Determinants for Substrate Specificity of Protein Phosphatase 2A

Protein phosphatase 2A- (PP2A-) catalyzed dephosphorylation of target substrate proteins is widespread and critical for cellular function. PP2A is predominantly found as a heterotrimeric complex of a catalytic subunit (C), a scaffolding subunit (A), and one member of 4 families of regulatory subunits (B). Substrate specificity of the holoenzyme complex is determined by the subcellular locale the complex is confined to, selective incorporation of the B subunit, interactions with endogenous inhibitory proteins, and specific intermolecular interactions between PP2A and target substrates. Here, we discuss recent studies that have advanced our understanding of the molecular determinants for PP2A substrate specificity.


Introduction
Cellular adaptation requires biochemical processes including post-translational mechanisms to modify existing proteins. Catalyzed by opposing kinases and phosphatases, reversible phosphorylation of serine, threonine, and tyrosine residues is now appreciated as a fundamental regulatory mechanism with the majority of phosphorylation (>99%) occurring on serine and threonine residues [1,2]. Due to their untapped therapeutic potential, protein phosphatases have been identified as promising targets for xenobiotic manipulation through rational drug design (reviewed in [3][4][5][6]). In particular, the ubiquitously expressed protein phosphatase 2A (PP2A) has been proposed as a target for the treatment of a number of pathologies ranging from neurodegenerative diseases such as Alzheimer disease to a variety of neoplasias [7][8][9].
Compared to other members of the phosphoprotein phosphatase (PPP) superfamily of serine/threonine phosphatases, a detailed understanding of the mechanism by which PP2A recognizes substrates and mediates site-specific dephosphorylation remains to be developed. Sequence and structural homology of the catalytic subunits of PPP family members has revealed a conserved catalytic mechanism in which a divalent metal cation activates a water molecule to hydrolyze phospho-serine/threonine without the formation of a phosphoenzyme intermediate (reviewed in [10][11][12]).
Despite a shared catalytic mechanism, substrate specificity within the PPP family is mediated by distinct mechanisms of substrate recognition. For example, the PPP calcineurin (also known as protein phosphatase 2B) has been shown to interact with two consensus sequences, PxIxIT and LxVP, found on nonsubstrate-interacting proteins and target substrates (reviewed in [11,13]). For protein phosphatase-1 (PP1), substrate specificity is conferred by incorporation of PP1-interacting proteins via a conserved docking motif with a general consensus sequence of RVxF (reviewed in [11,12]). At present, consensus sequences in PP2A substrates have not been identified. This review will focus on our emerging understanding of PP2A substrate specificity, which appears to involve additive effects of multiple discrete interactions.
PP2A is a highly conserved serine/threonine phosphatase which, depending on the tissue of origin and cell type, may account for up to 1% of cellular protein and the majority of serine/threonine phosphatase activity [14]. The physiological functions of PP2A have been implicated in all facets of cellular existence (reviewed in [15]). Further, PP2A functions as a critical tumor suppressor whose interruption leads to proliferative diseases [14]. The heterotrimeric holoenzyme is composed of a catalytic subunit (C) a scaffold subunit (A) and one member of four families of regulatory subunits (B) (Figure 1). The diversity of PP2A heterotrimers is achieved through expression of two C subunits, two A subunits and approximately fifteen B subunits in vertebrates. The B 2 Enzyme Research subunits are derived from four diverse gene families (B, B , B , and B ) that have little sequence similarity between families but maintain high sequence similarity within families. The B family (B55, PR55, PPP2R2) of regulatory subunits consists of four genes (α, β, γ, δ), the B -family (B56, PR61, PPP2R5) is comprised of five isoforms (α, β, γ, δ, ε), the B family (PR72, PPP2R3) includes three isoforms (α/PR72/130, β/PR59, γ/PR48), and the B family (PR93/PR110) is made up of three proteins (SG2NA, striatin, and mMOB1). There is some controversy as to whether the B family members, most notably SG2NA, are bona fide PP2A regulatory subunits that always associate with the AC dimer or whether they are merely regulated by association with the PP2A dimer. Given the large number of PP2A subunits, it is thought that each cell expresses a dozen or more distinct holoenzyme complexes which act on a diverse array of substrates. PP2A holoenzyme diversity has been the subject of several excellent papers [14,15,18].
Like PP1, regulatory subunit incorporation is thought to dictate the substrate specificity of the PP2A complex, however, only recently have molecular studies begun to develop insight into the mechanism by which the regulatory subunit acts [15]. The results from recent studies suggests a multitiered mechanism wherein PP2A substrate specificity arises from (1) subcellular localization of PP2A defined by the B subunit, (2) selective holoenzyme assembly by posttranslational modification, (3) interaction with specific endogenous inhibitors, (4) interactions between the B subunit and phosphosubstrates at sites distant from the active site, and (5) B-subunit residues which infiltrate the catalytic cleft of the C subunit. This paper will provide a summary of these studies and how the understanding of the determinants of PP2A substrate specificity has advanced.

Subcellular Localization of the Holoenzyme Complex.
The heterotrimeric holoenzyme is targeted to discrete subcellular locales dictated in part by which B-regulatory subunit is incorporated. The localization imparted by the B-regulatory subunit dictates the spatial sphere of influence of the holoenzyme complex for potential substrates. This mechanism of targeting PP2A activity is highlighted by extensive studies of the B family of regulatory subunits. For instance, the B family regulatory subunits target the holoenzyme to different cellular compartments in the brain [19]. Specifically, Bα and Bβ are primarily cytosolic where as the Bγ-regulatory subunit associates with a detergent-resistant protein fraction consistent with an interaction at the cytoskeleton [19].
Similar diversity has been observed in the B family of regulatory subunits. A C-terminal nuclear export signal common to B α, B β, and B ε which, when these regulatory subunits are incorporated into the PP2A holoenzyme, results in cytoplasmic localization of the heterotrimer [20]. B δ and B γ isoforms lack a similar sequence and are found primarily in the nucleus [21]. In cardiomyocytes, B α interacts with the protein ankyrin-B through its C-terminus which leads to localization at the cardiac M-line [22]. B γ on the other hand has been shown to target the holoenzyme complex Figure 1: Diversity of the B-regulatory subunits and the structure of the PP2A holoenzyme complex. (a) Cladogram of the human PP2A B-regulatory subunits generated using ClustalW [16] and shown with Dendroscope [17]. (b) Structure of the dimeric A scaffold subunit and C catalytic subunit (gray). The shaded portions of the A scaffold subunit indicate the known docking sites for the B regulatory subunits; green (HEAT repeat 1) for the B family of regulatory subunits, red (HEAT repeat 8) for the B family of regulatory subunits, and yellow (HEAT repeat 2-7) for the docking site area common for the B and B families. Cartoon structure of the B (green ribbon), B (red ribbon), B (purple rectangle), and B (tan circle) families of regulatory subunits. Cartoon structures were generated using the Protein Data Bank accession codes 3DW8 and 2NPP.
to subnuclear structures in cardiomyocytes where PP2A/B γ may regulate gene expression [23]. Loss of proper subcellular targeting of PP2A has been implicated in the biogenesis and aggressive phenotype of neoplastic growths. Specifically, a truncated form of B γ (ΔB γ), has been isolated from a melanoma cell line wherein the PP2A/ΔB γ complex is targeted to the trans-golgi network, blunts p53 responsiveness, contributes to genetic instability and increases metastatic motility [24][25][26][27].
Further complexity arises from differing localization imparted by alternative splicing of regulatory subunit genes.

Enzyme Research 3
This was first observed in a member of the B family, Bβ. Two neuron-specific isoforms of Bβ, Bβ1, and Bβ2, are generated by the use of an alternative 5 exon which results in the production of a divergent N-terminal extension on Bβ2 [28,29]. The N-terminal extension of Bβ2 directs the holoenzyme complex to the outer mitochondrial membrane (OMM) by targeting the mitochondrial translocase complex and forming an abortive complex resistant to import into the mitochondrial matrix [28,30]. The OMMdirected PP2A/Bβ2 complex promotes fragmentation of the mitochondria reticulum and increases cell susceptibility to proapoptotic insults through an unknown mechanism [31]. Alternative splicing of B-regulatory subunits directing subcellular localization of the PP2A holoenzyme has also been observed in both the B and B families. Two B ε isoforms differ by the inclusion or exclusion an N-terminal nuclear localization signal leading to isoform-specific nuclear or cytoplasmic localization [32]. The murine-specific B family member, B δ, is also subject to alternative splicing leading to isoform-specific nuclear or cytoplasmic localization through an as yet unidentified mechanism [33].
Localization is also imparted by interactions between the PP2A holoenzyme complex and other effectors. For example, the interaction between PP2A/B and shugoshin during meiosis is crucial for spatial and temporal regulation of sister chromatid disjunction. During meiosis, the cohesin complexes, which link the arms of bivalent chromosomes and the centromeres of sister chromatids, must be released in a stepwise fashion by the protease separase; the cohesin complex is firstly hydrolyzed along the arms of the bivalent chromosomes for completion of anaphase I and secondly at the centromeres of sister chromatids for completion of anaphase II. During anaphase I, the centromeric cohesin complex is protected from separase-dependent proteolysis by shugoshin [34,35]. Shugoshin recruits PP2A/B to the centromere which likely results in dephosphorylation of the cohesin complex leading to protection of the cohesin complex from separase-dependent proteolysis [36,37]. Through cocrystallization, Xu and colleagues, have revealed that dimeric human shugoshin 1 interacts with PP2A/B γ through a coiled-coil region across a broad composite surface of the C and B γ subunits [38]. Other PP2A-effector interactions have recently been reviewed elsewhere [14].

Selective Holoenzyme Assembly and Activation.
Incorporation of specific regulatory subunits is influenced by reversible posttranslational modification of the C subunit. Many groups have shown that the C-terminus of the C subunit is modified through phosphorylation and methylation on Y307 and L309, respectively [39][40][41][42][43][44][45][46]. Phosphorylation of Y307 is catalyzed by src kinase and is likely opposed by PP2A-catalyzed autodephosphorylation of this phosphotyrosine [46]. Phosphorylation of Y307 selectively inhibits recruitment of the B family and some B family members to the dimeric AC complex whereas B recruitment is not effected [44].
Methylation of the C subunit at the C-terminal L309 is catalyzed by the protein phosphatase methyltransferase (PPM1) and is opposed by the phosphatase methylesterase (PME-1) [47][48][49][50][51]. Reversible methylation of PP2A is absolutely critical as knocking-out PME-1 in mice changes the phosphoproteome and results in early perinatal lethality [52]. Further, methylation of the C subunit is a dynamic process which plays a role in cellular response to acute stimuli [53]. The recruitment of the B subunit to the AC dimer has been postulated to require methylation of the C subunit for some of the B-subunit families [39]. However, conflicting results have been reported that may reflect differences in experimental design and will be discussed further. Studies wherein PP2A/B is isolated from intact cells have revealed that methylation of the C subunit at L309 is required for incorporation of the B family of regulatory subunits into the holoenzyme complex [39,40,42,44,54,55]. Conversely, in vitro assembly of the PP2A/B holoenzyme complex does not require methylation of the C subunit for incorporation of the B family of regulatory subunits [56,57]. Similarly controversial, the requirement for C subunit L309 methylation was observed to be study-specific in in vitro PP2A/B timer formation [58,59]. Methylation was dispensable for isolation of the PP2A/B holoenzyme complex from intact cells [44]. The role of methylation of the C subunit in recruitment of B and B families of regulatory subunits is less controversial with methylation of the C subunit being dispensable [44,54]. For more information, the reader is directed to an excellent recent paper [60].
Posttranslational modifications which influence formation of the PP2A holoenzyme complex also occur on the B subunit. PP2A/B negatively regulates the ERK MAP kinase signal transduction pathway [61]. Through formation of a ternary complex of the early response gene product IEX-1, B γ1, and ERK, ERK mediates its own disinhibition by phosphorylation of B γ1 on S327 leading to B γ1 disassociation from the PP2A holoenzyme [62]. Since S327 is conserved among B -subunit family members, it is likely that other B subunits are regulated similarly. Additionally, Bα is likely phosphorylated on S167 to disrupt the Bα subunit from the AC dimer in early mitotic stages. However, PP2A/Bα activity is necessary to resolve the mitotic spindles and conclude mitosis; therefore, autodephosphorylation may occur on Bα to allow efficient PP2A/Bα heterotrimer formation and cell cycle progression [63]. Thus, phosphorylation of the Bregulatory subunits also influences holoenzyme assembly and, therefore, substrate specificity.
Phosphorylation of the B subunit of heterotrimeric PP2A also potentiates the catalytic activity of the holoenzyme complex. In response to activation of D1 dopamine receptors on striatal neurons, cAMP-dependent protein kinase A (PKA) phosphorylates B δ at S566 increasing activity of the PP2A/ B δ holoenzyme towards dopamine-and cAMPregulated neuronal phosphoprotein (DARPP-32) [64,65]. Dephosphorylation of T75 on DARPP-32 by PP2A/B δ disinhibits PKA-mediated phosphorylation of DARPP-32 at T34 which converts DARPP-32 into a potent PP1 inhibitor leading to changes in neuronal signaling. This circuit acts to attenuate phospho-T75 inhibition of T34 phosphorylation of DARPP-32. This circuit has been shown to be differentially regulated by psychomotor stimulants and antipsychotics acting on different striatal neuron subpopulations [66].  figure. (a), Structure of the PP2A/B γ holoenzyme (PDB 2NPP); inset highlights the infiltration of the B γ-subunit residue, E122 (green), into the catalytic core above. Below, sequence alignment of the B family of regulatory subunits with the conserved glutamate residue (green). (b), Structure of the PP2A/Bα holoenzyme (PDB 3DW8); inset shows infiltration of the Bα-subunit residues, K88 (green), into the catalytic core above. Below, sequence alignment of the B family of regulatory subunits with the conserved lysine residue (green).

Regulating PP2A through Protein Inhibitors.
While assembly of the many trimeric PP2A holoenzymes directs cellular localization and substrate specificity, further regulation is afforded through binding of specific protein inhibitors of PP2A. One such inhibitor SET (I 2 /TAF-1) is upregulated during the progression of chronic myelogenous leukemia through BCR/ABL activity and results in decreased PP2A activity. Importantly, restoring PP2A activity prevents disease progression in an animal model of leukemogenesis [67]. An additional PP2A inhibitor is the protein CIP2A (cancerous inhibitor of PP2A), overexpression of which is associated with several human malignancies. CIP2A associates with c-myc to protect its phosphorylated S62 from PP2A-directed activity stabilizing the c-myc protein and allowing it to promote oncogenesis [68].
Recently an interplay between the Drosophila serine/ threonine kinase Greatwall (gwl) and PP2A/Bδ was observed during mitotic entry in two separate studies [69,70].
PP2A/Bδ activity prevents mitotic entry by maintaining Cdc25 in a dephosphorylated and inactive state. Gwl reverses this inhibition through the phosphorylation of S67 of both α-endosulfine (Ensa) and cyclic adenosine monophosphate-(cAMP-) regulated phosphoprotein-19 (Arpp-19). Phosphorylation converts Ensa and Arpp-19 into very specific inhibitors of PP2A/Bδ activity and produces activation of Cdc25 leading to cell cycle progression. Similar cell cycle regulatory activity has been observed with the mammalian ortholog of gwl, MASTL; however, the MASTL-PP2A interaction has yet to be characterized [71].

Substrate Recruitment by Regulatory Subunit Interactions
Distant from the Active Site. Once targeted to specific subcellular locales, the PP2A holoenzyme must recruit and dephosphorylate target substrates. Recent structural studies have begun to suggest the mechanism by which the regulatory subunit of PP2A mediates initial binding to target substrates.

5
The B family of regulatory subunits adopts a seven-bladed β-propeller structure [72]. Other β-propeller proteins have been shown to bind ligands in the central depression on the top surface of the toroid [73]. Crystallization of PP2A/Bα revealed a cluster of acidic residues in this depression that is available to recruit potential substrates containing a basic motif [56]. In this same study, the acidic central depression of Bα was experimentally confirmed to bind the microtubuleassociated protein tau, an established PP2A/Bα substrate. Several conserved aspartate and glutamate residues in Bα engage in weak, electrostatic interactions across a large basic portion of tau and support dephosphorylation of tau at multiple sites through cycles of binding and unbinding [56].
Structurally divergent B family members may recruit substrates in a similar fashion as the B family. The crystal structure of PP2A/B γ shows that the B subunit contains 18 stacked α-helices which adopt 8 huntingtin elongation A subunit Tor-(HEAT-) like repeats [58,59]. A portion of these HEAT-like repeats interact with the A subunit of the holoenzyme to mediate regulatory subunit incorporation into the holoenzyme complex. Like the Bα subunit, an acidic patch is exposed in the B family of regulatory subunits and may mediate protein-protein interactions and substrate recruitment [59].

Substrate Recruitment by Regulatory Subunit Interactions
Near the Active Site. Structural studies of PP2A have revealed a conserved loop in the B family of regulatory subunits which infiltrates the catalytic core of the holoenzyme [58,59] (Figure 2(a)). At the tip of this loop is a conserved glutamate residue, E153 (B β numbering), which contacts through its carbonyl oxygen the catalytic subunit and through its carboxyl group a cocrystallized microcystin molecule in the active site. Mutational analysis revealed that E153 is an absolute requirement for efficient dephosphorylation of tyrosine hydroxylase (TH), a known PP2A/B β substrate, as well as other as yet unidentified cellular substrates of this PP2A holoenzyme [74]. Further, it was determined that E153 of B β interacts with R37 and R38 of TH to mediate dephosphorylation of both S31 and S40 on TH. Positively charged residues in the vicinity of target phospho-serine/threonine residues could represent a consensus sequence for Bsubunit-mediated dephosphorylation. Further, the infiltrating loop is likely a conformationally dynamic structure which is not sterically hindered by surrounding portions of the A, C, or B subunits. Since R37/38 are important for dephosphorylation of both upstream (S40) and downstream (S31) sites, it appears that the orientation of phosphopeptides relative to the catalytic cleft is not constrained by additional interactions. Collectively, the above observations support a model in which the sites of interaction between the substrate and the B regulatory subunit that are distant and near the active site together control substrate specificity. First, the interaction occurring at sites distant from the active site increases the local substrate concentration. Following this initial substrate recruitment, the interactions near the active site mediate site-specific dephosphorylation.
Although divergent in its sequence, an analogous structure from the β-propeller fold of the B family of regulatory subunits extends to the catalytic core of the holoenzyme [56] (Figure 2(b)). This loop places conserved residues of the B-family subunits very near the holoenzyme active site. Unpublished observations generated in our lab suggests that of these loop residues K87 of Bβ2 may play a similar role as E153 of B β in site-specific dephosphorylation of target substrates; however, further characterization of this substrate specificity loop is required.

Conclusion
PP2A is a ubiquitous protein phosphatase responsible for the dephosphorylation of many different intracellular targets. The diverse repertoire of potential substrates for PP2A is imparted by the incorporation of one of fifteen unique B-regulatory subunits. Recent studies have increased our understanding of the mechanisms by which the B subunit imparts specificity to the holoenzyme complex. Through selective incorporation of the B-regulatory subunit, the holoenzyme complex is recruited to discrete subcellular locales which define the sphere of influence for the phosphatase. Secondly, interactions between endogenous inhibitors and specific PP2A heterotrimers further restrict phosphatase activity. As shown for the B family, regulatory subunits mediate low-affinity interactions with substrates to increase the local concentration of substrates. Through a flexible substrate selectivity loop which contacts the catalytic subunit, interactions between the regulatory subunit and phosphosubstrate may mediate multiple nearby dephosphorylation events. With the current structural information available for the PP2A complexes, future high-resolution studies will further define the molecular mechanism of PP2A substrate specificity. As general, inhibitors of PP2A are either clinically irrelevant or toxic, as in the case of the small molecule inhibitor microcystin, novel methods to increase the specificity of PP2A inhibition or activation must be developed. A clearer understanding of the PP2A substrate specificity mechanisms will serve as the foundation for rational drug design of selective inhibitors and activators of specific PP2A holoenzyme complexes.