We thank de Vries et al. (1) for their insightful and informative remarks. They have highlighted that differences in the preparation of the recombinant TRAIL1 rather than possible interspecies variability may explain some conflicting reports on TRAIL toxicity to normal human cells (2, 3, 4, 5, 6, 7, 8) . They also suggest that the detection of serum levels of TRAIL in some human diseases might indicate that the human body can tolerate certain levels of TRAIL.
In our review (9) we had already pinpointed that differences in the structure of distinct preparations of recombinant TRAIL used in different studies should be investigated as a potential cause of toxicity in humans. In line with that, Lawrence et al. (5) have demonstrated that whereas soluble, untagged TRAIL induced no hepatocyte toxicity, a His-tagged preparation of TRAIL exerted strong hepatotoxicity. The authors linked this toxicity to the formation of multimeric aggregates (5) . It has been suggested that the tumor-cell selectivity of TRAIL may be lost when other than trimeric; therefore, aggregate forms of TRAIL are used (6) . Using keratinocytes, Leverkus et al. (10) described toxicity to these cells when tagged forms of TRAIL were used but not when untagged preparations were applied. Moreover, using His-tagged TRAIL Nesterov et al. (11) observed toxicity to PrECs. In contrast, previous reports using homogeneous preparations of TRAIL that only form trimers failed to show TRAIL-induced apoptosis on PrECs (12) . Altogether, these data suggest that tagged preparations of TRAIL should not be used against human cells. However, it should be emphasized that the effect of His-tagged TRAIL on PrECs described by Nesterov et al. (11) did not seem to be because of nonspecific toxicity but rather to a TRAIL-induced effect because it could be inhibited by the extracellular domain of TRAIL receptor. In addition, the TRAIL preparation they used did not seem to be more potent than the one used by Ashkenazi et al. (12) , because the effect of both preparations on renal cancer cell lines was comparable (11 , 12) .
On the basis of that, we could speculate that TRAIL aggregates could be toxic to some types of human cells but not to others. In fact, it has been suggested that aggregate TRAIL may effectively activate not only TRAIL-R1, but also TRAIL-R2, a type of receptor that is normally not activated by soluble TRAIL (6) . The resulting double activation might be undue and underlie the toxic effect observed, again pointing against the use of tagged TRAIL preparations in humans. Nonetheless, one aspect that we should keep in mind is that precluding preparations of TRAIL that may potentially form aggregates from the clinic could mean limiting the application of TRAIL only to TRAIL-R1-predominant tumor cells.
Cynomolgus monkey is an appropriate model to characterize the safety profile of TRAIL, as suggested by Kelley et al. (13) . However, their results do not rule out that some interspecies variability in TRAIL toxicity may exist, especially because the mechanisms that determine the cellular sensitivity to TRAIL remain only partially understood. If we assume that TRAIL sensitivity of cells is controlled by different molecules at multiple levels (14) , we have to admit that this sensitivity may vary not only among cell types with different molecular background, but also among cells from different species.
Regarding the role of circulating TRAIL, we disagree from de Vries et al. (1) that the levels of TRAIL in the serum of AIDS patients might indicate that the human body can tolerate certain levels of TRAIL. The function of soluble TRAIL remains unknown, and it could in fact act as an inhibitor of some possible undesired effect of membrane-bound TRAIL (15) . Therefore, we believe this cannot be used as an argument for the safety of TRAIL at this point.
In summary, de Vries et al. (1) have a point when they highlight the possible role of tagged forms of TRAIL in the toxic effects observed against human cells by some groups. Nevertheless, de Vries et al. (1) suggest that all of the variability in the toxic effects could be explained by the different preparations, what we believe should not be assumed in the light of the thus far incomplete knowledge on the determinants of sensitivity to TRAIL in human cells. Issues of safety will only be addressed in the context of well-designed trials of recombinant TRAIL in humans. Yet, we believe these clinical studies should be coupled to additional preclinical and exploratory studies so that we can additionally understand, and fully and safely explore the potential of TRAIL as an anticancer agent.
Footnotes
-
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 The abbreviations used are: TRAIL, tumor necrosis factor-related apoptosis-inducing ligand; PrEC, prostate epithelial cell.
- Received September 26, 2002.
- Accepted October 3, 2002.
References
Volume 9, Issue 2
