Oximes: Inhibitors of Human Recombinant Acetylcholinesterase. A Structure-Activity Relationship (SAR) Study

Acetylcholinesterase (AChE) reactivators were developed for the treatment of organophosphate intoxication. Standard care involves the use of anticonvulsants (e.g., diazepam), parasympatolytics (e.g., atropine) and oximes that restore AChE activity. However, oximes also bind to the active site of AChE, simultaneously acting as reversible inhibitors. The goal of the present study is to determine how oxime structure influences the inhibition of human recombinant AChE (hrAChE). Therefore, 24 structurally different oximes were tested and the results compared to the previous eel AChE (EeAChE) experiments. Structural factors that were tested included the number of pyridinium rings, the length and structural features of the linker, and the number and position of the oxime group on the pyridinium ring.

The SAR of the bis-pyridinium oxime inhibition of AChE obtained from this study will be useful in designing and subsequent synthesis of new peripherally acting reactivators. As these compounds contain quaternary ammonium moieties, they have limited permeability through the blood-brain barrier [9,10]. Furthermore, they may be used as prophylaxis against OPCs or as new drugs for the treatment of Myasthenia gravis, and in anesthetic practice to reverse the skeletal muscle relaxation induced by non-depolarizing neuromuscular blocking agents [11].

Results and Discussion
A large number of pyridinium oximes have previously been tested for their efficacy in reactivating hrAChE. These studies have shown which structural factors are important for activity. These factors included the number of quaternary ammonium fragments, the length and the structural features of the connecting linker between pyridinium rings, and the number and position of the oxime group in the pyridinium ring [12]. In this study, we followed up on our previous efforts to understand the correlation of important structure factors operating during reversible inhibition and in the reactivation process of the hrAChE enzyme. As in the previous EeAChE study [13], the reactivators ability to reversibly inhibit hrAChE is expressed as an IC 50 value.

Mono-Pyridinium Reactivators with Different Oxime Group Positions
The first described reactivator, pralidoxime (2-PAM), has been shown to inhibit AChE activity [3]. This was confirmed according to our results in Table 1, displaying an IC 50 value of 45 mM. Compound 2 with oxime group in the meta position shows a similar value of 41 mM ( Figure 1). Interestingly, 4-PAM fails to inhibit 50% of hrAChE activity with the concentrations used in this study. 2-PAM is considered a more potent reactivator than 4-PAM [14], but we find that 2-PAM also has a greater inhibitory potency. However, a reactivator with two quaternary ammonium functional group atoms has been shown to have higher affinity for both the intact and OPC inhibited AChE compared to these mono-quaternary compounds [15].

Bis-Pyridinium Reactivators with Various Connecting Chain Lengths
The connecting link between the two quaternary ammonium moieties atoms has a significant effect on the reactivation rate and overall toxicity [16]. To observe whether the same effect was present for inhibition potency, we assayed 12 bis-pyridinium reactivators with the oxime group at the para position and varying linker length in the hrAChE enzyme ( Figure 2). Compounds that are commonly shown to be AChE reactivators (1-5 carbons in connecting chains) have lower inhibitory potency. AChE inhibition increases with the length of chain as IC 50 values decrease from 21 mM for 4 to 1 mM for 8 (Table 2). Bis-pyridinium aldoximes with linkers composed of 2-10 methylene groups were tested by Kassa et al [17] and it was confirmed that these compounds were not able to reactivate AChE inhibited by tabun or cyclosarin. Remarkably, based on our current results, the extended linker decreases the IC 50 values to 55 µM for 13. Compounds with linkers of 9-12 methylene groups possess similar IC 50 values. This finding could be a result of the bulkier molecules which allows interaction with more amino acid residues occupying AChE active site. The distance of 10 carbons appears to be ideal in connecting the peripheral and catalytic sites of AChE [18]. Thus, bis-pyridinium reactivators can occupy both sites (peripheral aromatic site and catalytic anionic site) and confer the compounds inhibitory ability towards AChE as confirmed in our study. The connecting linker does not play a direct role in the dephosphonylation process; however, it is important in distribution, elimination, and AChE reactivation rates (e.g., in the binding mechanism) [19].

Bis-Pyridinium Reactivators with 3 or 4 Carbon Connecting Linker and Various Positions of the Oxime Group on the Pyridinium Ring
Tri or tetra-carbon linkers appear to be the most suitable for the reactivation process of soman, tabun, or cyclosarin-inhibited AChE [19] (Figure 2, n = 3,4). These derivatives of trimedoxime (compound 6) were prepared by Musilek et al. and superior reactivation ability was confirmed for them over trimedoxime [20]. Interestingly, the IC 50 for analogue 6 measured in hrAChE is 10 −2 M while other reactivators 16-20 showed values in the mM range (10 −3 M) ( Table 2). The highlighted inhibitor in this series was compound 16 with the oxime group in ortho position (2.2 mM). Similar results were observed for reactivators connected with 4 methylenes (7, 21 and 22) with IC 50 values also in the mM range. Overall, the results point to compound 21 (1.1 mM) with the oxime group in the ortho position.

Bis-Pyridinium AChE Reactivators-Substitutions and Double Bond in Linker
The insertion of a double bond increases the reactivation ability, however toxicity is also increased [20]. Incorporating an oxygen atom instead of a double bond may reduce toxicity. The IC 50 values are similar for 7 and 24, 3 × 10 −3 M and 8 × 10 −3 M, respectively. As a double bond is slightly shorter in length, the linker with a double bond ( Figure 3) is between 3 and 4 carbons and the IC 50 values are reflective of this structure (Table 3). In addition, a double bond does not permit rotation and therefore, passage of the reactivators with this moiety through the narrow AChE cavity may be more complicated. Substitution of oxygen in the connecting chain does not have a big impact on hrAChE inhibition in comparison to other compounds with similar linker size in this study.

SAR for Reversible Inhibition of hrAChE
The inhibition data for 21 bis-pyridinium compounds were used to build a statistical model of inhibitory potential. A principle component regression was performed to determine a SAR for inhibition (descriptors are detailed in the Experimental Section). By far, the largest determinant from this particular set of compounds is the length (>5 carbon atoms) as seen by the contribution of rotatable bonds, molecular weight and molar refractivity of the linker between pyridinium rings. The second major contributor to inhibition potency is the position of the oxime groups on the rings (ortho derivatives resulted in highest anti-AChE activity) ( Figure 4). More potent compounds were predicted from these calculations and will be assessed in future work. As several compounds in the data set are currently approved reactivators (they contain one to three carbon linkers), a drastic reduction in activity can be found among them (IC 50 ~10 −3 M) suggesting a possible negative correlation with reactivation.

Reversible Inhibition of EeAChE versus hrAChE
A set of 24 reactivators were assayed for reversible inhibition of EeAChE by Sepsova et al. [13]. The SAR results are similar to those for hrAChE as presented in this study. In general, the IC 50 values reported in this study for hrAChE are equal to or lower than those reported for the eel enzyme. In the group of bis-pyridinium reactivators with variable length linkers, compound 13 was the strongest inhibitor of hrAChE with an IC 50 of 55 μM, which is 50% lower than in the EeAChE enzyme (100 μM). Compounds 4-6, 10-15, 17, 18, 20 have IC 50 values ranging between two and 40-fold lower for hrAChE in comparison to EeAChE (Table 2 and Figure 5). The seven remaining compounds have equal or higher IC 50 values in hrAChE in comparison to EeAChE. Unfortunately, there is not enough information to determine a cause for the species differences in these IC 50 values. However, based on the inhibition data, the native hrAChE enzyme is more sensitive to inhibition by these oxime containing reactivators. For the most potent inhibitors, therapeutically achievable concentrations equivalent to IC 50 values of 10 −4 and lower are possible [10]. This means that general extrapolation of non-human AChE data needs to account for possible species differences. Although these data are determined from native, non-OPC adducted enzymes, they suggest that reversible inhibition should be included in assays for new lead compounds with reactivation potential.

Molecular Docking
Molecular docking studies were carried out on compound 13 in order to rationalize its plausible interactions within the active sites of hrAChE, EeAChE and Torpedo califonica AChE (TcAChE). The crystal structures of TcAChE with bis(7)-tacrine (PDB ID: 2CKM) and EeAChE (PDB ID: 1C2O) were chosen because of an assumption of a comparable binding mode for 13/hrAChE complex (for hrAChE complexed with fasciculin-2: PDB ID: 1B41).

Figure 5. Comparison of various IC 50 values of bis-pyridinium compounds in hrAChE and
EeAChE. In general, the hrAChE enzyme is more easily inhibited by oxime reactivators than the EeAChE enzyme.
Selection of TcAChE for docking simulation deserves further comment. In vitro assessment was carried out with EeAChE. However, this enzyme is available from the Protein Data Bank (PDB.org) only in very low resolution [21]. To the best of our knowledge, amino acid residues within both TcAChE and EeAChE active sites are conserved and the sequence corresponds one to another preserving important amino acid residues (e.g., for peripheral anionic site: Trp286 in EeAChE = Trp279 in TcAChE; for the active site: Trp86 in EeAChE = Trp84 in TcAChE; the catalytic triad is maintained and residues in mid-gorge are conserved). In order to decide how much the enzymes TcAChE and EeAChE are similar and/or different in quantitative terms, we performed a similarity study of the X-ray structural enzyme models prior to docking simulation. Using two aligning functions-"super" and "align"-available in Pymol viewer, root-mean square deviations (RMSD) of the active sites as well as of the whole enzyme structures were calculated. To approximate the enzyme active sites, residues included in a sphere of R = 8 Å, around His440 in the case of TcAChE and around His447 in the case of EeAChE, were selected. The radius was chosen empirically to encompass the key part of the enzyme active sites. Superimposition of the active site TcAChE model (mobile part) to the active site EeAChE model (stationary part) by the "super" algorithm provided optimal overlap with RMSD (super) = 0.400 Å. Matching residue pairs within the active sites, whose distance was minimized during the RMSD calculations, are listed in Table 4. Applying the "align" algorithm (Pymol) similar results were obtained: RMSD (align) = 0.414 Å ( Figure 6). Alignments restricted to 13 residues set as flexible in docking simulations (see below) yielded RMSD (super) = 0.586 Å and RMSD (align) = 0.536 Å ( Figure 6). Depending on the algorithm used, superimposing of the whole enzymes converged to these values: RMSD (super) = 0.616 Å (3056 atom pairs matched), RMSD (align) = 0.624 Å (3078 atoms pairs matched). Considering only alpha carbons, even lower RMSD values were achieved: C α RMSD(super) = 0.545 Å, C α RMSD(align) = 0.550 Å (Figure 7). In all RMSD calculations, the structures of both enzyme models were treated as rigid.  Overall superimposition of TcAChE on EeAChE colored by C α RMSD ("super" algorithm used, 449 C α carbon atoms matched). Dark blue is a good alignment; higher deviations are in orange/yellow/red. Residues omitted in the alignment are white. Minimum C α distance: 0.04 Å, maximum C α distance: 5.25 Å, average C α distance: 0.71 Å, C α RMSD = 0.545 Å. According to the results obtained via RMSD calculations, we hypothesized significant similarity of TcAChE and EeAChE enzymes. Afterwards, this assumption was re-evaluated in docking simulation rendered in complexes 13/TcAChE and 13/EeAChE. The original spatial conformation of amino acid residues are rendered in blue, amino acid residues in computationally determined alternate conformations are shown in yellow, and the rest of the enzyme is displayed in a cartoon representation. For the sake of clarity, original conformation of Trp279 is not depicted since no structural modification was observed.
Considering 13/TcAChE complex, docking calculations were carried out using the TcAChE crystal structure (Figure 8). The recorded binding energy for 13/TcAChE complex is −9.3 kcal/mol, which is slightly higher compared with the 13/hrAChE complex (−8.7 kcal/mol) but lower than it appeared for 13/EeAChE complex (−10.1 kcal/mol). The distal pyridinium ring exerts a similar spatial conformation as in the hrAChE, providing a hydrogen bond between Trp279 (3.5 Å, correspond with Trp286 in hrAChE) and oxime function at the peripheral anionic site of TcAChE. Next, the peripheral pyridinium moiety is stacked by a T-shaped cation-π interaction to Phe331 (3.7 Å), which is additionally stabilized directly by a parallel sandwiched π-π bonding to Phe288 (3.7 Å). Within the anionic site, Trp84 demonstrates a similar rearrangement as Trp86 within hrAChE with a rotation of ~94°. Moreover, these conformational changes of Trp84 lead to a stabilization of the proximal pyridinium ring via a cation-π interaction (3.4 Å). Supplementary π-π bonding is formed with Tyr121 in a T-shaped conformation which is slightly distorted from its original structure (~28°). The catalytic triad is more disrupted when compared to the 13/hrAChE complex yielding a drastically altered conformation of His440 (~74°). This change in the residue orientation results in hydrogen-bonding with oxime group of proximal pyridinium ring (3.2 Å). Glu199 is not affected in this study, while Ser200 shows a strong hydrogen bond with an oxime group (1.7 Å). Likewise, the 10-carbon spacer bridges the gorge making aliphatic-π interactions with Phe330 (3.2 Å) and Tyr334 (4.6 Å). Related to EeAChE, the corresponding ligand conformation of 13 exhibited larger geometrical differences that do not correlate with the assumption implied by the RMSD calculation results (Figure 9). The proximal pyridinium ring of 13 provided a parallel π-π interaction with Trp86 (3.7 Å) in the CAS of the enzyme active site. This might be additionally stabilized by the hydrogen bond between the OH group from Tyr133 and the oxime moiety (3.3 Å); the interaction was only observed for 13/EeAChE complex. The distal pyridinium ring did not approach Trp286-one of the most important aromatic residues in the PAS-and only provided a T-shaped π-π interaction with Tyr124 (4.2 Å). This might explain the decrease in binding affinity between hrAChE and EeAChE found in in vitro experiments. Furthermore, several hydrophobic interactions contributed to the linker orientation in the mid-gorge of EeAChE (e.g., Tyr341, Tyr337). The presence of 13 in the active site of EeAChE has no impact on catalytic triad. It is important to note that these results must be taken with caution due to low resolution of the EeAChE crystal structure. The results for the hrAChE docking calculation are shown in Figure 10. The original crystal structure positions of important amino acid residues are displayed in blue, those that were computationally reoriented in yellow, and residues having direct interactions with 13 are shown in magenta ( Figure 10). The top-scored docking pose for 13/hrAChE complex (−8.7 kcal/mol) shows important π-π and cation-π interactions ( Figure 10). The peripheral pyridinium moiety is oriented ring-to-ring facing Trp286 (3.6 Å). For this interaction to occur, the indole skeleton of Trp286 (yellow) undergoes a ~63° rotation relative to its conformation in the original position in the hrAChE crystal structure (blue Trp286). Tyr72 is also dramatically reoriented (~59°) to provide hydrogen bonding with the oxime group of the distal pyridinium moiety with a distance of 3.1 Å. Interestingly, Tyr72 is excluded from parallel π-π bonding and gives only a weak π-π, T-shaped interaction (4.2 Å). The largest structural alterations, upon the binding of 13 to hrAChE, are observed for Trp86 in the cation-π site that bind to the distal pyridinium ring (3.7 Å). The atoms of this residue show a ~60° rotation from their original position in the hrAChE crystal structure. The catalytic triad is only slightly affected by 13, a very frail π-π interaction is observed between the distal pyridinium ring and His447 (4.5 Å), which is slightly rotated. The side chain of Glu202 is close to the oxime group (3.7 Å) forming a hydrogen donating bond. In the gorge between both pyridinium units, the 10-carbon spacer spans the length of the active site gorge and is consistent with the idea of a dual binding site interaction as mentioned above. In the middle of the gorge, the aliphatic chain of 13 was surrounded by the phenyl rings of Phe297 (3.4 Å), Tyr337 (4.3 Å) and Tyr341 (3.4 Å) stabilizing 13/hrAChE. The distance between both moieties is approximately 20 Å which is in accordance with the distance between peripheral anionic site and catalytic site of hrAChE [22].
The results from the docking calculations presented in this study are in general agreement with previous docking studies for non-oxime based, bis-pyridinium-inhibitors of cholinesterases, which predicted similar orientations within active site of AChE [23,24]. The low IC 50 values for compound 13 further suggest a characteristic dual binding site that was elucidated by in silico studies. Shorter pyridinium ring based reactivators (e.g., trimedoxime) showed less activity presumably because they do not reach both binding sites in the hrAChE gorge. The docked poses of the 13 in hrAChE and TcAChE interact with similar residues, while 13/EeAChE complex revealed different binding pose unexpectedly to calculated RMSD for TcAChE and EeAChE. Although the conformational changes in the TcAChE enzyme appear to be more dramatic in the catalytic triad residues, the similar binding mode in the gorge suggests that inhibition of both enzymes would be similar in magnitude.

The Measurement of IC 50 of hrAChE
The activities of hrAChE were evaluated by the standard spectrophotometric Ellman's method. Acetylthiocholine iodide was used as a substrate and DTNB was used as the chromogen. The standard wavelength of 412 nm was used [26,27]. The absorbance was determined using a Helios Alpha (Thermo Scientific, Loughborough, UK) spectrophotometer. The results were then analyzed by using the standard statistical software package, Prisma 4.0 (GraphPad Software, La Jolla, CA, USA).
In vitro measurements were completed as follows: A solution of hrAChE (90 μL, activity was previously established) was pipetted into the cuvette. Subsequently, 10 μL of the selected reactivator in concentrations from 10 −1 to 10 −8 M were added. This mixture was then incubated for 10 min under laboratory temperature (20 ± 2 °C). Then, 200 μL of DTNB and 600 μL of phosphate buffer (0.1 M, pH 7.4) were added. The reaction was started by adding acetylthiocholine iodide (100 μL, for 1 μM). This procedure was repeated three times for each incubation.
Reactivators in higher concentrations may split DTNB; this process is known as oximolysis and produces false-positive results [28]. To eliminate this issue, a portion of hrAChE was replaced by distillated water. Subsequently the same portions of other reagents were added. Acquired measurements were then deducted from the calculated hrAChE activity values. Actual activity of the enzyme (the blind sample) was established for all concentration series. The reactivator was replaced by water in cuvette and obtained values were calculated as 100% of the enzyme's activity [13].
Structure activity relationships were calculated with Canvas [29] (version 1.5, Schrödinger, LLC, New York, NY, USA, 2011) from the Schrodinger software suite. Briefly, all default molecular descriptors were calculated for the 21 bis-pyridinium oxime compounds. Principle component analysis was performed to determine which features sample the most variance in the data. These descriptors are AlogP, electrotopological state, hydrogen bond acceptors, molar refractivity, molecular weight, polar surface area, and the number of rotatable bonds. These features were then used in principal component regression to build a model. The model was built with 15 compounds randomly selected from the original set of 20. The remaining 5 were then used in testing the generated model.

Conclusions
The main aim of this study was the determination of a structure-activity relationship for 24 oxime based compounds and their ability to inhibit hrAChE. In addition, we compared the inhibitory potency of these compounds as found in hrAChE to those previously determined in EeAChE. Our results revealed that the important structural factors for the design and synthesis of novel peripherally acting hrAChE inhibitors are firstly, linker length (>5 carbons), and secondly, the positions of the oxime groups on the pyridinium rings (ortho to the pyridinium nitrogen revealed the highest inhibitory potency). Docking calculations justify the low IC 50 values of 13 by predicting its binding pose that altered the conformations of residues in the catalytic triad while simultaneously occupying the mid-gorge and contacted PAS (peripheral active site) in both enzymes. We also showed that the hrAChE enzyme is generally more sensitive to inhibition by these compounds than the EeAChE enzyme. Moreover, reversible inhibition needs to be taken into account when screening new, bis-pyridinium oxime based compounds for reactivation potential.
As these compounds contain quaternary ammonium moieties which limit permeability through the blood-brain barrier, they could be used for medical conditions as peripherally acting agents. Inhibitors of peripheral hrAChE are commonly used in the clinic for treatment of myasthenia gravis or in anesthetic practice to reverse the skeletal muscle relaxation induced by non-depolarizing neuromuscular blocking agents. Several of the compounds presented in this study could be used as new therapeutic leads for these peripheral indications.