Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Chapter 11
Inverse Flash NanoPrecipitation for Biologics Encapsulation: Nanoparticle Formation and Ionic Stabilization in Organic Solvents Robert F. Pagels and Robert K. Prud’homme* Department of Chemical and Biological Engineering, Princeton University, Princeton, New Jersey 08544, United States *E-mail:
[email protected].
Inverse Flash NanoPrecipitation, or iFNP, is a recently developed scalable, controllable, and reproducible process to produce nanoparticles highly loaded with water soluble molecules, such as biologics. The nanoparticles produced by iFNP have a hydrophilic core and hydrophobic corona, and are dispersed in organic solvents. However, most nanoparticle applications, particularly biomedical applications, require that the particle be stable in aqueous environments. Here, we demonstrate that the size of these “inverted” nanoparticles can be controlled between 40 and 300nm, and that they can be stabilized by ionically gelling the poly(acrylic acid) core with metal cations. The effect of solvent and salt on crosslinking were investigated. A method to evaluate the crosslinking efficiency was developed, and a number of metals were found to be effective at crosslinking the particle core, including Ca2+, Zn2+, and Fe3+. Once stabilized, these particles may be further processed for applications in aqueous environments.
© 2017 American Chemical Society Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Introduction
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Delivery of Biologically Derived Therapeutics Over the last decade, the pharmaceutical industry has placed increasing emphasis on biologically derived therapeutics, or biologics. Biologics, including proteins and peptides, are the most rapidly growing segment of the pharmaceutical marketplace (1). Protein and peptide therapeutics are large, with molecular weights in the thousands to hundreds of thousands of Daltons, which imparts chemical complexity. Because of this chemical complexity, biologics can more specifically interact with targets in the body, leading to higher efficacy and fewer side effects (2). The same complexity that provides biologics with superior specificity and potency also limits the ways in which these new medicines can be administered to patients. Proteins and peptides have poor oral bioavailability and, in most circumstances, must be delivered by injection (3). These injections can be required daily or even multiple times a day, as is the case for insulin, the prototypical biologic. Once injected, some proteins, such as humanized antibodies, may have long circulation times. However peptide therapeutics, which lack tertiary structure, and non-exogenous proteins can be cleared by proteases on the order of minutes (4). In response to these limitations, polymeric drug delivery vehicles have been developed to alter the pharmacokinetics of biologics. The goals of these delivery vehicles include shielding the biologic from proteases and the immune system, and slowing the release of the biologic into the blood stream over time. Ideally, these vehicles can decrease the frequency of injection required for biologic drugs. The size scale of the vehicle determines its function. Nanoparticle based delivery systems, on the order of 100nm in size, administered intravenously can circulate in the bloodstream for several days, and can help target the delivery of the drug to specific tissues. Microparticles, on the order of 50μm in size, administered by subcutaneous or intramuscular injection, are capable of releasing drugs over a longer period of time (up to months) due to their larger size (5). There are two types of polymeric delivery vehicles: those comprised of water-soluble polymers, and those comprised of water-insoluble polymers. Water soluble polymers are chemically crosslinked to form hydrogels. Water soluble therapeutics, such as biologics, are released by diffusing through the hydrogel mesh (6). Hydrogel-based systems are ideal for large proteins which are structurally sensitive to the solvents and processing required for water-insoluble polymers. However, small proteins and peptides can rapidly diffuse through the hydrogel mesh, resulting in fast release upon injection (7). Water-insoluble polymers, such as poly(lactic-co-glycolic acid) (PLGA), can encapsulate biologics by forming impermeable shells around pores of the drug. As the polymer degrades, channels through which the biologic can diffuse are formed, and the drug is slowly released (8). While release from hydrophobic matrices can be much slower than that from hydrogel matrices, the design and processing of these constructs is more complex. Most commonly, biologics are loaded into PLGA through a solid-in-oil-in-water (S/O/W) emulsion or a water-in-oil-in-water (W1/O/W2) double emulsion (9). First, the biologic (either 250 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
as a solid, S, or aqueous solution, W1) is dispersed in an oil phase, typically dichloromethane, which contains the dissolved hydrophobic polymer. This solution is then emulsified in a second aqueous phase. When the solvent is removed, the polymer hardens around pockets of the encapsulated drug. The emulsion processes described above are difficult to optimize. An ideal drug delivery vehicle should have high drug loading (the weight percent of drug relative to the total mass of the vehicle), high encapsulation efficiency (the fraction of the drug used in the formulation that is actually encapsulated within the vehicle), and near-zero order release kinetics. For PLGA particles produced by emulsion processes, optimizing for one of these requirements results in losses in another (9). For example, increasing loading results in greater losses of biologic to the external aqueous phase during processing, and faster release. Slow and controlled release requires small, widely spaced pores of the biologic, and is more readily achievable at low drug loadings. There are examples of water soluble drugs that have been successfully formulated into PLGA particles, including some commercial products such as the Lupron Depot®. However, a near-universal process to form delivery vehicles which meet the requirements listed above has yet to be developed. Inverse Flash NanoPrecipitation for the Encapsulation of Water-Soluble Drugs Flash NanoPrecipitation (FNP), is an industrially-scalable and controllable continuous precipitation technique, which has been used to produce nanoparticles highly loaded with hydrophobic drugs (10). In the traditional FNP process, an amphiphilic block copolymer and a hydrophobic drug are molecularly dissolved in a water-miscible organic solvent. This solvent stream is rapidly mixed with an aqueous antisolvent using special mixing geometries, inducing high supersaturations of the hydrophobic drug. The drug nucleates and grows to form the particle core, and the growth of the core is halted by the self-assembly of the hydrophobic block of the copolymer onto the particle surface. Because the mixing time is faster than the particle assembly time, which is set by the characteristic time of block copolymer self-assembly, the resulting nanoparticles have a low polydispersity (11). Additionally, because the mixing can be done in a continuous process, FNP is highly scalable. Traditionally, FNP has been used with highly hydrophobic drugs (logP > 4), which results in high loadings and encapsulation efficiencies due to the low solubility in the aqueous antisolvent. Weakly hydrophobic materials with charged groups can be ion-paired with hydrophobic counter-ions to induce higher supersaturations in the aqueous antisolvent (12). However, until recently, there was not a method to encapsulate water soluble therapeutics such as biologics (see Table 1 for the material requirements of FNP). In a recent innovation, the FNP process has been “inverted” for the encapsulation of water soluble molecules (9). Inverse FNP, or iFNP, is used to produce particles with a hydrophilic core and a hydrophobic shell (Table 1 and Figure 1). In this process, a water-soluble drug and amphiphilic stabilizing block copolymer are dissolved in a polar solvent such as methanol or dimethyl sulfoxide (DMSO). Currently, poly(n-butyl acrylate)-b-poly(acrylic acid) (PBA-b-PAA) 251 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
has been used as the stabilizing polymer. This solvent stream is rapidly mixed with a non-polar antisolvent, such as chloroform (CHCl3) or acetone, which precipitates the drug and the PAA block of the copolymer.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Table 1. Comparison Between FNP and iFNP Component
FNP
iFNP
Suitable Drug
· Hydrophobic
· Hydrophilic
Stabilizing polymer
· Amphiphilic
Solvent
· Water-miscible solvent
· Polar solvent
· Good solvent for core material and both blocks of stabilizing polymer · Miscible with antisolvent Antisolvent
· Aqueous
· Non-polar solvent
· Antisolvent for core material and one block of stabilizing polymer · Miscible with solvent Resulting particle core
· Drug and hydrophobic polymer block
· Drug and hydrophilic polymer block
Resulting particle corona
· Hydrophilic
· Hydrophobic
Figure 1. Inverse Flash NanoPrecipitation (iFNP) method and the resulting particle structure.
252 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
The resulting particles are highly loaded (typically ≥ 50wt%) with the watersoluble molecule. As with traditional FNP, the encapsulation efficiency is very high provided that the therapeutic has limited solubility in the antisolvent stream. These inverted nanoparticles are sterically stable in the organic antisolvent due to the solvophilic PBA brush on the surface, however further processing is required to produce a particle that is stable in the aqueous environments required for drug delivery applications.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Processing Inverted Nanoparticles If placed directly into an aqueous environment, the inverted nanoparticles produced by iFNP will (1) burst or invert, releasing the water soluble contents, or (2) aggregate due to the hydrophobic polymer shell. Several processing steps must be taken to ensure that this does not occur (9). First, the anionic PAA core of the inverted particle is crosslinked or gelled to prevent the particles from disassembling in water. There are examples of nanoparticles containing PAA cores loaded with water soluble drugs in which the PAA is covalently crosslinked (13–15). In those systems, the drug is loaded after the PAA is gelled to ensure that drug is not affected or modified by the crosslinking chemistry. Much higher drug loadings are possible in iFNP than in post-loading processes, however this requires that the PAA crosslinking must be done in the presence of the drug. Covalent crosslinking might have side reactions with functional groups on the biologic, which would complicate FDA approval of the delivery system. Therefore, we have avoided covalent crosslinking and instead focused on ionic crosslinking. In ionic crosslinking, multivalent metal cations interact with the acid groups of the PAA to stabilize the particles. Acid/base interactions can also crosslink particle cores, which is commonly used in the formulation and delivery of siRNA and has also been used with highly charged peptides (16). Next, the inverted particles are processed into a final form that is suitable for drug delivery (Figure 2). There are two possible processing routes. First, the inverted nanoparticles can be coated with a second block copolymer in another FNP step to produce a hydrophilic polymer brush on the particle surface. Typically the hydrophilic block is poly(ethylene glycol) (PEG), which is commonly used in nanoparticle drug delivery applications to give long circulation times (17). Alternatively, the inverted nanoparticles can be used to replace the W1 phase of the double emulsion process. The nanoparticles, in addition to a small amount of additional hydrophobic polymer to help glue the particles together, are emulsified in an aqueous phase and the solvent is removed to form hardened microparticles. This is superior to the traditional W1/O/W2 or S/O/W emulsion processes because the polymer shell of the inverted nanoparticles limits drug losses to the external aqueous phase during processing. In addition, the shell could slow the release of the drug by limiting the formation of percolating pathways through the particle. Finally, the high loading of the inverted nanoparticles allows for the formation of similarly highly loaded microparticles (9). 253 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Figure 2. Possible processing routes for inverted nanoparticles produced by iFNP. The hydrophobic particle shell can be coated with an amphiphilic PEG-containing block copolymer to create nanoparticles that are sterically stabilized in water. The hydrophobic shell can also allow the particles to be well dispersed in a matrix of hydrophobic polymer, such as PLGA.
In the current form, iFNP is very promising for the production of nanoparticle and microparticle formulations of peptides and other water soluble therapeutics. In this chapter we discuss how the iFNP process was developed. The focus of this chapter is on the formation and ionic stabilization of the inverted nanoparticles. Due to the proof-of-concept nature of the work presented here, non-degradable model polymers were used, as well as model hydrophilic molecules which are not of actual therapeutic interest. This work has laid the foundation for all future studies on the drug delivery function of the particles produced by iFNP.
Experimental Methods Materials Optima ® chloroform (CHCl3), HPLC grade dimethyl sulfoxide (DMSO), and ACS grade glacial acetic acid, methanol (MeOH), and acetone were purchased from Fisher Scientific. Poly(n-butyl acrylate)-b-poly(acrylic acid) (PBA-b-PAA, 7.5-b-5.5kDa, PDI of 1.5) and poly(styrene)-b-poly(ethylene glycol) (PS-b-PEG, 1.6-b-5kDa, PDI of 1.1) were purchased from Polymer Source, Inc. (Dorval, Quebec). Lysozyme from chicken egg whites (≥90%, 43000units/mg solid), tobramycin, vancomycin hydrochloride from Streptomyces orientalis, diethylenetriaminepentaacetic acid gadolinium (III) dehydrate 254 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
(Gd-DTPA, 97%), L-glutathione (GSH, reduced, ≥98%), eosin Y (92% dye content), and tartrazine (~80% dye content) where purchased from Sigma Aldrich. The crosslinking salts were all ACS grade and purchased from Sigma Aldrich in the hydrate form listed in the crosslinking section of this chapter. DL-tryptophan was purchased from Eastman Chemical Company. Peptide I was supplied as a gift from Amgen. Deionized water (MQ) (18.2 MΩ·cm) was generated by a NANOpure Diamond UV ultrapure water system (Barnstead International, Germany). All materials were used as received. Nanoparticle Synthesis: Inverse Flash NanoPrecipitation Nanoparticles were produced by iFNP. The stabilizing block copolymer (BCP), poly(n-butyl acrylate)-b-poly(acrylic acid) (PBA-b-PAA) and the core material, if any, was dissolved in a good solvent at the appropriate concentration. A PBA3kDa-b-PAA12kDa molecular weight polymer was employed unless otherwise indicated. Solvents used include MeOH and DMSO, sometimes with the addition of small volume fractions of water. This solvent stream was rapidly mixed with an equal volume of antisolvent (CHCl3 or acetone) in a handheld confined impingement jet (CIJ) mixer (17). The effluent of the mixer was collected in an antisolvent bath such that the final solvent to antisolvent ratio was 1:9 (v:v). The resulting particles had a core containing the PAA and a shell of PBA. All nanoparticle formulations used this chapter are given in Table 2.
Table 2. iFNP Formulations [Parts have been reproduced with permission from reference (9). Copyright (2015) Elsevier.] Sample
Solvent Stream
Antisolvent Stream
Bath
Section: Examples of Core Materials Tartrazine
5mg/mL tartrazine, 10mg/mL BCP, DMSO
CHCl3
CHCl3
Eosin Y
5mg/mL Eosin Y, 10mg/mL BCP, DMSO
CHCl3
CHCl3
Gd-DTPAa
5mg/mL Gd-DTPA, 5mg/mL BCP, DMSO with 5v% H2O
CHCl3
CHCl3
Tobramycinb
10mg/mL BCP in DMSO
5.5mg/mL tobramycin in DMSO
Acetone
Peptide Ic
5mg/mL peptide I, 5mg/mL BCP, DMSO with 5v% H2O
CHCl3
CHCl3
Tryptophand
5mg/mL tryptophan, 5mg/mL BCP, DMSO with 5v% H2O and 5v% acetic acid
CHCl3
CHCl3 Continued on next page.
255 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Table 2. (Continued). iFNP Formulations [Parts have been reproduced with permission from reference (9). Copyright (2015) Elsevier.] Sample
Antisolvent Stream
Solvent Stream
Bath
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Section: Size Control Lysozymec
Base case: 5mg/mL lysozyme, 5mg/mL BCP, DMSO
Base case: CHCl3
Base case: CHCl3
Vancomycinc
Base case: 5mg/mL vancomycin, 5mg/mL BCP, DMSO with 5v% H2O
CHCl3
CHCl3
Section: Structural Analysis 2% H2O
10mg/mL BCP in deuterated MeOH with 2v% H2O
Deuterated CHCl3
Deuterated CHCl3
10% H2O
10mg/mL BCP in deuterated MeOH with 10v% H2O
Deuterated CHCl3
Deuterated CHCl3
Section: Metal Crosslinking Empty
10mg/mL BCP, MeOH or DMSO with 5v% H2O
CHCl3 or acetone
CHCl3 or acetone
Glutathionea (GSH)
5mg/mL GSH, 5mg/mL BCP, DMSO with 5v% H2O
CHCl3
CHCl3
a
Core material first dissolved in water. b Particles assemble by charge association. PBA7.5kDa-b-PAA5.5kDa used as BCP. d Core material first dissolved in acetic acid.
c
Nanoparticle Characterization The nanoparticle size was characterized by dynamic light scattering (DLS) with a Zetasizer Nano-ZS (Malvern Instruments). The particles were diluted 10x in antisolvent, and measurements were taken at 25°C using a 173o detection angle and a 632nm helium neon laser. Normal Mode analysis was used, and intensity average peak 1 diameters are reported. Zeta potentials were measured in a disposable folded capillary cell in phosphate buffered saline diluted tenfold. Transmission electron microscopy (TEM) samples were prepared by placing the nanoparticle solution on a grid and allowing it to dry under ambient condition overnight. The nanoparticles were imaged using a Philips CM100 TEM (Eindhoven, The Netherlands) with an accelerating voltage of 100 kV. 1H
NMR Structural Analysis
Nanoparticles were synthesized as described above, with 10mg/mL PBA3kDa-b-PAA12kDa in the MeOH solvent stream. Deuterated solvent (d-MeOH) and antisolvent (d-CHCl3) were used. Particles were made with increasing amounts of water in the solvent stream, which partitions to the hydrophilic core of the nanoparticles. Proton nuclear magnetic resonance (1H NMR) spectra were 256 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
recorded with a Bruker AVANCE III (500 MHz) spectrometer with a CryoProbe optimized for 1H detection, and used to interpret the particle structure.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Metal Crosslinking of Anionic Particle Cores Multi-valent metal salts were dissolved in the appropriate solvent (MeOH or DMSO) and then diluted ten-fold with the appropriate antisolvent (CHCl3 or acetone). In some cases, small amounts of acetic acid were added to solubilize the salt. This salt solution was mixed with an equal volume of nanoparticles in the same solvent/antisolvent mixture such that there was one charge equivalent of metal for every equivalent of acrylic acid. This solution was incubated at room temperature to allow the metals to enter the particle core and ionically interact with the PAA. After incubation for a predetermined amount of time, the particles were diluted ten-fold with DMSO. DLS was used to determine if the particles were stable, swelled, or dissolved in DMSO. Other good solvents, such as MeOH, can also be used to test for crosslinking. However, we found that MeOH could not distinguish between weakly and strongly crosslinked nanoparticles. DMSO was employed because it appears to be a more rigorous crosslinking test. Coating of Stabilized Particles for Aqueous Dispersion Chromium-crosslinked (4 days) glutathione (GSH) particles were diluted 4fold with acetone to decrease the density of the solvent phase. This solution was centrifuged for 15 minutes at 20,000rcf. The supernatant was decanted, and the pellet was washed two times with acetone to ensure that all of the water-immiscible CHCl3 was removed. The particles were re-suspended in acetone with PS1.6kDa-bPEG5kDa such the final concentrations were 5mg/mL of nanoparticles and 5mg/mL PS-b-PEG. In a second FNP step, this solution was rapidly mixed with an equal volume of water using a CIJ mixer. The mixer effluent was collected in a bath of water such that the final acetone:water volumetric ratio was 1:9.
Results and Discussion Examples of Core Materials Prior to encapsulation, each potential core material was tested in the iFNP without a stabilizing BCP. These controls resulted in large, visible aggregates. The lack of large aggregates when the BCP is present is evidence that the hydrophilic materials are successfully encapsulated in the nanoparticle core. We were able to encapsulate several water soluble molecules using iFNP. Synthetic small molecules that were encapsulated include the water-soluble dyes tartrazine and Eosin Y, as well as the gadolinium-based MRI contrast agent Gd-DTPA (Figure 3a). Biologically derived molecules that were encapsulated include tobramycin (a polysaccharide-based antibiotic), peptide I (a seven-amino acid long peptide supplied as a gift from Amgen), and tryptophan (a single amino acid) (Figure 3b). In addition to those shown in Figure 3, we also successfully 257 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
encapsulated the protein lysozyme, the peptide-based antibiotic vancomycin, and the tri-peptide glutathione, all of which will be shown in later sections. The materials encapsulated only share one thing in common: minimal solubility in non-polar organic solvents. They differ largely both in size (lysozyme is ~12,000Da while tryptophan is only ~200Da) and charge (tartrazine has a charge density of -1 per 178Da, while tobramycin has a charge density of +1 per 93Da). Despite these wide differences, most formulations resulted in ≤100nm particles. The tobramycin particles were larger. However, these were assembled through charge complexation (tobramycin is highly positively charged), not solvent based precipitation. This shows that iFNP can be used universally to encapsulate water soluble materials, provided that the appropriate solvent and antisolvent have been determined.
Figure 3. Size distributions of inverted nanoparticles with different core materials. (a) Synthetic small molecules tested in iFNP included the water soluble dyes tartrazine and eosin Y, and the MRI imaging contrast agent Gd-DPTA. (b) Biologically derived molecules tested in iFNP include the antibiotic tobramycin, a seven amino acid peptide (peptide I), and the amino acid tryptophan (Trp). Other biologically derived encapsulated molecules include the antibiotic vancomycin, the tripeptide glutathione, and the small protein lysozyme. Data for those formulations are given in later sections. [Adapted with permission from reference (9). Copyright (2015) Elsevier.]
Control over Inverted Nanoparticle Size In “traditional” FNP – that is, FNP in which water is the antisolvent and particles with hydrophobic cores are formed – the mixing Reynolds number (Re), total solids concentration, and ratio of core material to stabilizing block copolymer have been shown to control the resulting nanoparticle size (18–21). In general, high Re of mixing are employed such that mild fluctuations in mixing rate have little to no effect on the particle size (11, 20). Therefore, practically, only the concentrations of block copolymer and core material are varied to control particle 258 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
size. Because the core materials in traditional FNP are hydrophobic, water or aqueous buffers are invariably used as the antisolvent stream. Inverted FNP has a complementary set of variables that can be used to tune particle size. The core materials that can be encapsulated using iFNP fall under the broad category of “hydrophilic,” but this does not specify which non-aqueous solvent will best serve as the antisolvent stream. Additionally, as we look towards encapsulating larger peptides and proteins, there can be structural and chemical inconsistency within even a single molecule of core material (i.e. a single molecule can contain both solvophobic and solvophilic regions). This means that the choice of antisolvent can affect the nucleation and growth rate of the growing particle core, as well as the morphology of the precipitate. The choice of the good-solvent can also be an important factor in controlling the size of nanoparticles produced by iFNP. In most cases, solvent and anti-solvents are chosen such that they are completely miscible. If the solvent and antisolvent phase separate, the core material will stay in the good-solvent phase and nanoparticles will not be formed. The solvents which best precipitate hydrophilic materials (hexane, toluene, diethyl ether, dichloromethane, chloroform, etc.) are not water miscible, so a non-aqueous good solvent needs to be chosen. However, a small volume fraction of water may be required to dissolve some hydrophilic materials, even in polar solvents such as DMSO. The presence of small volume fractions of non-miscible solvents (specifically water in the case of iFNP) can also affect the final particle size. We produced particles with a lysozyme core and a PBA-b-PAA shell in which four variables were tested: (1) total mass concentration of polymer and lysozyme, (2) ratio of core material to polymer, (3) the choice of antisolvent (specifically ratios of chloroform and acetone), and (4) the volume percent of water in the DMSO good-solvent stream. The base formulation was 5mg/mL of PBA7.5kDa-b-PAA5.5kDa and 5mg/mL lysozyme (i.e. 10mg/mL total mass concentration, 50wt% core) with a DMSO good solvent (no water), and CHCl3 antisolvent (no acetone). Each variable was changed independently from this base formulation. Additionally, the effect of the ratio of core material to stabilizing polymer was tested with vancomycin, a peptide antibiotic, as the core material. The results are given in Figure 4. First, we observed that increasing the total mass concentration caused an increase in the nanoparticle size (Figure 4a). This is similar to the trend observed with traditional FNP, and has previously been attributed to an increase in the growth rate of the particles relative to the nucleation rate (18). This is predicted to occur at high supersaturations. We also observed increases in particles sizes caused by high loadings of core material (>40wt% for lysozyme, and >60wt% for vancomycin, Figure 4b). Once again, this has been reported previously for FNP with hydrophobic materials. This can be simply understood as a geometric effect – increasing the amount of core material (volume) relative to the stabilizing polymer (surface area) will result in fewer and larger particles. Interestingly, at loadings below 25wt% for lysozyme and 62.5wt% for vancomycin, there is an initial downward trend in particle size. There are several plausible explanations for this trend. First, the addition of core material may help the anionic PAA better pack in the core by shielding electrostatic 259 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
repulsions. Second, at low amounts of core material, the nucleation rate may be increasing more rapidly than the growth rate. Finally, solvophilic sections of the lysozyme or vancomycin may be able to partially stabilize the particle surface. The effect of good solvent and antisolvent composition has not been a significant variable for traditional FNP, but these solvent choices are important for iFNP. For example, Bilati and coworkers found that, when precipitating lysozyme from DMSO, the antisolvent composition changed the resulting precipitate size (22). However, in those experiments, the mixing times were slower than in iFNP and no stabilizing block copolymer was used.
Figure 4. The effect of iFNP process parameters on the resulting particle size with a lysozyme core and PBA-b-PAA shell. (a) Effect of total mass concentration at a set lysozyme loading of 50wt%. (b) Effect of loading at a constant total mass concentration of 10mg/mL. Lysozyme particles were made with a pure DMSO solvent stream, and vancomycin particles were made with 5v% water in the DMSO stream with a total mass concentration of 10mg/mL. [Reproduced with permission from reference (9). Copyright (2015) Elsevier.] (c) Effect of water in the DMSO solvent stream. (d) Effect of anti-solvent CHCl3/acetone composition. Error bars are +/- the standard deviation for three samples. 260 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
The addition of water to the DMSO solvent stream (Figure 4c) had a similar effect to increasing the amount of core material. This is because the water phase separates from the CHCl3 antisolvent and becomes an additional core material. High water contents (10v% and 20v%) produced milky nanoparticle solutions suggestive of an emulsion and consistent with the phase separated water partitioning to the particle core. The antisolvent composition had little effect on the particle size below 50v% acetone in CHCl3, however antisolvent streams consisting of mostly acetone resulted in a large increase in particle size (Figure 4d). In fact, a pure acetone antisolvent stream resulted in the formation of large visible aggregates that could not be analyzed by DLS. This is due to a mismatch in the lysozyme and PAA-block precipitation rates. Acetone is a worse antisolvent for lysozyme than CHCl3, resulting in lower supersaturations and slower nucleation and growth upon mixing. When the nucleation and growth of the core material is slower than the block copolymer self-assembly, empty polymer nanoparticles will be formed along with large aggregates of the core material. For traditional FNP this means that the core material needs to have a logP close to or greater than 4. The antisolvent needs to be chosen for iFNP such that the core material faces similarly high supersaturations upon mixing. Together, through independently tuning four process variables, we were able to produce nanoparticles ranging from 50 to 400nm in diameter. In general, the process variables that have previously been used to tune particle size in traditional FNP may be similarly employed in iFNP. Structural Analysis For traditional FNP with hydrophobic core materials, the nanoparticles have been shown to have a core-shell structure by transmission electron microscopy (TEM) (21). This is the first time that FNP has been used to make nanoparticles with hydrophilic cores and hydrophobic coronas, therefore we sought to verify that these particles also have the expected core-shell structure. The predicted particle structure is a core consisting of the insoluble PAA block and any other hydrophilic material, and a shell consisting of the soluble PBA block (Figure 1). In particle formulations with little or no water in the solvent stream, the particle core should be a solid, while the shell is solvated and mobile. Only mobile, solvated protons are visible in traditional 1H NMR, therefore only PBA proton peaks should be measured in formulations with low amounts of water (23). On the other hand, in formulations with high amounts of water the mobility of the core will increase, and proton peaks corresponding to PAA should become visible (Figure 5a). We made empty nanoparticles (no core material) using deuterated solvent (d-MeOH) and antisolvent (d-CHCl3), with increasing amounts of water in the solvent stream (2v% and 10v%). From here on, we will simply refer to these two samples as “2%” and “10%.” The particle formulations differed slightly in size – the 2% sample was 40nm and the 10% sample was 60nm. This increase in size is consistent with the lysozyme results given in Figure 4c, and corresponds to the water swelling the particle core (see Figure 5a). 261 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Figure 5. 1H NMR experiment to determine the structure of the inverted nanoparticles. (a) Adding water to the solvent stream during iFNP results in particles with swollen and more mobile cores. (b) 1H NMR spectra of the 2% particles (bottom) and 10% particles (top). The PBA proton peaks are unchanged by the addition of water, while the PAA peaks are only visible in the 10% formulation. In the 1H NMR spectra of the 2% particles, only the PBA protons are visible (Figure 5b). This supports the prediction that formulations with low water contents will have a solid core. The PAA peaks, which match other reported spectra (24), are visible in the 10% formulation, supporting the hypothesis that the water swells the core and increases the PAA mobility. Importantly, there is no significant change in the PBA proton peaks between the 2% and 10% formulations. This suggests that there is a clean core-shell interface, with little PBA entrapped in the PAA core. A dense hydrophobic brush is beneficial in the future processing of these particles. In nanoparticle delivery applications, this brush will collapse into a hard shell which is necessary to slow the release of drug into the external aqueous phase. Similarly, when processing these inverted particles into microparticle formulation, the hydrophobic shell will frustrate the formation of percolating channels and slow the release of drug. 262 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Particle Stabilization with Metal Cations The nanoparticles produced by iFNP are ideal for drug delivery applications. They can be packed with large amounts of hydrophilic materials, and the 1H NMR-verified shell of hydrophobic polymer will slow the rate of drug release as described above. However, for any medical application, these particles need to be processed into an aqueous solvent. The solvophobicity of the core materials holds the particles together in organic solvents, and this force is lost in water. Therefore, a stabilization step is required to keep the particles from bursting or falling apart in aqueous solvents. The carboxylic acid side chains in the PAA can be crosslinked to form a hydrogel core. Once gelled, the particle core may swell in water but should otherwise be stable to aqueous processing. There are many covalent methods to crosslink acid groups, however most water soluble drugs, and all peptides and proteins, will have acid groups on them as well. It is important to avoid chemically modifying the drug in any way, therefore we chose to ionically crosslink the PAA core with multivalent metal cations (Figure 6). Ionic crosslinking is reversible, so even if the metal cations interact with the drug, the drug will not be permanently modified (15). However, certain metal cations can participate in oxidation-reduction reactions, therefore care must still be taken in choosing metals for ionic crosslinking.
Figure 6. Crosslinking of anionic polymers with multivalent cations. The metal can bind to the PAA both through simple charge interactions as well as through chelation effects. For macroscopic hydrogels, the crosslinking density or mesh size of the gel can be determined by measuring the volume uptake of water (25). For our application we would prefer particle cores that swell little in water and have a tight mesh to help hold in the water-soluble core materials. Water cannot be used to test swelling of the iFNP produced particles because it would result in aggregation of the hydrophobic shell. Therefore swelling experiments were performed in DMSO which is a good solvent for both the core and the shell. After diluting the particles in DMSO, DLS was used to monitor particle swelling or dissolution. In DMSO, the particles may display three different behaviors (Figure 7). In the ideal case, the particles will appear the same size in both CHCl3 and DMSO, indicting tight crosslinking (many stable crosslinks formed by the metal). If only a small amount of metal has entered the particle core, or if the crosslinks are weak, the particles may swell significantly in DMSO which indicates loose crosslinking. Finally, if an insufficient amount of metal has entered the core, or if the metal-PAA interactions are weak, the particles may dissolve completely in DMSO. 263 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Figure 7. Particle behavior in a good solvent. Tightly crosslinked particles (small mesh size) will have a similar size in both CHCl3 and DMSO, while weakly crosslinked particles will swell. If there is insufficient or no crosslinking, the particles will dissolve completely in DMSO.
Preliminary Tests with Chromium (III) Initially, crosslinking tests were completed on empty nanoparticles (no additional core material, just the PAA block) using chromium (III) acetate hydroxide ((CH3CO2)7Cr3(OH)2). Chromium is not relevant for pharmaceutical applications, but was used because of its high solubility in organic solvents and its strong interactions with acid groups. Nanoparticles that were 60nm in diameter were made with a DMSO solvent stream and a CHCl3 antisolvent stream (final solvent ratio: 1:9 DMSO:CHCl3). The nanoparticle solution was diluted 2x with a chromium solution also in a 1:9 DMSO:CHCl3 mixture. The final particle concentration was 0.5mg/mL and the PAA:Cr3+ charge ratio was 1:1. The crosslinking was monitored by diluting the particles 10-fold in DMSO, mixing for 2 minutes, and then measuring their size on DLS. The results are given in Figure 8a. The three behaviors illustrated in Figure 7 were observed. At time points less than 20hrs no particles were detected, indicating that little chromium had entered the particle core. The particles swelled less in DMSO as the incubation time increased from 20 to 48hrs, indicating a transition from loose to tight crosslinking. In the final time point the particles were close to their original size in CHCl3. In order to increase the rate of crosslinking, CrBr3 was also tested because, from the hard-soft acid base (HSAB) theory, the chromium should be less strongly bound to the bromide anions. While CrBr3 did appear to crosslink the particles rapidly, when the solution was open to the air the more reactive chromium also 264 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
formed chromium oxide precipitates. This complicated the interpretation of the DLS results, so chromium acetate hydroxide was preferred over CrBr3. The stability of the crosslinks was tested by holding the particles in DMSO and measuring the size over 8hrs (Figure 8b). The particles continued to swell for the first two hours, then maintained a relatively stable size. The interactions between the chromium and PAA are strong, preventing the particles from fully dissolving.
Figure 8. Nanoparticle crosslinking with chromium acetate hydroxide. (a) The formation of particles stable in DMSO was observed after crosslinking for 20hrs. Crosslinking density increased over the next two days. (b) The particles crosslinked for 48hrs swelled in DMSO over 8hrs, but remained stable. Error bars are from three dilutions and measurements. Dashed lines are visual guides.
Particles were made with a glutathione (GSH) core and crosslinked with chromium acetate hydroxide. The presence of additional core material does not prevent the ability to crosslink the particles (Figure 9a). To further demonstrate the stability of the Cr3+-crosslinking, the GSH nanoparticles were coated with an equal mass of PS-b-PEG in a second FNP step. This step is similar to the coating of polystyrene-latex spheres with PEG containing block polymers, which has been done using FNP previously (26, 27). The coated nanoparticles were stable in water and only slightly larger than the uncoated particles (Figure 9a) and had a near-neutral zeta potential (Figure 9b). Without the chromium crosslinking, the nanoparticles would have burst during the second FNP step and coating with PEG would not be possible. The chromium also provided electron contrast to the otherwise organic nanoparticles, which was imaged by TEM (Figure 9c). The cores visible in TEM match the DLS data well. Chromium strongly crosslinks the PAA core of inverted particle produced by iFNP, however it is not ideal for biomedical applications. We sought to find other metal salts that could similarly gel the particle core but would be more biologically acceptable.
265 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Figure 9. Crosslinking and coating of GSH nanoparticles. (a) Inverted nanoparticles were 110nm as made in CHCl3, and swelled to 130nm in MeOH after crosslinking with chromium. Adding the PS-b-PEG coating increased the particle size to 160nm. (b) The coated nanoparticles had a near-neutral zeta potential (-3.8mV) in 0.1x PBS. (c) Uniform particles are observed by TEM. The chromium provides electron contrast to nanoparticle core.
Crosslinking with Other Metals For over half a century, researchers have ionically gelled PAA with metal cations (28). However, in almost every case, the crosslinking is done in water (29–33). An immediate hurdle is that the salts must be soluble in the organic solvents used in iFNP. Salts were screened for their solubility first in the typical good solvents used in iFNP (MeOH and DMSO) at the same charge concentration as PAA in that stream. Small amounts of acetic acid was added to help solubilize the salt in some cases. Those that dissolved were then diluted with 9 volume equivalents of antisolvent (acetone or CHCl3) and checked for precipitation over a half an hour. The results of this solubility screening are given in Table 3. Empty nanoparticles were made with each solvent/antisolvent combination given in Table 3. The MeOH/acetone did not produce nanoparticles, possibly because the 1:1 solvent composition in the CIJ mixer was not a strong enough antisolvent to precipitate the PAA block. Crosslinking was tested on the three formulations that did form well defined nanoparticles (DMSO/CHCl3, MeOH/ CHCl3, and DMSO/acetone). Metal solutions were mixed with an equal volume of empty nanoparticles made in the same solvent composition. Because the MeOH/acetone particles did not form well, zirconium isopropoxide isopropanol (Zr Iso.Iso.), iron nitrate, and basic aluminum acetate dissolved in MeOH/acetone was tested on nanoparticles made with DMSO/acetone. Iron chloride was also tested on the three nanoparticles both with and without the addition of acetic acid. The particles were mixed for five days, and then tested for crosslinking by diluting them ten-fold in DMSO and measuring their size using DLS. We wanted to quantitatively rank the crosslinking effectiveness of each formulation, so we developed a metric to grade the correlation function of each sample. 266 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Table 3. Solubility of metal salts in solvent/antisolvent mixtures. Check marks indicate solvent combinations in which the salt was soluble. Metal Salt
Acetone
CHCl3 DMSO
MeOH
DMSO
MeOH
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Al(OH)(CH3CO2)2 NH4Al(SO4)2 *12 H2O
b
b
b
Ca(CH3CO2)2
a
a
a
b
b
b
b
Ca(NO3)2 *4 H2O CoSO4 *7 H2O
a
Gd(III) acetylacetonate
b a
FeCl3 *6 H2O Fe(NO3)3 *9 H2O b
NH4Fe(SO4)2
b
b
b
SrCl2 *6 H2O Zn(CH3CO2)2 a
Zr(OCH(CH3)2)4*(CH3)2CHOH a
Addition of acetic acid required. formed when centrifuging.
b
Addition of acetic acid required, and small pellet
The correlation function measured by DLS gives how quickly the scattering intensity changes over time due to the Brownian motion of particles in solution (34). A good correlation function should have several features. First, it should start close to one, indicating that at short times the scattering signal is not changing (see the MeOH/CHCl3 sample in Figure 10a). Starting values much lower than one indicate that most of the measured signal is simply a product of noise. When testing crosslinking, this could mean either that the particles have completely dissolved, or that they have swelled so much with DMSO that there is little difference in the refractive index of the particle and solvent. Either way, correlation functions that start close to zero are an indication of poor crosslinking. The next characteristic of a DLS correlation function is a drop-off from one to zero. The correlation time where this drop-off occurs gives information about the size of the colloids diffusing through the solution. Smaller particles diffuse more rapidly, and the drop-off will occur at shorter times (see the MeOH/CHCl3 and DMSO/CHCl3 samples in Figure 10b). When testing crosslinking, short drop-off times indicate small amounts of swelling and strong crosslinking. Finally, as implied above, at long correlation times the correlation function should go to zero. Correlation functions that fluctuate above zero at long times indicates either the presence of large aggregates in the sample, or that the signal to noise ratio is low (see the DMSO/acetone and control samples in Figure 10a,b).
267 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Figure 10. Ranking correlation functions. a. The as-measured correlation function (correlation coefficient, C.C., versus correlation time) for the three FeCl3 samples and the no-salt control diluted in DMSO. The MeOH/CHCl3 sample starts closest to 1, indicating the strongest scattering signal. b. The correlation functions of the same samples normalized to start at 1. The MeOH/CHCl3 function drops off at a shorter time than the DMSO/CHCl3 sample, which was not as clear in (a). This shows smaller particles and tighter crosslinking. c. The metric used to rank the correlation functions. In order to quantitatively rank the metal salts, the three characteristics of a good correlation function described above were distilled into a single metric (Figure 10c). The first term gives how far the correlation function starts from one, and ranges from 0 (strong signal) to 1 (weak signal). The second term is the correlation coefficient at an arbitrarily long time (10,000μs), which should be close to zero. The third and final term gives the drop-off time by normalizing the start of the correlation function to 1 and finding when the correlation coefficient first crosses 0.5 (good correlation functions are monotonically decreasing, and therefore will only cross 0.5 once; however, particularly noisy samples may cross several times). In order to have the third term in the same range as the first two (0 for the best crosslinking to 1 for the poorest), we take the logarithm of the time and divide by 5 (log10(100,000μs)). Taken together, the lower the sum of these terms, the better the crosslinking. From this metric, we split the crosslinking conditions into three groups: good, weak, and poor (Table 4). Iron nitrate and zirconium isopropoxide isopropanol both resulted in strong correlation functions, however we could not confirm that the scattering came from crosslinked particles and not insoluble salts, so they are not included in Table 4. We first notice that the addition of acetic acid does not strongly affect the ability of the metals to crosslink the particle core (FeCl3 in MeOH/CHCl3 was a strong crosslinking agent with and without acetic acid). This is good because acetic acid was required for several salts to be soluble in the organic solvents. The solvent/antisolvent combination clearly has an effect on the crosslinking. This is particularly obvious with FeCl3, which was tested on all solvent 268 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
combinations (Figure 10a,b). In general, particles in which CHCl3 was the antisolvent had a greater chance for successful crosslinking than those with acetone as an antisolvent: 53% of the CHCl3 samples had good crosslinking compared to only 27% of the acetone samples. One potential explanation for this observation is that, in the CHCl3 case, the immiscible water is localized in the nanoparticle core. Solvating and swelling the particle core will help the salt partition into the PAA where it can interact with, and crosslink, the acid groups. On the other hand, acetone is water miscible, therefore there is less of a driving force for the water to swell the particles. Without water, the salts will not be able to penetrate the solid core, and only acid groups at the PBA/PAA interface will be accessible for crosslinking. The solvent also appears to play a role in the crosslinking efficiency: 70% of the MeOH samples had good crosslinking, compared to only 30% for the DMSO samples. This effect may be enhanced by the fact that there were no MeOH/ Acetone samples, but there were DMSO/Acetone samples. Nonetheless, DMSO has long been known to form complexes with different metal cations (35). The weaker crosslinking generally observed with DMSO samples may be because the DMSO interacts more strongly with salts than MeOH, slowing the rate at which the metals will enter the particle core. The choice of metal salt also clearly effects the degree of crosslinking. For example, calcium acetate, zinc acetate, and gadolinium acetylacetonate showed good or at least weak crosslinking for all solvent combinations tested. On the other hand, calcium nitrate, ammonium aluminum sulfate, and cobalt sulfate showed poor crosslinking for all solvent combinations tested. Winkleman et al. discussed the effects of the metal salt when crosslinking hydrated PAA films (36). In their work they found several features that can prevent metal salts from crosslinking PAA. First, metals that are strongly bound to their counter-ion fail to crosslink because the salt form of the cation is thermodynamically preferred to the crosslinked form. Similarly, cations that do not strongly interact with acetate groups will not crosslink PAA. Finally, metal salts may not crosslink due to kinetic effects. For example, salts that are strongly solvated by water will be slow to exchange anions. The salt characteristics described by Winkleman will play a role in our system as well, however we have added hurdles. The metal salts must be soluble in organic solvents, which immediately limits which salts can be used. Additionally, salts behave differently in organic solvents than in water. The low dielectric constant keeps the salts associated, slowing anion exchange. The ion exchange may be further slowed by complexation with the solvent itself. Finally, and perhaps most importantly, we did not control the pH of the system. Generally PAA is neutralized prior to crosslinking, commonly with NaOH (31, 36). Sodium is weakly bound to acetate groups, and the deprotonated form of PAA is primed to interact with metal cations. Unfortunately, the sodium salt of PAA is insoluble in DMSO and MeOH, therefore we used the block copolymer in the protonated form. In many cases this will push the equilibrium away from crosslinking. For example, when using a nitrate salt, the byproduct of crosslinking will be nitric acid. Nitric acid is more acidic than PAA, therefore this is unfavorable. In future work, it will be important 269 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
to understand how the addition of bases to modify the pH within the particle core can increase the strength and speed of crosslinking.
Table 4. Rankings of the Crosslinking Conditions Rank
Salt
Solvent Composition
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Good Crosslinking (1)
FeCl3 + acetic acid
MeOH/CHCl3
(2)
NH3Fe(SO4)2
DMSO/Acetone
(3)
Ca(CH3CO2)2
DMSO/CHCl3
(4)
FeCl3
MeOH/CHCl3
(5)
Gd (III) acetylacetonate
MeOH/CHCl3
(6)
Gd (III) acetylacetonate
DMSO/CHCl3
(7)
Zn(CH3CO2)2
MeOH/CHCl3
(8)
Ca(CH3CO2)2
MeOH/CHCl3
(9)
Gd (III) acetylacetonate
DMSO/Acetone
(10)
Al(OH)(CH3CO2)2
DMSO/Acetone
(11)
Al(OH)(CH3CO2)2
MeOH/CHCl3
(12)
Zn(CH3CO2)2
DMSO/CHCl3
(13)
CoSO4
MeOH/CHCl3 Weak Crosslinking
(14)
FeCl3
DMSO/CHCl3
(15)
Zn(CH3CO2)2
DMSO/Acetone
(16)
Ca(CH3CO2)2
DMSO/Acetone Poor Crosslinking
n/a
NH4Al(SO4)2
DMSO/CHCl3, MeOH/CHCl3, DMSO/Acetone
n/a
Ca(NO3)2
DMSO/CHCl3, MeOH/CHCl3, DMSO/Acetone
n/a
CoSO4
DMSO/CHCl3, DMSO/Acetone
n/a
NH4Fe(SO4)2
DMSO/CHCl3, MeOH/CHCl3
n/a
FeCl3
DMSO/Acetone
n/a
FeCl3 + acetic acid
DMSO/CHCl3, DMSO/Acetone
n/a
SrCl2
DMSO/Acetone
270 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Despite these hurdles, some salts relevant to biomedical applications did successfully crosslink the nanoparticles. In particular, salts of Ca2+, Zn2+, and Fe3+ all crosslinked well under specific conditions. Interestingly, Gd3+ also crosslinked the nanoparticles. Though toxic, gadolinium-containing compounds are used as MRI contrast agents (37). In future animal studies, gadolinium crosslinking may be used to help trace where the particles accumulate over time. Additional studies will also be necessary to determine how ions leak out of the particles over time both in storage and in use. The hydrophobic interface surrounding each particle should have a low ion permeability, however ion exchange across this layer may still prove to be an important variable in determining drug release rates.
Conclusions We have developed a process, iFNP, to form polymeric nanoparticles that are sterically stable in organic solvents. The iFNP process has been demonstrated on nine different water soluble compounds ranging in molecular weight and charge. With lysozyme as a model core material, we established four ways to control the nanoparticle diameters ranging from 50 to 300nm: the mass concentration of solids, ratio of core to stabilizing polymer, the antisolvent choice, and the water content of the solvent stream. With lysozyme, loadings up to 62.5wt% were achieved, while with vancomycin stable particles were formed even at 80wt% loadings. For future applications in aqueous environments, we have shown that the anionic core of the nanoparticles can be stabilized with metal cations. Initially chromium was used as a stabilizer, however the screening of several salt and solvent combinations revealed that more biologically compatible metals, including Fe3+, Zn2+, and Ca2+, may also be used. While the DLS correlation functions reveal that the particles are indeed crosslinked, future work on controlling the pH within the nanoparticle core may help increase the degree of crosslinking. The work described in this chapter is at the proof of concept stage. Moving forward, there are additional points which we are addressing. Most importantly, the polymers used here were not biodegradable. In order to determine the release kinetics of entrapped molecules, the work here will need to be translated to biodegradable polymers. We have recently demonstrated a completely biocompatible and biodegradable system through the replacement of PAA by poly(aspartic acid) and PBA by poly(lactic acid). We will be publishing these results shortly. The current iFNP process is suitable for the formulation and delivery of peptide therapeutics, antibiotics and vaccine components – all of which do not require maintenance of protein secondary structure. Many peptide therapeutics have short half-lives, therefore the ability to make particle-based formulations with high loadings, as we have shown here, is an important advancement in the drug delivery field. The ability of iFNP to encapsulate and release protein therapeutics with secondary structure is currently being addressed. We will be using concepts recently employed in the protein delivery field to achieve high protein concentrations without denaturation. Certainly, many proteins can tolerate 271 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
non-aqueous media contact, as evidenced by the use of enzyme as biocatalysts in organic solvent. However, some proteins will lose secondary structure. The rules for which proteins can successfully be processed by iFNP is an active research area in the group.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
Acknowledgments The authors would like to thank Istvan Pelczer, the Director of the NMR Facility in the Princeton Department of Chemistry, for his assistance in acquiring the NMR spectra of the nanoparticles. We would also like to thank Dr. Christina Tang for her help with TEM imaging. This work was supported by a grant from the Princeton Old Guard Fund and the Princeton Innovation Fund. Robert Pagels was supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGA 1148900. Support is acknowledged from Optimeos Life Sciences Inc.
References 1.
Peters, S. Biotech Products in Big Pharma Clinical Pipelines Have Grown Dramatically. Tufts CSDD Impact Report 2013, 15, 1. 2. Mitragotri, S.; Burke, P. A.; Langer, R. Overcoming the challenges in administering biopharmaceuticals: formulation and delivery strategies. Nat. Rev. Drug Discovery 2014, 13, 655–672. 3. Frokjaer, S.; Otzen, D. E. Protein drug stability: a formulation challenge. Nat. Rev. Drug Discovery 2005, 4, 298–306. 4. Sato, A. K.; Viswanathan, M.; Kent, R. B.; Wood, C. R. Therapeutic peptides: technological advances driving peptides into development. Curr. Opin. Biotechnol. 2006, 17, 638–642. 5. Lee, B. K.; Yun, Y. H.; Park, K. Smart nanoparticles for drug delivery: Boundaries and opportunities. Chem. Eng. Sci. 2015, 125, 158–164. 6. Peppas, N. A.; Bures, P.; Leobandung, W.; Ichikawa, H. Hydrogels in pharmaceutical formulations. Eur. J. Pharm. Biopharm. 2000, 50, 27–46. 7. Lu, S.; Anseth, K. S. Release behavior of high molecular weight solutes from poly(ethylene glycol)-based degradable networks. Macromolecules 2000, 33, 2509–2515. 8. Batycky, R. P.; Hanes, J.; Langer, R.; Edwards, D. A. A theoretical model of erosion and macromolecular drug release from biodegrading microspheres. J. Pharm. Sci. 1997, 86, 1464–1477. 9. Pagels, R. F.; Prud’homme, R. K. Polymeric nanoparticles and microparticles for the delivery of peptides, biologics, and soluble therapeutics. J. Controlled Release 2015, 219, 519–535. 10. Saad, W. S.; Prud’homme, R. K. Principles of nanoparticle formation by flash nanoprecipitation. Nano Today 2016, 11, 212–227. 11. Johnson, B. K.; Saad, W.; Prud’homme, R. K. Nanoprecipitation of pharmaceuticals using mixing and block copolymer stabilization. ACS Symp. Ser. 2006, 924, 278–291. 272 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
12. Pinkerton, N. M.; Grandeury, A.; Fisch, A.; Brozio, J.; Riebesehl, B. U.; Prud’homme, R. K. Formation of stable nanocarriers by in situ ion pairing during block-copolymer-directed rapid precipitation. Mol. Pharm. 2013, 10, 319–328. 13. Bontha, S.; Kabanov, A. V.; Bronich, T. K. Polymer micelles with crosslinked ionic cores for delivery of anticancer drugs. J. Controlled Release 2006, 114, 163–174. 14. Kabanov, A. V.; Vinogradov, S. V. Nanogels as Pharmaceutical Carriers: Finite Networks of Infinite Capabilities. Angew. Chem., Int. Ed. 2009, 48, 5418–5429. 15. Kim, J. O.; Nukolova, N. V.; Oberoi, H. S.; Kabanov, A. V.; Bronich, T. K. Block ionomer complex micelles with cross-linked cores for drug delivery. Polym. Sci., Ser. A 2009, 51, 708–718. 16. Gupta, K.; Ganguli, M.; Pasha, S.; Maiti, S. Nanoparticle formation from poly(acrylic acid) and oppositely charged peptides. Biophys. Chem. 2006, 119, 303–306. 17. Han, J.; Zhu, Z.; Qian, H.; Wohl, A. R.; Beaman, C. J.; Hoye, T. R.; Macosko, C. W. A simple confined impingement jets mixer for flash nanoprecipitation. J. Pharm. Sci. 2012, 101, 4018–4023. 18. D’Addio, S. M.; Prud’homme, R. K. Controlling drug nanoparticle formation by rapid precipitation. Adv. Drug Delivery Rev. 2011, 63, 417–426. 19. Shen, H.; Hong, S.; Prud’homme, R. K.; Liu, Y. Self-assembling process of flash nanoprecipitation in a multi-inlet vortex mixer to produce drug-loaded polymeric nanoparticles. J. Nanopart. Res. 2011, 13, 4109–4120. 20. Johnson, B.; Prudhomme, R. Flash NanoPrecipitation of organic actives and block copolymers using a confined impinging jets mixer. Aust. J. Chem. 2003, 56, 1021–1024. 21. Pustulka, K. M.; Wohl, A. R.; Lee, H. S.; Michel, A. R.; Han, J.; Hoye, T. R.; McCormick, A. V.; Panyam, J.; Macosko, C. W. Flash nanoprecipitation: particle structure and stability. Mol. Pharm. 2013, 10, 4367–4377. 22. Bilati, U.; Allemann, E.; Doelker, E. Development of a nanoprecipitation method intended for the entrapment of hydrophilic drugs into nanoparticles. Eur. J. Pharm. Sci. 2005, 24, 67–75. 23. Heald, C.; Stolnik, S.; Kujawinski, K.; De Matteis, C.; Garnett, M.; Illum, L.; Davis, S.; Purkiss, S.; Barlow, R.; Gellert, P. Poly(lactic acid)-poly(ethylene oxide) (PLA-PEG) nanoparticles: NMR studies of the central solidlike PLA core and the liquid PEG corona. Langmuir 2002, 18, 3669–3675. 24. Liu, A.; Honma, I.; Ichihara, M.; Zhou, H. Poly(acrylic acid)-wrapped multi-walled carbon nanotubes composite solubilization in water: definitive spectroscopic properties. Nanotechnology 2006, 17, 2845–2849. 25. Canal, T.; Peppas, N. A. Correlation between mesh size and equilibrium degree of swelling of polymeric networks. J. Biomed. Mater. Res. 1989, 23, 1183–1193. 26. Budijono, S. J.; Russ, B.; Saad, W.; Adamson, D. H.; Prudhomme, R. K. Block copolymer surface coverage on nanoparticles. Colloid Surf., A 2010, 360, 105–110. 273 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.
Downloaded by NORTH CAROLINA STATE UNIV on December 29, 2017 | http://pubs.acs.org Publication Date (Web): November 15, 2017 | doi: 10.1021/bk-2017-1271.ch011
27. D’Addio, S. M.; Saad, W.; Ansell, S. M.; Squiers, J. J.; Adamson, D. H.; Herrera-Alonso, M.; Wohl, A. R.; Hoye, T. R.; Macosko, C. W.; Mayer, L. D.; Vauthier, C.; Prud’homme, R. K. Effects of block copolymer properties on nanocarrier protection from in vivo clearance. J. Controlled Release 2012, 162, 208–217. 28. WALL, F.; DRENAN, J. Gelation of Polyacrylic Acid by Divalent Cations. J. Polym. Sci. 1951, 7, 83–88. 29. Gustafson, R.; Lirio, J. Binding of divalent metal ions by cross-linked polyacrylic acid. J. Phys. Chem. 1968, 72, 1502. 30. Habert, A.; Huang, R.; Burns, C. Ionically crosslinked poly(acrylic acid) membranes. 1. Wet technique. J. Appl. Polym. Sci. 1979, 24, 489–501. 31. Ikegami, A.; Imai, N. Precipitation of polyelectrolytes by salts. J. Polym. Sci. 1962, 56, 133. 32. Lahav, M.; Narovlyansky, M.; Winkleman, A.; Perez-Castillejos, R.; Weiss, E. A.; Whitesides, G. M. Patterning of poly(acrylic acid) by ionic exchange reactions in microfluidic channels. Adv Mater 2006, 18, 3174+. 33. Wei, Z.; He, J.; Liang, T.; Oh, H.; Athas, J.; Tong, Z.; Wang, C.; Nie, Z. Autonomous self-healing of poly(acrylic acid) hydrogels induced by the migration of ferric ions. Polym. Chem. 2013, 4, 4601–4605. 34. Pecora, R. Dynamic light scattering measurement of nanometer particles in liquids. J. Nanopart. Res. 2000, 2, 123–131. 35. MEEK, D.; STRAUB, D.; DRAGO, R. Transition Metal Ion Complexes of Dimethyl Sulfoxide. J. Am. Chem. Soc. 1960, 82, 6013–6016. 36. Winkleman, A.; Perez-Castillejos, R.; Lahav, M.; Narovlyansky, M.; Rodriguez, L. N. J.; Whitesides, G. M. Patterning micron-sized features in a cross-linked poly(acrylic acid) film by a wet etching process. Soft Matter 2007, 3, 108–116. 37. Caravan, P.; Ellison, J.; McMurry, T.; Lauffer, R. Gadolinium(III) chelates as MRI contrast agents: structure, dynamics, and applications. Chem. Rev. 1999, 99, 2293–2352.
274 Ilies; Control of Amphiphile Self-Assembling at the Molecular Level: Supra-Molecular Assemblies with Tuned ... ACS Symposium Series; American Chemical Society: Washington, DC, 2017.