Directed carbonylative (3+1+2) cycloadditions of amino-substituted cyclopropanes and alkynes: reaction development and increased efficiencies using a cationic rhodium system.

Urea-directed carbonylative insertion of Rh(I)-catalysts into one of the two proximal C e C bonds of aminocyclopropanes generates rhodacyclopentanone intermediates. These are trapped by N-tethered alkynes to provide a (3 þ 1 þ 2) cycloaddition protocol that accesses N-heterobicyclic enones. Stoichio- metric studies on a series of model rhodacyclopentanone complexes outline key structural features and provide a rationale for the ef ﬁ cacy of urea directing groups. A comprehensive evaluation of cycloaddition scope and a ‘ second generation ’ cationic Rh(I)-system, which provides enhanced yields and reaction rates for challenging substrates, are presented. (cid:1) 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). involving trans -1,2-disubstituted aminocyclopropanes.


Introduction
Synthetically flexible and modular entries to stereochemically rich N-heterocyclic scaffolds are of topical interest to the pharmaceutical sector. 1 Recently, we reported a strategy to access selectively amino-rhodacyclopentanones 3 by the carbonylative insertion of Rh(I)-catalysts into the more hindered CeC bond of amino-substituted cyclopropanes 1 (Scheme 1). 2 Specifically, we established that carbonyl-based N-protecting groups can direct oxidative addition (to 2) and CO-insertion to generate regioisomer 3 in a selective manner. Trapping of 3 with N-tethered alkynes provided a (3þ1þ2) cycloaddition strategy to generate N-heterobicyclic enones (4/5). 3 These investigations provided proof-ofprinciple for an approach that has the potential to enable a wide range of carbonylative cycloadditions for accessing directly 'sp 3rich' chiral scaffolds. 1,4 Indeed, to date, the catalysis platform outlined in Scheme 1 has served as the basis for related (3þ1þ2) cycloadditions involving alkenes, 5 and a (7þ1) cycloadditionfragmentation approach to substituted azocanes. 6 In this article we disclose our full studies on the development of urea-directed (3þ1þ2) cycloadditions involving alkynes. In addition to key mechanistic considerations, detailed studies on the generation of the amino-rhodacyclopentanones intermediates are outlined, and a simple modification to our original (3þ1þ2) cycloaddition protocol is disclosed, which provides enhanced yields and reaction rates for challenging substrates.

Contents lists available at ScienceDirect
Tetrahedron j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / t e t

Results and discussion
At the outset of our studies, catalytic protocols reliant on the intermediacy of rhodacyclopentanones were scarce and processes involving amino-substituted variants had not been developed. Pioneering studies by Wilkinson demonstrated Rh/CO insertion into cyclopropane to generate a dimeric rhodacyclopentanone. 7a Subsequent work by McQuillin examined the regioselectivity of this process for substituted variants. 7b An alternative approach was reported by Murakami and Ito, where rhodacyclopentanones were accessed by the insertion of Rh(I)-systems into the acyl-carbon bond of cyclobutanones. 8 This process has served as the basis for a series of methodologies, 9e11 however, carbonylative rhodacyclopentanone formation has not been as widely exploited in synthesis. 3a,12 Notable processes that harness this approach include carbonylative rearrangements of spiropentanes, as reported by Murakami,12b and (3þ1þ2) cycloadditions involving alkynes to generate carbocyclic systems, as reported by Narasaka. 3a In this latter process, the alkyne is invoked as a directing group. For the strategy outlined in Scheme 1, the most pertinent work is that of Chirik, who demonstrated efficient phosphinite-directed insertion of neutral Rh(I)-systems into alkyl-substituted cyclopropanes, albeit under non-carbonylative conditions. 13,14 With this report in mind, our initial goal was to establish the viability of using Ndirecting groups to control Rh/CO insertion.

Regioselective generation and key structural features of amino-rhodacyclopentanones
Preliminary studies examined the regioselectivity of Rh(I)insertion into carbamate-protected aminocyclopropane 6a (Scheme 2). A neutral Rh(I)-system, derived from [Rh(cod)Cl] 2 and PPh 3, delivered branched vinyl carbamate iso-7a via insertion into the less hindered CeC bond. In contrast, employment of [Rh(cod) 2 ] BF 4 as the pre-catalyst resulted in rapid and quantitative formation of linear vinyl carbamate 7a via Rh-insertion into the more hindered CeC bond. For [Rh(cod) 2 ]BF 4 , significant conversion to 7a was observed even at 60 C; this indicates that oxidative addition is reasonably facile. The faster rate of vinyl carbamate formation versus [Rh(cod)Cl] 2 may be due to an additional vacant coordination site facilitating b-hydride elimination. These results suggest that relatively Lewis acidic metal complexes (i.e., cationic vs neutral) are required to ensure coordination to the carbamate directing group and contrast Chirik's studies where neutral Rh(I)-systems were effective with strongly directing phosphinites. 13 To confirm the role of the directing group in the conversion of 6a to 7a, analogous regioselectivity studies on alkyl-substituted cyclopropane 8 were conducted (Scheme 3). Both neutral and cationic Rh(I)-systems delivered only branched products, resulting from Rhinsertion into the less hindered CeC bond; no detectable levels of linear product 10 were observed. Using [Rh(cod)Cl] 2 , alkene 9a was observed as the sole product, whereas [Rh(cod) 2 ]BF 4 delivered a mixture of branched adducts 9aec. 9bec presumably arise from Rh-catalysed isomerisation of alkene 9a. The sole formation of branched alkenes 9aec from alkyl-substituted cyclopropane 8 confirms the importance of the N-directing group for the conversion of 6a to 7a in Scheme 2.
Further studies were conducted to assess the capability of other N-protecting groups for directing Rh-insertion (Scheme 4). Accordingly, amide 6b and sulfonamide 6c were exposed to [Rh(cod) 2 ]BF 4 /PPh 3 under non-carbonylative conditions. The exclusive formation of linear products 7b/c was observed in both cases, thereby supporting a directed oxidative addition pathway.
The studies described above show that, under non-carbonylative conditions, only Lewis acidic cationic Rh(I)-catalysts can be utilised for protecting group directed oxidative addition. It was anticipated that, under carbonylative conditions, CO-ligated neutral catalysts may be sufficiently Lewis acidic because CO is a strong p-acceptor ligand, and this, in turn, should increase the electron deficiency of the Rh-centre. However, when carbamate-protected cyclopropane 6a was exposed to a neutral Rh-catalyst ([Rh(cod)Cl] 2 , PPh 3 ) under a CO atmosphere, formation of linear or branched vinyl carbamates 7a and iso-7a did not occur. The absence of 7a or iso-7a is suggestive of fast migratory insertion of CO at the stage of the incipient rhodacyclobutane to generate small quantities of the corresponding rhodacyclopentanone. Consequently, stoichiometric reactions were conducted with the aim of isolating and characterising neutral amino-substituted rhodacyclopentanones and thereby determining the regioselectivity of Rh/CO insertion. Wilkinson reported the synthesis of a dimeric rhodacyclopentanone from cyclopropane and [Rh(CO) 2 Cl] 2 ; addition of triphenylphosphine generated the corresponding monomeric complex. 7a  catalysts are effective for directed insertion into aminocyclopropanes. Cowie and co-workers have reported an X-ray crystal structure of a bimetallic rhodacyclopentanone derived from allene insertion into a methylene-bridging CO-ligated RheRu complex. 15 However, the structure shown in Scheme 5A is the first X-ray crystal structure of a rhodacyclopentanone derived from carbonylative cyclopropane ring expansion.
At 27 C, the 1 H and 13 C NMR spectra (CD 2 Cl 2 ) of rhodacyclopentanone 12 were broad and suggestive of the presence of two different species (Scheme 5B). Low temperature NMR data (À13 C) revealed a divergence and sharpening of the two sets of signals, indicating the presence of two components in dynamic equilibrium. The precise structure of the two components has not been determined, however, rhodacycle 12 may be in equilibrium with a monomeric species or, alternatively, two diastereomeric complexes, such as 12 and 13, may be interconverting. Addition of one equivalent of PPh 3 resulted in sharpening of the signals in the 1 H NMR spectrum to a single, new, rhodacyclopentanone complex 14 (Scheme 5A). The 31 P NMR spectrum showed a doublet at 16.2 ppm (J Rh-P ¼80 Hz) confirming coordination of PPh 3 to the Rh-centre. Addition of a second equivalent of PPh 3 resulted in no change in the aliphatic region of the 1 H NMR spectrum, whereas the 31 P NMR spectrum showed a doublet at 16.2 ppm and an additional singlet at À5.56 ppm (corresponding to unbound PPh 3 ) in a 1:1 ratio, thereby demonstrating that only one phosphine ligand is coordinated in the new rhodacyclopentanone complex. Wilkinson and co-workers reported that addition of excess PPh 3 to dimeric rhodacyclopentanone complexes (derived from cyclopropane) delivered monomeric complexes with two phosphine ligands bound to the Rh-centre. 7a In the present case, both the directing group carbonyl and a CO ligand remain bound to the Rh-centre, consequently only one phosphine ligand can be accommodated. The 13 C NMR spectrum of complex 14 (see the Supplementary data) confirmed the presence of the CO and PPh 3 ligands, and also showed that the carbamate directing group remains bound to the Rh-centre because (a) 3 J-coupling of C3 to phosphorus was observed (J¼10 Hz) and (b) there was little change in the chemical shift of C3 between the dimeric and monomeric complexes (dimer 12: 164.6 ppm vs monomer 14: 165.3 ppm vs starting material 11: 157.8 ppm). 13 C NMR coupling constants were used to assign the geometry of complex 14. An alkyl/acyl CeRh coupling constant ofw20 Hz was observed for C5 and C8 and a carbonyl-Rh coupling of 77 Hz was observed for C9; these coupling constants are consistent with related structures in the literature. 16 The CeP coupling constants for C5, C8 and C9 are 86.0, 3.0 and 11.5 Hz respectively, indicating that C5 has a trans relationship to PPh 3 (large coupling constants are reported for trans CeRheP relationships). 16 C8 and C9 have small CeP coupling constants which suggests a cis relationship to PPh 3 . The structure of 14 was supported further by the relatively small J Rh-P , indicative of a complex with a coordination number of five or six and containing an alkyl group trans to the phosphine. 17 We have recently reported an X-ray crystal structure of a related phosphinebound rhodacyclopentanone complex, the NMR data of which are consistent with those of 14. 5 For the strategy outlined in Scheme 1, a key factor is the coordinating strength of the N-directing group. Studies in the literature have compared the binding strength of various carbonyl groups with Lewis or Brønsted acids, 18 but these results are not directly transferable to transition metal complexes. Consequently, we synthesised a series of rhodacyclopentanone complexes (15aec), which were characterised by X-ray diffraction, and evaluated the relative donor strength of each directing group by comparing the stretching frequencies of the trans CO ligand in each case (Scheme 6). 19 From these studies the following ranking of directing group strength emerges: urea>>carbamate>amide. Attempted synthesis of an analogous complex containing a sulfonamide directing group was not successful (cf. Scheme 4) and so, at the present time, we have been unable to quantify the donor strength of this class of directing group.

Development of a neutral Rh(I)-system for (3D1D2) carbonylative cycloadditions of aminocyclopropanes and alkynes
The regioselectivity studies outlined above demonstrate that amino-substituted rhodacyclopentanones can be accessed in a selective manner utilising the directing group strategy outlined in Scheme 1. Our initial synthetic studies sought to incorporate this activation mode into (3þ1þ2) cycloadditions involving N-tethered alkynes (Table 1). Narasaka and Koga have reported (3þ1þ2) carbonylative cycloadditions of alkyl-substituted cyclopropanes, possessing tethered alkynes, to afford carbocyclic enones. 3a In this process, the alkyne was proposed to direct oxidative addition of the Rh(I)-catalyst into the more hindered CeC bond and relatively harsh reaction conditions were required (20 mol % [Rh], 160 C). It was anticipated that our carbonyl directed approach may allow related ring expansions of substituted aminocyclopropanes to proceed under milder reaction conditions. It is important to note that, for processes described here, a key aspect is the requirement that the directing group dissociates from the metal centre after rhodacyclopentanone formation to allow CeN rotation and coordination of the alkyne. Consequently, a range of directing groups were evaluated in the hope of fine-tuning the equilibrium between 18 and 19 (Table 1).
Early studies identified that, at 160 C in DCB, a BINAP-ligated neutral Rh(I)-system effected cycloaddition of amide 16a to 17a in 26% yield under an atmospheric pressure of CO (Entry 1). At this stage, the influence of the directing group was investigated, and it was found that cycloaddition efficiency increased progressively as the donor strength was increased to carbamate 16b and then urea 16c (Entries 2 and 3; cf. Scheme 6). Notably, complete consumption of 16c was observed within 24 h, whereas cycloaddition of 16b to 17b took 72 h. The structure of N-heterobicyclic enone 17b was confirmed by single crystal X-ray diffraction. To enhance reaction efficiency further, methoxy-urea 16d was prepared, however, cyclisation of this substrate was not effective and only trace quantities of target 17d were observed (Entry 4). N-Boc and N-tosyl directing groups were also completely ineffective. The urea directing group of 16c allowed the reaction temperature to be lowered to 130 C and further optimisation led to the conditions outlined in Entry 5, which use 7.5 mol % [Rh] and P(3,5-(CF 3 ) 2 C 6 H 3 ) 3 as ligand to deliver 17c in 53% yield. Under these conditions, cationic Rh-sources were considerably less effective (Entries 6 and 7, vide infra). Final optimisation involved switching the solvent from DCB to PhCN and, under these conditions, 17c was generated in 72% yield after 72 h (Entry 8).
The results of the cycloadditions of 16aec (Entries 1e3) merit further comment because they suggest that equilibration to 19 is not a key issue (because stronger directing groups are more effective). It is likely that the dimethylurea directing group is especially efficient because it is able to outcompete the alkyne moiety for coordination of the Rh(I)-catalyst, and thereby enhance the rate of oxidative addition. Indeed, for related processes involving alkenes, we have established that carbamate directing groups are preferred over ureas. 5 In these cases, the less strongly coordinating alkene component does not inhibit directed oxidative addition and so a strongly coordinating urea directing group is not required. Additionally, equilibration to the intermediate p-complex (cf. 19) is more challenging, such that a less strongly coordinating directing group leads to faster rates.
The scope of the reaction with respect to the alkyne component has been assessed by exposing aminocyclopropanes 16een to the optimised neutral Rh(I)-system (Table 2). Both alkyl-and arylsubstituted alkynes can be utilised, and a range of electron-rich and -neutral derivatives (16eei) cyclised to the target enones 17eei in moderate to good yield. The structure of cyclohexenone 17h was confirmed by X-ray crystallography. However, electrondeficient alkynes (e.g., 16j) and heteroaromatic variants (e.g., 16k and 16l) cyclised less efficiently and the target enones were isolated in low yield. The inefficiency observed for quinoline derivative 17k may be due, in part, to competitive coordination of the Rh-catalyst   20 In all cases, reaction times are long and, ultimately, this stimulated the development of a 'second generation' protocol, which is discussed later. Heterocyclic products of greater stereochemical complexity can be accessed by employing substrates with substitution on the alkyne tether (Table 3). Carbonylative cycloadditions of substrates 16oeq delivered cyclohexenones 17oeq in good yields and with moderate diastereoselectivity; the relative stereochemistry of the major diastereomers was determined by NOE experiments. Substitution on the alkyne tether resulted in increased rates of cycloaddition relative to unsubstituted substrate 16c, and, in certain cases, this enabled the employment of marginally lower temperatures (120 C for 16o vs 130 C for 16c). The increased rate of reaction for 16oeq may reflect a greater propensity for the incipient rhodacycle to adopt a conformation that allows alkyne insertion. The observed levels of diastereoselectivity for 16oeq are constant over the timeframe of the reaction. Additionally, no equilibration was observed when the diastereomers of 17q were separated and resubmitted to the reaction conditions. These observations suggest that diastereoselection occurs during cycloaddition rather than by equilibration (via epimerisation) of the products. The modest diastereoselectivities are unsurprising given that high diastereoselection likely requires the R 1 group to bias Rh-insertion into one of the two diastereotopic cyclopropane CeC bonds of 16oeq. In related (3þ1þ2) cycloadditions involving alkenes, a solution to this issue was developed which relies upon reversible rhodacyclopentanone formation. 5 Unfortunately, this strategy is not applicable to the current scenario, perhaps due to more rapid insertion of the alkyne (vs the alkene), which diminishes reversibility.

Development of a cationic Rh(I)-system for (3D1D2) carbonylative cycloadditions of aminocyclopropanes and alkynes
Preliminary mechanistic studies indicated that cationic Rh(I)systems are especially effective for directed oxidative addition (see Section 2.1), presumably because the more Lewis acidic Rh(I)centre (vs neutral Rh(I)-systems) enhances coordination to the Lewis basic directing group. Accordingly, we decided to re-examine cationic Rh(I)-systems in more detail and 1 H NMR profiling of the reactions outlined in Entries 5 and 6 in Table 1 was undertaken (Fig. 1). This revealed that the use of [Rh(cod)Cl] 2 effects slow but steady formation of product 17c. On the other hand, employment of [Rh(cod) 2 ]BF 4 promotes faster product formation over the initial stages of the reaction, but after this only gradual degradation of the starting material occurs. These data establish that cationic Rh(I)sources provide higher initial turnover frequencies but less stable catalysts (i.e., lower turnover numbers) than neutral systems.
To improve the performance of cationic Rh(I)-derived systems we assayed a range of ligands and coordinating solvents that might stabilise the catalyst further (Table 4). In the event, simply changing the reaction solvent from DCB to PhCN provided conditions that generated 17c in 72% yield after 9.5 h (Entry 1). Other coordinating solvents (e.g., valeronitrile and DMF) were less effective (Entries 2 and 3). 21 The difference in performance between PhCN and the more strongly coordinating valeronitrile is particularly striking and indicates that reversible binding of the solvent to the metal centre is a key factor. 22 Using PhCN as solvent, a range of cationic Rh(I)-sources were efficient and no strong dependency on the counterion was observed (Entries 4e6). The key benefits of the conditions outlined in Entries 1 and 4e6 are the significantly shorter reaction times (7.5e9.5 h vs 72 h for Table 1, Entry 8) and better substrate scope (vide infra) compared to the neutral Rh(I)system. These 'second generation' conditions also tolerate decreased catalyst loading, but this comes at the expense of longer reaction times (Entry 7). 23 The reaction scope has been explored using the conditions outlined in Table 4, Entry 6. We have focussed on substrates that were problematic using the 'first generation' protocol and a comparison of the original and improved catalysis conditions is presented in   Pronounced improvements were also seen with substrates 16iel and higher yields (63e88%) of the targets 17iel could be achieved in considerably shorter reaction times (18e48 h vs 108e192 h using [Rh(cod)Cl] 2 ). Quinoline 16k required a higher catalyst loading (10 mol % [Rh(cod) 2 ]OTf) for efficient conversion, presumably due to the aforementioned issues associated with the Lewis basic quinoline nitrogen. 24 Terminal alkyne 16m, which provided only a 10% yield of 17m under the 'first generation' conditions, cyclised in 23% yield using the cationic Rh(I)-system. Evidently, in this case, further optimisation is still required and this will be a focus of future studies. However, even under these modified conditions, electron deficient alkyne 16n does not participate and target 17n was not observed. Intriguingly, for 16o and 16q (see Table 3), where additional substitution is present on the alkyne tether, reaction rates were improved using [Rh(cod) 2 ]OTf, but the isolated yields and diastereoselectivities associated with 17o and 17q were lower than when [Rh(cod)Cl] 2 was employed (using [Rh(cod) 2 ]OTf, 17o: 57% yield, 4:3 d.r., 39 h; 17q: 45% yield, 1:1 d.r., 24 h). In these particular cases, the neutral Rh(I)-system is therefore preferred.

(3D1D2) cycloadditions of substituted aminocyclopropanes
Carbonylative (3þ1þ2) cycloadditions involving trans-1,2disubstituted aminocyclopropanes can potentially deliver two different regioisomeric products depending on the selectivity of oxidative addition. Carbonylative cycloaddition of methyl-substituted cyclopropane 16r using the neutral Rh(I)-system, yielded a 5:2 mixture of regioisomers 17r and iso-17r, resulting from competing oxidative addition of the Rh(I)-catalyst into either of the proximal cyclopropane CeC bonds (Scheme 7). For the major regioisomer 17r, which is derived from oxidative addition of the Rh(I)-catalyst into less hindered proximal cyclopropane CeC bond a, complete transfer of the cyclopropane stereochemistry of the major diastereomer of 16r was observed. The regiochemistry of 17r was determined by HMBC analysis, and the relative stereochemistry was corroborated by NOE experiments. Minor regioisomer iso-17r was obtained as a 1:1 mixture of diastereomers, perhaps due to epimerisation of the C6 stereocentre of the product under the reaction conditions. Using [Rh(cod) 2 ]OTf, a 76% yield of 17r/iso-17r was achieved, and because the reaction temperature could be lowered to 120 C (from 140 C), selectivity for 17r also increased (4:1 17r:iso-17r vs 5:2 17r:iso-17r at 140 C using [Rh(cod)Cl] 2 ). 25 Notably, the cationic conditions resulted in a decrease in the reaction time from 108 to 38 h. For 16s, where the steric bulk of the cyclopropane substituent is increased, selectivity for oxidative addition of the Rh(I)-catalyst into the less hindered proximal CeC bond a was increased and a 5:1 ratio of 17s:iso-17s was obtained using the neutral system. Again, complete transfer of the cyclopropane stereochemistry to the major regioisomer 17s was achieved. In this case, the cationic Rh-system increased both the reaction rate and regioselectivity, but only provided a modest improvement to the yield.

Derivatisations of the cycloaddition products
The utility of the N-heterobicyclic enone products described here is outlined in Scheme 8. Addition of the Gilman-cuprate derived from n-BuLi to enone 17c proceeded with excellent facial selectivity, in favour of the cis-ring junction, to deliver 18 in 56% yield. 26 The C2 stereocentre of 18 was not readily controlled and attempted epimerisation of this position under a variety of basic or acidic conditions did not enhance diastereopurity. Hydrogenation of 17c (Pd/C, H 2 ) proceeded smoothly under mildly basic conditions (Et 3 N) to deliver 19 in 80% yield. Again, the stereochemistry of the ring junction was readily controlled, presumably due to syn-addition of hydrogen to the less hindered face of 17c, which delivers initially the depicted diastereomer of 19. The C2 stereocentre of 19 was labile under the reaction conditions, such that prolonged reaction times (48 h vs 2 h) afforded predominantly the alternate (and presumably thermodynamically favoured) diastereomer (not depicted). Oxidative transformations are also feasible. For example, under transfer hydrogenative conditions (Pd/C, cyclohexene), 27 the cyclohexenone moiety of 17h underwent oxidation to provide phenol 20, albeit in moderate yield.

Conclusions
In summary, we outline studies on the selective generation and trapping of amino-substituted rhodacyclopentanones. A range of N-directing groups are effective at directing oxidative addition of Lewis acidic Rh(I)-systems. This underpins an efficient and controlled approach to the key metallacyclic intermediates. For prototypical cycloaddition processes involving N-tethered alkynes, efficient oxidative addition requires a strongly donating urea directing group to outcompete the alkyne for coordination of the Rh(I)-catalyst. Subsequent dissociation of the urea at the stage of the rhodacyclopentanone is relatively facile and allows alkyne insertion to provide the heterocyclic target. A 'first generation' neutral Rh(I)-system is effective at promoting (3þ1þ2) cycloadditions, but suffers from low yields and long reaction times in cases where the alkyne component is electron deficient or bears a heterocyclic substituent. A 'second generation' cationic Rh(I)-system provides increased efficiencies in these cases and the success of this protocol is strongly dependent upon the use of PhCN as solvent. Current studies are focussed on the development of catalyst systems that tolerate synthetically versatile directing groups (e.g., carbamates) and allow expansion of the approach to asymmetric processes and 6-ring cyclisations.

General experimental
Starting materials sourced from commercial suppliers were used as received unless otherwise stated. Dry solvents, where necessary, were obtained by distillation using standard procedures or by passage through a column of anhydrous alumina using equipment from Anhydrous Engineering based on the Grubbs' design. 28 Melting points were determined using a Reichert melting point table and temperature controller and are uncorrected. Infrared spectra were recorded in the range 4000e600 cm À1 on a Perkin Elmer Spectrum either as neat films or solids compressed onto a diamond window. NMR spectra were recorded using either a Varian 400 MHz or JOEL ECS 400 MHz spectrometer. Chemical shifts are quoted in parts per million (ppm), coupling constants (J) are given in Hz to the nearest 0.5 Hz. Other abbreviations used are s (singlet), d (doublet), t (triplet), m (multiplet) and br (broad). 1 H and 13 C NMR spectra were referenced to the appropriate residual solvent peak. Mass spectra were determined by the University of Bristol mass spectrometry service by either electron impact (EI þ ) or chemical ionization (CI þ ) using a Fisons VG Analytical Autospec spectrometer, or by electrospray ionization (ESI þ ) using a Br€ uker Daltonics Apex IV spectrometer. Experimental procedures and data for compounds that were reported in our earlier work are not included here. 2 X-ray crystallographic data for compounds 12, 15b, 15c, 17b and 17h have been reported previously. 2,5 4.2. General procedure for (3D1D2) carbonylative cycloadditions of aminocyclopropanes and alkynes using a neutral Rh(I)-catalyst system An oven-dried reaction tube, fitted with a magnetic stirrer, was charged with [Rh(cod)Cl] 2 (3.75 mol %), P(3,5-(CF 3 ) 2 C 6 H 3 ) 3 (15 mol %) and Na 2 SO 4 (20 mol %). The tube was fitted with a rubber septum and purged with argon. Aminocyclopropane substrate (100 mol %) in argon sparged anhydrous PhCN (0.07 M) was added via syringe. The reaction mixture was sparged with CO for ca. 2 min, then heated at the specified temperature (120e140 C as noted) under a CO atmosphere (1 atm) until complete consumption of starting material was observed by thin layer chromatography (36e192 h as noted). The mixture was cooled to rt and purified directly by flash column chromatography, under the conditions noted, to afford the target cyclohexenone.

General procedure for (3D1D2) carbonylative cycloadditions of aminocyclopropanes and alkynes using a cationic Rh(I)-catalyst system
An oven-dried reaction tube, fitted with a magnetic stirrer, was charged with [Rh(cod) 2 ]OTf (7.5 mol %) and P(3,5-(CF 3 ) 2 C 6 H 3 ) 3 (15 mol %). The tube was fitted with a rubber septum and purged with argon. Aminocyclopropane substrate (100 mol %) in argon sparged anhydrous PhCN (0.07 M) was added via syringe. The reaction mixture was sparged with CO for ca. 2 min, then heated at the specified temperature (120e130 C as noted) under a CO atmosphere (1 atm) until complete consumption of starting material was observed by thin layer chromatography (3e48 h as noted). The mixture was cooled to rt, concentrated in vacuo and purified by flash column chromatography, under the conditions noted, to afford the target cyclohexenone.

Experimental procedures and data for new compounds
4.4.1. N-Benzyl-N-cyclopropylbenzamide (6b). To a solution of Ncyclopropylbenzamide 29 (1.00 g, 6.20 mmol) in anhydrous THF (12.4 mL) was added NaH (1.24 g, 31.0 mmol). The suspension was stirred at rt for 1 h and then benzyl bromide (3.7 mL, 31.0 mmol) was added dropwise over 5 min. The reaction mixture was stirred at rt for 16 h. The solution was cooled to 0 C and then water (20 mL) and Et 2 O (20 mL) were added. The layers were separated and the aqueous portion was further extracted with Et 2 O (3Â20 mL). The organic extracts were combined, washed with brine (20 mL), dried over Na 2 SO 4 and concentrated in vacuo. The residue was purified by flash column chromatography (20% EtOAc/hexane) to afford the title compound 6b