Microfluidics-based self-assembly of peptide-loaded microgels: Effect of three dimensional (3D) printed micromixer design

Accepted Manuscript Microfluidics-based self-assembly of peptide-loaded microgels: Effect of three dimensional (3D) printed micromixer design Bruno C. Borro, Adam Bohr, Saskia Bucciarelli, Johan P. Boetker, Camilla Foged, Jukka Rantanen, Martin Malmsten PII: DOI: Reference:

S0021-9797(18)31444-9 https://doi.org/10.1016/j.jcis.2018.12.010 YJCIS 24387

To appear in:

Journal of Colloid and Interface Science

Received Date: Revised Date: Accepted Date:

26 October 2018 27 November 2018 3 December 2018

Please cite this article as: B.C. Borro, A. Bohr, S. Bucciarelli, J.P. Boetker, C. Foged, J. Rantanen, M. Malmsten, Microfluidics-based self-assembly of peptide-loaded microgels: Effect of three dimensional (3D) printed micromixer design, Journal of Colloid and Interface Science (2018), doi: https://doi.org/10.1016/j.jcis.2018.12.010

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Microfluidics-based self-assembly of peptide-loaded microgels: Effect of three dimensional (3D) printed micromixer design Bruno C. Borro1,*, Adam Bohr1, Saskia Bucciarelli2, Johan P. Boetker1, Camilla Foged1, Jukka Rantanen1, Martin Malmsten1,3 1Department

of Pharmacy, University of Copenhagen, Universitetsparken 2, DK-2100 Copenhagen,

Denmark 2Department

of Drug Design and Pharmacology, University of Copenhagen, Jagtvej 162, DK-2100

Copenhagen, Denmark 3Department

of Pharmacy, University of Uppsala, SE-75123 Uppsala, Sweden

*Corresponding author: Mail: [email protected]. Tel.: +4571378628 Short title: Microfluidics-based self-assembly of peptide-loaded microgels Keywords: antimicrobial peptides, biomaterials, drug delivery, microfluidics, microgels, 3D printing.

1

Abstract In an effort to contribute to research in scalable production systems for polymeric delivery systems loaded with antimicrobial peptides (AMPs), we here investigate effects of hydrodynamic flow conditions on microfluidic particle generation. For this purpose, rapid prototyping using 3D printing was applied to prepare micromixers with three different geometric designs, which were used to prepare Ca 2+-cross-linked alginate microgels loaded with the AMP polymyxin B in a continuous process. Based on fluid dynamic simulations, the hydrodynamic flow patterns in the micromixers were designed to be either (i) turbulent with chaotic disruption, (ii) laminar with convective mixing, or (iii) convective with microvortex formation. The physicochemical properties of the microgels prepared with these micromixers were characterized by photon correlation spectroscopy, laser-Doppler micro-electrophoresis, small-angle x-ray scattering, and ellipsometry. The particle size and compactness were found to depend on the micromixer geometry: From such studies, particle size and compactness were found to depend on micromixer geometry, the smallest and most compact particles were obtained by preparation involving microvortex flows, while larger and more diffuse microgels were formed upon laminar mixing. Polymyxin B was found to be localized in the particle interior and to cause particle growth with increasing peptide loading. Ca2+-induced cross-linking of alginate, in turn, results in particle contraction. The peptide encapsulation efficiency was found to be higher than 80% for all investigated micromixer designs; the highest encapsulation efficiency observed for the smallest particles generated by microvortex-mediated self-assembly. Ellipsometry results for surfaceimmobilized microgels, as well as results on peptide encapsulation, demonstrated electrolyte-induced peptide release. Taken together, these findings demonstrate that rapid prototyping of microfluidics using 3D-printed micromixers offers promises for continuous manufacturing of AMP-loaded microgels. Although the micromixer combining turbulent flow and microvortexes was demonstrated to be the most efficient, all three micromixer designs were found to mediate self-assembly of small microgels displaying efficient peptide encapsulation. This demonstrates the robustness of employing 3D-printed micromixers for microfluidic assembly of AMP-loaded microgels during continuous production.

2

1. Introduction Microgels are dispersions of loosely cross-linked polymeric colloids, which can be designed to display pronounced swelling transitions in response to a wide range of stimuli, e.g., ionic strength, pH, light, temperature, reducing conditions, and the presence of specific solutes

1,2

. Microgels have potential as

delivery systems for biopharmaceuticals, e.g., peptides, proteins, and nucleic acids, due to this swelling behaviour. In addition, they may protect such compounds from chemical and/or enzymatic degradation, and/or increase the physical stability by reducing unfavourable aggregation or conformational changes3. Commonly used approaches for preparing microgels include emulsion or suspension polymerization, formation by homogeneous nucleation, and complexation4. While these approaches provide excellent control of the microgel generation as such, if properly designed, they are essentially multi-step batch processes, and the active pharmaceutical ingredients are frequently loaded after particle formation for compatibility reasons, which is particularly prevalent for labile biomacromolecular drugs. Hence, these methods represent a challenge from a manufacturing point of view, because they are at risk of being either time-consuming, costly, or unsuitable for upscaling 5,6. Considering this, methods based on microfluidics have been explored as alternative approaches for microgel preparation as they allow for precise control of both size distribution and drug loading

7,8

.

However, the microfluidic devices commonly used for such investigations are prepared from poly(dimethylsiloxane) (PDMS), and are not readily scalable for industrial manufacturing due to both complicated and expensive fabrication methods and low production yield

5,9,10

. An emerging alternative is

the use of additive manufacturing (e.g., 3D printing) for the fabrication of micromixers, which are suitable for the preparation of nanocomplexes in scalable setups because they provide opportunities for parallelization of modules 11. These production geometries can also be designed to operate in a continuous mode, which is potentially more cost-efficient than traditional batch production

12

. Despite being less

sophisticated than PDMS-based microfluidic cells of complex designs, initial studies have demonstrated that the production of nanoparticles by the use of 3D-printed micromixers is highly reproducible, eventually resulting in comparable nanoparticle sizes to those obtained by using other methods 11. Hence, they offer an interesting alternative for parallelized microfluidic generation of nanomedicines with increased focus on the implementation of Quality by Design (QbD) principles 5. In contrast to the relative complexity of preparing mold-based silicon microfluidic cells, 3D printing provides a rapid, cheap, and flexible approach for the fabrication of devices, including microfluidic cells 11. Hence, we investigated the applicability of such 3D-printed micromixers for the preparation of microgels loaded with antimicrobial peptides (AMP). AMPs are short amphiphilic peptides, generally carrying a net

3

positive charge and containing a significant fraction of hydrophobic amino acids. They exert broadspectrum antimicrobial effects due to their surface activity, eventually resulting in lysis of bacterial membranes 13–15. Although considerable efforts have been dedicated to the identification and optimization of AMPs in order to optimize potency, selectivity, safety, and stability, drug delivery aspects have received considerably less attention 16. However, delivery systems may increase AMP performance in various ways, e.g., through (i) protecting them from proteolytic degradation in infected tissue

17

, (ii) reducing AMP

binding to anionic serum proteins, eventually resulting in prolonged blood circulation time and improved bioavailability

18

, and (iii) promoting selective cell internalization to achieve intracellular antimicrobial

effects without killing host cells 19. From these perspectives, microgels offer interesting opportunities as AMP delivery systems. In the present study, we investigated how the geometric design of 3D-printed microfluidic micromixers affects the generation of AMP-loaded microgels. The gels were composed of the biodegradable polymer alginate, which effectively cross-links in the presence of Ca2+ through formation of so-called egg-box structures (Figure S1A, Supplementary Material). While there may be a semantic issue if the particles obtained should be referred to as “microgels”, “polyelectrolyte complex particles”, or “polyplexes”, we refer to these as “microgels” throughout in the following discussion, as non-covalently cross-linked polymer structures are frequently observed in various systems, loosely referred to as “microgels” in literature. In investigating these systems, three different types of 3D-printed micromixers were designed according to principles of mixing by hydrodynamic focusing

20

. The micromixer geometries were designed to yield

different mixing patterns, which were verified by fluid dynamic simulations. Subsequently, the micromixers were used to prepare microgels at varying concentrations of polymyxin B, sodium alginate, and CaCl2. Polymyxin B (Figure S1B, Supplementary Material) is a cyclic lipopeptide with surfactant properties, and was chosen as a model antimicrobial peptide due to the potential stabilizing effect of its acyl moiety on particle formation

21

, as well as due to the potent antimicrobial effect displayed by this lipopeptide

Alginate, in turn, was selected due to its known biodegradability and low toxicity

23

22

.

; but also because it

forms very strong gels in the presence of calcium ions23. The structural gel features were subsequently characterized via dynamic light scattering, zeta potential measurements, and small-angle x-ray scattering (SAXS), while information on peptide encapsulation and release was obtained from bicinchonic acid assay and ellipsometry, respectively.

4

2. Materials and Methods 2.1 Materials Sodium alginate (CAS number: 9005-38-3; 1405-20-5;

), and

), polymyxin B sulfate salt (CAS number: (CAS number: 10035-04-8) were acquired from

Sigma Aldrich (Stockholm, Sweden). Polyamide (SLS PA 2200) and polypropylene (SI-25) powders, tailored for 3D printing, were obtained from Damvig 3D Printed Solutions (Taastrup, Denmark), while Titanium (TiAl6V4) powder was obtained from Materialise (Leuven, Belgium). All other chemicals used were of analytical grade, and obtained from Sigma-Aldrich. Ultrapure Milli-Q water (MQ, 18.2

) was used

for all experiments. 2.2 Design and 3D printing of micromixers Hydrodynamic focusing micromixers

20

were designed using the COMSOL Multiphysics software, version

4.4 (Stockholm, Sweden). The three micromixer geometries were designed to display different mixing profiles (Figure 1), and are referred to as (A) turbulent flow micromixer (TFM), (B) laminar flow micromixer (LFM), and (C) integrated compartment micromixer (ICM), respectively. In the TFM, sheat flows are introduced at angles of 90 degrees from the top and bottom of the central flow, respectively 11. The LFM, in turn, consists of a coaxial tube micromixer with two sheat flows injected from the outer layer. The ICM was designed as a planar micromixer with four sheat flows, each half introduced at the beginning and the end of the device, respectively. For 3D printing, the designs were exported as “.stl” files, loaded into the MakerWare software version 2.4.1 (Makerbot Industries, New York, NY, USA), and printed using the MakerBot Replicator 2 (Makerbot).

Figure 1. Technical drawings of the different types of micromixers: (A) turbulent flow micromixer (TFM), (B) laminar flow micromixer (LFM), and (C) integrated compartment micromixer (ICM).

5

2.3 Fluid dynamics simulations The Navier-Stokes equation (1) was employed to model the flow of fluids in the micromixers, following the finite element method 11: (1)

where

is the fluid velocity,

is the fluid density,

is the fluid pressure, and μ is the fluid dynamic

viscosity. The system of differential equations derived from this expression was solved using the parallel direct solver (PARDISO). COMSOL multiphysics, version 4.4 (Stockholm, Sweden) was used to design the inner geometry of the micromixer, using boundary conditions of (i) the corresponding velocities of the fluid at the inlets, and (ii) zero pressure at the outlets.

2.4 Microfluidic mixing Polymyxin B-loaded alginate-Ca2+ microgels were prepared employing the 3D printed micromixers, organized in either one of two different operating setups, depending on the specific micromixer design (Figure 2). The inlets were dedicated to solutions of 200 ppm sodium alginate, 0-48 60-600

polymyxin B, and

CaCl2, prepared in 10 mM Tris-HCl (pH 7.4). All inlets were connected to low protein-binding

Tygon 3350 silicone tubings (ISMATEC, Wertheim, Germany), which were further connected to gas-tight glass syringes (Hamilton, Reno, NV, USA) at the other end. The flow rates for each of the inlets (Table 1) were controlled using Harvard Apparatus Elite pumps (Harvard Apparatus, Cambridge, MA, USA).

6

Figure 2. Operating setups for the micromixers. The dual compartment setup (A) consists of two micromixers of the same type connected in series. For this setup, the outlet of the first micromixer was split into two via a y-junction (Mikrolab, Aarhus, Denmark), and connected to the lateral inlets of the second micromixer. This setup was employed for the TFM and LFM. The integrated compartment (B) was used for the ICM. The latter two inlets of this type of micromixer were connected to a y-junction to allow feeding of this stream with the use of a single syringe. Table 1. Flow rates (mL/min) for the inlets of the micromixers. Inlet component

TFM

LFM

ICM

Alginate

3.0

0.3

3.0

Polymyxin B

1.0

0.1

1.0

CaCl2

1.0

0.1

1.0

2.5 Size and zeta potential measurements Dynamic light scattering (DLS) and electrophoretic mobility measurements were used to estimate the average hydrodynamic radius (

), polydispersity index (PDI), and effective zeta potential of the microgels.

The measurement were performed using the Zetasizer Nano ZS (Malvern Instruments, Worcestershire, UK), and data was analysed using Malvern v7.02 software (Malvern Instruments). All measurements were performed in triplicates for the indicated solutions at 25 ◦C, λ = 633 nm, and 173◦ scattering angle.

2.6 SAXS measurements SAXS measurements were performed using a Xenocs BioXolver L (Xenocs SAS, Sassenage, France), equipped with a 250W liquid Gallium alloy X-ray source (MetalJet, Excillum, Kista, Sweden) and a motorized detector, allowing the sample-to-detector distance to be varied. This instrument produces an X-ray beam with a wavelength of λ = 1.34 Å. Samples were prepared in a 96-well tray and kept at 25°C using a thermostated sample holder (Xenocs SAS). Subsequently, the samples were automatically loaded into the exposure cell, connected to the same thermostat, using a sample handling robot (Xenocs SAS). 2D scattering images were collected in multiple 60 s exposures, and were automatically corrected for background radiation, transmission, and direct beam intensity. Subsequently, these were radially averaged into scattering intensities I(q), where q = (4π/λ)sin(θ/2) is the magnitude of the scattering vector and θ is the scattering angle, by using the software RAW 24,25. Two sample-detector distances d were selected to cover a broad qrange (d = 654 mm with q = 0.01 – 0.5 Å-1 and d = 1507 mm with q = 0.007 – 0.2 Å-1). In the configuration

7

with the short sample-to-detector distance, samples and the corresponding buffer were exposed for a total of 14 min, whereas in the case of d = 1507 mm, the total exposure time was 20 min. The 1D curves were subsequently averaged, background-corrected and merged to cover a total q-range of 0.007 – 0.5 Å-1 in RAW 24,25. The final scattering curves were brought to absolute scale (cm-1) using a pure water sample as a secondary standard 26. 2.7 Data evaluation The measured scattering spectra were fitted to a model based on the one presented by Mears et al. 27, assuming spherical particles with a core-shell structure. In this model, the total scattering in a dilute microgel dispersion is the sum of two scattering contributions, as expressed in Eq. 2. Here, the contribution corresponds to the scattering from the interface between the microgel particle and the solvent, which accounts for the width (“fuzziness”) of the outer layer as a convoluted radial box profile with a Gaussian distribution

28

. The

contribution arises from the fluctuations of chains and interchain

interactions within the microgel network, and is described by a Lorentzian function. With these assumptions: (2) where (3) and

(4)

where

is the thickness of the surface layer of the microgel particle, or fuzziness,

electron density between the solvent and the polymer, and

represents the mesh size of the microgel

network. The overall particle dimension obtained here is approximately equivalent to is the radius of a highly cross-linked core, and

is the difference in

, where

is the thickness of the microgel shell (fuzziness, Figure

3). (While SAXS spectra can be straightforwardly analyzed also in terms of particle inhomogeneity 29,30, both BCA results and zeta potential measurements provide strong support to peptide encapsulation into the particle interior. Hence, we believe that the applied core-shell model describes the investigated systems more correctly.)

8

9

Figure 3. Microgel core-shell model with its structural parameters: thickness of the shell,

is the radius of the high-density core, and

is the mesh size of the network,

is the

is the radius of gyration.

The fitting analysis was performed by creating customized equations using MATLAB release version 2017b (The MathWorks, Inc., Natick, MA, USA). The

contribution was fitted in the -range from

to

. This range was selected, considering that at sufficiently large q-values, the scattering is dominated by the Lorentzian term, which was verified by the linearity of plotting Subsequently, the

as a function of

. 31

contribution was subtracted from the total scattering intensity to obtain

.

2.8 Encapsulation efficiency The encapsulation efficiency (EE%) of the microgels was estimated by measuring the amount of free peptide in solution, combining this with the known amount of peptide introduced into the closed system during particle formation. This amount was determined using the Bicinchonic acid (BCA) assay

32

(G-

Biosciences, Overland, MO, USA), following the instructions of the manufacturer. In short, a working solution containing ten parts of a BCA solution (bicinchonic acid solution in 0.1 N NaOH) and one part of a copper solution [4 % (w/v) CuSO4.5H2O] was prepared beforehand. Aliquots of the sample (25 µL) were added to 200 µL of this solution, placed in a multi-well absorbance plate, and the absorbance at 570 nm was measured using a FLUOstar Optima plate reader (BMG Labtech, Ortenberg, Germany). The peptide encapsulation efficiency was calculated using Eq. 5 under the assumption of negligible material loss due to adsorption to tubes and micromixer walls: (5)

10

2.9 Release studies The release of polymyxin B from surface-immobilized microgel particles was investigated by null ellipsometry using an Optrel Multiskop (Optrel, Kleinmachnow, Germany) at 532 nm at an angle of incidence of 67.66°. The total adsorbed amount (Γ) was calculated using a refractive index increment of 0.154 cm3/g, as described previously 33. Cationic poly-L-lysine (150 kDa, Sigma Aldrich, St. Louis, MO, USA) was pre-adsorbed from water to negatively charged (-40 mV)34 silica substrates, oxidized and cleaned as described in detail previously 35, prior to microgel addition. This adsorption step, which resulted in a poly-Llysine adsorption of 0.045 ± 0.010 mg/m2, was followed by removal of non-adsorbed poly-L-lysine by rinsing with Tris (10 mM, pH 7.4) at 5 mL/min for 15 min. Subsequently, the negatively charged microgel particles were added at a concentration of 70 ppm, and the adsorption was monitored over 1 h, followed by rinsing with Tris buffer (10 mM, pH 7.4) for 30 min, to monitor peptide release and the resulting material loss at the interface. Particle disintegration and peptide release were further promoted by rinsing with Tris buffer (10 mM, pH 7.4, supplemented with 150 mM NaCl) for additional 30 min. The measurements were conducted in duplicate. 2.10 Statistics Statistical calculations were performed by using MATLAB and Statistics Toolbox release version 2017b (The MathWorks) by a one-way analysis of variance at a 0.05 significance level, followed by means comparison by applying the Fisher’s test.

3. Results 3.1 The micromixer flow pattern determines the microgel size Fluid dynamic simulations were used to model the flow patterns in the micromixers. Different mixing behaviors were observed, depending on the specific geometric design. In order to understand how the micromixer geometry affects the properties of the peptide-loaded microgels, we present simulation results together with information on the average hydrodynamic size and PDI of the resulting microgels, as measured by DLS (Figure 4). For the TFM, there was a chaotic disruption of the flow layers, and particle formation was achieved by turbulent mixing (Figure 4A), as previously reported for similar inlet flows using a comparable device 11. For a fixed system composition of 200 ppm alginate, 48 M polymyxin B, and 600 M CaCl2 in 10 mM Tris, pH 7.4, the microgels prepared with this micromixer displayed a hydrodynamic radius (

) of 132 ± 6 nm (Figure 4D), which is well in line with the size of microgels prepared by emulsion

polymerization 36. In addition, this size is sufficiently small to allow for their use in a wide range of drug delivery applications

37,38

. Significantly larger particles (p<0.001) were obtained when using the LFM (

=

11

179 ± 17 nm, Figure 4D), for which the simulations predicted no apparent disruption of the flow layers inside this type of micromixer (Figure 4B). This suggests that (i) only modest convective mixing is taking place inside the LFM, and (ii) mixing and particle formation occurs via interdiffusion. As evident from the larger particle size, diffusive mixing is less efficient in particle formation. For the third type of micromixer (the ICM), fluid dynamics simulations showed the formation of a microvortex at the end of the chamber (Figure 4C.1), where the latter streams seem to span into the mixture (Figure 4C.2). As for the LFM, flow layer disruption was not observed, presumably because the sheat flows were introduced parallel to the central flow, which was not the case for TFM. Nevertheless, microgel particles prepared with the ICM displayed the smallest particle size (

= 98 ± 8 nm, Figure 4D; p<0.001), demonstrating the importance of

microvortex formation for the generation of smaller-sized particles. For all systems, PDIs < 0.25 were obtained (Figure 4D), which is a result of confinement of flows in the center of the hydrodynamic focusing systems 20, as also verified in the simulations.

Figure 4. Effect of micromixer flow pattern on the microgel size. Mixing patterns modelled by the fluid dynamic simulations for the TFM (A), LFM (B) and ICM (C), respectively. The red and blue lines represent the streams from the central and lateral inlets, respectively, corresponding to polymyxin B and alginate; the green lines represent the Ca2+

12

inlets. (D)

(bars) and PDI (points) of microgels prepared at 200 ppm alginate, 48 µM polymyxin B, and 600 µM

CaCl2, using the TFM (red), LFM (blue), and ICM (grey), respectively. Data points represent mean values ± SD (n =3).

3.2 Composition effects on particle formation Subsequently, the effects of sheath solution composition on the size and zeta potential of the resulting microgels were investigated. The results indicate that irrespective of the micromixer design, the hydrodynamic size of the resulting microgels increased with the peptide concentration in the sheath solution (Figure 5A). In contrast, the zeta potential was largely unaffected by the peptide concentration (0.001 < p < 0.01), and remained distinctly negative in all cases (Figure 5B). Since polymyxin B is positively charged at pH 7.4

39,40

, this suggests that the peptide is incorporated in the interior of the microgel

structure, and that negatively charged alginate tails determine the zeta potential, irrespectively of the micromixer used. Furthermore, the microgel particles prepared at 60 µM CaCl2 were larger than those prepared at 600 µM (0.001 < p <0.1), demonstrating that a higher degree of cross-linking results in particle contraction, as commonly observed for microgels prepared by emulsion polymerization and other conventional methods4. While increasing the CaCl2 concentration reduced the negative zeta potential for all investigated systems, it remained distinctly negative, demonstrating that Ca 2+ binding to alginate is insufficient to cause charge reversal. It should also be noted that microgels prepared using the ICM at 60 µM CaCl2 displayed particle sizes smaller than those prepared at the highest CaCl 2 concentration using the other types of micromixers.

13

Figure 5. Hydrodynamic radius (A) and zeta potential (B) of microgels prepared from 200 ppm alginate using different micromixers at 60 (grey) and 600 (black) M CaCl2 at different concentrations of polymyxin B. Data points represent mean values ± SD (n =3).

3.3 The microgel structure The structure of the microgel particles was next investigated in further detail by SAXS measurements. First, the scattering data were fitted to Eq. 4 to obtain information on the mesh size of the microgel network. The slope of the scattering curves changed gradually according to the composition of the microgel, notably increasing with the concentration of polymyxin B and CaCl 2 (Figure 6A and B). Furthermore, for the dataset where the polymyxin B concentration was varied, the curves overlap as they approach the region where the fitting was performed ( =

). The fitting results (Figure 6C and D) revealed that microgels

prepared at a constant CaCl2 concentration, but different polymyxin B concentrations, displayed comparable mesh sizes. On the other hand, when keeping the polymyxin B concentration constant, the mesh size decreased significantly with increasing CaCl 2 concentration (0.001 < p < 0.01). The mesh size values did not seem to depend on the specific type of micromixer employed for particle formation. For all micromixers used, it can be noted that these mesh sizes are smaller than pore size values previously reported for alginate microgels prepared by a microfluidic approach involving a hydrodynamic flow focusing platform of cross-junction microchannels 41.

14

Figure 6. SAXS results on microgel structure. Representative scattering measurement curves are shown at different concentrations of polymyxin B (A) and CaCl2 (B) in 10 mM Tris, pH 7.4, at a constant alginate concentration (200 ppm). Also shown are results on mesh size for samples prepared using different types of micromixers, either varying the polymyxin B concentration at a fixed CaCl2 concentration of 600 µM (C), or varying the CaCl2 concentration at a fixed polymyxin B concentration of 48 µM (D). Data points represent mean values ± SD (n =3).

3.4 Laminar mixing results in fuzzier microgels SAXS data were also used to determine the radial profile of the microgels after fitting the data to Eq. 3. The fitting was performed by approximating the starting itineration values of Rg with experimental Rh results obtained with DLS. For samples prepared by using the TFM and ICM, the extracted values of fuzziness ( were found to be insignificant (

)

; Figure S2, Supplementary Material), suggesting that these microgels

display sharp interfaces. For samples prepared by the LFM, the results indicated a somewhat different structure. In this case, the thickness of the microgel shell decreased with increasing CaCl2 concentration, approaching zero at high CaCl2 concentrations (Figure 7). This illustrates the condensing effect of the crosslinking Ca2+ ions, as well as electrostatic screening, in line with the overall size decrease observed at high CaCl2 concentrations.

15

Figure 7. Box profile parameters of microgels prepared from 200 ppm alginate and 48 M polymyxin B, using the LFM, as a function of the applied CaCl2 concentration. Diamonds: Radius of gyration. Squares: Radius of the core. Triangles: fuzziness (shell width, 2). Data points represent mean values ± SD (n =3).

3.5 High encapsulation efficiency of polymyxin B Having characterized the structural features of the microgels, we next investigated the encapsulation efficiency of polymyxin B in the microgels. The results indicate a high degree of encapsulation (>80%) for all systems investigated (Figure 8). This can be attributed to the high binding affinity of alginate for cationic molecules through attractive electrostatic interactions 42, high affinity complexation between alginate and Ca2+

41

, attractive interactions between the hydrophobic tails of polymycin B, and the uniformity of

interactions taking place in hydrodynamic focusing systems 43. While the microgels prepared using all three micromixers showed comparable encapsulations efficiencies, it was found that particles prepared at the highest CaCl2 concentration displayed a significantly lower degree of encapsulation (0.0001 < p < 0.1) than the gels prepared at a ten-fold lower CaCl2 concentration for all micromixers, possibly due to competition for binding sites between calcium ions and the peptide. 39

16

Figure 8. Encapsulation efficiency of microgels prepared at constant alginate and polymyxin B concentrations of 200 ppm and 48 M in 10 mM Tris, pH 7.4, as estimated by using the BCA assay. Data points represent mean values ± SD (n =3).

3.6 Peptide release To study peptide release kinetics, microgel particles prepared at 48 M polymyxin B and 600 M CaCl2 were immobilized onto cationic poly-L-lysine-modified silica surfaces, and release monitored by ellipsometry. The results showed that within the time-frame investigated, peptide release was only partial (20-30 %) in 10 mM Tris buffer (Figure 9A and B). When increasing the ionic strength to 150 mM NaCl, however, peptide release was complete. Notably, the desorption kinetics displayed a leveling off during rising with Tris buffer, suggesting the existence of two fractions of polymyxin B molecules, presumably residing in the outer and the interior regions of the particles, respectively. When increasing the ambient solution to Tris NaCl buffer, i.e., physiological ionic strength, complete peptide release was observed for all micromixers. The latter was also indicated by the very low degree of encapsulation after exposure to Tris NaCl buffer for 1 h (Figure 9C), which is also in line with the findings from DLS of particle disruption (Figure S3, Supplementary Material).

17

Figure 9. (A) Representative kinetic curve, measured by ellipsometry, showing the release of polymyxin B from surface-immobilized microgels (200 ppm alginate, 48 M polymyxin B, and 600 M CaCl2, using the ICM) in 10 mM Tris buffer, pH 7.4 (i), and (ii) 10 mM Tris buffer supplemented with 150 mM NaCl, pH 7.4. (B) Peptide release after rinsing with either Tris buffer (grey) or Tris buffer supplemented with 150 mM NaCl (black) at 5 mL/min for 30 min. (C) Encapsulation efficiency before (grey) and after (black) exposure to 150 mM NaCl for 1 h, as estimated by using the BCA assay. Data points represent mean values ± SD (n =3).

4. Discussion In the present study, microgels were prepared through electrostatic complexation between negatively charged alginate chains and cationic polymyxin B, facilitated by hydrophobic interactions between the aliphatic tails of the latter, as well as by Ca2+-induced cross-linking, in a production setup with 3D-printed micromixers for continuous manufacturing. Here, it should be noted that microfluidic methods have been previously investigated for the preparation of highly monodisperse microgels

21,41,44

. From such studies, it

has been found that the properties of microgels are highly dependent on flow parameters (e.g., flow rates and ratios) of the mixing device employed for preparation 45. While numerous studies report how the flow behavior depends on the operating conditions

11,46

, less information is available on how the micromixer

design affects the latter. Hence, we employed 3D-printed micromixers of three different geometric designs and investigated the properties of the resulting peptide-loaded microgels. Results obtained using a combination of scattering techniques revealed that the hydrodynamic size and compactness of the microgels depend on the type of flow generated in the micromixer. Hence, the smallest particles were obtained when mixing involved microvortex formation and high shear flows, whereas larger and more diffuse microgels resulted from laminar mixing preparation. This is in agreement with previous studies,

18

which highlighted the importance of mixing speed and vortex formation to obtain small-sized nanoparticles 46

.

The microfluidic approach and the use of simple 3D printed micromixers adopted in this study permitted incorporation of the peptide in the interior of the microgel, as demonstrated by the negative zeta potential, also at high peptide load, despite a substantial increase in particle size with increasing peptide loading. The observed effect on the zeta potential is in agreement with previous results

47

, showing that the zeta

potential of anionic poly(ethyl acrylate-co-methacrylic acid) microgels after addition of variants of the cationic peptide EFK17 (EFKRIVQRIKDFLRNLV) remained strongly negative, even at high peptide-to-microgel ratios. Analogous effects were observed by Nordström et al. for such microgels after incorporation of the net

cationic

peptides

LL-37

(GKHKNKGKKNGKHNGWKWWW)

(LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES)

and

DPK-060

48

. However, in contrast to observations in the present investigation,

incorporation of net positively charged peptides in pre-formed anionic microgels resulted in osmotic deswelling and reduced particle size, depending on peptide loading and microgel charge. The latter findings are also in line with findings from several other studies on the effect of incorporating peptides and proteins into oppositely charged microgels 32,49. However, in contrast to these previous studies, polymyxin B is not post-loaded after particle formation in the present study, but plays a key role in the particle formation process, as suggested by the observed effects of the peptide concentration on the particle size (Figure 5A). Through the close contact between alginate and polymyxin B during the particle formation process, dense particles are obtained, as demonstrated by the small particles and small mesh sizes observed. The condensing effects were further emphasized by cross-linking of alginate by Ca2+ ions. As a result of such dense particle formation, the particle size increased as the peptide loading increases for excluded volume reasons. Analogous effects of peptide concentration were previously observed by Klodzinska et al.45, who reported an increase in size of hyaluronic-based microgels upon incorporation of a lysine-based αpeptide/β-peptoid peptidomimetic at higher peptide-polymer ratios. Cross-linking of microgels was effectively induced by complexation between alginate and Ca 2+, resulting in an increased compactness of the microgel. This is in accordance with previous findings on the effect of Ca2+ concentration on, e.g., structure50 and viscosity complexes

41

51

, showing that alginate-Ca2+ mixtures form dense

, effectively resulting in increased “cross-linking” density and network contraction with

increasing Ca2+ concentration. Reporting on such effects for alginate nanogels, Bazban-Shotorbani et al. varied the concentration of alginate while keeping the Ca2+ concentration constant, and found smaller particles sizes at higher Ca2+/alginate ratios, which is in line with the findings of the present study for alginate/polymyxin B/CaCl2 41. Thus, increasing the concentration of Ca2+ ions during microgel formation

19

results in a microgel network with higher connectivity. However, composition is not the only determinant for particle size. In fact, microvortex mixing had a more pronounced condensing effect on microgel size than the Ca2+ ions, as particles prepared at 60 µM and ICM were significantly smaller than those prepared at 600 µM using the other micromixers (0.001 < p < 0.01). Mesh sizes derived from SAXS measurements were found to mirror the results on particle size, thus being smaller at higher Ca2+ concentration. This effect is not unexpected, as it has previously been reported that the pore size of microgel networks is proportional to the particle size

52,53

. Quantitatively, however, the

derived mesh size values were 1-1.5 nm, considerably smaller than those previously reported for alginate microgels prepared by using a different microfluidic approach (12-26 nm)41. However, it should be noted that the latter microgels were prepared in the absence of any cationic solute. From electrostatic considerations, the presence of cationic polymyxin B in the currently investigated systems is expected to result in particle compaction. This is in line with findings of previous studies, in which alginate was complexed with polycations, e.g., chitosan

54,55

. In addition to attractive electrostatic interactions,

polymyxin B also contains a hydrophobic acyl chain, which drives the self-assembly of this lipopeptide in solution via hydrophobic interactions56. Within the alginate microgels, polymyxin B localizes primarily in the microgel interior in order to maximize such hydrophobic interactions, hence providing an additional driving force for the formation of dense particles with small mess sizes. The polymyxin B encapsulation efficiency for all microgels was higher than 80%. This is in line with previous findings for peptide-loaded microgels, prepared via hydrodynamic focusing mixing, reporting high drug encapsulation efficiency (88%) of a lysine-based α-peptide/β-peptoid peptidomimetic in hyaluronic-based microgels

45

. Together, this highlights the importance of this approach to achieve a desirable drug

encapsulation. The ellipsometry results indicate that release is strongly accelerated with increasing ionic strength, which was supported also by findings of the peptide encapsulation being essentially zero after exposure for 1 h in Tris-NaCl buffer, and of particle size reduction. This is in agreement with studies where alginate microgels were reported to give rise to a burst drug release in the presence of other cations in solution

57,58

. It should also be noted that while the microgels are efficiently disrupted to release all

incorporated polymyxin B at high ionic strength, light scattering results indicate that remaining alginate fractions, after peptide release, are larger than the individual alginate chains, presumably consisting of alginate chains cross-linked by Ca2+.

5. Conclusions

20

Peptide-loaded microgels can be readily prepared by complexation of polymyxin B with alginate, combined with Ca2+-induced cross-linking, using continuous microfluidic particle generation based on 3D-printed micromixers. As such, the latter offer an interesting alternative to more complex mold-based microfluidics set-ups. Although small and relatively uniform particles were obtained for widely different micromixer designs, mixing involving microvortex flow resulted in the smallest microgels (≈100 nm), whereas laminar (diffusion-based) mixing resulted in larger and less compact microgels. Polymyxin B localized in the particle core and caused particle growth with increasing peptide loading. On the other hand, Ca2+- induced crosslinking of alginate results in particle contraction. Polymyxin B encapsulation was found to be high (>80%), irrespective of the applied micromixer design in the investigated composition ranges, and peptide release was strongly accelerated at physiological ionic strength. Taken together, these findings extend on previous studies of microgels prepared by emulsion polymerization as carriers of antimicrobial peptides 36, as well as of microfluidic generation of polyelectrolyte complex particles

11

, by demonstrating that microfluidic

approaches based on simple 3D-printed microgels provide good opportunities for continuous generation of peptide-containing microgels, for which small particles and high degrees of peptide encapsulation can be achieved over wide composition ranges and for various microfluidic designs. Considering the latter, the approach taken thus allows considerations of the refined control over particle structure and composition, theoretically offered by multi-compartment set-ups, to be balanced against considerations of mixer simplicity, depending, e.g., on the relative importance of particle structure and process scalability in specific applications.

Acknowledgements This research was funded by NordForsk (Nordic University Hub project #85352 -Nordic POP, Patient Oriented Products) (BCB), the Novo Nordisk Foundation - Denmark - Interdisciplinary Synergy program SYNERGY (grant number NNF15OC0016670) (SB); and the Leo Foundation Center For Cutaneous Drug Delivery (Grant number 2016-11-01) (MM).

21

References 1.

Saunders, B. R. & Vincent, B. Microgel particles as model colloids: Theory, properties and applications. Adv. Colloid Interface Sci. 80, 1–25 (1999).

2.

Peppas, N. A., Bures, P., Leobandung, W. & Ichikawa, H. Hydrogels in pharmaceutical formulations. Eur. J. Pharm. Biopharm. 50, 27–46 (2000).

3.

Malmsten, M., Bysell, H. & Hansson, P. Biomacromolecules in microgels - Opportunities and challenges for drug delivery. Curr. Opin. Colloid Interface Sci. 15, 435–444 (2010).

4.

Pelton, R. & Hoare, T. Microgels and their synthesis: An introduction. in Microgel Suspensions: Fundamentals and Applications 1–32 (2011).

5.

Colombo, S. et al. Transforming nanomedicine manufacturing toward Quality by Design and microfluidics. Adv. Drug Deliv. Rev. 128, 115–131 (2018).

6.

Paliwal, R., Babu, R. J. & Palakurthi, S. Nanomedicine scale-up technologies: Feasibilities and challenges. AAPS PharmSciTech 15, 1527–1534 (2014).

7.

Utech, S. et al. Microfluidic generation of monodisperse, structurally homogeneous alginate microgels for cell encapsulation and 3D cell culture. Adv. Healthc. Mater. 4, 1628–1633 (2015).

8.

Seiffert, S. & Weitz, D. A. Controlled fabrication of polymer microgels by polymer-analogous gelation in droplet microfluidics. Soft Matter 6, 3184–3190 (2010).

9.

Volpatti, L. R. & Yetisen, A. K. Commercialization of microfluidic devices. Trends Biotechnol. 32, 347– 350 (2014).

10.

Holtze, C. Large-scale droplet production in microfluidic devices - An industrial perspective. J. Phys. D. Appl. Phys. 46, (2013).

11.

Bohr, A. et al. High-throughput fabrication of nanocomplexes using 3D-printed micromixers. J. Pharm. Sci. 106, 835–842 (2017).

12.

Srai, J. S., Badman, C., Krumme, M., Futran, M. & Johnston, C. Future supply chains enabled by continuous processing-opportunities and challenges May 20-21, 2014 continuous manufacturing symposium. Journal of Pharmaceutical Sciences 104, 840–849 (2015).

13.

Pasupuleti, M., Schmidtchen, A. & Malmsten, M. Antimicrobial peptides: Key components of the innate immune system. Critical Reviews in Biotechnology 32, 143–171 (2012).

22

14.

Brogden, K. A. Antimicrobial peptides: Pore formers or metabolic inhibitors in bacteria? Nature Reviews Microbiology 3, 238–250 (2005).

15.

Hancock, R. E. W. & Sahl, H. G. Antimicrobial and host-defense peptides as new anti-infective therapeutic strategies. Nature Biotechnology 24, 1551–1557 (2006).

16.

Nordström, R. & Malmsten, M. Delivery systems for antimicrobial peptides. Advances in Colloid and Interface Science 242, 17–34 (2017).

17.

Tiwari, A. K., Gajbhiye, V., Sharma, R. & Jain, N. K. Carrier mediated protein and peptide stabilization. Drug Delivery 17, 605–616 (2010).

18.

Veronese, F. M. & Mero, A. The impact of PEGylation on biological therapies. BioDrugs 22, 315–329 (2008).

19.

Silva, J. P. et al. Delivery of LLKKK18 loaded into self-assembling hyaluronic acid nanogel for tuberculosis treatment. J. Control. Release 235, 112–124 (2016).

20.

Lu, M. et al. Microfluidic hydrodynamic focusing for synthesis of nanomaterials. Nano Today 11, 778–792 (2016).

21.

Water, J. J. et al. Hyaluronic acid-based nanogels produced by microfluidics-facilitated self-assembly improves the safety profile of the cationic host defense peptide novicidin. Pharm. Res. 32, 2727– 2735 (2015).

22.

Storm, D. R., Rosenthal, K. S. & Swanson, P. E. Polymyxin and related peptide antibiotics. Annu Rev Biochem 46, 723–763 (1977).

23.

Paques, J. P., Van Der Linden, E., Van Rijn, C. J. M. & Sagis, L. M. C. Preparation methods of alginate nanoparticles. Adv. Colloid Interface Sci. 209, 163–171 (2014).

24.

Nielsen, S. S. et al. BioXTAS RAW, a software program for high-throughput automated small-angle Xray scattering data reduction and preliminary analysis. J. Appl. Crystallogr. 42, 959–964 (2009).

25.

Hopkins, J. B., Gillilan, R. E. & Skou, S. BioXTAS RAW: Improvements to a free open-source program for small-angle X-ray scattering data reduction and analysis. J. Appl. Crystallogr. 50, 1545–1553 (2017).

26.

Orthaber, D., Bergmann, A. & Glatter, O. SAXS experiments on absolute scale with Kratky systems using water as a secondary standard. J. Appl. Crystallogr. 33, 218–225 (2000).

23

27.

Mears, S. J., Deng, Y., Cosgrove, T. & Pelton, R. Structure of sodium dodecyl sulfate bound to a poly(NIPAM) microgel particle. Langmuir 13, 1901–1906 (1997).

28.

Stieger, M., Richtering, W., Pedersen, J. S. & Lindner, P. Small-angle neutron scattering study of structural changes in temperature sensitive microgel colloids. J. Chem. Phys. 120, 6197–6206 (2004).

29.

Matsui, S. et al. Small-angle X‑ ray scattering study on internal microscopic structures of poly(N‑ isopropylacrylamide-co-tris(2,2′- bipyridyl))ruthenium(II) complex microgels. Langmuir 31, 7228–7237 (2015).

30.

Seiffert, S. Origin of nanostructural inhomogeneity in polymer-network gels. Polymer Chemistry 8, 4472–4487 (2017).

31.

Saunders, B. R. On the structure of poly(N-isopropylacrylamide) microgel particles. Langmuir 20, 3925–3932 (2004).

32.

Bysell, H., Hansson, P., Schmidtchen, A. & Malmsten, M. Effect of hydrophobicity on the interaction between antimicrobial peptides and poly(acrylic acid) microgels. J. Phys. Chem. B 114, 1307–1313 (2010).

33.

Schmidtchen, A. et al. Membrane selectivity by W-tagging of antimicrobial peptides. Biochim. Biophys. Acta - Biomembr. 1808, 1081–1091 (2011).

34.

Malmsten, M., Burns, N. & Veide, A. Electrostatic and hydrophobic effects of oligopeptide insertions on protein adsorption. J. Colloid Interface Sci. 204, 104–111 (1998).

35.

Wong, A. K. Y. & Krull, U. J. Surface characterization of 3-glycidoxypropyltrimethoxysilane films on silicon-based substrates. Anal. Bioanal. Chem. 383, 187–200 (2005).

36.

Nordström, R. et al. Membrane interactions of microgels as carriers of antimicrobial peptides. J. Colloid Interface Sci. 513, 141–150 (2018).

37.

Zhou, T., Wu, W. & Zhou, S. Engineering oligo(ethylene glycol)-based thermosensitive microgels for drug delivery applications. Polymer (Guildf). 51, 3926–3933 (2010).

38.

Bysell, H., Månsson, R., Hansson, P. & Malmsten, M. Microgels and microcapsules in peptide and protein drug delivery. Adv. Drug Deliv. Rev. 63, 1172–1185 (2011).

39.

Gounani, Z., Asadollahi, M. A., Meyer, R. L. & Arpanaei, A. Loading of polymyxin B onto anionic mesoporous silica nanoparticles retains antibacterial activity and enhances biocompatibility. Int. J.

24

Pharm. 537, 148–161 (2018). 40.

Domingues, M. M. et al. Biophysical characterization of polymyxin B interaction with LPS aggregates and membrane model systems. Biopolymers 98, 338–344 (2012).

41.

Bazban-Shotorbani, S., Dashtimoghadam, E., Karkhaneh, A., Hasani-Sadrabadi, M. M. & Jacob, K. I. Microfluidic directed synthesis of alginate nanogels with tunable pore size for efficient protein delivery. Langmuir 32, 4996–5003 (2016).

42.

Li, Y. et al. pH sensitive Laponite/alginate hybrid hydrogels: Swelling behaviour and release mechanism. Soft Matter 7, 6231–6238 (2011).

43.

Lu, M. et al. A three-dimensional hydrodynamic focusing method for polyplex synthesis. ACS Nano 8, 332–339 (2014).

44.

Sadat Majedi, F. et al. Microfluidic assisted self-assembly of chitosan based nanoparticles as drug delivery agents. Lab Chip 13, 204–207 (2013).

45.

Kłodzińska, S. N. et al. Biopolymer nanogels improve antibacterial activity and safety profile of a novel lysine-based α-peptide/β-peptoid peptidomimetic. Eur. J. Pharm. Biopharm. 128, 1–9 (2018).

46.

Kim, Y. et al. Mass production and size control of lipid-polymer hybrid nanoparticles through controlled microvortices. Nano Lett. 12, 3587–3591 (2012).

47.

Singh, S. et al. Conformational aspects of high content packing of antimicrobial peptides in polymer microgels. ACS Appl. Mater. Interfaces 9, 40094–40106 (2017).

48.

Nordström, R. et al. Membrane interactions of microgels as carriers of antimicrobial peptides. J. Colloid Interface Sci. 513, 141–150 (2018).

49.

Bysell, H., Schmidtchen, A. & Malmsten, M. Binding and release of consensus peptides by poly(acrylic acid) microgels. Biomacromolecules 10, 2162–2168 (2009).

50.

Sarker, M., Izadifar, M., Schreyer, D. & Chen, X. Influence of ionic crosslinkers (Ca2+/Ba2+/Zn2+) on the mechanical and biological properties of 3D bioplotted hydrogel scaffolds. J. Biomater. Sci. Polym. Ed. 29, 1126–1154 (2018).

51.

Amici, E., Tetradis-Meris, G., de Torres, C. P. & Jousse, F. Alginate gelation in microfluidic channels. Food Hydrocoll. 22, 97–104 (2008).

52.

Sierra-Martin, B., Lietor-Santos, J. J., Fernandez-Barbero, A., Nguyen, T. T. & Fernandez-Nieves, A.

25

Swelling thermodynamics of microgel particles. in Microgel Suspensions: Fundamentals and Applications 71–116 (2011). 53.

Fernández-Barbero, A., Fernández-Nieves, A., Grillo, I. & López-Cabarcos, E. Structural modifications in the swelling of inhomogeneous microgels by light and neutron scattering. Phys. Rev. E - Stat. Physics, Plasmas, Fluids, Relat. Interdiscip. Top. 66, 10 (2002).

54.

Sarmento, B. et al. Alginate/chitosan nanoparticles are effective for oral insulin delivery. Pharm. Res. 24, 2198–2206 (2007).

55.

Patil, J. S., Kamalapur, M. V, Marapur, S. C. & Kadam, D. V. Ionotropic gelation and polyelectrolyte complexation: The novel techniques to design hydrogel particulate sustained, modulated drug delivery system: A review. Digest Journal of Nanomaterials and Biostructures 5, 241–248 (2010).

56.

Brandenburg, K., Rubinstein, I., Sadikot, R. & Önyüksel, H. Polymyxin B self-associated with phospholipid nanomicelles. Pharm. Dev. Technol. 17, 654–660 (2012).

57.

Lee, K. Y., Bouhadir, K. H. & Mooney, D. J. Controlled degradation of hydrogels using multi-functional cross-linking molecules. Biomaterials 25, 2461–2466 (2004).

58.

Khdair, A. et al. Nanoparticle-mediated combination chemotherapy and photodynamic therapy overcomes tumor drug resistance. J. Control. Release 141, 137–144 (2010).

26

Graphical abstract

27