Alternative Splicing of a Structured Partner Alters the Folding-Upon-Binding Trajectory of an Intrinsically Disordered Protein

Abstract

Folding-upon-binding of intrinsically disordered proteins (IDPs) is governed by a complex interplay of kinetic and thermodynamic factors shaped by the structure and conformational dynamics of both binding partners. Alternative splicing offers a natural way to remodel the conformational energy landscape of structured partners, yet how such biologically relevant changes influence the molecular recognition trajectories of interacting IDPs remains poorly understood. Here, using the small GTPase Rac1 and its oncogenic splice variant Rac1b as a model system, we integrate X-ray crystallography, isothermal titration calorimetry (ITC), and nuclear magnetic resonance (NMR) spectroscopy to investigate how the 19-residue insertion in Rac1b alters recognition of the disordered signaling effector POSH. We show that the insertion restricts POSH to partial folding-upon-binding and determine the crystal structure of the POSH-Rac1b complex at 1.77 Å resolution. The structure reveals that POSH stabilizes the otherwise dynamic switch regions of Rac1b in a signaling-competent conformation, while the insertion itself remains dynamic. NMR exchange experiments further delineate the molecular recognition trajectory of POSH upon binding to Rac1b, revealing a folding intermediate characterized by a 5.7-fold slower association rate and a 3-fold faster dissociation rate compared to Rac1. Together, these results demonstrate that the insertion, kinetically and entropically, destabilizes the effector-bound state of Rac1b, directly linking enhanced conformational dynamics to impaired downstream signaling. More broadly, our work illustrates how alternative splicing of folded proteins can reshape folding trajectories, binding kinetics, and thermodynamic landscapes of IDP-mediated interactions, thereby rewiring cellular signaling networks.

Introduction

Intrinsically disordered proteins (IDPs) or regions (IDRs) play central roles in cellular signaling and regulation by mediating dynamic, context-dependent protein–protein interactions. − Although IDPs populate heterogeneous conformational ensembles in their free form, − they often undergo folding transitions upon binding to partner proteins, forming structured complexes through the engagement of one or more molecular recognition elements (MREs). − The thermodynamics and kinetics of these transitions have largely been attributed to intrinsic features of the IDP sequence, particularly preformation of transient secondary structure elements. − However, how the conformational landscape of the folded binding partner influences IDP recognition remains poorly understood. Although a few studies have examined how the architecture of the binding pocket and single point mutations modulate IDP binding, − little is known about how broader changes in partner structure and dynamics shape molecular recognition mechanisms. In this context, alternative splicing offers a natural way to alter the conformational landscape of a structured binding partner. Insertions, deletions, or sequence substitutions can rewire dynamics and long-range contacts within folded proteins, thereby potentially modulating the recognition of IDPs, for example by altering which MREs bind, modifying the interaction kinetics, or shifting the balance between enthalpic and entropic contributions to binding.

The small GTPase Rac1 and its oncogenic splice variant Rac1b provide a compelling model system to investigate these structural, kinetic, and thermodynamic effects. Rac1 cycles between an inactive, GDP-bound state and an active, GTP-bound state, accompanied by conformational changes in two conserved regions, switch I and switch II (Figure A). In the GDP-bound state, switch I adopts an open conformation, whereas in the GMPPNP-bound state (where GMPPNP is a nonhydrolyzable GTP analogue; Figure B), switch I closes over the nucleotide and rearranges its side chains to expose a hydrophobic patch. This patch serves as a docking interface for signaling effectors, whose binding leads to activation of downstream signaling pathways. Rac1 preferentially recognizes effectors containing Cdc42/Rac interactive binding (CRIB) motifs, defined by the consensus sequence: ISXP­(X)_2–4FXHXXHVG, where X denotes any amino acid. ,

1.

The signaling effector POSH undergoes partial folding-upon-binding to Rac1b. (A) Domain organization of Rac1, highlighting the switch I (light blue) and switch II (salmon) regions. (B) Crystal structure of Rac1 bound to GMPPNP, a nonhydrolyzable GTP-analogue (PDB 3TH5). Rac1 is shown in beige, with switch I and II in light blue and salmon, respectively. GMPPNP is shown as gray sticks. The switch I loop adopts a closed conformation. (C) Crystal structure of the Rac1-POSH_319–371 complex bound to GMPPNP (PDB 9RFF). The two molecular recognition elements of POSH_319–371 are shown in green (MRE1) and blue (MRE2). The linker region between MRE1 and MRE2 is not resolved in the electron density (gray dashed line). (D) Domain organization of Rac1b, highlighting switch I (light blue), switch II (salmon), and the Rac1b-specific insertion (yellow). (E) Crystal structure of Rac1b bound to GMPPNP (PDB 1RYH). Switch I adopts an open conformation, while most of switch II and the insertion are not resolved in the electron density (gray dashed line). (F) ITC titration of Rac1b with POSH_315–380. ITC data from five titration experiments were merged by concatenation. Raw injection heats (top, DP: differential power) and the corresponding specific binding isotherms (bottom) are shown. The data were acquired at 35 °C and analyzed using the binding model “One set of sites” (solid line). (G) ITC titration of Rac1b with POSH_315–380 acquired at 5 °C. (H) Temperature dependence of the thermodynamic parameters derived from ITC titrations of Rac1b with POSH_315–380. Points represent experimental data and lines correspond to linear least-squares fits. Error bars reflect the standard deviation between duplicate experiments. (I) Comparison of the change in heat capacity for the interaction of POSH_315–380 with Rac1b (blue) and Rac1 (red, obtained previously). (J) Comparison of the conformational entropy for the interaction of POSH_315–380 with Rac1b (blue) and Rac1 (red, obtained previously).

One such effector is POSH (Plenty Of SH3s), a scaffold protein that promotes apoptosis through activation of the c-Jun N-terminal kinase (JNK) pathway. , POSH contains an N-terminal zinc-finger domain and four Src-homology 3 (SH3) domains connected by long IDRs. The interaction between POSH and Rac1 was originally identified in yeast-two-hybrid screens. Subsequent mapping localized the Rac1-binding region to the IDR between the second and third SH3 domains (residues 260–445), and more recently to a minimal interaction module encompassing residues 315–380 (hereafter named POSH_315–380). Although POSH_315–380 contains only a partial CRIB motif (specifically an ISPP submotif), it undergoes extensive folding upon binding to Rac1 via two MREs. The first element, MRE1 (residues 323–346), encompasses the partial CRIB motif and adopts an extended conformation, a β-strand and an α-helix that engages the closed conformation of the switch I loop (Figure C). The second element, MRE2 (residues 353–372), is connected to MRE1 by a flexible linker and folds into a β-hairpin structure that contacts both MRE1 and the switch II region of Rac1 (Figure C). It has been shown that POSH_315–380 binding proceeds through a hierarchical folding mechanism, initiated by contacts made by the partial CRIB motif, followed by sequential folding of MRE1 and MRE2 on the surface of Rac1.

In contrast, little is known about Rac1b, a splice variant frequently overexpressed in colorectal and breast carcinomas, − which contains a 19-amino acid insertion immediately following the switch II region (Figure D). A crystal structure of Rac1b bound to GMPPNP reveals an open switch I loop and poorly defined electron density for both switch II and the insertion, indicative of elevated conformational dynamics in regions critical for effector recognition (Figure E). Although Rac1b is predominantly GTP-bound, a signature of constitutive activation, it is paradoxically signaling-deficient, displaying markedly reduced affinity for effectors, including POSH. , This apparent uncoupling between its activation state and downstream signaling suggests that the insertion not only perturbs the nucleotide cycle, but also reshapes the conformational energy landscape of Rac1b in a manner that alters the folding-upon-binding trajectory of signaling effectors.

In this work, we combine X-ray crystallography, isothermal titration calorimetry (ITC), and nuclear magnetic resonance (NMR) spectroscopy to define how alternative splicing of Rac1 reshapes effector recognition. We demonstrate that, in Rac1b, the insertion truncates the molecular recognition trajectory of POSH_315–380, restricting binding to MRE1 and precluding engagement of MRE2. The crystal structure of the POSH-Rac1b complex reveals that effector binding induces closure of switch I and stabilization of switch II, restoring a signaling-competent conformation of Rac1b. NMR exchange experiments further show that the Rac1b insertion slows association and accelerates dissociation of the partial CRIB motif of POSH_315–380. Together, these results support a model in which the effector-bound state of Rac1b is both kinetically and entropically destabilized compared to Rac1, establishing a direct link between the enhanced conformational dynamics of Rac1b and its impaired signaling. More broadly, our work highlights alternative splicing of folded proteins as an important mechanism for rewiring signaling networks by reshaping the folding trajectories, binding kinetics, and thermodynamic landscapes of interacting IDPs.

Results

The Signaling Effector POSH Undergoes Partial Folding-Upon-Binding to Rac1b

To investigate how the intrinsic conformational dynamics of Rac1b influence its interaction with POSH_315–380, we first characterized the thermodynamics of binding using ITC. Titration of POSH_315–380 into Rac1b yielded a dissociation constant (K _d) of 265 μM at 35 °C (Figure F and Table S1), corresponding to an approximately 12-fold reduction in affinity relative to the Rac1-POSH_315–380 complex (K _d = 22 μM at 35 °C). To gain mechanistic insight into this weakened interaction, we carried out additional ITC experiments at 5 °C (Figure G). Measurements at intermediate temperatures yielded very weak ITC signal as the binding enthalpy approaches 0 kcal/mol at 25 °C. Binding enthalpies (ΔH) obtained from analysis of the ITC data at 5 and 35 °C were therefore used to determine the change in heat capacity, ΔC _p, from the slope of ΔH versus temperature (Figure H). The determined value of ΔC _p = −1.2 kJ·mol^–1·K^–1 is less than half that of the Rac1-POSH_315–380 complex (ΔC _p = −2.8 kJ·mol^–1·K^–1), indicating a substantially reduced burial of hydrophobic surface area for binding of POSH_315–380 to Rac1b compared to Rac1 (Figure I). We next estimated the conformational entropy contribution (ΔS _conf) to the total binding entropy using an empirical relationship originally proposed by Spolar and Record and recently adapted to complexes involving IDPs (see Methods). The entropic penalty associated with POSH_315–380 binding to Rac1b is approximately one-third of that observed for Rac1 (Figure J). Collectively, our ITC data suggest that binding of POSH_315–380 to Rac1b involves less extensive surface burial and conformational ordering compared to binding to Rac1.

POSH Binds to Rac1b Exclusively through the First Molecular Recognition Element

We next used NMR spectroscopy to probe how POSH_315–380 interacts with Rac1b with site-specific resolution. The ¹H–¹⁵N HSQC spectrum of POSH_315–380 shows minimal chemical shift perturbations upon the addition of Rac1b (Figure A). However, a gradual decrease in signal intensities is observed with increasing Rac1b concentrations (Figure B). Notably, this intensity reduction is pronounced within MRE1 but minimal within MRE2, showing that only MRE1 binds to Rac1b. This interpretation is further supported by ¹⁵N R _1ρ relaxation rate measurements of POSH_315–380 in the absence and presence of Rac1 or Rac1b. While Rac1 binding increases R _1ρ rates across both MRE1 and MRE2, Rac1b induces elevated relaxation rates primarily within MRE1, with only minor changes observed within MRE2 (Figure C). Together, these results suggest that the insertion in Rac1b hinders stable engagement and folding of MRE2 on its surface.

2.

The insertion in Rac1b precludes folding of the second molecular recognition element of POSH. (A) Zoom on the ¹H–¹⁵N HSQC spectrum of POSH_315–380 in the absence (gray) and presence (red) of a 100% molar ratio of Rac1b. (B) Intensity ratios, I/I ₀, obtained from titration of ¹⁵N-labeled POSH_315–380 with unlabeled Rac1b. I ₀ represents peak intensities in the ¹H–¹⁵N HSQC spectrum of free POSH_315–380, while I corresponds to peak intensities in the presence of Rac1b. Data are shown for different molar ratios of Rac1b: 23 (orange), 45 (blue), 76 (green), and 100% (red). (C) ¹⁵N R _1ρ relaxation rates of free POSH_315–380 (gray) and with a 45% molar ratio of Rac1b (blue) or Rac1 (black). Measurements were performed at a ¹H frequency of 700 MHz at 25 °C.

POSH Binding Induces Switch I Loop Closure and Switch II Stabilization in Rac1b

To elucidate the structural basis of the Rac1b-POSH interaction, we determined the crystal structure of their complex bound to GMPPNP at 1.77 Å resolution (Table S2). Given the low affinity of this interaction, we enhanced the likelihood of crystallization by designing a fusion construct, in which residues 319–348 of POSH (corresponding to a linker and MRE1) were fused to the C-terminus of Rac1b (Figure S1A). We have previously validated this approach on the Rac1-POSH_319–371 complex, where crystal structures of both fused and unfused POSH peptides showed nearly identical binding modes. The Rac1b-POSH_319–348 fusion complex crystallized as a domain-swapped dimer, with POSH_319–348 from one fusion protein engaging a symmetry-related Rac1b molecule (Figure S1B). The structure reveals that POSH_319–348 adopts an extended structure at its partial CRIB motif (the ³²⁶ISPP³²⁹ submotif), followed by a short β-strand that augments an existing β-sheet within Rac1b and finally a short α-helix that clamps the newly formed β-strand (Figure ). The entire molecular recognition element is well-defined in the electron density map, with clear definition of most side chains (Figure A). Notably, POSH_319–348 binds to a hydrophobic surface on Rac1b, adopting a conformation that closely resembles that of the corresponding region in the Rac1-POSH_319–371 complex (Figure B). In addition to hydrophobic contacts, POSH_319–348 binding is further stabilized by several hydrogen bonds with Rac1b, particularly involving residues within the partial CRIB motif and the β-strand of POSH_319–348 (Figure C).

3.

Crystal structure of the Rac1b-POSH_319–348 complex bound to GMPPNP. Rac1b is shown in beige, with the switch I and II regions highlighted in light blue and salmon, respectively. The Rac1b-specific insertion is shown in yellow, although only a few residues are well resolved in the electron density. GMPPNP is displayed as gray sticks, and MRE1 of POSH is shown in green. Two orientations of the complex, rotated 90° around the x-axis, are shown.

4.

POSH binding promotes closure of the switch I loop and induces ordering of the switch II region of Rac1b. (A) Unbiased electron density map (Fo-Fc) of POSH-MRE1. (B) Comparison of POSH binding to Rac1b (green) and Rac1 (pink, PDB 9RFF). The Rac1b surface is colored by hydrophobicity (ranging from dark orange for the most hydrophobic potentials, through white, to teal for the most hydrophilic potentials). (C) Detailed view of the Rac1b-POSH_319–348 interface, highlighting hydrogen bonds (cyan dashed lines) between POSH_319–348 and Rac1b. (D) Zoom on the nucleotide-binding site in the Rac1b-POSH_319–348 complex. The switch I loop adopts a closed conformation, stabilized by multiple hydrogen bonds between switch I and the γ-phosphate of GMPPNP. (E) Zoom on the switch II region showing its ordered conformation stabilized by multiple hydrogen bonds. (F) View of the catalytic water molecule (red sphere) and its hydrogen-bonding interactions (cyan dashed lines). The side chain of Q61 is displaced and lies outside hydrogen-bonding distance from the catalytic water (indicated by black dashed lines).

Binding of POSH_319–348 to Rac1b induces a closure of the switch I loop, stabilized by hydrogen bonds between the γ-phosphate of GMPPNP and the side chain hydroxyl groups of Y32 and T35, as well as the backbone amide of T35 of Rac1b (Figure D). This conformation closely resembles the switch I loop structure observed in GMPPNP-loaded Rac1 alone and in the Rac1-POSH_319–371 complex. In addition to switch I loop closure, POSH_319–348 binding promotes extensive stabilization of the switch II region of Rac1b, which is nearly fully resolved in the electron density map. Switch II adopts a short helical turn followed by a coil structure, a conformation stabilized by hydrogen bonds between the backbone amide of G60 and the γ-phosphate of GMPPNP, between the side chains of E62 and D65 and the side chain of K115, and between the backbone amide of L67 and the side chain of H123 of Rac1b (Figure E). To further explore the structural ensembles accessible to the insertion of Rac1b within the complex with POSH_319–348, we used a coil generator to sample this segment while keeping the remainder of Rac1b fixed to its GMPPNP-bound crystal structure. The resulting ensemble reveals extensive conformational flexibility, suggesting that it is the dynamic behavior of the insertion that precludes the interaction of MRE2 with Rac1b through steric interference (Figure S2).

In Rac1, the residue Q61 plays a crucial role in GTP hydrolysis by orienting and activating a water molecule for a nucleophilic attack on the γ-phosphate of GTP. Thus, it has been shown that mutations of this residue impair or abolish GTP hydrolysis. The side chain carbonyl and amide groups of Q61 facilitate hydrolysis by forming hydrogen bonds that increase the nucleophilicity of the attacking water. In the Rac1b-POSH_319–348 structure, unlike the crystal structure of Rac1b alone, Q61 is resolved in the electron density. However, its side chain carbonyl and amide groups are positioned at 3.6 and 4.6 Å from the catalytic water, respectively, distances that exceed typical hydrogen-bonding ranges (Figure F). This indicates that, although POSH_319–348 stabilizes the switch II region and renders Q61 ordered, it does not position Q61 correctly to support hydrolysis. Collectively, our crystal structure of Rac1b bound to POSH_319–348 suggests that POSH_319–348 stabilizes a signaling-competent, but hydrolysis-incompetent, conformation of Rac1b.

POSH Binds to Rac1b via a Structurally Distinct Folding Intermediate

To investigate how the intrinsic dynamics of Rac1b influence the interaction kinetics and folding rate of POSH_315–380 on the surface of Rac1b, we measured ¹⁵N chemical exchange saturation transfer (CEST) experiments of POSH_315–380 at 25 °C, using three different B ₁ saturating fields (9.2, 19.3, and 38.9 Hz), in the presence of Rac1b at a 20% molar ratio. The major state observed in the CEST profiles therefore corresponds to free POSH_315–380 and the minor state to the Rac1b-bound conformation of POSH_315–380. The CEST profiles across the POSH_315–380 sequence reveal that only residues within MRE1 display detectable minor states, while residues within MRE2 show only the major state (Figure S3). This confirms that MRE2 does not interact with Rac1b, consistent with the NMR signal intensities and the ¹⁵N R _1ρ relaxation rates of POSH_315–380 measured in the presence of Rac1b (Figure B,C).

The CEST data were analyzed using both 2-site and 3-site exchange models. The 2-site model describes a direct interconversion between free POSH_315–380 and a Rac1b-bound state (Figure A). In contrast, the 3-site model introduces a folding intermediate (I) in which POSH_315–380 is initially anchored to Rac1b followed by folding on the Rac1b surface at a rate that contributes to the observed CEST profiles (Figure B). These two exchange models represent simplifications of the linear 4-site model recently employed to describe the hierarchical folding trajectory of POSH_315–380 upon binding to Rac1, where folding of MRE2 accounts for an additional folding step. For each exchange model (2- and 3-site), we performed a global analysis of the CEST profiles across all residues in POSH_315–380 displaying minor states and across all three B ₁ fields. To ensure robustness of the data analysis, the dissociation constant at 25 °C was fixed at K _d = 336 μM, based on extrapolation of the Gibbs free energy derived from ITC measurements at 5 and 35 °C (Figure H). Direct determination of the binding affinity at 25 °C by ITC was not possible, as the binding enthalpy at this temperature is close to zero.

5.

Analysis of the experimental ¹⁵N CEST data of POSH_315–380 with Rac1b using 2- and 3-site exchange models. (A) Schematic representation of the linear 2-site exchange model used for analysis of the ¹⁵N CEST data. (B) Schematic representation of the linear 3-site exchange model. (C) Analysis of ¹⁵N CEST data of POSH_315–380 at a 20% molar ratio of Rac1b using a 2-site exchange model. Experimental data (red circles) were globally analyzed across all residues and all B ₁ fields (red lines). Selected CEST profiles for residues in MRE1 are shown (B ₁ = 9.2 Hz). Vertical gray lines represent the chemical shifts of the free state of POSH_315–380 (solid line) and the final complex (dashed line). Gray stars indicate resonances that are aliased into the spectral window. Data points shown in pink were excluded from the analysis (see Methods). (D) Analysis of ¹⁵N CEST data of POSH_315–380 at a 20% molar ratio of Rac1b using a 3-site exchange model. Experimental data (red circles) were globally analyzed across all residues in MRE1 and all B ₁ fields (red lines). Selected CEST profiles for residues in MRE1 are shown (B ₁ = 9.2 Hz). Vertical gray lines represent the chemical shifts of the free state of POSH (solid line), intermediate I (dash-dotted line) and the final complex (dashed line). Gray stars indicate resonances that are aliased into the spectral window. Data points shown in pink were excluded from the analysis (see Methods). (E) Chemical shift difference, Δω_FI, between intermediate I and the free state F of POSH_315–380 obtained from the analysis of the ¹⁵N CEST data using a 3-site exchange model. The secondary structure elements observed in the crystal structure of the Rac1-POSH_319–371 complex are shown below in gray. (F) Chemical shift difference, Δω_IC, between the complex C and intermediate I obtained from the analysis of the ¹⁵N CEST data using a 3-site exchange model.

The CEST profiles fit reasonably well to the 2-site exchange model (Figures C and S4), however, data fitting improves noticeably when using the 3-site model (Figures D and S5). This improvement is consistent across residues in MRE1 of POSH_315–380 (Figure S6A), and an F-test confirms that the 3-site model provides a statistically significant improvement in data fitting compared to the 2-site model (p < 0.0001, Figure S6B). To evaluate how well the experimental CEST data can determine the exchange parameters, we performed grid searches. First, we conducted one-dimensional searches on individual parameters (k _off, k _IC, and k _CI) describing the 3-site exchange model (Figure S6C). We then performed two-dimensional grid searches to examine correlations between the exchange parameters (Figure S6D). These analyses indicate that the parameters are well-determined by the data and show no significant correlations with each other, supporting the robustness of the 3-site model. Finally, the uncertainties of the 3-site exchange parameters were estimated using a bootstrap analysis (Table S3). Importantly, the excellent fit quality (reduced χ² = 1.3, Table S3) obtained when fixing the minor-state populations using the ITC-derived K _d value supports the thermodynamic consistency between the two independent techniques.

Our CEST data analysis supports a folding-upon-binding mechanism of POSH_315–380. In the first step, POSH_315–380 forms a folding intermediate (I) with Rac1b, characterized by rate constants k _on = 1.28 × 10⁵ M^–1s^–1 and k _off = 166 s^–1. The chemical shift differences, Δω_FI, between the free state and the intermediate I, demonstrate that the partial CRIB motif within POSH_315–380 serves as the initial anchoring site to Rac1b (Figure E). Thus, significant chemical shift perturbations are observed only for residues 325–333 surrounding the partial CRIB motif, while the remainder of MRE1 and all of MRE2 remain fully flexible, showing negligible chemical shift changes. In the second step, the rest of MRE1 folds on the surface of Rac1b characterized by rate constants k _IC = 141 s^–1 and k _CI = 49 s^–1, while MRE2 remains dynamic. This folding event is supported by the chemical shift differences, Δω_IC, between the intermediate I and the final complex C, which show pronounced perturbations across most residues of MRE1 (Figure F).

Taken together, our ITC, crystallographic, and NMR data support a model in which POSH_315–380 binding to Rac1b proceeds through a folding-upon-binding pathway that sequentially stabilizes the switch regions (Figure A). In the intermediate state, the partial CRIB motif within MRE1 anchors POSH_315–380 to Rac1b in close proximity to the switch I loop, likely inducing closure of this loop, while leaving the switch II and insertion flexible. In the subsequent step, folding of the remainder of MRE1 on the Rac1b surface locks switch I into a stably closed conformation and concomitantly stabilizes the switch II region, while the insertion remains dynamic. Thus, POSH_315–380 binding not only involves its own stepwise folding but also drives the progressive structural ordering of the switch regions of Rac1b.

6.

Intrinsic dynamics of Rac1b reduce the on-rate and enhance the off-rate of POSH_315–380 binding compared to Rac1. (A) Folding-upon-binding mechanism of POSH_315–380 to GMPPNP-loaded Rac1b derived from CEST experiments, highlighting how the conformational dynamics in the switch regions and insertion are altered upon POSH_315–380 binding. (B) Previously established folding-upon-binding mechanism of POSH_315–380 to GMPPNP-loaded Rac1, in which POSH_315–380 folds via two structurally distinct intermediates (A and B) before reaching the final bound state C.

Discussion

Rac1 has long been recognized as a central regulator of cytoskeletal reorganization and MAPK signaling, acting through precise cycles of GTP binding and hydrolysis. In contrast, Rac1b, which harbors a 19-amino acid insertion immediately following the switch II region, is constitutively GTP-loaded yet signaling-deficient. Elucidating the molecular mechanisms underlying this deficiency has important implications not only for cellular signaling, but also for the broader question of how intrinsic protein dynamics affect interactions with molecular partners. In this work, we have used a combination of ITC, X-ray crystallography, and NMR spectroscopy to delineate how the intrinsic dynamics of Rac1b reshape the molecular recognition trajectory of the signaling effector POSH.

Previous studies of POSH_315–380 binding to Rac1 have revealed a hierarchical folding trajectory in which POSH_315–380 transitions from a disordered state to a bound conformation (C) through two structurally distinct folding intermediates (A and B). The partial CRIB motif serves as the initial anchoring site (k _on = 7.28 × 10⁵ M^–1s^–1 and k _off = 56 s^–1), enabling sequential folding of MRE1 (k _AB = 72 s^–1 and k _BA = 64 s^–1) and MRE2 (k _BC = 89 s^–1 and k _CB = 94 s^–1), resulting in a fully structured effector complex (Figures C and B). In contrast, our present data show that Rac1b truncates this pathway involving only a single folding intermediate in which the partial CRIB motif of POSH_315–380 anchors to Rac1b, followed by folding of MRE1 on its surface, while MRE2 remains disordered in the complex (Figure A). The Rac1b-POSH_315–380 complex is characterized by a ∼5.7-fold slower association rate constant (k _on) and a ∼3-fold faster dissociation rate constant (k _off) compared to Rac1, quantitatively accounting for the reduced binding affinity of the Rac1b-POSH_315–380 complex. The slower on-rate is consistent with the reduced ability of POSH_315–380 to engage with Rac1b due to the open switch I loop and potential steric hindrance from the extensive conformational sampling of the switch II and insertion in Rac1b. The faster off-rate is consistent with the formation of weaker hydrogen bonds and/or hydrophobic interactions being established in the CRIB-anchored state for Rac1b compared to Rac1, most likely due to the intrinsic dynamics of both switch regions. Following the formation of the CRIB-anchored state of POSH_315–380 with Rac1b, folding of the remainder of MRE1 occurs on a slow time scale with rates comparable to the complex with Rac1.

Despite the highly dynamic nature of Rac1b in isolation, our crystallographic data demonstrate that POSH_319–348 binding induces closure of the switch I loop and stabilization of the switch II region. These changes recapitulate aspects of the Rac1-bound conformation and show that effector engagement can impose order on otherwise flexible regions, although associated with an entropic barrier. Stabilization is however incomplete as the insertion remains highly flexible, and crucially, the side chain of Q61 is misaligned, preventing proper activation of the catalytic water molecule required for nucleophilic attack on the γ-phosphate of GTP. Thus, while POSH_319–348 binding stabilizes Rac1b in a signaling-competent conformation, it fails to reestablish the structural prerequisites for efficient GTP hydrolysis, explaining why Rac1b remains GTP-loaded.

Together, these findings have several broader implications. First, they provide a mechanistic explanation for why Rac1b displays impaired effector recognition despite being constitutively GTP-loaded. The intrinsic dynamics of Rac1b impose kinetic and entropic barriers that reduce effector affinity and prevent complete folding-upon-binding. Second, our study illustrates how alternative splicing can rewire the binding properties of a signaling protein by destabilizing critical recognition events. Similar principles likely apply to disease-associated mutations or post-translational modifications that can alter the dynamics of regulatory regions. In such cases, the outcome is not simply a loss or gain of binding, but a more nuanced remodeling of recognition trajectories, folding landscapes, and kinetic parameters. This insight is especially relevant to cellular signaling networks, where interaction dynamics and kinetics represent an often-overlooked layer of regulation.

Conclusion

In conclusion, our work defines the structural and kinetic basis for the signaling deficiency of Rac1b. By integrating structural and kinetic approaches, we show that Rac1b dynamics truncate the hierarchical folding trajectory of POSH_315–380, destabilize effector binding, and impose entropic and kinetic penalties that explain the reduced affinity relative to Rac1. These findings reinforce the idea that enhanced intrinsic dynamics can impair, rather than enhance, effector recognition, offering a conceptual link between conformational plasticity and defective signaling. Beyond Rac1b, our results highlight a general mechanism by which alternative splicing, mutations, or post-translational modifications may tune signaling networks by reshaping protein dynamics, ultimately determining the outcome of protein–protein interactions in the cell.

Methods

Protein Expression and Purification

Human POSH_315–380 (UniProt Q7Z6J0, residues 315–380) was subcloned into a pET-28a-derived vector with an N-terminal thioredoxin (Trx)-6xHis-tag for improved expression and a TEV cleavage site. The protein was expressed and purified as described previously. Human Rac1 (Uniprot P63000, residues 1–177), Rac1b (Uniprot P63000–2, residues 1–196) and the Rac1b_1–196-POSH_319–348 fusion construct were subcloned into a pESPRIT vector containing an N-terminal 6xHis-tag and a tobacco etch virus (TEV) cleavage site. The proteins were expressed and purified following the same protocol as Rac1, described previously. Following purification, Rac1 and Rac1b were found in the GDP-loaded form and a nucleotide exchange reaction was carried out to load the protein with the nonhydrolyzable GTP analogue GMPPNP (Guanosine 5′-[β,γ-imido]­triphosphate trisodium salt hydrate powder, Sigma-Aldrich), as described previously.

Chemical Shift Titrations and Relaxation Measurements

The chemical shift assignment of POSH_315–380 was obtained from the Biological Magnetic Resonance Bank (BMRB) accession code 52979. All NMR experiments were carried out at 25 °C using protein samples in 50 mM sodium phosphate buffer pH 6.0, 150 mM NaCl, 5 mM MgCl₂, 10 mM dithiothreitol (DTT) with added protease inhibitor (Roche cOmplete) and D₂O (7% v/v). Chemical shift titrations of POSH_315–380 with GMPPNP-loaded Rac1b were performed by recording a series of ¹H–¹⁵N HSQC spectra of ¹⁵N POSH_315–380 (200 μM) with 0, 45, 76, and 100% (molar ratio) of Rac1b. In addition, ¹⁵N R _1ρ relaxation rates were measured at 25 °C and at a ¹H frequency of 700 MHz for free POSH_315–380 and with 45% (molar ratio) of both Rac1 and Rac1b using HSQC-based pulse sequences. A spin lock field of 1.5 kHz was used with the following relaxation delays: 1, 10, 30, 50, 70, 90, 130, 170, 210, and 250 ms. A technical replicate at 70 ms was recorded to estimate the uncertainty on the relaxation rates using a Monte Carlo approach.

Isothermal Titration Calorimetry

ITC measurements were performed using a MicroCal PEAQ-ITC calorimeter (Malvern Instruments). Protein samples were dialyzed against ITC buffer containing 50 mM HEPES pH 6.9, 150 mM NaCl, 5 mM MgCl₂, 2 mM TCEP. Duplicate titrations of POSH_315–380 (1000 μM) into GMPPNP-loaded Rac1b (250 μM) were carried out at 5 and 35 °C. Results from five replicate injections into the same Rac1b cell content were merged by concatenation using the Malvern MicroCal Concat software (5 × 7 injections of 5.6 μL each with 180-s intervals, stirring speed 750 rpm). All ITC data were analyzed using the “One set of sites” binding model within the PEAQ-ITC analysis software. The uncertainty in the thermodynamic parameters was obtained from the standard deviation of the duplicates. The contribution from conformational changes, ΔS _conf, to the total binding entropy was determined from an empirical equation developed by Spolar and Record and recently reparameterized for IDP complexes:

$$\mathrm{\Delta }{S}_{\mathrm{conf}}=-1.66\mathrm{\Delta }{C}_{\mathrm{p}}\ln \left(\frac{T_{\mathrm{S}}}{386}\mathrm{K}\right)+110\mathrm{J}·{\mathrm{mol}}^{-1}·{\mathrm{K}}^{-1}$$ 1

where ΔC _p is the change in heat capacity at constant pressure (derived from the change in ΔH with temperature) and T _S is the iso-entropic temperature.

CEST Experiments

The ¹⁵N CEST experiments of POSH_315–380 (500 μM) were acquired with a 20% molar ratio of Rac1b and at a ¹H frequency of 600 MHz at 25 °C. To accelerate data acquisition, we used a DANTE multifrequency irradiation scheme (D-CEST) with three different saturating fields (B ₁ = 9.2, 19.3, and 38.9 Hz with DANTE windows of 450, 800, and 1500 Hz) using a modification of the published pulse sequence with a semiconstant time chemical shift evolution in the indirect dimension. , The frequency of the saturating field was varied with a step size of 8, 20, and 30 Hz for spectra recorded with windows of 450, 800, and 1500 Hz, respectively, and all D-CEST spectra were acquired with a saturation period of 0.4 s. To avoid contributions to the CEST profiles from solvent exchange of the amide protons of POSH_315–380, we employed an external lock where a 3 mm NMR tube containing the sample was placed inside a 5 mm tube with D₂O.

Analysis of the CEST Data

The CEST data were analyzed using the program ChemEx, implementing two distinct exchange models. The first model is a 2-site exchange model, assuming that POSH transitions directly from its free state (F) to its final bound conformation with Rac1b (C):

$$\mathrm{POSH}(\mathrm{F})+\mathrm{Rac}1\mathrm{b}\begin{array}{c}{k}_{\mathrm{on}}\\ \rightleftarrows \\ k_{\mathrm{off}}\end{array}\mathrm{POSH}(\mathrm{C})\text{:}\mathrm{Rac}1\mathrm{b}$$ 2

The second model is a linear 3-site exchange model, assuming that POSH transitions through a structurally distinct intermediate (I) before adopting its final bound conformation:

$$\mathrm{POSH}(\mathrm{F})+\mathrm{Rac}1\mathrm{b}\begin{array}{c}{k}_{\mathrm{on}}\\ \rightleftarrows \\ k_{\mathrm{off}}\end{array}\mathrm{POSH}(\mathrm{I})\text{:}\mathrm{Rac}1\mathrm{b}\begin{array}{c}{k}_{\mathrm{IC}}\\ \rightleftarrows \\ k_{\mathrm{CI}}\end{array}\mathrm{POSH}(\mathrm{C})\text{:}\mathrm{Rac}1\mathrm{b}$$ 3

Uncertainties in the CEST profiles were estimated using the “scatter” option in ChemEx, and data points within ±B ₁ (in Hz) of the major state frequency were excluded from the analysis (indicated as pink points in relevant figures), as explained previously. In or near the slow exchange regime, separating the populations (p _C in the 2-site exchange model, and p _I + p _C in the 3-site exchange model) from the rate constants can be difficult. To ensure a robust data analysis and improve the precision of the extracted kinetic rate constants, we therefore fixed the populations to those calculated from the complex dissociation constant at 25 °C derived from the ITC experiments (Figure H) and the concentrations of the proteins used in the CEST experiments ([POSH_315–380] = 500 μM and [Rac1b] = 100 μM). This approach reduces the number of fitted parameters while maintaining thermodynamic consistency between the ITC and CEST data sets.

In addition, for the 3-site exchange model, the transverse relaxation rate of the intermediate state, R _2I, was assumed to be equal to that of the fully bound state, R _2C, when the signals from the two states (I and C) overlapped in the minor-state peak of the CEST profile. This applied to residues S325, S327, V330, L331 and S333. For both the 2-site and 3-site models, all residues with CEST profiles having minor states were analyzed simultaneously and across all three B ₁ field strengths. Data from residues I326 and L346 were excluded due to spectral overlap.

Uncertainties in the globally fitted exchange parameters were estimated using a bootstrap analysis, as described previously. A total of 1000 bootstrapped data sets were generated and globally fitted using the 3-site exchange model. The uncertainty of each global parameter was defined as the 68% confidence interval of its bootstrap distribution.

Crystallization of the Rac1b-POSH_319–348 Fusion Complex Bound to GMPPNP

High-throughput crystallization screening using sitting drop vapor diffusion was performed at the crystallization (HTX) platform at the European Molecular Biology Laboratory (EMBL), Grenoble. Screening was performed with the GMPPNP-loaded Rac1b-POSH_319–348 fusion construct at 4–20 mg/mL in 20 mM Tris–HCl pH 8.0, 150 mM NaCl, 5 mM MgCl₂, 2 mM DTT, 1 mM GMPPNP. Standard screens (Wizard I + II, Rigaku; Salt-Grid, Hampton; JCSG, Molecular Dimensions; PACT, Molecular Dimensions; PEGs I, Qiagen; and Classics Suite, Qiagen) and screens optimized specifically to promote protein complex formation (Morpheus and ProPlex; Molecular dimensions) at 4 and 20 °C were used, thereby testing a total of 1536 different conditions. Successful crystallization conditions were optimized manually (24-well plates, hanging drop) around the ProPlex D2 condition (8–14% 1-propanol; 8–14% PEG 5000 monomethyl ether (MME); 0.1 M MES pH 6.0, 6.5 and 6.8). Two concentrations (6 and 12 mg/mL) were tested for each condition with a protein:precipitant solution ratio of 1:1 (2 μL drops). The best diffraction data were obtained from crystals with 6 mg/mL Rac1b-POSH_319–348 fusion complex with 14% 1-propanol, 10% PEG 5000 MME, 0.1 M MES pH 6.8 as the precipitant solution grown at 20 °C. Crystals were harvested by hand, cryoprotected with 20% glycerol and diffracted on the MASSIF-1 beamline equipped with a Pilatus detector (Dectris) at the European Synchrotron Radiation Facility (ESRF, Grenoble, France). The structure was solved at 1.77 Å resolution.

Processing and Refinement of X-ray Crystallography Data

Data processing was performed with autoPROC (version 1.1.7, Global Phasing Ltd.), including the STARANISO program for anisotropic data analysis. Phases were found by molecular replacement with Phaser, using as a search model the structure of the Rac1-POSH_319–371 fusion complex (PDB: 9RFF) after deletion of the region containing the MRE2 β-hairpin motif (residues 353–369 of POSH). The initial solution was improved through cycles of manual adjustment in WinCoot (version 0.9.8.1 or 0.9.8.95) and refined using Refmac5 (version 5.8.0411 or 5.8.0425). The peptide chain of POSH_319–348 and the switch regions of Rac1b were manually built and/or readjusted. Regions of poor electron density were left unmodeled. Phaser and Refmac were all used as programs of the CCP4 suite (version 8.0.008 or successive versions) and crystallography applications were compiled and configured by SBGrid.

Protein structure data of the Rac1b-POSH_319–348 fusion complex have been deposited in the PDB database with accession code 9TJY.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.6c04072.

• Data collection and refinement statistics of the POSH-Rac1b crystal structure, conformational ensemble analysis of the Rac1b insertion, and details on the analysis of the ¹⁵N CEST data and thermodynamic parameters from ITC experiments (PDF)

4.

Amoéba, 38 Avenue de Frères Montgolfier, 69680 Chassieu, France

The authors declare no competing financial interest.