Highly Regioselective Synthesis of Substituted Isoindolinones via Ruthenium-Catalyzed Alkyne Cyclotrimerizations

(Cyclooctadiene)(pentamethylcyclopentadiene)ruthenium chloride [Cp*RuCl(cod)] has been used to catalyze the regioselective cyclization of amide-tethered diynes with monosubstituted alkynes to give polysubstituted isoindolinones. Notably, the presence of a trimethylsilyl group on the diyne generally led to complete control over the regioselectivity of the alkyne cyclotrimerization. The cyclization reaction worked well in a sustainable non-chlorinated solvent and was tolerant of moisture. The optimized conditions were effective with a diverse range of alkynes and diynes. The 7-silylisoindolinone products could be halogenated, protodesilylated or ring opened to access a range of usefully functionalized products.


Introduction
Substituted isoindolinones have recently generated considerable interest because of their diverse biological activities, including the inhibition of angiogenesis, [1] tumour necrosis factor production, [2] MDM2-p53 protein-protein interactions, [3] hypoxia-inducible factor-1a [4] and histone deacetylase. [5] The majority of existing protocols for isoindolinone synthesis require the construction of a g-lactam adjacent to a preformed aromatic core. [6] Recent examples include the one-pot transformation of 2-halobenzaldimines into chiral 3-substituted isoindolinones and the Ni-mediated cyclization of N-benzoyl aminals in the presence of a stoichiometric Lewis acid. [7,8] However, the inevitable limitation of these approaches is the accessibility of the arene starting material itself. The synthesis of polysubstituted arenes is often non-trivial, frequently requiring numerous steps, the use of protecting group strategies and/or functional group interconversions.
The transition metal-catalyzed [2+2+2] cyclotrimerization of alkynes is emerging as an elegant, atom efficient and convergent approach to the synthesis of highly substituted arenes. [9] The strategy allows for the regioselective synthesis of compounds that would be extremely difficult to make via traditional aromatic chemistry. The regioselectivity of a cyclotrimerization is normally controlled by tethering two or three of the alkyne components together, so this strategy is best suited to the synthesis of bicyclic and tricyclic ring systems. This allows for the assembly of substituted multiple-ring aromatic compounds from alkyne precursors in a single step.
Yamamoto and co-workers have previously recognized the potential of alkyne cyclotrimerizations for the synthesis of isoindolinones bearing substituents on the aromatic ring. [10] They reported the cyclization of amide-tethered diynes 1 with monoynes 2 using Cp*RuClA C H T U N G T R E N N U N G (cod) 3 as the catalyst to give regioisomeric isoindolinones 4 and 5 (Scheme 1). In general the regioselectivity of the cyclotrimerization was poor to moderate, with the exception of a single example bearing a methyl group at R 1 . In addition, a significant limitation of this method is the use of 1,2-dichloroethane (DCE) as solvent, a substance which is potentially detrimental to human health and is generally avoided within industry. [11] The aim of this study was to explore the regioselective synthesis of polysubstituted isoindolinones using more industrially viable reaction conditions, to establish the general applicability of the reaction, and to develop the synthetic potential of the cyclized products. On the basis of previously reported cyclizations we envisaged that the introduction of a trimethylsilyl group at R 1 in diyne 1 would direct the regioselectivity of the cyclisation reaction effectively with a broad range of monoynes. [10,12] The arylsilane unit present in the isoindolinone product could then be transformed using standard chemical techniques to access a variety of 7-substituted derivatives.

Diyne Synthesis
Initially several amide-tethered diynes 6 were prepared by the coupling of propargylic amines 7 with 3-(trimethylsilyl)propiolic acid 8, via the corresponding acid chloride (Scheme 2). [13] Where necessary the corresponding amines were prepared using literature procedures. [14][15] Optimization Various conditions were screened for the cyclotrimerization of diyne 6a with 1-hexyne 9a to form isoindolinone 10a, and the results are summarized in Table 1. All reactions were conducted for 16 h at which point Scheme 2. Synthesis of diynes 6a-e. Scheme 1. Isoindolinone synthesis as reported by Yamamoto and co-workers. [10]  [b] Conversion of 6a into 10a and 11 (determined by crude 1 H NMR without the use of an internal standard).
[c] Solvent dried over activated 4 molecular sieves and degassed.
[d] Cp*RuClA C H T U N G T R E N N U N G (cod) 3 was added to the reaction mixture at 0 8C, which was then allowed to reach room temperature.
[e] Diyne 6a in CPME was added dropwise over 3 h to a stirring solution of 9a and 3 in CPME.
conversion and selectivity were determined by analysis of the crude 1 H NMR spectrum. The cyclotrimerization of diyne 6a and alkyne 9a was examined using four different literature procedures. Neither RhClA C H T U N G T R E N N U N G (PPh 3 ) 3 nor Co 2 (CO) 8 were effective in catalyzing the alkyne cyclotrimerization, with no measurable conversion of diyne 6a (entries 1 and 2). [16] Treating diyne 6a with 5 mol% Grubbs first generation catalyst and 4 equivalents of 1-hexyne 9a in dried, degassed CH 2 Cl 2 resulted in formation of the target isoindolinone 10a with only 5% conversion (entry 3). [17] Treating diyne 6a with 1-hexyne 9a and 1 mol% Cp*RuClA C H T U N G T R E N N U N G (cod) in dried, degassed DCE also gave isoindolinone 10a, again with 5% conversion of 6a (entry 4). [10] Given that the latter procedure gave a similar conversion with a lower catalyst loading, Cp*RuClA C H T U N G T R E N N U N G (cod) was selected for subsequent optimization.
Interestingly, treating diyne 6a with 1-hexyne 9a and 1 mol% Cp*RuClA C H T U N G T R E N N U N G (cod) with no solvent (neat) at 0 8C gave isoindolinone 10a with a 50% conversion (entry 5). This suggests that using DCE as a solvent for this reaction is actually detrimental. In addition to the desired isoindolinone 10a, dimer 11 was also formed as a significant by-product. [12] Crucially, regioisomeric cyclotrimerization product 12 was not observed at all in the crude 1 H NMR spectrum. The reaction under neat conditions reached completion within 16 h when 3 mol% of catalyst 3 was used, and with a significant reduction in the proportion of homo-coupled product 11 produced (entry 6).
We were interested in using cyclopentyl methyl ether (CPME) as a solvent for this cyclization as it has been recently established as a safer and more environmentally benign alternative to many traditional organic solvents. [18] As shown in entry 7, when the reaction was conducted in CPME with 3 mol% of catalyst 3, diyne 6a was completely consumed within 16 h and an improved selectivity for the cross-coupled product 10a was observed. By comparison, the same reaction using only 1 mol% catalyst resulted in a comparable level of selectivity, but a lower conversion (entry 8). Reducing the number of equivalents of 1hexyne 9a to two resulted in the complete consumption of diyne 6a but also a significantly increased level of homo-coupling.
In an attempt to minimise the formation of dimer 11, diyne 6a was added dropwise over 3 h to a stirring solution of monoyne 9a and catalyst 3, [19] and this proved to be highly effective (entry 10). When using the 3-hour dropwise addition it was possible to reduce the number of equivalents of 1-hexyne 9a from four to two with no increase in homo-coupling (entry 11). A further reduction to 1.1 equivalents of 1-hexyne 9a did result in increased homo-coupling, but target isoindolinone 10a was still the major product (entry 12).
The cyclization of 6a and 9a was also effective when 2-MeTHF or MTBE were used as solvents, but in both cases a greater degree of homo-coupling of 6a was observed than with CPME (entries 13 and 14). The reaction proved to be relatively water tolerant, with a significant conversion and a reasonable selectivity observed when the reaction was conducted in the presence of 10% water (entry 15). Cyclization was even observed when the reaction was conducted in water as solvent (entry 16). This is important as it could enable the extension of the reaction to aqueous conditions for reactions of water-soluble substrates.
Following the optimization study the conditions described in entry 11 were taken as the "optimized" cy-A C H T U N G T R E N N U N G clization conditions as they required a reduced excess of monoyne and minimized the formation of dimer 11. Crucially this protocol did not require the CPME solvent to be either degassed or dried. This, together with the environmental benefits of CPME, makes this reaction a very practical method for the synthesis of isoindolinones. Dimer 11 could be readily separated from the desired product by flash column chromatography, and the optimized conditions described in entry 11 gave the target isoindolinone 10a in 81% isolated yield ( Table 2, entry 1). This reaction was also scaled up to a 500-mg scale and isoindolinone 10a was isolated in 66% yield (428 mg product).

Monoyne Scope
The cyclization of 6a was then examined with a variety of monoynes using the optimized conditions described above to determine how robust the reaction was for a range of different substrates. Diyne 6a cyclized with a wide range of monynes 9 as detailed in Table 2. Crucially, no evidence for the formation of regioisomeric isoindolinones was observed in any of the cyclization reactions. Alkyl monoynes 9a-e cyclized efficiently with 6a to give the corresponding isoindolinones 10a-e in good isolated yield (entries 1-5, 66-83%). Little formation of the undesired dimer 11 was observed, except in the reaction of tert-butylacetylene 9b, presumably due to high steric crowding about the monoalkyne. Carbamate 9f cyclized with 6a to give 10f in reasonable yield and with modest levels of homo-coupling (entry 6).
Ether 9g and acetal 9h both underwent cyclotrimerization with 6a, but with the formation of significant quantities of dimer 11. Propargylic alcohol 9i and methoxyacetylene 9j both failed to cyclize with diyne 6a, with only starting material being recovered in both cases. In addition to aliphatic monoynes, diyne 6a cyclized effectively with a broad range of aromatic monoynes. Electron-rich (entries 12, 13, 17 and 18), electron-poor (entry 16) and sterically hindered substrates (entries 12 and 14) could all be tolerated and products were isolated in good yields (79-93%) with low levels of diyne homo-coupling. For most of these examples longer reaction times (up to 24 h), and in some cases higher catalyst loadings, were required to drive the reaction to completion. However the reactions with ortho-substituted arylacetylenes 9l and 9n reached  12 and 14). Monoyne 9l also cyclized with exceptionally high selectivity for the cross-coupled product 10l over dimer 11, whereas ortho-bromo alkyne 9n gave a slightly lower selectivity. Although Yamamoto et al. have reported the [2+2+2] cycloaddition of an electron-deficient nitrile and an amide-tethered diyne to give a pyridine, [20] in our reaction nitrile 9s failed to cyclize with 6a to form any product via reaction of either the alkyne or the nitrile (entry 19). Only a limited quantity of 11 (~10%) was formed in this reaction suggesting that 9s may inhibit the catalyst. Heterocycle-containing alkyne 9t cyclized effectively with 6a to give the corresponding 2-pyridyl derivative 10t in a moderate 50% yield (entry 20). In contrast N-methylimidazole 9u failed to cyclize with 6a, with unreacted starting material being recovered (entry 21). Alkyne 9v cyclized with 6a to give boramide 10v in reasonable yield (entry 22). [21] Diyne Scope The cyclization of amide-tethered diynes bearing different N-substituents was examined and the results are summarized in Table 3. N-t-Bu diyne 6b proved to be an excellent substrate for the synthesis of 5,7-substituted isoindolinones. Treatment of 6b with 1hexyne 9a under the optimized reaction conditions gave isoindolinone 13a in 84% yield with little formation of the dimer 14a (entry 1). The cyclization of 6b with 9k required 4 mol% 3 and 24 h to reach completion, giving isoindolinone 13b in 89 % yield (entry 2). The reaction of 6b with 2-ethynyltoluene 9l proceeded in 94% yield without an elevated reaction time or an increased loading of catalyst 3, and also occurred with very little formation of dimer 14a (entry 3).
The N-H diyne 6c proved less effective for the synthesis of isoindolinones, with the cyclization of 6c and 1-hexyne 9a requiring 10 mol% Cp*RuClA C H T U N G T R E N N U N G (cod) 3 and 24 h to achieve a 90% conversion of diyne 6c (entry 4). Isoindolinone 13d was only formed in modest yield (51%) and significant formation of dimer 14b was observed. Under the same conditions the cyclization of 2-tolylacetylene 9l and N-H diyne 6c gave the desired isoindolinone 13e in a slightly higher yield with 90% conversion. Again, the reaction with 2-ethynyltoluene 9l proved to be unusually selective, with 13e and 14b formed in the ratio 7:1 (entry 5). The lack of a sterically bulky N-substituent is presumably responsible for both the reduced reactivity of N-H diyne 6c with monoynes and the high level of diyne homo-coupling observed in these reactions.
The cyclization of amide-tethered diynes bearing different alkyne substituents was also explored ( Table 4). With doubly substituted diynes 6d and 6e, no homo-coupling of the diyne was observed and dropwise addition of the diyne to the reaction was unnecessary (entries 1-3). With 10 mol% of Cp*RuClA C H T U N G T R E N N U N G (cod), methyl-substituted diyne 6d cyclized with 1-hexyne 9a to form a 9:1 mixture of regioisomeric isoindolinones 15a and 16a (entry 1).
The dependence of the cyclotrimerization on an SiMe 3 regiodirecting group was also investigated. Diyne 6f with a terminal methyl substituent reacted with 1-hexyne 9a under the optimized cyclization conditions to give isoindolinone 15d in 85% yield (entry 4). Crucially, there was no trace of the regioisomeric isoindolinone 16d by crude 1 H NMR. Similarly, diyne 6f cyclized with 2-ethynyl toluene 9l to give isoindolinone 15e in 94% yield, with no evidence for the formation of regioisomer 16e (entry 5).

Functional Group Manipulation of Cyclized Products
Conversion of the cyclized isoindolinone products into a number of synthetically interesting motifs was examined. Isoindolinone 10a was converted to aryl halides 17 and 18, in 79% and 90% yields, respectively, via an ipso substitution of the silyl group (Scheme 3). [22] Treatment of N-t-butylisoindolinone 13a with triflic acid resulted in a simultaneous deprotection of the lactam and protodesilylation within 30 min to give N-H isoindolinone 20 in good yield. [23] Alternatively, treatment of 13a with iodine monochloride followed by deprotection with triflic acid gave 7-iodoisoindolinone 19 in 83% yield. Thus, an Nt-Bu diyne can be used as an indirect method for the synthesis of N-H isoindolinones via this acid-mediated deprotection.
It was also possible to access a tetrasubstituted monocyclic benzene. Treatment of N-H isoindolinone 19 with di-tert-butyl dicarbonate gave N-Boc isoindolinone 21, which could be reduced with lithium borohydride to form N-Boc protected amino alcohol 22, [a] Reaction conditions: A solution of 6 in CPME was added to a stirring solution of 9 and 3 in CPME over 1 min at room temperature. [b] Isolated yield. [c] Determined by the analysis of crude 1 H NMR spectra. [d] Diyne 6f in CPME was added dropwise over 3 h to a solution of 9 and 3 in CPME.
[e] Evidence of limited homo-coupling of 6f was observed in the crude 1 H NMR spectrum. together with cyclic aminol 23, in a combined yield of 78% (Scheme 4). The preparation of mono-cyclic substituted arenes via tethered alkyne cyclotrimerizations has little precedent and such systems are somewhat difficult to access via traditional aromatic substitution reactions, highlighting the value of this strategy. [24]

Conclusions
In summary, we have demonstrated the regioselective synthesis of polysubstituted isoindolinones via the Cp*RuClA C H T U N G T R E N N U N G (cod)-catalyzed cyclotrimerization of amidetethered diynes and monoynes. This cyclization is effective with a wide range of structurally diverse monoynes and was demonstrated to work with a variety of different diynes. We have also demonstrated that the cyclization products could be converted into a range of functionalized isoindolinones and a tetrasubstituted benzene derivative.

Experimental Section
Full experimental details are provided in the Supporting Information.

Cp*RuClA C H T U N G T R E N N U N G (cod)-Catalyzed Cyclization of a Diyne and a Monoyne
A solution of 6a (500 mg, 1.86 mmol) in CPME (11 mL) was added dropwise over 3 h to a stirring solution of 1-hexyne 9a (0.43 mL, 300 mg, 3.7 mmol) and Cp*RuClA C H T U N G T R E N N U N G (cod) (21 mg, 3 mol%) in CPME (7.7 mL) at room temperature. The reaction mixture was stirred for a further 13 h before being filtered through a silica pad, eluting with ethyl acetate. The solvent was removed under vacuum to give the crude product, which was purified by flash column chromatography (13:1 petrol:ethyl acetate) to give 2-benzyl-5-butyl-7-(trimethylsilyl)isoindolin-1-one 10a; yield: 428 mg (1.