Crosslinked Fibroin Nanoparticles: Investigations on Biostability, Cytotoxicity, and Cellular Internalization

Recently, crosslinked fibroin nanoparticles (FNP) using the crosslinker 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) or the polymer poly(ethylenimine) (PEI) have been developed and showed potentials as novel drug delivery systems. Thus, this study further investigated the biological properties of these crosslinked FNP by labeling them with fluorescein isothiocyanate (FITC) for in vitro studies. All formulations possessed a mean particle size of approximately 300 nm and a tunable zeta potential (−20 to + 30 mV) dependent on the amount/type of crosslinkers. The FITC-bound FNP showed no significant difference in physical properties compared to the blank FNP. They possessed a binding efficacy of 3.3% w/w, and no FITC was released in sink condition up to 8 h. All formulations were colloidal stable in the sheep whole blood. The degradation rate of these FNP in blood could be controlled depending on their crosslink degree. Moreover, no potential toxicity in erythrocytes, Caco-2, HepG2, and 9L cells was noted for all formulations at particle concentrations of < 1 mg/mL. Finally, all FNP were internalized into the Caco-2 cells after 3 h incubation. The uptake rate of the positively charged particles was significantly higher than the negatively charged ones. In summary, the crosslinked FNP were safe and showed high potentials as versatile systems for biomedical applications.


FNP Characterization
To track the particles in biological systems, FNP were labeled with FITC, a xanthene dye. FITC is the most popularly used fluorescence labelling agent for cell-based experiments due to its water solubility, ease of conjugation, high quantum efficiency, and low nonspecific binding with biological systems [35]. Generally, FITC isothiocyanate groups could react with fibroin residual amine groups to form thioamide (thiourea) covalent bonds. However, the success of this process depends on various stringent factors such as fibroin concentrations and buffers. Moreover, the incomplete removal of unreacted FITC might cause high background fluorescence in cellular studies. Thus, thorough characterizations should be considered.
Our results demonstrated that FITC-bound FNP were successfully prepared with homogeneous distribution in the medium (Figure 1). In addition, no free dye staining the image background was Pharmaceuticals 2020, 13, 86 3 of 13 observed, suggesting that the unbound FITC was completely washed off. All formulations manifested similar FITC-binding efficiency of approximately 3.5% w/w, with no statistically significant difference. These qualitative results were in agreement with FITC dissolution studies. After 8 h in the sink condition at 37 • C, less than 0.5% of FITC was released from all FNP formulations. significant difference. These qualitative results were in agreement with FITC dissolution studies. After 8 h in the sink condition at 37 °C, less than 0.5% of FITC was released from all FNP formulations.
In terms of physical properties, Table 1 summaries the mean size, PI, and zeta potential of four investigated FNP formulations, including non-crosslinked FNP, EDClow-FNP, EDChigh-FNP, and PEI-FNP. No significant difference between the blank FNP and the FITC-bound FNP was noted, indicating that the FITC incorporation did not affect the particle properties. All formulations possessed similar particle size of approximately 300 nm and an acceptable size distribution with PI of < 0.3. On the other hand, the surface zeta potential of the non-crosslinked FNP and EDClow-FNP had negative values of −18 mV, which is the normal charge of fibroin in water, whereas the EDChigh-FNP and PEI-FNP had positive values of + 30 mV. Thus, the positively charged polymer PEI and the high amount of EDC successfully altered the FNP surface charge from negative to positive [24]. Moreover, all formulations were stable at 4 °C for at least 6-month storage, with no significant differences in mean particle size and zeta potential.  In terms of physical properties, Table 1 summaries the mean size, PI, and zeta potential of four investigated FNP formulations, including non-crosslinked FNP, EDC low -FNP, EDC high -FNP, and PEI-FNP. No significant difference between the blank FNP and the FITC-bound FNP was noted, indicating that the FITC incorporation did not affect the particle properties. All formulations possessed similar particle size of approximately 300 nm and an acceptable size distribution with PI of < 0.3. On the other hand, the surface zeta potential of the non-crosslinked FNP and EDC low -FNP had negative values of −18 mV, which is the normal charge of fibroin in water, whereas the EDC high -FNP and PEI-FNP had positive values of + 30 mV. Thus, the positively charged polymer PEI and the high amount of EDC successfully altered the FNP surface charge from negative to positive [24]. Moreover, all formulations were stable at 4 • C for at least 6-month storage, with no significant differences in mean particle size and zeta potential. Table 1. Mean particle size, polydispersity index, and zeta potential of blank fibroin nanoparticles (FNP) and FITC-bound FNP: The results are expressed in terms of mean ± SD, n = 3.

Formulation
Particle

Biostability and Hemolysis Study
Intravenous injection is one of the most common applications of nanoparticles [23]. Thus, the interaction between FNP and the circulatory blood components critically affects their efficiency. Fibroin has been proved by the U.S. Food and Drug Administration (FDA) as a biomaterial, and its film formulations are hemocompatible [36]. However, limited studies have explored this issue on FNP.
To this end, we reported three set of data, including the biostability in the whole blood, the degradation behaviors, and the hemolysis ability to the erythrocytes of four investigated FNP.
Generally, due to their high kinetics energy, most nanoparticles tend to aggregate in the biological systems before contacting the cells. These aggregates might result in blockage of the blood vessel and in altering the nanoparticle properties [37]. Thus, we first investigate the FNP colloidal stability in the cell culture media. The result showed that all FNP were aggregated within 1 h in the Dulbecco's Modified Eagle Medium (DMEM) medium without fetal bovine serum (FBS). Interestingly, in the complete medium with FBS, although their zeta potential was modified to become near 0 mV, they were colloidal stable for 3 days with no significant changes in particle size (data not shown). Similar result was observed when incubating FNP in the whole blood at 37 • C: all formulations were stable for 24 h without observable aggregation. Therefore, the superior colloidal stability of FNP in whole blood could be a result from stabilizing effects of serum protein, such as albumin, coating on the particle surface, which help prevent particle aggregation [38].
Secondly, FNP degradation, in terms of percentage weight loss, was investigated in the whole blood during a period of 7 days ( Figure 2). In the first 12 h, the average weight loss of all formulations was 5-7%, with no significant differences between them ( Figure 2B). Interestingly, after 12 h and until 7 days, the degradation rate followed FNP ≈ PEI-FNP > EDC low -FNP > EDC high -FNP. In general, FNP are degraded by proteolytic enzymes (i.e., proteases), in which most of them mainly cleave the amorphous, less crystalline, or non-compact regions of fibroin [39,40]. Therefore, higher crystallinity EDC high -FNP (i.e., more compact structure) [24] was more stable than lower ones, such as EDC low -FNP, PEI-FNP, and FNP. Notably, since all formulations possessed similar mean particle size and FITC-binding capacities, FNP polymorph was the main factor affecting FNP degradation rate. Our results suggested that the degradation rate of FNP can be controlled favorably to fit in the biomedical applications by adjusting the particle crystallinity through crosslinking reactions. For example, highly crosslinked EDC high -FNP with slow degradation rate can be used in clinical settings that required long treatment duration (i.e., cancer) to reduce the drug administration frequency. On the other hand, the fast degradation rate FNP can be utilized in short-duration treatments.

In Vitro Cytotoxicity
In this study, three representative cell lines, Caco-2, HepG2, and 9L, were chosen. Caco-2 is an intestinal cell line, which is beneficial in the study context of an oral exposition to the nanoparticles, whereas HepG2 is a hepatic cell line, which is crucial for intravenous exposition. Additionally, 9L, a In terms of hemolysis actions, non-crosslinked FNP have been proved to be nontoxic to the blood [41]. On the other hand, the positively charged particles (i.e., EDC high -FNP and PEI-FNP) might theoretically be more toxic to the erythrocytes than the negatively charged particles (i.e., non-crosslinked FNP and EDC low -FNP) [31,32]. Surprisingly, our results indicated that all formulations showed no potential toxicity to the red blood cells at a concentration of as high as 1 mg/mL. These data demonstrated that the crosslinked FNP are safe and suitable for systemic applications.

In Vitro Cytotoxicity
In this study, three representative cell lines, Caco-2, HepG2, and 9L, were chosen. Caco-2 is an intestinal cell line, which is beneficial in the study context of an oral exposition to the nanoparticles, whereas HepG2 is a hepatic cell line, which is crucial for intravenous exposition. Additionally, 9L, a gliosarcoma cell line, was used for the potential applications of FNP in cancer treatment.

Cellular Uptake and Flow Cytometry Study
Cellular interactions between FNP and investigated tissues are crucial for biomedical applications. For example, negatively charged particles might get internalized by the cells less than positively charged ones [32]. Therefore, to investigate the effects of FNP properties on cellular internalization, both qualitatively and quantitatively, the cellular uptake and flow cytometry studies were conducted on the representative Caco-2 cell line. Figure 4 illustrates the fluorescence images of various FITC-bound FNP formulations on Caco-2 cells. Clearly, after 3 h of incubation, regardless of their differences in the zeta potential and crystallinity, all formulations were internalized into the cell cytoplasm and, possibly, nucleus. Furthermore, flow cytometry results clarified that the cellular uptake was time dependent, with significant differences in the percentage of uptake cells on all formulations ( Figure 5). Interestingly, the amount of uptake cells varied based on formulations. After 3 h of incubation, the negatively charged non-crosslinked FNP and EDC low -FNP showed significantly less uptake compared to the positively charged EDC high -FNP and PEI-FNP (approximately 30%, 30%, 60%, and 90%, respectively). This might be explained by the cell-FNP interaction mechanisms. Cellular internalization commonly starts with the nanoparticle adhering on the cell membrane via various forces, such as van der Waals, electrostatic, hydrophobic, and ligand-receptor, before internalization. Thus, stronger forces result in better cell-FNP interactions and, therefore, increased endocytosis. As both EDC high -FNP and PEI-FNP possessed positive surface charges, they bound more strongly and efficiently to the negatively charged Caco-2 cell membrane surface, consequently enhancing internalization. Additionally, the PEI branched structure could penetrate and provide more contact points with the cell membrane, further increasing the PEI-FNP uptake. Another issue to note is that the soft (i.e., less tight) particles tend to be internalized more than rigid ones [43]. To this end, PEI-FNP, with lower crystallinity and softer structure [24], was taken up better than the higher crystallinity and rigid EDC high -FNP.  possessed positive surface charges, they bound more strongly and efficiently to the negatively charged Caco-2 cell membrane surface, consequently enhancing internalization. Additionally, the PEI branched structure could penetrate and provide more contact points with the cell membrane, further increasing the PEI-FNP uptake. Another issue to note is that the soft (i.e., less tight) particles tend to be internalized more than rigid ones [43]. To this end, PEI-FNP, with lower crystallinity and softer structure [24], was taken up better than the higher crystallinity and rigid EDChigh-FNP.  Although being highly internalized by the cells than negatively charged particles, the positively charged particles could possibly never get into contact with cells in tissues that contained a thick mucus layer such as the intestines because they are stuck in this mucosa [30]. Thus, choosing the right FNP properties for suitable administration routes should be greatly considered, as each formulation might interact differently with the biological systems. Our data also suggested that the uptake rate could be altered favorably, dependent on therapeutic applications, based on the particle structure and surface modifications.

Materials
Bombyx mori silkworm cocoons were collected from Bodin Thai Silk Khorat Co., Ltd, Nakhon Ratchasima, Thailand. EDC, branched PEI (molecular weight 25,000 Da), FITC, and 4′,6-diamidino-2-phenylindole (DAPI) were bought from Sigma-Aldrich, Singapore. Sheep whole blood was Although being highly internalized by the cells than negatively charged particles, the positively charged particles could possibly never get into contact with cells in tissues that contained a thick mucus layer such as the intestines because they are stuck in this mucosa [30]. Thus, choosing the right FNP properties for suitable administration routes should be greatly considered, as each formulation might interact differently with the biological systems. Our data also suggested that the uptake rate could be altered favorably, dependent on therapeutic applications, based on the particle structure and surface modifications.
To visualize the FNP in cellular uptake and flow cytometry studies, the fluorescent dye FITC was used to label the FNP to yield FITC-bound FNP (FITC-FNP). To this end, 100 µL of FITC solution (1 mg/mL in dimethyl sulfoxide (DMSO)) was added dropwise into the FNP dispersions in carbonate buffer, pH 9.0 (20 mg FNP/mL), and stirred at 4 • C for 24 h. To remove the unbound FITC, the mixtures were centrifuged at 31,514 × g for 30 min and re-dispersed in DI water. The process was repeated until no FITC was determined in the supernatant. All formulations were prepared freshly before experiments.

Particle Size and Zeta Potential
Dynamic light scattering (DLS) and phase analysis light scattering (PALS) methods (ZetaPALS ® analyzer, Brookhaven Instrument Corporation, Holtsville, NY, USA) were used to determine the mean particle size, size distribution (PI), and zeta potential, respectively. For DLS, the instrument was run at 632. most commonly used theory for calculating zeta potential from experimental data as it is suitable for nanoparticles of any shape and concentration [44]. The FITC-FNP were also imaged using a fluorescence microscope (Axio Observer Z1 model, Carl Zeiss, Oberkochen, Germany). All measurements were determined in triplicate.

FITC-Binding Efficiency
An indirect method was used to determine the amount of FITC that bound to FNP. The supernatants after each washing step were collected and combined, followed by measuring the fluorescence intensity by a fluorescence microplate reader (Synergy H1 Hybrid Reader, BioTek, Winooski, VT, USA) at excitation (Ex) and emission (Em) wavelengths of 490 nm and 525 nm, respectively. The unbound FITC concentrations were calculated using a calibration curve (range: 0.25-20.00 ng/mL, y = 0.0004x + 0.2444, x: FITC concentrations, y: fluorescence intensity, R 2 = 0.9983). The binding efficacy was determined using Equation (1).

FITC Dissolution Profile
In vitro dissolution profiles of FITC-FNP were performed using a shaker method. FITC-FNP, equivalent to 30 µg FITC, were dispersed in 50 mL of DI water at 37 • C and stirred at 200 rpm for 8 h. At each time point of 0.5, 1, 2, 4, and 8 h, 1 mL of sample was withdrawn and DI water was refilled. Then, the samples were centrifuged at 31,514 × g for 5 min, and the supernatant containing released FITC was fluorescently measured at Ex/Em = 490/525 nm. The FITC concentrations were calculated using the same calibration curve as mentioned in Section 3.4.2. The cumulative percentage of FITC released at time t (C t ) was calculated followed Equation (2).
where C t and C i are the concentrations of released FITC at the time points t and i, V 0 is the total volume of dissolution buffer (50 mL), V is the withdrawal sample volume at each time point (1 mL), M 0 is the initial amount of FITC (30 µg), and M i is the withdrawal total amount of FITC at the time point i.

Physical Stability
To determine the FNP long-term physical stability, the lyophilized powders were stored in a tight container in the dark for 6 months at 4 • C. After 6-month storage, the particle size and zeta potential were determined using ZetaPALS ® analyzer.

Biostability Study
To investigate the biostability of FNP in the biological blood, the weight loss experiment was conducted. To this end, 10 mg of each freeze-dried FNP was dispersed in 1 mL of sheep whole blood, followed by continuously shaking at 200 rpm, 37 • C, for 7 days. At each time point of 0, 1, 3, 6, 12, and 24 h and 2, 4, and 7 days, the blood was lysed with 1 mL of DI water and centrifuged at 31,514× g for 5 min. The precipitated particles were collected, re-dispersed in DI water, and centrifuged (washing steps) repeatedly until the supernatant was clear. The precipitates were then dried at 60 • C until constant weight. The remaining FNP weights were determined by an analytical balance, and the percentage of weight loss was calculated based on Equation (3). The respective blood samples with no FNP were used as controls.

Hemolysis
To investigate the in vitro hemolysis action of FNP, sheep red blood cells were used. Briefly, erythrocytes were collected by centrifugation the sheep whole blood at 2432 × g for 5 min. Then, the cells were washed twice in PBS and reconstituted at a concentration of 1% w/v (1% hematocrit) in PBS. Consequently, all formulations, at various concentrations, were incubated with 1 mL of the prepared erythrocytes at 37 • C for 30 min, followed by halting the reaction with ice for 5 min. Finally, the mixtures were centrifuged at 2432 × g for 5 min, and the hemoglobin presented in the supernatants was UV-Vis spectroscopically measured at 540 nm. The percentage of hemolysis were calculated following Equation (4).
where At, An, and Ap are the absorbance values of the test samples, the negative control (PBS), and the positive control (DI water), respectively.

In Vitro Cytotoxicity
The evaluation of the cell viability was assessed using the mitochondrial-based CellTiter 96 ® AQueous Non-Radioactive Cell Proliferation Assay of Promega (MTS assay) (Charbonnieres, France), following the manufacture's protocol. Cells were seeded 72 h before the treatments in a 96-well plate at 70,000, 80,000, and 20,000 cells/well for Caco-2, HepG2, and 9L cells, respectively. Subsequently, cells were incubated with FNP at various concentrations from 0.01 to 1 mg/mL for 24 h. Cells were then washed once with phosphate buffered saline and then incubated with fresh culture medium containing tetrazolium compound (MTS) and phenazine methosulfate (PMS) for 2 h at 37 • C. The absorbance was measured at 490 nm with a microplate reader (Synergy HT BioTek, Winooski, VT, USA). Mixture of the medium and MTS/PMS without cells served as a blank, and the untreated cells were the control and represented 100% viability. The percentage of cell viability was calculated based on Equation (5).

Cellular Uptake and Flow Cytometry Study
Prior to experiments, Caco-2 cells were cultured in a 6-well plate (with coverslips for cellular uptake study), with an initial amount of 150,000 cells/well. The medium was changed every even day until the cells reached 70-80% confluence. Then, cells were treated with FITC-FNP (1 µg/mL, equivalent to FITC) for 3 h for cellular uptake study and for 1 and 3 h for flow cytometry. Subsequently, cells were washed thrice with PBS and 5 mM EDTA pH 5.0 to remove the membrane-bound particles.
For cellular uptake study, cells were then fixed with 2.5% glutaraldehyde in PBS, permeabilized using 0.5% Triton-X 100 in DI water for 15 min, and nuclei counterstained with 300 nM DAPI in PBS for 3 min. Finally, cells were washed twice and the coverslip was mounted onto glass slides using glycerol 70% in PBS as a mounting medium. Untreated cells were used as a control. The slides were observed under fluorescence microscope (Axio Observer Z1 model, Carl Zeiss, Oberkochen, Germany) using 405-nm laser line with a band pass (BP) at 350/470 nm for the blue channel to detect DAPI and a 488-nm laser excitation using a BP 490/525 nm for the green channel to detect FITC.
For flow cytometry study, cells were then harvested by trypsinization and resuspended in PBS for measurements. The percentage of cells internalizing FITC-FNP were determined using a flow cytometer (Guava easyCyteTM5, Merck Millipore, Massachusetts, USA), on an average of 10,000 cells/cycle. The results were analyzed using In-Cytes software installed in the machine (Guava soft 3.2).

Statistical Analysis
All experiments were performed at least in triplicate. For quantitative results, the mean ± SD (standard deviation) was reported. One-way analysis of variance (ANOVA) and Student's t-test were used for statistical purposes, with the p-value of at least < 0.05 for significant comparisons.

Conclusions
This study investigated the in vitro biological properties of the crosslinked FNP utilizing EDC and PEI, with non-crosslinked FNP as a reference. All formulations showed a mean particle size of 300 nm, an adjustable zeta potential from -20 to + 30 mV, and a long-term stability of > 6 months. Additionally, all FNP were biostable in biological systems and showed no potential cytotoxicity to the red blood cells, 9L, Caco-2, and HepG2 cells. Depending on their crystallinity and/or surface modification, FNP degradation rates and cellular internalization rates were controlled favorably. Taking into account of their safety and tunable properties, these crosslinked FNP can be useful in various biomedical applications for both parenteral and non-parenteral routes.