The membrane-bound but not the soluble form of human Fas ligand is responsible for its inflammatory activity

Authors


Abstract

The ectopic expression of Fas ligand (FasL/CD95L) in tissues or tumors induces neutrophil infiltration and the destruction of the tissues or the rejection of tumors. It has been suggested thatthe infiltrated neutrophils are responsible for the latter phenomena. FasL is synthesized as a type II transmembrane protein, and soluble FasL is produced by a proteolytic mechanism from the membrane-bound form. We previously demonstrated that uncleavable membrane-bound FasL of mice induces IL-1β release from inflammatory cells, and suggested that the IL-1β enhances neutrophil infiltration. However, recent papers reported that human soluble FasL is directly chemoattractive to neutrophils in vitro and proposed that the soluble form of FasL is responsible for its inflammatory activity. Therefore, in this report, we investigated which form is responsible for the inflammatory activities of human FasL. We produced tumor cell lines expressing one or both forms of human FasL. Cells expressing both forms or only the membrane-bound form of FasL induced neutrophil infiltration when transplanted into the peritoneal cavity of syngeneic mice, while cells expressing only the soluble form did not. Purified soluble FasL failed to induce neutrophil infiltration in vivo. IL-1β release from inflammatory peritoneal exudate and acceleration of tumor rejection were also mediated by membrane-bound but not soluble FasL. These results indicate that the membrane-bound form of FasL is primarily responsible for its inflammatory activity.

Abbreviation:
FasL:

Fas ligand

1 Introduction

Fas ligand (FasL/CD95L) is a cytokine that induces apoptotic cell death and plays important roles in various aspects of the immune system, including cell-mediated cytotoxicity and self tolerance 1. FasL is produced as a 40-kDa type II transmembrane protein 2, and a portion of this membrane-bound FasL is then released in a soluble form by a proteolytic mechanism 3. In humans, soluble FasL retains cytotoxicity, although it is significantly weaker than the cytotoxicity of the membrane-bound form. For example, human soluble FasL induces apoptosis in Con A-activated but not freshly isolated peripheral blood T cells, while the membrane-bound form kills both 4. In contrast, mouse soluble FasL is essentially inactive in terms of the apoptosis inducing ability. It should be noticed that the recombinant mouse soluble FasL (termed WX1) used in this and other studies possesses artificially strong apoptosis inducing activity, probably because of its aggregating nature. WX1, like membrane-bound FasL, kills both activated and fresh peripheral blood T cells 4.

FasL is expressed in the eyes and testes, and is implicated in the immune-privileged property of these organs 5, 6. It has been suggested that FasL protectsthese organs from the detrimental effects of inflammation by inducing apoptosis in inflammatory cells. On the line of this idea, it was reported that the survival time of allogeneic islet transplants was significantly prolonged by the co-transplantation of syngeneic myoblasts that were genetically engineered to express FasL 7. Several reports have suggested that FasL expressed in tumor cells is involved in the tumor immune escape 810. However, controversially, FasL ectopically expressed in tissues by genetic engineering has been found to induce inflammation accompanied by massive neutrophil infiltration 11, 12. FasL-expressing tumor cells induced neutrophil infiltration and were rejected when transplanted in syngeneic mice 1315. Thus, FasL seems to induce or suppress inflammation depending on the situation. To manipulate these outcomes of FasL stimulation, it is important to understand the molecular mechanism underlying FasL-induced inflammation. We have found that a recombinant uncleavable membrane-bound form of mouse FasL induces the processing and release of IL-1β in inflammatory PEC consisting of mainly neutrophils 14. This FasL-induced IL-1β release involves caspase activation, and hence apoptosis in neutrophils. Furthermore, FasL-induced neutrophil infiltration was found to be suppressed in IL-1-deficient mice compared to wild-type mice. Thus, we concluded that IL-1β released from apoptotic inflammatory cells, at least in part, mediates FasL-induced neutrophil infiltration. If it is the case, the membrane-bound form of FasL should play a more important role than the soluble form in its inflammatory activity, because the former has stronger cytotoxic activity than the latter, as described above. In contrast, it was recently reported that the soluble form of human FasL is directly chemoattractive to neutrophils in vitro, and proposed that soluble FasL is responsible for its inflammatory activity 16, 17. To clarify which mechanism is central to FasL-induced inflammation, in this report we compared the in vivo inflammatory activity of the membrane-bound and soluble forms of human FasL.

2 Results

2.1 Establishment of tumor cells expressing the membrane-bound and/or a soluble form of human FasL

The expression plasmids for FasL producing both membrane-bound and soluble FasL (FasLDC), the uncleavable membrane-bound form only (FasLDC2), and a soluble form only (FasLS) (Fig. 1a) were generated (see Sect. 4.2). Several independent clones of cells expressing these mutants of FasL were generated from the FBL-3 mouse erythroleukemia cell line. The transfectants of FasLDC, FasLDC2, and FasLS were named FFL, FDC2, and FFS, respectively. Flow cytometry analyses using an anti-FasL mAb confirmed that the FFL and FDC2 cell lines expressed FasL on the cell surface, while FFS as well as its parental cell line FBL-3 and the control hygromycin B resistant cell line (FBH) did not (Fig. 1b). The cytotoxic activity of soluble FasL in the culture supernatant was determined using W4 cell line, a transfectant of mouse Fas (Apo-1/CD95) cDNA derived from WR19L lymphoma cell line. Comparable cytotoxicity was detected in the supernatant of FFL and FFS cells, whereas no or little activity was detected in the supernatant of the other cell lines (Fig. 1c). None of these supernatants showed cytotoxicity against WR19L cells (data not shown). It was confirmed by gel filtration analyses that the FasLS protein purified from culture supernatant of a FFS cell line existed as a trimmer, like naturally processed soluble FasL (data not shown).

Figure 1.

Preparation of transfectants expressing membrane-bound and soluble FasL. (a) A schematic diagram of constructs producing one or both of the membrane-bound and soluble forms of human FasL. FasLDC lacks most of the cytoplasmic region compared with wild-type FasL (top). FasLDC, like wild-type FasL, produces both membrane-bound and soluble FasL. In addition, FasLDC2 lacks exon 2, which contains the cleavage site. Thus, FasLDC2 is not cleavable, and exists in the membrane-bound form only. FasLS, a fusion protein consisting of the signal sequence of mouse Fas and the TNF-homologous region of human FasL, exists in the soluble form only. cDNA encoding these constructs were transfected into FBL-3 cell line. The transfectants were named as indicated. (b) FBL-3 and its transfectants were labeled with biotinylated anti-human FasL mAb (shaded area) or control hamster IgG (dotted line) followed by PE-streptavidin. Cells were then analyzed by flow cytometry. (c) FasL-sensitive W4.5 cells were cultured with the indicated dilution of culture supernatants of FBL-3 and its transfectants. After 24 h in culture, cell viability was determined. FBH: diamond, clone 1; square, clone 2; triangle, clone 3; circle, clone 4. FFL: square, clone 10; triangle, clone 21; circle, clone 39. FDC2: diamond, clone 20; square, clone 35; triangle, clone 40; circle, clone 50. FFS: diamond, clone 2; square, clone 3; triangle, clone 14; circle, clone 31.

2.2 Membrane-bound but not soluble FasL induces neutrophil infiltration

To investigate which form of FasL is responsible for the inflammatory activity, the established cell lines were transplanted into the peritoneal cavity of syngeneic mice. PEC were recovered 18 h later, and the proportion of Gr-1+ neutrophils was determined by flow cytometry. As shown in Fig. 2, neutrophil infiltration was significantly induced by cells expressing membrane-bound FasL (FFL and FDC2), but not by those producing soluble FasL only (FFS). This conclusion was confirmed by calculating absolute numbers of neutrophils in PEC (data not shown). To directly test whether soluble FasL induces the chemotaxis of neutrophils, naturally processed soluble FasL was purified from the culture supernatant of K562 cells transfected with FasLDC, and 0.1–10,000 ng soluble FasL was injected i.p. As shown in Fig. 3, no significant enhancement of neutrophil infiltration compared with that induced by diluent only was induced at any time up to 18 h after the soluble FasL injection. Based on these results, we concluded that membrane-bound FasL is responsible for the neutrophil infiltration induced by FasL-expressing tumor cells and that soluble FasL is not chemoattractive to neutrophils in vivo.

Figure 2.

Membrane-bound but not soluble FasL induces neutrophil infiltration. FBL-3 and the indicated individual clones of transfectants (4×106 cells) were injected into the peritoneal cavities of the indicated numbers of syngeneic mice. After 18 h, peritoneal cells were recovered and analyzed for the proportion of Gr-1+ cells by flow cytometry. Tumor cells and dead cells were excluded from analysis based on forward scatter, side scatter and propidium iodide staining. Vertical lines indicate standard deviations and asterisks indicate statistically significant (p<0.005) neutrophil infiltration compared to that observed by FBL-3 inoculation.

Figure 3.

Purified soluble FasL does not show neutrophil chemotactic activity in vivo. (a) The indicated amounts of purified soluble FasL in 200 μl PBS containing 1% syngeneic mouse serum or 1.5 ml 3% thioglycollate medium was injected i.p. into ddY mice. Peritoneal cells were recovered 4, 9, or 18 h later and analyzed for the proportion of Gr-1+ cells by flow cytometry. (b) The indicated amounts of recombinant IL-1β in 200 μl PBS containing 1% syngeneic mouse serum was injected i.p. into C57BL/6 mice. Peritoneal cells were recovered 4 and 9 h later and analyzed as described above. (a, b) Each bar indicates an average of n=3, except for those of thioglycollate injection (n=1). Vertical lines indicate standard deviations.

We previously found that FasL induces release of the active form of IL-1β from 4-h PEC, and proposed that this IL-1β is responsible for the FasL-induced neutrophil infiltration based on the previous report demonstrating that IL-1β induces neutrophil infiltration in vivo. To confirm the latter phenomenon, 1 or 5 ng recombinant IL-1β was injected i.p. Consistently with the previous report, injection of as little as 1 ng recombinant IL-1β induced massive neutrophil infiltration at 4 h after the injection.

2.3 Membrane-bound but not soluble FasL induces early tumor rejection

FBL-3 is a tumor cell line that spontaneously regresses within about three weeks when 1×106–5×106 cells are transplanted i.d. in syngeneic mice. This rejection is mediated by T cells, given that the same tumor cells grow progressively in nude mice 18. Using transfectants derived from this tumor cell line, we tested whether the introduction of FasL into tumor cells would cause immune escape or promote rejection. In Fig. 4, the tumor cells were transplanted into the dorsal skin of syngeneic mice. Tumor cells expressing membrane-bound FasL (FFL, FDC2) were rejected more quickly than the control FBH cells in wild-type mice. The same tumor cells transplanted into lpr mice, which do not express functional Fas, showed a slower time course of rejection that was similar to control FBH cells. The latter observation is consistent with the previous report 13, and indicates that the accelerated tumor rejection in wild-type mice was mediated by Fas-FasL interaction, where FasL was expressed on the tumor cells and Fas was expressed in the host. It also assures that the growth potential of FFL and FDC2 clones were comparable to the other clones. On the other hand FFS cells were rejected slowly with a time course similar to FBH in both wild-type and lpr mice. These results indicate that expression of FasL in tumor cells results in the promotion of tumor rejection rather than in tumor immune escape, and that membrane-bound but not soluble FasL is responsible for this phenomenon.

Figure 4.

Membrane-bound but not soluble FasL promotes tumor rejection. Transfectants (2.5×106 cells) were injected into the dorsal skin of 8-12-week-old wild-type (closed symbols) or lpr/lpr female C57BL/6 mice (open symbols). Tumor size was measured at 3-day intervals after tumor inoculation. Each point is a mean of six tumors in three mice. FBH: circle, clone 2; triangle, clone 3. FFL: circle, clone 10; triangle, clone 21. FDC2: circle, clone 20; triangle, clone 50. FFS: circle, clone 14; triangle, clone 31.

2.4 Membrane-bound but not soluble FasL induces IL-1β release from inflammatory peritoneal exudate cells

IL-1β release-inducing activity of different forms human FasL was examined. We first examined the presence of IL-1β in the peritoneal fluid collected 18 h after the injection of tumor cells using ELISA (Fig. 5a). IL-1β was found in the peritoneal fluid from mice received FFL and FDC2 cell lines, but not from those mice injected FFS cells. Similar results were obtained by in vitro experiments in which 4-h PEC were cultured with the same cell lines for 18 h, and the amounts of IL-1β in the culture supernatants were determined (Fig. 5b). In the experiments shown Fig. 5c, 4-h PEC and 4-day PEC consisting of mainly neutrophils and macrophages, respectively, were cultured with transfectants as well as purified soluble FasL. Consistently with our previous results, WX1, a recombinant mouse soluble FasL that has artificially strong apoptosis-inducing activity, elicited IL-1β release from 4-h but not 4-day PEC. In contrast, soluble human FasL failed to induce significant IL-1β release in both 4-h and 4-day PEC. This difference of WX1 and human soluble FasL correlates with their apoptosis inducing activity, namely, WX1 but not human soluble FasL induced apoptosis in 4-h PEC (14 and data not shown). On the other hand, FFL and FDC2 clones induced strong IL-1β release in both 4-h and 4-day PEC. We previously suggested that FasL does not induce IL-1β release in macrophages based on the experiments using WX1. However, FFL was found to induce IL-1β release in plastic adherent macrophages separated from 4-day PEC (data not shown). Therefore, membrane-bound FasL is capable of inducing IL-1β release in not only neutrophils but also macrophages.

Figure 5.

The membrane-bound but not soluble form of human FasL induces IL-1β release from 4-h and 4-day PEC. (a) The indicated transfectants (4×106 cells) were injected i.p. into C57BL/6 mice. After 18 h, the peritoneal cavities were washed with 1 ml PBS. The peritoneal wash was recovered, spun, and the supernatant was collected. (b) Four-hour PEC (5×105 cells) were cultured with indicated transfectants (1×105 cells) in 200 μl culture medium for 18 h. (c) Four-hour or 4-day PEC (1×106 cells) were cultured in the presence or absence of 4,000 U/ml WX1 or human soluble FasL, or 2.5×105 cells of indicated transfectants in 500 μl culture medium for 18 h. (a–c) The bars indicate the means of IL-1β concentration (triplicates) in peritoneal wash or culture supernatant determined by ELISA. Vertical lines indicate standard deviations.

3 Discussion

It was recently reported that soluble FasL was chemotactic for human and/or mouse neutrophils in vitro, and suggested that the inflammatory activity of FasL in vivo was mediated by soluble FasL 16, 17. However, we previously found that MethA tumor cells expressing uncleavable mouse FasL (MAFL) induced neutrophil infiltration when transplanted into the peritoneal cavity of syngeneic mice. The uncleavable FasL protein expressed in the MAFL was the chimeric mouse FasL in which the cytoplasmic, transmembrane, and part of extracellular region containing the cleavage site was replaced with corresponding portions of the CD40 ligand, and thus no soluble FasL was detected in the culture supernatant of MAFL. Recently, Hohlbaum et al. 19 also demonstrated that L5178Y thymic lymphoma cell line transfected with uncleavable membrane FasL of mouse origin induced neutrophil infiltration in vivo. They also showed that the transformants producing artificial mouse soluble FasL that retained cytotoxic activity (like WX1 described above) could induce neutrophil infiltration, whereas those producing non-cytotoxic soluble FasL (like natural soluble FasL of mice) could not. These results are contradictory to the idea that the soluble form of FasL is responsible for its inflammatory activity. However, it was still possible that the soluble form of human FasL plays an important role in its inflammatory activity, because (1) in vitro chemotactic activity to human and mouse neutrophils has been shown using human but not mouse soluble FasL, and (2) the former but not the latter is biologically active in terms of the apoptosis-inducing activity as described above. Therefore, here we compared the inflammatory activity of the membrane-bound and soluble forms of human FasL. Our results clearly indicated that tumor cells expressing the membrane-bound but not soluble form induce neutrophil infiltration in vivo (Fig. 2). It has been reported that 0.1–10 nM soluble FasL induces chemotaxis of mouse neutrophils in vitro16, 17. Since the molecular mass of natural soluble FasL is 27 kDa, 0.1–10 nM corresponds to 2.7–270 ng/ ml. Taking dilution and clearance in vivo into account, we administered 0.1–10,000 ng purified soluble FasL into mice i.p., and examined PEC after 4, 9, and 18 h (Fig. 3). However, significant neutrophil infiltration was not observed under any of the conditions tested. These results led us to conclude that natural soluble FasL in humans is not chemoattractive to neutrophils in vivo. Thus, in vitro chemotaxis assay may not always relevant for investigating chemotactic activity in vivo.

FBL-3 tumor cells transplanted in syngeneic mice develop tumors once, but spontaneously regress within 3 weeks. This rejection is mediated by T cells as described above. Since soluble FasL is able to induce apoptosis in Con A-activated T cells 4, but unable to induce inflammation, we expected that FFS tumor cells, which expressed only soluble FasL, might grow progressively in syngeneic mice. This was the reason why we chose FBL-3 cells as a parental cell line of transfectants in this study. However, FFS tumor cells were rejected just like control tumors (Fig. 5). The amount of soluble FasL produced by the FFS cells may not be sufficient to kill activated T cells effectively, since relatively high concentrations (more than 2,000 U/ml) of soluble FasL are required to induce apoptosis at significant levels in activated T cells 4. In any case, our results shown Fig. 4 indicated that the acceleration of tumor rejection is also mediated by the membrane-bound form of FasL.

We previously discovered that cells expressing FasL are capable of inducing apoptosis in 4-h PEC (mainly neutrophils), and cause the release of active IL-1β from the apoptotic cells 14. We further demonstrated that FasL-induced neutrophil infiltration is significantly suppressed in IL-1-deficient mice compared with wild-type mice. Therefore we concluded that activeIL-1β released from inflammatory cells upon FasL-induced apoptosis causes enhancement of neutrophil infiltration. As shown in Fig. 3b, we confirmed the ability of IL-1β to induce neutrophil infiltration in vivo. Consistent with this notion, the IL-1β release-inducing activity of FasL was also mediated by membrane-bound FasL (Fig. 5). As shown in our previous report 14 and Fig. 5c, mouse recombinant soluble FasL, WX1 induced IL-1β release in 4-h PEC. However, WX1 possesses artificially strong cytotoxicity as described above. On the other hand, human soluble FasL has limited cytotoxicity compared with membrane-bound FasL or WX1 4, 20, 21. Unlike membrane-bound FasL and WX1, natural human soluble FasL was unable to induce apoptosis in 4-h PEC (data not shown). Thus, IL-1β release-inducing activity of FasL correlates wellwith its apoptosis-inducing activity in neutrophils. Here, we found that macrophages (4-day PEC) also produce IL-1β in response to membrane-bound FasL but not WX1 (Fig. 5c). Since resident macrophages exist in the peritoneal cavity as well as various tissues, IL-1β released from macrophages may play an important role in the initiation of FasL-induced neutrophil infiltration.

In conclusion, the results shown here demonstrated that the membrane-bound form of FasL primarily mediates the activities of FasL in inducing neutrophil infiltration, promoting tumor rejectionin vivo, and IL-1β release-inducing activity. Soluble FasL showed none of these activities. These findings support our previous conclusion that IL-1β mediates FasL-induced neutrophil infiltration.

4 Materials and methods

4.1 Mice and reagents

Female C57BL/6 and ddY mice were purchased from SLC Inc. (Shizuoka, Japan) at 8 weeks of age, and used between 8–12 weeks of age. Thioglycollate medium without dextrose (Difco, Detroit, MI) was dissolved in water to 3%, autoclaved, and stored in the dark at room temperature. Recombinant mouse IL-1β was purchased from Pepro Tech EC Ltd., London, GB)

4.2 Plasmid construction and establishment of transfectants

The expression plasmid for human FasL lacking a cytoplasmic region (amino acids 8–69) (FasLDC) has been described previously 22. The expression plasmid for human FasL lacking exon 2 (amino acids 116–133, which contains the cleavage site) as well as amino acids 8–69 (FasLDC2) was generated by recombinant PCR as described previously 20, except that a sense deletion primer (ACAGAAGGA-GCCCAGTCCAC) and an antisense deletion primer (GTGGACTGGGCTCCTTCTGT) were used. The expression plasmid for a secretory FasL [FasLS, a fusion protein consisting of thesignal sequence of mouse Fas (amino acids 1–31) and TNF-homologous region of human FasL (amino acids 137–281)] was generated as follows. A cDNA corresponding to the signal sequence of mouse Fas wasamplified using pBOS-EA 23, which contains the mouse Fas cDNA as the template. The sense primer (SVEFS1472) contained a 5′ flanking sequence in the pEF-BOS vector 24 and the antisense fusion primer sequence was AGCTCCTTTTTTTCAGGGGGTAAGCTTTCGGAGATGCTAT. A cDNA corresponding to the TNF-homologous region of human FasL was amplified using pBOS-HFL1 25, which contains the human FasL cDNA, as the template. The sense fusion primer was ATAGCATCTCCGAAAGCTTACCCCCTGAAAAAAAGGAGCT, and the antisense primer (BOSA1) contained a 3prime; flanking sequence found in pEF-BOS. The PCR products were purified, mixed 1:1, and then amplified by secondary PCR with primers SVEFS1472 and BOSA1. The resultant DNA fragments were digested with XbaI and inserted into pEF-BOS.

The FBL-3 erythroleukemia cell line 18 derived from C57BL/6 mice was cotransfected with each of the expression plasmids described above together with pBL-HygB, which carriesthe hygromycin B resistance gene. The hygromycin B-resistant clones expressing FasL were selected by flow cytometry or using a cytotoxicity assay as described below.

4.3 Purified soluble FasL

Naturally processed soluble FasL of humans was purified from the culture supernatant of K562 cells transfected with FasLDC cDNA using an affinity column coupled to an anti-human FasL mAb (4A5)as described previously 20. The purified soluble FasL was active in that it killed Fas-expressing W4 cells 26. The specific activity of the purified soluble FasL determined by the cytotoxic assay described below was approximately 2×107 U/mg. There was no contamination of the membrane-bound form of FasL in this preparation based on a SDS-PAGE analysis, and a gel filtration analysis indicated that the soluble FasL was a trimmer 3, 20. Purified WX1, a recombinant soluble FasL of mouse origin were prepared as described previously 4.

4.4 Flow cytometry

Approximately 1×105 transfectants were pretreated with 1 μg/ml of FcBlockR (PharMingen, San Diego, CA) for 10 min, and stained with biotinylated anti-human FasL mAb (4H9) followed by PE-conjugated streptavidin on ice. Approximately 1×105 PEC were pretreated with FcBlockR as described above, and stained with FITC-conjugated anti-Gr-1 and PE-conjugated anti-B220 mAb (PharMingen) on ice. Cells were then analyzed using a FACSCalibur (Becton Dickinson, Mountain View, CA).

4.5 Cytotoxicity assay

The cytotoxic activity of soluble FasL was assayed as described previously 26, except that the W4.5 instead of the W4 cell line was used as the target. W4.5 is a sub-line ofW4 that has approximately fivefold greater FasL sensitivity. Samples were titrated in this assay in parallel with a standard recombinant mouse soluble FasL (WX1) 26, and units ofsamples were determined by comparing with the standard.

4.6 Assay for in vivotumor growth

The right and left sides of the dorsal skin of 8–12-week-old female C57BL/6 mice were each inoculated intradermally with 2.5×106 tumor cells. The long and short diameters of the tumors were measured using calipers at 3-day intervals. Tumor size was calculated as: tumor size (mm2) = long diameter (mm) × short diameter (mm).

4.7 Preparation and culture of PEC

FasL transfectants (4×106) were transplanted i.p. into syngeneic C57BL/6 mice. Alternatively, 1.5 ml 3% thioglycollate medium (Difco, Detroit, MI) was injected i.p. into C57BL/6 mice. Mice were killed 18 h after tumor transplantation, and 4 h or 4 days after thioglycollate injection. PEC were harvested, treated with red blood cell lysis buffer (17 mM Tris-HCl pH 7.2, 140 mMNH4Cl, 2% FCS) when necessary, and washed three times with 10% FCS-RPMI. Thioglycollate-induced PEC at 2×106–2.5×106 cells/ml were then cultured with 5×105/ml of FasL transfectants or soluble FasL in RPMI medium supplemented with 10% FCS, 10 mM Hepes, 1 mM sodium pyruvate, 50 μM 2-ME, 100 U/ml benzyl-penicillin potassium, and 100 μg/ml streptomycin sulfate in 96-well plates for 18 h.

4.8 Quantification of IL-1β

The concentration of IL-1β in the supernatants of peritoneal wash and PEC culture was measured using the mouse IL-1β ELISA DuoSet (Genzyme, Cambridge, MA) according to the manufacturer's protocol.

Acknowledgements

We thank Ms. R. Hayashi for secretarial assistance and Izumi Hashitani for technical assistance. This study was supported in part by Special Coordination Funds of the Science and Technology Agency of the Japanese Government, by grants-in-aid from the Ministry of Education, Science and Culture of Japan, by the Naito Foundation, and by the Hokkoku CancerResearch Foundation.

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH

  5. 5

    WILEY-VCH

Ancillary