The ability to produce biomaterials suitable for diverse applications is paramount to the success of regenerative medicine. An optimal biomaterial should be tailored for a specific purpose and control several reaction parameters, including, in part, the degree of inflammation, the level of vascularization and the degradation rate. In addition, the material should elicit an acceptable late peri-implant tissue reaction. In recent years, fibroin silk proteins from the domestic silk worm (Bombyx mori) have shown promise as biomaterials for applications in regenerative medicine. Silk from this source is a filamentous material about 1 mile long, consisting primarily of two proteins, an internal core of fibroin and an external gum of sericin. Silk fibroin can be transformed by simple methods into films, sponges, non-woven nets and solid or injectable gels (Altman et al., 2003; Vepari et al., 2007). This material also has history of use as suture threads.
Previous in vivo studies showed a low inflammatory response to both native and sericin-deprived silk fibres, concluding that, while fibroin is substantially inert, sericin was responsible for activating some proinflammatory molecules (Panilaitis et al., 2003). These results confirmed initial findings in which cases of an immune response towards whole silk fibres were attributed to their external coating of sericin (Zaoming et al., 1996). Correspondingly, sericin-deprived fibroin films exhibited a lower inflammatory response in comparison to other established reference biomaterials, such as polystyrene and poly-2-hydroxyethylmethacrylate (Santin et al., 1999). Depending on the fabrication method and on varying treatments, fibroin can exhibit different and peculiar interactions with cells (Servoli et al., 2005). Cast fibroin films seeded with mesenchymal stem cells (MSCs) were implanted intramuscularly, showing enhanced MSC proliferation and a lower inflammatory tissue reaction than similarly used collagen and poly-lactic acid films (Meinel et al., 2005). Injectible fibroin gels seeded with osteoblasts were found to also promote proliferation and resulted in the formation of new bone in a rat femur critical-size defect (Fini et al., 2005). This protein was also found to induce proliferation in adipose tissue and chondrocytes (Mauney et al., 2007; Wang et al., 2006).
Non-woven silk fibroin (SF) nets have been studied in our group in vitro (Fuchs et al., 2006, 2009; Unger et al., 2004, 2007). These studies revealed the potential of SF to function as a suitable scaffolding biomaterial for complex tissue-engineering applications with co-cultures of mesenchymal and endothelial cells. Human outgrowth endothelial cells from peripheral blood (OECs) and human microvasculature endothelial cells (HDMECs) formed, in co-culture with human osteoblasts (HOs), capillary-like structures within the SF scaffold. Further in vivo studies analysing SF scaffolds treated with FA alone for 30 min in the subcutaneous implantation model in mice focused on the inflammatory proteins involved in the early tissue response and indicated a mild inflammatory response to this biomaterial (Dal Pra et al., 2005).
Several studies have examined the tissue reaction to fibroin-based materials qualitatively. However, until now, no quantitative study exists on the in vivo vascularization and degradation of SF scaffolds. The in vivo vascularization of SF is of great importance for understanding the possible applications of this material, and could also be used to design in vitro experiments with human endothelial and mesenchymal cells on SF to predict expected outcomes after implantation. Another factor essential for the application of SF is to determine how differences in silk fibroin preparation, in particular the duration of formic acid treatment, could potentially influence the tissue reaction to these materials. To determine the effects of SF prepared with slight changes in the FA processing time, we examined the vascularization and degradation of SF scaffolds after implantation. This in vivo study was aimed at characterizing the tissue reaction over 180 days for these two SF scaffolds. We focused on quantifying the extent of vascularization of the implantation bed and monitored scaffold degradation. Ultimately, the aim is to use FA treatment time as a parameter to control in vivo vascularization and degradation, in order to produce tailored SF scaffolds for specific tissue-engineering applications.
2. Materials and methods
2.1. Production of SF-based 3D non-woven scaffolds
B. mori cocoons were degummed by boiling twice, for 1 h each, in an aqueous solution of Na2CO3 (1.1 and 0.4 g/l, respectively), in order to remove sericine, a protein gum surrounding the inner fibroin filaments. Degummed silk was then washed several times in distilled water and dried at room temperature. SF-based three-dimensional (3D) non-woven substrates were then prepared according to the described method (Armato et al., 2002), with a variation in the formic acid treatment duration. Briefly, degummed silk fibres were soaked in 98% formic acid at room temperature. The fibre suspension was shaken in the FA solution for either 30 or 60 min to produce two differently treated scaffolds. The acid solution was evaporated under atmospheric conditions, and the resulting non-woven material was repeatedly washed with double-distilled water and finally vacuum-dried. The resulting 3D non-woven micronets showed a porous ultrastructure, with visible differences depending on acid treatment (Figure 1).
2.2. Experimental design
This study protocol was approved by the Committee on the Use of Live Animals in Teaching and Research (Landesuntersuchungsamt) of the State of Rhineland-Palatinate, Germany (AZ: 23177-07/G06-1-016). 128 female 5 week-old Wistar rats were obtained from Charles River Laboratories, Germany. The animals were maintained for 1 week before use at the Laboratory Animal Unit, Institute of Pathology, Johannes Gutenberg University of Mainz, Germany, as follows. They were fed with regular rat pellets (Laboratory Rodent Chow, Altromin, Germany); water was available ad libitum and there was an artificial light–dark cycle of 12 h each. The rats were randomly distributed into two groups of 64 animals each. From each group, eight animals were used for each of the following ‘early response’ time points: 3, 10 and 15 days, as well as for the following ‘late response’ time points: 30, 60, 90, 120 and 180 days. The first group received the 3D non-woven micro-net silk fibroin (SF) produced by the 30 min FA treatment, while the second group received the same SF material that had been subjected to 60 min FA treatment. After intraperitoneal anaesthesia with 10 ml ketamine (50 mg/ml) containing 1.6 ml 2% xylazine, sterile SF materials of 5 × 5 mm size were implanted under sterile conditions in preformed subcutaneous pockets of the subscapular region according to an established method (Dal Pra et al., 2005). Briefly, the shaved skin of the rostral portion of the interscapular region was incised and the biomaterial was placed into the subcutaneous pocket under the thin skin muscle of the animal. The wound was subsequently stitched with 6.0 Prolene (Ethicon, NJ, USA). Each surgical procedure was carried out under strictly aseptic conditions and care was taken to close the operation situs immediately after biomaterial implantation and incisions were accurately stitched. All animals survived the indicated time periods without any complications.
2.3. Tissue preparation for histology and immunohistochemistry
The animals were sacrificed by an overdose of ketamine and xylazin at the corresponding time points, as indicated above. Immediately after death, the biomaterials were explanted together with the surrounding peri-implantation tissue and fixed in 4% buffered formalin for 24 h prior to histological and immunohistochemical analysis. The fixed explanted material was cut into three segments of identical dimensions that included the margins and centre of the biomaterial and were dehydrated in a series of alcohols, transferred to xylene and embedded in paraffin. From each segment, three consecutive 3–4 µm thick paraffin slices were then deparaffinized and rehydrated. One slice from each of the three sections was stained with Mayer's haematoxylin and eosin (H&E) to establish the most representative histological slide (i.e. margin, centre or other margin) for the animal in order to best analyse the biomaterial and surrounding tissue. This slide was selected for qualitative histopathological evaluation and the consecutive two sections were prepared for immunohistochemical detection of murine vessels. Endogenous peroxidase activity was quenched by treatment with 4% H2O2 in methanol. After blocking with horse serum, the sections were stained with anti-rabbit polyclonal von Willebrand Factor (vWF) antibody (GeneTex Inc., USA; diluted 1:400). Visualization was performed using the DAKO REAL™ EnVision™ detection system (DAKO, Glostrup, Denmark). As a negative control for vWF staining, the third section was treated similarly but the primary antibody was omitted. All control stains were negative. Finally, the samples were counterstained with haemalaun and viewed by light microscopy.
2.3.1. Morphological evaluation of the inflammatory response
Histopathological evaluation was conducted using a Nikon ECLIPSE 80i microscope (Nikon, Tokyo, Japan) by two independent examiners (S.G., C.O.) experienced in histomorphological analysis, who were blinded to the experimental group. The slide was divided into two parts, ‘inside the implant’ and ‘outside the implant’, which were evaluated independently. The SF scaffolds were assessed qualitatively for the following characteristic features: fibrotic capsule around the biomaterials; fibrosis; haemorrhage; necrosis; vascularization; neutrophils; lymphocytes; plasma cells; macrophages; giant cells; and degradation. Microphotographs were taken using a Nikon DS-Fi1/digital camera and a digital sight control unit (Nikon, Tokyo, Japan).
Histomorphometric analysis was performed using the software NIS-Elements (Nikon) according to the manufacturer's instructions. Briefly, images were obtained with a DS-Fi1/digital camera connected to an Eclipse 80i histological microscope (Nikon), equipped with an automatic scanning table (Prior, USA). A total scan, one large image assembled from 100–120 images (Figure 2) of the region of interest containing the SF biomaterial and the corresponding peri-implant tissue was taken using a × 100 magnification at a resolution of 2500 × 1200 pixels. For each animal the slide that was stained with the anti-mouse vWF antibody (see histological preparation) was used to detect the host vascularization of the scaffold. Using the NIS-Elements ‘annotations and measurements’ tool, the total biomaterial area and blood vessels were marked separately by the ‘area’ tool. The total number of vessels on each slide was determined. The number of vessels/mm2 was calculated from the total vessel number found in the biomaterial area divided by the total biomaterial area in mm2. For each time point, a mean number of vessels/mm2 was determined. The SF degradation was determined by surrounding the outer surface area of 50 randomly selected SF fibres within one scaffold with the ‘area’ tool. The mean surface area was determined for each animal, and the group was then averaged to obtain the mean surface area for each group of eight animals in the study. In order to assess the dynamic degradation, the mean SF fibre surface areas from days 15, 90 and 180 were compared to the mean SF fibre surface area of day 3, assuming no, or a very slow, degradation of SF fibres in the first 3 days after implantation.
3.1. 30 min FA-treated SF
Animals implanted with the 30 min FA-treated SF scaffolds showed no signs of haemorrhage or necrosis at the implantation site. The implantation of this material induced no capsule formation within the first 3 days after implantation (Figure 3A). At the same time, a mild neutrophil infiltration was observed in the implantation bed. However, neutrophil inflammation subsided 10 days after implantation (Figure 3B). Starting at day 10, SF fibres were well integrated (Figure 3B) within a cell-rich connective tissue and the SF-scaffolds began to be mildly vascularized by day 15 (Figure 3C). The dynamic vascularization of the scaffold was quantified at each time point, as shown in Figure 6A. At 15 days, 10 vessels/mm2 were observed lying within the SF scaffold (Figure 6A). This vascularization rapidly increased over the next 3 months after implantation (Figures 4A–C) and by day 90 there were 65 vessels/mm2; a greater than six-fold increase since day 15 (Figure 6A). Vascularization of the 30 min FA-treated SF continued to increase for the remainder of the observation period (Figures 4D–E), with the vessel density nearly doubling over the last half of the study to 126 vessels/mm2 by day 180 (Figure 6A). Thus, during the 180 day observation period, well-defined blood vessels developed and formed networks throughout the entirety of the biomaterial scaffold. Throughout the study, the amount of connective tissue between the SF fibres decreased and consequently the SF fibres appeared closer to each other, resulting in a tissue dominated by SF fibres and large-lumen vessels (Figure 4E).
The dynamics of SF fibre degradation was quantified and is shown in Figure 6B. The 30 min FA-treated SF did not undergo a complete degradation during the evaluation period. However, after 180 days of implantation, the SF fibres noticeably decreased in volume, losing up to 95% of their surface area when compared to the SF fibre area after 3 days. The degradation of SF was associated topographically with macrophages and multinucleate giant cells in concert with lymphocytes. Both multinucleated giant cells and macrophages had their highest concentrations at day 90 after implantation, indicating a more intensive degradation phase of the SF biomaterial within the first 3 months after implantation. Quantitative analysis supports this histological impression of early degradation. After 90 days, the area of the 30 min FA-treated SF fibres had lost 85% of its original day 3 area. Through the end of the study, the number of multinucleated giant cells declined, while macrophages persisted at a constant low level from day 120 to day 180. The mild persistence of macrophages and the decline of the multinucleated giant cells indicates the reduction of an ongoing foreign body reaction to SF in the later phase of the study. Time points after 6 months were not examined, since all inflammatory cells involved in the degradation process of SF showed a decreasing or stable level, such that no major changes in the peri-implant tissue reaction to SF were to be expected.
3.2. 60 min FA-treated SF
For the groups implanted with the 60 min FA-treated SF scaffolds, the formation of a moderate capsule rich in blood vessels and early inflammatory cells, such as granulocytes and a few macrophages, was observed at the 3, 10 and 15 day time points (Figures 3D–F). The capsule remained into the later time points of the study, effectively isolating the SF scaffold from the peri-implant tissue, although the thickness of this capsule decreased in thickness between days 60 and 90 after implantation (Figure 5A, B). With this capsule thinning, macrophages, multinucleated giant cells and lymphocytes initiated scaffold degradation and vascularization (Figure 5C). The SF scaffold, which appeared through the first 3 months as a bulk mass, underwent a fast degradation in the last 3 months until day 180 after implantation. At 180 days, the SF fibre surface area had decreased in area by 90% when compared to the initial 60 min FA-treated SF scaffolds at 3 days after implantation (Figure 7A). Corresponding to this late degradation, the vascularization of the longer FA-treated scaffolds increased between day 90 (32 vessels/mm2) and day 180 (86 vessels/mm2) after implantation (Figures 5C–E, 7B). After 180 days, the SF scaffold became a well-vascularized reticular tissue that was morphologically similar to the one shown for SF with a 30 min FA treatment at 180 days (Figure 5E). Interestingly, the mean SF fibre area at day 180 after implantation was 380 µm2, which was between the values observed for 30 min FA-treated SF at days 90 and 180.
In this study, we examined two SF scaffolds differing only in the duration of the formic acid processing time, either 30 or 60 min. Since the goal of this study was to determine whether the different formic acid treatment time influenced the tissue response, an extensive analysis of structural changes occurring in the different materials was not carried out. However, visual differences in scaffold morphology could be seen in the two different scaffolds and may be attributed to changes in dissolution of the fibroin filament and hence fibre entanglement formation, but also to structural changes induced by the FA. Previous studies have shown that dissolution of SF in formic acid for 60 min leads to an increase in both short- and long-range fibre order compared to aqueous processing, due to a transition from a primarily random coil secondary structure to one with a significantly more β-sheet character (Um et al., 2001, 2003). Thus by altering the time SF is exposed to formic acid, scaffolds with different morphological and molecular characteristics could be obtained.
Surprisingly, altering only this one processing parameter leads to a significantly different tissue reaction in vivo. Longer FA processing results in less penetration of the surrounding mesenchymal cells, with the scaffold existing more as a bulk mass encased within a thick capsule that forms as early as 3 days. In contrast, the SF with a shorter FA treatment shows no capsule formation and there is evidence of cell penetration into the scaffold, most likely due to greater scaffold porosity. This drastically different inflammatory response occurs as a result of a change only in the FA treatment time, while all other processing and materials preparation conditions remained the same. These differences in early inflammatory response led to different tissue reactions toward the two materials. The SF fibres treated for only 30 min begin to degrade rapidly, likely attributable to the ability of immune cells to penetrate the scaffold and break it down from within, as opposed to only being able to access the periphery as in the longer treatment case. The beginnings of degradation are not observed until 90 days in the scaffold treated for 60 min, due to the capsule formation. However, once the capsule is thinned, the degradation process for this material occurs very rapidly. The histological and quantitative findings indicate that after the breakdown of the capsule surrounding the 60 min FA-treated SF scaffolds begins around day 90, a fast dynamic degradation takes place. This degradation is similar to that observed within the first 90 days for SF that has been treated for only 30 min with FA. Thus, the degradation is delayed in the 60 min FA-treated case, due in large part to the capsule formation, but once the capsule thins at 90 days after implantation, the degradation proceeds similarly to that which begins right away for the 30 min FA-treated SF scaffolds.
The early inflammatory responses to the 30 min FA-treated SF consisted of mild neutrophil (PMN) infiltration as well as a mild macrophage infiltration in the implantation bed, beginning at day 3 of the study. The low-level acute inflammation was verified by the fact that neutrophil infiltration subsided in this group 10 days after implantation, indicating that it was likely related to operative trauma and the subsequent wound-healing process resulting from implantation of the SF scaffold (Anderson, 2003). Accordingly, this result indicates that the standard 30 min FA-treated SF does not induce a major inflammatory response and does not contain substances that could cause infection. The presence of macrophages, beginning at day 3 after implantation, indicates the involvement of mononuclear cells in SF degradation from the very early implantation time points.
The early capsule formation surrounding the 60 min FA-treated SF promotes its isolation from the host immune response and thus contributes to its persistence as a ‘place holder’ for up to 90 days. However, the capsule thickness decreases between 60 and 90 days post-implantation, enabling cells that are involved in SF degradation to access the SF scaffold and initiate its degradation from the periphery—a good response for a scaffold that is intended as a longer-term ‘place holder’. This results in SF thickness reduction and, correspondingly, total scaffold penetration by macrophages, multinucleated giant cells and lymphocytes, which are all involved in its degradation. Interestingly, the inner portion of the 60 min FA-treated SF becomes well vascularized 90 days later than the 30 min FA-treated SF. However, both scaffolds had a similar extent of vascularization at the end of the study, indicating eventual acceptance and integration for both of the scaffold types, independent of FA treatment.
Silk fibroin has great potential for use in tissue engineering for a variety of reasons. Initially, the natural origin of the polymer and reasonable production costs make it an attractive biomaterial. Add to that the present findings of a mild peri-implant tissue reaction, reasonable degradation and considerable vascularization rates, and this material becomes an ideal candidate for a number of therapeutic applications. The present findings demonstrate that simple changes in the FA processing time could enable the material to be tailored for a host of specific tissue-engineering applications, since FA processing was found to influence the in vivo degradation and vascularization. For cell-based tissue-engineering strategies in which mesenchymal and endothelial cells are combined, shorter FA processing may be advantageous, as the porous structure will allow the mesenchymal cells to spread between the fibres and build a 3D network on which endothelial cells can form vessel-like structures (Fuchs et al., 2006, 2009; Unger et al., 2004, 2007). On the other hand, filling a bone defect or void in tissue, necessitating a slowly degrading scaffold that can recede as tissue ingrowth begins from the periphery, may call for an SF scaffold that has had a longer FA processing time; the initial characteristic as a ‘place holder’ is followed by a degradation from the periphery that is initially delayed but then progresses rapidly, allowing the integration of the SF fibres into the host tissue.
We have previously shown that endothelial, osteoblast and co-cultures of these cells attach, grow and proliferate on the 30 min FA-treated silk fibroin scaffolds in vitro (Fuchs et al., 2009; Unger et al., 2007). In co-cultures of osteoblast and endothelial cells on the 30 min FA-treated scaffolds, endothelial cells form microcapillary-like structures that weave between the osteoblasts and the silk fibroin fibres, generating lumen-like structures. Similar in vitro studies are under way with the 60 min-treated SF scaffolds to determine whether structural changes of SF due to longer formic acid treatment have any harmful effects on endothelial and mesenchymal cells, and whether their functionality is inhibited. Additionally, further ongoing in vivo experiments with precultured 60 min FA-treated silk fibroin (SF) scaffolds with endothelial and mesenchymal cells will help to evaluate whether the initial capsule formation around this micro-net will result in an isolation of ex vivo precultured cells from the host, and thus lead to death of the transplanted cells.
The histological technique used to generate one large image (see Methods) allowed a complete conventional histological evaluation of the entire SF scaffold, including both the centre and margins of the SF micro-net. This technique significantly improves the ability to obtain a complete histological analysis of an implanted biomaterial, by not only looking at small portions of the explanted material and instead evaluating the material as a whole in one large image. This technique makes comparison studies between different animals and different treatment groups more precise and should also have further applications in other fields outside of biomaterials. In this study the dynamic vascularization of the two SF scaffolds was quantified by examining the entire vasculature within the total biomaterial area. The ability to scan an entire histological preparation and assemble 120 single smaller images into one large image, combined with the image analysis software, enabled us to develop a novel method to objectively examine and quantifiably evaluate the vascularization throughout the entire biomaterial scaffold.
Sufficient vascularization is the most desired criterion for successful integration of a biomaterial within the implant site for tissue-engineering applications. Accordingly, new strategies are being examined to improve in vivo vascularization of scaffolds after implantation by, for example, preculturing a biomaterial ex vivo with endothelial cells alone or in co-culture with mesenchymal cells (Fuchs et al., 2009; Unger et al., 2007; Egana et al., 1993; Hofmann et al., 2008). All of these approaches are aimed at achieving better scaffold vascularization following implantation. However, understanding the host-derived vascularization of a scaffold is necessary in order to determine whether cell-based strategies augment or improve biomaterial vascularization after implantation. Thus, quantification of the in vivo vascularization of scaffolds without added cells is essential in order to evaluate any additional contribution to the vascularization by ex vivo cultured cells. As demonstrated in this study, the structural and chemical properties of a biomaterial may also affect the degree and rate of vascularization, and this must be understood in an in vivo setting. The previous studies by our group using human endothelial cells of different origins and human osteoblasts precultured on 30 min FA-treated SF scaffolds have shown that osteoblasts support endothelial cells to form capillary-like structures within the scaffold in vitro (Fuchs et al., 2009; Unger et al., 2007). The ability to prevascularize these SF scaffolds, combined with the host-derived vascularization shown here, is promising for use in repair for tissue injury necessitating a rapidly vascularized scaffold.
As tissue engineering progresses into more complex in vitro and in vivo studies, new technological approaches are necessary to adequately evaluate cell culture and tissue samples. Current histological and immunohistological preparations are generally qualitative. However, more precise and quantitative information needs to be obtained in order to assess the feasibility and benefits of translating the many systems being developed into a clinical setting. This is apparent in current regenerative medicine that incorporates biomaterials and tissue-engineering strategies, where material structure and composition can affect cell ingrowth, immune response, vascularization and degradation, directly affecting the degree of success for these materials-based therapies. As we have shown here, a quantitative histological assessment of silk fibroin-based biomaterials, differing only in the time of FA treatment, further described the substantial differences in the resulting tissue reaction. These differences point to the possibility of tuning the in vivo tissue reaction of SF-based materials for diverse applications in tissue engineering that can be further evaluated now that these histological differences are understood on a qualitative and quantitative level. Studies using similar quantitative methods for analysing the vascularization of materials preseeded with cells both before and after implantation are ongoing. Such quantitative analysis is crucial for assessing the benefits of cell–material therapies, both in vitro and in vivo.
We examined at various time points up to 180 days the in vivo host tissue reaction to subcutaneous implantation of two silk fibroin micro-nets, differing only in the duration of formic acid treatment. Qualitative histological evaluation, combined with a method for quantifying the dynamic in vivo vascularization and degradation, revealed that the duration of formic acid treatment has a major influence on the tissue response to the material. SF scaffolds treated with FA for 30 min are well integrated in the peri-implant tissue and have some vascularization within the first 15 days, followed by a rapid increase over 90 and 180 days after implantation. SF fibres treated for 30 min with FA also underwent a continuous degradation, with most degradation occurring in the first 90 day after implantation. Strikingly, increasing the FA treatment time to 60 min leads to a very different tissue reaction. Within the first 90 days, the formation of a barrier-like fibrous capsule limits scaffold vascularization and degradation. With time this capsule is thinned and rapid degradation and vascularization occurs. The process of vascularization and degradation that begins at 90 days for the 60 min FA-treated SF resembles the dynamics of that which occurs over the first 3 months after implantation for the 30 min FA-treated SF. Thus, altering only the FA treatment time can significantly change the resulting tissue reaction and degradation properties, making SF widely applicable to various areas of tissue engineering for both soft and hard tissue replacement.
The authors would like to thank C. Braun, A. Valentino, U. Hilbig and K. Molter for their excellent technical assistance. This work was financially supported by grants from the European Commission (EXPERTISSUES Contract No. 500283-2).