Optimizing Self-Emulsifying Drug Delivery Systems for the Oral Delivery of a Hydrophobic Ion-Paired Lysozyme Complex

Abstract

Background: The oral delivery of biopharmaceuticals remains a major challenge for researchers and the pharmaceutical industry. Therefore, extensive research is ongoing to develop a viable delivery method, hence self-emulsifying drug delivery systems (SEDDSs) are being investigated because of their ability to protect the carried macromolecules in the gastrointestinal environment and facilitate absorption through the intestinal barrier. Objectives: To systematically investigate this promising method for the oral delivery of lysozyme (LYZ) and to model oral peptide/protein administration. Methods: LYZ/sodium dodecyl sulfate (SDS) hydrophobic ion pairs (HIPs) were prepared to enhance protein solubility and stability in SEDDSs. Different surfactants (Tween^® 20 and 80) and as co-surfactants (Span^® 20 and 80) were combined for the preparation of liquid SEDDSs according to a 2² full factorial design and samples of each combination were formulated based on a three-factor-constrained mixture design. The critical quality attributes (CQAs), droplet size, polydispersity index (PDI), and zeta potential were measured by dynamic light scattering (DLS). The process design space was determined by response surface methodology (RSM) and two-dimensional ternary contour plots. An in vitro release test was performed using the sample-and-separate approach. Results: Emulsions of SEDDSs with the optimal properties of droplet size < 200 nm, PDI < 0.4 and zeta potential < −10 mV were prepared. Consequently, a HIP load of 10 mg/g was achievable, exhibiting apparent first-order kinetics, with approximately 80% of the loaded LYZ released within 6 h. Conclusions: This study may contribute to better understanding of the effects and interactions of formulating materials for SEDDSs and their possible role in the oral delivery of biopharmaceuticals.

1. Introduction

The medicinal application and market share of various biopharmaceuticals, including peptides, proteins, and antibodies, are rapidly expanding due to their high efficacy, selectivity, and favorable safety profiles [1,2]. Biopharmaceuticals have revolutionized the treatment of diseases, for instance, diabetes mellitus [3,4], cardiovascular disorders, COPD [5,6], infectious diseases, and autoimmune or malignant conditions [7]. Despite their clinical success, most biopharmaceuticals are administered parenterally because of their poor oral bioavailability [8]. This administration route is associated with reduced patient adherence, especially for chronic therapies, due to its invasive nature, discomfort caused, need for medical personnel, as well as high manufacturing and healthcare costs, which limit the widespread use of these therapies [9].

Therefore, extensive research is ongoing to deliver biopharmaceuticals via alternative administration routes, among which oral delivery remains the most convenient and widely accepted one, despite its numerous challenges [5,9].

The main obstacles of the effective oral bioavailability of peptides are: high susceptibility to enzymatic degradation [10], pH-induced instability in the gastrointestinal (GI) tract [11], poor permeability across the intestinal epithelium due to their large molecular size and polarity, and first-pass metabolism following absorption [12]. According to the Biopharmaceutical Classification System (BCS) peptide and protein drugs typically fall into class III or IV, characterized by low permeability and/or low solubility [5,13].

To overcome these barriers, drug delivery methods that are capable of protecting the carried protein and enhancing intestinal absorption need to be utilized. Various strategies have been employed, including structural modifications [10], or the co-administration of excipients, such as permeation enhancers [14], pH modulators [3], and direct enzyme inhibitors [13].

However, formulation-based strategies using lipid-based drug delivery systems [12,15] have attracted particular interest due to their excellent biocompatibility and safety profiles [16,17]. Lipid-based systems can improve protein stability, enhance intestinal absorption [1] and potentially promote lymphatic uptake, which may be key to increasing oral bioavailability by avoiding first-pass metabolism in the liver [12,15].

Self-emulsifying drug delivery systems (SEDDSs) provide a novel and favorable approach for the oral administration of peptides and proteins due to their simple and cost-efficient preparation and ability to enhance drug solubilization and absorption [18]. SEDDSs are isotropic mixtures of drugs, lipids, emulsifiers, and one or more co-emulsifiers. Based on droplet size, SEDDSs are differentiated as self-microemulsifying drug delivery systems (SMEDDSs, 100–250 nm), or self-nanoemulsifying drug delivery systems (SNEDDSs, <100 nm) [19,20]. Upon contact with GI fluids, SEDDSs preconcentrates spontaneously form thermodynamically stable, oil-in-water (O/W) micro- or nanoemulsions, significantly improving drug solubilization and absorption [20,21,22].

Although SEDDSs have been successfully commercialized for several small-molecular drugs, their widespread application was not in the spotlight until the need for oral delivery of biopharmaceuticals emerged [23,24]. There have been attempts to formulate SEDDSs containing daptomycin, desmopressin, enoxaparin, exenatide, heparin [25], and leuprorelin [26], as well as several attempts to develop a method for the oral administration of insulin [27,28,29]. Nevertheless, their use for biopharmaceuticals has been limited by the hydrophilic nature of peptides and proteins, which hinders their stable incorporation into lipophilic formulations [30].

To overcome this limitation, hydrophobic ion pairing (HIP) has been employed to increase the lipophilicity of macromolecules through neutralizing their charge with oppositely charged counterions [31,32]. The resulting complex has a significantly higher lipophilicity, enabling the stable incorporation of proteins into lipid-based carriers [33]. This strategy is simple and cost-effective and does not result in chemical modification, thereby preserving biological activity and minimizing regulatory concerns [34].

Numerous studies, including work from our research group, have investigated the potential of the utilization of HIP complexation for peptides and proteins [34,35,36], but to our knowledge, no systematic investigation has been performed to evaluate the solubility of the obtained HIP complex in various SEDDSs formulations. In the present study, lysozyme (LYZ) was selected as a model protein. LYZ is a well-characterized antimicrobial enzyme consisting of 129 amino acids (MW = 14.4 kDa) [37], with an isoelectric point (pI) of 11.3, maintaining its activity over a broad pH range [38]. It is a natural part of the innate immune system, and can be found in human saliva, but its main source is chicken egg white, which provides plentiful and relatively cheap access to it. In recent years LYZ has gained new attention from a therapeutic perspective due to its antibacterial effect, and its antiviral, antifungal, immunomodulatory and anticancer activities were revealed [38,39].

Considering its broad therapeutic potential, lysozyme is a suitable model protein for investigating the oral delivery of biopharmaceuticals. Building on a previously established quality-by-design-based HIP complexation method developed by our research group, this study aimed to systematically develop and optimize a lysozyme-loaded self-emulsifying drug delivery system for the oral administration of lysozyme. To achieve a deeper understanding of the role and interactions of formulation components within this multicomponent system, a design of experiments (DoE)-based factorial approach was employed, reducing experimental time and labor intensity and ensuring product quality.

2. Materials and Methods

2.1. Materials

Tween 20^® (polysorbate 20) and Tween 80^® (polysorbate 80) purchased from Merck KGaA (Darmstadt, Germany) were used as surfactants, Span 20^® (sorbitan monolaurate) and Span 80^® (sorbitan monooleate) purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan) were used as co-surfactants, and Miglyol 810^® (medium-chain triglycerides) from Sasol Chemicals GmbH (Witten, Germany) was used as the oil component of the SEDDS formulation. The lyophilized powder of the chicken egg white lysozyme was purchased from MedChemExpress (Monmouth, NJ, USA) and sodium dodecyl sulfate (SDS) used in hydrophobic ion-pair formation was gifted by EGIS Pharmaceuticals Plc. (Budapest, Hungary). The chemicals potassium hydroxide and hydrochloric acid (Ph. Eur. 11.0) were used for pH adjustment. Purified water was used for the preparation of different solutions.

2.2. Methods

2.2.1. Mixture Design for the Preparation of Liquid SEDDSs

Miglyol^® 810 was used as the oil phase for SEEDS formulation, while 4 different compositions were created by the combination of emulsifiers (Tween 80 (−1) and Tween 20 (+1)) and co-emulsifiers (Span 80 (−1) and Span 20 (+1)) according to a 2² full factorial design (Table 1). The experimental plan was designed by TIBCO Statistica, v. 13 (TIBCO Software Inc, Palo Alto, CA, USA).

Table 1.

Composition of SEDDSs based on the mixture design experimental plan (%, w/w).

Materials	Concentration in Various Experimental Runs (%)
Composition 1	1 V	2 V	3 V	4 V	5 C
Miglyol 810	10	20	10	20	15
Tween 80	60	60	80	70	67.5
Span 80	30	20	10	10	17.5
Composition 2	1 V	2 V	3 V	4 V	5 C
Miglyol 810	10	20	10	20	15
Tween 20	60	60	80	70	67.5
Span 80	30	20	10	10	17.5
Composition 3	1 V	2 V	3 V	4 V	5 C
Miglyol 810	10	20	10	20	15
Tween 80	60	60	80	70	67.5
Span 20	30	20	10	10	17.5
Composition 4	1 V	2 V	3 V	4 V	5 C
Miglyol 810	10	20	10	20	15
Tween 20	60	60	80	70	67.5
Span 20	30	20	10	10	17.5

To examine the influence and interactions of each component on the characteristics of the formulations, a three-factor-constrained mixture design was used. The percentage of the oil phase (A), the emulsifier (B), and the co-emulsifier (C) were selected as independent variables with a total concentration of 100%. The low and high values of the factors were chosen according to the existing literature in this field [22,23,35]. Consequently, the amount of oil phase, surfactant and co-surfactant ranged from 10–20%, 60–80%, and 10–30%, respectively (Table 1), for each combination of surfactants and co-surfactants. The results acquired from the applied mixture design were evaluated with the help of response surface methodology (RSM) to identify the underlying interactions of the variables and determine the formulations that met the desired criteria.

The appropriate amount of each component was measured and mixed in a beaker with a magnetic stirrer at 700 rpm and 45 °C for 2 h. Following that, the homogenous preconcentrates were stored in the refrigerator (2–8 °C).

2.2.2. Self-Emulsifying Qualities of the Preconcentrates

The SEDDS preconcentrates were emulsified by dispersing them in purified water in a ratio of 1:100 on a heated magnetic stirrer (Multi-HS 6 Digital, VELP Scientifica, Usmate, Italy) at 150 rpm and 37 °C to mimic the gastrointestinal environment. The self-emulsifying ability of all samples were visually observed for the clarity, phase separation and coalescence of the oil droplets immediately after preparation. The self-emulsifying properties of the preconcentrates were also evaluated by measuring and recording the time required for spontaneous self-emulsification upon dilution. A classification system reported by Shafiq et al. [40] was used to grade the in vitro performance of the resulting emulsions. The definitions used for each grade can be seen in Table 2.

Table 2.

Emulsion grading system [40].

Grade A	Rapidly forming (within 1 min) nanoemulsion, with a clear or bluish appearance.
Grade B	Rapidly forming, slightly less-clear emulsion, with a bluish-white appearance.
Grade C	Fine, milky emulsion that forms within 2 min.
Grade D	Dull, grayish-white emulsion with a slightly oily appearance that is slow to emulsify (longer than 2 min).
Grade E	Formulation exhibiting either poor or minimal emulsification with large oil globules present on the surface.

2.2.3. Kinetic Stability Tests of SEDDSs

After determination of the CQAs, the 100 mL emulsions were divided into centrifuge tubes, each containing 25 mL. The samples were centrifuged at 10,000 rpm for 5 min using the Hermle Z323K high-performance refrigerated centrifuge (Hermle AG, Gossheim, Germany). After centrifugation, the samples were visually examined for any sign of instability such as phase separation, creaming, or turbidity. They were observed immediately after centrifugation and one week later, after being stored at 25 °C.

2.2.4. Determination of Droplet Size, PDI, and Zeta Potential

Besides visual observation, the CQAs (droplet size, PDI and zeta potential) of the prepared emulsions were determined by dynamic light scattering (DLS) using a Malvern nano ZS instrument (Malvern Instruments, Worcestershire, UK), equipped with a He-Ne laser (633 nm). Measurements were performed in disposable folded capillary cuvettes (Malvern Instruments, UK) at 25 °C, with an equilibration time of 120 s and back-scattering detection at 173°, with the general data processing model. The refractive index (RI) was determined before measurement (RI: 1.334).

2.2.5. Determination and Validation of Design Space

To determine the design space, the predicted responses of the chosen compositions were displayed as contour plots by TIBCO Statistica, v. 13 (TIBCO Software Inc, Palo Alto, CA, USA). Following that, the areas corresponding with the preliminarily determined acceptance criteria (e.g., droplet size < 200 nm, PDI < 0.4, zeta potential < −10 mV) were cropped from the two-dimensional contour plots for each composition and were overlapped into one ternary diagram using Inkscape software (Inkscape: Open Source Scalable Vector Graphics Editor, v. 1.4., Boston, MA, USA).

For validation, two points representing considerably different compositions were selected within each design space and the values predicted by the DoE model were used for comparison. The samples used for validation were investigated according to the methods described above. After examination, the observed and predicted values of each variable were compared using factorial ANOVA and post hoc LSD tests.

2.2.6. Hydrophobic Ion Pairing (HIP) of Lysozyme

The HIP complexation of proteins may be essential to enable their stability in lipid-based carriers. Beside pKa and chain length, toxicity is also an important consideration for selection of the appropriate ion-pairing agent. SDS is already commonly used in the food industry and non-parenteral pharmaceutical formulations, for instance as a tablet lubricant in a 0.5–2.0% concentration. Published data for the oral toxicity of SDS indicate an LD₅₀ of 1.29 g/kg in rats, with the probable human lethal dose estimated to be between 0.5 and 5.0 g/kg body weight. Based on the optimal 1:8 molar ratio of lysozyme to SDS, the theoretical SDS content corresponds to approximately 1.6 mg per 10 mg of HIP complex. Therefore, the amounts of SDS present in the HIP complex are several orders of magnitude lower than doses associated with toxicity [41].

A LYZ/SDS HIP complex was prepared using a 1:8 molar ratio at pH 8 according to the previously published methodology [34]. Briefly, LYZ and SDS were dissolved individually in phosphate buffer (PBS pH 6.8) in a concentration of 4 mg/mL and 0.645 mg/mL, respectively. The pH of the freshly made solutions was adjusted to 8.0 before complexation, using 1 M KOH (and 1 M HCl if necessary). After that, equal volumes of both solutions were measured into centrifuge tubes by pipetting and were mixed. The resulting final LYZ concentration was 2 mg/mL. The samples were left standing for 30 min and then centrifuged at 15,000 rpm for 15 min at room temperature using the Hermle Z323K high-performance refrigerated centrifuge (Hermle AG, Gossheim, Germany). Subsequently, the supernatant was removed, and the samples were dried in a ventilated oven (Memmert GmbH + Co. KG, Buechenbach, Germany) at 25 °C with a 50% fan speed for 3 h. The dried complexes were collected and stored in a freezer (at −18 °C).

2.2.7. LYZ/SDS HIP Complex Solubility Test

The selected compositions for validation were used for the solubility tests, which were carried out by dissolving the previously prepared LYZ/SDS complex in increasing concentrations of 5, 10 and 15 mg/g in the chosen SEDDS preconcentrates. The samples were stirred with a magnetic stirrer at 700 rpm and 40 °C for 4 h, until the mixtures became visually clear and transparent. The samples were stored in a refrigerator (2–8 °C) until further investigation. The droplet size, PDI and zeta potential of the HIP complex-loaded SEDDSs were determined again after emulsification using a Malvern nano ZS instrument (Malvern Instruments, Worcestershire, UK) as described in Section 2.2.4. The achieved drug load (${\mathrm{m}}_{\mathrm{L}\mathrm{Y}\mathrm{Z}}$ [mg]) was calculated by the following equation (Equation (1)):

$${\mathrm{m}}_{\mathrm{L}\mathrm{Y}\mathrm{Z}}={\mathrm{m}}_{\mathrm{H}\mathrm{I}\mathrm{P}}\times {\displaystyle \frac{{\mathrm{M}\mathrm{W}}_{\mathrm{L}\mathrm{Y}\mathrm{Z}}}{{\mathrm{M}\mathrm{W}}_{\mathrm{L}\mathrm{Y}\mathrm{Z}}+\mathrm{R}\times {\mathrm{M}\mathrm{W}}_{\mathrm{S}\mathrm{D}\mathrm{S}}}}$$ (1)

where m is the mass and MW is the molecular weight of the components, while R is the molar ratio of LYZ/SDS.

2.2.8. In Vitro Drug Release Test of HIP-Loaded SEDDSs

To carry out the drug release study, the sample-and-separate method was applied. The SEDDSs were loaded with the LYZ/SDS HIP complex in a concentration of 10 mg/g, as described in Section 2.2.7. The HIP-loaded SEDDSs preconcentrates were emulsified by dispersing them in PBS at pH 6.8 in a ratio of 1:100 on a heated magnetic stirrer (Multi-HS 6 Digital, VELP Scientifica, Usmate, Italy) at 150 rpm and 37 °C.

Samples with a volume of 2 mL were taken from the emulsions and put into Eppendorf tubes, which was followed by centrifugation at 16,500 rpm for 30 min at 4 °C using the Hermle Z323K high-performance refrigerated centrifuge (Hermle AG, Gossheim, Germany). After sedimentation, the supernatant was carefully removed and analyzed, leaving the nanoparticle- and HIP-containing pellet in the bottom. The pellet was resuspended with 1 mL PBS at pH 6.8 (137 mM NaCl) and the Eppendorf tubes were put in a water bath (37 °C) and mildly agitated with 100 rpm. During the dissolution test, samples were centrifuged at the 1, 3, 6 and 24 h timepoints, with the aforementioned conditions, after that the supernatants were carefully removed and the pellets were resuspended again, and the tubes were put back into the water bath.

The amount of free LYZ in the collected supernatants were determined using a UV spectrophotometric method at λ_max = 281 nm (Genesys 10 S UV-VIS Spectrometer, Thermo Fisher Scientific Inc., Waltham, MA, USA). As a zero baseline for the absorbance measurement, blank emulsions were prepared and centrifuged following the same procedure. As described in Section 3.2., 10 mg of HIP complex equal to 8.611 mg of LYZ was considered the theoretical maximum amount. If the applied dilution is taken into consideration, this corresponds to 0.1722 mg/mL in the emulsion samples. The unencapsulated LYZ concentration was determined from the absorbance of the emulsion at the zero timepoint (Equation (2)).

$$\mathrm{U}\mathrm{n}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{a}\mathrm{p}\mathrm{s}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d} \mathrm{L}\mathrm{Y}\mathrm{Z}\left(\%\right)={\displaystyle \frac{{\mathrm{C}}_{\mathrm{L}\mathrm{Y}\mathrm{Z} 0}}{{\mathrm{C}}_{\mathrm{L}\mathrm{Y}\mathrm{Z} \mathrm{m}\mathrm{a}\mathrm{x}}}}\times 100$$ (2)

where C_{LYZ 0} is the measured concentration of LYZ after the initial centrifugal separation. C_{LYZ max} is the theoretical maximum concentration of LYZ in the sample.

The amount of LYZ at the chosen timepoints was calculated with the measured absorbance of the samples and using the following equation (Equation (3)):

$$\mathrm{L}\mathrm{Y}\mathrm{Z}\left(\%\right)={\displaystyle \frac{{\mathrm{C}}_{\mathrm{L}\mathrm{Y}\mathrm{Z} \mathrm{t}}}{{\mathrm{C}}_{\mathrm{L}\mathrm{Y}\mathrm{Z} \mathrm{m}\mathrm{a}\mathrm{x}}-{\mathrm{C}}_{\mathrm{l}\mathrm{Y}\mathrm{Z} 0}}}\times 100$$ (3)

where C_{LYZ t} is the measured LYZ concentration at the given timepoint.

The kinetic model was fitted with Sigmaplot v.12.1 (Systat Software Inc., San Jose, CA, USA).

2.2.9. Physical Stability Studies of HIP-Loaded SEDDSs

To investigate the physical stability of liquid SEDDSs under simulated in vivo gastric and intestinal conditions, the selected liquid preconcentrates were diluted in a 1:100 ratio in 0.1 M HCl at pH 1.4 and PBS at pH 6.8 and emulsified by dispersing them on a heated magnetic stirrer (Multi-HS 6 Digital, VELP Scientifica, Usmate, Italy) at 150 rpm and 37 °C to mimic the gastrointestinal environment. Right after that, their droplet size, polydispersity index and zeta potential were determined as in Section 2.2.4. Separate samples were analyzed immediately after dilution (0 h) and continuously at the selected timepoints in both media. For the simulated gastric conditions, measurements were performed at 0, 1 and 2 h, reflecting typical gastric residence time. For the simulated intestinal conditions, measurements were conducted at 0, 2, 4, 6 and 24 h. The 0–6 h timepoints correspond to the average small intestine transit time, while the 24 h timepoint was included to evaluate the extended stability and to correlate the results with the duration of the in vitro drug release study.

3. Results and Discussion

3.1. Characterization of Liquid SEDDS

The aim of the present study was the systematic investigation of the effect on SEDDS composition on the solubility of HIP-complexed lysozyme. As toxicity is an important aspect of excipient selection, only nonionic surfactants were used in the SEDDS formulations, since ionic surfactants are more frequently associated with mucosal irritation [18,23]. Furthermore, a relatively high Hydrophilic-Lipophilic Balance (HLB) value is required to achieve rapid self-emulsification; however, excessive surfactant concentrations may increase the risk of GI irritation. The inclusion of appropriate co-surfactants enables a reduction in the total surfactant content while maintaining efficient self-emulsification and emulsion stability, thereby minimizing potential mucosal irritation [42]. The chosen components Tween 20^® and Tween 80^® (surfactants), Span 20^® and Span 80^® (co-surfactants), and Miglyol 810^®, the oil phase all possess GRAS status and commonly used in pharmaceutical formulations, including SEDDS [42].

3.1.1. Grading of Self-Emulsification and Kinetic Stability

The self-emulsifying performance of the liquid SEDDS preconcentrates was first evaluated by visual observation combined with measurement of the self-emulsification time, according to the method described in Section 2.2.2. [40]. The results are summarized in Table 3. Upon dilution under standardized conditions, formulations graded as A and B formed transparent or slightly bluish emulsions within a short emulsification time, whereas formulations graded as C and D exhibited longer emulsification times and a cloudy, milky appearance.

Table 3.

Grading and self-emulsification times (SEs) of each composition of emulsions based on the mixture design.

C1	Grading	SE Time (s)	C2	Grading	SE Time (s)
1 V	D	142 ± 12	1 V	D	129 ± 17
2 V	D	167 ± 9	2 V	B	32 ± 4
3 V	B	38 ± 6	3 V	A	15 ± 6
4 V	A	19 ± 4	4 V	B	45 ± 3
5 C	D	134 ± 15	5 C	B	28 ± 4
C3	Grading	SE Time (s)	C4	Grading	SE Time (s)
1 V	D	163 ± 15	1 V	D	124 ± 9
2 V	D	187 ± 21	2 V	A	16 ± 3
3 V	D	134 ± 12	3 V	A	12 ± 5
4 V	B	33 ± 5	4 V	B	31 ± 5
5 C	D	168 ± 16	5 C	A	18 ± 4

Results are expressed as mean ± SD (n = 3).

The visual examination confirmed that samples graded as A and B exhibited good kinetic stability as the same appearance was observed after one week following the preparation and centrifugation test. In contrast, emulsions graded as C and D already showed signs of phase separation after one week.

3.1.2. Droplet Size, Polydispersity Index and Zeta Potential

Visual observation was used to gain preliminary information regarding the quality of the formulations; however, this qualitative assessment was not used as a standalone method but was systematically combined with the droplet size, polydispersity index and zeta potential measurements obtained immediately after emulsification (Table 4). The consistency between visual appearance, emulsification time and colloidal characteristics supports the reliability of the self-emulsification assessment.

Table 4.

Droplet size, polydispersity index and zeta potential values of the prepared SEDDS mixtures. Results are expressed as mean ± SD (n = 3).

Sample	Droplet Size ± SD (nm)	PDI ± SD	Zeta Potential ± SD (mV)
C1
1 V	366.5 ± 8.2	0.585 ± 0.018	−15.53 ± 0.49
2 V	368.6 ± 20.0	0.486 ± 0.050	−24.00 ± 1.54
3 V	85.4 ± 3.3	0.981 ± 0.008	−9.82 ± 0.37
4 V	33.7 ± 0.6	0.572 ± 0.015	−12.13 ± 0.32
5 C	364.8 ± 4.1	0.443 ± 0.003	−17.97 ± 0.31
C2
1 V	224.9 ± 3.7	0.456 ± 0.059	−17.60 ± 0.20
2 V	96.4 ± 0.6	0.293 ± 0.007	−15.10 ± 0.79
3 V	25.0 ± 0.1	0.268 ± 0.021	2.59 ± 2.02
4 V	100.6 ± 0.5	0.453 ± 0.013	−9.38 ± 0.22
5 C	33.3 ± 0.2	0.396 ± 0.005	−8.89 ± 1.07
C3
1 V	372.0 ± 26.5	0.779 ± 0.021	−26.87 ± 6.70
2 V	443.4 ± 27.8	0.497 ± 0.023	−23.80 ± 0.27
3 V	129.2 ± 5.8	0.979 ± 0.020	−12.03 ± 0.15
4 V	22.6 ± 1.0	0.562 ± 0.021	10.45 ± 0.95
5 C	417.0 ± 5.6	0.489 ± 0.009	−18.37 ± 0.23
C4
1 V	237.0 ± 8.5	0.517 ± 0.083	−18.00 ± 0.36
2 V	42.8 ± 0.8	0.494 ± 0.017	−15.43 ± 0.55
3 V	16.3 ± 0.1	0.221 ± 0.006	−0.16 ± 0.12
4 V	116.5 ± 2.8	0.344 ± 0.046	−8.82 ± 0.30
5 C	19.2 ± 1.0	0.288 ± 0.016	−4.38 ± 0.52

From visual assessment, A- and B-graded samples formed either SNEDDSs or SMEDDSs, while D-graded samples exhibited a bigger droplet size, but were still within the submicron range. The zeta potential was negative in almost all cases (ranging from −0.1623 mV to −26.867 mV (detailed results can be found in Supplementary Materials Table S3). Based on the connection between the visually observable properties and the measured characteristics of the emulsions, the grading system could be used as a screening tool to rapidly identify formulations with acceptable self-emulsifying behavior and to exclude formulations with poor performance, thereby reducing unnecessary experimental workload.

3.1.3. Characterization of the Obtained Design Spaces

The effect of various components on the measured droplet size, PDI, and zeta potential was evaluated by response surface methodology using TIBCO Statistica, v. 13 (TIBCO Software Inc, Palo Alto, CA, USA). The interactions between the oil phase and surfactants/co-surfactants (AB and AC, respectively) were also evaluated, but the current experimental setup did not enable the estimation of the surfactant–co-surfactant (BC) interactions. The significance of the effects was evaluated with ANOVA tests. Regarding the response variables, the best fitting model was quadratic, and the corresponding polynomial equations of the fitted models are displayed in Table 5. For the purpose of the further analyses and visualization, ternary diagrams with two-dimensional contour plots were made.

Table 5.

Coefficients of the fitted quadratic model. Coefficients marked with ‘*’ show significant effect (p < 0.05). The applied model could not estimate the effect of BC.

Comp.	Variable	A	B	C	AB	AC	R2	adjR2	MS Res.
1	Droplet size	−29.0266	114.2129	385.168	-	911.3281	0.8909	0.5637	12732.59
	PDI	0.3962	0.9714	0.537667	−0.744	-	0.8493	0.3973	0.029992
	Zeta potential	−14.3722	−10.1791	−15.7468	-	−37.4699	0.9862	0.9448	1.671112
2	Droplet size	17.51843	−2.99135	19.32273	-	−1.19991	0.8703	0.4812	3323.411
	PDI	* 0.129482	* 0.26788	* 0.455258	* 1.01503	-	0.9998	0.9991	7.3 × 10−6
	Zeta potential	* −21.3737	2.6981	* −17.5368	-	17.9269	0.9994	0.9976	0.146652
3	Droplet size	−88.482	160.509	392.145	-	1327.301	0.8963	0.5854	14880.33
	PDI	0.225405	0.968785	0.726975	−0.388768	-	0.8166	0.2665	0.035728
	Zeta potential	33.021	−12.506	* −27.15	-	−109.211	0.9967	0.9866	2.948558
4	Droplet size	219.711	1.525	228.127	-	−795.512	0.9173	0.6694	2886.633
	PDI	0.475133	0.205267	0.491667	−0.10667	-	0.8695	0.4781	0.008696
	Zeta potential	−17.7653	1.2759	−17.1371	-	14.9747	0.8766	0.5063	27.30139

Coefficients represent: A—Oil phase, B—emulsifier, C—co-emulsifier.

It is an important observation that while droplet size and zeta potential were mainly influenced by the interaction between the oil phase and the low-HLB co-surfactant (AC), PDI was mostly influenced by the interaction of the oil phase with the high-HLB surfactant (AB), while the other interactions showed a smaller effect. There is also a difference in the positive and negative effects of the interactions depending on the combined surfactants. In case of samples containing Tween 80 (e.g., compositions 1 and 3), increasingly smaller droplets were observed at a higher oil content, while an increasing amount of surfactant or the oil/co-surfactant (AC) interaction greatly increases droplet size but concurrently decreases the zeta potential. To the contrary, in the case of the Tween 20-containing samples (e.g., compositions 2 and 4) the surfactant content (B) had a minimal influence on droplet size while AC interaction was associated with a highly negative coefficient of droplet size and a relatively high positive coefficient of zeta potential. A similar difference can be observed between the co-surfactants, since Span 80 (compositions 1 and 2) exhibited a bigger impact on droplet size and zeta potential than Span 20 (compositions 3 and 4). Moreover, their interaction with the oil phase and presumably with the surfactant caused overall differences in the coefficients, which may be in connection with the role of co-surfactants in stabilizing the formulation by assisting in the reduction of interfacial tension, which is supported by the negative coefficient of the AB interaction with PDI, which results in a more homogenous droplet size distribution. These results were in accordance with the effects of the components published in other articles [43,44,45,46]. The best model fittings were observed for composition 2, which greatly helped to identify significant coefficients (see Table 5). Further statistical results can be found in Tables S4–S7, and in Compositions in the Supplementary Material.

The design space was determined on the basis of the following limits: droplet size < 200 nm, PDI < 0.4 and zeta potential < −10 mV (Figure 1). These thresholds were selected based on the intended oral delivery performance and formulation robustness. Droplet size was limited to <200 nm, as previous studies on peptide- and protein-loaded lipid-based nanocarriers have demonstrated a strong correlation between smaller droplet sizes and enhanced mucus permeation and intestinal uptake, whereas larger droplets exhibit significantly reduced permeability. In particular, Griesser et al. reported that particles exceeding 200 nm exhibited only minor permeation across the mucus layer [15,44]. A PDI threshold of <0.4 was applied to ensure an acceptable degree of size uniformity, while accounting for the intrinsic variability introduced by the drug–HIP complex loading. Although a PDI below 0.3 is often cited as indicative of highly monodisperse systems, previous studies have shown that protein-based HIP complexes and drug incorporation into lipid-based nanocarriers can inherently increase polydispersity. Based on these observations, a PDI limit of 0.4 was considered a literature-supported compromise, allowing inclusion of drug-loaded systems that remain physically stable and reproducible [45,47]. For zeta potential, a limit of <−10 mV was selected as an indicator of electrostatic stabilization, reducing the likelihood of droplet aggregation. In addition, negatively charged droplets are considered advantageous for intestinal absorption due to reduced electrostatic interactions with the negatively charged mucus layer [44,47]. All 2D contour plots with their preliminary determined acceptance criteria can be found in Supplementary Material Figures S2–S5.

Figure 1.

2D contour plots with the chosen limits of (a) droplet size (blue), (b) PDI, (pink) and (c) zeta potential (orange) for the design space (composition 2).

3.1.4. Influence of HLB Value on the Size of the Design Space

The size of the design space considerably differs between compositions, while composition 3 did not result in a design space by the applied criteria (Figure 2). This could be due to the differences in the HLB values of the surfactants applied. To further evaluate this effect, the correlation between the HLB difference and the area of the obtained design space was investigated. The area of the design space was determined with the help of the built-in area measurement tool of the Inkscape software by setting the edited images to a standard size. The percentage value of the process design space was calculated as a percentage of the total space (Table 6).

Figure 2.

The acquired design spaces (marked area) for every composition of mixtures and the selected points for validation (1–4, (a–d)) and their place according to the applied three-factor-constrained mixture design (e).

Table 6.

Percentage value of acquired design space area for each emulsifier combination.

Surfactant/Co-Surfactant	Comp.	HLB	ΔHLB	Total %
Tween 80/Span 80	1	15.0/4.1	10.9	15.872
Tween 20/Span 80	2	16.7/4.1	12.6	37.289
Tween 80/Span 20	3	15.0/9.0	6.0	0
Tween 20/Span 20	4	16.7/9.0	7.7	8.034

A clear trend can be observed between the difference in HLB values and the resulting design space area, as the difference increases, the percentage of the design space area also increases. It was the largest (37.289%) in the case of composition 2, where the difference in HLB for the surfactant and co-surfactant is 12.6, while composition 3, wherein the difference in HLB was only 6.0, did not yield any design space area with the criteria set. These findings were also tested via response surface methodology, with the help of a factorial design, where the HLB values of the surfactants and the co-surfactants were used as factors and the percentage value of the design space area was the response (Figure 3). It can be observed from Equation (4) that the HLB of the co-surfactant has a more considerable effect on the design space, while the interaction of the two values was negligible.

$$y=15.2988+7.3628 x_1-11.2818{ x}_2$$ (4)

Figure 3.

The response surface of the effect of HLB value difference on the acquired design space area. HLB1 refers to the surfactant and HLB2 to the co-surfactant. Blue dots represent meesured values as stated in Table 6.

The fitting of the model had an R² = 0.9419 and an adjusted R² = 0.82571, which means there was a good correlation between the respective variables.

This finding expands the previous finding that using emulsifiers with one value above HLB 12 and the other below it is favorable [24], and concludes that a minimum difference of 6 between the HLB values is required to achieve proper self-emulsification.

3.1.5. Validation of the Design Space

The optimized formulations were selected from the predicted design space based on predefined acceptance criteria (Section 3.1.3.) and were subsequently used to validate the DoE model by comparing the predicted and experimentally observed responses. External validation was conducted by selecting 2–2 points in each design spaces randomly by hand (Table 7). The location of each point within the design spaces can be seen in Figure 2a–d and the values predicted by the DoE model within the established design spaces can be seen in Supplementary Material Table S9.

Table 7.

Compositions of SEDDSs for design space validation.

Composition	Code	Component	%
C1/Point 1# Replicate 1 and 2	1D11/1D12	Miglyol 810	18.5
		Tween 80	71.0
		Span 80	10.5
C1/Point 2# Replicate 1 and 2	1D21/1D22	Miglyol 810	19.5
		Tween 80	66.0
		Span 80	14.5
C2/Point 1# Replicate 1 and 2	2D11/2D12	Miglyol 810	18.5
		Tween 20	65.5
		Span 80	16.0
C2/Point 2# Replicate 1 and 2	2D21/2D22	Miglyol 810	14.0
		Tween 20	63.5
		Span 80	22.5
C4/Point 1# Replicate 1 and 2	4D11/4D12	Miglyol 810	11.5
		Tween 20	66.5
		Span 20	22.0
C4/Point 2# Replicate 1 and 2	4D21/4D22	Miglyol 810	18
		Tween 20	66
		Span 20	16

Overall, in most cases, the observed values fell in the 95% confidence intervals of the predicted values (Figure 4a–c). A significant difference was observed for samples 1D2 and 2D2 in their droplet size and zeta potential, and for 1D1 and 4D1 in their PDI (detailed statistical results can be found in Supplementary Material Tables S10 and S11 and Figure S8a–c), which corresponds well with the grading results (Table 8).

Figure 4.

Factorial ANOVA comparison of the observed and predicted values ((a)—droplet size; (b)—PDI; (c)—zeta potential) of the validation samples. Data marked with ‘*’’ show a significant difference from the predicted value (p < 0.05). Vertical bars denote 0.95 confidence intervals.

Table 8.

Grading and self-emulsification times (SEs) of the design space validation samples.

Sample	Grading	SE Time (s)	Sample	Grading	SE Time (s)	Sample	Grading	SE Time (s)
1D11	B	34 ± 6	2D11	A	18 ± 6	4D11	A	23 ± 5
1D12	A	28 ± 7	2D12	A	21 ± 4	4D12	A	17 ± 7
1D21	C	88 ± 9	2D21	D	129 ± 16	4D21	A	38 ± 6
1D22	C	73 ± 6	2D22	C	114 ± 8	4D22	B	45 ± 4

Results are expressed as mean ± SD (n = 3).

Although there are some shortcomings, these can be attributed to the selection of validation points, since the goal was also to test the robustness of the acquired design space, and compositions farther away from the center were selected. However, the model itself can be used to estimate the characteristics of optimized liquid SEDDSs, which holds the possibility of further improvement in the future.

3.2. Characterization of HIP Complex-Loaded SEDDSs

According to the review by Pandey and Kohli, the solubility of the API depends more on the overall solubilizing power of the SEDDS, rather than the individual components [20]. Therefore, in this investigation the emphasis was on the solubilizing capacity of the mixtures. Samples used for the validation were also used for the HIP complex loading, as the differences in the ratio of components may reveal the influence of this parameter on the solubility of HIP complex.

The investigations of the HIP-loaded SEDDSs (Table 9) showed an increase in droplet size with increasing drug load. This behavior can be attributed to the incorporation of the HIP complex into the formulation, which could promote the formation of larger droplets and may also result in the presence of partially or non-solubilized complex particles at higher loads. Moreover, an increase and a greater variability in PDI values were observed, indicating a broader droplet size distribution upon drug incorporation. Similar phenomena were reported previously by other studies as well. Hetényi et al. found that SEDDSs loaded with an insulin– and desmopressin–HIP complex exhibited higher PDI values compared to blank systems, which was attributed to the higher molecular weight of insulin (5.8 kDa) and to the structural characteristics in the case of desmopressin [47]. In the present study, lysozyme has a substantially higher molecular weight (14.4 kDa) than the proteins investigated in the aforementioned report, which further supports the possibility that molecular size and structural complexity contribute to the observed increase in PDI upon HIP loading. In addition, incorporation of azithromycin into blank SEDDS formulations resulted in increased droplet size and PDI, possibly due to the entering of the drug molecule into the oil–water interface [45]. In contrast, the absolute values of the zeta potential generally decreased with increasing HIP load. As the HIP complex has an overall net-zero charge, it is possible that the increasing amount incorporated masks the surface charge of the droplets or alters interfacial interactions, especially when the drug load exceeds its solubility in the formulation. A comparable trend was reported for SNEDDSs loaded with an insulin–SPC complex, where the interactions between the complex and the delivery system led to reduced absolute zeta potential values [46,48].

Table 9.

Droplet size, polydispersity index and zeta potential values of the selected HIP-loaded SEDDSs.

Sample	Droplet Size ± SD (nm)	PDI ± SD	Zeta Potential ± SD (mV)
1D11
5 mg/g	52.3 ± 2.9	0.749 ± 0.031	−13.266 ± 0.208
10 mg/g	51.6 ± 3.4	0.751 ± 0.025	−11.633 ± 0.153
15 mg/g	65.2 ± 7.2	0.686 ± 0.058	−0.0093 ± 0.050
1D21
5 mg/g	448.2 ± 22.4	0.561 ± 0.026	−13.833 ± 0.208
10 mg/g	491.7 ± 3.8	0.581 ± 0.005	−10.700 ± 0.173
15 mg/g	302.8 ± 19.1	0.544 ± 0.057	−0.002 ± 0.008
2D11
5 mg/g	93.5 ± 16.6	0.365 ± 0.092	−0.010 ± 0.084
10 mg/g	281.6 ± 81.0	0.359 ± 0.055	−0.007 ± 0.041
15 mg/g	626.5 ± 64.1	0.660 ± 0.044	−0.033 ± 0.008
2D22
5 mg/g	242.0 ± 9.7	0.421 ± 0.039	−13.467 ± 0.208
10 mg/g	249.0 ± 23.5	0.422 ± 0.049	−0.008 ± 0.052
15 mg/g	3921.7 ± 387.6	1.000 ± 0.000	−5.850 ± 0.245
4D11
5 mg/g	131.5 ± 8.1	0.739 ± 0.025	−14.967 ± 0.153
10 mg/g	352.2 ± 11.2	0.507 ± 0.042	−11.200 ± 0.656
15 mg/g	450.6 ± 51.3	0.513 ± 0.007	0.027 ± 0.069
4D21
5 mg/g	23.1 ± 0.3	0.321 ± 0.006	−0.025 ± 0.051
10 mg/g	34.7 ± 0.2	0.489 ± 0.005	−0.018 ± 0.054
15 mg/g	154.7 ± 70.2	0.291 ± 0.059	−3.367 ± 0.076

The highest achievable LYZ/SDS HIP loading was obtained in case of the 1D11 sample, while still maintaining a favorable droplet size (droplet size: 51.550 ± 3.367) and zeta potential (−11.633 ± 0.153), whereas the PDI value (0.751 ± 0.025) exceeded the predefined acceptance criteria. However, the elevated PDI can be expected, as was discussed previously. Despite the increased PDI, the formulation maintained a droplet size well below 200 nm to enable efficient mucus permeation and exhibited sufficient electrostatic stabilization, indicated by the zeta potential. The 10 mg of HIP complex corresponds to a LYZ load of 8.611 mg/g. The HIP-loaded SEDDSs showed no signs of instability or drug precipitation after 24 h while stored in a refrigerator (2–8 °C) (detailed statistical results can be found in Supplementary Material Table S12).

3.3. In Vitro Drug Release Study of LYZ/SDS HIP Complex-Loaded SEDDSs

To test the in vitro drug release of the HIP-loaded SEDDSs, one formulation was selected from each composition exhibiting the most favorable overall characteristics (1D11, 2D11 and 4D11), taking the predefined criteria into consideration, while also acknowledging the impact of drug loading as described in Section 3.2. Following the initial centrifugation of the emulsions, the unencapsulated amount of LYZ was determined from the supernatants (Equation (3)) as 30.48% for 1D11 and 33.42% for 4D11. In the case of sample 2D11, all of the drug load was released upon the first centrifugation, which could be due to its higher droplet size and near-zero zeta potential (Table 9), and thus overall lower stability. Accordingly, 2D11 was excluded from the release study. On the other hand, the dissolution of the other two compositions (Figure 5) follows first-order kinetics with good correlation (R² = 0.9223 for 1D11 and R² = 0.9886 for 4D11) and a dissolution rate constant (k) of 0.3994 and 0.3847 for the 1D11 and 4D11 samples, respectively.

Figure 5.

In vitro release profiles of LYZ/SDS HIP-loaded SEDDSs. Indicated values are means of three experiments ± SD.

Drug release from SEDDSs is mainly driven by simple diffusion from the oily phase into the aqueous phase. According to Bernkop-Schnürch and Jalil the only parameter controlling the release is the partition coefficient (log D) between the lipophilic phase of the SEDDS and the aqueous release medium [49]. Hydrophilic components in SEDDSs can facilitate drug release into the medium while hydrophobic components may slow it down [50]. This can explain the difference between the release kinetics of the two compositions. The higher Tween/Span ratio (Table 7) for 1D11 may be attributed to faster drug release. HIP complexation of the drug, while in our case necessary to incorporate and protect the hydrophilic LYZ, may also contribute to a more sustained release, since the dissociation of the complex takes time. The more stable the HIP complex, the more time it takes to release. However, the stability of HIPs is much lower in intestinal fluids and similar release mediums than in SEDDSs, due to their ionic strength and pH (PBS pH 6.8, 137 mM NaCl) [36].

Nevertheless, since there is no available standard pharmacopeial in vitro dissolution test for SEDDSs, the demand to develop standardized studies is significant. The current most-used membrane diffusion and sample-and-separate methods both have their limitations [51,52]. With the former it is quite difficult to determine the real release kinetic since the membrane has a huge effect on release control. While with the latter, the separation method can significantly influence the release profile. In addition, in both cases, sink conditions are often violated [49].

3.4. Physical Stability Study of LYZ/SDS HIP Complex-Loaded SEDDSs

To evaluate the physical stability under simulated gastrointestinal conditions, two formulations exhibiting the most favorable overall characteristics and in vitro release profiles (1D11 and 4D11) were selected for further investigation.

No increase was observed in droplet size during the study, instead a gradual decrease was detected over time (Figure 6a,d). This may be attributed to the progressive release of the HIP complex from the droplets, which is correlated with the described in vitro release behavior of the formulations, resulting in a reduction in the droplet diameter. The polydispersity index remained essentially unchanged for formulation 1D11 in both media (Figure 6b,e). In contrast, 4D11 showed a slight increase in PDI, which may be associated with a more pronounced reduction in droplet size and the resulting redistribution of droplet populations, leading to increased variability in size distribution. The differences in zeta potential observed (Figure 6c,f) across media can be explained by pH-dependent surface charge modulation and ionic strength-induced changes. In an acidic medium (0.1 M HCl, pH 1.4) near-neutral or slightly positive zeta potential values were measured, which may result from the protonation and neutralization of the negatively charged hydroxyl groups of fatty acid chains in the oil and surfactants. In PBS (pH 6.8), negative zeta potential values were observed (between −0.66 and −4.52 mV); however, their magnitude was lower than that in distilled water (Table 9), likely due to the higher ionic strength of the buffer. Similar behaviors of zeta potential have been reported in previous studies [45,53]. Importantly, the zeta potential values remained stable over time in both media.

Figure 6.

Physical stability studies for SEDDS formulations 1D11 and 4D11 in 0.1 M HCl at pH 1.4 ((a–c)—1D1G and 4D1G) and in PBS at pH 6.8 ((d–f)—1D1P and 4D1P). Indicated values are means of three experiments ± SD.

The absence of droplet size increase or aggregation, together with the stable zeta potential values throughout the study, supports the physical stability of the formulations under the simulated gastrointestinal conditions. Nevertheless, although the emulsified systems remained colloidally stable in an acidic medium, the incorporation of the HIP-loaded SEDDSs into an enteric-coated capsule or tablet may be considered in future developments to further protect the protein from potential gastric degradation and to ensure targeted release and absorption in the small intestine.

4. Conclusions

In this study, systematic investigation of complex self-emulsifying drug delivery systems was conducted with the purpose of contributing to better understanding of this drug delivery method. Based on the results it can be concluded that a difference of at least 6.0 between the HLB values of the applied surfactants and co-surfactants must exist to obtain a system suitable for self-emulsification with the chosen characteristics. The secondary focus was to develop SEDDSs with sufficient stability and capability to solve LYZ/SDS HIP complexes. Among the HIP-loaded SEDDS formulations, 1D11 exhibited the most favorable overall characteristics and was therefore identified as the optimal candidate. The highest drug complex loading was also achieved with this formulation, while providing the best balance between drug incorporation and the critical quality attributes investigated. Although increased drug loading was associated with broader droplet size distribution, this trade-off was considered acceptable, as droplet size and zeta potential remained within the predefined target ranges. Importantly, this observation highlights an inherent formulation limitation of HIP-loaded SEDDSs and provides a valuable direction for future optimization strategies.

Additionally, an in vitro release study was performed to gain insight into the release behavior of lysozyme from HIP-loaded SEDDSs. The release apparently followed first-order kinetics, with approximately 80% of the loaded protein being released within 6 h. Despite the well-recognized limitations and methodological challenges associated with in vitro release testing of SEDDSs, these results may provide valuable comparative information and suggest dissociation-driven release from the HIP SEDDS formulation. Moreover, physical stability testing confirmed that the selected formulations maintained their nanoscale droplet size and colloidal integrity under simulated gastrointestinal conditions.

The characterization of HIP-loaded SEDDSs is a step toward the oral administration of lysozyme and this method also allows the modelling of peptide biopharmaceutical delivery. Although achieving a higher drug load is still necessary, it provides a good basis to develop solid SEDDSs and prepare a LYZ-loaded tablet dosage form in the future.

Abbreviations

The following abbreviations are used in this manuscript:

ANOVA	Analysis of Variance
BCS	Biopharmaceutics Classification System
COPD	Chronic Obstructive Pulmonary Disease
CQA	Critical Quality Attributes
DLS	Dynamic Light Scattering
DoE	Design of Experiments
HIP	Hydrophobic Ion Pair(ing)
HLB	Hydrophilic–Lipophilic Balance
LCT	Long-Chain Triglycerides
LSD test	Least Significant Difference
LYZ	Lysozyme
MCT	Medium-Chain Triglycerides
NSAID	Non-Steroidal Anti-Inflammatory Drug
O/W	Oil-in-Water
PBS	Phosphate-Buffered Saline
PDI	Polydispersity Index
pI	Isoelectric Point
QbD	Quality-by-Design
RSM	Response Surface Methodology
SDS	Sodium Dodecyl Sulfate
SEDDSs	Self-Emulsifying Drug Delivery Systems
SMEDDS	Self-Microemulsifying Drug Delivery System
SNEDDS	Self-Nanoemulsifying Drug Delivery System

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/pharmaceutics18020275/s1, Table S1: 2² full factorial design of SEDDS mixtures; Table S2: Composition of SEDDSs based on the mixture design experimental plan for each composition (%, w/w); Table S3: Droplet size, polydispersity index, zeta potential values and grading of the prepared SEDDS mixtures. ± SD (n = 3); Table S4: Detailed statistical results of the 3-factor-constrained mixture design for composition 1; Table S5: Detailed statistical results of the 3-factor-constrained mixture design for composition 2; Table S6: Detailed statistical results of the 3-factor-constrained mixture design for composition 3; Table S7: Detailed statistical results of the 3-factor-constrained mixture design for composition 4; Table S8: Detailed statistical results of the 2² factorial design to estimate the effect of HLB value difference on the acquired design space area; Table S9: Compositions and predicted values of SEDDSs for design space validation; Table S10: Droplet size, polydispersity index, zeta potential values and grading of the selected validation SEDDS samples. ± SD (n = 3); Table S11: Post hoc LSD tests of the observed and predicted droplet size, PDI, and zeta potential values; Table S12: Droplet size, polydispersity index and zeta potential values of the selected HIP-loaded SEDDS. ± SD (n = 3); Table S13: Droplet size, PDI and zeta potential values for SEDDS formulations 1D11 and 4D11 in 0.1 M HCl pH 1.4 (1D1G and 4D1G) and in PBS pH 6.8 (1D1P and 4D1P). Indicated values are means of three experiments ± SD (DS-droplet size, PDI-polydispersity index, ZP-zeta potential); Figure S1: the applied three-factor-constrained mixture design; Figure S2: 2D contour plots with the chosen limits of (a) droplet size (blue), (b) PDI (pink) and (c) zeta potential (orange) for the design space in the case of composition 1; Figure S3: 2D contour plots with the chosen limits of (a) droplet size (blue), (b) PDI (pink) and (c) zeta potential (orange) for the design space in the case of composition 2; Figure S4: 2D contour plots with the chosen limits of (a) droplet size (blue), (b) PDI (pink) and (c) zeta potential (orange) for the design space in the case of composition 3; Figure S5: 2D contour plots with the chosen limits of (a) droplet size (blue), (b) PDI (pink) and (c) zeta potential (orange) for the design space in the case of composition 4; Figure S6: The acquired design spaces (marked area) for every batch of mixtures and the selected points for validation (C1–C4, a–d); Figure S7: The response surface of the effect of HLB value difference on the acquired design space area; Figure S8: Factorial ANOVA comparison of the observed and predicted values (a—droplet size; b—PDI; c—zeta potential) of the validation samples. Data marked with ‘*’ shows significant difference from the predicted value. Vertical bars denote 0.95 confidence intervals. Detailed statistical results, observes vs. predicted data plots and residual plots of the iterative determination of response surface equations.

Author Contributions

Conceptualization, K.K. and T.S.; methodology, K.K., T.S., E.R. and G.K.; validation, G.K., E.R., M.D. and T.S.; formal analysis, M.D. and N.A.; investigation, G.K., M.D. and N.A.; resources, T.S.; data curation, M.D. and T.S.; writing—original draft preparation, M.D.; writing—review and editing, K.K. and T.S.; visualization, M.D.; supervision, K.K. and T.S.; project administration, T.S.; funding acquisition, T.S. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data are available upon request from the corresponding author. The datasets presented in this article are not readily available because the data are part of an ongoing study that requires further time until the raw data can be transformed to publicly accessible file formats, and the file names and nomenclature used in the various measurements would be unified and the correct metadata could be provided to enable the correct submission of the raw files into a publicly available data repository. Until that time, requests to access the datasets should be directed to the corresponding author (sovany.tamas@szte.hu).

Conflicts of Interest

The authors declare no conflicts of interest.

Funding Statement

Project no. TKP2021-EGA-32 has been implemented with the support provided by the Ministry of Culture and Innovation of Hungary from the National Research, Development and Innovation Fund, financed under the TKP2021-EGA funding scheme. The publication was supported by University of Szeged OA Fund: Grant ID: 8476.