A Modular Approach to Functionalised Dyes

Abstract A modular approach to the synthesis of sensors is described. In this approach a central dye scaffold, prepared from the SNAr reaction between a halo-substituted azo-dye and a disubstituted phenol, was decorated with a representative carbohydrate or macrocycle using Sharpless click chemistry. Regiochemical issues in the click reaction are also addressed.

The realisation that heavy metals are implicated in the aetiology 1a-d of a number of disease states require the development of analytical techniques capable of determining their biodistribution. 1e-g Recently Quayle et al. 2 developed a modular approach to the synthesis of pyridine-containing crown ethers which possess a range of additional hard/soft ligating centres embedded within the macrocyclic core. These macrocycles were prepared with a view to optimising their binding to biologically relevant 'soft' metal ions such as Cu 2+ , Hg 2+ , and Zn 2+ as part of a much broader programme of research concerned with the development of molecular probes able to map the distribution of heavy metals within various tissue types. 3 Our design strategy for the construction of such probes ( Figure 1) centred upon the identification of a suitable scaffold which could be decorated by the introduction a recognition unit (capable of selectively binding to a given metal), a reporting unit (e.g., a dye or fluorescing agent which would relay the complexation event), and a carrier domain which would aid active transport 1c,d of the whole assembly to a specific organ or tissue. Key to this approach was the application of 'click chemistry' 4 for the decoration of the central scaffold thereby enabling a flexible method for sensor assembly. 5

Figure 1 Metal-sensor design concept
Since its popularisation by Sharpless 4a the copper(II)-catalysed azide-alkyne cycloaddition (CuAAC) reaction has found many applications in the fabrication of sensors, 6 drug discovery, 7 bioconjugation, 8 polymer science, 9a and supramolecular/macromolecular chemistry. 9b-e Given the robustness and generality of this reaction we deemed it ideal for the functionalisation of a central scaffold as required in our sensor fabrication. Herein we present our initial findings concerning the attachment of carbohydrate and azo-dye functionalities to a central scaffold using the CuAAC reaction. 10 In our initial model studies we wished to establish whether the readily available azo-dyes 11 1 (R = Me, Et) could be transformed into the scaffold 3 12 via a displacement reaction using phenol 2 as nucleophile. Given the ex-

Letter Syn lett
tensive literature on azo-dye chemistry it was a little surprisingly therefore to find that there is little precedent for such displacement reactions. 13 Fortunately, we were able to show that Ullmann-type 14 coupling reactions between 1 and 2 were achieved under the auspices of copper catalysis, either in the presence of the preformed copper(I)-NHC complex [(IPr)CuCl] 15a,b or, more simplistically, by CuI in the presence of picolinic acid (Scheme 1, Table 1). 16 The synthesis of macrocycles 17 12 and 13, which possessed an alkyne moiety for the down-stream click reaction, was also readily accomplished using the aziridine ring-opening-alkylation-macrocyclisation sequence which we have recently developed. 2 Accordingly, either (R)-or (S)alaninol were converted into either 8 or 9 in yields of 80-90% (Scheme 2). Macrocyclisation of 8 or 9 with 11 was best achieved by reaction with 11 (which itself was easily prepared from diol 10 as shown in Scheme 3) under conditions of high dilution (ca. 3 mM).

Letter Syn lett
In these particular examples the course of the macrocylisation reaction appeared to be relatively insensitive to the counteranion and proceeded efficiently when either of Cs 2 CO 3 , K 2 CO 3 , or KOH was employed as the base (Scheme 4). 18 In the case of the reaction between 9 and 11 macrocyclization, leading to 13 (isolated in 84% yield), was also accompanied by the formation of a small amount of a dimeric species 14 (5%) (Scheme 4). In addition the synthesis of alkyne 16, the remaining key component of our sensor model, proceeded without complication using the method developed by Mereyala,19 and afforded the α-D-glycoside in essentially quantitative yield (Scheme 4).

Scheme 3 Synthesis of functionalized aromatic spacer
Finally, conversion of diol 3 into either the bisazide 17 or the monoazide 18 was best accomplished using DPPA in the presence of DBU. 20 From an operational standpoint this process was highly efficient as the crude products from these reactions were usually of sufficient purity to be used in the pivotal click reactions.
With the substrates 12, 13, 16, 17, and 18 to hand, their pivotal metal-catalysed cycloaddition chemistry was next investigated. 21 After some experimentation we were delighted to observe that the click reaction between 17 and two equivalents of 16 proceeded efficiently when conducted at 50 °C, rather than at room temperature, in the presence of copper sulfate (10 mol%) and sodium ascorbate (20 mol%) in aqueous dioxane, afforded 19 in 64% isolated yield. Gratifyingly the azide-tagged dye 17 also underwent a double click reaction with the functionalised macrocycle 12 affording the conjugated dye 20 in 57% isolated under the same reaction conditions (Scheme 5). Likewise click reaction of the monoazide 18 with the acetylenic sugar derivative 16 (1 equiv), under our standard reaction conditions, afforded the conjugated dye 21 in good isolated yield (61%) after column chromatography. Attempted one-pot double click reaction between macrocycle 13, carbohydrate 16, and scaffold 22 was, however, less successful and resulted in the isolation of the mixed click product 23 (9%) together with the symmetrical bisadducts 24 (5%) and 25 (13%). Both adducts 24 and 25 were also prepared independently by the reaction of 22 with either 13 (in 19% isolated yield)

Letter Syn lett
57% isolated yield; Scheme 6). The use of aqueous THF as solvent proved optimal for these reactions, although it is not clear, at this stage, why the cycloaddition leading to 24 was so low yielding.

Scheme 6 Regiochemical outcome of click reactions using Cu(II) and Ru(II) catalysts
In passing it should be noted that, from extensive literature analogy, 22 we presumed that the use of copper(II) precatalysts in these Huisgen cycloaddition reactions afforded 1,4-disubstituted triazole derivatives, as shown. While the 1 H NOESY spectra of the adducts described above (Scheme 6) were in accord with such an outcome we were keen to prepare the alternate 1,5-regioisomers in order to confirm this assertion. Accordingly, the click reaction between the carbohydrate 16 and the bisazide 22 was repeated using the protocol reported by Fokin. 23 Reaction between 16 (2 equiv) and 22 (1 equiv) was best accomplished in 1,4-dioxane at 50 °C using a catalytic quantity of Cp*RuCl(PPh 3 ) 2 (2 mol%). After 15 hours at 50 °C the reaction was quenched and afforded the 1,5-disubstituted triazole 26 as the sole product in 63% isolated yield after column chromatography. The 1 H NMR spectra of 25 and 26 proved to be subtly different (Figure 2) and the 1 H NOSEY spectrum of 26 was again fully consistent with the proposed regiochemical outcome of this ruthenium-catalysed reaction (Scheme 6).
In conclusion we have developed a modular, convergent, synthetic approach to the synthesis of functionalised dyes which possess either mixed-donor macrocylic or carbohydrate motifs. The present study serves as a proof of concept and demonstrates the viability of click chemistry for the conjugation of recognition and carrier domains to a central scaffold for use in the synthesis of assembly of metal sensors. 24 Work is currently in progress towards the construction of molecular sensors, which incorporate fluorescent reporter domains capable of mapping the distribution of metals in biological systems.