Preparation and Evaluation of Eudragit L100-PEG Proliponiosomes for Enhanced Oral Delivery of Celecoxib

PEGylated Eudragit L100 (ELP)-containing proliponiosomes (PLNs) were developed for improved oral delivery of celecoxib (CXB). The successful introduction of PEG 2000 or 5000 to Eudragit L100 (EL) was confirmed via proton nuclear magnetic resonance analysis of which calculated molar substitution ratio of PEG to EL was 36.0 or 36.7, respectively. CXB, ELP, phospholipid, and non-ionic surfactants were dissolved in dimethyl sulfoxide and lyophilized to produce CXB-loaded PLNs (CXB@PLNs). The physical state of CXB@PLNs was evaluated using differential scanning calorimetry and powder X-ray diffractometry, which revealed that crystalline CXB was transformed into amorphous form after the fabrication procedure. The reconstitution of CXB@PLNs in aqueous media generated CXB-loaded liponiosomes with nano-sized mean diameters and spherical morphology. CXB@PLNs displayed enhanced dissolution rate and permeability compared to CXB suspension. In vivo pharmacokinetic studies performed on rats demonstrated the improved oral bioavailability of CXB@PLNs compared to that of CXB suspension. No serious systemic toxicity was observed in the blood biochemistry tests performed on rats. These results suggest that the developed PLNs could be promising oral delivery systems for improving the bioavailability of poorly water-soluble drugs, such as CXB.


Introduction
Celecoxib (CXB) is a nonsteroidal anti-inflammatory drug (NSAID) indicated for osteoarthritis, rheumatoid arthritis, ankylosing spondylitis, acute pain, primary dysmenorrhea, and familial adenomatous polyposis [1,2]. Given the selectivity to cyclooxygenase (COX)-2, CXB provides higher efficacy with less gastrointestinal (GI) side effects compared to non-selective COX inhibitors [3]. the surface of a 200-mesh carbon-coated copper grid and negatively stained with uranyl acetate (UA) solution (2%). The sample observation was conducted at an accelerating voltage of 80 kV. The average particle size with polydispersity index (PDI) and zeta potential values of the reconstituted LNs were measured by dynamic light scattering (DLS) and laser Doppler electrophoresis, respectively, using a Zetasizer Ultra (Malvern Instruments Ltd., Malvern, UK). The encapsulation efficiency (EE) of LNs was determined after removing unloaded CXB from the reconstituted LNs using a syringe filter (pore size: 0.45 µm; Minisart RC15; Sartorius). The CXB concentration in filtrate was quantified by the RP-HPLC method presented in Supplementary Materials (Section S1.1).

Release Test
The drug-releasing property of CXB@PLNs was evaluated by using a dialysis method. The formulations (CXB suspension, F1, and F2 containing 200 µg of CXB) were loaded into a regenerated cellulose tubing (MWCO: 2 kDa; CelluSep H1; Membrane Filtration Products). The tubes were immersed into release media (20 mL) of various pH conditions, including pH 1.2 (containing 40 mg of sodium chloride and 140 µL of 1 M hydrochloric acid solution), 4.5 (containing 24 mg of sodium acetate and 48 mg of acetic acid), 6.8 (containing 156 mg of sodium phosphate monobasic; adjusted with 1 M sodium hydroxide solution), and 7.5 (containing 156 mg of sodium phosphate monobasic; adjusted with 1 M sodium hydroxide solution), all of which were supplemented with SDS (0.4%, w/v) to maintain sink condition. CXB suspension was prepared by grinding drug powder with DDW containing 0.5% (w/v) sodium carboxymethylcellulose (NaCMC) using a sterilized mortar and pestle. The experiment was conducted in a shaking water bath at 37 • C (50 rpm). Aliquots (0.3 mL) of the buffers were collected at predetermined time points (0.5, 1, 2, 4, 6, 8, and 24 h), and an equal volume was replenished after each sampling. The collected samples were diluted with ACN and analyzed using the RP-HPLC method in Supplementary Materials (Section S1.1).
The Caco-2 permeability assays were conducted using a previously reported method by our group [2]. More detailed information can be found in Supplementary Materials.

Pharmacokinetic Study and Blood Biochemistry Test
The pharmacokinetic study was performed in a Sprague-Dawley rat model previously reported by our group [2]. Further information regarding the animal and experimental protocol is presented in Supplementary Materials. Pharmacokinetic parameters, including area under the plasma concentration-time curve from time zero to 24 h (AUC), peak plasma concentration (C max ), and time to reach C max (T max ), were calculated using non-compartmental analysis (WinNonlin, version 3.1, NCA 201; Pharsight, Mountain View, CA, USA). Relative bioavailability (%) was calculated by dividing the average AUC value of each formulation with that of CXB suspension.
The blood biochemistry test was also performed in Sprague-Dawley rats (body weight: 235 ± 5 g; Orient Bio). Blood samples were collected at 24 h after an oral administration of F1 or F2 (as Pharmaceutics 2020, 12, 718 5 of 14 CXB, 2 mg/kg) using the cannulation method presented in Supplementary Materials. The serum was separated by incubating blood samples at room temperature for 1 h and centrifuging at 16,000× g for 10 min. The levels of serum alanine transaminase (ALT), albumin (ALB), total protein (TP), blood urea nitrogen (BUN), and serum creatinine (SCr) were analyzed using Fuji Dri-Chem 3500s (Fujifilm Corp., Tokyo, Japan).

Statistical Analysis
All assays were conducted at least three times of which results were presented as the mean ± standard deviation (SD). The statistical analysis (one-way analysis of variance with Tukey's multiple comparison test) was performed using an SPSS statistics software (Version 21.0; IBM Corp, Armonk, NY, USA).

Synthesis and Characterization of ELPs
ELPs were synthesized via EDC/NHS coupling reaction, where the activated carboxylic acid group of EL was conjugated with the amine group of mPEG-NH 2 ( Figure 1a). Although it is well known that EDC/NHS reagents can form amide bonds at room temperature efficiently [28], ELPs were obtained only at the mildly heated condition. In our preliminary study, the reaction was also performed at room temperature only to result in precipitation during dialysis against DDW. However, ELPs prepared at 45 • C appeared to be translucent with no precipitation during the dialysis, which implies the successful conjugation of PEGs, considering the insolubility of EL in DDW.

Statistical Analysis
All assays were conducted at least three times of which results were presented as the mean ± standard deviation (SD). The statistical analysis (one-way analysis of variance with Tukey's multiple comparison test) was performed using an SPSS statistics software (Version 21.0; IBM Corp, Armonk, NY, USA).

Synthesis and Characterization of ELPs
ELPs were synthesized via EDC/NHS coupling reaction, where the activated carboxylic acid group of EL was conjugated with the amine group of mPEG-NH2 ( Figure 1a). Although it is well known that EDC/NHS reagents can form amide bonds at room temperature efficiently [28], ELPs were obtained only at the mildly heated condition. In our preliminary study, the reaction was also performed at room temperature only to result in precipitation during dialysis against DDW. However, ELPs prepared at 45 °C appeared to be translucent with no precipitation during the dialysis, which implies the successful conjugation of PEGs, considering the insolubility of EL in DDW.  Figure S2). Chemical shifts are assigned to the corresponding hydrogen atoms (red font).
The synthesis of ELPs was also confirmed using 1 H-NMR analysis. As shown in Figure 1b and Supplementary Materials Figure S1, the representative signal of PEG (d; 3.5 ppm) can be observed in both ELP spectra, where the acidic proton signal of EL (12.4 ppm) also disappeared (Supplementary Materials Figure S2). The PEG content of ELPs  Figure S2). Chemical shifts are assigned to the corresponding hydrogen atoms (red font).
The synthesis of ELPs was also confirmed using 1 H-NMR analysis. As shown in Figure 1b and Supplementary Materials Figure S1, the representative signal of PEG (d; 3.5 ppm) can be observed in both ELP spectra, where the acidic proton signal of EL (12.4 Figure S2). The PEG content of ELPs was quantified by using 1 H-NMR spectra of EL/PEG physical mixtures. The peak area ratio of representative EL and PEG signals (Y; methyl and methylene peaks of EL [a + b; 0.5-1.5 ppm] vs. ethylene peak of PEG [d; 3.5 ppm]) and corresponding weight ratio (X) were used to construct standard curve: Y = 5.226·X + 0.047 (R 2 = 1.000). The calculated PEG contents of ELP 2k and ELP 5k were 36.6 ± 0.1% (w/w) and 59.5 ± 0.2% (w/w), respectively. The molar substitution ratios of PEG to EL were almost the same for ELP 2k and ELP 5k , exhibiting 36.0 and 36.7, respectively, which could be attributed to the same molar feed ratios (i.e., approximately 80 PEGs per each EL chain). The substitution yields (%) were calculated to be 45.0% (= 36.0/80 × 100%) and 45.9% (= 36.7/80 × 100%) for ELP 2k and ELP 5k , respectively. Degrees of substitution (%) were determined as the molar ratio of conjugated PEGs to total carboxylic acid groups in each EL chain (approximately 672), which give 5.36% (= 36.0/672 × 100%) and 5.46% (= 36.7/672 × 100%) for ELP 2k and ELP 5k , respectively. To confirm the removal of unconjugated mPEG-NH 2 from the products, ninhydrin staining was performed (Supplementary Materials Figure S3). The contents (%) of unconjugated mPEG-NH 2 in ELP 2k and ELP 5k were estimated to be less than 1% and 2%, respectively.

Preparation and Characterization of CXB@PLNs in Solid State
The composition and visual appearance of CXB@PLNs are presented in Figure 2a, which were fabricated by lyophilization of the DMSO solution of those ingredients. Compared to the theoretical CXB content, the prepared CXB@PLNs exhibited 88.6 ± 3.3% (F1) and 91.0 ± 4.5% (F2) of loading efficiency. SEM observation revealed that CXB powder displayed a stacked plate-like morphology, which is consistent with the appearance of a known CXB polymorph (Form III) ( Figure 2b) [29]. Although ELP 2k -containing CXB@PLNs (F1) displays a more granular surface than those containing ELP 5k (F2), neither exhibited the characteristic shape of CXB powder. These results imply that CXB was converted to the amorphous form (or other crystalline polymorphs) during the preparation process.
The physical state of CXB@PLNs was further investigated using DSC. The DSC thermograms of ELPs, CXB, CXB@PLNs, and other excipients used are shown in Figure 2c. PEG 1000 exhibited a broad melting peak at around 30 • C, which correlates well with the previous report [30]. SPC also showed a subtle endothermic depression at approximately 20-30 • C, which corresponds to its melting temperature [31]. EL displayed a flat baseline, indicating the absence of a distinct phase transition within the tested temperature range. However, both ELP 2k and ELP 5k showed depressions at around 50-60 • C. Considering that the melting temperatures of PEG 2000 and 5000 are 49-52 • C and 60-64 • C, respectively, those endothermic peaks can be explained by the phase transition of the PEG chains in ELPs. The depression patterns of F1 and F2 ranging over 30-50 • C appear to be the combined melting peaks of PEG 1000, SPC, and ELPs with their weight percentages taken into account. Meanwhile, CXB exhibited a sharp endothermic peak at 163 • C, which corresponds to the melting point of Form III polymorph [29]. Interestingly, this characteristic peak disappeared in the DSC curves of F1 and F2, which indicates that the amorphization of CXB occurred during the preparation of CXB@PLNs.
A more detailed investigation of the crystallinity of CXB@PLNs was conducted using PXRD ( Figure 2d). PEG 1000 showed two distinctive diffraction peaks at 19.1 • and 23.2 • , which are derived from the (120) and (032) planes in Miller index notation, respectively [32]. SPC displayed a non-crystalline pattern, which is in good accordance with the gradual depression observed in DSC analysis. EL showed no sign of crystallinity, whereas ELP 5k displayed the characteristic peaks of PEG at 19.0 • and 23.2 • . This difference can be attributed to the introduced PEG chains in ELP 5k . Although ELP 2k showed very subtle elevation at those 2θ values, the almost flat diffractogram indicates the non-crystalline property of ELP 2k . As expected from SEM and DSC data, CXB exhibited a characteristic pattern of Form III polymorph, where sharp peaks at 5.2 • , 10.6 • , 13.0 • , 14.8 • , 16.1 • , 19.7 • , and 21.5 • can be observed [29]. However, CXB@PLNs exhibited diffraction patterns similar to that of PEG 1000 with no characteristic signals of CXB. This result indicates the loss of crystallinity of CXB during the fabrication of F1 and F2, supporting the DSC results.

Characterization of LNs Generated by Reconstitution of CXB@PLNs
CXB@PLNs were designed to be rapidly dispersed in the aqueous milieu, generating nanoscale LNs (Figure 3a). The phospholipid (SPC) and non-ionic surfactants (SHS and tween 80) were incorporated to CXB@PLNs, as they are known to form LNs together [33,34]. SPC is a mixture of diacyl-conjugated glycero-3-phosphocholines of which fatty acid chains mostly consisting of palmitate, stearate, oleate, linoleate, and linolenate [35]. SHS is a PEGylated fatty acid ester that has 15-PEG units conjugated to 12-hydroxystearate. Tween 80 possesses a structure of sorbitan monooleate decorated with 20-PEG units. Due to the amphiphilic structures, these three compounds can serve as not only drug solubilizers but also absorption enhancers [36]. ELPs were added to the PLNs as main PIs, as these polymers exhibited micelle-forming property. Our preliminary study revealed that ELP2k and ELP5k dispersions in DDW exhibited mean diameters of 49.4 ± 28.0 nm (PDI:

Characterization of LNs Generated by Reconstitution of CXB@PLNs
CXB@PLNs were designed to be rapidly dispersed in the aqueous milieu, generating nanoscale LNs (Figure 3a). The phospholipid (SPC) and non-ionic surfactants (SHS and tween 80) were incorporated to CXB@PLNs, as they are known to form LNs together [33,34]. SPC is a mixture of diacyl-conjugated glycero-3-phosphocholines of which fatty acid chains mostly consisting of palmitate, stearate, oleate, linoleate, and linolenate [35]. SHS is a PEGylated fatty acid ester that has 15-PEG units conjugated to 12-hydroxystearate. Tween 80 possesses a structure of sorbitan monooleate decorated with 20-PEG units. Due to the amphiphilic structures, these three compounds can serve as not only drug solubilizers but also absorption enhancers [36]. ELPs were added to the PLNs as main PIs, as these polymers Pharmaceutics 2020, 12, 718 8 of 14 exhibited micelle-forming property. Our preliminary study revealed that ELP 2k and ELP 5k dispersions in DDW exhibited mean diameters of 49.4 ± 28.0 nm (PDI: 0.46 ± 0.18) and 70.2 ± 9.8 nm (PDI: 0.45 ± 0.10), respectively. PEG 1000 was blended to promote the rapid dispersion of CXB@PLNs in aqueous media. Theoretically, PEGs with lower molecular weight may achieve faster dispersion in aqueous conditions. In the preliminary study, however, the CXB@PLNs containing PEG 600 exhibited sticky and paste-like properties after lyophilization (data not shown). 1). The PDI values less than 0.3 indicate that the LNs possess a narrow size distribution [37]. CXB@PLNs containing EL instead of ELPs were also prepared and evaluated based on DLS analysis only to find out that the resulting formulation exhibited incomplete reconstitution in DDW, thus discouraging further investigation (Supplementary Materials Figure S4). The generation of LNs was further confirmed by UA-staining assisted TEM imaging, where both F1-and F2-LNs displayed spherical morphology (Figure 3b). Interestingly, the mean diameters of F1-and F2-LNs estimated from the TEM images (42.0 ± 13.6 nm and 67.7 ± 14.3 nm, respectively) are in good accordance with the number-averaged particle sizes derived from DLS analysis (44.4 ± 4.7 nm and 70.0 ± 12.9 nm, respectively) (Supplementary Materials Figure S5). The zeta potential values of F1-and F2-LNs were 1.87 ± 0.13 mV and -5.41 ± 0.94 mV, respectively ( Table 1). The EE of F1-and F2-LNs determined using 0.45 μm filters exhibited comparable values of 80.8 ± 0.3% and 89.2 ± 2.5%, respectively. Considering the reduced particle size and positively-charged zeta potential, F1-LNs could be more effective in the oral delivery of CXB than F2-LNs [38,39].  Data are presented as mean ± SD (n = 3).

In Vitro CXB Release from CXB@PLNs
The in vitro release test was conducted at various pHs of 1.2 (fasted-state stomach), 4.5 (fed-state stomach), 6.8 (proximal small intestine), and 7.4 (distal small intestine) [40]. The release profiles of CXB suspension, F1, and F2 are presented in Figure 4. The CXB suspension prepared by grinding CXB powder in NaCMC solution was employed as a control. It is worth noting that both F1 and F2 exhibited significantly higher cumulative release than CXB suspension in all tested pH conditions (p < 0.05). These results imply that the bottom-up processing that induces amorphization of CXB is more efficient in enhancing the dissolution rate than the top-down method that does not accompany a change in drug crystallinity (i.e., grinding). The amorphous form of CXB can provide an improved Reconstitution of F1 and F2 in DDW produced LNs with mean diameters of 231.7 ± 1.5 nm and 292.0 ± 3.5 nm of which PDI values were 0.27 ± 0.01 and 0.25 ± 0.01, respectively (Figure 3b and Table 1). The PDI values less than 0.3 indicate that the LNs possess a narrow size distribution [37]. CXB@PLNs containing EL instead of ELPs were also prepared and evaluated based on DLS analysis only to find out that the resulting formulation exhibited incomplete reconstitution in DDW, thus discouraging further investigation (Supplementary Materials Figure S4). The generation of LNs was further confirmed by UA-staining assisted TEM imaging, where both F1-and F2-LNs displayed spherical morphology ( Figure 3b). Interestingly, the mean diameters of F1-and F2-LNs estimated from the TEM images (42.0 ± 13.6 nm and 67.7 ± 14.3 nm, respectively) are in good accordance with the number-averaged particle sizes derived from DLS analysis (44.4 ± 4.7 nm and 70.0 ± 12.9 nm, respectively) (Supplementary Materials Figure S5). The zeta potential values of F1-and F2-LNs were 1.87 ± 0.13 mV and -5.41 ± 0.94 mV, respectively ( Table 1). The EE of F1-and F2-LNs determined using 0.45 µm filters exhibited comparable values of 80.8 ± 0.3% and 89.2 ± 2.5%, respectively. Considering the reduced particle size and positively-charged zeta potential, F1-LNs could be more effective in the oral delivery of CXB than F2-LNs [38,39].

In Vitro CXB Release from CXB@PLNs
The in vitro release test was conducted at various pHs of 1.2 (fasted-state stomach), 4.5 (fed-state stomach), 6.8 (proximal small intestine), and 7.4 (distal small intestine) [40]. The release profiles of Pharmaceutics 2020, 12, 718 9 of 14 CXB suspension, F1, and F2 are presented in Figure 4. The CXB suspension prepared by grinding CXB powder in NaCMC solution was employed as a control. It is worth noting that both F1 and F2 exhibited significantly higher cumulative release than CXB suspension in all tested pH conditions (p < 0.05). These results imply that the bottom-up processing that induces amorphization of CXB is more efficient in enhancing the dissolution rate than the top-down method that does not accompany a change in drug crystallinity (i.e., grinding). The amorphous form of CXB can provide an improved dissolution rate compared to its crystalline form because of weaker intermolecular forces [41]. Meanwhile, F1 showed comparable (at pH 6.8 and 7.4) or significantly higher (at pH 1.2 and 4.5) cumulative release than F2 (p < 0.05). As the absorption of CXB occurs throughout the GI tract [4], the enhancement in the dissolution rate at any physiologically-relevant pHs may improve the oral bioavailability of CXB, as suggested by previous reports [42][43][44].
Pharmaceutics 2020, 12, x FOR PEER REVIEW 9 of 14 dissolution rate compared to its crystalline form because of weaker intermolecular forces [41]. Meanwhile, F1 showed comparable (at pH 6.8 and 7.4) or significantly higher (at pH 1.2 and 4.5) cumulative release than F2 (p < 0.05). As the absorption of CXB occurs throughout the GI tract [4], the enhancement in the dissolution rate at any physiologically-relevant pHs may improve the oral bioavailability of CXB, as suggested by previous reports [42][43][44].

Cytotoxicity of ELPs and Permeability of CXB@PLNs
The cytotoxicity of ELPs and CXB@PLNs was evaluated using MTS-based assay, where the polymers and formulations were incubated with Caco-2 cells for 24 and 48 h. As shown in Figure 5a, the average cell viability was higher than 96.2% within the tested concentration range (2-2000 μg/mL), indicating negligible cytotoxicity of ELPs. F1 and F2 also exhibited no significant cytotoxicity in Caco-2 cells even after 48 h of incubation at the highest concentration (as cell viability, 97.6 ± 2.0% and 96.6 ± 3.1%, respectively) (Figure 5b), which can be explained by the non-cytotoxic and biocompatible nature of ELPs and all other excipients used [35,36,45,46].
Absorptive transport of CXB suspension, F1, and F2 was evaluated in the Caco-2 cell monolayer model ( Figure S5b). Although F1 and F2 exhibited comparable apparent permeability (Papp) values of 17.4 ± 0.5 and 17.7 ± 1.6 nm/sec, respectively, both of them were significantly different from Papp of CXB suspension (6.1 ± 0.4 nm/sec) (p < 0.05). This could be attributed to the improved dissolution properties of CXB@PLNs (Section 3.4), which may have increased the dissolved CXB concentration in the apical side (A-side). In addition, the enhancing effects of phospholipid and non-ionic surfactants in CXB@PLNs may also have affected the increased permeability [36]. At the end of the transport experiment, trans-epithelial electrical resistance (TEER) values were measured to verify whether the increased permeability resulted from any damages to the cell monolayer. The average TEER values of CXB suspension, F1, and F2 at 120 min were 454 ± 72, 669 ± 262, and 643 ± 359 Ω·cm 2 , respectively, displaying no significant differences. This result suggests that F1 and F2 exhibited enhancement in permeability without causing serious damage to the monolayer compared to CXB suspension.

Cytotoxicity of ELPs and Permeability of CXB@PLNs
The cytotoxicity of ELPs and CXB@PLNs was evaluated using MTS-based assay, where the polymers and formulations were incubated with Caco-2 cells for 24 and 48 h. As shown in Figure 5a, the average cell viability was higher than 96.2% within the tested concentration range (2-2000 µg/mL), indicating negligible cytotoxicity of ELPs. F1 and F2 also exhibited no significant cytotoxicity in Caco-2 cells even after 48 h of incubation at the highest concentration (as cell viability, 97.6 ± 2.0% and 96.6 ± 3.1%, respectively) (Figure 5b), which can be explained by the non-cytotoxic and biocompatible nature of ELPs and all other excipients used [35,36,45,46].
Absorptive transport of CXB suspension, F1, and F2 was evaluated in the Caco-2 cell monolayer model ( Figure S5b). Although F1 and F2 exhibited comparable apparent permeability (P app ) values of 17.4 ± 0.5 and 17.7 ± 1.6 nm/sec, respectively, both of them were significantly different from P app of CXB suspension (6.1 ± 0.4 nm/sec) (p < 0.05). This could be attributed to the improved dissolution properties of CXB@PLNs (Section 3.4), which may have increased the dissolved CXB concentration in the apical side (A-side). In addition, the enhancing effects of phospholipid and non-ionic surfactants in CXB@PLNs may also have affected the increased permeability [36]. At the end of the transport experiment, trans-epithelial electrical resistance (TEER) values were measured to verify whether the increased permeability resulted from any damages to the cell monolayer. The average TEER values of CXB suspension, F1, and F2 at 120 min were 454 ± 72, 669 ± 262, and 643 ± 359 Ω·cm 2 , respectively, displaying no significant differences. This result suggests that F1 and F2 exhibited enhancement in permeability without causing serious damage to the monolayer compared to CXB suspension.

Pharmacokinetic Study and Blood Biochemistry Test
The pharmacokinetic properties of CXB@PLNs were evaluated in rats, where CXB suspension, F1, or F2 was orally administered at a CXB dose of 2 mg/kg. The plasma CXB concentration versus time profiles and pharmacokinetic parameters are presented in Figure 6 and Table 2, respectively. The AUC values of all groups were significantly different from among others (p < 0.05), showing rank order of CXB suspension (as relative bioavailability, 100%) < F2 (176%) < F1 (287%). Considering the low solubility of free CXB (≈1.23 ± 0.05 μg/mL at pH 6.8), the enhancement in oral absorption of CXB@PLNs can be mainly attributed to their improved dissolution properties [43]. Moreover, the generated LNs were designed to serve as both absorption enhancers and PIs in the GI tract, which may also have affected the increased oral bioavailability. A similar approach with CXB-loaded proliposomes composed of SPC, poloxamer 188, and sorbitol was reported by our group [2], where an improved dissolution of CXB was also translated to an enhanced oral absorption (1.73-fold, compared to free CXB). Meanwhile, the higher bioavailability of F1 than F2 can be explained by the difference in physicochemical properties (i.e., reduced mean diameter and positive surface charge) (Section 3.3), as well as the superior dissolution properties (Section 3.4). The average Cmax values of F1 and F2 increased 3.45-and 1.83-fold compared to that of CXB suspension, which could also be explained by the improvement in dissolution behaviors. The Tmax values of CXB suspension were highly variable, displaying a range of 1-6 h. However, F1 and F2 exhibited shorter values of 0.56 ± 0.31 and 0.63 ± 0.25 h, respectively. As these Tmax values of CXB@PLNs are much faster than that of Celebrex (2.75 ± 1.50 h) [2], CXB@PLNs could be a promising CXB delivery platform for acute pain management, which requires rapid onset time [47].

Pharmacokinetic Study and Blood Biochemistry Test
The pharmacokinetic properties of CXB@PLNs were evaluated in rats, where CXB suspension, F1, or F2 was orally administered at a CXB dose of 2 mg/kg. The plasma CXB concentration versus time profiles and pharmacokinetic parameters are presented in Figure 6 and Table 2, respectively. The AUC values of all groups were significantly different from among others (p < 0.05), showing rank order of CXB suspension (as relative bioavailability, 100%) < F2 (176%) < F1 (287%). Considering the low solubility of free CXB (≈1.23 ± 0.05 µg/mL at pH 6.8), the enhancement in oral absorption of CXB@PLNs can be mainly attributed to their improved dissolution properties [43]. Moreover, the generated LNs were designed to serve as both absorption enhancers and PIs in the GI tract, which may also have affected the increased oral bioavailability. A similar approach with CXB-loaded proliposomes composed of SPC, poloxamer 188, and sorbitol was reported by our group [2], where an improved dissolution of CXB was also translated to an enhanced oral absorption (1.73-fold, compared to free CXB). Meanwhile, the higher bioavailability of F1 than F2 can be explained by the difference in physicochemical properties (i.e., reduced mean diameter and positive surface charge) (Section 3.3), as well as the superior dissolution properties (Section 3.4). The average C max values of F1 and F2 increased 3.45-and 1.83-fold compared to that of CXB suspension, which could also be explained by the improvement in dissolution behaviors. The T max values of CXB suspension were highly variable, displaying a range of 1-6 h. However, F1 and F2 exhibited shorter values of 0.56 ± 0.31 and 0.63 ± 0.25 h, respectively. As these T max values of CXB@PLNs are much faster than that of Celebrex (2.75 ± 1.50 h) [2], CXB@PLNs could be a promising CXB delivery platform for acute pain management, which requires rapid onset time [47].
(Section 3.3), as well as the superior dissolution properties (Section 3.4). The average Cmax values of F1 and F2 increased 3.45-and 1.83-fold compared to that of CXB suspension, which could also be explained by the improvement in dissolution behaviors. The Tmax values of CXB suspension were highly variable, displaying a range of 1-6 h. However, F1 and F2 exhibited shorter values of 0.56 ± 0.31 and 0.63 ± 0.25 h, respectively. As these Tmax values of CXB@PLNs are much faster than that of Celebrex (2.75 ± 1.50 h) [2], CXB@PLNs could be a promising CXB delivery platform for acute pain management, which requires rapid onset time [47].  Blood biochemistry tests were conducted in rats, where clinical biomarkers that are related to the damages to the liver (ALT, ALB, and TP) and kidney (BUN and SCr) were assessed after an oral administration of the developed formulations at a CXB dose of 2 mg/kg. As shown in Table 3, no significant difference was observed between the control (no treatment) and the formulation-treated groups, indicating that oral administration of CXB@PLNs at the administered dose caused no serious systemic toxicities. Together with the negligible cytotoxicity observed in Caco-2 cells (Section 3.5), this result supports the safety of CXB@PLNs as oral delivery platforms.

Conclusions
CXB@PLNs were prepared using a simple lyophilization method for the oral delivery of CXB, which contain ELPs, novel polymeric PIs, synthesized by the PEGylation of EL. The physical state of CXB@PLNs was investigated using DSC and PXRD analyses, which suggest that the amorphization of CXB occurred during the fabrication of PLNs. Generating LNs in aqueous conditions, CXB@PLNs displayed enhanced dissolution rate and permeability compared to CXB suspension. In vivo pharmacokinetic studies demonstrated the higher oral bioavailability and faster T max of CXB@PLNs than those of CXB suspension. The blood biochemistry tests performed in rats confirmed no severe systemic toxicities of CXB@PLNs. Taken together, the developed PLNs could be promising platforms for improving the oral bioavailability of poorly water-soluble drugs, such as CXB.