A novel method to prepare tussah/Bombyx mori silk fibroin-based films

The possibility of using silk fibroin in biomaterials for tissue engineering is a subject of broad interest. In this study, Bombyx mori/tussah silk fibroin (BSF/TSF) blend films were prepared by solution casting using CaCl2/formic acid as a co-solvent and water as a rinse solvent. The morphology, crystallinity, thermal resistance, mechanical properties and water contact angle of the blend films as well as the biocompatibility were investigated. The BSF/TSF blend films displayed a smooth surface and specific nanostructure in their cross-section, originating from the nanofibril-preservation during fibroin dissolution. The water rinse process induced the formation of a stable β-sheet structure. The BSF film showed superior mechanical properties to the TSF film, and the blending with TSF led to a significant reduction in the strength and elasticity of blend films. However, adding the TSF component could regulate the hydrophilic properties and enhance cell growth on the blend films. The BSF/TSF blend films with specific nanostructure, stable secondary structure, appropriate mechanical properties, and good biocompatibility, are promising candidates for application in regenerative medicine.


Introduction
Silk is a natural polymer produced by domestic and wild silkworms, which has been used in the textile industry for thousands of years. 1,2 Recently, silk broin has been widely studied for potential application in biomedical and biotechnological areas because of its excellent mechanical properties, biological compatibility, and biodegradation. 3,4 Bombyx mori silk broin (BSF), the most investigated broin-based biomaterial, has been processed into various forms for engineering of bone, vascular, neural, skin, cartilage, ligaments, cardiac and bladder tissue. 5 Although giving generally promising results in terms of cell adhesion and proliferation, and tissue growth, the ability of BSF materials to promote early cell adhesion is insufficient due to the lack of specic functional molecules. 6 The incorporation of the Arg-Gly-Asp (RGD) sequence, a cell-attachment recognition signal, has been a widely used strategy to improve the celladhesive properties of BSF. 7 Compared with BSF, tussah silk broin (TSF) molecule inherently contains the RGD sequence and more amino acids with positive charge which favors cell adhesion. 8 Therefore, blending BSF and TSF is a potential strategy to prepare silk broin-based material with excellent mechanical properties and improve biocompatibility. To prepare BSF/TSF blend lms, the dissolution of BSF and TSF are required. A series of solvents has been successfully developed to dissolve BSF, such as aqueous lithium salt solution, CaCl 2 /ethanol/water solution. 9 Although the prior solvents is excellent solvent for BSF, TSF is insoluble in these solvents. 10 To date, only limited work on TSF dissolution and regeneration has been reported. 11 TSF has been reported soluble in melt Ca(NO 3 ) 2 at 105 C, 12,13 LiSCN at 40-55 C, 14 and ionic liquid. 11 In addition, TSF has been extracted directly from the glands of silkworms. [15][16][17] The as-cast BSF and TSF lms are water soluble due to their random coil structure, post-treatment using organic solvents is required, such as methanol, ethanol. 18,19 To develop highly biocompatible silk broin material, using less toxic solvents is an important direction. 20 Therefore, the main problem of the present silk broin dissolution and water-insoluble lm preparation is complicated and time-consuming, and is difficult to scale up, and the employment of toxic solvent.
Recently, we showed that CaCl 2 /formic acid could serve as new dissolving solvent for BSF, and water as post-treatment solvent directly. 21 In this study, we employed CaCl 2 /formic acid as a cosolvent to dissolve BSF and TSF at room temperature, and to prepare BSF/TSF blend lms using water as rinse solvent. We evaluated and discussed here the morphology, structure, thermal stability, and mechanical properties using a combination of SEM, FTIR, XRD, TGA and DSC, and mechanical testing of lms composed of BSF, TSF and their blends, as well as the lms' biocompatibility using cell culture experiment.

Materials
Bombyx mori silk bers and tussah silk bers were purchased from Zhejiang Province and Liaoning Province, China, respectively. All chemical regents (calcium chloride, formic acid, etc.) were bought from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China), and used without any further purication.

Preparation of BSF and TSF blend lms
The preparation process of BSF/TSF blend lms were illustrated in Fig. 1. The degumming process of silk was conducted according to our previous published procedures. 22 Raw Bombyx mori silk bers were boiled in 0.05 wt% Na 2 CO 3 solution for 30 min and rinsed thoroughly with deionized water to extract the glue-like sericin proteins. In addition, tussah silk bers were boiled in 0.5 wt% Na 2 CO 3 solution for 30 min and rinsed thoroughly with deionized water to extract the glue-like sericin proteins. The above steps was repeated thrice, then the degummed silk bers were dried at room temperature. Aer drying, the degummed BSF and TSF were dissolved in 10% CaCl 2 -formic acid solution at room temperature with ratio of 1/3, 1/1 and 3/1. The mixture solution was cast upon polystyrene dishes (diameter 90 mm) for drying. The as-cast lms were immersed in deionized water to remove salt ions. Aer drying, the desired BSF, BSF/TSF blend and TSF lms were prepared. The schematic process for lm preparation was showed in Fig. 1.

SEM imaging
The surface and cross-section of blend lms were observed with S-4800 SEM (Hitachi, Tokyo, Japan) at 3 kV. For cross-section imaging, the lms were fractured in liquid nitrogen to avoid deformation.

FTIR analysis
The structure of blend lms prepared by CaCl 2 -formic acid dissolution method were analyzed by FTIR on Nicolet 5700 (Thermal Nicolet Company, USA) in transmittance mode. For each measurement, each spectrum was obtained by the performance of 32 scans with the wavenumber ranging from 400 to 4000 cm À1 with a resolution of 4 cm À1 .

X-ray diffraction (XRD)
X-ray diffraction experiments were also measured on X Pert-Pro MPD (PANalytical, Netherlands) in transmittance mode to investigate the crystalline structure of samples. The incident beam wavelength was 0.154 nm. The intensity was nally corrected for changes in the incident beam intensity, sample absorption, and background.

Thermogravimetric measurements
Thermogravimetry/differential thermal analysis (TGA-DTA, PE-SII, America) condition were nitrogen ux at 30 ml min À1 , heating rate at 10 C min À1 and temperature range from 50 to 600 C, and the samples' weight was about 5 mg.

Mechanical properties
Blend lms samples, prepared similar as above, were cut into 50 mm Â 10 mm rectangle with thickness of 100-120 mm. The micrometer was used to measure thickness of these samples. Before using an automatic tensile tester (model 3365 electronic strength tester, Instron, Boston, USA) to investigate the mechanical properties of lms, these lms were kept for 24 hours at atmospheric conditions (20 C and 65 RH). During test process, distance between grips and test speeds were set to 20 mm and 10 mm min À1 , respectively. At the same time, the pre-tension was 0.2 cN. An average of twenty measurements was reported as the mean AE standard deviation for each sample. The formulas for breaking strength and extension at break were as follows: Extension at break (%) ¼ specimen elongation (mm)/original length (mm) Â 100%; The linear elastic modulus was calculated using a least squares tting between 0.1 N load and 2.5% strain past that point. Ultimate tensile strength was the highest stress value attained during the test and the elongation to failure was the last data point before a >10% decrease in load.

Contact angle
The drop of pure distilled water of volume 5 ml was placed on the lm surface using a syringe with a 22-gauge needle. The static contact angle was measured using a Kruss GmbH, Germany. The measurement of each contact angle was made within 10 s aer each drop to ensure that the droplet did not soak into the compact. The contact angles reported were the mean of 10 determinations. Smaller contact angles corresponded to increased wettability.

In vitro studies
Rabbit nucleus pulposus cells (rNPCs) culture was performed to characterize the biocompatibility of materials. Blend lms were formed in 24-well cell culture plate as above procedures. These materials were irradiated with g-ray before use. The rNPCs were provided by Second Affiliated Hospital of Soochow University, the density of rNPCs seeded on lms was 2.0 Â 10 6 / ml. The cell-seeded lms were incubated for 4 hours to allow the cells to attach to the surface of the lms, then a certain volume of Dulbecco's Modied Eagle Medium (H-DMEM) including 10% Fetal Bovine Serum (FBS) was added and continually cultured for 12 hours and 2 days. For the SEM observation, the cell-lms aer 12 hours and 2 days of culture were xed for 6 hours with 2.5% glutaraldehyde, dehydrated in graded series of ethanol, dried and coated with gold, nally were examined by SEM.

Results and discussion
Morphology The surface and cross-sectional morphology of the BSF/TSF blend lms were observed with scanning electron microscopy, as shown in Fig. 2. Overall, the blend lms showed relatively at surface as expected, indicating the good lm forming property of broin-CaCl 2 /formic acid solution. However, close observation inside the lms demonstrated different microstructure. The special nanostructure of BSF lm was studied in our previous study. 23 The BSF lm showed nanobrous structure originated from the nanobril-preservation dissolution behavior of broin in CaCl 2 /formic acid. The preserved nano-bril endowed the BSF lm with exceptional mechanical properties. Different with BSF lm, the TSF lm showed layer structure as previous reports. 24,25 The specic nanostructure of BSF existed, and the layer structure of TSF disappeared in the blend lms. Furthermore, the morphological phase separation was not observed, which had been reported in the BSF/TSF, 26 and in the TSF/carboxymethyl chitosan blend lms prepared using water as a co-solvent. 27 The possible reason was that formic acid was an excellent solvent for silk broin dissolution and regeneration, 28 but water was just a temporary solvent for silk broin in which gel formed easily. 29 The sol-gel transition of TSF was more faster than that of BSF, 30 which easily resulted in the phase separation in the blend lms. Instead, the uniformly distribution of BSF and TSF in formic acid favored the formation of compatible BSF/TSF blend lm. Therefore, CaCl 2 /formic acid was an excellent solvent for the dissolution and regeneration of BSF, TSF, and their hybrid, the resulting blend lms showed smooth surface and specic nanostructure, as well as the compatibility of BSF and TSF.

Structure characteristic
FTIR. FTIR was a widely used technique to analyze the molecular conformation of silk broin because the amide bands of IR spectra were known sensitive to the molecular change of silk broin. 31 The FTIR spectra of BSF/TSF blend lms was shown in Fig. 3A. In addition to the characteristic peaks of b-sheet conformation at 1623 cm À1 (amide I), 1527 cm À1 (amide II) and 1265 cm À1 (amide III), the appearance of peak at 1240 cm À1 (amine III) was attributed to the random coil conformation, 32 suggesting the predominantly b-sheet but with certain amount of random coil structure in water rinsed BSF lm. In the spectra of TSF, the strong absorption bands at 1622, 1520, 1240 and 965 cm À1 supported the mainly b-sheet structure. 33,34 The spectra of BSF/TSF blend lms revealed that there was a weak, or even absent molecular interactions between the two broins because no obvious shi of absorption bands or appearance of new absorption bands was observed. To further conrmed the result, the second-derivative procedures was used to distinguish individual overlapping absorptions. 35 Fig. 3B showed the second-derivative spectra of amide I region of BSF/TSF blend lms. Obviously, the numbers and positions of peaks were not changed, providing strong evidence that the secondary structure of BSF and TSF was not signicantly by the blending of two broin, which was in good agreement with previous reports. 36,37 X-ray diffraction curves. X-ray diffraction was carried out to study the crystalline structure of BSF/TSF blend lms. Fig. 4 showed the X-ray diffraction curves of blend lms with different BSF/TSF volume ratio, the pure BSF lm was characterized by diffraction peak at 2q values of 20.3 and 23.8 , corresponding to the spacing of 4.26Å and 3.73Å, respectively, attributing to bsheet. 27 The pure TSF showed diffraction peaks at 16.5 , 20.5 and 24.5 corresponding to the spacing of 5.37Å, 4.33Å and 3.63Å, attributing to b-sheet. 38 From the IR analysis, we knew that the molecular interaction between BSF and TSF was absent or very weak. Consistent with these results, the diffraction of the blend lms (b-d) exhibited diffraction peaks' characteristic of BSF and TSF simply overlapping in Fig. 4.
However, the diffraction peak at 16.5 characteristic of TSF disappeared when the BSF content was up to 75%, demonstrating that the blending of BSF/TSF had affected the crystal structure of TSF. The low concentration of TSF may reduce the probability of TSF chains' aggregation and b-sheet conformation. In FTIR spectra, the absorption peak at 965 cm À1 attributed to the b-sheet was hard to be observed when the BSF content was 75%.   Thermal analysis TGA. TGA was widely used technique to characterize silk broin 39 and other polymers. 40 Fig. 5 showed the thermogravimetric curves of BSF/TSF blend lms, which demonstrated the weight a continue weight loss (WL) to 600 C. Overall, the weight decrease behavior in TGA curves can be divided into three distinct regions. Region Ⅰ, the lms showed a little and slow weight loss with the temperature increased up to approximately 250 C. The rst weight loss about 10% was mainly due to the water evaporation. Region II, an approximate linear TGA curve was observed in range of 267-383 C, in which the weight loss decreased rapidly with temperature, ascribed to the breakdown of intermolecular interaction, as well as the cleavage of peptide bonds. Region III, the lm demonstrated continue decomposition but slow weight loss behavior.
DTG. As shown in Fig. 5, BSF and TSF lms showed different thermal decomposition peaks at 284.5 C and 358.3 C, respectively. As expected, the blend lms showed two decomposition peaks which was accurately corresponding to the decomposition of BSF and TSF, respectively. It illustrated that the two silk broins didn't form co-crystal structure, as reported previously. 41 The thermal decomposition process of blend lms was composed of two stages: BSF and TSF thermal decomposition. They had two different decomposition peak temperature near 300 C and 360 C, respectively, which were decided by their different crystal structure.
DSC. DSC thermograms of BSF/TSF blend lms were shown in Fig. 5C. The pure BSF lm showed an endothermic peak at 284.5 C without any trace of exothermic transition, and pure TSF lm showed an endothermic peak at 360.6 C, this behavior was due to b-sheet structure. 42,43 In addition, BSF/TSF blend lms appeared two endothermic peaks at 278.6 C and 360.6 C, which belonged to endothermic peaks of BSF and TSF, respectively. The result conrmed that BSF/TSF blend lms were co-existence and did not appear co-crystallization. DSC experiment did not evidence phase transitions, suggesting the crystalline structure of silk broin consistent with above XRD and IR results.
The dissimilarities in the thermal behavior of BSF and TSF could be straightforwardly interpreted by their different crystalline structure. The amino acid sequence in the crystalline region of BSF and TSF was considered to be (GAGAGS)n (G: glycine, A: alanine, S: serine) and -(ala)-, respectively. 44,45 So the TSF molecular chains would form a more compact crystalline structure than BSF, resulting in higher thermal decomposition peaks (Fig. 5). 46 The T d for BSF and TSF were 284 C and 358 C respectively. The T d of BSF was comparable to those of other biodegradable polymers like cellulose (T d ¼ 266 C), poly(hydroxy butyrate-valerate) (T d ¼ 287 C), polypropylene (T d ¼ 295 C) and polystyrene (T d ¼ 270 C), 47 all which were lower than that of TSF. As expected, the blending of TSF improved the T d of BSF, in turn, the adding of BSF decreased the T d of TSF (Table 1). But to our surprise, compared to the pure BSF and TSF, the residual masses of BSF/TSF blend lms at 600 C were all signicantly higher than that of pure BSF and TSF lms as previous report, 37 suggesting that the thermal stability of silk broin could be improved by blending two different silk broin. The underlying mechanism was unclear, possibly due to the interaction between BSF and TSF that was not evidenced here and needed further study.

Mechanical properties
Mechanical properties were of primary essential for determining the performance of materials expected to undergo various types of stresses during use. The representative stress-stain curves were presented in Fig. 6 and summaries of mechanical properties were   listed in Table 2. The mechanical properties of blend lms were obtained in the wet state, which was more important in practical applications of biomaterials than the dry state.
It can be found that wet lm of BSF (a) was stronger and tougher, while pure TSF (e) was weak and fragile. In addition, the Young's modulus showed huge difference between different This journal is © The Royal Society of Chemistry 2018 content of BSF in the blend lms, the Young's modulus of pure BSF lm was 33.63 AE 0.80 MPa, and the pure TSF was only 2.19 AE 0.83 MPa. From the Fig. 6 and Table 2, we also found that the breaking stress, breaking strain and Young's modulus of BSF/ TSF blend lms were continuously increasing along with the proportion of BSF, exhibiting the typical features of a miscible multi-component polymer material. This result also conrmed that TSF lm had a poor mechanical properties. 46

Contact angles
To investigate the hydrophilicity of BSF, TSF and BSF/TSF blend lms, water contact angle measurement was shown in Table 2. Smaller contact angle usually indicated that the material surface was more hydrophilic, enhancing the cell adhesion and proliferation. 48 Table 2 showed that the pure BSF lm had the biggest contact angle, when TSF contents were 25%, 50% and 75% in the blend lms, the contact angle decreased from 76

Biocompatibility
To evaluate biocompatibility to the BSF, TSF and BSF/TSF blend lms at different ratios, rNPCs were cultured on lms. The longterm differentiation and stability were affected by initial attachment. Therefore, the adhesion and proliferation of rNPCs on lms were examined by SEM. As shown in Fig. 7, the rNPCs grew in clusters, from the center to the surrounding area. When the culture period was 12 hours, more rNPCs can be observed on the surface of the pure TSF lms comparing with pure BSF lms. Toward the BSF/TSF blend lms, with the increase of TSF, more rNPCs attached on the blend lms. In addition, the morphology of rNPCs on pure BSF and BSF/TSF at ratio of 3/1 were granular, while rNPCs on the pure TSF, BSF/TSF blend lms at ratio of 1/1 and 3/1 began to grow from center to the surrounding area. When the culture time was 2 days, rNPCs began to spread growth. Although the number of rNPCs observed on all the lms increased from 12 hours to 2 days, with the increase of TSF content, rNPCs showed a better adhesion and proliferation properties. Which due to the TSF had more polar amino acids with positive charge and RGD sequence can support the cell growth that were absent from BSF, so the TSF showed a better compatibility comparing with BSF.

Conclusions
In summary, we provided a novel method for the preparation of TSF/BSF blend lms using CaCl 2 /formic acid as a co-solvent and water as post-treatment solvent. We showed that BSF and TSF was compatible in the BSF/TSF blend lms because CaCl 2 /formic acid was an excellent solvent for BSF and TSF. The BSF/TSF blend lms showed smooth surface and specic nanostructure in the cross-section. Water was a useful post-treatment solvent in regenerating stable silk broin lms through inducing the formation of b-sheet. The direct use of water as a post-treatment solvent for CaCl 2 /formic acid dissolved silk broin suggested that the dissolution behavior was different to the traditional dissolution method, since traditionally derived BSF and TSF was water-soluble. Interestingly, the BSF/TSF blend lms showed higher thermal stability compared to pure BSF and TSF lm, but no obvious molecular interaction was detected through FTIR and XRD analysis. In addition, the good mechanical properties in wet state, controllable hydrophilicity, as well as excellent biocompatibility of the pure BSF, TSF and their hybrid lms were also demonstrated. Thus, this study provided a useful method of preparing silk broin-based material for potential application in biomedical and biotechnological area.

Conflicts of interest
There are no conicts to declare.