acad Sci USA 1994; 91:4185–9.
46 Ramsdell F, Seaman MS, Miller RE, Picha KS, Kennedy MK, Lynch
DH. Differential ability of Th1 and Th2 T cells to express Fas ligand
and to undergo activation-induced cell death. Intern Immunol 1994;
6:1545–53.
47 Vignaux F, Vivier E, Malissen B, Depraetere V, Nagata S, Golstein P.
TCR/CD3 coupling to Fas-based cytotoxicity. J Exp Med 1995;
181:781–6.
48 Anel A, Buferne M, Boyer C, Schmitt VA, Golstein P. T cell receptor-
induced Fas ligand expression in cytotoxic T lymphocyte clones is
blocked by protein tyrosine kinase inhibitors and cyclosporin A. Eur J
Immunol 1994; 24:2469–76.
49 Grogg D, Hahn S, Erb P. CD4þ T cell mediated killing of major
histocompatibility complex class II-positive antigen-presenting cells.
III. CD4þ cytotoxic T cells induce apoptosis of APC. J Immunol 1992;
22:267–79.
50 Miskovsky E, Liu A, Pavlat W et al. Studies of the mechanism of
cytolysis by HIV-1 specific CD4þ human CTL clones induced by
candidate AIDS vaccines. J Immunol 1994; 153:2787–99.
51 Susskind B, Shornick M, Iannotti M, Duffy B, nee Tanden P, Siegel J,
Mohanakumar T. Cytolytic effector mechanisms of human CD4þ
cytotoxic T lymphocytes. Human Immunol 1996; 45:64–75.
52 Pandolfi F, Boyle LA, Trentin L, Kurnick JT, Isselbacher KJ, Gattoni
Celli S. Expression of HLA-A2 antigen in human melanoma cell lines
and its role in T-cell recognition. Cancer Res 1991; 51:3164–70.
53 Pandolfi F, Trentin L, Boyle LA, Stamenkovic I, Byers HR, Colvin RB,
Kurnick JT. Expression of cell adhesion molecules in human melanoma
cell lines and their role in lymphocyte mediated cytotoxicity. Cancer
1992; 69:1165–73.
54 Kurnick J, Kradin R, Blumberg R, Schneeberger E, Boyle L.
Functional characterization of T lymphocytes propagated from
human lung carcinomas. Clin Immunol Immunopathol 1986; 38:367–
80.
55 Kurnick J, Hayward A, Altevogt P. Helper and suppressor/inducer
activity of human T cells and their cloned progeny maintained in long-
term culture. J Immunol 1981; 126:1307.
56 Pandolfi F, Strong DM, Slease RB, Smith ML, Ortaldo JR, Herberman
RB. Characterization of a suppressor T-cell chronic lymphocytic
leukemia with ADCC but not NK activity. Blood 1980; 56:653–60.
57 Nicoletti I, Migliorati G, Pagliacci MC, Grignani F, Riccardi C. A rapid
and simple method for measuring thymocyte apoptosis by propidium
iodide staining and flow cytometry. J Immunol Methods 1991;
139:271–9.
58 Denizot F, Lang R. Rapid colorimetric assay for cell growth and survival.
Modifications to the tetrazolium dye procedure giving improved
sensitivity and reliability. J Immunol Methods 1986; 89:271–7.
59 Weller M, Frei K, Groscurth P, Krammer P, Yonekawa Y, Fontana A.
Anti-Fas/Apo-1 antibody-mediated apoptosis of cultured human glioma
cells. J Clin Invest 1994; 94:954–64.
60 Hahne M, Rimoldi D, Schroter M et al. Melanoma cell expression of
Fas (Apo-1/CD95) ligand: implication for tumor escape. Science 1996;
274:1363–6.
61 Yamauchi A, Taga K, Mostowski HS, Bloom ET. Target cell-
induced apoptosis of interleukin-2-activated human natural killer
cells: roles of cell surface molecules and intracellular events. Blood
1996; 87:5127–35.