Kinetic characterization of apoptotic Ras signaling through Nore1-MST1 complex formation

The mammalian canonical Hippo signaling pathway leading to cell apoptosis involves a core kinase cascade which includes the Hpo homologs MST1 and MST2 (Pan, 2010; Fallahi et al., 2016; Meng et al., 2016). Additionally, several models involving the small GTPase Ras and members of the tumor suppressor family Rassf have been proposed for mediating and regulating apoptosis, but the precise mechanism is not known yet and contradictory results were also reported with respect to the apoptotic roles of these proteins and their complexes (Praskova et al., 2004; Oh et al., 2006; Matallanas et al., 2007; Song et al., 2010; Guo et al., 2011; Rawat and Chernoff, 2015).

The Rassf protein family has ten members; Rassf1-8 consisting of multiple splice variants (Richter et al., 2009; Gordon and Baksh, 2011; Volodko et al., 2014). Rassf5 (also named Nore1 or RAPL) was the first member to be cloned (Vavvas et al., 1998). It is expressed as three transcripts (A–C) via differential promoter usage and alternative splicing, the longest form being Rassf5A (also and hereafter named Nore1A).

Among the Rassf proteins, the interactions of Nore1 with Ras and with MST1/2 were extensively studied (Khokhlatchev et al., 2002; Vos et al., 2003; Praskova et al., 2004; Wohlgemuth et al., 2005; Stieglitz et al., 2008; Bee et al., 2010; Park et al., 2010; Fausti et al., 2012; Ni et al., 2013; Hwang et al., 2014). In 293-T cells, Nore1-induced growth inhibition was found to be H-Ras dependent (Vos et al., 2003). The role and effects of Ras in apoptosis might be cell- and mechanism-dependent, as contradictory results exist in the literature: H-Ras was found to be both more (Walsh and Bar-Sagi, 2001) and less (Khokhlatchev et al., 2002) effective than K-Ras at promoting apoptosis in different cell lines.

TNF-α and TNF-related apoptosis-inducing ligand (TRAIL) stimulation determined complex formation between Nore1A and MST1, leading to apoptosis (Park et al., 2010). An important aspect of regulation by heterodimerization seems to be the temporal considerations, as Nore1 prevents activation if it binds to inactive MST2, but has no effect on active MST2 (Ni et al., 2013). Endogenous Nore1A and MST1 were found to be in a constitutive complex and serum stimulation induced association of the complex with endogenous H/K-Ras via the RBD domain of Nore1A (Khokhlatchev et al., 2002; Praskova et al., 2004). Moreover, the apoptotic efficiency of MST1 increased when the recombinant kinase was targeted to the membrane (Khokhlatchev et al., 2002). The complex Nore1A-MST1/2 can be considered as a Ras effector unit mediating the apoptotic effect of Ras. Additionally, Ras activation and binding to Nore1A inhibits the Nore1A-induced microtubule nucleation (Bee et al., 2010).

Beside its pro-apoptotic role, Nore1A is involved in a multitude of other Ras-mediated signaling pathways with anti-transformation effects, independently of MST1. It was shown to couple Ras to the control of protein turnover and to be a powerful Ras senescence effector (reviewed in Donninger et al., 2016). Thus, Ras was shown recently to qualitatively control the posttranslational modifications of p53 via the HIPK2 kinase and favor senescence over apoptosis through the novel Nore1A-HIPK2-p53 pathway (Donninger et al., 2015). Nore1A was also found to regulate p53 stability by destabilizing the major p53 antagonist, MDM2, in a Ras-dependent manner (Schmidt et al., 2016). Nore1A is also involved in the other major Ras-induced senescence pathway, which is regulated by the retinoblastoma (Rb) tumor suppressor (Barnoud et al., 2016). Nore1A-Ras also regulates the Wnt/β-catenin growth promoting signaling pathway (Schmidt et al., 2014). Other studies found that Nore1A could inhibit cell proliferation independently of both MST1/2 kinases and Ras-GTPases through a delay in cell cycle progression (Aoyama et al., 2004). Thus, Nore1A seems to have several effectors through which it exerts its tumor suppressor role, depending on the cell context.

All Rassf members contain a Ras binding domain (RBD) close to their N- or C-terminus, which potentially interacts with members of the Ras GTPase family of proteins. Rassf1-6 also contain a coiled-coil Salvador-Rassf-Hippo (SARAH) domain for interactions with the SARAH domain of other proteins such as MST and homo- and hetero-dimerization of the Rassf isoforms (Avruch et al., 2009; Volodko et al., 2014). All SARAH-mediated dimers were reported to be antiparallel coiled-coil-like structures (Hwang et al., 2007; Makbul et al., 2013; Ni et al., 2013; Hwang et al., 2014). Most of the in vitro studies involving Nore1A and/or other Rassf members, Ras and/or MST used only the isolated interacting domains. Nore1A is comprised of a proline rich domain at the N-terminus, followed by C1 and RBD domain and ending with the SARAH domain at the C-terminus. The C1 domain was found to form an intramolecular complex with the RBD domain, which gets disrupted by the competitive binding between Ras and RBD eliberating a lipid binding interface in the C1 domain (Harjes et al., 2006). The contact area of Nore1A RBD in complex with H-Ras was found to be extended compared with those of other Ras-effectors, having a unique H-Ras switch II binding site (Stieglitz et al., 2008). This second binding site affects mainly the dissociation rate constant, k_off, which results in a prolonged lifetime (10 s) compared with other Ras effectors (0.1–1s). As no catalytic activity has been described for Nore1A or any other Rassf, it is supposed to act as a scaffolding protein, relying on signals from Ras to MST or other effectors, regulating their activity, which could explain the exceptionally long lifetime.

Besides the SARAH domain at the C-terminus, MST also presents a kinase domain at the N-terminus, which is linked to SARAH via a long (100 amino acids) intrinsically disordered auto-inhibitory (hereafter named Inhibitory) domain (Creasy et al., 1996; Hwang et al., 2007; Constantinescu Aruxandei et al., 2011; Rawat and Chernoff, 2015). MST forms SARAH-mediated homodimers and the investigations of the homodimerization role has led to contradictory conclusions, but they generally indicate that it regulates the kinase activity (Creasy et al., 1996; O’Neill et al., 2004; Anand et al., 2008; Ni et al., 2013; Rawat and Chernoff, 2015).

As mentioned above, the previous in vitro studies of the interactions between Ras and Rassf involved only the RBD domain (Wohlgemuth et al., 2005; Stieglitz et al., 2008; Bee et al., 2010). The Rassf-MST heterodimers are formed via the SARAH domains. Therefore, it is important to evaluate the intramolecular effect of the Nore1A SARAH domain and the effect of MST interaction on the binding of Rassf to Ras. To the best of our knowledge, there is no quantitative analysis addressing this issue. Moreover, the Ras-Rassf-MST complex formed in vitro has never been characterized before. Here the influence of the Nore1A SARAH domain, of the MST1 Inhibitory-SARAH (Inh.-SARAH) and of increased ionic strength (200 mm NaCl) on Nore1A – active H-Ras interaction is investigated in molecular detail. The bacterial synthesis and the purification of the proteins are described in Supporting Information.

Both Nore1A RBD (residues 199-358) and full-length H-Ras (residues 1-189) bound to either GDP or non-hydrolysable GTP analogue, GppNHp, elute as monomers when loaded separately on a column in analytical gel filtration. Nore1A RBD and active H-Ras.GppNHp (but not GDP-bound) mixed together elute as a 1:1 complex (Figure S1A and B). Nore1A RBD-SARAH (residues 199-413) elutes as a dimer and active H-Ras binding to the RBD domain does not interfere with the dimerization via the SARAH domains, the proteins eluting as a single peak of a complex comprising two Nore1A RBD-SARAH and two H-Ras molecules (Figure S1C–E).

We have previously found that MST1 Inh.-SARAH (residues 330–487) eluted in size exclusion chromatography (SEC) at a volume that corresponds to a molecular mass more than double the calculated mass of the dimer. Most likely this is due to the intrinsically disordered nature of the Inhibitory domain, which induces a larger volume (Constantinescu Aruxandei et al., 2011). ProtParam calculates a molecular mass for the MST1 Inh.-SARAH monomer of 18.75 kDa and a mass for Nore1A RBD-SARAH equal to 25.14 kDa. Nore1A RBD-SARAH shows an elution volume corresponding to 50 kDa (Figure S2), i.e. close to the calculated mass of the dimer. The eluting profile of the MST1 Inh.-SARAH corresponds to 76 kDa, which we consider to represent the apparent mass of the MST1 Inh.-SARAH dimer unfolded in the inhibitory part. Hence, the apparent mass of MST1 Inh.-SARAH in SEC equals twice the calculated mass. Finally, 1:1 mixing of Nore1A RBD-SARAH and MST1 Inh.-SARAH results in an elution volume corresponding to 63 kDa, which equals the mass of one Nore1A RBD-SARAH plus the apparent mass of one MST1 Inh.-SARAH molecule (Figure S2). Therefore the elution of an eventual H-Ras – Nore1A RBD-SARAH – MST1 Inh.-SARAH ternary complex is expected to correspond to ~84 kDa molecular mass (considering one molecule of each), which would overlap with the MST1 Inh.-SARAH dimer peak. Therefore, in order to have a larger difference between the different complexes, MST1 Inh.-SARAH was fused to enhanced Cyan Fluorescent Protein (eCFP) with molecular mass of 30.6 kDa, at the N-terminus. The control experiment with eCFP showed that there was no dimerization of eCFP (data not shown). The fusion protein MST1 Inh.-SARAH – eCFP yielded the expected elution profile corresponding to the molecular mass of two eCFP molecules and the apparent mass of two MST1 Inh.-SARAH molecules (Figure S3A and B).

As can be seen from Figures S2 and S3, the heterocomplex is formed preferentially when MST1 Inh.-SARAH and Nore1A RBD-SARAH are mixed together, as the major peak in the elution profile turns out to be shifted to intermediate molecular mass. This means that the MST1 Inh.-SARAH and Nore1A RBD-SARAH homodimers are dissolved in favor of the MST1 Inh.-SARAH – Nore1A RBD-SARAH heterodimer formation.

Finally, a ternary complex comprising MST1 Inh.-SARAH, Nore1A RBD-SARAH and active H-Ras was formed preferentially when the three proteins were mixed together (Figure S3D–F), which indicates that H-Ras binding to RBD and the heterodimer formation via the SARAH domains are not competitive to each other. In fact, comparing the fractions of MST1 Inh.-SARAH homodimer in Figure S3A and S3D, H-Ras rather increases the Nore1A RBD-SARAH – MST1 Inh.-SARAH affinity.

The equilibrium dissociation constant K_D of the Nore1A RBD – H-Ras complex and of the Nore1A RBD-SARAH – H-Ras complex, respectively, were determined by two methods: guanidine dissociation inhibitor (GDI) experiments at 37°C and as the ratio k_off/k_on measured from stopped-flow experiments at 10°C. As the Nore1A construct consisting of RBD and SARAH was stable only in the presence of 200 mm NaCl, the effect of ionic strength on the Nore1A RBD – H-Ras interaction was also evaluated.

Figure 1 shows the dissociation rate constant of fluorescently labeled GppNHp bound to H-Ras, k_obs, obtained from the GDI experiments (Figure S4), as a function of Nore1A RBD-SARAH concentration, and the fit according to the quadratic Equation 1 (Figure 1) from which K_D (0.1 μm) was obtained.

Figure 1:

The observed dissociation rate constant obtained from the GDI measurements plotted against Nore1 RBD-SARAH concentration.

The fluorescence change due to mant-GppNHp (mGppNHp) dissociation from H-Ras was recorded (Figure S4). The ability of Nore1 RBD-SARAH to lower the dissociation rate of the nucleotide from Ras was used to determine the dissociation constant between Nore1 RBD-SARAH and H-Ras (Herrmann et al., 1995, 1996). The data were fitted to the quadratic equation shown below resulting in the dissociation constant K_D=0.1 μm.

(1)kobs=k−1−(k−1−k−2)([R0]+[E0]+KD)−([R0]+[E0]+KD)2−4×[R0]×[E0]2×[R0]

where R₀ is H-Ras and E₀ is Nore1 RBD-SARAH. More information on the GDI experiment is given in the Supplementary Material (Figure S4).

Two approaches for the GDI assay were used to see if the Nore1A RBD-SARAH – MST1 Inh.-SARAH heterodimer has any effect on the Nore1A – H-Ras complex: either (1) using constant concentration of MST1 Inh.-SARAH, and changing Nore1A RBD-SARAH concentration or (2) keeping Nore1A RBD-SARAH concentration constant and varying MST1 Inh.-SARAH concentration. In both cases, MST1 Inh.-SARAH did not have a significant effect on neither the observed dissociation rate constant nor on the K_D (0.13 μm versus 0.1 μm). In conclusion, MST1 Inh.-SARAH binding to Nore1A RBD-SARAH via SARAH domains does not change Nore1A RBD-SARAH – H-Ras affinity.

At the same ionic strength (200 mm NaCl), Nore1A RBD-SARAH affinity towards H-Ras was 4.6-fold higher compared to Nore1A RBD alone (K_D of 0.1 vs. 0.46 μm, Table 1). The increase in NaCl concentration from 0 to 200 mm resulted in ~10 fold decreased affinity (K_D from 0.05 to 0.46 μm) of Nore1A RBD – H-Ras complex (Table 1).

The kinetics of complex association and dissociation have been shown to be important in signal transduction processes (Batista and Neuberger, 1998; Piehler et al., 2000; Kiel et al., 2008; Szegezdi et al., 2012). The kinetics of Nore1A RBD and Nore1A RBD-SARAH binding to H-Ras was measured using stopped-flow experiments. The observed rate constant increased linearly with increasing Nore1A RBD or RBD-SARAH concentration and became saturated at high Nore1A RBD and Nore1A RBD-SARAH concentrations, respectively, indicating a two-step binding process (Figure 2). The first step represents the formation of an initial complex, followed by conformational changes yielding the final complex.

Figure 2:

Kinetics of Nore1 RBD-SARAH binding to H-Ras.mGppNHp.

The observed rate constant was plotted against Nore1 RBD-SARAH concentration. Fitting the data to the two-step binding equation (Equation 2) gave the initial complex dissociation constant K₁, the isomerization rate constant k₂ of the initial complex and the overall association rate constant k_on; the later was also obtained from the linear approximation k_obs=k_off+k_on[Nore1] of Equation 2 at low Nore1 RBD-SARAH concentrations, i.e. <<K₁ (Kiel et al., 2004).

(2)kobs=k+2⋅[E][E]+K1+k−2

The fit resulted in K₁=51 μm, k₂=201 s⁻¹, yielding k_on=3.9 μm⁻¹ s⁻¹ (Table S1). Similar values were obtained from the linear range (Table S1). The experiment was performed at 10°C in 50 mm Tris buffer, 5 mm MgCl₂, 2 mm DTE, and 200 mm NaCl.

The fit of the observed rate constant (Figure 2) obtained from the stopped-flow association measurements (Figure S5) gave a slightly higher k_on for Nore1A RBD-SARAH (3.9 μm⁻¹ s⁻¹) than for Nore1A RBD (2.8 μm⁻¹ s⁻¹) binding to H-Ras (Table 1 and Table S1).

The dissociation rate constantk k_off was measured by displacing mGppNHp-bound H-Ras in complex with Nore1A RBD or Nore1A RBD-SARAH with an excess of unlabelled GppNHp-bound H-Ras (Figure S6).

The dissociation rate constant was 0.15 s⁻¹ for H-Ras and Nore1A RBD-SARAH, half that of H-Ras and Nore1A RBD (Figure S6) under the same conditions and comparable to that under no salt conditions (k_off is 0.12 s⁻¹). Comparing the K_D (k_off/k_on) for the two Nore1A constructs, (RBD-SARAH and RBD) the SARAH domain increased 3.5-fold the Nore1A affinity towards H-Ras at 10°C (Table 1).

In conclusion, although the SARAH domain was not reported to interact with H-Ras directly, it increases the affinity between Nore1A and H-Ras, affecting both the association and dissociation rate constants. In the absence of a structure including the SARAH domain, it is difficult to speculate why SARAH has this effect. Interestingly, the isolated RBD domain had a higher thermodynamic stability compared with the domain within Nore1A RBD-SARAH (Figure S7A), which unfolded near the physiological temperature when measured by differential scanning calorimetry (DSC). The data suggest that the RBD domain within the longer construct has a different conformation than the domain taken separately, which could influence the Nore1A – H-Ras complex. The short link of only nine residues between the SARAH and RBD domain and the present data suggest that there might be intramolecular interactions between the two domains. Unfortunately, the constructs show irreversible aggregation/precipitation at high temperatures (Figure S7), therefore reliable thermodynamic parameters could not be extracted.

The increase of ionic strength from 0 to 200 mm NaCl induced a 5-fold decrease of k_on (from 10.6 to 2.1 μm⁻¹ s⁻¹) for Nore1A RBD binding to H-Ras (Table 1). The previous k_on determined by Stieglitz et al. was 11.9 μm⁻¹ s⁻¹, in agreement with the present value. The dissociation rate constant increased 2.5-fold in the presence of 200 mm NaCl compared to k_off measured in buffer without NaCl (k_off is 0.30 s⁻¹ and 0.12 s⁻¹, respectively). The K_D value calculated as k_off/k_on was almost 12-fold higher in the presence of 200 mm NaCl (from 0.012 μm to 0.140 μm).

The previous crystal structure and mutational analysis of Nore1A RBD in complex with H-Ras revealed that hydrophobic contacts between the N-terminal of RBD and switch II of H-Ras make≈30% of the total free binding energy (Stieglitz et al., 2008); disruption of these contacts leads to a drastic decrease of Nore1A RBD – H-Ras affinity, mainly by affecting the dissociation of the complex (e.g. H-Ras Y64A gives a 150-fold increase in the k_off and only a 8-fold decrease of k_on). This extended interface is one of the main factors responsible for the 40–160-fold longer average lifetime compared with other Ras-effector complexes, such as Ras-Raf and Ras-RalGDS (Stieglitz et al., 2008).

Nevertheless, electrostatic interactions such as that between Nore1A K283 (β2 strand) and H-Ras D38 (switch I) are also crucial for Nore1A RBD – H-Ras binding, disruption of the mentioned interaction resulting in a 240-fold decrease in affinity (Stieglitz et al., 2008). The data presented here show that the increase in ionic strength screens these interactions, leading to a decreased affinity between Nore1A RBD and H-Ras, mainly by affecting the association rate constant, k_on. The fact that k_on is more affected than k_off (5-fold vs. 2.5-fold) is consistent and also explains the contributions played by the electrostatic and hydrophobic forces to these constants:

1. The electrostatic interactions were shown to be essential for the formation of the initial encounter of Ras-effector complex (Vijayakumar et al., 1998; Kiel et al., 2004, 2008) and therefore to the k_on value. As mentioned above, the hydrophobic forces do not contribute much to the k_on value. Therefore the increased ionic strength is able to inhibit the association between Nore1A and H-Ras through electrostatic screening.

2. Once the final conformation of the complex is formed, the additional hydrophobic forces specific to this complex become dominant and therefore the ionic strength gets less relevant for k_off.

In conclusion, both the electrostatic (H-Ras switch I) and hydrophobic forces (H-Ras switch II) contribute to the K_D (k_off/k_on) of Nore1A RBD – H-Ras, but while the first are more relevant to the association (k_on), the latter are more relevant to the dissociation (k_off) and differentiate this complex from other Ras-effector complexes. For other Ras-effector complexes such as Ras-Raf the electrostatic interactions prevail at the binding interface (Kiel et al., 2004; Wohlgemuth et al., 2005), which explains the higher k_on and k_off compared to the Nore1A RBD – H-Ras complex. As expected, the ionic strength was found to impact the k_on value for these complexes (Kiel et al., 2008).

In summary, we have shown that H-Ras, Nore1A RBD-SARAH and MST1 Inh.-SARAH form a stable ternary complex in vitro and that Nore1A RBD-SARAH presents two binding sites for simultaneous binding: one to H-Ras via the RBD domain and one to MST1 via the SARAH domain. This suggests that external factors are not essential for H-Ras – Nore1A – MST1 complex formation in the Ras-mediated apoptotic signaling. Nevertheless, as the formation of H-Ras – Nore1A complex is shared by several growth-inhibitory signaling pathways the cell context and the status of the players involved decide if Hippo or other biological processes not involving MST1 are favored. The dynamics of the Nore1A – H-Ras complex is sensitively regulated by both electrostatic and hydrophobic interactions. Moreover, the presented data underline the importance of neighboring domains when investigating protein-protein interactions. The structures of Nore1A RBD-SARAH and RBD domain in complex with H-Ras might be significantly different from each other. Structures of longer Rassf constructs in complex with Ras are required in order to better understand the mechanism of these proteins.