Dear Editor,

FAS-associated death domain (FADD) protein is an adaptor protein that connects death receptors with caspase-8 at the cell membrane. We reported that FADD is, unexpectedly, mainly nuclear,1 and this was subsequently confirmed by another laboratory.2 However, a recent paper in this journal by the Strasser laboratory has challenged this conclusion.3 They developed a series of monoclonal antibodies against FADD, one of which, clone 12E7, produced a mainly cytoplasmic signal in HeLa and L929 cells.

In the Strasser paper, cells were formaldehyde fixed and permeabilized with saponin. I compared saponin versus the more commonly used Triton X-100 with respect to immunofluorescence with anti-FADD 12E7 (obtained from Dr. Strasser, WEHI, Melbourne, Australia). With saponin, 12E7 localized FADD primarily in the cytoplasm, and the Transduction Labs monoclonal localized it more variably to the cytoplasm in some cells and nucleus in others (see Figure 1 below). However, with Triton X-100, 12E7 generated a primarily nuclear signal in both MCF10a and HeLa, in agreement with our published results using the Transduction Labs monoclonal antibody and monospecific anti-mouse FADD and anti-human FADD polyclonals generated/characterized in our lab.

Figure 1
figure 1

(a) Immunofluorescent localization of FADD was performed using HeLa or MCF10a cells that were fixed with formaldehyde, followed by permeabilization with 0.2% Triton X-100 or 0.3% saponin; the latter detergent was included in all subsequent incubation and washing steps. Results are shown for 12E7 and the Transduction Labs monoclonal (TL), followed by detection with anti-rat or anti-mouse labeled with Alexa 488 (Molecular Probes). (b) Immunofluorescent localization of other nuclear proteins on Triton X-100 versus saponin-permeabilized cells. Antibodies were from the following sources: CtBP: from our laboratory;6 PML and BRCA1: from BD Pharmingen

Full size image

As both the Strasser and Transduction Labs antibodies produced detergent-dependent results, we tested a series of antibodies directed against known nuclear proteins for their behavior in saponin versus Triton X100-permeabilized cells. C-terminal binding protein (CtBP), PML, and BRCA1 antibodies produced the expected nuclear signal using Triton X-100 permeabilization. With saponin, the nuclear signal was much weaker in all cases and was replaced by a cytoplasmic signal with CtBP. These results are consistent with the view that saponin inadequately permeabilizes intracellular membranes, which hinders access of some antibodies to the nucleus, additionally trapping certain antibodies artifactually in the cytoplasm.

The Strasser paper also asserted that FADD was cytoplasmic, based on fractionation of hypotonically lysed HeLa cells. Whereas FADD was in the nuclear fraction of MCF10a cells in our experiments, it was in their soluble cytoplasmic fraction of HeLa cells. In their protocol, nuclei were washed and re-centrifuged twice prior to lysis in SDS sample buffer, while our nuclei were not washed. We have found that FADD (a 26 kDa protein, below the exclusion limit of nuclear pores) leaks from the nucleus during repeated washes (data not shown). This is true of many other known nuclear proteins. For example, DNA polymerase-alpha appears to be cytoplasmic under standard hypotonic lysis conditions, but can be retained in the nuclear fraction using sophisticated ultracold, nonaqeuous fractionation techniques;4, 5 nuclear fractionation is thus very prone to false negatives.

In summary, the results with several antibodies including 12E7 indicate a nuclear localization for FADD in adherent cell lines; this localization probably differs in lymphocytes, as the Strasser3 and Cidlowski2 labs have reported.