IL-4-stimulated TS1αβ cells (1×10⁷) were lysed for 20 min at 4 °C in lysis buffer (50 mM Tris HCl pH 8, 1% NP-40, 137 mM NaCl, 1 mM MgCl₂, 1 mM CaCl₂, 10% glycerol and protease inhibitor cocktail). Lysates were immunoprecipitated with the appropriate antibody and protein A-Sepharose was added for 1 h at 4 °C. After washing, immunoprecipitates were separated by SDS-PAGE, transferred to nitrocellulose and blocked with 5% non-fat dry milk in Tris-buffered saline (TBS, 20 mM Tris HCl pH 7.5, 150 mM NaCl) and incubated with primary antibody in TBS/0.5% non-fat dry milk. The membrane was washed with 0.05% Tween-20 in TBS and incubated with PO-conjugated secondary antibody. After washing, proteins were developed using the ECL system.

Anti-PP1α, anti-p85 PI3-K, anti-p110 PI3-K, anti-Hsp70, anti-Bcl-2, anti-Ras and anti-CD4 antibodies were from UBI (Lake Placid, NY), Calbiochem (La Jolla, CA) or Transduction Laboratories (Lexington, KY).

Peptides from Bcl-x_L and Bad were prepared as described [7,8] by automated spot synthesis (Abimed, Langerfeld, Germany) onto an amino-derivatized cellulose membrane, immobilized by their C-termini via a polyethylene glycol spacer and N-terminal acetylated.

The interaction of the targeting subunits with PP1 was competed by the F-x-x-R-x-R, (GDEFELRYRRAF) and R-I-V-A-F (NWGRIVAFFSF) peptides. Lysates were immunoprecipitated with anti-PP1α antibody and protein A Sepharose was added. The interaction was competed by incubation with the peptides (30 min, room temperature). After washing, immunoprecipitates were transferred to nitrocellulose and blotted with the corresponding antibody.

Using a murine T cell line that can be propagated independently in the presence of IL-2 IL-4 or IL-9 [9], we have recently described that PP1α is a Ras-activated phosphatase that dephosphorylates Bad, a pro-apoptotic member of the Bcl-2 family proteins, prior to induce apoptosis in response to IL-2 deprivation [5]. As shown in Table 1, two sequences (R-I-V-A-F or R-L-V-A-F) analogous to the R/K-x(0,1)-V-x-F PP1 binding consensus motif are present in the BH1 domain of three Bcl-2 family proteins. We have shown that Bcl-2, Bcl-x_L and Bcl-w require these motifs for binding and targeting PP1α to Bad. Based on biochemical competition studies, we demonstrate that Bcl-2 is a new PP1 interacting protein that binds PP1α through the R-(IL)-V-A-F sequence and targets the phosphatase to Bad [5]. In addition to this canonical docking consensus site, we have also characterized a new PP1c docking site in Bcl-x_L and Bcl-w [10]. The new docking site comprises a motif analogous to the 140-F-E-M-R-R-K-146 sequence identified as a PP1c binding site for the small cytosolic protein inhibitor-2 (I-2) (Fig. 1A) [11,12]. A new F-x-x-R/K-x-R/K consensus sequence similar to PP1α docking sites in I-2, Bcl-x_L and Bcl-w is also present in known PP1c binding proteins (Fig. 1A). Mutation of the critical F and R residues strongly reduces binding of Bcl-x_L to PP1c (Fig. 1B).

The new PP1c binding motif is also present in Bad, a proapoptotic member of the Bcl-2 family (Fig. 1B). Mutation of serine residue in the vicinity of the F-xx-R/K-xR/K motif affects binding of Bad to PP1c, although the affinity depends on the type mutation. When the Ser amino acid is replaced by an acidic Asp amino acid, mimicking a phosphorylated amino acid, the binding of the PP1c is inhibited. This result suggests that phosphorylation of Bad on the regulatory Ser-136 while allowing binding of Bad to 14–3–3 [13,14] may also inhibit binding to PP1c (Fig. 1B). Unlike Bcl-x_L or Bcl-w, this motif is located outside of the BH3 domain immediately upstream of the regulatory Ser-136.

The statistical analysis of sequences in the Swissprot Release 40 indicate that, while 16% of the data library contains one of the motifs, only 3.9% share the two motifs (Table 2). The reduced number of proteins with the two distinct PP1c docking motifs clearly supports the concept of a PP1 predictive signature. In addition, the finding that most characterized PP1-binding proteins share the two motifs sustained this concept and has motivated the Institut Pasteur to create and to maintain the http://PP1signature.pasteur.fr website as a new bioinformatic tool to identify novel putative PP1 interacting proteins. To experimentally validate this concept, we have demonstrated a PP1c association with four candidates, randomly selected in our web site (Fig. 2). We were able to detect association of PP1α to p85 PI3K, p110 PI3K, hsp70 and CD4, four molecules that share both motifs. As a control, in Ras immunoprecipitates we did not observed association of PP1α-Ras, a molecule that contains neither R/K-x(0,1)-V-x-F nor F-x-x-R/K-x-R/K motifs. Interestingly, the association of hsp70 to PP1α is almost undetectable upon competition of PP1α immunoprecipitates with exogenous peptides R and F (for sequence, see materials and methods) corresponding to the two motifs R/K-x(0,1)-V-x-F and F-x-x-R/K-x-R/K, respectively.

The recent development of penetrating peptides, in addition to the identification of peptide sequences surrounding PP1c docking sites in important regulatory proteins, suggests a new approach for phosphatase-derived drug research. This strategy might be based on the specific ‘peptide knock out’ of an intracellular pathway controlled by a specific PP1 holoenzyme. To develop this academic or therapeutic strategy as well as to facilitate academic research, we propose a new website: http://PP1signature.pasteur.fr, which allows the identification of putative PP1-interacting proteins containing the two distinct PP1c docking consensus motifs represented in the Swissprot library.