The localization of CD95L in different cell types within tumors has not been well defined, and its role in tumor growth is uncertain. In this study, CD95L expression and its contribution to tumor growth are evaluated using genetic polymorphisms and adoptive transfer in genetically deficient mice. CD95L was detected in tumors in vivo at levels up to 104-fold higher than those in cell culture and predominantly in host tumor-infiltrating macrophages, but not in tumor cells. Adoptive transfer into genetically deficient mice revealed that host CD95-CD95L function did not alter tumor growth, demonstrating that CD95L alone does not affect tumor growth.

CD95 (Fas or Apo-1) and CD95L (Fas ligand) are cell surface molecules that induce apoptosis in immune and other cell types and play a pivotal role in the regulation of the immune response (1). Mutations in the CD95 and CD95L gene lead to the phenotypic changes seen in lpr3 and gld mice, respectively (2). The presence of CD95L on tumor cells is thought to eliminate tumor-infiltrating lymphocytes and contribute to local immune suppression. Although there is evidence that some cancers express CD95L on their cell surface (3, 4), a number of laboratories have failed to confirm these observations (5, 6, 7). Furthermore, tumors transfected with genes expressing CD95L, originally predicted to accelerate engraftment, were unexpectedly rejected (5). These findings, in addition to a recent report that CD95L is not found on most human melanomas in vitro, have confounded the interpretation of the earlier studies and the role of CD95L in immune recognition of malignancies (8, 9). To address these questions, we have analyzed the expression of CD95L in tumors using a polymorphism that can distinguish between its host or tumor derivation in syngeneic animals, and we have defined its role in the regulation of tumor growth by adoptive transfer into mice genetically deficient in CD95 or CD95L.

Cells.

LL/2 (Lewis lung carcinoma, mouse, H-2b) and B16-F10 (melanoma, mouse, H-2b) cell lines derived from C57BL/6 mice were purchased from American Type Culture Collection (Manassas, VA) and maintained with DMEM (Life Technologies, Inc., Grand Island, NY) containing 4 mml-glutamine and 10% fetal bovine serum. The CT26 (colon carcinoma, mouse, H-2d) cell line, from BALB/c mice, was derived from frozen laboratory stocks (5) and maintained with RPMI 1640 (Life Technologies, Inc.) containing 4 mml-glutamine and 10% fetal bovine serum.

Mice.

Six-week-old female mice were used for the experiments. C57BL/6J(H-2b), C.B10-H2b/LilMcdJ (H-2b), BALB/cJ (H-2d), B6.SmnC3H-Faslgld, and B6.MRL-Faslpr mice were purchased from Jackson Laboratory (Bar Harbor, ME).

Flow Cytometric Analysis for CD95L.

Target cells (1 × 106 cells) were stained with antimouse CD95L antibody (MFL3; PharMingen, San Diego, CA) or isotype control IgG followed by biotinylated secondary antibody (PharMingen) with streptavidin-phycoerythrin (PE) (PharMingen) to detect the expression of mouse CD95L. Relative fluorescence intensity was measured by fluorescence-activated cell sorter (FACScan; Becton Dickinson, Franklin Lakes, NJ) analysis of 104 cells. CT26 cells stably transfected with a eukaryotic expression vector encoding mouse CD95L (CT26-CD95L cells) were used as a positive control (10).

Protein Immunoblot Analysis for CD95L.

Total cell lysates were analyzed by 4–15% gradient SDS-PAGE. Protein immunoblot analysis was performed with anti-CD95L antibody (1:500; N-20; Santa Cruz Biotechnology, Santa Cruz, CA) and with the secondary antibody, horseradish peroxidase-conjugated goat antibody to rabbit IgG (1:5000; Santa Cruz Biotechnology). Primary and secondary antibodies were incubated for 1 h or 40 min, respectively, at room temperature. The immunocomplexes were detected by chemiluminescence (enhanced chemiluminescence; Amersham Pharmacia Biotech, Piscataway, NJ) according to the manufacturer’s instructions. Matrix metalloproteinase inhibitor (PharMingen) was added to culture medium to inhibit the cleavage of CD95L from cell membrane.

Real-time Quantitative RT-PCR.

Total RNA was extracted from cultured cells or tumors by using an RNeasy kit (Qiagen, Hilden, Germany) treated with DNase I according to the manufacturer’s directions. A standard curve was generated for CD95L from pVR1012/mCD95L (10) and GAPDH and synthesized by RT-PCR with 1 cycle at 50°C for 30 min and 94°C for 2 min, followed by PCR amplification with 40 cycles of denaturation at 94°C for 45 s, annealing at 49°C for 45 s, and elongation at 68°C for 2 min. Primers were as follows: (a) sense primer, 5′-GCTCTAGAGCCACCATGGTGAAGGTCGGTGTGAACG GATTTGGC-3′; and (b) antisense primer, 5′-CGGGATCCTTACTCCTTGGAGGCCATG TAGGCCATGAGGTCCAC-3′. After each cDNA was inserted into pGEM-4Z (Promega, Madison, WI), CD95L and GAPDH RNA were synthesized by in vitro transcription (MEGAscript T7 Kit; Ambion, Austin, TX). The number of target RNA copies was determined by A260 nm, and purity was assessed by A260 nm:A280 nm and agarose gels. Standard curves were constructed with 10-fold serial dilutions ranging from 10 to 108 copies of these standard RNAs.

Primers were designed with the assistance of the Primer Express computer program (Perkin-Elmer Corp., Norwalk, CT) and purchased from Life Technologies, Inc. Fluorogenic probes were from MegaBases, Inc. (Evanston, IL). For mouse CD95L, the primers were as follows: (a) sense, 5′-AACCCCAGTACACCCTCTGAAA-3′; (b) antisense, 5′-GGTTCCATATGTGTCTTCCCATTC-3′; and (c) fluorogenic probe, 5′-TGTGGCCCATTTAACAGGGAACCCC-3′. For mouse GAPDH, the primers were as follows: (a) sense, 5′-GGGAAGCCCATCACCATC-3′; (b) antisense, 5′-GCACCGGCCTCACCC-3′; and (c) fluorogenic probe, 5′-TCCAGGAGCGAGACCCCACTAACATC-3′.

To prepare samples, LL/2 or B16-F10 cells (5 × 106) were injected into C57BL/6 mice, or CT26 cells were inoculated into BALB/c mice, respectively. Fresh tumor samples, 15–20 mm in diameter, were collected for RNA extraction after approximately 2 weeks. Real-time quantitative RT-PCR using an ABI PRISM 7700 Sequence Detection System (Perkin-Elmer Corp.) and Platinum Quantitative RT-PCR Thermoscript One-Step System (Life Technologies, Inc.) were performed according to the manufacturer’s protocol with 500 ng of total RNA and 40 units of RNase inhibitor, with 1 cycle of 60°C for 30 min and 95°C for 5 min, followed by PCR amplification with 45 cycles of 95°C for 15 s and 60°C for 1 min. Quantitation of the threshold cycle was performed using a standard curve to determine the starting copy number for target RNA.

Detection of CD95L Polymorphisms.

RT-PCR was performed with total RNA as described above from LL/2 and B16-F10 cells injected into C.B10-H2b/LilMcd mice using primers (sense, 5′-CGGGTAAATTGTCCCTTG-3′; antisense, 5′-TGACCCCGGAAGTATAC T-3′) with 1 cycle at 50°C for 30 min and 94°C for 2 min, followed by PCR amplification with 40 cycles of denaturation at 94°C for 45 s, annealing at 49°C for 45 s, and elongation at 68°C for 2 min. The RT-PCR product was digested by AluI (Life Technologies, Inc.) at 37°C for 1 h and 5′-end-labeled by T4 polynucleotide kinase (Promega) with [γ-32P]ATP (Amersham Pharmacia Biotech). Labeled DNA was applied to 8% PAGE and detected by autoradiography.

RNA in Situ Hybridization and Immunohistochemistry.

The protocol used has been described previously (11). Briefly, optimal protease digestion time was first determined using nonspecific incorporation of the reporter nucleotide as the guide (10 μm digoxigenin dUTP). Optimal protease digestion was followed by overnight incubation in RNase-free DNase (10 units/sample; Boehringer Mannheim, Indianapolis, IN), and one-step RT-PCR using the reverse transcriptase thermostabile (rTth) system and digoxigenin dUTP. The same primer sequence for CD95L mRNA was used with the real-time quantitative RT-PCR, as described. After 20 cycles, the slides were washed at high stringency (60°C for 10 min in 15 mm salt with 2% BSA). The digoxigenin-labeled target-specific cDNA was detected using the antidigoxigenin-alkaline phosphatase conjugate (Boehringer Mannheim, 1:200 in PBS for 30 min at 37°C) followed by exposure to the chromogens nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate (Enzo Biochemicals, Farmingdale, NY). Nuclear fast red, which stains negative cells pink, served as the counterstain. For immunohistochemistry, we used a CD68 monoclonal antibody with the BioGenex (BioGenex, San Ramon, CA) supersensitive peroxidase system and hematoxylin counterstain.

Comparison of Tumorigenesis in gld, lpr, and wt Mice.

The tumor-free actuarial rate was analyzed for C57BL/6, B6.SmnC3H-Faslgld, and B6.MRL-Faslpr mice. LL/2 or B16-F10 cells (1 × 106 or 1 × 104) were injected s.c. into each group of mice (n = 5), and the animals were observed for 30 days. The first day on which a palpable tumor was detected in each mouse was recorded as an event time on the survival curve using the Kaplan-Meier method and compared statistically between groups by the log-rank test.

Flow Cytometric Analysis and Protein Immunoblot Analysis for CD95L.

Three murine cancer cell lines, LL/2 (Lewis lung carcinoma), CT26 (colon carcinoma), and B16-F10 (melanoma), were analyzed for CD95L expression. By flow cytometry, minimal levels of CD95L protein were detected in cell culture, and the protein was not observed by immunoblot analysis (Fig. 1). Similar results were observed in several human malignant cell lines, including H522 (lung carcinoma); MB 231, MCF-7, and mDA-MB231 (breast carcinomas); HepG2 (hepatoma); ASPC-1 and Bxpc-3 (pancreatic carcinomas); DU145 and LNCAP (prostate carcinomas); HeLa (cervical epithelial carcinoma); SW620 and HT29 (colon carcinomas); and M316, M342, M347, M444, M449, and M720 (melanomas; data not shown).

Significant Increase in CD95L mRNA Levels in Vivo Detected by Real-time Quantitative RT-PCR.

Because expression patterns might differ between cultured cells and growing tumors in vivo, CD95L mRNA levels were examined directly in tumor tissue using real-time RT-PCR (Taqman ABI 7700; Perkin-Elmer Corp.). LL/2, B16-F10, or CT26 cells were injected s.c. into syngeneic recipients, and total RNA was extracted from 15–20-mm-diameter tumors after 2 weeks. Adjacent normal tissue was carefully excluded. CD95L mRNA levels in tumors were increased 10- to 104-fold in vivo compared with cultured cells (Fig. 2). This result demonstrated that CD95L was readily detected in tumors; however, its origin remained unclear.

Use of a Genetic Polymorphism to Define the Recipient Origin of CD95L mRNA in Tumors.

To define the origin of CD95L in tumors, an analysis was performed using two genetic polymorphisms that have been identified previously in mice (12). The C57BL/6 strain contains a CD95L gene with an adenine at position 550 that is replaced with a guanine in BALB/c mice. This single-base substitution produces an AluI restriction site in BALB/c mice. A congenic mouse strain, C.B10-H2b/LilMcd, carries the BALB/c genotype for CD95L with the C57BL/6 MHC haplotype (H-2b) and is thus syngeneic to LL/2 and B16-F10 cells. In the C.B10-H2b/LilMcd recipient, LL/2 and B16-F10 tumors grew similarly to those in the congenic C57BL/6 mouse. To detect the origin of CD95L in these tumors, an allelic analysis was performed. A 162-bp segment that spanned the polymorphic site was amplified by RT-PCR and digested with AluI, followed by 5′ end-labeling with [γ-32P]ATP and PAGE (Fig. 3,a). Products of the expected size were identified from LL/2 and B16-F10 tumors grown in C.B10-H2b/LilMcd mice. Upon restriction enzyme digestion, fragments of 118 and 44 bp, consistent with the BALB/c (host) genotype, were detected. This finding showed that the majority of CD95L in the tumor mass originated from the recipient mouse rather than donor tumor cells. Mixture experiments showed that tumor cell (C57BL/6) CD95L could be detected when it was present at levels as low as 10% of total CD95L mRNA (Fig. 3 b). This result was confirmed by cloning the PCR products into plasmids and AluI digestion. Of 20 colonies analyzed, all contained the recipient mouse CD95L DNA sequence (data not shown).

RT in Situ PCR and Immunohistochemistry in Vivo.

Expression of recipient CD95L in tumors was further characterized by performing RNA in situ hybridization and immunostaining with CD68 antibody. This analysis revealed that CD95L mRNA was present predominantly in CD68+ monocytic cells (Fig. 4 A–D). This distribution was also observed by immunostaining with CD95L and CD11b antibodies and, to a lesser degree, with anti-CD3ε (data not shown).

Similar Tumorigenesis in gld, lpr, and wt Mice.

The role of host CD95 and CD95L in the regulation of tumor growth was analyzed using mice genetically deficient for either gene product, with congenic CD95L deficient (gld/gld) or CD95 deficient (lpr/lpr) or wt C57BL/6 recipients. LL/2 or B16-F10 tumor cells (104 or 106) were injected sc into C57BL/6, B6.SmnC3H-Faslgld, and B6.MRL-Faslpr mice, and tumor growth was measured. Tumor-free survival curves indicated no significant difference in tumorigenesis among gld, lpr, or wt mice in all cases (Fig. 4 E). Similarly, no significant differences in tumor size were observed among these groups (data not shown). These data suggest that the CD95-CD95L system did not significantly affect tumor growth, morbidity, or survival.

The role of CD95L in immune surveillance of malignancies has been viewed with uncertainty. It is evident that CD95L promotes lymphocyte apoptosis and inhibits immune activation through its effects on activated T or B lymphocytes and monocytes (10, 13). The requirement for CD95L in allogeneic corneal transplantation further supported its role for immune suppression in vivo(14). Although some data suggested that CD95L facilitated allogeneic engraftment (15), and it was implicated in immune escape in a melanoma (4), several studies subsequently raised questions about its role in tissue rejection and in cancer immune evasion. In fact, CD95L was subsequently shown to exhibit a paradoxical proinflammatory effect in several independent studies, enhancing allograft or tumor transplant rejection when CD95L was expressed on the cell surface (5, 16). The discrepancies among these studies may be technical in part because the specificity of some antibody reagents has been questioned (9), and it has remained difficult to identify the cellular localization and function of CD95L in tumors. In this report, we find that CD95L in tumors is derived not from malignant cells but from nontransformed mouse cells, primarily tumor-infiltrating monocytes and, to a lesser extent, T cells. Furthermore, CD95 and CD95L mutant mouse strains reveal that their expression does not affect tumor growth.

The disparate roles of CD95L in cell death and in immune activation of distinct immune cells have been reviewed elsewhere (17). In part, this complexity may stem from the fact that CD95L, which exists in both membrane-bound and soluble forms, binds to extracellular matrix and can be retained at specific tissue sites in vivo, where its activity can be increased (18). In addition, the activity of CD95L can be modulated by other cytokines. The combination of CD95L with transforming growth factor β has been shown to suppress the proinflammatory effects of CD95L (19), and the two are often found in combination, both in immune privileged sites and in different malignancies in vivo(20).

Although the effects of CD95L in privileged sites made it an attractive candidate for suppression of antitumor immunity, the findings reported here demonstrate that it does not directly affect tumor growth. Thus, if malignant cells are able to use CD95L immune suppression to enhance tumor growth, it likely acts through an indirect mechanism and is not dependent solely on signaling through the CD95/CD95L pathway. The continued interest and reliance on the expression of Fas ligand in tumors for pathological or clinical diagnostic parameters underscore the need to define its cellular localization and function. These findings also suggest that strategies to inhibit immune-suppressive effects of CD95L alone are unlikely to improve immune-based therapies for cancer.

Fig. 1.

Flow cytometric analysis and protein immunoblot analysis for CD95L. a, three murine cancer cell lines, CT26 (colon carcinoma), B16-F10 (melanoma), and LL/2 (Lewis lung carcinoma), were analyzed for CD95L expression by flow cytometry. Minimal levels of CD95L protein were detected in cell culture. CT26 cells stably transfected with a eukaryotic expression vector encoding mouse CD95L (CT26-CD95L cells) were used as a positive control (1). b, protein immunoblot analysis was performed with anti-CD95L antibody for CT26, B16-F10, or LL/2 cell lines. CD95L protein was not observed in these three cancer cell lines but was readily detected in the positive control, CT26-CD95L. This CD95L-specific band was enhanced by adding matrix metalloproteinase inhibitor that inhibits the cleavage of CD95L from the cell surface.

Fig. 1.

Flow cytometric analysis and protein immunoblot analysis for CD95L. a, three murine cancer cell lines, CT26 (colon carcinoma), B16-F10 (melanoma), and LL/2 (Lewis lung carcinoma), were analyzed for CD95L expression by flow cytometry. Minimal levels of CD95L protein were detected in cell culture. CT26 cells stably transfected with a eukaryotic expression vector encoding mouse CD95L (CT26-CD95L cells) were used as a positive control (1). b, protein immunoblot analysis was performed with anti-CD95L antibody for CT26, B16-F10, or LL/2 cell lines. CD95L protein was not observed in these three cancer cell lines but was readily detected in the positive control, CT26-CD95L. This CD95L-specific band was enhanced by adding matrix metalloproteinase inhibitor that inhibits the cleavage of CD95L from the cell surface.

Close modal
Fig. 2.

Real-time quantitative RT-PCR for CD95L. A, standard curve construction for CD95L and GAPDH with 10-fold serial dilutions from 10 to 108 copies of synthesized standard RNAs. The copy number of CD95L mRNA from 500 ng of total RNA from control BALB/c mouse spleen was 5,906.17 ± 539.70, and the copy number of GAPDH from the same source was 1,255,623.00 ± 5,294.87. B, increased CD95L in three tumor types compared with cultured cells ranged from 1- to 104-fold. The number of copies of CD95L was normalized by GAPDH by calculating the number of CD95L per 10 million GAPDH for comparison. The data represent the mean and SE from five mice for LL/2, eight mice for CT26 and B16-F10, and three independent samples for cultured cells, respectively. P was determined by the Mann-Whitney U test. For reference, the normalized value of control BALB/c mouse spleen was 45,596.37 ± 1,445.66.

Fig. 2.

Real-time quantitative RT-PCR for CD95L. A, standard curve construction for CD95L and GAPDH with 10-fold serial dilutions from 10 to 108 copies of synthesized standard RNAs. The copy number of CD95L mRNA from 500 ng of total RNA from control BALB/c mouse spleen was 5,906.17 ± 539.70, and the copy number of GAPDH from the same source was 1,255,623.00 ± 5,294.87. B, increased CD95L in three tumor types compared with cultured cells ranged from 1- to 104-fold. The number of copies of CD95L was normalized by GAPDH by calculating the number of CD95L per 10 million GAPDH for comparison. The data represent the mean and SE from five mice for LL/2, eight mice for CT26 and B16-F10, and three independent samples for cultured cells, respectively. P was determined by the Mann-Whitney U test. For reference, the normalized value of control BALB/c mouse spleen was 45,596.37 ± 1,445.66.

Close modal
Fig. 3.

Detection of polymorphism and recipient origin of CD95L mRNA in tumors. a, polymorphism of mouse CD95L mRNA and RT-PCR amplification. Adenine 550 of the CD95L mRNA sequence in C57BL/6 mice was replaced by guanine in BALB/c and C.B10-H2b/LilMcd mice. This replacement produces an AluI restriction site (GAAGCT, 547–552) in BALB/c and C.B10-H2b/LilMcd mice. Primers were designed to span this digestion site, and a 162-bp segment was amplified by RT-PCR. b, the 162-bp RT-PCR product from mouse spleen was digested with AluI to 118- and 44-bp fragments from BALB/c or C.B10-H2b/LilMcd mice (Lanes 4 and 6), but this was not observed in the C57BL/6 mouse (Lane 2). In the case of LL/2 tumors (C57BL/6, syngeneic) grown in C.B10-H2b/LilMcd mice, the digestion pattern was similar to the recipient mouse strain, C.B10-H2b/LilMcd (Lane 5); however, the two restriction products were not observed in LL/2 tumors grown in C57BL/6 mice (Lane 1). A 10% contamination of C57BL/6 mRNA in BALB/c mice could be readily detected in this assay (Lane 3). CD95L in the B16-F10 tumor (C57BL/6, syngeneic) grown in C.B10-H2b/LilMcd mice also had the same polymorphism as the host C.B10-H2b/LilMcd (Lanes 9 and 10) or BALB/c mice (Lane 8), but not the C57BL/6 mouse (Lane 7).

Fig. 3.

Detection of polymorphism and recipient origin of CD95L mRNA in tumors. a, polymorphism of mouse CD95L mRNA and RT-PCR amplification. Adenine 550 of the CD95L mRNA sequence in C57BL/6 mice was replaced by guanine in BALB/c and C.B10-H2b/LilMcd mice. This replacement produces an AluI restriction site (GAAGCT, 547–552) in BALB/c and C.B10-H2b/LilMcd mice. Primers were designed to span this digestion site, and a 162-bp segment was amplified by RT-PCR. b, the 162-bp RT-PCR product from mouse spleen was digested with AluI to 118- and 44-bp fragments from BALB/c or C.B10-H2b/LilMcd mice (Lanes 4 and 6), but this was not observed in the C57BL/6 mouse (Lane 2). In the case of LL/2 tumors (C57BL/6, syngeneic) grown in C.B10-H2b/LilMcd mice, the digestion pattern was similar to the recipient mouse strain, C.B10-H2b/LilMcd (Lane 5); however, the two restriction products were not observed in LL/2 tumors grown in C57BL/6 mice (Lane 1). A 10% contamination of C57BL/6 mRNA in BALB/c mice could be readily detected in this assay (Lane 3). CD95L in the B16-F10 tumor (C57BL/6, syngeneic) grown in C.B10-H2b/LilMcd mice also had the same polymorphism as the host C.B10-H2b/LilMcd (Lanes 9 and 10) or BALB/c mice (Lane 8), but not the C57BL/6 mouse (Lane 7).

Close modal
Fig. 4.

Detection of CD95L mRNA using RT in situ PCR in macrophages in CT26 tumors in vivo and comparison of tumorigenesis among gld, lpr, and wt mice. A–C are serial sections (4 μm apart) representing the interface between the tumor (small arrow) and adjacent benign reactive tissue that contains a large number of mononuclear cells (large arrow). A shows the H&E stain. B demonstrates many mononuclear cells positive for CD68, indicative of activated macrophages. These same cells contained PCR-amplified CD95L cDNA (C). Tumor cells were negative for this target. The signal was lost in the macrophages with RNase digestion (data not shown), indicating that it was RNA based. D shows a section of normal spleen analyzed for CD95L expression by RT in situ PCR. Note that the cells that express CD95L show the ample cytoplasm and grooved nuclei typical of activated macrophages. E shows that the tumor-free actuarial rate for C57BL/6J (wt), B6.SmnC3H-Faslgld (gld), and B6.MRL-Faslpr (lpr) mice. Tumor-free survival curves indicated no significant difference in tumorigenesis among gld, lpr, or wt mice in all cases. LL/2 (top panels) or B16-F10 (bottom panels) cells (1 × 106 or 1 × 104) were injected s.c. into each group of mice (n = 5), which were observed for 30 days. The first day on which a palpable tumor was detected in each mouse was recorded as an event time on the survival curve using the Kaplan-Meier method and compared statistically between groups by the log-rank test.

Fig. 4.

Detection of CD95L mRNA using RT in situ PCR in macrophages in CT26 tumors in vivo and comparison of tumorigenesis among gld, lpr, and wt mice. A–C are serial sections (4 μm apart) representing the interface between the tumor (small arrow) and adjacent benign reactive tissue that contains a large number of mononuclear cells (large arrow). A shows the H&E stain. B demonstrates many mononuclear cells positive for CD68, indicative of activated macrophages. These same cells contained PCR-amplified CD95L cDNA (C). Tumor cells were negative for this target. The signal was lost in the macrophages with RNase digestion (data not shown), indicating that it was RNA based. D shows a section of normal spleen analyzed for CD95L expression by RT in situ PCR. Note that the cells that express CD95L show the ample cytoplasm and grooved nuclei typical of activated macrophages. E shows that the tumor-free actuarial rate for C57BL/6J (wt), B6.SmnC3H-Faslgld (gld), and B6.MRL-Faslpr (lpr) mice. Tumor-free survival curves indicated no significant difference in tumorigenesis among gld, lpr, or wt mice in all cases. LL/2 (top panels) or B16-F10 (bottom panels) cells (1 × 106 or 1 × 104) were injected s.c. into each group of mice (n = 5), which were observed for 30 days. The first day on which a palpable tumor was detected in each mouse was recorded as an event time on the survival curve using the Kaplan-Meier method and compared statistically between groups by the log-rank test.

Close modal

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

Supported by the NIH.

3

The abbreviations used are: lpr, lymphoproliferative; gld, generalized lymphoproliferative disorder; wt, wild-type; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; RT, reverse transcription.

We thank Ati Tislerics and Nancy Barrett for manuscript preparation, Ling Xu for preparation of the slides for histological analysis, and members of the Nabel Laboratory for helpful advice and comments.

1
Suda T., Takahashi T., Golstein P., Nagata S. Molecular cloning and expression of the Fas ligand, a novel member of the tumor necrosis factor family.
Cell
,
75
:
1169
-1178,  
1993
.
2
Watanabe-Fukunaga R., Brannan C. I., Copeland N. G., Jenkins N. A., Nagata S. Lymphoproliferation disorder in mice explained by defects in fas antigen that mediates apoptosis.
Nature (Lond.)
,
356
:
314
-317,  
1992
.
3
O’Connell J., O’Sullivan G. C., Collins J. K., Shanahan F. The Fas counterattack: Fas-mediated T cell killing by colon cancer cells expressing Fas ligand.
J. Exp. Med.
,
184
:
1075
-1082,  
1996
.
4
Hahne M., Rimoldi D., Schroter M., Romero P., Schreier M., French L. E., Schneider P., Bornand T., Fontana A., Lienard D., Cerottini J. C., Tschopp J. Melanoma cell expression of Fas(Apo-1/CD95) ligand: implications for tumor immune escape.
Science (Wash. DC)
,
274
:
1363
-1366,  
1996
.
5
Arai H., Gordon D., Nabel E. G., Nabel G. J. Gene transfer of Fas ligand induces tumor regression in vivo.
Proc. Natl. Acad. Sci. USA
,
94
:
13862
-13867,  
1997
.
6
Chappell D. B., Zaks T. Z., Rosenberg S. A., Restifo N. P. Human melanoma cells do not express Fas (Apo-1/CD95) ligand.
Cancer Res.
,
59
:
59
-62,  
1999
.
7
Favre-Felix N., Fromentin A., Hammann A., Solary E., Martin F., Bonnotte B. The tumor counterattack hypothesis revisited: colon cancer cells do not induce T cell apoptosis via the Fas (CD95, APO-1) pathway.
J. Immunol.
,
164
:
5023
-5027,  
2000
.
8
O’Connell J., Houston A., Bennett M. W., O’Sullivan G. C., Shanahan F. Immune privilege or inflammation? Insights into the Fas ligand enigma.
Nat. Med.
,
7
:
271
-274,  
2001
.
9
Restifo N. P. Countering the “counterattack” hypothesis.
Nat. Med.
,
7
:
259
2001
.
10
Arai H., Chan S. Y., Bishop D. K., Nabel G. J. Inhibition of the alloantibody response by CD95 ligand.
Nat. Med.
,
3
:
843
-848,  
1997
.
11
Nuovo G. J., Plaia T. W., Belinsky S. A., Baylin S. B., Herman J. G. In situ detection of the hypermethylation-induced inactivation of the p16 gene as an early event in oncogenesis.
Proc. Natl. Acad. Sci. USA
,
96
:
12754
-12759,  
1999
.
12
Kayagaki N., Yamaguchi N., Nagao F., Matsuo S., Maeda H., Okumura K., Yagita H. Polymorphism of murine Fas ligand that affects the biological activity.
Proc. Natl. Acad. Sci. USA
,
94
:
5986
-5990,  
1997
.
13
Iwata M., Harada K., Hiramatsu K., Tsuneyama K., Kaneko S., Kobayashi K., Nakanuma Y. Fas ligand-expressing mononuclear cells around intrahepatic bile ducts co-express CD68 in primary biliary cirrhosis.
Liver
,
20
:
129
-135,  
2000
.
14
Griffith T. S., Ferguson T. A. The role of FasL-induced apoptosis in immune privilege.
Immunol. Today
,
18
:
240
-244,  
1997
.
15
Bellgrau D., Gold D., Selawry H., Moore J., Franzusoff A., Duke R. A role for CD95 ligand in preventing graft rejection.
Nature (Lond.)
,
377
:
630
-632,  
1995
.
16
Allison J., Georgiou H. M., Strasser A., Vaux D. L. Transgenic expression of CD95 ligand on islet β cells induces a granulocytic infiltration but does not confer immune privilege upon islet allografts.
Proc. Natl. Acad. Sci. USA
,
94
:
3943
-3947,  
1997
.
17
Lenardo M., Chan K. M., Hornung F., McFarland H., Siegel R., Wang J., Zheng L. Mature T lymphocyte apoptosis: immune regulation in a dynamic and unpredictable antigenic environment.
Annu. Rev. Immunol.
,
17
:
221
-253,  
1999
.
18
Aoki K., Kurooka M., Chen J-J., Petryniak J., Nabel E. G., Nabel G. J. Extracellular matrix interacts with soluble CD95L: retention and enhancement of cytotoxicity.
Nat. Immunol.
,
2
:
333
-337,  
2001
.
19
Chen J-J., Sun Y., Nabel G. J. Regulation of the proinflammatory effects of Fas ligand (CD95L).
Science (Wash. DC)
,
282
:
1714
-1717,  
1998
.
20
Ferrone S., Finerty J. F., Jaffee E. M., Nabel G. J. How much longer will tumour cells fool the immune system?.
Immunol. Today
,
21
:
70
-72,  
2000
.