- Email: [email protected]
In vitro degradation of silk fibroin
ARTICLE IN PRESS
Biomaterials 26 (2005) 3385–3393 www.elsevier.com/locate/biomaterials
In vitro degradation of silk fibroin Rebecca L. Horan, Kathryn Antle, Adam L. Collette, Yongzhong Wang, Jia Huang, Jodie E. Moreau, Vladimir Volloch, David L. Kaplan, Gregory H. Altman Biomedical Engineering Department, Tufts University, 4 Colby Street, Room 153, Medford, MA 02155, USA Received 8 July 2004; accepted 8 September 2004
Abstract A significant need exists for long-term degradable biomaterials which can slowly and predictably transfer a load-bearing burden to developing biological tissue. In this study Bombyx mori silk fibroin yarns were incubated in 1 mg/ml Protease XIV at 37 1C to create an in vitro model system of proteolytic degradation. Samples were harvested at designated time points up to 12 weeks and (1) prepared for scanning electron microscopy (SEM), (2) lyophilized and weighed, (3) mechanical properties determined using a servohydraulic Instron 8511, (4) dissolved and run on a SDS–PAGE gel, and (5) characterized with Fourier transform infrared spectroscopy. Control samples were incubated in phosphate-buffered saline. Fibroin was shown to proteolytically degrade with predictable rates of change in fibroin diameter, failure strength, cycles to failure, and mass. SEM indicated increasing fragmentation of individual fibroin filaments from protease-digested samples with time of exposure to the enzyme; particulate debris was present within 7 days of incubation. Gel electrophoresis indicated a decreasing amount of the silk 25 kDa light chain and a shift in the molecular weight of the heavy chain with increasing incubation time in protease. Results support that silk is a mechanically robust biomaterial with predictable long-term degradation characteristics. r 2004 Elsevier Ltd. All rights reserved. Keywords: Degradation; Silk; Scaffold; Ligament; Mechanical properties
1. Introduction When damaged or diseased tissues such as bone, tendons, or ligaments are incapable of self-repair, a substitute biomaterial is often required to aid the healing process. Slowly degrading biomaterials which can maintain tissue integrity following implantation while continually transferring the load-bearing burden to the developing biological functional tissue are desired [1]. Such a phenomena, the gradual transfer (at predictable rates) of the load-bearing burden to the developing and/ or remodeling tissue, should support the restoration and maintenance of tissue function over the life of the patient. Many natural and synthetic biomaterials have been explored for load-bearing applications in vivo. Rapidly Corresponding author. Tel.: +1 617 627 4604; +1 617 627 3231. E-mail address: [email protected] (G.H. Altman).
fax:
0142-9612/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2004.09.020
degrading polymers such as collagen and polyester synthetics have been shown to transfer load-bearing responsibility to developing tissue, but prior to sufficient ingrowth and remodeling, resulting in mechanical failure of the repair [2–4]. Non-degradable materials such as polytetrafluorethylene, polyester, carbon fiber and polypropylene failed to support host tissue ingrowth and remodeling as a result of stress shielding in high load-bearing applications [5–9]. Thus, integrity was limited to that of the synthetic material’s fatigue properties, eventually resulting in failure [10]. Current focus has been placed on the development of a novel silk fibroin protein matrix that is biocompatible, mechanically robust and can be designed to desired mechanical specifications including ultimate tensile strength, yield point and stiffness [11,12]. Bombyx mori silkworm silk, the most common form of commercially available silk, contains at least two fibroin proteins, the light and heavy chains, 25 and 325 kDa, respectively. Combined, silk fibroin contains highly
ARTICLE IN PRESS 3386
R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
ordered crystalline domains and less-ordered intermediate domains that result in silk’s unique mechanical properties (i.e., strength and flexibility). Structural analysis of silk reveals numerous but small b-sheet crystals contained within the heavy chain. As well, it is also know that silk fibroin lacks typical cell binding domains but those domains may be engineered following extraction to improve the biocompatibility of the material in vitro [13]. In vivo, such a strategy is likely to alter rates of degradation as a function of (macrophage) cell recognition and binding, yielding the conversion of what is typically referred to as a non-degradable biomaterial to one that degrades over the long term. Recently, Li et al. [14] demonstrated degradation of regenerated silk films, consisting primarily of amorphous silk. Incubation in Protease XIV (cat #P8811, Sigma UK) led to the most significant degradation with results indicating an overall increase in crystallinity of the silk film over time (i.e., the amount of silk II decreased and the amount of silk I crystalline structure increased). Of interest, collagenase IA was shown to degrade silk II as well, but to a lesser extent. Historically, a-chymotrypsin has been shown to degrade silk [15,16]; however, achymotrypsin did not have an appreciable effect on the degradation of the silk films [14]. The rate of proteolytic degradation observed for naturally formed fibroin will likely differ from that observed in films, as dissolving the fibroin and reforming films alters the structure and composition of the silk polymer. Protease XXI has been shown to degrade silk films and fibers, altering surface roughness and strength over 17 days; however, the rate of degradation was shown to be significantly different between films and fibers [16]. We demonstrate here the in vitro proteolytic degradation of the silk fibroin and its predictable mechanical behavior including tensile and fatigue properties. While silk, as a FDA approved biomaterial, is defined by the USP as non-degradable because it retains greater than 50% of its tensile integrity 60 days post-implantation in vivo, as a protein, silk will be proteolytically degraded and resorbed in vivo over a longer time period (e.g., typically within a year) [12]. The development of an improved understanding of the effect of proteolytic degradation on mechanical integrity and mass loss, will provide greater insight into the appropriate silk scaffold design (e.g., void volume and exposed surface area) for various tissue engineering applications.
groups of 12 parallel fibers (20/22 denier) (RudolphDesco Co., NJ) were cabled, at 3.5 turns/cm, into a 36 fiber yarn. The yarn was extracted in a batch system with 0.3% detergent (Procter & Gamble, Cincinnati, OH) and 0.02 M sodium carbonate (Sigma, MO) solution at 90 1C for 60 min, rinsed in dIH2O and dried. 2.1.2. Enzyme Enzyme type and concentration used for degradation characterization were selected through a brief screening experiment. All enzymes were purchased from Sigma (St. Louis, MO). Yarns were incubated at 37 1C in (1) 0.25 mg/ml a-chymotrypsin, (2) 0.5 mg/ml a-chymotrypsin, (3) 0.5 mg/ml Protease XIV, and (4) 1.0 mg/ml Protease XIV. Enzyme solutions were changed daily for 1 week. Percent change in mass and ultimate tensile strength (UTS) were determined (described below) for each sample group. A specific enzyme and concentration was selected based on preliminary results for further use in the in vitro degradation model system. 2.1.3. In vitro degradation model system Silk matrices (25 cm in length) (N=4 per group and time point) were incubated at 37 1C in 3 ml solution of 1.0 mg/ml Protease XIV (from Streptomyces griseus, cat #P5147, Sigma, MO) in phosphate-buffered saline (PBS) or in PBS as a negative control. Each solution contained an approximately equivalent (75 mg) mass of silk. Solutions were changed and collected daily. At designated time points, groups of samples were rinsed in dIH2O and (1) prepared for scanning electron microscopy (SEM), (2) lyophilized and weighed, (3) mechanical properties determined using a servohydraulic Instron 8511 (Canton, MA), (4) dissolved and run on a gel, and (5) characterized with Fourier transform infrared spectroscopy (FTIR). 2.2. Analytics 2.2.1. Scanning electron microscopy (SEM) Samples were dehydrated at their respective time points and segments (1 cm in length) were affixed via carbon tape to the SEM sample holders. Yarns were sputter coated with Au for 45 s and examined using a JEOL JSM 840A Scanning electron microscope. Fibroin diameter was determined from individual SEM images (N=9 images per group and time point); a 3 3 grid was placed over each image and a diameter measurement was taken from each section with vernier calipers.
2. Materials and methods 2.1. Materials 2.1.1. Silk matrix B. mori silkworm silk yarns were generated and processed as previously described [11]. Briefly, three
2.2.2. Particle collection and image analysis Solutions were collected at 24 h intervals over 84 days and 1 ml of each sample’s solution was stored at 20 1C. One sample from each weekly interval was defrosted, vortexed and 10 ml of the solution was pipetted onto a hemocytometer. Samples were compared grossly on an
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
inverted light microscope (Axiovert S100, Zeiss, Germany) at 10 magnification to identify time points with substantial differences in size and number of particles. Sample images at days 14, 42, and 70 were captured using an inverted microscope (Axiovert S100, Zeiss, Germany), and digital image analysis: 3CCD color video camera (DXC-390, Sony, Tokyo, Japan), a frame grabber card (CG-7 RGB, Scion, Frederick, MD) and Scion-Image software version 1.9.1. Images were generated for each individual grid of a hemacytometer and particle major and minor diameters were measured from high-resolution prints using vernier calipers. The total number of particles at each time point as well as particle size (i.e., average diameter) distribution as a function of degradation time was determined. On day 42, yarns were removed from solution and dried over night in a laminar flow hood to prevent matting of the samples during blotting. Digital images of the samples were captured with a Canon Pro90IS (Lake Success, NY) with 10 optical zoom. 2.2.3. Mass loss Twice weekly, the incubated yarns were washed at 90 1C for 30 min in 0.3% detergent and 0.02 M sodium carbonate to remove protease absorbed on the yarn surface. Samples were agitated twice during washing (at 10 min intervals) to maintain detergent suspension and remove any aggregate on the surface of the silk. Each yarn sample was then rinsed via agitation in individual dishes of Milli-Q water for 60 s and blotted dry on a clean lint-free towel. Each group was blotted for an equal amount of time, resulting in a yarn that would leave no visible watermark on the towel. Yarns were dried completely in a vacuum oven at 60 1C for 3 h. Immediately following removal from the oven, the samples were massed and returned to new solution. Percent mass loss over time was determined. 2.2.4. Mechanical analysis Both single pull to failure and cyclic fatigue testing were performed. Samples were rinsed in Milli-Q water for 60 s, blotted and stored in a 6-well plate. All samples were tested within 1 h of harvest for pull to failure testing and within 9 days for fatigue testing (due to the amount of time required to cycle a sample 125,000 times at 1 Hz). Fatigue samples were kept dry at 23 1C, following harvest until test preparation. Samples were wet with PBS throughout preparation and testing. Customized clamps were used to anchor the samples with a gage length of 3 cm. Pull to failure testing was performed at 1800 mm/min (100%/s) on a servohydraulic Instron 8511 (Instron, Canton, MA). Material properties including UTS, stiffness and yield point were determined using Instron Series IX software (Canton, MA). Wavemaker32 v6.6 (Instron, Canton, MA) was utilized to generate a fourth quadrant H-sine function
3387
for cyclic testing. Cycles to failure at 1 Hz were determined with an 8 N load applied to yarns submersed in PBS at room temperature. Fatigue analysis was discontinued at a pre-determined 125,000 cycle limit for each sample.
2.2.5. SDS–PAGE Samples were dissolved in 9 M LiBr (1%w/v silk) at 50 1C for 24 h. The dissolved samples were dialyzed with milli-Q water (Millipore, Billerica, MA) in 3.5 kDa molecular weight cutoff Slide-A-Lyzer Dialysis Cassettes (Pierce, Rockford, IL). Milli-Q water (0.5 L per cassette) was changed after 1, 4, 12, 24 and 36 h. After 48 h of dialysis, the samples were removed from the cassettes and stored at 4 1C. Samples were reduced and run on a NuPage 4–12% Bis–Tris gel in MOPS buffer (Invitrogen, Carlsbad, CA). Reduced laminin was run as a molecular weight marker (400 and 200 kDa) on all gels. Gels were stained with the Easy Stain Commassie Blue Kit (Invitrogen, Carlsbad, CA). Digital images of the gel were captured with a Fluor-S MultiImager (Bio-Rad Laboratories, Hercules, CA). Relative band intensities were determined with MultiAnalyst v1.1 Build 34 software (Bio-Rad Laboratories, Hercules, CA).
2.2.6. Fourier transform infrared spectroscopy (FTIR) Incubated yarns were removed from solution on days 1, 21, 42, and 70, dried overnight in a laminar flow hood and stored dry at 23 1C until processing. The crystal structure was analyzed via FTIR on a TENSOR 37 with a MIRacle ATR attachment and ZnSe crystal (Bruker, Billerica, MA). Background measurements were taken twice with an empty cell and subtracted from the sample readings. Silk samples, approximately 1 cm in length, were orientated with the long axis of the yarn parallel to the infrared beam and intensity at wavenumbers from 1000 to 3000 cm 1 recorded.
2.2.7. Statistical analyses Analysis of variance (ANOVA) was performed on data sets when appropriate using a significance of p values o0.05. One-way ANOVA was used to evaluate differences over time for fatigue analysis of degraded samples. Post hoc comparison of means was accomplished with the Student–Newman–Keuls test to determine significance as a function of time. Two-way ANOVA was used to compare features between groups and/or over time; Tukey’s test for multiple comparisons was used to determine where significant differences existed. SPSS 6.0 software was used as needed and regression analysis was performed using Microsoft Excel.
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3388
3. Results In vitro degradation was observed in all silk samples incubated in protease. Increasing fragmentation of individual fibroin filaments from proteasedigested samples was evident by eye and SEM as a function of time of exposure to the enzyme, a feature absent in the PBS controls up to 10 weeks (Fig. 1(A, a–f)). Fibroin filament diameter decreased exponentially (R2=0.99) with time in protease to 66% of initial diameter at 10 weeks (Fig. 1B); no decrease in filament diameter was observed in the diameter of PBS controls. Particulate debris was present within 7 days of incubation with protease; no debris was observed in the solution of the PBS incubated silk samples even after 10 weeks. The total number of particles (Fig. 2) collected in the protease solution was greatest at 10 weeks, then 3 weeks and 1 day, respectively. The right tail of the size distribution histogram increased in length in solutions harvested at later time points.
Greater than 50% loss in mass after 42 days of in vitro degradation (Fig. 3A) was observed in the protease incubated matrices. The mass loss was consistent with a second-order polynomic trendline (R2=0.99). No significant differences in mass loss were observed in the PBS incubated samples over time. Digital photographs (Fig. 3B) demonstrate a gross change in the sample appearances reflecting mass over 42 days of proteolytic degradation. Gross observation revealed more fraying (i.e. loss of initial organized geometry) in the protease incubated yarns than the PBS incubated controls. UTS decreased exponentially (R2=0.985) with time as a function of proteolytic degradation (Fig. 4A); no significant change (p40.05) in UTS was observed for PBS incubated controls. UTS decreased by 50% of initial strength after 21 days of in vitro proteolytic incubation. Cycles to failure decreased logarithmically (R2=0.99) with increasing exposure to protease (Fig. 4B). Prior to degradation, initial mechanical integrity provided for greater than 125,000 cycles prior to failure; however, after 28 days of degradation, only 4.571.7
(a)
(b)
(c)
(d)
(e)
(f )
(A)
(B) Fig. 1. (A) SEM images of silk fibers incubated in PBS for (a) 1 day, (b) 21 days and (c) 42 days, or in 1.0 mg/ml protease solution for (d) 1 day, (e) 21 days and (f) 42 days. (B) Silk fibroin diameter as a function of time in PBS and in 1.0 mg/ml protease.
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3389
60 50
Day 14
# of Particles
Day 42
40
Day 77
30 20 10 0 4.3
5.7
7.1
8.6
10.0
11.4
12.9
14.3
21.4
28.6
Diameter (um) Fig. 2. Silk particle size (i.e., average diameter (mm)) distribution in protease solutions on days 14, 42, and 70.
90 80 70
Mass (mg)
60
y = -0.0298x 2 + 0.2391x + 75.244 R2 = 0.9947
50 40 30 20
Protease 1mg/mL
10
PBS control
0 0
7
14
21
28
35
42
Degradation time (days)
(A)
(a)
SDS–PAGE gel electrophoresis (Fig. 5) indicated the 25 kDa silk light chain was degraded with exposure to protease. After 10 weeks, the amount of 25 kDa band present decreased to 64% of the day 1 levels. The heavy chain smear was shifted to a lower molecular weight range with increasing protease exposure time indicating the degradation of both the light and heavy chains. The heavy chain smear and 25 kDa band remain consistent in size and intensity for all PBS incubated controls. FTIR spectra (Fig. 6) demonstrated predominately a b-sheet structure within the samples as anticipated [17–19], with only slight changes in structure evident over the 70 days of enzymatic exposure; no change was observed in control samples. The characteristic b-sheet peak at 1625 cm 1 (amide I) did not change in terms of position with time of digestion, while there was a slight decrease of the b-sheet characteristic peak at 1516 cm 1 (amide II) with time. This later change shows a general decrease in intensity and slight shift with time, with a small shoulder appearing at 1528 and at 1535 cm 1 (random coil silk). The slight shoulder initially present at 1650 cm 1 (amide I), indicative of silk I, disappears with time of enzymatic digest.
(b)
(B) Fig. 3. (A) Mass of silk matrix as a function of time in PBS or 1.0 mg/ ml protease. (B) Digital photograph of samples, initially the same mass (78 mg), after 42 days in (a) PBS control (77 mg) and (b) 1.0 mg/ml protease (30 mg).
cycles were completed before failure. Fatigue integrity was maintained (i.e., all samples exceeded 125,000 cycles) for all samples incubated in PBS.
4. Discussion Protease (Protease XIV) was selected for use in the degradation model system on the basis of preliminary data generated during the screening experiment and prior experiments described in the literature. Specifically, after 1 week of incubation in a-chymotrypsin the UTS and mass of the silk matrices remained unchanged (data not shown); however, matrices incubated in protease significantly decreased in mass and UTS following 1 week incubation, demonstrating the poten-
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3390 70
1
3
4
5
6
7
8
9 kDa
60
191
50
UTS (N)
2
97
-0.0359x
40
y = 54.754e 2 R = 0.9853
30
64 51
20 PBS control
10
39
Protease 1mg/ml
0 0
7
14
21
28
28
Degradation time (days)
(A) 1000000
19
Cycles to failure
100000
14
-0.3623x
y = 162494e 2 R = 0.9883
10000 1000
Fig. 5. 4–12% Bis–Tris gel with samples including: 191 kDa ladder (lanes 1 and 9), reduced laminin (lane 2), silk incubated in 1.0 mg/ml protease for 1 day (lane 3), 21 days (lane 4), 70 days (lane 5), or in PBS for 1 day (lane 6), 21 days (lane 7), and 70 days (lane 8).
100 10 1 0
(B)
5
10
15
20
25
30
Degradation time (days)
Fig. 4. (A) UTS of silk matrices incubated in PBS or in 1.0 mg/ml protease as a function of time. (B) Fatigue life of a silk matrix at 8 N (25% UTS of a single 36 fiber yarn) after in vitro degradation in 1.0 mg/ml protease.
tial of proteolytic degradation of silk fibroin. Li et al. [14] observed a number of free amino acids in an enzymatic solution due to the degradation of silk films incubated in the Protease XIV solution. Li et al.’s [14] results demonstrated the potential of protease to cleave silk non-discriminately (i.e., at multiple locations along the length of the protein) and therefore supported the use of protease in this study. However, unlike Li et al. [14], protease solutions were changed on a daily basis due to a significant loss in activity level when incubated at 37 1C for periods longer than 24 h (Sigma technical support). The frequent exchange of protease proven to be needed here is likely the result of the higher ordered structure of native fibroin as compared to reconstituted silk structures. Increased fragmentation of individual fibroin filaments over time in the protease-digested samples observed by SEM (Fig. 1A) corresponded with the observed decrease in mechanical integrity. As a greater number of fibrils were fractured, there was a decrease in the mechanically functional cross-sectional area and
therefore a decrease in load-bearing capacity. The change in diameter as well as visible particulate surface debris suggested that the fibroin was degraded through a surface erosion process (i.e. individual layers of silk fragmented from the exposed surface area). The exponential diameter change (Fig. 1B) suggested that the fibroin diameter will continue to change slowly with time, leading eventually to complete degradation. Larger diameter particles observed in the protease solution (Fig. 2) were likely aggregates of particles formed on the silk fibroin surface prior to complete fracture. The debris particles may have remained physically attached to each other or alternatively, they may have been bound together by protease absorbed to the silk surface. It is important to note that the fibroin was washed, according to the initial extraction protocol, every 3 days to remove any protease absorbed to the silk surface and therefore it is likely loose surface debris was washed away during processing. Any loose debris was removed with the wash solution and not included in the particulate solution assayed for size distribution. Particles cleaved from the silk and within the enzyme solution remained there for a maximum of 1 day following cleavage from the fibroin (as solutions were changed daily) and therefore would likely decrease in size over time if continually exposed to the enzyme. The increase with time in the total particle numbers collected from the solution per day correlates with the accelerated rate of mass loss and increased fibril damage observed at
ARTICLE IN PRESS
Absorbance
R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3391
(D) (C) (B) (A)
1800
1700
1600
1500
1400
1300
1200
Wavenumber (cm-1) Fig. 6. FTIR spectra of silk fibroin incubated in 1.0 mg/ml protease for (A) 1 day, (B) 21 days, (C) 42 days or (D) 70 days.
later time-points. Likely the increase in exposed surface area on fractured fibrils provided additional sites for enzymatic attack on silk not mechanically attached to the rest of the fibroin. Mass loss data (Fig. 3A) was fit to a second-order polynomic trendline; loss occurred gradually initially and as larger particles were observed in solution, the mass loss increased accordingly. The observed difference in degradation rate between mass loss and fibroin diameter is of interest, as the two would be expected to correlate. Digital images (Fig. 3B) verified the gross difference in total size and mass between fibroin incubated in protease and in PBS. A significant difference existed in the amount of silk remaining, as well as in the surface properties of the matrices. Similar to the changes observed on a microscopic level at earlier time points (Fig. 1A), the enzyme incubated samples were visibly altered at the macro-level with loose fibrils and multiple fracture sites. The observed rate of strength loss (Fig. 4A) in the protease incubated samples in this study was rapid in comparison to the expected rates in vivo, i.e., silk will lose 50% of its tensile strength approximately 100 days post-implantation [12]. It is important to note that these yarns were not subjected to fluid flow, which has been shown to significantly decrease the rate of degradation in other in vitro model systems [21]. The addition of flow to the model system would subject the silk surface to shear forces and increase the adhesion strength required between the enzyme and the silk as well as residence time in order to facilitate proteolytic degradation. Protease would likely be pulled from the surface prior to completely cleaving the particles, leading to increased time required to degrade the fibroin. It is expected that if the rate of strength loss was
decreased as a function of flow and shear force, the number of cycles to failure during loading would increase for the given time points. SDS–PAGE (Fig. 5) indicated both the 325 kDA heavy chain and the 25 kDa light chain were degraded with increased exposure time to protease. The intensity of the 25 kDa band was decreased with time, suggesting that the 25 kDa chain was cleaved. The heavy chain appeared as a smear on day 1, likely due to minor degradation of the heavy chain during cocoon reeling and processing [22] and due to the difficulty in maintaining the 325 kDa chain in solution during processing of the gel. The peptides within this smear decreased in size with increased exposure to protease. These results indicate that both the intermediary regions of the fibroin as well as regions within the heavy chain (possibly those between the crystalline regions) were susceptible to proteolytic degradation. FTIR data (Fig. 6) suggest that the structure is essentially unchanged, with perhaps some slight loss of crystalline (b-sheet) material either directly due to hydrolysis or due to loss of crystalline material into the solution as the non-crystalline domains are digested. The relative significance of the appearance of a shoulder at 1535 cm 1 in the later time points cannot be quantified. The disappearance of the slight shoulder present at 1650 cm 1 indicated hydrolysis of the noncrystalline domains. This study provides the first comprehensive evidence of native silk fibroin proteolytic degradation, an event often questioned following implantation of the biomaterial in vivo. This work provides a system with which to further study the degradation of native fibroin fiber thereby improving our mechanistic understanding of the phenomena both in vitro and more importantly, in vivo.
ARTICLE IN PRESS 3392
R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
Correlating the trend to rates of silk degradation in vivo reported in the literature [12], the silk yarn is expected to lose 100% of its tensile strength approximately 5 months post-implantation. However, degradation rates can likely be altered through scaffold design (i.e. yarn organization) and surface modification. The rate may change as the foreign body response to the material is altered, at the onset of angiogenesis, or as a function of the applied loading regime during healing [20]. The in vitro model system described here has been used to generate an approximation of the degradation rate of silk fibroin and to demonstrate its predictability. However, the complex in vivo environment specific to tissue type and location is required to determine the exact rate of degradation. A better understanding of the in vivo remodeling cascade is needed to develop improved in vitro–in vivo correlations in order to balance the rates of degradation with the rates of ingrowth and remodeling.
5. Conclusions This is the first paper to demonstrate a substantial loss in mechanical integrity can be correlated to mass loss in native silk fibroin. Fibroin was shown to proteolytically degrade with predictable rates of change in fibroin diameter, failure strength, cycles to failure, and mass. SEM indicated increasing fragmentation of individual fibroin filaments from protease-digested samples with time of exposure to the enzyme; particulate debris was present within 7 days of incubation. Gel electrophoresis indicated a decreasing amount of the silk 25 kDa light chain and a shift in the molecular weight of the heavy chain with increasing incubation time in protease. These results support that silk is a mechanically robust biomaterial with predictable long-term degradation characteristics. The ability to engineer a yarn to meet specific mechanical requirements [11], coupled with knowledge of the predictable degradation rates will allow for the design of scaffolds to meet numerous loadbearing tissue requirements. In particular, ligament scaffolds and bone substitutes can be designed with an appropriate factor of safety to match loading burdens at specific time points in vivo. An improved understanding of the in vivo environment and remodeling cascade’s role in the degradation of silk fibroin will provide the next logical step in modeling an appropriate long-term degradable scaffold for various tissue engineering applications.
Acknowledgements We thank the NIH (R01 AR46563-01) and Tissue Regeneration, Inc. (SBIR R44 AR049139-03) for sup-
port of this research. We also thank Dr. Herman Chernoff, Harvard University, for his support in statistical analysis.
References [1] Weitzel PP, Richmond JC, Altman GH, Calabro T, Kaplan DL. Future direction of the treatment of ACL ruptures. In: Gladstone JN, Andrews JR, editors. Orthopedic clinics of North America. Philadelphia: W.B. Saunders Co.; 2002. p. 653–61. [2] Dunn MG, Avasarala PN, Zawadsky JP. Optimization of extruded collagen fibers for ACL reconstruction. J Biomed Mater Res 1993;27:1545–52. [3] Durselen L, Dauner M, Hierlemann H, Planck H, Claes LE, Ignatius A. Resorbable polymer fibers for ligament augmentation. J Biomed Mater Res 2001;58:666–72. [4] Lu L, Peter SJ, Lyman MD, Lai HL, Leite SM, Tamada JA, Uyama S, Vacanti JP, Langer R, Mikos AG. In vitro and in vivo degradation of porous poly(DL-lactic-co-glycolic acid) foams. Biomaterials 2000;21:1837–45. [5] McCarthy DM, Tolin BS, Schwenderman L. In: Jackson DW, editor. The anterior cruciate ligament: current and future concepts. New York: Raven Press Ltd; 1993. p. 343–56. [6] Markolf KL, Pattee GA, Strum GM, Gallick GS, Sherman OH, Nuys V, Dorey FJ. Instrumented measurements of laxity in patients who have a Gore-Tex anterior cruciate ligament substitute. J Bone Joint Surg Am 1989;71:887–93. [7] Richmond JC, Manseau CJ, Patz R, McConville O. Anterior cruciate ligament reconstruction using a Dacron prosthesis. A long term study. Am J Sports Med 1992;20:24–8. [8] Bonnarens FO, Drez D. Biomechanics of artificial ligaments and associated problems. In: Jackson DW, editor. The anterior cruciate deficient knee, new concepts in ligament repair. St. Louis: CV Mosby; 1987. p. 239–53. [9] Kumar K, Maffulli N. The ligament augmentation device: an historical perspective. Arthroscopy 1999;15:422–32. [10] Frank BC, Jackson DW. The science of reconstruction of the anterior cruciate ligament. J Bone Joint Surg Am 1997;79: 1556–76. [11] Altman GH, Horan RL, Lu HH, Moreau J, Martin I, Richmond JC, Kaplan DL. Silk matrix for tissue engineered anterior cruciate ligaments. Biomaterials 2002;23:4131–41. [12] Altman GH, Diaz F, Jakuba C, Calabro T, Horan RL, Chen J, Lu H, Richmond J, Kaplan DL. Silk-based biomaterials. Biomaterials 2003;24:401–16. [13] Chen J, Altman GH, Karageorgiou V, Horan R, Collette A, Volloch V, Calabro T, Kaplan DL. Human bone marrow stromal cell and ligament fibroblast responses on RGD-modified silk fibers. J Biomed Mater Res 2003;67A:559–70. [14] Li M, Ogiso M, Minoura N. Enzymatic degradation behavior of porous silk fibroin sheets. Biomaterials 2003;24:357–65. [15] Tsukada M. Effect of a-chymotrypsin on the structure of silk fibroin. J Seric Sci Jpn 1986;55:120–6. [16] Arai T, Freddi G, Innocenti R, Tsukada M. Biodegradation of Bombyx mori silk fibroin fibers and films. J Appl Polym Sci 2004;91:2383–90. [17] Freddi G, Pessina G, Tsukada M. Swelling and dissolution of silk fibroin (Bombyx mori) in N-methyl morpholine N-oxide. Int J Bio Macromol 1999;24:251–63. [18] Ishida M, Asakura T, Yokoi M, Saito H. Solvent- and mechanical-treatment-induced conformational transition of silk fibroins studies by high-resolution solid-state carbon-13 NMR spectroscopy. Macromolecules 1990;23:88–94.
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393 [19] Valluzzi R, Szela S, Avtges P, Kirschner D, Kaplan DL. Methionine redox controlled crystallization of biosynthetic silk spidroin. J Phys Chem 1999;103:11382–92. [20] Muneta T, Lewis JL, Stewart NJ. Load affects remodeling of transplanted, autogenous bone–patellar tendon–bone segments in a rabbit model. J Ortho Res 1994;12:138–43.
3393
[21] Agrawal CM, McKinney JS, Lanctot D, Athanasiou KA. Effects of fluid flow on the in vitro degradation kinetics of biodegradable scaffolds for tissue engineering. Biomaterials 2000;21:2443–52. [22] Tsubouchi K, Yamada H, Takasu Y, Nakao H. Process for producing functional silk fibroin and utilization of the same. European Patent Office 2002; EP 1 241 178 A1.
Biomaterials 26 (2005) 3385–3393 www.elsevier.com/locate/biomaterials
In vitro degradation of silk fibroin Rebecca L. Horan, Kathryn Antle, Adam L. Collette, Yongzhong Wang, Jia Huang, Jodie E. Moreau, Vladimir Volloch, David L. Kaplan, Gregory H. Altman Biomedical Engineering Department, Tufts University, 4 Colby Street, Room 153, Medford, MA 02155, USA Received 8 July 2004; accepted 8 September 2004
Abstract A significant need exists for long-term degradable biomaterials which can slowly and predictably transfer a load-bearing burden to developing biological tissue. In this study Bombyx mori silk fibroin yarns were incubated in 1 mg/ml Protease XIV at 37 1C to create an in vitro model system of proteolytic degradation. Samples were harvested at designated time points up to 12 weeks and (1) prepared for scanning electron microscopy (SEM), (2) lyophilized and weighed, (3) mechanical properties determined using a servohydraulic Instron 8511, (4) dissolved and run on a SDS–PAGE gel, and (5) characterized with Fourier transform infrared spectroscopy. Control samples were incubated in phosphate-buffered saline. Fibroin was shown to proteolytically degrade with predictable rates of change in fibroin diameter, failure strength, cycles to failure, and mass. SEM indicated increasing fragmentation of individual fibroin filaments from protease-digested samples with time of exposure to the enzyme; particulate debris was present within 7 days of incubation. Gel electrophoresis indicated a decreasing amount of the silk 25 kDa light chain and a shift in the molecular weight of the heavy chain with increasing incubation time in protease. Results support that silk is a mechanically robust biomaterial with predictable long-term degradation characteristics. r 2004 Elsevier Ltd. All rights reserved. Keywords: Degradation; Silk; Scaffold; Ligament; Mechanical properties
1. Introduction When damaged or diseased tissues such as bone, tendons, or ligaments are incapable of self-repair, a substitute biomaterial is often required to aid the healing process. Slowly degrading biomaterials which can maintain tissue integrity following implantation while continually transferring the load-bearing burden to the developing biological functional tissue are desired [1]. Such a phenomena, the gradual transfer (at predictable rates) of the load-bearing burden to the developing and/ or remodeling tissue, should support the restoration and maintenance of tissue function over the life of the patient. Many natural and synthetic biomaterials have been explored for load-bearing applications in vivo. Rapidly Corresponding author. Tel.: +1 617 627 4604; +1 617 627 3231. E-mail address: [email protected] (G.H. Altman).
fax:
0142-9612/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2004.09.020
degrading polymers such as collagen and polyester synthetics have been shown to transfer load-bearing responsibility to developing tissue, but prior to sufficient ingrowth and remodeling, resulting in mechanical failure of the repair [2–4]. Non-degradable materials such as polytetrafluorethylene, polyester, carbon fiber and polypropylene failed to support host tissue ingrowth and remodeling as a result of stress shielding in high load-bearing applications [5–9]. Thus, integrity was limited to that of the synthetic material’s fatigue properties, eventually resulting in failure [10]. Current focus has been placed on the development of a novel silk fibroin protein matrix that is biocompatible, mechanically robust and can be designed to desired mechanical specifications including ultimate tensile strength, yield point and stiffness [11,12]. Bombyx mori silkworm silk, the most common form of commercially available silk, contains at least two fibroin proteins, the light and heavy chains, 25 and 325 kDa, respectively. Combined, silk fibroin contains highly
ARTICLE IN PRESS 3386
R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
ordered crystalline domains and less-ordered intermediate domains that result in silk’s unique mechanical properties (i.e., strength and flexibility). Structural analysis of silk reveals numerous but small b-sheet crystals contained within the heavy chain. As well, it is also know that silk fibroin lacks typical cell binding domains but those domains may be engineered following extraction to improve the biocompatibility of the material in vitro [13]. In vivo, such a strategy is likely to alter rates of degradation as a function of (macrophage) cell recognition and binding, yielding the conversion of what is typically referred to as a non-degradable biomaterial to one that degrades over the long term. Recently, Li et al. [14] demonstrated degradation of regenerated silk films, consisting primarily of amorphous silk. Incubation in Protease XIV (cat #P8811, Sigma UK) led to the most significant degradation with results indicating an overall increase in crystallinity of the silk film over time (i.e., the amount of silk II decreased and the amount of silk I crystalline structure increased). Of interest, collagenase IA was shown to degrade silk II as well, but to a lesser extent. Historically, a-chymotrypsin has been shown to degrade silk [15,16]; however, achymotrypsin did not have an appreciable effect on the degradation of the silk films [14]. The rate of proteolytic degradation observed for naturally formed fibroin will likely differ from that observed in films, as dissolving the fibroin and reforming films alters the structure and composition of the silk polymer. Protease XXI has been shown to degrade silk films and fibers, altering surface roughness and strength over 17 days; however, the rate of degradation was shown to be significantly different between films and fibers [16]. We demonstrate here the in vitro proteolytic degradation of the silk fibroin and its predictable mechanical behavior including tensile and fatigue properties. While silk, as a FDA approved biomaterial, is defined by the USP as non-degradable because it retains greater than 50% of its tensile integrity 60 days post-implantation in vivo, as a protein, silk will be proteolytically degraded and resorbed in vivo over a longer time period (e.g., typically within a year) [12]. The development of an improved understanding of the effect of proteolytic degradation on mechanical integrity and mass loss, will provide greater insight into the appropriate silk scaffold design (e.g., void volume and exposed surface area) for various tissue engineering applications.
groups of 12 parallel fibers (20/22 denier) (RudolphDesco Co., NJ) were cabled, at 3.5 turns/cm, into a 36 fiber yarn. The yarn was extracted in a batch system with 0.3% detergent (Procter & Gamble, Cincinnati, OH) and 0.02 M sodium carbonate (Sigma, MO) solution at 90 1C for 60 min, rinsed in dIH2O and dried. 2.1.2. Enzyme Enzyme type and concentration used for degradation characterization were selected through a brief screening experiment. All enzymes were purchased from Sigma (St. Louis, MO). Yarns were incubated at 37 1C in (1) 0.25 mg/ml a-chymotrypsin, (2) 0.5 mg/ml a-chymotrypsin, (3) 0.5 mg/ml Protease XIV, and (4) 1.0 mg/ml Protease XIV. Enzyme solutions were changed daily for 1 week. Percent change in mass and ultimate tensile strength (UTS) were determined (described below) for each sample group. A specific enzyme and concentration was selected based on preliminary results for further use in the in vitro degradation model system. 2.1.3. In vitro degradation model system Silk matrices (25 cm in length) (N=4 per group and time point) were incubated at 37 1C in 3 ml solution of 1.0 mg/ml Protease XIV (from Streptomyces griseus, cat #P5147, Sigma, MO) in phosphate-buffered saline (PBS) or in PBS as a negative control. Each solution contained an approximately equivalent (75 mg) mass of silk. Solutions were changed and collected daily. At designated time points, groups of samples were rinsed in dIH2O and (1) prepared for scanning electron microscopy (SEM), (2) lyophilized and weighed, (3) mechanical properties determined using a servohydraulic Instron 8511 (Canton, MA), (4) dissolved and run on a gel, and (5) characterized with Fourier transform infrared spectroscopy (FTIR). 2.2. Analytics 2.2.1. Scanning electron microscopy (SEM) Samples were dehydrated at their respective time points and segments (1 cm in length) were affixed via carbon tape to the SEM sample holders. Yarns were sputter coated with Au for 45 s and examined using a JEOL JSM 840A Scanning electron microscope. Fibroin diameter was determined from individual SEM images (N=9 images per group and time point); a 3 3 grid was placed over each image and a diameter measurement was taken from each section with vernier calipers.
2. Materials and methods 2.1. Materials 2.1.1. Silk matrix B. mori silkworm silk yarns were generated and processed as previously described [11]. Briefly, three
2.2.2. Particle collection and image analysis Solutions were collected at 24 h intervals over 84 days and 1 ml of each sample’s solution was stored at 20 1C. One sample from each weekly interval was defrosted, vortexed and 10 ml of the solution was pipetted onto a hemocytometer. Samples were compared grossly on an
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
inverted light microscope (Axiovert S100, Zeiss, Germany) at 10 magnification to identify time points with substantial differences in size and number of particles. Sample images at days 14, 42, and 70 were captured using an inverted microscope (Axiovert S100, Zeiss, Germany), and digital image analysis: 3CCD color video camera (DXC-390, Sony, Tokyo, Japan), a frame grabber card (CG-7 RGB, Scion, Frederick, MD) and Scion-Image software version 1.9.1. Images were generated for each individual grid of a hemacytometer and particle major and minor diameters were measured from high-resolution prints using vernier calipers. The total number of particles at each time point as well as particle size (i.e., average diameter) distribution as a function of degradation time was determined. On day 42, yarns were removed from solution and dried over night in a laminar flow hood to prevent matting of the samples during blotting. Digital images of the samples were captured with a Canon Pro90IS (Lake Success, NY) with 10 optical zoom. 2.2.3. Mass loss Twice weekly, the incubated yarns were washed at 90 1C for 30 min in 0.3% detergent and 0.02 M sodium carbonate to remove protease absorbed on the yarn surface. Samples were agitated twice during washing (at 10 min intervals) to maintain detergent suspension and remove any aggregate on the surface of the silk. Each yarn sample was then rinsed via agitation in individual dishes of Milli-Q water for 60 s and blotted dry on a clean lint-free towel. Each group was blotted for an equal amount of time, resulting in a yarn that would leave no visible watermark on the towel. Yarns were dried completely in a vacuum oven at 60 1C for 3 h. Immediately following removal from the oven, the samples were massed and returned to new solution. Percent mass loss over time was determined. 2.2.4. Mechanical analysis Both single pull to failure and cyclic fatigue testing were performed. Samples were rinsed in Milli-Q water for 60 s, blotted and stored in a 6-well plate. All samples were tested within 1 h of harvest for pull to failure testing and within 9 days for fatigue testing (due to the amount of time required to cycle a sample 125,000 times at 1 Hz). Fatigue samples were kept dry at 23 1C, following harvest until test preparation. Samples were wet with PBS throughout preparation and testing. Customized clamps were used to anchor the samples with a gage length of 3 cm. Pull to failure testing was performed at 1800 mm/min (100%/s) on a servohydraulic Instron 8511 (Instron, Canton, MA). Material properties including UTS, stiffness and yield point were determined using Instron Series IX software (Canton, MA). Wavemaker32 v6.6 (Instron, Canton, MA) was utilized to generate a fourth quadrant H-sine function
3387
for cyclic testing. Cycles to failure at 1 Hz were determined with an 8 N load applied to yarns submersed in PBS at room temperature. Fatigue analysis was discontinued at a pre-determined 125,000 cycle limit for each sample.
2.2.5. SDS–PAGE Samples were dissolved in 9 M LiBr (1%w/v silk) at 50 1C for 24 h. The dissolved samples were dialyzed with milli-Q water (Millipore, Billerica, MA) in 3.5 kDa molecular weight cutoff Slide-A-Lyzer Dialysis Cassettes (Pierce, Rockford, IL). Milli-Q water (0.5 L per cassette) was changed after 1, 4, 12, 24 and 36 h. After 48 h of dialysis, the samples were removed from the cassettes and stored at 4 1C. Samples were reduced and run on a NuPage 4–12% Bis–Tris gel in MOPS buffer (Invitrogen, Carlsbad, CA). Reduced laminin was run as a molecular weight marker (400 and 200 kDa) on all gels. Gels were stained with the Easy Stain Commassie Blue Kit (Invitrogen, Carlsbad, CA). Digital images of the gel were captured with a Fluor-S MultiImager (Bio-Rad Laboratories, Hercules, CA). Relative band intensities were determined with MultiAnalyst v1.1 Build 34 software (Bio-Rad Laboratories, Hercules, CA).
2.2.6. Fourier transform infrared spectroscopy (FTIR) Incubated yarns were removed from solution on days 1, 21, 42, and 70, dried overnight in a laminar flow hood and stored dry at 23 1C until processing. The crystal structure was analyzed via FTIR on a TENSOR 37 with a MIRacle ATR attachment and ZnSe crystal (Bruker, Billerica, MA). Background measurements were taken twice with an empty cell and subtracted from the sample readings. Silk samples, approximately 1 cm in length, were orientated with the long axis of the yarn parallel to the infrared beam and intensity at wavenumbers from 1000 to 3000 cm 1 recorded.
2.2.7. Statistical analyses Analysis of variance (ANOVA) was performed on data sets when appropriate using a significance of p values o0.05. One-way ANOVA was used to evaluate differences over time for fatigue analysis of degraded samples. Post hoc comparison of means was accomplished with the Student–Newman–Keuls test to determine significance as a function of time. Two-way ANOVA was used to compare features between groups and/or over time; Tukey’s test for multiple comparisons was used to determine where significant differences existed. SPSS 6.0 software was used as needed and regression analysis was performed using Microsoft Excel.
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3388
3. Results In vitro degradation was observed in all silk samples incubated in protease. Increasing fragmentation of individual fibroin filaments from proteasedigested samples was evident by eye and SEM as a function of time of exposure to the enzyme, a feature absent in the PBS controls up to 10 weeks (Fig. 1(A, a–f)). Fibroin filament diameter decreased exponentially (R2=0.99) with time in protease to 66% of initial diameter at 10 weeks (Fig. 1B); no decrease in filament diameter was observed in the diameter of PBS controls. Particulate debris was present within 7 days of incubation with protease; no debris was observed in the solution of the PBS incubated silk samples even after 10 weeks. The total number of particles (Fig. 2) collected in the protease solution was greatest at 10 weeks, then 3 weeks and 1 day, respectively. The right tail of the size distribution histogram increased in length in solutions harvested at later time points.
Greater than 50% loss in mass after 42 days of in vitro degradation (Fig. 3A) was observed in the protease incubated matrices. The mass loss was consistent with a second-order polynomic trendline (R2=0.99). No significant differences in mass loss were observed in the PBS incubated samples over time. Digital photographs (Fig. 3B) demonstrate a gross change in the sample appearances reflecting mass over 42 days of proteolytic degradation. Gross observation revealed more fraying (i.e. loss of initial organized geometry) in the protease incubated yarns than the PBS incubated controls. UTS decreased exponentially (R2=0.985) with time as a function of proteolytic degradation (Fig. 4A); no significant change (p40.05) in UTS was observed for PBS incubated controls. UTS decreased by 50% of initial strength after 21 days of in vitro proteolytic incubation. Cycles to failure decreased logarithmically (R2=0.99) with increasing exposure to protease (Fig. 4B). Prior to degradation, initial mechanical integrity provided for greater than 125,000 cycles prior to failure; however, after 28 days of degradation, only 4.571.7
(a)
(b)
(c)
(d)
(e)
(f )
(A)
(B) Fig. 1. (A) SEM images of silk fibers incubated in PBS for (a) 1 day, (b) 21 days and (c) 42 days, or in 1.0 mg/ml protease solution for (d) 1 day, (e) 21 days and (f) 42 days. (B) Silk fibroin diameter as a function of time in PBS and in 1.0 mg/ml protease.
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3389
60 50
Day 14
# of Particles
Day 42
40
Day 77
30 20 10 0 4.3
5.7
7.1
8.6
10.0
11.4
12.9
14.3
21.4
28.6
Diameter (um) Fig. 2. Silk particle size (i.e., average diameter (mm)) distribution in protease solutions on days 14, 42, and 70.
90 80 70
Mass (mg)
60
y = -0.0298x 2 + 0.2391x + 75.244 R2 = 0.9947
50 40 30 20
Protease 1mg/mL
10
PBS control
0 0
7
14
21
28
35
42
Degradation time (days)
(A)
(a)
SDS–PAGE gel electrophoresis (Fig. 5) indicated the 25 kDa silk light chain was degraded with exposure to protease. After 10 weeks, the amount of 25 kDa band present decreased to 64% of the day 1 levels. The heavy chain smear was shifted to a lower molecular weight range with increasing protease exposure time indicating the degradation of both the light and heavy chains. The heavy chain smear and 25 kDa band remain consistent in size and intensity for all PBS incubated controls. FTIR spectra (Fig. 6) demonstrated predominately a b-sheet structure within the samples as anticipated [17–19], with only slight changes in structure evident over the 70 days of enzymatic exposure; no change was observed in control samples. The characteristic b-sheet peak at 1625 cm 1 (amide I) did not change in terms of position with time of digestion, while there was a slight decrease of the b-sheet characteristic peak at 1516 cm 1 (amide II) with time. This later change shows a general decrease in intensity and slight shift with time, with a small shoulder appearing at 1528 and at 1535 cm 1 (random coil silk). The slight shoulder initially present at 1650 cm 1 (amide I), indicative of silk I, disappears with time of enzymatic digest.
(b)
(B) Fig. 3. (A) Mass of silk matrix as a function of time in PBS or 1.0 mg/ ml protease. (B) Digital photograph of samples, initially the same mass (78 mg), after 42 days in (a) PBS control (77 mg) and (b) 1.0 mg/ml protease (30 mg).
cycles were completed before failure. Fatigue integrity was maintained (i.e., all samples exceeded 125,000 cycles) for all samples incubated in PBS.
4. Discussion Protease (Protease XIV) was selected for use in the degradation model system on the basis of preliminary data generated during the screening experiment and prior experiments described in the literature. Specifically, after 1 week of incubation in a-chymotrypsin the UTS and mass of the silk matrices remained unchanged (data not shown); however, matrices incubated in protease significantly decreased in mass and UTS following 1 week incubation, demonstrating the poten-
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3390 70
1
3
4
5
6
7
8
9 kDa
60
191
50
UTS (N)
2
97
-0.0359x
40
y = 54.754e 2 R = 0.9853
30
64 51
20 PBS control
10
39
Protease 1mg/ml
0 0
7
14
21
28
28
Degradation time (days)
(A) 1000000
19
Cycles to failure
100000
14
-0.3623x
y = 162494e 2 R = 0.9883
10000 1000
Fig. 5. 4–12% Bis–Tris gel with samples including: 191 kDa ladder (lanes 1 and 9), reduced laminin (lane 2), silk incubated in 1.0 mg/ml protease for 1 day (lane 3), 21 days (lane 4), 70 days (lane 5), or in PBS for 1 day (lane 6), 21 days (lane 7), and 70 days (lane 8).
100 10 1 0
(B)
5
10
15
20
25
30
Degradation time (days)
Fig. 4. (A) UTS of silk matrices incubated in PBS or in 1.0 mg/ml protease as a function of time. (B) Fatigue life of a silk matrix at 8 N (25% UTS of a single 36 fiber yarn) after in vitro degradation in 1.0 mg/ml protease.
tial of proteolytic degradation of silk fibroin. Li et al. [14] observed a number of free amino acids in an enzymatic solution due to the degradation of silk films incubated in the Protease XIV solution. Li et al.’s [14] results demonstrated the potential of protease to cleave silk non-discriminately (i.e., at multiple locations along the length of the protein) and therefore supported the use of protease in this study. However, unlike Li et al. [14], protease solutions were changed on a daily basis due to a significant loss in activity level when incubated at 37 1C for periods longer than 24 h (Sigma technical support). The frequent exchange of protease proven to be needed here is likely the result of the higher ordered structure of native fibroin as compared to reconstituted silk structures. Increased fragmentation of individual fibroin filaments over time in the protease-digested samples observed by SEM (Fig. 1A) corresponded with the observed decrease in mechanical integrity. As a greater number of fibrils were fractured, there was a decrease in the mechanically functional cross-sectional area and
therefore a decrease in load-bearing capacity. The change in diameter as well as visible particulate surface debris suggested that the fibroin was degraded through a surface erosion process (i.e. individual layers of silk fragmented from the exposed surface area). The exponential diameter change (Fig. 1B) suggested that the fibroin diameter will continue to change slowly with time, leading eventually to complete degradation. Larger diameter particles observed in the protease solution (Fig. 2) were likely aggregates of particles formed on the silk fibroin surface prior to complete fracture. The debris particles may have remained physically attached to each other or alternatively, they may have been bound together by protease absorbed to the silk surface. It is important to note that the fibroin was washed, according to the initial extraction protocol, every 3 days to remove any protease absorbed to the silk surface and therefore it is likely loose surface debris was washed away during processing. Any loose debris was removed with the wash solution and not included in the particulate solution assayed for size distribution. Particles cleaved from the silk and within the enzyme solution remained there for a maximum of 1 day following cleavage from the fibroin (as solutions were changed daily) and therefore would likely decrease in size over time if continually exposed to the enzyme. The increase with time in the total particle numbers collected from the solution per day correlates with the accelerated rate of mass loss and increased fibril damage observed at
ARTICLE IN PRESS
Absorbance
R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
3391
(D) (C) (B) (A)
1800
1700
1600
1500
1400
1300
1200
Wavenumber (cm-1) Fig. 6. FTIR spectra of silk fibroin incubated in 1.0 mg/ml protease for (A) 1 day, (B) 21 days, (C) 42 days or (D) 70 days.
later time-points. Likely the increase in exposed surface area on fractured fibrils provided additional sites for enzymatic attack on silk not mechanically attached to the rest of the fibroin. Mass loss data (Fig. 3A) was fit to a second-order polynomic trendline; loss occurred gradually initially and as larger particles were observed in solution, the mass loss increased accordingly. The observed difference in degradation rate between mass loss and fibroin diameter is of interest, as the two would be expected to correlate. Digital images (Fig. 3B) verified the gross difference in total size and mass between fibroin incubated in protease and in PBS. A significant difference existed in the amount of silk remaining, as well as in the surface properties of the matrices. Similar to the changes observed on a microscopic level at earlier time points (Fig. 1A), the enzyme incubated samples were visibly altered at the macro-level with loose fibrils and multiple fracture sites. The observed rate of strength loss (Fig. 4A) in the protease incubated samples in this study was rapid in comparison to the expected rates in vivo, i.e., silk will lose 50% of its tensile strength approximately 100 days post-implantation [12]. It is important to note that these yarns were not subjected to fluid flow, which has been shown to significantly decrease the rate of degradation in other in vitro model systems [21]. The addition of flow to the model system would subject the silk surface to shear forces and increase the adhesion strength required between the enzyme and the silk as well as residence time in order to facilitate proteolytic degradation. Protease would likely be pulled from the surface prior to completely cleaving the particles, leading to increased time required to degrade the fibroin. It is expected that if the rate of strength loss was
decreased as a function of flow and shear force, the number of cycles to failure during loading would increase for the given time points. SDS–PAGE (Fig. 5) indicated both the 325 kDA heavy chain and the 25 kDa light chain were degraded with increased exposure time to protease. The intensity of the 25 kDa band was decreased with time, suggesting that the 25 kDa chain was cleaved. The heavy chain appeared as a smear on day 1, likely due to minor degradation of the heavy chain during cocoon reeling and processing [22] and due to the difficulty in maintaining the 325 kDa chain in solution during processing of the gel. The peptides within this smear decreased in size with increased exposure to protease. These results indicate that both the intermediary regions of the fibroin as well as regions within the heavy chain (possibly those between the crystalline regions) were susceptible to proteolytic degradation. FTIR data (Fig. 6) suggest that the structure is essentially unchanged, with perhaps some slight loss of crystalline (b-sheet) material either directly due to hydrolysis or due to loss of crystalline material into the solution as the non-crystalline domains are digested. The relative significance of the appearance of a shoulder at 1535 cm 1 in the later time points cannot be quantified. The disappearance of the slight shoulder present at 1650 cm 1 indicated hydrolysis of the noncrystalline domains. This study provides the first comprehensive evidence of native silk fibroin proteolytic degradation, an event often questioned following implantation of the biomaterial in vivo. This work provides a system with which to further study the degradation of native fibroin fiber thereby improving our mechanistic understanding of the phenomena both in vitro and more importantly, in vivo.
ARTICLE IN PRESS 3392
R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393
Correlating the trend to rates of silk degradation in vivo reported in the literature [12], the silk yarn is expected to lose 100% of its tensile strength approximately 5 months post-implantation. However, degradation rates can likely be altered through scaffold design (i.e. yarn organization) and surface modification. The rate may change as the foreign body response to the material is altered, at the onset of angiogenesis, or as a function of the applied loading regime during healing [20]. The in vitro model system described here has been used to generate an approximation of the degradation rate of silk fibroin and to demonstrate its predictability. However, the complex in vivo environment specific to tissue type and location is required to determine the exact rate of degradation. A better understanding of the in vivo remodeling cascade is needed to develop improved in vitro–in vivo correlations in order to balance the rates of degradation with the rates of ingrowth and remodeling.
5. Conclusions This is the first paper to demonstrate a substantial loss in mechanical integrity can be correlated to mass loss in native silk fibroin. Fibroin was shown to proteolytically degrade with predictable rates of change in fibroin diameter, failure strength, cycles to failure, and mass. SEM indicated increasing fragmentation of individual fibroin filaments from protease-digested samples with time of exposure to the enzyme; particulate debris was present within 7 days of incubation. Gel electrophoresis indicated a decreasing amount of the silk 25 kDa light chain and a shift in the molecular weight of the heavy chain with increasing incubation time in protease. These results support that silk is a mechanically robust biomaterial with predictable long-term degradation characteristics. The ability to engineer a yarn to meet specific mechanical requirements [11], coupled with knowledge of the predictable degradation rates will allow for the design of scaffolds to meet numerous loadbearing tissue requirements. In particular, ligament scaffolds and bone substitutes can be designed with an appropriate factor of safety to match loading burdens at specific time points in vivo. An improved understanding of the in vivo environment and remodeling cascade’s role in the degradation of silk fibroin will provide the next logical step in modeling an appropriate long-term degradable scaffold for various tissue engineering applications.
Acknowledgements We thank the NIH (R01 AR46563-01) and Tissue Regeneration, Inc. (SBIR R44 AR049139-03) for sup-
port of this research. We also thank Dr. Herman Chernoff, Harvard University, for his support in statistical analysis.
References [1] Weitzel PP, Richmond JC, Altman GH, Calabro T, Kaplan DL. Future direction of the treatment of ACL ruptures. In: Gladstone JN, Andrews JR, editors. Orthopedic clinics of North America. Philadelphia: W.B. Saunders Co.; 2002. p. 653–61. [2] Dunn MG, Avasarala PN, Zawadsky JP. Optimization of extruded collagen fibers for ACL reconstruction. J Biomed Mater Res 1993;27:1545–52. [3] Durselen L, Dauner M, Hierlemann H, Planck H, Claes LE, Ignatius A. Resorbable polymer fibers for ligament augmentation. J Biomed Mater Res 2001;58:666–72. [4] Lu L, Peter SJ, Lyman MD, Lai HL, Leite SM, Tamada JA, Uyama S, Vacanti JP, Langer R, Mikos AG. In vitro and in vivo degradation of porous poly(DL-lactic-co-glycolic acid) foams. Biomaterials 2000;21:1837–45. [5] McCarthy DM, Tolin BS, Schwenderman L. In: Jackson DW, editor. The anterior cruciate ligament: current and future concepts. New York: Raven Press Ltd; 1993. p. 343–56. [6] Markolf KL, Pattee GA, Strum GM, Gallick GS, Sherman OH, Nuys V, Dorey FJ. Instrumented measurements of laxity in patients who have a Gore-Tex anterior cruciate ligament substitute. J Bone Joint Surg Am 1989;71:887–93. [7] Richmond JC, Manseau CJ, Patz R, McConville O. Anterior cruciate ligament reconstruction using a Dacron prosthesis. A long term study. Am J Sports Med 1992;20:24–8. [8] Bonnarens FO, Drez D. Biomechanics of artificial ligaments and associated problems. In: Jackson DW, editor. The anterior cruciate deficient knee, new concepts in ligament repair. St. Louis: CV Mosby; 1987. p. 239–53. [9] Kumar K, Maffulli N. The ligament augmentation device: an historical perspective. Arthroscopy 1999;15:422–32. [10] Frank BC, Jackson DW. The science of reconstruction of the anterior cruciate ligament. J Bone Joint Surg Am 1997;79: 1556–76. [11] Altman GH, Horan RL, Lu HH, Moreau J, Martin I, Richmond JC, Kaplan DL. Silk matrix for tissue engineered anterior cruciate ligaments. Biomaterials 2002;23:4131–41. [12] Altman GH, Diaz F, Jakuba C, Calabro T, Horan RL, Chen J, Lu H, Richmond J, Kaplan DL. Silk-based biomaterials. Biomaterials 2003;24:401–16. [13] Chen J, Altman GH, Karageorgiou V, Horan R, Collette A, Volloch V, Calabro T, Kaplan DL. Human bone marrow stromal cell and ligament fibroblast responses on RGD-modified silk fibers. J Biomed Mater Res 2003;67A:559–70. [14] Li M, Ogiso M, Minoura N. Enzymatic degradation behavior of porous silk fibroin sheets. Biomaterials 2003;24:357–65. [15] Tsukada M. Effect of a-chymotrypsin on the structure of silk fibroin. J Seric Sci Jpn 1986;55:120–6. [16] Arai T, Freddi G, Innocenti R, Tsukada M. Biodegradation of Bombyx mori silk fibroin fibers and films. J Appl Polym Sci 2004;91:2383–90. [17] Freddi G, Pessina G, Tsukada M. Swelling and dissolution of silk fibroin (Bombyx mori) in N-methyl morpholine N-oxide. Int J Bio Macromol 1999;24:251–63. [18] Ishida M, Asakura T, Yokoi M, Saito H. Solvent- and mechanical-treatment-induced conformational transition of silk fibroins studies by high-resolution solid-state carbon-13 NMR spectroscopy. Macromolecules 1990;23:88–94.
ARTICLE IN PRESS R.L. Horan et al. / Biomaterials 26 (2005) 3385–3393 [19] Valluzzi R, Szela S, Avtges P, Kirschner D, Kaplan DL. Methionine redox controlled crystallization of biosynthetic silk spidroin. J Phys Chem 1999;103:11382–92. [20] Muneta T, Lewis JL, Stewart NJ. Load affects remodeling of transplanted, autogenous bone–patellar tendon–bone segments in a rabbit model. J Ortho Res 1994;12:138–43.
3393
[21] Agrawal CM, McKinney JS, Lanctot D, Athanasiou KA. Effects of fluid flow on the in vitro degradation kinetics of biodegradable scaffolds for tissue engineering. Biomaterials 2000;21:2443–52. [22] Tsubouchi K, Yamada H, Takasu Y, Nakao H. Process for producing functional silk fibroin and utilization of the same. European Patent Office 2002; EP 1 241 178 A1.