Formation of the indoline ring

Successful access to 30 provided the substrate for the 5-exo-trig radical cyclization which would form the indoline ring, and afford the benzyl protected seco form of the di-Boc- protected duocarmycin alkylation subunit 10.

The original chemistry pioneered by the Patel and Boger groups makes use of Bu3SnH to propagate the radical reaction.^{90, 91} Here it was decided to follow the Tietze adaption which employs TTMSS in place of the tin species, as it is reportedly easier to remove during purification.⁹⁶ The Tietze group conducts the reaction in benzene. Here it was decided to substitute benzene for the less toxic toluene, as this was expected to be easier to handle during the scale up of the synthesis.

This step proceeded without issue, using 25 mol % of AIBN, and 1.1 equiv. of TTMSS in toluene at 90 ^oC. The presence of reactive O₂ is known to be potential deleterious to the success of radical reactions, therefore this step was performed under an atmosphere of N2 and the solution degassed prior to heating. The reaction was carried out at a low concentration of 0.03 M. This is to favour the intramolecular cyclisation, and lessen the likelihood of any intermolecular side reactions. The product was easily isolated by subjection of the reaction mixture directly to flash chromatography, in excellent yield (90

%).

Scheme 2.30 Formation of the indoline ring.

83 Scheme 2.31 Likely mechanism of the 5-exo-trig radical cyclisation.

As discussed the reaction is 5-exo-trig radical cyclisation, and is most likely to proceed via the mechanism depicted in scheme 2.31. Here AIBN serves as a radical initiator. Upon heating, the AIBN decomposes via the concerted homolytic cleavage of the two carbon nitrogen bonds of the azo group. This is driven by the thermodynamically favourable release of diatomic nitrogen gas. The process also forms two isobutyronitrile radicals.

These radicals initially existed as a cage pair. This is to say they are solvated as one species via a ‘cage’ of surrounding solvent molecules. Some of the caged radical pairs couple together to form tetramethylsuccinonitrile. Others escape the cage by rapid diffusion and become available to initiate the reaction.

Although it is of course possible for the isobutyronitrile radicals to react directly with the substrate, the main pathway involves the propagation of the reaction via the stoichiometric excess of TTMSS. Here, the isobutyronitrile radical abstracts a hydrogen from the relatively weak hydrogen silicone bond of the TTMSS, producing the silane radical.

Subsequently the silane radical attacks the weak iodine carbon bond of 30, abstracting the halogen and providing the aryl radical. Formation of the aryl radical provides the starting point of the intramolecular cyclisation, and the subsequent attack of this species

84 at the alkene forms the indoline ring. This process is likely augmented by the donating effect of the para ether group, which increases the electron density at the radical centre.

The resulting alkyl radical is presumably reduced by either a molecule of unreacted TTMSS, or the isobutyronitrile formed by the generation of the silane radical. Both routes afford the desired product, and propagate the reaction by producing more radicals.

Various termination steps are possible, via the coupling of two radicals.

The reaction is selective for the 5-exo-trig cyclisation. No evidence of the completing 6-endo-trig reaction was observed. This is consistent with Baldwin’s rules.¹²⁴ Although, the more substituted radical product of the 6-endo-trig cyclisation is often more stable, and hence thermodynamically favoured; the dominance of the 5-exo-trig pathway is typically rationalised as being more kinetically favourable, due to better overlap of the molecular orbitals in the transition state.¹²⁵

In fact, in the case of this substrate, the 5-exo-trig pathway might be both kinetically, and thermodynamically favoured. This is because of the chlorine atom geminal to the radical centre of the cyclisation product. Although this group is electronegative, and thus might be seen to disfavour the electron deficient radical, it is important to remember that the alkyl radical is best described as being sp² hybridised. It is therefore conceivable, that the singularly occupied p orbital of the radical could overlap with a lone pair containing p orbital of the chloro group, and thus allow donation of electron density to the radical. This may have a net stabilising effect which is greater than the inductive stabilisation of the radical formed by the 6-endo-trig pathway.

As stated the substrate for this reaction is a mixture of E and Z isomers. This has no effect on the product as the carbon bonding to the chlorine becomes sp³ hybridised and achiral in the product. Chirality is however introduced at C-8 carbon (see scheme 2.30 for numbering). The reaction is racemic, as the aryl radical can attack either face of the alkene.

Introduction of the chiral centre produces diastereotopic effects in the ¹H NMR. The geminal hydrogens of the C-1 and C-9 carbons, are non-equivalent. This leads to complex second order effects in the splitting patterns of the five hydrogens comprising the substituted indoline ring. These appear as triplet integrating for one hydrogen at 4.13 ppm, neighbouring a complex multiplet at 4.06-3.89 ppm integrating for the remaining 4 hydrogens.

A characterised correlation pattern is observed when 10 is analysed by DEPT-edited HSQC experiment. The triplet at 4.13 ppm is correlated to a CH2 carbon at 52.3 ppm. This

85 carbon is also correlated with part of the neighbouring multiplet, approximately the 3.91 ppm region. These signals most likely correspond to the geminal diastereotopic pair of hydrogens on the C-9 carbon. This assumption is made based on the downfield position of the first hydrogen signal, as this pair is closest to the electronegative chlorine. Similarly the 4.07 ppm, and 3.92 ppm regions of the multiplet, are both correlated with the same CH₂ carbon at 47.6 ppm. This represents the second geminal diasterotopic pair of the C-1 carbon. Finally the 3.99 ppm region of the multiplet is correlated to a CH carbon at 40.7 ppm, corresponding to the hydrogen of the C-8 carbon of the chiral centre (see figure 2.5).

Partially overlapping with the upfield edge of the indoline multiple is a strong singlet integrating for the 3 hydrogens of the methyl ester, at 3.87 ppm. This correlated with a CH3 carbon at 52.2 ppm.

The two Boc groups are now again seen as two singlets at 1.48 ppm, and 1.39 ppm. A singlet at 5.27 ppm corresponds to the two alkyl hydrogens of the benzyl group. The aromatic benzyl hydrogens signals are overlapping with one of the indole hydrogens to form a complex multiplet 7.47-7.29 ppm, integrating for 6 hydrogens. This is again confirmed by the HSQC experiment. The second indole hydrogen is observed as broad singlet at 7.69 ppm. Again this signal was confirmed to correlate with a CH carbon at 97.4 ppm.

Figure 2.5 Expansion of the indoline region of the DEPT-edited HSQC of 10 at 298 K.DEPT phasing: Blue = CH or CH3 carbon. Red = CH2.

86 Scheme 2.32 Ester hydrolysis.

Scheme 2.33 Mechanism of methyl ester hydrolysis with LiOH.