Introduction

Organophosphorus (OP) compounds (OPCs) interacting with cholinesterases (ChEs) are represented by two distinct classes, namely, OP insecticides and nerve agents (NAs). OP insecticides pose both severe health and environmental risks (Gunnell et al. 2007). Thousands of cases of pesticides intoxications, including OPs, with fatal consequences have been estimated in some WHO reports referred to studies in last decades (Mew et al. 2017; Gunnell et al. 2017). The second group, NAs, also renders a serious threat to modern society. NAs are classified as the most toxic species of chemical warfare agents (CWAs), which are characterized as weapons of mass destruction and are easily accessible because of their well-documented synthetic procedures (Black and Harrison 2009; Watson et al. 2015). NAs can be further subdivided into three families (Fig. 1): the so-called G-agents and V-agents, distinguished by their physicochemical properties, and the “mysterious” A-agents (Novichoks) (Kuca and Pohanka 2010; Kloske and Witkiewicz 2019) (Fig. 2).

Fig. 1
figure 1

a NAs from the volatile G-family (tabun, sarin, soman, and cyclosarin), the nonvolatile V-family (VX, Russian VX, Chinese VX) and four representatives from the A-family (based on OPCW Schedule A). b Representatives of OP insecticides that were employed within this study (dichlorvos and both types of paraoxon)

Full size image
Fig. 2
figure 2

a Commercially available reactivators. b Molecular design of monoquaternary compounds employing a 2-(hydroxyiminomethyl)pyridin‐3‐ol scaffold (Saint-André et al. 2011) and charged PSLs based on pyridinium, quinolinium, and isoquinolinium moieties

Full size image

The toxic effects of OPCs are based on pseudo-irreversible inhibition of the enzyme acetylcholinesterase (AChE, E.C. 3.1.1.7) and require urgent medical attention. AChE inhibition results in excessive accumulation of acetylcholine (ACh) in synapses and subsequent overstimulation of cholinergic receptors (Marrs 1993). This leads to a fatal scenario associated with a cholinergic crisis and ultimately ends in death through respiratory arrest (King and Aaron 2015). Only immediate administration of causal antidotes, i.e., AChE reactivators, that act as strong nucleophiles can recover AChE activity (Wong et al. 2000). Surprisingly, none of the commercially available reactivators are promiscuous in terms of the AChE reactivation efficacy against all different types of OPs (Worek et al. 2004). There is an urgent need for novel, broad-spectrum reactivators that can address any OP-AChE conjugates (Elsinghorst et al. 2013).

Design

After a comprehensive literature survey along with in-depth structure–activity relationships (SARs), we designed a new family of monoquaternary reactivators. We were inspired by molecules reported in the recent literature and attempted to combine their desirable features (Gorecki et al. 2016, 2017). The designed monoquaternary reactivators reflect the so-called dual binding site strategy. The molecules are endowed with five methylene tether linkers that have been reported to be of optimal length for reaching the catalytic active site (CAS) of the enzyme (Mercey et al. 2012). A 2-(hydroxyiminomethyl)pyridine-3-ol moiety was selected as the CAS ligand for its ability to stabilize phosphyloxime after reactivation (Mercey et al. 2011; Saint-André et al. 2011). The originality of this idea also lies in the presence of a positive charge on the part of the molecule, which presumably anchors the reactivator to the enzyme domain known as the peripheral anionic site (PAS). This is in contrast to the usual approach in which the charged part of the molecule targets CAS. Quinolinium and isoquinolinium scaffolds are well-known PAS ligands (PSLs) with high affinity for both ChEs, AChE and butyrylcholinesterase (BChE; E.C. 3.1.1.8) (Malinak et al. 2020). Similarly, pyridinium fragments are frequently employed as PSLs; e.g., 4-carbamoylpyridinium can be seen in HI-6 and other reactivators (Sit et al. 2014; Zorbaz et al. 2018). In this study, we synthesized 15 novel compounds that were initially screened for their reactivation ability against three NAs (sarin, tabun, and VX) and two insecticides (paraoxon-ethyl and dichlorvos). Three highlighted candidates, K1371 (21), K1374 (24), and K1375 (27), were investigated in detail; their reactivation kinetics and molecular dynamics were evaluated. Finally, the preclinical perspective was assessed by in vivo experiments (testing of acute toxicity, pharmacokinetics, in vivo reactivation efficacy, the protective index and neurobehavioral outcomes using a functional observational battery) in mice.

Materials and methods

Production and purification of recombinant mAChE

Recombinant mAChE (L544Stop) was expressed in Chinese hamster ovary (CHO)-K1 cells as described previously (Carletti et al. 2013). Growth medium containing the secreted mouse enzyme was loaded into an affinity column (Sepharose-4B/procainamide) and eluted with a buffer containing 20 mM Tris–HCl (pH 8), 500 mM NaCl, 0,5 M tetramethylammonium iodide and 2 mM decamethonium chloride. The eluted enzyme was dialyzed overnight at 4 °C against a buffer containing 20 mM Tris–HCl and 500 mM NaCl. The recombinant enzyme was then loaded on a Superdex 200 Increase column (GE Healthcare) and eluted with a buffer containing 20 mM Tris–HCl and 500 mM NaCl. Fractions containing active mAChE were pooled and concentrated using a Vivaspin 30 centrifugal concentrator (molecular weight cutoff (MWCO): 30,000) (Sartorius). The concentrated enzyme was then dialyzed overnight at 4 °C against a buffer containing 10 mM HEPES (pH 7.5) and 10 mM NaCl, aliquoted and flash-frozen in liquid nitrogen.

In vitro reactivation kinetics

Materials

Acetylthiocholine iodide (ATCh) and 5,5′-dithio-bis-2-nitrobenzoic acid (DTNB, Ellman's reagent) were purchased from Sigma-Aldrich (Taufkirchen, Germany). The OP nerve agents tabun (GA), sarin (GB), cyclosarin (GF), VX and Russian VX (VR) (> 95% by GC–MS, 1H NMR and 31P NMR) were made available by the German Ministry of Defence (Bonn, Germany). PXE and PXM were supplied by Dr Ehrenstorfer (Augsburg, Germany). Obidoxime dichloride was purchased from Ferak (Berlin, Germany), and HI-6 dichloride was a kind gift from Dr Clement (Defence Research Establishment Suffield, Ralston, Canada). All other chemicals were from Merck Eurolab GmbH (Darmstadt, Germany) and were of the purest grade available. OP stock solutions (0.1% v/v) were prepared in acetonitrile, stored at RT, and diluted appropriately in distilled water immediately before use. Oxime stock (10 mM) and working solutions were prepared in DMSO.

Preparation of hemoglobin-free erythrocyte ghosts

Hemoglobin-free erythrocyte ghosts served as a source of erythrocyte hAChE and were prepared as described previously (Worek et al. 2012). Prior to use, thawed erythrocyte ghosts were homogenized on ice with a Sonoplus HD 2070 ultrasonic homogenizer (Bandelin Electronic, Berlin, Germany) twice for 5 s with a 20-s interval in between to achieve a homogenous matrix for kinetics studies. Recombinant mAChE was donated by Dr Nachon (IRBA, Bretigny, France).

Reactivation kinetics of oximes against OP-inhibited AChE

OP-inhibited hAChE was prepared by adding a small volume (< 1% v/v) of tabun, sarin, cyclosarin, VX, Russian VX, PXE or PXM to erythrocyte ghosts and incubating the mixture for 15 min at 37 °C to achieve > 95% inhibition of control activity. For removal of excess OP, the inhibited samples were dialyzed in phosphate buffer (0.1 M, pH 7.4) overnight at 4 °C, and residual enzyme activity was measured by incubation of OP-treated and control enzymes (30 min, 37 °C). Until use, aliquots were stored at −80 °C. OP-inhibited mAChE was prepared by adding a small volume (< 1% v/v) of tabun, sarin or PXE to the enzyme and incubating the mixture for 15 min at 37 °C to achieve > 95% inhibition of control activity. Then, the AChE was diluted 1000-fold, and aliquots were stored at -– 80 °C. Next, 150 µL of OP-treated AChE was mixed with an equal volume of phosphate buffer (0.1 M, pH 7.4 (containing 0.2% gelatin for AChE stabilization), and at t = 0, 3 µL of oxime (1–100 µM final concentration) was added to initiate reactivation. After specified time intervals (t = 1–12 min), aliquots were transferred to cuvettes containing phosphate buffer, DTNB (0.3 mM) and ATCh (0.45 mM) for measurement of enzyme activity at 412 nm for 3 min (Worek et al. 1999). All experiments were performed in duplicate.

Data analysis

The pseudo-first-order reactivation rate constants (kobs) were determined by linear regression analysis, and the dissociation constant KD and the reactivity constant kr were calculated from plots of oxime concentration versus kobs by nonlinear regression analysis using Prism 5.04 (GraphPad Software, San Diego, CA, USA) (Worek et al. 2004).

Molecular dynamics

The X-ray structure of the dimer of hAChE (PDB ID: 4EY4) was used as the initial model for molecular dynamics simulations. All crystallographic water molecules were conserved, but ligands, sugars and ions were removed. Missing residues in surface loops (260–261 and 493–494) were manually rebuilt using Coot (Emsley et al. 2010).

We developed a full force field (FF) for the tabun-serine conjugate. The modified O-[(R)-(dimethelyamino)(ethoxy)phosphoryl] serine was built using Chimera (Pettersen et al. 2004) with geometry optimization at the molecular mechanics level. The restrained electrostatic potential (RESP) charges of the modified residue were calculated using the RED Server (Vanquelef et al. 2011) following the recommended procedure, and quantum mechanics optimization of the fragments was achieved with the Gaussian09 package at the HF/6-31G* theoretical level. The new FF was included in the amber14sb FF (Maier et al. 2015). When parameters (bond distances, angles, and dihedral) were not available in amber14sb, they were assigned from the general AMBER FF (Wang et al. 2004) using Antechamber (Wang et al. 2006). The reactivators K1371, K1374 and K1375 were built using Elbow (Moriarty et al. 2009), and topology files were made with acepype (Sousa da Silva and Vranken 2012). Each reactivator was manually docked into the active site gorges of monomers A and B of tabun-inhibited hAChE using PyMOL (Schrödinger). The quinolinium subpart (or pyridinium for K1375) was modeled stacked between Tyr72 and Trp286 in monomer A or stacked only against Trp286 in monomer B, as seen in the structures of mAChE with a phenanthridinium ligand (Bourne et al. 2004). The hydroxypyridine bearing the oxime functional group was modeled in the gorge in the vicinity of Trp86 and His447. His447 was modeled in the alternative conformations seen in tabun-inhibited mAChE (Carletti et al. 2008) H-bonded to the adduct in monomer A or H-bonded to Glu202 in monomer B. The protonation states of His447 (unprotonated) and Glu202 (protonated) were set as recommended (Driant et al. 2017).

The molecular dynamics simulations were carried out using GROMACS version 2020 (Pronk et al. 2013). The protein system was immersed in a 16-Å layer triclinic periodic water box using the TIP5P solvation model (Mahoney and Jorgensen 2000). The system was neutralized by adding Na+ and Cl− counterions at a final concentration of 10 mM.

The molecular dynamics simulations were preceded by energy minimization (1000.0 kJ/mol/nm tolerance) followed by a 100-ps isothermal-isochoric NVT equilibration phase and then a 100-ps isothermal-isobaric NPT equilibration phase. A 300-K temperature coupling scheme using a separated Berendsen thermostat for ions/solvent and protein/ligand was applied with periodic boundary conditions. During the molecular dynamics simulations, the lengths of bonds including hydrogen atoms were constrained by the LINCS algorithm. The time integration step was set to 2 fs. Nonbonded interactions were cut off for distances larger than 12 Å. The particle mesh Ewald method was used for long-range electrostatic interactions (Darden et al. 1993).

Two 50-ns molecular dynamics simulations differing only by the seed used to generate the initial velocities were performed at 300 K under these conditions for each reactivator/enzyme complex, resulting in two trajectories (trajectory A and trajectory B). The trajectories were visualized using VMD (Humphrey et al. 1996) and analyzed using the tools provided in the GROMACS package.

Design of in vivo studies

All presented in vivo studies, including acute toxicity (LD50) pharmacokinetics, AChE reactivation, and protective ratio assays, were performed on young male and female adult mice (BALB/c; body weight 25 ± 5 g; Anlab Inc., Prague, Czech Republic). Groups of 20 mice were housed in the Faculty of Military Health Sciences vivarium at a temperature of 21 ± 1 °C under a 12-h/12-h light/dark cycle. Free access to standard feed (Cerea Corp., Pardubice, Czech Republic) and drinking water was ensured. The minimum acclimatization period was 12 days before the experiment. All experimental procedures and protocols were reviewed and approved by the Ethics Committee of the Faculty of Military Health Sciences, University of Defence in Brno.

Acute toxicity study – LD50

The acute toxicities (LD50 values) of the selected compounds (K1371, K1374 and K1375) were determined. The compounds were administered intramuscularly (in the hind limb) in several continuously recalculated doses. This study was performed using standard probit-logarithmic analysis of death occurring within 24 h. The established LD50 values were used for calculation of relatively safe therapeutic doses (5% of the LD50 values), which were applied in subsequent in vivo studies.

Pharmacokinetics study

K1371, K1374, K1375 and obidoxime were administered intramuscularly following the previously conducted toxicity study. The applied doses were 4 mg/kg K1371, 5.25 mg/kg K1374, 34 mg/kg K1375 or 8.05 mg/kg obidoxime; the solutions were prepared in situ using 20% Kolliphor® EL (Sigma-Aldrich) in 0.9% saline. An additional 3% DMSO was used for K1374 to ensure solubility.

Blood samples were collected directly from mice under deep terminal anesthesia by cardiac puncture into heparinized 1.5-mL tubes at 0, 3, 5, 15, 30, 45, 60, 120, and 240 min (n = 4). The samples were immediately centrifuged at 3,000 × g for 10 min (10 °C), and the plasma obtained was stored at − 80 °C. The animals were perfused transcardially, the skulls were opened, and the brains were carefully removed and stored at -80 °C until HPLC sample preparation (Karasova et al. 2013; Zdarova Karasova et al. 2020).

In vivo reactivation

The in vivo reactivation potential of K1371, K1374 and K1375 was verified by assessment of blood and brain ChE activity after sarin (128 µg/kg), tabun (287 µg/kg) or paraoxon (541 µg/kg) intoxication. For one reactivation study, 56 mice were used, and the mice were divided into seven groups (n = 8). The control group was sham-intoxicated (with i.m. saline) and sham-treated (with i.m. atropine). Atropine was administered to all intoxicated animals at a dose of 10 mg/kg, which was considered sufficiently effective and safe for rodents (on the basis of author experience). Six groups were intoxicated with sarin, tabun or paraoxon (i.m., 1 LD50) and treated after 1 min with atropine alone, a combination of atropine and a selected compound (K1371: 4 mg/kg; K1374: 5.25 mg/kg; K1375: 34 mg/kg), or a combination of atropine and standardly used/recommended oximes (obidoxime: 8.05 mg/kg; HI-6: 34 mg/kg). Blood and brain samples were collected 60 min after intoxication. For sample collection, the animals were deeply narcotized with a CO2 atmosphere, their thorax cavities were opened, and blood samples were collected directly by cardiac puncture into heparinized polypropylene tubes. The blood samples were promptly measured. Then, the animals were perfused transcardially to eliminate blood from the brain vessels. After perfusion, the skulls were opened, and the brains were carefully removed and stored at −80 °C until analysis. Within 3 days, the brains were homogenized in Tris–HCl buffer (0.02 M/L, pH 7.6, 1:10, w/v) with an Ultra-Turrax T25 Basic homogenizer (IKA, Staufen, Germany) to determine ChE activity. ChE activity was tested using a standard reactivation test described previously (Kassa et al. 2011). The percentage reactivation at a specific concentration was calculated according to the following equation:

$$R = \left( {1 - \frac{{\Delta A_{0} - \Delta A_{r} }}{{\Delta A_{0} - \Delta A_{i} }}} \right) \times 100\left[ \% \right]$$
(1)

Protective index evaluation for tabun

The ability of atropine alone and atropine in combination with one of the selected compounds to eliminate tabun-induced lethal effects in mice was determined. Briefly, the LD50 value of tabun and its 95% confidence limit in tabun-poisoned mice was assessed using probit-logarithmic analysis of death occurring within 24 h after i.m. administration of tabun at three to five different doses; six mice received each dose (Tallarida and Murray 1987) Then, the tabun-poisoned mice were treated intramuscularly with atropine alone or with atropine in combination with one of the selected compounds or standards/recommended oximes. The doses and administration routine followed those of a previously described in vivo reactivation study. The LD50 values of tabun and their 95% confidence limits in tabun-poisoned mice treated with antidotes were assessed by the same method. The efficacy of the tested antidotes is expressed as the protective ratio (LD50 value of tabun in protected mice/LD50 value of tabun in unprotected mice). Statistical significance was determined via one-way ANOVA with Scheffe’s post hoc test, and differences were considered significant when p < 0.05.

Results and discussion

Chemistry

The synthesis of these compounds can be subdivided into three main parts. Initially, preparation of the oxime key intermediate 1 started from commercially available 3-hydroxypicolinic acid (2), which was used to synthesize 3, as previously described (see Supplemental Information (SI) (Mercey et al. 2011; Soukup et al. 2018). Hydrogenation of the triple bond with Pd(OH)2/C also resulted in deprotection of the phenolic group. Hydroxyl functionality was consequently protected by tert-butyldimethylsilyl triflate (TBDMSOTf), affording 4. Finally, three successive reactions (ester reduction to aldehyde, deprotection of the silyl group, and oxime formation) were conducted, generating 1. Excellent overall yield of 25% was achieved after ten steps from the starting compound 2 to the key intermediate 1 (Scheme 1a).

Scheme 1
scheme 1

a The synthesis of the key intermediate 1 from commercially available 3-hydroxypicolinic acid 2. Reagents and conditions: ref. (Mercey et al. 2011; Soukup et al. 2018) (see SI); a) Pd(OH)2/C (20%), MeOH/EA (2:1), H2, RT, 1 h; b) 2,6-lutidine, TBDMSOTf, DCM, 0 °C → RT, 30 min; c) DIBAL-H (1 M solution in DCM), DCM, −78 °C, 15 min; d) TBAF (1 M solution in THF), THF, −30 °C, 10 min; e) NH2OH (50% in H2O), absolute EtOH, RT, 30 min. b The preparation of quinoline or isoquinoline-based PSLs. Reagents and conditions: f) phenylboronic acid, Pd(PPh3)4, K2CO3, dioxane, 95 °C, 48 h; g) NaNO3, H2SO4, 0 °C → RT; 20 h; h) Pd/C (10%), toluene, O2, 150 °C, 20 h; i) SnCl2.2H2O, absolute EtOH, 90 °C, 15 h; j) phenylboronic acid, TFA, AgNO3, K2S2O8, DCM, H2O, RT, 20 h; k) NH2OH.HCl, Cs2CO3, FeCl3, H2O, 110 °C, 48 h. c The final N-alkylation reaction yielding the desired monoquaternary reactivators. Reagents and conditions: l) MW, MeCN, 24 h, 90 °C

Full size image

All pyridine-based PSLs used in the synthesis were commercially available. Another synthetic procedure was applied for PSLs that were not commercially available (Scheme 1b). Compounds 18 and 19 were acquired from the corresponding aldehydes 16 and 17 by oxime formation and subsequent reduction of the oxime to a carbamoyl group (Gowda and Chakraborty 2011). Compound 6 was generated by Suzuki coupling with phenylboronic acid. 1,2,3,4-Tetrahydro derivatives 7 and 8 were submitted to nitration to yield the 7-nitro analogs 9 and 10, respectively (Zhu et al. 2003). These two analogs underwent aromatization to form 7-nitroisoquinoline (11) and 7-nitroquinoline (12). The 7-nitro group was readily reduced with tin(II) chloride to produce the final 7-amine-containing PSLs 13 and 14. 7-Nitroisoquinoline (11) was also used for the preparation of PSLs (15) through direct arylation (Seiple et al. 2010) with phenylboronic acid with subsequent reduction of the 7-nitro group to a 7-amino group.

Finally, microwave (MW)-assisted N-alkylation between oxime intermediate 1 and the respective PSLs allowed the formation of final adducts 20–34 (Scheme 1c and Table 1). Notably, several other quinoline-based PSLs (similar to the isoquinolines: quinoline, 2-phenylquinoline, and 4-carbamoylquinoline) were considered for synthesis. However, the final N-alkylation did not proceed, possibly owing to their low reactivity or steric hindrance (data not shown).

Table 1 List of prepared monoquaternary charged reactivators and their IC50 values for hAChE
Full size table

In vitro assays

The prepared compounds possessed moderate to high affinity for native human AChE (hAChE), with IC50 values ranging from 0.5 to 75 µM (Table 1). These results indicate a rather higher affinity for intact enzymes than that of clinically used standards, which could impact the reactivation process beneficially or detrimentally (Table 1). In particular, an affinity for AChE ensures the ability to reach the enzyme for reactivation. On the other hand, elevated affinity could result in increased toxicity due to enhanced AChE inhibition (Petroianu et al. 2007). All novel reactivators were screened for their reactivation ability at a single concentration (10 µM) towards OP-inhibited hAChE after 10 min, as this concentration is considered physiologically attainable after intramuscular (i.m.) administration. Sarin, VX, and tabun were selected as standard NAs, where e.g., sarin was misused in armed conflicts in Syria 2013 (Dolgin 2013). VX was used as an assassination tool in Kuala Lumpur, Malaysia 2017 (Paddock and Sang-Hun 2017). Dichlorvos and paraoxon were taken as insecticide representatives. Paraoxon is an active metabolite of parathion, which is along with dichlorvos still used in several countries in Asia. The reactivation ability was compared with that of pralidoxime, trimedoxime, obidoxime, and HI-6 (see Table S1; SI). Substantial activity was observed mainly for compounds 21, 24, and 27 (further coded as K1371, K1374 and K1375), whose efficacy is outlined in Fig. 3. The ability of the highlighted compounds to reactivate human BChE (hBChE) was further inspected (Fig. 4). hBChE is another member of the ChE family that exists mainly in human plasma. However, its physiological role is of minor importance, and it participates in detoxification processes (Lockridge 2015). On the other hand, BChE could be effectively utilized as a bioscavenger for OPCs, as has been recently proposed, and effective BChE reactivators might serve as pseudocatalytic scavengers that reactivate the OP-BChE conjugate (Broomfield et al. 1991; Kovarik et al. 2010; Sit et al. 2014). In general, the potency of the novel oximes in restoring hBChE activity was low to moderate, although it surpassed the efficacy of standard obidoxime for VX-, paraoxon- and dichlorvos-inhibited hBChE. Notably, K1375 was the most potent hBChE reactivator, whereas K1371 and K1374 were the most effective hAChE reactivators.

Fig. 3
figure 3

In vitro reactivation of OP-inhibited AChE at 10 µM for the three top-ranked compounds. The results are the mean from a minimum of three experiments. AChE activity was measured 10 min after application of the reactivator

Full size image
Fig. 4
figure 4

In vitro reactivation of OP-inhibited hBChE by the highlighted compounds at 10 µM. The results are the mean from a minimum of three experiments. hBChE activity was measured 10 min after application of the reactivator

Full size image

Selected compounds were further investigated for their cytotoxicity and potential to cross the blood–brain barrier (BBB), as poor central nervous system (CNS) availability is one of the major drawbacks of commonly used charged reactivators (Kobrlova et al. 2019). A noncellular parallel artificial membrane permeation assay (PAMPA) model simulating passive diffusion and a cell-based model using Madin-Darby Canine Kidney (MDCK) cells expressing various BBB transporters were utilized. Under both experimental conditions, the novel reactivators exhibited only a weak ability to cross the BBB, similar to clinically used standards (Table S3).

Notably, cell viability assays on hepatocellular carcinoma HepG2, neural astrocytoma 1321N1 and adherent epithelial CHO-K1 cell lines showed rather low toxicity (Table S4). Interestingly, the most potent reactivator, K1374, was the most toxic, with IC50 values ranging from 59–220 µM (for all used cell lines), whereas K1375 exhibited the least cytotoxicity, with IC50 values above 1 mM.

To further inspect reactivation ability, the kinetic parameters of the highlighted compounds (K1371, K1374 and K1375) were explored in vitro for tabun-, sarin-, cyclosarin-, VX-, RVX-, paraoxon-ethyl (PXE)- and paraoxon-methyl (PXM)-inhibited hAChE. The following parameters were evaluated: KD (µM), the apparent affinity of the reactivator towards the OP-AChE conjugate; kr (min−1), the rate by which the oxime cleaves the covalent bond between the enzyme and OP; and kr2 (µM−1 min−1), the bimolecular reactivation rate constant (Table 2). Compared to our initial screening, this assay yielded rather ambiguous results regarding the efficacy of the novel compounds. On the other hand, broad reactivation profiles were confirmed for the novel oximes. The parameters indicated that the high efficacy of K1371 and K1374 was mainly caused by augmented affinity for the inhibited enzymes (low KD values). The first-order rate constants (kr values) suggested a relatively weak or moderate ability of the set of developed monoquaternary reactivators to cleave the OPC-AChE conjugate. Therefore, standards containing charged warheads such as pyridinium-2-aldoxime (in HI-6) or pyridinium-4-aldoxime (in obidoxime) might still exhibit an elevated reactivation ability. On the other hand, we assumed that the PSLs in K1371 and K1374 were better accommodated to the PAS region of AChE than 4-carbamoylpyridinium (HI-6 and K1375) and pyridinium-4-aldoxime moieties (obidoxime). Thus, they allowed proper orientation of the reactivating moiety at the bottom of the gorge, and their stronger binding resulted in improved overall reactivation velocity (kr2). In particular, K1371 and K1374 were effective in the case of tabun-hAChE and both types of paraoxon-hAChE, whereas HI-6 completely failed. Furthermore, K1374 was the most efficient for the VX-inhibited enzyme. Pralidoxime was the least potent reactivator that did not overwhelm the reactivation ability of the novel monoquaternary reactivators.

Table 2 Determination of reactivation kinetics of OP-inhibited hAChE. Reactivation kinetics were determined in duplicate using six different concentrations
Full size table

As a next step, we used OPC-inhibited Mus musculus AChE (mAChE) to extrapolate in vitro data to data obtained from in vivo studies performed on mice. In this case, the novel compounds K1371, K1374 and K1375 were even more efficacious in reactivating tabun, sarin and PXE-mAChE conjugates than obidoxime and HI-6 (Table S2). The structural differences in the primary amino acid residue sequence made it responsible (90.4% identity between hAChE Uniprot entry Q9Y215 vs mAChE Uniprot entry O35348) to generate a slightly improved reactivation potency for mAChE-inhibited adducts.

Molecular dynamics

Reactivation of the tabun-hAChE conjugate by oximes is commonly the most difficult reaction to achieve due to i) poor accessibility, ii) the electrophilicity of the phosphorus atom, and iii) alternate conformations of the catalytic histidine that are less favorable for reactivation.(Worek et al. 2004; Carletti et al. 2008) However, K1371, K1374 and K1375 compared very well in terms of reactivation efficiency to the best tabun-hAChE reactivators known so far (Fig. 3). The origin of this remarkable efficiency warranted examination in a molecular model. Thus, we studied the binding of these reactivators to tabun-hAChE by means of molecular dynamics simulations. Our goal was to analyze the stability of the complexes formed between the inhibited enzyme and the oxime and to verify whether the orientation of the oxime function relative to the tabun adduct remained favorable for the reactivation reaction to proceed.

We simulated the dimeric enzyme, with both active sites occupied by one molecule of reactivator but with slightly different initial conformations. These conformations were based on the binding of apparent phenanthridinium molecules to mAChE (Bourne et al. 2004). In monomer A, the phenylquinolinium (or pyridinium for K1375) was stacked between Trp286 and Tyr72 of the peripheral aromatic site, and the catalytic histidine was in the usual "triad" conformation, i.e., at H-bonding distance from Glu334. Notably, we did not model the pyridinium of K1375 stacked between Trp286 and Tyr124, as is usually seen, for example, for HI-6 (Allgardsson et al. 2016), because the linker of K1375 is longer (5 atoms vs. 3 atoms) and thus less favorable to this kind of interaction. In monomer B, the phenylquinolinium (or pyridinium) was stacked against Trp286 in its most common conformation, and the catalytic histidine was in the alternate conformation seen in tabun-AChE conjugates (Carletti et al. 2008), i.e., at H-bonding distance from Glu202. These initial models enabled exploration of a larger conformational space for the peripheral-site binding of the reactivator. We performed two 50-ns simulations for each setup, thus totaling 200 ns of simulation for each individual molecule of reactivator.

For each complex, we monitored the root mean square deviation (RMSD) of the heavy atoms of the reactivator relative to the enzyme backbone and the sum of short-range Lennard–Jones and Coulombic interaction energies between the reactivator and the enzyme conjugate (Figure S1 and S2). The average values and standard deviation are reported in Table 3.

Table 3 Molecular dynamics trajectory analysis. Average values ± SDs
Full size table

The RMSDs for the ligands in monomer A were mostly below 2.0 Å, demonstrating that stacking of the peripheral pyridinium/phenylquinolinium between Trp286 and Tyr72 strongly anchors the molecules at the gorge entrance. By comparison, the RMSD values for the ligands in monomer B without double stacking were larger (1.2–4.8 Å range), revealing weaker anchoring.

The interaction energy was stable over the trajectories and not significantly different for the set of three molecules, remaining in a very favorable range (−50/−60 kcal/mol). The ranking we observed, with slightly stronger interactions for K1374 and slightly weaker interactions for K1375, correlated with the sizes of the peripheral parts of the molecules (pyridinium vs. phenylquinolinium); accordingly, larger aromatic systems allow more interactions with the PAS at the gorge entrance.

We monitored the distance between the tabun adduct (P atom) and the oxime functional group of the reactivator (O atom) as well as the angle formed between the atoms Ser203Oɣ, P and O of the oxime functional group to assess if their relative spatial orientation was favorable for nucleophilic attack. We considered that a distance of approximately 5 Å and an angle near 160° are favorable for nucleophilic attack from the opposite side of the catalytic serine (apical position) and reactivation by in-line displacement (Nachon et al. 2010; Allgardsson et al. 2016). An angle near 100° with a distance of approximately 5 Å remains favorable for adjacent nucleophilic attack followed by pseudorotation of the putative trigonal bipyramidal transition state (Nachon et al. 2010). We also monitored the distance between the geometric center of the aromatic rings of the reactivator and the geometric center of the aromatic residues lining the gorge to assess which residues were the main contributors to the interaromatic interactions. A representative snapshot of each molecule in both monomers is shown in Fig. 5.

Fig. 5
figure 5

taken from trajectory 1 at 8 ns for K1371, trajectory 2 at 24 ns for K1374, and trajectory 1 at 45 ns for K1375. The upper panels display the molecule in monomer A, and the lower panels display the molecule in monomer B. The reactivators are represented with balls and sticks, and the serine-tabun conjugate is represented with sticks, with carbon atoms in cyan. The residues lining the active site gorge are represented with sticks, with carbon atoms in white. Water molecules are displayed as lines. The black dashed lines render H-bonds. The distance d(P-Ox) between the phosphorus atom and the oxygen of the oxime functional group is given in Å

Representative snapshots of the molecular dynamics simulations of K1371, K1374 and K1375 bound to hAChE inhibited by tabun. The snapshots were

Full size image

For K1371, the P-O distance remained within 5 Å for both monomers in both trajectories, with an angle generally larger than 150°; both values supported a conformation favorable for the reactivation reaction (Table 3 and Figure S3/S4). This stability was provided by the interaction of the phenylquinolinium with Trp286 and that of the pyridine oxime with Tyr341 (Figure S3/S4). We noticed that K1371 was logically less stable in monomer B than in monomer A due to a weaker interaction with Tyr72.

The binding conformations appeared relatively unfavorable for K1374, with a P-O distance mostly longer than 7 Å, except for monomer B in trajectory 2 (Table 3 and Figure S5/S6). These nonoptimal conformations were stabilized by the formation of H-bonds between the oxime and the main-chain carbonyl of Tyr72 or His447 (Fig. 5; K1374 top panel) and were thus held away from the adduct. However, the conformation in monomer B and trajectory 2 proved that an optimal conformation could be reached (Fig. 5; K1374 bottom panel). During the course of the simulations, the supplementary amino group did not exhibit additional interactions with the protein. Instead, it formed H-bonds with surrounding water molecules, suggesting that this group is a minor contributor to the binding of the molecule.

Similar to that of K1371, the oxime function of K1375 remained mostly within 5 Å of the tabun adduct during the course of the simulations (Table 3 and Figure S7/S8). This binding conformation was stabilized by strong interactions of the isonicotinamide group with Trp286 (and Tyr72) at the gorge entrance and interactions of the pyridine-2-aldoxime with Tyr337 and Phe338 in the lower part of the gorge (Fig. 5; K1375). We also observed H-bond formations between the amide and His287 during the simulation. This dual anchoring was enabled by the length of the 5-carbon linker bridging the two aromatic moieties. By comparison, for HI-6 with a three-atom linker, the isonicotinamide moiety was stabilized by intercalation between Trp286 and Tyr124 and was thus situated lower in the gorge. The longer linker of K1375 provided it with higher flexibility than HI-6, allowing the oxime functional group to closely approach the tabun adduct.

In summary, the molecular dynamics showed that K1371, K1374 and K1375 were well stabilized within the active site gorge of tabun-inhibited hAChE via strong anchoring to the PAS of the enzyme and had linkers providing sufficient flexibility to let the oxime functional group approach the phosphorus atom of the tabun adduct at a distance favorable to the reactivation reaction.

In vivo experiments

Encouraged by the promising in vitro data, we investigated the acute toxicity and tolerability of the highlighted compounds (i.e., K1371, K1374 and K1375) to determine their median lethal dose (LD50) values after i.m. administration in mice. In accordance with the in vitro cytotoxicity data, K1375 was the least toxic, with an LD50 value comparable to that of HI-6 or methoxime (Table 4). In contrast, K1371 and K1374 showed distinctive toxicity profiles (Table S5; SI). Although K1371 and K1374 are permanently charged compounds, as are all the compounds in the series, the bulkiness of their PSL was responsible for their enhanced lipophilic character, which may have augmented their toxicity. The elevated lipophilicity also resulted in delayed mortality; most of the animals died after a prolonged period (more than 1 h). In the case of high doses of K1375, death occurred within the first hour (Table S5; SI). Notably, K1375 was found to be the compound with the weakest inhibitory activity towards native hAChE, which may also imply that it has low acute toxicity, as overdosing by AChE reactivators results in symptoms associated with AChE inhibition (Bartosova et al. 2006).

Table 4 Acute toxicities expressed as LD50 (mg/kg) values, determined after i.m. administration in mice
Full size table

Subsequently, the 5% LD50 values were used as therapeutic doses for further in vivo assays. Initially, pharmacokinetic (PK) profiles were established and compared with that of the standard obidoxime (Table 5, Fig. 6). Unfortunately, the CNS availability or BBB permeation ability was in line with the in vitro results, predicting very low permeability (Table S3). The tested compounds were minimally distributed to the CNS, presuming their central availability was below 1% of the plasma concentration, which is in accordance with that of clinically used bis-pyridinium standards (Bajgar et al. 2007).

Table 5 Pharmacokinetic parameters of the highlighted compounds and the standard obidoxime after i.m. administration in mice
Full size table
Fig. 6
figure 6

Pharmacokinetics of the tested compounds in plasma after i.m. administration in mice

Full size image

Generally, the blood concentration of OPC antidotes must increase quickly to ensure that a sufficient amount of oxime is available for reactivation after poisoning. It is also important that the blood levels are maintained to enable reactivation of the inhibited ChEs as long as possible. Thus, some trade-offs had to be accepted for the optimal reactivator. The hydrophilic compound K1375 possessed a PK profile similar to that of obidoxime with an elimination half-life (t1/2) = 9 min. On the other hand, the rate and extent of absorption were slightly lower than those of obidoxime. The more lipophilic compounds, such as K1371 and K1374, were endowed with delayed half-lives, enabling them to exert extended effects while preserving a relatively fast rate of onset (time to reach maximum plasma concentration (Tmax) = 15 min. However, the novel reactivators exhibited much lower bioavailability than obidoxime, which resulted in markedly reduced peak plasma concentration (Cmax) values at the administered doses. Distinct declines in bioavailability were evident for the relatively lipophilic oximes K1371 and K1374.

The highlighted compounds were further evaluated for their in vivo reactivation abilities (Fig. 7). Mice were intoxicated with OP (i.m.), a treatment was given 1 min after poisoning (i.m.), and blood reactivation was measured after 60 min. Atropine (10 mg/kg), a clinically used symptomatic drug, was coadministered with the oximes (King and Aaron 2015). The newly developed compounds displayed results comparable to those of obidoxime; only HI-6 was considerably more effective against sarin poisoning. K1375 yielded the highest efficiency among the novel reactivators, especially against PXE intoxication. This could be explained by the higher administered dose and thus the higher area under the concentration–time curve (AUC) value of K1375 compared to K1371 and K1374 and by the higher efficacy of K1375 towards inhibited BChE. Notably, reactivation ability was measured from the blood plasma, where both ChEs reside. Although acetylthiocholine (ATCh) was used for quantification of AChE activity by Ellman's assay, it can also be hydrolyzed well by BChE (Ellman et al. 1961; Lockridge 2015). Since BChE can act as a bioscavenger for free OPCs in blood, a non-selective ChE reactivator would likely provide higher potency to counteract OPCs (Rosenberg et al. 2017, 2018). Furthermore, because of the low toxicity of K1375, this compound could be applied at almost tenfold higher doses than the other compounds. On the other hand, it remains unclear whether the administered doses of 5% of the LD50 were sufficient and whether higher doses would deliver better therapeutic effects, since some authors have suggested administration of up to 25% of the LD50 values (Sit et al. 2018) and other dosages based on the no observed adverse effect levels (NOAELs) (Hepnarova et al. 2019; Calas et al. 2020).

Fig. 7
figure 7

In vivo blood reactivation of OP-poisoned mice. Reactivators (dose: 5% LD50) were intramuscularly coadministered with atropine (10 mg/kg) 1 min after intoxication, and AChE activity was measured 60 min after poisoning. HI-6 was used for comparative purposes only in the context of sarin intoxication due to its low efficacy against tabun and PXE

Full size image

Tabun intoxication countermeasures were selected for detailed investigation. Initially, we determined the in vivo therapeutic indexes of K1374, K1375 and obidoxime for protection of animals against lethal doses of tabun (Table 6). Surprisingly, obidoxime was considerably more effective than K1375 or K1374, while the efficacy of K1375 only slightly exceeded that of K1374. These results are somehow contradictory to our findings from the in vivo reactivation experiment (Fig. 7), perhaps because of the faster and more complete absorption of obidoxime (Fig. 6). Thus, rapid ChE reactivation is probably more critical than the amount of restored ChE activity over time. However, subsequently performed functional observational battery (FOB) assays highlighted the oxime K1374 as the drug that most effectively reduced the toxic effects of tabun poisoning. Tabun caused passive behavior of mice during handling and catching, nasal secretions, and piloerection during both time intervals of assessment. The unprovoked activity and mobility were significantly decreased, and these changes were accompanied by hyperkinesis, ataxia, and low muscular tonus. Although atropine administered alone was able to reduce some signs of tabun toxicity, coadministration of the oximes restored more of the observed functions (Tables S7-S10; SI). In summary, the oxime K1374, unlike obidoxime, improved overall activity and the distance between hind paws after a jump. In contrast to K203, a highly efficient anti-tabun experimental reactivator (Gorecki et al. 2019), K1374 decreased panic after the click response. After treatment with K1375, overall movement was initially slightly reduced, and arousal occurred 2 h after administration.

Table 6 Therapeutic/protective indexes of a selected treatment strategy to mitigate the acute lethal effects of tabun in mice
Full size table

Conclusion

In this study, monoquaternary reactivators with a permanent charge outside the reactivating warhead were developed, and their reactivation potency was assessed both in vitro and in vivo. Among the reactivators, three compounds, namely, K1371, K1374 and K1375, were highlighted for their improved in vitro properties and primarily versatile reactivation profiles, including for tabun-inhibited AChE. In vivo, the compounds turned out to be similarly effective as the clinically used standards in an OPC-intoxicated mouse model. We assume that the reduced efficacy of K1371 and K1374 can be attributed to their decreased bioavailability and reduced doses, as required because of their acute toxicity. On the other hand, because of its long half-life and thus its prolonged action, K1374 was the most effective in eliminating the toxic signs of tabun as seen in the FOB experiment, even though obidoxime exhibited a better protective index than K1374 against tabun poisoning. In contrast, the least toxic compound, K1375, proved to be the most effective and versatile reactivator in vivo from the series, although its activity was not as profound as in in vitro. This finding is important, because in vitro reactivation assay results are considered primary selection criteria in the search for highly active reactivators. Thus, we suggest that other essential factors, such as physicochemical properties, toxicity and pharmacokinetic profiles, must be considered in the search for potent AChE reactivators. Similarly, non-selective ChE reactivators could be more efficient in OP poisoning countermeasures as they would also contribute to accelerated BChE-mediated detoxification of an OP agent in the blood. Despite the fact that the availability of K1374 and K1375 in the brain did not surpass that of the currently marketed reactivators, these compounds exhibited significant and broad-spectrum reactivation ability superior to that of the clinical standards. We speculate that BBB permeation is a critical factor for reactivators, as a recent study describing highly potent in vitro reactivator with fast-onset CNS availability found that the efficacy of that reactivator could not surpass that of HI-6 in vivo for VX and sarin poisoning (Zorbaz et al. 2020). We also hypothesize that if blood and peripheral ChE are reactivated rapidly enough, central AChE inhibition could be markedly decreased or prevented. Taken together, the findings indicate that both compounds, K1374 and K1375, could indeed be valuable templates for OP countermeasures based on their broad-spectrum reactivation ability, although each of them is unique. Future in vivo assessments using higher applied doses will better reveal the real impacts of these compounds.

Associated content

The Supporting Information is available free of charge on the ACS Publications website at DOI: It contains information experimental sections devoted to chemical synthesis, NMR and LC–MS spectra, in vitro data analyses (reactivation potency, cytotoxicity, BBB permeation estimation), molecular dynamics, design of in vivo studies and sample preparation for the HPLC–MS study as well as data from the FOB.