Spectrophotometric determination of α-hydroxytropolone pK_a values: A structure-acidity relationship study

Abstract

α-Hydroxytropolones (αHTs) have a wealth of biological activity owing to their ability to serve as metalbinding fragments for many therapeutically valuable dinuclear metalloenzymes. They also have the potential to exist in as many as 4 protonation states under aqueous acidic or basic conditions. The following details how UV absorption can be used to generate pK_a values on a series of αHTs. The studies also provide some knowledge into how the acidity and basicity change with some different functional groups. These studies thus provide new strategies and knowledge that could be valuable in leveraging αHTs as metal-binding fragments in drug-development pursuits.

α-Hydroxytropolones (αHTs) are oxygenated troponoids that are useful as fragments in drug development due to their ability to inhibit several clinically relevant dinuclear metalloenzymes [1]. They also have a dynamic range of protonation states that could be accessible in the aqueous pH range. Based on measurements of dissociation constants of αHT (1) by Yui in 1956 (Scheme 1) [2], we know that αHTs are fairly acidic (pK_a2 = 6.7) and likely exist in their anionic form at physiological pH. Furthermore, under mildly basic conditions, 1 becomes dianionic (pK_a3 = 11.5), which could represent the ionization state through which αHTs bind to certain metalloenzymes [3]. While Yui did not measure pK_a1, Hosoya and coworkers found the pK_a of tropone (−1.02) and tropolone (−0.86) to be substantially higher than that of saturated ketones like acetone (pK_a = −7.5), and much more close to that of an amide (pK_a ~ −0.5) [4]. This can be attributed to the high contribution of the polarized ground state (i.e., the tropylium cation resonance form). Thus, αHTs would be expected to have a similar pK_a1 values.

Scheme 1.

α-Hydroxytropolone protonation states. α-Hydroxytropolone and its 4 different protonation states, along with pK_a values previously determined using potentiometric titration [2].

Over the last several years our lab has been developing chemistry to access new and structurally diverse αHTs [5], and leveraging this chemistry in various drug development pursuits [6]. Given these studies, we have become highly interested how pK_a values may differ throughout our library of compounds, as each of these values might tell us about a property that could be relevant to their binding. Thus, we became interested in developing a baseline understanding of how these structural changes would influence pK_a values. Attention was focused on UV–Vis-based determination rather than potentiametric titration mostly due to the low amounts of material needed as compared to previously described potentiametric titration methods [7].

We began our studies by evaluating pK_a values of 1 through UV–Vis analysis to provide baseline analysis and to confirm that the method would be consistent with literature values. UV Vis spectra of 1 were thus taken at a series of 24 different pH values ranging from approximately −1 to 13. We found it particularly valuable that the spectra differed significantly at the higher λ_max values (>340 nm), which helped more readily identify the spectra of the 4 different ionization states (Fig. 1A). Processing the UV–Vis data closely analogously to the methods outlined by Tomsho [8], spectral absorbance differentiation graphs were generated for low (pH −1.08 to 3.12, Fig. 1C), medium (pH 3.12–9.06, 1 to 1-, Fig. 1D), and high pH ranges (pH 9.06–12.93, 1- to 12-, Fig. 1E), which represented the transition from 1⁺ to 1, 1 to 1⁻, and 1⁻ to 1²⁻, respectively. Focusing on the high λ_max values for clarity, we identified wavelengths that provided local minima (most negative) and maxima (most positive). The absorbance difference of the spectra at 372 nm and 340 nm were plotted against lower pH, the absorbance difference of the spectra at 397 nm and 362 nm were plotted against a middle range of pH values, and the absorbance difference of the spectra at 420 nm and 380 nm were plotted against high pH values. pK_a was determined through analysis of the resulting sigmoidal curves through non linear regression curve fitting with GraphPad Prism software using the equation:

$$\begin{gathered} \text{Absorbance Difference}=(\text{Low Absorbance - High Absorbance} \\ \ast [10(\mathrm{pH}-{\mathrm{pK}}_a)])∕(1+10(\mathrm{pH}-{\mathrm{pK}}_a)) \end{gathered}$$

Fig. 1.

Overview of UV–vis-based pK_a Determination of HT 1. (A) Normalized UV–Vis traces of 1 across a range of pH. Highlighted in bold are spectra that reflect different protonation states for clarity. (B) Plot of absorbance differences determined by spectral difference graphs versus pH. pK_a values determined by curve fitting using GraphPad Prism. (C–E) Spectral difference plot at (C) low, (D) medium, and (E) high pH values, along with local minima and maxima values used for pK_a determination. For (C–E), refer to color key in (A) for pH range.

This analysis gave us strong fits (R² > 0.98) and provided us with measured pK_a2 and pK_a3 values of 6.4 and 11.6, respectively. These values were in close agreement with those determined previously Yui (6.7 and 11.5) [2]. We were also able to determine pK_a1 value of 0.3, which was slightly higher then that measured previously for tropone and tropolone [4], and could be the result of added resonance stabilization of the tropylium form provided by the additional hydroxyl group. This consistency gave us confidence that UV–Vis could be used to determine pK_a values across a broader range of structurally diverse αHTs.

We chose to evaluate 11 additional αHTs that provided us with a good representation of the structural diversity attainable through synthetic methods and strategies being developed in our lab [1d,5,6c,9], and a summary of these pK_a values is shown in Table 1. Determination of pK_a1 was limited to only one additional molecule, which was phenyl-containing αHT 5. All other molecules possessed electron-withdrawing groups that appear to lower the basicity below the limitations provided by aqueous pH range. All molecules provided fully formed sigmoidal curves that allowed us to confidently measurable pK_a2 values. Every pK_a2 measured was less then 7, suggesting that all of these molecules as well as close structurally homologs of them likely exist in an anionic state at normal physiological pH. Several of the molecules did not differ that substantially in pK_a2 from 1. Molecules possessing a single carbonyl appendage (2–4), a phenyl group (5) or a single thioether appendage (9) were all within 0.6 pK_a units of 1. αHT 7, which contained both an ester and a bromide, as well as 8, which contained two esters, decreased in acidity by over 1 pKa unit to about 5. A similar pK_a2 was also observed with two thioether groups (10). By far the lowest pK_a2 values were observed for molecules possessing sulfonyl groups. αHT 6 and 11, which possessed a single sulfonyl, were each over 2 pK_a units lower than 1, with pK_a values of about 4. Meanwhile compound 12, which possessed two sulfonyl groups, had a pK_a2 of 2.3, which was over 4 pK_a units less then that of 1.

Table 1.

Summary of α HT UV–Vis-determined pK_a values.

HT	pKa1	pKa2	pKa3	HT	pKa1	pKa2	pKa3
1	0.3	6.4	11.6	7	<0	4.8	10.9
2	<0	6.0	11.8	8	<0	5.0	10.6
3	<0	6.0	11.7	9	<0	5.9	>12
4	<0	5.9	11.9	10	<0	4.8	>12
5	0.1	6.5	>12	11	<0	3.8	11.0
6	<0	4.3	10.9	12	<0	2.2	7.6

The higher pK_a3 values of the αHTs were more measurable then those of pK_a1, and fully or near fully formed sigmoidal curves were obtained for all compounds except 5, 9 and 10, whose pK_a3 values exceeded 12. Despite their electron-withdrawing groups, compounds 2–4 had similar pK_a3 values then 1, if not higher. When an additional electron withdrawing group, such as a bromide (7) or an additional ester (8) was appended, pK_a3 did drop a bit to 10.9 and 10.6, respectively. A similar small decrease in pK_a3 was also observed with mono-sulfonyl-containing α HTs 6 and 11, which each had a pK_a3 values of 10.9. These pK_a3 values were within 1 pK_a unit of 1, whereas in both cases pK_a2 values were over 2 pK_a units lower than 1. Thus, for many of these molecules, pK_a3 values are not influenced by resonance-mediated anion stabilizing groups to the extent that they are for pK_a2, and indeed similar trends have been observed previously with certain catechol derivatives [10]. Furthermore, in many instances, pK_a3 may even be slightly higher then those of 1 (e.g. 2–4, 9, 10). We can only speculate at the present time why this change in acidity trend is observed. For compounds 2–4, it may be the case that once the molecule achieved an anionic state, the resonance stabilizing groups are occupied by higher electron density, and thus behave as donor groups to destabilize the dianionic state. In the case of thioether tropolones 9 and 10, the dianionic state of these molecules may cross an electron density threshold whereby unfavorable electron-repulsion with the thioether lone pairs exist.

The slight exception to this trend appeared to be those molecules with two resonance-stabilizing groups, namely 8 and 12. In both cases, the decrease in pK_a3 as compared to 1 was much closer to the decrease in pK_a2 (Fig. 2). Compound 8, for example, has the second lowest pK_a3 value despite ranking sixth in pK_a2 value. The impact of the 2nd electron-withdrawing group is perhaps best visualized comparing the pK_a values with 11, which has a substantially lower pK_a2 than 8 but has a higher pK_a3. Thus, it appears that while a single sulfonyl group helps stabilize the monoanionic form better than two esters, the dianionic form is stabilized more efficiently by two esters. When comparing 11 to 12, the additional sulfonyl group drops pK_a3 by 3.3 pK_a units, whereas pK_a2 differs only by 1.7 pK_a units. Thus, the 2nd sulfonyl group impacts pK_a3 moreso then pK_a2. A model whereby each resonance-stabilizing group delocalizes the electrons of one oxyanion can help explain this result (Fig. 2). Compound 12 is by far the most acidic molecule we tested, and with a pK_a3 of 7.6, it may even exist primarily in the dianionic form in certain biological systems.

Fig. 2.

Dianionic molecules containing one and two resonance stabilizing groups, and the change in pK_a2 and pK_a3 as compared to parent α HT 1.

In conclusion, we have demonstrated that UV–Vis can be a practical method for measuring experimental pK_a values of αHTs. We used this method in the evaluation of pK_a1, pK_a2 and pK_a3 of 12 αHTs. These studies provide some preliminary insight into how structural variations can alter the pK_a values of αHTs, as well as details of how UV–Vis can be useful for future determination of other αHT-based pK_a values. These knowledge and methods to access them should be useful in leveraging α HTs in drug-development pursuits.