STK25 directly activates LATS1/2 independent of MST/MAP4Ks

The Hippo pathway maintains tissue homeostasis by negatively regulating the oncogenic transcriptional co-activators YAP and TAZ. Though functional inactivation of the Hippo pathway is common in tumors, mutations in core pathway components are rare. Thus, understanding how tumor cells inactivate Hippo signaling remains a key unresolved question. Here, we identify the kinase STK25 as a novel activator of Hippo signaling. We demonstrate that loss of STK25 promotes YAP/TAZ activation and enhanced cellular proliferation, even under normally growth-suppressive conditions. We reveal that STK25 activates LATS via a previously unobserved mechanism, in which STK25 directly phosphorylates the LATS activation loop. This represents a new paradigm in Hippo activation and distinguishes STK25 from all other identified kinase activators of LATS. STK25 is significantly focally deleted across a wide spectrum of human cancers, suggesting STK25 loss may represent a common mechanism by which tumor cells functionally impair the Hippo tumor suppressor pathway.


Abstract
The Hippo pathway maintains tissue homeostasis by negatively regulating the oncogenic transcriptional co-activators YAP and TAZ. Though functional inactivation of the Hippo pathway is common in tumors, mutations in core pathway components are rare. Thus, understanding how tumor cells inactivate Hippo signaling remains a key unresolved question.
Here, we identify the kinase STK25 as a novel activator of Hippo signaling. We demonstrate that loss of STK25 promotes YAP/TAZ activation and enhanced cellular proliferation, even under normally growth-suppressive conditions. We reveal that STK25 activates LATS via a previously unobserved mechanism, in which STK25 directly phosphorylates the LATS activation loop. This represents a new paradigm in Hippo activation and distinguishes STK25 from all other identified kinase activators of LATS. STK25 is significantly focally deleted across a wide spectrum of human cancers, suggesting STK25 loss may represent a common mechanism by which tumor cells functionally impair the Hippo tumor suppressor pathway.
3 First discovered in Drosophila as a critical regulator of organ size, the Hippo tumor suppressor pathway has emerged as playing a major role in maintaining tissue homeostasis through the regulation of cell proliferation and survival (Zanconato et al. 2016). The key mediators of Hippo signaling are LATS1 and LATS2 (Large Tumor Suppressor) kinases, which function to negatively regulate the activity of the oncogenic transcriptional co-activators Yes-associated protein (YAP) and transcriptional co-activator with PDZ-binding motif (TAZ) (Zhang et al. 2009, Zhao et al. 2010. Upon stimulation of Hippo signaling, activated LATS kinases directly phosphorylate YAP/TAZ at conserved serine residues, which promote YAP/TAZ nuclear extrusion and subsequent degradation (Zhao et al. 2010). By contrast, in the absence of LATS activation, YAP/TAZ are free to translocate into the nucleus, where they bind to the TEAD/TEF family of transcription factors to promote the expression of numerous genes essential for cellular proliferation and survival (Wu et al. 2008, Zhang et al. 2008, Hong et al. 2005). Deregulation of LATS1/2 kinases, which leads to subsequent hyper-activation of YAP/TAZ, is sufficient to promote tumorigenesis in mouse models (Zhou et al. 2009, Nishio et al. 2015. Furthermore, amplification of YAP and/or TAZ has been found in a variety of human malignancies (Overholtzer et al. 2006, Fernandez-L et al. 2009).
Multiple signals lead to the activation of LATS kinases, including contact inhibition, cellular detachment, loss of actin cytoskeletal tension, serum deprivation, glucose starvation, signaling from G-protein coupled receptors, and cytokinesis failure (Zhao et  To identify novel upstream kinases that regulate LATS activity, we performed a focused RNAi screen to identify kinases that contribute to LATS phosphorylation and subsequent YAP phosphorylation. This approach identified STK25 as a novel upstream activator of the LATS kinases, whose loss significantly promotes YAP/TAZ activity. Mechanistically, we demonstrate that STK25 phosphorylates LATS at the activation loop motif in the absence of hydrophobic motif phosphorylation, which distinguishes it from all of the other known LATS activating kinases discovered to date.

A focused kinome screen identifies STK25 as a novel regulator of LATS-YAP phosphorylation
We performed a focused RNAi screen to identify kinases that are necessary for inducing YAP phosphorylation under conditions of activated Hippo signaling in the non-transformed IMR90 fibroblast cell line. We focused our kinome screen on members of the Sterile20 superfamily of kinases (which includes MST1/2), as members of this superfamily maintain structural similarities in spite of evolutionary divergence (Thompson and Sahai 2014 Fig. 1a-d). Depletion of STK25 significantly reduced levels of YAP S127 phosphorylation as assessed by quantitative immunoblotting relative to controls in this assay (Supplementary Fig. 1a-d). Importantly, this effect was reproduced in multiple cell lines, including both HEK293A and hTERT RPE-1 (Fig. 1a, Supplementary Fig. 1c, d). We performed phos-tag gel electrophoresis to assess overall levels of YAP phosphorylation and found that STK25-depletion led to significantly reduced levels of phosphorylated YAP, with consequent enrichment of unphosphorylated YAP, as compared to controls (Fig. 1b). Additionally, we found that TAZ, a mammalian paralog of YAP, was also enriched in an unphosphorylated status in STK25-depleted cells (Fig. 1c). We reproduced these results in cells treated with Latrunculin A, which is a fungal-derived actin-binding toxin that has a different mechanism of action than cytochalasin-class agents ( Supplementary Fig. 1e) (Morton et al. 2000).
We used multiple approaches to ensure that decreases in YAP phosphorylation following STK25 depletion via RNAi was not due to RNAi-induced off-target effects. First, we validated this finding with multiple distinct siRNA sequences targeting STK25 (8 total siRNAs, including three targeting the 3'UTR), and we observed that the degree of reduction in YAP phosphorylation strongly correlated with STK25 knockdown efficiency ( Supplementary Fig. 1f, g). Second, we used CRISPR-Cas9 to genetically knockout STK25 from HEK293A cells and found that STK25 KO clonal cells (generated with two different sgRNA sequences) similarly failed to induce YAP phosphorylation to the same extent as control cells expressing Cas9 and a 7 non-targeting sgRNA following DCB treatment (Fig. 1d). Finally, we demonstrated that overexpression of siRNA-resistant wild-type STK25, but not kinase-dead STK25 K49R , was sufficient to rescue the loss of YAP phosphorylation observed in our knockdown experiments ( Fig. 1e). Altogether, these data reveal that the kinase STK25 plays a previously unappreciated role in promoting YAP phosphorylation.

STK25 Depletion Promotes YAP Activation
We next analyzed if the decrease in YAP phosphorylation following STK25 depletion leads to a corresponding increase in the nuclear localization of active YAP. Depletion of STK25, either by RNAi-mediated knockdown or CRISPR-mediated gene knockout, led to a significant increase in nuclear YAP in multiple cell lines ( Fig. 2a- To confirm that nuclear localized YAP was active, we first assessed whether depletion of STK25 increased YAP/TAZ activity using a luciferase-based gene expression reporter assay. HEK293A cells were transfected with a reporter encoding a YAP/TAZ-responsive luciferase gene, in which eight TEAD-YAP binding sites are cloned into the promoter driving expression of firefly luciferase (Dupont et al. 2011). Using this approach, we found that depletion of STK25 resulted in a doubling of expression from the luciferase reporter, indicating that loss of STK25 promotes YAP/TAZ activity (Fig. 2i). We also assessed the expression of YAP-target genes in cells 8 depleted of STK25 and found that canonical YAP-target genes were significantly upregulated in IMR90 fibroblasts (Fig. 2g). Moreover, we found that YAP-target gene expression was also increased in a STK25 gene-dose dependent fashion in knockout MEFs for three well-established genes (Fig. 2h). Lastly, to assess gene expression in a comprehensive, unbiased fashion, we depleted STK25 from hTERT-RPE-1 cells and performed gene expression microarray analysis to obtain a list of genes that were significantly upregulated in cells lacking STK25 compared to controls. Gene Set Enrichment Analysis (GSEA) was performed using a curated set of 11 publically available, published gene sets for active YAP/TAZ as well as a composite gene set derived from RPE-1 cells expressing constitutively active YAP S5A or TAZ 4SA . This GSEA revealed that depletion of STK25 in RPE-1 cells results in a highly significant enrichment of active YAP/TAZ gene expression signatures (Fig. 2j, Supplementary Fig. 3a, b). Collectively, our data demonstrate that loss of STK25 promotes YAP/TAZ activation.

STK25 acts through LATS1/2 to inhibit YAP
Given that loss of STK25 leads to an overall decrease in phosphorylation of YAP, we predicted that overexpression of STK25 may have the opposite effect and promote phosphorylation of YAP. Indeed, we found that overexpression of wild-type STK25, but not a kinase dead mutant (STK25 K49R ) led to significant increases in levels of phosphorylated YAP relative to controls ( Supplementary Fig. 4a, b) and that this caused YAP to become enriched in the cytoplasm (Fig.   3a-c). We found that this effect of STK25 on YAP phosphorylation was LATS1/2-dependent, as overexpression of STK25 in HEK293A cells genetically depleted of LATS1/2 by CRISPR (LATS1/2 dKO) did not produce cytoplasmic enrichment of YAP ( Fig. 3d-f). We also depleted STK25 from both wild-type and LATS dKO 293A cells and found that while loss of STK25 9 reduces YAP phosphorylation and drives YAP into the nucleus in wild-type cells, there was no such effect in LATS dKO 293A ( Fig. 3g-i, Supplementary Fig. 4c), further indicating that STK25 functions through LATS1/2. Lastly, we found that knockdown of LATS1/2 via RNAi was sufficient to rescue YAP-target gene expression, even upon STK25 overexpression ( Supplementary Fig. 4d). Thus, STK25 depends on LATS1/2 to regulate YAP.
Interestingly, this STK25-LATS1/2 axis is independent of other identified upstream LATSactivating kinases. HEK293A cells depleted of MST1, MST2, and all members of the MAP4K family (MAP4K1-7) via CRISPR-Cas9 (henceforth called MM8KO 293A) still demonstrate YAP phosphorylation when grown to confluence, albeit at much lower levels than in wild-type 293A. However, MM8KO 293A cells depleted of STK25 showed significantly decreased levels of YAP phosphorylation (Fig. 3j). Additionally, overexpression of STK25 in MM8KO 293A was sufficient to induce a cytoplasmic shift of YAP (Fig. 3k, l). This suggests that STK25 acts to regulate YAP in a LATS1/2-dependent fashion and that this signaling is independent of MST1/2/MAP4Ks' activities.

STK25 stimulates LATS kinases by promoting phosphorylation of LATS on the activation loop motif
Since STK25 is sufficient to induce YAP phosphorylation in the absence of other upstream LATS-activating kinases, and given the structural similarities between STK25 and MST1/2, we tested if STK25 directly phosphorylates and activates LATS. We first assessed whether STK25 and LATS interact. We expressed HA-tagged LATS2 and FLAG-tagged STK25 in HEK293A cells and found that immunoprecipitated HA-tagged LATS2 co-precipitated FLAG-tagged STK25 (Fig. 4a). To assess whether LATS1 or LATS2 bind to endogenous STK25, we overexpressed either Myc-tagged LATS1 or HA-tagged LATS2, and then immunoprecipitated each respective LATS kinase. We found that endogenous STK25 co-precipitated with both LATS1 and LATS2, suggesting that STK25 interacts with both LATS kinases to promote their activation (Fig. 4b).
We next tested if STK25 could directly phosphorylate LATS. To do this, we carried out an in vitro kinase assay using HA-LATS2 as the substrate and either FLAG-STK25 or FLAG-MAP4K1 (positive control) as the kinase after proteins were immunoprecipitation-purified from transfected HEK293A lysates (Fig. 4c). As expected, we found that FLAG-MAP4K1 induced phosphorylation at the hydrophobic motif of LATS (LATS-HM), leading to subsequent phosphorylation at the activation loop (LATS-AL) (Fig. 4d). By contrast, STK25 did not induce phosphorylation at the LATS-HM, but was able to induce phosphorylation at the LATS-AL (Fig.   4d). This effect was dependent on STK25 kinase activity, as a kinase-dead STK25 K49R was unable to produce robust phosphorylation at the LATS-AL (Fig. 4d). Altogether, these data suggested that STK25 might be acting to activate LATS through direct phosphorylation at the activation loop site, bypassing phosphorylation at the hydrophobic motif.
To verify that STK25, and not other identified upstream activators of LATS kinases such as MST1/2 and members of the MAP4K family were responsible for the increase in phosphorylation at the activation loop of LATS, we utilized MM8KO 293A cells to IP purify STK25 and LATS2 for our in vitro kinase reactions as before. Once again, co-incubation of wild-type FLAG-STK25 with HA-LATS2 produced robust increases in phosphorylation at the activation loop of LATS2 as assessed by immunoblotting, which was decreased when LATS2 was co-incubated with kinase-dead STK25 K49R ( Supplementary Fig. 5a). As phosphorylation at the activation loop has been canonically associated with LATS auto-phosphorylation, we wished to further verify that STK25 kinase activity, and not LATS2 intrinsic kinase activity, was driving To accomplish this, we utilized LATS dKO 293A cells, into which kinase-dead LATS2 D809A was transfected. FLAG-STK25 and FLAG-MAP4K1 were also transfected into LATS dKO 293A cells. These proteins were IP-purified as before from the LATS dKO 293A cell lysates, which were then used for downstream in vitro kinase assays. Such a set-up ensured that we would be utilizing only LATS2 with no intrinsic kinase activity as our substrate, and that there would be no wild-type LATS1 or LATS2 that co-immunoprecipitated with either HA-LATS2 D809A , FLAG-STK25, or FLAG-MAP4K1 to confound interpretation of the data. Our results revealed that wild-type STK25, but not kinase-dead STK25 K49R , promoted the phosphorylation of LATS-AL, independent of LATS intrinsic kinase activity (Fig. 4e). Further, we also noted that while MAP4K1 was able to promote phosphorylation of kinase-dead LATS2 at the hydrophobic motif, it was unable to do so at the activation loop. These data reveal that STK25 activates LATS kinases through a mechanism that is completely distinct from what has previously been characterized for the MST/MAP4K kinases (Fig. 4e).
To assess whether the increase in phosphorylated LATS-AL correlates with an increase in LATS activity, we transfected STK25 either alone, or together with LATS2, in LATS dKO 293A cells and then assessed levels of YAP phosphorylation via immunoblotting. As we previously noted, overexpression of STK25 in LATS dKO 293A was insufficient to induce YAP phosphorylation; however, transfection of wild-type LATS2 was sufficient to increase levels of phosphorylated YAP, which was further increased upon co-transfection of LATS2 with STK25 WT , but not STK25 K49R (Supplementary Fig. 5b). Additionally, transfection of kinase-dead LATS2 abrogated this effect, which could not be rescued by co-transfection with STK25, strongly indicating once again that STK25 depends upon LATS kinases for its inhibitory effects on YAP ( Supplementary Fig. 5b).
To further validate our in vitro findings that STK25 is able to phosphorylate LATS at its activation loop, we assessed the effects that modulation of STK25 has on LATS-AL phosphorylation in cells. To do this, we depleted STK25 via siRNA transfection and then stimulated LATS activity by treating cells with 10 µM DCB, after which the levels of activated LATS were assessed by quantitative immunoblotting. We observed an increase in phosphorylated LATS-AL with DCB treatment, which was significantly decreased upon knockdown of STK25 (Fig. 4f). Next, we overexpressed STK25 together with LATS2 in cells and found that co-expression of LATS2 along with wild-type STK25, but not kinase-dead STK25, was able to increase levels of phosphorylated LATS-AL (Fig. 4g). Lastly, we grew to confluence either control HEK293A cells expressing Cas9 together with a non-targeting sgRNA or STK25 KO HEK293A expressing Cas9 together with a sgRNA targeting STK25, and then assessed levels of phosphorylated LATS1 at the hydrophobic motif (LATS-HM) and the activation loop (LATS-AL) by quantitative immunoblotting. We found a significant decrease in phosphorylation of the LATS-AL in our STK25 KO HEK293A, but to our surprise, we also noted a robust and seemingly compensatory increase in phosphorylation of the LATS-HM (Fig.   4h). Further, this data suggested that cells lacking STK25 are unable to appropriately 13 phosphorylate LATS-AL even when the LATS-HM is phosphorylated (Fig. 4h). Taken together, these data demonstrate that STK25 activates LATS through a previously unreported mechanism in which STK25 directly promotes phosphorylation of LATS-AL, independent from LATS intrinsic kinase activity.

STK25 Loss Impairs LATS Activation under Physiologic Conditions and Provides a Proliferative Advantage
We next analyzed the effects of STK25 in physiologically relevant contexts that are known to stimulate LATS activation. We first assessed both LATS and YAP activity in confluent, contact inhibited cells, as contact inhibited cells are known to have strong activation of LATS kinases (Zhao et al. 2012). Indeed, we found that control HEK293A cells efficiently activated LATS and induced YAP phosphorylation upon being grown to full confluence (Fig. 5a, Supplementary Fig.   6a). By contrast, HEK293A cells depleted of STK25 exhibited reduced levels of YAP phosphorylation as assessed by quantitative immunoblotting (Fig. 5a) and phos-tag analysis ( Supplementary Fig. 6b). We also observed reduced TAZ phosphorylation and subsequent stabilization of unphosphorylated TAZ levels following loss of STK25 ( Supplementary Fig. 6c).
These results were reproduced in STK25 KO 293A cells (Supplementary Fig. 6d). We also grew adherent cells in suspension for defined periods of time, as cell detachment is another known activator of LATS activity (Zhao et al. 2012). We found that depletion of STK25 significantly impairs the dynamics of YAP phosphorylation under these conditions, such that not only does the overall magnitude of YAP phosphorylation become blunted, but YAP gets phosphorylated more slowly and to a lesser extent in STK25-depleted cells compared to controls (Fig. 5c).
14 Together, these results demonstrate that STK25 plays a significant role in activating LATS kinases following cellular perturbations known to activate the Hippo signaling pathway.
We hypothesized that depletion of STK25, with subsequent YAP activation, would also provide proliferative advantages to cells. Indeed, we found that cells genetically depleted of STK25 have an increased growth rate in culture (Fig. 5b). We also found that depletion of STK25 allows cells to partially overcome cell cycle arrest induced by contact inhibition, thereby validating our findings that loss of STK25 prevents YAP phosphorylation under contact inhibited conditions (Fig. 5d). Further, it has previously been demonstrated that tetraploid cells, which arise from cytokinetic failures, fail to proliferate efficiently due to LATS activation and subsequent inactivation of YAP. Proliferation can be restored to tetraploid cells through restoration of YAP activity (Ganem et al. 2014). Indeed, we found that STK25 knockdown is also sufficient to restore proliferative capacity to tetraploid cells (Fig. 5e, f). Lastly, we wished to assess if loss of STK25 would have similar effects in vivo. To do this, we extracted protein samples from the livers of STK25-/-adult mice and their wild-type adult littermates, which we then analyzed via SDS-PAGE for YAP and TAZ levels (Fig. 5g). As expected, we found increased levels of YAP and TAZ protein in the STK25-/-mice livers compared to their wild-type controls, indicating that loss of STK25 inactivates Hippo signaling under physiologic conditions both in vitro and in vivo (Fig. 5g).
Our data suggest STK25 is a novel activator of the LATS kinases whose depletion leads to activation of YAP/TAZ and subsequent cellular proliferation. YAP/TAZ are known to be hyperactivated across a wide range of human malignancies, although the mechanisms leading to their 15 activation remain poorly understood. Our data suggest that loss of STK25 may represent one route through which cancer cells functionally activate YAP/TAZ. Interestingly, bioinformatic analysis of the TCGA reveals that focal deletion of STK25 is a very common event across many tumor subtypes, with deep deletions occurring in a significant proportion of multiple aggressive cancers (Fig. 6a, b). For example, STK25 is homozygously deleted from nearly 9% of all sarcomas (Fig 6a, b). We assessed if STK25 deletion status has an effect on the clinical course of disease and found that sarcoma patients with STK25 deletions have significantly shorter durations of survival than those with intact STK25 and that deletions of STK25 are more common in patients with recurrent disease than those without (Fig. 6c, Supplementary Fig. 7a).
Additionally, we found that this pattern of recurrent focal loss in human cancers is unique to STK25 among the currently identified LATS activating kinases (Supplementary Fig. 7b). Taken together, our findings suggest that loss of STK25 may be one mechanism through which human cancers functionally inactivate Hippo signaling to promote tumorigenesis and disease progression.

Discussion
It has been well established since the first knockout studies that deletions of both MST1 and We have now shown that STK25 is responsible for at least a portion of this remaining LATS activity, and that the mechanism through which it promotes LATS activity is novel and independent from other upstream Hippo kinases. This may explain the robustness with which loss of this singular kinase is able to potently activate YAP. Further, this lack of mechanistic redundancy may also explain why STK25 appears to be frequently focally deleted in a spectrum of human cancers, which is not observed with members of the MST/MAP4K signaling cascade (Supplementary Fig. 7b). that STK25 becomes released from the Golgi to associate with LATS kinases. Alternatively, loss of tension and consequent Golgi disruption may serve as a signal to recruit LATS kinases to sites of disturbed Golgi, where they can then associate with STK25 to promote LATS activity.
In conclusion, we define STK25 as a novel regulator of Hippo signaling, which activates LATS1/2 via a novel mechanism independent from the canonical MST/MAP4K signaling pathway. The novelty of this mechanism explains why loss of this single kinase is able to induce significant activation of YAP, which is unable to be compensated for by the presence of other upstream Hippo kinases. We posit that STK25 is a putative tumor suppressor gene, with data from human cancers supporting our claim; STK25 appears to undergo significant focal deletions 18 in a large variety of human cancers, and loss of STK25 in our cellular models promotes increased proliferation and resistance to stimuli that would normally induce cell cycle arrest.
Deletions or mutations of core Hippo pathway components are rare, and it remains to be discovered how cancer cells overcome this critical tumor suppressor pathway in the process of transformation. Our data demonstrate that loss of STK25 represents one potential route through which cancer cells might deregulate this critical tumor suppressor pathway to achieve pathologic capacity.
We would like to thank Kun-Liang Guan and members of his lab for technical advice and cell lines, Duojia Pan for reagents, and Bob Varelas and Anurag Singh for critical discussions about the work. We would also like to thank members of the Ganem     indicates "not significant"). f. YAP localization from d were quantified as before (n=4 biological replicates). g. Wild-type and LATS dKO HEK293A were transfected with the indicated siRNA, grown to confluence, then stained for YAP (Green), Tubulin (Red), and DNA (White). h. YAP intensities from g were quantified and nuclear:cytoplasmic ratios were calculated (n=225 over 3 biological replicates; ****p<0.0001, Kruskal-Wallis test with Dunn's post-test; N.S. indicates "not significant.") i. YAP localization from g were quantified as before (n=3 biological replicates). j. Immunoblot and quantification of phosphorylated YAP in MM8KO 293A cells transfected with the indicated siRNA (n=3; *p<0.05, paired t-test). k. MM8KO 293A cells were transfected with a vector encoding FLAG-tagged wild-type STK25 and were stained for YAP (Green), FLAG (Red), and DNA (White). l. YAP intensities from k were quantified in a FLAGsignal positive cell as well as an immediately adjacent FLAG-signal negative cell, and nuclear:cytoplasmic ratios were calculated (n=100 per group over 3 biological replicates; ****p<0.0001, Mann-Whitney test). All scale bars 20 µm, unless indicated otherwise.  Immunoblot and quantification of YAP and TAZ levels in these mice are presented (n=3 per genotype, **p<0.01, ***p<0.001, unpaired t-test). All data is presented as mean ± SEM unless otherwise indicated.