Surface Functionalization of Polymer Particles for Cell Targeting by Modifying Emulsifier Chemistry

Abstract

The oil in water emulsion/solvent extraction method is used to fabricate many FDA approved, polymer particle formulations for drug delivery. However, these formulations do not benefit from surface functionalization that can be achieved through tuning particle surface chemistry. Poly(vinyl alcohol) (PVA) is the emulsifier used for many FDA approved formulations and remains associated with the particle surface after fabrication. We hypothesized that the hydroxyl groups in PVA could be conjugated with biomolecules using isothiocyanate chemistry and that these modifications would endow the particle surface with additional functionality. We demonstrate that fluorescein isothiocyanate and an isothiocyanate derivatized mannose molecule can be covalently attached to PVA in a one-step reaction. The modified PVA polymers perform as well as unmodified PVA in acting as an emulsifier for fabrication of poly(lactide-co-glycolide) particles. Particles made with the fluorescein modified PVA exhibit fluorescence confined to the particle surface, while particles made with mannose modified PVA bind concanavalin A. In addition, mannose modified PVA increases particle association with primary macrophages by three-fold. Taken together, we present a facile method for modifying the surface reactivity of polymer particles widely used for drug delivery in basic research and clinical practice. Given that methods are established for conjugating the isothiocyanate functional group to a wide range of biomolecules, our approach may enable PVA based biomaterials to engage a multitude of biological systems.

Graphical Abstract

1. Introduction

The oil in water (O/W) emulsion/solvent extraction method for fabricating biodegradable polymer particles is widely practiced in biomedical laboratories around the world and utilized in 19 FDA approved extended release drug formulations^1–3, including Vivitrol, Risperdal, and Lupron depot^4–6. For many of these formulations, poly(lactide-co-glycolide) (PLG) is dissolved in dichloromethane (DCM) and this organic solution is emulsified in an aqueous solution of poly(vinyl alcohol) (PVA), which serves as the emulsifier. The emulsion is then added to a large volume of water and the DCM is extracted from the dispersed phase, which results in the formation of polymer particles. These polymer particles are primarily composed of PLG, but PVA polymers remain associated with the surface of the particle after purification and freeze drying⁷. While several techniques are established for forming biodegradable polymer particles, the O/W emulsion/solvent extraction technique remains one of the most utilized at both the laboratory and pharmaceutical scales because of its scalability and the ease of tuning particle parameters^2,8,9. Indeed, properties such as size, degradation rate, and drug release rate are easily addressed by modulating emulsion ratios, homogenization speeds, and PLG composition and molecular weight⁹. In contrast, tuning particle surface chemistry is challenging due to the tendency of the emulsifier (PVA for most FDA approved controlled release applications¹⁰) to remain associated with the particle surface.

Tuning particle surface chemistry is of interest because it can impact particle-cell interactions^11–13. A common exploitation of this principle is attaching ligands to the particle surface that support interactions with cells that express the cognate receptor¹⁴. For example, particles were functionalized with mannose in order to bind macrophages, which express lectins that bind mannose, and this strategy improved particle uptake¹⁵. In another study, polymer particles were coated with an RGD-containing peptide and this led to enhanced uptake by macrophages compared to control particles¹⁶. In addition to these applications, particle surface chemistry may also be leveraged to disrupt cellular processes; in one study, particles were functionalized with polyethyleneimine, which imparted pH buffering properties that allowed for endosomal escape of the particle into the cytosol¹⁷. As many drug delivery applications are hindered by off-target effects¹⁸, surface functionalization offers an avenue to deliver the drug to specific cells or even specific compartments within cells.

Particle surface functionalization is usually achieved through covalent linkage or passive adsorption of the chemical modifier. Although this approach is effective, it requires an additional reaction step following particle formation. Attaching the chemical modifier post particle formation also has the potential drawback of a chemical reaction yield that is hard to control and difficult to characterize¹². In addition, particle yield will likely be decreased by the additional purification steps. Modifying particle surface chemistry during particle formation would address these issues, and if it could be achieved with the O/W solvent/extraction method, new drug formulations with additional surface functionalities could be produced using the PLG/PVA system. Further, the currently FDA approved extended release formulations that might benefit from additional surface reactivity could be easily modified.

In the current study, our goal was to alter the surface chemistry of PLG particles by chemically modifying PVA prior to its addition to the O/W emulsion. As mentioned earlier, PLG particles made with the O/W emulsion/solvent extraction method, in which PVA is the emulsifier, contain PVA on the surface⁷. Thus, we hypothesized that functionalizing PVA for ligand presentation could achieve alterations in particle surface function. We investigated modifying PVA by attaching isothiocyanate containing molecules to the alcohol groups. It is established that isothiocyanates react with alcohols to form thiourethanes¹⁹, and we employed this approach to attach pendant groups on PVA (Figure 1A). Controlling surface chemistry of PLG particles with isothiocyanate-modified PVA, to our knowledge, has not been investigated. Importantly, methods for conjugating isothiocyanate groups to biomolecules are established²⁰; thus, our approach could allow for the modification of PVA with a wide range of molecules that interface with biological systems.

Figure 1. Synthesis of Modified PVA.

A) Isothiocyanates (ITC) were conjugated to PVA via a one-step reaction. B) PVA and ITC were dissolved in DMSO. The solution was stirred for either 5 or 24 hours at a predetermined temperature ranging between 25 and 100°C. The solution was then dialyzed against water. C) Photograph showing FITC-PVA produced when FITC was added to the reaction at 0.1, 1 and 10% by mol of OH groups present in PVA.

As a proof of principle, we investigated two model isothiocyanates. The first was fluorescein isothiocyanate, a fluorophore, which enabled us to investigate the spatial location of the isothiocyanate on the particle. The second was an isothiocyanate derivatized mannose, a monosaccharide that enhances particle-macrophage interactions and enabled us to investigate if particle associations with these cells could be modulated with this approach. Mannose was also selected because macrophages are involved in many biological processes, and there are many particle-based therapies that could benefit from increased interactions with macrophages^21,22.

In the work that follows, we demonstrate that the two isothiocyanates can be covalently attached to PVA through a mild, one-step reaction and detail how reaction conditions impact the extent of PVA modification for each isothiocyanate. We then investigate the use of the modified PVA in the production of PLG particles and study how particle surface reactivity changes in response to the modified PVA employed. Taken together, this work presents a facile method for controlling particle surface function that does not require a functionalization step after particle formation. Given the importance of surface chemistry in cell-material interactions, this work may aid in the development of new biomaterial-based therapies and diagnostics as well as the repurposing of existing FDA approved extended/controlled drug release formulations.

2. Materials and Methods

2.1. Materials

Poly (D,L-lactide-co-glycolide) (PLG) (50:50, ester terminated, M_W 7,000–17,000), Poly (vinyl alcohol) (PVA) (M_W 13,000–23,000, 87–89% hydrolyzed), dichloromethane (DCM), coumarin 6, fluorescein isothiocyanate isomer 1 (FITC) and mannan from Saccharomyces cerevisiae were purchased from Sigma (St. Louis, MO). α-D-mannopyranosyl 4-phenylisothiocyanate (MITC) was purchased from Synthose (Concord, ON, Canada). Ethanol was purchased from Decon Laboratories (King of Prussia, PA). Dimethyl sulfoxide (DMSO), annexin v binding buffer, fetal bovine serum (FBS), trypsin with 0.25% EDTA and pen/strep were purchased from Fisher (Hampton, NH). Ultrapure water was obtained from a Thermo Scientific Barnstead Nanopure system. Dimethyl sulfoxide d6 was purchased from Cambridge Isotope Labs (Tewksbury, MA). Concanavalin A rhodamine (ConA) was purchased from Vector Labs (Burlingame, CA). RAW 264.7 macrophages were obtained from ATCC (TIB-71). Dulbecco’s Modified Eagle’s Medium (DMEM) with 4.5 g/L glucose and L-glutamine and DMEM/F12 was purchased from Corning Cellgro (Corning, NY). Recombinant macrophage colony stimulating factor (MCSF) was purchased from Cell Guidance Systems (Cambridge, UK). Antibodies for flow cytometry, Trustain fcX (anti-CD16/32) and APC anti-mouse F4/80 antibody (clone BM8) were purchased from BioLegend (San Diego, CA, USA).

2.2. Modification of Poly (Vinyl Alcohol) (PVA)

The reaction diagram is shown in Figure 1. Modified PVA was prepared by dissolving 100 mg of PVA into 10 mL of DMSO along with a mass of isothiocyanate (fluorescein isothiocyanate [FITC] or α-D-mannopyranosyl phenyl isothiocyanate [MITC]) that corresponded to a predetermined mol% of OH groups present in the PVA. For these studies, this value ranged from 0.1–20 mol%. The solution was stirred continuously at a predetermined temperature (25, 60, 80, or 100°C) for either 5 or 24 hours. Then, the solution was added to dialysis tubing with a 3.5 kDa MWCO and dialyzed (protected from light) against 5 L of ultrapure water for 3 days, with exchanges of water twice a day. Following dialysis, the PVA product was frozen and lyophilized overnight. Final dried product was weighed, and mass yield was calculated via equation 1.

$$\mathit{\text{Mass}} \mathit{\text{Yield}} \left(\text{\%}\right)=\left(\frac{M_{\mathit{\text{product}}}}{M_{PVA}+M_{ITC}}\right)*100$$ (1)

Where M_product is the mass of PVA recovered from the reaction, M_PVA is the initial mass of PVA and M_ITC is the initial mass of the isothiocyanate.

The color of FITC-PVA recovered ranged from yellow to orange depending on the amount of FITC added to the reaction (Figure 1C). The coloring contrasts with unmodified PVA, which is white (data not shown).

2.3. Fourier Transform Infrared Spectroscopy (FTIR)

FTIR was carried out on a PerkinElmer Spectrum 100 machine. 5 mg of sample was spread on the sample holder and data was recorded from 650–4000 cm⁻¹. Transmittance was plotted against wavenumber (cm⁻¹). Presence of functional groups was analyzed from the data.

Proton Nuclear Magnetic Resonance (H-NMR)

H-NMR was carried out using a Bruker Avance IIIHD 400 MHz spectrometer. 10 mg of sample was dissolved in 0.7 mL of DMSO-d6. 32 scans were taken with a 20 ppm window centered at 6.175 ppm. 4 second acquisition time with 1 second recovery delay and 30-degree pulse widths were used.

The percentage of OH groups on PVA conjugated to an isothiocyanate was determined from the NMR spectra. Peaks characteristic of phenyl protons associated with the isothiocyanate were identified and integrated. This signal was compared to the signal integrated from peaks characteristic of methylene protons in the backbone of PVA. The ratio of signals was used to calculate the percentage of OH groups conjugated to an isothiocyanate. Peaks were considered detectable if they were 3 times larger than the baseline noise. Quantification was considered reliable down to approximately 0.01% (with regards to the proportion of OH conjugated to an isothiocyanate).

2.5. Poly(Lactide-co-Glycolide) (PLG) Particle Fabrication

PLG particles were formed using a single oil-in-water emulsification followed by solvent extraction method, reported previously by our group^23–25. Briefly, 50:50 PLG was dissolved in dichloromethane at 4% (wt/wt). 0.6 mL of the 4% PLG solution was added dropwise to 4 mL of an aqueous solution of 0.2% (wt/v) poly(vinyl alcohol) (PVA) or modified PVA and emulsified at 11,000 rpm for 5 minutes using a Kinematica PT3100D homogenizer. Particle formation occurred by adding the emulsion to 16 mL of ultrapure water and then stirring the mixture for 1 hour. The particles were then passed through a 40 μm filter and collected via centrifugation at 250×g. The particles were then washed 4 times by suspending them in ultrapure water and then collecting them via centrifugation (250×g). Washed particles were frozen at −20°C and lyophilized overnight. Recovered particles were stored under vacuum in a dry environment at room temperature. For coumarin 6 loaded particles, all conditions were the same except coumarin 6 was added to the organic phase at 0.3 mg/mL.

Particle mass yield was calculated by dividing the mass of recovered particles by the mass of PLG emulsified. Particle mass yield was calculated according to equation 2.

$$\mathit{\text{Mass}} \mathit{\text{Yield}} \left(\text{\%}\right)=\left(\frac{M_{PT}}{M_{PLG}}\right)\text{*}100$$ (2)

Where M_PT is the mass of particles recovered from the emulsion, M_PLG is the mass of PLG added to the emulsion.

2.7. Particle Size Measurements

Particles were reconstituted in ultrapure water, added to a 48-well plate, and allowed to settle prior to image acquisition. Particles were then imaged on an EVOS FL microscope at 20x magnification. Three representative images of each particle formulation were acquired. The Image J “Particle Analysis” function was used to measure particle size. At least 1000 particles were sized for each formulation. We report the mean particle size and the coefficient of variation (CV%). We validated this method previously with monodispersed polystyrene beads purchased from Duke Standards²³.

2.8. Confocal Microscopy

Particles were suspended in DI water at 10 mg/mL and 10 μL of this solution was added to a glass slide. Confocal images were taken on a Zeiss 700 confocal microscope using the 488 nm laser at 63x magnification.

2.9. Concanavalin A (ConA) Binding Assay

Particles (0.5 mg) were suspended in 0.5 mL of annexin v binding buffer. 10 uL of ConA (5 mg/mL) was added to the particle suspension, mixed well, and incubated at room temperature for 10 minutes. To remove unbound ConA, the particles were washed twice with binding buffer. Half of the samples were then viewed by fluorescent imaging using the EVOS FL microscope. The other half of samples were dissolved in DMSO and analyzed for fluorescence intensity using a Biotek Synergy H1 plate reader.

2.10. RAW 264.7 Macrophage Culture

RAW 264.7 macrophages (ATCC TIB-71) were subcultured via the ATCC protocol. Briefly, cells were seeded at 1 million cells in T-25 flasks in 3 mL of complete media (DMEM supplemented with 1mM sodium pyruvate, 1500 mg/mL sodium bicarbonate, 1% penicillin-streptomycin, and 10% fetal bovine serum). Cells grew for 2 days or until cells became 60–70% confluent. Cells were then rinsed with PBS and 1 mL of Trypsin with 0.25% EDTA was added for 5 minutes at 37°C. Trypsin was then diluted into 2 mL of complete media, cells were scraped with a rubber policeman, and transferred to a 15 mL centrifuge tube. An aliquot of this solution is diluted with Trypan blue and used to count the cells with a hemacytometer. The cell solution was then centrifuged at 400×g for 5 minutes. The cell pellet was reconstituted in complete media and seeded into flasks or 24 well plates as needed. Cells were seeded at 25,000/cm².

2.11. Bone Marrow Derived Macrophage Culture

Femurs and tibiae were isolated from female ICR mice. The bone marrow was extracted and passed through a 70 μm filter and then red blood cells were lysed. The bone marrow cells were then washed, counted, and seeded in non-tissue culture treated petri dishes at 500,000 cells/cm². Cells were cultured in DMEM/F12 supplemented with 10% FBS, 1% Pen/Strep, and 10 ng/mL macrophage colony stimulating factor. After 5 days, non-adherent cells were removed with gentle washing and discarded. Adherent cells were harvested using Trypsin with 0.25% EDTA for 5 minutes at 37℃. The adherent cells were then seeded into 48 well plates at 50,000/cm² for particle binding assays. Purity of the adherent cells was monitored with flow cytometry, which indicated 98% of the adherent cells express F4/80 and CD11b. Bone harvest was performed in accordance with NIH Guidelines for Care and Use of Animals and approved by the Institutional Animal Care and Use Committee at the University of South Carolina.

2.12. Assessing Particle Association with Macrophages Using Fluorescent Spectroscopy

RAW 264.7 macrophages were plated in 24 well plates and BMDMs were plated in 48 well plates and cultured for 24 hours prior to treatment. Then cells were treated with coumarin 6-containing particles (7.5 μg in 500 μL of DMEM) produced with unmodified PVA, 0.71% MITC-PVA or 5.4% MITC-PVA for 24 hours at a concentration of 15 μg/mL. In some wells, BMDMs were pre-treated with 4mg/mL of mannan, derived from Saccharomyces cerevisiae, for 15 minutes to antagonize the mannose receptor adding particles. Following particle treatment, unbound particles were then washed away with three washes in PBS. To visualize particle binding in situ, cells were imaged with an EVOS FL microscope. To quantify particle binding, washed cells were incubated with DMSO for 10 minutes to solubilize the particles and extract the coumarin 6. The DMSO solution was then collected and added to wells of a 96 well plate. Fluorescence of the DMSO solution was measured with a Biotek plate reader. A serial dilution of fluorescent particles dissolved in DMSO was used to construct a standard curve, which was used to interpolate mass of particles bound to the cells. Interpolation from the standard curve was necessary because the amount of coumarin 6 in each particle formulation varied slightly.

2.13. Assessing Particle Association with Macrophages Using Flow Cytometry

BMDMs were plated in 6 well plates at a density of 50,000 cells/cm² for 48 hours prior to particle treatment. To antagonize the mannose receptor, certain wells were treated with 4mg/mL of mannan from Saccharomyces cerevisiae for 15 minutes, prior to particle treatment. Cells were then treated with coumarin 6-labeled particles produced with either unmodified PVA, or 5.4% MITC-PVA. Cells were incubated with particles for 24 hours or 3 hours at a concentration of 15 μg/mL in 2 mL of complete media (DMEM/F12 supplemented with 10% FBS, and 1% penicillin-streptomycin) in each well of culture plate. Cells were then washed three times with PBS to remove non-associated particles. Cells were then collected by adding 348 μL of Trypsin with 0.25% EDTA for 5 minutes at 37°C. Then, trypsin was diluted with 616 μL of complete media and cells were scraped with a rubber policeman and transferred to a 5 mL conical tube. Cells were spun down at 300 ×g and 4°C for 5 minutes and then supernatant was decanted. The cells were suspended in 0.5 μg of TruStain fcX (anti-CD16/32) diluted in 50 μL of MACS buffer (PBS, 0.5 mM EDTA, 30% BSA) for 15 minutes on ice prior to the addition of 0.125 μg of APC anti-mouse F4/80 antibody in 100 μL of MACS buffer for 30 minutes on ice. Cells were washed with MACS buffer to remove unbound antibody and then fixed with 0.5 mL of fixation buffer (4% paraformaldehyde in 1X PBS solution), followed by a second wash and filtration through a flow tube filter top. A FACS Aria flow cytometer (BD Biosciences) was used for data acquisition. Fluorescence minus one (FMO) controls were used to set gates for F4/80+ cells and cells associated with C6-labled particles. Flow cytometry data was analyzed using BD FACSDiva software.

2.14. Statistics

Statistical analyses were carried out using Graphpad Prism software. Where appropriate, an unpaired t-test or a One-Way ANOVA followed by a Tukey’s multiple comparisons test was carried out to assess the differences between means. Error bars represent the standard deviation (SD). Specific details about statistical analysis are given in the figure legends.

3. Results

3.1. Modifying PVA with isothiocyanates

We first determined if FITC could be conjugated to PVA as depicted in Figure 2A. FITC (10%) was added to a solution of PVA, stirred for 5 hours at 25°C and then dialyzed against water as depicted in Figure 1. FTIR characterization of the product (termed FITC-PVA) indicated the presence of an aromatic C=C at 1600 cm⁻¹, which was also present in FITC, but not PVA (Figure 2B). In addition, the PVA-FITC spectra indicated the absence of the isothiocyanate (NCS) group present in FITC at 2100 cm⁻¹. Importantly, both of these features are absent from the PVA spectra. Taken together, the data indicates that FITC was conjugated to PVA, possibly through the formation of a thiourethane bond.

Figure 2. Characterization of FITC-PVA.

A) Chemical structures of FITC, PVA and FITC-PVA, B) FTIR spectra of FITC-PVA (green), PVA (red), and FITC (blue). Absorption peaks at 2100 cm⁻¹ (NCS) and aromatic C=C at 1600 cm⁻¹ are delineated with dotted lines. C) NMR spectra of FITC-PVA, PVA and FITC. Resonances associated with protons in FITC and PVA are marked with “a” and “b”, respectively. Additionally, the protons are labeled on the chemical structures in A.

NMR characterization of the FITC-PVA revealed resonances in the 6–10 ppm region, which were also present in FITC, but not PVA (Figure 2C). These peaks were shifted or broadened in the FITC-PVA compared to FITC, suggesting that FITC was conjugated to the PVA and that unreacted FITC was removed during dialysis. The peaks between 6–7 ppm indicate the 6 protons on the phenyl rings of FITC marked with “a”, while the peaks between 1–2 ppm indicate the two protons present in the carbon backbone of PVA marked with “b” (Figure 2C and A). FITC-PVA contained both of these features, and integration of signal in these two regions indicated that 4.8% of the OH groups in the FITC-PVA were conjugated to FITC and, thus, 48% of the FITC reacted.

We next determined if MITC could be conjugated to PVA. MITC (20%) was added to a solution of PVA, stirred for 24 hours at 100°C, and then dialyzed against water as depicted in Figure 1. NMR of the MITC-PVA revealed aromatic resonances in the 7–8 ppm region, which were also present in MITC, but not PVA (Figure 3A). These peaks indicate the four protons in the phenyl linker of MITC marked with “a” (Figure 3B). Integration of the signal due to phenyl linker protons and the signal corresponding to the two protons on the carbon backbone of PVA (1–2 ppm, marked with “b”) indicated 5.4% of the OH groups in the MITC-PVA were conjugated to MITC and, thus, 27% of the MITC reacted.

Figure 3. Characterization of MITC-PVA.

A) NMR spectra of MITC-PVA (green), PVA (red) and MITC (blue). Peaks associated with characteristic protons of MITC and PVA are labeled with “a” and “b” respectively. B) Chemical structures of MITC-PVA, PVA and MITC. Protons associated with the peaks in the NMR spectra are labeled with “a” and “b”.

In all, we conducted 13 reactions between PVA and either FITC or MITC to understand the effect of reaction conditions (Table 1). Reactions between FITC and PVA carried out at 25°C for 5 hours led to relatively high amounts of FITC reacted, which increased with the amount of FITC in the reaction (Trials 1–3). In contrast, the amounts of MITC that reacted under the same time and temperature conditions were lower and could not be improved by increasing the amount of MITC in the reaction (Trials 4–6). Increasing the reaction time to 24 hours did not improve the amount of MITC reacted (Trial 7). However, by maintaining the amount of MITC in the reaction at 5% and the reaction time at 24 hours, we found that heating improved the amount of MITC that reacted (Trials 8–12). Finally, in trial 13, we found that increasing the amount of MITC in the reaction to 20% increased the amount of MITC that reacted. In general, for these 13 reactions, we found that the reaction occurs more readily between PVA and FITC than with MITC and that extent of reaction is dependent on the isothiocyanate concentration.

Table 1.

Reactions of isothiocyanates with PVA

Trial	ITC	Amt of ITC in Rxn (%)	Rxn Time (hr)	Temp (°C)	Mass Yield (%)	OH conjugated (NMR) (%)	ITC reacted (calculated) (%)
1	FITC	0.1	5	25	37	≤ 0.01*	---
2	FITC	1	5	25	28	0.04	4
3	FITC	10	5	25	25	4.8	48
4	MITC	1	5	25	51	≤ 0.01*	---
5	MITC	10	5	25	55	0.12	1.2
6	MITC	20	5	25	19	0.12	0.6
7	MITC	5	24	25	46	ND**	---
8	MITC	5	24	60	54	0.13	2.6
9	MITC	5	24	80	22	0.57	11
10	MITC	5	24	80	28	0.60	12
11	MITC	5	24	100	34	0.93	19
12	MITC	5	24	100	28	0.71	14
13	MITC	20	24	100	16	5.4	27

^*Weak signal. Detectable but not quantifiable.

^**Not detected

3.2. Fabrication of PLG particles with modified PVA

We next investigated if modifying the PVA with isothiocyanates impacted its performance as an emulsifier when making PLG particles. We utilized a single emulsion, solvent extraction protocol that yielded 3.9 μm particles with a CV of 34% and a mass yield of 52% when unmodified PVA was used as the emulsifier (Table 2). Modifying the PVA with either FITC or MITC at low (0.04%−0.71%) or high (4.8%−5.4%) levels did not significantly impact mass yield of particles, indicating that its emulsifying properties remained intact. In general, particle size was not impacted by modifying PVA, but particles made with the highly modified FITC- and MITC-PVA did exhibit a trend of larger size compared to particles made with unmodified PVA. In addition, the 5.4% MITC-PVA particles exhibited a higher CV than the other formulations.

Table 2.

Particle fabrication with isothiocyanate modified PVA

PVA	Mass yield (%)	Size (μm)	CV (%)
unmodified	52 ± 3	3.9 ± 0.3	34 ± 3
0.04% FITC	50 ± 4	3.7 ± 0.1	36 ± 1
4.8% FITC	45 ± 6	5.0 ± 0.2	30 ± 1
0.71% MITC	55 ± 5	2.8 ± 0.1	36 ± 1
5.4% MITC	54 ± 7	4.7 ± 0.9	53 ± 2

3.3. Location of FITC on PLG particles produced with FITC-PVA

We next investigated the location of the FITC molecules within the PLG particles produced with FITC-PVA. Phase contrast microscopy indicated that particles made with unmodified PVA and FITC-PVA were spherical in shape (Figure 4A). Confocal microscopy revealed a thin corona of signal delineating the outside of the particle (Figure 4B), indicating FITC-PVA was confined to the particle surface and not encapsulated within the bulk of the particle. In addition, particles made with 4.8% FITC-PVA exhibited more signal than those made with 0.04% FITC-PVA. As expected, particles made with unmodified PVA exhibited no fluorescence.

Figure 4. Imaging of PLG particles made with FITC-PVA.

A) Phase contrast images and B) confocal images of particles made with unmodified PVA, 0.04% FITC-PVA, or 4.8% FITC-PVA.

3.4. Bioactivity of mannose on PLG particles produced with MITC-PVA

We next investigated if the MITC remained bioactive when MITC-PVA was used to produce PLG particles. To accomplish this, particles were incubated with rhodamine labeled Concanavalin A (ConA), a lectin that binds mannose. The particles were then washed to remove unbound ConA and imaged using epifluorescence microcopy (Figure 5A–C). Qualitatively, 5.4% MITC-PVA particles exhibited more signal than 0.71% MITC-PVA particles. As expected, particles made with unmodified PVA did not exhibit signal, indicating they did not bind ConA. To quantify the extent of ConA binding, particles were dissolved with DMSO and analyzed using a fluorescence plate reader. The fluorescence intensity of the 5.4% MITC-PVA particles was significantly higher than the particles made with unmodified PVA (Figure 5D).

Figure 5. ConA binding to MITC-PVA particles.

Fluorescence images of rhodamine-ConA binding to particles made with A) unmodified PVA, B) 0.71% MITC-PVA, or C) 5.4% MITC-PVA. D) Fluorescence intensity of solutions of particles solubilized in DMSO and analyzed on a plate reader. Data analyzed by One-Way ANOVA with Tukey’s multiple comparisons test. * indicates p<0.05. Data combined from three independent experiments.

3.5. Particle association with macrophages measured with fluorescence spectroscopy.

Macrophages express several lectins that bind mannose; thus, we investigated if particles produced with MITC-PVA associated more readily with macrophages than particles produced with unmodified PVA. To accomplish this, we treated RAW 264.7 macrophages with fluorescent particles produced with unmodified PVA, 0.71% MITC-PVA, or 5.4% MITC-PVA for 24 hours. Cells were then washed to remove unbound particles and imaged using phase contrast and epifluorescence microscopy (Figure 6). Qualitatively, more particles were apparent with the 5.4% MITC-PVA particles compared to the 0.71% MITC-PVA particles or the unmodified PVA particles.

Figure 6. Binding of MITC-PVA particles to RAW 264.7 macrophages.

Phase contrast, epifluorescence, and merged images of RAW 264.7 macrophages treated with fluorescent particles produced with unmodified, 0.71% MITC-PVA, and 5.4% MITC-PVA for 24 hours followed by removal of unbound particles with a washing. Data is representative of three independent experiments. Scale bar is 30 μm.

To confirm the increased binding of MITC-PVA particles observed in Figure 6, RAW 264.7 macrophages treated with particles were incubated with DMSO to dissolve the particles and extract the fluorescent dye. The cells treated with particles exhibited no fluorescence after incubation with DMSO (Figure 7A versus B). The DMSO solution was then analyzed on a plate reader and mass of particles bound was interpolated from a standard curve. The mass of particles was significantly higher for the 5.4% MITC-PVA particles compared to the 0.71% MITC-PVA and the unmodified PVA particles (Figure 7C). Finally, to confirm that the enhanced particle interactions held for primary cells, we repeated the experiment with bone marrow derived macrophages (BMDM) from mice. As expected, more binding was observed with 5.4% MITC-PVA particles compared to unmodified PVA particles (Figure 7D).

Figure 7. Quantification of particle binding to RAW 264.7 macrophages and bone marrow derived macrophages (BMDMs).

Images of RAW 264.7 macrophages with bound fluorescent particles A) before and B) after incubation with DMSO. Mass of particles bound to C) RAW 264.7 macrophages and D) BMDMs. E) Fold change in fluorescent intensity of fluorescent particle binding to BMDMs upon co-treatment with mannan. For RAW 264.7 macrophages, data analyzed by One-Way ANOVA with Tukey’s multiple comparisons test. * indicates p<0.05, ** indicates p<0.01, and *** indicates p<0.001. For BMDM, data analyzed by unpaired t-test. ** indicates p<0.01. Data combined from three independent experiments for each cell type.

We then investigated if 5.4% MITC-PVA particle binding to BMDMs could be decreased by co-treatment with mannan, a high affinity ligand for the mannose receptor. There was a trend for mannan to decrease the association of 5.4% MITC-PVA particles with the cells (Figure 7E). As expected, the mannan did not have an impact on the binding of unmodified particles.

3.6. Particle association with BMDMs measured with flow cytometry.

To quantitatively assess the impact of MITC-PVA on particle association with BMDMs, we conducted flow cytometry on BMDMs treated with 5.4% MITC-PVA or unmodified PVA particles for 3 or 24 hours. Representative bivariate blots of APC (F4/80) versus FITC (Coumarin 6 particles) for untreated macrophages (Figure 8A) and macrophages treated with coumarin 6 containing particles (Figure 8B) are shown. The macrophages positive for particle association (% positive cells) are viewed in quadrant 2 of Figure 8B. Interestingly, the % positive cells were not different between the macrophages treated with 5.4% MITC-PVA or unmodified PVA particles at the 3 hour time point (Figure 8C). In addition, the median fluorescence intensity of macrophages that were positive for particle association at 3 hours was not different between the two particle types (Figure 8E), indicating MITC-PVA did not lead to more particle associations per macrophage. We also investigated these indices at 24 hours (Figure 8D and F), which is the same timepoint studied in the plate reader assay shown in Figure 7. Interestingly, there was a slight, but significant decrease in the % positive cells for MITC-PVA versus unmodified PVA (Figure 8B) but there was no difference in median fluorescence intensity for the two groups (Figure 8F).

Figure 8. Flow cytometry of macrophages treated with particles.

Representative bivariate plots for A) untreated macrophages and B) macrophages treated with coumarin 6 particles. Macrophages associated with particles are found in quadrant 2 (Q2). % positive cells are shown for C) 3 hours and D) 24 hours of particle treatment. The 3 hour experiment includes cells that were co-treated with mannan and particles. Median fluorescence intensity of the cells associated with particles are shown for E) 3 hours and F) 24 hours. 3 hour time point data analyzed with One-Way ANOVA followed by a Tukey’s multiple comparisons test. * indicates p<0.05. ** indicates p<0.01. The 24 hour data analyzed by unpaired t-test. * indicates p<0.05. Data represents 3 replicates for each condition.

Finally, we determined if MITC-PVA particle association with macrophages could be decreased by co-treatment with mannan, a high affinity ligand for the mannose receptor. Interestingly, mannan decreased the % positive cells for unmodified particles by a small, but significant amount, while 5.4% MITC-PVA particles were not significantly decreased (Figure 8C). Median fluorescence intensity was significantly decreased by a small amount for both particle types with mannan inhibition (Figure 8E).

4. Discussion

Our goal was to alter the surface functionality of polymer particles made by the O/W emulsion/solvent extraction method, with PVA as the emulsifier, by chemically modifying PVA prior to particle fabrication. We demonstrate that PVA can be modified with isothiocyanates and that the modified PVA acts as an effective emulsifier for producing PLG particles. We achieved about 5% OH conjugation for both FITC and MITC but found that the reaction occurs more readily with FITC than with MITC. Modification of PVA with FITC and MITC does not impact PLG particle morphology or mass yield, indicating modified PVA maintains its properties as an emulsifier in our system. The most common methods for modifying particle surface chemistry involve attaching surface ligands after particle fabrication²⁶. To our knowledge, we are the first to show that PLG particle surface chemistry can be controlled by modifying PVA prior to particle formation. Finally, we show that MITC modified-PVA increases particle binding to ConA (a mannose binding lectin) and macrophages (a cell that expresses mannose binding lectins), demonstrating that the pendant group added to the PVA remains functional and can endow the particle with new biological activity. Taken together, we present a facile method for biomolecule presentation by biodegradable/biocompatible particles using a wide-spread fabrication technique, in situations where the biomolecule contains (or can be modified with) an isothiocyanate group.

In this study we focused on a single PVA that was 87 to 89% hydrolyzed with an average molecular weight between 13,000 and 23,000. This polymer is commonly used as the emulsifier for the production of PLG particles for biomedical applications^7,27–29 and, thus, we employed it here. However, it is important to consider how hydrolyzation and molecular weight might impact particle characteristics and PVA reactions with isothiocyanates. To begin, PVA hydrolyzation is a key parameter in the production of PLG particles. This is because extent of hydrolyzation impacts the emulsifier properties of PVA with poor emulsifier properties leading to poor particle formation. Partially hydrolyzed PVAs between 76 and 89%, such as the one we used, are superior emulsifiers compared to PVAs with higher % hydrolyzation³⁰. Consistent with this, mass yield of PLG particles decreases by 60% when 98.5% hydrolyzed PVA is used in place of 88% hydrolyzed PVA³¹. PLG particles can also be produced with PVAs that are hydrolyzed as low as 30%³², however that study did not address mass yield. Taken together, it is clear that PVA that is 87 to 89% hydrolyzed is effective in producing high yields of PLG particles. Since our goal was to produce PLG particles, and high yield was desirable, we did not investigate reactions between isothiocyanates and fully hydrolyzed PVA or those with low extents of hydrolyzation. However, since the reaction occurs between the isothiocyanate and the alcohol group, we expect that the reactions described in this article could be carried out with a range of partially hydrolyzed PVA, including fully hydrolyzed PVA.

PVA molecular weight is also an important consideration in the production of PLG particles. The two most common molecular weight ranges for this purpose are 13,000 to 23,000^7,27–29 (the one we used) and 30,000 to 70,000^28,33,34. At a given concentration in the continuous phase, higher molecular weight PVA leads to the formation of smaller particles than lower molecular weight PVA³⁵; however, this effect can also be achieved by increasing the concentration of a low molecular weight PVA in the continuous phase^7,27,33. It is proposed that the PVA molecular weight effect on particle size is due to increasing the viscosity of the continuous phase, which inhibits movement of oil droplets and coalescence prior to particle formation³³. Similarly, increasing the concentration of a low molecular weight PVA in the continuous phase increases viscosity, but the increased concentration may also promote more efficient transport of PVA to the oil-water interface⁷, which would inhibit coalescence of oil droplets and favor smaller particles. With regards to the amount of PVA that coats the surface of the particle, it has been reported that increasing the amount of PVA in the continuous phase increases the residual amount of PVA on the surface of the particles^7,27,33; but these studies did not address molecular weight. The thickness of a monolayer coating of PVA has been estimated on the scale of 10 nm³⁶; however, this could be greater if there are multiple layers of PVA, which might be supported by strong intermolecular interactions between chains supported by hydrogen bonding³¹. While we did not investigate the effect of PVA molecular weight on reactions with isothiocyanates, we postulate that the reaction might be inhibited with higher molecular weight PVA owing to stronger inter and intramolecular interactions of the longer polymer chains.

We modified PVA by conjugating an isothiocyanate to alcohol groups on the polymer. PVA has been modified previously in a variety of ways. PVA has been tosylated³⁷, reacted with acrylamides³⁸, and phosphorylated³⁹. In addition, PVA has been modified by esterification, carbamation, and etherification⁴⁰. Many of these methods are used for modifying PVA for applications in paper making, textiles, and adhesives. However, PVA has also been modified for biomedical purposes^41–43. For example, PVA has been succinylated, and then reacted with glycidyl methacrylate and carbamate for producing hydrogels⁴⁰. In addition, one group used isocyanate to modify PVA for making hydrogels⁴⁴. With regards to the broad applicability of conjugating isothiocyanate-containing molecules to PVA, it is important to consider that methods for attaching the isothiocyanate functional group to biomolecules are established²⁰. Thus, our approach is expected to enable the modification of PVA with a wide range of molecules and is not limited to MITC and FITC.

We produced PLG particles with modified PVA and demonstrated this altered particle surface functionality. While there have been successful efforts to functionalize the surface of PLG particles (and other polymer particles) using the O/W emulsion/solvent extraction method, functionalization involves adding ligands to the microparticles after fabrication⁴⁵. The most common method is to use carbodiimide chemistry to attach a ligand with an amide bond to the carboxylic acid group on PLG using NHS/EDC chemistry^15,17,46. Although this is a widely used reaction, the downside to using it after particle fabrication is that it can be difficult to control the reaction yield and to characterize the polymer particle due to the relatively small proportion of surface ligand relative to the bulk polymer¹². In addition, the extra purification steps may lead to loss of particle and drug. The methodology presented here of attaching chemical modifiers to the PVA prior to particle formation obviate the need for a second surface-functionalization step and its associated challenges.

We used two methods to quantitatively investigate the impact of MITC-PVA on PLG particle association with macrophages. One involved analyzing the fluorescent extract of particles associated with macrophages using fluorescence spectroscopy. The other was flow cytometry. Interestingly, the results of the two methods did not agree. Fluorescence spectroscopy indicated that MITC-PVA enhanced particle association with macrophages by 2- to 3-fold (dependent on cell type) compared to unmodified particles. In contrast, flow cytometry indicated that MITC-PVA particles and unmodified particles associate with macrophages to a similar degree. A possible explanation for this discrepancy is that in the current study, the fluorescence spectroscopy assay is measuring both particle uptake and particles bound to the surface of macrophages, while flow cytometry is measuring only particle uptake. We propose this possibility because of the different sample preparations involved in each method. While fluorescence spectroscopy involves solubilizing the fluorescent particles as they are associated with macrophages adherent to a tissue culture plate, flow cytometry entails treating the cells with trypsin, which cleaves extracellular proteins, including those that might bind particles, and scraping the macrophages off the plate. In addition, there are several wash steps as macrophages are labeled with fluorescent antibodies. The enzymatic and mechanical manipulations required of macrophages preceding flow cytometry may remove surface bound particles, leaving only the internalized particles for detection. Based on the data, it is possible that MITC-PVA enhances the binding of the PLG particles used in the study, but not uptake. Congruent with our findings, two groups have shown an increase in macrophage association with mannose functionalized particles using the fluorescence spectroscopy method we used^47,48; however, another study found that mannose functionalization increases particle association with macrophages using flow cytometry⁴⁹. A key difference between the aforementioned three studies and our work is particle size. Our particles are 4 to 5 microns, while the particles from the other studies were submicron. The size dependance on macrophage particle uptake has been studied⁵⁰ and, on average, a macrophage only takes up 1 particle when the size is above 3 microns. Thus, internalization may not be a major of mode of association between macrophages and the particles used in this study. However, a limitation of the flow cytometry study is external fluorescence was not quenched in any of the samples prior to analysis and, thus, we are not able to say with certainty if the signal detected is due to particles bound, internalized, or both.

It is important to address that there was a small, but significant decrease in particle association detected by flow cytometry for the MITC-PVA particles compared to the unmodified PVA particles at 24 hours. The average sizes of the unmodified and MITC-PVA particles were 3.9 microns and 4.7 micron, respectively. The particles are spherical and, thus, the volume of a MITC-PVA is approximately 1.75 times that of an unmodified PVA particle (calculated by ratio of the radii cubed). This size differential may have played a role in the small, but significant decrease in uptake at 24 hours between the two particles formulations.

Finally, we must address the modest decrease in particle binding when macrophages were co-treated with particles and mannan in both the fluorescence spectroscopy assay and flow cytometry. Mannan has been shown to inhibit uptake of mannose decorated particles⁵¹, which is why we selected it as an inhibitor for our study. However, receptors that bind mannan on macrophages, may not be involved in macrophage interactions with the 4–5 micron particles utilized in this study, which are larger than the particles in the aforementioned study⁵¹. In addition to mannose receptor, Fc receptors, complement receptors and macrophages scavenger receptors have been implicated in macrophage uptake of particles^52,53, and it’s possible that macrophage interactions with the particles in this study occurs via one or more of these receptors.

The goal of this study was to further the functionality of PLG particles by investigating the conditions under which PVA can be modified with isothiocyanates and determine if the modified PVA impacts the surface reactivity of PLG particles. This goal was achieved by systematically investigating the impact of reaction conditions on the extent of PVA modification (Table 1) and demonstrating that the surface of FITC-PVA particles fluoresce (Figure 4) while the surface of MITC-PVA particles bind concanavalin A (Figure 5). In addition, we provide strong evidence that MITC-PVA increases particle association with macrophages using a fluorescence plate reader assay (Figure 7). However, with an eye towards cell targeting, this study is not without limitations. First, we were unable to detect an increase in particle association with flow cytometry and second, we did not determine the receptors involved in particle binding. With that being said, it is easily conceived how this work will enable future studies, by our lab and others, to develop particle formulations with further targeting capabilities. While outside the scope of the current investigation, future work should address the impact of particle size on targeting and investigate glycosides other than MITC, including branched oligosaccharides containing MITC, which may exhibit tighter binding to the cell surface⁵⁸.

Macrophages play key roles in human health and disease and an ability to increase association of drug loaded particles with these cells could improve many medical treatments⁵⁴. For example, tumor associated macrophages promote tumor growth, and delivering drugs to tumor regions in which these cells reside is a widely investigated strategy in cancer immunotherapy⁵⁵. Lupron depot is a sustained release PLG particle formulation FDA approved for prostate cancer^56,57. It is an intramuscular injection that delivers leuprolide acetate systemically. Delivery of particles with enhanced capacity to associate with macrophages within the tumor microenvironment might improve efficacy and reduce side effects. As Lupron depot particles are made using the O/W emulsion/solvent extraction fabrication method with PVA as the emulsifier⁵⁷, using the techniques described herein to associate particles with macrophages within prostate tumors is an interesting repurposing/modifying of an FDA approved extended release drug formulation.

4. Conclusion

Fluorescein isothiocyanate (FITC) and α-D-mannopyranosyl 4-phenylisothiocyanate (MITC) can be conjugated to poly(vinyl alcohol) (PVA) using a one-step reaction. Under the conditions tested, the reaction occurs more readily with FITC than with MITC and that the extent of reaction is dependent on the initial isothiocyanate concentration. It is possible to achieve approximately 5% conversion of the hydroxyl groups on PVA with both FITC and MITC. Modification of the PVA does not significantly impact mass yield or size of poly(lactide-co-glycolide) particles produced, indicating the emulsifier properties are not significantly impacted for the system studied. FITC-PVA is localized to the surface of particles and MITC-PVA endows the particle with an ability to bind concanavalin A. MITC-PVA also enhances particle association with a macrophage cell line (RAW 264.7) and primary macrophages cultured from mouse bone marrow when studied using fluorescence spectroscopy. However, these differences are not detected with flow cytometry. We propose that this discrepancy is due to a loss in particle association with the cells during the cell processing for flow cytometry. While outside the scope of this study, future work should determine the impact of size on particle binding to macrophages and investigate functionalization of PVA with branched oligosaccharides, which might enable tighter binding to the cell surface. Taken together, this work demonstrates a facile method for modifying the surface reactivity of polymer particles that can be produced using the oil-in-water emulsion/solvent evaporation method, in which PVA is employed as the emulsifier and the chemical modifier contains an isothiocyanate functional group. Given that methods are established for attaching the isothiocyanate functional group to a wide range of biomolecules, our approach may enable PVA-based materials to engage a multitude of biological systems.