Kinetic regulation of the Mammalian Sterile 20-like Kinase 2 (MST2)

Abstract

Canonically, MST1/2 functions as a core kinase of the Hippo pathway and non-canonically during both apoptotic signaling and with RASSFs in T-cells. Faithful signal transduction by MST1/2 relies on both appropriate activation and regulated substrate phosphorylation by the activated kinase. Considerable progress has been made understanding the molecular mechanisms regulating activation of MST1/2 and identifying downstream signaling events. Here we investigated the ability of MST2 to phosphorylate a peptide substrate and how that activity is regulated. Using a steady state kinetic system, we parse the contribution of different factors to substrate phosphorylation including the domains of MST2, phosphorylation, caspase cleavage, and complex formation. We find in the unphosphorylated state the SARAH domain stabilizes interactions with a peptide substrate and promotes turnover. Phosphorylation drives activity of MST2 and, once activated, MST2 is not further regulated by complex formation with other Hippo pathway components (SAV1, MOB1A, and RASSF5). We also show that the phosphorylated, caspase cleaved MST2 is as active as full-length suggesting caspase stimulated activity arises through non-catalytic mechanisms. The kinetic analysis presented here establishes a framework for interpreting how signaling events and post-translational modifications contribute to signaling of MST2 in vivo.

INTRODUCTION

The Sterile 20-like kinase MST1/2 functions in multiple signal transduction pathways. Most notably, MST1/2 is the first kinase in the core kinase cassette of the Hippo pathway, a tumor-suppressive pathway and, during development, an anti-proliferative pathway. MST1/2 is also activated in response to apoptotic stimuli in FAS mediated signaling and during immune regulation¹. MST1/2 is comprised of a kinase domain, linker region, and C-terminal SARAH-domain and its kinase activity relies on the phosphorylation of its activation loop (T183 for MST1 and T180 for MST2), an event that is predominantly a consequence of trans-autophosphorylation^2–4. Mechanistically, autophosphorylation of MST1/2 is triggered by events that increase the proximity of MST1/2 to itself⁵.

During Hippo signaling, multiple events promote MST1/2 activation, most of which are mediated by the SARAH-domain of MST1/2 including recruitment of MST1/2 to the membrane^6–9, homodimerization of MST1/2^2,4,10–13, or higher-order complex formation with SAV1^5,14,15. This coiled-coil domain can homo- or hetero-dimerize with other SARAH-domains¹⁶. SAV1 contributes to Hippo signal transduction in a SARAH-domain dependent manner by helping recruit MST1/2 to the membrane and forming a higher-order complex and in a SARAH-domain independent manner by inhibiting the phosphatase complex STRIPAK, which dephosphorylates and inactivates MST1/2, and scaffolding MST1/2 to its substrate LATS1/2^15,17,18. In contrast, SARAH-domain mediated hetero-dimerization of RASSFs with MST1/2 inhibits autophosphorylation in vitro⁴. Recent work has also revealed both of these binding partners may modulate signal transduction through additional, SARAH-domain independent routes.

Following activation, MST1/2 is primed for downstream signal transduction in the Hippo pathway. Active MST1/2 phosphorylates its linker region^19,20, and the phosphorylated linker forms the binding sites for MOB1A^19,20 (Figure 1). Active MST1/2 also phosphorylates MOB1A and the other core Hippo kinase, LATS1/2²¹, the kinase ultimately responsible for inhibition of the transcriptional co-factor YAP following phosphorylation^22–25. In cells, expression of MOB1A promotes Hippo signal transduction, evidenced by increased phosphorylation of the down-stream transcription co-factor YAP²⁶. Numerous roles have been postulated for how MOB1A contributes to pathway activity. The two main theories are either MOB1A promotes LATS1/2 phosphorylation by mediating a ternary complex between MST1/2 and LATS1/2^{19,20,27–29} or relieves autoinhibition of LATS1/2 following complex formation^20,30–33. Whether or not phosphorylation of the linker domain or complex formation with MOB1A influences MST1/2 function or kinase activity remains unclear.

Figure 1.

Hippo signal transduction. Cartoon diagram of the core kinase cassette during Hippo signal transduction. Phosphorylation sites are indicated by a letter “P” enclosed in a circle, the color of which corresponds to the kinase responsible for the phosphorylation. The interaction between RASSFs and MST1/2 is indicated by a straight line with no function assigned as the effect of this interaction on MST1/2 activity varies.

The effect of RASSF family members on MST1/2 activity is complicated as RASSFs both promote and inhibit MST1/2 in cells^{4,14,16,34 35,36 15,43} (Figure 1). In vitro RASSF5 (also known as Nore1 or RAPL) and RASSF1A inhibit autophosphorylation of MST1/2 by blocking homodimerization of MST1/2 SARAH domains but in cells can stimulate the activity of MST1/2. This disparity suggests additional factors contribute to these outcomes including RASSFs mediating cross-talk with other signaling pathways³⁷, inhibiting MST1/2 activity³⁸, preventing dephosphorylation of MST1/2^39,40, or functional differences between the six RASSF homologs that bind MST1/2. Recently, the interaction between MST1 and either RASSF1A or RASSF5A was determined to involve, in addition to the SARAH-domains, other domains of each protein including the kinase domain of MST1⁴¹. This work revealed that our current understanding of the nature of this complex as well as the functional outcomes of it are still evolving. The SARAH-domain of RASSF1A was recently shown to bind atypically to the kinase domain Aurora Kinase A and tune its catalytic activity. This result provides further evidence that SARAH-domains may have roles outside of canonical coiled-coil interactions⁴².

During apoptotic signaling, MST1/2 is cleaved by caspase 3 which results in a fragment containing the kinase domain and 8 amino acids of the linker (Figure 2). Cleaved MST1/2 then translocates to the nucleus and phosphorylates apoptotic specific substrates such as H2B⁴³. When isolated from cell lysates, these MST1/2 fragments displayed both higher levels of activity and different substrate selectivity than full-length^10,43–48 supporting the notion that the linker region is an inhibitory domain and tunes interactions with substrates.

Figure 2.

Purification of MST2 variants in defined phosphorylation states. (A) Schematic of protein variants studied including MST2-FL, MST2-KL, MST2-K, and caspase cleaved (MST2-CC). (B) The homogeneity of each MST2 variant used was analyzed using Coomassie stained native page (Top) and the fraction of activation loop phosphorylation (pT180) analyzed by both Western blot using a phospho-specific antibody and Coomassie stained SDS-PAGE (Bottom).

The current picture of the kinetic regulation of MST1/2 is incomplete. Whether or how intradomain interactions, phosphorylation, and complex formation contributes to MST1/2 regulation remains to be determined. Cellular studies have identified and highlighted the range of outcomes but whether or how these events directly regulate MST1/2 kinase activity require further attention. In this work, we used purified proteins to establish a reconstituted kinase assay to directly monitor kinase activity of MST2. Using this steady-state system we performed quantitative enzymology to determine the factors regulating kinetic activity of MST2 and directly parse the contributions of each domain, phosphorylation, and complex formation to kinetic regulation of MST2. We find that following activation loop phosphorylation that complex formation with other components of the Hippo pathway (MOB1A, RASSF5, and SAV1) do not alter the activity of MST2 and the linker region is not inhibitory. We also identify a potential new role for the SARAH-domain of MST2 in modulating kinase activity.

MATERIALS AND METHODS

Expression and purification of MST2.

Nucleotides corresponding to human full-length MST2 (MST2-FL) (residues 1-484)(UniProtKB Q13188-1), kinase linker (MST2-KL) (residues 1-433), caspase cleaved (MST2-CC) (residues 1-322), or kinase domain (MST2-K) (residues 1-314) were cloned into a modified pBAT4⁴⁹ downstream from N-terminal hexahistidine and SUMO tags (Figure 2). MST2 variants were co-expressed with MBP-tagged λ-Phosphatase, encoded in pRSF-Duet (EMD-Millipore, MA) in T7 Express cells (New England BioLabs, MA) or in Rosetta2(DE3) (EMD-Millipore, MA) for MST2-KL. Cells were grown at 37°C to OD₆₀₀ of 0.8 at which time 0.5 mM Isopropyl β-d-1-thiogalactopyranoside (IPTG) was added to induce protein expression, and then cells were further grown at 20°C overnight. Cells were lysed in 50mM Tris pH8.0, 400mM NaCl supplemented with protease inhibitors (Sigma-Aldrich, MO). Clarified lysates were incubated with nickel-charged Profinity IMAC resin (Biorad, CA) for one hour at 4°C, and protein eluted with 125mM imidazole. Purified protein was incubated with SENP Protease (made in house) to remove affinity tags. Cleaved protein was further purified by anion exchange and size-exclusion chromatographies. The final protein sample was concentrated to approximately 10mg/mL in 20mM Hepes pH 7.5, 200mM NaCl, and 1mM dithiothreitol (DTT) and flash frozen in liquid nitrogen. To generate phosphorylated variants, following ion exchange, MST2 variants were incubated with 5mM ATP and 10mM MgCl₂ for 15-20 minutes at room temperature, and the reaction quenched by addition of 20 mM Ethylenediaminetetraacetic acid (EDTA). MST2-K and MST2-CC variants were purified further by MonoQ (Pharmacia, NJ) and MST2-FL and MST2-KL purified by size-exclusion chromatography.

Native PAGE.

Gel samples were prepared at a final concentration in 15% glycerol. 10μL of 3μM samples were loaded on Mini-PROTEAN TGX Gels (BioRad, CA) and run for 3 hours at 90V at 4°C in 192mM glycine and 25mM Tris base pH 8.3. Gels were stained with Coomassie Brilliant Blue for 20 minutes and imaged on an Odyssey IR Imaging System (LI-COR, NE).

Western blots.

Equal molar concentrations of MST variants were analyzed by Western blotting using a primary antibody that recognized phosphorylated Thr180 (Cell Signaling, lot 2) followed by the secondary antibody IRDye 800CW goat anti-rabbit (LI-COR, lot C91211-01). Blots were scanned on an Odyssey IR imaging system (LI-COR).

Expression and purification of MST2 binding partners.

Genes encoding the full-length human RASSF5 isoform D (hRASSF5, UniProtKB Q8WWW0-2), the full-length human MOB1A (UniProtKB Q9H8S9-1) with two site specific substitutions T12A and T35A (hMOB1A^T2A), or the SARAH domain (residues 531-608) of Salvador from Drosophila melanogaster (dSAV-SARAH)(UniProtKB Q9VCR6)were cloned into a modified pBAT vector downstream from H6 and SUMO tags⁴⁹. Each of the three binding proteins were purified using the same basic protocol, as previously described¹⁶. Each protein was expressed in T7 Express cells (New England BioLabs, MA) grown in Terrific Broth at 37°C following induction with 0.5mM IPTG. Cells were lysed in 50mM Tris pH 8, 400mM NaCl, 10% glycerol. Proteins were purified by Nickel charged Immobilized Metal Affinity Chromatography (BioRad, CA). The affinity tags were removed during overnight incubation with SENP (made in-house), and the untagged protein further purified by gel-filtration chromatography, concentrated to a minimum of 100μM, and flash frozen in liquid nitrogen. Modifications to this protocol included purification and concentration of each protein prior to gel-filtration. hRASSF5 was concentrated by 45% w/v ammonium sulfate precipitation; hMOB1A^T2A was purified on anion-exchange chromatography; dSav-SARAH was purified by anion exchange chromatography and then concentrated by 60% w/v ammonium sulfate precipitation.

Binding reactions.

Phosphorylated MST2-FL was coupled to cyanogen bromide resin (Cytiva, Sweden) at 3mg/mL according to manufacturer’s directions. 40 μL binding reactions contained 20μL column volume (CV) of either blank or MST2-FL conjugated resin and 100μM of either dSAV-SARAH, hMOB1^T2A, hRASSF5, or Bovine Serum Albumin (BSA) (RPI, IL) in 50mM Hepes pH 7.5, 400mM NaCl, 10% glycerol, 1mM DTT, 10mM MgCl₂ and was incubated for 1 hour at room temperature. Resin was collected, washed three times with 1 mL of 20mM Tris pH8, 600mM NaCl, 1mM DTT, and resuspended in 35 μL 5X SDS-Loading Buffer. Reactions were analyzed by Coomassie stained SDS-PAGE. Gels were imaged using an Odyssey IR Imaging System (LI-COR, NE). Bands were quantified using ImageJ⁵⁰. Band intensities between different gels were normalized to a common loading control, and then bound protein calculated by subtracting the intensity the bands in the lanes with blank resin from the intensities of the bands in lanes with pMST2-FL resin. Data was plotted using Prism9 (Graphpad Software, CA). P-values were determined using unpaired t-test in R-studio version 1.4.1717⁵¹.

In vitro kinase assays.

Radiometric kinase assays were carried out as previously described⁵². Briefly, 25μL reactions contained 50mM HEPES pH 7.5, 50mM NaCl, 10mM MgCl₂, 1mM DTT, 10% glycerol, 0.1mM NaF, 0.1mM Na₃VO₄, 0.2mg/ml BSA with either fixed or varying concentrations of either the peptide substrate (Biotin-GGGENWYNTLKRKK-NH2) (Genscript) or ATP spiked with 0.8μCi[γ³²P]-ATP (Perkin Elmer). The sequence of the peptide was based on a previously determined consensus sequence for the MST family of kinases to which an additional three glycine residues were included after the N-terminal biotin to act as a linker between the tag and target sequence⁵³. All kinase reactions were performed at 30°C and quenched by adding 10μL of 100mM EDTA followed by a 10-minute incubation with 10μL of 20mg/mL avidin (Prospec, Israel) at room temperature. The reaction mixture was then transferred to a 30K Omega Nanosep centrifugal concentrator device (PALL, NY) and washed three times with 100μL of wash buffer (0.5M sodium phosphate, pH 7.5, 0.5M NaCl). The membrane unit was placed in 2mL scintillation fluid (RPI, IL) and counted using a liquid scintillation counter (Beckman, CA). The turnover of limiting substrate was less than 10% for all measurements. Each reaction was performed in duplicate, and measurements were within 20%. Data were fit in Prism (GraphPad Software, CA) to a non-linear curve using the Michaelis-Menten equation to obtain apparent K_m and k_cat values. Concentrations of ATP and peptides used for each experiment are listed in Table 2. The linear range of activity for each MST2 variant with respect to either time or enzyme concentration was established in a series of assays with varying reaction times or enzyme concentrations. Each reaction contained the assay buffer supplemented with 500μM cold ATP and 300μM peptide substrate and enzyme concentrations ranged from 0.5-16nM.

Table 2.

Concentration of ATP or peptide used in Michaelis-Menten experiments.

Protein	Curve	Concentration (μM)
Unphosphorylated MST2-FL	Km ATP	10, 50, 100, 250, 500, 750, 1000
Unphosphorylated MST2-FL	Km peptide	30, 70, 100, 200, 400, 600, 800
Phosphorylated MST2-FL	Km ATP	10, 50, 100, 150, 200, 300, 400
Phosphorylated MST2-FL	Km peptide	10, 20, 100, 150, 200, 400, 800
Unphosphorylated MST2-KL	Km ATP	50, 100, 250, 750, 1000
Unphosphorylated MST2-KL	Km peptide	20, 100, 200, 400, 800, 1200
Phosphorylated MST2-KL	Km ATP	10, 50, 100, 150, 200, 300, 400
Phosphorylated MST2-KL	Km peptide	10, 50, 75, 100, 200, 400, 600
Unphosphorylated MST2-K	Km ATP	50, 100, 250, 500, 750, 1000
Unphosphorylated MST2-K	Km peptide	40, 100, 200, 400, 600, 800
Phosphorylated MST2-K	Km ATP	12.5, 25, 50, 100, 400, 600, 800
Phosphorylated MST2-K	Km peptide	10, 50, 75, 100, 200, 400, 600
Phosphorylated MST2-CC	Km ATP	10, 50, 100, 200, 400, 800, 1000
Phosphorylated MST2-CC	Km peptide	10, 40, 100, 200, 400, 600
Phosphorylated MST2-FL + Mob1 2TA	Km ATP	10, 50, 100, 150, 200, 300, 400
Phosphorylated MST2-FL + Mob1 2TA	Km peptide	7, 14, 36, 72, 144, 288, 575
Phosphorylated MST2-FL + dSav SARAH	Km ATP	10, 50, 100, 150, 200, 300, 400
Phosphorylated MST2-FL + dSav SARAH	Km peptide	10, 20, 30, 50, 100, 200, 400
Phosphorylated MST2-FL + RASSF5	Km ATP	10, 50, 100, 150, 200, 300, 400
Phosphorylated MST2-FL + RASSF5	Km peptide	10, 20, 30, 50, 100, 200, 400
Phosphorylated MST2-FL + BSA	Km ATP	10, 50, 100, 150, 200, 300, 400
Phosphorylated MST2-FL + BSA	Km peptide	10, 20, 30, 50, 100, 200, 400

RESULTS

Steady State Analysis of MST2

We set out to parse the contributions of phosphorylation, each domain of MST2, and binding partners to the ability of MST2 to phosphorylate a peptide substrate. First, we expressed and purified three different length variants of MST2 corresponding to full-length (MST2-FL), the kinase-linker (MST2-KL), and kinase domain (MST2-K) (Figure 2A). Proteins were co-expressed with Lambda Phosphatase to yield unphosphorylated, inactive enzymes and to produce active, phosphorylated proteins, ATP and Mg⁺² were added during purification. The homogeneity of each sample was monitored by native gel electrophoresis (Figure 2B), and the extent of activation loop phosphorylation by Western blot (Figure 2C). Each variant was homogenous and had comparable levels of activation loop phosphorylation. We then determined steady-state kinetic parameters for MST2 variants using a direct radiometric-assay that monitored phosphoryl-transfer of [γ-³²P]ATP onto a biotinylated peptide substrate bearing a consensus sequence for the Sterile 20-like kinase family⁵³ (Table 1) and fit the data to the Michaelis-Menten equation. Assay conditions were determined that maintained a linear relationship between enzyme activity with respect to both enzyme concentration and time (Figures 3,4) indicating that the enzymes used neither degraded nor changed phosphorylation state during the reaction.

Table 1.

Enzymatic parameters of different MST2 variants and MST2 containing complexes.

Enzyme	KmappATP (μM)	Kmapppeptide (μM)	kcat (min−1)	kcat/KmappATP (μM−1min−1)
Unphosphorylated MST2-K	84 ± 33a,b	625 ± 335a,b	74 ± 22a,c	0.9a,c
Unphosphorylated MST2-KL	84 ± 27a,b	805 ± 140a,b	62 ± 6a,c	0.7a,c
Unphosphorylated MST2-FL	195 ± 44	139 ± 18	130 ± 6	0.7
pMST2-K	100 ± 22	27 ± 4	1640 ± 54	16.4
pMST2-KL	102 ± 8	35 ± 9	1652 ± 90	16.2
pMST2-FL	158 ± 42	32 ± 3	1570 ± 34	10.0
pMST2-CC	167 ± 34	41 ± 4	1801 ± 42	10.8
pMST2-FL + BSA	78 ±7	43 ± 10	1668 ± 47	21.5
pMST2-FL + hMOB1A2TA	79 ± 14	37 ± 5	1370 ± 77	17.4
pMST2-FL + dSav-SARAH	91 ± 7	60 ± 8	2047 ± 50	22.5
pMST2-FL + RASSF5	76 ± 10	52 ± 10	1567 ± 60	20.6

^aSaturating conditions of peptide were not possible so values reported are estimated thresholds

^bValues represent lower limits of measurement

^cValues represent upper limits of measurement.

Figure 3.

Enzyme behavior of MST2 variants is linear with respect to enzyme concentration. Representative plots showing the linear relationship between velocity (nM of product/min) and enzyme concentration (nM) for both phosphorylated or unphosphorylated MST2-FL (A, B), MST2-KL (C, D), MST-K (E, F), pMST2-CC (G), or pMST2-FL in the presence of BSA (H). Each data point was performed in duplicate and is shown separately.

Figure 4.

Enzyme behavior of MST2 variants is linear with respect to time. Representative plots showing the linear relationship between [Product]/[Enzyme] with respect to time (minutes) for phosphorylated or unphosphorylated MST2-FL (A, B), MST2-KL (C, D), MST-K (E, F), pMST2-CC (G), or pMST2-FL and BSA (H). Each data point was performed in duplicate and is shown separately.

The activity of MST2 is regulated by phosphorylation and the SARAH domain

To determine the contribution of each domain to activity we compared the activities of unphosphorylated MST2 variants of differing lengths (MST2-FL, MST2-KL, and MST2-K). Unphosphorylated MST2-FL was more active than either of the unphosphorylated variants lacking the SARAH-domain (MST2-KL and MST2-K) (Figure 5, Table 1). This increased activity arises from both an improved rate, which is at least 2-fold faster (k_cat = 130min⁻¹), and improved interactions with the peptide substrate, which is at least 5-fold lower (K_m^app_peptide = 139μM). These results suggest the SARAH-domain promotes activity of the unphosphorylated MST2 kinase. Technical limitation restrict our analysis to approximations of the changes in activity between these different variants. Determination of K_m and k_cat values requires addition of saturating amounts of both substrate and ATP to the reaction and owing to the extremely high K_m values for unphosphorylated MST2-K and MST2-KL these conditions were not possible. The kinetic parameters for these variants, therefore, could not be accurately measured and instead only thresholds determined.

Figure 5.

Representative curves of steady-state kinetic analyses for each MST2 variant. Each data point was performed in duplicate and is shown separately for either unphosphorylated (A) or phosphorylated (B) MST2 variants.

We then compared the activities of phosphorylated versions of these three variants (pMST2-FL, pMST2-KL, and pMST2-K). Phosphorylation stimulated the activity of each variant (Figure 5, Table 1) by both increasing turnover and improving interactions with the peptide substrate. This change was more pronounced for variants lacking the SARAH-domain than for MST2-FL suggesting that the positive contribution of the SARAH-domain to activity of unphosphorylated MST2 is lost following kinase domain activation. In fact, the catalytic efficiencies of phosphorylated variants lacking the SARAH-domain (pMST2-KL and pMST2-K) were 50% higher than pMST2-FL (16.2 and 16.4μM⁻¹ min⁻¹ compared to 10.0μM⁻¹ min⁻¹). This modest improvement is driven primarily from differences in the K_m^app_ATP (100 or 102μM vs 158μM) suggesting that in the phosphorylated state the presence of the SARAH-domain may interfere with interactions with ATP.

Caspase cleavage of MST2 does not alter MST2 activity

In cells, caspase cleavage of MST1/2 stimulates its activity, a change previously attributed to removal of the linker domain which was, therefore, deemed inhibitory^10,44,47,48. Our kinetic characterization, however, revealed no basis for inhibition by the linker as there were no significant differences in the catalytic efficiency between pMST2-KL and pMST2-K. To understand if caspase cleavage changed the kinetics of substrate phosphorylation, we directly monitored activity of a MST2 variant that exactly corresponds to the phosphorylated caspase-cleavage product (pMST2-CC) (Table 1, Figure 2, 5). The catalytic parameters of pMST2-CC are essentially identical to those of pMST2-FL. Our results suggest that changes in MST2 activity observed in cells following caspase cleavage cannot be attributed to changes in catalysis. We note that the trend of improved catalytic efficiency observed for phosphorylated variants lacking the SARAH domain (pMST2-K and pMST2-KL) is not conserved in pMST2-CC suggesting the modest changes observed are likely not significant, especially as the kinetic parameters of the two biologically relevant variants (pMST2-FL and pMST2-CC) are essentially the same.

Activity of MST2 is unaffected by binding partners

During Hippo signal transduction, MST1/2 binds either SAV1 or RASSF5 through mutually exclusive interactions mediated by its SARAH-domain⁵⁴. Unphosphorylated MOB1A binds the phosphorylated linker of MST1/2^19,27. We wanted to determine whether or how these proteins tune substrate phosphorylation by MST2, and to do so we repeated our radiometric assays using pMST2-FL in the presence of hRASSF5, hMOB1A, and SAV1^4,54,55. First, we purified variants of each binding protein compatible with our radiometric assay. We used full-length RASSF5A. For MOB1A, since it is a substrate of MST2 and its phosphorylation weakens interactions with MST2^19,20,27, we used a variant that would remain unphosphorylated during the course of the reaction, by mutating both potential phosphorylation sites to alanine, hMOB1A^T2A. Attempts to purify either full-length SAV1 or the SARAH-domain of SAV1 were unsuccessful, instead we used the SARAH domain from the Drosophila homolog (dSAV-SARAH). Drosophila Salvador shares both sequence, structural, and functional similarity with SAV1, and previous studies of mixed systems (Drosophila Salvador and human MST2) have displayed similar behavior as equivalent mammalian systems^{5,15,16,54,56,57}. Then we confirmed each binding partner bound MST2 as expected. Complex formation with pMST2-FL was confirmed by pull-down following incubation with each binding partner at concentrations at least 10-fold above previously reported K_ds^{15,16,19,20,27,41,58} (Figure 6).

Figure 6.

Complex formation of MST2-FL with binding partners. Mixtures containing either dSAV-SARAH (orange), hMOB1AT2A (pink), hRASSF5 (green) or BSA (white) were incubated with either blank or pMST2-FL coupled cyanogen bromide resin. Complexes were isolated, and the amount of protein bound was analyzed by Coomassie-stained SDS-PAGE (A). Reactions were performed in triplicate. The average value of normalized fraction of bound protein is graphed; Individual data points for each replicate are shape-coded; Error bars represent standard deviation. Significant differences were calculated using an unpaired t-test; p-vales corresponding to ≤0.01 (**), or ≤0.001 (***).

We then repeated our kinetic analysis of pMST2-FL but now in the presence of each binding partner, at concentrations sufficient to ensure near complete complex formation^{15,16,19,20,27,41,58}. Addition of excess protein could result in macromolecular crowding and alter excluded volumes, diffusion rates, or viscosity; each of these outcomes could manifest as an apparent change in the kinetics of MST2⁵⁹. To first control for macromolecular crowding, we performed our kinetic analysis of pMST2-FL in the presence of BSA at an equivalent mass amount as of the binding partners (Table 1, Figure 7). Indeed, we found that while excess BSA did not alter either K_m^app_peptide or k_cat, it did decrease K_m^app_ATP by two-fold. The activity of pMST2-FL in the presence of BSA, therefore, became the baseline we used to compare the effects of binding partners on MST2 activity. We then found that addition of binding partners (hRASSF5, hMOB1A^T2A, or dSAV-SARAH, hMOB1A^T2A) did not further tune the activity of pMST2-FL. In each condition tested, the kinetic parameters are within error of the pMST2-FL in the presence of BSA. Our results are in agreement with a previous, qualitative study on the contribution of RASSF5 on pMST2-FL¹³.

Figure 7.

Representative curves of steady-state kinetic analyses for MST2 in the presence of binding partners. Each data point was performed in duplicate and is shown separately for pMST2-FL in the presence of either dSAV-SARAH (orange), hMOB1AT2A (pink), hRASSF5 (green), or BSA (black).

DISCUSSION

While the mechanism of activation of MST1/2 is relatively well established, how kinase activity is further regulated is less so. Our kinetic analysis reveals that the main determinant of MST2 activity is its phosphorylation state. Phosphorylation improved all kinetic parameters monitored, a trend consistent with the common activation mechanism of kinases in which phosphorylation of the activation loop stabilizes the active conformation of the enzyme and improves access of substrates to the catalytic cleft⁶⁰. These trends are consistent with, although greater in magnitude than, previous studies that monitored substrate phosphorylation as an indirect readout of autophosphorylation and activation of MST1⁶¹.

We also uncovered a novel role for the SARAH-domain of MST2 in modulating its kinase activity. Previous discussions of how the MST2 SARAH-domain contributes to signal transduction, to this point, have largely focused on how the domain regulates autophosphorylation and activation of MST2 but our data suggests the SARAH-domain may additionally tune substrate phosphorylation. We find that the SARAH-domain stabilized interactions with the substrates and improved turnover for unphosphorylated MST2 and modestly weakened interactions with ATP without affecting turnover for phosphorylated MST2. These data suggest the positive effect of the SARAH-domain on the activity of unphosphorylated MST2 is superseded by activation loop phosphorylation. Futures studies are warranted to investigate a molecular mechanism for how the SARAH-domain can promote kinase activity of unphosphorylated MST2. The ability of a SARAH-domain to tune kinase activity is not without support; a recent study revealed a non-canonical interaction the activity of Aurora Kinase A is tuned by a direct interaction between its kinase domain and the SARAH-domain of RASSF1A⁴². While the presence of the SARAH-domain slightly lowered the catalytic efficiency of pMST2-FL, the addition of other SARAH-domain proteins (dSAV1-SARAH and RassF5), which would presumably bind MST2 SARAH-domain under the conditions tested, neither relieved this inhibitory affect nor further tuned MST2 activity. These results, coupled with the observation that pMST2-FL and pMST2-CC had the same catalytic activity, suggest that the role of the SARAH-domain in tuning activity of phosphorylated MST2 is minimal if any.

Once phosphorylated, pMST2-FL is not tuned by complex formation with other members of the Hippo pathway. In cells, these binding partners do, however, contribute to signal transduction by MST2 implying these changes must be attributed to non-kinetic factors. All three binding partners examined mediate larger assemblies which may influence the cellular localization of MST2 or mediate interactions with additional regulatory proteins that could influence downstream signaling without altering the kinase activity of MST1/2. For example, SAV1 increases the level of active MST2 by promoting activation and preventing dephosphorylation of active MST1/2, thus promoting downstream signal transduction^5,6,15,17; hMOB1A may stabilize a ternary complex with LATS1/2 which would increase the effective concentration of a substrate (LATS1/2) for MST1/2^19,20; and the RASSFs link MST1/2 to other signaling pathways. All of these mechanisms would entail alterations to Hippo pathway activity without a catalytic change in the activity of activated MST2, and any additional analysis of the contribution of these complexes to downstream studies should focus on these non-catalytic contributions.

We were surprised by the lack of difference in activity between pMST2-K and pMST2-FL since the linker region appears to be inhibitory in cells^10,62. Due to technical limitations, we cannot speculate on whether the unphosphorylated linker does act as a competitive substrate inhibitor (Table 1, Figure 5). Instead, we directly compared the activity of the exact caspase cleaved fragment (pMST2-CC) to pMST2-FL and found no difference in the catalytic properties between these two physiological variants. The increased activity of the caspase-cleaved product in cells most likely arises, therefore, from non-catalytic changes. Three previously suggested theories on the increased activity are consistent with both the present and previous data. Following caspase cleavage MST1/2 can translocate to the nucleus resulting in co-localization of enzyme with its substrate, caspase cleavage changes the substrate preference of the enzyme, or caspase cleavage reduces the likelihood of dephosphorylation of the activation loop^{3,43,44,46–48,63}. Future work will be necessary to characterize regulation of activity of pMST2-CC towards full length substrates.

Methodological differences between the current and previous studies most likely account for the reported differences in activity between full-length and caspase cleaved MST2. Of note, our current study uses purified proteins in defined phosphorylation states rather than proteins isolated from cells that most likely have incomplete and varying levels of phosphorylation. Further the present study uses a peptide substrate, rather than a physiological protein substrate and, therefore, would not detect changes in activities toward substrates that arise from interactions outside the catalytic cleft. Additionally, much of the early work on the inhibitory nature of the linker domain focused on MST1, so sequence variation between the two homologs may accounts for the differences. However, the caspase-cleaved fragments of MST1 and MST2 are 91% identical, and the differences cluster to the αK-helix rather than the linker so it is not readily apparent whether or how these variations would modulate kinase activity.

We set out to understand how substrate phosphorylation by MST2 is regulated by performing steady-state kinetics using different length MST2 variants in defined phosphorylation states and in the presence of binding partners. We evaluated the contribution of both internal regulation (phosphorylation or domain contributions) and external factors (protein:protein interactions or caspase cleavage) on substrate phosphorylation by MST2. By using purified recombinant proteins in defined phosphorylation states, we were able to parse the contribution of specific domains, phosphorylation states, and binding partners to regulating the activity of MST2 to phosphorylate a peptide substrate. While there is still much to discern about the contributions of MST1/2 to the complicated signaling landscape of cells, the work presented here will provide key constraints to help interpret how signal transduction is tuned in different cellular contexts and highlights potential new regulatory nodes needing further exploration.

DATA AVAILABILITY

All processed data are included in this manuscript. Requests for raw data, further information, or reagents contained within the manuscript are available upon request from the corresponding author.