Dynamic Long-Range Interactions Influence Substrate Binding and Catalysis by Human Histidine Triad Nucleotide Binding Proteins (HINTs), Key Regulators of Multiple Cellular Processes and Activators of Antiviral ProTides

Abstract

Human histidine triad nucleotide binding (hHINT) proteins catalyze nucleotide phosphoramidase and acyl-phosphatase reactions that are essential for the activation of antiviral proTides, such as Sofosbuvir and Remdesivir. hHINT1 and hHINT2 are highly homologous but exhibit disparate roles as regulators of opioid tolerance (hHINT1) and mitochondrial activity (hHINT2). NMR studies of hHINT1 reveal a pair of dynamic surface residues (Q62, E100) which gate a conserved water channel leading to the active site 13 Å away. hHINT2 crystal structures identify analogous residues (R99, D137) and water channel. hHINT1 Q62 variants significantly alter the steady-state k_cat and K_m for turnover of the fluorescent substrate (TpAd), while stopped-flow kinetics indicate the K_D also changes. hHINT2, like hHINT1, exhibits a burst-phase of adenylation, monitored by fluorescent tryptamine release, prior to rate-limiting hydrolysis and nucleotide release. hHINT2 exhibits a much smaller burst-phase amplitude than hHINT1, which is further diminished in hHINT2 R99Q. Kinetic simulations suggest that amplitude variations can be accounted for by a variable fluorescent yield of the E•S complex from changes in the environment of bound TpAd. Isothermal titration calorimetry measurements of inhibitor binding shows that these hHINT variants also alter the thermodynamic binding profile. We propose that these altered surface residues engender long-range dynamic changes that affect the orientation of bound ligands, altering the thermodynamic and kinetic characteristics of hHINT active site function. Thus, studies of the cellular roles and proTide activation potential by hHINTs should consider the importance of long-range interactions and possible protein binding surfaces far from the active site.

Graphical Abstract

Introduction

Histidine triad nucleotide binding (HINT) proteins are small, catalytically efficient enzymes known for their nucleotide phosphoramidase and acyl-phosphatase activities, which are critical in the regulation of the μ-opioid pathway and chronic pain as well as the management of mitochondrial calcium gradients.(1-4) In humans, there are three HINT enzymes, hHINT1, hHINT2 and hHINT3, which have been found to localize in the cytosol, mitochondria, and nucleus, respectively.(5, 6) hHINT1 bears 61% sequence homology with hHINT2 and previous active site alignments of hHINT1 and hHINT2 crystal structures show that nearly all side chain residues are identical with a very small root mean squared deviation (PDB IDs: 3TW2, 4INI RMSD = 0.21Å).(5, 7) Not surprisingly, hHINT1 and hHINT2 are also known to possess similar substrate specificities (k_cat/K_M) that differ by less than three-fold for model substrates such as adenosine tryptamine phosphoramidate monoester (TpAd) (Figure 1).(7)

Figure 1.

(A) The full catalytic mechanism for hHINT1 reaction with substrate TpAd where R-NH₂ represents the first product tryptamine generically known as P₁, and NMP, which represents the second product generically known as P₂.(2) (B-E) Crystal structures of each enzymatic form occurring during reaction with the TpAd substrate mimic TpGMPS that allows for capture of the E-NMP intermediate (PDB IDs: 1KPA, 5IPC, 5IPD, 5IPE).(21) The catalytic residue 112 in the E•S species is modeled as the catalytic histidine for illustrative purposes.

Despite the remarkably high structural homology and similar substrate specificity of hHINT1 and hHINT2, it has been demonstrated that each enzyme has a distinct role relevant to human health in the cytosol and the mitochondria. hHINT1 resides in the cytosol and active site inhibition of the enzyme in mouse models has been shown to both dramatically prevent and reverse opioid tolerance.(8, 9) The pharmacologic impact of hHINT1 active site inhibitors is understood to originate from disruption of a key origin of opioid tolerance, the cross-talk between the μ-opioid (MOR) and the N-methyl D-aspartate receptor (NMDAR). Clinical case studies involving numerous hHINT1 polymorphisms have established these variants as a direct cause of Charcot–Marie–Tooth disease, as well as yielding axonal neuropathy with neuromyotonia.(10-13) In some cases, these missense mutations cause amino acid residue changes at the active site, but some are also noted at the hHINT1 dimer interface as well as the protein surface away from the active site.(12-14) Outside of its role in the central nervous system (CNS), hHINT1 also helps regulate the microphthalmia transcription factor (MITF), which when dysregulated, has been shown to lead to melanoma and colon cancer.(15, 16) While the active site of the enzyme is known to be crucial in the hHINT1-MITF protein-protein interaction, an important posttranslational phosphorylation site distant from the hHINT1 active site (Y109) has also been shown to modulate the interaction, but clear mechanistic explanations have not yet emerged.(16)

Less in known about hHINT2 in the mitochondria, but it is clear that the enzyme helps regulate the mitochondrial calcium uniporter (MCU). This regulation has implications in mitochondrial protein acetylation levels and the onset of steatosis, a state of increased cellular fat retention.(17-19) A recent investigation of 90 diverse eukaryotic genomes indicates that the evolutionary jump from one to two subcellularly localized HINTs occurred through gene duplication not once, but in four parallel independent occurrences in premetazoan ancestors, highlighting the importance of at least two separate hHINTs in the development of higher ordered life forms.(20)

The steps in the catalytic cycle from substrate binding to product release have been determined for the substrate TpAd with hHINT1 (Figure 1A).(2) However, such mechanistic studies have not been carried out for hHINT2 and might be of value in determining why these two otherwise similar enzymes exhibit such distinct roles in human metabolism. hHINT1 is known to first bind TpAd (Figure 1B and 1C), which is followed by rapid adenylation of the catalytic nucleophile, H112 (Figure 1D). Upon the formation of the adenylated intermediate, the tryptamine P₁ product is rapidly released. Due to its low binding affinity, the release of tryptamine is essentially irreversible. The second phase of the ping-pong catalytic cycle is dependent on rate-limiting hydrolysis of the adenylated intermediate by an activated water molecule, followed by partially rate-limiting release of the adenosine monophosphate P₂ product from the active site (Figure 1E). It should be noted that the water molecule that participates in hydrolysis of the intermediate has been a topic of study and is understood to come from the bulk solvent and not from a conserved internal water channel that resides just behind the active site.(21, 22)

While the conserved water channel does not appear to provide the hydrolytic water in the reaction mechanism, the channel has been a topic of speculation in previous structural and computational analyses of hHINT1 and hHINT2.(21, 22) To date, its functional influence on catalysis has not yet been demonstrated experientially. In the current study, we have combined data from NMR, steady-state and transient kinetics, binding measurements, computations, and mutagenesis with the goal of identifying differences between hHINT1 and hHINT2 that might allow them to be uniquely regulated. The studies show that both isozymes employ a previously unrecognized strategy of regulation originating far from the active site at the protein surface origin of the conserved hHINT water channel. Subtle differences in the way this novel method is implemented result in isozyme-specific substrate interactions and reaction cycle dynamics as well as the thermodynamics of binding an active site inhibitor. The long-range communication demonstrated here gives added importance to the effects of posttranslational modifications others have found at residue positions remote from the active site,(16) as well as the many characterized protein-protein interactions between hHINTs and the proteins they regulate in the CNS and mitochondria.(3, 8, 9, 19, 23)

EXPERIMENTAL PROCEDURES

Chemical reagents were purchased from Sigma Aldrich and ChemImpex. The Pierce BCA assay and the BioRad Bradford assay were purchased and utilized for protein concentration determination and a Thermo Fisher Scientific Nanodrop 2000 was utilized for protein absorbance measurements.

Inhibitor and Substrate Synthesis.

Both the substrate TpAd and inhibitor 1 were synthesized in house as previously described.(1, 24) Compounds were purified by cation exchange and flash chromatographies and their purity was confirmed by ¹H-NMR, ³¹P-NMR, and mass spectrometry.

Protein Expression and Purification.

Recombinant full length hHINT1 was produced as described previously with cDNA in MCSG7 vector. The hHINT1 construct has a cleavage His tag. hHINT2 with a 36 residue N-terminal truncation (to replicate the cleavage of its mitochondrial localization signal that is removed upon its transport to the mitochondria in vivo) was produced as described.(5, 7) Briefly, the hHINT2 cDNA was inserted into the pMCSG9 vector with a TEV cleavable N-terminal His-6 tagged maltose binding protein. Rosetta2 (DE3) pLysS competent Escherichia coli were transformed and with the relevant plasmid and cultured in TB media. Protein expression was induced by addition of 1 mM IPTG at OD₆₀₀ = 0.6 and continuing growth at 37 °C for 4 hours. Harvested cells were lysed, centrifuged, and hHINT proteins were purified by Ni-NTA chromatography. Where applicable, the N-terminal His6-MBP was cleaved with 2% by weight TEV protease, and the hHINT2 protein product was purified via SEC-FPLC using a was a HiPrep 16/60 Sephacryl S100 HR, GE Healthcare column. The size exclusion column serves to both finish the purification and also to confirm that the HINT proteins were well-folded and dimeric. Both hHINT1 and hHINT2 can readily form homodimers.(25) The New England Biolabs Q5 Mutagenesis Kit was utilized in the construction of hHINT1 Q62A and Q62R variants as well as hHINT2 R99A and R99Q variants, using the NEBaseChanger web tool for mutagenesis primer design.

Isothermal Titration Calorimetry.

Isothermal titration calorimetry (ITC) experiments were conducted in a MicroCal Auto-ITC200 system (GE Healthcare life sciences). All titration experiments were performed at 25 °C in 10 mM Tris, 150 mM NaCl, pH 7.2 (ITC buffer ). hHINT1 and hHINT2 were exchanged into ITC buffer using Micro Biospin6 columns (BioRad, USA) so that all experimental buffer was identical. hHINT protein concentration (5 - 600 μM) depended on its affinity for the ligand and a 10:1 ligand to protein ratio was utilized.

Steady-State Kinetics Assay.

The previously described assay (1, 2) measures the production of fluorescence occurring on hydrolysis and release of the fluorescent indole-bearing sidechain product, which liberates it from the intramolecular quenching by the substrate nucleobase. Measurements were taken in a quartz cuvette in a temperature-controlled Varian/Cary Eclipse fluorimeter with excitation at 280 nm and measurement of emission at 360 nm (25 °C). hHINT enzymes (5 nM) was prepared in degassed 20 mM HEPES, 1 mM MgCl₂, pH 7.2, substrate concentration was varied, and the linear fluorescence slopes were measured and converted to velocities using fluorescence standard curves of the reagents. Measurements were made in triplicate and curves were fitted to Y = V_max*X/(K_m + X) where Y is the fluorescence intensity and X is time (Graphpad Prism) to obtain the Michaelis-Menten k_cat and K_m.

Stopped-Flow Kinetics Assays.

The transient adenylation burst rate constants and subsequent linear steady state rate constants were measured using an Applied Photophysics SX.18MV spectrophotometer in fluorescence mode with excitation at 280 nm. Fluorescence above 320 nm was recorded. TpAd (2.5 - 62.5 μM) in degassed 20 mM HEPES, 1 mM MgCl₂, pH 7.2 was reacted with the hHINT enzymes. hHINT2 experiments were conducted at 4 °C to intentionally decrease the rate of adenylation, since, at 25 °C, the fluorescence increase occurs within the dead time of the stopped-flow instrument (pseudo-first-order rate constant ≥ 600 s⁻¹). hHINT1 measurements were also determined at 25 °C to facilitate comparison to published hHINT1 rate constants. All measurements were collected in triplicate, averaged, and fitted to Equation 1 using non-linear regression fitting (Applied Photophysics Pro-Data Viewer v.4.2.12). In Equation 1, FU corresponds to the relative fluorescence, A is the burst amplitude, k_obs is the observed exponential rate of adenylation, t is time, V_std is the observed velocity of the linear steady state phase, and C is the extrapolated maximum of the exponential phase.

$$[FU[t=A{e}^{-k_{obs}t}+V_{std}t+C$$ (Equation 1)

When k_obs was plotted versus substrate concentration, a hyperbolic curve describing the first phase of the hHINT ping-pong reaction was obtained and fit with Equation 2 where k_obs is the observed exponential rate of adenylation at a given [S], k₂ is the maximum velocity of the adenylation reaction, and K_D is the [S] that yield half of the maximum adenylation velocity and is equal to the dissociation constant for substrate binding to the enzyme (k₋₁/k₁). Note that in Equation 2, the term k₋₂ can, in effect, be ignored since tryptamine has no affinity for the active site once cleaved.

$$k_{{}_{obs}}=\frac{k_{{}_2}[S[}{K_{{}_D}+[S[}+k_{-2}$$ (Equation 2)

Circular Dichroism Spectroscopy.

Circular dichroism spectra (195-260 nm) were recorded using a JASCO J-815 spectropolarimeter in a 1 mm quartz cuvette at 25 °C. hHINT proteins (10μM) were in 200 mM Tris, 200 mM NaF, pH 7.4 buffer. The α-helix and β-strand content of the protein was estimated using the K2D3 webserver.(26)

¹⁵N-CPMG NMR Spectroscopy.

¹⁵N labeled hHINT1 was expressed in minimal media supplemented with ¹⁵NH₄Cl as its nitrogen source and purified using the same techniques as the non-labeled proteins used in this work. CPMG experiments were conducted both at 1.64 and 19.9 T using ν_CPMG pulse rate of 50, 100, 150, 200, 250, 300, 350, 400, 500, 600, 700, 800, 900, and 1000 Mhz with duplicate data at 50, 100, 300 and 500 Mhz. The time spacing between the 180° pulse centers is equal to 2τcp, which is expressed as the frequency (ν_CPMG) equal to 1/(4τcp). Data were fit with ShereKhan molecular exchange rate fitting toolset(27) using the Bloch-McConnell two-state equation for residues Q62 and E100 at both NMR frequencies tested.

hHINT1 Q62A Crystallography.

hHINT2 Q62A crystals were grown in 16 % w/v PEG3350, 0.2 M sodium citrate, 0.1 M sodium cacodylate pH 7.0 under hanging drop diffusion. Data collection was conducted on the BL14.1 BESSY beamline at Helmholtz-Zentrum Berlin. Further details can be found in Table S2.

Molecular Dynamics Simulations.

The apo-hHINT1 (PDB ID: 1KPA) and apo-hHINT2 (PDB ID: 4NJZ) were used as starting points and the requisite hHINT1 Q62A and hHINT2 R99Q variants were generated using Visual Molecular Dynamic’s (VMD’s) mutagenesis tool. The three conserved internal water molecules identified in the crystal structures (W1, W2, W3) were retained while all other crystallographic waters were removed. The QwikMD VMD plugin was used to help generate the solvation box of TIP4P waters with net neutral [NaCl₂] of 150 mM. Standard CHARMM parameters and libraries were used to add hydrogens, minimize the structure, equilibrate for 1.25 ns, and conduct the 50 ns production run at 300 K. The water channel in Figure 2 was visualized with the PyMol CAVER plugin(28) and W1 exchange events were counted with the help of an in-house distance script used to track the loss of a water molecule from the channel over time.

Figure 2.

(A) ¹⁵N-CPMG NMR data of hHINT1 and the fitted R₂ regressions for residues E100 and Q62 at 700 MHz and 850 MHz field strength. The combined Bloch–McConnell fit of the data yielded a k_ex = 638 ± 68 s⁻¹ for these paired residues. (B) Crystal structure of the hHINT1 E•P complex with product AMP bound. The dynamic residues Q62 and E100 as well as the catalytic H112 are displayed as sticks. Three internal waters (red spheres) connect the 13 Å span between the active site and the dynamic ion-dipole pair (PDB ID: 3TW2). (C) Crystal structure of the hHINT2 E•P complex with product AMP bound. R99 and D137 as well as the catalytic H149 are displayed as sticks (PDB ID: 4INI). The internal waters that connect the 13 Å span between the active site and the Q62, E100 salt bridge pair are analogous to those in hHINT1.

Simulations of Kinetic Time Courses.

The fluorescence time course obtained when reacting enzyme with substrate was simulated using KinTek Explorer Pro v.6.3 using the reaction scheme shown in Figure 1A with the addition of an explicit tryptamine dissociation step. It was assumed that the fluorescence of TpAd and tryptamine in solution are unaffected by the presence of the hHINT enzyme, so that the initial and final fluorescence intensities are enzyme independent (i.e., the same concentration-to-fluorescence yield scaling factors are used for free TpAd and for tryptamine for both hHINT1 and hHINT2 reactions). The final fluorescence intensity represents complete cleavage the starting TpAd and release of tryptamine, and thus, is the same in all reactions. It was used to adjust for minor shifts in the baseline of the stopped-flow instrument. After this adjustment, the initial fluorescence intensity for the two hHINT enzymes was used to independently scale the fluorescence yield of the bound substrate. Approximate rate constants for the steps were entered into the program based on the values from non-linear regression fitting and steady-state measurements. The simulation program then progressed automatically to the best fit values. Entering different starting rate constants was used as a check for false minima in the simulation. Determination of the effect of incrementing the final values in the simulation on the root mean square deviation from the data allow an estimation of the error in the values.

RESULTS

CPMG-NMR Reveals a Dynamic Residue Pair Distant from the hHINT1 Active Site.

A simple inspection of hHINT X-ray crystallography B-factors indicates that, despite the apparent conformational homogeneity between hHINT1 and hHINT2 structures, both enzymes possess unique dynamic motifs throughout their amino acid sequence (Figure S1). hHINT1 is known to be a small and robust enzyme, making it an ideal candidate for dynamics analysis by NMR.(29) The high expression yield of hHINT1 allowed ¹⁵N-CPMG NMR to be employed to probe the millisecond-microsecond timescale motions of the nitrogen backbone atoms within its structure as a first step toward characterizing the dynamics of the enzyme system.(29, 30) The NMR dispersion decay rates of two surface residues Q62 and E100 positioned at the end of a water channel behind the enzyme active site were particularly notable (Figure 2A). The hHINT1 crystal structure reveals that these two surface residues participate in an ion-dipole interaction and are positioned 13Å from the active site (Figure 2B). Clustering and fitting their R₂ decay rates in both orthogonal NMR field strengths tested (700 and 850 MHz) provided a k_ex value of 638 ± 68 s⁻¹ for these two residues. This exchange rate describes the speed at which the backbone nitrogen atoms oscillate between distinct conformational states.(29, 30)

Inspection of the crystal structure of hHINT2 revealed a similar interaction at the analogous residues R99 and D137 (Figure 2C). Both enzymes possess a similar water channel bridging the active site to a surface residue pair, but the ion-dipole interaction found in hHINT1 is replaced by a salt bridge within hHINT2 (Figure 2B). 50 ns molecular dynamics (MD) simulations were conducted with hHINT1 and hHINT2, revealing that the motions of the Q62, E100 motif in hHINT1 and R99, D137 motif in hHINT2 allow for the release and exchange of the internal water molecule immediately adjacent to the residue pair. While these MD simulations did not reveal an obvious mechanistic relationship between these distant dynamic residues and the active site, they corroborate the dynamic nature of these surface residue pairs and reveal their ability to gate the release of a water molecule from this channel that connects to the active site (Figure S1B).

Disruption of the hHINT1 Q62, E100 Residue Pair Alters Steady-State and Transient Kinetic Rate Constants for Substrate Turnover.

To probe the role of the Q62, E100 residue pair on hHINT binding and catalysis, the hHINT1 variants Q62A and Q62R were prepared. hHINT1 Q62A was created to ablate the ion-dipole interaction with E100, while Q62R was generated to assess the impact of a salt bridge as found in the analogous hHINT2 R99, D137 motif. With these hHINT1 variants in hand, their impact on catalytic activity was tested by determining the steady-state and pre-steady-state burst-phase kinetics of catalysis.

The continuous spectrophotometric assay utilized for hHINT kinetic studies relies on an observable increase in the fluorescence of the indole moiety within the P₁ product tryptamine of TpAd upon release after the first chemical step of the kinetic mechanism (Figure 3A).(1) The kinetic assay is made possible by the fact that the fluorescence of the indole moiety of TpAd is quenched or partially quenched until the first steps of the enzymatic reaction occurs and the tryptamine moiety is released into solution. The quenching of the fluorescence of TpAd in solution is due to internal π-π stacking between the indole sidechain and nucleobase, while that of bound substrate is quenched by an interaction in the active site (see Discussion). Upon release of the tryptamine product, an easily detected increase in fluorescence at 360 nm is observed (Figure 3A).

Figure 3.

(A) General scheme of substrate TpAd quenched in solution due to π - π stacking between the tryptamine indole and nucleobase or interactions in the hydrophobic active site. (B) Representative hHINT1 burst-phase and steady-state reaction time course (2.5 μM hHINT1, 75 μM TpAd, 4.3 °C) and the non-linear regression (NLR) fitting function used to analyze the data. [FU]_t is fluorescence generated as a function of time, A is the burst amplitude, k_obs is the observed exponential rate of adenylation, V_std is the linear steady-state velocity, t is time, and C is the extrapolated maximum of the exponential phase.

The hHINT continuous fluorescence assay has been utilized for steady-state measurements(1, 7) as well as for stopped-flow transient kinetic analysis to define the individual rate constants of the hHINT1 reaction coordinate.(2) During the first catalytic step after substrate binding, the active site catalytic nucleophile, H112 is adenylated followed by rapid release of the tryptamine product. For the substrate TpAd, past studies have shown that this first chemical step for hHINT1 is described by the rate constant k₂ with a value of 203 s⁻¹ (25 °C).(2) An example of the process for determination of this rate constant is illustrated for the hHINT1 reaction in Figure 3B and hHINT2 in Figure 4. First, the transient fluorescent time course (Figure 3B) of the reaction was measured at varying substrate concentrations. Then, the time course was fit using the equation shown in Figure 3B (Experimental Procedures, Equation 1) which accounts for the exponential rate of adenylation (k_obs) followed by the slower steady-state phase (V_std) during which the product P₂ is released and the enzyme cycles. The plot of k_obs versus the substrate concentrations tested (example shown in Figure 4) was then fit with Equation 2 (see Experimental Procedures), yielding the rate constant of adenylation (k₂) under the assumption that k₋₂ = 0. The dissociation constant (K_D) for the rapid association of substrate required for adenylation is also derived from this plot.(1, 2)

Figure 4.

(A) Time courses of hHINT2 reacting with various concentrations TpAd. (B) k_obs values obtained by fitting the time courses using Equation 1 plotted against their respective [TpAd]. The solid line is the non-linear regression fit using Equation 2 (a hyperbola offset by k₋₂) allowing determination of the k₂ = 89.7 s⁻¹ and K_D = 5.44 μM for hHINT2 at 4 °C. The extrapolated maximum of the hyperbola describes the sum of the forward and reverse adenylation rates, but because the reverse reaction (k₋₂) is effectively zero due to low affinity for the tryptamine product that is released, the hyperbola instead maximizes at k₂.

Steady-state kinetic analysis revealed that the hHINT1 Q62R mutation had little effect on either the steady-state k_cat or K_m (Table 1). While the rate of adenylation (k₂) of hHINT1 Q62R relative to the wild-type for the substrate TpAd was also not significantly affected, a more profound five-fold decrease in the K_D for substrate binding was observed (Table 1).

Table 1.

Steady-State Rate Constants and Burst Phase Adenylation Rate Constants Determined by Non-Linear Regression Fitting for hHINT1 and hHINT1 Variants

	hHINT1 WTa	hHINT1 Q62Ab	Q62A/WT Ratio	hHINT1 Q62Rb	Q62R/WT Ratio
Steady-State (25 °C)
kcat (s−1)	2.1 ± 0.1	0.45 ± 0.02	0.21	1.4 ± 0.09	0.67
Km (nM)	130 ± 20	15 ± 3	0.12	136 ± 30	1.05
kcat/Km x106 (s−1M−1)	15 ± 3	28 ± 4	1.87	11 ± 2.4	0.73
Burst Phase (25 °C)
k2 (s−1)	203 ± 13	172 ± 6	0.85	180 ± 9	0.89
KD (μM)	13 ± 2	4.9 ± 0.5	0.38	2.9 ± 0.2	0.2

^aPreviously published by our group(2)

^bThis study

The kinetic impact of the hHINT1 Q62A variant was much more dramatic as seen in Table 1. The steady-state k_cat value was five-fold lower than the wild-type value. Nevertheless, the nearly ten-fold lower value for the K_m resulted in a substrate specificity value (k_cat/K_m) twice as high as seen for the wild-type. While hHINT1 Q62A exhibited a similar k₂ (172 ± 6 s⁻¹) compared to hHINT1 wild-type (203 ± 13 s⁻¹), the K_D value for the hHINT1 Q62A experienced a modest 2.5-fold reduction.

Disruption of the hHINT2 R99, D137 Residue Pair Induces a Modest Decrease in Steady State Substrate Turnover.

hHINT2 variants R99A and R99Q were generated to perturb the salt bridge interaction found in crystal structures (Figure 2C), analogous to the dynamic hHINT1 residue pair found by NMR, in an effort to determine the role they might have in dictating the ligand binding and kinetic properties of hHINT2. Unlike hHINT1 Q62A, which exhibited a significant ten-fold decrease in K_M and a five-fold decrease in k_cat compared to the wild-type hHINT1, hHINT2 R99A exhibited essentially no change in the K_M value and only a modest, but reproducible, two-fold decrease in the k_cat value when compared to the wild-type enzyme (Table 2). As for hHINT2 R99Q, little change was observed specifically in the steady-state kinetic parameters when compared to the wild-type enzyme. While there are only minor effects on the fitted k₂ and K_D value for both variants, the effect on the amplitude of the burst phase of the non-linear regression fit proved to be quite dramatic as described below.

Table 2.

Steady-State Rate Constants and Burst Phase Adenylation Rate Constants Determined by Non-Linear Regression Fitting for hHINT2 and hHINT2 Variants

	hHINT2 WT	hHINT2 R99A	R99A/WT Ratio	hHINT2 R99Q	R99Q/WT Ratio
Steady-State (25 °C)
kcat (s−1)	5.77 ± 0.34	2.65 ± 0.19	0.46	6.01 ± 0.26	1.04
Km (nM)	910 ± 98	683 ± 42	0.75	840 ± 155	0.92
kcat/Km x106 (s−1M−1)	6.3 ± 1.0	3.88 ± 0.81	0.65	7.20 ± 1.10	1.18
Burst Phase (4 °C)
k2 (s−1)	89.7 ± 5.1	86.4 ± 6.1	0.96	93.2 ± 5.0	1.04
KD (μM)	5.44 ± 0.58	6.00 ± 0.47	1.10	5.45 ± 0.60	1.00

While all steady-state experiments were conducted at 25 °C, the burst-phase kinetics for hHINT1 and hHINT2 were measured at different temperatures (hHINT1, 25 °C; hHINT2; 4 °C) due to the k₂ of hHINT2 slightly exceeding the limitations of the instrument (k >600 s⁻¹ at 25 °C). When measured at 4 °C, the two enzymes exhibit similar values for k₂ (see simulation results below).

hHINT2 Exhibits a Diminished Transient Burst Amplitude Which is Altered by R99Q.

The amplitude term of the equation for a single exponential plus slope (Equation 1) that describes the hHINT kinetic time course provides the total increase in fluorescence during the transient burst-phase of the catalytic cycle (Figure 3B). Typically, the burst amplitude for the rapid rate of hHINT adenylation is expected to correlate with the concentration of enzyme reacted, since the first turnover event of the burst-phase should generate an equimolar concentration of fluorescent product before the slower steady-state equilibrium is achieved. Interestingly, wild-type hHINT2 was only observed to produce 26% of the fluorescence signal expected from the amount of enzyme used and the fluorescence signal measured in our standard curves for tryptamine fluorescence (Figure 5, Table 3). The R99Q variant was observed to further diminish this amplitude while the burst-phase kinetics were largely unchanged (Figure 6).

Figure 5.

Example of KinTek Explorer simulations (red dashed line) overlayed upon experimental fluorescence burst-phase data of hHINT1 (blue) and hHINT2 (grey). The parameters of the simulated kinetic model are given in the Table 3. The steps associated with k₁, k₋₁, k₃, and k₅ in this model are fast and could be varied over a wide range without affecting the quality of the simulation. The rate constants for the k₄ and k₅ steps can be exchanged without affecting the simulation indicating they share the role of being the rate-limiting steps. Reaction conditions: 2.5 μM enzyme, 75 μM TpAd in 20 mM HEPES, 1 mM MgCl₂, pH 7.2, 4.3 °C. NMP = adenosine monophosphate, Tp = tryptamine.

Table 3.

Rate Constants Determined for hHINT Enzyme Reactions by Time Course Simulation

Reaction step	Rate constant	hHINT1	hHINT2
E+TpAd ⇆ E•TpAd	k 1	>9.5 μM−1 s−1	>50 μM−1 s−1
	k −1	>60 s−1	>60 s−1
E•TpAd ⇆ E-NMP•Tp	k 2	56.5 ± 2.5 s−1	60.8 ± 2.5 s−1
	k −2	0	0
E-NMP•Tp ⇆ E-NMP + Tp	k 3	>2000 s−1	>1640 s−1
	k −3	0	0
E-NMP + H20 ⇆ E’-NMP	k 4 a	0.41 ± 0.03 s−1	1.34 s−1 ± 0.14 s−1
	k −4	0	0
E’-NMP ⇆ E + NMP	k 5 a	>100 s−1	>34 s−1
	k −5	0	0

^aThe values of k₄ and k₅ can be interchanged without altering the fit of the simulation

Figure 6.

Stopped-flow fluorescence time courses for hHINT2 (black) or hHINT2 R99Q (blue) reacting with TpAd. Non linear regression fits to Equation 1 are shown in red and the fit values are shown in the insert table Reaction conditions: 0.8 μM enzyme, 40 μM TpAd in 20 mM HEPES, 1 mM MgCl₂, pH 7.2 at 4.3 °C.

Diminished hHINT Kinetic Burst Phase Amplitudes Do Not Result from Differences in Protein Concentration or Folding.

The origin of the attenuated amplitude of the burst-phase kinetics seen with hHINT2 was further investigated to eliminate the possibility that it was simply due to inaccurate enzyme concentration measurements or inactivity. The concentration of the protein was determined by bicinchoninic acid (BCA) assay, Bradford assay, and A₂₈₀ measurements. When compared, the protein concentrations determined from each assay were found to be within 7% of each other, which is significantly smaller than the up to 75% observed reduction in the burst amplitude (data not shown). (31, 32) To assess the structural heterogeneity of the purified enzyme, circular dichroism (CD) spectroscopy was employed to examine the secondary structure of hHINT2 in solution in comparison to that of hHINT1 and crystal structures.(33) Analysis of the CD spectra show that hHINT2 shares a high degree of secondary structural homology with hHINT1 with a calculated α and β character of 26.01% and 30.44%, respectively (Figure S2). The calculated α and β character for hHINT1 was found to be 26.52% and 29.04%, respectively. CD analysis of the variants used in this work confirmed essentially unchanged secondary structure compared to their wild-type counterparts (Figure S2).

Size exclusion chromatography (SEC) also demonstrated that all hHINTs (dimer MW = 28 kDa) elute with a similar retention time, without aggregation (Figure S3).(34) In addition, the stoichiometric n-value determined by ITC binding analysis for an inhibitor bound to HINT1 and HINT2 described later in this study are consistent with a 1:1 enzyme to inhibitor binding stoichiometry. The potential for a pH dependent effect on the burst-phase amplitude observed for the hHINT2 reaction was also explored, however, no effect on the amplitude was observed over a pH range of 5.6 to 9.0 (data not shown). Taken together, these data indicate that purified hHINT2 and the variants tested are well folded and the active sites are readily available for ligand binding under our experimental conditions.

Kinetic Simulations Demonstrate that a Quenching in the hHINT2 E•S Complex Can Account for the Decreased Burst Amplitude.

The non-linear regression fitting of the time course with an exponential function plus a slope and a constant gives reasonably accurate values for k₂ and K_D in the case of hHINT1 because the burst is fast relative to the steady-state turnover (Figure 3B). However, in the case of hHINT2 the faster steady-state turnover (compare steady-state slopes in Figure 5) means that a second turnover may begin before the exponential phase is complete, making the exponential fit less accurate. Also, Equation 1 assumes that the maximum steady state rate begins at the start of the reaction, which is not correct. These simplifications could potentially cause the apparent amplitude of the exponential phase to be inaccurate. This possibility was tested by simulating the time course via numerical integration based on the previously reported model for the hHINT reaction displayed in this work in Figure 1A.(2) It was found that no combination of rate constants could reproduce the large difference in amplitudes observed for the exponential phases in the hHINT1 and hHINT2 reactions seen in Figure 5.

An alternative explanation for the amplitude differences was tested by simulating the time courses under the assumption that the TpAd exhibits different fluorescent yields when bound to hHINT1 versus hHINT2. It was further assumed that the fluorescence yield of TpAd in solution and the tryptamine product P₁ are not altered by the presence of either enzyme. As shown in Figure 5 (dashed lines), excellent simulations are obtained under the constraint that TpAd fluorescence is nearly quenched when bound to hHINT1 but is able to exhibit ~75% of the fluorescence of unquenched tryptamine when bound to hHINT2. Crystal structures of hHINT1 substrates bound to catalytically inactive variants have previously indicated that the substrate binds in an elongated manner when bound, so that self-quenching is not expected (Figure 7). Quenching within the active site must then derive from the environment provided by the protein near the tryptamine moiety prior to cleavage of TpAd, which would be very sensitive to the proximity and orientation of nearby aromatic residues and the hydrophobicity of the site. The requirement to assume different fluorescence yields for the TpAd when bound to hHINT1 versus hHINT2 suggests that the active site environments of the enzymes differ to an appreciable degree in the substrate bound state.

Figure 7.

Fluorescence emission spectra of tryptamine excited at 280 nm in solvents with different dielectric constants.

The full list of kinetic constants from the simulations of the time courses for this model are summarized in Table 3. Simulation showed that substrate binding and tryptamine release are fast compared to the other reactions and can be fit over a wide range of values. However, the simulations were particularly sensitive to the adenylation step (k₂) and the rate-limiting E-NMP hydrolysis and release steps in agreement with the reported mechanism of hHINT1.(2) Interestingly, the analysis shows that the exponential burst-phase occurs with the same rate constant within error for the hHINT1 and hHINT2, demonstrating the similarity of the reactions catalyzed by the two enzyme in this portion of the reaction cycle. The adenylation rate constant for hHINT1 determined from the simulation (Figure 5) is similar to that obtained from linear regression fitting. The simulated rate constant for the linear phase of hHINT2 was 3.25-fold higher than that for hHINT1, also in reasonable agreement with the ratio of k_cat values obtained from steady-state kinetics measurements as well (Table 1 and 2).

Both the non-linear regression fitting and the kinetic simulation are in agreement that after a rapid adenylation, the overall reaction cycle for both enzymes is rate limited by the hydrolysis and/or final product release steps. Based upon these simulations, a key plausible difference between these two enzymes is a differential fluorescence quenching effect on the TpAd substrate that is stronger when bound to hHINT1 and diminished in hHINT2, leading to the higher initial hHINT2 burst-phase fluorescence amplitude values, and in turn, smaller apparent burst amplitudes (Figure 5). Interestingly, the fact that hHINT2 R99Q can further alter the burst-phase amplitude implies that the variant may influence this hydrophobic active site quenching effect at a distance. This action at a distance correlates with the modest decrease in steady-state turnover exhibited by the hHINT2 R99A variant (2-fold), as well as the more dramatic manner in which the hHINT1 Q62A was able to influence so many properties of the active site environment (k_cat = 5-fold decrease, K_m = 10-fold decrease, K_D = 2.5-fold decrease).

Tryptamine Reporter Fluorescence Varies with the Hydrophobicity of the Environment.

The emission spectra of fluorescent indole compounds such as tryptamine, the reporter in this assay, are known to both quench and shift their λ_max due to a change in their dielectric environment. (35-37) Indeed, when tryptamine was placed in an array of solvents with varying degrees of hydrophobicity, the fluorescence emission spectra of tryptamine are observed to shift and change in intensity (Figure 7, Table S1). Together they demonstrate the ability for the hydrophobic environment to efficiently quench tryptamine, a phenomenon we propose accounts for the difference in amplitudes between hHINT1 and hHINT2, and a phenomenon that the hHINT2 R99Q variant subtly modulates (Figure 6).

hHINT1 and hHINT2 Inhibitor Binding Affinities Differ and Can Be Modulated by Distant Mutations.

To assess whether the similar steady-state k_cat/K_m values of hHINT1 and hHINT2 are reflected in similar binding affinities for a competitive inhibitor, the K_D values for the previously reported competitive inhibitor 1 (see Figure 8, Table 4) were measured by isothermal titration calorimetry (ITC).(24) Surprisingly, hHINT2 exhibited a K_D = 9.43 μM for 1, whereas hHINT1 has previously been reported by our group to exhibit a K_D = 0.230 μM for the same inhibitor (Figure 8, Table 4). The 40-fold difference in binding affinity between the enzymes equates to a difference of 2.29 kcal/mol of binding energy, which falls within the range of modest differences in hydrogen bonding interactions or an alternative hydrophobic force of equal scale.(38-40) Although hHINT2 experiences a smaller entropic penalty of binding (hHINT1 −TΔS = 8.19 ± 0.13 kcal/mol versus hHINT2 −TΔS = 3.69 ± 0.17 kcal/mol), the significantly less favorable enthalpic contribution observed for hHINT2 outweighs this benefit (hHINT1 ΔH = −17.31 ± 0.05 kcal/mol vs hHINT2 ΔH = −10.54 ± 0.16 kcal/mol), resulting in a higher net free energy value.

Figure 8.

Isothermal titration calorimetry enthalpograms for the interaction between the hHINT inhibitor 1 and hHINT1 Q62A, hHINT2, or hHINT2 R99Q. hHINT1 WT data is referenced from the previously published analysis.(24)

Table 4.

Isothermal Titration Calorimetry Thermodynamic Binding Constants for the Interaction Between Inhibitor 1 and hHINT Proteins

	hHINT1 WTa	hHINT2 WTb	hHINT1 Q62Ab	hHINT2 R99Qb
ΔH (kcal/mol)	−17.31 ± 0.05	−10.54 ± 0.16	−11.50 ± 0.29	−14.72 ± 0.10
-TΔS (kcal/mol)	8.19 ± 0.13	3.69 ± 0.17	3.27 ± 0.21	7.68 ± 0.55
ΔG (kcal/mol)	−9.13 ± 0.11	−6.84 ± 0.32	−8.22 ± 1.38	−7.67 ± 0.56
KD (μM)	0.23 ± 0.01	9.43 ± 0.45	0.83 ± 0.24	6.94 ± 0.51
n	0.98	0.80	0.95	1.1

^aPreviously published by our group(24)

^bThis study

To test whether the hHINT mutations distant from the active site which impacted kinetic constants (hHINT1 Q62A) and burst amplitudes (hHINT2 R99Q) could, in turn, extend to inhibitor binding affinity, ITC binding experiments with inhibitor 1 were performed and compared against their wild-type counterparts (Figure 8). For hHINT1 Q62A, the K_D value for 1 notably increased by four-fold. The primary cause for the increase in the K_D value was a modest reduction in the entropic penalty of binding (Δ = −4.91 kcal/ml) accompanied by a significant loss in the beneficial enthalpic binding contribution, which was similar to the result from hHINT2 (Δ = 5.81 kcal/mol). Only a modest 35% reduction in the K_D value of hHINT2 R99Q was observed compared to wild-type hHINT2. Even though this mutation did not result in a significant change in -ΔG, the hHINT2 R99Q mutation significantly altered the entropic (Δ = 3.99 kcal/mol) and enthalpic (Δ = −4.18 kcal/mol) contributions to the -ΔG. Thus, the distant hHINT1 Q62-E100 interaction and complimentary hHINT2 R99, D137 interaction appear to affect inhibitor binding thermodynamics in addition to the spectroscopic and kinetic alterations we observed.

hHINT1 Crystal Structures Cannot Directly Detect Variant Dynamics.

In an effort to better understand the structural constraints regarding the mutations at Q62 in hHINT1 and their impact on catalytic function and inhibitor binding, we determined the crystal structure of hHINT1 Q62A (Figure S4, Table S2). Despite ¹⁵N-CPMG NMR clearly indicating that the Q62-E100 ion-dipole interaction is highly dynamic, and the Q62A variant kinetics and inhibitor binding thermodynamics showing significant changes, no significant change to the loops, sidechains or perturbations of the water channel in position or B-factors were observed in the structure. At 2.27 Å resolution, the crystal structure does not have sufficient resolution to observe hydrogen atoms, so it is possible that the hydrogen bonding network within the water channel that connects the active site to the Q62, E100 motif had rearranged in response to the mutation. Thus, with this crystallographic evidence alone, we could only verify that the location of the water oxygen atoms were nearly identical to those of WT hHINT1 (RMS deviation of 0.294 Å), leaving protein dynamics as the likely origin of the phenomena imparted by the mutation.

DISCUSSION

It has been shown here using a variety of biophysical and biochemical techniques that the reaction cycles of hHINT1 and hHINT2 can be described and simulated using the same chemical steps. Analysis of the kinetic time courses show that the initial adenylation reactions of the two enzymes occur with the same rate constant within error. However, the subsequent hydrolysis and/or product release reactions occur more rapidly for hHINT2. Crystal structures of the two enzymes indicate high structural homology, and that both enzymes possess homologous water channels connecting a remote surface with the active site. Pairs of surface residues gate these water channels, and exchange of water with bulk solvent occurs rapidly. Detailed evaluation of the amplitudes and rate constants of the kinetic phase during turnover of a fluorescent model substrate, as well as direct determination of inhibitor binding constants, suggests that the interaction of the enzyme with the bound substate differs in the two enzymes due to: (i) differences in the hydrophobic interactions in the substrate environment, and (ii) perturbation of the environment due to the dynamics of the water channel and its gating residue pairs. The origin of these perturbations and the potential bearing on the distinct roles of the two hHINT isozymes are discussed here.

Origin of Altered Substrate Fluorescence Quenching.

Despite the apparently similar rate constants for the initial adenylation reactions of hHINT1 and hHINT2, the fluorescence amplitudes of the first phase of kinetic time courses (dominated by the covalent intermediate formation and release of tryptamine) are quite different. The numerical integration simulation of the time courses revealed that the observed amplitude variance is due to fluorescence yield differences, which is consistent with a difference in the substrate environment. The potential origin of the enhanced quenching effect seen for the TpAd-hHINT1 complex can be appreciated from the crystal structure (Figure 9A) of a catalytically inactive hHINT1 variant, H112N.(41) The structure shows that the tryptamine reporter of the substrate binds proximal to an active site tryptophan (W123), which might effect quenching by π-π stacking. However, the indole rings of W123 and TpAd are not optimally aligned for quenching. Instead, the quenching appears to originate from a tight, hydrophobic CH-π bond (2.7 Å) between the indole ring of W123 and methylene linker group of the model substrate. Previous studies have shown the importance of CH-π bonds in enzymes, notably work by Choy and colleagues which demonstrated the role of a conserved CH-π bond involving a tryptophan residue in the active site of the enzyme Protein Tyrosine Phosphatase 1B, which controls a long-range allosteric interaction. (42-44)

Figure 9.

(A) The E•S complex between TpAd (green) and a catalytically inactive hHINT1 (white), exemplifying the proximal relationship between the indole of the bound substrate and the indole of the active site W123. One aliphatic hydrogen within the tryptamine methylene linker makes a hydrophobic CH-π bond with the pyrrole moiety of the W123 indole (top right). PDB ID: 5KLY. (B) Side view of inhibitor 1 bound to the hHINT1 active site. Two aliphatic hydrogens within the sidechain linker make slightly offset CH-π bonds with each ring of the indole (bottom right). Hydrogens have been predicted and added with the H-add function of PyMol. PDB ID: 5I2E. Expanded views of the complexes are shown to the right.

Inspection of the crystal structure of inhibitor 1 bound to the active site of hHINT1 (PDB ID: 5I2E), also suggests that the binding affinity is influenced by CH-π bonding (Figure 9B).(24) The methylene linker of the inhibitor 1 sidechain displays aliphatic CH-π bonds with the pyrrole moiety of the indole of W123, similar to the CH-π bond observed for the methylene linker of the E•S complex (Figure 9A). In fact, there are two potential CH-π bonds found between the inhibitor and indole moiety of W123 (Figure 9B).(42, 43) It should be noted a tryptophan exists in the same location in hHINT2 (W160).(5) The observed fluorescence yield of the bound substrate or inhibitor complex would be expected to be very sensitive to the precise alignment of the putative CH-π bond. Mutation of W123 was considered for this study, however, previous work from our group demonstrated its absolute requirement for the folding and structural integrity of the enzyme.(2)

The sensitivity of the substrate environment to long-range perturbation is dramatically illustrated by mutations of the surface water channel gating residues. For example, the R99Q variant of hHINT2 subtly alters the quenching of the bound substrate, presumably by altering the CH-π bond to W123 despite being more than 13 Å away at the other end of the water channel. The alteration of the kinetic behavior and thermodynamic binding constants of an inhibitor when using hHINT2 R99Q and hHINT1 variants is also consistent with active site modulations influenced by the water channel. It should be noted that the alteration in the thermodynamic binding constants and kinetic constants reflect different aspects of the reaction cycle. The former is reflective of intact inhibitor binding, whereas the latter quantifies the hydrolysis and/or release of the non-fluorescent portion of the substrate. The effect of the surface variants on intact substrate fluorescence, inhibitor binding, and kinetic parameters indicates an influence on many aspects of the active site structure. Although we were unable to obtain the structure of the hHINT1 Q62A-substrate complex, previous studies have demonstrated that no significant changes are observed between the crystal structures of the wild-type hHINT1-substrate and hHINT1-product complexes.(21) Inspection of the crystal structure of hHINT1 Q62A revealed no significant perturbations by the mutation relative to wild-type hHINT1. Consequently, unobserved dynamic changes both in the water channel and surrounding residues may be responsible for the observed alterations in kinetic behavior.

Role of the Water Channel in Hydrolysis.

Their observation in the crystal structure indicates that the waters in the water channel have fixed positions, and thus have previously been assumed to be locked into the core of the enzyme. However, the MD simulations presented here indicate that the solvent in the water channel can exchange with the bulk solvent (Figure S1). Previous experimental and density functional theory calculations have indicated that the water that participates in hHINT catalytic hydrolysis derives from the bulk solvent, and thus, it is tempting to speculate that the water channel delivers the required water to initiate the second half of the ping-pong reaction.(21, 22) However, the active site terminus of the water channel is adjacent to the active site catalytic histidine residue. Consequently, it is not spatially positioned to deliver water for the required in-line water addition characteristic of the hydrolysis reaction.(21, 22)

Potential Biological Functions of the Dynamic Amino Acid Pair and Water Channel.

The existence of the long-range interaction between the distant enzyme surface and the active site 13 Å away may find relevance in several biological functions of hHINT1. Crystal structures show that the hydroxyl group of Y109 is only 4.9 Å from the side-chain amide of Q62 in hHINT1.(16) Phosphorylation of hHINT1 Y109(16) has been shown to modulate the affinity of the binding interaction between hHINT1 and Microphthalmia-associated transcription factor (MITF) with implications for the expression of the oncogenic genes for which the transcription factor is responsible.(3, 16, 45) Another example of the potential effects of surface residues near the water channel of hHINT1 is found in numerous clinical case reports and genomic analyses which reveal hHINT1 missense mutations as a direct cause of Charcot-Marie-Tooth disease (CMT).(10-13) CMT leads to axonal neuropathic pain and the hyperexcitable state of nerves known as neuromyotonia. One reported missense mutation on the protein surface, I63N, stands out in the context of our study.(13) I63N lies immediately adjacent to Q62 and thus could potentially have an effect on active site function by a mechanism much like that studied here using Q62 variants however further experimentation are required.

The crystal structure of the hHINT1 dimer shown in Figure 10 indicates that the dynamic residue pair Q62 and E100 is exposed on the protein surface and potentially available for effector or partner protein binding. The surface area of hHINT1 available for interactions of this type is quite limited due to the small size of the enzyme (14 kDa) and the amount of surface occluded by the dimer interface. The hHINT1 Q62, E100 motif is located in one of the few available surface regions that does not directly overlay the active site.

Figure 10.

Cartoon representation of the hHINT1 dimer, each monomer shown in green and cyan, based on the crystal structure of the protein complex with AMP which is shown as sticks bound in the active site (PDB ID: 3TW2). The dynamic residue pair (Q62, E100) is shown as space filling spheres.

hHINT1 has indeed been shown to participate in a range of protein-protein interactions that affect metabolic processes in humans. For example, hHINT1 is increasingly appreciated as a regulator of the μ-opioid receptor via interactions with N-methyl D-aspartate receptor (NMDAR).(8, 46-48) hHINT1 binding at the cytosolic C-terminal domain of NMDAR has been demonstrated as a key complex in this regulatory mechanism of opioid tolerance. Other proteins have been shown to assist in this regulatory complex by immunoprecipitation with hHINT1. These proteins include calmodulin, protein kinase C γ and the σ1-receptor.(46) Our group has also previously shown how hHINT1 active site inhibition with competitive inhibitors such as inhibitor 1 can disrupt this cross-talk between NMDAR and the μ-opioid receptor. The mechanism by which hHINT1 causes these effects in unknown, but the existence of a communicative pathway between the active-site and the potential protein interaction surface would account for many of the observations.(9)

Another case in which hHINT1 has been shown to play an important regulatory role through protein-protein interactions involves aminoacyl-tRNA synthetases, notably in its association with Lysyl-tRNA synthetase (LysRS).(3) hHINT1 can associate with LysRS and downregulate one of the intermediate products of the LysRS mechanism, lysyl-AMP, by cleaving the molecule as substrate.

Less is known regarding the protein-protein interactions of hHINT2, but expression levels of the enzyme have been shown to impact mitochondrial influx of ions and small molecules via the mitochondrial Ca²⁺ uniporter (MCU) and mitochondrial permeability transition pore (MPTP) with implications to the onset and ability to treat hepatocellular and pancreatic carcinomas. (19, 49, 50)

Importantly, further defining inherent differences between hHINT1 and hHINT2 by interrogating the mechanics of the long-range interaction between the potential effector binding surface and the active site could facilitate the design of selective hHINT1 inhibitors; an attractive goal in light of growing evidence for the therapeutic analgesic properties and regulation of opioid tolerance by hHINT1.(8, 9, 47, 48) The 40-fold difference in binding affinity found between hHINT1 and hHINT2 for inhibitor 1 (Figure 8) is an encouraging sign that highly selective inhibitors of hHINT1 might be possible to design. These results highlight how nuanced differences can exist between closely related enzymes that may be responsible for tuning their biological function.

Taken together, the results of catalytic, kinetic and structural studies of HINTs provide new insight into how water channels in enzymes can be used to regulate catalysis through mechanisms such as transmission of surface dynamics. Alteration of these dynamics via mutation are found to affect active site catalysis in the case of hHINT enzymes and also may play similar roles in other systems.

CONCLUSION

Implementation of ¹⁵N CPMG-NMR resulted in the identification of a dynamic surface residue motif (Q62, E100) 13Å from the hHINT1 active site at the solvent exposed end of a conserved water channel. Disruption of this dynamic site with mutations (most notably Q62A) proved to have a significant impact on the steady-state and transient kinetics, demonstrating the existence of a communication link between the distant motif and active site function. While mutations at a structurally analogous site in the closely related enzyme hHINT2 (R99Q) did not have as significant an effect on the rate constants, fluctuations in the kinetic time course burst amplitudes were observed, again linking the distant motif to active site behavior. Kinetic simulations and dielectric solvent studies strongly suggest these fluctuations in fluorescent burst amplitudes are the result of differing degrees of fluorescent quenching of the E•S complex in each enzyme variant. The distant surface motif also proved to play a role in the thermodynamics of binding for competitive inhibitor 1, further demonstrating the important link between the active site and this distant motif. Future investigations of hHINT1 and hHINT2 should consider the functional effects of this long-range interaction, and the role it may play in the host of regulatory protein-protein interactions of the hHINTs that define their cellular roles in both natural and disease processes, as well as activation of therapeutic proTides.