The to and fro of Rho

Abstract

The functions of Ras members are largely governed by the structural dynamics of the nucleotide-binding switch regions. In this issue of Structure, Lin et al. (2021) take a close look at the dynamic equilibrium of RhoA, the founding member of the Rho family of Ras GTPases.

Ras GTPases sit at the center of vital cellular signaling pathways and execute their functions primarily by way of their alternating structural states. A key determinant of Ras structural dynamics is the identity of a bound guanosine nucleotide (GDP versus GTP). However other factors, some intrinsic to specific Ras proteins, can tune these sensitive dynamics. The mobile regions of Ras proteins form a good portion of the nucleotide-binding pocket and are called the “switches.” The switches provide a key interface for binding with other signaling proteins and Ras regulators. A prominent working hypothesis is that the dynamic equilibria of switch conformations determine Ras effector specificity. A corollary hypothesis is that differences in dynamics between Ras isoforms or specific mutants, such as those found in disease, drive biological phenotypes and pathologies. These ideas have motivated detailed studies of K-Ras, H-Ras, and N-Ras dynamics because of their prominent roles in lethal cancers.

In this issue of Structure, a study by Lin and colleagues characterizes the structural dynamics of the Ras protein RhoA, the founding member of the Rho subfamily of Ras GTPases. Rho proteins such as Rac1, Cdc42, and RhoA are important regulators of actin and microtubule cytoskeletons that modulate cell polarity, migration, cycling, differentiation, and gene transcription (Hall, 2012). Dysregulation of these processes leads to disease states, ranging from neurodegenerative diseases to cancer. Understanding the detailed dynamics of Rho proteins has the potential to inspire new therapeutic strategies. Using NMR, X-ray crystallography, and molecular dynamics simulations, Lin and colleagues show that GTP-bound RhoA is dynamic in the switch regions, populating at least two states including an inactive, conformation referred to as “state 1” that presents with switch 1 extended away from the body of RhoA such that it loses most of its contacts with the nucleotide (Figure 1A). These data contrast with many structures showing that GTP-bound Ras proteins tend to present in “state 2” where the protein is more compact because switch 1 fully engages with the nucleotide (Figure 1B). Nevertheless, GTP-bound, state 1 Ras proteins have been seen before. K-Ras also behaves this way, surprisingly in contrast to closely related H-Ras (Parker et al., 2018). The authors also find that cancer-associated mutations, such as RhoA G14V, shift these dynamics.

Figure 1. Conformational states of Ras can predict biochemical activities.

In all panels, switch 1 in yellow and switch 2 in green. Nucleotide in sticks and colored by element. Active site magnesium in magenta.

(A) State 1 is defined by a switch 1 conformation where interactions with the nucleotide are lost. PDB: 6V6M.

(B) State 2 is defined by closure of switch 1 and engagement with the nucleotide. PDB: 5B2Z.

(C) Hyperextension of switch 1 leads to rapid nucleotide exchange (RNE). PDB: 6BOF.

(D) Mobility of switch 2 decouples its interactions with switch 1, allowing a more stable conformation of switch 1. This is the proposed mechanism for enhanced binding between RBD-containing proteins (gray) and Ras mutants that cause switch 2 mobility. All crystallographic monomers from 6MNX were superimposed for the figure. RBD from 4G0N in gray was superimposed based on common K-Ras sequences between 6MNX and 4G0N.

A general theory for predicting which dynamics will give rise to particular biochemical or high-level biological effects remains a work in progress. On the one hand, K-Ras mutations such as V14I and A146T give rise to large movements in switch 1, more extreme than seen in state 1 structures, and this consistently results in high rates of GDP dissociation and GEF-stimulated nucleotide exchange in K-Ras (Figure 1C) (Bera et al., 2019; Poulin et al., 2019). One might guess that state 1 conformations could also lead to increased rates of GDP dissociation, but to date, Ras isoforms or mutants showing increased population of state 1 do not show substantial increases in GDP dissociation rates or nucleotide exchange, consistent with data from the current report. On the other hand, the P29S substitution located in switch 1 of Rac1, a relatively common mutation found in melanoma, causes a similar phenotype. However, structures of this mutant have not shown an extended switch 1 (Krauthammer et al., 2012).

Another plausible hypothesis is that prominence of state 1 in dynamic equilibria could impair interactions with downstream effectors that use this interface, such as those containing a Ras-binding domain (RBD). So far, evidence for this is weak and, in fact, some Ras proteins showing state 1 conformations activate Ras signaling, the opposite of what would be expected if state 1 disrupts interactions with downstream signaling proteins such as RAF (Lu et al., 2018). Instead, dynamic behavior in switch 2 appears to predict for increased interactions with certain RBD-containing proteins (Zhou et al., 2020). One explanation for this might be that switch 2 and switch 1 normally interact in a way that limits RBD binding. However, if switch 2 becomes highly mobile, switch 1 and 2 interactions become decoupled, leading to stabilization of switch 1, enabling enhanced interactions with RBD (Figure 1D).

Another possibility is that Ras dynamics alter the behavior of Ras-containing multimeric signaling complexes. Signalosomes in which RhoA might participate have been proposed including the “zonular signalosome” that regulates tight junctions (Citi et al., 2014). Such complexes have been proposed as mobile scaffolds that function to integrate the large networks of signaling components found in Ras pathways. From the data available, these complexes appear to be highly dynamic. Presumably this makes them highly responsive to stimuli, meeting biological needs. While this property may have biological advantages, it also makes them difficult to study because they are not easily isolated. In the absence of feasible experimental approaches, structural information can enable computational approaches, such as simulation. The study by Lin et al. (2021) provides detailed structural information that might be integrable into larger modeling efforts.

Finally, enhanced movements in switch 1 could have implications for direct therapeutic targeting strategies in diseases where dysregulated RhoA causes disease, such as lymphomas and gastric cancer (Kataoka and Ogawa, 2016). The discovery of Ras G12C inhibitors such as sotorasib and adagrasib suggests a possible avenue for achieving direct RhoA targeting. These inhibitors bind to a transient pocket in Ras, known as the switch 2 pocket, that forms when switch 2 moves away from the protein (Ostrem et al., 2013) (Figure 2A). This movement is highly analogous to switch 1 movements observed in the current study. Another feature of clinically advanced switch 2 pocket compounds is that they are obligate covalent inhibitors; they rely on reacting irreversibly with Cys 12, an oncogenic mutation. Interestingly, state 1 structures of RhoA expose a native cysteine that may be targetable in a fashion similar to Cys 12 (Figure 2B). While the rationale for developing such inhibitors against RhoA is not immediately apparent, especially since targeting a native cysteine may not impart selectivity as it does in the case of K-Ras G12C, the current study raises the intriguing possibility that covalent RhoA switch 1 pocket-directed tool inhibitors may be feasible.

Figure 2. State 1 RhoA may be targetable by covalent inhibitors.

Nucleotide in sticks and colored by element.

(A) K-Ras G12C switch 2 pocket binders (blue sticks) take advantage of switch 2 movements that create a pocket next to their nucleophilic target, Cys 12 (red). PDB: 6UT0 showing MRTX849 bound to K-Ras G12C.

(B) Proposed switch 1 pocket in state 1 RhoA. Opening of switch 1 exposes native Cys 20 (red). PDB: 6V6M.