Prior to any optimization effort to improve this chemotype’s ability to inhibit TGR and effectively act as an anti-schistosomal agent we needed to establish a mechanistic link between NO donation and TGR inhibition. As such, our first considerations involved the synthetic deconstruction of the molecular structure of 2
. Various synthetic pathways for oxadiazole-2-oxides have been reviewed by Gasco and coworkers.22,23
Additionally, Maloney and coworkers have recently described the synthetic elaboration of the small molecule oxadiazole-2-oxides described in this report.44
From the outset, an examination of the N
-oxide moiety was of paramount importance. To better understand this key structural feature of the compound we examined analogues that lack (derivative 4
) and regiochemically transpose (derivative 5
) the N
-oxide (). As opposed to furoxan (2
), both 4
gave no appreciable NO release with TGR nor was either analogue active versus TGR or an effective ex vivo
worm killing agent. Our next consideration was to determine the importance of the nitrile functionality. To explore the structure activity relationships (SAR) of this moiety, we considered analogues along the scale of oxidation states (i.e. alkyl to acyl). Several analogues were accessible intermediates of synthesis of 2
(including derivatives 7
) while others required a separate synthetic effort (derivatives 6
The SAR of these analogues provided the first indication of the complexity surrounding the relationship between NO release, TGR inhibition and worm killing (described through visual examination of cultured worms treated with 10 μM of compound over a 72 hour time-course). The methyl analogue 6
was inactive as a NO donor and a TGR inhibitor and was essentially inactive in the worm killing evaluation (). The hydroxyl analogue 7
regained the ability to inhibit TGR (IC50
of 11.3 μM) but was only a weak NO donor and possessed minimal ability to kill ex vivo
worms. The aldehyde 8
and carboxylic acid 9
were noted to be potent inhibitors of TGR (IC50
of 0.11 μM and 0.63 μM, respectively). Surprisingly, these analogues possessed only modest NO donating abilities and were ineffective worm killing agents. Interestingly, the aldehyde moiety in 8
would be capable of entering into electrophilic/nucleophilic reactive mechanisms with cysteine (and selenocysteine) residues in a similar fashion to the nitrile of 2
. The only analogue in this SAR series that retained both TGR inhibition and worm killing ability was the amide 10
, albeit at slightly lower potency and efficacy. With these results in place it seems reasonable to consider the 3-cyano-1,2,5-oxadiazole-2-oxide core as the pharmacophore for these agents in terms of TGR inhibition, NO donating potential and worm killing efficacy.
Structure activity relationships of the N-oxide moiety.
SAR of Nitrile Substitutions
With a verified and consistent definition of the pharmacophore, we next turned to a systematic evaluation of the phenyl ring through the appraisal of standard substitutions (including methyl, methoxy, halo, trifluoromethyl, nitro, hydroxyl and phenyl) and replacements (including heterocycles furan and thiophene). The results are detailed in and show clear evidence that the majority of analogues preserving the core 3-cyano-1,2,5-oxadiazole-2-oxide moiety (11–18
) retain TGR inhibition and worm killing capacity (11–18
were additionally noted to be capable NO donating molecules). The SAR trends revealed that electron-withdrawing substitutions (halo, trifluoromethyl and nitro) provided slight potency enhancements and the meta-position seemed favored [in general, substitutions at the ortho position were not synthetically viable (compound 21
was the lone exception)]. Importantly, there was general agreement between TGR inhibition and worm killing efficacy. For instance, analogues 12–17
possessed TGR IC50
between 2 μM and 4 μM and were at least as efficacious as furoxan (2
) in terms of worm killing. Conversely, several analogues with TGR IC50
values between 9 μM and 18 μM and were generally less active against worms in culture. From this phenyl ring scan there were several intriguing lessons. The unique difference between substitutions at the meta and para positions was further highlighted by the hydroxyl analogues 19
. Analogue 19
possessed good TGR inhibition but was inactive as a worm killing agent, while 27
was among the least potent analogues for TGR inhibition yet retained respectable activity versus worms. Replacement of the phenyl ring with heterocycles furan and thiophene (analogues 28
, respectively) resulted in derivatives with slightly improved potencies for TGR inhibition, good NO release profiles and good worm killing efficacy. Another important exploration involved the substitution of the 4 position of the oxadiazole ring with a substituted carbonyl. Data from our initial studies on compounds with acyl substitutions on the oxadiazole ring system (such as in 3
) and analogues presented here (such as aldehyde 8
) demonstrate that such functionality confers significant potency enhancements in terms of TGR inhibition. These analogues, however, are devoid of the key 3-cyano-1,2,5-oxadiazole-2-oxide pharmacophore and were markedly less active versus cultured worms. We were, therefore, eager to examine an analogue that possessed both the nitrile moiety and a ketone moiety. As such, we synthesized a novel oxadiazole-2-oxide (analogue 30
) that satisfies these requirements (it should be noted that 30
required an entirely novel synthetic route spanning a total of 9 transformations44
). The inhibition of TGR by 30
was in alignment with our previous results for 3
(potency of 63 nM). Unfortunately, like the aforementioned analogues, 30
was devoid of activity versus worms in culture.
SAR of Phenyl Ring Substitutions and Substitutions on the 3 and 4 Carbons of the Oxadiazole Ring.
The results of these studies provide solid evidence that inhibition of TGR by oxadiazole-2-oxides must be coupled with the ability of these small molecules to act as a NO donating compound in order to effect parasite death. A major question, however, concerns the role that molecular NO plays in TGR inhibition and ultimately worm killing. That NO can modify protein structure and function by the act of S-nitrosylation was first shown in the early 1990s.45,46
Becker et al subsequently showed that human GR can be inhibited by NO donating compounds and an active site cysteine residue is oxidized as a result.47
Additionally, S-nitrosylation of cysteine residues on coxsackieviral protease 2A, 3C and rhinovirus 3C is recognized to affect coxsackievirus and rhinovirus, respectively.48–50
Given the aforementioned precedence and the intriguing presence of 13 cysteine residues and one selenocysteine in TGR, we set out to examine the potential role of S-nitrosylation (or Se-nitrosylation) during TGR inhibition by 2
. To explore this possibility, we took advantage of the well known biotin-switch assay introduced by Snyder and coworkers.51,52
A cartoon representation of the principal experiment is outlined in . Here, recombinant TGR was incubated with and without NADPH, furoxan (2
) and glutathione (GSH). After incubation the reactants were removed from TGR on a spin column and by acetone precipitation of TGR. All cysteine (or selenocysteine) residues that remain in their reduced sulfhydryl form are then capped by methyl methanethiosulfonate (MMTS). TGR was then treated with reduced ascorbate to return all S-nitrosylated cysteines (or Se-nitrosylated selenocysteines) to their reduced form, followed by reaction with the thiol-modifying reagent biotin-HPDP to allow visualization of labeled protein. After purification of the protein on a spin column and acetone precipitation it was analyzed by native-Western blotting using a HRP-streptavidin conjugate to identify biotinylated proteins (). The results shown clearly indicate that recombinant TGR is S-nitrosylated by 2
and that the nitrosylation depends on the protein activity (i.e. requires NADPH) (). The experiment was performed with pure, recombinant TGR and analyzed on a native gel; additional bands are likely the result of both monomeric and dimeric proteins and multiple S-nitrosylation events. No TGR labeling was observed when incubation was with GSH or in the absence of NADPH.
Figure 3 A. Cartoon representation of the Biotin-switch assay (see ref 49 and 50). B. Detection of S-nitrosylation in recombinant S. mansoni TGR after reaction with furoxan (2) 20 μM (lane 1) and 10 μM (lane 2) ± NADPH. The molecular weight (more ...)
The results from this experiment strongly suggest that NO plays a significant role in TGR inhibition and likely in parasite killing. Based upon this knowledge, our thoughts regarding the optimization of the oxadiazole-2-oxide chemotype focused on ways to optimize the NO donating qualities of individual compounds. Our previous attempts to optimize the compound through a standard phenyl ring scan () provided evidence that the electron-withdrawing nature of the substituent at the 4-position of the oxadiazole ring was beneficial. While this makes mechanistic sense, even the most electron deficient phenyl rings provided only modest potency enhancements in terms of TGR inhibition and no discernable benefit in worm killing. Our next consideration was to explore analogues capable of delivering two equivalents of NO. To this end, we synthesized and evaluated several bis-oxadiazole-2-oxide derivatives linked through substituted phenyl rings and heterocycles. The results for these agents are shown in and demonstrate the validity of the theory. All bis-oxadiazole-2-oxide analogues were capable of TGR inhibition at potencies exceeding furoxan (2) and worm killing profiles also demonstrated similarity to furoxan (2) and in some cases modest improvements (analogues 33, 34, and 35 were examined in a separate experiment). Worm death will not occur until the GSH/GSSG ratio reaches a critical point and the overall worm redox balance is unrecoverable, making it difficult to derive correlations between worm killing and TGR inhibition below a certain level.
SAR of Bis Oxadiazole-2-Oxides
With a growing understanding of this chemotype’s potential and the advancements of several analogues with improved properties, it was important to consider the mechanism of NO donation from oxadiazole-2-oxides. A mechanistic rationale for NO donation that is initiated through nucleophilic attack (presumably by the sulfhydryl moiety of a cysteine residue or by selenocysteine) at either the 3 or 4 position of the oxadiazole ring and subsequent rearrangement of the heterocycle in a manner that allows release of the nitroxyl anion (NO−
) has been presented.22,23
An enzymatic oxidation is posited to transform this agent to NO (NO·
). It is important to note that the nitroxyl anion (NO−
) exists as a triplet that reacts very rapidly with triplet dioxygen to form peroxynitrite ion. Thus, it is unlikely that the transformation from NO−
occurs via oxidation by molecular oxygen. Additionally, the previously reported mechanism27
suggests a permanent ligation between the attacking nucleophile and the rearranged carbon scaffold of furoxan (2
). While this mechanism is plausible, there are several aspects that are not consistent with the data from our study. Foremost, this mechanism does not account for the release of NO (rather than NO−
) during a reaction with cysteine (i.e. in an environment lacking an enzyme capable of oxidizing NO−
). A second issue is the stoichiometry of the processes. Our previous data suggests that TGR acts catalytically on furoxan (2
) in terms of NO release. A 10 μM concentration of 2
in the presence of 15 nM of TGR and NADPH resulted in the generation of ~ 15 mole % of NO relative to 2
; i.e. ~ 1500 nM NO from only 15 nM of TGR -or- 100 equivalents of NO per enzyme equivalent. Obviously TGR has the ability to attack furoxan (2
), effect molecular rearrangement and release of NO, and subsequently release from the heterocycle scaffold in such a manner that it retains the ability to re-enter the cycle with another molecule of furoxan (2
). Presumably, this cycle continues until a critical concentration of NO is present and the enzyme is sufficiently S-nitrosylated as to lose activity. Given these inconsistencies, we sought evidence of alternate mechanisms.
Recently, Chakrapani et al presented experimental evidence of NO donation from C-nitroso compounds through a first-order homolytic C-N bond scission event.52
Interestingly, this mechanism required a geminal nitrile moiety to allow homolysis of the C-N bond. Moreover, the reactivity between nitrile groups and free thiols is well established and explored.54
To further explore the mechanistic release of NO, we examined selected analogues ability to donate NO as judged by the aforementioned ABTS method (see the supporting information section for these data). The results confirmed that analogues must be capable of generating NO in the presence of TGR and NADPH to be effective worm killing agents. The ABTS method provides a measure of the oxidation products of NO rather than NO itself. To more exactly quantify the release of NO, we examined the release of NO quantitatively through a well known chemiluminescent assay that relies upon the reaction between NO and ozone to produce an activated NO2
species that produces an intense chemiluminescence emission.54
Additionally, selected analogues were characterized for their ability to react with free cysteine (as judged by consumption of starting material detected across an HPLC gradient). For these experiments, we selected a number of compounds with specific SAR in terms of TGR inhibition and worm killing ability (). Furoxan (2
) was noted to react fully with cysteine and produce approximately 0.6 equivalents of NO per equivalent of 2
. Gratifyingly, bis-oxadiazole-2-oxide analogues 33
were both cysteine reactive and produced more NO equivalents relative to furoxan (2
) (1.55 molar equivalents for 33
and 1.19 molar equivalents for 34
). Analogue 6
, with the replacement of the nitrile with a methyl group, was not capable of reacting with cysteine nor was 6
capable of NO donation. Most interestingly, derivatives 3
(with the bis-thiophene ketone moiety) and 5
(with a transposed N-O moiety, but retaining the nitrile) did react with cysteine, but did not generate NO. An additionally interesting result was that compound 30
(with both thiophene ketone and nitrile moieties) reacted with cysteine and was capable of releasing NO. From this result, the failure of 30
to produce worm death is unexplained.
Evaluation of furoxan (2) and selected analogues for reactivity with free cysteine as judged by consumption of reagent and the molar ratio of NO release per stoichiometric equivalent of reagent.
Mechanistically, it is not possible to identify the location of the initial nucleophilic attack, but the ultimate result is addition into the oxadiazole ring system alpha to the nitrile (). The ring system would likely open resulting in a quaternary alkyl-thio-nitroso-nitrile species that is highly similar to the C-nitroso compounds shown by Chakrapani et al to be capable of homolytic C-N bond cleavage and release of NO (path 1). Alternatively, this intermediate could undergo transnitrosylation and release of R-SNO and a byproduct of the original oxadiazole-2-oxide heterocycle (path 2). It is further possible that this intermediate could rearrange to form an isoxazol-5(4H)-imine type intermediate prior to homolytic C-N bond cleavage and release of NO (path 3). A second mechanistic issue is the ultimate release of the attacking nucleophile from the core oxadiazole scaffold. While plausible mechanisms can be envisioned for this de-ligation event, there is currently no experimental evidence for how it would proceed. Interestingly, we tested this point through the synthesis of a biotinylated version of furoxan (2) (conjugated through an ether linkage at the 4-position of the phenyl ring) and subsequent treatment of this analogue with recombinant TGR and NADPH. Gel analysis of the resulting mixture provided evidence that TGR was not cross-linked to biotin, further suggesting that TGR reacts with oxadiazole-2-oxides in a catalytic rather than terminally reactive mechanism (data not shown).
Schematic description of potential mechanisms for NO release by oxadiazole-2-oxides via reaction with reactive thiols (or with selenol).
These insights into the interaction of oxadiazole-2-oxides with TGR suggest that the preeminent mechanistic event is the nucleophilic attack by either cysteine or selenocysteine residues on TGR. Selenium has a unique role in biology and the biosynthesis of selenocysteine residues and, ultimately, selenoproteins has been the subject of numerous reviews.56
To date, 25 human proteins have been characterized that contain selenocysteine residues and each retain a specified cellular function (over 300 selenoprotein families have been found in nature).57
A minimalist view of atomic sulfur and selenium highlight several divergent properties that translate into important biochemical differences for selenocysteine and cysteine residues. Primary among these variations are the relative pK
a values for both amino acids (8.5 for cysteine versus 5.2 for selenocysteine) and the stronger nucleophilicity of selenocysteine.58–60
Evolutionary analyses clearly suggest that selenocysteine supports unique roles in selenoproteins that can not easily be maintained by cysteine.61
The strikingly low pK
a of selenocysteine results in full anionic character for this residue at physiological pH. In several instances, the divergent activities of selenocysteine and cysteine residues have been exploited. For instance, Filipovska and coworkers recently reported the rational design of gold (I) containing N
-heterocyclic carbene complexes that target the selenocysteine-containing TrxR and are selectively toxic to cancer cells.62
Arnér and coworkers have found that forms of selenium compromised thioredoxin reductase-derived apoptotic proteins (SecTRAPs) may have prooxidant toxic properties in addition to a loss of thioredoxin reducing capacity.63
While there is not yet any direct evidence, it seems reasonable to assume that the enhanced nucleophilicity associated with the selenocysteine residues (including the one found in TGR) makes this moiety a likely candidate for the initial attack on furoxan (2
) and other oxadiazole-2-oxides. As such, it was important to explore the selectivity of this chemotype with regards to its ability to act as a general electrophile for other selenocysteine containing enzymes. Additionally, it was essential to consider if these compounds inhibit related host enzymes, most significantly GR and TrxR. To resolve both questions we assayed several analogues versus human GR and rat TrxR (rat TrxR is a selenoprotein64
while human GR is not). We found that furoxan (2
) does not inhibit recombinant human GR (IC50
> 50 μM) or recombinant rat TrxR (IC50
> 400 μM) (). Of interest was the discovery that three of the agents tested (the aldehyde 8
and phenyl ring analogues 15
) were found to inhibit both hGR and rat TrxR, albeit with reduced affinity. The aldehyde derivative 8
may yield insight into the mechanism of these agents and the discovery of inhibitors of both hGR and rat TrxR is an important finding on its own accord. More important for this study, however, is the fact that several of the anti-schisotosomal analogues were found not to be inhibitors of the TGR-related mammalian host enzymes. Taken as a whole, the data presented here provides strong evidence supporting the selective action by oxadiazole-2-oxides against schistosomes and suggests that the interaction of this chemotype with TGR is dependent upon both binding affinity and appropriately aligned reactivities.
Comparative inhibition of selected analogues versus TGR, human GR and rat TrxR.
We were further interested in determining if the oxadiazole-2-oxide scaffold has suitable properties for oral bioavailability. Since therapies for schistosomiasis are intended to be single dose agents (as opposed to chronic treatments), we were initially and primarily interested in determining selected analogues’ potential for passive membrane permeation and stability in liver microsomes. Profiles of these agents binding to the human Ether-à-go-go Related Gene (hERG) potassium channel and inhibition of selected members of the cytochrome P450 class of enzymes were also of interest. Our own investigations of these agents suggested highly variable solubilities in water. We were interested to go beyond conjectural evaluations of this property (i.e. HPLC retention times) and establish more reliable data for selected analogues. Consequently, solubility measurements (in both μM and μg/mL) and LogD determinations were ascertained. The results of these profiles are shown in . Several compounds had only modest water solubility including furoxan (2), 16, 31, 32, and 34. Analogues containing the thiophene ring were noted as having better solubility profiles, including 29 and 35. Interestingly, the addition of a single fluorine moiety to the bis-oxadiazole-2-oxide analogue 31 resulted in an impressive gain in solubility (see derivative 33).
Solubility and Selected ADME Properties for Chosen Analogues
Beyond solubility profiling, several analogues were found to have modest ability to permeate artificial Caco-2 cell monolayers [apical to basolateral (A→B) permeability was typically greater than basolateral to apical (B→A)]. Furoxan (2) was not found to pass through Caco-2 cells, however, several analogues (including 16, 29, 33 and 35) were noted to possess adequate Caco-2 profiles. While these data does not guarantee successful realization of these analogues as orally available agents, it does dispel concerns that the oxadiazole-2-oxide scaffolds are incapable of passive membrane permeation. Analysis of these analogues in liver microsomes (over 1 hour in the presence/absence of NADPH) suggests that these compounds are only moderately stable and may possess short half-lives in intact animal models. However, the variable data (for instance compound 29 versus 33) suggests that more robust analogues may be found through additional modification. Further, the hERG and Cyp profiles for these agents were generally reasonable. Taken as a whole, these data suggests that properly substituted oxadiazole-2-oxides may possess efficacy and ADMET properties adequate for clinical application.