Bombyx mori raw silk fibers (Huzhou, China) were degummed three times in 0.05 wt% Na₂CO₃ at 98˚C - 100˚C for 30 min, and then dried in an oven after rinsing. The extracted SF fibers were dissolved in 9.3 M LiBr at 60˚C ± 2˚C for 2 h, Ca(NO₃)₂ at 100˚C ± 2˚C for 3 h and a ternary solvent of CaCl₂:CH₃CH₂OH:H₂O (1:2:8 molar ratio) at 72˚C ± 2˚C for 1 h; the resulting dissolution products were dialyzed (MWCO 9000), respectively, in deionized water for 4 days. The SF solutions were obtained after filtration and then stored at 4˚C. The SF solutions derived from the three ionic liquids were diluted to 2%; then, 2-morpholinoethanesulfonic acid (MES), NHydroxysuccinimide (NHS) and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) (all obtained from Sigma-Aldrich) were added to the SF solutions at 20%, 10% and 20% of the SF weight, respectively. The mixtures were poured into stainless steel vesseles and frozen at −40˚C for 6 h, followed by lyophilization for 48 h using a Virtis Genesis 25-LE Freeze Dryer.

Collagenase IA (from Clostridium histolyticum, EC 3.4.24.3, Sigma-Aldrich,) was dissolved in phosphatebuffered saline (PBS, 0.05 M, pH 7.4) at 1.0 U/ml. The scaffolds were cut into squares (3 × 3 cm, n = 3 per time point) and weighted. The samples were incubated in 5 ml of enzyme solution for 1, 3, 6, 12 and 18 days at 37˚C under slow shaking, and in PBS under otherwise identical conditions as a control. The degradation products and remains were collected for analysis at 1, 3, 6, 12 and 18 days, and the degradation solution was replaced with a fresh enzyme solution every 3 days. The samples at each time point were rinsed with deionized water and then lyophilized. Quantitative changes were expressed as the percentage of weight retained relative to the initial dry weight.

After enzymatic degradation, the scaffolds were lyophilized and examined by scanning electron microscopy (SEM, S-570, Hitachi, Japan). For the molecular conformation measurements, Fourier-transform infrared spectroscopy (FTIR) analysis was performed using a Nicolet 5700 spectrometer (Thermo Scientific, USA). The infrared spectra were analyzed in the amide I region (1595 - 1705 cm⁻¹) using Opus 6.5 software (Bruker, Germany). Deconvolution was performed using a Lorentzian line shape with a half-bandwidth of 25 cm⁻¹ and a noise reduction factor of 0.3. The FTIR spectra were curve-fitted to measure the relative areas of the amide I region components [4,19].

The samples were run on a 5% - 15% polyacrylamide gel in running buffer (0.25 M Tris-HCl, 10% SDS, 0.5% bromophenol blue, 50% glycerol and 5% 2-mercaptoethanol, pH 8.3). The stacking gel contained 5% acrylamide, 0.1% ammonium persulfate and 0.1% SDS in 1.0 M Tris-HCl buffer (pH 6.8), and the gradient separating gel contained 8 to 15% acrylamide, 0.1% ammonium persulfate and 0.1% SDS in 1.5 M Tris-HCl buffer (pH 8.8). Pre-stained protein served as the MW markers (25 - 200 and 43 - 300 kDa) for examining the SF MWD values; the gels were stained with a Easy Stain Coomassie Blue Kit (Invitrogen, Carlsbad, CA). The degradation products were examined to analyze the polypeptides released during degradation. The MW markers were 10 - 200, 10 - 100 and 10 - 50 kDa for the degradation products at 3 d, 6 d and 18 d, respectively.

For the free amino acids determination of the degradation products, the 5 ml of 8% sulfosalicylic acid solution was added to an equal volume of the degradation solution and centrifuged at 15,000 rpm for 15 min to remove the protein. The collected supernatant was diluted with 0.02 N HCl and filtered with Millipore 0.22-μm syringe filters (Milford, USA) [11]. A 20 mL filtrate was analyzed using an amino acid analyzer (L-8800, Hitachi, Japan).

To measure the amino acids contents of the samples remaining after degradation, the samples were hydrolyzed in 6 N HCl at 110˚C for 24 h and then analyzed using an amino acid analyzer.

The SF scaffolds (20 × 20 mm) were implanted into the skin defects in the back of SD rats (180 g - 200 g, SPF grade, male) with 5 rats in each group. Pentobarbital sodium (30 mg/kg body weight) was administered presurgically. Full-thickness wounds (approximately 20 mm × 20 mm) were created on the upper back of each rat, and the scaffolds were implanted as dermal substitutes into the defect sites, followed by covering with thin split-thickness skin grafts. The wounds were then closed with 6-0 silk sutures and covered by Vaseline carbasus and dry carbasus. Finally, the wounds were treated with circular bandages. The Specimens were harvested at 28 days and fixed in 4% formaldehyde in PBS at room temperature before embedding in paraffin. The sections were stained with hematoxylin and eosin (H & E) and were observed under an optical microscope (Olympus BH-2, Japan). The H & E images were used to assess the remaining area of each scaffold. The scaffold degradation ratio was calculated using SigmaScan Pro5.0 software (IBM, USA) according to a previously reported method [20]. All animal experiments were in accordance with the Management Ordinance of Experimental Animal of China ([2001] No. 545) and were approved by the Jiangsu Province in experimental animals management rules ([2008] No. 26).

The silk fibroin consists of 6 heavy chains (~390 kDa), 6 light chains (~26 kDa) and a P25 chain (~25 kDa) [21-23]. As shown in Figure 1, the SDS-PAGE analysis indicated that the SF solutions produced by the different solvents all have sequential bands from about 20 to 300 kDa, indicating that the SF has been partially degraded and hydrolyzed into a mixture of polypeptides with various MWs. The CaCl₂-derived SF solution has a lower MW level (under 72 kDa), that is, at 35 - 72 kDa (Figure 1(b)). The MWDs of the SF samples derived from the LiBr and Ca(NO₃)₂ were similar in the low-MW region, but the band of the LiBr-derived SF was stronger than the band for Ca(NO₃)₂ over 300 kDa (Figures 1(a) and (c)). This result sugguests that LiBr is mild with respect to the degradation of SF, while the CaCl₂ ionic liquid exhibits the strongest degradation behavior. In this study, the MW levels derived from the three solvents were, in descending order, LiBr > Ca(NO₃)₂ > CaCl₂.

As shown in Figures 2(a-0), (b-0) and (c-0)), the SF scaffolds derived from the three solvents exhibit porous morphologies, where the pore shapes are irregular polygons and fusiform. The pore wall thicknesses of the three scaffolds are similar. The large pores have a long diameters in the range of approximately 100 - 600 μm and short diameters of approximately 50 - 200 μm, while the long diameters and short diameters were both approxi-

mately 10 - 100 μm for the small pores. The average pore diameters of the three scaffolds were similar.

The scaffolds began to collapse after 6 days of incubation, however, partial scaffold integrity was maintained. For example, a shrunken pore structure could still be observed in the LiBr-derived scaffold (Figures 2(a-6), (b-6) and (c-6)). With further degradation, all of the scaffolds were unable to maintain integrity and collapsed. The residual CaCl₂-derived scaffold was powder-like, but the remains of the LiBr-derived scaffold and Ca(NO₃)₂-derived scaffold were still sheet-like (Figures 2(a-18), (b-18) and (c-18)). This result indicates that the LiBr-derived scaffold and Ca(NO₃)₂-derived scaffold were able to maintain the structural integrity for longer times, while the CaCl₂-derived scaffold was more susceptible to enzymatic degradation.

Figure 3 showed that the mass loss of the SF scaffolds increased with the enzymatic degradation time. After 3 days, the residual weights of the scaffolds derived from LiBr, Ca(NO₃)₂ and CaCl₂ were 90.33%, 88.29% and 85.68%, respectively. Following further degradation, the difference of residual weights among the three scaffolds significantly increased, yielding residual weights of 81.92%, 79.38% and 66.25% at 6 days and 46.33%, 44.12% and 26.25% at 12 days. At 18 days, the residual weight of the CaCl₂-derived scaffolds was only 6.34%, showing the most rapid degradation rate in comparison to the LiBr and Ca(NO₃)₂-derived scaffolds. The mass loss data were fitted with linear trendlines, giving linearly dependent coefficients of a, b and c (Figure 3) valued at 0.9827, 0.9872 and 0.9874, respectively. These results indicate that the mass loss trend was approxi-