This study examined the properties of PP6R3, as an example of the conserved family of SAPS domain subunits that are specific for binding to PP6 phosphatase. The PP6R3 co-precipitated half the endogenous PP6c in different cell lines, showing it was the major binding partner for PP6, compared to the other SAPS subunits, and other proteins known to bind PP6c, such as alpha4 [22
] and TIP [25
], a.k.a. TAB4 [26
]. Expression of the human SAPS subunits in yeast allowed for the co-precipitation of endogenous Sit4 with each of the SAPS. However, only PP6R3 fully rescued strains deleted for all yeast SAPS with restoration of budding, indicative of cell cycle progression [16
]. We conclude that PP6R3 is a dominant and functionally conserved cellular partner for PP6c, making it a suitable representative of the family of SAPS proteins.
Our hypothesis is that the SAPS region of PP6R3 recognizes PP6c using alpha helical repeats that present residues in the inter-helical loops for subunit-subunit interaction. This follows in part from what is known about the PP2A complex with its scaffold A or PR65 subunit [6
], but diverges in specific details. The N terminal 513 residues of PP6R3 are sufficient for stable association with PP6c, showing truncation of the C terminal 350+ residues of PP6R3 does not eliminate, or even diminish binding of PP6c. The 1-513 fragment of PP6R3 supports co-precipitation of PP6c, but a 1-355 fragment is not sufficient, suggesting that residues between 355 and 513 are critical, either for direct contact with the catalytic subunit, or for maintenance of the conformation of the SAPS domain. The inability of the 1-355 fragment to co-precipitate PP6c does not imply that it is not required or involved, just that it alone is insufficient. Clearly, point mutations within the 1-355 region (at 204, 205, 259, 262) were effective at nearly eliminating PP6c binding to full length PP6R3. This at least implies that these residues are somehow required and even dominant, though not alone sufficient, for stable subunit-subunit association. We speculate that there are contacts between PP6c and two regions of the SAPS domain, involving residues in the 200-265 region as well as residues in the 355-513 region. Multiple points of contact may well be necessary to achieve the specificity for PP6 vs. PP2A. Determination of the co-crystal structure will be needed to visualize the spatial organization of the SAPS domain and interactions with PP6c.
As a preliminary step in understanding the SAPS domain organization, and as a guide for mutagenesis we used sequence alignments and jury modelling methods to produce models of the SAPS region of PP6R3. The sequences of SAPS domains are sufficiently conserved so as to allow discovery of orthologs in species from yeast to mammals [14
], and have been assigned a Pfam http://pfam.sanger.ac.uk/
designation, however the sequences do not allow a match to any structure based on available algorithms. The SAPS domain is predicted to have regions of alpha helical secondary structure, and the CD spectrum of an isolated SAPS domain indicates predominantly alpha-helical organization. We turned to jury modelling methods to produce hypothetical structures for the SAPS domain, using other known helical repeat proteins such as importins, beta-catenin, PP2A scaffolding PR65 subunit and golgin p115. These models resemble one another in that they are made of multiple helices, but each has different helix segments, inter-helix loops, and positioning of the helices relative to one another. Because previous studies of PP2A binding to its helical scaffolding (PR65) or regulatory subunit alpha4 showed basic residues on the phosphatase catalytic subunit were required, and these presumably paired with acidic side chains in other subunit [27
], we focused on acidic residues in PP6R3 as likely sites for contacts with PP6c. Reduced PP6 binding due to charge reversal mutations was taken as evidence for involvement of specific residues in subunit-subunit association. This analysis implicated E204, E205, E2659 and E262 as possible participants. Helical repeat models for SAPS domain based on beta-catenin (Figure ), or on golgin p115 (Figure ) position residues 204, 205, 259 and 262 proximal to one another, on the same side of the protein surface, in contrast to the 3D model based on PR65 A subunit (Figure ) that positions these loops on opposite sides. Therefore, results of our mutagenesis studies suggest either model B or C in Figure for arrangement of the alpha helices in the SAPS domain.
The sensitivity of PP6R3 to proteolytic cleavage about 330-350 residues from the N terminus was noted with trypsin digestion of purified recombinant GST fusion protein, and with endogenous protease cleavage of various sized FLAG-tagged proteins expressed in intact cells, and with chymotrypsin digestion of FLAG-tagged PP6R3 immunoprecipitated from cells. In the 3D model based on the golgin p115 structure (Figure ) residues 320-340 are not in a helix, but instead predicted to be in an exposed surface loop (Figure ), whereas in the beta-catenin structural model (Figure ) this segment is mostly in a helical conformation. There are two tandem Pro-Pro sequences at 323-324 and 338-339 likely to prevent alpha helix formation and the intervening sequence contains KKS and TWG as possible sites for trypsin and chymotrypsin. This putative surface loop and the juxtaposition of the 204, 205, 259 and 262 residues makes the golgin p115 structure, compose of ARM repeats, our favored model for the SAPS domain in PP6R3. We speculate that some structural element in the 350-513 sequence region, following this loop in the golgin p115 structure, constitutes a required contact site for binding to PP6c. The low sequence similarity between the golgin p115 and other ARM repeat proteins, despite ARM structures that are remarkably similar, tells us that the sampling of ARM, HEAT, ANK and PUM repeat proteins in the PDB database is still limited. This may account for why the SAPS domain sequence did not match any known structures.
Finally, binding of the PP6 catalytic subunit to the SAPS region of PP6R3 leaves the C terminal region of PP6R3 for interaction with the third subunit of the PP6 trimer, one of the ankyrin-repeat subunits (ARS), proteins known previously as Ankrd28, Ankrd44 and Ankrd52. This hypothetical arrangement is similar to, but different from, the actual 3D structures determined for PP2A heterotrimers by X-ray crystallography. Most obvious, the entire A subunit for PP2A is HEAT repeats, while no predictions suggest the same is true for PP6 subunits, where the SAPS domain only includes the N terminal half of the protein. The PR65 or A subunit scaffold for PP2A is an open arc of side-by-side helices in the AC dimer, but bends into a more closed horseshoe or letter "C" shape in the ABC trimer structures (see Fig. ). The B regulatory subunits or proteins such as small t antigen evidently induce the conformational change in the scaffold subunit [28
]. Other evidence prior to the determination of the crystal structure indicated that small t antigen affected the conformation of the PP2A scaffold subunit [29
]. The protease sensitivity of PP6R3 at ~330 residues from N terminus suggests that there may be a flexible or disordered and exposed junction. In PP6 trimers the subunit made of ankyrin repeats is expected to contact PP6c to alter substrate specificity while primarily tethered to the C terminal region of PP6R3, which alone is sufficient for stable association. The structures of known ankyrin-repeat proteins such as ankyrin, IκB, and MYPT1 show that neighboring helical repeats produce curvature in the overall structure [1
]. Thus, we imagine that the PP6 catalytic and ARS subunits could come into proximity to one another by their mutual binding to PP6R3. Such a model, like what is seen with PP2A, would predict that substrate specificity and possible regulation of activity arises from interaction of the catalytic and regulatory subunits (ARS) brought together by tethering to a common scaffold (SAPS subunit).