Linear Oligosulfoxides: Synthesis and Solubility Studies

Abstract

Synthesis of three linear oligosulfoxides containing up to six sulfoxide groups was achieved by multiple S_N2 reactions between an alkanethiol and alkyl tosylate to give a linear oligosulfide followed by oxidation of the oligosulfide with sodium periodate to give an oligosulfoxide. The challenge of complete avoidance of partial oxidation and over oxidation was easily overcome using the sodium periodate oxidation conditions. Although sulfoxide is a highly polar functional group, the oligosulfoxides were found to have limited solubility in many solvents including DMSO and water, which disobeys the “like dissolves like” rule. The surprising solubility pattern of oligosulfoxides was discussed in the context of the drastically different solubility patterns of polyethylene glycol (PEG), poly(butylene oxide), and poly(methylene oxide). According to a dissolution model, solubility properties of linear oligomers including the oligosulfoxides and PEGs may be heavily affected by their conformations and the suitability of their conformations in water for maximizing attractive interactions between them and water. Based on these hypotheses, the limited solubility of the present oligosulfoxides may not imply the low solubility of similar molecules.

Graphical Abstract

Introduction

Simple linear molecules containing multiple sulfoxide groups, although have not yet received much attention from synthetic chemists, are important on several fronts. Many plants and fungi including chives, garlic, onion and edible mushroom contain such compounds, and are believed to be responsible for their pungent aroma and flavor.¹ In human, several metabolites of the chemical warfare agent bis(2-chloroethyl)sulfide, which is more commonly called sulfur mustard and has a US military code HD, contain multiple sulfoxide groups. These metabolites can be detected in urine and can serve as biomarkers for diagnosis of HD exposure.^{2, 3} In chemistry, simple linear compounds containing multiple sulfoxides have been used in phase transfer catalysis^{4, 5} and as ligands in organometallic chemistry.⁶ Recently we became interested in evaluating the suitability of simple linear oligosulfoxides for applications similar as polyethylene glycols (PEGs) in biomedical research and medicine. The sulfoxide group may be even more biocompatible than the dialkyl ether group. For example, methionine – one of the essential amino acids – can exist as methionine sulfoxide in living systems.^7–10 DMSO is widely used as a solvent in drug delivery.^{11, 12} Therefore, there is a good chance that simple linear oligosulfoxides are also biocompatible. Because the sulfoxide group is highly polar, according to the “like dissolves like” rule, oligosulfoxides should be highly soluble in water and other polar solvents. In addition, it was reported that surfaces coated with oligosulfoxides were resistant to non-specific protein adsorption.^{13, 14} With these analyses, simple linear oligosulfoxides could have many properties of PEGs that are needed for biomedical applications. Therefore, like PEGs, linear oligosulfoxides could be conjugated to drugs to increase their solubility in water. They may also be used in nanomedicine as coating agents for nanoparticle stabilization and as hydrophilic linker for bioconjugation.^{15, 16} Recently PEGs were found to be immunotoxic.^{17, 18} Linear oligosulfoxides or their co-polymers with PEGs may be able to provide a solution for the problem. For these reasons, we decided to synthesize several model linear oligosulfoxides to evaluate the solubility pattern of this class of potentially highly useful compounds.

Results and discussion

Our first target molecule was set to be compound 1a (Scheme 1). This molecule contains six sulfoxide groups, one dialkyl ether group and one hydroxyl group, and should be highly polar and potentially highly soluble in polar solvents including water. Such simple linear oligosulfoxides having more than four sulfoxide groups have not been reported in the literature.^{13, 14} Our proposed synthesis route relied on the facile S_N2 reactions between an alkanethiol and alkyl tosylate to form an oligosulfide followed by oxidation although such a route had the potential problem of incomplete oxidation and over oxidation.¹⁹ As shown in Scheme 1, compound 2²⁰ was selectively protected with TBDPS-Cl to give 3 in 68% yield. Compound 3 was then tosylated to give 4, which was coupled via S_N2 reactions with 1,2-ethanedithiol to give 5 using potassium tertiary butoxide as the base. As expected, due to the high acidity of thiols (pKa ~11 as compared to pKa ~15 of ethylene glycol) and the soft nature of the thiolate nucleophiles, the reaction proceeded at room temperature with high yield (87%). No elimination product of the tosylate was observed. Next, the TBDPS groups were removed with TBAF to give 6, which was selectively methylate at one end to give compound 7. Despite the unavoidable formation of the dimethylated symmetric side product, we were able to obtain 7 in 75% yield. Compound 7 was then protected with TBDPS-Cl again to give 8, which was then oxidized with 6.6 equivalents sodium periodate at room temperature over 48 hours to give the target compound 9. An isolate yield of 83% was obtained. As indicated by MS (Supporting Information), all the six sulfide groups in 8 were oxidized to sulfoxides. None of them were over oxidized to sulfones. With compound 9, the highly hydrophilic target molecule oligosulfoxide 1a was obtained by treating with excess triethylamine trihydrofluoride. Excess triethylamine was also added to ensure that the reaction conditions were not too acidic, under which the sulfoxides could be further oxidized to sulfones.²¹ At the end of the reaction, the excess fluoride ion was quenched with excess methoxytrimethylsilane, which converted the fluoride ion to the non-toxic and volatile fluorotrimethylsilane. This allowed easy removal of all excess reagents and side products.^{22, 23} Compound 1a was given as white solid. As indicated by its MS (Supporting Information), compound 1a has six sulfoxide groups. No impurities containing sulfides or sulfones were detectable. ¹H and ¹³C NMR also showed that the compound was highly pure (Supporting Information).

Scheme 1.

Synthesis of oligosulfoxide 1a.

It is noted that controlled oxidation of sulfide to sulfoxide without over oxidation of sulfoxide to sulfone had been documented to be challenging.^{19, 24–26} Obviously, this problem is predicted to be more serious for controlled oxidation of compounds that contain multiple sulfides. Fortunately, in our synthesis during the conversion of 8 to 9, we did not meet much trouble. Our lab has been using sodium periodate for the oxidation of sulfide for several other projects, and we knew this oxidation reaction is highly efficient.^27–32 Therefore, we believed that there was a good chance for us to achieve controlled oxidation by using exact equivalents of this oxidizing agent. Indeed with slightly excess sodium periodate, we were able to oxidize all the six sulfide groups to sulfoxides efficiently without over oxidizing any of them to sulfones. During the synthesis of 1a, we also tried to oxidize 7 to 1a directly under similar conditions without going through compounds 8 and 9. Probably due to the low solubility of partially oxidized compounds that contained one to six sulfoxide groups during the reaction as indicated by the low solubility of 1a described later, the oxidation could not be controlled well. We got mixtures of compounds with insufficient oxidation and over oxidation. For this reason, we protected 7 with the hydrophobic TBDPS group to increase the solubility of the intermediates during the oxidation reaction. With this strategy, we were able to control the oxidation very well and pure compound 9 could be obtained. In compound 1a, there are three methylene groups between the second and third sulfoxide groups from either end of the molecule. Two methylene groups would make the molecule even more hydrophilic and therefore more preferred. The reason for us to use three was the instability of the molecule corresponding to 4 needed for the synthesis. When we followed a reported procedure to synthesize the compound,³³ it decomposed during flash column chromatography and we were unable to identify a feasible method for its purification. The instability of the molecule corresponding to 4 that has a sulfur atom at the position β to the electrophilic carbon of a tosylate may be caused by the ease of formation of a cationic thiiranium ion intermediate via an intramolecular S_N2 reaction.

With pure compound 1a in hand, we evaluated its solubility in common organic solvents and water. It was surprising that 1a is virtually insoluble in many organic solvents and has limited solubility in highly polar solvents including DMSO and water. As shown in Table 1, the compound has less than 0.2 mg/ml (~0.02%) solubility in non-polar solvents including hexanes, toluene, and diethyl ether (entries 1-3). This was within expectation considering the high polarity of 1a and the “like dissolves like” rule. However, the compound also has less than 0.2 mg/ml solubility in relatively polar solvents including dichloromethane, ethyl acetate, and THF (entries 4-6). What was surprising was that the compound also has limited solubility in highly polar solvents such as acetonitrile, acetone, methanol, DMF and DMSO (entries 7-11). Even in water, the solubility of the compound is only 150 mg/mL, which corresponds to 13% (entry 12). This solubility pattern is drastically different from PEGs, which have good solubility in all the above solvents except for hexanes and diethyl ether. In particular, PEGs with molecular weights similar to 1a, which exist as liquid, are miscible with water, while compound 1a has a solubility of only 13%.

Table 1.

Solubility of oligosulfoxides 1a-c in common solvents.

Entry	Solvent	Solubility of 1a	Solubility of 1b	Solubility of 1c
1	Hexanes	< 0.2 mg/mL	< 0.2 mg/mL	< 0.2 mg/mL
2	Toluene	< 0.2 mg/mL	< 0.2 mg/mL	< 0.2 mg/mL
3	Diethyl ether	< 0.2 mg/mL	< 0.2 mg/mL	< 0.2 mg/mL
4	Dichloromethane	< 0.2 mg/mL	< 0.2 mg/mL	43 mg/mL (3.1%)
5	Ethyl acetate	< 0.2 mg/mL	< 0.2 mg/mL	< 0.2 mg/mL
6	THF	< 0.2 mg/mL	33 mg/mL (3.6%)	68 mg/mL (7.1%)
7	Acetonitrile	21 mg/mL (2.6%)	20 mg/mL (2.5%)	22 mg/mL (2.7%)
8	Acetone	18 mg/mL (2.2%)	30 mg/mL (3.9%)	40 mg/mL (4.9%)
9	Methanol	16 mg/mL (2.0%)	24 mg/mL (2.9%)	175 mg/mL (18.1%)
10	DMF	30 mg/mL (3.1%)	45 mg/mL (4.6%)	36 mg/mL (3.7%)
11	DMSO	10 mg/mL (0.9%)	31 mg/mL (2.7%)	140 mg/mL (11.3%)
12	Water	150 mg/mL (13.0%)	305 mg/mL (23.4%)	> 2 g/mL (> 66.7%)

To investigate the relationship between the number of sulfoxide groups in linear oligosulfoxides and solubility of the compounds, oligosulfoxides 1b-c, which have five and three sulfoxide groups, respectively, were synthesized. Similar conditions for the synthesis of 1a were used for the synthesis (Scheme 2). The solubility of the compounds was determined. As shown in Table 1, similar as 1a, compound 1b has limited solubility in hexanes, toluene, diethyl ether, dichloromethane and ethyl acetate (entries 1-5). It has slightly better solubilities than 1a in the more polar solvents THF, acetone, methanol, DMF and DMSO (entries 6 and 8-11), and similar solubility as 1a in acetonitrile (entry 7). Compound 1c, which has only three sulfoxide groups, also has limited solubility in hexanes, toluene, diethyl ether and ethyl acetate (entries 1-3 and 5), but its solubility in dichloromethane, THF, acetone, methanol and DMSO (entries 4, 6, 8-9 and 11) is significantly higher. In acetonitrile and DMF, its solubility is similar as 1a-b (entries 7 and 11). Similar as 1a, both 1b-c have the highest solubility in water although the solubility of 1b is still far less than PEGs (entry 12). Among the three oligomers, 1b is much more soluble than 1a, and 1c is far more soluble than 1b. Therefore, in water, solubility decreases with increased number of sulfoxide groups in the linear oligosulfoxides. The same trend can be seen in most of the other solvents for the three compounds as well.

Scheme 2.

Synthesis of oligosulfoxides 1b-c.

Although the far lower than expected solubility of 1a-b in water are difficult to explain, we would like to give some insights on the factors that could be at play. We assume that dissolution of linear oligosulfoxides in water (and similarly in other solvents) involves the following steps: (1) Converting the oligomers in their pure solid or liquid state to ideal gas state without changing their conformations. This step has positive change of enthalpy and positive change of entropy. (2) Converting the conformations of the oligomers in their pure solid or liquid state to their preferred conformations in ideal gas state. This step has zero or negative change of enthalpy. (3) Converting the conformations of the oligomers in ideal gas state to their preferred conformations in water. The preferred conformations of the oligomers in water are not necessarily the preferred conformations in gas state or the conformations that would favor the attractive interactions between the oligomers and water molecules to the highest extend. Instead, they are the results of a balance between the two. This step has zero or positive change of enthalpy. (4) Converting the arrangement of water molecules in their pure state to the arrangement in the final state of oligomer solution. This step provides the spaces in water for molecules of the oligomers to fit in, and has negative changes of entropy and positive changes of enthalpy. (5) Placing the oligomers into the spaces in water. This step involves installing net attractive interactions between the oligomers and water molecules such as van der Waals forces, dipole-dipole forces and hydrogen bonding. It has negative change of enthalpy.

Based on the above dissolution model, the drastically lower solubility of 1a-b in water than DMSO may be a result of steps (1), (3) and (5). In step (1), in terms of each sulfoxide group, 1a-b would have less entropy gain than DMSO. In step (3), 1a-b would most likely have a positive change of enthalpy, while DMSO would most likely have a zero change of enthalpy. In step (5), the specific conformations adopted by 1a-b in water could impose barriers for optimal attractive interactions (e.g. dipole-dipole interaction and hydrogen bonding) between the molecules of oligomers and water, while DMSO is a smaller molecule, and its conformation is less likely to do so. The dissolution model can also provide insights on the much lower solubility of 1a-b in water than PEGs. In this case, the difference in step (1) is most obvious because the sulfoxide groups in 1a-b are much more polar and have a much stronger dipole-dipole attractive force than the dialkyl ether groups in PEGs. As a result, it takes much more energy to separate 1a-b molecules than PEG molecules in this step. However, it is possible that steps (3) and (5) may play an even more important role for the different solubilities of 1a-b and PEGs. In step (3), it might take less energy for PEGs than for 1a-b to convert their conformations in ideal gas state to those in water. In step (5), the conformation of PEGs in water might be more ideal for dipole-dipole interaction and hydrogen bonding with water than that of 1a-b.

The drastically different solubility patterns of PEGs from their analogous oligomers including poly(butylene oxide) and poly(methylene oxide) (or paraformaldehyde), which differ from PEGs by only one methylene group, can also provide important insights on the solubility patterns of 1a-c. PEGs are highly soluble in water and many other polar organic solvents such as toluene, dichloromethane, THF, acetonitrile, acetone, methanol, DMF and DMSO. However, poly(butylene oxide) and poly(methylene oxide) are hydrophobic and insoluble in water.³⁴ In the context of the above dissolution model, the changes of enthalpy and entropy in steps (1), (2) and (4) for the cases of PEGs, poly(butylene oxide) and poly(methylene oxide) should be similar. Therefore, the difference in solubility between them would most likely come from steps (3) and (5). In step (3), it might take little or no energy for PEGs to change their conformation in the ideal gas state to their conformation in water, while it might take more energy for poly(butylene oxide) and poly(methylene oxide) to do so. In other words, the conformations of PEGs in ideal gas state and in water might be similar. In step (5), the conformation of PEGs in water might be more suitable for attractive interactions (e. g. dipole-dipole interactions and hydrogen bonding) between PEG molecules and water than those of poly(butylene oxide) and poly(methylene oxide). For the solubility of 1a-b in water, the situation is more like that for poly(butylene oxide) and poly(methylene oxide), and not like that for PEGs.

According to the above dissolution model and analysis, conformations of the linear oligomers such as 1a-c, PEGs, poly(butylene oxide) and poly(methylene oxide) in their pure state and in water (and other solvents as well) may play a critical role on their solubility. In step (3), the change of enthalpy for converting the conformation of an oligomer in ideal gas state to that in water may be so high that the energy cannot be compensated by the other steps. In step (5), the conformation of an oligomer in water may be so unfavorable for attractive interactions with water that little driving forces can be provided for the dissolution process. In contrast, the situation is very different for smaller molecules such as DMSO and diethyl ether. In those cases, conformation may play little roles on solubility. In step (3), the change of enthalpy is small, and in step (5), conformation has little effect on the attractive forces between the molecules and water. Therefore, for smaller molecules, their solubility is more affected by step (1), and the attractive forces between them and water without unfavorable effects from their conformation in step (5). As a result, for small molecules, the “like dissolves like” rule is usually held, while for linear oligomers, the rule may not be obeyed.

Usually, the change of enthalpy for converting a molecule from one conformation to another is unconsciously considered small. Such a perception is reasonable because most conformation changes can occur rapidly at room temperature. Therefore, it is natural to underestimate the effect of conformations on solubility. However, the heat of dissolution of many substances in water (e.g. −17.6 kJ/mol for DMSO) and the strength of hydrogen bonds (e.g. ~20 kJ/mol for –O-H•••O–) are also low in values, and the values are close to the difference of enthalpies for different conformations of many molecules (e.g. ~25 kJ/mol for full-eclipsed and anti-staggered conformations of butane). Other intermolecular forces such as van der Waals forces and dipole-dipole forces are even weaker than hydrogen bonding. Therefore, the notion that the changes of enthalpy for conformational transitions of linear oligomers during the dissolution process are critical factors that determine the solubility of the molecules is reasonable.

Considering the importance of conformations of linear oligomers on their solubility, the low solubility of 1a-b may not necessarily indicate low solubility of other similar oligosulfoxides. The reason is that conformations are easily affected by subtle structural variations. Therefore, the present work should not be used to prevent future studies on searching more soluble linear oligosulfoxides. Even if all linear oligosulfoxides have limited solubility in water, which are unlikely, due to their other PEG-like properties such as linearity, flexibility, neutrality, high polarity, resistance to protein binding, being only hydrogen bond acceptor, and potential high biocompatibility, they may still be valuable in biomedical research and medicine. In particular, if installation of an oligosulfoxide fragment into an immunogenic site of PEGylated drugs or nanocarriers of drugs and diagnostic tools could prevent undesirable immune responses, the emerging PEG immunotoxicity problem could be overcome by the use of oligosulfoxides.^{17, 18}

Conclusion

In summary, three simple linear oligosulfoxides containing up to six sulfoxide groups have been synthesized. Using the facile sulfide-formation S_N2 reaction between an alkanethiol and alkyl tosylate, the backbones of these molecules were easily assembled. Controlled oxidation of the multiple sulfides to sulfoxides was achieved with sodium periodate. Side products with insufficient oxidation or over oxidation could be completely avoided. As expected, the oligosulfoxides (1a-c) are insoluble in non-polar solvents. What was surprising was that the solubility of 1a-b, which contain six and five highly polar sulfoxide groups, respectively, in polar solvents including DMSO and water is also low, and is far lower than PEGs with similar molecular weight. The unusual solubility patterns of 1a-b in water are discussed in the context of the drastically different solubility properties of PEG, poly(butylene oxide) and poly(methylene oxide).³⁴ According to a dissolution model, conformational changes of the oligomers during the dissolution process, and the suitability of the conformations of the oligomers in water for maximizing attractive interactions between the oligomers and water (e.g. dipole-dipole interaction and hydrogen bonding) are critical factors that determine the solubility patterns of the oligomers. Due to these conformation factors, linear oligomers are less likely to obey the “like dissolves like” rule, which is different from small molecules. For the latter, conformation plays little roles, and small molecules are more likely to obey the “like dissolves like” rule. Finally, even if oligosulfoxides have limited solubility in water, due to their other PEG-like properties, they may still find wide applications in medicine.