Protein encapsulation by electrospinning and electrospraying

Given the increasing interest in the use of peptideand protein-based agents in therapeutic strategies, it is fundamental to develop delivery systems capable of 55 preserving the biological activity of these molecules upon administration, and which can provide tuneable release profiles. Electrohydrodynamic (EHD) techniques, encompassing electrospinning and electrospraying, allow the generation of fibres and particles with high surface area-to-volume ratios, versatile architectures, and highly controllable release profiles. This review is focused on exploring the potential of different 60 EHD methods (including blend, emulsion, and co-/multi-axial electrospinning and electrospraying) for the development of peptide and protein delivery systems. An overview of the principles of each technique is first presented, followed by a survey of the literature on the encapsulation of enzymes, growth factors, antibodies, hormones, and vaccine antigens using EHD approaches. The possibility for localised delivery using 65 stimuli-responsive systems is also explored. Finally, the advantages and challenges with each EHD method are summarised, and the necessary steps for clinical translation and


Fabrication of monolithic and core-shell fibres and particles
Typically, electrospinning refers to the fabrication of continuous fibres from polymer 165 suspensions, solutions, or emulsions extruded through a conductive spinneret under the application of a strong electric field (melts can also be processed). An adjustment of the solution properties and processing parameters allows the production of particles instead of fibres, and this process is called electrospraying [22]. The standard setup for EHD processing (Figure 1) includes a high voltage power supply, a syringe pump, a spinneret 170 made from a conductive material, and a collector consisting of a grounded metal plate [25]. The spinneret can comprise a single needle (monoaxial electrospinning; Figure   2A), which yields monolithic fibres from a single solution ( Figure 2B). Alternatively, a coaxial spinneret, with two concentrically nested needles (Figure 2C), can be used to process two fluids into core/shell fibres ( Figure 2D). It is also possible to process three 175 or more fluids (so-called multi-axial electrospinning).
The polymer solution is pumped through the spinneret at a regulated and specified flow rate. Without the application of an electric field, the solution exits the spinneret in the form of droplets, so as to minimise surface tension [22]. Once an electric force is applied to the spinneret, electrostatic charge builds up near the orifice as the polymer 180 solution exits the needle, which, together with the potential difference between the spinneret and the grounded collector, causes the droplet to deform into a conical shape termed the 'Taylor cone' (Figure 2E) [26]. Further increases in the strength of the electric field cause greater charge accumulation at the droplet surface, generating a repulsive force that will eventually overcome the liquid surface tension and culminate in the 185 formation of a jet [27]. This jet either then accelerates towards the collector, narrowing as it is drawn under the electrical field and yielding fibres, or undergoes Coulombic explosion. The latter results in the formation of droplets, and ultimately to particles being deposited on the collector. In either case, as the solution travels towards the collector, the solvent is evaporated, resulting in dry structures [28]. 190 EHD techniques thus allow the production of particles and continuous fibres with sizes on the nano-or micro-scale [29,30]. Proteins and other therapeutic agents can be encapsulated within these structures using different variations of these methods, of which the most common are blend, emulsion, and coaxial electrospinning/spraying [31].
Importantly, due to the rapid speed of the drying process, the distribution of an active 195 ingredient within the solution(s) is maintained and propagated to the solid state. This generally results in amorphous solid dispersions of the therapeutic agent within the fibres or particles forming, which usually increases the dissolution rate of the drug. It also means that the presence of two different solutions in the two needles of a coaxial spinneret will generate a core/shell structure if the processing conditions are 200 appropriately set. tip of the spinneret and liquid surface upon application of an electric field. As charge builds up, the liquid droplet progressively loses its spherical shape and the Taylor cone is formed, followed by the ejection of a jet from the surface. (F) Schematic illustration of coaxial vs. emulsion electrospinning: both result in core-shell fibre architectures, but while coaxial electrospinning is performed with a coaxial spinneret emulsion electrospinning makes use of a monoaxial needle. Monoaxial EHD processing

Blend electrospinning and electrospraying
The simplest method for the development of a drug delivery system using EHD processing relies on the dissolution or dispersion of the drug of choice with a polymer carrier, usually in a volatile solvent (so called 'blend' processing) [31]. A high electrical 215 potential, usually 5-20 kV for monoaxial electrospinning/spraying of a single fluid, is applied to yield monolithic materials with the drug typically dispersed on the molecular level within the fibre or particle product.
The release profile of a drug from a delivery system can be governed by several different mechanisms. Most commonly, drug release is controlled by the diffusion of the 220 active pharmaceutical ingredient through the carrier matrix, dissolution of the carrier matrix within the release media, or erosion of this matrix in the case of insoluble materials [22]. Other mechanisms such as osmosis and swelling may also regulate the release of the encapsulated active ingredients [32]. Since proteins have a number of ionisable sites, they are often charged during the EHD process, and thus will tend to migrate towards 225 the surface of the jet via dielectrophoretic motions [33,34]. Such surface enrichment of the encapsulated drug is common in blend electrospinning/spraying, and is likely to lead to an initial burst of release [35]. Blend fibres have been used successfully for the delivery of antimicrobial peptides and other small molecules [36][37][38]. However, for labile biomolecules like proteins, issues may arise from the typical use of organic solvents in 230 blend EHD processing. A prolonged exposure to these is likely to cause protein misfolding or aggregation, and consequent loss of activity. This limits the use of blend nanofibres/particles for protein delivery [39].

Emulsion electrospinning and electrospraying
Emulsion EHD processing is a relatively new approach for the fabrication of core-235 shell nanostructures using a monoaxial nozzle (Figure 2F), either through a water-in-oil emulsion (hydrophilic droplet phase and lipophilic continuous phase) or vice-versa [40].
This emulsion is used to directly encapsulate compounds in the core of the nanostructure [41]. More specifically, the process involves the electrospinning/spraying an emulsion of two or more immiscible fluids typically stabilized by surfactants such as Pluronics ® , 240 Span ® 60, or Span ® 80 [42]. The product, unlike a blend fibre, consists of two or more phases that do not mix in the EHD process. Conventional lipophilic polymers include polyesters such as PCL, PLA, PLGA, polyurethanes (PUs), and polystyrene, while classical hydrophilic polymers comprise poly(vinyl alcohol) (PVA), poly(ethylene oxide) (PEO), polyvinylpyrrolidone (PVP), cellulose derivatives, chitosan, and alginates [31]. 245 The former will dissolve freely in organic solvents, while the latter are soluble in water or polar liquids.
For hydrophilic molecules such as proteins, water-in-oil emulsions are used. The active ingredient of interest is dissolved in an aqueous phase, followed by dispersion of these aqueous droplets in an organic solvent. Since organic solvents are more volatile, 250 they evaporate at a faster rate, which increases the viscosity of the lipophilic phase. This causes the aqueous phase to concentrate in the centre of the jet, thereby resulting in core-shell structures ( Figure 2F) [43]. Emulsion EHD not only avoids contact of the protein with organic solvents, but also favours their localisation within the centre of the fibres/particles, contributing to a minimisation of burst release effects [44]. Another 255 advantage of this technique is the attainment of a core-shell structure without the use of a complex coaxial spinneret.
The primary disadvantage of emulsion EHD processing is the difficulty in generating stable and uniform core-shell structures using solutions with low surface tension, which characterises most of the solvents used in conventional electrospinning and 260 electrospraying [22]. Emulsion instability is also a common issue. Furthermore, the emulsification process often involves ultrasonication or other mechanical mixing processes to disperse the aqueous phase within the organic solvent, which may affect the structural integrity and functionality of an encapsulated biomolecule.

Experimental considerations 265
In order for an EHD process to be successful, the optimisation of several solution, operational, and environmental parameters is required, and these will also influence the morphology and structure of the products obtained. A brief summary of the main criteria to be taken into consideration are detailed below, but for more extensive information other recent reviews can be consulted [22,45]. 270 Solution parameters include viscosity, conductivity, and surface tension, which are determined by the chosen solvent and the polymer's molecular weight and concentration [45,46]. The viscosity of the solution reflects the degree of molecular entanglement in the liquid [47]. It should be high enough to allow the formation of the Taylor cone, but sufficiently low to avoid clogging of the spinneret. Lower solution viscosities are obtained 275 by using reduced polymer concentrations and molecular weights. If the viscosity is sufficiently high, then a continuous jet is formed and fibres ultimately generated. However, at low viscosities, the solution viscosity is not enough to overcome the liquid surface tension, causing the jet to break into small charged droplets rather than forming continuous fibres. This is electrospraying, and results in particles. Solvent evaporation 280 during the droplets' trajectory towards the collector decreases droplet size and leads to the accumulation of more charges at their surface, further breaking them into smaller particles (atomisation) [22]. Intermediate viscosity values usually give rise to beaded fibres, halfway between electrospraying and electrospinning (Figure 3).
Surface tension is another very important parameter in EHD processing. Since it 285 opposes the jetting effect of the applied electrical force, solutions with increased surface tension may be difficult to spin, easily giving rise to droplets or beaded fibres. In turn, higher solution conductivity facilitates the EHD process, since it promotes greater charge accumulation at the liquid surface and thereby facilitates Taylor cone and liquid jet formation [48]. Another aspect that must be taken into account is the volatility of the 290 solvent of choice. The solvents used in EHD processing should be sufficiently volatile in order to evaporate completely before reaching the collector, so as to avoid fused fibres or particles and other defects like porous structures. However, if the solvent is excessively volatile, premature evaporation may hamper the generation of products with smaller dimensions, as the time for fibre elongation or particle atomisation decreases 295 [22]. The diameter of the ejected products also increases with increased viscosity and decreased conductivity. Clogging of the spinneret may also occur with highly viscous and low conductivity solutions.
Operational or processing parameters comprise the solution flow rate, the distance from spinneret to collector, and the applied voltage. Increasing the solution flow 300 rate may result in larger particle/fibre diameters, since there is a higher amount of liquid being ejected per unit of time. Very high flow rates may lead to incomplete solvent evaporation and to structural defects (wrinkles, pores, fused fibres/particles). On the contrary, if the flow rate is too low then beaded fibre morphologies or discontinuous processes can result [22,49]. The distance from the spinneret to the collector has a direct 305 influence on the trajectory of the charged polymer jet/droplets: if it is too short, there is insufficient time for the solvent to fully evaporate, and defects may form in the products.
If the distance is too long, the yield of the EHD process decreases, since the particles or fibres start depositing on other surfaces as they seek the shortest route to dissipate their charge. Longer distances normally give rise to smaller diameters, as the period of time 310 available for elongation or atomisation is longer. There is interplay between the spinneret-to-collector distance and the applied voltage, since both affect the electrical field gradient. The effect of the applied voltage on product morphology is complex, however, and contradictory results have been reported in the literature [22,50,51]. The applied voltage needs to be high enough to promote the generation of the Taylor cone 315 and ensure stable ejection of the polymer jet (without dripping), but kept under a certain threshold to avert instabilities like multiple jet formation.
Environmental parameters are temperature and humidity, which have a significant influence on solution viscosity, surface tension and conductivity, as well as solvent evaporation rates. High temperatures decrease the viscosity and increase the 320 conductivity of the solution [52], but equally result in faster solvent evaporation rates.
Conversely, solvent evaporation is delayed with increased humidity values [22].
The effects of each parameter on the morphology of the obtained products are summarised in Figure 3. Figure 3. A schematic illustration of the influence of EHD processing parameters on the morphology of the obtained products. A) Effect of solution parameters on the EHD products; B) influence of solution and processing parameters on the diameter of the fibres obtained in electrospinning.

Coaxial EHD 330
Coaxial EHD processing uses a spinneret composed of two nested concentric needles [46] (Figure 2C, F). This allows the simultaneous dispensing of two solutions: the core solution is pumped through the inner needle, while the shell solution exits through the outer needle, generating core-shell structures in a single step. As EHD processing is rapid, the core and shell components will have distinct and separate 335 compartments. Coaxial EHD is particularly useful because it can be performed successfully with only one of the fluids independently processable into solid products via EHD, expanding the range of polymers that can be worked with [31]. Most frequently, a spinnable shell solution is used to encapsulate an otherwise unspinnable core material.
Coaxial processes permit the encapsulation of one or more therapeutic agents within stratified particle carriers or nano/microfibres. This approach can be particularly beneficial with protein active ingredients, since it is possible, for instance, to use an 345 aqueous protein-loaded core solution and a polymer shell solution in an organic solvent.
This minimises contact between the protein and the organic solvent, and thus should help to prevent degradation. The use of the coaxial technique can also help to prevent a burst of release and extend drug release [31]. Potential caveats of this method include a limited rate of production and a complex experimental process requiring extensive 350 optimisation. Multi-axial EHD processing (using more than two fluids) can also be performed with multiple concentric needles, generating multi-layered products. Highly tuneable and exquisite architectures can be produced, but this comes at the expense of an increasingly complex experimental process.

Experimental considerations 355
To achieve successful coaxial electrospinning/spraying, solution, processing, and environmental parameters all need to be considered similarly to monoaxial work (see Section 2.1.3). However, there are some additional considerations to be taken into account when performing coaxial EHD, which are listed in Table 1. Table 1. Summary table of key parameters for successful coaxial electrospinning and spraying [22].

Considerations
Impact Solution parameters

Rate of solidification
 Core and shell solutions should have similar rates of solidification to allow stable co-flow and avoid needle clogging.
Viscoelasticity  Core-shell fibres/particles will form if one solution has sufficient viscoelasticity. The other solution may be composed of a nonspinnable solution.

Miscibility and volatility
 High miscibility of the inner and outer solutions may cause the encapsulated agent to leach from the core to the shell.  Marked volatility differences between the core and shell solvents can result in structural defects (e.g. pore formation).
Flow rate  The shell flow rate must be greater than that of the core to allow efficient encapsulation. Larger core:shell flow rate ratios result in bigger core components.  The core to shell flow rate ratio is usually 1:3 to 1:10.
365 Table 2 summarises the advantages and disadvantages associated with blend, emulsion, and coaxial EHD processing.  66], and even cells [67][68][69] for therapeutic delivery. Electrospinning produces fibres with high interconnected porosity and adjustable pore size, surface functionalisation potential, and 375 structural similarity to the extracellular matrix (ECM), highlighting its applicability in drug delivery, particularly in tissue regeneration [24]. In turn, electrospraying gives rise to nearly monodisperse particles and high encapsulation efficiencies, unlike other frequently used techniques such as spray drying and traditional solvent evaporation (i.e., fabrication of micro-or nanoparticles via single or double emulsification) [24,70,71]. 380 Importantly, EHD processing is cost-effective, robust and scalable, which paves the way for its industrial application [22].
It is essential to choose a polymer or polymer mixture that is compatible with the drug selected for delivery [72]. It should be kept in mind that electrospinning and electrospraying usually require the use of organic solvents, which can destabilise the 385 structural integrity and compromise the therapeutic activity of unstable biomolecules, such as proteins [22]. In this case, emulsion or co-/multi-axial EHD methods may be preferred, since these techniques could protect protein drugs from a direct interaction with the organic phase.
As mentioned in Section 2.1.1., the release of drugs encapsulated in polymer 390 delivery systems is controlled by several different processes: the diffusion of the drug through the polymer matrix, the dissolution of the polymer matrix itself, and, in the case of an insoluble polymer template, the degradation of the polymer matrix by erosion [72].
Choosing polymers with distinct density, molecular weight, and hydrophobicity is fundamental to achieve the required release profiles. For example, PEO is a hydrophilic, 395 fast-dissolving, polymer that can be utilised when an immediate drug release is required, while PLGA is largely insoluble in water but degrades slowly by ester hydrolysis, giving more sustained release of the encapsulated drug [22]. A prevailing issue characteristic of most protein delivery systems is the inherent difficulty in controlling the initial release of the incorporated agent(s). The latter often comprises a burst of release, which is 400 usually not desirable for several reasons: i) it is generally not controllable or predictable; ii) elevated drug levels from the burst effect may exceed therapeutic concentrations and induce toxicity; iii) higher drug loading or further administrations are required in order to maintain therapeutic levels over an extended period of time after the initial burst release, which is not only cumbersome to the patient, but also economically inefficient [73,74]. 405 Therefore, the development of new strategies that can avert initial burst effects and promote sustained drug release is now crucial. Co-/multi-axial EHD processes allow for the production of layered structures that may delay protein diffusion and, hence, prevent burst releases.
EHD processing of polymer solutions is a valuable method that can be used for the 410 generation of protein delivery systems, both for local and systemic administration.
Appropriate control of the polymer solution, processing, and environmental parameters will result in a highly reproducible and adjustable system that can then be tailored to fulfil specific clinical needs. The key advantages of EHD are summarised in Table 3. Table 3. A summary of the advantages of EHD processing for protein encapsulation and delivery.

EHD processing
 Single-step methodology  Significant encapsulation efficiency (up to 100%) [75]  High drug loading capacity (up to 60% in weight) [75]  Particle/fibre uniformity [22]  Capability of processing over 100 polymers with distinct properties [76]; non-spinnable polymers can be used in coaxial or multi-axial techniques  Scaffolds with high interconnected porosity resulting in a large surface area, ranging from 10-100 m 2 /g [77]  Adjustable release profile (both hydrophilic and hydrophobic polymers, variations in molecular weight, and different geometries may be used) [78]  Cost-effective production and reproducibility [79,80] Early reports demonstrated the potential of EHD processing for the encapsulation of bovine serum albumin (BSA) [81,82], enzymes [82,83], and growth factors [84], which catalysed interest in electrospinning and electrospraying for protein delivery. The next section will provide a comprehensive review of different EHD formulations developed for 420 this purpose, including encapsulation of enzymes, growth factors, antibodies, vaccine antigens, and hormones.

Bovine serum albumin encapsulation
A lot of studies directed at protein encapsulation by EHD processing are performed using BSA as a model protein, due to its well-known properties, stability, easy 425 accessibility, and low cost. Jiang et al. [82] encapsulated BSA in electrospun dextran membranes using water as solvent, proving that the protein could be directly incorporated into an aqueous polymer solution without compromising the stability of the electrospinning process. The structure and morphology of the fibres was preserved for BSA concentrations up to 10 wt.%. Additionally, sodium dodecyl sulphate-430 polyacrylamide gel electrophoresis (SDS-PAGE) and circular dichroism (CD) analyses indicated that there was no protein degradation or denaturation. However, the use of water as the solvent in electrospinning is often challenging, since its low volatility and high surface tension hamper the stability of the EHD process and the production of uniform fibres [22]. 435 Hence, in another report, core-shell nanofibres were produced by monoaxial emulsion electrospinning, in which aqueous solutions of BSA and methyl cellulose (used as a protein stabiliser) were mixed with poly(D,L-lactic acid) (PDLLA) in chloroform solution and ultrasonicated to form emulsions before EHD processing [85]. The authors found that the aqueous phase: organic phase ratios used in the emulsification 440 procedures had a significant influence on the fibre diameter and structure. The fibre diameter was inversely proportional to the aqueous phase fraction (i.e., thinner fibres were obtained for higher aqueous: organic ratios). These observations were expected, since an increase in the aqueous content of the spinning solution will also enhance its conductivity, leading, as previously mentioned, to smaller diameters. Moreover, fibres 445 with a more defined and complete core-shell structure were obtained with lower volume fractions of the aqueous phase (1.0%). The latter also resulted in a reduced burst effect and more sustained BSA release over time. It was suggested that, because the nanofibres produced from emulsions with higher aqueous: organic volume ratios have smaller diameters, the increased fibre surface area-to-volume ratios and thinner shell 450 components may have led to faster protein release. However, several concerns were raised regarding the effect of ultrasonication on the protein's native structure and conformation. SDS-PAGE analyses did not reveal the occurrence of any aggregation or degradation, but size-exclusion chromatography and Fourier-transform infrared (FTIR) spectroscopy showed that BSA suffered significant modifications in its secondary 455 structure during ultrasonication. The deleterious effects of ultrasonication were also noted after encapsulation of lysozyme in electrospun PDLLA fibres, although to a lesser extent, with a bioactivity loss of ~16% after emulsification [86]. Subsequent studies have replaced ultrasonication with high speed magnetic stirring to minimise mechanical stress, and surfactants like Tween 20 ® and Span80 ® were introduced to improve protein stability 460 during emulsification [87,88].
A comparison of the encapsulation of BSA in blend and coaxial electrospinning techniques was performed in two reports, studying fibre morphology and structure, protein distribution, and release profiles [89,90]. Both studies used a shell of PCL in trifluoroethanol (TFE) and an aqueous core of PEG and BSA for coaxial electrospinning, 465 whereas the blend fibres were a composite of both polymers and the protein. The solutions used for blend and coaxial electrospinning had the same PCL, BSA, and PEG concentrations. It was found that, unlike the coaxially electrospun nanofibres, the blend fibres presented an irregular beaded morphology. Furthermore, laser scanning confocal microscopy (LSCM) analyses demonstrated that rhodamine B (in the PCL solution) and 470 fluorescein isothiocyanate (FITC)-conjugated BSA (in the core) were evenly distributed throughout the coaxial nanofibres, while in blend fibres they were mainly concentrated in the bead structures [90] (Figure 4). The release profiles were also distinct between the two fibre types: a burst release effect was observed for both blend and core-shell nanofibres, but the latter elicited a more extended protein release profile, over more than 475 30 days. The work by Ji and co-workers corroborated previous observations [91] showing that the addition of PEG to the nanofibres accelerates protein release, presumably due to its hydrophilic nature facilitating fibre dissolution and the penetration of larger amounts of release medium into the fibres. shellfor dual drug delivery. Rhodamine B (RhoB) was incorporated in the PLGA shell, while FITC-conjugated BSA was encapsulated in the gelatine layer of the fibres. A burst release effect followed by a sustained release profile was observed for both molecules, and the release of FITC-BSA was faster when RhoB was present in the shell layer. The authors attributed this phenomenon to the hydrophilic character of RhoB, which 495 promotes the penetration and retention of higher amounts of water in the fibres and, therefore, enhances the diffusion and release of BSA. In addition, the tri-layered scaffolds were associated with higher levels of cell metabolic activity after culture with adipose-derived stem cells than monolithic PLGA and gelatine (core)-PLGA (shell) fibres. PLGA electrospun scaffolds tend to shrink when in contact with cell culture media 500 [93]. The authors suggest that the addition of the PCL core layer in the triaxial fibres helped reduce the contraction of the fibrous scaffolds more efficiently than the uniaxial and coaxial systems, therefore providing a larger surface area for cell attachment and proliferation.
BSA and lysozyme were successfully encapsulated in PLGA microparticles by 505 emulsion electrospraying in the presence of a surfactant (Pluronic ® F127) [94]. In this report, the use of sonication to generate an emulsion and surfactant addition were found to improve the BSA encapsulation efficiency (76.6% was achieved using probe sonication for the emulsification and a Pluronic ® F127 concentration of 10%, compared to an encapsulation efficiency of 20% attained by vortex emulsification and absence of 510 surfactant). SDS-PAGE, FTIR, CD, and enzyme-linked immunosorbent assay (ELISA) studies indicate that the secondary structure of the protein was not majorly affected.
Likewise, encapsulation in PLGA microparticles preserved the enzymatic activity of lysozyme at 92%. The PLGA concentration and the presence of the surfactant significantly influenced the protein release profiles from the different formulations. For 515 instance, the initial burst was suppressed in the release profile of BSA-loaded microparticles prepared from 10% (w/v) PLGA containing 10% Pluronic ® F127, from which BSA was released in a sustained fashion over 35 days. However, protein burst release occurred from particles generated with lower PLGA (6%) and surfactant (0 and 5%) concentrations. 520 In a later study, Zamani and co-workers [95] performed both emulsion and coaxial electrospraying of PLGA microparticles for the encapsulation of BSA. The protein encapsulation efficiency was superior for coaxially electrosprayed microparticles (approximately 70%) compared to that of emulsion electrospraying (up to 50%), but, interestingly, the latter presented lower burst effects during in vitro protein release 525 studies. The authors believed this may be due to a greater extent of adsorption of the protein to the hydrophobic chains of PLGA during the emulsification process, which arises from a large organic-aqueous interface that is not present in coaxial electrospraying. Importantly, it was also emphasised that such protein-polymer interactions had previously been associated with loss of native protein structure and 530 consequent denaturation.

Enzyme encapsulation
The encapsulation of enzymes in electrospun fibres and electrosprayed particles can be a simple way of evaluating the effects of EHD processing on protein structure and 535 bioactivity, due to the commercial availability of many standardised enzyme activity assays. One such protein is lysozyme, an anti-bacterial enzyme for which the substrate is peptidoglycan, one of the major components of the cell wall of Gram-positive bacteria [96]. Liu and co-workers created a composite scaffold by combining electrosprayed lysozyme nanoparticles with electrospun PLGA and PEG-PLGA monolithic fibres [97]. 540 The electrosprayed lysozyme solution was prepared by dissolving the enzyme in varying ratios of ethanol (EtOH) and water. Optimal enzymatic activity preservation (approximately 100%) was attained with an EtOH:H2O ratio of 30:70 v/v, as measured after electrospraying. The electrosprayed lysozyme nanoparticles (with or without PEG) were then dispersed in a PLGA solution for electrospinning. Lysozyme release from the 545 composite mats was low, with only up to 25% of the enzyme loading being detected after 56 days. The authors found that a large fraction of the protein was present within the fibres in the form of insoluble aggregates after the 56 day period, which may explain why such a small amount of the enzyme was detected in the release medium. Of note, lysozyme bioactivity was not assessed after release from the electrospun mats, but it is 550 possible that such structural alterations may have compromised its catalytic activity.
A few strategies have been developed to improve lysozyme stability in suitable solvents for electrospinning and spraying, such as the conjugation of lysozyme with a fatty acid (oleate), which can provide amphiphilicity and, thus, enhance the solubility of the enzyme in organic solvents [91,98]. The enzyme was encapsulated in electrospun 555 PCL-PEG blend fibres, and its solubility in dimethylsulfoxide (DMSO) was improved from 12.6 ± 4.2 mg/mL to 21.1 ± 3.5 mg/mL after ionic conjugation with oleate [91].
Furthermore, the activity of the encapsulated lysozyme released from the fibrous meshes was retained at ca. 90% for over 7 weeks. In a report by Puhl et al. [99], electrospinning of lysozyme crystals in a polymer suspension, rather than a solution or emulsion of the 560 protein, was performed ( Figure 5). The authors proposed that, because protein crystals are more thermodynamically stable and present a smaller solvent-exposed surface area compared to a solution, they could be more suitable for EHD processing in organic solvents. The preparation of lysozyme crystals and suspension in pure PCL and composite PCL-PEG and PCL-PLGA solutions for electrospinning allowed for a relatively 565 homogenous distribution of the protein within the fibres [99]. The bioactivity of the lysozyme released from PCL and PCL-PEG fibres was efficiently preserved; however, the acid degradation products from PLGA in PCL-PLGA fibres and consequent alteration of the release medium pH led to a loss of up to ~30% in bioactivity over a period of 11 weeks. In another report [100], emulsion electrospinning was used for the encapsulation 570 of lysozyme in PCL-PEO composite fibres. After a release period of 24h, the highest catalytic activity of the enzyme was seen when the emulsification process was performed with a concentration of 0.4% (v/v) of Span80 ® and a decrease in the sonication amplitude.

580
Various studies have revealed that the release profiles of lysozyme could be tailored via the ratio of hydrophilic (PEO/PEG) to hydrophobic (PCL/PLGA) polymers. Larger amounts of enzyme were released for higher PEO:PCL [100,101] and PEG:PCL [91,99] ratios. It is thus possible to conclude that, at optimised hydrophilic: hydrophobic polymer ratios, both PEO and PEG can act as porogenic agents, since their rapid dissolution in 585 aqueous media increases fibre porosity and enhances protein diffusion and release from the polymer matrix. However, in work by Liu et al. [97], this tendency was only verified for a PEG:PLGA ratio of up to 10:100 (w/w), as the lysozyme release was slower at a 20:100 ratio. The authors suggested that this phenomenon resulted from a change in the thermodynamic properties of the scaffold with a higher PEG concentration in the mixture, 590 which facilitated the interaction between PEG and PLGA chains and slowed down enzyme diffusion and release. It should be noted that the lysozyme encapsulation efficiency was not presented in this study, meaning the lower amount of released lysozyme could also be a result of reduced encapsulation efficiency in the electrospun mats with higher PEG content. 595 Complex delivery devices may be generated by the combination of various drug carriers in a single construct. For instance, horseradish peroxidase (HRP)-loaded liposomes were prepared and subsequently electrospun into PVA-PCL blend or coreshell fibres, resulting in the incorporation of these vesicles into fibrous scaffolds [39]. The core-shell fibres consistently provided better results, including in terms of improved 600 liposome stability and preservation of HRP activity (average value of 62.3% compared to 9.59% in blend fibres).
In addition to their dissolution in solvents prior to EHD processing, proteins can also be incorporated into particles or scaffolds after electrospraying or electrospinning. One example of this was provided by Ma and co-workers, who prepared silica supraparticles 605 (Si-SPs) to be used as an inner ear drug delivery system [102]. Alginate spheres encapsulating several individual silica particles were prepared by electrospraying. The alginate template was then removed, resulting in Si-SPs with an average diameter of ~550 μm. The Si-SPs were incubated with a FITC-lysozyme solution after electrospraying, and it was shown by LSCM that the enzyme not only interacted with the 610 Si-SPs surface, but was also found in the interior of the particles, due to their porous nature. Protein loading capacities of up to 15 µg of lysozyme per particle were achieved.
Sustained release profiles over 40 days were obtained for both lysozyme, used as a model protein, and brain-derived neurotrophic factor.
Examples of other enzymes encapsulated in electrospun scaffolds include alkaline 615 phosphatase (ALP) [90] and α-chymotrypsin [103,104]. Importantly, the applications of enzyme-containing electrospun systems are not limited to model studies or drug delivery.
In work by Dai and colleagues [105], laccase, an oxidase, was loaded into PDLLA fibres via emulsion electrospinning and explored for bioremediation applications. The activity of the encapsulated laccase was reduced to 67% of that of the free enzyme, presumably 620 due to denaturation during the emulsification process. Nevertheless, this catalytic activity was conserved at up to 50% after ten consecutive reactions of the same electrospun mats, highlighting the relatively high operational stability of this system. Similar results were obtained in a separate study [106], where emulsion electrospinning was used for the encapsulation of trypsin into PCL fibres for potential industrial applications. The 625 incorporation of FITC-conjugated trypsin during the electrospinning process enabled the observation of a relatively uniform distribution of the enzyme within the nanofibres, and catalytic activity after encapsulation was preserved at up to ~66%. The encapsulated enzyme demonstrated higher thermal and storage stability compared to the free enzyme counterpart. Moreover, the operational stability of this system was also satisfactory, with 630 59% of trypsin's enzymatic activity being conserved after five consecutive reactions.
The studies presented above emphasise the suitability of enzyme-loaded electrospun/electrosprayed systems for both medical and industrial applications. It is worth mentioning, however, that the preservation of catalytic activity during EHD processing remains a challenge in a significant number of these devices, which may limit 635 their practical applications. Because this enzymatic activity is highly dependent on the protein's three-dimensional structure and conformation, the use of harsh processing conditions (e.g. contact with organic solvents or extensive organic-aqueous interfaces, ultrasonication) during the electrospinning or electrospraying process should be avoided.

Growth factor delivery 640
Growth factor delivery is a strategy widely used in TE to promote cell migration, growth, and differentiation [107]. Nevertheless, the administration of growth factors in vivo is highly inefficient, due to their low stability, short half-lives, and rapid inactivation [108,109]. In addition, overdosing with these protein agents is associated with detrimental side effects, including an increased risk of cancer development [110]. As 645 such, the creation of delivery devices capable of preserving the bioactivity of growth factors during processing and storage, and providing sustained release profiles, is pivotal for these molecules to be more extensively used in clinical applications [111].
In the last decade, a lot of effort has been put into fabricating growth factor delivery systems based on scaffolds and microcarriers produced by EHD processing. One of the 650 main areas of focus in these studies is neural TE, in which both mechanical and chemical cues are important determinants for the success of the construct. For example, an aligned electrospun PCL scaffold was combined with BSA-loaded electrosprayed PLGA core-shell spheres in order to provide distinct nano-and micro-scaled topographical cues for the promotion of neural cell growth [112] (Figures 6 and 7). In other work, Hu and 655 co-workers developed an aligned nanofibrous scaffold composed of nerve growth factor (NGF)-encapsulating PCL nanofibres [88]. The fibres were produced using emulsion electrospinning, and their alignment was achieved by collection on a cylindrical mandrel rotating at high speed (~3000 rpm). The co-encapsulation of NGF and BSA allowed sustained release of the growth factor over 28 days. Moreover, the protein-loaded 660 scaffolds successfully promoted the adhesion, proliferation and neurite extension of PC12 cells (a rat adrenal phaeochromocytoma cell line widely used as a neuronal cell model, as their phenotype is similar to that of sympathetic ganglion neurons after NGFmediated differentiation [113]). A similar approach was used in another study, with the polymer matrix composed of PLGA [114]. 665 Coaxial electrospinning has been used to produce nerve guidance conduits (NGCs) encapsulating NGF and tested in a rat sciatic nerve model (Kuihua et al., 2014;. After excision of a sciatic nerve segment, the NGCs were used as reparative grafts, and shown to improve nerve functional recovery when compared to scaffolds without NGF. In a different report, NGF-loaded PLGA microspheres were incorporated 670 into methacrylated hyaluronic acid fibrous scaffolds [117]. Scaffolds with aligned or random fibres were obtained by collecting them on a rotating mandrel at a high (10 m/s) or low (0.5 m/s) speed, respectively. The microspheres were generated with a water-inoil-in-water (W/O/W) double emulsion and mixed with the polymer solution before electrospinning. This delivery system successfully stimulated neurite extension in dorsal 675 root ganglion neurons, demonstrating that the encapsulated NGF retained its bioactivity.
Importantly, the direction of the neurite growth depended on the orientation of the fibres in the scaffold: in randomly oriented fibres, the neurites grew in several directions, while in aligned scaffolds the neurites tended to grow in the direction of alignment. Encapsulation efficiencies over 80% were achieved for NGF and GDNF, as well as sustained release profiles over 42 days after a relatively small burst release (up to 16.7% 705 and 26.6% for NGF and GDNF, respectively, in the first 24h). In vitro, the NGF and GDNF released from the fibrous scaffolds had only a small loss of bioactivity compared to freshly dissolved growth factors, implying that the electrospinning process does not impair their functionality [121]. Bone and cartilage TE is another major area of research where EHD methods have been heavily explored. One of the most widely used growth factors in this context is bone 715 morphogenetic protein 2 (BMP-2). BMP-2 is osteoinductive, stimulating progenitor cell proliferation and differentiation into osteogenic phenotypes [123,124], and essential in bone fracture recovery [125]. Furthermore, BMP-2-based therapy has been approved by US) and InductOS ® (Wyeth, Europe) [127]. As such, BMP-2 has been incorporated in a great variety of biomaterials, frequently associated with inorganic, osteoconductive components like hydroxyapatite, for sustained delivery [128][129][130]. For example, hydroxyapatite nanoparticles incorporated in silk fibroin (SF) and chitosan blend solutions have been used to form the shell of fibres produced by coaxial electrospinning, 725 while the core was composed of an aqueous solution of BMP-2 [131]. These scaffolds demonstrated osteoconductive and osteoinductive properties in vitro and in vivo. BMP-2 has also been combined with other drugs in several delivery systems for the maximisation of cellular and tissue responses. Coaxial electrospinning was used to encapsulate BMP-2 and dexamethasone in poly(L-lactide-co-caprolactone) (PLLACL)-730 collagen [132] and PLLA-zein [133] based nanofibres. In the first report [132], core-shell (shell: PLLACL-collagen and dexamethasone; core: BMP-2 in PBS stabilised by BSA) and blend nanofibres were compared in terms of the release profiles of BSA over 21 days (however, BMP-2 release was not evaluated). It was shown that the core-shell structured fibres provided a slower release of BSA than blend fibres of the same 735 composition. In the second study [133], PLLA and BMP-2 composed the core of the fibres, while the shell was constituted of zein, a biocompatible protein extracted from corn, and dexamethasone. The encapsulation of BMP-2 within the fibre core allowed for a more sustained release during 21 days, but also caused the retention of a large amount of protein (45-55%) inside the nanofibres. In both studies, mesenchymal stem cell (MSC) 740 adhesion, growth, and expression of osteogenic markers were promoted more efficiently by the dual drug-loaded scaffolds than blank or single drug-loaded nanofibres.
Recently, Cheng and co-workers successfully developed another dual-release system, where BMP-2-loaded PVA (core)/SF-PCL (shell) nanofibres were produced by coaxial electrospinning and subsequently coated with connective tissue growth factor 745 (CTGF) using a layer-by-layer (LbL) technique (Figure 8) [134]. Both BMP-2 and CTGF are important chemical cues in bone regeneration, but while the former is present throughout the whole regenerative process CTGF appears to be more influential in the early stages. Accordingly, the objective was to achieve a fast release of CTGF and a sustained release of BMP-2, promoting both osteogenic and angiogenic responses for 750 bone regeneration. A burst release of approximately 60% of CTGF was achieved over the first 24h, while BMP-2 was released in a controlled fashion over 40 days. electrospraying [136]. Protein powders (human serum albumin, BMP-7, and VEGF) were first micronized to reduce particle size and subsequently mixed in an aqueous solution with PEG, followed by lyophilisation. The effect of trehalose as a protein stabiliser in the mixture was also evaluated. A PLGA in chloroform solution was added to the lyophilised particles and the dispersion homogenised by magnetic stirring and sonication. After 770 electrospraying, only trace amounts of BMP-7 and VEGF from the microparticles could be detected by ELISA, indicating denaturation or aggregation. Nevertheless, after critical procedures such as micronisation and vortexing with chloroform, their bioactivity levels were comparable to freshly prepared BMP-7 and VEGF controls in cell-based assays, where they retained in vitro efficacy. It is possible that, during extraction or release from 775 the microparticles, reversible modifications occur in the growth factor structure that hinder their ELISA detection, but do not compromise their bioactivity. The addition of trehalose to the formulation resulted in slower growth factor release, but had no influence on in vitro bioactivity.
Cartilage TE is a particularly challenging field, since the naturally low cellular content 780 and the avascularity of this tissue both impair its intrinsic regenerative capacity. In order to promote the homing of chondrocytes to the lesion site and improve graft-host integration, a blend PLGA-PCL mesh was developed for the encapsulation of insulin-like growth factor 1 (IGF-1) [137]. The IGF-1-loaded meshes supported chondrocyte growth and the deposition of cartilage ECM components such as glycosaminoglycans (GAGs) 785 and type II collagen in vitro. In vivo, IGF-1 seemingly improved graft integration and ECM deposition compared to blank meshes. Another scaffold, composed of TGF-β1-loaded PLGA microspheres incorporated into electrospun PCL nanofibres, was also evaluated for its chondrogenic potential [138]. The authors demonstrated that this fibrous construct was able to support MSC attachment, proliferation, and chondrogenic differentiation, with 790 improved production of GAGs and type II collagen.
VEGF and platelet-derived growth factor (PDGF) have been explored in electrospinning both for bone regeneration [139][140][141] and cardiovascular TE. Two studies developed dual-delivery systems incorporating both growth factors using modified coaxial and DSDP electrospinning to create double-layered [142] and multi-795 layered [143] polymer vascular scaffolds (Figure 9), respectively. The chemical compositions of these meshes involved both natural (chitosan and gelatin) and synthetic polymers (poly(ethylene glycol)-b-poly(lactide-co-ε-caprolactone), PLGA, and PCL). In both cases, the encapsulation of VEGF and PDGF was tailored so that the former would be released faster than the latter, leading to differential release profiles. The objectives 800 of this were to promote an early recruitment of vascular endothelial cells (ECs), stimulated by VEGF, and the later homing of vascular smooth muscle cells (SMCs), in which PDGF plays a more determinant role. In addition, the authors intended to avoid PDGF-induced SMC hyperproliferation, which is seemingly inhibited by VEGF [144].
After in vivo implantation in a rabbit model, both scaffolds were able to promote the 805 homing of ECs and SMCs, and the combined action of VEGF and PDGF allowed the inhibition of SMC hyperproliferation. It should be noted, however, that the encapsulation efficiencies for both growth factors were quite low in both the double-layered and the multi-layered devices (under 20%).

815
Skin regeneration is another potential application for electrospun TE constructs.
PDGF and CTGF have been incorporated in electrospun scaffolds for wound dressing purposes [145,146]. Basic fibroblast growth factor (bFGF)-loaded PEG-PDLLA scaffolds fabricated with an emulsion electrospinning technique have been tested as skin patches in a diabetic mouse model [147]. These constructs markedly accelerated wound closure 820 and healing, promoting skin re-epithelialisation, vascularisation, and the growth of skin appendages like hair and sebaceous glands. The release of bFGF from PLGA-cellulose acetate electrospun fibres was modulated in another study using DSDP and emulsioncoaxial electrospinning [148]. Bi-layered and tri-layered scaffolds were produced by DSDP electrospinning of a bFGF emulsion in PLGA and a cellulose acetate solution (bi-825 layered scaffolds were composed of a bFGF+PLGA layer and a cellulose acetate layer, whereas in tri-layered scaffolds the bFGF+PLGA stratum was "sandwiched" by two cellulose acetate layers; see Figure 10a). Core-shell fibres were generated by coaxial electrospinning, where the core comprised the bFGF and PLGA emulsion and the shell a solution of cellulose acetate (Figure 10b). The hybrid scaffolds provided a more 830 sustained release of the protein than that observed for the scaffolds without cellulose acetate. The slowest bFGF release profile was achieved from the tri-layered fibrous scaffold, although the growth factor release was rather incomplete (~40% after approximately one month).

840
Similarly, epidermal growth factor (EGF) has been encapsulated in electrospun scaffolds for wound healing applications [149,150], and multiple delivery studies of EGF with bFGF [151,152] and VEGF and PDGF [152] were performed. Albright and colleagues designed TGF-β1-loaded PCL-collagen electrospun nanofibres, which were subsequently coated with gentamycin/clindamycin antibiotic-containing micelles using a 845 LbL approach [153]. The objective was to use these fibrous meshes as skin patches for TGF-β1-mediated wound healing, and simultaneous prevention of bacterial growth and lesion infection. Importantly, the encapsulated TGF-β1 seemingly maintained its bioactivity in vitro, and the incorporation of both gentamycin and clindamycin in the micelle-coated scaffolds successfully inhibited the growth of a Staphylococcus aureus 850 strain.
The existence of such an extensive list of studies applying EHD processing for the generation of tissue engineered constructs demonstrates the potential of this technique for successful outcomes and clinical translation. Although a few reports showed that growth factors may be loaded into emulsion electrospun fibres and still retain a significant 855 portion of their bioactivity, there is a tendency for the use of coaxial electrospinning/spraying instead. Indeed, the incorporation of these molecules in the core of coaxially produced fibres or particles is a more effective way of preventing their contact with organic solvents or organic-aqueous interfaces than using an emulsion, thereby minimising protein denaturation and loss of functionality. 860

Antibody delivery
To the best of our knowledge, the first study reporting the use of EHD processing for antibody encapsulation was performed by Gandhi et al. [154]. In this work, anti-αvβ3 integrin immunoglobulin G (IgG) antibodies, promising antiangiogenic agents, were incorporated into monoaxial electrospun PCL nanofibres. BSA was encapsulated 865 together with the antibodies, as a potential stabilising excipient. The release profiles of BSA and the antibody from all tested formulations were typically associated with significant burst releases followed by slower release over 30 days. Faster protein release was observed where lower PCL concentrations (11 wt.%) were used in the spinning solutions, and at low antibody: BSA ratios (1:100, w/w). Additionally, the antigen-binding 870 activity of the antibodies to αvβ3 integrins expressed by human umbilical vein endothelial cells (hUVECs) was shown to be maintained after release from the nanofibres.
The extensively used protein active ingredient bevacizumab (commercial name Avastin ® ) is an anti-VEGF antibody used in the treatment of age-related macular degeneration and several types of cancer, including glioblastoma, metastatic colorectal 875 cancer, and non-small cell lung cancer [155]. Bevacizumab was encapsulated in PCL core-shell fibres by Angkawinitwong and co-workers [156]. Using coaxial electrospinning, the authors found that the pH of the core solution in which the antibody was dissolved and electrospun played a crucial role in functional performance. Two distinct pH values were tested: 6.2 (pH of the commercial bevacizumab formulation, 880 Avastin ® ) and 8.3 in Trizma ® buffer (the isoelectric point of the antibody). In general, the core solution with pH 8.3 led to well-defined core-shell structures, zero-order release over two months without any burst effect, and efficient preservation of bevacizumab's structural integrity and VEGF-binding activity. The pH of the core solution affects the antibody charge, which, in turn, influenced its behaviour during the coaxial 885 electrospinning process: at pH 6.2, bevacizumab has a positive charge, and tends to migrate towards the solution surface upon application of an electric field. In contrast, at the isoelectric point of the antibody, it will be neutral, hence remaining in the fibre core.
This preferential distribution in the fibre core efficiently protects bevacizumab from contact with the shell solvent (TFE) and losing bioactivity, and promotes a gradual 890 release with no burst observable. Conversely, the localisation of more antibody molecules near the fibre surface results in a more extensive interaction with the organic solvent and a burst of release. Recent work further demonstrated that coaxially electrospun mats composed of (PVA-bevacizumab)-(PCL-gelatin) core-shell nanofibres were able to decrease the number of blood vessels formed in a chicken embryo 895 chorioallantoic membrane (CAM) model, suggesting the maintenance of antiangiogenic activity [157].

Vaccines
EHD processing techniques, particularly electrospraying, have been widely studied for the formulation of protein-based vaccines. A few of these reports have focused 900 specifically on the development of oral vaccine formulations. Oral vaccination, compared to the typical subcutaneous immunisation route, can be associated with benefits such as improved vaccine accessibility and distribution (due to the possibility for selfadministration), and stimulation of both systemic and mucosal immune responses [158].
The naturally occurring polymer chitosan has attracted much attention in this regard, 905 since it is mucoadhesive [159]. This characteristic favours its retention in the mucosal system for prolonged periods of time [160], maintaining locally high concentrations of the antigen and thus improving immune recognition and response. In a study by Suksamran Release of OVA from the microparticles in in vitro studies was incomplete, albeit sustained, over the time of the study (24h), with a maximum of ~60% of protein release for uncoated particles and ~40% for chitosan-coated particles.
OVA was also used as a model antigen in a different study [162], where it was encapsulated in coaxially electrosprayed acetalated dextran (Ace-DEX) microparticles 920 and tested in vitro and in vivo with soluble mirabutide, an adjuvant with an immunomodulatory structure similar to peptidoglycan [163]. Ace-DEX samples with approximately 20, 40, and 60% cyclic acetal coverage (CAC) levels were tested, in order to attain tuneable degradation and protein release kinetics. Using BSA as a model for OVA, 91.8% of the encapsulated protein was released from the Ace-DEX microparticles 925 with 20% CAC after a 48h period, whereas Ace-DEX with higher CAC levels led to much slower protein release. OVA encapsulation efficiencies over 70% were obtained for all microparticle formulations, but the highest value was observed for the 60% CAC Ace-DEX formulation (99.8%). Immunisation studies in a mouse model showed considerably enhanced serum anti-OVA IgG titres compared to mice given soluble OVA, with the most 930 potent antibody responses seen with the lowest CAC level formulation. Splenocytes from the immunised mice were harvested for ex vivo pro-inflammatory cytokine expression analyses upon exposure to OVA. The production of tumour necrosis factor (TNF)-α, interleukin (IL)-2, and interferon (IFN)-γ was also greater in cells isolated from mice vaccinated with encapsulated OVA compared to soluble OVA treatment. 935 Electrosprayed PVA-coated PLGA nanoparticles were utilised for the encapsulation of subunit antigens in the form of cytomegalovirus peptides, and the properties of this system were compared to particles fabricated by double emulsion methods [164] (Figure   11). The production of electrosprayed particles with nanoscale diameters was achieved by decreasing solution flow rate (optimal value of 10 μL/h) and increasing solution 940 conductivity (Figure 11a, b). The group found that 80% of the chosen antigen (pp65495-503 peptide) was released from the electrosprayed particles over the first 10 days, following a burst release of ca. 20%. The antigen encapsulation efficiency was generally superior for electrosprayed particles than those produced by double emulsion methods.
In vitro TNF-α and IFN-γ production by CD8 + T cells that were exposed to the antigen-945 loaded electrosprayed nanoparticles was comparable to those exposed to the soluble antigen counterparts.

Hormone delivery
Insulin is a protein hormone that is secreted by pancreatic β-cells, and is responsible for cellular glucose uptake and carbohydrate, lipid, and protein metabolism [165]. 960 Deficiencies in the production of insulin or the development of insulin resistance are associated with diabetes, a chronic disease that affects millions of people worldwide, with an incidence and prevalence that have been increasing steadily over recent decades (World Health Organization, "Diabetes," 2020). Exogenous insulin administration, crucial in the management of diabetes, is given mainly subcutaneously. 965 This, coupled with the required high frequency of injection required, can lead to low patient compliance [167]. For this reason, a growing number of studies have explored electrospinning and electrospraying for alternative insulin administration routes.
One such study focused on using fish sarcoplasmic proteins as a polymer carrier to encapsulate insulin into nanofibres via blend electrospinning for oral delivery [168,169]. 970 A high encapsulation efficiency of 98.6% was achieved, as well as a loading capacity of 14%. Maximum insulin release was obtained after a period of 3h, and the fibres were capable of protecting the encapsulated protein from α-chymotrypsin degradation. CD studies suggested that, even though the fibres were processed by blend electrospinning and insulin was in direct contact with an organic solvent (hexafluoroisopropanol, HFIP), 975 the protein structure was not affected. The same trend was observed in another report, in which semi-interpenetrating networks of gelatine and insulin were produced by blend electrospinning in HFIP [167]. In this study, it was proven that insulin retained its signalling activity in vitro after electrospinning, as it was capable of inducing Akt phosphorylation and adipogenic differentiation of pre-adipocytes. Moreover, the 980 encapsulated hormone was able to better permeate through a porcine buccal mucosa model compared to free insulin. The mucoadhesive properties of chitosan may also be advantageous in this context. Lancina et al. [160] produced blend fibres of chitosan, PEO, and insulin for transbuccal delivery, again using HFIP as the solvent. Sustained insulin release was achieved for 24h and its activity in vitro was maintained, as assessed 985 by Akt phosphorylation asssays. These studies demonstrate that it is possible to produce insulin delivery systems by blend electrospinning without compromising protein structure or bioactivity.
Coaxial electrospinning has recently been employed for this same purpose, and core (insulin)-shell (PLGA) fibres were generated for application in wound healing in diabetic 990 patients [170]. The core-shell fibres were compared with blend fibres of the same composition. LSCM studies using recombinant enhanced green fluorescent protein (reGFP) as a model revealed that protein distribution was more uniform in core-shell fibres than in the blend counterparts. Additionally, in an in vivo model, wound closure was faster when the core-shell fibre scaffolds were used than with the blend fibres. 995 Further examples of protein hormone encapsulation within electrospun fibres include growth hormone (GH)-loaded systems [171]. GH is commonly used to treat children with growth impairment, in order to normalise their height, body composition, and pubertal development [172]. It is also given to adults with GH deficiency [173]. Similarly to insulin, the most frequent method of administration of GH is via recurrent subcutaneous 1000 injections, owing to its short half-life [174]. A sustained GH release system was created by stabilisation of the hormone within sugar glass nanoparticles, which were subsequently encapsulated in PCL or poly(ester urea) electrospun fibres. GH release was observed for a period of 6 weeks with minimal burst, and the released protein maintained its bioactivity in in vitro cellular assays [175]. 1005

Chemokine delivery
Shafiq and co-workers [176] developed hybrid PCL-collagen vascular grafts by coelectrospinning from two separate spinnerets, preparing collagen fibres loaded with one of two chemotactic factors: stromal cell-derived factor 1α (SDF-1α), which promotes progenitor cell homing, angiogenesis, and tissue repair [177,178], or substance P (SP), 1010 a neuropeptide capable of promoting angiogenesis [179] and modulating MSC migration [180] and cytokine secretion [181]. Both grafts had suitable mechanical and biochemical properties for supporting the growth of ECs and SMCs in vivo, allowing for endothelialisation and vascularisation after implantation. Zamani et al. focused on the encapsulation of SDF-1α in coaxially electrosprayed PLGA microparticles to develop an 1015 injectable protein delivery system for cardiac tissue regeneration [182]. This core-shell system (Figure 12) showed sustained release of SDF-1α over 40 days, after a relatively small burst release (26.6% and 16.4% with and without BSA co-encapsulation, respectively). Moreover, in in vitro transwell assays, the chemotactic activity of the released SDF-1α towards MSCs was preserved. In a recent study [183], SDF-1α or 1020 granulocyte colony-stimulating factor (G-CSF) were separately encapsulated in poly(ethylene oxide terephthalate)/poly(butylene terephthalate) fibres via emulsion electrospinning. Even though beaded fibres were observed, rather than uniform, beadless meshes, these chemokine-loaded scaffolds supported the growth of MSCs.

3.7.
Stimuli-responsive protein delivery systems 1030 Several strategies have been proposed for localised drug delivery. Active targeting usually relies on the functionalisation of a drug delivery vehicle with specific moieties that are preferentially recognised by a subset of cells or tissues present at the target site [184]. In passive targeting, the therapeutic agents reach the target organ or tissue by passive mechanisms (e.g. the enhanced permeation and retention [EPR] effect in tumour 1035 tissues) [185]. An effective way of localising delivery is the use of stimuli-responsive systems, in which drug release is triggered by internal or external cues such as changes in local pH or temperature [186].
Sukarto and Amsden developed a temperature-responsive system using electrosprayed P(TMC-CL)2-PEG (poly(1,3-trimethylene carbonate-co-ε-caprolactone)-1040 b-poly(ethylene glycol)-b-poly(1,3-trimethylene carbonate-co-ε-caprolactone)) microspheres loaded with four different proteins: BSA, lysozyme, BMP-6, and TGF-β3 [187]. The microspheres were then loaded into a N-methacrylate glycol chitosan hydrogel to be used as a controlled delivery system for cartilage repair. P(TMC-CL)2-PEG solidified at temperatures lower than 10 °C, allowing its easy dispersion within the 1045 hydrogel, but it became a viscous liquid at body temperature, which could then trigger the release of the encapsulated active factors. Sustained release profiles were attained over up to two months, even though initial burst effects were observed. The isoelectric point of each protein was thought to influence its release. Proteins which were negatively charged at the physiological pH (TGF-β3, BSA) were released faster than those which 1050 were positively charged (lysozyme, BMP-6), even though the chitosan hydrogel matrix also had a positive charge. The authors suggested that the positively charged lysozyme and BMP-6 may interact with P(TMC-CL)2-PEG or with its negatively charged degradation products, thus explaining the slower release from the microspheres.
EHD products composed of pH-responsive polymers allow the release of 1055 microencapsulated compounds in a pH-dependent manner, which has potential applications for oral protein delivery. The GI tract is characterised by a wide range of pH values throughout its length, with an acidic environment in the stomach and a typically neutral-slightly basic pH in the small and large intestines [188]. Thus, fibres or particles that are insoluble at acidic pH but soluble in neutral conditions can be used for targeted 1060 GI delivery. Importantly, such systems are capable of protecting the encapsulated drugs from inactivation or degradation within the acidic milieu of the stomach, allowing their selective and controllable release in the gut [189].
A pH-sensitive device for peroral protein delivery was produced using HRP and ALP as model enzymes [87]. The polymer of choice was Eudragit ® L100, an anionic co-1065 polymer of methacrylic acid and methyl methacrylate that is soluble in biological fluids with pH > 6, but not in acidic conditions [190]. Emulsion electrospinning was used for the encapsulation of HRP or ALP into Eudragit ® L100 fibres, with very high encapsulation efficiencies for both enzymes (>94%). As expected, protein release was shown to be greatly dependent on the pH of the release medium: only ~5% of the enzymes were 1070 detected in a medium at pH 2, but an increase to pH 7 led to release of approximately 100% of the encapsulated proteins in 1h (Figure 13). This was attributed to the fast dissolution of the fibres at neutral pH. Among the several processing parameters that were analysed in this study, the authors found that the aqueous: organic phase ratio may have an influence on bioactivity preservation after electrospinning. This influence was 1075 only observed for ALP, which had a more accentuated loss of catalytic activity when the aqueous phase was increased from 5 to 20% v/v (80% compared to 50% of bioactivity, respectively). This effect was presumably due to more extensive organic: aqueous interfaces and protein-solvent interactions at higher aqueous phase volumes.
Conversely, the enzymatic activity of HRP was preserved at 90% regardless of aqueous 1080 phase content. Additionally, it was found that protein stability during storage was improved by fibre encapsulation and lyophilisation, compared to solution formulations. of the fibrous structure at pH 2 (undissolved) and pH 7 (dissolved). EEHRP5 and EEAP5 are emulsion electrospun HRP and AP formulations with an aqueous phase of 5% (v/v); EEHRP20 and EEAP20 are formulations with an aqueous phase of 20%. Reprinted from [87] with permission from Elsevier. Copyright © 2017, Elsevier B.V.
A different example of a stimulus-responsive system was provided by Yan et al., who 1090 fabricated ultrasound-sensitive electrosprayed microbubbles for the local delivery of tissue plasminogen activator (tPA) in ischemic stroke [191,192]. tPA is an enzyme that promotes clot breakdown and prevents thrombosis. Using coaxial electrospraying, microbubbles with a gaseous core (air or sulphur hexafluoride, SF6) and a shell composed of phosphatidylcholine and PEG were produced. tPA was also incorporated 1095 into the microbubble shell (Figure 14). The coaxial electrospraying process allowed the generation of microbubbles with minimal aggregation [191]. Furthermore, ca. 80% of the bioactivity of tPA was preserved even at the harshest spraying conditions (voltage of 14 kV). Microbubble bursting and tPA release was dramatically enhanced by the exposure to 2 MHz ultrasonic waves of higher amplitudes, therefore validating the ultrasound-1100 sensitivity of this delivery system [192]. The abovementioned examples illustrate how EHD processing can be applied for the development of tailored, stimuli-responsive, materials for targeted delivery and controlled release while maintaining protein bioactivity. These characteristics hold promise for oral protein-based therapies, which are more patient-friendly than parenteral 1110 administrations, and for long-acting local delivery, potentially minimising systemic side effects and the need for frequent dosing. Structural and bioactivity preservation over the course of the release study (2 months) was observed for the formulation with a core solution of pH 8.3 [156] a Calculated only for BSA in coaxial fibres without PEG. b Optimised values obtained by using probe sonication (vs. vortex) for the emulsification process and a surfactant (Pluronic ® F127) concentration of 10%

Discussion and future outlook
The mechanisms governing fibre and particle production by EHD processing and their applicability in various medical and industrial fields have been increasingly explored 1120 over the last decade. The relatively simple experimental set-up and low cost of electrospinning and electrospraying have led to a growing number of studies making use of these techniques for TE and drug delivery purposes. Traditional techniques for the production of scaffolds and nano/microparticles include, for instance, freeze-drying (or lyophilisation), solvent casting/particulate leaching, emulsification, and spray drying. 1125 However, application of these methods for protein encapsulation is often challenging, due to the requirement for aggressive processing conditions that may result in loss of structure and biological activity. For instance, spray drying typically requires high processing temperatures and may exert shear, interfacial stress, and dehydration stresses on the encapsulated proteins [193]. Similarly, the organic solvents used in 1130 solvent casting are unsuitable for protein encapsulation [194]. During lyophilisation, proteins undergo different kinds of stress both while freezing (e.g. temperature lowering, generation of extensive aqueous-ice interfaces) and drying (e.g. dehydration) [195]. As mentioned previously, emulsification leads to the formation of aqueous-organic interfaces that can also affect protein conformation and bioactivity [196]. Conversely, 1135 EHD processing is typically performed under ambient temperature and pressure conditions, thereby avoiding heat-induced denaturation and inactivation of the incorporated proteins. Although during extrusion through the spinneret shear forces may manifest, and a strong electric field is applied, these do not seem to have any deleterious effect on protein stability. 1140 Further, the versatility of the technique has enabled the creation of multiple geometries as a way of regulating protein distribution within the delivery system and the release profile of the active agent(s). Overall, EHD methods have great potential for the encapsulation and delivery of protein active ingredients, but care is needed to ensure that bioactivity is not compromised upon processing. Blend, emulsion, and co-/multi-axial 1145 electrospinning and electrospraying are all different variations of EHD processing, and the determination of the most appropriate method for the encapsulation of a certain drug should take several parameters into consideration.
The studies discussed throughout this review, together with those additionally presented in Table 4, allow us to draw some conclusions on the most suitable strategies 1150 for protein EHD encapsulation. Owing to the poor stability of protein drugs in non-native conditions, particularly when in direct contact with organic solvents, blend electrospinning and electrospraying are usually not employed for water-soluble protein encapsulation (even though there are exceptions, such as insulin, which is stable even when in contact with HFIP). Instead, these protein agents should be solubilised in 1155 aqueous solvents and encapsulated either via emulsion or co-/multi-axial EHD methods.
Emulsion electrospinning has been broadly used to this end, and promising results were attained in terms of structural and functional protein preservation. However, the emulsification process can be detrimental, decreasing protein stability and resulting in loss of activity of the biological agent. These adverse effects may arise from the contact 1160 of the encapsulated proteins with organic-aqueous interfaces or from the ultrasonication procedure that is frequently employed to achieve more uniform emulsions. The use of appropriate surfactants and the replacement of ultrasonication with high speed stirring may improve the stability of the protein drugs throughout the emulsification protocol.
In general, there seems to be an increasing preference for coaxial or multi-axial 1165 techniques, as they allow for protein encapsulation in a hydrophilic, possibly nonspinnable, core that is incorporated into a polymer shell. Since the core and shell solutions are kept separated right until they exit the spinneret for fibre or particle formation, the contact between the protein agents and organic solvents used in the shell solution is minimised. Furthermore, even though core-shell structures can also be 1170 achieved with emulsion electrospinning/spraying, there is a tendency for the production of more uniform and consistent core-shell geometries with coaxial methods. The obtained release profiles are highly dependent on the distribution of the protein drug within the delivery system, therefore favouring the coaxial approach. In spite of these advantages, the thorough and laborious optimisation steps involved in coaxial and multi-1175 axial methods must be taken into account. The most suitable method for a given application must consider the physicochemical properties of the encapsulated agent, the desired characteristics of the final product, and the complexity of the experimental setup ( Table 2).
In order for these protein delivery systems to be used as therapeutic tools, their 1180 manufacturing procedure must be upscaled for industrial production. The industrial implementation of electrospinning and electrospraying requires the consideration of several factors. Firstly, it is essential to increase the volume of production from laboratory to industrial scales [197]. A recent review by Vass et al. [198] provides a very comprehensive overview of different methods for scaling-up the electrospinning process, 1185 including multi-needle, free surface (needleless), melt, and alternating current electrospinning approaches. Methods for the scale-up of electrospraying have received less attention, but some progress has already been made towards this goal. Strategies like high-throughput nozzles with open-channel architectures [199], multi-nozzle electrospraying [200], and needleless multi-pore electrospraying [201] have been 1190 developed to increase production rates. In the latter, solution flow rates of up to 10.5 mL/h were achieved. Recently, a novel high-throughput technology termed electrospraying assisted by pressurized gas (EAPG) was employed for the encapsulation of algae [202] and fish oil [203,204], to be used as nutraceutical supplements. In this method, atomisation of the polymer solution is performed by a 1195 pneumatic injector using compressed air, which nebulizes upon application of an electric field. After solvent evaporation, dry particles are then collected in a cyclone as a freeflowing powder [203]. Scaled-up electrospinning and electrospraying production is already performed routinely by a few companies such as InoCure (https://inocure.cz/), which is focused on applying these technologies towards drug delivery, cell culture, and 1200 TE.
In addition, it is fundamental that the production can be carried under current good manufacturing practice (cGMP) conditions. The facilities are now available, with companies like Bioinicia focusing on EHD products for biomedical, pharmaceutical, and cosmetic applications and requiring cGMP certification (https://bioinicia.com). 1205 Furthermore, it is necessary to guarantee batch-to-batch reproducibility. To this end, tight control of the processing and environmental conditions (temperature and humidity) is fundamental, and is already available in several electrospinning/spraying devices [197].
Another potential obstacle in industrial EHD processing is the difficulty in producing fibres or particles with complex geometries (e.g. core-shell) in a high-throughput fashion. 1210 This will require specially designed spinnerets or collectors and a very thorough and continuous quality control process [197]. Finally, it should be taken into account that these techniques often involve the use of large quantities of organic solvents, adding environmental and safety concerns. Strict policies for residue treatment and solvent recovery must be applied to ensure environmental safety, bearing in mind that the 1215 presence of residual solvents in the products may jeopardise their biomedical application.
Industrial scale EHD fabrication of protein delivery systems has not received much attention to date. However, a few researchers have explored the production of oral formulations (tablets) from protein-loaded electrospun fibres. Vass and co-workers have 1220 published a number of reports where β-galactosidase (lactase) was encapsulated within polymer fibres with varying compositions and further processed into tablets with excellent preservation of bioactivity during formulation and storage [205][206][207]. Feeding rates of up to 400 mL/h were achieved, corresponding to a productivity of approximately 270 g/h of dry material [205]. The tablets produced were able to preserve the enzymatic activity of 1225 lactase at nearly 100% after six months [205] and one year of storage [206,207] at room temperature, a remarkable improvement over liquid formulations of the enzyme. These studies demonstrate the potential of EHD processing for the development of oral solidstate protein active ingredient formulations.
Solid-state protein formulations offer several advantages compared to liquid 1230 formulations, since proteins in solution or suspension are usually prone to structural modifications, hydrolytic degradation, and require low-temperature storage and distribution (the abovementioned cold-chain) [208]. Unlike classical drying methods such as lyophilisation or spray drying, EHD processing is a cost-effective technique that is not associated with high processing temperatures or freeze-thawing cycles, therefore 1235 contributing to better preservation of the protein conformation and bioactivity [207].
It is hence possible to conclude that EHD processing holds great promise for the production of protein-drug delivery systems, as it provides a gentle encapsulation mechanism with a plethora of possible geometries and allows the development of solidstate formulations, potentially at an industrial scale. Nevertheless, there is still a long way 1240 to go before these formulations can be widely used in biomedical scenarios. Clinical trials need to be performed for the evaluation of their safety and efficacy profiles, and regulatory barriers must be overcome before their commercialisation is authorised.
Moreover, the industrial implementation of these methods will require further increases in productivity, reproducibility, standardised quality control protocols, and environmental 1245 safety policies to ensure regulatory and economic viability.

Conclusions
This review has demonstrated that blend, emulsion, and co-/multi-axial electrospinning and electrospraying are suitable for the encapsulation and delivery of a range of protein active ingredients, including enzymes, growth factors, antibodies, and 1250 protein-based vaccines. This range of proteins has widely varying structural characteristics, physicochemical properties, and biological roles, but all can be incorporated within electrospun fibres or electrosprayed particles with considerable functionality and structural integrity preservation. EHD methods are exceptionally versatile, since a large range of natural and synthetic polymers can be employed, either 1255 individually or in a blend. This results in products with characteristics that can be incrementally varied to provide the desired three-dimensional structures, mechanical properties, degradation rates, and release profiles. In general, co-/multi-axial electrospinning and electrospraying tend to be the most suitable techniques for the encapsulation of proteins, although promising results have also been achieved by 1260 emulsion EHD. Several high-throughput alternatives to the low-throughput methods used in most laboratory research have been proposed over the last few years, enabling increased productivity rates and paving the way for industrial implementation of electrospinning/spraying. Thus, EHD techniques have broad potential in the pharmaceutical and biomedical fields, particularly in drug delivery and tissue 1265 engineering, although extensive medical and operational research will be required before widespread clinical application can take place.