S

K-T gap, rarefaction analysis indicates tol. 17, 319 (1991); G. Keller, ibid. 13, 239 (1988); 24. K. G. MacLeod and W. N. Orr, Paleobiology 19, 235
GeoL Soc. Am. Bull. 101, 1408 (1989). (1993).
that anywhere from 3 to all 10 of the 8. C. R. Marshall, Paleobiology 16,1 (1990). 25. A. V. Dhondt, Zitteliana 10, (1983).
missing ammonites may have become ex- 9. ,_ ibid. 20, 459 (1994). 26. K. G. MacLeod, Geology 22, 139 (1994).
tinct before the last 1.5 m of the Creta- 10. D. Strauss and P. M. Sadler, Math. Geol. 21, 411 27. D. M. Raup, Paleobiology 1, 333 (1975).
ceous (P = 0.05). We are unable to dis- (1989). 28. S. H. Hurlbert and J. D. Archibald, Geology 23, 881
11. M. S. Springer, Paleobiology 16, 512 (1990). (1995).
tinguish between these two end-member 12. C. R. Marshall, Geology 23, 731 (1995). 29. T. Birkelund, Bull. Geol. Soc. Den. 40, 33 (1993).
possibilities. Thus, there was at least a 13. J. Smit and A. J. T. Romein, Earth Planet. Sci. Lett. 30. The absence of ammonites in the marls of Member V
minor biological change, and perhaps a 74,155 (1985). is unlikely to be due to preservation failure, given that
14. J. F. Mount and P. Ward, J. Sediment. Petrol. 56, ammonites are found in the similar marls of Member
fairly large extinction event, associated 228 (1986). ll, which occurs lower in the section.
with the regression that peaked shortly 15. K. G. MacLeod and P. D. Ward, Geol. Soc. Am. 31. Ammonite fossils have been found, on average,
before the K-T boundary. The fossil record Spec. Pap. 247, 509 (1990). once every 18 m2 of outcrop in the top 1.3 m of the
16. P. D. Ward, W. J. Kennedy, K. G. MacLeod, J. F.
from outside the Bay of Biscay basin does Mount, Geology 19, 1181 (1991). Cretaceous at Hendaye, France. Echinoids are re-
not help distinguish between these possi- 17. B. Mathey, in Palaeontology and Evolution: Extinc- covered at a comparable rate of one specimen per
19 m2 of outcrop. At least one echinoid per collect-
bilities, as only one of the 10 missing tion, M. A. Lamolda, E. G. Kauffman, 0. H. Walliser, ing season, but no ammonites over seven seasons,
Eds. (Sociedad Espanola de Paleontologia, Madrid,
species is known to have survived to the Spain, 1988), pp. 142-147. has been recovered in about 50 m2 of outcrop
K-T boundary elsewhere [Hoploscaphites 18. P. D. Ward and W. J. Kennedy, J. Paleontol. Mem. excavated within the pre-K-T gap (1.3 to 3.0 m
below the K-T boundary). At these recovery rates,
constrictus in Denmark (29)]. 34,1 (1993). the probability that ammonites were as abundant in
19. Based on graphic correlation of the major lithological
The ammonites in the Bay of Biscay are boundaries. the excavated interval as they were above the pre-
K-T gap is only 0.06; however, the probability that

Downloaded from www.sciencemag.org on December 17, 2014

preserved as impressions only and hence 20. Some of the echinoid species present in the Bay of the ammonites were a third as abundant in the
can only be seen on bedding surfaces. Biscay Cretaceous sections are also known from the excavated interval as they were above the pre-K-T
Within the marly pre-K-T gap there are Tertiary (A. B. Smith, personal communication). gap is 0.39.
21. K. G. MacLeod, J. Paleontol. 68,1048 (1994).
essentially no bedding surfaces exposed. 22. Forty-two diagnosable ammonite fossils are known 32. We thank A. Smith, M. Foote, D. Jablonski, D. Ja-
cobs, B. Runnegar, and two anonymous reviewers
Hence, it is difficult to confirm that there from the interval (about 30 fossils per meter of sec-
for their comments. Supported by NSF grant EAR-
was a drop in abundance, if not diversity, tion), compared with a recovery of 1.5 diagnosable 9258045 to C.R.M. and by EAR-8904797 and EAR-
ammonite fossils per meter over the preceding
during the interval, or whether their ab- 200 m of section. 8843926 to P.D.W.
sence in the pre-K-T gap simply reflects 23. D. M. Raup, Philos. Trans. R. Soc. London B 325,
collection failure (30). However, near 421 (1989). 29 May 1996; accepted 3 September 1996
Hendaye in southern France, it is relative-
ly easy to excavate bedding surfaces in the
top 2 m of the pre-K-T gap. Quantitative
analysis of collecting intensity at this lo-
Melanoma Cell Expression of Fas(Apo-1/CD95)
cality does indeed support the hypothesis Ligand: Implications for Tumor Immune Escape
that, minimally, the abundance of ammo-
nites within the pre-K-T gap dropped in M. Hahne, D. Rimoldi, M. Schr6ter, P. Romero, M. Schreier,
comparison with the last 1.5 m of the L. E. French, P. Schneider, T. Bornand, A. Fontana, D. Lienard,
Cretaceous (31). J.-C. Cerottini, J. Tschopp*
Of the 13 molluscan species known
from the last 1.5 m of the Cretaceous, only
six are known from two or more fossils: Malignant melanoma accounts for most of the increasing mortality from skin cancer.
five ammonites and T. argentea. Statistical Melanoma cells were found to express Fas (also called Apo-1 or CD95) ligand (FasL). In
analysis of the fossil records of these six metastatic lesions, Fas-expressing T cell infiltrates were proximal to FasL+ tumor cells.
species failed to reject the null hypothesis In vitro, apoptosis of Fas-sensitive target cells occurred upon incubation with melanoma
that they became extinct at the K-T tumor cells; and in vivo, injection of FasL+ mouse melanoma cells in mice led to rapid
boundary (Fig. 4). Thus, using the results tumor formation. In contrast, tumorigenesis was delayed in Fas-deficient Ipr mutant mice
of the six species as a proxy for all 13 in which immune effector cells cannot be killed by FasL. Thus, FasL may contribute to
found in the last 1.5 m of the Cretaceous, the immune privilege of tumors.
as well as any of the 10 missing ammonite
species that may have survived the pre-
K-T gap, we cannot reject the hypothesis
that all these species became extinct as a Membrane-bound FasL (mFasL) induces lytic pathways used by cytolytic T cells to
result of the impact. However, the data are rapid cell death of Fas-sensitive cells (1). kill target cells (2-4), but is also a key
also consistent with a range of gradual FasL is not only one of three major cyto- element in the elimination of activated T
extinction scenarios (12) not enumerated cells during the downregulation of the im-
here. M. Hahne, M. Schroter, P. Schneider, T. Bornand, J. mune response (5, 6). Similar to the struc-
Tschopp, Institute of Biochemistry, University of Lau- turally homologous tumor necrosis factor
sanne, CH-1 066 Epalinges, Switzerland.
REFERENCES AND NOTES D. Rimoldi, P. Romero, M. Schreier, J.-C. Cerottini, Lud- a (TNFcx) (7, 8), processing of the mem-
1. D. Jablonski and D. M. Raup, Science 268, 389 wig Institute of Cancer Research, Lausanne branch, Uni- brane-bound form of FasL by metallopro-
versity of Lausanne, CH-1 066 Epalinges, Switzerland. teases results in shedding of the extracel-
(1995). L. E. French, Department of Dermatology, University of
2. V. L. Sharpton et al., Nature 359, 819 (1992). Geneva, Medical School, CH-1 211 Geneva 4, Switzerland. lular portion (sFasL) (9-11). Patients with
3. L. W. Alvarez, W. Alvarez, F. Asaro, H. V. Michel, A. Fontana, Clinical Immunology, Department of Inter- diseases characterized by pathological cell
Science 208,1095 (1980). nal Medicine, University Hospital, CH-8044 ZOrich,
4. D. A. Russell, Annu. Rev. Earth Planet. Sci. 7, 163
Switzerland.
death, such as alcoholic hepatitis, contain
(1979). D. Lienard, Centre Pluridisciplinaire d'Oncologie, CHUV, high concentrations of sFasL in their se-
5. D. Jablonski, Science 253, 754 (1991). CH-1011 Lausanne, Switzerland. rum (12). In the process of screening se-
6. P. W. Signor and J. H. Lipps, Geol. Soc. Am. Spec.
Pap. 190, 291 (1982). *To whom correspondence should be addressed. E-mail: rum samples from patients with other dis-
7. J. l. Canudo, G. Keller, E. Molina, Mar. Micropaleon- jurg.tschopp@ib.unil.ch eases, elevated concentrations of sFasL
SCIENCE * VOL. 274 * 22 NOVEMBER 1996 1 363
 !l,iRf|M5gBbs~EI_U 0-.
*'i9;lfi*'**'~'Si
;Zt
g
~XKXAmS'e'"'
2 i K'
*~*&0+^'¾gj " t ~ S*7:~:7~ 7 ..'. .
7Y0 '.; 'lS 't ''2X '':.7.' 7'''t<i''
':';
'*''t0'0 ''

were also observed in sera from 18 of 35 tase-polymerase chain reaction (RT- FasL expression on melanoma with the
patients with malignant melanoma (Fig. PCR) ( 9). Melanoma-expressed FasL was severity of the disease.
1A). active, as Fas-sensitive A20 B lymphoma The expression of FasL expression may
Although FasL expression was initially cells, but not the FasL-resistant A20R help to maintain the integrity of immune
thought to be restricted to activated T cells (20), underwent apoptosis when cul- privileged sites. In many of these sites, includ-
lymphocytes (13), the ligand is also ex- tured with melanoma cells (Fig. IC). Lit- ing the eye, inner ear, testis, or brain, consti-
pressed in nonlymphoid cells such as cer- tle or no Fas was expressed by these mel- tutive FasL expression is observed (14, 16),
tain epithelial cells, Sertoli cells, and neu- anoma cells (19); they were also resistant and FasL+ cells may kill Fas-sensitive T cells
rons (14-16). We therefore considered to apoptosis mediated by recombinant entering such sites (15, 16). In this con-
the possibility that sFasL detected in these FasL (Fig. ID). text, FasL-expressing melanoma cells might
sera was derived from cleavage of melano- To exclude the possibility that FasL induce apoptosis of Fas-sensitive, tumor-
ma mFasL and not exclusively from acti- expression in melanoma cell lines was sim- infiltrating cells. With the use of the ter-
vated T cells directed against the tumor ply a consequence of prolonged in vitro minal deoxytransferase-mediated deoxyuri-
(17, 18). Substantial quantities of FasL culture conditions, FasL expression was dine 5'-triphosphate nick end labeling
were found in lysates of a series of human studied by immunohistochemistry in sec- (TUNEL) technique that detects single
melanoma cell lines (Fig. I B). As in jurkat tions of metastatic melanoma lesions from strand DNA breaks (21), apoptotic cells
T cells, two molecular species, one corre- seven individuals. In all patients, tumor were identified in regions where tumor-
sponding to mFasL (40 kD) and the other cells expressed FasL but not Fas (Fig. 2, A infiltrating T lymphocytes were found (Fig.
to sFasL (27 kD), were detected in some to F). In contrast, the majority of cells 3, A and B).
melanoma cells by Western blot analysis, infiltrating the tumor mass were Fas-posi- If FasL-expressing melanomas could
indicating that cellular FasL can consist of tive (Fig. 2C). No FasL was found in kill infiltrating immune effector cells
homotrimeric (mFasL3) as well as partly normal melanocytes of the skin, indicating through FasL-Fas interaction, cells from
processed heterotrimeric complexes that FasL upregulation probably occurs lpr mice that express little or no Fas (1)
(mFasL2-sFasL or mFasL1-sFasL2). FasL during tumorigenesis (Fig. 2, G and H). would be expected to be less sensitive to
expression by the human melanoma cell However, the limited number of patients FasL-dependent apoptosis and, conse-
lines was confirmed by reverse transcrip- studied does not yet permit us to correlate quently, be more efficient in the elimina-
tion of tumor cells. We tested this hypoth-
esis by subcutaneously injecting B16-F1O
o_ CD
(22) mouse melanoma cells that express
A ' c= 'O
substantial quantities of FasL protein (Fig.
kD ol-
l N N
o ur, a),0o o
omwF
,-
c

N N 0 X S
a
0s0x ca. (
4A) and message (Fig. 4B) into young
83 - syngeneic C57BL/6 mice carrying the Ipr
62- mutation in the fas gene. Tumor growth in
47- l B l | I * x 8: mutant mice was compared to that in
32- wild-type mice (Fig. 4C). In a representa-
tive experiment, 4 of 10 wild-type mice
I I | ||
carried a palpable tumor on day 4 after
B injection, but only 1 of 10 lpr mice was
M r- N.rl N N N1- (0 affected. At day 6 after injection, all wild-
47-
v-N N N N N N N N --3 type mice had a palpable tumor, whereas
half of the lpr mice were still devoid of
tumors. Tumor size differences became
32-
D smaller thereafter. To ascertain that the
25-5-111 increased rejection of melanoma cells was
- 75- indeed due to the absence of Fas and not
Fig. 1. (A) Detection of soluble FasL 8 due to the hyperactive immune system
(sFasL) in serum samples from patients * 501 found in lpr mice (1), melanoma cells were
(252, 256, and so forth) with melanoma. ° also injected into gld mutant mice that
Serum aliquots were subjected to SDS- ° 25- lack functional FasL (1). Unlike in Ipr
PAGE (10%) under nonreducing condi- 0 _ ,
mice, tumor growth in gld mice was rapid,
tions and immunoblot analysis using an ,n9° we K
anti-FasL antibody (32). Serum-derived and comparable to that of wild-type mice
sFasL migrates as a oligomeric complex Melanoma cell lines Control cell lines (Fig. 4C), thus supporting a crucial in-
under nonreducing conditions (-70 kD). volvement of the Fas system in tumor
The control lanes correspond to serum samples from a normal donor with no detectable sFasL rejection. The differences in tumor estab-
(Control) and from a patient with alcoholic hepatitis with very high levels of FasL (Hepatitis), lishment were even more pronounced
respectively (12). Immunodetection of sFasL was inhibited in the presence of peptide (100 pLg/ml) when B16-F1O cells (H2b) were injected
corresponding to the epitope recognized by the antibody (33). (B) FasL expression in melanoma into allogeneic MRL mice (H2k). At the
cell lines derived from patients with malignant melanoma (32). Triton extracts were analyzed as in high cell numbers injected, MRL mice did
(A). The transmembrane form of FasL is 40 kD; some cells show the proteolytically cleaved sFasL not reject the allogeneic tumor cells dur-
product (27 kD). (C) In vitro killing of Fas-sensitive cells by melanoma cells. Fas-bearing A20 B ing the observation period (mice were
lymphoma cells were cultured with melanoma cell lines derived from various patients (190, 248, and killed after 12 days), with tumor growth
so forth). Cell death was Fas-dependent, since only the FasL sensitive A20 lymphoma cell line but not
the FasL-resistant variant cell line A20R was killed (34). Controls: Melanoma effector cells are rates comparable to syngeneic C57BL/6
replaced by A20 cells and FasL, respectively. (D) Melanoma cell lines resist killing by FasL. Various mice (Fig. 4D). Five days after injection,
cell lines were incubated with recombinant FasL. Controls: A20 cells and mutant A20R cells were all MRL mice had growing tumors whereas
used as target cells. 70% of MRL-1pr mice were still tumor-free
1 364 SCIENCE * VOL. 274 * 22 NOVEMBER 1996
(Fig. 4D). Melanoma cells freshly isolated mor infiltrating lymphocytes and periph- cells, such as transforming growth factor-
from these tumors were cytolytic and eral blood from many melanoma patients 3 (TGF-P) or interleukin-10 (27, 28),
killed target cells via a Fas-dependent lytic (23-25). Nearly a dozen CTL-defined downregulation of MHC molecuLles ex-
pathway (Fig. 4E). melanoma peptide antigens have been pressed by tumor cells (29), or abnormal T
Amnple evidence exists showing that identified (26). Yet, melanoma cells, like lymphocyte signal transduction (30).
mel anomas can be imnMUnogenic: some tu- most other cancer cells, are able to avoid Combining the results presented here with
mors spontaneously regress, and they often immune destruction in most instances. the proposed role of FasL in creating im-
have large lymphocytic infiltrates (18). The mechanisms responsible for such an mune privileged sites (15, 1 6), an addi-
Melanomal-specific cytolytic T lympho- immune privilege could include the secre- tional mechanism may be considered:
cytes (CTLs) have been isolated from tul- tion of soluble inhibitory factors by tumor FasL-expressing melanoma cells may pro-
tect themselves against tumor-infiltrating
Fas-sensitive immune effector cells. The
delay in tumor formation in lpr mice is in
agreement with this premise, attribuLting
an important role to the Fas system in the
modulation of the anti-tumor immune re-
sponse. These data may also explain a
previous finding showing that the iMmu-
nosuppressive state induced by melanoma
cells is TGF-3-independent and requires
direct contact between tumor cell and T
cell (28). The molecular basis of this ob-
servation remains unknown and may be
complex, but may be similar to the one
protecting allogeneic islets from immune
rejection, when these are cotransplanted
with myoblasts ectopically expressing FasL
(31 ). As proposed for FasL-expressing tes-
tis grafts or stromal cells of the anterior
chamber of the eye, FasL-expressing mel-
anoma cells may kill Fas-sensitive activat-
ed T lymphocytes (15, 16). However, oth-
er explanations, including the downregu-
lation of an early uinspecific response by

Fig. 2. Expression of FasL in human malignant melanoma but not in melanocytes (A) Melanomas are r..- X
positive for the MAGE antigen (35). (B) Expression of FasL in malignant melanoma (M) from patients Fig. 3. (A) In situ detection of T cells infiltrating a
190 (B) and 225 (D), respectively. Higher magnification (F) suggests that FasL can be retained in the melanoma lesion (M). T cells extravasating a blood
cytoplasm, in agreement with its subcellular localization in activated T cells (36). (C) Cells infiltrating vessel (L, lumen) were stained using an antibody-
the tumor mass (marked with M). A substantial portion of the infiltrating cells are Fas-positive (arrow). CD3 specific antibody (closed arrow). (B) In situ
(E) Control: The antibody to FasL, Al 1, was replaced by rat Ig. Normal human skin was stained for detection of apoptotic cells (open arrow) in the
FasL in (G) and for the melanocyte-specific antigen HMB45 (arrow) in (H). Keratinocytes (K) express tumor. Apoptotic cells were detected through the
low amounts of FasL and melanocytes expressed no detectable FasL. Scale bars are each 20 yim: staining of DNA fragments with the TUNEL assay
bar in (E) applies to (A) through (E); bar in (H) also applies to (G). (37). Scale bar, 20 yLm.
SCIENCE * VOL. 274 * 22 NOVEMBER 1996 1 365
 .....................................-malmometeam
ill .:-" I.- ..A. .11'i 1:

Fig. 4. Consequences of A B 100

Fas-deficiency on tumor .L., 0
C
growth (38). (A) Expression O ( m
of FasL in melanoma B16- kD m + 75 *w-r
Fl0cell lines as detected by 47 _ *pr
immunoblots and (B) by RT-
PCR. Controls: In (A) inclu- -as F

sion of peptide (100 p.g/ml) 32 - FasL

corresponding to the 2-25-
epitope detected by the an- -
tibody to FasL (+peptide), 0
and in (B) amplification of 0 1 2 3 4 5
FasL cDNA (543 bp) in Neuro-2a-FasL but not in Neuro-2a-mock transfectants,
respectively, and amplification of actin cDNA (in B). (C) Tumor progression in Days afte r injection Days after injection
C57BU6 (WT), C57BU6-Ipr, and C57BU6-gld mice. The number of mice (in %) E
with palpable tumors are indicated. Data are representative of three to five experiments with 9 to 13 micE
group. Tumor size was measured daily. (D) Tumor progression in MRL and MRL-/pr mice. Data are repre,sen- A20
T
A20R
tative of three experiments with 9 to 12 mice per group. (E) Lytic activity of B1 6-Fl 0 cells on Fas-sensitive A20 75- 5
B lymphoma target cells. Melanoma cells were isolated from the tumor of C57BU6 mice 5 days after injec,tion.
Controls: Melanoma effector cells were replaced by A20 cells and FasL, respectively. Melanoma cells do ncotkill T
the variant, FasL-resistant A20R cells. 00
0 25

killing Fas-expressing granulocytes, mac- 23. L. M. Muul, P. J. Spiess, E. P. Director, S. A. Rosen-

berg, J. Immunol. 138, 989 (1987).
rophages, or NK-cells recognizing B16-F1O 24. M. Herin et al., Int. J. Cancer 39, 390 (1987).
melanoma cells that express few MHC 25. T. Boon, J. C. Cerottini, B. Van den Eynde, P. van der -N 5 )oe+t
molecules (19), are as likely. Preliminary Bruggen, A. Van Pel, Annu. Rev. Immunol. 12, 337 Melanoma to target cell ratio
results show that tumors other than ma- (1994).
26. A. Van Pel et al., Immunol. Rev. 145, 229 (1995).
lignant melanoma express FasL, suggesting 27. B. Mukherji and N. G. Chakraborty, Curr. Opin. On-
that FasL expression may be a more gen- col. 7, 175 (1995). anti-mouse Ig and avidin-peroxidase-conjugate
(TAGO, Burlingame, CA). Fas antigen was detect-
eral strategy used by tumor cells to escape 28. D. Huber, J. Philipp, A. Fontana, J. Immunol. 148,
ed with a mouse mAb to human Fas (10 p.g/ml). In
immune rejection. Thus, pharmacological 277 (1992).
29. S. Ferrone and F. M. Marincola, Immunol. Today 16, the negative control, the FasL antibody was re-
products that render infiltrating T cells 487 (1995). placed with rat 1g. The mAb to human melanoma
insensitive to FasL-induced killing may (HMB45) was from ENZO Diagnostics Inc., New
30. K. Zier, B. Gansbacher, S. Savadori, ibid. 17, 39 York, USA.
help break the immunological unrespon- (1996). 36. B. Lowin, C. Mattman, M. Hahne, J. Tschopp, Int.
31. H. T. Lau, M. Yu, A. Fontana, C. J. Stoeckert, Sci-
siveness to melanoma and, possibly, pro- ence 273,109 (1996).
Immunol. 8, 57 (1996).
vide a complementary approach in the 37. Frozen sections were stained as in Fig. 2. T cells
32. Serum aliquots (1 RI) or cellular NP-40 (1 %) extracts were detected using the anti-CD3 specific antibody
therapy of malignant melanoma. from 1 x 106 melanoma cells were separated by Leu-4 (Becton-Dickinson, Oxnard, CA). TUNEL
SDS-PAGE (10%) under nonreducing conditions and staining was done according to kit instructions
immunoblofted (ECL system, Amersham) using the from Boehringer, Mannheim, Germany.
REFERENCES AND NOTES affinity-purified rabbit anti-FasL antibody PE62 (33). 38. FasL was detected by Western blot analysis as in
33. M. Hahne et al., Int. Immunol. 7, 1381 (1995). Fig. 1. Total RNA was isolated from the B1 6-Fl 0 cell
1. S. Nagata and P. Golstein, Science 267, 1449 34. Cell lines were derived from tumor samples ob-
(1995). line and reverse transcribed to cDNA (Pharmacia,
2. D. Kagi et al., ibid. 265, 528 (1994).
tained from melanoma patients by surgical exci- ZOrich, Switzerland), and amplified by PCR generat-
sion. Samples were excised from lymph node (pa- ing a 543-bp fragment of FasL (forward: 5'-CACTC-
3. B. Lowin, M. Hahne, C. Mattmann, J. Tschopp, Na- tients 215, 252, and 256) or subcutaneous metas-
ture 370, 650 (1994). AAGGTCCATCCCTCTG-3'; reverse primer: 5'-TA-
tases (all other patients). The killing activity was GCTGACCTGTTGGACCTTGC-3') or a 445-bp
4. M. Y. Braun, B. Lowin, L. French, H. Acha-Orbea, J. assessed by incubating 1 x 105 melanoma cells
Tschopp, J. Exp. Med. 183, 657 (1996). fragment of actin (forward: 5'-ATCAAGATCCTGAC-
with 1 x 105 1251-UdR labeled target cells [A20 (20) CGAGCG-3'; reverse: 5'-TACTTGCGCTCAGGAG-
5. J. H. Russell, B. Rush, C. Weaver, R. Wang, Proc. and A20R, respectively] in 96-well plates 2 days
Natl. Acad. Sci. U.S.A. 90, 4409 (1993). GAGC-3'). The following conditions were used: 1
after seeding (3). The FasL-resistant variant cell line cycle at 94° for 5 min, then 30 cycles at 940 for 1 min,
6. G. G. Singer and A. K. Abbas, Immunity 1, 365 A20R was generated by continuous culture in me-
(1994). 55° for 1 min, and 720 for 2 min. The products were
dium containing recombinant FasL (39). The re- resolved on a 2% agarose gel. The lytic activity of
7. K. M. Mohler et al., Nature 370, 218 (1994). lease of 1251-UdR was then determined after a 16
8. G. M. McGeehan et al., ibid., p. 558. freshly isolated tumor cells was assayed as in Fig. 1.
hour co-incubation. Fas sensitivity of various cells B16-Fl 0 melanoma cells (1 x 105, American Type
9. M. Tanaka, T. Suda, T. Takahashi, S. Nagata, EMBO was determined by adding 5 p1 of supernatant from
J. 14,1129(1995). Tissue Culture) were injected subcutaneously into
recombinant FasL producing Neuro-2a cells (39) to 4-week old C57BL/6, C57BU6-lpr, C57BL/6-g/d
10. S. M. Mariani, B. Matiba, C. Baumler, P. H. Kram- 1 x 105 labeled cells in a total volume of 200 RI.
mer, Eur. J. Immunol. 25, 2303 (1995). (Jackson Laboratory, Bar Harbor, ME), MRL and
35. The metastatic melanoma of patient 190 was re- MRL-lpr mice (Harlan, Zeist, Nederland).
11. N. Kayagaki et al., J. Exp. Med. 182,1777 (1995). moved by surgery and snap-frozen in liquid nitro-
12. M. Hahne et al., in preparation. 39. A. Rensingehl et al., Eur. J. Immunol. 25, 2253
gen cooled isopentane. Ten-micrometer sections (1995).
13. T. Suda, T. Takahashi, P. Golstein, S. Nagata, Cell were cut and mounted on gelatin-coated slides.
75,1169 (1993). 40. S. Carrel, M. Schreyer, G. Spagnoli, J.-C. Cerottini,
After fixation in 4% paraformaldehyde for 10 min, D. Rimoldi, Int. J. Cancer 67, 417 (1996).
14. L. E. French et al., J. Cell Biol., 335 (1996). sections were rinsed with phosphate-buffered sa-
15. D. Bellgrau et al., Nature 377, 630 (1995). 41. We thank C. Kamel, F. Beerman, and K. Burns for
line (PBS) and endogenous peroxidase quenched helpful discussions; S. Hertig and S. Aslan for tech-
16. T. S. Griffith, T. Brunner, S. M. Fletcher, D. R. Green, with 0.3% H202 and 0.1% azide. Supernatants
T. A. Ferguson, Science 270, 1189 (1995). nical and secretarial assistance; and P. H. Kram-
(1 :1 dilution) of the monoclonal antibody (mAb) to mer and F. Lejeune for reagents. Supported by
17. G. Klein and T. Boon, Curr. Opin. Immunol. 5, 687 FasL, Al 1 (33) (Alexis Corp., San Diego, CA) was grants of the Swiss National Science Foundation
(1993). added to the section for 1 hour, and after rinsing in (J.T.) and of Human Frontier Science (M.H.). This
18. S. A. Rosenberg et al., N. Engl. J. Med. 319, 1676 PBS, peroxidase-conjugated goat antibody to rat
(1988). Ig (TAGO, Burlingame, CA) was added for 0.5 study is dedicated to Dr. Stephan Carrel (Ludwig
19. M. Hahne et al., data not shown. Institut of Cancer Research, Lausanne branch).
hour. After rinsing, antibody location was revealed With his death, melanoma research has lost a high-
20. S. Hahn et al., Eur. J. Immunol. 25, 2679 (1995). by the AEC chromogen. For the detection of mel- ly noted authority.
21. Y. Gavrieli, Y. Sherman, S. A. Ben-Sasson, J. Cell anoma cells, a mouse mAb to MAGE-1, detecting a
Biol. 119, 493 (1992). subpopulation of melanoma cells (40), was used
22. I. J. Fidler, Nature New Biol. 242, 148 (1973). (20 p.g/ml), and detected with a biotinylated goat 22 July 1996; accepted 12 September 1996

1 366 SCIENCE * VOL. 274 * 22 NOVEMBER 1996