A Tunable Silk Hydrogel Device for Studying Limb Regeneration in Adult Xenopus Laevis

The fourth presented antibody, collagen type II, was detected in the cartilage center of the controls, but not the device animals. Collagen type II is one the two major components found in the typical extracellular matrix (ECM) of hyaline cartilage [4]. The presented image for col2 in the control animal (Fig 4) is what is generally observed in the normal X. laevis spike formation, that is a distinct increase in fluorescence intensity for col2 in the spike center compared to the peripheral tissue [17]. On the other hand, the staining profile for the device animal shows little or no increase of intensity at the core compared the peripheral tissue. This would suggest that animals that wore the device exhibit a major decrease in one of its major ECM components compared to the controls.

Fig 5 shows the histology of animals with the three different types of outer sleeves implemented in the device design. It should be noted that typical spike formation for X. laevis will have a hyaline cartilage center with peripheral tissues composed of skin, mucosal glands, and fibrocartilage [15]. The different types of tissue observed in typical spike formation have been well documented histologically [14,15,33], and the histology from these early studies was used as an aid in identifying the observed tissue types. The spike formation of the device animals where a cloth sleeve was used had higher calcified tissue volume and surface area than cloth sleeve control animal, but were otherwise morphologically similar to the controls. This was reflected in the decrease in new bone surface area to volume ratio for the cloth device animals in comparison to controls (Table 1). The stiff outer sleeve design showed differences morphologically to the other device designs (Fig 5). The key differences were that the animals wearing the devices showed atypical tissue formation in the tissue around the old bone growth and in the hyaline cartilage spike. These spikes were shorter and thicker at the base than in the controls. Histologically the soft dragon skin device resulted in longer and narrower spike formation as well as more tissue integration between the spike cartilage core and the surrounding tissue [15].

The current method of attachment used both a rubber wrap and two surgical stitches to secure the device to the amputation site with minimal damage to surrounding tissue. However, in order to understand the effect that this method of attachment on biological outcomes, both methods were evaluated individually. Thus, the devices were attached to the animals for 24 hours using either a wrap or two stitches. Micro-CT was used to image changes in calcified bone formation at 2 months post amputation. Both methods led to some calcification and showed signs of calcified tissue forward of the amputation plane. The stitching attachment method showed more calcified tissue behind the amputation plane than the wrap. The wrap attachment showed signs of calcified tissue formation within the new tissue of the cartilage spike. A small domain of calcified tissue was observed in the spike formed (Fig 5) which was confirmed using micro-CT volume analysis and is well beyond the amputation plane.

One of the major changes in device design was the addition of the hydrogel inserts (Fig 6A). Micro-CT was used to compare calcified tissue formation between a device animal with no hydrogel insert (Fig 6A, no hydrogel) and a device animal with a hydrogel insert (Fig 6A, with hydrogel). At about 2 months post device attachment the device animal (for this situation it was a control and an experimental animal) with the attached hydrogel insert device had more calcified tissue compared to the control. Volume analysis estimated bone density of the calcified tissue growth at 0.00022 μm-1 and 0.00399 μm-1 for no hydrogel and hydrogel, respectively. The morphological difference that resulted from the addition of the hydrogel insert to the hydrogel base was also characterized (Fig 6B). The animals had the same stiff outer sleeve and utilized the wrap method of attachment. The animal in Fig 6B with hydrogel had only a hydrogel base, while the animal in Fig 6B hydrogel had both a hydrogel base and sleeve. Micro-CT was also compared with the histology. The biggest difference was at the base of the spike; in Fig 6B, no hydrogel the amputation site was clearly defined with dense pockets of calcified tissue growth, while the hydrogel animal showed a large mass of disorganized bone marrow like tissue, calcified tissue, and hyaline cartilage.

A device was generated that could be attached to X. laevis to study limb regeneration in a controlled setting, and without damage to the surrounding healthy tissue. Utilizing a hydrogel system as the medium of microenvironment manipulation negated the need for a water tight seal, which was the main cause of necrosis in past work [20]. The method of attachment, which utilized both stitching the device to the animal and then further securing it with a wrap, allowed for secure attachment while minimizing physical damage caused by compression.

A hydrogel insert provided important options for the device and in future work. Specifically, the hydrogel provides mechanical force, a material interface, and a substrate for drug delivery. All three of these pathways of micro-environmental manipulation can affect the final outcomes of regeneration [1,3,7]. In the present work, the hydrogel was able to handle the physical forces placed on the hydrogel during device attachment. This is due to the enzymatic crosslinks, which provided elastic behavior. This contrasts with the prior system that utilized a liquid reservoir [20]. Moreover, the different material and mechanical properties the hydrogel provides to the wound bed can be easily tuned by varying the concentration of silk or the silk molecular weight [26]. This tunability can be applied to both the static and dynamic aspects of the hydrogel insert. While the final stiffness did not change significantly by varying the silk concentration used in the hydrogel in the present work, due to the limited ranges studied, the pre-attached stiffness and degree of contraction did change and this can have a significant effect on the biological outcomes. The dynamic aspects of the different materials and mechanical properties also play an important role in the regenerative process [34]. In particular, in skin injuries the amount of pressure a bandage applies and how that pressure changes over time define initial wound closure as well as the degree of scar formation [35]. Thus, it is critical to have dynamic control over the device mechanical properties. Past work with silk HRP hydrogels has shown that changing the molecular weight can have the greatest impact on the mechanical stiffness of the hydrogel [26].

Due to the nature of the crosslinking and the natural anionic nature of the silk protein, the resultant hydrogel was sensitive to changes in pH, buffer solution, and the ionic content of the environment. This result in hydrogel insert contraction observed during device attachment. The contraction is a fundamental property of charged hydrogels and so this is not a surprising result. However, this contraction will influence the biology and should be matched to the target experimental goals. The hydrogel swelling and de-swelling can also be tuned through the molecular weight of the polymer used or a change in the composition of the solute used in the gelation process [27]. The current study explored two of the possible routes in tuning the silk HRP hydrogel insert. The overall amount of contraction was influenced by the concentration and of silk in the hydrogel as well as composition of the pre-gelation solution.

The increase in concentration of silk would have resulted in an increase in both polymer density and available reactive side chains (i.e. tyrosine) in the silk hydrogels. Both of these factors would define the ability of the hydrogel to swell and contract, since both would limit the ability of silk to accommodate water in the polymer mesh [32]. However, this observed decrease in hydrogel contraction in relation to increase in concentration is probably due to an increase of bulk polymer density rather than crosslinking density. The latter option is both sterically hindered due to the heterogeneous nature of the silk polymer chain and would also require an increase in enzyme concentration to compensate for the increased number of reactive groups. Increased polymer density versus crosslinking density would explain the impact of variation in concentration on initial gel stiffness more than its final stiffness (Fig 2A). An increase in crosslink density should result in a larger increase in final stiffness than that observed [36].

The second parameter that was varied in the work to tune the contraction of the hydrogel was the composition of the pre-gelation solution. The addition of the salt in the FW. The presence of salt can effect both the structural stability of the enzyme as well as the enzymatic reactivity. Low concentrations (the amount present in FW) can stabilize the overall tertiary structure of the protein (known as salting in) [37]. This increased structural stability would increase both the enzyme ability to crosslink and its working half-life [38,39]. Salt is also known to dramatically increase the enzymatic activity of HRP in both aqueous and non-liquid environments, such as the one presented in FW [40] or at the benchtop [41]. This could account for the increase in contraction speed that was observed in those situations.

While the magnitude of increased stiffness due to contraction was not large, it was statistically and biologically significant. The resulting change in material and mechanical forces can affect cell responses during wound healing. Different key biological pathways needed in proliferation, metalloprotease activation, and cell migration for wound closure and later stages of wound healing are activated in a mechanotransductive manner [42]. In vivo studies showed that application of cyclic tensile force on skin flaps in rats showed an increased rate of healing based on the speed and degree of revascularization of the injured tissue [43]. Further work demonstrated that application of a tensile force had an increased impact on angiogenesis in conjunction with the up regulation of hypoxia induced factor alpha 1 [44]. This gene is known to up regulate the production of vascular endothelial growth factor 1, and other essential proteins used in regenerative pathways [45]. Future work will focus on utilizing this device to help expand the role of mechanical forces on such areas of the wound healing process.

The spike formations observed between the animals with and without devices were morphologically evaluated. Histological and antibody staining showed qualitative differences in the anatomical structure between the experimental and control animals. This was further quantified through volume analysis of the calcified tissue. Both sets of animals showed signs of calcified tissue in the spikes. However, the quality of calcification and amount differed between the two groups. First, the controls usually displayed more of a calcified callus beyond the amputation plain that showed no signs of bone remodeling. The amount of callus formed in these controls increased over time, a trend not shown in the calcified tissue formed in the experimental animals. The calcified tissue in the device animals was more complex, with a filled in porous structure that extended beyond the amputation plane. Calcification of hyaline cartilage tissue occurs when hyaline cartilage undergoes mechanical forces and the cartilage closest to the site of mechanical forces becomes structurally reinforced by forming a calcified callus [4]. This event is known as chondrocalcinosis and is an adaptive response to the environment [4,46]. This could explain the observed fibrocartilage callus formed in the control groups. The callus would also develop over time and is similar to that seen in the controls. The observed calcified tissue for the experimental group is not an adaptive mechanism alone. The architecture was similar to that observed in trabecular bone [4]. This type of bone is formed through remodeling of calcified tissue, thus vascularization and innervation are needed. Further work is needed to understand the actual biological mechanisms, yet the complexity of the calcified tissue in the device animals appears to be a pro-regenerative response, above and beyond adaptation to the environment alone.

No matter which biological pathways are utilized to induce regeneration, innervation and vascularization will be required prerequisites in the process. Vascularization provides a route for the delivery of nutrients and oxygen, and a means to remove toxic waste from the tissue [1]. In relation to limb formation, the event which initiates the transformation of cartilage tissue to calcified bone is the vascularization of the articular cartilage tissue [4]. The immunostaining here showed that the device animals stained positive for smooth muscle actin. In this anatomic location, the only cells with high expression of this type of actin would be the endothelial cells that line blood vessels. The positive stain was found deep in the spike, even beyond the location of the calcified tissue. The nerve marker also stained positive in the same location as SMA. Tissue innervation has been demonstrated to be essential for cell migration and pattern formation for limb regeneration in axolotl [47]. Monaghan et al., demonstrated that the blastema can only be formed at an innervated amputation site [48]. The blastema provides the progenitor cells that are required for limb regeneration in this species. Therefore, it follows that for limb regeneration is dependent upon innervation.

The ECM plays major roles in tissue structure, cellular communication, and cell differentiation [12]. All of these are critical factors for successful tissue regeneration [12]. Each tissue type has a particular ECM composition designed to allow that tissue to perform its given function [12]. If the ECM composition is altered, then the function and tissue type could be altered. The results indicated a possible decrease in a major component of the typical spike ECM composition, which could point to a major alteration in the overall tissue in the device animals. Immunohistology is a qualitative assessment of the changes observed in the resultant tissue. More quantitative measures will be needed before definitive conclusions can be drawn on the observed morphological changes present in the device animal group. Yet, these findings, in the aggregate, suggest that the tissue in the resultant spike for the device animals has deviated from what is typically seen in this tissue for adult X. laevis [15]. Furthermore, this shift from the norm is one that shows an increase in innervation, vascularization, and proliferation; all requirements for regeneration [13].

One of the most defining features of the experimental animals was the level of connection that the usually isolated cartilage tissue of the spike core had with the surrounding tissue. This level of connection was seen to some degree in all the animals with devices in the stiff and soft dragonskin outer sleeve versions. Furthermore, the spike core of the experimental animals was usually connected to the old bone marrow. Device animals also had cartilage formation that surrounded the old bone growth behind the amputation plane. The expression of nerve and vasculature protein markers also tended to enter the spike core through this cartilage bridge. All the pro-regenerative changes that were observed in device animals looked similar to endochrondral ossification as seen in bone fracture healing [4]. Fracture healing occurs in three major steps: inflammation, repair, and remodeling [1,4]. The initial inflammatory response is a destructive phase in which the body destroys any foreign entities and removes damaged tissue. In the following reparative phase, hyaline cartilage forms around the fractured bone site and forms a calcified shell. The cartilage tissue then undergoes vascularization and innervation which leads to calcification and then transformation to progenitor bone cells [4]. At this point the final stage of wound healing occurs in which the newly formed bone is remodeled to its final dense and porous architecture [4]. Almost every major observed difference seen in the animals with devices can be placed into the context of these fracture healing steps, and is summarized in Table 2. Recent work from Dawson et al observed similar trends between mice fingertip amputation and fingertip bone fracture healing [49]. The key to complete regeneration in adults could be linked to manipulating the biophysical cues that controls wound resolution.

Future work will explore this possibility, but what should be noted is that this device provides the required platform to explore these types of questions.

Another key aspect to consider is that the success of bone repair relies on the ability of cells from the periosteum and bone marrow to reach the fracture site [1]. In studies of limb regeneration in X. laevis, progenitor cells from the limb bud of regenerative tadpoles were delivered to the amputation site of non-regenerative froglets through the use of a fibrin hydrogel [18]. One of the aspects of the hydrogel system that aided success was that the hydrogel helped keep the wound site open and biologically active. This aided the body in making that needed connection to the marrow cavity [18]. This could be one of the underlying mechanisms behind the device pro-regenerative abilities seen in the present work as well.

The present device went through multiple variations in design. Both the mode of attachment and the outer sleeve design had an effect on spike formation, with the hydrogel insert appearing to be the predominant factor. Since the inner hydrogel insert was in direct contact with the wound site, the local environment at the amputation site will be influenced by the different material properties of that hydrogel. Future work with the device can be broken down into two parts: characterization of the biological pathways which result from the application of the device, and technical characterization and optimization of the device design to provide the best system for future research. The fundamental mechanisms that define static and dynamic nature of the HRP silk hydrogel insert with and without the device will be needed. The presented research begins to probe and classify what drives the contraction force. The hydrogel is a weakly ionic, heterogeneously cross-linked, polymer mesh that has negative charged groups. This makes it sensitive to the ionic content of the system, as seen in the contraction during attachment. Initial variations in concentrations and gelation solution show possible methods of tuning the hydrogel to manipulate both the static and transient properties. These features can be exploited to offer defined contractive forces during healing to examine these relationships systematically. As mentioned earlier, the input of such forces is known to offer positive outcomes in skin healing [3].

Due to the nature of the present study, as the focus was on device design rather than biological pathway analysis, it is unclear whether the observed morphological phenomena were caused by activation of wound healing or regenerative biological pathways. Many of the known signs of these pathways being activated can be seen in the upregulation of certain genes within the first 2 weeks of injury[1,33,50]. This can be assessed using quantitative PCR. This information would clarify the device’s direct effect on the wound site. Clarification of old versus new tissue growth at the amputation site is also required for future device optimization. Bone could be labeled using tetracycline [51], calcein [52], and Brdu [52] to provide a more objective means of categorizing what is going on morphologically in the observed biological outcomes.

In the more technical area of device design, subsequent steps will include utilizing the hydrogel insert for drug delivery. Preliminary in situ experiments have demonstrated typical diffusion driven hydrogel release profiles for lithium chloride and progesterone, two compounds known to promote regeneration [53,54]. The release profiles are presented in supporting material S2 Fig. Still, more sophisticated methods will be needed to obtain more precise spatial and temporal control over any type of compound delivery. Past work on other pH sensitive hydrogels have been able to accomplish similar goals using streptavidin-biotin conjugation [55,56]. Characterization of the hydrogel dynamics will also play a critical role in this area of device design as well. The type and speed of release of any loaded compound in the hydrogel insert can be impacted by how it contracts. Past work in other pH sensitive “smart hydrogels” demonstrated a biphasic release profile by such a manipulation [57]. This level of control would be a very desirable trait in future applications, and is essential in activation of regenerative bio-pathways.

Future work will utilize the drug release capabilities of the hydrogel insert to connect observed morphological changes in the resultant spike formation to potential deviations in its underlying biological mechanism. One method which this could be done is to have the hydrogel insert deliver a given antagonists or agonists of a growth factor that is known to play a major role in an established regenerative pathway (e.g. a TNF-α inhibitor during the early stages of inflammation [5]). The changes in the outcome from the addition of the compound could be assess at both the macroscopic (e.g. variations in morphological structure and protein expression) and microscope (e.g. variations in typical DNA and RNA expression) level. Comparing the resultant deviations in both its phenotypic and genetic profile would help connect what is observed morphological to what is happening mechanistically.

The purpose of the work was to develop a device that can be used to probe the biological mechanisms behind limb regeneration. The device was designed for frog amputation studies and added a layer of control and analysis through the incorporation of a silk HRP hydrogel insert in contrast to prior solution reservoirs; adding both mechanical inputs to the healing process and reduced damage to native tissue due to the attachment. The device also presented an initial set of metrics with which to evaluate the observed gross anatomical outcomes, and showed that the hydrogel insert had an effect on biological outcomes, in this case spike formation from the amputated limb. While additional work is critical to categorize possible biological mechanisms involved in these outcomes, these tools allow options to tune and evaluate this system through quantitative measures.

The authors would like to thank Dr. Gary Sahagian, Dr. Min Fang, and Dr. David Wilbur. Dr. Sahagian provided access to the Micro-CT used in the present research, while Dr. Fang provided the needed technical training to operate the machine. Dr. Wilbur provided access to and training for the ICP-AES used.

NIH (R01 AR005593, R01 AR061988) funding for DK and ML and W. M. Keck Foundation to ML and DK. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.