PPP1R12A is a recycling endosomal phosphatase that facilitates YAP activation

Yes-associated protein (YAP) is a transcriptional coactivator that is essential for the malignancy of various cancers. We have previously shown that YAP activity is positively regulated by phosphatidylserine (PS) in recycling endosomes (REs). However, the mechanism by which YAP is activated by PS in REs remains unknown. In the present study, we examined a group of protein phosphatases (11 phosphatases) that we had identified previously as PS-proximity protein candidates. Knockdown experiments of these phosphatases suggested that PPP1R12A, a regulatory subunit of the myosin phosphatase complex, was essential for YAP-dependent proliferation of triple-negative breast cancer MDA-MB-231 cells. Knockdown of PPP1R12A increased the level of phosphorylated YAP, reduced that of YAP in the nucleus, and suppressed the transcription of CTGF (a YAP-regulated gene), reinforcing the role of PPP1R12A in YAP activation. ATP8A1 is a PS-flippase that concentrates PS in the cytosolic leaflet of the RE membrane and positively regulates YAP signalling. In subcellular fractionation experiments using cell lysates, PPP1R12A in control cells was recovered exclusively in the microsomal fraction. In contrast, a fraction of PPP1R12A in ATP8A1-depleted cells was recovered in the cytosolic fraction. Cohort data available from the Cancer Genome Atlas showed that high expression of PPP1R12A, PP1B encoding the catalytic subunit of the myosin phosphatase complex, or ATP8A1 correlated with poor prognosis in breast cancer patients. These results suggest that the “ATP8A1-PS-YAP phosphatase” axis in REs facilitates YAP activation and thus cell proliferation.

The Hippo pathway is evolutionarily conserved from Drosophila to mammals, and plays an essential role in regulating cell proliferation, organ size, and tissue homeostasis 1 .It integrates various extracellular and intracellular cues, including cell-cell contact, mechanical force, and energy status.It consists of the large tumor suppressor LATS1 and LATS2 (hereafter LATS1/2) kinase module and the YAP transcription module 2 .The activity of YAP depends on its phosphorylation status.Unphosphorylated YAP translocates from the cytosol to the nucleus where it binds to TEAD family transcription factors, and induces genes that promote cell proliferation, migration, and survival 3 .In contrast, phosphorylated YAP, generated by active LATS1/2, is sequestered in the cytoplasm and degraded 4,5 .Not surprisingly, dysregulation of the Hippo pathway has been implicated in tumorigenesis in several types of cancers 6 .
Phospholipids are essential components of cellular membranes.PS is the most abundant anionic phospholipid, accounting for up to 10% of the total phospholipids in cells 7 .PS in the plasma membrane (PM) plays key roles in various phenomena such as the coagulation cascade, clearance of apoptotic cells, and recruitment of signalling molecules 8 .PS is also highly enriched in the cytosolic leaflet of the RE membrane through the function of PS-flippase ATP8A1 9 , and regulates endosomal membrane trafficking by recruiting PS-binding proteins, such as evectin-2 10 and EHD1 9 .Besides its role in membrane trafficking, PS in the RE membrane plays a critical role in YAP signalling 11,12 .Knockdown of ATP8A1 increased the levels of phosphorylated YAP and phosphorylated LATS1 (the active form of LATS1), and abolished the presence of YAP in the nucleus 11 .Furthermore, enhanced

PPP1R12A facilitates dephosphorylation of YAP
We next asked whether PPP1R12A was critical to YAP activation in MDA-MB-231 cells.Knockdown of PPP1R12A increased the inactive form of YAP (the phosphorylated YAP at Ser 127 4 ) (Fig. 2a and Supplementary Fig. S1), and significantly reduced the nuclear localization of YAP (Fig. 2b).PPP1R12A knockdown also reduced mRNA expression of CTGF, a YAP-regulated gene involved in cell proliferation (Fig. 2c).These results suggested that PPP1R12A functioned to activate YAP, in line with the notion that PPP1R12A was required for YAP-dependent proliferation of MDA-MB-231 cells (Fig. 1b and c).Although the phosphorylated YAP increased, the active form of LATS1 (the phosphorylated LATS1 at Ser 909 21 ) was not affected in PPP1R12A-knockdown cells (Fig. 2a).Thus, PPP1R12A may directly dephosphorylate YAP but not LATS1.

Endogenous PPP1R12A localizes to REs in MDA-MB-231 cells
Given that PPP1R12A was identified as a PS-proximity protein candidate, PPP1R12A may localize to REs through its binding to PS and mediate dephosphorylation of YAP at REs.Several reports suggested that PPP1R12A resided at various subcellular locations, including the cytosol, nucleus, plasma membrane ruffles, and microfilaments [22][23][24][25] , however, its localization in MDA-MB-231 cells has not been examined.
We found that endogenous PPP1R12A localized to perinuclear regions and some bundle-like structures in the cytoplasm (Fig. 3a).These immunoreactive signals were not observed in PPP1R12A-knockdown cells, confirming the specificity of the staining.PPP1R12A co-localized well with the RE proteins, transferrin receptor (TfnR) and Rab11, but not with other organelle proteins [GM130 (Golgi), LAMP1 (lysosomes), and EEA1 (early endosomes)] (Fig. 3b and c, and Supplementary Fig. S2).These results demonstrated that PPP1R12A localized to REs in MDA-MB-231 cells.PPP1R12A also co-localized with YAP in the cytoplasmic bundles (Fig. 3d).Together with the RE localization of PPP1R12A, these data suggested that PPP1R12A interacted with YAP on the RE membrane.

ATP8A1/PS is involved in the recruitment of PPP1R12A to membranes
Purified recombinant PPP1R12A has been shown to bind anionic phospholipids such as PS, through its C-terminal region (Ser 667-Ile 1004 of chicken PPP1R12A, corresponding to Ser 668-Lys 1030 of human PPP1R12A) 26 .PP1B, the catalytic subunit of the phosphatase complex, bound to the N-terminal region of PPP1R12A 27,28 (Fig. 4a).We expressed either the N-or C-terminal half of human PPP1R12A with a myc-tag in MDA-MB-231 cells and examined their localizations.The N-terminal region (1-667) accumulated exclusively in the nucleus, whereas the C-terminal region co-localized with TfnR as endogenous PPP1R12A (Fig. 4b).Thus, the C-terminal region was essential for PPP1R12A localization to REs.
The ability of the C-terminal region to bind anionic phospholipids in vitro may be relevant to PPP1R12A localization to REs.Therefore, we examined if knockdown of ATP8A1, an RE-localized PS flippase, affected the www.nature.com/scientificreports/membrane association of PPP1R12A by cell fractionation.In control cells, PPP1R12A was recovered exclusively in the microsomal fraction after ultracentrifugation of the cell lysate, indicating that PPP1R12A was associated with membranes (Fig. 4c and Supplementary Fig. S3).PPP1R12A could not be fully recovered after ultracentrifugation even in the presence of protease inhibitors, indicating that PPP1R12A was unstable in the cell lysate.Nonetheless, in ATP8A1-knockdown cells, the amount of PPP1R12A increased in the cytosolic fraction and decreased in the microsomal fraction.These results suggested that membrane association of PPP1R12A was in part regulated by the levels of PS in the RE membrane.

Correlation of high ATP8A1/PPP1R12A/PP1B expression and poor prognosis in breast cancer patients
High YAP levels correlate with poor prognosis in triple-negative or estrogen receptor-negative breast cancer patients 29,30 .Therefore, we sought to ask whether ATP8A1, PPP1R12A, or PP1B, which we suggested as www.nature.com/scientificreports/YAP-activating factors in the previous and the present study, was related to the prognosis of breast cancer patients.Kaplan-Meier survival curves from TCGA cohort data indicated that high expression of ATP8A1, PPP1R12A, or PP1B correlated with shorter overall survival periods for breast cancer patients (Fig. 5a).In contrast to PPP1R12A and PP1B, the other phosphatases (PP1A, PP1C, PPP1R8, PPP1R9A, PPP1R10, PPM1G, PPP2R1A, and PPP2R5D), which were found not to be critical to the proliferation of MDA-MB-231 cells, did not show significant correlations between their high expression levels and shorter survival periods.

Discussion
In the present study, we showed that PPP1R12A localized to REs (Fig. 3b) and activated YAP in MDA-MB-231 cells (Fig. 2).In contrast, in other cell types, PPP1R12A was shown to inactivate YAP through the activation (or dephosphoryaltion) of Merlin (also known as NF2), the activator of the Hippo pathway 31,32 .The discrepancy may be explained by the fact that MDA-MB-231 cells are homozygous null mutants for Merlin 33 , so that PPP1R12A is unable to activate the Hippo pathway.Of note, Merlin is deleted or mutated in several types of cancers 34 , and thus we suggest that PPP1R12A promotes cell proliferation by acting predominantly on YAP in such Merlinnegative cancer cells.Incubation of ovarian cancer cell lines with platelets induced the association of PPP1R12A with YAP, coincidentally with YAP dephosphorylation and cellular resistance to anoikis, a type of apoptosis induced by loss of cell adhesion or inappropriate cell adhesion 18 .These observations, together with the present results (Fig. 2a-c), emphasized the role of PPP1R12A in facilitating YAP activation through YAP dephosphorylation.Intriguingly, another Ser/Thr phosphatase PPM1A was recently shown to directly dephosphorylate YAP at Ser 127 35 .Whether there is an interplay between PPP1R12A and PPM1A in the process of YAP activation, remains to be elucidated.
We previously showed that knockdown of ATP8A1 increased the phosphorylation of both YAP and LATS1 11 , contrasting with the present result that knockdown of PPP1R12A increased the phosphorylation of YAP, but not of LATS1 (Fig. 2a).These data suggested that PS in the RE membrane did not only facilitate YAP dephosphorylation by recruiting PPP1R12A to REs, but also regulated LATS1 dephosphorylation by recruiting other unidentified phosphatases.Several PS-proximity phosphatases (PP1A, PP1C, PPM1G, PPP1R8, PPP1R9A, and PPP1R37), the individual knockdown of which did not show any proliferation defect (Fig. 1b), are still such candidates: they may redundantly contribute to LATS1 dephosphorylation.
Rab11, a small GTPase localized to REs, has been shown to transport YAP to epithelial junctions, thereby suppressing YAP translocation into the nucleus and subsequent YAP-dependent transcription 36,37 .A recent study using alveolar type 2 epithelial cells, showed that adaptor protein complex 3 (AP-3)-mediated membrane traffic constitutively transported ATP8A1 out of REs to the lamellar bodies, which resulted in YAP inactivation 13 .Conversely, genetic AP-3 loss in alveolar type 2 epithelial cells and the pearl mouse model caused ATP8A1 retention in REs and concomitant accumulation of PS in the RE membrane and YAP activation, which may underlie progressive lung fibrosis associated with Hermansky-Pudlak syndrome type 2. Therefore, REs can regulate YAP activity, either negatively by membrane traffic out of REs or positively by dephosphorylation of YAP with the RE-localized PS-sensitive phosphatase (Fig. 5b).The balance of these two counteracting factors on REs may determine the phosphorylation/dephosphorylation status of YAP.
Cohort data available from TCGA showed that high expression of PPP1R12A, PP1B, or ATP8A1 correlated with poor prognosis in breast cancer patients (Fig. 5a).Recent studies have shown that high YAP expression correlates with shorter survival periods in triple-negative or estrogen receptor-negative breast cancer patients 29,30 .Combined with the present results using MDA-MB-231 cells, the active "ATP8A1-PS-YAP phosphatase (PPP1R12A/PP1B)" axis may account for the poor prognosis of breast cancer patients through YAP activation.

Figure 1 .
Figure 1.PPP1R12A is essential for the YAP-dependent proliferation of breast cancer cells.(a) Schematics of biotinylation of proteins proximal to PS with BirA*-2xPH (the PS biosensor).The list of phosphatases identified by the BioID method using BirA*-2xPH 11 is shown.(b) MDA-MB-231 cells were treated with the indicated siRNA for 48 h and then replated.After 24 h, the cells were transfected with the same siRNA again and further incubated for 96 h.The cell number was counted, normalized to the mean value of control siRNA (siCtrl)treated cells, and shown in a bar graph.(c) MDA-MB-231 cells were treated with the indicated siRNA, and the cell number was counted as in (b).(d-f) MDA-MD-231 cells were treated with the indicated siRNA as in (b), and the mRNA levels of PPP1R12A (d), LATS1 (e), and LATS2 (f) were determined by qRT-PCR and shown in a bar graph.GAPDH was used as an internal control.(g) MCF-7 cells were treated with the indicated siRNA as in (b), and the cell number was counted, normalized to the mean value of siCtrl-treated cells, and shown in a bar graph.(h) MCF-7 cells were treated with the indicated siRNA as in (b), and the mRNA level of PPP1R12A was determined by qRT-PCR.GAPDH was used as an internal control.Data represent the mean ± s.d.from duplicate or triplicate experiments with data points plotted.n.s., not significant; *p < 0.05; **p < 0.01 vs. siCtrl, one-way ANOVA with Tukey-Kramer post hoc test.

Figure 2 .
Figure 2. PPP1R12A knockdown increases phosphorylated YAP and reduces the nuclear localization of YAP.(a) MDA-MB-231 cells were treated with the indicated siRNA for 48 h and then replated.After 24 h, the cells were transfected with the same siRNA again and further incubated for 48 h.Cell lysates were prepared and analyzed by western blot.(b) Cells were treated as in (a), fixed, permeabilized, and stained for YAP.Nuclei were stained with DAPI.Scale bars, 10 µm.YAP nuclear enrichment values were determined by calculating the mean nuclear YAP intensity ratio to the mean cellular YAP intensity in individual cells.Data from the indicated number (n) of cells from two independent experiments were shown in box-whisker plots.(c) MDA-MB-231 cells were treated with the indicated siRNA for 48 h and then replated.After 24 h, the cells were transfected with the same siRNA again and further incubated for 72 h.The mRNA level of CTGF was determined by qRT-PCR.GAPDH was used as an internal control.Data are representative of two independent experiments performed in triplicate.The mean values ± s.d. were shown in a bar graph with data points plotted.*p < 0.05; **p < 0.01 vs. siCtrl, one-way ANOVA with Tukey-Kramer post hoc test.The uncropped blots in (a) are shown in Supplementary Fig. S1.

Figure 3 .
Figure 3. Endogenous PPP1R12A localizes to REs in MDA-MB-231 cells.(a) Cells were treated with the indicated siRNA for 48 h and then replated.After 24 h, the cells were transfected with the same siRNA again and further incubated for 48 h.Cells were then fixed, permeabilized, and stained for PPP1R12A.Nuclei were stained with DAPI.(b) Cells were fixed, permeabilized, and co-stained for PPP1R12A and the indicated organelle markers.Nuclei were stained with DAPI.(c) The Pearson's coefficient between PPP1R12A and the indicated organelle marker is shown.Data from two independent experiments and the indicated number (n) of cells were shown in box-whisker plots.(d) Cells were fixed, permeabilized, and co-stained for PPP1R12A and YAP.Nuclei were stained with DAPI.Arrowheads indicate the co-localization of PPP1R12A and YAP in bundle-like structures.(b, d) Magnified images of the perinuclear areas were shown in the right panel.Scale bars, 10 µm.Representative micrographs in (b) are also shown in Supplementary Fig. S2.

Figure 4 .
Figure 4. ATP8A1/PS is involved in the recruitment of PPP1R12A to membranes.(a) Structure of PPP1R12A.The N-terminal region (1-667 aa) binds to PP1B.The C-terminal region (668-1030 aa) has an affinity to anionic lipids, such as PS.(b) Myc-tagged PPP1R12A truncation mutant (1-667 aa or 668-1030 aa) was transiently expressed in MDA-MB-231 cells.Cells were then fixed, permeabilized, and stained for Myc-tag and TfnR (an RE marker protein).Nuclei were stained with DAPI.Magnified images of the perinuclear areas were shown in the right panel.Scale bars, 10 µm.(c) MDA-MB-231 cells were treated with the indicated siRNA for 72 h.The post-nuclear supernatant (PNS) was then spun at 100,000 × g for 1 h at 4 °C.The resultant supernatant (sup), pellet (ppt), and the PNS were subjected to western blot analysis.Data are representative of two independent experiments.The data from the other experiment are shown in Supplementary Fig. S3.The uncropped blots are shown in Supplementary Fig. S1.