What can we learn from CD95 mutations?

Read Also

Human.

Germinal mutations in APT1 have been reported in patients with autoimmune lymphoproliferative syndrome type Ia (ALPS, also called Canale-Smith syndrome) [88-90]. ALPS patients exhibit chronic lymphadenopathy and splenomegaly and expanded populations of double-negative α/β T lymphocytes CD3⁺CD4⁻CD8⁻), and often develop autoimmunity [88, 89, 91, 92]. In agreement with the notion that CD95 behaves as a tumor suppressor, ALPS patients display an increased risk of Hodgkin and non-Hodgkin lymphoma [93]. The predominance of post-germinal center (GC) lymphomas in patients with either germ line or somatic CD95 mutations can be explained by the fact that, inside germinal centers of the secondary lymphoid follicles, the CD95 signal plays a pivotal role in the deletion of self-reactive maturing B lymphocytes [94]; in addition, APT1 belongs to a set of rare genes (i.e., PIM1, c-myc, PAX5, RhoH/TTF, and Bcl-6) subject to somatic hypermutation [95, 96], which may affect its biological function. In addition to post-GC lymphomas, tumors of various histological origins have been shown to exhibit significant numbers of mutations in the CD95 gene (reviewed in [51]). Extensive analysis of CD95 mutations and their distribution in APT1 reveals that, with some exceptions, most are gathered in exons 8 and 9, which encode the CD95 intracellular region (Figure 3) [97]. Remarkably, most of these mutations are heterozygous, mainly localized in CD95-DD, and lead to inhibition of the CD95-mediated apoptotic signal. Indeed, in agreement with the notion that CD95 is expressed at the plasma membrane as a pre-associated homotrimer [19, 20], formation of heterocomplexes containing wild-type and mutated CD95 prevents FADD recruitment and dominantly abrogates the initiation of the apoptotic signal.

Extensive analysis of the positions of CD95 mutations described in the literature has revealed mutation “hot spots” in the CD95 sequence (Figure 3). Among these hot spots, arginine 234, aspartic acid 244, and valine 251 account for a considerable proportion of the documented CD95 mutations. Indeed, among the 189 mutations annotated in the 335 amino acids of CD95, 30 (~16%) are localized in one of these three amino acids (Figure 3). The pivotal roles played by these amino acids in stabilization or formation of intra- and inter-bridges between CD95 and FADD may explain the existence of these hot spots. For instance, both R234 and D244 contribute to homotypic aggregation of the receptor and FADD recruitment [59]. Nevertheless, the observation of death-domain hot spots contradicts the study of Scott and colleagues, who demonstrated that the region of CD95-DD that interacts with the FADD-DD extends over a dispersed surface and is mediated by a large number of low-affinity interactions [60].

Most ALPS type Ia patients affected by malignancies do not undergo loss of heterozygosity (LOH), leading some authors to hypothesize that preservation of a wild-type allele may contribute to carcinogenesis [98, 99]. In the same vein, expression of a unique mutated CD95 allele blocks the induction of apoptotic signals, but fails to block non-apoptotic signals such as NF-κB and MAPK [98, 99], whose induction promotes invasiveness in tumor cells [97, 100]. In addition, mutations in the intracellular CD95-DD result in more highly penetrant ALPS phenotype features in mutation-bearing relatives than mutations in the extracellular domain. These results suggest that unlike DD mutations, CD95 mutations localized outside the DD somehow block apoptotic signaling but fail to promote non-apoptotic pathways that may contribute to disease aggressiveness.

 Mouse models.

Three mouse models exist in which either CD95L affinity for CD95 is reduced (due to the germline mutation F273L in CD95L, called generalized lymphoproliferative disease [gld], which decreases CD95L binding to CD95) [101, 102]), the level of CD95 expression is down-regulated (due to an insertion of a retrotransposon in intron 2 of the receptor gene, these mice are called lymphoproliferation [Lpr] [103-105]), or DISC formation is hampered (due to a spontaneous mutation inside the CD95 DD at position 238, specifically, replacement of the valine 238 with asparagine; these mice are called lpr^cg for lpr gene complementing gld [106]). These mice have provided valuable insights into the pivotal role played by CD95 and CD95L in immune surveillance and immune tolerance [107]. In an attempt to simplify, some authors associated the phenotypes observed in these lpr, lpr^cg or gld mice with the complete loss of CD95 or CD95L [108]. However, conclusions must be drawn with caution, due to subtle differences between the phenotypes of spontaneous mouse models and genetically engineered mice. Indeed, in Lpr mice, insertion of an early transposon in intron 2 of CD95 causes premature termination of the CD95 transcript [104], which is leaky; consequently, CD95 mRNA and protein can be detected in mice homozygous for the spontaneous mutation [109, 110]. Also, the DD mutation in Lpr^cg mice reduces FADD recruitment but does not abrogate it [111]. Furthermore, CD95 can still interact with CD95L harboring the gld mutation, albeit somewhat more weakly than wild-type CD95L [112]. Finally, lpr, lpr^cg, and gld mice over-express CD95L relative to their wild-type counterparts [113]. Using T lymphocytes from ALPS type Ia patients or Lpr mice, we confirmed that far less intact CD95 is required to activate NF-κB than to induce apoptosis; therefore, although a single wild-type allele cannot restore cell death induction in these cells, it is sufficient to transduce NF-κB and MAPK cues [98, 99]. Overall, these observations support the idea that the biological roles ascribed to the CD95/CD95L pair, based on the analysis of these patients and mouse models, may correspond to the additive effects of the receptor’s inability to induce cell death and its tendency to implement non-apoptotic signals.

A recent study elegantly showed that elimination of the remaining allele in cancer cells leads to the induction of an unconventional cell death program called “death induced by CD95R/L elimination” (DICE) [114].

These findings highlight the fact that distinct activation thresholds exist in the process of CD95 engagement. Although complete loss of CD95 expression in cancer cells leads to cell death, one wild-type allele (low activation threshold) is sufficient to elicit non-apoptotic signaling pathways, and the second allele (high activation threshold) is required to implement the canonical apoptotic signal [98, 115]. However, this rule suffers from an exception: metalloprotease-cleaved CD95L implements non-apoptotic signals in cells expressing two wild-type alleles of CD95 [48, 50, 116, 117] (further discussed in 3.6.2). In summary, because the characterization of CD95/CD95L biological roles has been carried out mainly by considering the default of apoptosis in ALPS type Ia patients and mouse models, we believe it is important to carefully reconsider these conclusions by integrating the notion that exposure of these cells to CD95L will also lead to a chronic activation of non-apoptotic signaling pathways [99]. To better appreciate the complexity of the pathophysiological roles of CD95 and its ligand, it is therefore more appropriate to use conditional and tissue-specific CD95 and CD95L KO mice.