TIBS-1114; No. of Pages 8

Regulation of mammalian Ste20 (Mst)kinasesSonali J. Rawat1,2 and Jonathan Chernoff1

1 Cancer Biology Program, Fox Chase Cancer Center, Philadelphia, PA, USA2 Department of Biochemistry and Molecular Biology, Drexel University College of Medicine, Philadelphia, PA, USA

Review

Initially identified as mammalian homologs to yeastSte20 kinases, the mammalian sterile twenty-like (Mst)1/2 kinases have been widely investigated subsequent totheir rediscovery as key components of the Hippo tumorsuppressor pathway in flies. To date, our understandingof Mst substrates and downstream signaling outstripsour knowledge of how these enzymes are controlled byupstream signals. While much remains to be discoveredregarding the mechanisms of Mst regulation, it is clearthat Mst1 kinase activity is governed at least in part byits state of dimerization, including self-association andalso heterodimerization with various other signalingpartners. Here we review the basic architecture of Mstsignaling and function and discuss recent advances inour understanding of how these important kinases areregulated.

Hippo pathway architecture in Drosophila andmammalian cellsThe serine/threonine-specific protein kinases Mst1/2 arethe defining components of the Hippo signaling pathway.This pathway controls organ size and tissue homeostasisby regulating apoptosis and cell proliferation [1,2]. Mst1/2were initially discovered in mammalian cells as membersof the Ste20 family [3,4] and shortly thereafter were bio-chemically isolated as kinases activated by extreme stress[5]. Subsequently, an ortholog of these kinases, as well asother core components of what came to be known as theHippo pathway, were discovered in Drosophila melanoga-ster by genetic screens designed to identify genes thatregulate organ size. Following these pioneering studiesin flies, conditional gene-deletion studies in mice confirmeda conserved role for Mst1/2 as a regulator of organ size andas a potential tumor suppressor [6–9]. Since then, a greatdeal of attention has been paid to defining the elementsand regulation of the Hippo pathway in both flies andmammals. However, it is becoming increasingly clear thatwhile most of the Hippo pathway machinery is highlyconserved in multicellular organisms, the organizationand functions of this pathway differ substantially in vari-ous model systems. In this review, we focus on the regula-tion of Mst1/2 in mammals, in particular with respect to

0968-0004/

� 2015 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tibs.2015.01.001

Corresponding author: Chernoff, J. (Jonathan.Chernoff@fccc.edu).Keywords: serine/threonine protein kinases; signal transduction; tumor suppressor;dimerization.

how this process differs from what has been learned fromDrosophila.

Four proteins – Hpo, Sav, Wts, and Mats – constitutethe core components of the Hippo pathway in Drosophila,homologous to Mst1/2, WW45, Lats1/2, and Mob1, respec-tively, in mammals (Box 1). In mammals, Mst1/2, in con-junction with WW45, phosphorylates Mob1 and Lats1/2,leading to their activation [10,11]. Activated Lats1/2 phos-phorylates and inactivates a transcriptional coactivator,Yes-associated protein (Yap), and/or its partner Taz bypromoting its cytoplasmic sequestration [12,13]. Yap isan oncogene that enhances transcription of genes involvedin cell proliferation by partnering with the TEAD family oftranscription factors; inactivation of Yap by the Hippopathway kinase cascade suppresses cell proliferationand promotes apoptosis [1,13,14].

While the core kinase cascade of the Hippo pathwayleading from the protein kinase Hpo/Mst to the transcrip-tional coactivator Yap/Yki is well established and highlyconserved between insects and mammalian organisms, theupstream regulation of this pathway appears to be orga-nized differently in different model organisms. Geneticexperiments in Drosophila have uncovered several up-stream regulators of the Hippo pathway, including theapical membrane proteins Merlin (Mer), Expanded (Ex),and Kibra (Figure 1A). However, it is important to notethat, in mammalian cells, direct links between Mst1/2 andthese membrane proteins have not been established, andrecent evidence strongly suggests that the mammalianHippo pathway deviates significantly from the fly model.In some cases, it is unclear whether the central kinase forwhich the pathway is named is a necessary componentof the signaling module in mammalian cells (Box 1). Forexample, genetic experiments in mice have linked Merlinto the Hippo pathway components Lats and Yap but not toMst1/2 [15]. Rather than acting upstream of Mst1/2, it hasbeen suggested that Merlin instead acts in parallel toMst1/2 to activate Lats [16]. According to this model,Merlin directly binds to Lats and recruits it to the plasmamembrane, where it is subsequently phosphorylated byactive Mst1/2 (Figure 1B). While these findings help toclarify the heretofore obscure biochemical connectionbetween Merlin and Mst, questions remain regardingtheir physiological relevance. One such question is whythe overgrowth phenotype of Mer mutants in Drosophila isso much milder than that of Hpo, Sav, or Wts. The moststraightforward interpretation of the data is that the roleof Mer in Hippo pathway activation is relatively minor

Trends in Biochemical Sciences xx (2015) 1–8 1

Box 1. Mst-less Hippo signaling: a cautionary note

regarding nomenclature

Strictly defined, the core Hippo pathway comprises the four

signaling proteins Hpo/Mst, its partner Sav/WW45, its substrate

Wts/Lats, and the Wts/Lats binding partner Mob/Mats. However, in

recent years the term has also been more loosely applied to any

signaling cascade that results in inactivation of the transcriptional

coactivator Yki/Yap, regardless of whether Hpo/Mst is involved. As it

seems both formally incorrect and misleading to refer to the Hippo

pathway without involvement of the protein for which the pathway

is named, we recommend restricting the use of this term as

originally defined. It would be more appropriate to use a more

general term, such as the Yap pathway, to describe pathways that

regulate Yap phosphorylation independently of Mst (Figure I).

WW45(Sav)

W(

Mst1/2(Hpo)

Lats1/2(Wts)

Mob1(Mats)

P

P

Yap/TAZ(Yki)

P

_

TEAD (Sd)

?

TiBS

Figure I. The core Hippo pathway. The core kinase cascade of the Hippo

pathway is shown in mammals and Drosophila (Drosophila orthologs for the

core components are shown in parentheses). The core kinase cascade in

mammals consists of Mst1/2, WW45, Lats1/2, and Mob1. Upon activation by

upstream signals, Mst1/2 associates with WW45 and phosphorylates Mob1 and

Lats1/2, leading to their activation. Activated Lats1/2 phosphorylates

transcriptional coactivator, Yap/Taz, and leads to its inactivation. Inactivated

Yap loses its ability to bind transcription factor TEAD and this leads to

induction of apoptosis and inhibition of cell proliferation. Pointed arrowheads

indicate activation and blunted ends indicate inhibition. Solid lines indicate

direct activation and dashed lines indicate unknown mechanism.

Review Trends in Biochemical Sciences xxx xxxx, Vol. xxx, No. x

TIBS-1114; No. of Pages 8

compared with that of the core components, but this isa question that requires further examination given theimportant role of Merlin in human cancer.

Interestingly, disruption of the MST1 gene in humans[17] or in mice [6] is not associated with tissue overgrowthbut rather with low numbers of naı̈ve T cells, mediated bysignaling failure from an Mst1 complex that is normallyactivated downstream of T cell receptor ligation or chemo-kine stimulation [18]. This Mst-regulated signalingpathway, which affects T cell polarization and clusteringof the integrin lymphocyte function-associated antigen-1(LFA-1), contains components that are not typical of thecore Hippo signaling pathway as initially defined in fliesand is independent of Yap/Taz [6]. Here, the small GTPase

2

Rap1 serves as an upstream activator, coupling to Mst1 viathe Rap1-binding protein Rassf5b (also known as NORE1bor RAPL), and the downstream signal output is clusteringof LFA-1. It has been proposed that the regulation of LFA-1clustering is mediated by Mst1-catalyzed phosphorylationof the exchange factor DENND1C, which then activatesthe vesicle transport GTPase Rab13, thus facilitating thedelivery of LFA-1 to the leading edge of lymphocytes[19]. Mst1 also affects lymphocyte motility, a function thatis thought to be mediated by Mst-catalyzed phosphoryla-tion of Mob1, which then activates the exchange factorDock8 and its target small GTPases Rac and Rho [20](Figure 2). Thus, in this setting, Mst is regulated andfunctions in a signaling cascade that is, in almost allrespects, not congruent with the Hippo pathway as com-monly understood. Beyond species-specific variations,these differences may reflect a general separation of Mstregulation and usage in epithelial versus lymphoid cells.

As emerging genetic data from mammalian systemsshow that Mst plays key roles in organ development andimmune function, and that these kinases participate insignaling pathways that fundamentally differ from thatof the Drosophila Hippo pathway, it is worth reviewingthe biochemical mechanisms governing Mst activation,with an emphasis on recent advances in this area.

Mst1/2: domain structure and dimerizationMst1 and Mst2 are highly homologous by sequence and areidentically organized, with an N-terminal kinase domainand a C-terminal regulatory region (Figure 3). The C-terminal regulatory region can be further divided intotwo functional domains, an autoinhibitory domain, and acoiled-coil dimerization domain known as a Salvador–Rasassociation (RA) domain family (RASSF)–Hippo (SARAH)domain [3]. Mst1/2 contains two caspase-cleavage sitesbetween the kinase and the autoinhibitory domain [21].Caspase-mediated cleavage of Mst1/2 at these sitesremoves the autoinhibitory and SARAH domains andactivates Mst1/2 during apoptosis [22]. When transientlyoverexpressed, Mst1/2 itself induces apoptosis, leading tofurther Mst activation via caspase-mediated cleavage[4,22]. Interestingly, the active, cleaved form of Mst1/2 isfound at substantial levels in normal mouse liver (but notin the spleen) in the absence of apoptosis [7]. How thesetissue-specific cleaved forms are generated, and why theyfail to trigger cell death, is currently unknown, but it hasbeen speculated that these truncated activated forms ofMst1/2 play an important role in hepatocytes in maintain-ing a differentiated, nonproliferative state [7].

Mst1 and Mst2 have long been shown to form homo-dimers, although the functional significance of these homo-dimers remains a matter of debate. Homodimerization wasfirst suspected because immunoprecipitates from epitope-tagged exogenous Mst1-transfected cells contained a bandthat co-migrated with endogenous Mst1 [4]. Later, immu-noprecipitation and gel-filtration experiments demonstrat-ed that endogenous Mst1 and Mst2 could self-associate[3,23]. Further studies showed that the extreme C-termi-nal domain of Mst1 and Mst2 (later termed the SARAHdomain) is required for homodimerization of Mst1 andMst2 [3,23,24]. Although the SARAH domain is the major

YkiYki

Sd

Fat

PWts W

Mats

P

Sav

SHpo

ExKibra Mer

YAP

YAP

P

TEAD

MER

P

LATSSP

Mst

(A)

(B)

Cytoplasmicreten�on

Sav

PYki Cytoplasmic

reten�on

TiBS

Figure 1. Upstream links to the Hippo pathway in flies and mammals. (A) In Drosophila, Expanded (Ex), Merlin (Mer), and Kibra, on activation by Fat, activate the Hpo–Sav

complex. Activated Hpo–Sav complex phosphorylates and activates the Wts–Mts complex. Activated Wt–Mts phosphorylates Yki, leading to its cytoplasmic retention and

inactivation. In the absence of Hpo signaling, Yki binds transcription factor Sd and promotes cell proliferation and inhibition of apoptosis. (B) In mammals, Merlin

interaction with Lats leads to recruitment of Lats to the plasma membrane. At the plasma membrane, Lats is phosphorylated and activated by Mst. Activated Lats

phosphorylates Yap leading to its inactivation and cytoplasmic retention. Pointed arrowheads indicate activation and blunted ends indicate inhibition. Solid lines indicate

direct interactions and dashed lines indicate unknown mechanisms.

Review Trends in Biochemical Sciences xxx xxxx, Vol. xxx, No. x

TIBS-1114; No. of Pages 8

region of interaction in the Mst1/2 homodimer, a recentstudy showed that the kinase domain of Mst1 and Mst2could also interact to form weak dimers in solution, with adissociation constant of 36 mM [23]. Using heteronuclearNMR spectroscopy, Hwang et al. showed that the SARAHdomain of each monomer comprises two helices: a shortN-terminal helix (residues 433–437; h1) and a long C-terminal helix (residues 441–480; h2). In addition, theyshowed that the SARAH domains of the two monomersinteract in a head-to-tail manner to form an antiparallelhelix dimer in which helix h1 of one monomer is folded

toward helix h1’ of the other monomer. The homodimerinterface is mainly stabilized by hydrophobic interactions,but hydrogen bonds and electrostatic interactions betweenthe helices also stabilize the dimer interface [25].

Unlike Mst1/2 in mammals, the N-terminal kinasedomains of Drosophila Hpo interact with each other andthis interaction has been shown to be critical for Hpoautophosphorylation and trans-kinase activity. SARAHdomain-mediated dimerization was not shown to have amajor impact on the kinase activity of Hpo in flies. Theseobservations suggest that dimerization of the kinase

3

Mob1M b1

Dock8

Rho

Migra�on

Rac

DENND1C

Rab13

1CP

TCR

Rap1

3

Rassf5

Mst1

LFA

111 P

TiBS

Figure 2. Hippo pathway-independent Mst signaling. In lymphocytes, Mst1/2 is activated by chemokines and T cell receptor (TCR) ligation. These signals are propagated by

activation of the small GTPase Rap1, which binds to the adaptor protein Rassf5. Rassf5 contains a Salvador–Ras association (RA) domain family (RASSF)–Hippo (SARAH)

domain, which mediates interaction with Mst1/2. Mst has been shown to phosphorylate the guanine nucleotide exchange factor (GEF) DENND1C, leading to activation of

the small GTPase Rab13 and subsequent clustering of the integrin lymphocyte function-associated antigen-1 (LFA-1). Activated Mst1 also phosphorylates the Rac/Rho GEF

Dock8, promoting cell migration.

Review Trends in Biochemical Sciences xxx xxxx, Vol. xxx, No. x

TIBS-1114; No. of Pages 8

domain is more important for the activation of DrosophilaHpo than for Msts in mammals [26].

In addition to forming Mst1–Mst1 and Mst2–Mst2homodimers, these kinases are known to form heterodi-mers with other SARAH domain-containing proteins suchas the RASSF proteins. The RASSF proteins are tumorsuppressor proteins implicated in a diverse range of bio-logical functions such as apoptosis, autophagy, the cellcycle, DNA repair, and microtubule dynamics [27–37].The RASSF proteins comprise ten members (RASSF1–10), all of which contain a conserved RA domain [38]. Ofthese ten members, only RASSF1–6 contain a SARAHdomain and, not coincidentally, only these six membershave the ability to bind Mst1/2 [38]. Mst1 binding to all sixSARAH domain-containing RASSF members has beenshown by a yeast two-hybrid screen, although not all ofthese interactions have been corroborated with endoge-nous proteins in mammalian cells [38,39]. Among the sixRASSF proteins, the interaction between Rassf5 and Mst1/2 has been extensively studied. Two recent studies com-pared the crystal structure of the Mst1/2–RASSF5 hetero-dimer with that of the Mst1/2 homodimer. Similar to Mst1homodimers, Mst1–RASSF5 heterodimers form an anti-parallel helix; however, there were several key structuraldifferences between the Mst1–RASSF5 heterodimer andthe Mst1 homodimer. First, the short N-terminal helix h1present in the Mst1 homodimer was missing from theMst2–RASSF5 heterodimer. Second, the h2 helix of theMst1 SARAH domain was longer in the Mst2–RASSF5complex than in the Mst1 homodimer. Third, a sharpdistortion of the a helix of the Mst1 monomer was observedin the Mst1–RASSF5 heterodimer at P453 and P472.Interestingly, this study also showed that Mst1–Rassf5heterodimers are more stable than the Mst1 homodimer,as there are more extensive polar and nonpolar interac-tions in the heterodimer [40]. As discussed in the nextsection, the higher affinity for heterodimer formation mayhave implications regarding Mst activation and mayalso raise questions about the abundance of endogenous

4

Mst/Mst homodimers versus heterodimeric complexeswith other SARAH domain-containing proteins.

Regulation of Mst1/2 activityMst1 and Mst2 can be activated by various apoptotic andstress stimuli, such as staurosporine, UV radiation, hydro-gen peroxide, tumor necrosis factor alpha (TNF-a), retinoicacid, okadaic acid, and several anticancer drugs [22,41–45]. Although several mechanisms such as phosphor-ylation, SARAH-mediated dimerization, and interactionwith other proteins have been proposed to regulateMst1/2 activity, much remains to be learned regardingthe mechanisms of Mst1/2 activation. In the followingsections, we discuss some important mechanisms of regu-lation of Mst1/2.

Phosphorylation

In response to apoptotic or stress stimuli, Mst1/2 is autop-hosphorylated at multiple sites in the activation loop,which results in the activation of Mst1/2 [46,47]. Amongthese autophosphorylation sites, phosphorylation at T183and T180 is essential for Mst1 and Mst2 activation, re-spectively [48]. These sites can also be trans-phosphory-lated by the Ste20 family kinase TAO kinase-3, resulting inMst activation [49]. Beside autophosphorylation, Mst1/2are also regulated by trans-phosphorylation by proteinkinases such as Akt and c-Abl. Akt and Mst are thoughtto operate in opposition, as the former promotes cell sur-vival and the latter apoptosis. Akt binds to the C terminusof Mst1 and phosphorylates Mst1 at the T120 and T387residues. Phosphorylation of Mst1 by Akt prevents cas-pase-mediated cleavage of Mst1 and hence prevents acti-vation of Mst1 and its downstream target FOXO3[50,51]. Recently it was shown that the mammalian targetof rapamycin (mTOR) signaling pathway also regulatesMst1 phosphorylation at T120 and that this phosphoryla-tion results in loss of Mst1 function in prostate cancer cells[52]. Interestingly, the protein phosphatase M familymembers pleckstrin homology (PH) domain leucine-rich

Kinase AID SARAHN C 1 487 300 331 394 431

T387

Y433

T183

T120

D326 D349TiBS

Figure 3. Schematic representation of the domain structure of Mst1/2. Full-length Mst1/2 comprises an N-terminal kinase domain (green), a C-terminal Salvador–Ras

association (RA) domain family (RASSF)–Hippo (SARAH) domain (yellow), and an autoinhibitory domain between the kinase and the SARAH domain (blue). Mst1 is

regulated by phosphorylation at T120, T183, T387, and Y433 and caspase cleavage at D326 and D349 (equivalent phosphorylation and caspase cleavage sites are present in

Mst2).

Review Trends in Biochemical Sciences xxx xxxx, Vol. xxx, No. x

TIBS-1114; No. of Pages 8

repeat protein phosphatase 1/2 (PHLPP1/2) have beenshown to reverse the phosphorylation of Mst1 at T387,the Akt target phosphorylation site. PHLPP1/2 interactwith and dephosphorylate Mst1 at the inhibitory T387 site,thereby promoting Mst1 activation and Mst1-inducedapoptosis [53]. PHLPP1/2-induced dephosphorylation ofMst1 at T387 has also been shown to be required forcellular contact inhibition [54].

Akt also phosphorylates Mst2 at T117 and T384 resi-dues in response to mitogens and oncogenic Ras[55,56]. Akt-induced Mst2 phosphorylation inhibits Mst2kinase activity in several ways: by preventing Mst2 asso-ciation with its activator RASSF1A; by promoting Mst2interaction with its inhibitor Raf-1; and by preventingcaspase-mediated cleavage of Mst2 [55,56]. Akt-inducedMst1/2 phosphorylation and inactivation could be an im-portant mechanism by which Akt promotes cell survival.

c-Abl-mediated tyrosine phosphorylation has also beenshown to regulate Mst1 and Mst2 activity. Two groupsrecently demonstrated that protein kinase c-Abl phosphor-ylates Mst1 at Y433 and Mst2 at Y81 in response tooxidative stress, leading to Mst1 and Mst2 activationand induction of neuronal cell death [57,58]. However,these studies relied primarily on overexpressionapproaches and it is unclear whether c-Abl is an importantregulator of Mst1/2 under physiological conditions.

Homodimerization

In COS cells, endogenous Mst1 is found in high molecularweight complexes of approximately 145–443 kDa, with anaverage of 200 kDa, suggesting that Mst1 exists primarilyin multimeric complexes [3]. Exogenously expressed Mst1readily forms homodimers and these dimers require thepresence of the SARAH domain [3]. Many studies haveinvestigated the importance of Mst1/2 homodimerizationin regulating Mst1/2 activity, with contradictory conclu-sions. Creasy et al. demonstrated that dimerization hasno effect on the kinase activity of Mst1, as a deletionmutant lacking only the SARAH (but not the autoinhibi-tory) domain had the same kinase activity as wild typeMst1 [3]. However, Anand et al. showed that a dimeriza-tion-defective Mst1 mutant had a decreased ability tobind to and phosphorylate FOXO1, an important down-stream target of Mst1 [24]. These findings suggest thatMst1 dimerization is required for Mst1 kinase activity,at least toward FOXO1. Another study demonstratedthat full-length Mst2, but not the isolated kinase domainof Mst2, undergoes autophosphorylation at T180. As

autophosphorylation at T180 is important for Mst2 activa-tion, this study suggested that Mst2 homodimerizationis important for regulation of Mst2 activity [23]. In anotherstudy, reconstitution of Raf1 in Raf1�/� cells caused Mst2homodimers to disassemble. Further, Raf1 reconstitutionabrogated Mst2 kinase activity, suggesting that Mst2homodimerization is required for Mst2 activity [59]. Whilethese cell-based studies suggest that Mst dimerizationand activity are related, none are definitive, as it is difficultto separate dimerization per se from other binding events(e.g., to FOXO1) or phosphorylation events (e.g., Raf1).In our view, what is needed to resolve this question isa ‘cleaner’ biochemical approach using recombinantenzymes.

Heterodimerization with RASSF proteins

SARAH-mediated heterodimerization of Mst1/2 withRASSF proteins has been shown to both positively andnegatively regulate Mst1/2 activity [48,60]. For example,as mentioned above, in T cells Rassf5 complexes with Mst1and overexpression of an activated form of the smallGTPase Rap1 activates Mst1 [18]. Importantly, gene-dele-tion studies confirmed that this interaction is required forMst1 activation by T cell receptor or chemokine stimula-tion [18]. Supporting this model, one group showed thatoverexpression of RASSF1A in 293T cells increased thekinase activity of Mst1/2 and promoted Mst1-mediatedapoptosis [60]. By contrast, another group reported thatoverexpression of RASSF1A in cells suppressed autoacti-vation of Mst1 [48]. It is possible that these apparentlycontradictory results are related to the difference inRASSF1A expression level between the two studies, asRASSF1A was expressed at close to physiological levelsin the first study but was much higher in the second.

At least four mechanisms have been proposed for Mst1/2activation by RASSF1 (Figure 4). First, RASSF1A has beenshown to activate Mst1/2 by preventing protein phospha-tase 2A-mediated dephosphorylation of Mst1 at Thr183and Mst2 at Thr180. In addition to preventing dephos-phorylation, RASSF1A has been shown to stabilize theMst2 protein [61]. Third, RASSF1A binding to Mst2 hasbeen shown to release Mst2 from an inhibitory complexwith Raf-1, leading to the activation of Mst2. Finally,RASSF1A has been shown to promote the interaction ofMst2 with its substrate Lats1, leading to the activation ofLats1 and downstream Yap1 [62].

As with RASSF1, there are contradictory reports re-garding Mst1 regulation by RASSF2. Song et al. reported

5

Rassf1

Mst1/2

PP2A

Mst1/2P

Ac�ve

Inac�ve Ac�ve

Mst2

Raf-1

Mst2

Rassf1

Mst2

Proteindegrada�on

Rassf1 Mst2

Lats

Lats Mst2

P

P

Inac�ve

Ac�ve

Rassf1

(A) (B)

(C) (D)

TiBS

Figure 4. Mst1/2 regulation by Rassf1. Mechanistic models of Mst1/2 activation by Rassf1. (A) Rassf1 can activate Mst1 and Mst2 is by preventing dephosphorylation by

protein phosphatase 2A (PP2A). (B) Rassf1 can activate Mst2 by releasing it from its inhibitory complex with Raf-1. (C) Rassf1 can activate Mst2 by preventing its

degradation. (D) Rassf1 can activate Mst2 by promoting its interaction with its substrate, Lats. Pointed arrowheads indicate activation while blunted ends indicate inhibition.

Review Trends in Biochemical Sciences xxx xxxx, Vol. xxx, No. x

TIBS-1114; No. of Pages 8

that RASSF2 activates Mst1-mediated histone H2B phos-phorylation in vitro but that it inhibits Mst1-mediatedFOXO3 phosphorylation in intact cells [63]. These resultssuggest that regulation of Mst1 activity by RASSF2 inintact cells may be more complex than a simple associationbetween the two proteins. It is possible that, in intact cells,RASSF2 recruits other proteins into the Mst1–RASSF2complex, resulting in inhibition of Mst1 activity, or affectsMst1 substrate specificity.

The interaction of Mst2 with RASSF5 exemplifies oneother aspect of regulation by heterodimerization; namely,temporal considerations. In this case, it has been reportedthat the effects of RASSF5 binding to Mst2 depend onwhether the association occurs before or after Mst2 isactivated. In this model, if the two proteins interact beforeMst2 activation, disruption of the Mst2 homodimer byRASSF5 prevents subsequent Mst2 activation; if after,there is no effect on Mst2 activity. Thus, RASSF5 canact as a context-dependent negative regulator of Mst2depending on whether it binds to Mst2 before or afterMst2 activation [23] (Figure 5).

Regulation by non-SARAH domain protein interactions

Mst2 has been shown to interact with Raf-1, a key compo-nent of the Ras/Raf/MEK/ERK mitogen-activated proteinkinase signaling pathway that plays a crucial role in theregulation of cell proliferation, differentiation, and surviv-al [64–66]. Raf-1 interaction has been shown to negativelyregulate Mst2 kinase activity [55]. O’Neill et al. demon-strated that Mst2 was constitutively phosphorylated atactivating threonine residues in Raf1�/� cells but not inRaf1+/+ cells. Reconstitution of Raf-1 or a kinase-dead form

6

of Raf-1 in Raf1�/� cells abrogated Mst2 phosphorylationand activation, suggesting that Raf-1 inhibits Mst2 inde-pendent of Raf-1 kinase activity. Further examinationrevealed that Raf-1 inhibits Mst2 activation by two mech-anisms: first, by preventing homodimerization of Mst2;and second, by recruiting a phosphatase that removesactivating phosphorylations from Mst2 [59,67]. WhileMst2 is negatively regulated by Raf-1, conversely Mst2regulates Raf-1 [55,68]. Mst2 knock down has been shownto enhance the inhibitory phosphorylation of Raf-1 atSer259, suggesting that, in response to mitogen stimula-tion, Mst2 positively regulates Raf-1 activity by suppres-sing the Ser259 inhibitory phosphorylation of Raf-1 [69].

Antioxidant enzymes such as thioredoxin and peroxir-edoxin-1 (Prdx1) have also been shown to bind Mst1 and toregulate Mst1 activation by oxidative stress [70–72]. Simi-lar to the model of Mst2/RASSF5 interaction (Figure 5),Chae et al. suggested that association with thioredoxininactivates Mst1 by preventing its homodimerizationand autophosphorylation. In this schema, H2O2 alleviatesinhibition of Mst1 by disrupting the interaction of Mst1with thioredoxin [70]. Prdx1 represents a second antioxi-dant enzyme that interacts with Mst1 [71,72]. In thiscase, however, the interaction is promoted, rather thandisrupted, by oxidant stress [72]. While one study showedthat association of Mst1 with Prdx1 was required for H2O2-induced Mst1 activation and induction of apoptosis [71],another reported that Mst1 interaction with Prdx1 ledto phosphorylation and inactivation of Prdx1. In thissetting, inactivation of Prdx1 by Mst1 resulted in theaccumulation of H2O2 in cells, which might reinforceMst1 activation by positive feedback [72].

Ac�ve Inac�ve Inac�ve Ac�ve

T180 P

T180 P

T180 P

T180 T180 T180

Mst2

Rassf5 Rassf5 Rassf5

TiBS

Figure 5. Regulation of Mst2 by Rassf5. Rassf5 can act as an activator or inhibitor of Mst2 depending on whether it binds Mst2 before or after activation-loop

phosphorylation. When Rassf5 binds Mst2 before activation-loop phosphorylation, it blocks Mst2 homodimerization and hence activation, but when Rassf5 binds Mst2 after

activation-loop phosphorylation it promotes Mst2 activation and function.

Review Trends in Biochemical Sciences xxx xxxx, Vol. xxx, No. x

TIBS-1114; No. of Pages 8

Concluding remarksAs central components of a recently discovered tumorsuppressor pathway, the Mst kinases have deservedlygarnered much attention. Based on pioneering experi-ments in Drosophila, the central role of Mst1/2 in regulat-ing Yap has been established, but it is now apparent thatin mammals Mst1/2 have roles other than Yap activationand that Yap can be activated independently of Mst1/2.These findings mean that Mst signaling is more complexthan initially supposed. It is clear that much work needs tobe done in terms of understanding how Msts are activated,particularly in mammalian systems, and to determinewhether the two isoforms of Mst have unique regulationor functions. Other open questions remain regarding therelationship of Mst dimerization and activity and howsuch dimerization is regulated. Given the multiplicity ofrelevant genetic-model systems now available and recentadvances in the structural analysis of the Mst kinaseand SARAH domains, a more complete picture of theworkings of this important kinase should soon come intoview.

AcknowledgmentsThe authors thank Jeffrey Peterson for valuable comments on themanuscript. This work was supported by grants from the NationalInstitutes of Health (NIH) (R01 CA148805 and R01 CA098830) to J.C.and an appropriation from the State of Pennsylvania to the Fox ChaseCancer Center (P30 CA006927).

References1 Zhao, B. et al. (2010) The Hippo–YAP pathway in organ size control and

tumorigenesis: an updated version. Genes Dev. 24, 862–8742 Zeng, Q. and Hong, W. (2008) The emerging role of the Hippo pathway

in cell contact inhibition, organ size control, and cancer development inmammals. Cancer Cell 13, 188–192

3 Creasy, C.L. et al. (1996) The Ste20-like protein kinase, Mst1,dimerizes and contains an inhibitory domain. J. Biol. Chem. 271,21049–21053

4 Creasy, C.L. and Chernoff, J. (1995) Cloning and characterization of ahuman protein kinase with homology to Ste20. J. Biol. Chem. 270,21695–21700

5 Taylor, L.K. et al. (1996) Newly identified stress-responsive proteinkinases, Krs-1 and Krs-2. Proc. Natl. Acad. Sci. U.S.A. 93, 10099–10104

6 Zhou, D. et al. (2008) The Nore1B/Mst1 complex restrains antigenreceptor-induced proliferation of naive T cells. Proc. Natl. Acad. Sci.U.S.A. 105, 20321–20326

7 Zhou, D. et al. (2009) Mst1 and Mst2 maintain hepatocyte quiescenceand suppress hepatocellular carcinoma development throughinactivation of the Yap1 oncogene. Cancer Cell 16, 425–438

8 Song, H. et al. (2010) Mammalian Mst1 and Mst2 kinases play essentialroles in organ size control and tumor suppression. Proc. Natl. Acad. Sci.U.S.A. 107, 1431–1436

9 Oh, S. et al. (2009) Crucial role for Mst1 and Mst2 kinases in earlyembryonic development of the mouse. Mol. Cell. Biol. 29, 6309–6320

10 Praskova, M. et al. (2008) MOBKL1A/MOBKL1B phosphorylation byMST1 and MST2 inhibits cell proliferation. Curr. Biol. 18, 311–321

11 Chan, E.H. et al. (2005) The Ste20-like kinase Mst2 activatesthe human large tumor suppressor kinase Lats1. Oncogene 24,2076–2086

12 Hao, Y. et al. (2008) Tumor suppressor LATS1 is a negative regulator ofoncogene YAP. J. Biol. Chem. 283, 5496–5509

13 Zhao, B. et al. (2007) Inactivation of YAP oncoprotein by the Hippopathway is involved in cell contact inhibition and tissue growth control.Genes Dev. 21, 2747–2761

14 Zhao, B. et al. (2008) TEAD mediates YAP-dependent gene inductionand growth control. Genes Dev. 22, 1962–1971

15 Zhang, N. et al. (2010) The Merlin/NF2 tumor suppressor functionsthrough the YAP oncoprotein to regulate tissue homeostasis inmammals. Dev. Cell 19, 27–38

16 Yin, F. et al. (2013) Spatial organization of Hippo signaling at theplasma membrane mediated by the tumor suppressor Merlin/NF2. Cell154, 1342–1355

17 Nehme, N.T. et al. (2012) MST1 mutations in autosomal recessiveprimary immunodeficiency characterized by defective naive T-cellsurvival. Blood 119, 3458–3468

18 Katagiri, K. et al. (2006) Spatiotemporal regulation of the kinase Mst1by binding protein RAPL is critical for lymphocyte polarity andadhesion. Nat. Immunol. 7, 919–928

19 Nishikimi, A. et al. (2014) Rab13 acts downstream of the kinase Mst1 todeliver the integrin LFA-1 to the cell surface for lymphocyte trafficking.Sci. Signal. 7, ra72

20 Mou, F. et al. (2012) The Mst1 and Mst2 kinases control activation ofRho family GTPases and thymic egress of mature thymocytes. J. Exp.Med. 209, 741–759

21 Lee, K.K. et al. (2001) MST, a physiological caspase substrate, highlysensitizes apoptosis both upstream and downstream of caspaseactivation. J. Biol. Chem. 276, 19276–19285

22 Graves, J.D. et al. (1998) Caspase-mediated activation and induction ofapoptosis by the mammalian Ste20-like kinase Mst1. EMBO J. 17,2224–2234

23 Ni, L. et al. (2013) Structural basis for autoactivation of human Mst2kinase and its regulation by RASSF5. Structure 21, 1757–1768

24 Anand, R. et al. (2008) Biochemical analysis of MST1 kinase:elucidation of a C-terminal regulatory region. Biochemistry 47,6719–6726

25 Hwang, E. et al. (2007) Structural insight into dimeric interaction of theSARAH domains from Mst1 and RASSF family proteins in theapoptosis pathway. Proc. Natl. Acad. Sci. U.S.A. 104, 9236–9241

26 Jin, Y. et al. (2012) Dimerization and cytoplasmic localization regulateHippo kinase signaling activity in organ size control. J. Biol. Chem.287, 5784–5796

27 Donninger, H. et al. (2007) The RASSF1A tumor suppressor. J. Cell Sci.120, 3163–3172

7

Review Trends in Biochemical Sciences xxx xxxx, Vol. xxx, No. x

TIBS-1114; No. of Pages 8

28 Dammann, R. et al. (2000) Epigenetic inactivation of a RAS associationdomain family protein from the lung tumour suppressor locus 3p21.3.Nat. Genet. 25, 315–319

29 van der Weyden, L. and Adams, D.J. (2007) The Ras-associationdomain family (RASSF) members and their role in humantumourigenesis. Biochim. Biophys. Acta 1776, 58–85

30 Vos, M.D. et al. (2003) RASSF2 is a novel K-Ras-specific effector andpotential tumor suppressor. J. Biol. Chem. 278, 28045–28051

31 Eckfeld, K. et al. (2004) RASSF4/AD037 is a potential ras effector/tumor suppressor of the RASSF family. Cancer Res. 64, 8688–8693

32 Allen, N.P. et al. (2007) RASSF6 is a novel member of the RASSFfamily of tumor suppressors. Oncogene 26, 6203–6211

33 Clark, G.J. et al. (2012) RASSF family proteins. Mol. Biol. Int. 2012,938916

34 Dallol, A. et al. (2004) RASSF1A interacts with microtubule-associatedproteins and modulates microtubule dynamics. Cancer Res. 64,4112–4116

35 Shivakumar, L. et al. (2002) The RASSF1A tumor suppressor blockscell cycle progression and inhibits cyclin D1 accumulation. Mol. Cell.Biol. 22, 4309–4318

36 Vos, M.D. et al. (2004) A role for the RASSF1A tumor suppressor in theregulation of tubulin polymerization and genomic stability. Cancer Res.64, 4244–4250

37 Vos, M.D. et al. (2000) Ras uses the novel tumor suppressor RASSF1 asan effector to mediate apoptosis. J. Biol. Chem. 275, 35669–35672

38 Avruch, J. et al. (2012) Protein kinases of the Hippo pathway:regulation and substrates. Semin. Cell Dev. Biol. 23, 770–784

39 Khokhlatchev, A. et al. (2002) Identification of a novel Ras-regulatedproapoptotic pathway. Curr. Biol. 12, 253–265

40 Hwang, E. et al. (2014) Structural basis of the heterodimerization of theMST and RASSF SARAH domains in the Hippo signalling pathway.Acta Crystallogr. D: Biol. Crystallogr. 70, 1944–1953

41 Lu, M.L. et al. (1996) UV irradiation-induced apoptosis leads toactivation of a 36-kDa myelin basic protein kinase in HL-60 cells.Proc. Natl. Acad. Sci. U.S.A. 93, 8977–8982

42 Lehtinen, M.K. et al. (2006) A conserved MST–FOXO signalingpathway mediates oxidative-stress responses and extends life span.Cell 125, 987–1001

43 Watabe, M. et al. (2000) Activation of MST/Krs and c-Jun N-terminalkinases by different signaling pathways during cytotrienin A-inducedapoptosis. J. Biol. Chem. 275, 8766–8771

44 Lee, K.K. et al. (1998) Proteolytic activation of MST/Krs. STE20-related protein kinase, by caspase during apoptosis. Oncogene 16,3029–3037

45 Kakeya, H. et al. (1998) Caspase-mediated activation of a 36-kDamyelin basic protein kinase during anticancer drug-inducedapoptosis. Cancer Res. 58, 4888–4894

46 Glantschnig, H. et al. (2002) Mapping of MST1 kinase sites ofphosphorylation. Activation and autophosphorylation. J. Biol. Chem.277, 42987–42996

47 Graves, J.D. et al. (2001) Both phosphorylation and caspase-mediatedcleavage contribute to regulation of the Ste20-like protein kinase Mst1during CD95/Fas-induced apoptosis. J. Biol. Chem. 276, 14909–14915

48 Praskova, M. et al. (2004) Regulation of the MST1 kinase byautophosphorylation, by the growth inhibitory proteins, RASSF1and NORE1, and by Ras. Biochem. J. 381, 453–462

49 Boggiano, J.C. et al. (2011) Tao-1 phosphorylates Hippo/MST kinasesto regulate the Hippo–Salvador–Warts tumor suppressor pathway.Dev. Cell 21, 888–895

50 Jang, S.W. et al. (2007) Akt phosphorylates MstI and prevents itsproteolytic activation, blocking FOXO3 phosphorylation and nucleartranslocation. J. Biol. Chem. 282, 30836–30844

8

51 Yuan, Z. et al. (2010) Phosphoinositide 3-kinase/Akt inhibits MST1-mediated pro-apoptotic signaling through phosphorylation ofthreonine 120. J. Biol. Chem. 285, 3815–3824

52 Collak, F.K. et al. (2012) Threonine-120 phosphorylation regulated byphosphoinositide-3-kinase/Akt and mammalian target of rapamycinpathway signaling limits the antitumor activity of mammalian sterile20-like kinase 1. J. Biol. Chem. 287, 23698–23709

53 Qiao, M. et al. (2010) Mst1 is an interacting protein that mediatesPHLPPs’ induced apoptosis. Mol. Cell 38, 512–523

54 Jung, S. et al. (2014) PHLPP1 regulates contact inhibition bydephosphorylating Mst1 at the inhibitory site. Biochem. Biophys.Res. Commun. 443, 1263–1269

55 Romano, D. et al. (2010) Proapoptotic kinase MST2 coordinatessignaling crosstalk between RASSF1A, Raf-1, and Akt. Cancer Res.70, 1195–1203

56 Kim, D. et al. (2010) Regulation of proapoptotic mammalian ste20-likekinase MST2 by the IGF1–Akt pathway. PLoS ONE 5, e9616

57 Liu, W. et al. (2012) Regulation of neuronal cell death by c-Abl–Hippo/MST2 signaling pathway. PLoS ONE 7, e36562

58 Xiao, L. et al. (2011) The c-Abl–MST1 signaling pathway mediatesoxidative stress-induced neuronal cell death. J. Neurosci. 31, 9611–9619

59 O’Neill, E. et al. (2004) Role of the kinase MST2 in suppressionof apoptosis by the proto-oncogene product Raf-1. Science 306,2267–2270

60 Oh, H.J. et al. (2006) Role of the tumor suppressor RASSF1A in Mst1-mediated apoptosis. Cancer Res. 66, 2562–2569

61 Guo, C. et al. (2011) The tumor suppressor RASSF1A preventsdephosphorylation of the mammalian STE20-like kinases MST1 andMST2. J. Biol. Chem. 286, 6253–6261

62 Matallanas, D. et al. (2007) RASSF1A elicits apoptosis through anMST2 pathway directing proapoptotic transcription by the p73 tumorsuppressor protein. Mol. Cell 27, 962–975

63 Song, H. et al. (2010) Role of the tumor suppressor RASSF2 inregulation of MST1 kinase activity. Biochem. Biophys. Res. Commun.391, 969–973

64 Kolch, W. (2000) Meaningful relationships: the regulation of the Ras/Raf/MEK/ERK pathway by protein interactions. Biochem. J. 351,289–305

65 Avruch, J. et al. (2001) Ras activation of the Raf kinase: tyrosine kinaserecruitment of the MAP kinase cascade. Recent Prog. Horm. Res. 56,127–155

66 Baccarini, M. (2002) An old kinase on a new path: Raf and apoptosis.Cell Death Differ. 9, 783–785

67 O’Neill, E. and Kolch, W. (2005) Taming the Hippo: Raf-1 controlsapoptosis by suppressing MST2/Hippo. Cell Cycle 4, 365–367

68 Romano, D. et al. (2014) Protein interaction switches coordinate Raf-1and MST2/Hippo signalling. Nat. Cell Biol. 16, 673–684

69 Kilili, G.K. and Kyriakis, J.M. (2010) Mammalian Ste20-like kinase(Mst2) indirectly supports Raf-1/ERK pathway activity viamaintenance of protein phosphatase-2A catalytic subunit levels andconsequent suppression of inhibitory Raf-1 phosphorylation. J. Biol.Chem. 285, 15076–15087

70 Chae, J.S. et al. (2012) Thioredoxin-1 functions as a molecular switchregulating the oxidative stress-induced activation of MST1. Free Radic.Biol. Med. 53, 2335–2343

71 Morinaka, A. et al. (2011) Oligomeric peroxiredoxin-I is an essentialintermediate for p53 to activate MST1 kinase and apoptosis. Oncogene30, 4208–4218

72 Rawat, S.J. et al. (2013) The tumor suppressor Mst1 promotes changesin the cellular redox state by phosphorylation and inactivation ofperoxiredoxin-1 protein. J. Biol. Chem. 288, 8762–8771