Solubility study of tobramycin in room temperature ionic liquids: an experimental and computational based study

Abstract

Herein, we present a computational and experimental study assessing the solubility of tobramycin 1 in a series of hydrophilic room temperature ionic liquids (RTIL). We show that hydrophilic ionic liquids, in particular those containing acetate, [OAc]⁻, enable significantly enhanced solubilisation of tobramycin in comparison to conventional molecular solvents.

---

Tobramycin 1, Fig. 1, an aminoglycoside, is one of the main treatments for chronic Pseudomonas aeruginosa infections of patients suffering from cystic fibrosis (CF).¹ Under biologically relevant conditions, the amine functionalities of tobramycin are protonated, yielding a polycationic species which displays poor penetration of biological membranes, including the sputum which lines the lungs of CF sufferers and in which bacterial colonies accumulate.² In order to enhance delivery of tobramycin 1 through the sputum of the CF lung, and to overcome developing bacterial resistance, targeted modifications to this molecule are required. As tobramycin is a multi-functionalised molecule consisting of one primary alcohol, four secondary alcohols, and five primary amine moieties, procedures to selectively modify tobramycin 1 are complex, generally requiring a number of protection and deprotection steps to reach useful analogues. Commonly, the amino groups are masked using tert-butoxycarbonyl (Boc) protecting groups or using chelating metals such as zinc acetate, before any selective chemical modifications can be performed.^3–5 Due to its highly polar character, tobramycin 1 has a very low solubility in most organic solvents, including ether and chloroform and exhibits only limited solubility in protic organic solvents such as ethanol (∼0.01%) and polar aprotic solvents like dimethyl sulfoxide (DMSO) (∼0.01%).^6,7 Tobramycin's limited solubility profile in common organic solvents limits drastically the type of reagents and choice of conditions that can be employed towards its chemical modification. In this respect, tobramycin 1 possesses chemical properties which are similar to that of nucleosides, as the latter's poor solubility in organic solvents and multiple functionalities present similar challenges.

Fig. 1 Tobramycin 1 and amine pK_as.

Most commonly, nucleoside modifications must employ polar aprotic, high boiling organic solvents in addition to long reaction times and extensive workup procedures. These issues are very similar to considerations that are necessary for tobramycin 1 chemistry. However, it has been demonstrated that the solubility issues observed for nucleosides could be overcome by the use of room temperature ionic liquids (RTILs) as an alternative reaction medium.⁸ RTILs, consisting of weakly associated ions, are liquid at room temperature, and possess numerous properties unobserved in standard organic solvents, such as negligible vapour pressure, non-flammability, high thermal stability and ability to solubilise a wide range of organic and inorganic compounds.^9,10 An additional property which is particular to RTILs is the possibility for an almost infinite combination of cation and anion resulting in RTILs with tailor-made properties, such as polarity, hydrophilicity and solute solubility.¹¹

With such opportunities that designer solvents offer, RTILs have been applied in a broad range of chemical processes, including that of dissolving plant mass, to access valuable raw materials. Carbohydrates and their associated products are poorly soluble in organic solvents and, RTILs with their large solvating capacity have been found to be highly useful in the solubilisation of biomass and the isolation of carbohydrates.^12,13 Access to cellulose is hindered by its insolubility in most conventional polar and apolar solvents, due to its fibril structure. Therein, the RTILs, 1-butyl-3-methylimidazolium acetate [C₄MIM][OAc] 3 and 1-butyl-3-methylimidazolium bromide [C₄MIM]Br 16 were shown to be highly effective at dissolving cellulose¹⁴ and facilitating the isolation of the sugar alcohols, xylitol and mannitol, both of which are used as commercial sweeteners.^15,16

Based on these literature precedents and many others, it was hypothesised that due to the functional similarities of tobramycin 1 to carbohydrates and nucleosides, it would be possible to identify a range of RTILs as an alternative reaction media that would demonstrate enhanced solubilisation of tobramycin 1 and offer novel opportunities for chemistry. In the first instance, a series of molecular dynamics calculations (see ESI†) was performed in order to determine likely candidate systems for study. This information was valuable to guide the choice of IL employed in the solubility study, helped to predict the charge distribution and to rationalize the empirical experiments. The ionic liquids studied were all based around the 1-butyl-3-methylimidazolium cation, [C₄MIM]⁺, as the choice of anion predominantly affects the chemical properties of the ionic liquid. For this study, four different anions were considered: bromide, benzoate, bis(trifluoromethylsulfonyl)imide ([NTf₂]⁻), and acetate, Scheme 1. Initially the partial radial distribution functions between amine groups on tobramycin 1 and H-bond accepting sites on the various anions were calculated. These functions provide an indication as to the potential for H-bond formation, and, therefore, the potential solvation strength of the corresponding IL. These H-bonding sites were identified as the bromide itself, the oxygen atoms of the benzoate and acetate, and the nitrogen atom on the [NTf₂]⁻ anion, Scheme 1. Fig. 1 shows the numbering scheme utilised for the amine moieties on tobramycin 1, and Fig. 2 shows the partial distribution functions between the various amine groups and the relevant sites on the anions studied. For all the anions studied except the [NTf₂]⁻, a sharp peak located between 3 and 4 Å is observed, indicating the presence of relatively strong hydrogen bond interactions.

Scheme 1 [C₄MIM]⁺ cation and the anions used in this study with the hydrogen bonding acceptor sites highlighted in red.

Fig. 2 Partial RDFs between amine groups of tobramycin 1 and anion H-bonding sites.

In the radial distribution function (RDF) associated with the [NTf₂]⁻ anion, the first distinctive peak is further away, between 4 and 6 Å, than found for the other anions studied. This suggests a much weaker interaction with the amino functionality of tobramycin 1, as is expected for this type of anion.¹⁷ In nearly all cases (i.e. for almost all of the NH₂ groups considered) the acetate anion sits closest to tobramycin 1 indicating the strongest H-bond interaction and, therefore, the highest solvent potential. In contrast, for the benzoate anion, despite displaying chemical similarity to the acetate anion, presents a slightly longer first peak in the RDF. This suggests a weaker hydrogen bond interaction, and is likely to be due to the associated steric hindrance from the phenyl ring. The bromide anion displays its first distinctive peak slightly further than that of the benzoate anion, indicating weaker H-bonding interactions.

From the modelling data obtained, it was observed that apart from [NTf₂]⁻, the more basic ionic liquids, acetate and benzoate, displayed strong interactions with tobramycin 1 and, therefore, would likely act as suitable solvents. Consequently, it was decided to synthesise a series of ionic liquids incorporating variants based upon a carboxylic acid core. To examine the relationship between solvation properties and the polarizability and basicity of the “oxo”-based anions, the properties of the formate and mesylate based RTILs were also considered.

The ionic liquids tested consisted of imidazolium, pyrrolidinium or methoxylethyl imidazolium based systems and were used to ascertain if any additional fine tuning effects of the solubility could be observed by modulating the cation. In particular, 1-methoxyethyl-3-methyl imidazolium methane sulphonate [MOEMIM][OMs] 7 ionic liquid was considered in this study as it has been successfully employed in nucleoside chemistry.⁸ These series offered opportunities to assess whether additional H-bonding sites on the cation improved solubility. Furthermore, hydrophobic ionic liquids containing either the tris(perfluoroalkyl)trifluorophosphate ([FAP]⁻) or [NTf₂]⁻ anions were included to compare the experimental data and predictions from the modelling (synthesis and structure of the ionic liquids are reported in the ESI†).

The solubility study was performed on both dried ionic liquids and air equilibrated forms to investigate the effect of water content on the overall solubility. Table 1 presents the solubility of tobramycin 1 as weight percentage in the tested ionic liquids and at different water content. As the dried bromide containing ionic liquids are solids at room temperature, these were not included in the study. The data in Table 1 show that in its free base form, tobramycin 1 could not be solubilised in the hydrophobic ionic liquids consisting of the [NTf₂]⁻ and [FAP]⁻ anions, in both the dried and air-equilibrated samples. This was a predicted outcome as the electron delocalization within the anion results in poor hydrogen bonding ability with tobramycin 1 amino functionalities.

Table 1 Solubility of tobramycin 1 in ionic liquids either dried or air equilibrated^a

Ionic liquid	Wt% (solubility)	Water% (dried)	Wt% (solubility)	Water% (air Eq)
a Each sample performed in triplicate and the average quantity of tobramycin 1 solubilized and recorded as the weight percentage (wt%) at 298 K, air equilibrated samples were left exposed to the atmosphere for 48 h. Dried samples were dried under high vacuum with stirring for 48 h at RT [C4MPYRR][OAc] 2 (1-butyl-1-methyl-pyrrolidinium acetate), [C4MIM][OAc] 3 (1-butyl-3-methyl-imidazolium acetate), [C4MPYRR][OBz] 4 (1-butyl-1-methyl-pyrrolidinium benzoate), [C4MIM][OOCH] 5 (1-butyl-1-methyl-pyrrolidinium formate), [C4MPYRR][OOCH] 6 (1-butyl-3-methyl-imidazolium formate), [MOEMIM][OMs] 7 (1-methoxyethyl-3-methyl imidazolium methane sulphonate), [C4MIM][OBz] 8 (1-butyl-3-methyl-imidazolium benzoate), DMSO 9 (dimethylsulphoxide), DMF 10 (dimethylformamide), [C4MPYRR][TFA] 11 (1-butyl-1-methyl-pyrrolidinium trifluoroacetate), [C4MIM][TFA] 12 (1-butyl-3-methyl-imidazolium trifluoroacetate), [C4MIM][NTf2] 13 (1-butyl-3-methyl-imidazolium bistriflimide), [C4MPYRR][NTf2] 14 (1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl)imide) [C4MPYRR][FAP] 15 (1-butyl-1-methylpyrrolidinium tris(pentafluoroethyl)trifluorophosphate).
[C4MPYRR][OAc] 2	45.9	12.0	42.1	21.3
[C4MIM][OAc] 3	35.2	0.9	31.5	23.5
[C4MPYRR][OBz] 4	25.2	0.7	26.7	12.8
[C4MIM][OOCH] 5	23.5	0.5	21.4	18.5
[C4MPYRR][OOCH] 6	20.1	3.6	19.4	21.5
[MOEMIM][OMs] 7	21.3	0.2	20.4	24.2
[C4MIM][OBz] 8	13.4	0.9	22.9	10.5
DMSO 9	5.1	n/a	5.1	n/a
DMF 10	1.1	n/a	1.1	n/a
[C4MPYRR][TFA] 11	0	0.5	0	2.5
[C4MIM][TFA] 12	0	0.4	0	1.8
[C4MIM][NTf2] 13	0	1.6	0	1.6
[C4MPYRR][NTf2] 14	0	0.9	0	1.2
[C4MPYRR][FAP] 15	0	0.1	0	0.2

---

Modulation of the tobramycin's solubility by the anion followed the order of [TFA]⁻ < [HCOO]⁻ < [OBz]⁻ < [OAc]⁻ (Table 1), for the dried hydrophilic ionic liquids. With the exception of the [OMs]⁻ containing IL 7, this was consistent with increased pK_a of the corresponding conjugated acid at 0.23, 3.77, 4.20 and 4.75 respectively;^19–21 with, in general, the larger the pK_a, the greater the strength¹⁸ of the hydrogen bond being formed. As such, a weaker conjugated acid containing IL will form stronger H-bonding interactions with the amino groups of tobramycin 1, which can account for the observed trend in solubility. The ionic liquids 11 and 12, which incorporate [TFA]⁻ and [OMs]⁻ anions, solubilised tobramycin poorly, as a consequence of weaker H-bonding interaction with tobramycin 1. Benzoate based ionic liquids 4 and 8 displayed a lower capacity for the solvation of tobramycin 1 than anticipated, and this in comparison to RTIL 2 and 3, Table 1. This outcome can be attributed to steric hindrance as a result of the bulky aromatic ring of the benzoate anion, reducing proximity to hydrogen bond donor groups on tobramycin 1, increasing the bond length, and consequently reducing the strength of the H-bonding interactions. This is confirmed by the RDF data, Fig. 2, which shows the peak for the benzoate anion at greater distance than for the acetate anion, indicating a greater distance from the N26 amine of tobramycin 1. Of the anions studied, the acetate anion can be considered the most basic and, therefore, with the greatest hydrogen bond formation ability, accounting for the highest solvation of tobramycin 1 in ILs 2 and 3. It can be observed that dried [C₄MPYRR][OAc] 2 solubilised the greatest quantity of tobramycin 1 in 45.9 wt%, correlating with the modelling and theoretical data, Table 1. Of note, the water content of this IL is high at 12%. However, additional drying caused it to become too viscous to stir. Overall, the choice of cation showed a moderate effect upon the solubility of tobramycin 1. In dried [C₄MPYRR][OBz] 4 and [C₄MPYRR][OAc] 2, an increase of 10.7 and 11.8 wt%, respectively, was observed in comparison to their imidazolium cation based ILs, this unusual effect of the cation on tobramycin's solubility may be associated with a weaker interaction between the anion and the cation. The imidazolium based ILs contain an acidic cation that can H-bond with the anions that is not present in the pyrrolidinium ILs, allowing the anion to be more available for hydrogen binding with tobramycin hydrogen-bond donor groups, Table 1. Additionally, the non-planar conformations of pyrrolidinium and tobramycin might bring further opportunity for favourable van der Waals interactions.²² Interestingly, the water content of the ionic liquids, Table 1, had only a slight influence upon the solubility of tobramycin 1 within a given IL. For the majority of the ionic liquids studied, the solubility decreased ever so slightly as the water content increased. These results are in agreement with that observed in the dissolution of cellulose, in which the level of solubility has been reported to be negatively influenced by the water content of the ionic liquids, due to competing H-bond interactions.²³ One result that differs from this trend is that of the benzoate based ionic liquids 4 and 8, where solvation increases with increased water content. One possible explanation is that presence of water increases the polarity of the medium slightly which helps mitigate the influence of the large aryl group. The correlation of solvation as a function of the anion's basicity can be partially reversed by introducing additional opportunities for H-bonding in other parts of the IL, as observed in the dried [MOEMIM][OMs] 7 IL, which displayed moderate solubility of tobramycin at 21.3 wt% whilst the pK_a of [OMs]⁻ is −1.9. Here additional H-bonding interactions between tobramycin and the ethoxyl ether chain on the cation are likely to be responsible for the observed enhanced solvation.

The study was further extended to investigate whether the RTILs could be applied to other aminoglycosides, which also display poor solubility in conventional molecular solvents. As an exemplar, [C₄MIM][OAc] 3 was chosen to study the solubility of neomycin sulphate and gentamycin sulphate, and compared to their solubility in water, Table 2. As expected, the solubility of neomycin sulphate and gentamycin sulphate was significantly enhanced in [C₄MIM][OAc] 3 when compared to water.

Table 2 Solubility of neomycin sulphate and gentamycin sulphate in air equilibrated [C₄MIM][OAc] 3 at rt

Aminoglycoside	Solubility Wt%
	[C4MIM][OAc] 3	Water
Neomycin sulfate	53.2	5.0 (ref. 24)
Gentamycin sulfate	49.8	11.5 (ref. 25)

---

In order to demonstrate that tobramycin 1 solubilised in ionic liquids was still reactive towards electrophiles, a standard N-Boc protection with Boc anhydride was undertaken on tobramycin 1 in [C₄MPYRR][OAc] 2, which had shown the highest solubility. In conventional protection methodology, where water (100 eq.) and DMSO (170 eq.) are both used as solvent, tobramycin is reacted with 5 eq. of Boc anhydride for 4 h at 60 °C, and the five amino groups of tobramycin 1 are protected in 94% yield.²⁶ Utilising RTIL 2 (23 eq.) in the presence of 5 equivalents of Boc anhydride at 60 °C, no reaction was observed with tobramycin (1 eq.) until a catalytic amount of 4-(dimethylamino)pyridine (DMAP) was added to the reaction mixture. After 12 h the penta-N-Boc tobramycin derivative was isolated after work up and extraction into dichloromethane (DCM) in 34% yield, partially protected derivatives of tobramycin remained within the IL/aq layer. The requirement for DMAP (pK_a 9.7) can be explained by its ability to behave as acyl transfer reagent and as such enable the acylation of less reactive nucleophilic groups, such as amines and alcohols.²⁷ In ILs, the amino functionalities of solute tobramycin are thought to be engaged in strong hydrogen bonding, reducing their reactivity towards the non-activated dicarbonate reagent, thus giving opportunities for fine-tuning site-specific chemistry on such poly-functional reagents. The use of IL/water mixtures have also recently shown the ability to promote organic transformations superior to that of their organic solvent counter parts and was also attempted.²⁸ The reaction of tobramycin 1 with Boc anhydride in [C₄MPYRR][OAc] 2 was therefore repeated but with 1 volume equivalent of water added and after workup the penta-N-Boc tobramycin was isolated in 16% yield. In this instance the yield was low; however, no DMAP was required to promote reaction, wherein competing H-bonding interactions between the IL and water, may allow increased opportunity of the amino functionalities of tobramycin to react successfully with the anhydride.

In conclusion, the solubility of tobramycin 1 in a series of ionic liquids was performed, and offered a significant improvement over that of conventional high boiling point/toxic molecular solvents. We demonstrated that the use of such IL was not limited to tobramycin but was applicable to other aminoglycosides, and that the experimental work was in full agreement with the theoretical computational study. Furthermore importantly in such ILs and IL/water mixtures, tobramycin 1 retains its reactivity towards electrophiles, although in the case of N-Boc protection, lower yields were achieved. As such, future simulations to study the mode of solvation within IL/water mixtures and further development of ionic liquids with improved solvation properties for this class of antibacterials can be envisaged to expand on the class of reagents and conditions which can be applied to aminoglycosides.

Notes and references

1. R. M. Shawar, D. L. MacLeod, R. L. Garber, J. L. Burns, J. R. Stapp, C. R. Clausen and S. K. Tanaka, Antimicrob. Agents Chemother., 1999, 43, 122877.
2. L. Szilagyi, Z. S. Pusztahelyi, S. Jakab and I. Kovacs, Carbohydr. Res., 1993, 247, 99.
3. F. Liang, S. Wang, T. Nakatani and C. Wong, Angew. Chem., 2004, 43, 6496.
4. N. W. Luedtke, T. J. Baker, M. Goodman and I. Yitzhak, J. Am. Chem. Soc., 2000, 122, 12035.
5. S. Hanessian, A. Kornienko and E. E. Swayze, Tetrahedron, 2003, 59, 995.
6. Y. Chen, L. Zhu, Y. Gao and J. Wang, J. Chem. Eng. Data, 2009, 54, 1403.
7. L. Zhu and J. Wang, J. Chem. Eng. Data, 2008, 53, 2532.
8. V. Kumar, V. S. Parmar and A. V. Malhotra, Biochimie, 2010, 92, 1260.
9. T. Welton, Chem. Rev., 1999, 99, 2071.
10. J. Kumar, A. Chauhan and S. A. Chauhan, Tetrahedron, 2005, 61, 1015.
11. S. Keskin, D. Kayrak-Talay, U. Akman and O. Hortac, J. Supercrit. Fluids, 2007, 43, 150.
12. M. E. Zakrzewska, E. Bogel-Łukasik and R. Bogel-Łukasik, Energy Fuels, 2010, 24, 737.
13. L. Conceic, E. Lukasikb and R. Bogel-Łukasik, RSC Adv., 2012, 2, 1846.
14. J. Vitz, T. Erdmenger, C. Haenscha and U. Schubert, Green Chem., 2009, 11, 417.
15. J. M. Tanzer, Int. Dent. J., 1995, 45, 65.
16. M. Ikeda, A. K. Bhattacharjee, T. Kondoh, T. Nagashima and N. Tamaki, Biochem. Biophys. Res. Commun., 2002, 291, 669.
17. J. G. Huddleston, A. E. Visser, W. M. Reichert, H. D. Willauer, G. A. Broker and R. D. Rogers, Green Chem., 2001, 3, 156.
18. C. Laurence and M. Berthelot, Perspect. Drug Discovery Des., 2000, 18, 39.
19. J. P. Guthrie, Can. J. Chem., 1978, 56, 2342.
20. D. Harris, Quantitative Chemical Analysis, 2010, 8, 12.
21. Sigma Material Safety Data Sheet. T. Fujimori, K. Fujii, R. Kanzaki, K. Chiba, H. Yamamoto, Y. Umebayashi and S. Ishiguro, J. Mol. Liq., 2007, 131, 216.
22. J. Vitz, T. Erdmenger, C. Haenscha and U. S. Schubert, Green Chem., 2009, 11, 417–424.
23. K. Tateishi, H. Ogino, A. Waki, T. Ohishi, M. Murakami, H. Asoh and S. Ono, Electrochemistry, 2013, 81, 440.
24. http://www.scbt.com/datasheet-3573.html.
25. http://www.scbt.com/datasheet-203334.html.
26. K. Michael, H. Wangy and Y. Tor, Bioorg. Med. Chem., 1999, 7, 1361.
27. V. Lutz, J. Glatthaar, C. Würtele, M. Serafin, H. Hausmann and P. Schreiner, Chem.–Eur. J., 2009, 15, 8548.
28. P. Narayana Reddy, P. Padmaja, B. Reddyc and G. Rambabua, RSC Adv., 2015, 5, 51035.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra23078d

---