Synthesis and metal binding properties of N-alkylcarboxyspiropyrans

Spiropyrans bearing an N-alkylcarboxylate tether are a common structure in dynamic, photoactive materials and serve as colourimetric/fluorimetric cation receptors. In this study, we describe an efficient synthesis of spiropyrans with 2–12 carbon atom alkylcarboxylate substituents, and a systematic analysis of their interactions with metal cations using 1H NMR and UV-visible spectroscopy. All N-alkylcarboxyspiropyrans in this study displayed a strong preference for binding divalent metal cations and a modest increase in M2+ binding affinity correlated with increased alkycarboxylate tether length.


Introduction
Spiropyrans are a class of spiro-fused indolochromene (e.g., C4SP, Scheme 1) which exist in photo-controlled equilibrium with their zwitterionic, fully-conjugated merocyanine isomer [1] (e.g., C4MC, Scheme 1). The controllable nature of this isomerisation and the dramatic differences in physical and chemical properties displayed by the isomeric forms, coupled with the facile elaboration of the spiropyran core, have made spiropyran-merocyanine systems a common motif in molecular switch and sensing applications [2]. In this respect, the difference in optical properties between colourless, non-fluorescent spiropyran and coloured, fluorescent merocyanine has been extensively exploited [2].
Spiropyran-metal cation binding occurs as a result of stabilisation of the zwitterionic merocyanine isomer via phenoxide-metal complexation [17] (Figure 1). Commonly, merocyanines undergo photoreversion to their corresponding spiropyran under visible light irradiation and metal complexation is usually achieved either in darkness or under UV irradiation. Greater stabilisation of merocyanine-metal cation complexes is often achieved through the addition of extra ligation sites and this effect is particularly pronounced in structures bearing an 8′-OMe substituent with respect to binding divalent metal cations [18] (Scheme 1). The extension of this basic bidentate ligand with further substituents has been employed to generate structures with bespoke binding characteristics (e.g., metal ion specificity, control of complex stoichiometry, greater binding affinity) [17]. Typifying this approach, Natali et al. synthesised butanoate-tagged spiropyran C4SP and demonstrated its high affinity for Zn 2+ and Cu 2+ through the involvement of the N-alkylcarboxylate in metal binding [3] (Scheme 1).
Given the evident importance and widespread application of N-alkylcarboxyspiropyrans as building blocks in the synthesis of photoactive materials -and as interesting molecules in them-selves -we describe herein an efficient synthesis of N-alkylcarboxyspiropyrans (analogous to C4SP) with alkyl chain lengths from C2 to C12, and a systematic analysis of their metalbinding properties across a range of metal cations. It was envisaged that spiropyrans bearing incrementally-spaced carboxylate groups on conformationally flexible tethers would present trends in their metal-binding characteristics and that such information would have relevance to the application of N-alkylcarboxyspiropyrans generally.
Initially, we followed the conditions employed by Natali et al. in the three-step synthesis of butyric acid derivative C4SP, which involved: (i) alkylation of 2,3,3-trimethylindolenine (1) with ethyl 4-bromobutyrate (2; R = Et, n = 3) in chloroform; (ii) condensation of the resulting indolium salt 3 with 3-methoxy-5-nitrosalicylaldehyde (5) in ethanol; (iii) basic ester Scheme 2: General synthesis of N-alkylcarboxyspiropyrans.  hydrolysis [3] (Scheme 2). Unfortunately, in our hands the alkylation was extremely sluggish, and both this step and the final ester hydrolysis were low yielding. Consequently, we pursued a more efficient approach, reacting trimethylindolenine (1) directly with bromoalkanoic acids 2a-g, obviating the requirement for ester hydrolysis (Table 1). N-Alkylation of trimethylindolenine (1) with bromoalkanoic acids/esters is often slow and the solvent choice is crucial in identifying viable, robust conditions. Reported protocols have employed chloroform [3], acetonitrile [19], acetone [20], nitromethane [21], 1,2dichlorobenzene [22], toluene [23], reaction in the absence of solvent [24] or use of microwave irradiation [25]. We found that refluxing acetonitrile was relatively effective, if slow (completion in ca. 60 h), and that a more rapid process occurred if the solvent was allowed to evaporate during the course of the reaction (completion in ca. 20 h). The reaction in the absence of solvent was less effective, perhaps due to inefficient stirring of the small reaction volume (ca. 0.3 mL) or the absence of polar aprotic solvent-mediated acceleration of this S N 2 process ( Table 1, compare entries 6, 7 and 8).
The relative instability of indolium salts is well-documented [26] and our attempts to purify compounds 3 by silica gel chromatography or recrystallisation were unsuccessful. In light of this, and given that our alkylation reactions were relatively clean, crude indolium salts 3 were then condensed with 3-methoxy-5-nitrosalicylaldehyde (5) to give the required spiropyrans C n SP (Table 1). Commonly, this has been achieved through simple reflux in ethanol; however, this transformation proved relatively ineffective for alkylcarboxyindolium salts 3 and consequently, we applied alternative conditions using MEK and piperidine [27]. This two-step approach was used to synthesise a range of N-alkylcarboxyspiropyrans derived from bromoalkanoic acids of varying chain length in two-step yields from 60-82% (Table 1, entries 3, 7, and 9-11).
Our attempts to synthesise C4SP using our two-step protocol were undermined by the ineffective alkylation of trimethylindolenine (1) with 4-bromobutyric acid (2c). Under our alkylation conditions, intramolecular lactonisation of 4-bromobutyric acid to γ-butyrolactone (8) was more rapid than N-alkylation and no indolium product was observed (Scheme 4). Consequently, to obtain C4SP, we employed an optimised version of the three-step procedure of Natali et al. [3], wherein lactonisation is avoided through the use of ethyl 4-bromobutyrate, and basic ester hydrolysis is required as a final step (

Binding studies
Analysis of spiropyran-merocyanine equilibria with respect to metal binding has frequently employed UV-visible spectroscopy. Merocyanine-metal cation complexes absorb strongly in the visible range (often 450-600 nm), giving rapid access to clear, quantitative data at low concentrations and without interference from paramagnetic transition metal cations (a feature which causes significant problems with NMR analysis). Conversely, use of bound merocyanine-metal cation absorbance intensity as the sole metric for quantification of metal binding of different metal cations should be undertaken with caution. Direct comparison of absorbance intensity measurements (as a surrogate for concentration) is only valid if the molar extinction coefficient, ε, remains constant for all merocyanine complexes of dif-ferent metals, and indeed for different merocyanine complexes of the same metal. We thus sought initial validation of UV-vis data by direct comparison with data from 1 H NMR, derived from merocyanine complexes of two non-paramagnetic metal cations, Mg 2+ and Zn 2+ . 1 H NMR allows rapid assessment of spiropyran:merocyanine via measurement and comparison of integrals, and hence enables calculation of [merocyanine]. In turn, [merocyanine] can be applied to quantify the corresponding UV data. Related approaches have employed IR spectroscopy [31] and partial least squares regression analysis [32] as adjuncts to UV-vis spectroscopy in the analysis of merocyanine binding across ranges of transition metal cations.
We prepared solutions of C2SP-C12SP in CD 3 CN to which were added aqueous solutions of either Zn(NO 3 ) 2 ·6H 2 O or Mg(NO 3 ) 2 ·6H 2 O. These samples were allowed to equilibrate in darkness at room temperature overnight and were then analysed by 1 H NMR spectroscopy. Following this, the samples were diluted, allowed to re-equilibrate in darkness, and then analysed by UV-vis spectroscopy. Similar Zn 2+ and Mg 2+ samples were prepared with N-methyl spiropyran 9 (Figure 2, prepared according to a reported procedure [33]), which served as a carboxylate-free control. The results from these experiments, and example spectra, are shown in Figure 3 and Figure 4, respectively, from which two trends are apparent: (i) ε remains relatively constant for a given compound, irrespective of divalent metal cation; (ii) ε decreases slightly with increasing chain length. Unexpectedly, 1 H NMR analysis of C3SP in the presence of Zn 2+ and Mg 2+ produced complex and intractable spectra and it would appear that these metal salts are able to promote degradation of this compound.
Irrespective of the precise fate of C3SP under these conditions, without NMR evidence of merocyanine formation, this compound was necessarily omitted from this study.
On the basis that the nature of the metal cation appeared to have minimal influence upon the merocyanine extinction coefficient, we were able to use UV-vis spectroscopy to rapidly assess the impact of binding various metal cations upon compounds C2SP-C12SP and 9 by applying the NMR-derived ε values (average of ε for MC-Zn 2+ and MC-Mg 2+ for each compound) to calculate approximate [merocyanine] from merocyanine absorbance intensity ( Figure 5). Although this is based on the analysis of only two metal cations, studies using dynamic modelling have displayed similar, limited impact of different metal cations upon merocyanine ε values [34] so we have cautious confidence in the validity of this approach. All compounds, including control compound 9, showed selectivity for M 2+ over M + which follows the pattern previously observed for merocyanines bearing an 8′-OMe substituent [18]. Broadly speaking, the effects upon metal binding of incremental increase in alkylcarboxylate chain length were subtle. The general trends were: (i) compounds bearing longer tethers produced higher concentrations of MC-M 2+ complex per unit metal ion (presumably on the basis that carboxylates on longer tethers with greater conformational flexibility are able to interact more effectively with cations, despite their inherent entropic penalty [35]); (ii) affinity for metals was in the order Zn 2+ > Mg 2+ > Co 2+ > Ni 2+ . Binding to Ni 2+ was considerably less effective than for other divalent metal cations and this was particularly pronounced for non-carboxylate 9, wherein 9-Ni 2+ complexation did not exceed background [merocyanine]. Merocyanine complexation of Cu 2+ was also investigated; however, the absorbance wavelength for MC-Cu 2+ (≈400 nm) overlapped with that of the larger spiropyran absorbance (≈370 nm) and we were unable to extract meaningful data from this. This is consistent with related studies [3] and, furthermore, spiropyranbased Cu 2+ detection can be complicated by Cu 2+ -catalysed spiropyran dimerisation [36]. In the light of the results from Natali et al. [3] -in which evidence for involvement of a carboxylate ligand in metal binding was presented but no comparison was made against a control -it is reassuring to note that incorporation of such a ligand into the merocyanine structure did increase the affinity for Zn 2+ , Mg 2+ and Co 2+ over that seen in the non-carboxylate control compound 9.
It is apparent from these data that the length of the carboxylate tether can also affect the spiropyran-merocyanine equilibrium in the absence of metal cations ( Figure 5 and Figure 6). All compounds tested existed as mixtures of spiropyran and merocyanine isomers at 0.1 mM in acetonitrile in darkness. For shorter-chain compounds C2SP and C4SP, and control compound 9, this resulted in an approximate 2:8 MC:SP ratio. Longer chain carboxylates C6SP-C12SP, however, showed higher merocyanine concentrations and these data appear to indicate a sharp threshold between 4-C and 6-C carboxylates. Correspondingly, we synthesised 5-C carboxylate C5SP (using our standard protocol; 77%) and assessed its MC:SP equilibrium behaviour under similar conditions. In this case, [C5MC] resembled [merocyanine] values for the shorter chain compounds and reinforced the idea of a threshold distance required for carboxylate-mediated merocyanine stabilisation. It is as yet unclear how a long carboxylate tether might stabilise a merocyanine structure whilst structures bearing a superficially simi-lar tether do not. To probe involvement of the carboxylate/ carboxylic acid moiety (e.g., in forming inter-or intramolecular hydrogen bonds) we prepared C6 ester derivative 10 ( Figure 7)   is the defining factor in merocyanine stabilisation and that only beyond a threshold distance does hydrogen bond donation of the carboxylic acid become important. This is a key point in designing functional materials based on N-alkylcarboxyspiropyrans.

Conclusion
In summary, we have developed a high-yielding, two-step synthesis of N-alkylcarboxyspiropyrans via N-alkylation of trimethylindolenine with bromoalkanoic acids, then condensation of the resulting indolium salts with methoxynitrosalicylaldehyde. This protocol is generally effective and enabled the synthesis of a range of spiropyrans bearing different length alkylcarboxylate tethers; however, it is ineffective with bromoacetic acid and bromobutyric acid, where decarboxylation and intramolecular lactonisation respectively compete with N-alkylation. In these cases, we have developed alternative, though lower-yielding procedures.
N-Alkylcarboxyspiropyrans can function as colourimetric/ fluorimetric receptors for metal cations via complexation of the merocyanine isomer. Consequently, we have assessed the metal binding behaviour of spiropyrans bearing N-acetic acid through to N-dodecanoic acid tethers by 1 H NMR and UV-vis spectroscopy. All compounds tested displayed a strong preference for divalent over monovalent metal cations and a modest increase in M 2+ binding affinity correlated with increasing alkycarboxylate tether length.
This paper details a clear, effective protocol for the synthesis of N-alkylcarboxyspiropyrans and a thorough analysis of the effect of metal cations upon their spiropyran-merocyanine equilibria. Consequently, the results of this study will impact upon the design and synthesis of dynamic functional materials based on spiropyran-merocyanine units, with particular relevance to cases where such materials are used in the presence of metal cations (e.g., in biological media).

Supporting Information
Supporting Information File 1 Full experimental details.