Accessing 1,2-Substituted Cyclobutanes through 1,2-Azaborine Photoisomerization

Abstract

We provide a seminal example of the utility of the 1,2-azaborine motif as a 4C+1N+1B synthon in organic synthesis. Specifically, conditions for practically scalable 1,2-azaborine photoisomerization in a flow reactor are reported to furnish aminoborylated cyclobutane derivatives. C-B bond functionalizations to furnish a diverse set of highly substituted cyclobutanes were also achieved.

Graphical Abstract

A continuous flow approach was applied to 1,2-azaborine photoisomerization. The resulting azaborabicyclohexane product was derivatized into a series of functionalized aminocyclobutanes

1,2-Azaborines are structures resulting from the conceptual replacement of two adjacent carbon atoms in benzene with a boron and a nitrogen atom. The introduction of the boron and nitrogen heteroatoms isosteric to a C=C bond unit (also known as BN/CC isosterism) expands the structural diversity of classic organic compounds, and as a consequence, produces new chemical space enabling new functions and properties.^[1] 1,2-Azaborines possess a combination of structural and electronic properties^[2] that particularly distinguish them among both aromatic and heteroaromatic moieties with utility to biochemistry^[3] and materials science.^[4,5] We have recently initiated a program to further extend the applications of 1,2-azaborines to the field of organic synthesis, specifically as a 4C+1N+1B synthon (Scheme 1). In view of the relatively easy access to the N-TBS-B-Cl-1,2-azaborine reagent,^[6] we envision that structural complexity could be built from this motif (taking advantage of its unique electronic structure) that would have been otherwise more difficult to accomplish. In our initial foray into this research, we demonstrated that 1,2-azaborines can engage in thermal [4+2] cycloaddition^[7] reactions, producing cyclic aminoboranes with 4 contiguous stereocenters (Scheme 1). In this communication, we show that N-TBS-B-Cl-1,2-azaborine can be readily converted into aminoborylated cyclobutanes with regio- and stereocontrol (Scheme 1, equation (1)). Given the scope of methods available to separately manipulate boron and nitrogen functionalities, the transformation illustrated by equation (1) represents a synthetically useful strategy for accessing a diverse array of cyclobutane derivatives.

Scheme 1.

Developing the 1,2-azaborine motif as a 4C+1N+1B synthon in organic synthesis.

We recently reported the photo-induced valence isomerization of 1,2-azaborines to selectively furnish the corresponding BN isosteres of Dewar benzene.^[8] Inspired by certain ring-opening strategies applied to similar heterobicyclic photoproducts,^[9] we envisioned cleavage of the B-N bond in 1 (Scheme 2) could lead to an unfused cyclobutane bearing vicinal boron and nitrogen substituents with pre-determined cis relative stereochemistry.^[10,11] The presence of boron in place of the usual carbonyl group found in the systems alluded to above was subsequently viewed as an opportunity to readily functionalize the corresponding position on the cyclobutane ring by fundamentally different means.

Scheme 2.

Examples of heterobicyclic precursors to isolated all-carbon four-membered rings (top) and proposed general sequence leading from 1,2-azaborine photoisomer 1 to 1,2-disubstituted cyclobutanes (bottom).

To demonstrate this possibility, we undertook to establish a synthetic sequence that would overall transform a 1,2-azaborine photoisomer into a series of 1,2-disubstituted cyclobutanes specifically through manipulations of the boron unit.

The first step in this endeavor comprised retooling the preparative photoisomerization process itself: the scale of the previously reported reactions did not exceed 0.1 mmol, and the reaction time for complete conversion was determined for only one batch reactor configuration (a standard J. Young NMR tube) under irradiation from a high-pressure mercury lamp.^[8a] This particular setup was deemed ill-suited for carrying out synthesis of the photoisomer in the substantially greater quantities needed to facilitate the present work. Simple adjustment by increasing the size of the reaction vessel though was anticipated to result in decreased reaction efficiency as a consequence of diminished relative depth of light penetration into the reaction medium.^[12] For the desired scale-up of the photoisomerization reaction then, we considered it would be more advantageous to optimize conditions for a continuous flow reactor configuration instead.^[12]

An apparatus suitable for this purpose was assembled in accordance with Berry and Booker-Milburn’s general design.^[13] Briefly, a length of UV-transmissive FEP tubing was coiled in a single layer around the outside of a quartz immersion cooling well housing a 450 W, medium-pressure mercury lamp. (A Pyrex filter with a UV cut-off of 280 nm was also placed around the lamp since 1,2-azaborine photoisomers had been previously observed to decompose when subjected to prolonged irradiation by 254 nm light.^[8a]) Substrate solution would be injected into the tubing using a syringe pump, and the reactor outflow would pass directly into a collection vial maintained under a dry nitrogen atmosphere.

Along with the above change in photoreactor type, we also chose to modify the 1,2-azaborine substrate from those studied previously^[8] to better align with our main objective of developing a post-isomerization cyclobutane derivatization sequence. Specifically, tert-amyloxy-substituted 2 was selected to maintain strong steric shielding of the corresponding photoisomer while still allowing subsequent ligand exchange at the boron center that would lead to a boronic ester (vide infra). Solutions of 2 in cyclohexane were thus injected into the flow photoreactor to determine a combination of concentration and reactor residence time (τ) that would afford complete conversion comparable to that of the original NMR experiments. Since the two variables share an inverse relationship in this regard, i.e., higher concentrations require longer residence times and vice versa, we compromised between maximizing concentration and minimizing residence time and settled on τ = 50 min for substrate solutions of 0.25 M. Following this protocol, the 5 mL test reactor we had constructed produced ~1.2 mmol of photoisomer 3 per hour with a typical isolated yield of 83% (Scheme 3).

Scheme 3.

Flow photoisomerization of 1,2-azaborine 2. O^tAm = tert-amyloxy.

With sizeable quantities of 3 in hand, we subjected the photoisomer to catalytic hydrogenation to reduce the all-carbon ring of the bicycle to the cyclobutane. Since this reaction proceeded equally well using the crude cyclohexane solution of 3 obtained directly from the photoreactor, the two processes were generally carried out without isolation of intermediates to prepare azaborabicyclohexane 4 in reasonable overall yield (Scheme 4). Addition of pinacol then led to B-N bond cleavage and displacement of tert-amyl alcohol. Mindful that beyond this point the N-silyl group would likely be incompatible with many reaction conditions, its direct conversion^[14] to a more robust amide functionality was simultaneously achieved by co-addition of benzoyl chloride. Single crystal X-ray diffraction analysis^[15] of the resulting benzamidocyclobutyl boronic ester (5) confirmed the molecule had retained the cis relative stereochemistry of the nitrogen and boron groups originally imparted by the diastereoselective photoisomerization of 2.

Scheme 4.

Sequential isomerization/hydrogenation of 2 to azaborabicyclohexane 4 and subsequent ring opening to form aminoborylated cyclobutane 5.

Late-stage functionalization methods have recently been developed to readily install substituents on some of the carbon positions of the 1,2-azaborine heterocycle.^[16] These synthetic methods when combined with our photoisomerization conditions allow the possibility to 1) generate cyclobutane derivatives bearing quarternary carbon centers and to 2) introduce additional functional groups to the aminoborylated cyclobutane motif. Specifically, we demonstrate here that C3-substituted^[16a] 1,2-azaborines 2’ are also suitable substrates for this transformation, producing highly substituted aminoborylated cyclobutanes 4’ and 5’ in a diasterstereoselective fashion.

As shown in Scheme 5, B-Mes and B-Ph substituents are compatible with our photocyclization reaction, affording isolable azaborabicyclohexanes 4’a-BMes and 4’a-BPh in 73% and 75% yield, respectively. The use of a B-OR substituted 1,2-azaborine enables access to tertiary boronic ester-bearing aminoborylated cyclobutanes (entries 5’a-5’e). Alkyl chloride (entry 5’b), acetal (5’c), carboxylic acid ester (5’d), and benzyl (5’e) groups are all tolerated, with the corresponding products obtained in moderate yields from the 3-step sequence.

Scheme 5.

Diastereoselective synthesis of hindered aminoborylated cyclobutanes. Isolated yields reported as the average of two runs.

Further elaboration of the C-B bond is possible. For example, oxidation of 5 and 5’a under standard conditions^[11d] yielded N-protected, cyclobutyl β-amino alcohols 6 and 6’a (Scheme 6).^[17] Boronic esters 5 and 5’a can also be readily converted into potassium trifluoroborate salts 7 and 7’a following Lloyd-Jones’s KF/tartaric acid protocol (Scheme 6).^[18]

Scheme 6.

Diastereoselective synthesis of hindered aminoborylated cyclobutanes. Isolated yields reported as the average of two runs.

Furthermore, dual iridium photoredox/nickel catalysis^[19] proved effective for (hetero)arylation of borate salt 7^[20] after some modest modification of reported conditions (see Supporting Information for other conditions surveyed).^[21] As shown in Scheme 7, both phenyl and pyridyl bromides coupled with 7 to afford 2-(hetero)arylated amidocyclobutanes (8) in moderate to good yields.^[22]

Scheme 7.

Substrate scope of dual iridium photoredox/nickel-catalyzed cross-coupling of 7. Isolated yields reported as the average of two runs at 0.35 mmol scale. Values in parentheses are the trans:cis diastereomeric ratios of isolated material. [a] Using aryl iodide. [b] Isolated yield of trans diastereomer only. [c] From crude mixture ¹H NMR analysis.

In all cases, the trans diastereomer formed as the major product.^[23] Minor amounts of the cis product diastereomer were also obtained. Their origin remains to be concretely identified,^[24] but we tentatively propose cis diastereomers may arise in the case of 7 owing to a weak coordinative directing effect exerted by the β-amide substituent^[25] on the nickel catalyst that ultimately results in C(sp³)–C(sp²) bond formation on the same side of the cyclobutane ring.^[26]

Additional C-B bond functionalizations can be performed with the secondary aminoborylated cyclobutane 5. For example, compound 5 can be alkylated with methylvinyl ketone in the presence of an Ir catalyst under photoredox conditions to produce 9 (Scheme 8a).^[27,28] Compound 5 can also be fluorinated stereoselectively with Selectfluor (Scheme 8b).^[28,29] Finally, we demonstrate that the azaborabicyclohexane 4’a-BOnC₁₂H₂₅ intermediate is also a competent starting material for C-B bond functionalization reactions,^[30] e.g., in stereospecific oxidation (Scheme 8c).

Scheme 8.

Additional C-B bond functionalizations.

In summary, we have provided the first demonstration of the synthetic utility of the 1,2-azaborine reagent in the preparation of functionalized cyclobutane derivatives. We developed flow conditions to support scale-up of regio- and diastereoselective 1,2-azaborine photoisomerization. The quantities of photoisomer thus obtained were applied to the development of a route to a cis-1,2-aminoborylated cyclobutane and analogues bearing quaternary carbon centers. Further derivatizations at the borylated position to produce a series of 2-functionalized aminocyclobutanes were also achieved. Going forward, we aim to explore the synthesis of other richly functionalized structural motifs by way of using 1,2-azaborines as a versatile synthon.