Synthesis and kinetic resolution of substituted tetrahydroquinolines by lithiation then electrophilic quench

Treatment of N-Boc-2-aryl-1,2,3,4-tetrahydroquinolines with n-butyllithium in THF at –78 °C resulted in efficient lithiation at the 2-position and the organolithiums were trapped with a variety of electrophiles to give substituted products.


Introduction
The tetrahydroquinoline ring system is of great importance in natural products and medicinal compounds. 1 Substituted 1,2,3,4-tetrahydroquinolines are present in alkaloids such as angustureine, cuspareine, galipinine, martinellic acid, virantmycin, many with (for example) antiviral, antibacterial, antifungal, antimalarial, or antitumour activities. 1 The majority of syntheses of tetrahydroquinolines involve the reduction of quinolines or dihydroquinolines, 2 a Povarov type reaction, 3 or a cyclization process to make one of the bonds in the partially saturated ring. 4 This oen leads to tetrahydroquinolines that are monosubstituted at positions in the partially saturated ring, for example at C-2, rather than 2,2-disubstituted products. The ability to prepare 2,2-disubstituted tetrahydroquinolines would be attractive, opening up a greater diversity of products and exploring more chemical space. However, there are few examples of their enantioselective preparation from achiral compounds. You and co-workers reported an isolated example of an asymmetric reduction of a substituted quinoline and intramolecular trapping with an indole (Scheme 1a). 5 Zhao, Shi and co-workers developed an asymmetric Povarov reaction (Scheme 1b). 6 Hopkins and Wolfe reported a palladium catalyzed carboamination to give 2,2-dialkyl tetrahydroquinolines (Scheme 1c). 4d Enantioselective intramolecular N-arylation by Cai and co-workers was found with copper catalysis (Scheme 1d). 7 Here we report the use of deprotonation followed by electrophile trapping as a convenient approach to 2,2-disubstituted tetrahydroquinolines (Scheme 1e). This strategy has found only limited use in a racemic sense with 2-cyanotetrahydroquinolines. 8 Our research group has been studying the lithiation then electrophilic quench of N-Boc heterocycles, with a particular recent focus on piperidines, 9 and tetrahydroisoquinolines. 10, 11 We show in this study that we can extend our lithiation chemistry with N-Boc-2-aryl-heterocycles to tetrahydroquinolines. These substrates have been found to undergo highly enantioselective kinetic resolution. 12, 13 The organolithium is congurationally stable at À78 C and can be trapped to give 2,2-disubstituted tetrahydroquinolines with excellent enantioselectivities.

Results and discussion
We needed access to N-Boc-2-aryltetrahydroquinolines to test our lithiation chemistry. These were prepared from the quinolines 1a-e by reduction with sodium cyanoborohydride in acetic acid using a known method, 14 followed by Boc protection of the amine to give the novel products 2a-e (Scheme 2).
We were aware from earlier studies that the rate of lithiation is dependent on the orientation of the Boc group. 9d Therefore we probed the rate of rotation of the Boc group in compound 2a by using VT-NMR spectroscopy (see ESI †) and found activation parameters for Boc rotation, giving DG ‡ z 47 kJ mol À1 at À78 C for each rotamer. This suggests that the Boc group rotates quickly (t 1/2 z 1 s) even at À78 C. This was conrmed by ReactIR spectroscopy, which showed rapid lithiation (within a few minutes) at this temperature ( Fig. 1 and ESI †).
The intermediate organolithium could be quenched with a selection of electrophiles to give the 2,2-disubstituted tetrahydroquinoline products 3a-i (Scheme 3). Generally good yields of the products were obtained. The only exception to this was the use of the electrophile methyl cyanoformate, which gave the product 4 rather than the expected product 3a. This is discussed further below.
Kinetic resolution of the tetrahydroquinoline 2a was studied using (À)-sparteine as the chiral ligand in PhMe and adding n-BuLi to this mixture. 15 Moderately good enantiomer ratios of the recovered tetrahydroquinoline 2a and the product 3a were obtained by this method. However improved results were found by pre-mixing the n-BuLi and (À)-sparteine before adding the tetrahydroquinoline 2a (Scheme 4). The deprotonation was relatively slow under these conditions and despite being a kinetic resolution, it was best to use 1.2 equiv. n-BuLi to achieve a suitable rate of reaction. The recovered tetrahydroquinoline 2a was isolated with high enantiomer ratio (er 97 : 3) and this equates to a selectivity factor (k rel ) ¼ 20.  The opposite enantiomer of the recovered starting material 2a could be obtained by using (+)-sparteine in the kinetic resolution (Scheme 5). Several runs were conducted, all with very good enantioselectivities. Recrystallisation of the product 3a from run 1 gave material that was suitable for single crystal X-ray analysis. This conrmed the absolute conguration to be (R)-3a as indicated (see ESI †), and as expected based on previous ndings of the stereoselectivity preference for BuLi$sparteine in the deprotonation of N-Boc-piperidines. 15,16 The kinetic resolution was extended to the tetrahydroquinolines 2c-e (Scheme 6). High enantiomer ratios of the recovered tetrahydroquinolines were obtained in each case, particularly if more than just a slight excess of sparteine was added to the reaction mixture. The quenched products could be separated from the recovered starting material to give the desired tetrahydroquinolines (S)-2c-e.
The enantioenriched 2-aryltetrahydroquinolines 2a and 2c-e were treated with n-BuLi in THF at À78 C and the resulting organolithiums were found to be congurationally stable. Aer electrophilic quench, the products 3a-b, 3e, 3g, and 8-10 were obtained with high enantiomer ratios (Scheme 7). A slight loss of enantioselectivity was noticeable on using iodomethane as the electrophile, possibly as this reacts more slowly allowing some racemization on warming prior to quench. 16b The tetrahydroquinoline 3g was recrystallised and its absolute conguration was conrmed by single crystal X-ray analysis. The major diastereomer of the oxazolidinones 10 was conrmed by X-ray analysis. For X-ray data, see ESI. † Another electrophile that we were interested in testing was a trialkylborane, as this could result in a borate intermediate that should be prone to rearrange. 17 However we found that, instead of quenching the organolithium, addition of triethylborane promoted Boc group migration to give the tetrahydroquinoline 11 (Scheme 8). 18 The absolute conguration of the product 11 was conrmed by single crystal X-ray analysis showing that the rearrangement occurred with retention of conguration starting with the highly enantioenriched tetrahydroquinoline 2a. We speculate that the borane coordinates to the carbonyl oxygen atom to effect the migration and this must be preferable to direct reaction of the organolithium on the boron atom. The same reaction occurred with BEt 3 as a catalyst (0.2 equiv. gave product 11, 59% yield), or even in the absence of any catalyst (42% yield of 11 on using BuLi then warming without any BEt 3 ). We were able to remove the Boc group from the nitrogen atom in the products by using triuoroacetic acid (TFA), for example from compound 3g to give the secondary amine 12 with only minimal loss of enantiopurity.
We were surprised to nd that the electrophile methyl cyanoformate gave the product 4 rather than the expected product 3a. We were concerned that there may have been competitive ortho lithiation, 19 although this would be contrary to the formation of the products 3a-i. We recently uncovered an example of such an unusual change in regioselectivity with a benzylic organolithium on changing the electrophile, 11c and wanted to probe this reactivity further. Treatment of the deuterated tetrahydroquinoline 3c under the same conditions (1.2 equiv. n-BuLi in THF) followed by addition of MeOCOCN returned recovered starting material 3c, indicating a large kinetic isotope effect. 20 This suggests that the benzylic proton in 2a is indeed removed and not the ortho proton. Forcing the deprotonation with 3 equiv. n-BuLi for 1 h before addition of MeOCOCN still gave predominantly recovered 3c but did give a small amount (8% yield) of the product 4 (no deuterium present). Treatment of the tetrahydroquinoline 2b with n-BuLi then MeOCOCl gave the expected 2,2-disubstituted product 13 (Scheme 9). However, with MeOCOCN as the electrophile, the ortho substituted product 14 was obtained in moderate yield as a mixture in which the major product had deuterium at the 2position. Hence the reaction must proceed by initial deprotonation alpha to the nitrogen atom. Most electrophiles react at C-2 to give the products 3a-i. With methyl cyanoformate, substitution occurs at the ortho position and then rearomatisation takes place (see ESI †). The transfer of the proton (or deuterium) is likely to occur non-selectively and indeed on using enantioenriched tetrahydroquinoline 2a (er 92 : 8, prepared as described in Scheme 5), the product 4 was formed with low selectivity (er 61 : 39).
Calculations (using DFT/B3LYP-GD3BJ/6-311G**; see Computational methods below) initially focused on Boc rotation of tetrahydroquinoline 2a, for which we found an activation Gibbs energy of 48.7 kJ mol À1 in fair agreement with the experimental value of about 44 kJ mol À1 at 298 K. Subsequent calculations focused on the structures of the intermediate organolithium. In particular, we studied the complexation of the intermediate lithiated species with MeOCOCl and MeO-COCN. This gave insight into the potential reason for the change in regioselectivity. The minimised structures had the lithium atom coordinated to the carbonyl oxygen atom and close to C-2 when coordinated to THF or MeOCOCl ( Fig. 2a and  b). On the other hand, there was clearly an h 3 co-ordination of the lithium atom when MeOCOCN was bound (Fig. 2c). An alternative explanation could be that released cyanide could affect the regiochemistry, however an experiment in which MeOCOCl was added to the organolithium aer addition of one equivalent of NaCN returned only the alpha-substituted product 3a. Thus, we surmise that a change in structure of the organolithium on complexing the different electrophiles (MeOCOCl or MeOCOCN) must be inuencing the regiochemistry on reaction with the electrophile, although the precise way that this happens will be subject to further study. 21 Scheme 8 Formation of secondary amine products.

Computational methods
All calculations were performed using density functional theory, employing the B3LYP 22 functional as implemented in the D.01 version of Gaussian 09. 23 Calculations included dispersion corrections using the GD3-BJ 24 method. All calculations used the 6-311G(d,p) 25 basis set. Solvent was included via the PCM method 26 as implemented in Gaussian with the default parameters for THF.
The starting positions of coordinated solvent and electrophile molecules were varied to obtain the lowest energy structures. Frequency calculations were performed on all optimized structures to conrm that these were true minima. One transition state calculation was performed, for which a single imaginary frequency was found, as expected. For the calculations on 2a no imaginary frequencies were found. The two complexes presented in Fig. 2a and b also showed no imaginary frequencies. The complex presented in Fig. 2c showed a single imaginary frequency of À16.5 cm À1 . Inspection shows this mode is largely a torsional mode of the ligands around the lithiumligand bond. Re-running the calculation with a ner integration grid and tighter optimization convergence led to a structure without imaginary frequencies. This latter structure is reported in the ESI. † All Gibbs energies reported were evaluated at 298.15 K and standard pressure. For the precise keywords used in each of the calculations see the ESI. †

Conclusions
In conclusion, we have developed a rapid access to 2,2-disubstituted tetrahydroquinolines from 2-aryltetrahydroquinolines by deprotonation with n-butyllithium followed by electrophilic quench. The reaction proceeds with retention of conguration on using enantiomerically enriched starting materials. Surprisingly, methyl cyanoformate reacted at the ortho position of the 2-aryl substituent and this change in regioselectivity on change in the electrophile is proposed, on the basis of DFT studies, to result from a small change in the structure of the organolithium intermediate. Excellent enantioselectivities in the kinetic resolution were obtained in the presence of the chiral ligand sparteine. This chemistry therefore provides a new method to prepare highly enantiomerically enriched 2-aryltetrahydroquinolines and tetrahydroquinolines that are fully substituted at the 2 position.

Conflicts of interest
There are no conicts of interest to declare.