A Microfluidic Co‐Flow Route for Human Serum Albumin‐Drug–Nanoparticle Assembly

Abstract Nanoparticles are widely studied as carrier vehicles in biological systems because their size readily allows access through cellular membranes. Moreover, they have the potential to carry cargo molecules and as such, these factors make them especially attractive for intravenous drug delivery purposes. Interest in protein‐based nanoparticles has recently gained attraction due to particle biocompatibility and lack of toxicity. However, the production of homogeneous protein nanoparticles with high encapsulation efficiencies, without the need for additional cross‐linking or further engineering of the molecule, remains challenging. Herein, we present a microfluidic 3D co‐flow device to generate human serum albumin/celastrol nanoparticles by co‐flowing an aqueous protein solution with celastrol in ethanol. This microscale co‐flow method resulted in the formation of nanoparticles with a homogeneous size distribution and an average size, which could be tuned from ≈100 nm to 1 μm by modulating the flow rates used. We show that the high stability of the particles stems from the covalent cross‐linking of the naturally present cysteine residues within the particles formed during the assembly step. By choosing optimal flow rates during synthesis an encapsulation efficiency of 75±24 % was achieved. Finally, we show that this approach achieves significantly enhanced solubility of celastrol in the aqueous phase and, crucially, reduced cellular toxicity.

Abstract: Nanoparticlesa re widely studied as carrier vehicles in biological systems because their size readily allows access throughc ellularm embranes.Moreover, they have the potential to carry cargo molecules and as such, these factors make them especially attractive for intravenous drug delivery purposes.Interest in protein-based nanoparticles has recently gained attraction due to particleb iocompatibility and lack of toxicity.H owever,t he production of homogeneous protein nanoparticles with high encapsulatione fficiencies, withoutt he need for additional crosslinking or furthere ngineeringo ft he molecule, remains challenging.Herein, we present am icrofluidic3 Dc o-flow device to generate human serum albumin/celastrol nanoparticles by co-flowinga na queous protein solution with celastrol in ethanol.This microscale co-flow methodr esulted in the formationo fn anoparticles with ah omogeneous size distributiona nd an average size, which could be tuned from % 100 nm to 1 mmb ym odulating the flow rates used.We show that the high stabilityo ft he particles stems from the covalentc ross-linking of the naturally presentc ysteiner esidues within the particles formed during the assembly step.By choosingo ptimal flow rates during synthesis an encapsulatione fficiency of 75 AE 24 % was achieved.F inally,w es how that this approach achieves significantly enhanced solubility of celastrol in the aqueous phasea nd, crucially,r educed cellular toxicity.
Ta rgeted delivery and controllable release of activep harmaceuticals are major objectives to improve the safety and efficacy of potential drugs.These important properties can be enhanced by using suitable drug carriers, such as nanoparticles.Nanoparticles are an attractive class of carrieri nt his context becauset hey can solubilize therapeutic cargo, whichc an prolong the circulation lifetimes of drugsand ability to extravasate to tumour sites. [1]These therapeutic cargo carriers need to be very biocompatible to decrease the risk of unwanted complications.T hus, protein nanoparticles, which intrinsically have minimal immunogenicity and biocompatibility,h ave attracted al ot of interest. [2]An additional benefit of using proteinsf or nanoparticlef ormation is that they can be selectively modified with specific ligands for targetingp urposes. [3]Previously, proteins have been appliedt oi ncrease stability in microdroplets [4] and form protein-based nanoparticles from bovine serum albumin (BSA), [5] human serum albumin (HSA) [6] and b-lactoglobulin. [7]nterestingly,a lbumin nanoparticles have been shownt op enetrate throught he blood-brain barrier, [8] which broadens the areas where therapeutica gents could be delivered.
There are aw ide variety of methods available for nanoparticle formation including nano emulsification and spray drying. [9]owever, these methods requirec hemical cross-linking or proteins to form fibrillarn etworks.Another,p opularm ethod for nanoparticle formation is desolvation, in which ap rotein or polymer is in aqueous media and ad esolvating agent,s uch as ethanol, is added drop by drop. [10]In this methodt he introduction of the desolvatinga gent to the protein solution, de-stabilises the protein structure and exposes itsb uried hydrophobic and reactive residues.T his promotes interactions between protein molecules, so that the proteinsc lump together in small aggregate nanoparticles. [11]Recently,amethod to introduce intermolecular disulfide bonds betweenH SA molecules to avoid the use of toxic crosslinkersw as reported. [12]This approach relied on an additional denaturing step prior to nanoparticle formation,w hich reduced someo ft he HSA's disulfide bridges and further promotedc ysteine-cysteine interactions between different HSA molecules during nanoparticle formation.
Herein, we presentamicrofluidicco-flows trategy for nanoparticle formation without the need for additional cross-linkers.This co-flow method is basedo nt he desolvation method, but insteado fd ropwise addition, the desolvating agent flows adjacent to the protein solution in am icrofluidic chip.In this approach, mixing is much faster with no graduali ncreaseo ft he desolvating agent concentration.Furthermore, we encapsulated ah ighly lipophilicd rug, celastrol, into HSA nanoparticles.
Celastrol is an atural compound extracted from herb Tripterygium wilfordii,adirect modulator of progesteronea nd cannabinoid receptors [13] that elicits ap otent anti-inflammatory response [14] and shows promise as at reatment for Alzheimer's disease, [14] obesity, [15] rheumatoid arthritis [16] and cancer. [17]Due to its hydrophobicity,celastrol is difficult to use in aqueous solutions.T herefore, it would be as ignificant advance if this molecule could be encapsulated into am ore hydrophilic shell.With our approach, we show encapsulation of celastrol with high efficiency within HSAnanoparticles, which increases celastrol's solubility and simultaneously dramatically decreases its cytotoxicity.
To showcase the variety of this co-flow method to produce nanoparticles, we initially investigated the formation of BSA nanoparticles by using ethanola st he desolvating agent.B SA was introduced from the middle inlet and surrounded by ethanol (Figure 1a and 1b).Additionally,aw ater stream was added as the outer layer.T his water layer has two purposes:f irst, the water pinches the ethanola nd protein stream so that there will be fast diffusion between the streams.Secondly, the water layer will ensure that the protein does not come in contact with the hydrophobic polydimethylsiloxane (PDMS) walls to avoid surface adherence.To completely envelop the protein and ethanol streamsw ith water a3 Dc o-flow design wasu sed (Figure 1c).In ac onventional 2D microfluidic chip the fluid streamsf low adjacent to one another,w hereas in 3D devices one stream flows within the another (Figure 1c).
Our 3D co-flow device was appliedt os tudy BSA nanoparticle formation with two different concentrations and six different flow rates.Figure 2s hows the size distributions measured with dynamic light scattering( DLS;F igures 2a and 2d)a nd averaged iameters (Figures 2b and 2e)f rom BSAn anoparticles prepared from 1mgmL À1 (Figures 2a and 2b)a nd 10 mg mL À1 (Figures 2d and 2e)B SA solutions.T his study revealed that whilst the concentration has only al imited effect on the average size, it has al arger effect on the size distributions.Furthermore, polydispersity in the nanoparticle samples produced with a1 0mgmL À1 BSA solution was larger,s howing an increasei nr ecorded polydispersity indexes (PDI;T able S1).This is also evident from the DLS size distributions for which broader peaks are observed (Figure 2d).Furthermore, larger standard deviations were recorded for the average diameters (Figure 2e).Am ore profound effect was achieved by changing the EtOH to protein flow-rate ratio.The highert he EtOH flow rate, the highert he concentration of EtOH in the resulting mixture, and subsequently the biggert he nanoparticles formed.With 1mgmL À1 BSA solution the average size increases exponentially.H owever,i nt he case of 10 mg mL À1 BSA solution, the increase in average diameter is more linear.T his can be explained by the 10-fold increase of protein molecules as the EtOH concentrationr emains constant.T hus, the curve is shifted to the right as BSAc oncentration is increased.The morphologyo fB SA nanoparticles was examined by transmission  electron microscopy (TEM).Figure 2s hows BSA nanoparticles made from 1mgmL À1 (Figure 2c parts i and ii)a nd 10 mg mL À1 solutions (Figure2fp arts i and ii), using a1:1 flow ratio.The particles were highly spherical for both protein concentrations.However,b ased on the TEM images the polydispersity seems to be larger in the case of nanoparticles made from 10 mg mL À1 solution,w hich confirms the resultso btained from DLS.
To determinew hether the BSA nanoparticles remain stable, the zeta potential for the two different solutions was measured in phosphate-buffered saline (PBS;p H7.3)f or particles created with 1:1f low rate ratio (Figure 3a).Generally,z eta potential values that are not in the range À30 AE 30 mV are generally considered to have sufficient repulsive force to attain better physicalc olloidal stability. [18]From the data obtained, both samples have ar elatively high zeta potential( À46.1 for 1mgmL À1 and À29.2 for 10 mg mL À1 )a nd are stable in solution.To further elucidate our understanding of the nanoparticle formation,a n8 -anilinonaphthalene-1-sulfonic acid (ANS) assay was conducted.ANS binds to hydrophobic cavities found on the protein surfacea nd increases its fluorescencei ntensity upon binding. [19]Thus, the more hydrophobic residues are exposed on the sample, the higher the fluorescences ignal.In Figure 3b the fluorescence signal of a3mm BSA solution and a3mm nanoparticle based BSA solution is shown.The ANS fluorescences howedd ecreased signal with BSA nanoparticles, which indicatest hat the hydrophobic pockets are less exposed in the nanoparticles comparedt on ative BSA.This suggestst hat the interactions between hydrophobic residues are drivingp rotein nanoparticle formation.
Typically,w hen forming protein nanoparticles, the use of toxic crosslinkers is employed to stabilize the system.H owever, such systemsc an have adverse health effects and thesec rosslinked nanoparticles cannot always be used safely for in vivo deliverya pplications.H ere, the protein was reduced with glutathione (GSH) to increase the number of free cysteines and promote the formation of intermolecular disulfide bridges. [12]o prove the involvement of disulfide bridges in the nanoparticle stability,t hree different protein solutions were prepared: BSA, BSA reduced with glutathione (GSH) and BSA in which the free cysteine was blocked with ac ysteines electivec arbonylacrylic linker (CAA). [20]To tal conversion from free cysteine to blocked cysteineb yu sing aC AA linker was observed (see Figures S3 andS 4).All three solutions were used to prepare BSA nanoparticles and the stability was measured over time in PBS solution (pH 7.3).Initially,t he size distributions were quite similar,a sd etermined by DLS (Figure 3d).However,t he differences in stabilityo ver time are evident (Figure 3e).Both BSA and reduced BSA remained as nanoparticles throughout the duration of the experiment.However,a se xpectedt he nanoparticles in which the formation of disulfide bonds was blockedd isassembled within 24 h.Furthermore, nanoparticles made from native BSA showedadecreasing linear trend in average size during the stability measurement.This indicates lower stability relative to nanoparticles made from reduced BSA.
Following the characterisation of nanoparticles formed by using the co-flow,aprotein-drug particle was established,w ith potentialf or clinical application.W ec hose to use HSAi nt his study since it is the mosta bundant protein in human plasma and is therefore widely used for drug delivery. [21]Moreover, due to its structurals imilarity to BSA, it could be assumedt o work similarly to BSA in the co-flow method.Celastrol was chosen as ac argo molecule due to its potential as at herapeutic molecule. [13]However,i ts use is still limited due to its highly lipophilic nature and cytotoxicity.T hus, it would be of significant relevance if this molecule could be encapsulated into a more hydrophilic and less toxic shell.H SA/celastrol nanoparticles were created by using the previously established strategy by co-flowing HSA with celastrol in EtOH (Figure 4a).Similarly, to BSA, HSA was partly reduced prior to microfluidic co-flow to increaset he number of intermolecular disulfide bridges and furthert oi ncrease the stability of the produced nanoparticles.This time there are two contributingf actors for nanoparticle formation.In addition to the desolvating factor of EtOH to protein, now the aqueous protein solutioni sd esolvatingc elastrol due to its low solubility in water.W ew ere expecting celastrol to aggregate due to the addition of water and furtheri nteract with the exposed hydrophobic residues of HSA burying the lipophilic cargo into ap olar shell.If celastrol and protein did not interact, we would expect to see two populations of nanoparticles, whichwould likely be evidentinthe size distributions obtainedw ith DLS.Furthermore, amorphous protein and crystalline celastrol would have different morphologies.Similart o BSA-based nanoparticles, the EtOH to protein solution ratio was investigated for HSA/celastrol co-nanoparticles (Figure 4b).Even though the size distribution (Figure 4a)w as slightly broader than in the case of HSA alone, only one population was achieved, which suggestsa ni nteraction between HSA and celastrol.A4:1 flow rate ratio gave ad istribution with an average size of 105.0 AE 0.7 nm (PDI = 0.026) for HSA and 122.5 AE 0.9 nm (PDI = 0.107) forH SA with celastrol (Figure4a), which are ideal for drug delivery purposes. [1]Furthermore, similarly to the BSA-based nanoparticles (Figures 2c and 2f), the produced HSA/celastrol particles were spherical as determinedf rom the TEM images (Figure 4d).However,t he higher contrast in the TEM images suggests higherd ensity of the nanoparticles than in case of pure BSA nanoparticles.This could be due to the addition of celastrol insidet he nanoparticlem atrix.To quantify the amount of encapsulated celastrol and get an estimate of the encapsulation efficiency (EE%) HPLC was used.An EE% of 75 AE 24 %w as achieved by comparing the amount of celastrol injectedi nt he co-flow device to the amount free in solution after nanoparticle formation.
In addition to encapsulation, the release kinetics of the cargo molecules is important for drug delivery applications.The release should not happenb efore the nanoparticleh as reachedi ts target.The release profiles of celastrol were followed by using HPLC.Twoe xperimentsw ere conducted:i n the first, the nanoparticles were placed in PBS (pH 7.3) and in the second experiment, the nanoparticles were mixed with 10 %h uman serum for 24 h( Figure 4e).The release kinetics for both samples follows the same trend over a1 0hperiod reaching % 10 %r elease, but after 24 ht he celastrol concentrationi n the solution drops to 0i nt he sample containing human serum.This could mean either that the celastrol concentration is too high and it forms aggregates, which are then not detected, or that celastrol is not stable in human serum.F inally,t he cell toxicityo ff ree celastrol and HSA/celastrol nanoparticles were examined (Figure 4f)a nd show high toxicity with free celastrol with an EC 50 of 125 AE 34 nm.H owever,w hen celastrol is encapsulated within HSA nanoparticles its toxicity is greatly reduced (EC 50 approximately 1600 nm).Indeed, using encapsulated celastrol, ad ose 10 times higherc an be tolerated by cells.Cell toxicity is an important consideration for biomedical applications and should be minimizedi no rder to reduce potential side effects that occur as ar esult off target interactions with healthyc ells.Even with the best available targetings trategies only ap ortion of the injected drug molecules end up in the target tissue.Thus, reducing the cytotoxicity of ad rug significantly decreases potential side effects.
Protein nanoparticles have gainedc onsiderable attention owing to their high binding capacity of various drugs and low immunogenic response, which minimises adverse side-effects.Herein, am icrofluidic platform for generating protein nanoparticles for drug delivery was presented.Am icrofluidic co-flow methodw as established in which sub-micrometre-sized protein particles were created.By using this co-flow method, HSA nanoparticles were formed with homogeneous size distribution and we showed that by varying the flow rates of the different components, the average size of the nanoparticles can be modulated from % 100 nm to 1 mm.The nanoparticles were stabilized by intermolecular disulfide bondsb yr educing HSA prior to co-flowing andb yg iving them time to re-oxidise after the co-flow,w hich eliminates the need for toxic crosslinkers.We further demonstrate that highly lipophilic celastrol can be encapsulated into our nanoparticles, which increases its solubility in aqueous solutions and decreases its cell toxicity.Future work will determine the potentialo ft hese particles for targeted drug delivery.

Figure 1 .
Figure 1.Microfluidic co-flow device design:S chematic representation of the microfluidic co-flow methodfor synthesis of protein nanoparticles.(a) CAD design from the co-flow device,i nwhich water flows from the outer channel, ethanolfrom the middle channel, and proteinf rom the inner channel.(b) The three solutions meet in the middle of the device to form the nanoparticle.(c) The 3D channelgeometry gives co-flow layers.

Figure 2 .
Figure 2. BSA nanoparticle characterisation: Proteinnanoparticleformation was characterized using two different BSA concentrations.(a-c)1mgmL À1 and (d-f) 10 mg mL À1 solutionsw ere used and characterised with (a, b, da nd e) DLS and TEM (c and f).Size distributions (a and d) and average size (b and e) werer ecorded for six different flow ratios (ethanol/protein flow rate).And TEM images were taken for (c iand ii)1mg and (f ia nd ii)10mgmL À1 BSA nanoparticles with 1:1f low ratio.

Figure 3 .
Figure 3. Stability of BSA nanoparticles :(a) Zeta potential measurement for nanoparticles made from 1and 10 mg mL À1 BSA solutions by using a1:1 flow ratio (ethanol/water).(b) ANS binding to free BSA and BSA nanoparticles.(c) Stability in aqueous solution( PBS, pH 7.3)was examined with three different samples: native BSA (BSA, blue), BSA with blocked free cysteine (BSA + CAA,orange) and GSH reducedB SA (BSA + GSH,green).(d) Size distributionsf or these three samples followingt heir formationb yu sing the co-flow method and (e) average diametera fter 1,2 and 3d ays of incubation in 23 AE 2 8C.The error bars in panel erepresentt he standard deviations of the size distributions.

Figure 4 .
Figure 4. Production of HSA/celastrol hybrid nanoparticles.(a) Celastrol is encapsulated within HSA nanoparticles by the microfluidic co-flow device.This was achieved by adding celastrolt ot he EtOH phase and HSAast he protein phase.(b) TEM images of HSA/celastroln anoparticles formed by using a4:1 flow rate ratio (EtOH/protein).(c)The resulting hybridn anoparticles have slightly broader size distributions.(d) The average diameter sizes were comparable to those of pure HSA nanoparticles.(e) Stability of HSA/celastrol nanoparticles in PBS and in 10 %h uman serum, followedo ver a24htime period.The releaseofcelastrol from the nanoparticles to the outside environment was determined by HPLC.(f) Cell viability in RAW2 64.7 murine macrophages with different concentrations of celastrol, either free in solution or incorporated into HSA nanoparticles.