Hydrophobic Ion Pairing of Peptide Antibiotics for ... - ACS Publications

Oct 18, 2017 - as V × [API:IP]sld, where V is the volume of the system and. [API:IP]sld is the concentration of API:IP complex above its saturation c...
0 downloads 13 Views 3MB Size
Article Cite This: Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Hydrophobic Ion Pairing of Peptide Antibiotics for Processing into Controlled Release Nanocarrier Formulations Hoang D. Lu, Paradorn Rummaneethorn, Kurt D. Ristroph, and Robert K. Prud’homme* Department of Chemical and Biological Engineering, Princeton University, Princeton, New Jersey 08544, United States S Supporting Information *

ABSTRACT: Nanoprecipitation of active pharmaceutical ingredients (APIs) to form nanocarriers (NCs) is an attractive method of producing formulations with improved stability and biological efficacies. However, nanoprecipitation techniques have not been demonstrated for highly soluble peptide therapeutics. We here present a model and technique to encapsulate highly water-soluble biologic APIs by manipulating API salt forms. APIs are ion paired with hydrophobic counterions to produce new API salts that exhibit altered solubilities suitable for nanoprecipitation processing. The governing rules of ion pair identity and processing conditions required for successful encapsulation are experimentally determined and assessed with theoretical models. Successful NC formation for the antibiotic polymyxin B requires hydrophobicity of the ion pair acid to be greater than logP = 2 for strong acids and greater than logP = 8 for weak acids. Oleic acid with a logP = 8, and pKa = 5, appears to be a prime candidate as an ion pair agent since it is biocompatible and forms excellent ion pair complexes. NC formation from preformed, organic soluble ion pairs is compared to in situ ion pairs where NCs are made in a single precipitation step. NC properties, such as stability and release rates, can be tuned by varying ion pair molecular structure and ion pair-to-API molar ratios. For polymyxin B, NCs ≈ 100−200 nm in size, displaying API release rates over 3 days, were produced. This work demonstrates a new approach that enables the formation of nanoparticles from previously intractable compounds. KEYWORDS: nanocarrier, hydrophobic ion pairing, antibiotics, controlled release

T

fastest growing segment of the pharmaceutical market, the development of new methods to efficiently encapsulate hydrophilic APIs into NCs, using direct water precipitation methods, is of significant interest. The ion pairs are processed into stable NCs through Flash NanoPrecipitation. The ion pairing technique is quite general, but as examples, we show gentamicin (LogP = −4.21) and polymyxin B (LogP = −5.6). These cationic APIs, which are used to manage P. aeruginosa bacterial pulmonary cystic fibrosis infections, adhere to pulmonary mucus and do not effectively reach sites of infections after drug administration. 10,11 Encapsulation, especially in a muco-diffusive PEG NC, would increase efficacy.12 Soluble, or relatively soluble, compounds have been encapsulated by covalent conjugation to create hydrophobic pro-drugs.13−16 After encapsulation and delivery, covalent linkages are cleaved, thereby producing the original API for therapeutic activity. However, covalent modification of APIs results in the creation of a new molecular entity, which requires additional FDA approval. An alternative route to reduce API

he controlled delivery of active therapeutic ingredients from nanocarriers (NCs) can result in improved bioavailability, reduced toxicity, sustained activity, simplified dosing regimens, improved patient adherence, and enhanced overall efficacy.1 A major driver for NC formulations is the desire to use the nanoparticle to target delivery and then to release cargo at the desired site. NC formation through direct precipitation methods is attractive because these methods form NCs with high mass loadings in a scalable and continuous fashion, as in the case of Flash NanoPrecipitation (FNP).2−4 In nanoprecipitation processes, hydrophobic active pharmaceutical ingredients (APIs) are typically dissolved in organic solvents and mixed with water as an antisolvent to induce precipitation and NC self-assembly. However, such direct precipitation methods are only feasible for highly water insoluble compounds (LogP > 4) and cannot be used to encapsulate hydrophilic peptides and biologics into NC form. Currently, water-soluble biologics are typically encapsulated through water-in-oil-inwater emulsion (W/O/W) or liposomal processes that require multiple steps, suffer from poor encapsulation efficiencies, or exhibit low drug mass loadings.5−7 Processes involving coacervate complexation of biologics with oppositely charged polymers have been demonstrated.8,9 Coacervates are sensitive to ion exchange and dissolution, so a second coating process is required to achieve controlled release. Since biologics are the © XXXX American Chemical Society

Received: September 20, 2017 Revised: October 18, 2017 Accepted: October 24, 2017

A

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics water solubility is to ion pair the active ingredient with a hydrophobic counterion.17−21 An advantage of this approach is that the resultant products are not considered new molecular entities and do not require full FDA reapproval. In most instances, API salt forms have been engineered for enhanced water solubility or crystal stability; the use of hydrophobic ion pairs to create less soluble drug forms is less well-explored. This technique has recently been used for the precipitation of hydrophobic small molecules (LogP = 2−5) into NCs, but not for highly water-soluble (LogP > 0) biologics.22,23 It is wellknown that highly anionic oligonucleotides (DNA, siRNA) can be precipitated by complexation with cationic polymers or with cationic lipids. 24 However, the lower entropy of the oligonucleotide chains and charge spacing of less than 0.17 nm/negative charge enable this complexation. The species we are demonstrating have much lower charge densities, so the success of ion pairing and the stability of the ion pair against ion exchange with physiological media are unanticipated. We have employed Flash NanoPrecipitation, a continuous and scalable method of forming NCs with high API mass loads (>50−90%), to encapsulate the APIs. The FNP process relies upon rapid micromixing (O(ms)) of a solvent and antisolvent stream to create high supersaturations, which drive high nucleation rates. The hydrophobic APIs precipitate into NCs, while the hydrophobic block of the stabilizing polymer absorbs onto NC surface to arrest particle growth. This results in highly loaded NCs with a narrow size distribution. We have described the process in detail elsewhere.21 The outline of the article is as follows. We present the two approaches to forming hydrophobic ion pairs: pre-ion pairing and in situ ion pairing. A screen of 15 ion pair reagents is presented, and NCs are formed from nine of these ion pair reagents. From these results, the rules governing precipitation and complex formation are developed. A mathematical model of the concentrations, dissociation constants, and pKas is presented. The stability and release characteristics are presented, which show different release mechanisms dependent on the ion pair agent.

USA) were used to separate encapsulated drug from unencapsulated drugs. Ion Pairing Screening. The 15 ion pairing (IP) reagents listed above were screened to assess if hydrophobic API salt forms would be formed and would precipitate from solution. Ion pair and API drug properties were calculated with Molinspiration software (Molinspiration Cheminformatics, Slovakia). APIs were dissolved in water at 5 mg mL−1 concentration and mixed with IPs dissolved in water at varying concentrations to yield different API to IP ratios. Ratios are defined based on positive charges of APIs with negative charges on the IP, and exact compositions are given in Table S1. The screening process determines whether a particular IP may be used with a given API for NC formation. The IP process may be performed either prior to or during Flash NanoPrecipitation; in the case of the latter, we refer to as “in situ ion pairing”. These two ion pairing methods are described in Figure 1.

Figure 1. Mechanism of ion pairing-based Flash NanoPrecipitation. FNP of pre-ion paired API. Hydrophilic active pharmaceutical ingredients are created as a hydrophobic salt by pre-pairing the API with a hydrophobic counterion. Pre-ion paired APIs are dissolved in organic solvent alongside an amphiphilic block copolymer and rapidly micromixed against water. Hydrophobic API ion pairs precipitate into NCs and self-assemble into NCs. FNP with in situ ion pairing. Hydrophilic APIs are dissolved in water and rapidly micromixed against organic solvent containing a hydrophobic API counterion and an amphiphilic block copolymer. Ion pairing of the API occurs during mixing, producing a hydrophobic API ion pair that precipitates in the presence of an amphiphilic block copolymer, resulting in the formation of NCs.



EXPERIMENTAL SECTION Materials. Sodium hexanoate, sodium decanoate, sodium myristate, sodium oleate (OA), pamoic acid disodium salt (PA), benzenesulfonic acid monohydrate, sodium 2-naphthalenesulfonate, (1R)-(−)-10-camphorsulfonic acid, sodium 1,2ethanesulfonate, sodium 1-heptanesulfonate, sodium 1-octanesulfonate monohydrate, sodium 1-decanesulfonate, sodium dodecyl sulfate (SDS), sodium dodecylbenezenesulfonate (DBS), and sodium deoxycholate were obtained from SigmaAldrich (St. Louis, MO, USA) and used as ion pairing reagents. Gentamicin sulfate (MP Biochemicals, Santa Ana, California, USA) and polymyxin B sulfate (Calbiochem, San Diego, CA, USA) were used as hydrophilic and cationic antibiotic APIs for encapsulation. The block copolymers 1.6 kDa polstyrene-block5 kDa polyethylene glycol (PS−PEG) and 5 kDa polycaprolactone-block-5 kDa polyethylene glycol (PCL−PEG) were obtained from Polymersource (Montreal, Quebec, Canada) and used as NC formation and stabilizing agents. Bicinchoninic acid (BCA) assay reagents were obtained from Thermo Scientific (Waltham, MA, USA) and used to quantify protein concentrations. Corning 96-well, clear, circular flat bottom, half area microplates were used for the BCA assays. Amicon 50 kDa centrifugal ultrafilters (EMD Millipore, Billerica, MA,

Pre-ion Paired APIs. In the case of pre-ion pairing, API and IP are mixed in water to form a water-insoluble precipitate (referred to as an API:IP complex). The water is then removed, and the API:IP complex is dissolved into an organic solvent (along with a block copolymer) for use in the organic stream in FNP. In this case, the API:IP complex is treated as a component in the organic stream of normal FNP. The screening process for the pre-ion paired method FNP involves four sequential tests. 1. The first test examines the solubility of the IP agent. IP is added to DI water at a concentration that would be required for a 1:1 API:IP charge ratio if the API were B

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics dissolved at 5 mg mL−1. Substantial insolubility in water is required for successful ion pairing. 2. The second test examines whether an in situ API:IP complex can be formed in a single step. IP in an organic, but water miscible, solution mixed against the aqueous API solution. The final solution is checked visually to see if precipitation has occurred. In this test, the IP solution is at a concentration such that mixing equal volumes of IP solution and API solution will form a 1:1 API to IP charge ratio. Four API:IP ratios are investigated: 200 μL of API solution is mixed with 100 μL of IP solution for a 1:0.5 charge ratio; 200 μL of API solution is mixed with 200 μL of IP solution for a 1:1 charge ratio; 200 μL of API solution is mixed with 400 μL of IP solution for a 1:2 charge ratio; and 200 μL of API solution is mixed with 800 μL of IP solution for a 1:4 charge ratio. 3. The third test examines whether the API:IP complex can be solubilized in the organic phase for subsequent FNP NC formation. Pellets of the precipitate formed in Test 2 are tested to see if they are soluble in a water-miscible organic solvent such as tetrahydrofuran (THF) or more polar solvents such as dimethyl sulfoxide (DMSO) or methanol (MeOH). 4. The final test assesses if the hydrophobic API:IP complex will reprecipitate upon FNP mixing with water as the antisolvent. The organic solvent containing API:IP from Test 3 is diluted 10-fold in water, and precipitate formation is examined. If a precipitate forms, then the construct is suitable for use in FNP. In Situ Ion Pairing. In this route, the ion pairing complexation occurs in situ during FNP. The hydrophilic API enters in the aqueous stream, and the stabilizing block copolymer and hydrophobic IP enter via the water-miscible organic stream. The results of the solubility screening show three different solvent compositions for six IPs. Pamoic acid, sodium dodecyl sulfate, and decyl sulfate dissolve in 50% DMSO/THF; oleic acid and myristatic acid dissolve in 50% MeOH/THF; and dodecyl benzyene sulfate dissolves in 20% water/THF. Nanocarrier Formation by Flash NanoPrecipitation. Flash NanoPrecipitation was used to self-assemble PEG-coated NCs containing an encapsulated API:IP complex. For the preion paired method, particles were formed by rapid micromixing of hydrophobic−hydrophilic block copolymer PCL−PEG and API:IP complex dissolved in organic solvent against deionized (DI) water through a confined impingement jet (CIJ) mixer.25 For the in situ ion pairing method, the organic stream containing PCL−PEG and anionic IP dissolved in organic solvent were rapidly micromixed in the CIJ mixer against the water stream containing cationic hydrophilic API dissolved in water. For both methods, the organic and water streams were mixed in a 1:1 volume ratio, and the resulting mixed streams were subsequently diluted 10-fold into water. Concentrations of PCL−PEG stabilizer and hydrophilic API (gentamicin and polymyxin B) were kept constant, while IP concentrations were varied to form NCs with a range of API:IP ratios. This ratio is the ratio of the molar concentration of cationic functional groups in the API to the molar concentration of anionic functional groups in the IP. Exact compositions of the stabilizer block copolymers, APIs, and IPs used are given in Table S1. Nanocarrier Characterization. Particle sizes and polydispersities were characterized by dynamic light scattering

(Malvern Zetasizer Nano, Malvern Instruments). Sizes were determined with backscattering measurements and reported as Z-average and intensity-weighted distributions. Particles were diluted 10-fold into two different collection buffers, DI water and 1× phosphate-buffered saline (PBS) for sizing measurements and NC stability assessments. PBS has an ionic strength of 163 mM and mimics the ionic strength of serum. Incubation in PBS assesses the stability of the NC under conditions where ion exchange might destabilize the NC ion paired complex. NC stability was assessed by taking DLS measurements of the particles at four time points: t = 0, 3, 24, and 72 h. Stability measurements were conducted at room temperature. NC drug encapsulation efficiencies were determined by separating encapsulated API from dissolved/unencapsulated drugs in the NC sample through ultrafiltration across a 50 kDa membrane (Amicon Ultra, EMD Millipore) and determining the concentration of dissolved/unencapsulated API in ultrafiltration flow-through. Encapsulation efficiency is calculated as the concentration of API within the flow-through fraction divided by the concentration of API within the unseparated NC sample, and then subtracting this ratio from 1 (eq 1). Drug concentrations were determined with the BCA assay (Spectramax i3x, Molecular Devices) with absorbance measurements at 562 nm, using a standard curve of the same drug (unencapsulated) as the reference. API fractional release was determined by dialyzing NC sample in 100-fold 1X PBS. Aliquots were taken from within the dialysis bag at different time points: t = 0, 1, 2, 3, 5, and 7 days. The drug concentration in each aliquot was determined by the BCA assay at 562 nm. The released fraction is the API concentration at each time point divided by the initial API concentration before dialysis (eq 2). encapsulation efficiency = 1 −

release fraction =



[API]in solution [API]initial

[API]not in NCs [API]total

(1)

(2)

RESULTS AND DISCUSSION Hydrophobic Ion Pair Screening. Gentamicin and polymyxin B are highly charged hydrophilic compounds, with water solubility of >100 and >50 mg mL−1, respectively (Figure 2), and with minimal solubility in polar organic solvents such as tetrahydrofuran. A variety of water-soluble counterion salt IPs were chosen for this investigation, to determine the physical properties of the counterions required for successful complex formation (Figure 3). A range of carboxylate and sulfonate salt IPs whose acid forms have pKa ranging 5 to −1.9 were tested, as well as salts with a wide range of solubilities and LogP values as high as 7.6 or as low as −4.92. API was dissolved at 5 mg mL−1 in water in 1× volume and rapidly mixed with 1× volume of dissolved IP to give a final mixture containing a 1:1 API:IP charge ratio. Mixtures were visually observed to assess if precipitates rapidly formed. Out of the 15 IPs tested, 11 IPs resulted in the formation of precipitates (Figure 4, Tables S1 and S2). Control experiments where API or IP alone were mixed did not result in precipitation, thereby demonstrating API:IP interactions are responsible for the phase formation. As a general trend, the IPs that caused precipitation have higher MW, are more hydrophobic, and of a stronger acidic form compared to IPs that did not cause precipitation. These effects C

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 4. Screening of ion pair properties for FNP-based NC assembly with polymyxin. The physical properties of ion pairs that failed to precipitate APIs (blue diamonds), that could precipitate APIs but failed to produce NCs (green circles), and that could precipitate APIs and produce NCs (red squares) are detailed.

Figure 2. Model hydrophilic active pharmaceutical ingredients tested within this study. Gentamicin (LogP = −4.21) is an aminoglycoside that functions by inhibiting protein synthesis, and polymyxin B (LogP = −5.62) is a macrocyclic peptide that functions by forming pores on cell membranes, which are both active against Gram-negative bacteria.

4.0; this effect of pKa highlights differences in the ability of these two IPs to charge-interact with the APIs. Thus, even though sodium decanoate is a more hydrophobic IP with a LogP of 2.01, sodium octanesulfonate with a LogP of 0.85 is a better IP agent due to its greater ability to ionically complex with APIs. As a control, APIs prepared without IP complexation were not soluble in THF, highlighting that association with hydrophobic IPs are necessary for solubility in organic solvents. In the FNP process, API:IP complexes dissolved in organic solvents must rapidly precipitate when mixed against the water antisolvent. To assess if the nine remaining API:IP salt complexes would behave accordingly, complexes dissolved from the previous step were rapidly mixed and diluted 10-fold into water. The results of the tests are shown in Figure 4, where the axes are the pKa of the IP agent and the calculated LogP of the compounds. Four of the IPs tested resulted in no observable precipitation upon mixing (blue diamonds in Figure 4), while four of the IPs tested resulted in rapid precipitation and successful NC formation (red squares). The IPs that failed to produce precipitates at this stage were the least hydrophobic among the group, with LogP at or below 0.85, while IPs that

can be explained by the facts that more strongly hydrophobic IPs produce more hydrophobic API:IP complexes and that IPs that ionically interact more strongly with the APIs produce stronger API:IP complexes. Both hydrophobicity and strong ionic interactions are necessary. After successfully identifying water-insoluble API:IP complexes, we determine if these new salt forms can be dissolved in THF for NC processing. Pellets formed from the previous step were incubated with THF at a 1× volume equal to the starting volume of API in water. Of the 11 API:IP complexes tested, all complexes except sodium decanoate and sodium deoxycholate dissolved in THF. Interestingly, the IP sodium octanesulfonate, which has a hydrophobic eight carbon chain pendant group, was sufficient for API:IP dissolution into THF, while sodium decanoate, which has a hydrophobic nine carbon chain pendant group, was not. The pKa of the acid form of sodium octanesulfonate is −0.40, while that of sodium decanoate is

Figure 3. Ion pairs screened and tested within this study. A wide variety of anionic salts were used and paired against gentamicin and polymyxin B. Salts that could precipitate APIs into NCs in the FNC process are named in bold. Salts that could precipitate APIs out of water, but not into NCs, in the FNC process are italicized. D

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 5. Modeling of API precipitation conditions by utilizing eq 10. (A) Fraction of API that is precipitated with varying saturation solubilities of the API:IP complex and with varying tendencies of complex formation. Saturation solubility is noted with [API:IP]sat and has units of mol L−1. Complexation strength is noted with Ks and has dimensions of mol−1 L−1. The amount of API and IP included in the reaction is at a one to one ratio at 1 M. Calculations are detailed in Supplementary Methods. (B) Fraction of API that is precipitated with varying saturation solubilities of the API:IP complex and with varying the amount of API in the reaction. Amount of API and IP included in the reaction is at a one to one ratio and with [API]0 with dimensions of M. Precipitation yields can be increased by decreasing complex saturation solubility, by increasing complexation strength, or by increasing the concentration of the reaction.

have LogP at or above 1.86 successfully form precipitates. Increasing difficulty in deprotonation, represented by increased pKa, requires greater hydrophobicity for successful NC formation. The first ionization of polymyxin occurs at pH = 10.23. Therefore, the ΔpKa gap between oleic acid and polymyxin is over 5 units. In our previous study of hydrophobic APIs, we found that ΔpKa > 2 values were required for stable NC formation.26 For the very strongly basic polymyxin B and gentamicin, that criteria, ΔpKa > 2, is satisfied with all of the IP counterions. These results provide general experimental insight on candidate IP selection for hydrophobic ion pairing. Model of IP Complex Equilibria. We develop a quantitative model of complexation and precipitation for basic and acidic APIs and IPs and compare the model with experimental results. The goal is to have rules that can predict the properties of successful IP agents and determine concentration ratios for successful FNP NC formation. For the complexation of an ionized basic API and acidic IP Ks

+



[API:IP] ⇌ [API ] + [IP ] +

For strongly disassociating and fully ionized API and IP, the starting concentrations of ionized compounds would be equal to the initial concentrations of the salt forms added

(3)

(8)

[IP−] = [IP−]0 − [API:IP]sat − [API:IP]sld

(9) (10)

Substitution of eqs 9 and 10 into eq 6 yields a general expression that relates the starting concentrations of aqueous API and IP with the moles of solid precipitated API:IP complex

(4)

{([API]0 − [API:IP]sat − [API:IP]sld )([IP]0 − [API:IP]sat − [API:IP]sld )}/[API:IP]sat = K s

(11)

Solving for the precipitated API:IP complex, the expression for the solid precipitate is obtained

(5)

[API:IP]sld =

Correspondingly, at or above saturation, the concentrations of aqueous API, IP, and API:IP complex are described by the equilibrium relationship of Ks and that of [API:IP]sat. While this equation is commonly reduced by combining the saturation concentrations and Ks terms to provide a solubility product, it is useful to separate these terms when assessing precipitation yields [API+][IP−] = Ks [API:IP]sat

[IP−]0 = [IP]0

[API+] = [API+]0 − [API:IP]sat − [API:IP]sld

Concentrations in square brackets are given in equivalents of anionic or cationic charge, so [API+] indicates API in the units of molarity, with the compound having a single cationic charge. At or above saturation of ion paired complex, the concentration of aqueous API:IP is equal to the aqueous saturation concentration of API:IP [API:IP] = [API:IP]sat

(7)

After exceeding the solubility of the API:IP complex, precipitates form. We denote the amount of precipitate formed as V × [API:IP]sld, where V is the volume of the system and [API:IP]sld is the concentration of API:IP complex above its saturation concentration prior to precipitation. This relationship would hold if the volume of the precipitated solid phase is much lower than that of the initial volume of reaction such that the volume of the solution phase is essentially constant. By mass balance, and combining eqs 7 and 8, the concentration of aqueous soluble ionized API and IP are as follows:



[API ][IP ] = Ks [API:IP]

[API+]0 = [API]0



1 ([API]0 + [IP]0 − 2[API:IP]sat 2

[API]0 2 + [IP]0 2 − 2([API]0 [IP]0 ) + 4K s[API:IP]sat ) (12)

This expression holds for concentrations above the API:IP saturation concentrations. For the complexation of the strongly basic polymyxin B and gentamicin at neutral pH ≈ 7 and at concentrations where precipitates had formed, these conditions are satisfied. Thus, under these assumptions and conditions, the bulk concentration of precipitated API:IP is a function of the starting concentrations of API and IP, the equilibrium constant

(6) E

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics Ks, and aqueous saturation solubility [API:IP]sat, all of which are readily determined. Whereas the Ks[API:IP]sat term is typically simplified to a Ksp solubility product, separation of these two terms provides mechanistic insight on precipitation yields. Ksp is a function of complexation strength and molecular actions between the API and IP (the charged groups between the API and IP), while the [API:IP]sat is a function of the hydrophobicity of the IP complex. Inspection of eq 12 provides insight on how to design precipitation reactions for NC processing, as shown in Figure 5. For a system with Ks and initial API and IP concentrations, the fraction of initial API that precipitates increases as the [API:IP]sat values decrease (Figure 5a). If the complex is below a critical [API:IP]sat value, no solids are formed. Highly hydrophobic IPs result in the formation of API:IP complexes with lower saturation solubilities, which is in agreement with our experimental results that higher LogP complexes are better precipitate formers. For a system with a defined [API:IP]sat and constant starting API and IP initial concentrations, the fraction of initial API that precipitates increases as the Ks of the complex decreases (Figure 5a). If the complex exhibits Ks above a critical value, no solids are formed. These results highlight that IPs with strong counterion interactions with the API maximize complex formation, which is reflected in the experimental results observed, where IPs with lower acid pKa are more effective in forming precipitates, even if the IP itself is less hydrophobic. For a system with a specific [API:IP]sat and Ks, the fraction of initial API that precipitates increases as the initial concentrations of API and IP increase (Figure 5b). Below a critical starting API and IP initial concentrations, no precipitates are formed. The complex can be driven to nearly complete complexation of API by increasing the IP concentration, which corresponds to Le Chatlier’s principle. However, this only occurs if individually the species are soluble. If one species precipitates or micellizes, then adding more of that species does not further drive the reaction. In that case, the individual ion concentrations in eq 3 would be given by the solubility product of the insoluble solid phase or by the CMC of the micellizing species. In that case an additional set of solubility products and mass balances would have to be added to the balances given by eqs 3−12. If there are two solid phases that can be formed, that is, an insoluble ion pair complex (API:IP) and a solid phase of the IP, then NCs can be formed, but the stoichiometry of the core of the NC will not be determined by the stoichiometry of the ion pair, but by the molar ratio of the complex and insoluble IP. These results provide a model to guide IP choice and precipitation reaction conditions for NC processing, and it supports the experimental results observed. Flash NanoPrecipitation of API:IP Complexes. The four IP candidates (OA, DDS, DS, PA) that passed the screening process were used to form NCs with the FNP process. Gentamicin and polymyxin B were precipitated with IPs and dissolved into THF as described above and mixed with PCL− PEG dissolved in THF to yield an organic THF stream containing both polymeric stabilizer and API:IP complex. This stream was rapidly mixed against water within a confined impingement jet to precipitate API:IP complex in the presence of the amphiphilic PCL−PEG to form NCs. DLS analysis of the resultant samples demonstrated that NCs 50−200 nm with narrow polydispersity were formed (Figure 6a,b, Table S1). No particles, only PCL−PEG micelles, were formed when only stabilizer and API or when only stabilizer and IP was used in

Figure 6. NC formation of APIs. Size distributions of NCs formed (A) using gentamicin (Ge) as a preformed API:IP complex, (B) using polymyxin B (Py) as a preformed API complex, (C) and with in situ ion pairing of polymyxin B. Compositions of NC formulations are given in the SI (Table S1).

the FNC process, with the exception of OA (Table S1). Large aggregates were formed when only API and IP were included during the FNC process, demonstrating that the inclusion of all three components, stabilizer, API, and IP, are simultaneously needed for NC formation. These results demonstrate that while water-soluble APIs cannot be formulated into NCs, hydrophobic API:IP complexes can be formed into NCs with the water-precipitation process. NC formation using preformed API:IP complex is, therefore, demonstrated. However, there are advantages to in situ API:IP NC complex formation since NC formation could be accomplished in a single step. This single step would simplify processing. To test in situ complexation and NC formation, polymyxin B dissolved in water was impinged against organic solvent containing IP and PCL−PEG in the FNC process. The IPs previously identified as capable of forming NCs using preformed complexes were studied using this method. Out of the IPs assessed, all converted polymyxin into hydrophobic forms in situ during mixing to yield stable NCs with narrow size F

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics distributions (Figure 6c). By comparing Figure 6b,c, where 6b is for the preformed IP and 6c is the in situ ion paired polymyxin, one sees the NC sizes produced by either process are essentially equivalent. Control experiments, where PCL− PEG and IP only or PCL−PEG and polymyxin B only were used, resulted in empty micelles and not in NC formation. Control experiments where IP and polymyxin B only were used resulted in the formation of aggregations, together demonstrating that all three components, stabilizer, API, and IP, are required for in situ complexation and NC formation. To measure the encapsulation efficiency of polymyxin B, free unencapsulated drug was separated from that which was encapsulated in NCs through ultrafiltration across a 100 kDa membrane and characterized with BCA analysis. The encapsulation efficiency is based on the relative ratio of free drug compared to the total amount of drug included in the FNC system encapsulation efficiency = 1 −

[polymyxin B]free [polymyxin B]total

(13)

At a 1:1 IP to API charge ratio, polymyxin B was encapsulated with efficiency greater than 95% for all IPs tested (Figure S1). At excess IP, high encapsulation efficiency was retained, but at a 0.5:1 IP to API charge ratio, encapsulation efficiency dropped significantly, to ∼70% in the case of sodium oleate. These results demonstrate that decreasing the amounts of IP relative to API that is present can cause decreased encapsulation efficiencies, by increasing the relative amount of solubilized API that is not retained in the core by ion pairing. NC Stability. NC stability was assessed by measuring the size distributions of the NCs when diluted into water or PBS, all at room temperature. NCs made with all IPs tested were fully stable when dispersed in DI water, showing no size change over time up to 72 h after NC formation (Figures S2−S4). When diluted 10-fold into PBS, however, particles exhibited size changes over time, which are driven by ion exchange between the anionic IP and the Cl− and PO4−3 ions in PBS. The ion exchange releases the bound API, which is soluble in the external phase. In the case of pamoic acid IPs, particles shrank in size, which suggests simple dissolution of encapsulated API from the NC core (Figure 7a). In the case of sodium dodecylbenzenesulfonate, sodium dodecyl sulfate, and sodium oleate, particles increased in size over time. These results are consistent with Ostwald ripening of particles, where complexes and API are released into solution but subsequently redeposit onto particles of larger size, resulting in particle size growth over time.27 The least hydrophobic of these IPs, sodium dodecyl sulfate, demonstrated the greatest size change rates over time, while those IPs with greater hydrophobicity and lower pKa exhibited size changes at a slower and lesser extent. Increasing the IP ratio above 1:1 increased stability. IP ratios above 1:4 resulted in NCs that did not change size, even in PBS, over 72 h. This is consistent with the model, which would show that increasing IP would drive higher complexation (Figure 8). API Release from NC. API release rates from NCs are a critical determinant of dosing frequency, in vivo active concentrations, and overall therapeutic effectiveness. The release rates of polymyxin B were assessed by enclosing NCs in an 8 kDa MWCO dialysis bag and diluting into a thousandfold excess volume of PBS to simulate an “open sink” condition at room temperature. Encapsulated polymyxin within dialysis

Figure 7. Stability of NCs when diluted into PBS over time. (A) Average sizes of polymyxin B NCs formed with various ion pairs during FNP processing. (B) Average sizes of polymyxin B NCs formed using sodium dodecyl sulfate ion pair during FNP processing. (C) Average sizes of polymyxin B NCs formed using sodium oleate ion pair during FNP processing. See SI Figure S5 for additional polydispersity index information. Size data are not shown for particles that had increased beyond the effective measurement range of the DLS by 48−72 h.

bags was measured over time, using the BCA assay, to determine the amount of API that had been released (Figure 9a). Figure 9 shows the release results for the three anionic groups studied: SDS, DBS, and OA, and it shows API:IP ratios for SDS and OA of 1:0.5 to 1:4. The release profiles for SDS and OA have different characteristics. The SDS has high water solubility, the “soft” sulfate anion couples well with the polymyxin B amines. The release is tunable and almost linear (i.e., first order release) over 3−9 days depending on the SDS−polymyxin ratio. This indicates a strong but mobile ionic interaction between the SDS and API. At the high concentration in the NC core, the SDS hydrophobic groups contribute to the stability of the complex. The hydrophobic surfactant chain tails prevent water and counterion entry that would weaken the ion pairing. This is G

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 8. Size distributions of NCs diluted into PBS when using (A,B) 1:1 API to IP charge ratio and (C,D) 1:4 API to IP charge ratio during NC assembly, when using sodium dodecyl sulfate and sodium oleate as ion pairs for polymyxin.

equivalent to “salt-bridges” in the hydrophobic cores of proteins. The salt bridges are merely ion pairs that would rapidly ion exchange if exposed to the external aqueous phase, but when in the hydrophobic core of the protein, they substantially contribute to the stability of the protein. The release process for the micellizing IP involves: (i) ion exchange to destabilize the ion pair, (ii) release the hydrophilic API, (iii) dissolve the micellizing IP agent, and (iv) expose the remaining IP:API to exchange. The oleic acid release follows a different pattern. Rather than continuous release, there is initial release but then containment with no further release. Again, the NC is created by the ionic interactions between polymyxin B and the carboxylic acid headgroup of the oleic acid molecule. The data in Figure 9c show a relatively rapid release within the first day, then a plateau where, essentially, no more API is released. As the concentration of hydrophobic IP (oleic acid) is increased, this plateau decreases to lower and lower amounts of the API being released. At 1:1 IP:API ratios, 90% of the API is released in 1 day; at 4:1 ratio 25% is released at 1 day, and that amount remains constant over 7 days. The mechanism of release appears to be (i) API at the surface of the NC is exchanged by counterions in the release media, and the API escapes, but the hydrophobic IP agent is retained on the NC surface, and (ii) accumulation of insoluble oleic acid (or oleate) at the NC surface creates a barrier for further API release. The solubility of oleic acid in water at ambient temperature is ∼0.3 × 10−6 g/g.28 The water−oleate remaining at the surface of the NC forms a liquid crystal phase.29 The lamellar structure of the liquid crystal phase, that is, the stacked C18 bilayers, provides a barrier for the release of the hydrophilic polymyxin B. This is similar to the barrier properties of a lipid bilayer in a liposome, which would encapsulate and prevent release of a hydrophilic, high molecular weight species. Figure 9c shows that, within increasing oleic acid concentration, less polymyxin

B is released before the remaining oleic acid phase can form the impermeable liquid crystal phase. While this permanent encapsulation with oleic acid may seem problematic, it actually offers the promise of novel delivery routes. Salentinig et al. note that at physiological pH the oleic acid has a limited degree of ionization that supports multilamellar layer formation.30 Using SAXS they show that, at pH below 6.8, the liquid crystal structure is eliminated, and an unstructured fluid phase is produced. This pH dependence indicates that in an endosome or locally inflamed tissue, at a lower pH, the liquid crystal phase may be disrupted, and the encapsulated polymyxin B may be released. This would clearly enable targeted release into sites of inflammation or into endosomes that may afford significant advantages for some drug applications. Together, the results for SDS and oleic acid ion pairing show that changing the phase behavior of the ion pairing agent can change the release characteristics of the soluble compound. It is possible to tune release profiles from continuous release from minutes to a week. The liquid crystal behavior of oleic acid provides a mechanism to encapsulate at high efficiency and high loading and to have release provided by pH shifts or possibly enzymatic action. The biological activity of released polymyxin B, after complexation, exposure to organic solvents during FNP, and decomplexation, is an important consideration. Here, it should be noted that polymyxin B is a locked macrocycle and does not undergo large structural changes, so we expect biological efficacy to be maintained. Future work will include measuring the activity of the polymyxin following release to offer more insight into how reversible HIP complexation may protect macromolecules such as peptides or proteins, with complex structures, from degradation and loss of function. H

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

The release profiles depend on the IP agent. For the micellizing SDS or dodecylbenzenesulfonate IPs, release can be tuned over a 7 day period with uniform release. For the more hydrophobic oleic acid, a process of sealing the NC surface occurs, which leads to containment of the soluble API in the core. Further investigations of the mechanism of this secondary containment and its use in triggered release are being pursued. The details of the structure and stoichiometry of the anionic IP interactions with the polycationic antibacterial agents is an interesting open question. The stability with SDS is especially interesting, given the extremely high solubility of SDS in solution; SDS is soluble in aqueous solution up to 8 mM prior to micellization. In our previous study with small molecule APIs with single ionic sites, the ion pairing was shown to be 1:1. Polymyxin B has four secondary amines and 11 amide nitrogens. The interactions with SDS are more likely to involve nonstoichiometric association. This SDS interaction may be similar to the SDS−protein interactions, which are the basis for SDS electrophoresis. The denatured protein backbone would be similar to the peptide backbone in polymyxin B. The SDS binding to proteins is highly cooperative and results from both ionic interactions and cooperative alkyl chain interactions among the SDS molecules associated with the backbone, as has been elucidated in the classic paper by Turro et al.31 The combination of ionic interactions and surfactant tail association is well-known to lead to intricate structures, which have been demonstrated for ionic polymers32 and DNA24 by X-ray diffraction. These more complex structures are not strictly stoichiometric but are determined by geometrical arrangement. It will be interesting to see if these peptide ion paired complexes display similar structural details.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.molpharmaceut.7b00824. Supplementary methods, figures, and tables (PDF)



Figure 9. Release rates of polymyxin NCs. Fraction of polymyxin B released when (A) using 1:1 API to IP charge ratio for various IPs. Fraction of polymyxin B released when (B) using sodium dodecyl sulfate IP and (C) sodium oleate IP, at various API:IP ratios. In (B) and (C), the charge ratio is given as 1:x, where x is the number of ion pair charge equivalents of the positively charged API.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Robert K. Prud’homme: 0000-0003-2858-0097



Notes

The authors declare no competing financial interest.

PERSPECTIVE AND CONCLUSIONS We have presented a method of transiently changing API water solubility to enable API processing with nanoprecipitation methods and present a theoretical framework for optimizing processing conditions. The rules for ion pair selection are based on the hydrophobicity and the pKa of the IP. We show that there are two processing routes; the IP:API complex can be preformed and isolated as a single component and then assembled into NCs via Flash NanoPrecipitation, or the IP:API complex can be formed in a precipitation where the hydrophobic IP agent, the stabilizing polymer, and the hydrophilic API can be assembled into the final NC during a single precipitation step. The former process has the advantage of knowing the stoichiometry of the IP:API a priori; the latter process has the advantage of being a simple, single precipitation.



ACKNOWLEDGMENTS We are grateful for support from the Princeton University Center for Health and Wellbeing (to H.D.L.) and Woodrow Wilson School of Public and International Affairs Program in Science, Technology, and Environmental Policy (to H.D.L.). This work was supported by the Princeton University SEAS grant from the Old School Fund (to R.K.P.). This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. #DGE-1656466 awarded to K.D.R.



REFERENCES

(1) Solaro, R.; Chiellini, F.; Battisti, A. Targeted Delivery of Protein Drugs by Nanocarriers. Materials 2010, 3 (3), 1928−1980.

I

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics (2) Immordino, M. L.; Dosio, F.; Cattel, L. Stealth liposomes: review of the basic science, rationale, and clinical applications, existing and potential. Int. J. Nanomed. 2006, 1 (3), 297−315. (3) Steichen, S. D.; Caldorera-Moore, M.; Peppas, N. A. A review of current nanoparticle and targeting moieties for the delivery of cancer therapeutics. Eur. J. Pharm. Sci. 2013, 48 (3), 416−427. (4) Galindo-Rodriguez, S. A.; et al. Polymeric nanoparticles for oral delivery of drugs and vaccines: a critical evaluation of in vivo studies. Crit. Rev. Ther. Drug Carrier Syst. 2005, 22 (5), 419−464. (5) Li, S.-D.; Huang, L. Pharmacokinetics and Biodistribution of Nanoparticles. Mol. Pharmaceutics 2008, 5 (4), 496−504. (6) Owens Iii, D. E.; Peppas, N. A. Opsonization, biodistribution, and pharmacokinetics of polymeric nanoparticles. Int. J. Pharm. 2006, 307 (1), 93−102. (7) Pattni, B. S.; Chupin, V. V.; Torchilin, V. P. New Developments in Liposomal Drug Delivery. Chem. Rev. 2015, 115 (19), 10938− 10966. (8) Gaudana, R.; et al. Design and evaluation of a novel nanoparticulate-based formulation encapsulating a HIP complex of lysozyme. Pharm. Dev. Technol. 2013, 18 (3), 752−759. (9) Patel, A.; Gaudana, R.; Mitra, A. K. A novel approach for antibody nanocarriers development through hydrophobic ion pairing complexation. J. Microencapsulation 2014, 31 (6), 542−550. (10) Muheem, A.; et al. A review on the strategies for oral delivery of proteins and peptides and their clinical perspectives. Saudi Pharm. J. 2016, 24 (4), 413−428. (11) Sohn, J. S.; et al. Polymer prodrug approaches applied to paclitaxel. Polym. Chem. 2010, 1 (6), 778−792. (12) Lu, H. D.; et al. Modulating Vibrio cholerae Quorum-SensingControlled Communication Using Autoinducer-Loaded Nanoparticles. Nano Lett. 2015, 15 (4), 2235−2241. (13) Singh, R.; Lillard, J. W. Nanoparticle-based targeted drug delivery. Exp. Mol. Pathol. 2009, 86 (3), 215−223. (14) Bao, G.; Mitragotri, S.; Tong, S. Multifunctional Nanoparticles for Drug Delivery and Molecular Imaging. Annu. Rev. Biomed. Eng. 2013, 15 (1), 253−282. (15) Ansell, S. M.; et al. Modulating the Therapeutic Activity of Nanoparticle Delivered Paclitaxel by Manipulating the Hydrophobicity of Prodrug Conjugates. J. Med. Chem. 2008, 51 (11), 3288−3296. (16) Patnaik, A.; et al. A phase I study of pivaloyloxymethyl butyrate, a prodrug of the differentiating agent butyric acid, in patients with advanced solid malignancies. Clinical cancer research 2002, 8 (7), 2142−2148. (17) Langer, R. Drug delivery and targeting. Nature 1998, 392 (6679), 5−10. (18) Zhang, L.; et al. Development of Nanoparticles for Antimicrobial Drug Delivery. Curr. Med. Chem. 2010, 17 (6), 585− 594. (19) Johnson, B. K.; Prud’homme, R. K. Flash NanoPrecipitation of Organic Actives and Block Copolymers using a Confined Impinging Jets Mixer. Aust. J. Chem. 2003, 56 (10), 1021−1024. (20) D’Addio, S. M.; Prud’homme, R. K. Controlling drug nanoparticle formation by rapid precipitation. Adv. Drug Delivery Rev. 2011, 63 (6), 417−426. (21) Saad, W. S.; Prud’homme, R. K. Principles of nanoparticle formation by flash nanoprecipitation. Nano Today 2016, 11 (2), 212− 227. (22) Bruno, B. J.; Miller, G. D.; Lim, C. S. Basics and recent advances in peptide and protein drug delivery. Ther. Delivery 2013, 4 (11), 1443−1467. (23) Jain, A.; et al. Crit. Rev. Ther. Drug Carrier Syst. 2013, 30 (4), 293−329. (24) Rädler, J. O.; et al. Structure of DNA-Cationic Liposome Complexes: DNA Intercalation in Multilamellar Membranes in Distinct Interhelical Packing Regimes. Science 1997, 275 (5301), 810−814. (25) Johnson, B. K.; Prud’homme, R. K. Chemical processing and micromixing in confined impinging jets. AIChE J. 2003, 49 (9), 2264− 2282.

(26) Pinkerton, N. M.; et al. Formation of Stable Nanocarriers by in Situ Ion Pairing during Block-Copolymer-Directed Rapid Precipitation. Mol. Pharmaceutics 2013, 10 (1), 319−328. (27) Liu, Y. Ostwald ripening of beta-carotene nanoparticles. Phys. Rev. Lett. 2007, 98 (3), 036102. (28) Khuwijitjaru, P.; Adachi, S.; Matsuno, R. Solubility of saturated fatty acids in water at elevated temperatures. Biosci., Biotechnol., Biochem. 2002, 66 (8), 1723−1726. (29) Borné, J.; Nylander, T.; Khan, A. Phase behavior and aggregate formation for the aqueous monoolein system mixed with sodium oleate and oleic acid. Langmuir 2001, 17 (25), 7742−7751. (30) Salentinig, S.; Sagalowicz, L.; Glatter, O. Self-assembled structures and p K a value of oleic acid in systems of biological relevance. Langmuir 2010, 26 (14), 11670−11679. (31) Turro, N. J.; et al. Spectroscopic probe analysis of proteinsurfactant interactions: the BSA/SDS system. Langmuir 1995, 11 (7), 2525−2533. (32) Antonietti, M.; Conrad, J.; Thuenemann, A. PolyelectrolyteSurfactant Complexes: A New Type of Solid, Mesomorphous Material. Macromolecules 1994, 27 (21), 6007−6011.

J

DOI: 10.1021/acs.molpharmaceut.7b00824 Mol. Pharmaceutics XXXX, XXX, XXX−XXX