Tumor Suppressor LATS1 Is a Negative Regulator of Oncogene YAP*

LATS (large tumor suppressor) or warts is a Ser/Thr kinase that belongs to the Ndr/LATS subfamily of AGC (protein kinase A/PKG/PKC) kinases. It is a tumor suppressor gene originally isolated from Drosophila and recently isolated from mice and humans. Drosophila or mice mutant for LATS develop tumors in various tissues. Recent studies in Drosophila demonstrate that LATS is a central player of an emerging tumor suppressor pathway called the Hippo-LATS/Warts pathway that suppresses tumor growth by regulating cell proliferation, cell growth, and cell death. Although tremendous progress has been made toward understanding the roles of LATS in tumorigenesis, the kinase substrates of LATS or downstream target proteins mediating LATS function remain largely unknown. In this study, we have provided convincing evidence that the LATS1 tumor suppressor can bind to and phosphorylate transcription regulator and oncogene YAP in vitro and in vivo. We have also identified HX(R/H/K)XX(S/T) as the consensus phosphorylation sequence for LATS/Ndr kinase substrates. Significantly, we have discovered that LATS1 inactivates YAP oncogenic function by suppressing its transcription regulation of cellular genes via sequestration of YAP in the cytoplasm after phosphorylation of YAP. Finally, by using microarray analysis, we have also identified many oncogenes or tumor suppressor genes up-regulated or down-regulated by YAP. These research findings will have profound impacts on our understanding of the molecular mechanism of the LATS tumor suppressor and the emerging Hippo-LATS/Warts pathway.

In this study, we have discovered that LATS1 can interact with and phosphorylate YAP both in vitro and in vivo. Significantly, we have provided convincing evidence that LATS1 can negatively regulate transcription regulation and transformation functions of YAP by inhibiting its nuclear translocation via phosphorylation of multiple sites in YAP. By using peptide kinase assays, we have also identified the consensus phosphorylation sequence for LATS1 and Ndr1 kinase substrates. Our findings clearly indicate that LATS/Ndr has substrate specificity similar to those of other AGC kinase members.
Cell Culture, Transfection, and Immunofluorescence-Cell culture, cell transfection, and immunofluorescence staining are as described (8). Images were obtained with a Nikon Eclipse TE-2000U Inverted Fluorescent Microscope.
Fusion Protein Production and GST Pull-down Assay-GST fusion proteins were produced and purified as described (8). For the GST pull-down assay, about 50 -200 g of protein lysate expressing LATS1/2-FLAG, Ndr1/2-FLAG, or YAP-HA was mixed with 10 g of GST (control) or LATS1 or YAP GST fusion proteins on beads and incubated at 4°C with rotating for 2 h. The beads were washed four times with 1% Nonidet P-40 lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1.0% Nonidet P-40, and 1ϫ proteinase inhibitor), resuspended in 2ϫ SDS sample buffer, boiled, and centrifuged, and the supernatants were subjected to SDS-PAGE and Western blot analysis using either ␣-FLAG or ␣-HA antibody.
Lentivirus Production, Purification, Titration, and Infection-One day before transfection, 5 ϫ 10 6 293T cells were plated on a 150-mm plate coated with 0.1 mg/ml poly-L-lysine and incubated at 37°C overnight. Then 7.5 g of LATS1-Myc-WPI or YAP-HA-WPI or Mst2-FLAG-WPI transfer vector were mixed with 5.6 g of PAX (packing) and 1.9 g of MD2G (envelope) plasmids and 37.5 l of Lipofectamine 2000. After incubation for 20 min at room temperature, the mixture was added dropwise into each plate. The medium was replaced with 18 ml of OPTI-MEM I medium (serum-free) containing 10 mM sodium butyrate (demethylation of plasmids to increase gene expression) 20 -24 h after transfection. Two days after transfection, the media containing lentivirus were collected, passed through a 0.45-m filter, and concentrated using a Centricon-20 ultrafiltration column (Millipore Corp.). For virus titration, 4 ϫ 10 4 HeLa cells were plated into each well of a 24-well plate. One day after plating, 8 g/ml Polybrene and a series dilution of lentivirus were added into each well. GFP-positive cells (cells containing lentivirus) and GFP-positive plus GFP-negative cells were then counted under a fluorescent and light microscope, respectively. The percentage of GFP-positive cells was calculated. The virus titer was calculated according to the following formula: transduction units/ml ϭ (cell number at the time of infection) ϫ (percentage of GFP-positive cells) ϫ (dilution factor). For infection of MCF10A or HeLa cells with lentivirus, 4 ϫ 10 5 cells were plated into each well of a 6-well plate 1 day prior to infection. One day after incubation at 37°C, cell numbers were counted, and YAP, Mst2, or LATS1 lentivirus was then added together with 8 g/ml of Polybrene into each well at a multiplicity of infection of 2. Two days after infection, the cells were either collected for protein analysis or microarray analysis or expanded to be used for soft agar assays (see below).
RNA Interference Knockdown of LATS-A TriFECTa kit containing a Dicer substrate siRNA duplex targeting human LATS1 or LATS2 mRNA and a negative control siRNA with scrambled sequence absent in the human genome were purchased from Integrated DNA Technologies (Coralville, IA). Transfection of HeLa cells with 10 nM siRNA was performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Four days post-transfection, protein was extracted, and knockdown of LATS1 and LATS2 was confirmed by Western blot. For immunostaining of YAP after LATS knockdown, HeLa cells were replated into a well of a 24-well plate containing a coverslip coated with fibronectin. The cells were subject to immunostaining using rabbit ␣-YAP polyclonal antibody after they were cultured at 37°C overnight.
RNA Isolation, Microarray Analysis and Real Time Quantitative Reverse Transcription (qRT)-PCR-MCF10A cells grown in a 6-well plate were infected with lentivirus expressing vector (control) or YAP-HA with a multiplicity of infection of 2. Duplicate MCF10A cells in different wells of a 6-well plate (biological replicate) were infected with YAP lentivirus. Two days after infection, total RNAs were extracted with TRIzol reagent (Invitrogen) according to a standard protocol. To clear up residual genomic DNA, total RNA was further cleaned using an RNeasy Mini Kit (Qiagen). The quality of RNA is examined by electrophoresis on 1.5% agarose gel and Agilent 2100 Bioanalyzer (version B/02.02) analysis.
For microarray analysis, 250 ng of total RNA were labeled with either Cy3-CTP or Cy5-CTP (PerkinElmer Life Sciences) and subsequently amplified with the Low RNA Input Linear Amplification kit (Agilent) according to the manufacturer's instructions. The RNAs were labeled, amplified, and used for microarrays as the following: Array 1, Control/Cy3-CTP, YAP-1/Cy5-CTP; Array 2, Control/Cy3-CTP, YAP-2/Cy5-CTP (biological replicate); Array 3, YAP-1/Cy3-CTP, Control/Cy5-CTP (flip color). About 825 ng of each Cy3-or Cy5-labeled complementary RNA were mixed and hybridized to a 4ϫ44K whole human genome oligonucleotide (60-mer) array (Agilent) at 65°C for 17 h in a microarray rotating oven (SciGene). The slides were washed and subsequently scanned on an Agilent Microarray Scanner G2565BA. Scanned images were analyzed with the Agilent Feature Extraction Program (version 9.1). YAP up-regulated and down-regulated genes were selected according to the following selection criteria. Only those genes whose transcription had changed more than 2-fold (log ratio Ն0.3 or ՅϪ0.3) in 2 of 3 microarrays with a p value of Ͻ0.0001 were further evaluated. GeneOntology analysis in GeneTools (available on the World Wide Web) and a Pubmed gene search were used to identify the functions of each differentially expressed gene.
The SuperScript III Platinum SYBR Green One-step qRT-PCR kit (Invitrogen) was used for real time qRT-PCR to quantify the level of each differentially expressed gene identified from microarray analysis. In brief, triplicates of 0.2 g of total RNA extracted from MCF10A cells infected with lentivirus expressing vector (MCF10A-control) or YAP (MCF10A-YAP) were mixed with SuperScript III RT/Platinum, 2ϫ SYBR Green reaction mix, and gene-specific forward and reverse primers in 25 l. RT-PCR was run at 1 cycle of 50°C for 5 min, 95°C for 5 min, and 40 cycles of 95°C for 15 s, 60°C for 30 s, and 40°C for 1 min using an ABI PRISM 7700 sequence detection system. 18 S rRNA was used as internal control of RNA amounts in each sample. The mRNA level of each gene in MCF10A-YAP cells relative to that in MCF10A-control cells was calculated as the following formula: 2 Ϫ(YAP⌬Ct Ϫ Control⌬Ct) , where ⌬Ct ϭ cellular gene average Ct Ϫ rRNA average Ct. The mean and S.D. were calculated from Ct values of triplicate real time RT-PCRs for each RNA sample.
Kinase Assays-COS-7 cells transfected with wild-type (WT) or kinase-dead (KD) LATS-FLAG or Ndr-FLAG alone or together with Mst2-Myc were lysed in 1% Nonidet P-40 lysis buffer supplement with 1 mM DTT and 1ϫ phosphatase inhibitor (Sigma). For an in vitro immunoprecipitation-kinase assay, 100 g of protein lysate was mixed with 2 g of ␣-FLAG (M2) monoclonal antibody together with Protein G beads and incubated at 4°C for 3 h. The beads were washed twice with 1% Nonidet P-40 lysis buffer, 1 mM DTT; once for 10 min with 1% Nonidet P-40 lysis buffer, 500 mM NaCl, 1 mM DTT at 4°C (rotating); and twice with 20 mM Tris-HCl (pH 7.4), 1 mM DTT, 1ϫ phosphatase inhibitor. The washed beads were then mixed with 2 g of YAP-GST substrates in a kinase buffer (20 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , 5 mM MnCl 2 , 1ϫ phosphatase inhibitor, 2 mM DTT, 10 M ATP, 5 Ci of [␥-32 P]ATP) and incubated at 30°C for 30 min. The reaction was stopped by adding 7 l of 5ϫ SDS sample dye, boiled at 100°C for 5 min, and subjected to SDS-PAGE. After electrophoresis, the proteins were transferred to a nitrocellulose membrane, stained with 1% Ponceau S for visualization of fusion protein substrates, and exposed to autoradiograph film for 0.5-2 h. To test phosphorylation of YAP by LATS1 in vivo, YAP-HA was transfected alone or together with LATS1-FLAG/Mst2-Myc into COS-7 cells. The cells were lysed as described above. Ten g of cell lysate was untreated or treated with 1 unit of calf intestinal alkaline phosphatase at 37°C for 1 h and subjected to Western blot analysis using ␣-HA antibody.
For in vitro peptide kinase assays, the precipitated and washed beads were mixed with a 3 mM concentration of each peptide in a kinase buffer and incubated at 30°C for 30 min. The reaction mix was spotted on a P81 circle filter paper (Whatman) and washed four times with 1% phosphoric acid, dried, and subjected to cpm radioactivity counting. P81 spotted with reaction mix without peptide was used as background control. The relative kinase activity of specific peptide is calculated as the percentage of cpm for this peptide relative to the cpm for wildtype YAP-S127 peptide.
Protein Fractionation and Soft Agar Assays-Protein fractionation of cytoplasmic and nuclear proteins and soft agar assays were as described (8,32). Colony numbers were counted using the colony count program in the Gel Doc EQ system (Bio-Rad).

Interaction of LATS1 and YAP in Vivo and in Vitro-
We first used co-immunoprecipitation assays to examine whether LATS1 interacted with YAP in vivo. FLAG-tagged LATS1 and HA-tagged YAP expression vectors were transfected either alone or together into COS-7 cells. When cell lysates were precipitated with ␣-HA antibody, LATS1-FLAG was found in the YAP-HA immune complex (Fig. 1A). To eliminate the possibility that LATS1 interaction with YAP is due to their high level of expression in the cells, we also examined the interaction of endogenous LATS1 with YAP in HeLa cervical carcinoma cells expressing high levels of both LATS1 and YAP. Interestingly, endogenous LATS1 was detected in the immune complex precipitated with ␣-YAP antibody (Fig. 1B).
Next, we used a GST pull-down assay to examine whether LATS1 directly interacted with YAP in vitro. As expected, LATS1-FLAG was able to bind strongly to YAP-GST in vitro (Fig. 1C). We also examined whether other members of the Ndr kinase family in mammals, LATS2, Ndr1, and Ndr2, interacted with YAP. Like LATS1, LATS2-FLAG was able to bind to YAP-GST, whereas both Ndr1 and Ndr2 were unable to interact with YAP-GST (Fig. 1C), suggesting that LATS1 and LATS2 are the only two members of the Ndr kinase family in mammals that interact with YAP.
Since LATS and Ndr are conserved only in their C-terminal kinase domain, this suggests that LATS may interact with YAP with its N-terminal domain. To confirm this possibility and further understand the molecular mechanism of LATS-YAP interaction, we attempted to map the domains in both LATS and YAP responsible for their interactions. YAP belongs to Group I of the WW domain proteins that bind their partners through their WW domains to PPXY, PPRXXP, or PPPPP motifs in their binding partner (33). LATS1 contains two PPXY ( 373 PPXY 376 and 556 PPXY 559 ) motifs and two PPPPP motifs ( 236 PPPPP 240 and 247 PPPPP 251 ) within a P-stretch region ( Fig. 2A). To examine which motif is responsible for its binding to YAP, we first performed a GST pull-down assay using cell lysate expressing YAP-HA and GST fusion proteins purified from bacteria expressing a series of LATS1 deletions. Only the region between aa 526 and 655 in LATS1, which is immediately upstream of the kinase domain and covers the PPPY 559 motif, interacted with YAP (Fig. 2B). Mutation of Tyr to Phe at aa 559 in LATS526 -655 completely abolished its binding to YAP (Fig.  2C), indicating that the PPPY 559 motif in LATS1 is responsible for its interaction with YAP.
To map the regions in YAP responsible for its interacting with LATS1, we first performed a GST pull-down assay using cell lysate expressing LATS1-FLAG and GST fusion proteins purified from bacteria expressing a series of YAP deletions. As expected, only the region (aa 150 -300) covering two WW domains in YAP bound to LATS1-FLAG (Fig. 2, A and D). To determine which WW domain in YAP is responsible for its binding to LATS1, we created YAP150 -300 mutants with mutations that disrupt single (⌬WW1 (W199A/P202A) or ⌬WW2 (W258A/P261A)) or double (⌬WW1-2 (W199A/P202A/W258A/ P261A)) WW domains. Only disruption of both WW domains in YAP abrogated its binding to LATS1 (Fig. 2, A and E). In conclusion, our results suggest that LATS1 binds to both WW domains of YAP through its PPPY 559 motif.
Phosphorylation of YAP by LATS1-We used an immunoprecipitation-kinase assay to examine whether YAP is a substrate of LATS1 kinase. LATS1-WT-FLAG or LATS1-KD-FLAG was transfected alone or together with Mst2-Myc into COS-7 cells. LATS1-WT-FLAG or LATS1-KD-FLAG was precipitated from cell lysates expressing these constructs and subject to kinase assay using YAP-GST as a substrate. Although precipitated LATS1-KD-FLAG has no activity toward YAP-GST in vitro, LATS1-WT-FLAG can phosphorylate YAP-GST with low activity (Fig. 3A). Since it has been shown recently that Mst2, the mammalian homolog of Drosophila Hippo, can phosphorylate and activate LATS1 kinase activity (34), we co-transfected LATS1 together with Mst2. As expected, co-transfection of Mst2-Myc together with LATS1-WT-FLAG rather than LATS1-KD-FLAG into COS-7 cells significantly increases the FIGURE 1. Interaction of LATS/Ndr with YAP. A, co-immunoprecipitation of LATS1 and YAP. LATS1-FLAG and YAP-HA were transfected alone or together into COS-7 cells. About 500 g of protein lysate were precipitated with 2 g of ␣-HA (Y11) polyclonal antibody. Precipitated proteins were subject to Western blot (WB) with ␣-FLAG (M2) monoclonal antibody. About 10 g of protein lysate was also subjected to Western blot with the indicated antibodies to examine whether LATS1-FLAG and YAP-HA are expressed in COS-7 cells. B, interaction of endogenous LATS1 and YAP in HeLa cells. About 500 g of HeLa cell lysate was incubated with rabbit IgG (control) or ␣-YAP polyclonal antibody. The precipitated YAP immunocomplexes were subject to Western blot with ␣-LATS1 (Y03) polyclonal antibody along with 50 g of HeLa cell lysate (1/10 input) precipitated with ␣-LATS1 antibody. C, interaction of YAP with LATS/Ndr in vitro. About 10 g of GST (control) or YAP-GST were incubated with 100 g of cell lysate from LATS1/2-overexpressing cells or 20 g of cell lysate from Ndr1/2overexpressing cells. GST fusion proteins were precipitated with glutathione-Sepharose beads and subject to Western blot with ␣-FLAG antibody along with one-tenth amounts of cell lysate used for GST pull-down assays (1/10 input). The membrane was stained with 1% Ponceau S to visualize the GST fusion proteins used in the GST pull-down assays. FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9 phosphorylation of YAP-GST by precipitated LATS1-FLAG in vitro, whereas Mst2 alone was unable to phosphorylate YAP-GST (Fig. 3A). To confirm whether LATS1 can also phosphorylate YAP in vivo, we transfected YAP-HA alone or together with LATS1-FLAG/Mst2-Myc. LATS1 was able to phosphorylate YAP and caused a band shift of YAP-HA. This band shift of YAP-HA is due to its phosphorylation by LATS1, because it was abolished after treatment of cell lysate with calf intestinal alkaline phosphatase (Fig. 3B). Therefore, YAP is a true substrate of LATS1 kinase.

Phosphorylation and Regulation of YAP by LATS1
Next, we wanted to know whether other members of Ndr kinases can also phosphorylate YAP. First, we transfected WT or KD LATS1/2-FLAG or Ndr1/2-FLAG together with Mst2-Myc into COS-7 cells. After precipitation by ␣-FLAG antibody, the in vitro kinase assay was performed using YAP-GST as substrates. Interestingly, all of the precipitated WT LATS1/2 or Ndr1/2 rather than their KD counterparts were able to phosphorylate YAP in vitro (Fig. 3C). We also examined whether LATS/Ndr phosphorylate YAP in vivo by co-transfection of YAP-HA together with WT or KD LATS-FLAG/Mst2-Myc or Ndr-FLAG/Mst2-Myc. Significantly, co-transfection of WT other than KD of both LATS1 and LATS2 together with YAP-HA causes phosphorylation of YAP-HA. On the other hand, co-transfection of YAP-HA with WT or KD Ndr1 or Ndr2 had no effect on YAP-HA (Fig. 3D), suggesting that YAP is the substrate of LATS rather than Ndr in vivo.
Identification of the LATS1 Phosphorylation Sites in YAP-To identify the sites in YAP phosphorylated by LATS1, we first performed in vitro kinase assays using precipitated WT or KD LATS1-FLAG as kinase and YAP GST fusion proteins purified from bacteria expressing different regions of YAP as substrates. All of the YAP deletion fusion proteins, including YAP1-150-GST, YAP150 -300-GST, and YAP300 -504-GST, were significantly phosphorylated by wild-type LATS1 (Fig. 4A), suggesting that there are at least three phosphorylation sites in YAP. To further identify the phosphorylation sites, we created YAP deletions and used these deletion fusion proteins as substrates for in vitro kinase assay. YAP1-50-GST and YAP200 -300-GST proteins were the only two fragments that were not phosphorylated by LATS1 (Fig. 4B). Since YAP1-100-GST and YAP150 -300-GST were phosphorylated by LATS1, phosphorylation sites may be located in fragments of aa 50 -100 and 150 -200. By closer examination of all of the potential Ser/Thr phosphorylation sites (three sites in aa 50 -100 and five sites in aa 150 -200), we identify a phosphorylation motif, HXRXXS, located at Ser 61 (fragment 50 -100) and Ser 164 (fragment 150 -200), that is present in both fragments (Fig. 4B). Further screening of this motif in YAP identified a total of five sites with this motif located at Ser 61 , Ser 109 , Ser 127 , Ser 164 , and Ser 397 (Fig. 4, B and C). These five phosphorylation sites can perfectly explain the results of all of the kinase assays (Fig. 4B, see asterisk at phosphorylation sites).
To confirm whether our hypothesis was true, we created YAP mutant fusion protein with single, double, or triple phosphorylation site mutations and used these proteins as substrates for LATS1 kinase. Mutations of all of the potential phosphorylation sites into alanines (S61A/S109A/S127A) rather than a single (S61A, S109A, or S127A) or double phosphorylation site (S109A/S127A) completely abolished the phosphorylation of YAP1-150-GST by LATS1. As expected, mutation of Ser 164 (S164A) or Ser 397 (S397A) in YAP150 -300 and YAP300 -504, respectively, also abolished their phosphorylation by LATS1, suggesting that all of the five potential phosphorylation sites in YAP can be phosphorylated by LATS1. To further confirm whether YAP does contain five phosphorylation sites, we also created full-length YAP fusion proteins with all five phosphorylation site mutations (5ϫSA). As expected, mutations of all of the phosphorylation sites (YAP-5ϫSA-GST) abolished the phosphorylation of YAP by LATS1 in vitro (Fig. 4E). This in vitro kinase assay result was further confirmed in vivo. Mutations of all phosphorylation sites (YAP-5ϫSA-HA) abrogated its band shift/phosphorylation caused by overexpression of LATS1/Mst2 (Fig. 4F). In summary, these results clearly demonstrate that LATS1 phosphorylates YAP through five sites with a phosphorylation motif HXRXXS.

Identification of Consensus Phosphorylation Sequence for LATS/Ndr
Kinase Substrate-Sequence alignment of human (hYAP), mouse (mYAP) and Drosophila (dYAP, same as Yokie) YAP showed that, like hYAP, mYAP contains all of the five phosphorylation sites identified in hYAP. However, dYAP/Yokie contains only two sites (Ser 111 and Ser 168 ) with the HXRXXS motif corresponding to Ser 109 and Ser 127 in hYAP. A third potential phosphorylation site (Ser 250 ) with the HXKXXS (Lys replaces Arg) motif, which aligned with the Ser 164 phosphorylation site in hYAP, was also identified in dYAP (Fig. 5A) (see results below). To identify the consensus phosphorylation sequence for LATS1 substrates, we used the sequence surrounding the Ser 127 phosphorylation site as a template and synthesized peptides with mutations at each conserved residue in the motif (Fig. 5B) to test their activity as the substrates of LATS1 kinase. Since His, Arg, and Lys all belong to the same group of positive charged amino acids and are often interchangeable in phosphorylation motifs, we first mutated His at the Ϫ5-position into Arg (H122R) or Lys (H122K). We also mutated His 122 into other amino acids, such as Asn (H122N) or Ala (H122A) to see whether they abolished their function as substrates (Fig. 5B). Significantly, like other mutations (H122N or H122A), mutation of His into Arg (H122R) or Lys (H122K) also completely eliminated its phosphorylation by LATS1 kinase (Fig. 5C), suggesting that the His residue in the motif is essential for the phosphorylation of LATS1 substrate by LATS1. On the other hand, mutation of Arg at the Ϫ3-position into Lys (R124K) or His (R124H) had  FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9 only a modest effect or no effect (Fig. 5C), whereas mutation of Arg into Ala (H125A) completely abolished its phosphorylation by LATS1. Since LATS1 is a Ser/Thr kinase, we also examined whether LATS1 can phosphorylate threonine residue using peptides with a mutation from Ser to Thr (S127T). The phosphorylation of S127T by LATS1 dropped to 33% compared with its wild-type counterpart with serine as the phosphorylation donor site. This suggests that LATS1 can phosphorylate both serine and threonine with a preference for serine. This is consistent with the previous publications indicating that most of Ser/Thr kinases have preference for serine for their phosphorylation (35). Since other Ndr/LATS family members, such as Ndr1 and Ndr2, were also able to phosphorylate YAP in vitro (Fig. 3C), we tested whether they had the same consensus phosphorylation motif. Consistently, the same phosphorylation pattern was obtained for Ndr1 (Fig. 5D). According to the above results, we can conclude that the consensus phosphorylation sequence for LATS/Ndr family kinase substrates is HX(R/H/ K)XX(S/T) (Fig. 5E). We did an alignment of LATS/Ndr consensus phosphorylation sequences with those of other members of the AGC kinase family. Surprisingly, all of the consensus sequences are very similar (Fig. 5E). Most of the AGC kinases except DMPK prefer Ser to Thr for phosphorylation. In addition, all of the consensus phosphorylation sequences for AGC kinase substrates contain an Arg at the Ϫ3-position but are variable (Arg, Lys, His, Leu/Ile, or X) at other positions. However, LATS/Ndr is the only kinase in the AGC kinase family that has His in its consensus phosphorylation sequence at the Ϫ5and Ϫ3-positions (Fig. 5E).

Phosphorylation and Regulation of YAP by LATS1
Phosphorylation of YAP by LATS1 Regulates Its Subcellular Localization-Previous studies indicate that phosphorylation of YAP may affect its subcellular localization (36). To test whether phosphorylation of YAP by LATS also changes its subcellular localization, we transfected YAP-HA or YAP-5ϫSA-HA alone or together with LATS1-FLAG/Mst2-Myc into COS-7 cells. Consistent with a previous report (37), wildtype YAP was distributed in both the cytoplasm and nucleus. Although overexpression of LATS1 alone can sufficiently sequester YAP in the cytoplasm with low efficiency (data not shown), co-transfection of LATS1 together with Mst2 completely sequestered YAP in the cytoplasm in most of the cells examined (Fig. 6A). On the other hand, YAP-5ϫSA, a YAP mutant without any LATS1 phosphorylation sites, was mostly in the nucleus and remains unchanged after co-expression of LATS1/Mst2 in the cells. These results strongly suggest that the phosphorylation of YAP by LATS1 inhibits its translocation into the nucleus. We also examined whether overexpression of LATS1 affects subcellular localization of endogenous YAP by infection of HeLa cells with lentivirus expressing vector or LATS1 and Mst2 by protein fractionation. Consistent with FIGURE 4. Mapping of the phosphorylation sites in YAP. A and B, phosphorylation of YAP deletion proteins by LATS1. A series of YAP fusion proteins with deletions were used as substrates for an in vitro kinase assay using LATS1-WT-FLAG activated by Mst2 or LATS1-KD-FLAG. Potential phosphorylation sites were shown in the schematic graph. Black bars in B indicate that these two fusion proteins were not phosphorylated by LATS1. C, alignment of the sequences of phosphorylation sites in YAP. Conserved residues are highlighted. D, phosphorylation of YAP mutants by LATS1. YAP fusion proteins with single (S61A, S109A, S127A, S164A), double (S109/127A), or triple (S61A/S109A/S127A) mutations were used as substrates for an in vitro kinase assay using activated LATS1 kinase. E, phosphorylation of YAP with phosphorylation site mutations by LATS1 in vitro. In vitro kinase assays were carried out using activated LATS1 kinase and WT and 5ϫSA (Ser residues in all of the five phosphorylation sites were mutated into Ala) YAP as substrates. F, phosphorylation of YAP with phosphorylation site mutations by LATS1 in vivo. YAP-HA or YAP-5ϫSA-HA was transfected alone or together with LATS1/Mst2 into COS-7 cells. About 10 g of protein lysates was used for Western blot with the indicated antibodies.
results obtained from the YAP overexpression experiment stated above, endogenous YAP was mostly localized in both cytoplasm and nucleus. However, overexpression of LATS1/ Mst2 caused exclusive localization of endogenous YAP in the cytoplasm. As cytoplasmic and nuclear protein fraction controls, ␣-tubulin and PCNA are localized in the cytoplasm and nucleus, respectively (Fig. 6B). We have also examined whether endogenous LATS affect subcellular localization of endogenous YAP. As expected, after knockdown of LATS (LATS1 and LATS2) in HeLa cells, the endogenous YAP originally localized in both cytoplasm and nucleus was redistributed exclusively into the nucleus (Fig. 6C). These results clearly demonstrate that phosphorylation of YAP by LATS1 inhibits its translocation to the nucleus, which leads to sequestration of YAP in the cytoplasm.
Identification of Cellular Genes Regulated by YAP by Microarrays-In Drosophila, Yokie/YAP was identified as an oncogene that promotes cell proliferation and inhibits apoptosis by up-regulation of cyclin E and dIAP transcription (15). Recently, mammalian YAP was also identified as an oncogene, and overexpression of YAP caused transformation of MCF10A immortalized mammary cells. However, mammalian YAP has no effect on the expression of both cyclin E and cIAP (25), suggesting that the downstream genes mediating YAP-induced transformation in human cells may be different from those in Drosophila. To identify the cellular genes up-regulated or down-regulated by YAP during transformation of mammary cells, we infected MCF10A cells with lentivirus expressing vector (MCF10A-control) or YAP-HA (MCF10A-YAP). GFP was also co-expressed by lentivirus vector and was used as a marker for infected cells. As shown by immunofluorescent images of infected cells, over 98% of MCF10A were infected by lentivirus (Fig. 7A). Interestingly, infection of MCF10A by YAP rather than vector caused a dramatic increase in YAP protein level and morphological changes (Fig. 7, A and B). Although MCF10Acontrol cells grow as epithelial-like cells, MCF10A-YAP cells lost cell-cell contact and displayed mesenchymal (fibroblast)like morphology, a phenotype called epithelial-to-mesenchymal transition (EMT) (Fig. 7A).
To identify the genes regulated by YAP, we exacted RNAs from MCF10A-control and MCF10A-YAP cells and performed an expression profile analysis using the Agilent 4ϫ44K whole human genome oligonucleotide (60-mer) array. In this microarray, many cellular genes with important biological functions have 3-10 probes (spots) on each array. After careful examination of signals from all of the probes for each gene and statistic analysis (t test) and GeneOntology analysis using GeneTool (see "Experimental Procedures" for details), 17 upregulated and 18 down-regulated genes associated with cell proliferation, cell death, cell migration, cell adhesion, and EMT FIGURE 5. Identification of consensus phosphorylation sequence for LATS substrates. A, alignment of YAP phosphorylation site sequences from different organisms. B, peptide sequences surrounding YAP-S127 and its mutants. The conserved residues in WT YAP-S127 and each mutated residue were in boldface type. The amino acid position for each residue in YAP is shown at the top. The phosphorylation site at Ser 127 is regarded as position 0 in the sequence. C, in vitro phosphorylation of peptide substrates by LATS1. In vitro kinase assays were carried out using immunoprecipitated, activated LATS1 and various peptide substrates in the presence of [␥-32 P]ATP. Circle P81 filter papers were used for phosphorylated peptide capture and radioactivity quantification. Phosphorylation activity by LATS1 is calculated as a percentage of cpm for each mutant peptide compared with cpm of WT peptide. Error bars, ϮS.D. of two independent kinase assays for each peptide substrate. D, in vitro phosphorylation of peptide substrates by Ndr1. The in vitro kinase assays were carried out as described for C except that Ndr1 was used as the kinase for the assay. E, sequence alignment of AGC family kinases. Conserved residues are highlighted. The phosphorylation site (S/T) was regarded as position 0. Note that only DMPK prefers Thr to Ser for phosphorylation. FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9 JOURNAL OF BIOLOGICAL CHEMISTRY 5503 have been identified and are listed in Table 1. We used qRT-PCR to confirm some of the important genes significantly changed after overexpression of YAP. About 70% of genes identified in microarray analysis as differentially expressed genes were also confirmed by qRT-PCR assays. As a positive control, YAP was 16-fold higher in MCF10A-YAP cells compared with MCF10A-control cells (Fig. 7C). For all of the up-regulated genes examined, integrin ␤2 (ITGB2) is the gene that was most significantly up-regulated (over 12-fold increase) by YAP. Many hormones or hormone-binding proteins, such as FGF1, IGFBP3, and PDGF␤, were also enhanced after overexpression of YAP (Fig. 7C). Surprisingly, although YAP was originally identified as a transcription co-activator, the transcription of some of the tumor suppressors, such as p57, RASSF4, prolactin, and BMP2, a member of the transforming growth factor ␤ family, were significantly (3-10-fold) down-regulated by YAP (Fig. 7C).

Suppression of YAP-induced Transcription Regulation and Cell
Transformation by LATS1-It has been shown that overexpression of Drosophila LATS inhibits cell proliferation and induces apoptosis by inhibiting Yokie trans-activation activity (15). Our experimental results demonstrate that LATS1 can phosphorylate YAP and inhibit its translocation to the nucleus, suggesting that phosphorylation of YAP by LATS1 may repress its transcription regulation of cellular genes in the nucleus. To test this possibility, we infected MCF10A-YAP cells with lentivirus expressing LATS1 and Mst2. Real time qRT-PCR analysis of mRNA levels for genes regulated by YAP, including FGF1, IGFBP3, integrin ␤2, PDGF␤, BMP2, p57, prolactin, and RASSF4, shows that overexpression of LATS1/Mst2 significantly inhibited YAP-induced transcription activation or repression of these cellular genes (Fig. 7C). We also tested whether LATS1 can inhibit YAP-induced MCF10A cell transformation using soft agar assays. As shown in Fig. 8, A and B, no colony was formed on soft agar for MCF10Acontrol cells, further confirming that MCF10A cells are nontumorigenic, immortalized mammary cells that do not display anchorage-independent growth. However, overexpression of YAP caused MCF10A cell transformation, as shown by the formation of many colonies on soft agar (Fig. 8, A and B). As expected, although infection of MCF10A-YAP cells with lentivirus expressing LATS1 alone significantly suppressed YAP-induced cell transformation, infection of MCF10A-YAP with lentiviruses expressing LATS1 together with Mst2 completely abolished cell transformation (Fig. 8, A and B). On the other hand, YAP-5ϫSA lacking LATS phosphorylation sites had higher transformation activity (more colonies formed) and became resistant to LATS1-or LATS1/Mst2-induced transformation inhibition (Fig. 8, A and B).

DISCUSSION
Although tremendous progress has been made toward our understanding the emerging Hippo-LATS/Warts tumor suppressor signaling pathway in the regulation of cell growth, cell proliferation, and cell death in Drosophila, little is known about this pathway in mammals. In this study, we have provided convincing evidence that human LATS1 is a negative regulator of FIGURE 6. Phosphorylation of YAP by LATS1 enhances its localization in the cytoplasm. A, immunofluorescent staining of YAP and LATS1. YAP-HA or YAP-5ϫSA-HA was transfected alone or together with LATS1-FLAG/Mst2-Myc into COS-7 cells. Transfected cells were fixed with 3.7% formaldehyde and permeabilized in 0.2% Triton X-100 in phosphate-buffered saline. For cells transfected with YAP-HA or YAP-5ϫSA-HA alone, the fixed cells were stained with Y11 ␣-HA polyclonal antibody followed by AF555 ␣-rabbit IgG (red). For cells transfected with YAP-HA or YAP-5ϫSA-HA together with LATS1-FLAG/Mst2-Myc, the fixed cells were co-stained with ␣-HA polyclonal antibody (YAP-HA or YAP-5ϫSA-HA) together with ␣-FLAG (LATS1) monoclonal antibody followed by AF555 ␣-rabbit IgG (red) and AF488 ␣-mouse IgG (green). B, overexpression of LATS1/Mst2 induces relocalization of endogenous YAP in cytoplasm. HeLa cells were infected with LATS1 and Mst2 lentiviruses. Cytoplasmic and nuclear protein fractions were subject to Western blot (WB) to examine the subcellular localization of endogenous YAP using ␣-YAP polyclonal antibody. ␣-Tubulin and PCNA were used as cytoplasm and nuclear markers, respectively, for the purity of the protein fractionation. The same protein lysates were also used to examine the expression of LATS1-Myc and Mst2-FLAG by Western blot using ␣-Myc (9E10) and ␣-FLAG (M2) antibodies, respectively. C, subcellular localization of endogenous YAP after knockdown of LATS by siRNA. HeLa cells were transfected with control-siRNA or LATS1-siRNA together with LATS2-siRNA (10 nM). To confirm LATS knockdown, 4 days after transfection, the proteins were extracted and subject to Western blot using rabbit ␣-LATS1 (Y03) or ␣-LATS2 (BL2213) polyclonal antibody. Cells from the same transfection were replated onto a coated coverslip and subject to immunofluorescent staining as described in A using rabbit ␣-YAP polyclonal antibody (H-125) followed by AF488 ␣-rabbit IgG.
YAP/Yokie. LATS1 together with Mst2 can significantly inhibit both YAP-induced transcription regulation and YAP-induced cell transformation of mammary immortalized cells. Consistent with our findings, YAP was also recently found to be inhibited by Drosophila LATS/Warts and human LATS2 through phosphorylation and to play important roles in the regulation of animal organ size and cell contact inhibition (38,39). Together with the finding that Mst1/2 can phosphorylate and activate LATS1/2 (34), we can conclude that part of the Hippo-LATS/Warts signaling pathway, Mst1/2 3 LATS1/2 3 YAP, is conserved in both Drosophila and mammals.
LATS is the central player in the Hippo-LATS/Warts signaling pathway. Previous studies have shown that both fly and mammalian LATSs are involved in tumorigenesis by regulating cell proliferation, cell growth, and apoptosis (6 -7, 14, 22). However, the signaling pathways by which LATS suppresses tumor growth are not fully understood. Since LATS is a kinase and the kinase activity is essential for its tumor suppressing and other functions (6, 7, 28 -31), identification of the kinase substrates of LATS is crucial to elucidate the biological and biochemical functions of LATS. So far, Yokie is the only protein identified as the substrate of LATS kinase in Drosophila (15). However, how LATS phosphorylation regulates Yokie activity and whether mammalian LATS1/2 and their homologs Ndr1/2 also phosphorylate the mammalian Yokie homolog, YAP, remains to be explored. In this study, we have shown for the first time that human LATS1 and LATS2 can physically interact with and phosphorylate YAP in vitro and in vivo, whereas LATS homologs Ndr1 and Ndr2 were unable to bind to YAP and could only phosphorylate YAP in vitro. We have also mapped the domains in LATS1 and YAP responsible for their interactions. We found that the PPXY 559 in LATS1 and both WW domains in YAP are essential for their interactions (Fig. 2, A-E). Sequence analysis shows that there is another PPXY motif located at aa 373-376 ( 373 PPPY 376 ) in LATS1 ( Fig. 2A). However, this motif is not conserved and cannot be found in other LATS, such as LATS2 or Drosophila LATS. Therefore, it is possible that the structure surrounding this motif prohibits it from binding to YAP. By examining the sequence of Ndr1/2, we could not find any PPXY motif. This explains why LATS rather than Ndr could bind to YAP. However, the kinase domain of LATS has high homology to that  FEBRUARY 29, 2008 • VOLUME 283 • NUMBER 9

Phosphorylation and Regulation of YAP by LATS1
of Ndr, suggesting that they may phosphorylate the same substrate. As expected, both Ndr1 and Ndr2 were able to phosphorylate YAP in vitro. Since Ndr was unable to interact with YAP, they were unable to phosphorylate YAP in vivo. Therefore, LATS1 and LATS2, but not Ndr1 and Ndr2, are the true kinases phosphorylating YAP.
Although Yokie was found to be phosphorylated by Drosophila LATS, the phosphorylation sites in Yokie have not been identified (15). Identification of phosphorylation sites in Yokie/ YAP is very important to elucidate their roles in the Hippo-LATS/Warts pathway. By making a series of deletions and mutations, we have identified five phosphorylation sites with the same HXRXXS motif in YAP. Importantly, three of these five phosphorylation sites are also conserved in Drosophila Yokie. By using a peptide kinase assay, we have also identified the consensus phosphorylation sequence, HX(R/H/K)XX(S/T), for both LATS1/2 and Ndr1/2 kinase substrates. Interestingly, Dbf2, a yeast homolog of LATS, was found to preferentially phosphorylate peptides containing the RXXS motif (40), which covers part of our consensus phosphorylation sequence. These findings strongly suggest that all of the members of LATS/Ndr kinase family in all organisms may have the same phosphorylation motifs. By searching for proteins having this motif using ScanSite (available on the World Wide Web), we were able to identify many potential substrates of LATS/Ndr with this motif. YAP/Yokie is the only kinase substrate identified so far for LATS. Identification of the consensus phosphorylation sequence will greatly facilitate our efforts to find more kinase substrates of Ndr/LATS kinases that may be important in mediating the emerging Hippo-LATS/Warts pathway and carrying out various biological functions, such as tumor suppression, organ size control, and neuron tilling, etc. Moreover, this consensus phosphorylation sequence is very similar to those of other AGC kinase family members (Fig. 5E). Due to the similarity of their consensus phosphorylation sequences, many members of the AGC kinase family can often phosphorylate the same substrate. For example, Akt, p70S6K, and RSK1 AGC kinases were able to phosphorylate Tau on the same phosphorylation site (41). However, LATS and Ndr kinases are unique and are the only two kinase kinases that have His at the Ϫ5and Ϫ3-positions of its consensus phosphorylation sequence. Specifically, His at the Ϫ5-position of the consensus phosphorylation motif is crucial for their substrates to be phosphorylated (Fig. 5, C and D). Ndr family kinases have a unique feature that is not found in any other AGC kinases. They have an insert of about 30 -60 amino acids located between subdomains VII and VIII of the kinase domain. This unique feature possibly distinguishes them from other AGC kinases and restricts their substrate specificity. This may also be why only one kinase substrate of LATS has been identified so far.
Interestingly, one of the five phosphorylation sites in YAP, Ser 127 , was also found to be phosphorylated by another AGC kinase, Akt1/PKB (35). Akt1/PKB has a consensus phosphorylation sequence, RXRXXS. Although more than 100 substrates have been identified for Akt/PKB (42), YAP is the only substrate that has the HXRXXS phosphorylation motif (35). We also tested the possibility of whether LATS/Ndr and Akt/PKB share their substrates, as was found in other AGC family kinases. Our peptide kinase assays clearly demonstrate that the His residue in the HXRXXS motif is crucial and that replacement of His with Arg, which mimics the Akt/PKB consensus phosphorylation motif, completely eliminated its phosphorylation by LATS1/Ndr1 (Fig. 5, C and D). Therefore, LATS/Ndr and Akt/ PKB do not share their substrates. Although our findings and the results shown in Drosophila genetic studies clearly demonstrated that YAP is the true substrate of LATS, it remains to be tested whether Akt is also the kinase phosphorylating YAP in our experimental system. Although Yokie/YAP was found to be phosphorylated by Drosophila LATS, the significance of this phosphorylation for Yokie/YAP function is still unknown. In this study, we provided convincing evidence that phosphorylation of YAP by LATS1 sequestered YAP in the cytoplasm, which would prohibit it from functioning as a transcription regulator in the nucleus.
Overexpression of LATS1/Mst2, which phosphorylates YAP, caused retention of YAP in the cytoplasm, whereas mutations of phosphorylation sites in YAP (YAP-5ϫSA) allowed YAP to be localized mostly in the nucleus and resistant to a LATS1-induced effect on its subcellular localization (Fig. 6A). Contrary to the effect of LATS1 on YAP, phosphorylation of Tyr 433 in YAP by Yes tyrosine kinase promotes its translocation to the nucleus to repress Runx2-induced transcription activation (43). Therefore, phosphorylation of YAP plays a very important role in regulating its function. The phosphorylated Ser 127 in YAP has been shown to be the binding site for 14-3-3 (35). Therefore, it is likely that phosphorylation of YAP by LATS1 increases its interaction with cytoplasmic protein 14-3-3. However, it remains to be explored whether phosphorylation of all of the phosphorylation sites of YAP by LATS1 is essential for the full inhibition of YAP function.
Both Drosophila and mammalian YAP/Yokie has been identified as an oncogene important for tumorigenesis (15,25,26). Overexpression of Yokie/YAP caused enhanced cell proliferation, reduced apoptosis, and tumor formation in Drosophila and mice and transformation of immortalized mammary cells (15,25). In addition, our studies and previous publications also showed that overexpression of YAP induced many cellular changes, such as disruption of actin cytoskeletal organization and cell-cell adhesion, an increase in cell migration, and the formation of EMT 3 (Fig. 7A) (25), all of which are the phenotypes for invasive tumor cells. Although up-regulation of cycle E and dIAP transcription has been proposed to be partially responsible for Drosophila Yokie/YAP-induced tumor formation, induction of cyclin E and dIAP1 cannot fully account for all of the phenotypes caused by overexpression of Yokie/YAP in Drosophila (11). In mammalian cells, it has been shown that cycle E and cIAP1 are not induced after overexpression of YAP (25). Although YAP was originally identified as a transcription co-activator and regulates many transcription factors, such as PEBP2, p73, P53-BP-2, and TEAD/TEF (36, 44 -46), the downstream genes mediating YAP-induced cell transformation and many cellular changes, 3 Y. Hao and X. Yang, unpublished results. such as EMT, remain to be identified. By using the whole human genome oligonucleotide microarray, we have successfully identified many genes regulated by YAP. These genes are involved in the regulation of cell proliferation, cell death, cytoskeletal organization, cell migration, and cell adhesion. FGF1 and PDGF␤ activated by YAP are growth hormones that induce increased cell proliferation, cell migration, and EMT (47,48), suggesting that YAP may up-regulate growth hormone transcription to induce cell transformation. Interestingly, we have also found that overexpression of YAP dramatically induces ␤2 integrin, a member of the integrin family. Integrins are proteins essential for cell adhesion, cell migration, and anchorage-dependent cell proliferation and cell survival (49,50). It has been shown that overexpression of ␤4 integrin caused anchorage-independent growth and transformation of rodent fibroblast, whereas loss of ␤4 integrin in breast carcinoma cells abrogated anchorage-independent growth and tumor formation in nude mice (51). Integrin ␤2 is a cell surface protein predominantly expressed in leukocyte and present in low levels in other tissues (49). Like other integrins, ␤2 integrin has also been shown to be involved in cell-cell and cell-matrix adhesions of leukocytes (49). However, whether ␤2 integrin is involved in tumorigenesis has not been reported so far. This is the first report that integrin ␤2 is significantly induced in mammary cells after cell transformation. Therefore, ␤2 integrin may be a very important downstream protein mediating YAP-induced anchorage-independent growth and transformation of mammary cells. It will be very interesting to examine whether overexpression of ␤2 integrin can also transform MCF10A mammary cells or down-regulation of integrin in MCF10A-YAP cells can block their anchorage-independent growth and transformation or other cellular changes.
Surprisingly, we have also identified many genes significantly suppressed by YAP. These genes are usually negative regulators of cell proliferation or tumor invasion. For example, p57 and RASSF4 have been shown to be tumor suppressors that inhibit tumor cell proliferation and tumor formation and induce apoptosis (52,53), whereas prolactin has been shown to suppress cell proliferation and tumor invasion and EMT (54). Therefore, down-regulation of these genes by YAP will induce increased cell proliferation and cell migration, reduced apoptosis, and formation of EMT. In fact, it has also been previously shown that overexpression of YAP inhibited Runx2-mediated activation of osteocalcin transcription (43). Therefore, YAP may also cause cell transformation and enhanced cell proliferation and tumor invasion by inhibiting transcription of several negative regulators of cell proliferation and cell migration and inducers of apoptosis. Most importantly, our studies strongly suggest that although both mammalian and Drosophila YAP/Yokie can induce tumor formation by regulating transcription of downstream genes, they may use different pathways to accomplish their function as oncogene. Further characterization of these downstream genes of YAP will have great impact on our understanding the signal transduction pathway of this emerging Hippo-LATS/Warts pathway during tumorigenesis.
In conclusion, we have provided convincing evidence that human tumor suppressor LATS is a negative regulator of oncogene YAP. LATS1 inactivates YAP oncogenic function by phos-phorylating YAP, which inhibits its translocation into the nucleus to regulate cellular genes important for cell proliferation, cell death, and cell migration. Most significantly, by using YAP phosphorylation sites as a template, we were able to identify HX(R/H/K)XX(S/T) as the consensus phosphorylation sequence of the LATS/Ndr kinase substrate. Given the fact that the Hippo-LATS/Warts tumor suppressor signaling pathway and Ndr/LATS kinases are very important not only in tumorigenesis but also in various biological processes, our research findings will have a profound impact on our understanding of the molecular mechanism and the signal transduction pathway underlining their biological functions.