Interaction analysis of the heterotrimer formed by the phosphatase 2A catalytic subunit, α4 and the mammalian ortholog of yeast Tip41 (TIPRL)

Authors


N. I. T. Zanchin, Centro de Biologia Molecular Estrutural, Laboratório Nacional de Luz Síncrotron, R. Giuseppe Máximo Scolfaro, 10.000, Campinas – SP, PO Box 6192, CEP 13084-971, Brazil
Fax: +55 19 3512 1004
Tel: +55 19 3512 1113
E-mail: zanchin@lnls.br

Abstract

Type 2A serine/threonine phosphatases are part of the PPP subfamily that is formed by PP2A, PP4 and PP6, and participate in a variety of cellular processes including transcription, translation, regulation of the cell cycle, signal transduction and apoptosis. PP2A is found predominantly as a heterotrimer formed by the catalytic subunit (C) and by a regulatory (B, B′ or B′′) and a scaffolding (A) subunit. Yeast Tap42p and Tip41p are regulators of type 2A phosphatases, playing antagonistic roles in the target of rapamycin signaling pathway. α4 and target of rapamycin signaling pathway regulator-like (TIPRL) are the respective mammalian orthologs of Tap42p and Tip41p. α4 has been characterized as an essential protein implicated in cell signaling, differentiation and survival; by contrast, the role of mammalian TIPRL is still poorly understood. In this study, a yeast two-hybrid screen revealed that TIPRL interacts with the C-terminal region of the catalytic subunits of PP2A, PP4 and PP6. Τhe TIPRL-interacting region on the catalytic subunit was mapped to residues 210–309 and does not overlap with the α4-binding region, as shown by yeast two-hybrid and pull-down assays using recombinant proteins. TIPRL and α4 can bind PP2Ac simultaneously, forming a stable ternary complex. Reverse two-hybrid assays revealed that single amino acid substitutions on TIPRL including D71L, I136T, M196V and D198N can block its interaction with PP2Ac. TIPRL inhibits PP2Ac activity in vitro and forms a rapamycin-insensitive complex with PP2Ac and α4 in human cells. These results suggest the existence of a novel PP2A heterotrimer (α4:PP2Ac:TIPRL) in mammalian cells.

Abbreviations
3-AT

3-amino-triazol

GST

glutathione S-transferase

RBCC

ring finger B-box coiled coil

TIPRL

TOR signaling pathway regulator-like

TOR

target of rapamycin.

Type 2A phosphatases are part of the PPP subfamily that is formed by PP2A, PP4 and PP6, the mammalian orthologs of yeast Pph21/22, Pph3 and Sit4, respectively. These are serine/threonine phosphatases with a wide range of substrates acting in a variety of cellular processes such as transcription, translation, regulation of the cell cycle, signal transduction and apoptosis [1–4]. PP2A has been described as a holoenzyme formed by a catalytic (C), a regulatory (B, B′ or B′′) and a scaffolding (PR65/A) subunit [1–4]. Although dimers formed by AC subunits have been described in vivo, the prevalent form of the PP2A holoenzyme is the trimeric A:B:C complex. The number of B-type subunits is still growing with new members continuously being discovered. The subunit composition of the holoenzyme determines its subcellular localization, activation state and substrate specificity [1–4]. PP4 forms either a heterotrimer with the subunits PP4R2 and PP4R3 or a heterodimer with PP4R1 [5], and specific subunits of PP6 (PP6R1, PP6R2 and PP6R3) have also been characterized recently [6].

In addition to the regulatory and scaffolding subunits described above, mammalian type 2A phosphatases share the α4 protein as a common regulator, which binds directly to the catalytic subunits and displaces other regulatory subunits [7–10]. α4, the mammalian ortholog of yeast Tap42, was initially identified in association with the B-cell receptor Igα[11] and has been implicated in the regulation of B- and T-cell differentiation [12,13], vertebrate embryonic development and cell death [14]. α4 was shown to interact directly with the catalytic subunits of PP2A, PP4 and PP6 [10] and with the ring finger B-box coiled coil (RBCC) proteins MID1 and MID2 [15,16], and has also been found to participate in kinase/phosphatase signaling modules with S6K [17] and CaCMKII [18]. These α4-containing complexes exemplify mechanisms of PP2A regulation which are independent of the canonical A and B regulatory subunits.

Type 2A phosphatases are key players in the yeast target of rapamycin (TOR) signaling pathway [3]. Although Tap42 was characterized as a regulator of the TOR pathway in yeast cells [19], the role of α4 in the mTOR-dependent control of cell growth is still unclear. The yeast Tip41 protein was identified in a yeast two-hybrid screen as a binding partner for Tap42 and genetic analyses suggested that it functions as a negative regulator of the rapamycin-sensitive signaling pathway by competing with Sit4 for Tap42 [20]. The fission yeast homolog of Tip41 has been characterized as a regulator of the activity of type 2A phosphatases, possibly through its interaction with Tap42 [21]. Therefore, characterization of TOR signaling pathway regulator-like (TIPRL; TIP41), the mammalian ortholog of Tip41, may provide clues to better understand the regulation of type 2A phosphatases and mTOR signaling.

In this study, starting from yeast two-hybrid analyses, we identified the interaction of TIPRL with the C-terminal region of the catalytic subunits of type 2A phosphatases. TIPRL forms a heterotrimeric complex with PP2Ac and α4 and does not compete with α4 for PP2Ac binding, which contrasts with the model described previously for their respective yeast orthologs [20]. Reverse two-hybrid assays revealed that single amino acid substitutions on TIPRL including D71L, I136T, M196V and D198N can block its interaction with PP2Ac. TIPRL inhibits PP2A activity in vitro and the PP2Ac/TIPRL complex is not affected by rapamycin treatment of human cells. Our results suggest that TIPRL, α4 and PP2Ac constitute a novel heterotrimeric phosphatase holoenzyme.

Results

TIPRL interacts with the C-terminal region of the catalytic subunits of type 2A phosphatases

A yeast two-hybrid screen using TIPRL as bait revealed its interaction with the catalytic subunits of type 2A phosphatases. A human leukocyte cDNA library fused to the GAL4 activation domain of pACT2 was screened using the yeast two-hybrid system with TIPRL fused to lexA as bait. pACT2 was rescued from 88 positive clones and the cDNAs were identified by DNA sequencing. Ten cDNAs from the 88 positive clones encoded catalytic subunits of the type 2A phosphatases PP2Acα (one cDNA), PP2Acβ (three cDNAs), the C-terminal region of PP2Acα/β (one cDNA), PP4c (three cDNAs) and PP6c (two cDNAs). Initial mapping of the region of PP2Ac involved in TIPRL binding was obtained from the cDNAs that showed positive interaction with TIPRL. The extension of these cDNAs is shown in Fig. 1A. Complete cDNAs were isolated only for PP2Acα and PP2Acβ. An additional PP2Acβ cDNA was truncated at residue 14. A fourth type of PP2Ac cDNA, encoding residues from position 210 to the C-terminus, may correspond to both PP2Acα and PP2Acβ because they show identical amino acid sequence in this region. Two different cDNAs encoding PP4c were isolated, including from residues 175 and 195 to the C-terminus. The cDNAs encoding PP6c comprise from residues 106 and 171 to the C-terminus, respectively.

Figure 1.

 TIPRL interaction with catalytic subunits of type 2A phosphatases in the yeast two-hybrid system. (A) Schematic representation of the cDNAs encoding catalytic subunits of type 2A phosphatases isolated in the yeast two-hybrid screen using the TIPRL as bait. PP2Ac is represented by a black bar for comparison. Numbers on the left of the gray bars indicate the first amino acid in the respective activation domain-phosphatase catalytic subunit fusion. The PP2Ac isoforms α and β share identical amino acid sequences in the C-terminal region comprising residues 210–309. (B) Two-hybrid assay for expression of the HIS3 reporter gene. Strain L40 carrying the yeast two-hybrid vectors encoding the indicated DNA-binding domain (DB) and activation domain (AD) fusions were plated on synthetic minimal medium lacking tryptophan and leucine (left, SD-WL) and, on minimal medium supplemented with 10 mm 3-AT lacking tryptophan, leucine and histidine (right, SD-WLH +10 mm 3-AT). The phosphatase cDNAs fused to the activation domain were: PP2Acα and PP2Acβ: full length, PP4c: residues 175–307, PP6c: residues 106–305 and PP2AcCT: residues 210–309. As negative controls, the activation domain-phosphatase cDNA fusions were assayed in combination with pBTM-NIP7, encoding a DNA-binding domain fusion with an unrelated protein. Plasmids pBTM-NIP7 (DB-NIP7) and pACT-NOP8 (AD-NOP8) were used as a positive control.

The interaction between TIPRL and the catalytic subunit of type 2A phosphatases was verified by retransforming the prey plasmids into the L40 strain containing plasmids pTL1-TIPRL encoding the lexA–TIPRL fusion protein (Fig. 1B). This assay was performed with the complete PP2Acα and PP2Acβ cDNAs, with the longest PP4c and PP6c cDNAs, encompassing residues 175–307 and 106–305, respectively, and the shortest cDNA, corresponding to the C-terminal residues 210–309 of PP2Acα/β (named PP2AcCT). As negative controls, the cDNA clones in pACT2 were tested for self-activation using an unrelated bait (Nip7p). The interacting proteins Nip7p and Nop8p were used as a positive two-hybrid control [22]. This assay confirmed the activation of HIS3 and lacZ (not shown) expression in the clones containing lexA–TIPRL and the catalytic subunit of the phosphatases fused to the GAL4 activation domain (Fig. 1B), indicating specific interactions between TIPRL and PP2A catalytic subunits.

The cDNAs of the phosphatase catalytic subunits tested in the yeast two-hybrid system were subcloned into the plasmid pGEX-5x2 in frame with glutathione S-transferase (GST) and the resulting fusion proteins were used to test their interaction with His–TIPRL using recombinant proteins expressed in Escherichia coli. In this experiment, His–TIPRL was pulled down by all GST–phosphatase fusion proteins tested, but not by GST alone (Fig. 2A). Residues 210–309 corresponding to the C-terminal region of PP2Acα and PP2Acβ were sufficient for this interaction (Fig. 2A). The interaction between recombinant PP2Acα and endogenous TIPRL from HEK293 was tested in a GST pull-down assay using glutathione–Sepharose-immobilized GST–PP2Acα or GST and a HEK293 cell extract. TIPRL was able to bind to GST–PP2Acα, but not to GST alone, which further confirms the specificity of this interaction (Fig. 2B).

Figure 2.

 Analysis of TIPRL interaction with catalytic subunits of type 2A phosphatases. (A) GST pull-down assay using recombinant proteins. GST fusions of the indicated phosphatase catalytic subunits were coexpressed with His–TIPRL in E. coli. GST fusion proteins were isolated from extracts by binding to glutathione–Sepharose beads. Bound proteins were resolved by SDS/PAGE and detected by immunoblotting with the indicated primary antibodies or by Coomassie Brilliant Blue staining. The phosphatase cDNAs fused to GST were: PP2Acα and PP2Acβ: full length, PP4c: residues 175–307, PP6c: residues 106–305 and PP2AcCT: residues 210–309. His-tagged TIPRL copurified with each one of the GST–phosphatase fusions but not with GST alone. (B) GST or GST–PP2Ac immobilized on glutathione–Sepharose beads were incubated with HEK293 cell extracts and bound proteins were eluted by boiling in SDS/PAGE sample buffer. GST and TIPRL were detected by immunoblot analysis. TIPRL was detected in association with GST–PP2Ac but not with GST alone. (C) Analysis of TIPRL subcellular distribution. HEK293 cells were treated with 50 nm of the PP2Ac inhibitor okadaic acid (OA) or with vehicle (dimethylsulfoxide) for 3.5 h in serum-free medium and the nuclear (N) and cytoplasmic (C) fractions were separated and probed with specific antibodies. 7.5 µg of total protein extract were loaded on each lane. Both TIPRL and PP2Ac are found predominantly in the cytoplasm and their subcellular distribution was not affected by okadaic acid.

Analysis of TIPRL protein expression by immunoblot analysis identified similar levels in the immortalized cell lines HeLa, HEK293 and K562 (not shown). Cell fractionation experiments showed that the subcellular distribution of TIPRL in HEK293 cells was predominantly cytoplasmic, coinciding with that of PP2Ac (Fig. 2C), which further supports their functional relation. Inhibition of type 2A phosphatase activity by okadaic acid treatment did not alter the subcellular distribution of either TIPRL or PP2Ac (Fig. 2C).

Identification of TIPRL residues important for interaction with PP2Acα

Analysis of the TIPRL amino acid sequence did not reveal structural domains that could support a strategy for construction of deletion mutants to map the regions responsible for PP2Acα binding. Therefore, a reverse two-hybrid approach was employed to find interaction-deficient mutants of TIPRL that may provide information on the sites of interaction or contact regions between TIPRL and PP2Acα. A PCR-based random mutagenesis strategy [23] was used to generate a library of mutant TIPRL cDNAs which was transformed into strain L40 carrying pACT2–PP2Acα, along with the linearized pTL1 vector in which the region of the TIPRL cDNA comprising nucleotides 127–319 was removed. Recombination between a PCR product and the remaining residues of the TIPRL cDNA would reconstitute TIPRL coding sequence. As a first step, the screen involved identification of interaction-deficient mutants as determined by loss of the His3+ phenotype and loss of activation of the lacZ reporter gene. Subsequently, clones showing loss of interaction were submitted to a round of immunoblot analysis to exclude those that did not express the full-length lexA–TIPRL fusion protein. Using these criteria, 6 clones of 65 transformants tested were selected for DNA sequencing analysis in order to identify the mutations in the TIPRL cDNA. Each clone showed single amino acid substitutions including D71L, Y79H, I136T, M196V, D198N and Y214C. These clones were retransformed into the L40 strain carrying plasmids expressing activation domain fusions to full-length PP2Acα, PP2Acβ and PP4c and tested for the activation of the reporter gene HIS3 by growth on selective medium lacking histidine and supplemented with 10 mm 3-amino-triazol (3-AT). This assay confirmed loss of interaction for the mutants D71L, I136T, M196V and D198N, whereas mutants Y79H and Y214C still showed some activation of the reporter gene (Fig. 3A). Similar results were obtained for the three different catalytic subunits tested, which was expected, because they should share an equivalent interaction mechanism. Mutant Y79H behaved differently in this respect, because it appears to have a reduced affinity for PP2Acα, but not for PP2Acβ or PP4c. Two independently isolated clones contained mutations at very close positions (M196V/D198N), strongly supporting the hypothesis that these residues are located on TIPRL regions responsible for interaction with PP2Acα. In addition, a multisequence alignment showed that residues D71, I136 and D198 corresponded to conserved positions on the TIPRL sequence (Fig. 3B).

Figure 3.

 Yeast two-hybrid analysis of TIPRL interaction-deficient mutants. (A) L40 derivative strains containing pACT2-PP2Acα and the indicated TIPRL mutant cDNAs fused to the lexA DNA-binding domain of pTL1 were plated on synthetic minimal medium lacking tryptophan and leucine (upper panel, SD-WL) and, on minimal medium supplemented with 10 mm 3-AT lacking tryptophan, leucine and histidine (lower, SD-WLH +10 mm 3-AT). TIPRL mutants D71L, I136T, T138S, M196V and D198N have lost or show reduced interaction with the catalytic subunits of PP2Acα, PP2Acβ and PP4c. Amino acid substitutions Y79H and Y214C have less pronounced effects on these interactions. (B) Amino acid sequence alignment of TIPRL orthologs. Arrowheads indicate the amino acids that are substituted in TIPRL variants that have lost interaction with PP2Ac in the yeast two-hybrid system. * and : indicate conserved residues and conserved amino acid substitutions, respectively. Hsa, Homo sapiens; Xla, Xenopus laevis, Dre, Danio rerio; Dme, Drosophila melanogaster; Ath, Arabidopsis thaliana; Sce, Saccharomyces cerevisiae.

Ternary complex formation by TIPRL, PP2Ac and α4

Because the yeast ortholog of TIPRL has been described as a Tap42 interacting protein [20], it was surprising that no cDNA encoding α4 was isolated in the yeast two-hybrid screen using TIPRL as bait. Furthermore, a direct assay using lexA–TIPRL and GAL4 activation domain-α4 in the yeast two-hybrid system did not indicate an interaction between these two proteins (data not shown). However, the identification of type 2A phosphatase catalytic subunits as binding partners for TIPRL suggested that TIPRL and α4 might be physically and functionally connected through the type 2A phosphatase catalytic subunits. GST pull-down assays were performed using E. coli extracts containing His–α4, which were incubated with GST–PP2Acα, GST–TIPRL or GST alone immobilized on glutathione–Sepharose beads and extracts of a coexpression assay containing His–α4 and His–PP2Acα, which were incubated with GST–TIPRL immobilized on glutathione–Sepharose beads. Under these conditions, the association between His–α4 and GST–TIPRL takes place only in the presence of His–PP2Acα, clearly showing the existence of a ternary complex involving these proteins (Fig. 4A). A second experiment was performed in which α4 was fused to GST and immobilized on glutathione–Sepharose beads. As expected, His–TIPRL associated only with GST–α4 in the presence of His–PP2Acα (data not shown). Similar results were obtained using the PP2Ac-binding domain of α4, α4Δ222 [24], instead of the full-length protein (Fig. 4B), which further confirms that the TIPRL–α4 association is mediated by PP2A and suggests that no direct interaction between TIPRL and α4 is needed to stabilize this complex.

Figure 4.

 Ternary complex formed by TIPRL, PP2Ac and α4. (A) GST, GST–TIPRL, GST–PP2Acα and His–α4 were expressed separately in E. coli and His–PP2Acα was coexpressed with His–α4 in E. coli. Bound proteins were resolved by SDS/PAGE and detected by immunoblotting with an antibody for α4 or by Coomassie Brilliant Blue staining. His–α4 associated with GST–TIPRL only in the presence of His–PP2Acα. GST and GST–PP2Acα were used as negative and positive controls, respectively; I: input; B: bound. (B) The experiment shown in (A) was repeated using a C-terminal deletion of α4 (α4Δ222) instead of full-length protein to show that only the PP2Ac-interacting domain of α4 is sufficient to assemble the ternary complex. (C) TIPRL does not compete with α4 for PP2Ac binding. GST–PP2Ac was coexpressed with His–α4 in E. coli and the complex was affinity-purified on glutathione–Sepharose beads. Samples of the complex incubated with recombinant His–TIPRL (right) or of the control without His–TIPRL (left) were analyzed by SDS/PAGE (10%) and visualized by Coomassie Brilliant Blue staining. TIPRL interacted with the PP2Ac:α4 complex previously formed. (D) In vivo association of TIPRL and α4. Endogenous TIPRL was immunoprecipitated from HEK293 cell extracts and immunoprecipitates were probed with antibodies for TIPRL and α4. The * indicates staining of IgG heavy chain and is shown as a loading control. (E) The left panel shows western blot analyses of the elution profiles of PP2Ac, TIPRL and α4 fractionated by gel filtration chromatography. MWM indicates the elution positions of molecular mass markers are shown above the panels. The profiles of the three proteins overlap over a region that coincides with the expected molecular mass of the ternary complex (∼ 110 kDa). The TIPRL peak fractions indicated in the bottom of the left panel were fractionated by ion exchange chromatography. The elution profiles of PP2Ac, TIPRL and α4 from the ion-exchange chromatography are shown in the right panel. Only relevant fractions are shown.

GST pull-down assays indicated that TIPRL and α4 bind simultaneously to PP2Ac. This was confirmed using sequential binding experiments. Initially, GST–PP2Acα was coexpressed with either His–TIPRL or His–α4 and the GST–PP2Acα:His–TIPRL and GST–PP2Acα:His–α4 complexes were affinity-purified on glutathione–Sepharose columns. Subsequently, the GST–PP2Acα:His–TIPRL complex was incubated with His–α4 and the GST–PP2Acα:His–α4 complex was incubated with His–TIPRL. Binding of His–TIPRL to the previously formed GST–PP2Acα:His–α4 complex is shown in Fig. 4C. In the reciprocal experiment, binding of His-α4 to the previously formed GST–PP2Acα:His–TIPRL complex was also observed (data not shown). Because of the lower levels of expression of GST–PP2Acα relative to His–α4 or His–TIPRL, the recovered dimeric complexes were stoichiometric, and binding of the third protein without displacing the one that was previously associated with the complex was interpreted as an evidence of simultaneous binding to PP2Acα.

The results of these in vitro binding experiments suggested that although TIPRL and α4 do not interact directly, they may be associated in vivo in a ternary complex with PP2Ac. In agreement with this hypothesis, α4 was specifically detected in TIPRL immunoprecipitates from HEK293 cell extracts (Fig. 4D). To obtain further evidence on the TIPRL:PP2Ac:α4 association in vivo, HEK393 cell extracts were submitted to gel-filtration chromatography and TIPRL, PP2Ac and α4 were detected by western blotting (Fig. 4E). PP2Ac elutes in two major peaks, one of which, with molecular size in the range above 158 kDa, overlaps with only α4, whereas the second overlaps with both α4 and TIPRL (Fig. 4E, left panel). The elution profiles of the three proteins overlap in several fractions corresponding to the expected molecular mass of a ternary complex (∼ 110 kDa), which is in agreement with the existence of such a complex in mammalian cells. The TIPRL peak fractions from the gel-filtration chromatography were further fractionated on an ion-exchange column. The elution peaks of the three proteins correspond to the same fractions, further indicating that they are associated.

Regulation of PP2Ac activity by TIPRL

α4 has been characterized as a regulator of type 2A phosphatases [7–9]. The finding that TIPRL interacts with catalytic subunits of type 2A phosphatases suggests that it might also directly regulate PP2Ac activity. In order to test this hypothesis, in vitro assays were performed in which the activity of PP2A core enzyme (A and C subunits) was measured in the presence of His–α4 or His–TIPRL using the phosphopeptide RRA(pT)VA as a substrate. Because His–α4 and His–TIPRL are able to bind PP2Ac simultaneously, the effect of both proteins was also assayed. Under these conditions, His–α4 and His–TIPRL acted as PP2A inhibitors, but no additive effect on PP2A inhibition was observed in the presence of both His–α4 and His–TIPRL compared with the inhibitory effect of each single protein (Fig. 5A).

Figure 5.

 Regulation of PP2Ac by TIPRL and α4. (A) In vitro assay of PP2A core enzyme activity in the presence of purified His–TIPRL and/or His–α4 using the Promega phosphatase assay system. Activities are expressed as a fraction of the positive control (without TIPRL and α4). (B, C) PP2Ac interaction with TIPRL or α4, respectively, is not affected by okadaic acid treatment. GST–PP2Acα bound to glutathione–Sepharose beads was incubated with okadaic acid (1 µm) and His–TIPRL or His–α4. Bound proteins were resolved by SDS/PAGE (10%) and probed with antibodies for the histidine and GST tags.

To verify whether phosphatase inhibition was due to occlusion of the active site, in vitro binding assays were performed in the presence of the PP2Ac inhibitor okadaic acid. These assays showed that binding of His–TIPRL or His–α4 to GST–PP2Ac was not affected by previous incubation of GST–PP2Ac with okadaic acid (Fig. 5B,C), and also that okadaic acid was not able to induce dissociation of the copurified complexes His–TIPRL:GST–PP2Ac and His–α4:GST–PP2Ac (data not shown). Previously reported okadaic acid-induced dissociation of the α4:PP2Ac complex [25] was interpreted as evidence that the binding site for α4 might overlap the active site of the catalytic subunit. However, the results obtained in this study indicate that α4 and TIPRL are allosteric regulators of PP2Ac rather than inhibitors, which is in agreement with published observations showing that α4 binds PP2Ac on the surface opposite to the active site [26], and that it has opposing allosteric effects on PP2Ac and PP6c [27].

Rapamycin pathway-independent association of TIPRL, PP2Ac and α4 in human K562 cells

Although in yeast Tap42 and type 2A phosphatases are key players in the TOR pathway [19], the role of α4 and PP2Ac in the mammalian rapamycin-sensitive pathway remains controversial [7,8,9,14,25,28]. To test TIPRL involvement in the mTOR pathway, α4 or PP2Ac were immunoprecipitated from K562 cell extracts following rapamycin treatment. TIPRL coimmunoprecipitated specifically with α4, which further confirms the existence of a TIPRL:PP2Ac:α4 complex in vivo (Fig. 6). However, none of the pairwise interactions tested (PP2Ac:TIPRL, PP2Ac:α4, TIPRL:α4) was affected by rapamycin treatment. These observations support the existence of a TIPRL:PP2A:α4 heterotrimer in human cells, whose assembly is independent of the mTOR signaling pathway (Fig. 6).

Figure 6.

 Association of TIPRL, PP2Ac and α4 is not affected by rapamycin treatment. K562 cells were treated with 200 nm rapamycin or dimethylsulfoxide for 3.5 h in serum-free medium (A) or in the presence of 10% fetal calf serum (B, C). TIPRL (A) PP2Ac (B) and α4 (C) were immunoprecipitated from whole cell extracts (WCE), resolved on SDS/PAGE (10%) and probed with antibodies for α4, PP2Ac and TIPRL. The interactions PP2Ac:TIPRL (A) and PP2Ac:α4 (B), as well as the PP2Ac-mediated TIPRL:α4 association (C) were specifically detected and were not affected by rapamycin treatment.

Discussion

The interaction analyses presented in this study show that TIPRL interacts specifically with the C-terminal region of the catalytic subunits of type 2A phosphatases. Residues 210–309 of PP2Ac are sufficient for interaction with TIPRL. The TIPRL region that interacts with PP2Ac was investigated by using a reverse yeast two-hybrid approach, which identified amino acid substitutions in four independently isolated mutants (D71L, I136T, M196V and D198N) that block their interaction with type 2A phosphatases. TIPRL shows a subcellular distribution that coincides with PP2Ac in human HEK293 cells and inhibits its activity in vitro. Okadaic acid does not affect TIPRL interaction with PP2Ac, suggesting that its binding surface on PP2Ac does not involve the active site. These findings characterize TIPRL as a novel allosteric regulator of type 2A phosphatases, a role that has been attributed to date only to the α4 protein. The fission yeast ortholog of Tip41 was characterized as a regulator of type 2A phosphatases [21], which is in agreement with our results.

Because both TIPRL and α4 interact with the catalytic subunits of type 2A phosphatases, we examined the possibility of their simultaneous association, and showed that TIPRL forms a ternary complex with α4 and PP2Ac in mammalian cells and that this complex can be reconstituted in vitro from purified, recombinant proteins. The 3D arrangement of the binding sites for TIPRL and α4 on the surface of PP2Ac shows that they are in close proximity, but not overlapping, which allows the assembly of the TIPRL:PP2Ac:α4 complex (Fig. 7A). Genetic mapping of the interaction sites shows that α4 and TIPRL bind PP2Ac approximately on the same regions as PR65/A and B-type subunits, respectively. α4 and PR65/A bind to overlapping sites on the surface of PP2Ac in a mutually exclusive fashion, requiring complementary charged residues [26]. The α4-binding surface on PP2Ac was mapped to two separated regions, comprising residues 19–22 and 150–164 [17], which are represented in blue in Fig. 7. The interaction of PP2Ac with the regulatory B subunit requires the extreme C-terminal region of the catalytic subunit [29] and the interaction site for TIPRL was mapped to the C-terminal third of PP2A, showing that the TIPRL-binding region on PP2A is in close proximity to, possibly overlapping, the B-subunit-binding region. These similarities suggest that the overall shape and subunit arrangement of the TIPRL:PP2Ac:α4 complex might resemble that of the canonical A:B:C complex, although their assembly and regulation appear to be different. In the A:B:C complex, the A subunit binds to C and enhances its binding to B, whereas α4 and TIPRL appear to bind PP2Ac independently. There is also no evidence of physical contact between TIPRL and α4 in the ternary complex, which contrasts with the existence of an A:B interface [30,31].

Figure 7.

 PP2A catalytic subunit regions responsible for α4 and TIPRL binding. (A) The structure of PP2Acα downloaded from PDB (code 2IE3) is shown in ribbon (left) and space filling models (right). The regions responsible for binding to α4 and to TIPRL1 are shown in blue and violet, respectively. Residue Glu42 (yellow) is critical for interaction with α4 [26]. (B) Multiple sequence alignment of PP2A catalytic subunit orthologs. * and : indicate conserved residues and conserved amino acid substitutions, respectively. Active site residues are colored red.

Important differences between the yeast and mammalian models have been found. First, yeast Tip41 was reported to compete with Sit4 for Tap42 binding [20], whereas TIPRL and α4 can bind simultaneously to PP2Ac. In addition, the rapamycin-insensitive assembly of the TIPRL:PP2Ac:α4 complex also contrasts with yeast studies [19] and with some studies involving mammalian cells [7,8,14], although several studies have already reported that rapamycin treatment has no effect on the assembly of the PP2Ac:α4 complex [9,25,28]. While this manuscript was in preparation, similar observations were published by McConell et al. [32] regarding the rapamycin-insensitive binding of TIPRL to type 2A phosphatases. The effect of rapamycin on the stability of these complexes might depend on the cell line, because some cell lines are more sensitive to rapamycin than others. The mTOR pathway is constitutively active in the K562 cell line due to the expression of the BCR/Abl kinase, and this cell line responds to rapamycin treatment by dephosphorylating the ribosomal protein S6 [33]. However, no effect of rapamycin on the stability of the TIPRL:PP2Ac:α4 complex was observed in this cell line, although it cannot currently be ruled out that rapamycin responsiveness is not at the level of complex stability, but rather at the level of activity or substrate specificity. The apparent discrepancies between studies in yeast and mammalian cells indicate that the TOR signaling pathway is not as conserved as previously thought. Most probably, the TIPRL:PP2Ac:α4 complex participates in other signaling pathways, including the ataxia talangiectasia mutated/ataxia telangiectasia and Rad-3-related (ATM/ATR) pathway [32], but its targets remain to be identified.

In conclusion, our results show that TIPRL directly binds the catalytic subunits of type 2A phosphatases, but not α4, and that it regulates the activity of PP2A. These findings contrast with the model proposed for the yeast counterparts [20], but agree with recently published studies involving the human proteins [5,32]. In addition to previous studies, we have mapped the TIPRL-binding region on PP2Ac and identified some of the residues on TIPRL which are responsible for phosphatase binding. Finally, we report for the first time the ternary association of PP2Ac, α4 and TIPRL.

Experimental procedures

Plasmid construction

A list of the plasmid vectors used in this work is found in Table 1. The TIPRL cDNA (NM_152902) was amplified from a fetal brain cDNA library (Clontech Laboratories, Inc., San Diego, CA) and cloned into pTL1 (EcoRI–BamHI sites), pET–TEV (NdeI–BamHI sites) and pET–GST–TEV (NcoI–BamHI sites). pTL1, pET–TEV and pET–GST–TEV are derivatives of pBTM116 and pET28a (Novagen, Darmstadt, Germany) that have been previously described [34]. pTL1–TIPRL encodes TIPRL containing an N-terminal lexA fusion and harbors the bacterial kanamycin marker. pET–TIPRL and pET–GST–TIPRL encode N-terminal hexahistidine and GST fusions, respectively, separated by a TEV protease cleavage site. Construction of plasmids pET28a–α4 and pET28a–α4Δ222 was described previously [24]. To construct a plasmid encoding a GST–α4 fusion protein, the cDNAs of GST, digested with XbaI–SalI, and α4, digested with SalI–BamHI, were subcloned into the plasmid pET29a (Novagen) previously digested with XbaI–BamHI. The phosphatase cDNAs isolated from positive two hybrid interactions were subcloned into the EcoRI and XhoI sites of pGEX-5x2 (GE Healthcare, Piscataway, NJ). The PP2Acα cDNA was also subcloned into pProEx HtB (Invitrogen, Carlsbad, CA) using the same restriction sites. Full-length PP4c cDNA was amplified from a human leukocyte cDNA library and cloned into pACT2 between the EcoRI and XhoI sites.

Table 1.   Plasmid vectors used in this study.
PlasmidRelevant featuresRef.
pTL1–TIPRLexADBD::TIPRL TRP1 2 µmThis study
pACT2–PP2AcαGAL4AD::PP2Acα LEU2 2 µmThis study
pACT2–PP2AcβGAL4AD::PP2Acβ LEU2 2 µmThis study
pACT2–PP2Acβ (14–309)GAL4AD::PP2Acβ(14–309) LEU2 2 µmThis study
pACT2–PP4c (175–307)GAL4AD::PP4c(175–307) LEU2 2 µmThis study
pACT2–PP4c (196–307)GAL4AD::PP4c(196–307) LEU2 2 µmThis study
pACT2–PP6c (106–305)GAL4AD::PP6c(106–305) LEU2 2 µmThis study
pACT2–PP6c (171–305)GAL4AD::PP6c(171–305) LEU2 2 µmThis study
pACT2–PP2AcCT (210–309)GAL4AD::PP6c(210–309) LEU2 2 µmThis study
pACT2–PP4cGAL4AD::PP4c LEU2 2 µmThis study
pBTM116–NIP7lexADBD::NIP7 TRP1 2 µm[22]
pGAD–NOP8GAL4AD::NOP8 LEU2 2 µm[22]
pET28a–α4His6:: α4, kanR[24]
pET28a–α4Δ222His6:: α4Δ222, kanR[24]
pET–TIPRLHis6::TIPRL, kanRThis study
pProEx–PP2AcαHis6::PP2Acα, ampRThis study
pGEX−5X2GST, ampRGE Healthcare
pGEX−5X2–PP2AcαGST::PP2Acα, ampRThis study
pGEX−5X2–PP2AcαGST:: PP2Acα, ampRThis study
pGEX−5X2–PP4c (175–307)GST::PP4c(175–307), ampRThis study
pGEX−5X2–PP6c (106–305)GST::PP6c(106–305), ampRThis study
pGEX−5X2–PP2AcCT (210–309)GST::PP2AcCT(210–309), ampRThis study
pET–GST–TIPRLGST::TIPRL, kanRThis study

Yeast two-hybrid assays

Yeast two-hybrid screens were performed using yeast strain L40 harboring plasmid pTL1–TIPRL. Expression of the lexA–TIPRL fusion was verified by western blot with an antibody for lexA (Invitrogen) and self-activation of the reporter genes was tested with L40 strain transformed with plasmids pTL1–TIPRL and pGAD424. Large- and library-scale sequential transformations of L40/pTL1–TIPRL with a human leukocyte cDNA library cloned into pACT2 (BD Bioscience Clontech) were performed using the PEG/lithium acetate protocol as previously described. Positive clones (His3+) were initially selected on SD-WLH (synthetic complete medium lacking tryptophan; leucine and histidine) supplemented with 5 mm 3-AT. Subsequently, His3+ clones were subjected to a second round of selection based on the activation of the reporter gene lacZ, using X-Gal filter assays. The large-scale sequential transformation yielded 54 positive clones from 4.5 × 104 clones tested and the library scale transformation yielded 1100 positive clones from 6.0 × 106 clones tested. Plasmid pACT2 was rescued from 88 positive clones and the TIPRL-interacting proteins identified by DNA sequencing followed by BLAST analyses (http://www.ncbi.nlm.nih.gov/BLAST/).

Random mutagenesis and reverse two-hybrid assays

A PCR-based random mutagenesis strategy [23] was used to generate a library of mutant TIPRL cDNAs, using high MgCl2 concentrations (3 mm) to reduce DNA polymerase fidelity. This library was transformed into L40/pACT2–PP2Acα strain along with pTL1–TIPRL linearized with NdeI and the cotransformants were selected by plating on SD-WL (synthetic complete medium lacking tryptophan and leucine). Interaction-deficient mutants were selected by replica plating on SD-WL and SD-WLH +10 mm 3-AT to identify clones showing a His phenotype. His colonies were subsequently tested for activation of the lacZ reporter gene using an X-Gal filter assay. The colonies in which the yeast two-hybrid markers were no longer activated were tested for expression of the full-length TIPRL by western blotting using an antibody for lexA (Invitrogen). Colonies negative for two-hybrid interaction and expressing the full-length lexA–TIPRL fusion protein were selected for further analysis. The plasmid pTL1–TIPRL was rescued from these strains and analyzed by DNA sequencing to identify the amino acid substitutions that abrogate TIPRL–PP2Ac interaction. Loss of interaction was confirmed by retransforming pTL1 containing the mutant variants of TIPRL into L40 strains carrying pACT2–PP2Acα, pACT2–PP2Acβ, and pACT2–PP24c.

GST pull-down assays

Bacterial cells [E. coli BL21(DE3)] were grown on Luria–Bertani medium containing the appropriate antibiotics at 50 µg·mL−1 and recombinant protein expression was induced with 0.5 mm isopropyl thio-β-d-galactoside for four hours at 25 °C. Bacterial pellets were suspended in ice cold lysis buffer [NaCl/Pi (140 mm NaCl, 2.7 mm KCl, 10 mm Na2HPO4, 1.8 mm KH2PO4), containing 0.5% Igepal CA-630, 1 mm dithiothreitol and 1 mm phenylmethylsulfonyl fluoride], treated with lysozime, sonicated and centrifuged (20 000 g, 10 min, 4 °C). The supernates were incubated with glutathione–Sepharose beads (GE Healthcare) for 30–60 min at 4 °C. Subsequently, the beads were washed three times with ice-cold lysis buffer and bound proteins were eluted with SDS/PAGE sample buffer. For semi in vivo binding experiments, glutathione–Sepharose beads containing GST–PP2Ac or GST were prepared as above and, before elution in SDS/PAGE sample buffer, the beads were incubated with HEK293 cell extracts for 30–60 min at 4 °C and washed three times with High Salt Buffer (50 mm Tris/HCl pH 8.0, 250 mm NaCl, 0.5% Igepal CA-630, 5 mm EDTA) supplemented with protease inhibitors (Complete, EDTA-Free, Roche Molecular Biochemicals, Indianapolis, IN).

Phosphatase assays

Activity of PP2A core enzyme (A and C subunits, Promega, Madison, WI) was assayed using the Serine/Threonine Phosphatase Assay Kit (Promega) according to the manufacturer's instructions. The colorimetric measurements were taken in an ELx 800 microplate reader (Bio Tek Instruments, Winooski, VT) at 630 nm 0.5 units of PP2A were assayed in triplicate in the presence of 25 ng of His-tagged α4 and/or TIPRL, for 15 min at 30 °C. His-tagged TIPRL was purified by affinity chromatography on a Ni-NTA column (Qiagen, Valencia, CA), and his-tagged α4 was purified as described previously [22]. TIPRL and α4 were quantified by direct absorbance at 280 nm.

Cell culture, fractionation and immunoprecipitation assays

HEK293 and K562 cells were cultured in minimal essential alpha medium supplemented with 10% fetal bovine serum, with 100 U·mL−1 penicillin and 100 µg·mL−1 streptomycin at 37 °C in a humidified atmosphere with 5% CO2. Total cell extracts were prepared by suspending cells in high salt buffer supplemented with protease inhibitors (Complete, EDTA-Free, Roche) and NaF 50 mm. After 30 min on ice, lysates were cleared by centrifugation (20 000 g, 10 min, 4 °C). Cytoplasmic fractions were obtained by suspending cell pellets in low salt buffer (10 mm Tris/HCl pH 7.4, 320 mm sacarose, 2 mm MgCl2, 3 mm CaCl2, 0.4% Igepal CA-630, 0.5 mm dithiothreitol) supplemented with protease inhibitors (Complete, EDTA-Free, Roche) and centrifuging at low speed (800 g, 5 min, 4 °C) after 10 min on ice. Nuclear fractions were prepared from this low-speed centrifugation pellet as described above for total cell extracts. Protein extracts were quantified using the BCA assay kit (Pierce, Rockford, IL) and BSA as standard. Rabbit polyclonal antibodies for TIPRL (A300–663A), PP2Ac (A300–732A) and α4 (A300–471A) were used in immunoprecipitation assays, and antibody for murine MafB (A300–612A), which does not recognize any protein in human cells, was used as a negative control (Bethyl Laboratories, Montgomery, TX). Protein extract (500 µg) was incubated with 1 µg of antibody and Protein A–Sepharose beads (Pharmacia, Uppsala, Sweden) for 3 h at 4 °C. The Protein A–Sepharose beads were washed three times with ice-cold high salt buffer and the bound proteins were eluted with SDS/PAGE sample buffer. Western blots were performed as described previously [35].

Gel-filtration and ion-exchange chromatography of HEK293 cell extracts

HEK293 cells were suspended in High Salt Buffer supplemented with protease inhibitors (Complete, EDTA-Free, Roche), incubated on ice for 30 min and sonicated. Cell lysates were centrifuged at 20 000 g for 10 min at 4 °C and the supernates were centrifuged at 100 000 g for 60 min at 4 °C. The supernate of the high-speed centrifugation (S100 extract) was loaded on a Superdex 200 16/60 column (GE Healthcare) and isocratic elution was performed in high salt buffer at 1 mL·min−1. The column was calibrated using ferritin (440 kDa), aldolase (158 kDa), albumin (66 kDa) and ribonuclease (13.7 kDa) and the void elution volume was determined using blue dextran. Fractions from the gel filtration chromatography corresponding to the TIPRL elution peak were dialyzed against ion exchange buffer A (50 mm Tris/HCl pH 8.0, 50 mm NaCl, 5 mm EDTA, 0.1% Igepal CA-630) for 16 h at 4 °C and fractionated on a Mono Q column pre-equilibrated in the same buffer. The column was washed with 10 mL of buffer A and proteins were eluted with a linear gradient of 0–100% buffer B (50 mm Tris/HCl pH 8.0, 1 m NaCl, 5 mm EDTA, 0.1% Igepal CA-630) in 15 mL at 1 mL·min−1. Chromatography was performed on an ÄKTA FPLC (GE Healthcare).

Structural modeling and amino acid sequence alignment

The coordinates of the PP2Acα structure were downloaded from PDB (code 2IE3) and the figures generated using PYMOL (http://www.pymol.org). Multiple sequence alignment was performed with the clustalw algorithm (http://www.ch.embnet.org/software/ClustalW.html).

Acknowledgements

We thank Tereza C. Lima Silva, Adriana C. Alves and Zildene G. Corrêa for technical support, and Carlos R. C. Paier and Daniela S. Razolli for helping with yeast reverse two-hybrid screening. JHCS was recipient of CNPq fellowship. This work was supported by FAPESP grant 06/02083-7 (NITZ) and the FAPESP SMolBNet (00/10266-8) and CEPID/CBME (98/14138-2) programs.

Ancillary