Choice of biological source material supersedes oxidative stress in its influence on DJ-1 in vivo interactions with Hsp90

Abstract

DJ-1 is a small but relatively abundant protein of unknown function that may undergo stress-dependent cellular translocation and has been implicated in both neurodegenerative diseases and cancer. As such, DJ-1 may be an excellent study object to elucidate the relative influence of the cellular context on its interactome and for exploring whether acute exposure to oxidative stressors alters its molecular environment. Using quantitative mass spectrometry, we conducted comparative DJ-1 interactome analyses from in vivo cross-linked brains or livers and from hydrogen peroxide-treated or naïve embryonic stem cells. The analysis identified a subset of glycolytic enzymes, heat shock proteins 70 and 90, and peroxiredoxins as interactors of DJ-1. Consistent with a role of DJ-1 in Hsp90 chaperone biology, we document destabilization of Hsp90 clients in DJ-1 knockout cells. We further demonstrate the existence of a C106 sulfinic acid modification within DJ-1 and thereby establish that this previously inferred modification also exists in vivo. Our data suggest that caution has to be exerted in interpreting interactome data obtained from a single biological source material and identify a role of DJ-1 as an oxidative stress sensor and partner of a molecular machinery notorious for its involvement in cell fate decisions.

Figure 1

Quantitative differential profiling strategy employed in this work for the comparison of DJ-1 interactomes from brain and liver. (A) Biological source materials in DJ-1 multiplex interactome comparisons were tissues obtained from wild-type (+/+) or DJ-1 knockout (−/−) mice. Following time-controlled transcardiac perfusion, cross-linking protein complexes were stringently purified with protein A agarose-conjugated antibodies recognizing DJ-1, followed by alkylation and trypsinization. Tryptic sample digests were side-by-side iTRAQ labeled and subsequently combined. Two-dimensional liquid chromatography of peptides was coupled to online ESI-MS/MS. (B) On the basis of their iTRAQ signature, mass peak ratios of proteins were grouped into four different candidate categories: (i) Tissue-Independent Specific Binder Category, (ii) Tissue-Dependent Specific Binder Category, (iii) Tissue-Independent Nonspecific Binder Category, and (iv) Tissue-Dependent Nonspecific Binder Category. Please note that the assignment to these groups is merely intended to facilitate the grouping of proteins and required the ratio of the respective signature mass peaks to exceed an arbitrarily selected threshold of >3.0.

Figure 2

Alternative strategies used to quantitatively compare DJ-1 interactomes from DJ-1 deficient (−/−), heterozygote (+/−), and wild-type ES cells (+/+) treated with H₂O₂ or untreated. The formaldehyde cross-linking step was achieved by direct addition of the cross-linking solution to adherent cells. Cross-linked protein complexes were stringently purified with protein A agarose conjugated antibodies recognizing DJ-1, followed by alkylation and trypsinization. Tryptic sample digests were either (i) analyzed by consecutive two-dimensional LC, ESI tandem mass spectrometry, and spectral counting quantitation or (ii) side-by-side iTRAQ labeled, subsequently combined, LC fractionated, and analyzed by ESI–MS/MS.

Figure 3

Direct evidence of DJ-1 Cys106 sulfinic acid modification after H₂O₂ treatment of ES cells. Four-plex comparison of the relative contribution that samples made to the identification of two DJ-1-derived peptides. Parallel co-IPs from DJ-1 deficient, DJ-1 heterozygote, and wild-type ES cells treated or untreated with H₂O₂ were followed by iTRAQ-based quantitation and ESI–MS/MS analyses. iTRAQ labeling reactions were set up as indicated in Figure 3, i.e., iTRAQ114 label: DJ-1^−/−; iTRAQ115 label: DJ-1^−/+; iTRAQ116 label: wild-type; and iTRAQ117 label: H₂O₂-treated wild-type ES cells. (A) CID spectrum assigned to tryptic peptide surrounding the Cys106 residue (GLIAAICAGPTALLAHEVGFGCK). iTRAQ signature mass peak ratios document contribution of this peptide primarily from the H₂O₂-treated wild-type IP eluate, with minor contributions from DJ-1 heterozygote and untreated wild-type ES cells and absence of this peptide in the DJ-1 knockout eluate. (B) Representative CID spectrum assigned to an unmodified DJ-1-derived peptide indicating relatively similar contribution of this peptide from all DJ-1 eluate samples except the negative control. Insets: high resolution graphs depicting iTRAQ signature mass peaks.

Figure 4

Expression levels of DJ-1 candidate interactors are a poor predictor for DJ-1 specific coenrichment. Comparison of relative protein levels of DJ-1 and its candidate interactors in different biological source materials. Western blotting analysis using equal protein amounts from brain and liver tissues of DJ-1 knockout and wild-type mice and from DJ-1 knockout, DJ-1 heterozygote, and wild-type ES cells was used to compare the protein levels of a selection of the candidate DJ-1 interactors. Despite the highly selective coenrichment of Hsp90 with DJ-1 only in ES cell-derived samples, similar levels of protein expression can be observed for Hsp90 and Hsc70 (HspA8) in protein extracts derived from brain, liver, and ES cells.

Figure 5

DJ-1 influences the Hsp90-dependent stabilization of Raf-1 and Akt. (A) DJ-1-deficient, DJ-1 heterozygote, and wild-type ES cells were grown in the presence of carrier solvent or 1 μM of the Hsp90 inhibitor geldanamycin (GA) for 16 h and subsequently analyzed by Western blotting with antibodies directed against DJ-1 or Hsp90-client proteins Raf-1 and Akt. A blot probed with an antibody against actin served as a loading control. Note the absence of an effect of GA treatment on the expression of DJ-1 and the profound increase in Raf-1 and Akt in the presence of DJ-1. (B) Densitometric analyses confirm significant stabilization of Raf-1 and Akt in DJ-1 expressing cells when compared to DJ-1 knockout. Bar diagram summarizing analyses of three biological replicates of experiment shown in (A) (depicted as percentage of the signals observed for untreated wild-type cells). Differences in Raf-1 and Akt signal levels observed in DJ-1 deficient and DJ-1 expressing ES cells were statistically significant with a p-value below 0.05.