Protein phosphatase 2A Aα regulates Aβ protein expression and stability

Protein phosphatase 2A (PP2A) represses many oncogenic signaling pathways and is an important tumor suppressor. PP2A comprises three distinct subunits and forms through a highly regulated biogenesis process, with the scaffolding A subunit existing as two highly related isoforms, Aα and Aβ. PP2A's tumor-suppressive functions have been intensely studied, and PP2A inactivation has been shown to be a prerequisite for tumor formation. Interestingly, although partial loss of the Aα isoform is growth promoting, complete Aα loss has no transformative properties. Additionally, in cancer patients, Aα is found to be inactivated in a haploinsufficient manner. Using both cellular and in vivo systems, colorectal and endometrial cancer cell lines, and biochemical and cellular assays, here we examined why the complete loss of Aα does not promote tumorigenesis. CRISPR/Cas9-mediated homozygous Aα deletion resulted in decreased colony formation and tumor growth across multiple cell lines. Protein expression analysis of PP2A family members revealed that the Aα deletion markedly up-regulates Aβ protein expression by increasing Aβ protein stability. Aβ knockdown in control and Aα knockout cell lines indicated that Aβ is necessary for cell survival in the Aα knockout cells. In the setting of Aα deficiency, co-immunoprecipitation analysis revealed increased binding of specific PP2A regulatory subunits to Aβ, and knockdown of these regulatory subunits restored colony-forming ability. Taken together, our results uncover a mechanism by which PP2A Aα regulates Aβ protein stability and activity and suggests why homozygous loss of Aα is rarely seen in cancer patients.

The serine/threonine protein phosphatase 2A (PP2A) 2 is an important tumor suppressor protein which negatively regulates many oncogenic signaling pathways (1,2). It is a heterotrimeric enzyme comprised of a scaffolding A subunit, catalytic C sub-unit, and one of several regulatory B subunits. The A subunit consists of two closely related isoforms, designated A␣ and A␤, which are 86% identical (3). The A subunit structure is made up of 15 tandem Huntingtin Elongation A subunit Tor (HEAT) repeats and the high sequence similarity between A␣ and A␤ suggests that these two proteins have similar protein structures (3)(4)(5). However, the A␤ protein includes an N-terminal extension of 12 amino acids, which is not present in A␣ (4) (Fig. S1). Both isoforms are targets of viral antigens that have been implicated in the initiation of cellular transformation. The Polyoma middle T viral antigen binds to both A␣ and A␤, whereas only A␣ binds simian virus 40 (SV40) small T antigen, highlighting that there may be structural differences between the two isoforms (4,6). Both A␣ and A␤ have been identified to function as tumor suppressors, however the mechanisms of inactivation are unique to each isoform (6 -8). Interestingly, whereas complete loss of A␤ results in transformation, A␣ functions as a tumor suppressor in a haploinsufficient manner. These data are reflective of what has been seen in large sequencing cohorts, including The Cancer Genome Atlas (TCGA). These studies have revealed that although the A␣ isoform is altered in 35% of human cancers, homozygous deletions of A␣ are exceedingly rare, occurring in only 0.3% of patient tumors (Fig. S2, A and B). In contrast, deletions of A␤ are more common, as the A␤ gene PPP2R1B is located within the chromosomal region 11q23, a region commonly deleted in cancer (9 -13).
To define the molecular basis for why homozygous A␣ deletion appears to be unfavorable for cancer cell growth, we used a combination of biochemical and cellular assays to examine the functional ramifications of complete loss of the A␣ subunit. CRISPR-Cas9 mediated homozygous deletion of the A␣ subunit was growth suppressive across multiple cellular contexts. We examined the expression levels of various PP2A subunits in control and A␣-deficient cells and found that A␣ loss lead to a robust increase in expression of the A␤ scaffold subunit as a result of increased A␤ protein stability. Knockdown of A␤ in the A␣ knockout cells was lethal, suggesting that a minimum amount of PP2A activity is necessary for cell survival. Co-immunoprecipitation of A␤ protein in the presence and absence of A␣ revealed that there was an increase in specific A␤ holoenzymes, including B56␥ and PR130, upon A␣ deletion. Modulating specific A␤ holoenzymes by knockdown of B56␥ restored colony growth, indicating that B56␥-A␤ holoenzymes are at least partially responsible for the growth-suppressive effects of . The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains Figs. S1-S6 and Tables S1 and S2. 1 To whom correspondence should be addressed. cro ARTICLE A␤ upon homozygous A␣ loss. Together, these findings highlight why complete A␣ loss is rarely seen in patients and reveals a unique mechanism through which A␣ regulates A␤ protein levels and subsequent phosphatase activity.

Knockout of PP2A A␣ decreases colony formation
To eliminate PP2A A␣ protein, we used CRISPR-Cas9 to create insertion and deletion mutations in exon 5, corresponding to HEAT 5 of the A␣ protein ( Fig. 1A; Fig. S3). The double nicking strategy using mutant Cas9n was utilized to limit off-target effects, and two separate knockout clones for each cell line (designated as KO.1 and KO.2) were used to verify the on-target specificity of our findings (14, 15). Colorectal cancer and endometrial cancer cell lines were chosen for these analyses because both heterozygous mutations as well as heterozygous loss of the A␣ subunit are recurrent in these contexts, indicating that haploinsufficiency of A␣ in these cancers may be a mechanism of PP2A inactivation (The Cancer Genome Atlas) (16,17). Specifically, two colorectal cancer cell lines, SW620 and HCT-116, and two patient-derived endometrial cell lines, UT89 and UT150, were selected for the generation of A␣ knockouts. Knockout clones were first screened by DNA sequencing and subsequent Western blot analysis confirmed the knockout of A␣ protein. Because of the high sequence homology between A␣ and A␤, an antibody targeting both isoforms was used for the detection of A␣. Knockout of A␣ resulted in an almost complete loss of total A subunit protein expression across all cell lines analyzed, consistent with reports that the A␣ isoform is the predominantly expressed isoform in most tissues (Fig. 1, B-E) (4). Consistent with previous literature, depletion of the A␣ subunit also resulted in decreased levels of the PP2A C subunit expression (Fig. 1, B-E) (18,19).
To determine the effects of A␣ subunit knockout on cell growth, clonogenic colony formation assays were performed. Interestingly, in these assays the loss of A␣ subunit resulted in an approximate 20 -30% decrease in colony formation across all cell lines and knockout clones analyzed (Fig. 1, F and G). Taken together, these findings suggest that although heterozygous A␣ loss is tumor promoting, the complete loss of A␣ is not growth promoting, but instead growth suppressive.

Loss of PP2A A␣ results in altered protein expression of specific PP2A family members
To understand why knockout of the A␣ subunit was growth suppressive, we investigated how the loss of A␣ protein impacted other PP2A subunit family members. To do so, we performed Western blot analysis of regulatory subunits, includ-ing striatin, striatin 3, PR130, PR70, B55␣, B56␥, and B56␦, as well as A␣'s closely related isoform A␤ (Fig. 2, A-D; Fig. S4). To analyze A␤ protein levels, an antibody directed toward the unique N-terminal region was used. Interestingly, the protein expression levels of only select regulatory subunit isoforms were affected by A␣ knockout. Specifically, the expression of B55␣ and B56␦ dramatically decreased upon A␣ knockout, whereas other isoforms, including B56␥, PR130, and the striatin proteins were unaffected. Remarkably, although A␣ and A␤ have been proposed to have distinct functions, the protein levels of A␤ increased ϳ5-fold across all A␣ KO cell lines (Fig. 2, A-D). The increase in A␤ protein expression was not associated with changes in A␤ mRNA levels, as measured in SW620 and UT150 (Fig. S5), suggesting that the increase in A␤ expression occurred posttranscriptionally. Taken together, these results indicate that loss of the more abundantly expressed scaffold isoform, A␣, has widespread effects on PP2A subunit protein levels across all subunit families (scaffolding A, regulatory B, and catalytic C) and suggests that distinct PP2A subunits, in particular A␤, may be directly dependent on the A␣ scaffold for their stability.

Knockout of PP2A A␣ decreases tumor growth and alters protein expression in vivo
To determine the effects of A␣ depletion on tumor growth in vivo, we performed an SW620 xenograft model using control

PP2A A␣ regulates A␤ protein expression and stability
and A␣ KO.1 cells (Fig. 3). SW620 was selected for downstream analyses as it grows efficiently in vivo and is a well-characterized cell line, part of the NCI-60 panel. In this model, consistent with growth in vitro, the SW620 A␣ KO.1 tumors grew significantly slower than the control tumors ( Fig. 3A; Table S1). Subsequent analysis of PP2A subunit expression by Western blotting showed a trend consistent with what was observed in cell culture (Fig. 3B). Importantly, there was significantly less A subunit and C subunit protein in these tumors, confirming that the A␣ subunit knockout was maintained (Fig. 3, B and C). Additionally, there was significant up-regulation of the A␤ subunit, consistent with data observed in cell culture models (Fig. 3, B). Finally, quantification of the regulatory subunits showed no significant changes in the striatin, PR70 family, or B56␥ regulatory subunits consistent with in vitro analyses (Fig. 3, B and D). There was a trend of decreased B55␣ levels in the A␣ knockout tumors, but this change did not reach statistical significance (Fig. 3, B and D). B56␦ levels could not be measured in vivo as the antibody did not detect adequate bands in any group (data not shown). In summary, these results showed that A␣ knockout caused a reduction in tumor growth. Additionally, A␣ knockout resulted in the up-regulation of A␤ and decrease of C subunit expression in vivo.

Knockout of PP2A A␣ alters A␤ and C subunit protein stability
Based on the altered PP2A subunit expression upon A␣ depletion, we characterized the effects of A␣ expression on the stability of select PP2A subunits. To determine whether A␤ protein stability was altered in the absence of A␣ protein, we treated control and knockout SW620 cells with cycloheximide and monitored A␤ protein expression by Western blotting over time (Fig. 4A). From these Western blot analyses, we graphed and calculated the A␤ protein half-life (Fig. 4, B and C). In control cells, the calculated half-life of A␤ was ϳ6 h. Conversely, in A␣ knockout cells, the half-life of A␤ increased to ϳ23 h, suggesting that the increase in A␤ protein seen in A␣  Table S1. B, protein from SW620 control (n ϭ 9) or A␣ KO.1 (n ϭ 8) tumors was isolated and analyzed for expression of PP2A subunits by immunoblotting. Because of the number of samples, multiple Western blots were run to analyze all samples, and control no. 3 was run on each blot of quantification. Representative Western blot analysis of four control and KO.1 tumors is shown. C, quantification of A and C subunit levels determined by immunoblotting. Control: n ϭ 9; KO.1: n ϭ 8; error bars, mean Ϯ S.D. (multiple t tests with Holm-Sidak multiple comparisons test; p values: **** Ͻ0.0001). D, quantification of striatin, striatin 3, PR130, PR70, B55␣, and B56␥ levels determined immunoblotting. Control: n ϭ 9; KO.1: n ϭ 8; error bars, mean Ϯ S.D.

PP2A A␣ regulates A␤ protein expression and stability
KO cells was a result of increased protein stability (Fig. 4, A-C). We also examined whether C subunit stability was altered, based upon our observation of decreased C subunit protein expression in A␣ knockout cells. Similarly, we treated control and knockout SW620 cells with cycloheximide and monitored C subunit protein levels by Western blotting over time (Fig.  4D). In control cells, the C subunit expression was only decreased 40% within 24 h, so a half-life could not be calculated (Fig. 4, E and F). Conversely, the calculated half-life of the C subunit in A␣ knockout cells was ϳ2 h, suggesting that the stability of the C subunit may be partially dependent on the presence of the A␣ subunit (Fig. 4, E and F). Additionally, B55␣ was also found to be less stable in the absence of A␣ (Fig. S6). Re-expression of exogenous A␣ protein restored A␤ and C subunit protein expression levels (Fig. 4, G-I). Finally, cotreatment of control SW620 cells with cycloheximide and MG-132, a proteasome inhibitor, rescued A␤ protein degradation, indicating that A␣ regulation of A␤ protein stability is mediated through the proteasome (Fig. 4, J and K). Additionally, cotreatment of SW620 A␣ knockout cells with cycloheximide and MG-132 rescued C subunit protein degradation, indicating that the A␣ regulation of C subunit stability is also mediated through the proteasome (Fig. 4, L and M). Taken together, these data indicate that knockout of A␣ alters the protein stability of both the A␤ and C subunit, suggesting that A␣ may play a role in their regulation.

A␤ is essential for the survival of A␣ knockout cells
The A␤ subunit of PP2A is an established tumor suppressor and complete knockdown of A␤ leads to cellular transformation (6). To determine whether the baseline differences in growth observed in control and A␣ knockout cells resulted from the up-regulation of A␤ and subsequent A␤ holoenzymes, we generated stable knockdown cell lines with two different shRNAs targeting A␤ in the SW620 control and KO.1 cells (Fig.  5, A and B). Stable cell lines were generated using lentiviral transduction followed by selection in puromycin. Surprisingly, after selection, although all cell lines were resistant to puromycin, only the control cells had significant knockdown of A␤, indicating that complete removal of all PP2A scaffolding subunit may not be tolerated in cells (Fig. 5, A and B). To determine whether the reduction of A␤ in A␣ knockout cells impacted cell viability, we performed acute knockdown of A␤ using two distinct sequence-specific siRNAs targeting A␤ in SW620 control and KO.1 cells (Fig. 5, C and D). Acute depletion of A␤ resulted in a significant decrease in A␤ protein expression in both the control and KO.1 cells (Fig. 5, C and D). However, there was a significant decrease in viability in only the A␣ knockout cells upon knockdown of A␤ by siRNA as measured by MTT and Western blotting for cleaved PARP and cleaved caspase 3 (Fig.  5, E and F). Together, these data indicate that there is a minimum amount of scaffold required for cell survival and are suggestive that the up-regulation of A␤, and the A␤ holoenzymes formed, allows for cell survival in the absence of the A␣ subunit expression.

Knockdown of B56␥ restores colony formation of A␣ knockout cells
Homozygous deletion of A␣ reduced tumor growth and was associated with up-regulation of A␤, suggesting that increased activity of A␤ containing holoenzymes may be tumor suppressing in this context. To determine the tumor-suppressive A␤ containing holoenzymes, we performed co-immunoprecipitation experiments to identify which B subunits were interacting with A␤. Using an A␤-specific antibody, we immunoprecipitated A␤ and its binding partners in the presence and absence of A␣ using control and A␣ KO SW620 cells and measured the interactions of select PP2A subunits by immunoblotting (Fig. 6,  A and B). To quantify interactions, the amount of each protein was first normalized to the amount of A␤ to control for the expression differences in A␤ between the two cell lines. After normalization, the resulting values were graphed relative to the SW620 controls to determine whether binding ratios were altered upon knockout of A␣ (Fig. 6B). Through this analysis, it was determined that specific regulatory subunits, including PR130, B56␦, and B56␥, were more highly bound to A␤ in the absence of A␣, although the increase in PR130 binding did not reach statistical significance. Interestingly, the loss of regulatory subunits PR130 and B56␥ have previously been implicated in cellular transformation (8,20), leading us to hypothesize that the increased formation of holoenzymes containing these regulatory subunits may contribute to the growth-suppressive phenotype seen upon loss of A␣. Conversely, knockdown of B56␦ has been shown to be unfavorable to transformation and resulted in a further reduction of colony growth (8). To determine whether the colony growth phenotype depends on specific regulatory subunits, we generated stable knockdown lines using specific shRNAs, focusing on the regulatory subunits previously implicated in cellular transformation, PR130 and B56␥, which also displayed increased binding to A␤ in the absence of A␣, and one regulatory subunit B55␣, which displayed unaltered binding to A␤ upon A␣ loss. A nontargeting shRNA con-

PP2A A␣ regulates A␤ protein expression and stability
struct was used as a control (Fig. 6, C and D). Stable knockdown of B55␣ and B56␥ was efficiently achieved in both control and A␣ KO SW620 cells. However, stable knockdown of PR130 was only obtained in the control SW620 cells, paralleling the results seen with stable knockdown of A␤, raising the possibility that PR130/A␤ holoenzymes are essential for cell viability in the

PP2A A␣ regulates A␤ protein expression and stability
absence of A␣. To determine the effects of the regulatory subunit knockdown on cell growth, clonogenic colony formation assays were performed (Fig. 6, E and F). Interestingly, the loss of each regulatory subunit resulted in similar growth changes in both the control and A␣ KO cells; regulatory subunit knockdown resulted in either decreased (shB55␣) or increased (shB56␥) colony growth. However, only the knockdown of B56␥ in the A␣ KO cells restored colony growth to baseline, suggesting that the B56␥/A␤ holoenzymes are critical for the growth-suppressive effects seen upon A␣ loss (Fig. 6F). Taken together, these data show that there is an increase in the formation of specific A␤-containing holoenzymes in the absence of A␣, and the increase in A␤-B56␥ holoenzymes is at least partially responsible for the decreased colony growth seen upon loss of A␣.

Discussion
The goal of this study was to elucidate the functional ramifications and potential clinical relevance of the complete loss of the PP2A A␣ subunit. Here, we demonstrate that complete loss of A␣ suppressed cell and tumor growth and describe a compensatory mechanism by which PP2A A␣ regulates A␤ protein stability and activity, which may suggest why homozygous loss of A␣ is rarely seen in patients.
Previous literature suggests that the A␣ and A␤ scaffolding subunits have distinct functions within a cell (4,6,7). Specifically, in cell-based transformation assays, the exogenous overexpression of the A␣ isoform was unable to compensate for the loss of A␤ on transformation (6). However, our results indicate that there are some overlapping functions between the two scaffold subunits. In a transformed, tumorigenic cell, loss of A␣ is tolerated because of a compensatory increase in A␤ protein. Subsequent removal of the A␤ caused a decrease in cell viability, indicating that the complete loss of PP2A scaffolding subunits is not tolerated. Thus, the overlapping functions of A␣ and A␤ are likely essential functions of PP2A necessary for cellular survival, whereas the distinct actions of the two scaffolds may represent more specialized activities. The interplay between the two subunits is more complicated on a physiological level, as evidenced by A␣ knockout mouse models. In mice, complete loss of A␣ is embryonically lethal and total body inducible knockout of A␣ in adult mice is also lethal, suggesting that A␤ is not able to completely compensate for A␣ during development or in normal physiology (21,22).
To date, there is very little known about what mechanisms regulate A␣ and A␤ expression. Transcription factors responsible for controlling the expression of A␣ and A␤ have been described for both isoforms (23,24) and may help explain their differential expression levels and tissue distribution. Our results suggest that these proteins are also regulated posttrans-lationally. Here we show that genetic loss of A␣ led to a compensatory up-regulation of A␤ through increased A␤ protein stability. When A␣ is present, A␤ undergoes rapid degradation via the proteasome and loss of A␣ greatly increased the half-life of the A␤ protein. Further exploration into the ubiquitination sites of both isoforms and the identification of the E3 ligases targeting these proteins will be critical in understanding the posttranslational mechanisms regulating the expression of the PP2A scaffolding subunits. Additionally, studies on proteasomal regulation of the scaffolding subunits may give further insight into tissue-specific expression differences that exist between the two scaffolds. Ubiquitination is a mechanism of regulation for other PP2A subunits, as proteasomal degradation contributes to the brain-specific expression of the BЈ␤ regulatory subunit (25).
Furthermore, we show that CRISPR/Cas9 mediated-knockout of A␣ resulted in altered expression of not only the closely related A␤ isoform, but of other PP2A subunits, including the C subunit and B55␣. The loss of C subunit expression upon loss of the A␣ scaffold has been well-documented (18,19). This finding has potential implications beyond PP2A, as similar mechanisms likely occur in other multimeric proteins, including other protein phosphatases, such as PP4 and PP6. Therefore, when altering protein expression with CRISPR/Cas9 careful characterization of not only the target protein but other subunit family members or interactors may be important in the interpretation of results obtained when using these methodologies. Interestingly, this may not apply to all regulatory subunits as depletion of B56␣ was shown to not affect expression of other subunits (26). Additionally, the decrease in C subunit and B55␣ subunit half-life upon A␣ loss may indicate that these proteins rely on A␣ for stability and could provide insight into how PP2A holoenzymes are stabilized.
Finally, we probed the A␤ interactome in the presence and absence of A␣ to determine whether there were alterations in regulatory B subunit binding to A␤ in this context. Indeed, we determined that specific regulatory subunits had increased A␤ complex formation in the absence of A␣, which may indicate that there are affinity differences between PP2A subunits and the two A scaffold isoforms. Further, we examined the dependence of the growth-suppressive effect of A␤ up-regulation on the expression of these regulatory subunits and determined that the removal of B56␥ restored colony growth in A␣ knockout cells, suggesting that A␤-B56␥ holoenzymes display tumor-suppressive activity. Further exploration into the PP2A substrates responsible for the growth-suppressive effect of A␤-B56␥ holoenzymes will be critical for understanding the signaling pathways regulating A␤-B56␥-dependent cell growth.

PP2A A␣ regulates A␤ protein expression and stability
In conclusion, our results demonstrate a novel mechanism by which PP2A A␣ regulates A␤ protein and activity. The knockout of A␣ and subsequent increase in A␤ results in decreased cell and tumor growth and the selective removal of B56␥ in these cells ameliorates the growth-suppressive effects caused by A␣ loss. Taken together, these data suggest that homozygous loss of A␣ is rarely seen in patients because of an increase in tumor-suppressive A␤ holoenzyme activity.

Antibodies and immunoblot analysis
Antibodies used in this study can be found in Table S2. Proteins from whole cells were lysed in RIPA buffer (Thermo Fisher Scientific). All lysis buffers were supplemented with protease and phosphatase inhibitors (Roche). Protein concentrations of cell extracts were determined by Pierce BCA Protein Assay kit (Thermo Fisher Scientific) and equal quantities of protein were separated by SDS-PAGE 12% polyacrylamide gels (Bio-Rad) and transferred to nitrocellulose membranes (Bio-Rad). Primary antibodies were detected with goat anti-mouse (Abcam, Cambridge, MA) or donkey anti-rabbit (GE Healthcare) conjugated to horseradish peroxidase using the Bio-Rad ChemiDoc XRS chemiluminescence imager. Densitometry quantification was performed within the Bio-Rad Image Lab software.

Generation of CRISPR/Cas9-mediated knockout cell lines: sgRNA design and cloning
The MIT CRISPR tool was used to design the pairs of sgRNAs against the gene coding for PP2A A␣ (PPP2R1A). A ϳ200-bp sequence within exon 5 of PPP2R1A was submitted to the CRISPR design tool for sgRNA design. The top and bottom strands of the sgRNA were purchased from Integrated DNA Technologies and cloned into PX461-pSpCas9n(BB)-2A-GFP (Addgene plasmid 48410), following the previously described protocol (15). The guide RNA sequences used to cleave PPP2R1A were as follows: Guide1a, GCTGCGGCCCGCCG-CACCATGGG; Guide 1b, CAAGCTGGGGGAGTTTGC-CAAGG (Fig. S3).

Transfection and isolation of knockout clones
Plasmids were transfected into the target cell lines using Lipofectamine 3000 (Thermo Fisher Scientific). 72 h post transfection, GFP-positive cells were sorted 1 cell/well into 96-well plates using FACS and incubated for ϳ3 weeks. At this time, each 96-well plate was split into two 48-well plates (one master plate and one replica plate).

Genotyping and validation of isolated clones
The replica plate was used to isolate genomic DNA using QuickExtract DNA reagent. Genomic DNA was amplified using PCR primers spanning exon 5 (forward: TACTTCC-GGAACCTGTGCTC; reverse: CCAGGAAGCAAAA-CTCA-CCT) and sent for Sanger sequencing to identify deletions. Protein isolated from clones with deletions in exon 5 were analyzed for the presence of knockouts by immunoblotting.

Xenograft tumor formation
All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) at Case Western Reserve University. Animal use and care was in strict compliance with institutional guidelines and all experiments conformed to the relevant regulatory standards by Case Western Reserve University. 5 ϫ 10 6 SW620 cells or A␣ KO.1 SW620 cells were injected subcutaneously in 50% Matrigel into the right flanks of 6-to 8-week-old female Balb/c nu/nu mice. Tumor volume was assessed by caliper measurement every other day. Tumor tissue was both formalin-fixed and snap frozen in liquid nitrogen for immunoblot analysis.

qPCR analysis
RNA was isolated from cells (Roche, 11828665001) and cDNA was prepared (Roche, 5893151001). mRNA levels were determined using SYBR Green (Roche, 04887352001) and primers directed at control or target genes and measured on the Roche LightCycler 480. Actin was used as a reference gene (forward: CCCACACTGTGCCCATCTAC; reverse: GCTTCTC-CTTAATGTCACGC). Three independent primer sets specific to PPP2R1A (A␣) or PPP2R1B (A␤) were used for this study.

Co-immunoprecipitation analysis
Cell lines were plated to 70% confluency in 150-mm plates. After 24 h, cells were harvested and co-immunoprecipitation was performed per Dynabeads Co-Immunoprecipitation Kit protocol (Thermo Fisher, 14321D). A␤ antibody (Novus) was coupled at a concentration of 7 g/mg of Dynabeads. Fresh conjugated beads were prepared for each biological replicate, three biological replicates were performed for each experiment.