Essay: Reversible protein phosphorylation

Reversible protein phosphorylation is extremely important in regulating many intra- and inter-cellular processes and is orchestrated by protein kinases and protein phosphatases, which attach and remove a phosphate group from a substrate protein, respectively [1]. This post-translational modification can alter the activity, conformation and subcellular localization of target proteins in various signaling pathways, thereby affecting processes like cell proliferation, differentiation, apoptosis, DNA replication, growth and cell division [2]. In a normal cell, the function of one third of the cellular proteins is controlled by phosphorylation [3]. The majority of phosphorylations occurs on Ser/Thr residues with 86.4% on Ser and 11.8% on Thr residues, followed by 1.8% on Tyrosine (Tyr) residues [4]. Paradoxically, 385 Ser/Thr kinases exist in the cell which are counteracted only by 21 Ser/Thr phosphatases, leading to the misconception that phosphatases are less specific enzymes. [3]. Specificity is however achieved through the formation of large protein phosphatase families made up by many multisubunit holoenzymes and assembled from just a limited number of catalytic subunits. In addition, transient phosphorylations and interactions with regulatory and inhibitory proteins regulate activity and substrate specificity. So, phosphatases are clearly subjected to strict regulation and are equally important as kinases in maintaining proper protein phosphorylation balance [5].
PP2A: a family of specific serine/threonine phosphatases
Protein Phosphatase type 2A or PP2A represents such a large family of phosphatases and constitutes the majority of Ser/Thr phosphatase activity in the cell together with protein phosphatase 1 (PP1) [6]. PP2A is essential, ubiquitously expressed and highly conserved (from yeast to mammals) [7]. Importantly, its dysfunction has been linked to diverse pathologies, like Alzheimer’s disease, diabetes, intellectual disability, cancer, etc. [8-11].
In mammalian cells, PP2A exists as either a dimer (PP2A-AC) or a trimer (PP2A-ACB) [5] (Figure 1). The 36 kDa catalytic subunit (PP2Ac, C) can associate with a 65 kDa structural or scaffold subunit (PR65/A, A) to form the dimeric complex PP2A-AC [12, 13]. For both A and C subunits two isoforms exist in the cell, α and β. Although they possess high sequence similarities, these isoforms are functionally non-redundant [14-17]. Furthermore, Aα accounts for 0.1% of the total cell protein [3]. The core dimeric structure can exist in this form and is estimated to constitute one third of the total PP2A pool [5]. However, the dimer also associates with one of multiple regulatory B-type subunits with a molecular weight ranging from 48 to 130 kDa, to form a trimer. The B subunit is considered to be the major regulator to control the activity of PP2A, defining substrate specificity and subcellular localization [18]. The B-type subunits are grouped into four different families, namely B55/PR55/B, B56/PR61/B’, PR72/B” and PR93/PR110/B’. These families are diverse and do not show any sequence similarities [5]. Every family contains at least 4 different isoforms with varying tissue distribution, subcellular localization and developmental expression. Together, the B regulatory subunits comprise a minimum of 26 different transcripts and splice variants encoded by 15 different genes [3]. As a consequence, the combination of the different isoforms of A, B and C can give rise to at least 96 different PP2A holoenzymes, all presumably with different functions [23] in diverse cellular processes such as transcription, translation, DNA replication, apoptosis and cell division [5].
The structural A subunit bridges PP2Ac and the regulatory B-type subunit. It consists of a unique structure which contains 15 tandem repeats of 39 amino acids, known as HEAT (Huntington/elongation/A-subunit/TOR) motifs [19]. This 15-HEAT-repeat-molecule adopts a flexible L-shape structure [20]. The catalytic subunit binds with HEAT repeats 11-15, whereas regulatory subunits bind to HEAT repeats 1-8 [21, 22]. The different crystal structures of A in the monomeric (A [20]), dimeric (A-C [13, 23] and A-PR70 [24]) and trimeric (Bα [25], B’γ [19, 26, 27], PR70 [28]) forms reveal that the A subunit undergoes drastic conformational changes upon C and B binding, from a hook shape to a horseshoe-like conformation, probably contributing to the stability of the holoenzymes.
The B55/PR55/B family exists of five isoforms (, , 2,  and ) encoded by four different genes (PPP2R2A, PPP2R2B, PPP2R2C and PPP2R2D). B, and B are widely expressed, while B, B2 and B are highly enriched in the brain. Structurally, the B subunits contain five degenerate WD-40 repeats, i.e. minimally conserved sequences of approximately 40 amino acids that typically end with tryptophan-aspartate, involved in protein-protein interactions [5]. B55/PR55/B subunit-containing complexes regulate apoptosis [29], mitosis [30], transforming growth factor-β (TGF-β) signaling [31], extracellular signal-regulated kinase (ERK) signaling [32], angiogenesis [33], DNA damage signaling [34] and tau dephosphorylation [35].
The B56/PR61/B’ subunits represent the largest family (isoforms α to ε) and is encoded by five different genes (PPP2R5A, PPP2R5B, PPP2R5C, PPP2R5D and PPP2R5E), of which some are alternatively spliced or translated. B’ and B’γ isoforms are abundant in heart, while B’β and B’δ have a high expression in brain. These family members possess a highly conserved central region, which is 80% identical, and very divergent N- and C-terminal domains. It is believed that the central core mediates interactions with A and C, while the divergent ends regulate the substrate specificity and subcellular targeting of the specific B’ holoenzyme [5]. Based on amino acid similarity and evolutionary conservation, the B56/PR61/B’ family can be divided in two subgroups, B’αβε and B’γδ [36]. These subunits possess a unique feature among the different families, namely they are phosphoproteins (except B’γ1) [37]. Holoenzymes harboring a B’ subunit have other substrates in mitosis/meiosis [30, 38], apoptosis [29], development [39] and dopaminergic signaling in the brain [40]. Importantly, they are also key regulators of several oncogenic targets like Akt, Wnt and c-myc [5, 41-44], and are therefore considered as the main tumor suppressive PP2A B subunit family.
Human PR72/B” subunits are encoded by three different genes (PPP2R3A, PPP2R3B and PPP2R3C), of which the first two give rise to two alternative splice variants. All are ubiquitously expressed, except B”α2 and B”2, which are only found in heart and skeletal muscle. The PR72/B” subunits bind and are regulated by Ca2+ ions [45]. These PP2A holoenzymes have functions in non-canonical Wnt signaling [32], epidermal growth factor (EGF) signaling [46], DNA synthesis [5, 47], pocket protein dephosphorylation [48] and neuronal signaling [40].
The fourth B-type family exists of the three highly homologous proteins PR110/B’’’ (striatin), PR93/B’’’ (S/G2 Nuclear Antigen, SG2NA) and zinedin, encoded by three genes (STRN, STRN3 and STRN4). All members are highly expressed in the central and peripheral nervous system, but are also present in many other tissues. They all contain four protein-interaction domains: a Ca2+-calmodulin-binding domain, a caveolin-binding domain, a coiled-coil domain and a WD-repeat domain [49]. Striatin family members serve as scaffold proteins to assemble multiple diverse and large signaling complexes that are involved in numerous functions like e.g. organ size control and development through the Hippo signaling [50], apoptosis [51], cell cycle control [52], Golgi assembly [53], estrogen receptor α signaling [54], cell migration [55], and neural development [56]. All striatin family complexes encompass PP2Ac and A subunits. Together with germinal center kinase III kinases and other components, they form STRIPAK or striatin-interacting phosphatase and kinase complexes [49, 57]. It is believed that PP2A negatively regulates the kinases within this complex. In addition, separate STRIPAK-like complexes have been discovered that are not yet known to contain both a kinase and PP2A. Interestingly, STRIPAK complexes have been linked to diseases like diabetes, autism, cancer, heart disease and cerebral cavernous malformation [49].
So far, studies addressing the functional roles of different PP2A complexes mainly relied on overexpression and/or depletion of specific B-type subunits in cellular models in vitro. Only a limited number of in vivo studies have been performed in mammals, showing a role for B56/PR61/B’δ in the central nervous system [58] and for B56/PR61/B’γ in cardiomyocyte maturation and survival [59]. This at least demonstrates that despite significant homology within a given B subunit family, individual B-type subunits serve non-redundant functions in a complex mammalian system.
PP2A biogenesis: a complex and partially understood process
During the formation or biogenesis of PP2A, the C subunit is synthesized as an inactive enzyme in order to prevent promiscuous/unrestricted activity of free PP2Ac immediately after translation [60]. This is necessary because uncontrolled phosphatase activity would pose a risk to the cell as long as it is not restrained by the interaction with other subunits. As a result, an activation mechanism exists that is coupled to the assembly of the complete holoenzyme [61]. This biogenesis of active PP2A trimers is a complex step-by-step process that is tightly controlled by several essential PP2A regulators and modulators, including Phosphatase Two A Phosphatase Activator (PTPA), leucine carboxyl methyltransferase 1 (LCMT1), protein phosphatase methylesterase 1 (PME1), Type-2A interacting protein (TIPRL1) and α4.
The currently proposed model for PP2A biogenesis is shown in Figure 2. When PP2Ac is translated in an inactive form, it is believed that α4 binds to it to prevent its degradation [62]. Inactive PP2Ac can then associate with the A subunit, which competes with α4 to form an inactive PP2A-AC that is stabilized by complex formation with PME1. PME1 has a dual role, stabilizing inactive PP2Ac by binding to it and preventing premature PP2Ac methylation via its methylesterase activity [63]. In a next step, PTPA is responsible for the activation of the complex. PTPA induces a conformational change in PP2Ac, potentially via its peptidyl-prolyl cis/trans isomerase activity on Proline 190 [64], resulting in the dissociation of PME1 [14]. After activation by PTPA, the PP2Ac C-terminal tail is able to be methylated on the free carboxyterminus of the C-terminal Leucine (Leu) 309 by LCMT1 [65]. This modification, on its turn, promotes association with most/certain B-type subunits to form an active holoenzyme [18] (See further). Recently, a role for the A subunit was suggested in facilitating Leu309 methylation by LCMT1 through stabilization of a proper protein fold and an active conformation of PP2Ac, via limiting the space of PP2Ac-tail movement for enhanced entry into the LCMT1 active site, and by conferring weak electrostatic interactions with LCMT1 through the N-terminal HEAT repeats [66]. How TIPRL1 is involved in this process is speculative: it could play a role in the assembly process, or rather later on, in neutralizing PP2Ac upon stress-induced holoenzyme disassembly and, subsequently, making it available for reassembly (as part of the adaptive responses to the stress) or promoting its degradation.
The role of PP2Ac post-translational modifications in holoenzyme assembly
All PP2A-like phosphatases, including PP4 and PP6, possess a conserved C- terminal tail in their C subunit. Different post-translational modifications (phosphorylation and methylation) of this 304-TPDYFL-309 tail can influence the binding affinity with certain B-type subunits, and therefore, might either play a positive or a negative role in the assembly of a certain holoenzyme. Moreover, fluctuations in the methylation and/or phosphorylation levels of the C subunit tail might provide a theoretical basis for dynamic subunit exchange in PP2A complexes, resulting in altered PP2A targeting and substrate specificity [18]. Indeed, variations in C-tail methylation patterns have been demonstrated in both physiological and pathological circumstances [67].
As introduced in the biogenesis model, a vital post-translational modification of the flexible 304-TPDYFL-309 motif of the C subunit is the methylation of the carboxy-terminal Leu309. In vivo, 70 to 90% of PP2Ac is estimated to be methylated [18]. It is catalyzed by the S-adenosylmethionine (SAM)-dependent enzyme LCMT1 and reversed by the methylesterase PME1. These PP2Ac modifying enzymes bear indispensable functions since knockout of both LCMT1 [68] and PME1 [69] in mice is lethal. While PME1 is mainly localized in the nucleus, LCMT1 is predominantly present in the cytoplasm and also in the Golgi apparatus and early endosomes. So, spatial control of methylation/demethylation exists [18]. PME1 fulfills a dual role: it reverses the Leu309 methylation on the C-terminal tail of PP2Ac [70] and binds/stabilizes the newly synthesized inactive PP2A in the early steps during biogenesis [63, 70, 71]. The PP2A-LCMT1 structure reveals that the C-terminal tail occupies the deep active site pocket in the Lid domain of LCMT1. In addition, contacts with the active site of PP2A are necessary to facilitate binding of the C-tail to LCMT1. Like that, conformational changes of the LCMT1 active site are induced to form the C-tail binding pocket and the binding affinity between LCMT1 and PP2Ac is increased. So, a tight link between PP2A methylation and the active conformation of PP2A exist, highlighted by the facts that (1) mutations in the PP2A active site abolish its methylation, (2) two highly potent phosphatase (active site) inhibitors okadaic acid (OA) and microcystin LR prevent PP2A methylation and (3) the C subunit conformation in the PP2A-LCMT1 complex is very similar to that in the PP2A-ACB trimer [72]. According to mutational studies, PP2Ac subunit methylation is essential for binding of the B55/PR55/B subunits but is not absolutely required for interaction with the PR65/A, B56/PR61/B’, PR72/B” and PR93/PR110/B’ subunits (Figure 3). Moreover, B’δ, PR72, PR70 and the striatins [73] can still associate with a deletion mutant of the entire C-tail, suggesting that these B-type subunits do not need stabilizing contacts with the catalytic subunit within the holoenzyme. For B’αβε subunits, Leu309 methylation is not essential but facilitates the incorporation in the holoenzyme. For B’γ, the PP2Ac C-terminal determinants facilitating its incorporation in the holoenzyme have not been reported. In these cases, a contact with PP2Ac is required [74]. Recently, the crystal structure of an A-PR72 PP2A dimere was published, confirming that at PR72 is able to associate with the A subunit without the need for C subunit binding [24].
The C-terminal tail motif also undergoes phosphorylation of Tyr307 and Thr304. Either Tyr or Thr phosphorylation leads to inactivation of PP2A activity [61]. Tyr307 can be phosphorylated by different Tyr kinases, such as pp60v-src, pp56lck, and the epidermal growth factor and insulin receptors. This Tyr307 phosphorylation is increased by OA-mediated PP2A active site inhibition, implying that there is a reactivation mechanism for PP2A via auto-dephosphorylation [75]. This also implies that PP2A could function as a phosphotyrosine phosphatase. The formation of B55/PR55/B and B56/PR61⁔B’-containing PP2A complexes is inhibited by Tyr307 phosphorylation, possibly because addition of a bulky phosphate on Tyr307 prevents access of the tail into the C-tail binding pocket of LCMT1, therefore preventing methylation [76]. On the other hand, only the formation of B55/PR55⁔B-containing PP2A complexes is inhibited by Thr304 phosphorylation, which occurs during mitosis [REFERENCE] and does not affect Leu309 methylation (Figure 3).
PP2A alterations in disease
Alterations in PP2A function and/or activity have been reported in many human diseases, including Alzheimer’s disease, diabetes, intellectual disability, cancer, etc. [8-11]. For the puspose of this thesis, I will here discuss in more detail the pathological role of PP2A alterations in human cancer and intellectual disability.
1. Cancer
Cancer is a generic term for a group of diseases, characterized by uncontrolled growth, invasion and potentially metastasis of cells. The transformation from normal into malignant cell, called tumorigenesis, is a multistep process in which genetic or epigenetic alterations activate oncogenes and/or inactivate tumor suppressors. When cells become malignant, they acquire essential common characteristics or hallmark capabilities, reviewed by Hanahan and Weinberg [77], which include: sustained chronic proliferation, growth repression evasion, apoptosis evasion, replicative immortality, angiogenesis induction, invasion and metastasis, avoidance of immune destruction, deregulation of cellular energetics, tumor promoting inflammation, and genomic instability and mutations. These hallmarks are common, but the order in which they are acquired can differ between cancer types and subtypes. Moreover, the role of a specific genetic aberration may vary substantially between tumors, in the sense that it can partially or fully contribute to the acquisition of a certain hallmark, or that this event may aid in the simultaneous acquisition of several distinct hallmarks. Nevertheless, independently of how and when these hallmarks are acquired, at the biological endpoint of cancer, the same hallmarks will be shared by all types of human tumors.
PP2A, a well-established tumor suppressor
Since the large family of PP2A holoenzymes is involved in nearly every cellular process, it comes of no surprise that malfunctioning of PP2A can contribute to the acquisition of the hallmarks of cancer. PP2A is indeed considered as an important tumor suppressor in the cell. Many observations firmly establish this:
(1) The pharmacological PP2A inhibitor Okadaic Acid (OA) is a potent tumor promoter, underscoring a negative role for PP2A in tumorigenesis. OA is a polyether fatty acid produced by marine dinoflagellates and causes diarrhoetic shellfish poisoning. In 1998, it was found to bind to PP2Ac, via a hydrophobic cage not found in other Ser/Thr phosphatases, and to potently inhibit its phosphatase activity [2]. Moreover, when injected into mice, OA causes cancerous laesions on the skin and tumors in the liver and stomach [78].
(2) The expression of viral tumor antigens by small DNA tumor viruses plays an important role in cell transformation, since they can complex with cellular proteins involved in signal transduction and growth control, affecting their normal functioning [42]. Moreover, these viral tumor antigens can inactivate tumor suppressors, including p53, retinoblastoma (Rb) and PP2A. For instance, small t antigen of simian virus 40 (SV40 st) or polyoma virus are able to inactivate PP2A by binding to the A subunit and substituting for the B subunit in the holoenzyme [79].(3) In a cellular model of transformation, the so-called HEK-TER system reported by Hahn et al. 1999, inhibition of PP2A by SV40 st is essential to cause the full transformation of immortalized human cells by hTERT, SV40 large T (LT) and oncogenic Ras. [80]. Transformation was evaluated in vitro by anchorage-independent growth and in vivo by tumor formation in immune-deficient mice. With an increasing amount of cell divisions, the telomeres become increasingly shortened, leading to genomic instability and cell death. Expression of the catalytic subunit of telomerase hTERT ensures a stable length of the telomeres, thereby preserving chromosome integrity and cell viability. SV40 LT inhibits the tumor suppressors p53 and Rb and expression of oncogenic Ras stimulates cell growth (MAPK pathway), survival (Akt pathway) and migration (Ral pathway). At least two of these Ras effectors need to be activated in order for human cells to become tumorigenic. Also, sole expression of Ras in normal cells causes oncogenic stress-induced senescence, therefore p53 and Rb need to be inactivated to cause cellular transformation [81]. Of note, members of the Ras signaling pathways are the most mutated oncogenes in human tumors [80]. Together, co-expression of hTERT, Ras and SV40 LT results in the immortalization of human cells. To achieve complete transformation PP2A needs to be inhibited by SV40 st [82]. Indeed, PP2A-binding defective mutants of SV40 st are not able to cause transformation within the HEK-TER system [37], plus reduction of Cα levels (accompanied by the degradation of multiple PP2A subunits) nearly completely mimics the tumorigenic phenotype caused by SV40 st expression, both confirming that PP2A is the major cellular target of SV40 st [82, 83]. By manipulating the levels of A in the HEK-TER system, it was revealed that a 50% reduction of Aα could completely transform cells, while complete deletion causes apoptotic cell death [84]. Interestingly, the main holoenzyme affected by 50% loss of Aα is AB’γC, as opposed to other PP2A complexes which were less affected by Aα suppression [84]. This observation indicates that dynamic subunit exchange could occur in vivo due to competition of the B-type subunits for binding to A.
The HEK-TER system was also exploited to identify specific PP2A subunits involved in cell transformation by scoring to what extent loss-of-function of a certain subunit could mimic the tumorigenic phenotype caused by SV40 st. Three regulatory B subunits, B’α, B’γ and PR72/PR130, and the biogenesis regulator PTPA were identified as being important for protection against cellular transformation [85]. Several key oncogenic proteins were disturbed in these tumorigenic cells: B’α and PR72/PR130 suppression caused increased expression of c-myc, while suppression of B’γ resulted in Akt activation and increased β-catenin-dependent transcription (canonical Wnt pathway). All three pathways were affected by PTPA suppression, consistent with its role in PP2A biogenesis and holoenzyme formation. Indeed, a decreased level of C-tail methylation was detected together with reduced interaction between A and C [85]. Importantly, while PTPA suppression was able to very closely mimic cell transformation caused by SV40 st expression, suppression of the individual B-type subunits could only partially transform cells. In addition, overexpression of B’γ could only partially rescue the SV40 st tumorigenic phenotype, suggesting that SV40 st likely affects multiple tumor suppressive PP2A complexes [84]. Several additional studies confirm the importance of PP2A’s tumor suppressive function on the Wnt, Akt and c-myc pathway. For instance, constitutive PI3K signaling [86], a combination of constitutive active Akt and Rac1 [87], and oncogenic c-myc (T58A mutant, unable to be dephosphorylated on Ser62 by PP2A) [88] replaced SV40 st in anchorage-independent growth assays. However, it is likely that additional oncogenic pathways controlled by PP2A are involved. For instance, loss of the cyclin-dependent kinase (CDK) inhibitor p27/Kip1 also substitutes for SV40 st [89]. Finally, either LCMT1 knockdown or PME1 overexpression cause cell transformation (evaluated via anchorage-independent growth) only in combination with B’γ suppression, resulting in upregulation of Akt and p70/p85 S6 kinase pathways [90]. Also, Cancerous Inhibitor of PP2A (CIP2A), a cellular inhibitor of PP2A activity towards oncogenic c-myc, is able to completely mimic the tumorigenic phenotype of SV40 st [91].
(4) The most convincing evidence sustaining a tumor suppressive role of PP2A is the in vivo data from two PP2A knockout (KO) mice, obtained in our laboratory. Both Ppp2r5d (encoding PR61⁔B’δ) and Ppp2r4 (encoding PTPA) KO mice show indeed spontaneous cancer phenotypes (unpublished work).
Genetic alterations of PP2A in tumor cells
Not surprisingly, a growing amount of PP2A-inactivating mechanisms in cancer is being discovered, including mutations in PP2A subunit encoding genes, aberrant expression of PP2A subunits and regulatory proteins involved in PP2A biogenesis (PTPA, PME1 and TIPRL1) and overexpression of cellular PP2A inhibitors.
Mutations in and aberrant expression of PP2A subunit encoding genes
In different types of cancer, the A and B-type subunits have been found to be mutated or abnormally expressed.
(1) Reduced expression of Aα was found in 43% of gliomas [92]. In addition, several mutations of the Aα subunit, identified at low frequency (E64D in lung cancer, E64G in breast carcinoma and R418W in melanoma) cause defects in holoenzyme formation through haploinsufficiency [93-95]. In accordance, a knock-in mouse expressing the E64D mutation showed a 50 to 60% increase in the incidence of lung cancer induced by benzopyrene [16]. In the last few years, however, it became apparent that the frequency of Aα mutants is much higher in certain subtypes of endometrial cancer, ranging from 20% to 40% [96-103]. This suggests an important role for Aα (and PP2A) in the pathogenesis of uterine cancer. Many of these mutations (P179L/R, R182W, R183G/Q/W, R249H, S256F/Y, W257C/G, R258C/H/Y) are recurrent and cluster together in HEAT repeats 5 and 7. (2) Aβ mutations have been discovered in colorectal, lung, breast and gastric cancers and some of these can result in increased activity of RalA, due to complete loss of RalA dephosphorylation, and tumor formation [17]. The Aβ gene locus is prone to loss-of-heterozygosity and subsequent loss of this locus is reported in cervix, ovary, stomach, bladder and breast carcinomas and in melanoma [104]. (3) Mutations in B-type subunits occur at low frequency in a wide variety of cancers. For example, in melanoma (B’γ), c-kit-positive Acute myeloid leukemia (Bα and multiple B’), lung cancer (Bα and B’γ) and breast cancer (Bα), these mutations result in defects in paxillin dephosphorylation, double strand DNA repair, p53 functioning and Akt dephosphorylation [34, 105-110]. To further elaborate on this, a search in the cBioportal for Cancer Genomics database (http://cbioportal.org) for aberrations in the different PP2A subunits revealed some additional interesting observations. Firstly, Cβ and Bα are obviously deleted in a large number of cancers, for instance in up to 13.1% of metastatic prostate adenocarcinoma and up to 14.7% of prostate adenocarcinomas, respectively. Another obvious observation is the amplification of B’α and B’δ in up to 12.4% in liver and breast cancer and up to 10% in esophageal carcinoma, respectively. Bβ is mostly amplified (most frequently in kidney renal clear cell carcinoma (16.9%)) and mutated (highest frequency in colorectal adenocarcinoma (5.6%)). PR72/PR130 has the highest frequency of amplification (13.5%) and mutation (9.4%) in lung squamous cell carcinoma. Cα was found overexpressed in kidney renal clear cell carcinoma (16.1%). B’β,γ,ε and PR110/B’’’ (striatin), PR93/B’’’ (SG2NA) and zinedin show mixed aberrations, not exceeding frequencies above 8%.
Altertions in regulators of PP2A biogenesis
α4 is found highly expressed in hepatocellular carcinoma (HCC), lung cancer, and breast cancer [111]. PME1 is also overexpressed in several cancers like glioblastoma, endometrial cancer, breast cancer, melanoma, head and neck cancer, etc. with a frequency of 5-10% (cBioportal). Thus, increased PP2A demethylation may be tumorigenic. Heterozygous loss of PPP2R4, the gene encoding PTPA, was discovered at strikingly high frequencies in uterine, lung, liver, renal breast and pancreatic cancer and in melanoma (cBioportal). This loss may be associated with the presence of less active PP2A in the cell, responsible for the tumor formation. Besides heterozygous loss of PPP2R4, other alterations, like mutations, were found at a lower frequency in various cancers (cBioportal).
Overexpression of cellular PP2A inhibitors
By far the most frequent PP2A aberration in cancer is the overexpression of cellular inhibitors of PP2A, the most studied being SET (Suvar 3-9/Enhancer of zeste/Trithorax) and CIP2A. This occurs in a variety of cancer types, often with high frequency and correlating with cancer aggressiveness and poor outcome. In my review article, ‘Cellular inhibitors of Protein Phosphatase PP2A in cancer’ (Chapter 3, [112]), the mechanisms of action, regulation and function of these PP2A inhibitors are highlighted, and it is discussed how this knowledge might be exploited for therapeutic use.
Together, the above findings highlight the multiple mechanisms by which PP2A can be inactivated in cancer, underscoring the fact that PP2A is an important tumor suppressor. Specifically for hematological malignancies, these mechanisms causing aberrant PP2A functioning are reviewed in ‘The basic biology of PP2A in hematologic cells and malignancies’ (Chapter 4 [113]). Although reluctantly, the combined data have increased the interest in exploiting ‘PP2A reactivation’ as a therapeutic strategy in human cancers, as discussed later on [114]. Nevertheless, the clinical utility of using the “PP2A status” as a stratification marker in human tumors cells still largely remains underestimated and underexploited. This is particularly relevant in cancer types where PP2A aberrations occur with high frequency, such as certain subtypes of endometrial cancer.
Epidemiology of endometrial cancer
Endometrial cancer is the most common gynecological malignant disease; others are ovarian, cervical, vulval and vaginal cancer. Worldwide, endometrial cancer is the sixth most common malignant disorder with approximately 290,000 new cases annually and it is the fourth most common cancer in Belgian women after breast, colorectal and lung cancer ([115], Stichting Tegen Kanker). In 2010, 1450 (5%) new cases and 214 deaths were reported in Belgium. Newly diagnosed patients have a 5-year relative survival rate of 80%, ranging from 94.5% for early disease to 19.5% for advanced disease [116]. Over 90% of cases occurs in women older than 50 years of age, with 63 as a median age [117]. The most common symptoms patients suffer from include abnormal uterine bleeding and vaginal discharge, presenting in 90% of the cases. Patients with advanced disease may also experience abdominal or pelvic pain, abdominal distension, early satiety, or change in bowel or bladder function, symptoms often associated with advanced ovarian cancer [118]. Of all the gynecologic cancers, only cervical cancer can be detected by a standard screening test, called the Papanicolaou or Pap test, and this early in the disease when treatment is most effective. For endometrial cancer, the diagnosis is based on the histology of endometrial tissue obtained during a biopsy with the Pipelle aspiration catheter. To determine the stage of the disease, clinical examination, transvaginal ultrasound, computed tomography (CT) scan, magnetic resonance imaging (MRI), and integrated positron emission tomography and computed tomography (PET/CT) scan can be performed to evaluate the presence of metastases. Preoperative staging aids in determining the proper treatment options for patients and is based on the International Federation of Gynecology and Obstetrics (FIGO) system. 72% of patients are diagnosed with stage I disease, meaning that the tumor is confined to the body of the uterus. When the tumor has invaded the cervical stroma, but does not extend beyond the uterus, patients suffer from stage II endometrial cancer, as is the case in 12% of patients. Stage III defines local and/or regional spread (with for example vaginal involvement and lymph node metastasis), present in 13% of patients. 3% suffers from stage IV disease, meaning that the tumor invaded the bladder and/or bowel mucosa, with or without the presence of distant metastases, mostly in the lung [115, 119]. Tumors may additionally be grouped according to their level of differentiation in G1 (well differentiated), G2 (moderately differentiated) and G3 (poorly or undifferentiated) [115].
Traditionally, endometrial carcinoma is classified into two main groups, as defined by Bokhman [120]: type I and type II (Table 1).
Type I endometrial carcinoma (87-90% of cases [121]) is composed of low grade endometrioid carcinoma, which are estrogen dependent and develop from endometrial hyperplasia [98]. Most women are diagnosed after menopause and the stage at diagnosis is commonly FIGO stage I and II. Established risk factors are early onset of menstruation, nulliparity, late menopause and unopposed estrogen exposure due to estrogen replacement therapy to control menopausal symptoms, estrogen-producing tumors, tamoxifen treatment, polycystic ovarian syndrome, etc. [118]. In addition, insulin resistance, associated with obesity (causing high estrogen production in adipose tissue) and diabetes, can co-occur. Excessive insulin levels will stimulate insulin growth factor pathway, leading to tumorigenesis [122]. Insulin also inhibits the production of globulin, a sex hormone binding protein, resulting in increased levels of the sex hormones androgen and estrogen. It also promotes synthesis of androgens in the ovaries. An epidemic of obesity in high-income countries is likely the reason for a difference in incidence of endometrial cancer in high-income vs. low-income countries (5.5% vs. 4.2%). However, the specific mortality is higher in the latter, due to a higher prevalence of the aggressive, high-grade cancer type. Other risk factors are age and a family history of Hereditary Non-Polyposis Colon Cancer (HNPCC), also known as Lynch Syndrome, which is an inherited cancer syndrome causing mainly colon cancer, but also other cancer types, most commonly endometrial cancer. Inactivating mutations in DNA mismatch repair (MMR) genes cause this disease. Patients with HNPCC have a 20 to 70% risk for developing endometrial cancer depending on the specific MMR mutation [123]. Interestingly, a reduced risk is associated with the use of combination oral contraceptive pills or progesterone secreting intra-uterine devices. Smoking also reduces the risk for endometrial cancer, especially in postmenopausal women [118]. Type I endometrial cancer is typically diagnosed at early stage before extra-uterine metastasis and, therefore, has an overall favorable prognosis. Surgery is curative in many cases. This type is characterized by high frequency genomic alterations affecting PIK3CA, PIK3R1, PTEN, KRAS, FGFR2, ARID1A (BAF250a), and CTNNB1 (β-catenin), as well as microsatellite instability (MSI) due to epigenetic silencing of MMR genes [124]. So, mainly the PI3K/AKT/mTOR (cell growth and survival) and Wnt/β-catenin signalling (gene transcription and development) pathways drive tumorigenesis in these tumors.
Type II endometrial carcinoma includes mainly the uterine serous adenocarcinoma (2.9-10.5% of cases [121]) and the clear-cell adenocarcinoma (2.2-3.2% of cases [121]). Serous adenocarcinoma is estrogen independent and arises in atrophic endometrium and endometrial polyps from preinvasive lesions called serous endometrial intraepithelial carcinoma [98]. Although this type constitutes only a small percentage of all endometrial cancers, it accounts for a disproportionately high number of deaths due to its high aggressiveness and high tendency to metastasize. These tumors are typified as FIGO stage III and IV. Moreover, type II endometrial carcinomas are highly resistant to conventional chemotherapy and recurrence is almost inevitable. These tumors tend to be aneuploid and are characterized by frequent genomic alterations affecting TP53 (p53), PPP2R1A (PP2A Aα), HER-2/ERBB2, PIK3CA, and PTEN; plus dysregulation of E-cadherin, p16 and BAF250a. So, again the PI3K/AKT/mTOR and Wnt/β-catenin signaling seem to be important, together with inactivation of tumor supressors p53 and PP2A.
Important to emphasize is that an overlap exists between type I and type II endometrial tumors and that heterogeneity is present within each of these types. To clarify, low-grade endometrioid and high-grade serous adenocarcinoma integrate well in the Bokhman model, but 10 to 19% of endometrioid adenocarcinomas are high-grade and have features that are intermediate between type I and II or are more resembling those of type II endometrial cancer [125]. Likewise, not all serous adenocarcinomas show prototypical type II characteristics. Moreover, none of the mutations in any of the disease genes identified so far can be exclusively assigned to a specific type. For instance, mutations in TP53 are present in 90% of the serous type and 10% of low-grade and 30% of high-grade endometrioid adenocarcinomas [125]. Consequently, the Cancer Genome Atlas Research Network recently proposed a genomic classification, which can aid in reclassification of the different subtypes of endometrioid and serous carcinomas and the identification of potential targets for targeted treatment of different subgroups of the disease (Table 2). From these data it is clear that mutations in PPP2R1A are associated with poor outcome and aggressive behavior of both the endometrioid and serous types of endometrial adencarcinoma.
The golden standard treatment of endometrial carcinoma is surgery, including total hysterectomy (surgical removal of uterus and cervix) and bilateral salpingo-oophorectomy (surgical removal of fallopian tubes and ovaries). The type and FIGO stage determines the precise surgical procedure and the need for additional lymph node dissection and omentectomy. Moreover, the need for adjuvant therapies is evaluated based on information obtained from surgery [115, 117, 119, 126]. These adjuvant therapies include radiotherapy, chemotherapy, hormone and targeted therapy. Regarding radiotherapy and chemotherapy, depending on the FIGO stage and grade of the tumors, pelvic radiotherapy, vaginal brachytherapy and/or chemotherapy are performed, possibly in combination [117]. The combination of the chemotherapeutics carboplatin and paclitaxel is the standard adjuvant therapy for stage III endometrial cancer and is also standard first-line therapy for metastatic or recurrent disease [127]. Hormonal therapy is recommended for endometrioid histologies. The main hormone treatment for endometrial cancer uses progesterone-like drugs called progestins which slow down the growth of cancer cells. Progestins represented the first-line treatment for endometrial cancer in the 1970s and are still used in appropriate settings, particularly for low-grade tumors that recur long after primary therapy. Tamoxifen, which is an anti-estrogen drug often used to treat breast cancer, may also be used in the treatment of advanced or recurrent disease. Furthermore, gonadotropin-releasing hormone agonists are a way to lower estrogen levels in women who still have functioning ovaries. Aromatase inhibitors can block estrogen production in adipose tissue. These drugs are used in the treatment of breast cancer, but may also be utilized to treat endometrial cancer. The use of antiangiogenic agents, like bevacizumab, is currently under investigation. Recent understandings of the signaling pathways underlying tumorigenesis in endometrial cancer led to the evaluation of targeted therapies against these affected signaling molecules. For instance, the PI3K/Akt/mTOR pathway can be targeted at different levels via multiple drugs, for example with Akt, mTORC1, isoform-specific PI3K, Pan-PI3K, dual PI3K/mTOR and mTORC1/2 inhibitors [128]. Currently, however, the benefit of blocking the PI3K/Akt/mTOR pathway is disappointing and limited by cross-talk with the Ras/Raf/MEK pathway. Therefore, dual blockage might be more efficient in the future [129].
PP2A aberrations in endometrial cancer
From the previous chapter it is clear that until the molecular pathogenesis of endometrial (serous) carcinoma is better understood, therapeutic interventions to improve the clinical outcomes of these patients remain empirical. The high frequency of mutations in PPP2R1A, hinting towards an important role of aberrant PP2A functioning in the pathogenesis, may be one of the opportunities that could be further exploited in this respect.
Historically, a publication from 2010, reporting PPP2R1A mutations in 7% of ovarian clear-cell carcinoma [130], prompted the evaluation of the PPP2R1A mutational status in other gynecological cancers via whole-genome sequencing and targeted (exome 5 and 6) approaches [98-101, 103, 131-133]. These studies revealed that somatic mutations in PPP2R1A occur at high frequency (18.4-43.2%) in the serous type of endometrial cancer, while the percentage is low (2.5-6.9%) in the endometrioid subtype. Since PPP2R1A mutations are specifically correlated with high grade endometrial carcinoma, both serous and endometrioid of origin (genomic classification of The Cancer Genome Atlas Research Network), these findings link the presence of these mutations to aggressiveness of the tumors and poor patient outcome [125]. The heterozygous mutations identified include P179L/R, R182W and R183G/Q/W, located in HEAT-repeat 5, and R249H, S256F/Y, W257C/G and R258H, located in HEAT-repeat 7; they all have been recurrently found in ovarian and in endometrial cancer (Table 3). Because structural and functional studies of PP2A indicate that the mutated HEAT-repeats 5 and 7 directly contact the B subunits, it was hypothesized that the mutations might affect PP2A holoenzyme assembly [21, 22, 94, 95, 134]. The mutations described in endometrioid endometrial carcinoma and in ovarian carcinoma seem to more frequently involve residues R182 and R183, while in the serous endometrial tumors residues S256 and W257 and P179 seem to be much more affected. However, the significance of this uneven distribution is unclear [124]. Another interesting finding is the fact that PPP2R1A mutations occur in serous endometrial carcinomas but not (or barely) in ovarian serous carcinomas. This suggests that, although both tumor types share several clinical and pathological features, their pathogenesis is likely different. Therefore, the PPP2R1A status could also be used to distinguish these two serous cancer types [98, 99].
Lastly, no correlation was found between the occurrence of mutations in PPP2R1A and other genes frequently mutated in serous endometrial cancer, like TP53, KRAS and PIK3CA [101, 131].
Worthwhile to mention is that PPP2R1A is also found mutated in endometrial carcinosarcoma. Two studies report frequencies of 26.8% (15/56) [137] and 21.4% (9/42) [100]. The recurrent mutations included P179R, S256F, R183W and S219L. Carcinosarcoma is a rare cancer of the endometrium, with a prevalence of less than 2%. It is a highly aggressive cancer, likely to present in an advanced stage [121]. Furthermore, PPP2R1A mutations were detected in 20% (4/20) of patients diagnosed with undifferentiated carcinoma. This relatively uncommon neoplasm is classified as a type of endometrioid carcinoma, although undifferentiated carcinoma displays a more aggressive behavior [97].
Besides PPP2R1A mutations, two mutations in PPP2R5C, the gene encoding B’γ, were identified in endometrioid endometrial carcinoma, namely T182M and L168V [100].
PME1 overexpression was recently also detected in the endometrioid adenocarcinoma cell lines, RL95-2, Ishikawa, and ECC-1; and in 83% (24/29) of type I endometrial carcinomas [138]. This overexpression results in decreased PP2A activity, increased cell proliferation and metastases by causing ERK and Akt hyperphosphorylation. Increased PME1 levels also led to increased anchorage-independent growth and tumor formation in vivo. In addition, an enhanced interaction with PP4c, a tumor promoter, was found. However, whether PME1 overexpression activates PP4 to promote cell proliferation or inhibits PP4 to counterbalance PP2A inhibition is unclear [138].