Generation and Reactivity of 2-Amido-1,3-diaminoallyl Cations: Cyclic Guanidine Annulations via Net (3+2) and (4+3) Cycloadditions

Abstract

Toward a method for direct conversion of alkenes to cyclic guanidines, we report that 1,3-dipolar cycloadditions of 2-amido-1,3-diamino allylic cations with alkenes provides a new method for direct cyclic guanidine annulation. Generated under oxidative conditions, the 2-amido-1,3-diaminoallyl cations, react as 1,3-dipoles providing rapid access to 2-amino imidazolines through net (3+2) cycloadditions. The utility is demonstrated through a concise synthesis of the oroidin alkaloid, phakellin. The described 1,3-dipole also participates in net (4+3) cycloadditions with dienes.

Graphical Abstract

Cyclic guanidines (2-amino imidazolines) play an important role in medicinal chemistry and biochemistry and are found in numerous natural products.¹ The hydrogen bonding capability of this polar, basic moiety is useful in the design of small molecule drugs in particular for interactions with RNA and DNA.² Numerous natural products including saxitoxin³ and monanchorin,⁴ members of the pyrrolo-2-aminoimidazole⁵ alkaloid family including phakellin, and several drug leads⁶ possess cyclic fused or bridged guanidines (Figure 1a). Reported methods for the synthesis of cyclic guanidines include cyclization of diamines⁷ and metal-promoted hydroaminations of propargyl guanidines.⁸ Tepe and co-workers reported a N-bromosuccinimide initiated, one-pot annulation of a Boc-guanidine to an alkene in their synthesis of dibromphakellin, indolo-phakellin,⁹ and a proteasome inhibitor derivative, bromo-indolo-phakellin.¹⁰ However, methods for direct cyclic guanidine annulation of alkenes are rare but include intramolecular Pd-mediated oxidative diaminations,¹¹ intramolecular Pd- and Ag-catalyzed animations,¹² halogen catalyzed intramolecular diamination of terminal olefins,¹³ and radical-mediated net cycloadditions with diaziridinimines.¹⁴ Our continued interest in the synthesis and biosynthesis of various pyrrole aminoimidazole alkaloid family members^5,15 led us to consider direct routes for cyclic guanidine annulation of alkenes.

Figure 1.

(a) Natural products and other bioactive compounds bearing a cyclic guanidine (b) previously developed aza-oxyallylic and diaza-oxyallylic cations for (3+2) cycloadditions (c) current studies describing novel 2-amido-1,3-diazaallylic cations for (3+2) and (4+3) dipolar cycloadditions.

Oxyallyl cation intermediates and nitrogen-substituted oxyallyl cation intermediates have been extensively studied for (3+2) cycloadditions with alkenes.¹⁶ Aza-oxyallyl cations were previously developed for 1,3-dipolar cycloadditions (Fig. 1b)¹⁷ following proposals for this type of dipolar intermediate by Sheehan in the 1960s.¹⁸ The Jeffrey group was the first to demonstrate the utility of aza-oxyallyl cations generated under basic conditions for (4+3) and (3+2)¹⁹ cycloadditions with alkenes. Subsequently, the same group used oxidative conditions for 1,4-diamination and also for (3+2) cycloaddition (Fig. 1b)²⁰ and the Wu²¹ group extended the use of these 1,3-dipoles to (3+2) dearomative cycloadditions with indoles. Liao employed the aza-oxyallyl cation in their formal synthesis of mini-fiensine²² and Xing reported its use for conversion of al-kynes to pyrrolidinone derivatives.²³ These cycloadditions all employed chemical oxidants to generate the 1,3-dipoles, however, Luo described the use of electrochemical oxidation for their generation.²⁴ Zhao reported rapid access to benzodiazepines via formal (4+3) cycloaddition reactions of aza-oxyallyl cations with anthranils,²⁵ the Huang group²⁶ reported (3+2) cycloadditions with styrenes, and very recently the Kim group used aza-oxylallyl cations for (3+2) cycloadditions leading to dihydrobenzoxazine-fused spi-rooxazoline-4-ones.²⁷

Building on these precedents, we envisioned generation of a 2-amido-1,3-diaza allylic cation 1 that might undergo a formal cycloaddition with alkenes (Fig. 1c). Herein, we report the generation and application of this intermediate for direct functionalization of alkenes through net (3+2) cycloadditions and also (4+3) cycloadditions with dienes. Mechanistic details of this reaction and competing pathways were uncovered through isolation of several interesting by-products.

We anticipated that an appropriately substituted guanidine, on treatment with a base and an oxidant, might generate a 2-amido 1,3-diaza-allylic cation (1). Previous studies by Kikugawa²⁸ and Jeffrey^19b suggested that an N-alkoxy group stabilizes and likely promotes generation of oxy-substituted aza-allylic cations likely through increasing NH acidity but also through electron donation stabilization of the incipient allylic cation. Thus, two N-benzyloxy groups and a N-tosyl group on guanidine 2 were employed, with the latter expected to increase the NH acidity and enable a facile deprotonation and subsequent oxidation of the resulting anion to deliver the targeted 2-amido-1,3-diaza-allylic cation 1.^20a

Our studies of the proposed cycloaddition began with 1,3-dimethyl indole (3a) as a substrate, building on conditions developed by Jeffrey for related reactions with aza oxyallylic cations.^20b We initiated our studies employing sodium tetrafluoropropoxide (NaTFP) as base and PhI(OAc)₂ (PIDA, added as a CH₃CN solution) as oxidant in CH₂Cl₂, a solvent required to solubilize guanidine 2. Addition of guanidine 2 followed by indole 3a indeed provided the expected (3+2) cycloadduct 4a in 64% yield after 3 h (Table 1, entry 1; structure confirmed by X-ray crystallography, see SI for details). A solvent screen (entries 2–4) revealed that the solvent mixture of 2,2,3,3-tetrafluoro-1-propanol²⁹, dichloromethane, and acetonitrile (1:4:8) led to the highest yield of cycloadduct 4a (90% yield, entry 5). In efforts to expand the solvents useful for this transformation, CH₃CN alone was also studied and when combined with dropwise addition of PIDA in CH₃CN, this condition provided cycloadduct 4a in 80 % yield (entry 6). Reversing the addition order, by adding guanidine 2 as a solution in CH₃CN to a mixture of indole 3a, oxidant, and base produced the cycloadduct 4a in significantly reduced yield (20%, entry 8) likely due to competitive oxidation of the electron rich indole. Importantly, solid Na₂CO₃ (Table 1, entry 9) led to 4a in 86% yield (entry 9) while no base led to cycloadduct, albeit in reduced yield (52%, entry 10). However, as expected, no reaction was observed in the absence of oxidant PhI(OAc)₂ (entry 11) and the use of PhI(OCOCF₃)₂ did not lead to improvement (entry 12). Importantly, the use of only CH₃CN as solvent and solid Na₂CO₃ provided 70% yield of cycloadduct 4a (entry 13). We found in general that this cycloaddition does not require the use of fluorinated solvents and the more readily available base, Na₂CO₃, can be used (see yields in red in Schemes 1 and 2).

Table 1.

Optimization of (3+2) cycloaddition with indole 3a

entry	oxidant (equiv)	solvent	% yieldb
1	1.5	CH2Cl2	64
2	1.5	TFP	42
3	1.5	CH2Cl2/ CH3CN (1:2)	66
4	1.5	TFP/ CH3CN (1:8)	73
5	1.5	TFP/CH2Cl2/CH3CN (1:4:8)	90
6	1.5	CH3CN	80
7	1.0	TFP/CH2Cl2/CH3CN (1:4:8)	52c
8	2.0	“	20d
9	1.5	“	86e
10	1.5	“	52f
11	-	“	0
12	1.5	“	50g
13	1.5	CH3CN	70h

^bIsolated, purified yields.

^cEmployed 1.0 equiv each of indole 3a and NaTFP.

^dGuanidine 2 (1.2 equiv) as an CH₃CN solution was added to a mixture of NaTFP (2.0 equiv), oxidant, and indole.

^eSolid Na₂CO₃ (1.2 equiv) was employed as base.

^fNo base was added.

^gPhI(OCOCF₃)₂ (1.5 equiv) used as oxidant.

^hSolid Na₂CO₃ (1.2 equiv) employed as base. TFP = 2,2,3,3-tetrafluoro-1-propanol; NaTFP = sodium 2,2,3,3-tetrafluoro-1-propoxide.

Scheme 1. Guanidine annulations via net (3+2) cycloadditions of guanidine 2 and alkenes 3.

^aStarting material (~10–20%) was recovered. ^bPrepared by Representative Procedure B (see SI for further details). ^cPrepared using only CH₃CN as solvent with Na₂CO₃ as base (cf. entry 13, Table 1).

Scheme 2. (4+3)/(3+2) Cycloaddition reactions of dienes 7 with the 1,3-dipole derived from guanidine 2.

^aPrepared using only CH₃CN as solvent with Na₂CO₃ as base (see SI for more details, Table S1, entry 6)

Employing the optimized conditions, we investigated several 1,3-disubstituted indoles and these substrates underwent cycloaddition efficiently to provide tricyclic adducts 4b-e in 82–86% yields (Scheme 1). Bulky substituents at C-3, e.g. i-propyl and cyclohexyl, were tolerated providing the corresponding cycloadducts 4b and 4e, along with a pendant ester producing cycloadduct 4d (86%). The use of C3-unsubstituted substrates 3f,g led to extensive production of non-cyclized indolo guanidines 5,6 in addition to the expected (3+2) adducts 4f,g. 2,3-Disubstituted indoles had diminished rates of cycloaddition but still led to good mass recovery providing cycloadduct 4h (73%; confirmed by X-ray crystallography, see SI for details) along with recovered starting material (21%). Use of a removeable substituent on the indole nitrogen, e.g. N-benzyl group, was also tolerated providing the N-benzyl adduct 4i (87%).

Several 6-membered oxygen and nitrogen containing heterocycles were also studied, providing the corresponding N-Boc piperidine-fused guanidine 4j (54%) and the 1,4-dioxane fused-guanidine 4k (36%) accompanied by ~20% starting material in both cases. p-Methoxy styrene (3l) also participated providing cycloadduct 4l in 38% yield. Other alkenes were studied but did not participate in this cycloaddition under the described reaction conditions (see SI for further details).

We also studied the potential for (4+3) cycloadditions of the 2-amido diaminoallylic cation 1 with dienes. Following extensive optimization with furan (7a, see SI for further conditions studied), which previously demonstrated good reactivity in related cycloadditions,^{19g, 20a} both (4+3) and the (3+2) adducts, 8a and 9a respectively, were obtained under optimized conditions. Excess diene was sometimes required due to instability to the oxidant, PIDA. With optimized conditions, application to various dienes led to several novel bridged guanidines 8, in some cases accompanied by intervening (3+2) cycloadducts 9 (Scheme 2). Cyclopentadiene (7b) and 6,6-dimethyl fulvene (7g) provided both bridged and fused cyclic guanidines through net (4+3) and (3+2) cycloadditions, respectively. Several furans and pyrroles 7c-g, h also participate. 3-Bromofuran (7d) provided the bridged ether 8d (46%) equipped with a vinyl bromide for further functionalization. Similarly, dihydroxy furan (7f) participates providing both cycloadducts 8f (52%) and 9f (20%) showing tolerance for unprotected alcohols. The N-silyl bromopyrrole 7h led to regeneration of the pyrrole ring following (3+2) cycloaddition due to HBr elimination. Cyclohexadiene (7i) gave only the (4+3) cycloadduct 8i in 54% yield while 2-acetoxy cyclohexadiene (7j) led to poor yields of the bridged cycloadduct 8j (14%) likely due to acetate hydrolysis under the basic reaction conditions.

Several lines of evidence gathered during this study suggest a stepwise, ionic mechanism for the described net cycloadditions (Scheme 3). With indole substrate 3, nucleophilic addition of the C-3 olefinic carbon to the allylic cation 1 provides a stabilized carbocation, iminium 10. Cycloadditions with C3-unsubstituted indoles 10 (R² = H) enables deprotonation of the iminium, driven by rearomatization, to give indole 5 which suggests a competitive elimination of intermediate 13 and a stepwise mechanism. Interestingly, indole 5 had lost one benzyloxy group which we propose derives from a retro-ene process from anionic intermediate 14. However, with C3-substituted indoles 10 (R² ≠ H), rearomatization through elimination is not possible and thus these substrates provide higher yields of (3+2) cycloadducts 4. In this case, initial adducts were determined to be a ~1:1 mixture of oxime 11 (~1:1, E/Z) and cycloadduct 12. This would be anticipated for a stepwise mechanism proceeding through iminium 10 followed by ring closure by either the pendant N-Ts or N-OBn substituted amines. In early studies, the crude reaction mixture was directly loaded on SiO₂ prior to NMR analysis and under these conditions, oxime 11 is converted to the more stable tosylimine 12. This was verified by isolation of oximes 11 (~1:1, E/Z) and exposure to mild acid leading to quantitative conversion to cycloadduct 12.

Scheme 3. Proposed mechanisms for by-products suggestive of stepwise guanidine annulation.

^aInitial synthesis of isomeric tricyclic adducts 11 and 12 and a presumed retro-ene process leading to indolo guanidine 5 and benzaldehyde.

The N-O bonds of indolo-guanidine 4a were selectively cleaved by Mo(CO)₆ to furnish tosyl indolo-guanidine 16 in 78% yield (Scheme 4).²¹ Global deprotection was achieved with excess SmI₂ to provide guanidine 15 in 92% yield. Providing further structural confirmation for oximes 11 (~1:1 E/Z), treatment with excess SmI₂ led to the same cyclic guanidine 15. However, we cannot exclude the possibility that rearrangement to adduct 4a precedes deprotection under these conditions.

Scheme 4.

Methods for partial and complete deprotection of indole 4a and correlation to oximes 10

We targeted the synthesis of the N-tosyl variant 18 of the [2.3.0] N-phenyl bicyclic guanidine (see Fig. 1) previously developed as an antihypertensive agent.^6f Selective removal of the benzyloxy groups in this case was achieved with SmI₂ (12.0 equiv) to obtain alkene 17 (ORTEP, inset Scheme 5). Subsequent hydrogenation produced the N-Ts bicyclic guanidine 18.

Scheme 5.

Applications of the 1,3-dipolar cycloaddition to bridged guanidine 18

As a model study for the projected ring annulation to palau’amine, we applied this method to the pyrrolo-imidazole alkaloid, phakellin (22, Scheme 6). Indeed, the tricyclic alkene 19,³⁰ delivered the desired (3+2) cycloadduct 20 (45%) along with elimination product 21 (21%). Global deprotection of guanidine 20 with SmI₂ provided rac-phakellin (22) in 60% yield. Interestingly, the elimination by-product 21 bears structural resemblance to acyclic guanidines present in several PAIs including the recently isolated 1-N-methylugibohlin³¹ (inset, Scheme 6).

Scheme 6.

Total Synthesis of rac-phakellin (22)

In conclusion, we developed a new method for direct guanidine annulation of alkenes leading to a variety of bicyclic guanidines through net (3+2) cycloadditions employing guanidine-derived 1,3-dipoles. Net (4+3) cycloadditions are also possible through these intermediates leading to bridged guanidine systems. The by-products isolated in addition to the observed greater reactivity of electron rich alkenes, supports a stepwise, net cycloaddition. Further studies of this novel 1,3-dipole, the 2-amido diamino allylic cation 1, including computational studies of this intermediate and reaction pathways, methods to increase the reactivity of these 1,3-dipoles, and further applications of its utility are underway.