Proposed mechanism for conversion o f a hydroxy group to the chloride using the PhsP-CCLt complex [156].

A major problem with the use of PhsP-CCU complex is formation of the by­ product Ph3P=0 , which is particularly difficult to isolate from the product. However,

it was found with previously synthesised chloroethylaminoanthraquinones that isolation o f the desired products could be carried out by first making a hydrogen chloride salt o f the CAQ. This salt was then dissolved in a warm CH3OH/CH2CI2

mixture and stirred at 60 °C. A mixture of EtOAc-EtOH (1:1) was added dropwise until precipitation o f the CAQ-product was observed. Excess Ph3P and Ph3P= 0

remained in the EtOAc-EtOH solvent mixture, and isolation o f the desired product was possible. All the CAQs developed by this method were afforded in 70-85 % yields [33]. This method was also successfully applied to the conversion of several HAQs developed in this study. Purification by column chromatography is highly problematical since the CAQs and Ph3P= 0 elute firom the column together regardless

o f the solvent system applied. This may be attributed to the phosphoryl oxygen of Ph3P=0 , which is a good proton acceptor, and therefore is likely to form complexes

with a wide variety of organic compounds which act as proton donors [157].

As an alternative solution to remove Ph3P= 0 and thereby making purification

easier, polymer supported Ph3P was investigated as a reagent [158]. However, no

reaction took place, which may have been due to the steric bulkiness o f the polymer support or interaction of the support itself with the chromophore. The latter must be considered following the effective use of polymer supported Ph3P in the conversion

o f hydroxylated Boc-protected sidechains (configuring the same hydroxyl groups as on the aminoanthraquinones). This suggests that the polymer and the chromophore may interact in a way that prevents complex formation and conversion.

Solvents such as CHCI3, DMF and toluene at elevated temperatures were also investigated in the Ph3P-CCl4 reaction, but no improvement was observed.

l,2-bis(diphenylphosphino)ethane (diphos) has been reported as a good reagent

that produces a phosphine oxide by-product that is readily removed from the reaction mixture by filtration [159]. Unfortunately, however, the HAQs investigated were not converted to the desired CAQs. In addition, several previously undetected by­ products were observed by TLC monitoring using diphos as reagent.

Conversion o f the CAQ to a hydrochloride salt is only useful if the reaction has gone to completion. It is otherwise impossible to isolate the hydroxylated aminoanthraquinone from the chloride. HAQs containing secondary alcohols were found to be less susceptible to treatment with Ph3P-CCl4 complex, which is likely to

Chapter 2 61

be due to the steric constraints of the secondary alcohol as compared to the primary alcohol group. As a result, completion of reactions was not achieved.

It has been demonstrated that this type of reaction proceeds more rapidly in a more polar solvent such as acetonitrile [155]. However, the HAQs were not soluble in acetonitrile, and the use of this solvent was therefore not rewarding. Various CHzCb-acetonitrile mixtures were also attempted without any improvement in completion of the reaction. This led to the investigation o f other reagents suitable for conversion o f alcohols to halides. These attempts included thionyl chloride (SOCI2),

which has been used in the preparation of acridine mustards [90-92, 94], mesylation of the alcohol and subsequent treatment with NaCl [90, 92, 166] or tetra-«- butylammonium chloride [166]. Excess of SOCI2 resulted in decomposition of the

HAQ, and using equimolar SOCI2, with or without equimolar EtgN, in dry CH2CI2

led to precipitation o f the HAQ within a few minutes. Mesylation and conversion to chloride by treatment with a chloride salt (either NaCl or tetra-«-butylammonium chloride) led to a mixture o f products, none of which was identified as the desired CAQ.

A novel method for the development of CAQs was devised in order to obtain the desired target compounds. The synthetic route to these compounds involved 5 steps: (1) Boc-protection of the hydroxylated amino sidechain, (2) mesylation of the hydroxyl group, (3) conversion of the mesylate intermediate to the chloride by treatment with tetra-«-butylammonium chloride, (4) deprotection o f the Boc group, and (5) substitution onto the l-(dimethylamino)ethylamino-4-fluoro-5,8- dihydroxyanthraquinone chromophore (for synthetic route, see Scheme 2.11).

Despite the multi-step nature to prepare CAQs by this method, it proved an effective means by which to obtain these novel target compounds, which were important for the structure-activity studies in Chapters 3-5.

2.2.4 Synthesis of Anthraquinone Based Di-A-oxides

Synthesis o f AQ4N has successfully been achieved by the use o f either w-CPBA [151] or oxa-aziriridine as the oxidising agent [150]. In the synthesis of the novel di- A-oxides, w-CPBA was employed in the oxidation of the tertiary amine functionalities, which afforded crude products in 61-90 % yield (Table 2.4). Based on the carbon content from micro-elemental analysis (CHN), the di-A-oxides were estimated to contain 3-7 % impurities. The crude di-A-oxides were not purified.

The reaction mechanism for the A^-oxidation may proceed through two pathways (Scheme 2.17). Pathway (ii) has been described for the reaction between a tertiary amine and persulphuric acid [162], but pathway (i) is more likely to occur in these reactions. Such a proposal is based on the work-up procedure of the products. After completion o f reaction, as monitored by TLC, the reaction mixture was diluted with hexane. After 3 h o f stirring at 4 °C, the precipitated solid was filtered and washed successively with ice-cold hexane, ether, CH2CI2 and EtOAc. The mass spectra of the

crude di-A-oxides showed no evidence of the presence o f w-CPBA. If pathway (ii) was dominating, the two products would behave like a salt and strongly adhere to each other. Consequently, it is unlikely that the benzoate by-product would have been washed away with the above-mentioned solvents. The di-A-oxides were also analysed by ^H-NMR and ^^C-NMR, but due to the impurities (likely to be mono-A- oxide derivatives), no conclusive assessment could be made o f these compounds.

NHR