Altered protein dynamics and a more reactive catalytic cysteine in a neurodegeneration-associated UCHL1 mutant

Abstract

A mutant of ubiquitin C-terminal hydrolase L1 (UCHL1) detected in early-onset neurodegenerative patients, UCHL1_R178Q, showed higher catalytic activity than wild-type UCHL1 (UCHL1_WT). Lying within the active-site pocket, the arginine is part of an interaction network that holds the catalytic histidine in an inactive arrangement. However, the structural basis and mechanism of enzymatic activation upon glutamine substitution was not understood. We combined x-ray crystallography, protein NMR analysis, enzyme kinetics, covalent inhibition analysis, and biophysical measurements to delineate activating factors in the mutant. While the crystal structure of UCHL1_R178Q showed nearly the same arrangement of the catalytic residues and active-site pocket, the mutation caused extensive alteration in the chemical environment and dynamics of more than 30 residues, some as far as 15 Å away from the site of mutation. Significant broadening of backbone amide resonances in the HSQC spectra indicates considerable backbone dynamics changes in several residues, in agreement with solution small-angle X-ray scattering (SAXS) analyses which indicate an overall increase in protein flexibility. Enzyme kinetics show the activation is due to a $k_{cat}$ effect despite a slightly weakened substrate affinity. In line with this, the mutant shows a higher second-order rate constant ($k_{inact}/K_i$) in a reaction with a substrate-derived irreversible inhibitor, Ub-VME, compared to the wild-type enzyme, an observation indicative of a more reactive catalytic cysteine in the mutant. Together, the observations underscore structural plasticity as a factor contributing to enzyme kinetic behavior which can be modulated through mutational effects.

Graphical Abstract

Introduction

Ubiquitin (Ub) is a post-translational modifier used in cellular signaling, protein degradation, immune response, and various other cellular processes.^1–3 Ub is covalently attached primarily to lysine residues of substrate proteins in a process involving three enzymes: the ATP-requiring E1 Ub-activating enzyme, E2 Ub-conjugating enzymes, and E3 Ub-ligating enzymes or E3 Ub ligases. Attachment of Ub to a substrate protein is further elaborated by successive addition of more Ub groups using one of the seven internal lysines (or the N-terminal Met1 amino group) of Ub itself to produce architecturally distinct types of poly-Ub chains, which serve as biological signals enabling specific outcomes for the tagged protein.³ A regulatory mechanism to poly-Ub chain formation involves its reversal at the hands of deubiquitinases (DUBs), which hydrolytically remove Ub from poly-Ub chains and/or completely from the tagged protein thereby regulating the signal while ensuring the recycling of Ub back to the free Ub pool in the cells.

DUBs belong to two mechanistic classes – cysteine proteases and metalloproteases. Within the cysteine-protease class, the eukaryotic DUBs are divided into six groups based on the structure of their catalytic domain: the Ub carboxy-terminal hydrolase (UCH) family, Ub-specific protease (USP) family, ovarian tumor protease (OTU) family, Machado-Joseph domain (MJD) family, motif interacting with Ub-containing novel DUB (MINDY) family and zinc finger with UFM1-specific peptidase domain protein (zUFSP) family^4–9, whereas the metalloprotease DUBs are represented solely by the zinc-dependent JAMM/MPN (JAMM)¹⁰ domain enzymes.¹¹

The UCH family consists of four members: UCHL1, UCHL3, UCHL5 and Bap1. These different UCH DUBs act with different rates despite high sequence and structural homology of their catalytic domains, and a conserved catalytic triad.¹² For example, UCHL1 has been shown to be significantly less active than UCHL3 even though the two share nearly 50% sequence identity, hence a high structural similarity, showing C-α root mean square deviation (rmsd) of 0.84 Å over the entire length of their polypeptide chains.^13,14 This difference in their activity may be attributed to the misaligned catalytic residues in UCHL1, in which the Cys-His pair is separated by ~8 Å, a distance too far for the histidine to effectively activate the catalytic cysteine.¹⁵ Upon binding to Ub, the catalytic residue pair in UCHL1 assumes a typical arrangement observed in productive cysteine proteases.¹⁶ In contrast, the His-Cys dyad in UCHL3 is pre-organized in an active arrangement having a distance of ~4 Å, typically observed in productive cysteine proteases.^13,17 The requirement for the Ub binding to induce productive active-site conformation in UCHL1 as opposed to a constitutively preorganized active site in UCHL3 could at least in part explain the difference in their catalytic rates. It also underscores the importance of conformational plasticity and protein dynamics as a factor in UCHL1’s enzymatic function.

The importance of UCHL1 in the biological system remains elusive even though it is historically one of the oldest DUBs to have been identified. While earlier studies pointed to its protective role in the central nervous system (CNS), of late, its expression in other tissues has been linked to a variety of cancers.^18–21 UCHL1 is predominantly expressed in neurons, accounting for more than 1–2% of the total soluble protein content in the brain.²² Its neuroprotective function may be related to its role in preventing neuronal proteins from aggregating and regulating mono-Ub levels in neurons.^23–27 The UCHL1 protein level is reduced in the hippocampus of an Alzheimer’s disease (AD) mouse model and its expression has been shown to rescue cognitive function associated with beta-amyloid toxicity.²⁸ It is also linked to Parkinson’s disease, although its exact role is unclear. Abnormal expression of UCHL1 has been associated with Parkinson’s disease ^29,30 In another study, the role of UCHL1 in Parkinson’s Disease (PD) was linked to its regulation of the pyruvate kinase, where the loss of UCHL1 was shown to diminish ATP and pyruvate levels, which alleviates PD pathology.³¹

Mutations of UCHL1 have been found in neurodegenerative disease patients (Fig. S1). The I93M mutation has been found to lead to an increased risk of PD through a reduced hydrolase activity, which results in decreased aberrant protein regulation.³² A polymorphic variant, S18Y, is associated with the reduced risk of developing PD.^33–36 Recently, whole-exome sequencing of two Turkish siblings born of a consanguineous union and affected with progressive neurodegeneration, such as childhood-onset blindness, cerebral ataxia, and other upper motor neuron dysfunctions, revealed homozygous missense mutation within the UCHL1 gene, E7A, which shows impaired Ub binding and, consequently, a loss of hydrolase activity.²⁵

More recently, whole-exome sequencing of a Norwegian monozygotic twin affected with similar early-onset neurodegenerative symptoms revealed compound heterozygous variants in UCHL1 with R178Q and A216D mutations.³⁷.³⁸ Unlike the Turkish case, however, the patients showed normal levels of cognition even in their late fifties. The A216D mutation likely caused a folding defect resulting from the placement of the charged aspartate in a hydrophobic region of the protein (Fig. S1). Indeed, while the R178Q variant (UCHL1_R178Q) was readily detectable in the patient fibroblasts, albeit at a reduced level compared to the cells from a sibling carrying the wild type alleles, the A216D mutant was undetectable at the protein level despite comparable transcript levels. Intriguingly, UCHL1_R178Q showed higher catalytic activity than wild-type UCHL1 (UCHL1_WT). The authors proposed that A216D could be a loss-of-function mutant causing neurodegeneration while the higher activity of UCHL1_R178Q could be playing a neuroprotective role in maintaining cognition in these patients despite the neurodegenerative syndromes. In this manuscript, we sought to characterize this mutant using structural and biochemical analysis to better understand the mechanism of catalytic activation in the mutant. We started with the notion that the increased activity of the mutant may be due to a better aligned catalytic in UCHL1. We therefore sought to crystallize the mutant and perform further biochemical studies guided by the structure. We also measured the binding affinity of the UCHL1 variant to Ub through biolayer interferometry to determine whether binding affinity plays a role in this increased activity of UCHL1_R178Q. We find that the effect of the mutation is more complex than shown by the static snapshot of the crystal structure, rather manifesting in unexpected changes in protein dynamics as indicated by solution state NMR studies.

Results

UCHL1_R178Q crystallized in the P42₁2 space group with two copies of the protein in the asymmetric unit. The structure was solved by molecular replacement (MR) using the structure of the UCHL1 (PDB ID: 2ETL)¹⁵ monomer as the search model. Multiple rounds of refinements of the MR solution showed interpretable electron density in the $2\text{m}{F}_O-\text{D}{F}_C$ map corresponding to the Arg-to-Gln mutation at the 178^th position in the protein. Crystallographic data collection and refinement parameters are listed in Table S1. Further refinement produced the final model with an $R_{\text{work}}/R_{\text{free}}$ of 0.222/0.264. Three sulfate ions along with 110 ordered water molecules were also included in the final model. The electron density corresponding to the cross-over loop (L8), spanning from residues 151–155 of UCHL1_R178Q, was not observed in this structure and it was left out of the final model.

Superposition of the crystal structures of apo UCHL1_WT and UCHL1_R178Q reveals a near identical overall arrangement with a root mean square deviation (rmsd) of 0.27Å for Cα atoms (Fig. 1A). The overall dimeric arrangement of UCHL1_R178Q subunits in crystals is nearly identical to that UCHL1_WT, with the same interactions between the two protomers. However, similar to UCHL1_WT, UCHL1_R178Q is monomeric in solution, as indicated by similar migration pattern observed as WT UCHL1 in size-exclusion chromatography, which was further conformed by small angle X-ray scattering (see below).

Figure 1. Structure of UCHL1_R178Q.

(A) Comparison of the overall crystal structures of the mutant (pink) and the WT UCHL1 (green). (Inset) Superposition of the two structures in the active-site region showing the catalytic triad (in stick representation). Ordered water molecules in the active site area are shown as green (wt) and pink (mutant) spheres. (B) 2mFo-DFc electron density map corresponding to the Gln178 residue (contoured at σ = 1.0). (C) Surface Electrostatic potential map of the UCHL1 variants. (D) Hydrogen-bonding network surrounding the catalytic histidine of UCHL1_R178Q (left) and UCHL1_WT (right).

The mutation is readily accommodated in the active site of the enzyme without any discernible effect in the position of the surrounding residues (Fig. 1). The presence of the glutamine residue in the mutant does change the electrostatic environment in the active site resulting from the removal of a positive charge due to the arginine replacement (Fig. 1C). The substitution of Arg178 by Gln178 affects the hydrogen-bonding network in the active-site area. The hydrogen bond previously observed between the Asn159 with Arg178 in UCHL1_WT is absent in UCHL1_R178Q (Fig. 1D). In general, the structure of UCHL1 has not changed to any appreciable extent in the crystalline state, including other residues lining the active-site pocket. The water network around the active site has also largely remained unaffected.

Given the observation that the crystal structures of UCHL1_WT and UCHL1_R178Q are nearly identical to each other in the active-site region including the Cys-His dyad, we wanted to test if the altered dynamics of the protein in solution is a factor behind the increased activity of UCHL1_R178Q when compared to UCHL1_WT. We thought we could indirectly infer altered solution dynamics by comparing the stability of the proteins through measurement of melting temperature.^39–42 Proteins with higher melting temperatures are more stable, partly because of a more rigid arrangement of their structural elements, whereas more dynamic (flexible) variants are likely to be less tightly packed and may unfold more readily.⁴³ The melting temperatures of both UCHL1_WT and UCHL1_R178Q, determined by observation of temperature-dependent change in far-UV circular dichroism (CD) signals as the protein transitions from folded to unfolded state, were nearly identical (UCHL1_WT at 52.4 °C and UCHL1_R178Q at 50.9 °C, Fig. S2). Even though at least one hydrogen bond is lost due to the mutation (Fig. 1D), the thermal stability of the mutant remains practically unchanged. This implies the solution conformations of the proteins may not exactly reflect the static snapshot of the lowest energy state captured in the crystal structure.

The absence of any significant difference in thermal stability of the two UCHL1 variants means the R178Q mutation does not affect the global solution conformation in any drastic way. However, subtle, site-specific dynamics effects cannot be ruled out. The effect of R178Q mutation in solution structure and dynamics of UCHL1 was investigated by multidimensional heteronuclear nuclear magnetic resonance (NMR) spectroscopy. Superposition of the two-dimensional (2D) ¹⁵N-¹H heteronuclear single quantum coherence (HSQC) spectra of UCHL1_WT and UCHL1_R178Q showed significant spectral changes with many cross peaks exhibiting sizeable chemical shift perturbations (CSPs), indicating changes in the microenvironments around the amide groups of these effected residues. We recorded a set of backbone triple resonance NMR experiments to assign the chemical shifts of backbone amide hydrogen (H_N), nitrogen (N), carbonyl carbon (C’), and Cα and Cβ atoms of UCHL1_WT and UCHL1_R178Q. It transpired that more than 40 residues (excluding prolines) exhibited unfavorable dynamics potentially on the μs to ms timescale, leading to severe line broadening of the ¹⁵N-¹H correlations in the HSQC spectrum of UCHL1_R178Q and inability to complete the chemical shift assignments of the corresponding residues (Fig. 2A). Nevertheless, we calculated the composite backbone amide CSPs for those that can be unambiguously assigned (Materials and Methods). The composite CSPs were plotted as a function of sequence number and mapped onto the crystal structure of UCHL1_WT in complex with Ub (Fig. 2B and 2C). The chemical shift perturbations and changes in backbone dynamics were widespread across the entire structure of UCHL1, including regions that make direct contact with Ub, such as Gly210, Val212 and Phe214 that are more than 20 Å away from the R178Q mutation site (Fig. 2C). Additionally, α-helix 2 (α2), the beginning of α-helix 3 (α3) where the catalytic Cys90 lies, β-strands 2 and 3 (β2 and β3) and the loop connecting β-strands 4 and 5 exhibited severe line broadening. At the center of these substantially impacted regions is the R178Q mutation site.

Figure 2. Solution state NMR spectroscopy analysis of the impact of R178Q mutation on UCHL1.

(A) Overlay of ¹⁵N-¹H HSQC spectra of UCHL1_WT (orange) and UCHL1_R178Q (blue). Residues that exhibited significant chemical shift perturbations are indicated alongside the corresponding cross peaks; those that were broadened beyond detection in the spectrum of UCHL1_R178Q are indicated alongside the cross peaks of UCHL1_WT with underlines. The cross peak of UCHL1_R178Q is highlighted with a red circle and the cross peaks of the catalytic C90 and H161 were also significantly broadened, which are indicated in red fonts. (B) Chemical shift perturbation (CSP) as a function of residue number of UCHL1. CSP is defined as ${\left[{\text{Δδ}{(}^1\text{H})}^2+{\left(0.165\cdot \text{Δδ}{(}^{15}\text{N})\right)}^2\right]}^{1/2}$. Dashed line and long dashed line correspond to the mean CSP plus one and two standard deviations ($\sigma$), respectively. Residues that were broadened beyond detection are indicated by a gray bar. Residues that exhibited weak backbone connectivity in the backbone triple resonance experiments but had traceable ¹⁵N-¹H HSQC cross peaks based on the criterion of minimal chemical shift displacement without overlapping were included in the CSP analysis, and their values are highlighted in red. The positions of the individual secondary structure elements are indicated on the top of which the α-helices and β-strands are shown in red rectangles and yellow arrows, respectively. (C) Structural mapping of the CSP induced by the R178Q mutation. The backbone amide nitrogen atoms of the residues that exhibited 2σ and 1σ above the mean CSP are shown in orange spheres with two and half van der Waals radius, respectively; those that were broadened beyond detection are shown in semitransparent pink spheres with 0.75 van der Waals radius. The catalytic residues, C90, H161 and D176, are shown in ball-and-stick representation; R178 is shown in spheres; the carbon, nitrogen, oxygen and sulfur atoms are shown in white, blue, red and gold, respectively.

While the changes in protein dynamic induced by the R178Q mutation are intriguing, they also limited our ability to make quantitative descriptions regarding the changes in solution structure and dynamics by NMR spectroscopy. We therefore resorted to synchrotron-based small-angle X-ray scattering and wide-angle X-ray scattering (SAXS/WAXS) to investigate global structures and dynamics of the two variants. The solution structures of UCHL1_WT and UCHL1_R178Q deviated slightly from the respective crystal structures according to the comparison of the experimental SAXS/WAXS profiles and the back-calculated ones based on the crystal structures. Specifically, UCHL1_WT exhibited more differences in the high-q range suggesting differences in the side chain packing between the crystalline and solution states (Fig. 3A).⁴⁴ Nevertheless, UCHL1_WT and UCHL1_R178Q exhibited the same radius of gyration (${\text{R}}_{\text{g}}$) and overall global structures according to the pair-wise distance distributions, P(r) (Fig. 3B and 3C). Further examination of the Kratky plots revealed marked differences in the high-q range where UCHL1_R178Q exhibited an elevated tail that is indicative of the more pronounced flexibility compared to WT.⁴⁵ Collectively, the SAXS/WAXS data indicated that the R178Q mutation does not result in partial unfolding of the domain structure but the internal dynamics is increased, which may be responsible for the line broadening observed in the 2D ¹⁵N-¹H HSQC of UCHL1_R178Q (Fig. 2A). This is also in line with the CD thermal melt results (Fig. S2).

Figure 3. SAXS/WAXS analysis of UCHL1 variants.

(A) Comparison of experimental and back-calculated SAXS/WAXS profiles of UCHL1_WT and UCHL1_R178Q. The back-calculated SAXS/WAXS profiles were generated by the online FoXS server (https://modbase.compbio.ucsf.edu/foxs/) using the previously reported crystal structure of UCHL1_WT (PDB entry: 2ETL) and the crystal structure of UCHL1_R178Q reported in this study as inputs. The missing termini and loop residues were modeled by the built-in Modeller function within ChimeraX with default setting. Five models were built for UCHL1_WT and UCHL1_R178Q. The data points of UCHL1_WT were multiplied five-fold to nudge the overall profile upward for better separation from that of UCHL1_R178Q. The differences between the experimental and back-calculated data were estimated by the ${\text{χ}}^2$ values determined by FoXS for the five models of UCHL1_WT and UCHL1_R178Q. The average and standard deviation of the five ${\chi }^2$ values were shown. (B) Guinier plots and residuals of linear fits within the Guinier regions (bottom panel). The estimated ${\text{R}}_{\text{g}}$ values are shown on the upper right corner. (C) Pair-wise distance distributions, P(r), showed very similar overall structures of the UCHL1 variants. The estimated values of the maximum global dimension, ${\text{D}}_{\text{max}}$, are shown on the upper right corner. (D) Kratky plots of the UCHL1 variants showed a slightly more pronounced flexibility in UCHL1_R178Q evidenced by the more elevated tails in the q-range between 0.2 and 0.5.

How do changes in protein dynamics relate to higher catalytic activity? To assess the impact of the R178Q mutation on catalytic activity we performed Michaelis-Menten kinetics analysis of UCHL1-catalyzed hydrolysis of Ub-Rhodamine (Ub-Rho).⁴⁶ All kinetics data were compiled in Table S2. The $K_{\text{M}}$ for the UCHL1_WT was ~30 nM, consistent with previous reports²⁵, while UCHL1_R178Q showed a markedly increased $K_{\text{M}}$ of ~170 nM (Fig. 4A). In contrast, the $k_{cat}$ of UCHL1_R178Q was higher than that of UCHL1_WT by more than 5-fold (Fig. 4A). Thus, despite a weaker substrate affinity, the mutant is more active than UCHL1_WT mainly due to a $k_{cat}$ effect. We asked if the activation is caused by the removal of the arginine side chain from the active site that is resulting in relief of some sort of inhibitory effect of the side chain. To this end, we tested the activity of the Arg178-to-Ala mutant (Fig. 4B). Interestingly, the activity of the Arg-to-Ala mutant is also higher than the wild-type enzyme, suggesting that Arg178 may play a role in dialing down the enzymatic activity of UCHL1.

Figure 4. HA-WT-Ub-VME has increased inactivation efficiency with UCHL1_R178Q relative to WT UCHL1 and other mutants.

(A) Michaelis-Menten kinetics of UCHL1_R178Q and UCHL1_WT with Ub-Rho substrate ($K_{\text{M}}$ and $k_{cat}$ in Table S2). (B) Relative rate analysis performed on active site mutants of UCHL1. (C) Measurement of covalent adduct formation of Ub-VME with UCHL1 Cys90 to derive $k_{\text{inact}}/K_{\text{I}}$ values for UCHL1_WT, UCHL1_S18Y, and UCHL1_R178Q. Ub-Rho reaction progress curves were fitted with a non-linear equation (see Materials Methods) to extract $k_{obs}$ at different concentration of Ub-VME (right). Plots of $k_{obs}$ as a function of Ub-VME concentration were fitted with linear equations, slopes of which provided $k_{inact}/K_I$.

In the crystal structure of apo UCHL1_WT, Arg178 is seen in a hydrogen-bonding network interacting directly with Asn159, which in turn is hydrogen-bonded to Glu60, which seems to hold the histidine in its unproductive orientation (Fig. 1D). Moreover, Arg178 seems to intervene between the catalytic dyad by positioning itself in between the Cys-His pair. We suspected that the removal of this arginine might facilitate the catalytic dyad to rearrange more easily into a productive configuration by allowing His161 to come closer to Cys90. To test this notion, we mutated Glu60 to glutamine hoping to weaken its hold on His161, thereby activating the enzyme (Fig. 4B). The Glu-to-Gln mutant is expected to have a loss of charge effect without introducing a steric change in the environment of His161. This would lead to the removal of the electrostatic attraction between Glu60 and His161, hence increasing the dynamics of His161. This mutant however has a lower catalytic activity in the Ub-Rho hydrolysis assay than the wild-type enzyme (Fig. 4C). Once again, this shows the crystal structure may not reflect the exact arrangement in solution. Additionally, the glutamate may be playing a role in contributing to the active-site electrostatics that is perturbed with a neutral sidechain.

The higher $k_{cat}$ of the mutant indicates a more preorganized active site for better transition-state stabilization, or a more reactive cysteine in the mutant. To probe the cysteine reactivity factor, we used a Ub-based suicide inhibitor, Ub-VME (commercially available HA-tagged Ub-VME was used. HA: the human influenza hemagglutinin epitope), which carries a reactive electrophile at the C-terminus of Ub in place of the scissile peptide bond that would be targeted by the catalytic cysteine in a ubiquitinated protein substrate⁴⁷. Ub-VME reacts with the catalytic cysteine of DUBs in a similar manner to the hydrolysis reaction, except that the nucleophilic attack of the catalytic cysteine is stalled as a covalent complex. The resulting irreversible inhibition produces a thioether product that mimics the thioester acyl-enzyme intermediate arising during the hydrolysis reaction. We reasoned that the rate of covalent modification with this substrate-like inhibitor would provide a measure of the cysteine reactivity.^48,49 To this end, the irreversible covalent inhibition was characterized by determination of $k_{inact}/{\text{K}}_{\text{I}}$, where $K_{inact}$ is the rate constant of covalent reaction between UCHL1 and Ub-VME under saturating conditions and $K_l$ is the equilibrium dissociation constant of UCHL1 binding to Ub-VME. To this end, we performed a series of Ub-Rho hydrolysis assay for UCHL1 in presence of varying Ub-VME concentrations and extracted the apparent first-order rate constant ($k_{obs}$) by a nonlinear fit of the progress curves (Fig. 4C). The $k_{obs}$ for each Ub-VME concentration was plotted against the corresponding inhibitor concentration and the slope of the linear regression of these data points provided $k_{inact}/{\text{K}}_{\text{I}}$. The difference in $k_{inact}/{\text{K}}_{\text{i}}$ between UCHL1 and its mutant reflects both binding and relative reactivity of their catalytic cysteine. We observed the $k_{inact}/{\text{K}}_{\text{i}}$ value of UCHL1_R178Q to be ~50-fold higher than UCHL1_WT (Fig. 4C). We also performed the same analysis for UCHL1_S18Y, which is known to have a similar reactivity as UCHL1_WT. Indeed, the $k_{inact}/{\text{K}}_{\text{I}}$ for UCHL1_S18Y was similar to that of UCHL1_WT with a relatively modest increase in this parameter (8.41×10⁻³ M⁻¹s⁻¹ compared to 2.96×10⁻³ M⁻¹s⁻¹). Using biolayer interferometry (BLI), we observed that the binding affinity of UCHL1_R178Q for Ub is lower than that of UCHL1_WT by a factor of two (Fig. S3, Table S2). Thus, UCHL1_R178Q binds slightly weaker than UCHL1_WT, consistent with $K_{\text{M}}$ values from enzyme kinetics data, but has gained considerable reactivity of its catalytic cysteine towards a substrate-like irreversible inhibitor. Together these results indicate that despite having an identical arrangement of the catalytic residues in the crystal structure, UCHL1_R178Q exhibits substantial differences in solution dynamics, which somehow translate to a higher reactivity of the catalytic cysteine.

Discussion

The lower DUB activity of UCHL1 when compared to its closely related paralog UCHL3 can be due to the misaligned catalytic Cys-His pair in UCHL1.^13,50 In UCHL3, the catalytic histidine is within 3.8 Å of the catalytic cysteine, a distance corresponding to a weak hydrogen-bonding interaction between the $S^{\gamma }$ and N^δ-H atoms required for extraction of the thiol proton by the imidazole group, while the same pair of atoms in UCHL1 are 8.2 Å apart.^13,15 Ub binding to UCHL1 induces conformational changes leading to alignment of the Cys-His dyad into a productive arrangement where the distance becomes close to 4 Å, primarily through the movement of the histidine induced by interactions of Ub at regions distal from the active site. Ub binds UCHL1 at two major sites: with its C-terminal tail at the active site and its N-terminal β1-β2 loop binding at a surface patch 17 Å away from the catalytic cysteine. The distal binding by the Ub β1-β2 loop ‘pushes’ a surface exposed phenylalanine (Phe214) into the protein interior of UCHL1, which triggers an internal conformational relay ultimately bringing the histidine next to the catalytic cysteine. These observations underscore conformational plasticity of UCHL1 in modulating its catalytic activity. Any factor, mutation or otherwise, that alters the protein dynamics is expected to affect the catalytic rate.

We showed that the crystal structure of UCHL1_R178Q remained largely similar to UCHL1_WT. The water hydrogen-bonding network in the active site also remained similar, hence the water network does not seem to contribute to the increased activity. The hydrogen-bonding network of residues in the vicinity of the catalytic site does change with the removal of the arginine, specifically, the one between Asn159 and Arg178 in UCHL1_WT is lost upon mutation. Before we solved the structure, we expected that the loss of the hydrogen bond may weaken interactions holding the catalytic histidine in its misaligned position, so in the mutant the active-site residues would be already aligned prior to Ub binding, preorganized like UCHL3. Clearly, that’s not the case. The structure of the mutant shows that the same arrangement can be maintained in the crystals despite the loss of the hydrogen bond, thus indicating that the effect of the mutation likely manifests at the level of solution dynamics. In line with this notion, we observed substantial changes in the environment and dynamics of as many as 30 UCHL1_R178Q residues, some even 20 Å away from the site of mutation, including Phe214. The increase in dynamics of many residues deduced from NMR analysis is consistent with higher overall protein flexibility of the mutant as derived from the SAXS analysis. However, we are unable to map the exact mechanism of activation resulting from the increased dynamics. One possible factor contributing to the catalytic activation could be different protonation states of the catalytic cysteine and surrounding active site residues in UCHL1_WT compared to UCHL1_R178Q, a subject of future investigations using computational means.

Compared to the Parkinson’s disease-associated I93M mutation (UCHL1_I93M), which also has essentially the same crystal structure as that of UCHL1_WT, the observed CSPs in UCHL1_R178Q were much smaller than those observed in UCHL1_I93M.⁵¹ Nevertheless, UCHL1_I93M does not significantly exhibit significant NMR line broadening as observed in UCHL1_R178Q. Most of the backbone amide ¹⁵N-¹H correlations in UCHL1_I93M could be assigned to make pairwise CSP analysis with UCHL1_WT. Note that the loop preceding the C-terminal β-strand 6 (β6) exhibited the largest CSPs (Fig. 2B). Phe214 is located in the loop and it is implicated in a cascade of side chain rearrangements upon Ub binding to enable appropriate side chain alignment within the active site.⁵² It is plausible that the R178Q mutation allosterically modulates the side chain dynamics and solution structures of the residues involved in the cascade of conformation rearrangements to enable more efficient side chain alignment compared to UCHL1_WT and therefore better enzyme efficiency. Disease-associated mutations in UCHL1 and its paralog, BRCA 1-assocaited protein 1 (BAP1) generally lead to reduced or loss of DUB activity, accompanied by reduced folding stability and loss of compact structures.^40,51 In particularly, several cancer-associated mutations in BAP1 result in a significantly more expanded global dimension of the UCH domain and allosterically perturbation of the integrity of the local structures.⁴⁰

The higher catalytic activity of the mutant is due to an increased $k_{cat}$ despite a modest loss in substrate affinity indicated by a slightly higher $K_{\text{M}}$, which is consistent with the position of the arginine residue relative to Ub binding site; It is not used in substrate binding. The Ub-VME inhibition study showed a substantial $k_{inact}/K_i$ increase in the mutant compared to the wild-type enzyme and the difference is mainly due to an increase in the rate of covalent reaction, with a relatively minor effect on Ub binding affinity (BLI results). Thus, the mutation has increased the apparent reactivity of the catalytic cysteine. However, the increase is disproportionately higher than what was observed in the $k_{cat}$ value, suggesting that the effect of the mutation on enzymatic activity is more complex than just the rate of covalent inhibition. Nevertheless, our studies show that enhanced catalytic activity is at least partly contributed by a more reactive catalytic cysteine in the mutant. We speculate that the rather extensive changes in dynamics observed in NMR indicate that the transition to a catalytically productive state is more facile in the mutant due to increased protein flexibility. While the conformational relay leading to active-site alignment starts at the surface exposed Phe214, it eventually involves residues changing positions within the tightly packed interior of the protein. Having a more flexible interior may mean a lower energy penalty for such a conformational transition.

As mentioned before, the patients carrying the R178Q mutation also have another loss of function allele in the second copy of UCHL1 in which alanine at position 216 has changed to an aspartate. The A216D mutant was quickly degraded because it likely caused a folding defect in the protein. It seems the loss of UCHL1 function resulting from this mutation could have had a much more drastic effect if it had not been for the higher catalytic activity of the R178Q allele. Biallelic variants in UCHL1 have previously been reported to aid in the progression of early onset neurodegenerative disorder⁵³, however the synergy of heterozygous loss of function variant A216D and activating variant R178Q highlights the neuroprotective effect of the activating mutant. It has been proposed that the activating mutant may serve a neuroprotective function explaining the normal cognition levels in the patients even in their late fifties. This raises an intriguing possibility of improving the catalytic activity of UCHL1 for better cognition in patients with neurological disorders. An implication of this is the possibility of finding small molecule activators of UCHL1 as novel therapeutics to provide cognitive improvement in neurodegenerative diseases.

Materials and Methods

Protein Expression and Purification

UCHL1 variants were purified as previously described.⁵⁴ Briefly, pGEX-6p-1 (GE Healthcare) carrying UCHL1 was transformed into BL21 (DE3) Escherichia coli (E. coli) cells (Novagen) and protein production was induced with 250 μM Isopropyl ß-D-1-thiogalactopyranoside (IPTG). The resulting proteins was then purified using a two-step purification method: GST-affinity chromatography followed by size-exclusion chromatography on a Superdex 75 column. Protein purity assessed by SDS-PAGE analysis at each stage of the purification.

Recombinant double His-tagged UCHL1 in pET-15b was expressed in BL21 (DE3) E. coli strain. 1 L of LB media was inoculated with 1% of the overnight inoculum and allowed to grow until an OD₆₀₀ of ~0.6. Upon induction with IPTG, cells were incubated at 18 °C for 16–18 hours to enable protein expression. The cells were then harvested and resuspended in 1X PBS buffer supplemented with 0.4 M KCl. The cell resuspension, incubated with lysozyme for 30 minutes on ice, was lysed and the cellular debris pelleted by ultra-centrifugation (@ 100,000 x g) for one hour at 4°C. The clarified protein was purified by Ni²⁺-affinity chromatography using Ni-NTA resin (GE Healthcare), which was followed by size-exclusion chromatography (Superdex 75). The hexa-histidine tags at both the N- and C-termini of UCHL1 were left uncleaved for the purpose of biolayer interferometry (BLI) experiments (described below). The purity of the purified proteins at each stage was assessed by SDS-PAGE analysis.

The human Ub expression vector, pRSETA-Ub, a kind gift from Genentech Corporation (USA), was transformed into the BL21(DE3) strain of E. coli. Protein expression was carried out following the same protocol as used for the expression of double His-tagged UCHL1. The cells were then harvested and resuspended in 1X PBS buffer supplemented with 0.4 M KCl, followed by heating at 70–80°C for 30 minutes before ultracentrifugation. The cleared supernatant was loaded onto a self-packed column of SP Sepharose Fast Flow resin (GE Healthcare) and eluted with a gradient of NaCl, up to 1M. Fractions containing pure Ub, confirmed by SDS-PAGE analysis, were pooled, concentrated, and exchanged into protein storage buffer (50 mM Tris pH 7.4, 50 mM NaCl and 1 mM DTT).

For NMR and SAXS/WAXS studies, the DNA sequence corresponding to the wild type human UCHL1 was cloned into a pET23a expression vector with a C-terminal His-tag. The R178Q construct was generated by site-directed mutagenesis and confirmed by DNA sequencing (Mission Biotech). The two plasmids were individually transformed into E. coli BL21 (DE3) for protein overexpression. The transformed cells were grown in M9 media with ¹⁵N-labeled NH₄Cl and ¹³C-labeled D-glucose (Cambridge Isotope Laboratory, USA) as the sole nitrogen and carbon sources, respectively. The cell cultures were shaken at 37 C until the optical density at 600 nm (OD₆₀₀) reached 0.6–0.8, at which point 0.5 mM IPTG was added to trigger recombinant protein overexpression followed by incubation at 16 °C for 18–20 hours. The cell pellets were harvested by centrifugation and resuspended in lysis buffer (50 mM sodium phosphate (pH 8.0), 300 mM NaCl, 5 mM β-ME, 0.001 % benzonase, 0.01 % TritonX-100, and 0.3 % (w/w) lysozyme). The cells were lysed by a NanoLyzer N2 (NanoLyzer) at 18 kpsi, and the cell debris was removed by centrifugation. The target recombinant proteins were purified by Ni-NTA (Roche). The lysate in purification buffer (50 mM sodium phosphate (pH 8.0), 300 mM NaCl, and 5 mM β-ME) was loaded onto 2 ml of Ni-NTA slurry in a PD-10 column (Cytiva, USA) followed by extensive wash by purification buffer. The target proteins were eluted by elution buffer (purification buffer plus 250 mM of imidazole) followed by size-exclusion chromatography using an HiLoad 16/600 Superdex 75 column (Cytiva, USA) in NMR buffer (10 mM sodium phosphate (pH 7.4), 137 mM NaCl, 2.7 mM KCl, 1 mM TCEP, 0.5 mM EDTA, and 0.02 % NaN₃).

Protein Crystallization and Structure Determination

Protein crystallization trials were set up using the vapor diffusion technique with a sparse matrix screening protocol. Crystals of the UCHL1 R178Q grew in 0.1M HEPES, 2.4M Ammonium Sulfate, pH 7.4. Small cube crystals grew in two days. These crystals were harvested and rapidly cooled without any cryo-protectant to liquid nitrogen temperatures for shipment to the Advanced Photon source. Diffraction data were collected at the Advanced Photon Source at Argonne National Laboratory and processed using HKL2000⁵⁵. The UCHL1 R178Q crystals diffracted to 2.07 Å. The structure of the mutant was determined by molecular replacement using UCHL1 (PDB ID: 2ETL¹⁵) as search model. The molecular replacement solution was obtained in P42₁2 space group using the program Phaser⁵⁶ in Phenix⁵⁷. A 2F_c – 2F_o map following rigid body refinement showed reasonable density for the model. The structure was then iteratively modelled and refined using COOT⁵⁸ and Phenix⁵⁷. The final refined structure was validated using MolProbity⁵⁹ and deposited in the protein data bank (PDB ID: 8DY8).

Ub VME $k_{inact}/K_I$ Assays

UCHL1_WT, UCHL1_S18Y, and UCHL1_R178Q were diluted to 2.5 nM in 50 mM Tris-HCl (pH 7.6) buffer containing 0.5 mM EDTA, 5 mM DTT, and 0.1% (w/v) BSA. HA-WT-Ub-VME underwent 1:1 serial dilution from a top concentration in the same buffer. Ub-Rho (Boston Biochem, catalog no. U-555) was diluted to 450 nM in the same buffer to make the Ub-Rho stock. Twenty μL of the Ub-Rho stock solution was first added to each well in a 384-well plate followed by 10 μL of HA-WT-Ub-VME. To initiate the reaction, 20 μL of UCHL1_WT, UCHL1_S18Y, or UCHL1_R178Q was added and fluorescence measurements were immediately recorded on a Synergy Neo 2 Multi-Mode Reader (BioTek) at excitation and emission wavelengths of 489 nm and 530 nm, respectively. Progress curve raw data were input into Prism 8, and a baseline correction analysis was completed to obtain all of the time zero points at the origin for fitting purposes. Each progress curve underwent fitting to the equation $\text{Y}={\text{V}}_0\times \left(1-e^{-kobst}\right)/k_{obs}$. The $k_{obs}$ values for each progress curve was graphed versus the concentration of HA-WT-Ub-VME. The slope of the linear fit was determined to be $K_{inact}/K_l$.

UCHL1 Michaelis-Menten Kinetics

Reactions were performed in black 384 well plates (Fisher 12566624) in a final volume of 50 μL. DUBs were diluted in reaction buffer (50 mm Tris pH 7.6, 0.5 mm EDTA, 5 mm DTT, 0.1 % (w/v) BSA) to a concentration of 4 nm (final concentration in well 2 nm). To each well was added 25 μL of DUB containing solution. Reactions were initiated by the addition of 25 μL of varying concentrations of Ub-Rho (Boston Biochem U-555, Final concentrations: 1000, 500, 250, 125, 62.5, 31.25, 15.625, 7.8125, 3.90625, and 0 μM). Reactions were read immediately (${\lambda }_{\text{ex}}=485\text{nm}$, ${\lambda }_{\text{em}}=535\text{nm}$) for 10 hours. Readings were performed on a Synergy Neo2. Data was exported to excel, and a standard curve was created by determining the average maximum fluorescence intensity at each concentration. This was plotted as a function of concentration, and the raw fluorescence data was converted to nM product produced (nm/sec⁻¹). The first 10% of the data points were used to calculate the initial velocity for each concentration of substrate, and this data was imported into GraphPad Prism 8 (GraphPad Software, San Diego, CA (USA), www.graphpad.com). The rates of each technical replicate for all concentrations were plotted as a function of substrate concentration. Data was fit using non-linear regression to determine the $V_{max},k_{cat}$, and $K_{\text{M}}$.

Biolayer Interferometry

Concentration of the N and C-terminally His-tagged UCHL1_R178Q were determined by nanodrop absorbance at 280 nm (A280) to enable dilution of the proteins into BLI buffer (1x PBS containing 0.05% (v/v) Tween 20 and 0.1% (w/v) BSA) to a concentration of 25 μg/mL. WT Ub was buffer exchanged into 1xPBS using PD10 columns (GE Healthcare). The concentration of the WT Ub was determined by a BCA assay and diluted 2 μM into BLI buffer due to low extinction coefficient. 1:1 serial dilution of Ub was performed to make the different concentrations of analyte used in the BLI experiment. 40 μL of each solution was added to a 384 tilted well plate. One Ni-NTA biosensor at 25°C was used for each $K_D$ measurement, dipping the UCHL1 protein loaded tip into wells that contained the lowest concentration of Ub mutant first. A reference sensor was also included in each experiment to use as a baseline to account for non-specific binding. These experiments were carried out in technical triplicate. His-tagged UCHL1 protein was allowed to load to the biosensor for 300 s, followed by cycle of Ub association and dissociation at 120 s and 100 s, respectively. ForteBio data analysis software (v8.2) was used to collect raw data for the association and dissociation curves. Averages of the association responses (from 110 s – 115 s) were calculated and plotted in Prism 8 program. These data were fit to a non-linear regression one site – specific binding model to determine a$K_D$.

NMR spectroscopy

Uniformly ¹⁵N and ¹³C labeled UCHL1_WT and UCHL1_R178Q were prepared in the NMR buffer at a protein concentration of 1 mM with 10% (v/v) D₂O. The NMR spectra were collected using an NMR spectrometer operated at a proton Larmor frequency of 600 MHz equipped with a triple resonance TCI cryo-probe at 298K. In addition to 2D ¹⁵N-¹H HSQC, three-dimensional HNCACB, CBCA(CO)NH, HNCA, HN(CO)CA and HNCO were acquired by non-uniform sampling (NUS) in sampling densities of 50 %. All spectra were processed by NMRPipe⁶⁰, and the NUS datasets were processed by SMILE reconstruction⁶¹ built in NMRPipe. Manual assignments of the backbone chemical shifts were achieved by using NMRFAM-SPARKY⁶². The composite backbone amide chemical shift perturbations (CSPs) were calculated by the following equation:

$$CSPs(ppm)=\sqrt{\text{Δ}{\delta }_{HN}^2+{\left(0.165\times \text{Δ}{\delta }_N\right)}^2}$$

where $\text{Δ}{\delta }_{\text{HN}}$ and $\text{Δ}{\delta }_N$ are the chemical shift differences between UCHL1_WT and UCHL1_R178Q for the amide proton and amide nitrogen, respectively. The CSPs were plotted as a function of Prism 9 (GraphPad, USA), and mapped on the crystal structure of UCHL1 in complex with Ub (PDB entry 3KW5) using PyMol 2.0 (Schrodinger Scientific, USA). The backbone NMR chemical shift assignments of UCHL1_WT and UCHL1_R178Q were deposited in the Biological Magnetic Resonance Data Bank (BMRB) under the accession codes of 52137 and 52138, respectively.

Size-exclusion chromatography-coupled small-angle X-ray scattering and wide-angle X-ray scattering (SEC-SAXS/WAXS)

SEC-SAXS/WAXS data of UCHL1 variants were collected using the BL13A beamline of the Taiwan Photon Source (TPS) at the National Synchrotron Radiation Research Center (NSRRC) in Hsinchu, Taiwan as described previously.^63–65 The protein concentrations were set to 14.0 and 10.0 mg/m for UCHL1_WT and UCHL1_R178Q, respectively. The protein samples were injected into a high-performance liquid chromatography (HPLC) system (Agilent, USA) to be separated by a size-exclusion column (BioSEC 30, Agilent, USA) to remove oligomeric population from the main monomeric population during the SAXS/WAXS data collection. The SAXS/WAXS data were collected at a sample flow rate of X ml/min with a momentum transfer (q) range between 0.008 and 0.5 for SAXS, and 0.4 and 1.5 for WAXS. An in-house program was used to perform data reduction, solvent subtraction and merging of the SAXS and WAXS data. The SAXS/WAXS data were analyzed using ATSAS package⁶⁶. The ${\text{R}}_{\text{g}}$ values were estimated using Guinier approximation for the low-resolution scattering (${\text{qR}}_{\text{g}}\le 1.3$ for globular scatters)

$$\text{I}[\text{q}]=\text{I}(0)e^{\left(-{\text{q}}^2{\text{R}}_{\text{g}}^2/3\right)}$$

where $\text{I}\left(\text{q}\right)$ is scattering intensity and $\text{I}\left(0\right)$ is the forward scattering intensity. The theoretical SAXS data of UCHL1 were back-calculated using FoXS⁶⁷ with the crystal structure of UCHL1 (PDB entry: 2ETL) as input for comparison with the experimental data. The missing termini and loop residues were modeled by the built-in Modeller⁶⁸ function within ChimeraX⁶⁹ with default setting. Five models were built for UCHL1_WT and UCHL1_R178Q.

The data were plotted using Prism 10 (GraphPad, USA). The SAXS profiles of UCHL1_WT and UCHL1_R178Q were deposited in the Small Angle Scattering Biological Data Bank (SASBDB) under the accession codes of SASDTP2 and SASDTQ2, respectively.

Research Highlights.

Altered protein dynamics and a more reactive catalytic cysteine in a neurodegeneration-associated UCHL1 mutant

1. The crystal structure of the mutant, UCHL1_R178Q (2.1 Å) reveals details of active-site structure including the catalytic water wedge which remains nearly the same as the wild-type enzyme. The arginine mutation to glutamine, while resulting in a loss of a hydrogen bond in the network, does not affect the lowestenergy conformation of the enzyme, as typically captured by static snapshots in X-ray structures.

2. Solution NMR shows extensive alteration of the chemical environment and backbone dynamics of as many as 40 residues, many of which are severely broadened indicating microsecond to millisecond dynamics. Several of these resonances correspond to residues far away from the site of mutation, including some involved in ubiquitin-induced conformational relay that brings the catalytic histidine into the correct location for catalysis. Solution x-ray scattering measurements (SAXS) echo a similar result of an increase in the overall flexibility of the mutant compared to the wild-type enzyme.

3. Michaelis-Menten enzyme kinetics show the activation is primarily a $k_{cat}$ effect, despite a somewhat weakened substrate affinity of the mutant. On further exploring this we found that the higher $k_{cat}$ effect is due to at least in part a higher reactivity of the catalytic cysteine of the mutant. The second-order rate constant of the R178Q mutant, as characterized by the $k_{inact}/K_I$ parameter, is significantly higher than that of the wild-type enzyme. Given the BLI results showing ubiquitin binding to the UCHL1 mutant is slightly weaker than the wild-type enzyme, the difference in $k_{inact}/K_I$ parameter can be safely taken to reflect a more reactive catalytic cysteine in the mutant.