Tele -substitution Reactions in the Synthesis of a Promising Class of 1,2,4-Triazolo[4,3-a]pyrazine-Based Antimalarials

We have discovered and studied a tele substitution reaction in a biologically important heterocyclic ring system. Conditions that favour the tele -substitution pathway were identiﬁed: the use of increased equivalents of the nucleophile or decreased equivalents of base, or the use of softer nucleophiles, less polar solvents and larger halogens on the electrophile. Using results from X-ray crystallographic and isotope labelling experiments, a mechanism for this unusual transformation is proposed. We focused on this triazolopyrazine as it is the core structure of the in vivo active anti-plasmodium compounds of Series 4 of the Open Source Malaria consortium.


Introduction
Nucleophilic substitution is a widely employed method for functionalising electron-deficient aromatic systems. Most commonly, a halide or other leaving group is simply displaced by an incoming nucleophile, known as direct or ipso-substitution. 1 Under some circumstances however, a leaving group may be displaced from an aromatic system by a nucleophile entering at a different position on the ring, for example at the carbon adjacent to the leaving group (cine-substitution 2 ) or even further away (telesubstitution, 3 Figure 1A). We report here our discovery, and mechanistic studies, of a telesubstitution reaction in a [1,2,4]triazolo [4,3a]pyrazine system, 4 which is at the core of a series of molecules with significant potential for the future treatment of malaria. 5 The first example of a tele-substitution reaction was reported in 1930 ( Figure 1B). 6 In this case, the reaction of 2-(chloromethyl)furan (1) with NaCN resulted in the attachment of the nitrile group not in place of the chlorine atom but, instead, distant from the expected electrophilic site on the opposite side of the furan ring (2). Other examples of tele-substitution reactions have since been reported for a variety of aromatic systems ranging from simple pyrazine rings 7 ( Figure 1C) to more complex triazolopyrazine ring systems 8,9 (Figures 1D and 1E), the latter being of particular relevance  [1,2,4]triazolo [1,5-a]pyrazine and E) [1,2,4]triazolo [4,3-a]pyrazine ring systems.
to the present work. Despite these and other reports, [10][11][12][13] tele-substitution reactions are not well understood; they remain hard to predict and appear to be strongly substrate dependent. Interestingly, many of the known examples of tele-substitution involve aza-aromatic ring systems which are common in medicinal chemistry and drug discovery campaigns. Given the isomeric nature of the ipso-and tele-substituted products, and the sometimes cursory level of characterisation in medicinal chemistry articles (where compound identity may be demonstrated using only a 1 H NMR spectrum and an LCMS trace) it is important, as we have discovered, to be aware of the possibility of this underappreciated reaction in order to avoid drawing conclusions from erroneous SAR data.
Here, we illustrate this with our studies on the tele-substitution reactions of the [1,2,4]triazolo [4,3-a]pyrazine (hereafter referred to as 'triazolopyrazine') heterocyclic system. These nitrogen-rich, electron-deficient heterocycles are important building blocks for the development of new medicines and have shown a wide variety of biological activities ( Figure 2). We have an interest in this motif because it forms the core of Series 4 of the Open Source Malaria (OSM) consortium, 14 represented here by compound 10 which possesses in vitro 15 (IC 50 = 38 nM) and in vivo 16 antimalarial activity. Compound 11 has been reported to have nanomolar potency as an inhibitor of the kidney urea transporter UT-A1. 17 Compound 12 was recently patented in 2016 as a renal outer medullary potassium channel (ROMK) inhibitor. 18 Sitagliptin (13) was approved by the FDA in 2006 as an antidiabetic drug (dipeptidyl peptidase (DPP)-IV inhibitor). 19 Compound 14 is a lead molecule (IC 50 <100 nM), that acts as an inhibitor of bromodomain and extra-terminal motif (BET) proteins for cancer treatment. 20 Compound 15 is patented as an N -methyl-d-aspartate subtype 2B (NM-DAR2B) receptor antagonist. 21   The synthesis of members of OSM Series 4 relies on a routine S N Ar reaction involving the nucleophilic displacement of a chlorine atom from a triazolopyrazine core (e.g. 16). When the synthesis of thioether analogue 17 was attempted using the standard conditions for this reaction ( Figure 3A), in addition to this expected product, a compound with a significantly lower TLC retention factor was observed and isolated. This was later identified as the tele-substituted isomer 18. Since the 8-isomer 18 is a main product that was formed in 83% yield and due to the similarity of the 1 H NMR spectra of these two isomers (Figure 4), the tele-substituted isomer 18 was initially misassigned as the desired product 17. After the reaction had been repeated and examined more thoroughly compound 17 was successfully isolated as a minor product with 8% yield. The diagnostic spectroscopic difference between these isomers lies in the peaks arising from the hy- drogen atoms at positions 5 and 8 on the triazolopyrazine ring; the correspondence between the NMR spectra and the structures was confirmed using X-ray crystallographic (vide infra) and deuteration experiments. In a medicinal chemistry context, this spectroscopic similarity is a hazard for the understanding of structure activity relationships: the original evaluation of this synthetic product had concluded that 17 was inactive (IC 50 > 10 µM) in a malaria parasite killing assay (in vitro against P. falciparum 3D7 strain), when in fact it was 18 that had been evaluated in its place. Compound 17 was later tested and found to have reasonable potency (IC 50 = 1.04 µM).
According to the generally accepted ipsosubstitution reaction mechanism, the first step is nucleophile attack on the carbon atom to which halogen is attached (19, Figure 3B). The resulting intermediate (20) expels chloride, leading to the ipso-substituted product (21). On the other hand, a plausible mechanism for the tele-substitution reaction could involve the initial attack of the nucleophile at the 8-position (22, Figure 3B), followed by loss of the 8-position proton as part of the elimination of the chloride (23). Since mechanistic studies on tele-substitution reactions are scarce, we sought better understanding of the process operating in this case.
To better define the scope of tele-substitution in this triazolopyrazine system, 8-and 6halogenated variants of the triazolopyrazine core were synthesised from the corresponding dihalopyrazines following literature procedures 23 and subjected to the same reaction conditions as the original 5-chloro triazolopyrazine. The 8-halogenated cores (25-27, Figure 5A) reacted to give the expected ipso-substituted products only (28)(29)(30)(31)(32)(33)(34)(35)(36), while the 6-halogenated analogues (37 and 38, Figure 5B) resulted only in degradation of starting material without formation of any substituted product. While there is limited literature precedence, dihalopyrazines (e.g. 39-41, Figure 5C) have been shown to give exclusively ipso-substituted products (42-44 respectively). With these experiments showing that the tele-substitution reaction is observed only with the 5-halogenated cores (Fig-ure 3A), the following mechanistic discussion will focus on that system.    Figure 5: Reactions of halogenated triazolopyrazine isomers and pyrazines. A) 8-Isomer; B) 6-Isomer; C) Pyrazine; Conditions: a KOH, 18-crown-6, toluene, room temperature (reactions involve measuring small amounts of hygroscopic KOH, which can contribute to reproducibility challenges, thus experiments were performed in duplicate and are reported as average values); b silica, toluene, reflux (more details in Table 1).
A) Influence of triazolopyrazine structure and nucleophile. a The nature of the nucleophile plays a crucial role in the outcome of the reaction (Table 1). When compared to reactions with alcohols, the use of more nucleophilic amines and thiols led to significantly more tele-substituted products (Entries 1-6, 12-17 and 21-26). This trend may explain why tele-substituted isomers were apparently not Table 1: Influence of triazolopyrazine structure, leaving halogen X and nucleophile on the reaction outcome.   Figure 6 for details). R 1 = CH 2 CH 2 Ph. ND: not determined.
seen in the literature synthesis of related structures 24 in which the incoming nucleophile was restricted to alcohols.
a When the conditions employed with alcohols and thiols (KOH, 18-crown-6) were used with amine nucleophiles, the reaction progress was comparatively slow so the base was replaced with silica, which gave better The nature of the leaving halogen also influences the outcome, with tele-substitution favoured in the order I > Br > Cl (compare conversion; for convenience the rate was made comparable to those seen with the other nucleophiles by raising the reaction temperature, as the reaction at room temperature was not complete after 2 weeks. : Unexpected product 50 via ring opening and rearrangement from the reaction of iodotriazolopyrazine and an amine nucleophile, with proposed mechanism for this product (see Figure S2 in SI for more details).

ratio in Entries 4, 15 and 24).
In cases where a larger substituent is in position 3 of the triazolopyrazine core (e.g. a (4-OMe)Ph group compared to a hydrogen atom), and the leaving halogen is either a Br or I atom, the distribution of ipso-to tele-substituted products is favoured towards the latter (compare Entries 12 and 15 or 21 and 24). Similar experiments in which the leaving halogen is a Cl atom show little to no change in distribution of products (compare Entries 1 and 4). Further investigation of the substituent at the 3-position led to the conclusion that bulkiness does not affect the reaction (i.e. substitution with (4-OMe)Ph is comparable to that of the larger (3,5-tBu)Ph or 9-anthracene; Entries 4, 10 and 11 respectively).
Substrates with electron donating (EDG) and electron withdrawing (EWG) groups on the phenyl ring at the 3-position of the core were studied in order to evaluate the influence of electronic effects on the distribution of products. Experiments on bromotriazolopyrazines showed that EDGs tend to promote the tele-substitution pathway of the reaction, while EWGs lead to ipso-products only (Entries 15 and 18-20). Interestingly, chloro-triazolopyrazines do not follow this pat-tern and show no dependence on the electronic effects from the substituent in the 3-position (Entries 4, 7, 8 and 9).
From the experiments summarised in Table  1, two gave surprising results. The reaction between the iodo-triazolopyrazine core 45n and the thiol nucleophile (Entry 25) in addition to the 8-substituted compound 47b, isolated in 13% yield, gave dehalogenated product 49 in 74% yield. This product was not observed for any other reaction substrates bearing a chlorine or bromine atom. This type of the dehalogenation reaction has not previously been reported in the literature. The other unexpected product was isolated from the reaction between the iodo-triazolopyrazine core 45n and the amine nucleophile (Entry 26). In addition to the isolation of the major tele-substituted isomer 47d and dehalogenation product 49, a minor byproduct was obtained in 17% yield, the structure of which was determined by single crystal X-ray diffraction (see SI) to be based on a 5-(1H-imidazol-2-yl)-1H-1,2,4-triazole core instead of the expected triazolopyrazine structure (50, Figure 6). It is possible that compound 50 could be formed via initial nucleophile attack at the 8-position of the pyrazine ring (51), followed by the pyrazine ring opening (52) and rearrangement (53) leading to 50. While the analogous reaction utilising the chlorinesubstituted triazolopyrazine (Entry 6) did not lead to this rearranged product, it was formed in trace amounts when the bromo-substituted triazolopyrazine was employed (Entry 17). This trend may either be due to a sub-optimal bond geometry (i.e. pseudo-equatorial I atom) arising from the larger halogen atom or from a better match of orbital energies for elimination (in the case of the chlorine leaving group).
B) Influence of solvent. With the reaction between 45i and the alcohol nucleophile (Table 1, Entry 1) giving significant quantities of both isomers, this was used as the model reaction to investigate further the influence of solvent on the reaction outcome ( Table 2). A screen of aprotic solvents clearly showed that solvents with higher dielectric constants lead to less telesubstitution and also lower the overall yield of the reaction. Protic solvents are inherently unsuitable for this reaction as they can easily themselves react with the halogenated triazolopyrazine. This was demonstrated when water was used as the solvent, giving the product 48a in 94% yield, by result of tele-substitution with H 2 O.
C) Influence of excess alcohol and base. By using the same model reaction above, the effect of alcohol and base equivalents was investigated. It was found that the use of an excess of nucleophile resulted in a shift of the reaction outcome drastically towards the formation of the 8-isomer (47a, Figure 7A). These observations suggest that the use of a softer nucleophile (here one in which the anion is surrounded by a "solvent shell" of OH bonds arising from excess nucleophile) leads to greater formation of the 8-isomer. Similarly, when fewer equivalents of base were used, a higher proportion of telesubstitution was again observed ( Figure 7B). D) Influence of water and temperature. In order to evaluate the impact of the level of water present on tele-substitution, the reaction between 45a (unsubstituted on the triazole ring) and piperidine was conducted in toluene with various levels of water, as well as in water itself (H 2 O and D 2 O). The isolated yields of the 5-(55) and 8-isomer (56) were identical for ex- periments in both wet and dry toluene ( Table 3, Entries 1 and 3, for X-ray single crystal structure of 45a and 56 see the SI). At room temperature the reaction took 14 days to complete (Entry 2), but the outcome was comparable to that when heating under reflux conditions. When molecular sieves were included in the reaction mixture (using dried toluene) the ratio of products changed, though it is possible that this could arise from catalytic activity at the zeolite surface itself (Entry 4). 25,26 Performing the reaction in H 2 O (Entry 5) gave a comparable result to that in wet toluene. This is counter to the example where the alcohol nucelophile was out-competed by the solvent water to give the tele-substitution product (vide supra). It could be concluded that the presence of water in the solvent and the reaction temperature do not alter the distribution of products in the studied reaction.
E) Isotope labeling experiments. Following the observation that no hydroxy-substituted product was identified in the reaction between the halogenated triazolopyrazine core 45a and an amine nucleophile in the presence of water, deuteration experiments were peformed to gain insight into the reaction mechanism. This reaction was carried out in D 2 O giving two compounds, 57 and 58 ( Figure 8A). The examination of products with 1 H NMR and 2 H NMR spectroscopy showed incorporation of one D Table 2: A) Reaction used to study the influence of solvent; B) Product isolated when H 2 O was employed as a solvent. Results of the reaction in different solvents (reactions performed in duplicate). All solvents were dried over molecular sieves (3 Å) for 48 h before application. All reactions proceed to complete consumption of bromo-triazolopyrazine as indicated by TLC. Total yield reported is the sum of both isomers. Product 48a typically observed to form in ∼15% yield but was not isolated in these reactions. R = CH 2 CH 2 Ph.  but not for the triazolopyrazine system investigated here. In order to prove that deuteration occurs at the 3-position as a parallel reaction to the main substitution, compounds 45a, 55 and 56 were heated under reflux in D 2 O without piperidine to give corresponding monodeuterated products 59, 57 and 60 respectively ( Figure 8B). The deuterium exchange at the 3-position could be explained by the relatively high acidity of the hydrogen in C-H bond on the triazole, though pKa values have not been reported, a prediction model estimates pKa of similar structures to be around 29, compared to > 35 for the C-H bond of pyrazine. 29 The second D atom in 58 was at the 5-position, thus confirming that the proton which takes the place of the leaving group in the tele-substitution reaction comes from the solvent and not from the substrate (see the proposed mechanism for 19 in Figure 3B). Deuteration position assignment was based on 1 H NMR spectra comparison of non-deuterated compounds 55 and 56 with deuterated 57 and 60, as well as 2D NMR data for 55 and 56. Importantly, the amine products 55 and 56 were found to be not interconvertible when each product separately was subjected to the reaction conditions for 3 days, as no conversion of one isomer into another could be detected by TLC. Thus the ratios of products observed in these telesubstitution reactions arise from a kinetic difference rather than one that has a thermodynamic origin.

Biological activity
As mentioned above, 5-substituted triazolopyrazines (e.g. 17) showed antiplasmodium activity, while an 8-substituted isomer (18) proved to be inactive. Based on the structural similarity of these triazolopyrazines to kinase inhibitors, 30 we evaluated several compounds in the preliminary KINOMEscan ® assay (at 1 µM concentration). The results revealed complementary activity of ipso-and tele-isomers, for example 47b has higher potency against Table 3: Results of the reaction with wet and dry solvent. 3Å molecular sieves were used to dry the toluene. Water levels were measured with a Karl-Fischer titration apparatus immediately before the experiment. a Reaction time 14 days. b Products were partially deuterated (Figure 8).   serine/threonine-protein kinase 3 (STK3) compared to 46d (Figure 9, see SI for full screening results). Thus the occurrence of this telesubstitution reaction allows the generation of two biologically active compounds with complentary activities from a single reaction.

Conclusion
Tele-substitution reactions are simple to achieve in the triazolopyrazine ring system, and it is important to be aware of the possibility of such isomers forming, given the wide biological relevance of many of these structures. The tele-substitution reaction occurs only in 5-halogenated triazolopyrazine cores, while 8-or 6-halogenated cores tend to give ipso-substitution or degradation respectively. The tele-substitution pathway of the reaction is also made more likely by the use of stronger nucleophiles, triazolopyrazines with bulkier halogens and the use of less polar solvents. As concluded from the isotope labeling experiments, the hydrogen atom that takes the place of the halogen derives from solvent and not from substrate. The product ratios arise from a kinetic difference in the reactions rather than a thermodynamic difference in product energies, where, broadly, a combination of hard nucleophile and hard electrophile promotes ipso-substitution while a softer combination promotes tele-substitution (for a graphical summary see Figure 10). Computational studies to rationalise and predict substitutions of these kinds are non-trivial (in part because of the possibility of direct 31 vs stepwise 32-34 substitution) but are ongoing and will be reported in due course.

General Procedure A. Preparation of halogenhydrazinylpyrazines
Mono or dihalogenopyrazine (70 mmol, 1 equiv.) was dissolved in ethanol (100 mL), then hydrazine monohydrate was added (140 mmol, 2 equiv.) and the mixture was heated at reflux overnight. The solvent was removed under reduced pressure. Equal amounts of EtOAc (100 mL) and H 2 O (100 mL) were added, the EtOAc layer was separated and the aqueous layer was washed with EtOAc (30 mL × 3). The combined organic phases were washed with brine (30 mL), dried over Na 2 SO 4 and evaporated under reduced pressure to give the desired com-pound, which was used in the subsequent reaction without further purification (for reaction schemes of general procedures see SI, Figure  S1).

General Procedure C. Preparation of halogeno-3-aryl-[1,2,4]triazolo[4,3-a]pyrazine
Adopted from the literature procedures. 23 To a stirred suspension of halogeno-hydrazinylpyrazine (7.0 mmol, 1.0 equiv.) in ethanol (100 mL) was added aldehyde (7.7 mmol, 1.1 equiv.) and the mixture heated at reflux overnight. After the full consumption of starting material as indicated by TLC, the reaction was cooled in an ice bath and chloramine T trihydrate (9.1 mmol, 1.3 equiv.) was added portionwise while stirring over 1 h. After consumption of the intermediate was confirmed by TLC, cold H 2 O (100 mL) was added to the reaction mixture. The solution was stirred for 10 min, then filtered through a sintered glass filter (P3 porosity) and washed with H 2 O (30 mL × 3) followed by Et 2 O (30 mL). The solid was dried in vacuo to give desired product that was used without further purification.
General Procedure D. Coupling of alcohol or thiol with halogen-heterocycle To a suspension of halogen-heterocycle (0.40 mmol, 1 equiv.) in toluene (10 mL) was added 18-crown-6 (0.032 mmol, 0.08 equiv.) and alcohol or thiol (0.40 mmol, 1 equiv.) followed by KOH (1.20 mmol, 3.0 equiv.). The reaction mixture stirred for 2-24 h at room temperature. Upon completion as indicated by TLC, the reaction mixture was directly subjected to the purification by FCC on silica and flushed at the beginning with hexanes (in order to wash out toluene from the column) followed by a gradient of EtOAc (30% to 100%) in hexanes (unless specified in the compound preparation) to give the desired product.
General Procedure E. Coupling of amine with halogen-heterocycle To a suspension of halogen-heterocycle (0.40 mmol, 1.0 equiv.) in toluene (10 mL) was added amine (1.20 mmol, 3.0 equiv.) followed by silica (0.5 g). The reaction was heated at 80°C overnight. Upon completion of the reaction as indicated by TLC, the solvent was evaporated in vacuo and the mixture purified by FCC on silica using a gradient of EtOAc EtOAc (30% to 100%) in hexanes (unless specified in the compound preparation) to give the desired product.

Supporting Information Available
The Supporting Information is available free of charge on the ACS Publications website at DOI: The following files are available free of charge.
• ORTEP diagrams for the X-ray structures and crystal data; experimental details for biological activity evaluations and copies of 1 H and 13 C{ 1 H} NMR spectra of novel compounds. (PDF) • The archive of laboratory notebook with all experiments described in the article and raw NMR data for all novel compounds. (ZIP) • The KINOMEscan ® assay report on the biological activity of compounds 46d and 47b. (XLS) • X-ray crystal data of 45a, 50, 56. (CIF) • The structural information in strings formant for all compounds described in the article with reference codes to the laboratory notebook. (XLS)