Synthesis, Structural Analysis, and Reactivity of Bridged Orthoamides by Intramolecular Schmidt Reaction

The intramolecular Schmidt reaction is a useful method for the synthesis of lactams.¹ Most commonly, this reaction affords fused lactams bearing a nitrogen atom at a ring fusion position by preferential migration of the carbon bearing the azide-containing side chain. We have recently shown that certain substitution types lead to the alternative regioisomeric product containing a bridged ring system incorporating nitrogen at a bridgehead position (Scheme 1).² Such compounds are noteworthy since they contain distorted amide bond linkages, which are of interest from theoretical, mechanistic, and reactivity perspectives.³ However, these methods are substrate dependent and provide mixtures of bridged and fused lactams, even in the most favorable cases. We now show that intramolecular Schmidt reactions can be reliably steered toward bridged heterocycles containing orthoamides in high yields. The method is broad in scope and allows for systematic study of compounds that are analogous to elusive tetrahedral intermediates of amide addition reactions.⁴

Scheme 1.

We hypothesized that (1) the formation of bridged lactams using the Schmidt reaction of ketones and azides was hampered by the necessity of forming a strained non-planar amide, (2) that this could be mitigated by relying instead on formation of a relatively stable iminium ether intermediate, and (3) that intramolecular capture would provide an easy road to a stable – and unusual – orthoamide product. That we had previously demonstrated that ketals also participate in azido-Schmidt chemistry^5a combined with a single report of a bridged ring system formed in this way^5b suggested that reactions of these species might favor bridged products to a greater degree.

To maximize the chance for success, we focused first on alkyl azides that had already been shown to react preferentially by path b (Scheme 1).² Thus, Lewis or protic acid treatment of a series of ketals derived from 1,3-dioxolane afforded bridged orthoamides in excellent yield (Table 1). The products were stable to the reaction and chromatographic isolation conditions. As hypothesized, the transformation proceeded with efficient control of regiochemistry, while the ketones corresponding to 1a–f provided a ca. 5:2 mixture of bridged and fused lactams.^2a These results demonstrate that this approach accommodates a variety of ketal types. In contrast, the behavior of enol ether 1f shows that success depends upon intramolecular capture of the intermediate oxonium ion and that the [4.3.1] ring system containing a hemiaminal is unstable.

Table 1.

Formation of bridged orthoamides from ketals.

entry	azide	product	conditionsa	yield (%)b
1	(1a) R, R = H	(2a)	A	88
2	(1b) R, R = -(CH2)4-	(2b)	A	71
3	(1c) n = 1, R, R = H	(2c)	B	77
4	(1d) n = 1, R, R = Me	(2d)	B	75
5	(1e) n = 2, R, R = H	(2e)	B	70c
6			C	63

^aA: BF₃•CH₃CN, 5.0 equiv, 0 °C to rt; B: BF₃•CH₃CN, 3.0 equiv, −78 °C to rt; C: TfOH, 5.0 equiv, 0 °C.

^bIsolated after chromatography

^cDetermined by ¹H NMR, see SI (Supporting Information) for details.

Subsequent work showed that a range of conformationally locked azido alkyl 1,3-dioxolanes substituted with aromatic rings in the 2 position could easily be transformed into analogous products (Table 2, entries 1–4). Throughout, these reactions proceeded with complete control of regiochemistry.⁶ Even the conformationally flexible azide 1k (entry 5) gave exclusively bridged product, in sharp contrast to the corresponding ketone, which only afforded fused lactam. A number of substrates with heteroatoms in the 2-position underwent the reaction in good to excellent yields (entries 6–8). Finally, a conformationally flexible azide with a shorter azide chain cleanly furnished the [4.2.1] bridged scaffold (entry 9). It is likely that in this case the fused product did not form due to strain developing en route to the corresponding a four-membered-lactam containing fused product.

Table 2.

Azide scope.^a

entry	azide	R1	R2	n	ketal	yield (%)
1	(1g)	t-Bu	4-(OMe)C6H4	1	(2g)	92
2	(1h)	t-Bu	3,4,5-(OMe)3C6H2	1	(2h)	86
3	(1i)	t-Bu	3,5-(OMe)2C6H3	1	(2i)	86
4	(1j)	t-Bu	4-(NO2)C6H4	1	(2j)	94
5	(1k)	H	C6H5	1	(2k)	59
6	(1l)	H	OMe	1	(2l)	52b
7	(1m)	H	SMe	1	(2m)	50
8	(1n)	H	SPh	1	(2n)7	78b
9	(1o)	H	C6H5	0	(2o)	91

^aBF₃•CH₃CN, 3.0–5.0 equiv in CH₂Cl₂

^bTMSOTf.

We have proposed that chairlike cyclohexanones afford bridged products only when the azide-containing side chain and the diazonium cation are both pseudoaxial (Scheme 2, intermediate C).^5c Here, the N₂⁺ leaving group is axial, leading to migration of the antiperiplanar alkyl group. We propose that the enhanced regioselectivity observed with ketal-containing vs. ketone-containing azides results from steric and electronic repulsion between the diazonium cation in the D conformation relative to C. In both cases, the most stable conformation of the O-alkyl side chain is likely to be as drawn; the N₂⁺ leaving group in D experiences a disfavorable syn-1,5 interaction with the O-alkyl group that is absent in C. In the analogous reactions of ketones, where the OR group is replaced by either a proton or a Lewis acid, such effects are minimized relative to the present case (and the role of attractive non-bonded stabilization between N₂⁺ and R₂ is even more important). Interestingly, the Deslongchamps principle of maximal overlap could also come into play here. Thus, if the side chain orientation proposed predominates in the transition state leading to product, it is only possible for an oxygen electron lone pair to be antiperiplanar to the migrating bond in C but not in D (where the O-alkyl group is instead antiperiplanar to the darkened bond in D).^4b Finally, the success of cases lacking an alkyl group at R₂ indicates that the reactive conformation in these examples is exclusively B and not A.

Scheme 2.

The α-amino ketal functionality is stabilized by a non-planar arrangement of atoms. The X-ray crystal structure of 2n reveals that the N1–C1 bond (1.448 Å) as well as the C1-O1 (1.405 Å) and C1–O2 (1.416 Å) bonds are shorter than typical N–C_sp3 (1.469 Å) and C_sp3–O (1.432 Å) bonds (Figure 1).^4a The C1-C2 bond length of 1.543 Å is slightly longer than the average C_sp3-C_sp3 bond (1.530 Å).^4a The torsion angles between N_lp and C1-O1 of 69.5° and between N_lp and C1-O2 of 46.1° are consistent with the absence of that N_lp→σ*_C-O interactions in this system. By contrast, there exists a reasonably good arrangement between O1_lp and N1-C1 bond (~148°) and between O2_lp and N1-C1 bond (~158°). While the lengths of C1-O1, C1-O2 and C1-C2 bonds are in agreement with an anomeric effect resulting from O_lp→σ*_C1-N1 and O_lp→σ*_C1-C2 interactions, the shortened N1-C1 bond seems to be characteristic to the tetrahedral intermediate stabilized by scaffolding effects of a medium-sized ring.⁴

Figure 1.

X-ray structure of 2n. Selected bond lengths [Å] and angles [°]: N1-C1 1.448, C1-O1 1.405, C1-O2 1.416, C1-C2 1.543, C6-N1-C1-O1 −38.1, C7-N1-C1-O1 177.0, C7-N1-C1-O2 61.5, C6-N2-C1-O1 −153.7

We begun to examine the reactivity of these α-amino ketals, (Scheme 3). For example, we found that acid hydrolysis affords 9-membered ring amino esters (these can be converted into the bridged lactams by treatment with base; see SI). In contrast, chromatography-stable hemiaminal ethers are formed following reduction (DIBAL-H), while reaction with Me₃Al delivers a “twisted enamine”. As expected, α-amino ketals also undergo reactions at nitrogen.

Scheme 3.

Conditions: (a) 2i, THF, 60 °C, 12 h, 89%.(b) 2c, PhH, 80 °C, 15 h, 67%. (c) 2d, CH₂Cl₂, 40 °C, 15 h, 56%. (d) 2h, rt, 20 h, 96%.

In summary, we have established a direct synthesis of orthoamides, which are analogs of tetrahedral intermediates derived from amide addition reactions. Continuing efforts to study this class of compounds are underway.