Microfluidics-generated immunomodulatory nanoparticles and formulation-dependent effects on lipopolysaccharide-induced macrophage inflammation

Abstract

Nanoparticles (NPs) have emerged as a highly useful and clinically translatable drug delivery platform for vast therapeutic payloads. Through the precise tuning of their physicochemical properties, NPs can be engineered to exhibit controlled drug release properties, enhanced circulation times, improved cellular uptake and targeting, and reduced toxicity profiles. Conventional bulk methods for the production of polymeric NPs suffer from the ability to control their size and polydispersity, batch-to-batch variability, significant preparation times, and low recovery. Here, we describe the development and optimization of a high-throughput microfluidic method to produce cargo-less immunomodulatory nanoparticles (iNPs) and their formulation-dependent anti-inflammatory properties for the modulation of lipopolysaccharide (LPS)-induced macrophage responses. Using poly(lactic acid) (PLA) as the core-forming polymer, a rapid and tunable microfluidic hydrodynamic flow focusing method was developed and optimized to systematically evaluate the role of polymer and surfactant concentration, surfactant chemistry, and flow rate ratio (FRR) on the formation of iNPs. A set of iNPs with 6 different surface chemistries and 2 FRRs was then prepared to evaluate their inherent anti-inflammatory effects using bone marrow-derived macrophages stimulated with the Toll-like receptor (TLR) 4 agonist, LPS. Finally, a lyophilization study was performed using various cryoprotectants and combinations to identify preferable conditions for iNP storage. Overall, we demonstrate a highly controlled and reproducible method for the formulation of iNPs using microfluidics and their formulation-dependent inherent anti-inflammatory immunomodulatory properties, which represents a potentially promising strategy for the management of inflammation.

Introduction

Nanoparticles (NPs) have attracted significant attention as a versatile drug delivery system [1]. In recent years, the U.S. Food and Drug Administration (FDA) has approved or established in clinical trials, over 50 different NPs ranging from polymeric, liposomal, metallic, and other complex systems for use in cancer, allergy, radiotherapy, anemia, and other applications [2]. In particular, NPs prepared from biodegradable and biocompatible aliphatic polyesters such as poly(lactic-co-glycolic acid) (PLGA) or poly(lactic acid) (PLA) hold significant promise due to their tunable physicochemical characteristics that enable the delivery of a diverse set of therapeutic payloads (i.e. small molecules, peptides/proteins, nucleic acids, etc.) for controlled release applications, enhanced circulation times, improved cellular uptake and targeting, and reduced toxicity profiles [3, 4].

A critical feature that affects the translational potential of NPs is the ability to reproducibly and efficiently control NP physicochemical properties to suit a desired biomedical application. Two common methods for polymeric NP generation are emulsification and nanoprecipitation [5]. The emulsification method involves the input of high shear stress into a mixture of a non-water miscible volatile organic solvent containing the polymer and an aqueous solution. Subsequently, the organic solvent is evaporated resulting in the formation of NPs. Nanoprecipitation utilizes a mixture of a water-miscible organic solvent containing the polymer and an aqueous solution. The formation of NPs using the nanoprecipitation method occurs rapidly and in a single step. However, these conventional bulk NP formulation methods suffer from several limitations including difficult scale-up, batch-to-batch variability, wide size distributions and high polydispersity, complex and long purification procedures, and low yields. Therefore, improved NP formulation methods are critically needed to enable the full potential of NPs to treat human diseases.

There have been emerging advances in the controlled drug delivery field by fabricating NPs using microfluidic devices [6–8]. Microfluidic technology is advantageous compared to bulk NP preparation methods, especially regarding controllability and reproducibility [7, 8]. The unique shape and geometry of microfluidic devices allow for the reduction in reagent use, control of fluid mixing, and reduced reaction time [9–11]. Specifically, microfluidic devices offer the potential to produce NPs with precise physicochemical properties in a high throughput manner with narrow size distributions at high recoveries [12–15]. Microfluidic methods allow for easy expansion of the number of NP formulations tested and the generation of small quantities of NPs for initial screening experiments yet are scalable to gram quantities without the need to modify formulation parameters.

Recently, our group reported cargo-less immunomodulatory NPs (iNPs) prepared from poly(lactic acid) (PLA) and their inherent anti-inflammatory properties [16]. Without incorporation of small molecules or biologic immunomodulators, we showed the iNP surface chemistry-, and polymer composition-dependent dampening of proinflammatory innate immune responses to Toll-like receptor (TLR) agonists. iNPs demonstrated broad inhibitory potential against pro-inflammatory cytokine secretions from macrophages induced by TLR agonists such as lipopolysaccharide (LPS; TLR4 agonist) and unmethylated CpG oligodeoxynucleotides (CpG ODN; TLR9 agonist). Further, these results directly correlated with improved survival of mice using a lethal LPS-induced endotoxemia mouse model. In another study, Saito et al. studied the differential cellular interactions of PLGA- and PLA-based iNPs with inflammatory monocytes and neutrophils, which affected disease scores using the experimental autoimmune encephalomyelitis (EAE) model, a mouse model of multiple sclerosis [18]. Getts et al. evaluated the size-dependent effects of tolerogenic NPs (tNPs) using polystyrene beads (PSB) (100 nm to 4.5 μm) conjugated to the antigenic proteolipid peptide PLP_139–151 for the treatment of EAE [17]. 500 nm PLP_139–151-PSB were most effective at reducing disease scores, whereas both smaller and larger NPs were less effective. The effect was not only size-dependent but also surface chemistry-dependent, where tolerance was more efficiently induced using anionic poly(ethylene-alt-maleic acid) (PEMA) as a surfactant compared to poly(vinyl alcohol) (PVA) [35]. The differences in immunomodulatory potential due to alterations in the physicochemical properties of both iNPs and tNPs led us to hypothesize that through systemic modification of iNP physicochemical properties that the anti-inflammatory properties could be tuned, thus enabling fundamental relationships between iNP properties and immune responses to be studied.

In this paper, we report the optimization of a microfluidic method using hydrodynamic flow focusing for the generation of nanometer-sized iNPs and their formulation-dependent immunomodulatory effects using primary murine macrophages. The formation of PLA-based iNPs is achieved through a flow focusing microfluidic method in combination with solvent evaporation leading to nanoprecipitation [8]. The effect of polymer concentration, surfactant concentration, surfactant chemistry, and flow rate ratio (FRR) were used to prepare a focused library of iNPs for immunomodulatory activity screening. Finally, we evaluated the impact of several cryoprotectants on iNP properties following lyophilization. The utilization of a microfluidics platform for iNP formation will enable a broad set of iNPs to be studied for their inherent immunomodulatory properties, which will aid in applications involving inflammation and for targeted drug delivery.

Materials and Methods

Materials

Acid-terminated poly(DL-lactide) (PLA) (approximately MW 11.3 kDa) with low inherent viscosity between 0.16–0.25 dL/g (Product No. B6014-1) was purchased from Lactel Absorbable Polymers (Birmingham, AL). Poly(ethylene-alt-maleic anhydride) E400 (PEMA, MW 400 kDa, Part No. 105564) and PEMA E60 (MW 60 kDa, Part No. 105568) were received as a gift from Vertellus^™ (Indianapolis, IN). Sodium hyaluronate (HA, MW 301,000–450,000, Part No. HA500K-1) was purchased from Lifecore Biomedical (Chaska, MN). Poly(vinyl alcohol) (PVA, MW 30,000–70,000, Part No. P8136), poly(acrylic acid) (PAA, MW 450,000, Part No. 181285), poly-L-glutamic acid sodium salt (PGA, MW 3,000–15,000, Part No. P4636), LPS from Escherichia Coli serotype O111:B4, D-mannitol, sucrose, D-(+)-trehalose dihydrate, acetone, and BAY 11-7082 (NF-κB inhibitor) were obtained from Sigma-Aldrich (St. Louis, MO).

Microfluidics system

iNPs were prepared using a microfluidics system (Dolomite, Royston, UK) equipped with a 5-input chip (150 μm channel) (Part No. 3200711). Polymer, surfactant, and acetone solutions were all 0.2 μm sterile filtered, degassed using sonication for 30 minutes, and loaded into three pressure pumps (Part No. 3200016) attached to either 1–50 μL/min (Part No. 3200098) or 30–1000 μL/min flow rate sensors (Part No. 3200097) (Figure 1). FEP tubing of OD 1.6mm and ID 0.25mm (Part No. 3200302) connected the pressure pumps to the 5-input chip. To control fluid flow through the chip, 2-way in-line valves (Part No. 3200087) were installed. T-connectors (Part No. 3000397) were inserted to connect the polymer and acetone line to facilitate system cleaning. A Meros high-speed digital microscope was utilized to view the laminar flow through the junction of the 5-input chip.

Figure 1.

Schematic representation of the microfluidic platform utilizing a 5-input flow-focusing chip. Modified with permission from Dolomite (Royston, UK). Figure was created with BioRender.com.

Microfluidic preparation of iNPs through hydrodynamic flow focusing

A three-inlet flow-focusing microfluidic device was utilized to generate monodisperse iNPs with varying sizes and low polydispersity (Figure 1). Two streams of aqueous surfactant solution converged with the PLA (organic; acetone) solution at the same or higher pressure to generate a stable laminar flow at the chip junction to prevent backflow. Once the laminar flow was established, the pumps were switched to flow rate control mode to ensure consistent flow rates during iNP formation. The organic-aqueous solvent exchange occurred within the output channel to allow for nanoprecipitation of the iNPs. Hardening of the iNPs occurred continuously as organic solvent diffused. iNPs were subsequently collected into a secondary container. Flow rates were established using the Flow Control Center Software. The FRR was defined as the ratio of surfactant to polymer flow rates and was varied from 0.5 to 4 (Equation 1). iNPs were synthesized with various surfactants: poly(vinyl alcohol) (PVA), PEMA E400 (E400), PEMA E60 (E60), poly(acrylic acid) (PAA), poly(glutamic acid) (PGA), and hyaluronic acid (HA) (Figure S1).

$$\text{FRR}=\frac{{\text{FR}}_{\mathit{\text{Surfactant}}}}{{\text{FR}}_{\mathit{\text{Polymer}}}}$$ (Equation 1)

Evaluation of polymer concentration, surfactant concentrations, and flow rate ratio (FRR) effect on iNP physicochemical properties

iNP were prepared using the microfluidic method described above. PLA was first dissolved in acetone at a concentration of 0.25%, 0.5%, or 1% w/v. To examine the effect of polymer concentration, 0.1% w/v E400 was used. E400 and PVA surfactant solutions were prepared at 0.01%, 0.1%, or 1% w/v to examine the effect of surfactant concentration, while using 0.25% w/v PLA. Both the PLA and E400 (or PVA) flow rate was set at 50 μL/min to generate an FRR of 1, and iNPs were collected for 5 minutes. FRR studies were performed using 0.25% PLA and 0.1% of each respective surfactant. FRRs were set at 0.5, 1, 2, and 4, and iNPs were collected for 5 minutes. Residual acetone was evaporated for 1 hour using an in-house vacuum. The size, polydispersity index (PDI), and zeta potential of the iNPs were determined as described below. All iNPs were formulated the same day for in vitro use.

Single emulsion iNP formulation

iNPs were prepared using the oil-in-water (o/w) single emulsion solvent evaporation (SE) technique using a similar method as described [16]. Briefly, 200 mg of PLA was dissolved in ethyl acetate at a concentration of 80 mg/mL. To this, 20 mL of 1% E400 was added and sonicated at 100% amplitude for 30 seconds using a Cole-Parmer 500-Watt Ultrasonic Processor (Vernon Hills, IL) equipped with a ¼” tip. The resulting o/w emulsion was then poured into 80 mL of 0.5% E400 and stirred overnight. The nanoparticles were then collected by centrifugation at 12,000 × g for 20 min at 4°C and washed with 40 mL of 0.1M sodium bicarbonate/carbonate buffer. The centrifugation and washing steps were repeated three more times, where the last wash cycle used 40 mL of MilliQ water. A mixture of sucrose and mannitol was added to the nanoparticle suspension as cryoprotectants to achieve a final concentration of 4% and 3% w/v, respectively. The nanoparticles were then frozen at −80°C for at least 2 hours prior to lyophilization using a Freezone 4.5L −50C Complete Freeze Dryer System (Labconco, Missouri, USA) for 2 days prior to use in subsequent experiments.

Dynamic light scattering (DLS) to measure iNP physicochemical properties

iNP size, zeta potential, and PDI were measured using a Zetasizer Nano ZSP (Malvern Instruments Inc., Westborough, MA) as previously described [18]. The dispersing medium consisted of MilliQ water (pH 6). It has previously been shown that at physiological pH and electrolyte concentrations that the zeta potential of nanoparticles is oftentimes not high enough to efficiently stabilize the dispersion [19]. Furthermore, the incorporation of physiologically-relevant electrolyte concentration into the measurements may introduce a ‘screening’ of the iNP charges, thus minimizing any potential measured differences between the formulations. The Z-average sizes were recorded as the average of at least three measurements.

Scanning electron microscopy (SEM)

The surface morphology of iNP-PVA₁ was examined using scanning electron microscopy (SEM) using an FEI Quanta 200 (FEI, Hillsboro, OR). Excess surfactant was removed from the iNP suspension by centrifuging at 12,000 × g and washing with MilliQ water twice followed by lyophilization. Lyophilized iNP was placed on carbon adhesive tape, mounted onto an aluminum stub, and dried for at least 2 hr. Samples were sputter-coated using gold and visualized at an accelerating voltage of 12.5 kV and 10 mm working distance.

BMMØ isolation and differentiation

Bone marrow-derived macrophages (BMMØs) were isolated from the tibias and femurs of 5–7-week-old female C57BL/6 mice as previously reported [16]. Complete RPMI 1640 media was prepared with 2 mM L-glutamine (Life Technologies, Carlsbad, CA), penicillin (100 units/mL), streptomycin (100 μg/mL), and 10% heat-inactivated fetal bovine serum (FBS) (VWR, Radnor, PA). To induce BMMØ differentiation, the medium was further supplemented with 20% L929 (ATCC) cell conditioned medium (containing M-CSF). BMMØs were allowed to differentiate for 8 days, with cell conditioned media changes on days 3 and 6. Versene (Invivogen) was used for BMMØ cell lifting. Cell count and viability were obtained using a trypan blue exclusion dye and the EVE^™ automated cell counter (NanoEntek, Waltham, MA).

Cytotoxicity assay

The cytotoxicity of iNPs was evaluated using an MTS assay (Abcam, Cambridge, MA). Day 8 BMMØs were cultured in a 24-well plate at a density of 1×10⁵ cells/well overnight to allow for cell adherence at 37°C and 5% CO₂. BMMØs were treated with 300 μg/mL iNPs for 3 hours, excess iNPs were washed away using fresh medium, and treated with or without LPS for an additional 24 hours (Figure 4B). Cells were again washed with fresh medium and 50 μL of MTS solution was added to each well and incubated for 1 hour. The optical density (O.D.) of the solution was measured using a SpectraMax iD3 microplate reader at 490 nm. The percentage of cell viability was measured as the ratio of O.D. at 570 nm to no treatment control.

Figure 4.

Preparation of iNP library and in vitro cytotoxicity assessment. (A) A set of 12 iNPs was prepared using 2 FRRs (0.5 and 4) with 6 different surfactant chemistries. Similarly sized iNPs could be prepared regardless of FRR, besides iNP-E600.5 and iNP-E4000.5. (B) Schematic of in vitro experimental setup. Bone marrow-derived macrophages are treated with 300 μg/mL of iNPs for 3 hours prior to washing away the excess iNPs. LPS (100 ng/mL) is then added and the cells are incubated for 48 hours prior to the assessment of cytotoxicity or cytokine secretions. (C) MTS assay of various iNP formulations. No statistical differences between groups were observed using a 1-way ANOVA with a Tukey’s post-hoc test. All panels were repeated for a total of at least 3 independent experiments.

Cytokine secretions by BMMØs

To evaluate the modulation of proinflammatory cytokine secretions from BMMØs treated with various iNPs, BMMØs (1 × 10⁵ cells/well) were seeded in a sterile 24-well plate and treated with 300 μg/mL of the 12 different iNP formulations for 3 hours at 37°C and 5% CO₂. BAY 11-7082 was used as a control where cells were pre-treated at 10 μM for 30 min prior to LPS treatment. Excess iNPs were next removed by washing twice with PBS followed by replacing with complete medium containing 100 ng/mL of LPS. After 48 hours, cell culture supernatants were collected and analyzed using an enzyme-linked immunosorbent assay (ELISA) (Biolegend, San Diego, CA) for IL-6 and TNFα following the manufacturer’s protocols.

Cryoprotectant study

The effect of various concentrations of sugars as cryoprotectants for lyophilization were evaluated using freshly prepared iNP-E400₄. iNP-E400₄ was washed twice with MilliQ water and resuspended in 800 μL of MilliQ water in a 2 mL vial followed by the addition to 200 μL of a 5X cryoprotectant solution (or 200 μL MilliQ water as control) to reach a final concentration of 1% sucrose, 5% sucrose, 1% mannitol, 5% mannitol, 1% trehalose, 5% trehalose, or mix (4% sucrose and 3% mannitol). The iNPs were then frozen at −80°C for at least 2 hours prior to lyophilization for 2 days prior to use in subsequent experiments. Samples were reconstituted using 1 mL of MilliQ water and evaluated for size, PDI, and zeta potential using DLS as described above. All measurements were performed in triplicates and the results were compared to measurements taken before adding cryoprotectant and lyophilization.

Statistical analysis

Statistical analyses were performed using Prism 9 (GraphPad, San Diego, CA). Results were reported as mean ± standard deviation (SD). Student’s t-test was used to determine the significance of parametric data between two groups. Significant differences between multiple treatment groups were determined by one-way ANOVA along with Tukey’s multiple comparison test. Statistical comparison between variances were performed using an F-test. In all cases unless otherwise noted, p < 0.05 was considered to be statistically significant.

Results

In this study, we conducted a series of experiments to develop and optimize a microfluidics-based method for the formulation of PLA-based iNPs and investigated their formulation-dependent immunomodulatory properties. This systematic approach was developed to overcome several noted limitations of conventional bulk NP formulation methodologies as well as to establish the differential effects of various formulation parameters on biological activity.

Microfluidic iNP formulation and characterization

We developed a hydrodynamic fluid flow focusing microfluidic method (Figure 1) to generate iNPs using PLA as the core-forming polymer and six types of surfactants (one neutral and five anionic) that differed in their chemistries (Figure S1). First, PLA dissolved in acetone and a surfactant-containing aqueous phase was used to form stable converging streams for iNP generation. The PLA concentration was varied at 0.25%, 0.5% and 1% w/v and a single surfactant, E400, at 0.1% was used (Figure 2A,B). At an FRR of 1, variation of the PLA concentration did not alter the particle size appreciably and all iNP formulations were approximately 400 nm. The zeta potential was highly negative ranging from −46 to −42 mV and PDI was between 0.18 and 0.21 for all formulations (Figure 2A). Notably, precipitation was observed in the outlet channel of the chip for PLA concentrations greater than 0.25% w/v (Figure 2B). This precipitation resulted in clogging of the chip that prevented further iNP generation over time. The 0.25% w/v PLA condition was subsequently chosen as precipitation was minimal.

Figure 2.

Impact of polymer concentration and surfactant type and concentration on iNP size, zeta potential, and PDI. (A,B) PLA concentration was varied from 0.25 to 1% w/v using a fixed 0.1% w/v E400 surfactant solution. (C,D) E400 surfactant concentration was varied from 0.01 to 1% w/v using a fixed PLA concentration of 0.25% w/v. (E,F) PVA surfactant concentration was varied from 0.01 to 1% w/v using a fixed PLA concentration of 0.25% w/v. Corresponding chip junction images are presented in panels (B,D,F). All iNPs were prepared using a fixed FRR of 1. Statistical differences between groups were determined using a 1-way ANOVA with a Tukey’s post-hoc test. ** p<0.01. All panels were repeated for a total of at least 3 independent experiments.

Next, we evaluated the role of surfactant concentration on iNP physicochemical properties. The PLA concentration was fixed at 0.25% w/v and the concentration of E400 was varied at 0.01%, 0.1%, and 1% w/v using an FRR of 1 (Figure 2C,D). Increasing the E400 solution concentration from 0.01% to 0.1% marginally increased the particle size (189 to 265 nm), whereas the use of the 1% E400 solution led to aggregation and a measured size of greater than 3500 nm (Figure 2C). Following similar trends, the PDI increased from 0.10 to 0.56 as the E400 concentration increased. Interestingly, the zeta potential was more neutral for the 1% E400 iNPs compared to the 0.01% and 0.1% E400 iNPs. Similar to the 1% w/v PLA, the 1% w/v E400 iNPs demonstrated high precipitation within the chip (Figure 2D). PVA was used to examine if it had similar effects on iNP properties as E400, where the same concentrations was used with 0.25% w/v PLA solution at an FRR of 1 (Figure 2E,F). iNPs prepared using PVA displayed particle sizes between 180–260 nm with zeta potentials between −26 to −7 mV and PDIs below 0.1 (Figure 2E). There was no precipitation formed in the chip at any PVA concentration tested (Figure 2F). SEM imaging confirmed the spherical topology, monodispersity, and homogeneity of iNPs, as shown through a representative iNP-PVA image (Figure S2). Lastly, we directly compared the reproducibility of iNP formation by the bulk emulsification method (n=18) versus the microfluidics method (n=19) using F-test to assess differences in the variance for size and PDI (Figure S3). iNPs prepared using the microfluidics method displayed significantly different variances with regard to size and PDI compared to iNPs prepared using emulsification. Based on these collective findings, subsequent experiments using microfluidics utilized 0.25% w/v PLA and 0.1% surfactant concentrations as their physicochemical properties were deemed appropriate and precipitation in the chip was minimal.

Impact of surfactant chemistry and flow rate ratio on iNP formation

The FRR of surfactant and polymer solutions controls the width of the laminar flow at the junction where the solution streams converge, which directly affects the mixing of the surfactant and polymer solutions. The effect of FRR on iNP properties was established using six surfactants (E400, E60, PAA, PGA, HA, and PVA) (Figures S1 and 3). The nomenclature for various iNPs prepared was defined as iNP-Surfactant_FRR (Figure 1). iNP-E400 and iNP-E60 displayed the most dynamic size range (200 nm to 1250 nm) based on FRR, with PDIs from 0.10 and 0.30. As the FRR decreased, the particle size and PDI increased. This inverse relationship allowed for the reproducible and fine-tuning of particle sizes in an FRR-dependent manner. For iNPs prepared with PAA, PGA, PVA, and HA, the particle size intervals were narrower, less dependent on FRR, and ranged from 100 nm to 300 nm with PDIs from 0.02 to 0.30.

In vitro cytotoxicity and immunomodulatory properties of iNPs

To assess the toxicity profiles and demonstrate the immunomodulatory effects of the microfluidics-generated iNPs, two types of iNPs were prepared for each of the six surfactant solutions at FRRs of 0.5 and 4 (Figure 4A and Table 1). Two FRRs were examined for cytotoxicity as it encompassed the range of physicochemical properties achievable. BMMØs were treated with iNPs (set of 12) prior to LPS stimulation (Figure 4B). BAY 11-7082 was used as a control due to its inhibitory activity on IκB kinase (IKK) [20], which completely eliminated the secretion of IL-6 and TNFα through modulation of NF-κB (nuclear factor κB) signaling. Using an MTS assay, the effects of iNP treatment on cell viability were determined (Figure 4C). All iNPs evaluated did not result in any significant differences in viability compared to the untreated controls. Formulations prepared with FRRs between 0.5 and 4 were expected to be non-toxic despite differences in physicochemical properties. As an example, HA had a limited size range as particle size was similar for both FRR 0.5 and 4, and demonstrated similar lack of toxicity. Smaller iNPs, less than 250 nm, generally led to greater suppression of IL-6 and TNFα secretions, and iNP zeta potentials did not correlate with the suppressive capabilities of the set of iNPs evaluated (Figure S4). However, further investigation identified that the FRR used to generate iNPs was a determining factor for the measured reductions in proinflammatory cytokine secretions. All iNPs prepared with an FRR of 4 were highly efficient at reducing both IL-6 and TNFα secretions to less than 20% of the LPS controls (Figures 5A–C and S4). This was in stark contrast to iNPs prepared with an FRR of 0.5, although several iNPs shared similar size distributions between the two FRRs. The most effective set of iNPs were those prepared using E60, PGA, HA, E400, and PAA at an FRR of 4. Conversely, iNP-PAA_0.5 displayed the most significant increase in TNFα (117% of LPS control) and only minor reductions in IL-6 secretions (52% of LPS control), despite its small size. iNP-E400_0.5 and iNP-E60_0.5 were the largest and least effective at reducing cytokine secretions. To simplify the data interpretation, we calculated a “composite cytokine suppression ratio” (CCSR) as an unweighted average of the relative measured IL-6 and TNFα secretions for iNPs compared to the LPS control (Figure 5D). As these cytokines, in part, can be used to predict the severity of inflammation-associated disease outcomes such as sepsis [21], this metric enabled the direct comparison of the anti-inflammatory effects for all iNPs produced.

Table 1.

Characterization of particle size (nm), PDI, and zeta potential (mV) of the 12 iNPs used in all in vitro assays. Data is presented as mean ± SD of three independent experiments.

iNP	Surfactant	FRR	Particle Size (nm)	PDI	Zeta Potential (mV)
iNP-E4000.5	0.1% PEMA E400	0.5	869.2 ± 144.9	0.108 ± 0.043	−47.9 ± 4.9
iNP-E4004	0.1% PEMA E400	4	229.3 ± 18.6	0.217 ± 0.021	−57.3 ± 8.6
iNP-E600.5	0.1% PEMA E60	0.5	1080.5 ± 88.9	0.347 ± 0.191	−50.9 ± 6.0
iNP-E604	0.1% PEMA E60	4	209.1 ± 13.5	0.139 ± 0.051	−53.5 ± 9.0
iNP-HA0.5	0.1% HA	0.5	249.5 ± 44.4	0.299 ± 0.025	−48.5 ± 1.2
iNP-HA4	0.1% HA	4	295.0 ± 82.3	0.346 ± 0.153	−63.5 ± 12.2
iNP-PAA0.5	0.1% PAA	0.5	274.8 ± 56.8	0.260 ± 0.027	−35.3 ± 4.0
iNP-PAA4	0.1% PAA	4	192.5 ± 11.0	0.270 ± 0.023	−46.8 ± 8.5
iNP-PGA0.5	0.1% PGA	0.5	149.9 ± 47.1	0.144 ± 0.060	−52.6 ± 5.3
iNP-PGA4	0.1% PGA	4	100.6 ± 17.5	0.099 ± 0.016	−59.7 ± 10.4
iNP-PVA0.5	0.1% PVA	0.5	228.6 ± 75.1	0.097 ± 0.032	−21.6 ± 6.5
iNP-PVA4	0.1% PVA	4	148.1 ± 32.5	0.051 ± 0.023	−20.6 ± 4.7

Figure 5.

iNP property-dependent modulation of inflammatory cytokine secretions. (A,B) Bone marrow-derived macrophages are treated with 300 μg/mL of iNPs for 3 hours prior to washing away the excess iNPs. LPS (100 ng/mL) is then added and the cells are incubated for 48 hours prior to the assessment of cytokine secretions (IL-6 and TNFα) using ELISA. # p<0.05 compared to LPS only control. & p<0.05 compared to corresponding iNP at FRR 4. (C) Scatterplot displaying the relationship between iNP properties on IL-6 and TNFα cytokine secretions as a percentage of LPS control. (D) Calculation of the composite cytokine secretion ratio (CCSR) for iNPs evaluated in this study. CCSR is calculated as an arithmetic average of the percentages of cytokine secretions relative to LPS control. Statistical differences between groups were determined using a 1-way ANOVA with a Tukey’s post-hoc test. n.s. (not significant from BAY 11-7082 control). * p<0.05. ** p<0.01. **** p<0.0001 compared to BAY 11-7082. All panels were repeated for a total of at least 3 independent experiments.

Assessment of cryoprotection strategies for iNPs

Four cryoprotectant strategies were evaluated using iNP-E400₄ at multiple concentrations with seven conditions evaluated in total (1% sucrose, 5% sucrose, 1% mannitol, 5% mannitol, 1% trehalose, 5% trehalose, and mix (4%sucrose and 3% mannitol)). iNP-E400₄ was utilized as a representative iNP to evaluate the effect of various cryoprotection strategies as preliminary studies demonstrated that iNP surfactant composition did not contribute to cryoprotectant efficacy (data not shown). All cryoprotectants performed similarly and enabled retention of similar size and PDI as the iNP-E400₄ before lyophilization (Figures 6A,B). iNP-E400₄ without the addition of cryoprotectants (water only) resulted in significant increases in particle size and PDI, indicating aggregation of the iNP. Interestingly, the choice of cryoprotectant led to differences in the measured zeta potentials (Figure 6C). The addition of cryoprotectants led to slight neutralization of iNP zeta potentials between −25 mV to −35 mV. Overall, there was marginal differences in the physicochemical properties of the iNPs after lyophilization, demonstrating the effectiveness of the cryoprotectants.

Figure 6.

Effects of various cryoprotectants on iNP properties following lyophilization. iNP-E400₄ was synthesized using 0.1% E400 and 0.25% PLA at FRR 4. (A) Particle size (nm). (B) PDI. (C) Zeta potential (mV). Water – Lyophilized iNP without the addition of cryoprotectants and only used MilliQ H2O. Representative data from 3 independent experiments.

Discussion

Nanomedicine has rapidly evolved in the past decades to demonstrate its useful application in numerous areas, such as diagnosis, prevention, and treatment of diseases. Several NP-based formulations have reached clinical trials and gained FDA approval [2]. Although many advances have been made regarding NP formulation, bulk formulation strategies suffer several limitations such as high polydispersity, large batch-to-batch variability, and low recovery. Several groups have employed microfluidics strategies to create diverse libraries of NPs for biological applications. Poly(ethylene glycol) (PEG)-PLA polymers and corresponding LIG-PEG-PLGA polymers (as a model targeting ligand for cancer cells) were systematically combined to enable a library of NPs to be generated to study targeting ligand densities [22]. Another study loaded hydrophilic N-acetylcysteine into PLGA nanoparticles [23]. In the present study, we sought to develop and optimize a microfluidic method for the generation of iNPs to increase the throughput of formulations for the evaluation of their inherent immunomodulatory properties. Employing a microfluidics method allows for the generation of distinct libraries comprised of a variety of iNP formulations, reduction in synthesis time, and production of small quantities of iNPs for biological screening experiments yet is scalable without the need to modify many formulation parameters.

Throughout the course of our investigation, optimal formulation parameters to prepare iNPs were identified. Polymer and surfactant concentrations were crucial to producing high-quality iNPs with narrow PDIs (Figure 2). The type of surfactant used to produce iNPs altered the surface chemistry, which affected the particle size, PDI, and zeta potential. Increasing the FRR from low (0.5) to high (4) generally decreased the particle size and PDI, while the particle size ranges that could be achieved were dependent on the type of surfactant used (Figure 3). For example, using E400 or E60 surfactants, the greatest particle size range (approximately 1000 nm) was achieved, whereas all other iNPs prepared using PAA, PGA, PVA, and HA displayed narrower particle size ranges (approximately 100–150 nm). The optimization parameters established here defined the procedures and protocols for microfluidic nanoparticle synthesis. Further optimization of the microfluidic parameters is anticipated to result in a wider range of particle sizes through modification of polymer solvent and concentration, surfactant concentration, as well as FRR or microfluidic chip type.

Figure 3.

Impact of surfactant chemistry and FRR on iNP size and PDI. A set of 6 different surfactants were evaluated including poly(ethylene-alt-maleic acid) E400 and E60, hyaluronic acid (HA), polyglutamic acid (PGA), polyacrylic acid (PAA), and polyvinyl alcohol (PVA). Polymer and surfactant concentrations were fixed at 0.25% and 0.1% w/v, respectively. All panels were repeated for a total of at least 3 independent experiments.

One challenge with our current microfluidic setup was the low nanoparticle production rate, where a maximum flow rate of the 0.25% w/v polymer solution of 50 μL/min corresponded to 0.125 mg iNP/min or 180 mg iNP/day. Consequently, in most of our experimental testing, we performed a run time of 8 minutes to produce 1 mg of iNPs. Given the modularity of the microfluidic platform to allow for scale-up, achieving a flow rate of 5 mL/min will be possible through utilizing a different flow rate sensor. This increase in polymer flow rate would improve the iNP production rate up to 12.5 mg/min (i.e. 18 g/day), a 100-fold potential increase in output. Furthermore, it is possible that the increased flow rate may decrease the level of precipitation observed at the chip junction as in Figure 2, which may allow for even higher production rates. For in vitro examination, since the required iNPs needed was minimal (0.45 mg), iNPs were synthesized on the same day. Various batches were synthesized and demonstrated controllable physicochemical properties (Table 1). Moreover, the microfluidics-based formulation produced iNPs with less variable physicochemical properties than corresponding iNPs formulated using the bulk emulsification method (Figure S3). The versatility of microfluidic iNP formulation in combination with traditional high-throughput screening methods offers a significant opportunity to develop improved nanomedicines in shorter timeframes [24].

In this study, we hypothesized that iNP physicochemical properties could be tuned to affect macrophage inflammatory responses upon LPS stimulation. iNP properties such as polymer composition, surface chemistry, and size were modified and anticipated to have direct effects on their immunological activity as similarly observed for other NPs [25, 26]. We used IL-6 and TNFα as readouts to determine iNP formulation-dependent effects on inflammatory responses as these cytokines have been shown to contribute to cytokine storm-induced mortality in sepsis [27–29]. Interestingly, the immunomodulatory potential of iNPs was correlated with FRR but not with particle size or zeta potential (Figures S4, S5). The FRR-dependent differences in immunomodulatory potential may be associated with different rates of mixing between the surfactant and polymer solutions (Figure 5) [30]. Nanoprecipitation involves the nucleation and aggregation of polymers to form iNPs (Figure 1). Using higher FRR results in more rapid mixing and faster exchange of the polymer-solvent. Therefore, the rate of nucleation would be expected to be faster at higher FRRs due to the higher supersaturation [22]. Casey et. al previously showed that soluble E400, E60, and PVA did not reduce proinflammatory cytokine secretions [16] and another group demonstrated the molecular weight-dependent anti-inflammatory effects of HA [31]. Dashtimoghadam et al. reported that hydrophobically modified chitosan nanoparticle compactness can be altered based on FRRs during microfluidic nanoparticle synthesis [32]. Thus, it is conceivable that the higher FRRs used in our study resulted in more compact iNPs by increasing the lactic acid (as a result of PLA degradation) concentration per volume of iNP, which would affect associated iNP-mediated immunomodulation. To facilitate comparison between the various iNP formulations, we developed and calculated the CCSR for each iNP evaluated to rank their abilities to reduce pro-inflammatory responses (Figure 5D). Notably, all iNPs evaluated at an FRR of 4 were more effective than those prepared at an FRR of 0.5 (besides iNP-PVA_0.5) and not significantly different from the BAY 11-7082 control (Figures 4, 5). This result was contrary to our hypothesis and was unlike our previous study using bulk (single emulsion) prepared 500 nm iNPs where PLA-PEMA was more effective than PLA-PVA at suppressing proinflammatory cytokine secretions from BMMØs induced by LPS or CpG ODN [16].

Microfluidic-generated iNPs were prepared using 10–100-fold lower surfactant concentrations compared to iNPs produced using bulk methods [16]. As a result, iNPs could be used in vitro without washing away the excess surfactant solution. This allows its same-day use and expediting the synthesis procedure for the purposes of high-throughput formulation testing. With this, there were no observed negative effects of excess surfactant on the toxicity or immunomodulatory potential of iNPs (Figures 4C, 5). Beyond using the microfluidic system for the generation of small batch sizes, eventual large batch synthesis would be performed for future in vivo testing. To ensure the long-term stability, biological activity, and safety of large batch sizes, the use of cryoprotectants and undergoing lyophilization is crucial. We observed that it was not feasible to directly lyophilize the output iNP and surfactant mixture from the microfluidic device even with surfactant concentrations as low as 0.01% and inclusion of 4% sucrose and 3% mannitol as cryoprotectants due to difficulties associated with iNP reconstitution (data not shown). Rather, removal of excess surfactant using a centrifugation procedure was required prior to lyophilization with cryoprotectants to allow for homogeneous iNP suspensions following reconstitution. Our studies demonstrated that removal of excess surfactant is necessary and suggested that the use of more efficient methods for iNP purification such as tangential flow filtration may provide an appropriate solution for the purification of large iNP batch sizes. We hypothesized that utilizing different cryoprotectants may prevent iNP aggregation during lyophilization. Thus, investigation into the effect of concentration of three various cryoprotectants on iNP properties after lyophilization was evaluated (Figure 6). Several factors including the concentration of iNP, freezing rate, the cryoprotectant concentration, surface chemistry of the iNP, and polymers used during formulation can be tuned to obtain a lyophilized product that upon reconstitution does not possess significantly different physicochemical properties from the fresh iNP solution [33]. Although the particle size and PDI of iNP-E400₄ was retained for all cryoprotectant conditions, the zeta potential was altered more significantly in some cases. The increased zeta potential of iNPs after cryoprotectant additions may be attributed to cryoprotectant association with the iNP surface. Taken together, our data support that iNP physicochemical properties were retained following lyophilization and that previously utilized cryoprotectants such as 4% sucrose and 3% mannitol (i.e. 70 mg sugars/mL iNP suspension) [18] could be replaced with up to 7-fold lower concentrations of sugars such as 1% trehalose (10 mg sugar/mL iNP suspension).

The immunomodulatory activity of iNPs is predicted to be driven by polymer composition and cellular uptake. Lactic acid, the degradation product of iNPs, is a byproduct of glycolysis that has been shown to be a potent modulator of antigen-presenting immune cell phenotypes [34]. Lactic acid derived from PLGA degradation has been shown to alter dendritic cell (DC) phenotype and reduce pro-inflammatory NF-κB signaling as well as other pathways resulting in an inactivated state [35, 36]. The surface chemistry of NPs affects cellular interactions and uptake [37, 38]. Gold NPs (GNPs) were modified with hydrophobic and aromatic amino acids, and it was shown that the uptake could be controlled in a surface composition-dependent manner [39]. Using hexapeptide modifications on GNPs, tryptophan, phenylalanine, and tyrosine end groups led to the most significant uptake as opposed to leucine, isoleucine, and valine. Notably, the same hexapeptide-modified GNPs attenuated NF-κB/AP-1 and IRF3 signaling induced by LPS [40]. Interestingly, GNPs alone (limited cellular interactions) or hexapeptides alone did not alter inflammatory signaling, which indicated that it was the combination of the hexapeptide and GNP that was responsible for the activity. Future direction of the microfluidics system involves examination of various polymer types on their effect on cellular uptake to improve the immunomodulatory effect of iNPs.

Conclusion

Microfluidic platforms are a promising new technology to develop diverse libraries of NPs rapidly and in a controlled and reproducible manner. Here, we sought to develop and optimize a microfluidic method to overcome some of the challenges associated with conventional methods used for NP formulation. Our microfluidics system has demonstrated its modular ability for scale-up to fit necessary demands, have limited batch-to-batch variability, lowered polydispersity, and simplified purification process for high-throughput purposes. Our results demonstrate that microfluidics is a methodology useful to generate a variety of iNPs to evaluate the impact of formulation parameters on their physicochemical properties. A set of twelve iNPs were prepared for additional biological testing where their cytotoxicity and ability to modulate macrophage responses under LPS stimulation were measured. FRR was found to be a critical feature associated with the immunomodulatory potential of iNPs. As storage stability is crucial for the implementation of iNPs widely, we evaluated the impact of various cryoprotectants to maintain the size, PDI, and zeta potential of iNPs. Our results are encouraging evidence that NP property-dependent immunomodulation can be specifically evaluated and potentially used in conjunction with drug loading to induce controlled immune responses in vivo. The versatility of this platform allows for its future use in other inflammatory models or targeted drug delivery applications.

Supplementary Material

Figure S1

Figure S1. Six iNP formulations utilizing one of the six various surfactants on the iNP exterior. Chemical structures of (1) poly(ethylene-alt-maleic acid) E60/E400, (2) poly(vinyl alcohol) (PVA), (3) poly(acrylic acid) (PAA), (4) poly(glutamic acid) (PGA), and (5) hyaluronic acid (HA) utilized for iNP synthesis.

Figure S2

Figure S2. Representative SEM image of iNP-PVA₁ prepared using microfluidics.

Figure S3

Figure S3: Comparison of two different nanoparticle synthesis methods: emulsification and microfluidics. Both nanoparticles were synthesized using E400 and PLA and measured for particle size (nm) and PDI. (A) A statistical F test to compare variance was performed, where particle size measurements for emulsification (n=18, SD=144.4) and microfluidics (n=19, SD=51.99) were statistically different (p<0.0001). (B) A statistical F-test to compare variance was performed, where PDI measurements for emulsification (n=18, SD=0.07031) and microfluidics (n=19, SD=0.0414) were statistically different (p<0.0315).

Figure S4

Figure S4. Correlation between particle size and zeta potential of iNPs on pro-inflammatory cytokine secretions. Data was obtained from experiments presented in Figure 5.

Figure S5

Figure S5. Correlation between FRR of iNPs on pro-inflammatory cytokine secretions. Data was obtained from experiments presented in Figure 5.