Synthesis and antiproliferative activity of controls 42, and 43. 43

In order to assess the impact of a C-terminal carboxyl group in isolation from the effects of the amino acid residue, it was decided to synthesise control compounds 42, and 43; these being the seco N-acetyl alkylation subunit with the natural C-terminal ester, and its carboxyl counterpart respectively (scheme 3.11).

Compounds 42, and 43, were accessed via the same synthetic route used to synthesize 11, up until the common di-Boc protected intermediate 10. Exhaustive Boc-deprotection and acetylation with acetyl chloride gave the benzyl protected precursor to 42. Benzyl ether cleavage by catalytic transfer hydrogenation as before, either immediately, or following methyl ester hydrolysis afforded 42 and 43 respectively (scheme 3.11). The desired structures were confirmed by a combination of accurate mass spectrometry, ¹H NMR and DEPT-edited-HSQC experiment, and further characterised by ¹³C NMR.

Antiproliferative activity of the racemates was assessed by MTS assay as before in the HL-60 cell line. While the intact ester 42 returned a mean IC50 of 25 nM (95 % CI: 11-37 nM, triplicate), in line with its spirocyclized counterpart 41, and over 1000 fold more potent

Scheme 3.11 Synthesis and structure of 42 and 43.

148 than the amino acid analogues, the carboxylic acid 43, showed no activity at all even at the highest concentration tested (500 µM). This clearly demonstrated that the presence of the ionisable C-terminal carboxyl group was having a detrimental effect on cytotoxicity, most likely by inhibiting passive diffusion through the cell membrane.

Given the complete lack of activity shown by 43, it now seemed surprising that the amino acid-duocarmycin conjugates had shown any activity at all. Thus it would appear, in comparison to 43, the amino acid structures seemed to be offering some mitigation against the negative impact on cellular activity imposed by the free carboxylic acid. As discussed previously in relation to the variation in activity observed between the different amino acid analogues, it seems unlikely that potential differences in the pKa could account for the difference in activity between 43 and the amino acid-duocarmycin conjugates.

It is possible that passive diffusion of the unionised fraction of the amino acid analogues is faster than that of 43 due to the increased hydrocarbon scaffold provided by the amino acid. However, it is also intriguing to speculate that perhaps the amino acid analogues are able to benefit, at least partially, from active transport processes used by cells to control intracellular amino acid homeostasis. It is possible that recognition by such transport proteins, may be providing greater cell permeability when compared to 43. This might also offer insight into why the β-alanine analogue appeared considerably less active than the alanine analogue. For example, perhaps HL-60 cells do not possess the same capacity for the active transport of β-alanine, as they might for alanine. However this is entirely speculative, although some amino acid related drugs have been shown to utilise transporters whose endogenous function is to mediate the uptake of amino acids. For example the phenylalanine derived nitrogen mustard melphalan, has been shown to cross cell membranes via recognition of the LAT-1 amino acid transporter.¹⁸⁴

3.11 Synthesis and antiproliferative activity of the first extended amino acid-duocarmycin conjugate.

The complete inactivity of 43 in the MTS assay, clearly demonstrated that a C-terminal carboxyl group had the potential to abate entirely the cytotoxicity of the duocarmycin alkylation subunit. However, the surprising activity of the amino acid-duocarmycin conjugates, although still over 1000 fold lower than that of the ester 42, was encouraging.

Clearly the additional structure of the amino acid scaffold, was providing some mitigation against the detrimental effect of the carboxyl group. As discussed, this may in part be due to recognition by amino acid transporters, but could also represent faster passive diffusion due to the increased hydrocarbon structure. This appeared evident in the observation that

149 the phenylalanine analogue containing the most hydrophobic sidechain appeared to be the most active of the amino acid conjugates. It was therefore intriguing to question how much activity would be recovered by replacing N-terminal acetylation, with the coupling of an additional hydrophobic group.

The group chosen was 5-methoxyindole. Coupling of this group via the commercially available 5-methoxyindole-2-carboxylic acid, would provide an amino acid-duocarmycin conjugate which was more similar in structure to the full natural product, which possesses an N-terminal trimethoxyindole unit. Such an analogue would be expected to benefit from faster passive diffusion, and increased affinity for the minor groove.

It was decided to synthesis the 5-methoxyindole equivalent of the zwitterionic lysine analogue 39. This had been one of the least potent of the original analogues, and it was therefore particularly interesting to see how much activity could be recovered by N-terminal extension. Furthermore, it was still anticipated that if cell permeability was improved by the additional indole subunit, that the cationic side chain of lysine might provide further favourable electrostatic interactions with the polyanionic backbone of DNA.

As such, the solid phase methodology was employed to synthesize the extended amino acid-duocarmycin conjugate 44 (Scheme 3.12). As before, the synthesis was conducted on a 0.038 mmol scale, beginning from a commercially preloaded lysine 2-chlorotrityl resin. Synthesis proceeded as already described, with the replacement of acetylation of the indoline nitrogen with the coupling of 5-methoxyindole-2-carboxylic acid, using HATU.

Cleavage was affected using 10 % TFA, 10 % TIPS, in DCM for 2 hours, to give one product.

Benzyl deprotection was affected as previously described by catalytic transfer hydrogenation. However, it is noteworthy to highlight that this was not as straightforward as with previous analogues. During initial attempts it was found that the extended lysine analogue was unexpectedly susceptible to completing dehalogenation. The particular susceptibility of the lysine extended analogue to this side reaction is not known. It was observed that benzyl deprotection was much slower than with previous analogues, and the later treatment with additional equivalents of Pd/C and ammonium formate to drive the reaction to completion, promoted the initiation of the side reaction. Furthermore, performing the reaction with lower quantities of both the catalyst and hydrogen donor resulted in no reaction. It was questioned whether the amine side chain could be poisoning the catalyst, or promoting dehalogenation. However, attempts to perform benzyl deprotection prior to removal of Boc-protection of the sidechain made no improvement.

This was achieved by cleavage of the Boc-protected product, using 1 % TFA. In the end

150 limiting the reaction time, and separating the desired product from unreacted starting material by reverse phase HPLC, gave the best recovery (4.4 mg, 20 % yield). The desired structure was confirmed by a combination of accurate mass spectrometry, ¹H NMR and DEPT-edited-HSQC experiment.

Antiproliferative activity was assessed by MTS assay as before in the HL-60 cell line, returning a mean IC₅₀ value of 374 nM (95 % CI: 200-701 nM, triplicate). Therefore, addition of the methoxyindole subunit had increased the potency of the extended lysine analogue 44 by over 1000 fold, compared to the truncated counterpart 39, the IC₅₀ of which was estimated to fall between 300 μM and 500 µM, but could not be calculated due to lack of activity at the available concentrations returning an incomplete dose response curve.

Clearly the methoxyindole unit was very effective at limiting the detrimental impact of the zwitterionic C-terminal lysine residue. This is consistent with expected faster passive diffusion of unionised fractions promoted by the increased hydrophobic structure.

However, improved activity is also likely to have resulted from more efficient DNA alkylation. This being a result of both improved non-covalent affinity for the minor groove,

Scheme 3.12 Synthesis of the extended lysine analogue 44.

151 and more effective activation of the cyclopropane caused by the increased ridged length of the compound (see chapter one).

Despite the dramatic improvement, 44, was still over 14 times less active than the truncated alkylation subunit possessing the C-terminal ester, 42. Furthermore, it was over 1000 fold least active then the full natural product, to which it bears closer structural homology, with duocarmycin SA reportedly returning IC50 values of as low as 0.006 nM (murine L1210).¹⁸⁵

It was therefore clear that the C-terminal amino acid was still having a detrimental effect on cytotoxicity. This may still be through limited cell permeability, but might also be the result of disruption to minor groove binding, or perhaps more likely a combination of both factors.