Reaction optimizations
The SN2 reaction is known to be very sensitive to the substrate structures, as well as reaction conditions.11-13 While initial attempts to run the new reaction were promising by masspectrometry analysis, the yields were far from being synthetically useful. Taking into account the established condition knowledge of the SN2 reaction we first interrogated stoichiometry and ratio of the reactants, temperature, temperature source, and solvents employing high throughput experimentation (HTE).14-15 HTE methods used were parallel reactions in 96, 48, and 24-well format, parallel heating in a metal block, parallel TLC analytics, and stacked injection into SFC. The methods are described in more detail in the SI. We chose the model reaction of p-chloro benzyl isocyanide with benzyl bromide, a good electrophile in SN2 reactions and good visibility of educts and product in TLC (Fig.2). Next, we investigated the result of additives in the SN2 reactions. Biphasic phase transfer catalyst (PTC) were often used in SN2 reactions to increase yields and conversion.16 We screened 16 different common PTCs (SI). The addition of iodine salts is often described as advantageous in the SN2 reactions as it converts the less reactive chloride leaving groups into the more reactive iodo leaving group. After thorough optimization of all parameters the optimized conditions involved the microwave heating at 105oC for 3 h of 1:2 ratio of isocyanide, alkyl halide, 20 mol% KI catalyst, and 1 equivalent of water in acetonitrile in the presence of 2 equivalent of the inorganic base K2CO3 (Fig.2).
Scope and limitations
The substrate scope for this reaction is very broad (Fig. 3-5). With the optimized conditions in hand, we interrogated the scope of the halide with respect to the leaving group, sterical bulkiness, electronic nature and diversity. Amongst the halide leaving group, chloride, bromide and iodide reacted well according to the well-established leaving group trend I>Br>Cl. To test the functional group tolerance, we successfully reacted 21 different alkyl halides with adamantyl isocyanide on a mmol scale (Fig.3). Adamantyl isocyanide is a solid, non-smelling, bench stable powder which has been synthesized recently on a mol scale.17 A variety of alkyl halides with different functionalities were well tolerated. The small methyl group can be easily introduced (21a), whereas bulky alkyl groups or alkyl groups with b-branching do not react. Long chain alky groups can be introduced (17a), also with a terminal phthalic amide amine protecting group (12a), whereas Boc-protecting groups were found to be not stable under the microwave conditions (SI). For several alkylation products, single crystals revealed X-ray structures which support the structural identity (4a, 8a, 12a, 16a). Allyl (5a), and benzyl (1a, 3a, 4a, 6a, 13a) groups react well due to the conjugated nature of the pentagonal bipyramidal transition state as suggested by the classical SN2 literature. Specifically, to mention is bis benzylchloride derived 13a which can be mono alkylated in 32% yield, and can be potentially further reacted through the unreacted benzylchloride. Also, the nature of the heterocyclic structures which could be reacted is quite diverse, including benzimidazole (7a), pyrazole (8a), triazole (9a), phthalimide (12a), coumarin (11a), thiophene (15a), and quinoline (19a). Especially to mention are 15a and 18a, which are formed from bifunctional (hetero)aromatic benzylchloride benzaldehydes. The aldehyde functionality can be further derivatized as will be shown below. We also found substrates which did not give the expected products or gave very low yields (<30%, SI).
The evaluation of the isocyanides also revealed a broad scope (Fig.4). We reacted 20 different isocyanides with methyl iodide in satisfactory to good yields. Benzylic (23a, 24a, 25a, 29a, 30a), aromatic (31a, 33a, 34a, 35a, 36a, 37a), aliphatic (27a, 42a) and heteroaromatic (26a, 28a, 32a) isocyanides all worked well. When isocyanides with a basic side chain were reacted, we observed the double alkylation and a quaternary amine salt formation (38a, 39a). Noteworthy, also a-amino acid isocyanides (40a, 41a) worked well.
We performed a number of mixed examples to further elaborate the scope and usefulness of the reaction (Fig.5). Highly substituted 47a is especially noteworthy, as it comprises a combination of a sterically hindered a,a-disubstituted cyclopropyl benzyl isocyanide with a bifunctional 4-formylbenzyl chloride. The new method is also applicable to the facile synthesis of diverse lipid derivatives (56a, 60a, 61a) which could be of interest in lipidomics applications. Bulky isocyanides (47a, 50a, 54a) and phenyl isocyanides with bulky o-substitutents (43a, 51a, 52a, 53a, 55a) reacted nicely. Amide 55a is accessible with a free compatible benzylic hydroxyl group. 4-Methylpentenoic acid (pyroterebic acid) ester or amides are common in biologically active isoprenoid compounds from plants. Compound 54a is a pyroterebic acid amide and it comprises an unprecedented synthesis. Another example of incorporation of an isoprenoid side chain (homo geranyl acid) is exemplified in 60a. It is conceivable that this methodology can be used to incorporate isotope labeled carboxy-C via the isocyanide. In summary, complex structures can be accessed from simple available building blocks in one step.
Scaling and late-stage functionalization
To further stress the reaction performance, we evaluated the robustness of this reaction towards pharmaceutical late-stage diversification on an actual drug.18 Late-stage-functionalization is a drug discovery technique to selectively derivatize already complex ‘drug-like’ molecules and is used to further improve their properties.18 Phenoxybenzamine (dibenzyline) is an alpha blocker used for the treatment of hypertension. To establish the usefulness of our novel SN2 reaction we reacted dibenzyline with adamantyl isocyanide and were able to isolate the expected amide product in 40% yield (Fig.6).
Having demonstrated a robust substrate scope for this novel isocyanide to amide transformation, we considered a variety of applications. First, we performed the thiophene carbaldehyde on a gram scale in fair yields. For this we reacted 1.59 gram of the bifunctional 5-(chloromethyl)thiophene-2-carbaldehyde with 1.61 gram adamantyl isocyanide on a 10 mmol scale (Fig. 6). The product 15a could be isolated in 51% yield (1.57 gram). We envisioned that the aldehyde group can be further functionalized to create molecule of high complexity in just a few steps. To increase the complexity of the products we used multicomponent reactions (MCR) for further derivatization of the thiophene carbaldehyde.7, 10, 19 The carbaldehyde 15a is of interest to test further reactivity due to its unprotected aldehyde group based on the functional group compatibility of the reaction. Thus, we used 15a, each in a Ugi-4CR, a Groebke Blackburn Bienaymé (GBB‐3CR) reaction, and a Ugi tetrazole reaction to exemplify rapid increase of molecular complexity (Fig. 6). The Ugi-4CR product 1b was obtained in 72% yield in one step from easily available building blocks. Noteworthy, an alkynyl amide is introduced in a straight forward mild manner. Electrophilic alkynyl amides are often used in covalent drug discovery targeting cysteines and an alkynylamide substructure can be found in the FDA approved Acalabrutininb Bruton's tyrosine kinase targeting drug.20 Next, we investigated aldehyde 15a as a substrate in the GBB‐3CR reaction. The GBB-3CR is a popular method to synthesize highly substituted bicyclic imidazo heterocycles which already have proven their value as drugs and candidates.21 Thus, we reacted 2-aminopyridine with aldehyde 15a and cyclohexyl isocyanide in a GBB-3CR, under microwave conditions in methanol to obtain complex heterocycle 1c in 36% yield. Lastly, we performed a Ugi tetrazol reaction employing aldehyde 15a. Tetrazoles are often used as advantageous carboxylic acid bioisosteres, and can be broadly obtained by multicomponent reaction chemistry.19
In summary, the new SN2 reaction turned out to be scalable, useful in late-stage-functionalization, and can yield highly interesting intermediates for allowing further chemistries to increase structural diversity in a quasi-exponential complexity increase, in just three steps: isocyanide synthesis, SN2 reaction, further aldehyde reaction.
Mechanism and chemical space
Preliminary observations support a SN2-type mechanism (Fig.7A). Accordingly, the nucleophile isocyanide attacks from the backside to form a trigonal bipyramidal transitions state I and kicks out the leaving halogen anion. The intermediately formed nitrilium ion II undergoes water attack on the isocyanide-C III, and through tautomerization reveals the final amide IV upon hydrolysis.
Several lines of evidence support a SN2 mechanism: sterically hindered substrates such as neopentyl iodide or isobutylbromide do not give any reaction product; the reaction is strongly solvent dependent and runs well in the polar solvent DMF which are believed to stabilize the transitions state, but not in apolar toluene or protic methanol; the reaction rate depends on the nature of the nucleofuge as reported in the SN2 literature I>Br>Cl (SI). To exclude a possible radical mechanism, we performed the reaction in the presence of 2x stoichiometric amounts of the radical quencher TEMPO, and did not find any difference in the reactivity (SI).22-23 While running the reaction in the absence of water and direct injection in the mass spectrometer we could observe a strong peak corresponding to the bromo nitrilium ion (Fig.7B). In conclusion, there is strong evidence that the reactions run according to a SN2 mechanism. In the classical amide coupling approach the carbonyl is a carbocation synthon, while in the SN2 approach the rare amide carbanion synthon is the result of an Umpolung (Fig.7D). The isocyanide is commonly synthesized from its primary amine precursor (Ugi method: formylation > dehydration or Hoffman reaction).10, 24 Alternatively, the isocyanide can be produced from an aldehyde or ketone precursor through reductive amidation with formamide (Leukart Wallach) and dehydration (Fig.7C).25-26 Phenomenologically, the overall transformation of this SN2 reaction corresponds to a coupling of a primary amine with a C1 synthon derived from chloroform (Hoffmann) or formic acid (Ugi) with an alkyl halide or coupling of an aldehyde/ketone through a NC synthon derived from formamide (Leukart-Wallach) with an alkyl halide (Fig 7.C). Due to the large number of commercially available primary amines and aldehydes and ketones as isocyanide precursors, and alkyl halides, the reaction can be of considerable synthetic utility. Noteworthy, in classical SN2 reactions mostly very simple nucleophiles are used (such as halides, CN-, thio- or alcoholates), whereas the herein described SN2 reaction can make use of the great structural diversity of isocyanides (Fig.4). This is leading to a strong increase in structural complexity upon coupling with alkyl halides (e.g. 60a). Next, we asked the question whether the new reaction can access a chemical space different from the classical amide coupling. For this we investigated the commercial availability of the corresponding carboxylic acid needed to form the target amides and compared them with the corresponding halide (SI). Surprisingly, in 52% cases the corresponding carboxylic acids were not commercially available at all. Noteworthy, in the remaining 48% the carboxylic acid was on average 2.3 times more expensive than the corresponding chloride. It turned out that the chemical space accessible by the two orthogonal amide syntheses is very different and only 12% are overlapping (i.e. can be synthesized by both methods). In conclusion, our herein reported novel SN2 reaction is of high synthetic value as it allows to access a chemical space which otherwise can only accessed through time-consuming and lengthy multistep syntheses and leads to a strong increase in molecular complexity, otherwise uncommon in SN2 reactions.