Self-Assembled Silk Fibroin-Based Aggregates for Delivery of Camptothecin

A water-soluble hydrolysate of silk fibroin (SF) (~30 kDa) was esterified with tocopherol, ergocalciferol, and testosterone to form SF aggregates for the controlled delivery of the anticancer drug camptothecin (CPT). Elemental analysis and 1H NMR spectroscopy showed a degree of substitution (DS) on SF of 0.4 to 3.8 mol %. Yields of 58 to 71% on vitamins- and testosterone-grafted SF conjugates were achieved. CPT was efficiently incorporated into the lipophilic core of SF aggregates using a dialysis–precipitation method, achieving drug contents of 6.3–8.5 wt %. FTIR spectra and DSC thermograms showed that tocopherol- and testosterone-grafted SF conjugates predominantly adopted a β-sheet conformation. After the esterification of tyrosine residues on SF chains with the vitamin or testosterone, the hydrodynamic diameters almost doubled or tripled that of SF. The zeta potential values after esterification increased to about −30 mV, which favors the stability of aggregates in aqueous medium. Controlled and almost quantitative release of CPT was achieved after 6 days in PBS at 37 °C, with almost linear release during the first 8 h. MCF-7 cancer cells exhibited good uptake of CPT-loaded SF aggregates after 6 h, causing cell death and cell cycle arrest in the G2/M phase. Substantial uptake of the CPT-loaded aggregates into MCF-7 spheroids was shown after 3 days. Furthermore, all CPT-loaded SF aggregates demonstrated superior toxicity to MCF-7 spheroids compared with parent CPT. Blank SF aggregates induced no hemolysis at pH 6.2 and 7.4, while CPT-loaded SF aggregates provoked hemolysis at pH 6.2 but not at pH 7.4. In contrast, parent CPT caused hemolysis at both pH tested. Therefore, CPT-loaded SF aggregates are promising candidates for chemotherapy.


Introduction
Cancer is becoming the mayor cause of death worldwide in the present century. In 2018, an estimate of 18.1 million new cancer cases were detected and 9.6 million cancer patients died [1,2]. Lung cancer is the most diagnosed and lethal cancer with 11.6% and 18.4% of the total cases and cancer-related mortality. It is followed by breast and prostate cancer as well as colorectal cancer in terms of incidence and mortality [1]. Conventional chemotherapy is still widely used in medical treatment of most cancers and tumors usually in combination with surgery, radiotherapy, and novel nanomedicines. These include genetically designed and personalized drugs, prodrugs, multiresponsive drug delivery systems, and thermal phototherapy, among others [2][3][4]. The anticancer drugs irinotecan and topotecan, two CPT derivatives, are used to treat lung cancer, colon cancer, metastatic or resistant

Preparation of Ergocalciferol-, Tocopherol-, and Testosterone-Grafted SF Aggregates
Ergocalciferol-, tocopherol-, and testosterone-grafted SF were prepared via the esterification of tyrosine fragments of SF with testosterone and vitamin hemisuccinate Nhydroxysuccinimide esters to prevent undesired cross-linking of SF as previously reported [25]. Ergocalciferol-grafted SF (SF1) was obtained as a brown powder. Tocopherolgrafted-SF (SF2) and testosterone-grafted SF (SF3) appeared as white powders after lyophilization (characterization data of SF1-SF3 is included in the Supplementary Material). These SF conjugates formed aggregates of particles in aqueous medium when stirred overnight at 1 mg/mL in water or PBS.

CPT Loading in SF Aggregates
Hydrophobic CPT was incorporated in the lipophilic core of ergocalciferol-, tocopherol-, and testosterone-grafted SF aggregates using a dialysis-precipitation method with lyophilization [4,16]. To this end, 10 mg of SF1, SF2, or SF3 and 1 mg of CPT in DMSO were stirred overnight at 25 • C in darkness. The formation of CPT-loaded SF aggregates and removal of excess CPT was carried out by dialysis against distilled water (2 L) for 5 h with one replacement. CPT-SF1 (slightly brown), CPT-SF2 and CPT-SF3 (white powders) were obtained after lyophilization.

CPT Content and Sustained Delivery
The content of CPT in CPT-loaded SF aggregates and related parameters (loading efficiency and yield of aggregates), as well as in vitro released CPT during drug release studies were determined by UV spectrophotometry (PerkinElmer Ltd., Buckinghamshire, UK), based on calibration curves of CPT in DMSO and PBS (pH 7.4) (Figures S1-S3) (CPT ε DMSO 368 = 21,006 M −1 cm −1 , ε PBS 370 = 26,930 M −1 cm −1 ). CPT delivery studies were conducted at 37 • C in PBS (pH 7.4). First, 5 mg of CPT-SF1, CPT-SF2 or CPT-SF3 in 2 mL of PBS were dialyzed against 10 mL of PBS. The 10 mL PBS was replaced with fresh media, and CPT delivered was determined using UV spectroscopy (CPT λ emission = 370 nm).

Characterization
Molecular weight determinations of SF were performed with a Viscotek GPCmax gel permeation chromatograph (GPC) provided with a PFG column from PSS (Mainz, Germany) (300 mm × 8 mm and 5 µm particle size) and a Viscotek TDA 305 Triple Detector Array (Malvern, Kassel, Germany) [25]. Calibration was performed by triple detection (refractive index, right angle light scattering and viscometer) with PSS polystyrene standards [25]. The purified SF sample was eluted with 10 mM LiBr in DMF and a flow rate of 0.75 mL/min at 60 • C.
ATR-FTIR spectra of SF conjugates were obtained using a Perkin Elmer Spectrum 100 FT-IR spectrophotometer (PerkinElmer Ltd., Buckinghamshire, UK) with an ATR accessory and a resolution of 4 cm −1 [25].
Differential scanning calorimetry (DSC) profiles of the samples were obtained using a TA Instruments DSC Q2000 (TA Instruments, Eschborn, Germany) with around 5 mg samples. A 10 • C/min ramp and 20 mL/min nitrogen flow were used for all runs. Samples were cooled and heated from 0 to 300 • C in sealed aluminum pans with holes. Thermogravimetric analyses (TGA) were carried out on a TGA Q5000 instrument (TA Instruments, Eschborn, Germany). First, 5 mg of samples on platinum pans were heated from 40 to 800 • C. A 10 • C/min ramp and 25 mL/min of nitrogen flow were used for the measurements. 1 H NMR spectra were registered at 25 • C in solution (DMSO-d 6 , δ = 2.51 ppm) [31] in a Bruker Avance III 300 spectrometer (Bruker, Faellanden, Switzerland) at 300 MHz frequency.
UV-Vis spectra were recorded on a Perkin Elmer Lambda 25 UV/VIS spectrophotometer (PerkinElmer Ltd., Buckinghamshire, UK). Dynamic light scattering (DLS) determinations were carried out on a Malvern Zetasizer Nano ZS (Malvern Instruments Ltd., Malvern, UK) operated at 173 • . Measurements were made in triplicate at room temperature. The SF aggregates in water or PBS at 0.5 mg/mL were filtered through a 0.45 µm nylon filter and measured in a DTS1070 disposable cuvette. Transmission Electron Microscopy (TEM) images were registered using a Jeol JEM-2011 FasTEM (Jeol Ltd., Tokyo, Japan) at 100 kV [25]. SF solutions were placed on Pioloform-coated copper grids, using negative staining with 1% uranyl acetate for contrast enhancement. TEM images were processed, and the size of aggregates and particles were measured using Gatan Digital Micrograph 3.10.1 (Pleasanton, CA, USA).
Atomic Force Microscopy (AFM) images (2 µm × 2 µm) were taken with an MFP 3D-Stand Alone AFM (Asylum Research, Oxford, UK) with the Olympus cantilever OMCL-AC160TSA (Olympus Europa SE & Co. KG, Hamburg, Germany) at a resonant frequency of 300 kHz and spring constant of 26 N/m, 50-70% set point, and scan rate of 1 Hz [25]. A 70 µL droplet of 1 mg/mL water solution of SF conjugates was placed on a silicon wafer and spin coated at 40 Hz for 6 s. AFM images were processed and statistical parameters of particles were obtained using Gwyddion 2.45 (Brno, Czech Republic).

Viability Assay
To determine the effect of the nanoparticles on the viability of the human breast adenocarcinoma cell line MCF-7, 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide assay (XTT) (Sigma-Aldrich, St. Louis, MO, USA) was used. Cells were grown in Dulbecco's modified Eagle's medium (DMEM), with 10% fetal bovine serum, 1% penicillin-streptomycin, and 1% L-glutamine. The MCF-7 cell line was a gift from Prof. Dr. Barbara Krammer, Paris-Lodron-University, Salzburg, Austria. Cell suspensions were cultivated in 96-well tissue culture plates in triplicate (4 × 10 4 cells/well). The day after, cells were exposed to dispersions of SF aggregates loaded or not with CPT at various concentrations in serum-free medium. After an additional incubation period of 48 h, the medium was discarded, and the cells were supplemented with complete DMEM with 50 µL of XTT. After 3 h at 37 • C in the dark, the absorbance at 490 nm was measured with a GloMax ® Multimode Microplate Reader (Promega, Madison, WI, USA). Three independent experiments were performed. Samples were normalized to untreated controls. Data were analyzed using GraphPad Prism 6.0 (San Diego, CA, USA).

Cell Uptake
For cell uptake analysis, 3 × 10 4 cells in complete growth medium were seeded into a 8-well µ-slide (ibidiTreat, Ibidi, Graefeling, Germany). The next day, medium was aspirated, and CPT-loaded SF aggregate dispersions at 0.1 mg/mL in serum-free medium were added. After 4 h of incubation at 37 • C in the dark, cells were exposed to 50 nM LysoTracker Yellow HCK-123 (Invitrogen, Waltham, MA, USA) for an additional 2 h. Subsequently, cells were imaged using an Olympus IX73 inverted microscope. For the CPT-loaded SF aggregates, the DAPI channel (λ excitation = 345 nm and λ emission = 455 nm) was used, and for the LysoTracker Yellow HCK-123, the FITC channel (λ excitation = 494 nm and λ emission = 518 nm) was used.

Annexin V/PI Assay
MCF-7 cells (1 × 10 5 cells/well) were plated overnight in 12-well culture plates. Subsequently, the medium was removed and serum-free medium was added, containing the compounds at various concentrations. After 48 h, supernatants were transferred to Eppendorf tubes. Cells were harvested with 500 µL of Accutase ® solution (Sigma-Aldrich, St. Louis, MO, USA) and mixed with the supernatant. The suspension was centrifuged at 1500 rpm for 5 min and washed twice with 1 mL of DPBS. The obtained pellet was stained for 15 min in 98.5 µL of 1X Annexin V Buffer with 1 µL of Annexin V-APC (Immuno-Tools, Friesoythe, Germany) and 0.5 µL of 1 mg/mL propidium iodide (PI, Sigma-Aldrich, St. Louis, MO, USA). Subsequently, 200 µL of 1X Annexin V Buffer were added and measured with the CytoFLEX flow cytometer (Beckman Coulter, Brea, CA, USA). At least three different experiments were conducted and analyzed with the Kaluza 1.5a software (Beckman Coulter, Brea, CA, USA).

Cell Cycle Analysis
First, 5 × 10 5 cells/well in 6-well plates grown overnight were treated with nanoparticles at different concentrations for 72 h. Afterwards, the supernatants were transferred to Eppendorf tubes, and cells were detached with 500 µL Accutase ® solution. After centrifugation (5 min at 1500 rpm), the cell pellet was washed with 1 mL of DPBS. The supernatant was discarded, and the pellet was re-suspended in 100 µL DPBS. To fix the cells, 1 mL of ice-cold 70% ethanol was added.

Spheroid Generation
First, 1 × 10 4 cells were seeded in a BIOFLOAT FLEX (faCellitate, Mannheim, Germany) coated non-TC treated U-bottom 96-well plate. Subsequently, the plate was centrifuged for 5 min at 300× g, to facilitate cell aggregation and subsequent spheroid formation.

Spheroid Uptake
After an incubation period of 4 days, the spheroids were treated with 0.1 mg/mL of the indicated compounds. After 72 h, the spheroids were washed and stained with PI for 10 min to visualize dead cells. Cells were counterstained using Hoechst 33342 (Fluka, Buchs, Switzerland). Fluorescence imaging was performed with an Olympus IX73 inverted microscope provided with DAPI channel for Hoechst 33342 and the CPTloaded SF aggregates (λ excitation = 345 nm and λ emission = 455 nm) and Cy3 channel for PI (λ excitation = 550 nm and λ emission = 565 nm).

Cell Viability Analysis of Spheroids
After an incubation period of 4 days, the spheroids were treated with increasing concentrations of the compounds in serum-free medium. To measure the cell viability with the CellTiter-Glo ® 3D Cell Viability Assay (Promega, Madison, WI, USA), the spheroids were deposited into a 96-well plate after 72 h and processed as described in manufacturer's protocol. The Luminescence was measured with the GloMax (Promega, Madison, WI, USA).

Hemolysis Assay
First, 4 mL of whole blood were obtained from a healthy human donor, which were directly drawn into Vacuette ® EDTA tubes (Greiner-Bio-One, Kremsmuenster, Austria) to prevent coagulation. The blood was centrifuged for 5 min at 2500× g to obtain a red blood cell (RBC) pellet, the plasma was discarded and the RBCs were washed twice with 150 mM NaCl (Merck, Darmstadt, Germany) and in triplicate with PBS at pH 7.4. Consequently, the RBCs were diluted 1:50 with PBS at pH 6.2 or pH 7.4 and 200 µL RBC suspension were incubated with the nanoparticles at 37 • C for 24 h, without CO 2 . Samples incubated with 1% Triton-X (Merck, Darmstadt, Germany) or PBS pH 7.4 instead of nanoparticles served as positive and negative control, respectively. Samples were centrifuged at 500× g for 5 min, and 100 µL of the supernatant was placed into a flat-bottom 96-well plate. Absorbance was determined with a GloMax microplate reader (Promega, Madison, WI, USA) at 405 nm. The percent hemolysis was determined as follows: % hemolysis = (Absorbance sample − Absorbance negative control) Absorbance positive control × 100

Statistical Analyses
All data are normalized and reported as average value ± standard deviation. Measurements were evaluated using a one-way ANOVA with Tukey's post-test for multiple comparison analysis (Statgraphics Plus 5.1). Statistically significant sample means were defined with a significance threshold set at 5% (p < 0.05), where * or no marker at all is used for means with significant differences (p < 0.05), and ns is used for means with no significant differences (p > 0.05). Figure 1 illustrates the structures of vitamin-and testosterone-grafted SF conjugates synthesized via the ester formation of tyrosine moieties in SF with vitamins and testosterone. Tyrosine residues of SF were chosen as the target point for the synthesis of SF conjugates because of the higher nucleophilicity of its -OH groups and the high tyrosine content in SF (approximately 5 mol %) [27,32]. Approximately 70-80% of tyrosine residues in SF were esterified with the vitamin hemisuccinates, while testosterone hemisuccinate reacted with about 9% of the tyrosine fragments (Table 1). Testosterone hemisuccinate was much less reactive than vitamin hemisuccinates in the esterification of SF, as previously observed in the esterification of cellulose ethers [16]. DS values estimated from 1 H NMR for SF1 and SF2 were similar to the values calculated from elemental analyses. The effect of the hydrophobic functionalization of SF on the SF structure, morphology, and properties was thoroughly studied.

Structural Characterization and Morphological Studies
IR spectroscopy confirmed the functionalization of SF with vitamins and testosterone hemisuccinates. C=O peaks of ester bonds and free carboxylic acid groups in vitamins and testosterone hemisuccinates are observed between 1728-1746 cm −1 and 1695-1702 cm −1 , respectively ( Figure S4) [16,33,34]. Once the tyrosine residues in SF were esterified with vitamins and testosterone hemisuccinates, C=O peaks of the vitamin-and testosteronegrafted SF ester bonds were observed at 1732-1739 cm −1 (Figure 2a, characterization data in Supplementary Material) [16,25]. SF with intense peaks observed at 1535 and 1645 cm −1 (silk I) must adopt random coil and β-turns structures [19]. Synthesized SF1-SF3 must consist of a β-sheet structure, as evidenced with intense IR peaks at 1515 and 1625 cm −1 (silk II) [19]. The effect of vitamins and testosterone on the thermal behavior and related properties of SF was also studied (Figure 2b,c). SF1-SF3 exhibited an endothermic peak at

Structural Characterization and Morphological Studies
IR spectroscopy confirmed the functionalization of SF with vitamins and testosterone hemisuccinates. C=O peaks of ester bonds and free carboxylic acid groups in vitamins and testosterone hemisuccinates are observed between 1728-1746 cm −1 and 1695-1702 cm −1 , respectively ( Figure S4) [16,33,34]. Once the tyrosine residues in SF were esterified with vitamins and testosterone hemisuccinates, C=O peaks of the vitamin-and testosteronegrafted SF ester bonds were observed at 1732-1739 cm −1 (Figure 2a, characterization data in Supplementary Material) [16,25]. SF with intense peaks observed at 1535 and 1645 cm −1 (silk I) must adopt random coil and β-turns structures [19]. Synthesized SF1-SF3 must consist of a β-sheet structure, as evidenced with intense IR peaks at 1515 and 1625 cm −1 (silk II) [19]. The effect of vitamins and testosterone on the thermal behavior and related properties of SF was also studied (Figure 2b,c). SF1-SF3 exhibited an endothermic peak at 66-71 • C with a peak enthalpy (∆H) of 149-213 J/g and related weight reduction of 2.4-4.2% (Figure 2b,c, and Figures S5 and S6). This peak might be ascribed to the loss of SFbound water [35]. Parent SF showed an endothermic peak at 80 • C with a related enthalpy of 209.4 J/g and 7.9% weight loss (Figure 2b and Figure S7) [25]. The glass transition of native SF was detected at 182 • C, which is near the 178 • C for amorphous SF [35,36]. This glass transition characteristic of SF (silk I) is associated to changes in the β-turns and random coils [35]. It was not detected in the SF1-SF3 conjugates. Partial pyrolysis of SF was detected in native and modified SF as endothermic peaks at 266-280 • C, with ∆H of 43 J/g and 100-133 J/g (Figure 2b). The total pyrolysis of raw and modified SF provoked a major weight loss of 91-96% (Figure 2c and Figures S5-S7).
H28 in ergocalciferol; 4′CH3-, 8′CH3-, and 12′CH3-in tocopherol), and at 5.27-5.35 ppm (=CH-CH= group H6 and H7 in ergocalciferol) (Figure 2 and Figure S9) [16]. The proton NMR spectra of SF3 also showed the characteristic testosterone peaks at 0.84-0.86 ppm (CH3groups H18 and H19 in testosterone) and at 5.34 ppm (O=CC-CH= group H4 in testosterone) ( Figure S10). However, the low signal to noise ratio in the 1 H NMR spectra of SF3 resulted in an underestimated integration of tyrosine aromatic protons, and a proper DS estimation was not suitable. DLS studies revealed the effect of SF esterification with vitamins and testosterone on the hydrodynamic size of SF aggregates and their zeta potential in water. Table 2 shows that the esterification of SF with the vitamins or testosterone provoked a significant change of their hydrodynamic parameters. Almost doubled or tripled hydrodynamic  Figure 1). 1 H NMR spectroscopy of vitamin-and testosterone-grafted SF in solution also corroborated the functionalization with vitamins and testosterone hemisuccinates of SF tyrosine amino acids (Figure 2d). The characteristic tyrosine protons appearing at 6.7-8.2 ppm ( Figure S8) were taken as a reference to calculate the DS of SF1 and SF2, with the aid of the distinctive vitamin signals at 0.82-1.02 ppm (CH 3 -groups H18, H21, H26, H27, and H28 in ergocalciferol; 4 CH 3 -, 8 CH 3 -, and 12 CH 3 -in tocopherol), and at 5.27-5.35 ppm (=CH-CH= group H6 and H7 in ergocalciferol) (Figure 2d and Figure S9) [16]. The proton NMR spectra of SF3 also showed the characteristic testosterone peaks at 0.84-0.86 ppm (CH 3 -groups H18 and H19 in testosterone) and at 5.34 ppm (O=CC-CH= group H4 in testosterone) ( Figure S10). However, the low signal to noise ratio in the 1 H NMR spectra of SF3 resulted in an underestimated integration of tyrosine aromatic protons, and a proper DS estimation was not suitable.
DLS studies revealed the effect of SF esterification with vitamins and testosterone on the hydrodynamic size of SF aggregates and their zeta potential in water. Table 2 shows that the esterification of SF with the vitamins or testosterone provoked a significant change of their hydrodynamic parameters. Almost doubled or tripled hydrodynamic diameters and doubled zeta potential values were observed after the esterification of tyrosine residues on SF chains. Table 2. Hydrodynamic parameters of SF aggregates; AFM average diameters of raw and modified SF dried particles; CPT contents in weight (wt %) and CPT encapsulation efficiency (EE %) of CPT-loaded SF aggregates. SF1-SF3 aggregates presented hydrodynamic sizes and zeta potential values in water of 554 to 653 nm and −28.3 to −32.2 mV, respectively. It seems that the intermolecular hydrophobic interaction of vitamins or testosterone groups grafted on SF chains caused stabilization of the SF aggregates in aqueous medium, with zeta potentials around −30 mV [37]. In general, the hydrodynamic sizes increased in PBS for native SF and SF aggregates when compared to the same aggregates dispersed in distilled water, except for SF1 aggregates. On the other hand, the hydrophobic loading of CPT in the inner core of SF aggregates resulted in a significant hydrodynamic sizes reduction of 11% to 59%, which was probably due to the stronger hydrophobic interactions of CPT molecules with vitamins and testosterone grafted on SF chains.

Sample dh a (nm)/(PDI) dh b (nm)/(PDI)
The strong hydrophobic interactions between CPT and lipophilic vitamins and testosterone allowed achieving a high CPT content in the prepared CPT-SF1 to CPT-SF3 aggregates, with encapsulation efficiencies (EE % = ( weight o f CPT loaded in particles weight o f f eeding CPT ) × 100) ranging from 60% to 84% (Table 2). Particularly interesting was the high CPT content found in CPT-SF3 aggregates and the associated high encapsulation efficiency achieved (67%) in spite the low testosterone content related to a DS of 0.4 mol % in SF3 conjugates. However, CPT-SF3 aggregates exhibited the smaller hydrodynamic diameters in aqueous medium. This might be due to the additional π − π interactions between the phenyl groups of tyrosine residues in SF with the aromatic backbone of CPT. SF1-SF3 dried aggregates appeared on TEM as irregular grain aggregates of around 50-800 nm (Figure 3a), with most of the small SF particles observed as agglomerates with undefined borders ( Figure S11). AFM showed SF nanoaggregates with 45 to 75 nm sizes for SF1-SF3 (Table 2, Figure 3b, Table S1 and additional AFM images in Supplementary Material). It must be noted that the size and distribution of dried SF aggregates is better calculated from AFM measurements due to a proper statistical analysis of SF samples (a representative number of SF particles is present in the 2 µm × 2 µm AFM images). Furthermore, the spin-drying process of AFM samples ensures a better dispersion of SF aggregates before measurements.

Drug Release
CPT delivery of CPT-loaded SF aggregates in PBS (pH 7.4) at 37 °C is presented in Figure 4. All CPT releases appeared linear with a slope 1.23 to 2.44%/h during the initial 8 h (Figure 4 and Table S2). The almost linear CPT release rate profiles observed during the initial 8 h can be due to a frequent dialysate replacement with fresh PBS. A frequent replacement of dialysate that contains the poorly soluble CPT with fresh release medium must affect the diffusion rate of CPT from the core of the SF aggregates and through the dialysis membrane. CPT was almost quantitatively released after 6 days. The almost quantitative CPT release after 6 days in PBS is due to the repeated replacement of the dialysate in every time point studied, which forced the diffusion of almost all CPT encapsulated in the SF aggregates after 10 replacements of the medium release (10 mL of PBS). CPT-SF2 and CPT-SF3 achieved a CPT release of approximately 97%, while the bigger CPT-SF1 aggregates carrying more CPT released 76% of the encapsulated anticancer drug. Furthermore, all the CPT release profiles adjusted well to the Weibull model that describes the drug release from a matrix [38,39]. To this end, kinetics data were fitted following a SWeibull2 distribution (Cumulative Release (%) = a − (a − b)*exp(−(k*Time(hours)) d ), with adjusted R 2 ranging from 0.9760 to 0.9918 and d values of 1.1 and 1.2 associated to a complex drug release mechanism (Figure 4 and Table S3) [39]. The CPT release kinetic data were also adjusted to the Korsmeyer-Peppas model (log(Cumulative Release (%)) = k*log(Time(hours) + m) to get an insight into the molecular interactions and factors dominant on the CPT release behavior (Table S3 and Figure S12) [16,38,40]. The slope of the linear fitting ranged from 0.72 to 0.78, which is related to a non-Fickian drug release mechanism (anomalous diffusion mechanism) [40]. Further evaluation of the interaction between the anticancer drug with the hydrophobically modified SF matrix was carried out according to the changes of the IR peak of CPT at 1737 cm −1 in the CPT-loaded SF aggregates (Table S4 and Figure S13). To this end, the intensity of IR adsorption peaks at 1732 cm −1 and 1739 cm −1 in blank and CPT-loaded SF aggregates were normalized to the characteristic 1515 cm −1 adsorption peak in all spectra. The IR adsorption peaks at 1732 cm −1 and 1739 cm −1 of CPT-loaded SF aggregates exhibited a 0.4 to 1.6 increase in intensity when compared to blank SF1-SF3 aggregates ( Figure S13). It might be due to the overlapping of the C=O adsorption peak at 1732 cm −1 and 1739 cm −1 of blank SF1 and SF2, or SF3 aggregates respectively, with the intense CPT adsorption in the CPT-loaded aggregates. A significant shift of the CPT adsorption peak from 1737 cm −1 in parent CPT to 1732 cm −1 in

Drug Release
CPT delivery of CPT-loaded SF aggregates in PBS (pH 7.4) at 37 • C is presented in Figure 4. All CPT releases appeared linear with a slope 1.23 to 2.44%/h during the initial 8 h (Figure 4 and Table S2). The almost linear CPT release rate profiles observed during the initial 8 h can be due to a frequent dialysate replacement with fresh PBS. A frequent replacement of dialysate that contains the poorly soluble CPT with fresh release medium must affect the diffusion rate of CPT from the core of the SF aggregates and through the dialysis membrane. CPT was almost quantitatively released after 6 days. The almost quantitative CPT release after 6 days in PBS is due to the repeated replacement of the dialysate in every time point studied, which forced the diffusion of almost all CPT encapsulated in the SF aggregates after 10 replacements of the medium release (10 mL of PBS). CPT-SF2 and CPT-SF3 achieved a CPT release of approximately 97%, while the bigger CPT-SF1 aggregates carrying more CPT released 76% of the encapsulated anticancer drug. Furthermore, all the CPT release profiles adjusted well to the Weibull model that describes the drug release from a matrix [38,39]. To this end, kinetics data were fitted following a SWeibull2 distribution (Cumulative Release (%) = a − (a − b)*exp(−(k*Time(hours)) d ), with adjusted R 2 ranging from 0.9760 to 0.9918 and d values of 1.1 and 1.2 associated to a complex drug release mechanism ( Figure 4 and Table S3) [39]. The CPT release kinetic data were also adjusted to the Korsmeyer-Peppas model (log(Cumulative Release (%)) = k*log(Time(hours) + m) to get an insight into the molecular interactions and factors dominant on the CPT release behavior (Table S4 and Figure S12) [16,38,40]. The slope of the linear fitting ranged from 0.72 to 0.78, which is related to a non-Fickian drug release mechanism (anomalous diffusion mechanism) [40]. Further evaluation of the interaction between the anticancer drug with the hydrophobically modified SF matrix was carried out according to the changes of the IR peak of CPT at 1737 cm −1 in the CPT-loaded SF aggregates (Table S4 and Figure S13). To this end, the intensity of IR adsorption peaks at 1732 cm −1 and 1739 cm −1 in blank and CPT-loaded SF aggregates were normalized to the characteristic 1515 cm −1 adsorption peak in all spectra. The IR adsorption peaks at 1732 cm −1 and 1739 cm −1 of CPT-loaded SF aggregates exhibited a 0.4 to 1.6 increase in intensity when compared to blank SF1-SF3 aggregates ( Figure S13). It might be due to the overlapping of the C=O adsorption peak at 1732 cm −1 and 1739 cm −1 of blank SF1 and SF2, or SF3 aggregates respectively, with the intense CPT adsorption in the CPT-loaded aggregates. A significant shift of the CPT adsorption peak from 1737 cm −1 in parent CPT to 1732 cm −1 in CPT-SF1 aggregates was observed. It is probably due to slightly stronger interactions between CPT and the SF1 polymer matrix. On the other hand, CPT-SF2 and CPT-SF3 showed a small shifting of CPT peak to 1739 cm −1 . The key factor that determines the CPT release rate and total CPT delivered after 6 days (drug delivery capability) seems to be the structure of the synthesized SF aggregates, which affects the CPT-polymer matrix interactions. SF1 aggregates with β-turns, random coils, and β-sheet conformers (silk I and silk II structure) are able to better accommodate the hydrophobic CPT in the lipophilic core, which facilitates the CPT-polymer matrix interactions when compared to the SF2 and SF3 aggregates (silk II structure), with a predominant content of β-sheet conformers and β-sheet nanocrystals. That is why the SF1 aggregates showed more CPT encapsulated and slower CPT release rate when compared to the SF2 and SF3 aggregates. CPT-SF1 aggregates was observed. It is probably due to slightly stronger interactions between CPT and the SF1 polymer matrix. On the other hand, CPT-SF2 and CPT-SF3 showed a small shifting of CPT peak to 1739 cm −1 . The key factor that determines the CPT release rate and total CPT delivered after 6 days (drug delivery capability) seems to be the structure of the synthesized SF aggregates, which affects the CPT-polymer matrix interactions. SF1 aggregates with β-turns, random coils, and β-sheet conformers (silk I and silk II structure) are able to better accommodate the hydrophobic CPT in the lipophilic core, which facilitates the CPT-polymer matrix interactions when compared to the SF2 and SF3 aggregates (silk II structure), with a predominant content of β-sheet conformers and βsheet nanocrystals. That is why the SF1 aggregates showed more CPT encapsulated and slower CPT release rate when compared to the SF2 and SF3 aggregates.  Figure 5 displays the viability of MCF-7 cells (normalized to untreated control) once treated with CPT-loaded SF aggregates or parent CPT. Non-loaded SF was not toxic to MCF-7 cells at 0.2 mg/mL, with a viability of (99 ± 5)% ( Figure S14). Similarly, SF1-SF3 aggregates were not cytotoxic to MCF-7 (89-94% relative cell viability) at 0.1 mg/mL (Figure S14).

Cytotoxic Activity and Cell Uptake
CPT hydrophobically loaded in SF aggregates provoked a significant cytotoxic effect, which was not observed with blank SF aggregates. Particularly, CPT-SF2 with a CPT concentration between 10 −2.8 and 10 −2.2 mg/mL exhibited similar to slightly better cytotoxic activity than parent CPT evaluated at 10 −2.6 to 10 −2 mg/mL. However, all CPT-loaded SF aggregates showed a slight to moderate cytotoxic effect (relative cell viability of 85% to 60%) when applied at CPT concentrations between 10 −4 and 10 −3 mg/mL.  Figure 5 displays the viability of MCF-7 cells (normalized to untreated control) once treated with CPT-loaded SF aggregates or parent CPT. Non-loaded SF was not toxic to MCF-7 cells at 0.2 mg/mL, with a viability of (99 ± 5)% ( Figure S14). Similarly, SF1-SF3 aggregates were not cytotoxic to MCF-7 (89-94% relative cell viability) at 0.1 mg/mL ( Figure S14).

Cytotoxic Activity and Cell Uptake
CPT hydrophobically loaded in SF aggregates provoked a significant cytotoxic effect, which was not observed with blank SF aggregates. Particularly, CPT-SF2 with a CPT concentration between 10 −2.8 and 10 −2.2 mg/mL exhibited similar to slightly better cytotoxic activity than parent CPT evaluated at 10 −2.6 to 10 −2 mg/mL. However, all CPT-loaded SF aggregates showed a slight to moderate cytotoxic effect (relative cell viability of 85% to 60%) when applied at CPT concentrations between 10 −4 and 10 −3 mg/mL. Polymers 2021, 13, x FOR PEER REVIEW 12 of 18  Figure 1). Mean ± standard deviation (n = 3). * represent means with significant differences (p < 0.05).
To determine the effect of CPT-loaded SF aggregates on cell cycle distribution and cell death induction, Annexin V and PI staining were conducted after 48 h of treatment. CPT-loaded aggregates induced cell cycle arrest in the G2/M phase at a concentration of 0.27 mg/mL. In addition, this concentration resulted in significant cell death. CPT-SF nanoaggregates showed a superior cytotoxic effect on MCF-7 cells when compared to parent CPT, with about 30% to 56% of cell death in CPT-loaded SF aggregates and ~ 20% in parent CPT (Table S5). Interestingly, parent CPT provoked cell cycle arrest in the G2/M phase but did not induce cell death at the given concentration (Tables 3 and S5).   Figure 1). Mean ± standard deviation (n = 3). * represent means with significant differences (p < 0.05).
To determine the effect of CPT-loaded SF aggregates on cell cycle distribution and cell death induction, Annexin V and PI staining were conducted after 48 h of treatment. CPT-loaded aggregates induced cell cycle arrest in the G2/M phase at a concentration of 0.27 mg/mL. In addition, this concentration resulted in significant cell death. CPT-SF nanoaggregates showed a superior cytotoxic effect on MCF-7 cells when compared to parent CPT, with about 30% to 56% of cell death in CPT-loaded SF aggregates and~20% in parent CPT (Table S5). Interestingly, parent CPT provoked cell cycle arrest in the G2/M phase but did not induce cell death at the given concentration (Table 3 and Table S5). Table 3. MCF-7 cell cycle arrest parameters at control and 0.27 mg/mL of CPT-loaded SF nanoaggregates and parent CPT.
Polymers 2021, 13, x FOR PEER REVIEW 13 of 18 CPT 40 ± 2 4 ± 2 55 ± 1 Figure 6 shows the hemolytic activity (% hemolysis) of CPT-loaded SF nanoaggregates and parent CPT on human red blood cells (RBC) at approximately 1 µg/mL of CPT in PBS (pH 6.2 and 7.4). Figure 6. Percentage of hemolysis of RBC treated with CPT-loaded SF nanoaggregates and parent CPT in PBS at pH 7.4 and 6.2 (see structures in Figure 1). Mean ± standard deviation (n = 3). ns represent means with no significant differences (p > 0.05), but significant differences exist between non-marked bars (p < 0.05).
For that reason, SF aggregates can be considered as good candidates for the in vivo delivery of CPT. CPT-loaded SF aggregates might be directly injected into the tumor. Alternatively, CPT carrying SF aggregates could selectively accumulate in the cancer tissues via passive diffusion and the EPR effect [41,42]. Particularly, CPT-SF2 and CPT-SF3 soft aggregates might exhibit extended serum residence times and moderate reticuloendothelial system (RES) clearance [43]. Once CPT-SF2 or CPT-SF3 aggregates are inside the cancer cells or in the slightly acidic cancer tissue, the CPT cargo might be delivered faster because of enzyme-promoted and acidic hydrolysis degradation of SF-tocopherol and SFtestosterone carriers. Additional in vivo assessment of the bioavailability, biodistribution, Figure 6. Percentage of hemolysis of RBC treated with CPT-loaded SF nanoaggregates and parent CPT in PBS at pH 7.4 and 6.2 (see structures in Figure 1). Mean ± standard deviation (n = 3). ns represent means with no significant differences (p > 0.05), but significant differences exist between non-marked bars (p < 0.05).
For that reason, SF aggregates can be considered as good candidates for the in vivo delivery of CPT. CPT-loaded SF aggregates might be directly injected into the tumor. Alternatively, CPT carrying SF aggregates could selectively accumulate in the cancer tissues via passive diffusion and the EPR effect [41,42]. Particularly, CPT-SF2 and CPT-SF3 soft aggregates might exhibit extended serum residence times and moderate reticuloendothelial system (RES) clearance [43]. Once CPT-SF2 or CPT-SF3 aggregates are inside the cancer cells or in the slightly acidic cancer tissue, the CPT cargo might be delivered faster because of enzyme-promoted and acidic hydrolysis degradation of SF-tocopherol and SFtestosterone carriers. Additional in vivo assessment of the bioavailability, biodistribution, antitumor activity, toxicity, and pharmacokinetics of the CPT-loaded SF aggregates must be carried out before their medical application.
To evaluate the uptake and localization of the CPT-loaded SF aggregates on MCF-7 cancer cells, fluorescence microscopy analyses were performed. Indeed, CPT-SF1 (Figure 7), CPT-SF2 and CPT-SF3 (Figures S16 and S17) internalized in cells (CPT associated with blue fluorescence) and accumulated in the lysosomes (LysoTracker HCK-123 associated with green fluorescence) [44]. The lysosomal accumulation of the CPT-loaded SF aggregates might facilitate the acidic hydrolysis of the vitamin-tyrosine and testosterone-tyrosine ester bonds in the SF aggregates. Thus, the co-delivery of vitamins, testosterone, and CPT inside the cancer cells is expected due to SF aggregates disintegration in the cancer cell lysosomes. antitumor activity, toxicity, and pharmacokinetics of the CPT-loaded SF aggregates must be carried out before their medical application.
To evaluate the uptake and localization of the CPT-loaded SF aggregates on MCF-7 cancer cells, fluorescence microscopy analyses were performed. Indeed, CPT-SF1 ( Figure  7), CPT-SF2 and CPT-SF3 (Figures S16 and S17) internalized in cells (CPT associated with blue fluorescence) and accumulated in the lysosomes (LysoTracker HCK-123 associated with green fluorescence) [44]. The lysosomal accumulation of the CPT-loaded SF aggregates might facilitate the acidic hydrolysis of the vitamin-tyrosine and testosterone-tyrosine ester bonds in the SF aggregates. Thus, the co-delivery of vitamins, testosterone, and CPT inside the cancer cells is expected due to SF aggregates disintegration in the cancer cell lysosomes.

Spheroid Uptake and Cytotoxic Activity
MCF-7 cells were grown in 3D-like architecture as spheroids to assess the uptake and cytotoxicity of synthesized CPT-loaded SF nanoaggregates as a step forward to their intended medical use against human breast cancer. Fluorescence microscopy images (Figures 8b and S18) show a good uptake of CPT-loaded SF nanoaggregates in MCF-7 spheroids (CPT associated blue fluorescence). It caused an extensive cell dead (PI related red fluorescence) after 72 h.

Spheroid Uptake and Cytotoxic Activity
MCF-7 cells were grown in 3D-like architecture as spheroids to assess the uptake and cytotoxicity of synthesized CPT-loaded SF nanoaggregates as a step forward to their intended medical use against human breast cancer. Fluorescence microscopy images (Figure 8 and Figure S18) show a good uptake of CPT-loaded SF nanoaggregates in MCF-7 spheroids (CPT associated blue fluorescence). It caused an extensive cell dead (PI related red fluorescence) after 72 h.
Cell death in MCF-7 spheroids treated with the CPT-loaded SF nanoaggregates was observed as a change of the spheroids borderline after 72 h when compared with no significant shape change of control ( Figures S19 and S20).

Conclusions
In this research, ergocalciferol-, tocopherol-, and testosterone-grafted SFs were prepared via the esterification of tyrosine fragments on SF, and CPT was efficiently encapsulated for its sustained delivery. The vitamins and testosterone grafting of SF attained a DS between 0.4 and 3.8 mol %, as determined by elemental analysis and 1 H NMR spectroscopies. FTIR spectra and DSC thermograms showed that tocopherol and testosteronegrafted SF conjugates predominantly adopted a β-sheet conformation. The functionalization of SF with the vitamins and testosterone resulted in the almost doubled hydrodynamic sizes and zeta potentials of vitamins-and testosterone-grafted SF aggregates as compared to parent SF. TEM images showed 50-800 nm rounded and irregular flakelike SF aggregates, which is in good agreement with DLS results. The dialysis-precipitation method allowed efficient CPT encapsulation in the SF aggregates, with a CPT content

Conclusions
In this research, ergocalciferol-, tocopherol-, and testosterone-grafted SFs were prepared via the esterification of tyrosine fragments on SF, and CPT was efficiently encapsulated for its sustained delivery. The vitamins and testosterone grafting of SF attained a DS between 0.4 and 3.8 mol %, as determined by elemental analysis and 1 H NMR spectroscopies. FTIR spectra and DSC thermograms showed that tocopherol and testosteronegrafted SF conjugates predominantly adopted a β-sheet conformation. The functionalization of SF with the vitamins and testosterone resulted in the almost doubled hydrodynamic sizes and zeta potentials of vitamins-and testosterone-grafted SF aggregates as compared to parent SF. TEM images showed 50-800 nm rounded and irregular flake-like SF aggregates, which is in good agreement with DLS results. The dialysis-precipitation method allowed efficient CPT encapsulation in the SF aggregates, with a CPT content of 6.3-8.3 wt %. A significant shrinkage of SF aggregates in PBS was observed once CPT was encapsulated in the hydrophobic core. Controlled and almost quantitative CPT release was achieved after 6 days in PBS at 37 • C, with almost linear release during the first 8 h. The antiproliferative effect of CPT on MCF-7 cancer cells seems unaltered after encapsulation in the SF aggregates. Particularly, CPT-loaded tocopherol-grafted SF aggregates showed slightly more cytotoxicity against MCF-7 cells than parent CPT. CPT-loaded SF aggregates provoked significantly reduced MCF-7 cell viability, with much lower IC 50 values than parent CPT in MCF-7 spheroids. It might indicate a better uptake of the loaded SF aggregates. All CPT-loaded SF aggregates caused significant cell cycle arrest in the G2/M phase with significant cell death, whereas parent CPT provoked significant cell cycle arrest in the G2/M phase and slight necrosis. No hemolytic activity was observed for the blank SF aggregates at pH 7.4 and 6.2. CPT-loaded SF aggregates provoked no hemolysis at pH 7.4 and caused similar or slightly lower hemolysis of human RBC than parent CPT at pH 6.2.
Fluorescence microscopy imaging confirmed MCF-7 cells uptake and lysosomal accumulation of CPT-loaded SF aggregates as well as good cellular uptake in the MCF-7 spheroids. Accordingly, CPT-loaded SF aggregates are good candidates for antitumor therapy. Future studies may focus on the in vivo evaluation of antitumor activity, toxicity, biodistribution, and other pharmacokinetic parameters.