Serine-Threonine Protein Phosphatases: Lost in Translation

Abstract

Protein synthesis is one of the most complex and energy-consuming processes in eukaryotic cells and therefore is tightly regulated. One of the main mechanisms of translational control is post-translational modifications of the components of translational apparatus. Phosphorylation status of translation factors depends on the balanced action of kinases and phosphatases. While many kinase-dependent events are well defined, phosphatases that counteract phosphorylation are rarely determined. This mini-review focuses on the regulation of activity of translational initiation factors by Serine/Threonine phosphatases.

Overview of initiation of protein synthesis

Protein synthesis (or translation) is a recurring process going on through initiation, elongation, termination and recycling. As one of the most energetically expensive cellular processes, it is tightly regulated, predominantly at the initiation step by phosphorylation/dephosphorylation of the eukaryotic initiation factors (eIFs) [1].

Initiation of translation starts with the binding of the eukaryotic initiation factors (eIF) eIF3, eIF1, eIF1A, eIF2-GTP-Met-tRNAi ternary complex, and probably eIF5, to the 40S ribosomal subunit resulting in a formation of a 43S preinitiation complex. Next, 43S attachment to mRNA is mediated by eIF4F complex, which comprises of a cap-binding protein (eIF4E), an RNA helicase (eIF4A) and a scaffolding subunit (eIF4G). The majority of eIF4A exists as free form while only a small fraction is present as the eIF4F subunit [2]. The higher eIF4A abundance, relative to eIF4F, has led to models invoking recycling of eIF4A through the eIF4F complex during translation initiation [3]. 43S complex then scans the 5′ untranslated region of the mRNA until the initiation AUG codon in an optimum context. During scanning, mRNA secondary structure is unwounded by eIF4A, eIF4B and eIF4F. Then, eIF5B promotes the displacement of the eIFs and the subsequent joining of 40S subunit with the 60S subunit to form translationally competent 80S ribosomes (Fig.1) (described in details [4, 5]). eIF4E activity is strictly regulated by eIF4E Binding Protein (4E-BP), which in its active hypo-phosphorylated form binds and sequesters eIF4E from the eIF4F complex. Dephosphorylated 4E-BP therefore acts as a repressor of cap-dependent translation. Activity of 4E-BP is controlled by the mechanistic target of rapamycin (mTOR) serine threonine kinase [6]. PI3K/AKT/mTOR signaling pathway incorporates signals from growth factors stimulation and hormone receptors activation, cellular responses to nutrient status, hypoxia, and other metabolic alterations [7]. mTOR forms 2 complexes mTORC1 and mTORC2. mTORC1 is comprised of mTOR, regulatory-associated protein of mTOR (Raptor), G-protein β-subunit-like protein/LST8 (GβL), and proline-rich AKT substrate 40 kDa (PRAS40). Two major mTORC1 substrates are 4E-BP and ribosomal protein S6 kinase (S6K). S6K affects protein synthesis by phosphorylating ribosomal protein S6 (rpS6), initiation factor eIF4B, tumor suppressor protein PDCD4 (eIF4A repressor), and translation elongation factor eEF2 kinase. While phosphorylation of the initiation factors and their effectors is well studied, dephosphorylation of the corresponding residues and phosphatases that involved in this process are much less appreciated. Here, we will review initiation factors and their immediate regulators that are known (or suggested) substrates of phosphatases. The focus will be on Serine (Ser) and Threonine (Thr) phosphatases as these residues are more common to be functionally important for the regulation of translation initiation factors.

Figure 1. Simplified cartoon of the formation of a translation initiation complex in eukaryotes.

Translation initiation starts with binding of the eukaryotic initiation factors eIF3, eIF1, eIF1A and eIF2-GTP-Met-tRNAi ternary complex to the 40S ribosomal subunit resulting in formation of a 43S preinitiation complex. 43S complex loading on mRNA is mediated by eIF4F. 43S complex then scans the 5′ untranslated region until the initiation AUG codon. During scanning mRNA cap-proximal region is unwound in an ATP-dependent manner by eIF4F with eIF4B (not pictured). Recognition of the initiation codon and 48S initiation complex formation leads to displacement of eIF1 to allow eIF5-mediated hydrolysis of eIF2-bound GTP. The scheme omits potential ‘closed loop’ structure involving poly(A)-binding protein (PABP) and eIF4G that though to be important for optimal mRNA translation, particular under competitive conditions. eIF1A is marked as 1A due to space limitation.

Serine/Threonine phosphatases

Serine/Threonine phosphatases comprise three major families: phosphoprotein phosphatases (PPPs), metal-dependent protein phosphatases (PPMs), and the aspartate-based phosphatases [8]. For a clarity purpose, we will focus on phosphatase-mediated regulation of protein synthesis through describing functions of individual translational factors rather than individual phosphatases family. For the same reason we will omit mechanistic aspects of Ser/Thrphosphatases functions as this subject is covered extensively in other chapters of this special issue.

Two cellular Serine/Threonine phosphatases that belong to PPP family and account for the majority of phosphatase activity in eukaryotes are Protein Phosphatase 1 (PP1) and Protein Phosphatase 2A (PP2A) [9]. PP1 catalytic subunit forms stable complexes with one out of approximately 200 PP1-interacting proteins that control its biogenesis and functions. Thus, PP1 functions as the catalytic subunit of a large number of multimeric holoenzymes, each with its own subset of substrates and mechanisms of regulation [10]. PP2A holoenzymes consist of a dimer formed by a catalytic C subunit (PP2A-C) and a scaffolding A subunit (PP2A-A) which are targeted to a substrate by a regulatory B subunit (reviewed in [11]). In humans, B-type subunits are encoded by 15 genes, representing four classes (B55/PPP2R2; B56/PPP2R5; B72/PPP2R3 and STRN/PPP2R6) that give rise to at least 23 isoforms, thus increasing a number of potential holoenzymes [12]. PP2A and PP1 activity is responsive to specific extracellular stimuli or stresses and is regulated at multiple levels including post-translation modifications of catalytic and regulatory subunits, association with specific cellular inhibitory proteins and etc. [10, 12, 13]. Both phosphatases can be inhibited by Okadaic acid (OA).

Subfamily of Protein Serine/Threonine Phosphatases that depend on Mn2+/Mg2+ for their catalytic activity (PPMs) are insensitive to OA inhibition.

As different phosphatases are discussed in the context of protein synthesis regulation, it is important to keep in mind that demonstrating that a protein is a true substrate of a phosphatase is challenging. The interactions between phosphatases and their substrates are likely transient and therefore are difficult to detect, particularly by conventional immunoprecipitation of endogenous proteins. While inhibition of Ser/Thr phosphatase by toxins is widely used to demonstrate PP2A involvement, all these inhibitors target the catalytic subunit and cannot distinguish between different holoenzymes. They also have potential to inhibit both PP2A and PP1 phosphatases depending on concentration. For example, OA has IC50 0.02–0.5nM for PP2A and 10–200nM for PP1. Calyculin A has IC50 0.5–1nM and 2–2.8nM respectively [12]. Therefore, depending on the experimental setup some findings described below might be reevaluated (see Table 1 for details). Similar caution has to be taken with an in vitro phosphatase assay, that is usually carried out with a catalytic subunit only. Taken out of the cellular context a catalytic subunit likely lacks phosphatase specificity observed in vivo.

Table 1.

Translation initiation factors as substrates of Ser/Thr phosphatases.

	Protein	Phosphates, reference	Comments
1	eIF2α (Ser51)	GADD34/PP1[16]	In vitro phosphatase assay; Immunoprecipitation
		CreP/PP1[18]	In vitro phosphatase assay; Immunoprecipitation; siRNA
		GADD34/CreP/PP1 [19]	mice with induced mutations in Ppp1r15a and Ppp1r15b that lack functional GADD34 or CReP and compound mice lacking both genes
2	eIF2β	PP1 [20]	In vitro phosphatase assay; Immunoprecipitation;
3	eIF2Bε (Ser535)	PP1 [25]	In vitro phosphatase assay using a peptide containing GSK3b targeted phosphorylation site
		PP2A [26]	Macrophages isolated from trif−/− mice were used for in vitro phosphatase assay; siRNA; 1nM OA
4	eIF4E (Ser209)	PP2A [30]	100 nM OA; PP2A activator FTY720
		PP2A [31]	100 nM OA; siRNA; In vitro phosphatase assay
5	eIF4B (S406)	PP2A [38]	25nM OA and 5 nM microcystin;
6	4E-BP	PP2A [46]	In vitro phosphatase assay; 4E-BP was phosphorylated in vitro using 32P, not know what residues were phosphorylated/dephosphorylated, (α4 inhibits PP2A)
		PP2A [49]	In vitro phosphatase assay (α4 targets PP2A to 4E-BP) WB-total 4E-BP
		PP2A, PP1 [41]	6µM OA; WB-total 4E-BP
		PP2A [48]	α4 inhibits PP2A; WB-total 4E-BP
		PP2A [45]	2.5 mM OA, 10nM calyculin A; WB- Ser65, Thr37/Thr46
		PP2A [42]	In vitro phosphatase assay; WB total 4E-BP
		PP2A [44]	100 nM OA; siRNA; In vitro phosphatase assay; WB-total 4E-BP
		PP2A [51]	SV40 ST over expression, WB-S65
		PP2A [53]	SV40 ST over expression; WB-total 4E-BP
		PP2A/PP1/PPM1[54]	1µM OA; Thr37/Thr46
		PPM1G [55]	In vitro phosphatase assay; WB-Thr-37/46; Ser65; siRNA; knockdown of PPM1B had similar effect on 4E-BP phosphorylation, but was not pursued in this study
7	S6K	PP2A[57]	Immunoprecipitation;
		PP2A [58]	Immunoprecipitation of cross-linked PP2A-C complexes;
		PP2A [59]	Immunoprecipitation; WB S6K Thr389
		PP2A [61]	siRNA; 50nM calyculin A; siRNA; WB S6K Thr389
		PP2A[60] *	Knockout flies; Regulatory subunit identified; Immunoprecipitation; corroborated in human cells; WB S6K Thr389
		PP2A [62]	siRNA; WB S6K Thr389
		PP2A [63]	Over expression of regulatory B55 subunit; SV40 ST over expression; WB S6K Thr421/Ser424
		PP2A [64]	WB S6K Thr389

^*Of particular interest

Regulation of eIF2 by Ser/The phosphatases

eIF2α. eIF2 is a heterotrimeric complex composed of eIF2α, eIF2β and eIF2γ (Fig.2A). GTP binding to eIF2γ is required for Met-tRNAi binding. GTP is hydrolyzed upon the initiation codon recognition and eIF2 is released as inactive eIF2/GDP. For another round of initiation to occur, guanine exchange factor (GEF) eIF2B mediates GDP replacement by GTP. Under various stress conditions, eIF2α subunit of eIF2 is phosphorylated at Ser51, becoming a competitive inhibitor of eIF2B, which results in a global inhibition of translation (reviewed in [14]). The importance of eIF2α Ser51 phosphorylation was confirmed genetically, as mice with Ser51 to Ala substitution die at birth [15].

Figure 2. (A) Regulation of eIF2 activity.

The guanine nucleotide-exchange activity of eIF2B is a prerequisite for the formation of the ternary eIF2/GTP/tRNAi complex. It is not clear which phosphatase (PP1 or PP2A) counteracts inhibitory phosphorylation of eIF2Be. Inhibitory phosphorylation of eIF2a in response to stress stimuli (by HRI (heme-regulated inhibitor), PKR (protein kinase R), PERK (PKR-like endoplasmic reticulum kinase) or GCN2 (general control non-depressible 2)) leads to eIF2B sequestering which is released when Ser51 is dephosphorylated by PP1 phosphatase. See text for more details.

(B) Regulation of 4E-BP activity. 4E-BP is phosphorylated on at least four residues by mTOR kinase (and likely by other kinases; see text for details). T46 and T37 are phosphorylated first following S65 and T70 phosphorylation that results in release of eIF4E from 4E-BP, and eIF4E association with both capped mRNA and eIF4G. 4E-BP dephosphorylation is likely mediated by PP2A phosphatase, though involvement of an additional phosphatase is highly possible.

While global translation is attenuated by this phosphorylation, some mRNAs with upstream Open Reading Frames (uORFs) are still translated efficiently as ribosomes are preferentially recruited to the downstream AUGs of these mRNAs. Growth Arrest and DNA Damage-inducible gene 34 (GADD34 encoded by PPP1R15A gene) is one of these mRNAs. GADD34 forms a complex with PP1 that dephosphorylates eIF2α, thus restoring general translation [16, 17]. Later GADD34 homolog, named Constitutive Repressor of eIF2α Phosphorylation (CReP, PPP1R15B gene) was identified [18]. GADD34 knockout mice had no apparent phenotype, however, a double knockout of CReP/GADD34, resulted in embryonic lethality [19]. Importantly this phenotype was rescued by crossing GADD34/CReP double knockout mice with the homozygous eIF2α Ser51Ala mice [19]. Thus genetic and biochemical evidence conclusively define the mechanism of Ser51 dephosphorylation of eIF2α.

eIF2β. The connection between eIF2 and PP1 deepened further when in vitro experiments demonstrated that eIF2β interacts with PP1. Primary PP1-binding RVxF-motif and another PP1-binding site were identified in eIF2β and it was suggested that eIF2β might be PP1 substrate or function as a substrate-specifier for PP1 [20]. Also, brining PP1 to the vicinity of 43S preinitiation complexes through interaction with eIF2β might be beneficial for dephosphorylation of other translational components. As overexpression of wild-type eIF2β or eIF2β with a mutated RVxF-motif did not significantly affect protein synthesis rate, more experiments are necessary to validate this hypothesis.

eIF2B is a heteropentamer whose ε catalytic subunit promotes GDP/GTP exchange on eIF2 [21]. Several phosphorylation sites were identified on all five subunits, but only sites on eIF2Bε were confirmed to be functionally important. Phosphorylation at Ser712 and Ser713 is required for basal eIF2B GEF activity [22], while phosphorylation of eIF2Bε by the glycogen synthase 3β protein kinase (GSK3β ) on Ser535 and Ser539 negatively regulates eIF2Bε activity [23, 24]. Originally it was suggested that PP1 is responsible for the dephosphorylation of Ser535 and 539 residues as a recombinant PP1 catalytic subunit dephosphorylated and activated eIF2B in vitro (eIF2B peptide containing Ser535 and 539 was used in in vitro assay) [25]. Recently, PP2A has also emerged as eIF2Bε phosphatase. Using TRIF (an adaptor that mediates interactions between Toll-like receptors (TLRs) and intracellular signaling molecules) knockout mice, it was shown that TLR-TRIF signaling dephosphorylates and activates eIF2Bε [26]. In the same work knockdown of PP2A-C in primary macrophages resulted in the increased eIF2Bε phosphorylation on GSKβ-targeted residues, supporting the hypothesis that PP2A directly dephosphorylates eIF2Bε. More experiments are required to conclusively implicate either PP1 or PP2A in the regulation of eIF2B activity.

Regulation of eIF4F activity by Ser/The phosphatases

eIF4E: Cap-binding protein eIF4E is a limiting component of eIF4F and its expression levels are believed to be critical determinants of the rates of eukaryotic translation. Although eIF4E is phosphorylated at Ser209 by MAP kinase interacting kinases (MNK1 and MNK2), this phosphorylation is not essential for normal development as MNK1,2 double knockout mice had no phenotype [27]. However, eIF4E is overexpressed in many cancers and eIF4E phosphorylation appears to enhance their transformation potential [28] underlining functional importance of this phosphorylation under pathological conditions (reviewed in [29]). Current evidence points at PP2A as eIF4E phosphatase. First, in blast crisis cells, Ser209 phosphorylation of eIF4E was sensitive to OA (100nM), while PP2A activator FTY720 strongly decreased Ser209 phosphorylation [30]. In another study, increased eIF4E phosphorylation was induced by the same concentration of OA or by PP2A-C siRNA [31]. Importantly, PP2A inhibition by OA or siRNA was counteracted either by the presence of the MNK inhibitor or by MNK genes deficiency. Furthermore, in vitro assays demonstrated that PP2A can directly dephosphorylate MNK1(Thr179/Thr202) and eIF4E(Ser209) [31]. It will be important to identify a holoenzyme that is responsible for eIF4E dephosphorylation in order to understand regulation of eIF4E phosphorylation in details, particularly in various cancers.

eIF4B is required for the efficient translation of mRNAs with structured 5′ UTRs [32]. Insulin-induced eIF4B phosphorylation correlates with an observed increase in translation [33], while heat shock and serum starvation cause its dephosphorylation. Identified phosphorylation sites include Ser422 (phosphorylated by S6K and p90 ribosomal protein S6 kinase (RSK) [34] or by PKB [35]), Ser406 (depends on MEK and mTOR activity [35]), Ser496 (by Pim1 and Pim2 kinases [36]), and Ser504 (by casein kinase [37]). Phosphorylation of these residues was demonstrated to modulate eIF4B translational activity. While Ser406 phosphorylation is required for optimal translational activity, it was reported that in neurons eIF4B is rapidly dephosphorylated at Ser406 upon neuronal stimulation to promote neuron-specific translation. In this case Ser406 dephosphorylation causes a decrease in the binding of eIF4B to brain cytoplasmic RNAs (non-protein-coding RNAs), freeing eIF4B for initiation [38]. This rapid eIF4B dephosphorylation was attributed to PP2A, as it was inhibited by 25nM OA (See Table 1).

4E-BP acts as a repressor of cap-dependent translation as it sequesters eIF4E from eIF4F complex by competing with eIF4G for its binding (Fig.2B). Among three 4E-BP isoforms (4E-BP1–3) 4E-BP1 is largely expressed in most tissues and the phosphorylation sites are conserved between all isoforms, we therefore will describe 4E-BP1 studies referring to the protein as 4E-BP for simplicity. 4E-BP phosphorylation upon mitogen stimulation occurs at multiple sites and proceeds in a hierarchical order: Thr37 and Thr46 first, then Thr70 and Ser65 last [6, 39] (Fig.2B). Only fully phosphorylated 4E-BP (Ser65 positive) is incompetent of eIF4E binding. Phosphorylation at Ser65 (and sometimes at Thr70) is therefore often used as functional readout of 4E-BP activity, as phosphorylated at Thr37 and Thr46 4E-BP still binds eIF4E. Differentially phosphorylated 4E-BP isoforms can be separated on high-density SDS PAGE according to their phosphorylation marks with the uppermost band representing fully phosphorylated (inactive) 4E-BP. As mentioned before, mTORC1 is considered as principal 4E-BP kinase for all four residues, although other kinases might be involved as well [40].

Overwhelming evidence points at PP2A as 4E-BP activator, though direct interaction between 4E-BP and PP2A has yet to be demonstrated. 4E-BP dephosphorylation in cardiac myocytes following hydrogen peroxide treatment was inhibited by 6μM OA, leading to suggestion that PP2A was responsible for 4E-BP activation, although this concentration would inhibit both PP2A and PP1 [41] In fibroblasts, β1 integrin treatment activated Src kinase, resulting in inhibition of PP2A and subsequently 4E-BP (through phosphorylation) [42]. It was also shown that PP2A-C knockdown resulted in downregulation of 4E-BP at both mRNA and protein levels that might (at least at the protein level) be attributed to ubiquitation-dependent degradation of phosphorylated 4E-BP [43]. Similar outcome was reported in human umbilical vein endothelial cells where TNFα-and cycloheximide (CHX)-induced apoptosis led to rapid 4E-BP degradation. It was suggested that CHX treatment inhibits PP2A and increases the activity of p38-MAPK (which was considered as 4E-BP kinase in this work) therefore resulting in increased 4E-BP phosphorylation with subsequent degradation [44]. PKC signaling in rat intestinal epithelial cells was shown to increase PP2A activity and to promote 4E-BP dephosphorylation at Thr46 and Ser65. PKC effect on 4E-BP was blocked by treatment with OA[45].

FLAG-tagged PP2A-C dephosphorylated GST-4E-BP in vitro, and this dephosphorylation was prevented by an overexpression of α4, a binding partner of PP2A [46]. α4(also known as immunoglobulin-a-binding protein 1 (IGBP1)) forms a stable complex with PP2A-C, independently of the A or B-type subunits. The regulatory effect of α4 on PP2A activity is ambiguous, with some reports showing inhibition and some activation of the phosphatase activity [47]. It has been suggested that this discrepancy might be due to different substrates used in these studies, and that α4 affects the PP2A-C substrate specificity rather than its activity [47]. In erythroblasts α4 inhibited PP2A activity leading to increased 4E-BP phosphorylation [48], while in COS-1 cells α4 increased PP2A phosphatase activity and decreased 4E-BP phosphorylation [49], implicating that α4 functions might also be cell-type specific

The downregulation of Ser/Thr phosphatase activity by viral proteins is believed to be a principal mechanism through which viruses affect multiple intracellular signaling pathways. Some PP2A holoenzymes can be inhibited by SV40 small T antigen (ST) [50]. We used SV40-ST overexpression to demonstrate PP2A’s role in 4E-BP activation in FGF-treated chondrocytes. Chondrocytes uniquely respond to FGF with growth inhibition and suppression of protein synthesis that is mediated by 4E-BP. Overexpression of SV40-ST prevented 4E-BP dephosphorylation suggesting that 4E-BP is PP2A substrate[51] Another viral protein -Merkel cell polyomavirus (MCV) ST -was found to preserve 4E-BP hyperphosphorylation as well [52]. However, in this case the situation is more complicated as abolishing MCV ST-PP2A interaction did not affect 4E-BP phosphorylation. MCV is related to SV40 and shares similar gene structures, but these viruses differ in their carcinogenic mechanisms. It is also not clear why during SV40 lytic infection in monkey cells, 4E-BP phosphorylation was decreased rather than prevented [53]. Surprisingly, it was further shown that 4E-BP phosphorylation status was not only controlled by SV40-ST, but required the SV40-ST PP2A binding domain as well [53]. Yu et al. hypothesized that SV40-ST mediated PP2A inhibition allows activation of another phosphatase that targets 4E-BP [53]. Additional experiments are needed to precisely understand the effect of viral proteins on 4E-BP dephosphorylation in context of their interaction with PP2A.

Gardner et al. reported that 4E-BP is rapidly dephosphorylated in the presence of glycolytic inhibitor in ex vivo rat retinas stimulated with insulin [54]. This dephosphorylation was mostly prevented by either 1µ M OA (PP1/PP2A inhibition) or by cadmium (PPM1 inhibition) suggesting that more than one phosphatase might be involved in dephosphorylation of all four residues, or maybe, 4E-BP phosphatase has to be activated (directly or not) by another phosphatase.

Echoing this data, a member of PPM1 family -PP2Cgamma (or PPM1G-protein phosphatase, Mg2+/Mn2+ dependent, 1G) was also implicated in 4E-BP dephosphorylation [55]. 4E-BP co-immunoprecipitated with PPM1G and purified PPM1G dephosphorylated 4E-BP in vitro and in a human colon carcinoma (HCT 116) cell line. Interestingly, PPM1G exhibited higher affinity for dephosphorylated rather than hyperphosphorylated 4E-BP raising a question about how this enzyme would cycle to the next substrate, and how the same 4E-BP molecule can be phosphorylated again. More evidence is needed to demonstrate that 4E-BP is a direct PP2A/or PP1/or PPM1G substrate including identification of a regulatory subunits responsible for 4E-BP targeting.

Regulation of S6K by PP2A phosphatase

S6K is another well-known mTORC1 substrate, whose activity is controlled by multiple phosphorylation events. S6K activation begins with the phosphorylation of Ser411, Ser418, Ser421 and Ser424, allowing mTOR to phosphorylate Thr389[56] S6K dephosphorylation was the subject of investigation for many years with all experimental evidence pointed at PP2A as S6K phosphatase. The first report that S6K physically interacts with PP2A was published almost two decades ago[57] S6K was similarly detected in the complexes isolated with cross-linked (compared to non-cross-linked) PP2A [58]. Also, S6K was shown to be associated with all three PP2A holoenzyme subunits: regulatory (B55α), scaffolding (Aβ), and catalytic (Cα) when TGF-β induced G1 arrest in epithelial cells [59].

Treatment of Drosophila Schneider 2 cells with Calyculin A resulted in a 7-fold increase in the basal level of S6K phosphorylation at Thr389 (human numbering). Similar effect was observed with PP2A-C knockdown [60]. Genetic evidence provided further insight into the mechanism of S6K dephosphorylation by PP2A, as PP2A-B’ knockout flies had elevated S6K phosphorylation, were lean and had shorter life span exhibiting phenotypes typical of defective insulin signaling [61]. It was demonstrated that PP2A-B’ interacts with S6K. The human homolog of PP2A-B’, PPP2R5C (B56γ ), was also shown to counteract dS6K phosphorylation, indicating a conserved mechanism in mammals. Knockdown of PPP2R5C resulted in increased S6K phosphorylation in HeLa cells. Following work confirmed that PP2A holoenzyme containing B56γ subunit targets S6K [62]. Phosphatase activity toward S6K was shown to be important in cerebellar granule neuron precursors (CGNPs) for their proliferation.

Another PP2A holoenzyme containing PPP2R2C (B55γ) subunit was implicated in S6K dephosphorylation in human glioma cells where its overexpression inhibited cell growth [63]. Direct interaction between B55γ and S6K was not detected, however B55γ overexpression induced binding of PP2A-C and S6K. It is not clear why a different regulatory subunit was involved in S6K dephosphorylation, however it is worth mentioning that phosphorylation of Thr421/Thr424 but not Thr389 (as in previous work) was monitored in this work. Salt-inducible kinase 2 (SIK2) was shown to interact with B55γ and to contribute to the survival of glioma cells under glucose depletion by inhibiting S6K phosphorylation at Thr389 [64]. Further experiments are necessary to address the sensitivity of other S6K phosphorylation sites to modulation of B55γ and B56γ expression.

Concluding remarks

To fully comprehend a phosphatase role in protein synthesis (or other biological processes), first, it is important to determine if a given phosphorylation site is functionally important and can be experimentally tested. While phosphorylation-mediated mechanisms of regulation are well established for some proteins (eIF2(Ser51) and 4E-BP(Ser65)), the functional importance of many other phosphorylation sites has just started to emerge, particular for multi-subunit factors, such as eIF3. For instance, phosphorylation of eIF3f at Ser46 and Thr119 was shown to enhance its association with the core eIF3 subunits leading to the inhibition of translation during apoptosis [65]. One of eIF4G phosphorylation sites (Ser1186) was demonstrated to be important for its interaction with the eIF4E kinase, MNK1 [66]. Obviously, until phosphorylation sites are functionally defined, it will be challenging to address the importance of their dephosphorylation. Some recently published data might also be considered in the future. For example, mass spectrometry analysis of proteins associated with activated PP2A in regulatory T-cells identified eIF2Bε, eIF3a, eIF3I, eIF3m, eIF4B, eIF4G1 and Raptor among others proteins, suggesting that some of them might be PP2A substrates [67]. PolyA binding protein (PABP), was shown to interact with α4 in a yeast two-hybrid system and in human cell lines, suggesting that it might be a target of PP2A [68].

Understanding detailed mechanism of phosphatase-mediated regulation requires identification of a holoenzyme responsible for dephosphorylation. At this point eIF2α dephosphorylation might be considered as an example of a well-defined (biochemically and genetically) regulatory mechanism. For the majority of the proteins specific holoenzymes and the mechanisms of their activation/regulation still need to be investigated. Unfortunately, phosphatases are difficult to study as they often act pleiotropically. Blocking their activity usually leads to broad molecular effects and often lethality. Additionally, interaction with a phosphatase might be transient and therefore requires more sophisticated biochemical approaches. Presence of several functional phosphorylation sites brings further ambiguity. For example, while substantial amount of data points at PP2A as the 4E-BP phosphatase, it was also demonstrated that Thr37/Thr46 phosphorylation is not responsive to PP2A activity [31]. One possible explanation might be that when multiple sites are involved, more than one phosphatase are responsible for their dephosphorylation. At this point the role of PP1 (or another phosphatase) might be explored further, as OA concentrations used in many reports would inhibit both PP2A and PP1 (Table.1). More experiments -particularly with mutations of single residues -are needed to address the selectivity of PP2A toward different sites.

Another essential consideration in phosphatases functions is the activity of a kinase that phosphorylates a residue of interest. In some cases, it was demonstrated that dephosphorylation results solely from phosphatase activation rather than from kinase inhibition as the basal kinase activity was not changed. Thus, inhibitory FGF signaling in chondrocytes stimulated PP2A activity without any effect on mTORC1 activity [51]. Similarly, in insulin stimulated rat retinas treated with a glycolytic inhibitor, 4E-BP was rapidly dephosphorylated with no reduction in mTORC1 activity detected [54]. The cases like these can serve as experimental models to digest the mechanism of PP2A activation.

Dysregulation of translation initiation is linked to aberrant proliferation, alterations in immune response and believed to be a hallmark of some cancers [1]. Understanding how initiation factors are regulated by phosphatases is an important direction of future experimental work as phosphatase holoenzymes variety and specificity provide wide range of opportunities for the development of drugs that selectively target the affected pathways.

Highlights.

• Translational factors activity often depends on dephosphorylation.

• Only eIF2α dephosphorylation is well-defined (biochemically and genetically).

• 4E-BP dephosphorylation/activation is likely mediated by PP2A.