To investigate potential variables leading to these divergent findings, we carried out experiments designed to replicate closely those described by Lau et al. (1). Primary skeletal myoblasts were isolated from C3H mice and transduced with a retroviral vector that directs murine FasL expression from the LTR promoter. Functional FasL expression was confirmed by cytotoxicity to Fas-expressing Jurkat cells (8). Unexpectedly, transduced myoblasts underwent rapid apoptosis during differentiation, which suggests that skeletal myoblasts express Fas, in contrast with the findings of Lau et al. (1). Although shown earlier to occur in postnatal cardiac and skeletal muscle tissues (9), Fas expression had, to our knowledge, not been examined in cultured cells. With the use of an antibody to mouse Fas, Jo2 (10), we confirmed Fas expression in myoblasts of C3H (Fig. 1) and C57BL/6 strains (11).

To avoid Fas/FasL-mediated self-destruction of myoblasts, we generated primary myoblasts from Fas-deficient C57BL/6 lpr mice, the mouse counterpart to human autoimmune lymphoproliferative syndrome (ALPS). When transduced with the FasL vector, lpr myoblasts did not self-destruct on differentiation in vitro. Nontransduced or FasL-transduced lpr myoblasts were injected under the kidney capsule of congenic C57BL/6 mice. Mice were killed 1, 3, 7, 14, and 26 days after transplantation, and their kidneys were removed for histological examination (Fig. 2). Kidneys transplanted with nontransduced myoblasts appeared normal at all time points (Fig. 2, A, C, and E). In contrast, each kidney transplanted with FasL-expressing myoblasts had a prominent white abscess that was abundant in neutrophils; these abscesses appeared by day 1, were pronounced by day 3, and disappeared by day 26 (Fig. 2, B, D, and F, respectively). Moreover, in contrast with the findings of Lau et al. (1), co-implantation of allogeneic C3H islets of Langerhans with congenic FasL-expressing myoblasts led to accelerated destruction of the islets (Fig. 2G). Untransduced myoblasts differentiated and persisted for at least 26 days, but FasL myoblasts were destroyed by the granulocytic infiltrate (Fig. 2, H and I). It is therefore unclear how Lau et al. (1) were able to generate FasL-expressing myoblasts that persisted as differentiated multinucleated myotubes for more than 80 days in vivo. Moreover, our data show that islet destruction is not prevented but accelerated by FasL-expressing myoblasts, presumably through a bystander effect mediated by infiltrating neutrophils.

Our findings are in direct conflict with those reported by Lau et al. (1). Subtle technical differences can perhaps be invoked; however, it is unclear from our studies how myoblasts that express both Fas and FasL can avoid apoptosis while differentiating or be available to induce apoptosis of invading lymphocytes, as proposed by Lau et al. Even if clones of non-Fas-expressing myoblasts were selected by Lau et al. (1) and were therefore spared from apoptosis, it remains an enigma how such FasL-expressing myoblasts escaped the granulocytic response that we observed and which resulted in premature elimination of both Fas-deficient myoblasts (lpr) and islets. It has been suggested that the exact quantity of FasL expressed may be critical in determining whether immunoprotection or immunodestruction occurs (12). However, we and others have found that, although low amounts of FasL expression do not result in granulocytic infiltration, these amounts also do not protect against T cell-mediated allograft rejection (6, 13). FasL-specific activity is known to vary as a result of polymorphisms (14); however, both Lau et al. (1) and we used the C57BL/6 form of FasL known to have reduced cytotoxic potential. Taken together, these findings suggest that, although FasL may have a role along with other factors in the immune privilege of the eye and testis, expression of endogenous FasL alone is unlikely to suffice.

In support of this conclusion, most cell types and tissues that have been genetically engineered to express FasL have been shown to undergo destruction by neutrophils (5, 6). Thus, FasL expression has complex consequences (15), and further investigation of the effects of dosage, cell context, and microenvironment are warranted. Our observations, although discouraging for transplant purposes, suggest other applications for FasL and new approaches for defining the molecular determinants requisite for immune protection.

Sang-Mo Kang
Department of Surgery,
University of California,
San Francisco, CA 94143, USA
Andreas Hofmann(^*)
Department of Molecular Pharmacology,
School of Medicine, Stanford University,
Stanford, CA 94305-5332, USA
David Le
Department of Surgery,
University of California
Matthew L. Springer
Department of Molecular Pharmacology,
School of Medicine, Stanford University
Peter G. Stock
Department of Surgery,
University of California
Helen M. Blau
Department of Molecular Pharmacology,
School of Medicine, Stanford University
E-mail: hblau{at}cmgm.stanford.edu
(^*) Present address: Institut für Biochemie I, Medizinische Fakultät der Universität zu Köln, 50931 Köln, Germany.

REFERENCES AND NOTES

1. H. T. Lau, M. Yu, A. Fontana, C. J. Stoeckert Jr., Science 273, 109 (1996) [Abstract] .
2. T. S. Griffith, T. Brunner, S. M. Fletcher, D. R. Green, T. A. Ferguson, ibid. 270, 1189 (1995) [Abstract/Free Full Text].
3. D. Bellgrau et al., Nature 377, 630 (1995) [CrossRef] [Medline].
4. M. Hahne et al., Science 274, 1363 (1996) [Abstract/Free Full Text].
5. K. Seino, N. Kayagaki, K. Okumura, H. Yagita, Nature Med. 3, 165 (1997) [CrossRef] [Web of Science] [Medline] .
6. J. Allison, H. M. Georgiou, A. Strasser, D. L. Vaux, Proc. Natl. Acad. Sci. U.S.A. 94, 3943 (1997) [Abstract/Free Full Text] ; S.-M. Kang et al., Nature Med. 3, 738 (1997) [CrossRef] [Web of Science] [Medline] .
7. H. Arai, S. Y. Chan, D. K. Bishop, G. J. Nabel, Nature Med. 3, 843 (1997) [CrossRef] [Web of Science] [Medline] .
8. A. Hofmann and H. M. Blau, Somat. Cell Mol. Genet., in press.
9. H. Yamada, M. Nakagawa, I. Higuchi, R. Ohkubo, M. Osame, J. Neurol. Sci. 134, 115 (1995) [CrossRef] [Web of Science] [Medline] ; R. Watanabe-Fukunaga , et al. J. Immunol. 148, 1274 (1992) [Abstract] .
10. J. Ogasawara, et al., Nature 364, 806 (1993) [CrossRef] [Medline] .
11. S.-M. Kang et al., data not shown.
12. H. T. Lau and C. J. Stoeckert, Nature Med. 3, 727 (1997) [CrossRef] [Web of Science] [Medline] .
13. D. Vaux, personal communication.
14. N. Kayagaki et al., Proc. Natl. Acad. Sci. U.S.A. 94, 3914 (1997) [Abstract/Free Full Text] .
15. D. R. Green and C. F. Ware, ibid., p. 5986.
16. We thank S. Nagata for supplying cDNA for Southern blotting probes; C. Goodnow for supplying lpr mice; Z. Lin, P. Kraft, and N. Bracey for technical assistance; and N. Asher, A. Estellés, and D. Vaux for helpful discussion. Supported by a Howard Hughes Medical Institute Physician Research Fellowship to S.M.K., a postdoctoral fellowship from the Deutsche Forschungsgemeinschaft to A.H., an NSRA (F32 HL08991) to M.L.S., and grants from NIH (R01-CA59717, R01-HD18179, and MERIT award R37-AG09521) to H.M.B.

Response: Kang et al. raise several issues with regard to engineered expression of FasL on myoblast as a means of giving immunoprotection to islet allografts. In our initial studies, we observed the expression of FasL on myoblasts from C57BL/6 mice cultured over a long term; when differentiated in vitro, these cells did not undergo apoptosis and continued to express functional FasL. In contrast, Kang et al. report that FasL expression of C3H myoblasts results in apoptosis after differentiation. It is unclear why FasL expression differs among different myoblast populations; perhaps apoptosis resistance is acquired during multiple passage, or susceptibility to FasL-induced apoptosis may be strain dependent. We have transfected nonobese diabetic (NOD) mice myoblasts and observed spontaneous cell death; however, when myoblasts were preselected with prolonged culture with soluble FasL, we obtained apoptosis-resistant cells that permitted subsequent functional expression of FasL. Preliminary co-transplantation experiments with allogeneic islets with these NOD FasL⁺ myoblasts have not resulted in prolonged survival. This cell mortality may be a result of the greater complexity in the killing of primed T cells that infiltrate and destroy the islets, as one would expect to find in the diabetic NOD recipient.

With regard to the issue of neutrophilic infiltration, we have reexamined the histology of the composite grafts from our initial study on day 3 after transplantation and have observed local inflammation, but islets and myoblast were present. Histology at day 7 revealed local pockets of neutrophilic infiltration, but again, islets and myoblast were identified. By the fifth week after transplantation [as we originally observed (1)], there was resolution of the inflammation, and fused muscle cells were seen on histology. The prolongation of islet allograft survival we observed appears to be a bystander effect of local expression of FasL in which there is no specificity in the killing of infiltrating T cells. Thus, there may be a race between the muscle cells and islets to survive the initial inflammation and still effect apoptosis of infiltrating activated T cells directed against the allogeneic islets. Under such circumstances, transplantation of borderline numbers of islets required for correction of hyperglycemia would not result in long-term correction of the diabetic state, because the initial inflammation would result in some attrition of islets. In this regard, as noted in our initial studies (1), we observed fluctuation in glucose in diabetic mice receiving the highest numbers of FasL expressing myoblast (2 × 10⁶). In retrospect, this may have represented the initial inflammation, which subsided 3 weeks after transplantation, as reflected in stabilization of glucose in the blood (1). The amount of FasL expression may be critical in this regard in that there may be a balance between FasL inducted local inflammation and immunoprotection (2).

We agree that immune privilege is more than the expression of FasL and that there are other factors at work, especially in light of our inability to extend these findings to the NOD mouse model of spontaneous diabetes. However, our initial studies showing that even a bystander effect can prolong islet allograft survival (1) suggest that expression of FasL in the context of alloantigen or autoantigen (as in the case of islets) may enable specific killing of T cells that are activated toward such co-expressed antigens. Although muscle cell expression of FasL in a local fashion may not be applicable across all stains or species (because of self apoptosis or destruction by neutrophils in a confined space), ectopic FasL expression (in the context of an alloantigen or autoantigen on engineered cells such as muscle) administered systemically may effect specific attenuation of an immune response. This attenuation has been demonstrated by Arai et al. with the use of an allogeneic tumor cell engineered to express FasL (3). These recent findings may help define a role for the use of engineered FasL expression in the modulation of immune response.

Henry T. Lau
Department of Pediatric Surgery,
Johns Hopkins Hospital,
Baltimore, MD 21287-3716, USA
C. J. Stoeckert
Department of Hematology,
Children's Hospital of Philadelphia,
Philadelphia, PA 19104, USA

REFERENCES

1. H. T. Lau, M. Yu, A. Fontana, C. J. Stoeckert Jr., Science 273, 109 (1996) .
2. H. T. Lau and C. J. Stoeckert, Nature Med. 3, 727 (1997) .
3. H. Arai, S. Y. Chan, D. K. Bishop, G. J. Nabel, ibid., p. 843.