RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Interaction of Model Aryl- and Alkyl-Boronic Acids and 1,2-Diols in Aqueous Solution WILLIAM A. MARINARO,1 RICHARD PRANKERD,2 KAISA KINNARI,1 VALENTINO J. STELLA1 1 2

Department of Pharmaceutical Chemistry, The University of Kansas, Lawrence, Kansas 66047 Monash Institute of Pharmaceutical Sciences, Monash University, Parkville, Victoria 3052, Australia

Received 1 October 2014; revised 18 November 2014; accepted 11 December 2014 Published online in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.24346 ABSTRACT: The goal of this work was to quantitate ester formation between alkyl and aryl boronic acids and vicinal-diols or 1,2-diols in aqueous solution. As used here, 1,2-diols includes polyols with one or more 1,2-diol pairs. Multiple techniques were used including apparent pKa shifts of the boronic acids using UV spectrophotometry (for aryl acids) and titration (for aryl and alkyl acids). Isothermal microcalorimetry was also used, with all reactions being enthalpically favored. For all the acids and 1,2-diols and the conditions studied, evidence only supported 1:1 ester formation. All the esters formed were found to be significantly more acidic, as Lewis acids, by 3–3.5 pKa units than the corresponding nonesterified boronic acid. The equilibrium constants for ester formation increased with increasing number of 1,2-diol pairs but stereochemistry may also play a role as sorbitol with five possible 1,2-diol pairs and five isomers (taking into account the stereochemistry of the alcohol groups) was twice as efficient at ester formation compared with mannitol, also with five possible 1,2-diol pairs but only three isomers. Alkyl boronic acids formed esters to a greater extent than aryl acids. Although some quantitative differences were seen between the various techniques used, rank ordering of the structure/reactivity was consistent. Formulation implications of ester C 2015 Wiley Periodicals, Inc. and the American Pharmacists Association J formation between boronic acids and 1,2-diols are discussed.  Pharm Sci Keywords: esters; boronic acid; mannitol; stability; 1,2-diols,; acid base equilibria; isothermal calorimetry; structureproperty relationship (SPR)

INTRODUCTION The objective of this study was to determine the equilibrium constants for the reversible esterification of model aliphatic and aromatic boronic acids with vicinal-diols or 1,2-diols. As used here, 1,2-diols includes polyols with one or more 1,2-diol pairs. Various experimental techniques were used to quantitate the equilibrium constant for ester formation, including the shift in apparent pKa (pKa ) values determined using UV spectral changes with pH of aromatic acids in the presence of 1,2diols; by titrimetry; and enthalpic changes on ester formation assessed by isothermal microcalorimetry. Boron nuclear magnetic resonance (NMR) was also used to directly detect ester formation. Scheme 1 shows the reaction scheme for the ionization of a boronic acid in the presence of a diol. A general scheme would be seen for any aliphatic and/or aromatic boronic acids in the presence of various 1,2-diols. Boronic acids are Lewis acids in that they do not give up a proton but are acidic by virtue of having an electron-pair acceptor boron atom. They are therefore able to interact through a pair of electrons on a base like a hydroxide ion or release a proton on reaction with water. See the two ionization equilibria in Scheme 1.

The interest in boronic acid derivatives as drugs has increased since the introduction and approval of bortezomib, a drug used to treat multiple myeloma,1 and boronophenylalanine, or BPA,2 a clinically useful agent for boron neutron capture cancer therapy. The low aqueous solubility and poor chemical stability of bortezomib was an issue in its formulation development.3–8 For example, the intravenous and current subcutaneous formulation of bortezomib is a lyophilized formulation from 1% mannitol where mannitol ester formation was confirmed.3–5 The solubility of bortezomib in water is about 0.5– 0.6 mg/mL, whereas in the presence of 1% mannitol, a solubility of 4 mg/mL has been reported because of ester formation.4,5 Except for the work of Yan et al., Springsteen and Wang, and a few others,9–16 relatively limited quantitative information on boronic acid ester formation with 1,2-diols is available. However, the interaction of boronic acid derivatives and sugars has long been used for separation and analytical purposes as well as in designing drug delivery devices, especially those used to sense glucose changes. This work addresses the relationship between the structures of the 1,2-diols and boronic acids and ester formation in aqueous solution.

EXPERIMENTAL Materials

Correspondence to: Valentino J. Stella (Telephone: +785-864-3755; Fax: +785864-5736; E-mail: [email protected]) William A. Marinaro’s present address is Formulation Sciences, Merck and Company, Inc., Summit, New Jersey 07901. Kaisa Kinnari’s present address is Tampere University Hospital, Department of Pharmacy, Tampere 33521, Finland Journal of Pharmaceutical Sciences  C 2015 Wiley Periodicals, Inc. and the American Pharmacists Association

All chemicals were analytical grade or ACS reagents and used without further purification unless otherwise noted. Glycerol, potassium hydroxide (pellets) (KOH), potassium hydrogen phthalate (99+%) (KHP), ethylene glycol (enzyme grade), and D-sorbitol (laboratory grade) (sorbitol) were purchased from Fisher Scientific Company (Fair Lawn, New Jersey). Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

1

2

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Equipment

Scheme 1. Reaction scheme for the ionization of a boronic acid in the presence of a diol. A general scheme would be seen for any aliphatic and/or aromatic boronic acids in the presence of various 1,2-diols.M.

A Lambda 6 UV/VIS Spectrophotometer with MultiCal UV Data Manager software was from PerkinElmer Inc. (Waltham, Massachusetts). A DL53 Titration System with LabX Light Titration V 2.1 software was from Mettler-Toledo GmbH (Schwerzenbach, Switzerland). An Isothermal Titration Calorimeter Model CSC 4200 with ITCRun system control software and BindWorks 3.1 data analysis software was from Calorimetry Science Corporation (Lindon, Utah). Bruker Avance 400 MHz Nuclear Magnetic Resonance Spectrometer equipped with a broadband probe with TopSpin 1.2 software was from Bruker BioSpin GmbH (Rheinstetten, Germany). Quartz NMR tubes were purchased from New Era Enterprises, Inc. (Vineland, New Jersey). Methods

Phenylboronic acid (97%), isobutylboronic acid (97%) (IBA), and cyclohexylboronic acid (96%) (CyHBA) were purchased from Frontier Scientific Inc. (Logan, Utah), and ethylboronic acid (98%) (EtBA) was purchased from Alfa Aesar (Pelham, New Hampshire). Xylitol (99+%) and meso-erythritol (99%) (erythritol) were purchased from Acros Organics (Geel, Belgium). The diol, 2,3-butanediol (mixture of stereoisomers) was purchased from TCI America (Portland, Oregon), 1,3-propanediol (99.6+%) was purchased from Sigma–Aldrich Company (St. Louis, Missouri), whereas 4-methoxybenzeneboronic acid (98%) (4-MBBA) was purchased from Lancaster Synthesis Ltd. (Windham, New Hampshire). Ultrapure water (UHP water) was produced by refluxing Milli-Q water for more than 45 min with dilute alkaline potassium permanganate solution. This was then distilled with a 1.6-m vacuum-jacketed spinning band column at a rate of 1 L/8 h. The structures of the various 1,2-diols and the various boronic acids studied are shown in Figure 1.

Boronic Acid Apparent pKa Value Change with 1,2-Diol Concentration Apparent pKa Measurement with UV Spectrophotometry. The aromatic boronic acid, 4-MBBA was dissolved in 200 mL of both 320 mM NaOH and 320 mM HCl to a concentration of 100 :M. For experiments performed in the presence of a polyol, the subject polyol was dissolved in the NaOH and HCl solutions to the stated concentration, generally between 1 mM and 1 M. The solutions were then titrated with one another under magnetic stirring to pH values of generally 2–12, by monitoring with an ARIS pH meter with a calibrated microelectrode. Aliquots of 2 mL were taken at intervals of approximately 1 pH unit; pH values were measured and recorded to two significant figures. The samples were then analyzed by UV spectrophotometry in quartz cuvettes with a 500-:L capacity and a 1-cm path length. Because of the large change in UV absorbance at a

Figure 1. Structures of various 1,2-diols and boronic acids used to evaluate equilibrium ester formation in aqueous solution. Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

DOI 10.1002/jps.24346

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

3

50 mM, and analyzed by potentiometric titration according to the method as described above. To determine whether an adjustment in the time between aliquots had altered the estimated pKa value of a boronic acid in the presence of mannitol, samples of 5 mM 4-MBBA in 0.5 M mannitol were analyzed by potentiometric titration according to the method described above; one set of samples was analyzed by a method that altered the aliquot addition rate from 5–20 to 30–45 s. Direct Measurement of Boronic Acid to 1,2-Diol Esterification

Figure 2. UV spectra of 100 :M 4- 4-MBBA under acid conditions (solid line), and under alkaline conditions (dashed line).

wavelength of 235 nm between the neutral and the anionic form of 4-MBBA, as illustrated in Figure 2, the absorbance of each sample was determined in triplicate at this value. The pH values of each sample were checked after analysis to make sure that pH values had not deviated during the experimental time frame. Apparent pKa Measurement with Potentiometric Titration The boronic acid of interest was dissolved in water or a solution of 1,2-diols in water. The concentration of boronic acid in the sample was 5 mM unless otherwise stated, and the concentration of 1,2-diols generally ranged from 0.01 to 0.5 M. The KOH solution was made from KOH pellets, which were washed with water to remove surface carbonates and the solution was kept under argon. The KOH solution concentration of approximately 0.20 M was standardized by endpoint titration with KHP on the day of the titration experiments. The potentiometric titrations were performed in a titration vessel containing 25 mL of sample solution; it was maintained at 25◦ C via the water-jacketed vessel with circulating temperature-controlled water. The solution was stirred during titrations at a rate set to 25% of maximum for the instrument. The pH electrode was calibrated the day of the titration experiments at pH values of 4, 7, and 10. Aliquots of KOH solution were added at a rate of one addition every 5–20 s as determined by an algorithm in the method that measured the stability of the pH meter reading; in this method, the operator-controlled variable, labeled dE[mV], was set to 0.5. The volume of the aliquot was between 5 and 50 :L as determined by an algorithm in the method that targets aliquot volumes that will evenly space pH endpoints; in this method, the operator-controlled variable, labeled dE(set) [mV], was set to 4.0. The total titrant (KOH) volume added and the pH for that volume were recorded and analyzed. Titrations were run in duplicate unless otherwise noted. To determine whether the concentration of the boronic acid in the sample has an effect on the estimated pKa of the boronic acid, samples of IBA in water were made to 1, 5, 10, and DOI 10.1002/jps.24346

Isothermal Titration Microcalorimetry. A boronic acid of interest was dissolved in 50 mM NaOH to a concentration of 1 mM unless otherwise stated. This sample solution was then loaded into the clean isothermal titration microcalorimetry (ITC) reactor cell (volume, 1275 :L). The titrant solution consisted of the 1,2-diols of interest at a concentration of 15–200 mM in 50 mM NaOH; it was degassed under negative pressure with magnetic stirring. The titrant solution was then loaded into a clean dry 250-:L titrant syringe. The syringe was fitted to the ITC, the stirrer was then set to run at 300 rpm, and the sample cell was allowed to equilibrate. Aliquots were added at a volume of 5–10 :L to a total titrant solution volume of 250 :L, generally 10 :L aliquots by 25 injections. The injections were spaced at 400 s equilibration times unless otherwise noted. The reference cell contained UHP water for all experiments. Experiments were run in duplicate unless otherwise noted. The titrant and sample solutions were kept under argon during storage at room temperature. For each experiment, an analogous blank titration was run in which the same titrant solution was titrated into 50 mM NaOH rather than the sample solution. These blanks accounted for the heat of dilution of the titrant, and the peak area for the blank was then subtracted from the peak area for the experiment prior to data analysis.

Ionic Strength Effect. An ITC experiment was run to determine the effect of ionic strength on ester formation of 4-MBBA with mannitol. Titrations were performed with mannitol as the titrant and 4-MBBA as the sample as described above; however, the ionic strength was altered. Instead of the titrant and sample solutions being made with 50 mM alone, solutions were made with 50 mM NaOH in addition to 0.055 or 0.135 M NaCl to bring the final ionic strength to 0.0105 and 0.185, respectively. The equilibrium constants generated from these experiments were compared with the titrations performed in 50 mM NaOH, which has an ionic strength of 0.05. Boron-11 NMR Solutions of 120 mM EtBA were made with 0.32 M NaOH and 0.32 M HCl and titrated with one another to a pH of 12.0. These solutions were made with varying concentrations of mannitol; 0, 30, 60, 90, and 110 mM. The samples were then placed in quartz NMR tubes and analyzed with the Bruker NMR spectrometer described above with the broadband probe tuned to boron-11. All spectra were referenced to boron trifluoride etherate (BF3 ·Et2 O) in methylene chloride (CH2 Cl2 ) at 0 ppm. The resulting peaks in the spectra were deconvoluted with TopSpin 1.2 software and integrated. Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

4

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

RESULTS AND DISCUSSION Effect of 1,2-Diols on Apparent pKa Values of Boronic Acids Variation in Apparent pKa Values with 1,2-Diol Concentration Using UV Spectrophotometry The Henderson–Hasselbalch equation for a weak acid (Eq. (1)) relates: pH = pKa + log

[base] [acid]

(1)

the pH of a solution to the concentration of an acid and its conjugate base in that solution through the pKa for that acid. As previously shown in Figure 2, there is a large difference in the molar absorption of 4-MBBA at 235 nm between the neutral and anionic species. This spectral change may be substituted into Eq. (1) for acid and base concentration to yield Eq. (2). pH = pKa + log

absmax − abssample abssample − absmin

(2)

where absmax is the maximum absorbance of 4-MBBA at the studied concentration at one extreme of pH, absmin is the minimum absorbance of 4-MBBA at the other extreme of pH at the studied concentration, and abssample is the actual absorbance of 4-MBBA at a particular pH between the extremes. The results for one titration may then be graphed with absorption at 235 nm on the y-axis and pH on the x-axis to yield a sigmoidal plot as in Figure 3. This plot may then be fit to Eq. (2) to yield a pKa value of 9.15. This same technique was applied to solutions of 4-MBBA with various concentrations of 1,2-diols, as an example the pKa determination of 4-MBBA in the presence of 5 mM mannitol is also illustrated in Figure 2 yielding an apparent pKa, pKa , value of 7.99.

Figure 4. pKa of 4-MBBA in aqueous solution versus the mannitol concentration in that solution from the UV spectrophotometric method. Insert shows the plot of Ka,obs versus the mannitol concentration in the same solutions. Squares are the raw data and the solid line is the curve fit using Eqs. 7 and 8, allowing the estimates of Ka1 , Ka2 , KB1 , and KB2 .

The pKa values obtained for 4-MBBA in the presence of varying concentrations of mannitol are plotted in Figure 4 on the y-axis and the concentration of mannitol on the x-axis. As the concentration of mannitol increased the pKa decreased sharply, but the value reached a minimum at an approximate pKa value of approximately 6.2. This pKa value corresponded to the pKa for the 4-MBBA mannitol ester, and the shape of the curve suggested an equilibrium process, as expected. It would be advantageous to mathematically describe this process. The covalent equilibrium formation of 1,2-diol esters between 1,2-diols and boronic acids is illustrated by Scheme 1. The individual equilibrium expressions, which govern these processes, are expressed in Eqs. (3)–(6). [− BOH][H+ ] [B]

(3)

[− BPOH][H+ ] [BP]

(4)

[BP] [B][P]

(5)

Ka1 = Ka2 =

K B1 = K B1 =

Figure 3. Absorbance at 235 nm versus pH of 100 :M 4-MBBA in aqueous solution alone (circles, with curve fit in solid line) with a pKa of 9.15, and 4-MBBA in the presence of 5 mM mannitol (squares, with curve fit in dashed line) with a apparent pKa, pKa , of 7.99. Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

[− BPOH] [− BOH][P]

(6)

where Ka1 is the acid dissociation constant of the free boronic acid (acting as a Lewis acid), Ka2 is the acid dissociation constant of the boronic acid ester, KB1 is the equilibrium constant between the neutral boronic acid and the 1,2-diol, and KB2 is the equilibrium constant between the anionic boronic acid and the 1,2-diol; [− BOH] is the molar concentration of free boronate anion, [H+ ] is the molar concentration of the hydronium ion, [B] is the molar concentration of the free neutral boronic acid, [− BPOH] is the concentration of the anionic boronate 1,2-diol DOI 10.1002/jps.24346

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

5

ester, [BP] is the molar concentration of the neutral boronic acid 1,2-diol ester, and [P] is the molar concentration of the free 1,2-diol. Thus, the observed or apparent acid dissociation constant (Ka,obs ) may be expressed as Eq. (7). K a,obs =

([− BOH] + [− BPOH])[H+ ] [B] + [BP]

(7)

In turn, Eqs. (3)–(6) may be substituted in to Eq. (7) to yield Eq. (8): K a,obs =

Ka1 + Ka2 K B1 [P] 1 + K B1 [P]

(8)

which assumes that the total 1,2-diol concentration is approximately equal to the free polyol concentration. This is the case when the 1,2-diol molar concentration is much greater than the boronic acid concentration and the magnitude of the equilibrium constants are moderate, as it appears in this work. Using mannitol as an example, the data were plotted and the fit is illustrated in Figure 4, inset. An acid dissociation constant (Ka ) for the 4-MBBA mannitol ester of 6.70 × 10−7 , which corresponded to a pKa value of 6.17, was estimated. In the same fit, the equilibrium constant between the neutral species (KB1 ) was determined; in this case, the value was 10.4 M−1 . Using the relationship expressed in Eq. (9), the equilibrium constant between the anionic species may be determined; this was found to be 9.7 × 103 M−1 . K B2 =

Ka2 K B1 Ka1

(9)

This fitting procedure was sufficient for what might be called high-equilibrium constant 1,2-diols such as mannitol, xylitol, and sorbitol (see later results). In the case of weaker interacting 1,2-diols, such as erythritol, glycerol, or ethylene glycol, KB1 is a low number such that the denominator in Eq. (8) approaches unity, and Ka,obs may be expressed as the linear Eq. (10). K a,obs = Ka1 + Ka2 K B1 [P]

(10)

Thus, when Ka,obs was plotted versus 1,2-diol concentration the slope of the line was the product of Ka2 and KB1 . In order to solve for KB1 , it was assumed that Ka2 was the same for all 4-MBBA 1,2-diol esters. This assumption is reasonable based on observations with high-equilibrium 1,2-diols and literature reports.10,12 The equilibrium constant values thus determined for a series of 1,2-diols are listed in Table 1. This series consists of linear Table 1. Equilibrium Constants and SDs for Ester Formation with Both Anionic 4-MBBA (KB2 ) and Neutral 4-MBBA (KB1 ) with a Series of Linear 1,2-Diols in Aqueous Solution According to Scheme 1 Polyol (Number of 1,2-Diol Pairs)

KB1 (M−1 ) ± SD

KB2 (M−1 ) ± SD

Mannitol (5) Xylitol (4) Erythritol (3) Glycerol (2) Ethylene glycol (1) 2,3-Butanediol (1)

10.4 ± 3.0 5.8 ± 1.7 3.1 ± 0.2 × 10−1 4.5 ± 0.2 × 10−2 2.8 ± 0.1 × 10−3 3.8 ± 0.2 × 10−2

9.7 ± 2.8 × 103 5.4 ± 1.5 × 103 2.8 ± 0.2 × 102 41.5 ± 1.8 2.6 ± 0.1 35.1 ± 2.0

DOI 10.1002/jps.24346

Figure 5. A semilog plot of the equilibrium constants (SDs, within size of symbol) for ester formation with both anionic 4-MBBA, SQUARES) AND NEUTRAL 4-MBBA CIRCLES) WITH A SERIES OF LINEAR 1,2-DIOLS IN AQUEOUS SOLUTION VERSUS NUMBER OF 1,2-DIOL PAIRS, USING DATA TAKEN FROM TABLE 1. THE OPEN SYMBOLS ARE FOR 2,3-BUTANEDIOL.

nonreducing carbohydrates or 1,2-diols of increasing size, beginning with ethylene glycol, and continuing to mannitol. A significant increase in the equilibrium constants with an increase in the number of 1,2-diol pairs is seen. A semilog plot of KB1 and KB2 versus the number of 1,2-diol pairs is shown in Figure 5. A surprisingly linear structure– reactivity relationship is readily seen with mannitol showing a small negative deviation. An alternative plot using just the number of hydroxyls showed an identical trend. There are several possible explanations for these observations. Increased bulk on the polyol backbone may favor the formation of boronic acid esters for many reasons. Using a kinetic argument, the bulky backbone should do little to slow the forward reaction, the attack of the hydroxyls on the boron atom to form the ester, especially if the backbone contains hydroxyls. However, it will serve to inhibit the reverse reaction, the attack of water on the ester to yield the boronic acid. The hydrolytic stability of commonly used bulky boronic acid esters in synthetic chemistry is attributed to this effect so much so that they can be purified by chromatography.17,18 Also, the bulk on the backbone may cause relief of steric strain that favors a cyclization reaction. The Thorpe–Ingold effect is a well-known analogy in which sterically hindered geminal substituents favor a cyclization reaction.19–22 Data in Table 1 also allow one to compare the equilibrium constants for ethylene glycol and 2,3-butanediol (see structures in Fig. 1) with 4-MBBA. Both of these diols contain hydroxyls that are, ideally, geometrically identical in relation to one another. The only difference between them is that 2,3-butanediol contains a methyl group geminal to each hydroxyl. This provides bulk on the backbone, and a greater than 13-fold increase in the equilibrium constants was observed. That is, 2,3-butanediol, behaves closer to that of glycerol. This is illustrated by the open symbols in Figure 5. This experiment suggested that the effect of steric bulk on the 1,2diols backbone had a significant effect on the magnitude of the equilibrium constant for ester formation. The use of sterically Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

6

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Figure 6. Structures of the only 4-MBBA, ethylene glycol ester structure, and the three possible isomers of 4-MBBA mannitol ester.

hindered1,2-diols such as pinacol have long been used as stable protecting groups for boronic acids. As the number of 1,2-diol pairs increase in the series of polyols illustrated by the structures shown in Figure 1, there was also a greater number of possible 4-MBBA polyol ester isomers. This should increase the observed equilibrium constant for the larger polyols. For example, the 4-MBBA ethylene glycol ester has one possible isomer, whereas the 4-MBBA mannitol ester has multiple possible isomers Figure 6, even assuming that ester formation via a five-membered ring is favored. Each isomer has its own discreet equilibrium constant associated with it, and thus the observed or measured equilibrium constant all of the equilibrium constants for all individual isomers. Mart´ınezAguirre et al.12 suggested that the equilibrium constants may also be related to the acidity/basicity of the 1,2-diols. The UV method to determine pKa values for aryl boronic acids provided insight into many aspects of aryl boronic acid– polyol interactions; however, it could not be used to study similar effects in alkyl boronic acids because of their lack of UV spectra, and thus the lack of spectral change between the acid and conjugate base. Additional methods were therefore employed. Variation in pKa Values with 1,2-Diol Concentration Using Potentiometric Titration Potentiometric titration is a virtually universal method to determine the Ka and thus the pKa values of weak acids and bases that have sufficient solubilities. It was therefore applied to the study of boronic acids and in particular the study of alkyl boronic acids, where changes in UV spectra were negligible on ionization. With each aliquot, these experiments generated a plot of total titrant volume versus pH as illustrated in Figure 7; in this case, the experiment with 5 mM 4-MBBA alone in solution will be used as an illustrative example. As the analyte concentration, sample volume, and titrant concentration are also known, the molar concentrations of acid and base, [HA] and [A− ], can be estimated. Because many of the titrations would yield useful information in the alkaline region, Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

Figure 7. Plot of titrant (0.2 M KOH) volume versus pH for the titration of 5 mM 4-MBBA in aqueous solution.

for pH values above approximately 10, a base correction term was added. It was therefore necessary to estimate hydroxyl ion concentration, [OH− ], so that the electroneutrality of the solution could be accounted for, this is also called the charge balance through the modified Henderson–Hasselbalch equation. A pKa value was thus calculated for each aliquot. The most reliable data were obtained from aliquots at a pH value closest to the pKa of the solute; thus, the mean and SD of the data generated ±0.5 pH units from the estimated pKa was used as the final value for each sample, in this case 9.29 ± 0.02 for 4-MBBA alone in solution. As described earlier, a change in the apparent pKa, pKa , value was observed for 4-MBBA and other boronic acids studied when 1,2-diols were present in solution. Similar data to that seen in Figure 4 using UV spectral changes were observed allowing estimation of the Ka and pKa values for the free boronic acid and the ester, as well as allowing an estimate of the ester formation equilibrium constants. Table 2 lists the results of four different boronic acids (see Fig. 1 for the structures) with mannitol. As before, pKa1 represents the free boronic acid pKa, pKa2 is the experimentally extrapolated pKa for the corresponding mannitol ester, KB1 is the equilibrium constant for the neutral species, and KB2 is the equilibrium constant for the anionic species. Several patterns

Table 2. pKa Values for the Free Boronic Acid (pKa1 ) and the Boronic Acid Mannitol Ester (pKa2 ), and Ester Formation Equilibrium Constants for Both the Anionic Boronic Acid (KB2 ) and the Neutral Boronic Acid (KB1 ) with Mannitol, as Determined by Potentiometric Titration for Two Aryl Acids (4-MBBA PhBA) and Two Alkyl Acids (IBA and CyBA) Boronic Acid

pKa1

pKa2

KB1 (M−1 ) ± SD

KB2 (M−1 ) ± SD

4-MBBA PhBA IBA CyHBA

9.32 8.88 11.51 10.56

6.25 5.78 8.07 7.27

3.4 ± 0.8 2.4 ± 0.2 2.7 ± 0.2 4.9 ± 1.1

4.0 ± 0.9 × 103 3.0 ± 0.3 × 103 7.1 ± 0.4 × 103 9.4 ± 2.1 × 103

DOI 10.1002/jps.24346

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Table 3. pKa Values for IBA Esters (pKa2 ), and Ester Formation Equilibrium Constants for Both the Neutral IBA (KB1 ) and the Anionic IBA (KB2 ) with the Corresponding 1,2-Diols, as Determined by the Potentiometric Titration Method 1,2-Diols

pKa2

KB1 (M−1 ) ± SD

KB2 (M−1 ) ± SD

Sorbitol Mannitol Xylitol

8.00 8.07 8.23

5.5 ± 0.1 2.7 ± 0.2 2.2 ± 0.1

1.7 ± 0.4 × 104 7.1 ± 0.4 × 103 4.2 ± 0.2 × 103

IBA itself has a pKa, value of 11.51.

immediately emerge. First, the decline in pKa from the free boronic acid to the ester is consistently between 3.0 and 3.4 units. This holds regardless of the initial pKa of the boronic acid or its structural features, such as whether they are alkyl or aryl boronic acids. Also, the equilibrium constants for ester formation under basic as well as acidic conditions are in the same order magnitude, with larger values for the alkyl boronic acids compared with the aryl acids. These two observations are very important in establishing the general nature of these effects; thus, the observations made in the previous section with 4-MBBA using UV methods may be applied to boronic acids as a whole. This is especially important in light of the fact that most boronic acids currently in development as therapeutics are alkyl boronic acids. Table 3 lists data for a specific model alkyl boronic acid, IBA, with three 1,2-diols with high-equilibrium constants. Again, roughly a 3–3.5-fold unit decrease in pKa value was found upon the formation of an ester from the initial IBA (pKa value of 11.51). Also of note is that the order of the equilibrium constants, sorbitol > mannitol > xylitol, follows the trend seen with 4-MBBA as determined by a number of other techniques. The comparison of sorbitol and mannitol also supports the hypothesis that an increase in the possible number of conformational isomers of a boronic acid polyol ester will increase the equilibrium constant between the boronic acid and the polyol. Although both mannitol and sorbitol have the same number of 1,2-diol pairs, again assuming only five-membered rings being formed, there is a greater number of nonidentical isomers possible for sorbitol versus mannitol, five versus three, respectively. That is, for mannitol, there are three isomers because ester formation at either the 1,2 and 5,6-hydroxyls or 2,3 and 4,5hydroxyls produce identical structures. There are five isomers because sorbitol varies from mannitol in its stereochemistry at the 2-hydroxyl. This one feature, a change of stereochemistry at one hydroxyl, causes an increase in possible conformational isomers, and the result is that the equilibrium constant more than doubles. This mechanism appears to be very significant and will need to be explored further. The ability to manipulate the ionization state of a drug molecule in aqueous solution has obvious implications for drug formulation in areas such as solubility, dissolution rate, and chemical stability. The manipulation of boronic acid ionization with 1,2-diol excipients is particularly appealing as it may be performed through a change in pKa value and pH rather than a change in formulation pH alone in the absence of 1,2-diol. That is, this effect may be used in order to increase the solubility of a drug at a particular pH, decrease the pH of an existing formulation, or aid in the formation of salts. Other applications are certainly possible, and will undoubtedly be explored by formuDOI 10.1002/jps.24346

7

lators as boronic acid moieties become more popular in modern pharmaceuticals. Effect of Boronic Acid Concentration on Apparent pKa Values, Potentiometric Titration To test the hypothesis that concentration of boronic acids might alter some of their solution properties, the pKa values of IBA in the concentration range of 1–50 mM were determined. At 1-mM IBA, a pKa value of 11.30 was estimated, whereas in the concentration range of 5–50 mM, pKa values of 11.50–11.58 were estimated and found to be not statistically different from one another. Although the 1 mM might be considered an outlier, this is more likely because of the limitations of the data at this low concentration. That is, there appears to be no significant effect of boronic acid concentration on the measured pKa value, suggesting no evidence for boronic acid self-association in the concentration range studied here. Effect of Time to Equilibrium on Apparent pKa Values, Potentiometric Titration There was no statistically significant effect of the time to equilibrium on the measured pKa value of 4-MBBA in the presence of 0.5 M mannitol. The pKa value determined for 4-MBBA in this solution is 6.46 (±0.01) with a minimum equilibration time between aliquots of 5 s, and 6.48 (±0.01) with a minimum equilibration time of 30 s. Hypothetically, as the alkaline aliquots are added and the solution pH increases, the equilibrium constant between the boronic acid and mannitol should increase. If the equilibrium was slow relative to the addition time, this would alter the apparent pKa value determination. This did not occur, suggesting one of two possibilities; the overall equilibrium at the measured pH values does not change significantly during the experimental range, or the rate of approach to equilibrium is sufficiently rapid so that it is reached within the allotted 5-s equilibration period. This suggests that, experimentally, the 5-s equilibration period is sufficient. The issue of the kinetics of boronic acid ester hydrolysis is the basis for a later paper. Change in Enthalpy for Boronic Acid Ester Formation with 1,2-Diols Isothermal Titration Microcalorimetry The raw data in ITC produced a plot of heat flow versus time, as depicted in Figure 8, which is the plot obtained for the titration of mannitol into 4-MBBA. The injections are clearly separated by the 400-s equilibration time. This is critical for proper data analysis as the heat flow must return to baseline before the next injection. The early injections elicited large responses and the later injections a decreased response, because as the experiment continued, most of the boronic acid molecules were already esterified. Therefore, additional 1,2-diol did not result in any further measurable heat of reaction. The residual heat flux response of the later injections is largely because of heat of dilution. The peaks for each injection are then integrated to yield a value of total heat flux for that injection. The heat of dilution per injection generated from the corresponding control experiment is then subtracted from this value, and the resulting corrected total heat may be plotted versus injection number as in inset to Figure 8. This data may then be fit to Eq. (11), which describes Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

8

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Figure 8. Raw data from an isothermal calorimetry experiment for the titration of 15 mM mannitol into 1 mM 4-MBBA under basic pH conditions. Inset is the cumulative heat per injection data from the same experiment (circles) fit to Eq. (11) to yield the curve shown (line).

the equilibrium process for a 1,2-diol with independent multiple ester sites possible.  Q = VH [L] +

1 + [M]nK −

√

(1 + [M]nK − [L]K)2 + 4K[L] 2K

an approximate pH of 11.7, a pH well above the pKa of 4-MBBA. There are several relevant features of this data. The rank order of equilibrium constants for the 1,2-diols is identical to the rank orders observed with the previous two methods that are determined from the apparent pKa shifts. This supports the assumption made that the pKa value shift is a result of ester formation, and not a solvent effect because of the addition of the 1,2-diol. The equilibrium constant values also support these conclusions; they are in the same order of magnitude as the values obtained for interaction with the anionic forms of boronic acids with the previous methods. It should be noted that a possible cause of the numerical difference in values between the equilibrium constants determined various methods, are different conditions used and the assumption in the model that the 1,2-diols themselves do not ionize. This assumption is generally valid for the pKa change methods as most of the critical data are taken at pH values between 6 and 10. However, in the ITC experiment, the static pH value is 11.7, this is nearer the literature values for the pKa of mannitol, 13.1, and some of the other 1,2-diols. A small but significant amount of 1,2-diol anion may be in solution and affect the calculated equilibrium constant values. The calculated value for n supports a 1:1 boronic acid to 1,2-diols interaction in all of these cases. For the less strongly interacting 1,2-diols,



(11) where Q is the total heat, V is the volume of the reaction cell, [L] is the ligand concentration, in this case the 1,2-diol, [M] is the receptor concentration, in this case the boronic acid, H is the enthalpy for ester formation, K is the equilibrium constant, and n is the number of ester formation sites. As the y-axis variable is total heat, the x-axis variable is a ratio of 1,2-diol and boronic acid concentration (depicted as injection number for graphical reasons), and the volume of the reaction cell is known, the enthalpy of ester formation, the equilibrium constant, and n may be estimated. Several ITC experiments were performed in order to confirm observations in the previously mentioned pKa studies as well as to explore other aspects of boronic acid to polyol ester formation. Table 4 lists the result of the titrations of a series of 1,2-diols with 4-MBBA. Note that the values are assumed to be for the anionic interaction (KB2 ) as the experiments were performed at Table 4.

Equilibrium Constants (K), n, and Enthalpy of Ester Formation (H) for the Titration of the Listed 1,2-Diol into 4-MBBA at pH 11.7

1,2-Diols (M−1 )

K ± SD n ± SD H (kJ/mol) ± SD

Sorbitol

Mannitol

6.3 ± 0.6 × 1.1 ± 0.1 −27 ± 0.6

103

Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

2.8 ± 0.04 ± 1.0 ± 0.01 −22 ± 1.9

103

Erythritol

Glycerol

2,3-Butanediol

120 ± NA 0.84 ± NA −24 ± NA

24 ± 0.7 NA NA

13 ± 0.8 NA NA

DOI 10.1002/jps.24346

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

9

Table 5. Equilibrium Constant (K) for Ester Formation, n, and Enthalpy of Ester Formation (H) for the Titration of Mannitol into the Listed Boronic Acids, 4-MBBA, IBA, and CyHBA, at pH 11.7 Using ITC Boronic Acid

4-MBBA

IBA

CyHBA

K (M−1 ) ± SD 2.8 ± 0.04 × 103 6.6 ± 0.3 × 103 3.6 ± 0.0 × 103 n ± SD 1.0 ± 0.0 1.0 ± 0.04 0.8 ± 0.03 H (kJ/mol) ± SD −22 ± 1.9 −29 ± 0.06 −27 ± 0.43

such as glycerol and 2,3-butanediol, these values cannot be confirmed; however, it is chemically impossible for higher-order interactions to form with triols and diols. Table 5 lists the ITC results of the titration of mannitol into three boronic acids, 4-MBBA and two alkyl boronic acids, IBA and CyHBA. This series of experiments again serves to support the general nature of the interaction. ITC behavior is analogous in aryl and alkyl boronic acids to that noted earlier. The changes in enthalpy of interaction for each case are generally in the same range for all the 1,2-diols studied here. The favorable free energy and exothermic enthalpic energy changes observed mean that there is a very significant opposing entropic contribution consistent with a bimolecular process. Therefore, the increase in the equilibrium constants with an increase in the number of 1,2-diol pairs is less entropically unfavorable. This is consistent with many interpretations such as the more favored entropy because of the larger number of possible isomers, already restrictive rotations, and possible relief of steric strain with the larger polyols. Further ITC studies on a larger pool of 1,2-diols would be needed to further elucidate these thermodynamic effects and their interpretation. An ITC titration with the 1,3-diol, 1,3-propanediol into 4MBBA was also performed. This elicited only a heat of dilution effect, suggesting that no measurable amount of ester formation took place at the concentrations used. This is significant, as the only ester that may form with this diol is a six-membered ring. The data suggest a very low propensity for six-membered boronic acid esters to form in aqueous solution. However, with other polyols and under other conditions, the six-membered esters may very well form; but in these studies, no direct or indirect evidence was observed to support this behavior. Ionic Strength Effect The results for the ITC titrations performed with 4-MBBA and mannitol at different ionic strengths were performed. The equilibrium constants do not significantly differ between the three ionic strength environments, 0.05, 0.105, and 0.185, nor is there any developing pattern. This experiment may serve as a control to relate some of the results of other experiments performed at a range of ionic strength values. Boron-11 NMR Direct evidence for ester formation was confirmed using boron11 NMR data. Others have confirmed ester formation using mass spectral analysis for other boronic acids and 1,2-diols.14,23 The raw spectra for a boron-11 NMR experiment are illustrated in Figure 9. The peak at 4.4 ppm results from the nonesterified EtBA) and the broad peak at 7.6 ppm results from the presumed boronate ester. As expected, the peak at 7.6 ppm increases and the peak at 4.4 ppm decreases with the addition of mannitol DOI 10.1002/jps.24346

Figure 9. Raw boron-11 NMR spectra for EtBA in the presence of increasing amounts of added mannitol. The arrows show the direction of each peak upon the addition of mannitol. The peak at 4.4 ppm is the nonesteried EtBA peak, which decreases upon the addition of mannitol. The peak at 7.6 ppm is the EtBA mannitol ester, which grows upon the addition of mannitol, 0–110 mM.

from 0 to 110 mM. Integration of the peaks proved challenging because of the relatively poor sensitivity of boron-11 NMR and the broadness of the peaks, so a quantitative interpretation of this data was not attempted.

CONCLUSIONS The formation of boronic acid esters with 1,2-diols resulted in a lower pKa value of 3–3.5 compared with pKa value for the parent boronic acid. This effect was seen for both alkyl and aryl boronic acids, regardless of the initial boronic acid pKa value. Also, in relatively dilute aqueous solution, a 1:1 boronic acid 1,2-diol ester formed. However, with an amenable boronic acid at high concentration of acid and/or 1,2-diols, a higher-order interaction may form but no evidence was seen in the present study. The comparison of heats of interaction of ethylene glycol and 1,3-propanediol by ITC suggests that the formation of sixmembered boronic acid polyol esters does not significantly occur in dilute aqueous solution. Polyols with multiple1,2-diol pairs have higher equilibrium constants with all the boronic acids studied. Possible explanations for this observation includes relief of steric bulk of the larger 1,2-diols exerting a strong effect as demonstrated by the comparison of ethylene glycol and 2,3-butanediol. The increased bulky methyl groups may improve the interaction by slowing the reverse reaction, hydrolysis with water, as well as making the cyclic product more energetically stable compared with less bulky 1,2-diols. Other considerations such as statistical considerations, with an increase in the larger number of possible esters and isomer, resulting in less unfavorable entropy. Already restricted rotations may also play a role. All of these hypotheses deserve further study and testing. The increased number of conformers meant that the individual isomer equilibrium constants may be significantly different resulting in a larger and additive observed total interaction constant. This Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

10

RESEARCH ARTICLE – Pharmaceutics, Drug Delivery and Pharmaceutical Technology

hypothesis is also supported by the comparison of sorbitol to mannitol. The ability to decrease or manipulate the apparent pKa value of boronic acids as a result of ester formation with 1,2diols has obvious formulation implications. Not only may the esters have altered solubility compared with the parent boronic acid but the lower pKa value of the ester may allow one to formulate boronic acids at a more physiologically acceptable pH value. There was no discernable change in the interaction of boronic acids with 1,2-diols or the pKa of the boronic acids themselves with increasing concentration of the boronic acid, suggesting no major self-association in aqueous solutions at the concentrations studied here.

ACKNOWLEDGMENTS The authors would like to thank Alana Toro-Ramos for help with some of the initial UV spectrophotometric studies. Sara Neuenswander helped with NMR studies. Help with the automated potentiometric titration method from Dr. Adrian Russell of the Centre for Drug Candidate Optimisation at Monash University is very much appreciated.

REFERENCES R (bortezomib) 1. Millennium Pharmaceuticals, Inc. 2004. VELCADE for injection—Prescribing information. Rev 1 ed., Cambridge, Massachusetts. 2. Yoshino K, Suzuki A, Mori Y, Kakihana H, Honda C, Mishima Y, Kobayashi T, Kanda K. 1989. Improvement of solubility of p-boronophenylalanine by complex formation with monosaccharides. Strahlenther Onkol 165:127–129. 3. Plamondon L, Grenier L, Adams J, Gupta S. 2006. Formulation of boronic acid compounds. Patent US 7109323 B2. 4. Gupta SL, Stella VJ, Waugh W. 2005. Formulation of boronic acid compounds. Patent US6958319. 5. Gupta SL, Stella VJ, Waugh W. 2005. Formulation of boronic acid compounds. Patent US6713446 B2. 6. Wu S, Waugh W, Stella VJ. 2000. Degradation pathways of a peptide boronic acid derivative, 2-Pyz-(CO)–Phe–Leu-B (OH) 2. J Pharm Sci 89:758–765. 7. Byrn SR, Tishmack PA, Milton MJ, van de Velde H. 2011. Analysis of two commercially available bortezomib products: Differences in assay active agent and impurity profile. AAPS Pharm Sci Tech 12:461–467. 8. Bolognese A, Esposito A, Manfra M, Catalano L, Petruzziello F, Martorelli MC, Paggliuca R, Ottiero M, Scalfaro M, Rotoli B. 2009. An NMR study of the bortezomib degradation under clinical use conditions. Adv Hematol 2009:1–5.

Marinaro et al., JOURNAL OF PHARMACEUTICAL SCIENCES

9. Yang W, Gao X, Wang B. 2003. Boronic acid compounds as potential pharmaceutical agents. Med Res Rev 23:346–368. 10. Yan J, Springsteen G, Deeter S, Wang B. 2004. The relationship among pKa, pH, and binding constants in the interactions between boronic acids and diols—It is not as simple as it appears. Tetrahedron 60:11205–11209. 11. Springsteen G, Wang B. 2002. A detailed examination of boronic acid–diol complexation. Tetrahedron 58:5291–5300. 12. Mart´ınez-Aguirre MA, Villamil-Ramos R, Guerrero-Alvarez JA, Yatsimirsky AK. 2013. Substituent effects and pH profiles for stability constants of arylboronic acid diol esters. J Org Chem 78:4674– 4684. 13. Shull BK, Spielvogel DE, Head G, Gopalaswamy R, Sankar S, Devito K. 2000. Studies on the structure of the complex of the boron neutron capture therapy drug, L-p-boronophenylalanine, with fructose and related carbohydrates: Chemical and 13 C NMR evidence for the $-D-fructofuranose 2, 3, 6-(p-phenylalanylorthoboronate) structure. J Pharm Sci 89:215–222. 14. Islam TM, Yoshino K, Sasane, A. 2003. 11B NMR study of p-carboxybenzeneboronic acid ions for complex formation with some monosaccharides. Anal Sci 19:455–460. 15. Schmidt E, Dooley N, Ford SJ, Elliott M, Halbert GW. 2012. Physicochemical investigation of the influence of saccharide-based parenteral formulation excipients on L-p-boronphenylalanine solubilisation for boron neutron capture therapy. J Pharm Sci 101:223–232. 16. Iwatsuki S, Nakajima S, Inamo M, Takagi HD, Ishihara K. 2007. Which is reactive in alkaline solution, boronate ion or boronic acid? Kinetic evidence for reactive trigonal boronic acid in an alkaline solution. Inorg Chem 46:354–356. 17. Sugihara JM, Bowman CM. 1958. Cyclic benzeneboronate esters. J Am Chem Soc 80:2443–2446. 18. Perun TJ, Martin JR, Egan RS. 1974. Cyclic phenylboronates as hydroxyl protecting groups in the synthesis of monoesters of macrolide aglycones. J Org Chem 39:1490–1493. 19. Jager J, Graafland T, Schenk H, Kirby AJ, Engberts JBFN. 1984. Intramolecular-catalyzed sulfonamide hydrolysis. VII. The Thorpe– Ingold effect in the intramolecular carboxyl-catalyzed hydrolysis of sulfonamides. J Amer Chem Soc 106:139–143. 20. Davey AE, Parsons AF, Taylor RJK. 1989. Reactions of dithiane derived alcohols: Promotion of the Thorpe–Ingold effect by trimethylsilyl and tert-butyl substituents. J Chem Soc Perkin Trans I 10:1853–1858. 21. McIntyre S, Sansbury FH, Warren S. 1991. The Thorpe–Ingold effect in the synthesis of tetrahydro-furans and pyrans by competitive cyclization with phenylthio migration. Tetra Let 32:5409– 5412. 22. Forbes MDE, Patton JT, Myers TL, Maynard HD, Smith DW Jr, Schulz GR, Wagener KB. 1992. Solvent-free cyclization of linear dienes using olefin metathesis and the Thorpe–Ingold effect. J Amer Chem Soc 114:10978–10980. 23. Yoshino K, Kotaka M, Okamoto M, Kakihana H. 1979. Boron11 NMR study of the complex formation of borate with catechol and L-dopa. Bull Chem Soc Japan 52:3005–3009.

DOI 10.1002/jps.24346