Zein’s biocompatibility, biodegradability, and ability to effectively encapsulate drugs make it an ideal matrix for drug delivery systems. However, the disadvantages of zein, such as its rapid clearance, limited solubility, and potential immunogenicity, necessitate PEGylation to overcome these challenges, hence ensuring safe and efficient delivery of therapeutic agents.

Here, we propose the covalent PEGylation of zein through its carbodiimide coupling with Mal-PEG5k-COOH. Based on reports in the literature, glutamine and asparagine account for 22% of the zein structure [26]. For that reason, the carboxylic acid group of Mal-PEG5k-COOH was used for the carbodiimide conjugation to target the glutamine and asparagine’s primary amines of zein. The amino acid cysteine in zein’s structure, on the other hand, accounts for less than 1% of the zein structure [27], reducing the probability of the Michael-type addition cross-reactivity between the thiol chemical group of cysteine and the maleimide moiety found on Mal-PEG5k-COOH.

To verify the success of the zein PEGylation reaction, the ATR-FTIR spectra of zein-PEG5k-Mal (ZP1-ZP6 conjugates), zein, Mal-PEG5k-COOH and stoichiometric control physical mixtures (zein and Mal-PEG5k-COOH) were analyzed. Generally, the ATR-FTIR spectrum of zein showed -NH2 stretching bands present at around 3300 cm− 1 which were typical for the unconjugated zein (band a), along with the characteristic bands of the zein protein backbone of C = O of amide I stretching and amide II N-H bending coupled to -C-N stretching at 1643 cm− 1 (band c) and 1542 cm− 1 (band d, respectively [28]. Alternatively, Mal-PEG5k-COOH spectra showed an imminent PEG-related band at around 1100 cm− 1 (band e), along a band at around 1700 cm− 1 corresponding to the C = O stretching vibration of carboxylic acid groups (band b). Zein bands (a, c, d) were found in the spectra of all zein-PEG5k-Mal conjugates. However, no new bands were observed in the spectra of the conjugates as the new amide bonds formed overlap with the protein amide peaks [29]. The success of the PEGylation reaction was confirmed both by the disappearance of the band characteristic of the carboxylic acid stretching vibrations of Mal-PEG5k-COOH (band b), and the increase of its band characteristic of the PEG spectra (band e). This band characteristic of the PEG spectra was, however, absent in the ZP1 spectrum, but present in the spectra of conjugates ZP2 to ZP6 (Fig. 2A). This undetectable amount of conjugated PEG in ZP1, which suggests the failure of an efficient conjugation, could be a consequence of the low amount of Mal-PEG5k-COOH used in the reaction, as reports have shown that the key factor affecting the conjugation efficiency is the ratio of carboxylic acid functionalities (-COOH) to the amine (-NH2) groups of the protein [30], where an excess of Mal-PEG5k-COOH is necessary to prevent intermolecular crosslinking within the protein [31].

ZP2-ZP6 were then analyzed by NMR to evaluate the conjugation efficiency of Mal-PEG5k-COOH to zein, determining the degree of PEGylation and amount of available maleimide groups for subsequent reaction with ANG2. 1H peak of ethylene glycol fragment from PEG (-CH2-CH2) were present at around δ = 3.67 ppm as shown in Fig. 2B (peak a) [32]. Whereas, maleimide peak in the Mal-PEG5k-COOH was found at around δ = 6.87 ppm (peak b) (Fig. 2B). Regarding zein, the protons from its aromatic amino acid fragments at around δ = 6.75 ppm (peak c) were considered (Fig. 2B). Peak c denotes two protons corresponding to the aromatic ring from tyrosine. By comparing the zein representative peaks (tyrosine) to Mal-PEG5k-COOH representative peaks (ethylene glycol) integrals, the percentage of conjugation of zein to Mal-PEG5k-COOH was calculated for each of the conjugates (Table 3).

(A) ATR-FTIR and (B)1H-NMR spectra of zein, Mal-PEG5k-COOH, physical mixture of zein and Mal-PEG5k-COOH, or ZP conjugates. (A)a, c and d represent the major distinctive bands of zein used for following the conjugation reaction, while b and e are characteristic bands of Mal-PEG5k-COOH. (B)a and b represent peaks characteristic of Mal-PEG5k-COOH, while c represents a peak characteristic of zein. Peaks a and c were used to calculate the conjugation efficiency of Mal-PEG5k-COOH to zein

Considering the good balance between PEGylation efficiency value and a reaction time of only 24 h, ZP4 was selected to proceed with further studies.

ANG2 peptide offers enhanced brain targeted drug delivery by efficiently shuttling drug delivery systems across the BBB, facilitating the accumulation of therapeutic agents in the brain parenchyma and, consequently, improving treatment efficacy. As such, here, we propose the modification of covalently PEGylated zein with ANG2 to enable an efficient brain targeted drug delivery, harnessing the advantages of both technologies for improved therapeutic outcomes in central nervous system disorders. Thiol-maleimide click chemistry was used to conjugate the maleimide chemical group of zein-PEG5k-Mal (ZP4) with the thiol chemical group of a cysteine-modified ANG2.

The efficiency of ZP4 chemical conjugation with ANG2 (giving ZP4-ANG2) was indirectly assessed by quantification of the free thiol chemical groups of the peptide, projecting on the unbound cysteine-modified ANG2 present in the supernatant collected from the conjugates’ washings. This quantification was performed using the Ellman’s assay. The conjugation efficiency of ANG2 was calculated as a function of the initial peptide mass and was found to be 96.43 ± 0.30%. Thus, this conjugation process demonstrated high grafting efficiency of ANG2 to the PEGylated zein. This implies that for every maleimide chemical group present in zein-PEG5k-Mal, a corresponding ANG2 moiety was successfully conjugated.

The percentage of ANG2 in the ZP4-ANG2 conjugate was also retrieved from the NMR spectra obtained for ZP4 (Fig. 2B). The percentage of covalently PEGylated zein in ZP4 was rounded down to 25% (Table 3), obtained as a function of the percentage of maleimide chemical groups on the conjugated Mal-PEG5k-COOH. Considering from the Ellman’s assay that most of these maleimide moieties on ZP4 were occupied by ANG2, the peptide covered approximately 25% of the zein protein surface.

Several ZNP nanoformulations loaded with DTX were produced through a straightforward nanoprecipitation process a posteriori from a successful covalent PEGylation and ANG2 modification of zein. ZNP formulations were produced with 5%, 10% or 20% ZP4-ANG2 (ZP4-ANG2 5%, ZP4-ANG2 10% and ZP4-ANG2 20% ZNPs, respectively). Control nanoformulations without ANG2 were also produced, based on covalently PEGylated zein (ZP4 ZNPs) or only pure zein (NF ZNPs).

Figure 3A highlights the physicochemical properties of the different ZNP nanoformulations. The PEGylation of the ZNPs (ZP4 ZNPs) caused a decrease in the hydrodynamic diameter of the nanoparticles compared with the pure zein nanoformulation (NF ZNPs; approximately 140 nm versus 80 nm). This can possibly be due to the amphiphilic nature of PEG, which led to it acting as a surfactant reducing the surface tension of the particles, changing its physical properties and therefore, decreasing its hydrodynamic diameter [33]. This decrease in hydrodynamic diameter improved the nanoformulation, considering that an average size smaller than 100 nm is generally more suitable for the purpose of BBB crossing [9]. Regarding PDI, all nanoformulations presented values up to around 0.3, which is generally considered to indicate a relatively monodisperse or narrowly distributed particle size population, suggesting uniform size distribution. The PEGylated ZP4 ZNPs presented a decrease in the zeta-potential to near neutrality compared with the pure zein NF ZNPs nanoformulation (approximately 20 mV versus 3 mV). This has been rendered useful to increase the circulation time of the nanoparticles [34], and to possibly lead to improved penetration through the BBB, like in the case of GBM [35]. A possible reason for this is due to the stealth effect of the PEG on decreasing the electrostatic potential and shifting the position of the shear plane outwards from the particle surface [36]. The positive zeta-potential of NF ZNPs was attributed to the protonation of the -NH2 groups of zein (isoelectric point around 6.2) upon dilution, which would have taken place at the dispersant’s pH of 4.5 [37].

TEM analysis depicted spherical-shaped particles with consistently smooth surfaces across all nanoparticle formulations, indicating uniformity in morphology. Additionally, the size distribution was found to be identical for all nanoparticle formulations, as illustrated in Fig. 3B.

(A) Summary of average hydrodynamic diameter, PDI and zeta-potential values for the different nanoformulations of zein loaded with DTX. Data presented as mean ± STD (n = 3, each n corresponding to a different nanoformulation batch). (B) TEM images of (i) DTX-loaded NF ZNPs, (ii) ZP4 ZNPs and (iii) ZP4-ANG2 ZNPs

AE and DL values were determined by HPLC analysis, and presented an increase for the PEGylated nanoformulations (average of 50–70% AE and 4–6% DL) in comparison to the non-PEGylated ZNPs (average of 25% AE and 2% DL) (Table 4). This increase is in concordance to the decrease of hydrodynamic diameter of the NPs composed by PEGylated zein. This could primarily be because smaller nanoparticles offer a larger surface area-to-volume ratio [38], allowing for more efficient interaction and entrapment of the encapsulated drug.

After demonstrating successful PEGylation and functionalization with ANG2 of zein, and the ability of the chemical conjugate to form nanoparticles of relatively uniform size around 100 nm with a high DTX loading value, it was necessary to prove that the encapsulated drug retains its anti-tumor activity. For this purpose, we utilized GBM cells as a cellular model for efficacy testing, although it is expected that our brain targeting nanosystem will enable the encapsulation of a variety of drugs with different therapeutic targets within the scope of brain diseases.

The cytotoxicity of ZNPs was tested against the U-87 MG GBM cell line over 48 h and 72 h (Fig. 4), using a DTX concentration ranging from 0.0005 µM to 5 µM. Owing to the mechanism of action of DTX of arresting the cell cycle, which has a timespan exceeding 24 h [39], the studies were performed for longer time periods.

Increasing the concentration of DTX loaded into ZNPs beyond 0.0005 µM incurred significant reduction of metabolic activity on GBM cells, where a reduction below the threshold of 70% [40] was seen after both 48 h and 72 h. This cytotoxicity was evident upon direct treatment with pure zein, PEGylated zein, as well as PEGylated and ANG2 functionalized ZNPs (5%, 10% and 20% ANG2), comparable to the profile of free DTX at the same used concentrations. This drug-associated cytotoxicity strongly suggests the maintenance of the anti-tumor activity of DTX upon loading into ZNPs.

To further confirm that the imposed metabolic activity impairment was a result of the treatment with DTX rather than the nanoparticle matrix itself, unloaded pure zein, PEGylated zein, as well as PEGylated and ANG2 functionalized ZNPs were tested against U-87 MG GBM cells for 72 h at the same concentrations used for the loaded nanoformulations (Figure S2). Data have shown no significant impairment in metabolic activity for all formulations up to a concentration equivalent to 0.5 µM of loaded DTX, therefore confirming the safety of their matrix.

In general, the metabolic activity of GBM cells decreased in a dose-dependent manner after treatment, and lower levels were reached at 72 h compared to 48 h. This could possibly be due to mechanism of action of DTX of arresting the cell cycle, which normally takes around 72 h [41]. PEGylated and ANG2 functionalized ZNPs caused a similar impact on metabolic activity compared to pure zein and PEGylated zein ZNPs. Moreover, PEGylated and ANG2 functionalized ZNPs demonstrated a similar or slightly lower decrease on cell metabolic activity compared to the non-nanoparticulate drug control, free DTX. This was anticipated since free drug molecules are immediately available to easily interact with the cell layers, thereby exerting a faster anti-proliferative effect. Whereas, for the nanoparticulate samples, including PEGylated and ANG2 functionalized ZNPs, drug molecules become only available upon DTX diffusion across the zein matrix and, in a later stage, matrix erosion.

Cytotoxicity of the different DTX-loaded ZNP nanoformulations and free DTX control against U-87 MG GBM cells after (A) 48 h and (B) 72 h incubation (concentrations related to loaded DTX). Data presented as mean ± STD (n = 3, each n corresponding to a different nanoformulation batch). All comparisons were performed using two-way ANOVA followed by the Dunnett’s test (*p < 0.05, **p < 0.01, or ***p < 0.001 relative to the free DTX group)

The primary goal of chemically modifying the zein matrix with ANG2, a pioneering approach, was to facilitate brain-targeted drug delivery, which has remained one of the major challenges in pharmaceutical technology.

Studies in a hCMEC/D3 BBB in vitro model (Fig. 5A) were performed to assess the blood-to-brain permeability of the different ZNP nanoformulations, thus allowing us to extrapolate their ability to reach the targeted site of action, namely the brain parenchyma.

ZNP nanoformulations were labeled with C6, which was used as a drug surrogate, as extensively reported in literature [22]. Different concentrations of C6 analogous with pre-performed permeability studies were tested [17, 21]. A concentration of 600 µg/mL of C6 labeled nanoparticles was chosen after confirmation of lack of significant 24 h cytotoxicity in hCMEC/D3 BBB endothelial cells, as confirmed through resazurin assays (Figure S3). The 24 h timepoint was anointed as the final timepoint of the study. Pre-incubation TEER values of the hCMEC/D3 cell monolayer ranged from 5 to 35 \( \text\text\text\), being consistent with TEER values pre-established for this specific in vitro model (Fig. 5B) [24]. An increase of TEER values against the blank was observed up till the day of treatment (day 8), indicating the maturation of the endothelial cells’ tight junctions [42] and the formation of a monolayer.

At 4 h and 8 h, no biologically relevant differences in percentage of permeability were found between all testing conditions. However, at 24 h, significant differences were found in percentage of permeability values, namely ZP4-ANG2 10% ZNPs > ZP4-ANG2 20% ZNPs > ZP4-ANG2 5% ZNPs > ZP4 ZNPs > NF ZNPs. In contrast to previous studies indicating that PEGylation reduces cellular interactions [34], it was deduced that PEGylated ZNPs (both non-functionalized ZP4 ZNPs and ANG2 functionalized ZP4-ANG2 ZNPs) exhibited enhanced permeation characteristics at the 24 h timepoint when compared to non-PEGylated ZNPs (NF ZNPs; approximately 3-times increase in percentage of permeability (Fig. 5C) and Papp (Fig. 5D)). This could possibly be due to the stability enhancing effects of PEG in biologically mimetic environments [43], consequently improving the ZNPs’ likelihood to traverse the BBB monolayer intact. Alternatively, this enhanced permeability could be attributed to the slightly smaller hydrodynamic diameter of the PEGylated ZNPs [44], which accelerates their cellular uptake and transcellular permeability.

Remarkably, ZP4-ANG2 ZNPs exhibited significantly higher blood-to-brain permeability across the BBB in vitro model compared to non-PEGylated NF ZNPs (around 3-times increase in percentage of permeability for ZP4-ANG2 5% and ZP4-ANG2 20% ZNPs, and 4-times increase for ZP4-ANG2 10% ZNPs (Fig. 5C)) and PEGylated ZP4 ZNPs (around 1.5-times increase in percentage of permeability for ZP4-ANG2 5% ZNPs, and 2-times increase for ZP4-ANG2 10% and ZP4-ANG2 20% ZNPs (Fig. 5C)), suggesting the key role of receptor-mediated transcytosis via ANG2 binding to LRP-1. Increasing the percentage of ANG2 in the nanoformulation from 5 to 10% and 20% significantly improved percentage of permeability (Fig. 5C) and Papp (Fig. 5D). Nonetheless, the increase in ANG2 content in the final nanoformulation from 10 to 20% yielded no significant change in in vitro blood-to-brain permeability, as shown by permeability percentage values (Fig. 5C), suggesting the possibility of receptor saturation [45]. The most substantial difference in permeability between formulations was observed at the 24 h timepoint, thereof denoting that the influence of ANG2 becomes more pronounced past the 8 h timepoint.

Throughout the permeability assay, TEER values showed no significant fluctuations across all nanoformulations (Fig. 5C). This consistency indicates that the integrity and intactness of the tight junction-held BBB in vitro monolayer were preserved throughout the entire experiment. If there had been any disruption of the monolayer, we would have observed a similar increase in permeability levels across all nanoformulations, which did not occur.

These results clearly demonstrated the ability of ANG2 functionalization to ameliorate the accumulation of the nanoformulations into the brain. With the use of a PEGylated and ANG2 functionalized zein matrix, the in vitro BBB permeability of the nanoformulation was efficiently maximized, with an overall increase of up to 4 times. This demonstrates an enhanced capacity of traversing the BBB, with potential to increase the efficacy of delivering therapeutics to the brain.

Blood-to-brain in vitro permeability study. (A) Scheme of the hCMEC/D3 BBB in vitro model. (B) Monitorization of TEER for the hCMEC/D3 BBB in vitro model from day 2 to day 8. (C) Cumulative permeability percentage and TEER values of the model after incubation with the different nanoformulations for 4 h, 8 h and 24 h. Statistical comparison relative to the NF ZNPs group. (D) Calculated Papp for the 24 h time-point. Data presented as mean ± STD (n = 3, each n corresponding to a different nanoformulation batch). All comparisons were performed using two-way ANOVA followed by the Dunnett’s test (*p < 0.05, **p < 0.01, or ***p < 0.001)