Digit specific denervation does not inhibit mouse digit tip regeneration

Abstract

It is long-established that innervation-dependent production of neurotrophic factors is required for blastema formation and epimorphic regeneration of appendages in fish and amphibians. The regenerating mouse digit tip and the human fingertip are mammalian models for epimorphic regeneration, and limb denervation in mice inhibits this response. A complicating issue of limb denervation studies in terrestrial vertebrates is that the experimental models also cause severe paralysis therefore impairing appendage use and diminishing mechanical loading of the denervated tissues. Thus, it is unclear whether the limb denervation impairs regeneration via loss of neurotrophic signaling or loss of mechanical load, or both. Herein, we developed a novel surgical procedure in which individual digits were specifically denervated without impairing ambulation and mechanical loading. We demonstrate that digit specific denervation does not inhibit but attenuates digit tip regeneration, in part due to a delay in wound healing. However, treating denervated digits with a wound dressing that enhances closure results in a partial rescue of the regeneration response. Contrary to the current understanding of mammalian epimorphic regeneration, these studies demonstrate that mouse digit tip regeneration is not peripheral nerve dependent, an observation that should inform continued mammalian regenerative medicine approaches.

1. Introduction

A nerve requirement for successful epimorphic regeneration has been known for almost 2 centuries (Todd, 1823), and nerve-dependent regeneration has been documented in all vertebrate groups ranging from fish to mammals (reviewed in (Farkas and Monaghan, 2017)). The best studied model is the regenerating limb of salamanders where blastema formation fails in the absence of nerves (Tanaka, 2016). The neurotrophic influence on limb regeneration is required for both blastema cell proliferation and limb patterning (Kumar et al., 2007; Nacu et al., 2016), and is mediated by innervation-dependent production of growth factors (reviewed in (Pirotte et al., 2016; Satoh et al., 2015)). The regenerating mouse digit tip represents a mammalian example of epimorphic regeneration (Borgens, 1982) and innervation has been linked to growth factor stimulated blastema cell proliferation suggesting conservation of nerve dependency with salamander limb regeneration (Johnston et al., 2016; Rinkevich et al., 2014; Takeo et al., 2013). However, a detailed analysis of axonal regeneration during mouse digit tip regeneration identified that axon regeneration was severely impaired, and quantitative analyses indicated that the blastema had reduced levels of both regrowing axons and Schwann cells (Dolan et al., 2019). As such, blastema formation in the absence of regrowing axons is inconsistent with the conclusion that digit tip regeneration in mammals is innervation dependent and that nerve dependency in mammalian regeneration represents a conserved vertebrate trait.

Studies demonstrating nerve dependent digit tip regeneration utilized a surgical procedure known as sciatic denervation (Johnston et al., 2016; Mohammad and Neufeld, 2000; Rinkevich et al., 2014; Takeo et al., 2013) which effectively eliminates axon regrowth to the amputated digit tip, but also severely limits limb use due to paralysis. An important consideration is that limb paralysis induced by sciatic denervation is also used as a model to study the role of mechanical load on skeletal and muscular homeostasis (Gross et al., 2010). During zebrafish fin regeneration, which is also nerve dependent (Simoes et al., 2014), decreasing mechanical load by induced paralysis without denervation is inhibitory for the regeneration response (Recidoro et al., 2014). Similarly, mouse digit tip regeneration was recently shown to be mechanical load dependent; continuous hindlimb unloading completely inhibits wound healing, blastema formation, and skeletal re-growth, but can be rescued upon restoring mechanical load (Dolan et al., 2022). Therefore, the innervation requirement for mammalian epimorphic regeneration may be complicated by changes in mechanical load caused by denervation induced paralysis.

To study the specific role of innervation on digit tip regeneration, we developed a surgical procedure in which denervation is restricted to the individual digit to avoid deficits in mechanical load associated with limb paralysis (Cirnigliaro et al., 2017). Herein, we demonstrate that denervated digits undergo blastema formation and complete digit tip regeneration in the absence of peripheral innervation. However, digit tip regeneration is attenuated in denervated digits, and we show that this is due to an innervation dependent delay in amputation wound healing. Furthermore, innervation dependent attenuation of regeneration can be rescued in part by applying a wound dressing known to enhance the rate of wound closure (Simkin et al., 2015b). Taken together, these results show that the role of innervation in mouse digit tip regeneration is restricted to stimulating wound closure, and that epimorphic regeneration in mammals is not innervation dependent. These findings support the conclusion that innervation is not a requirement for successful appendage regeneration in mammals, and that nerve dependency in appendage regeneration is not a conserved vertebrate trait.

2. Materials & methods

2.1. Animals, digit tip amputations, and Dermabond® application

Adult 8-week-old, female CD-1 mice were purchased from Texas Institute for Genomic Medicine (College Station, TX) and used for all denervation experiments. Digit tip amputations of the 2nd and 4th hindlimb digits were performed as described previously (Dawson et al., 2019; Simkin et al., 2013). For specific studies, 5 μL of Dermabond® (Ethicon) was applied to the amputation wound immediately following digit tip amputation, allowed to dry for 1 min, and then re-applied (Simkin et al., 2015b). All animal use and techniques were compliant with standard operating procedures and approved by the Texas A&M University Institutional Animal Care and Use Committee.

2.2. Digit specific denervation surgery

Mice were anesthetized with isoflurane and hindlimb paws cleaned using betadine. Digit specific denervations were performed by creating an incision with micro-scissors on the ventral surface of the digit, which stretched from the proximal head of the first phalange to the fat pad. Digital nerves were transected at the distal fat pad, pulled proximally with forceps, and then a second transection was made creating a gap the size of the original incision. Finally, the open wound was closed using Dermabond®. The 2nd and 4th left and right hindlimb digits were used for experimental and sham operations, respectively. Digit tips were amputated five days after digit specific denervation or sham operation.

2.3. Micro-computed tomography (μCT), image processing, and quantification of bone volume, bone length, and rate of new bone formation

Terminal phalanx (P3) bone volume and length were measured using a vivaCT 40 (SCANCO Medical) as previously described (Dawson et al., 2018). μCT files were saved as a DICOM image stack, and subsequently uploaded to ImageJ where we used the 3D viewer plugin to create 3D renderings of bones. Bone volume was measured using the volume fraction tool in the BoneJ plugin in ImageJ. Bone length was measured using the point tool also in ImageJ. Bone volume and length are normalized to each individual digit’s volume or length based on scans taken within 24 h after amputation to account for variation in digit size and amputation level. Depending on the experiment, for statistical analysis we used either a paired or unpaired t-test using GraphPad Prism 7 (GraphPad Software).

2.4. Bone mineral density measurements

Bone mineral density (BMD) and fat composition were determined to follow any denervation-induced paralysis and mechanical unloading of the limb, as has been well-described in nerve injury (Cirnigliaro et al., 2017). BMD was assessed in vivo using the small animal ultra-high-resolution imaging and dual-energy x-ray absorptiometry (DXA) system Faxitron UltraFocus (FAXITRON BIOPTICS, LLC, Tucson, AZ) using software V2.4.2U. Mice were anesthetized and scanned at −5, 1, 10, 20, 30, and 42 DPA. The scans were analyzed with the Faxitron Bioptics LLC software using the Arbitrary ROI Tissue Statistics analysis. Analysis was performed on the whole femurs (left and right). For each region analyzed the data points collected included bone mineral density (BMD) (mg/cm²), lean weight (g), and fat percent (%). The covariance is less than 0.30% for all DXA scans. For statistical analysis we used an unpaired t-test using GraphPad Prism 7 software.

2.5. Digit processing, Mallory’s trichrome staining, and wound closure

Digits were collected from mice and fixed in buffered zinc formalin (Anatech Ltd) for 24 h at room temperature. Digits were decalcified using Decalcifier I (Surgipath), a 10% formic acid solution, for 24 h. Decalcified digits were processed through a graded ethanol series, xylenes, and immersed in paraffin wax. Digits embedded in paraffin wax were serial sectioned at 4–5 μm thickness. Before to histological staining, slides were incubated at 60 °C for 45 min, followed by incubation at 37 °C for 2 h, with subsequent deparaffinization with xylenes, a graded ethanol series, and eventual submersion in water. Mallory trichrome staining was performed to illustrate general histology. Slides were mounted using Permount Mounting Medium (Thermo Fisher Scientific). All slides were imaged using a Olympus BX60 microscope with an Olympus DP72 camera utilizing the DP2-BSW software. Digits were collected at specified time points for histological analysis to assess for wound closure. Wound closure was assessed using Mallory’s Trichrome staining. An amputation wound was considered closed if the entire epidermis covered the amputation injury in all sections on the slide.

2.6. Immunohistochemistry and EdU studies

Antigen retrieval was performed using heat retrieval performed in 1 Å~ citrate buffer solution (Dako), except for Runx2 where antigen retrieval was performed using EDTA buffer (pH 8.0; Thermo Fisher Scientific). Slides were blocked using Protein Block (Dako) for 1 h at room temperature. Incubation with primary antibody/antibodies was performed overnight at 4 °C; they were then washed in tris buffered saline with Tween R20 solution (Sigma-Aldrich Co.) and incubated in secondary antibody/antibodies for 1 h at room temperature. Slides were then incubated in a phosphate buffered saline (Sigma-Aldrich) and DAPI (Invitrogen) solution, dried, and mounted with Prolong Gold (Invitrogen). Samples were imaged using an Olympus BX61 microscope with a Hamamatsu ORCA-ER camera via the Slidebook software (Intelligent Imaging Innovations Inc.) EdU (5-ethynyl-2′-deoxyuridine) was injected intraperitoneal (10 μL/kg) 3 h prior to specimen collection. EdU detection is based on a copper catalyzed reaction between an azide and an alkyne, and was performed after washing of the secondary antibody immunohistochemistry but before DAPI was applied. Primary antibodies used were, rabbit anti-beta III tubulin (1–500; Abcam; Ab18207), rabbit anti-runx2 (1:250; Sigma Aldridge Co.; HPA022040), rat anti-CXCR4 antibody (1:500; R&D Systems; MAB21651); rabbit anti-Cathepsin K (1:100; Abcam; Ab19027); rabbit anti-S100-beta antibody (1:500; Abcam; Ab52642). Secondary antibodies used were goat anti-rabbit, chicken, or rat Alexa Fluor-488 or 568 (1:500; Invitrogen).

2.7. Quantification of β3T, S100β, EdU, and cathepsin K

Quantification of β3T (beta–III–tubulin) in the P3 bone marrow was described previously (Dolan et al., 2019). For quantification, the blastema region was defined as the area distal to the bone stump, proximal to the wound epidermis, ventral to the nail epithelium, and dorsal to the underlying epidermis. To quantify S100β (S100 calcium-binding protein B)-Schwann cells, EdU-positive cells, and cathepsin K-osteoclasts, digits were imaged at 200X magnification using a Olympus BX61 microscope with a Hamamatsu ORCA-ER camera. Using the Slidebook software, a region of interest (ROI; μm²) was manually determined and automatically quantified. For S100β, the area of S100β (μm²) was automatically determined then normalized to the ROI. For EdU, the number of EdU⁺ cells was automatically determined in the ROI, and was then normalized to the number of DAPI⁺ cells within the ROI. Cathepsin K-immunopositive signal (μm²) was manually determined, and only multi-nucleated osteoclasts attached to the proximal bone were included for analysis. Cathepsin K-immunopositive signal was then normalized to the ROI. Three sections per slide were analyzed, then S100β/EdU/Cathepsin K was averaged and combined with other digits for statistical analysis which consisted of an unpaired t-test using GraphPad Prism 7 software. Individuals were blinded to experimental groups during image collection and quantification.

3. Results

3.1. Digit specific denervation removes peripheral axons without perturbing mechanical loading

The neuroanatomy of the digit tip is highly organized, with autonomic nerves innervating the P3 bone marrow and sensory nerves innervating the connective tissue surrounding the P3 bone (Dolan et al., 2019). Amputation of the digit tip transects sensory axons without damaging autonomic axons, and upon subsequent digit tip regeneration, sensory axons fail to regenerate whereas autonomic axons remain (Dolan et al., 2019). Peripheral nerve ablation was performed at the individual digit level (Fig. 1A) and validated based on the absence of beta–III–tubulin (β3T⁺) immunostained axons 5 days after ablation surgery, although remains expressed by dorsal epidermal cells (Dolan et al., 2019) (Fig. 1B). Maintenance of a denervated state was quantitated 21 days post-surgery by measuring β3T⁺ immunostained axons in the P3 bone marrow (BM) as described in an earlier study (Dolan et al., 2019). The level of innervation in denervated digits (0.19 ± 0.1 μm) was reduced by 17-fold compared to sham-operated digits (3.23 ± 0.2 μm; P < 0.0001) (Fig. 1C–E). Consistent with previous studies (Jessen and Mirsky, 2019; Sulaiman and Gordon, 2009), we found that denervated and sham-operated digits contained similar levels of S100β⁺ immunostained Schwann cells (P = 0.1902) indicating that digital denervation does not deplete the Schwann cell population (Fig. 1F–H). To determine if digit denervation altered mechanical loading of the limb, as has been shown previously following nerve injury (reviewed in (Cirnigliaro et al., 2017)), dual-energy x-ray absorptiometry (DXA) was used to measure changes in bone mineral density (BMD), lean weight, and fat percentage of hindlimb femurs of sham and denervated animals over a 7-week period. DXA analysis shows that there are no differences in BMD (Fig. 1I), lean weight (Fig. 1J), and fat percentage (Fig. 1K) at any timepoint indicating that bone volume and muscular mass was not modified by digit specific denervation. Taken together, the data show that digit specific denervation is an effective method for removing peripheral innervation to the digit tip without altering mechanical load.

Fig. 1.

Digit specific denervation ablates peripheral nerves in the digit tip but does not alter mechanical load

(A) Diagram of the digit specific denervation surgery. The digit is innervated by two ventral digit nerves (purple). An incision is made from the base of the first phalangeal element (P1) to the P2/P3 joint exposing the two digital nerves which are subsequently the nerves which is transected and removed (purple dashed lines). (B) Five days after digit specific denervation (den) beta–III–tubulin-immunopositive (β3T⁺) axons are not observed in the P3 bone marrow (BM) or surrounding connective tissue. The white dashed line represents the boundary between the epidermis and underlying connective tissues. The solid white line represents the amputation plane. (C) Quantification of β3T⁺ axons in the P3 BM of sham control and digit specific denervated digits at 21 days post-denervation (n = 8 digits/group; mean ± SD; **** = P < 0.0001). (D, E) β3T⁺ axons (green) are present in the BM of sham control digits (D) but absent in denervated digits (E) at 21 days post denervation. (F) Quantification of S100β-immunopositive (S100β⁺) Schwann cells in the connective tissue of sham control and digit specific denervated digits at 21 days post-denervation (n = 7–8 digits/group; mean ± SD). (G, H) S100β⁺ Schwann cells (green) are shown in the connective tissue of sham control (G) and digit specific denervated (H) digits at 21 days post-denervation. (D, E, G,H) White dashed line identifies the boundary of bone (b); Blue = DAPI stained nuclei; Scale bars = 100 μm. (I–K) DXA analysis of in vivo BMD (I), lean weight (J), and fat percentage (K) of hindlimb femurs subjected to digit specific denervation surgery compared to sham controls showing that denervation does not influence mechanical loading (n = 5–10 femurs/group; mean ± SD).

3.2. Digit tip regeneration is attenuated but not inhibited by digit denervation

To determine the effect of denervation on digit tip regeneration, digit specific denervated and sham control digits were amputated and monitored using micro-computed tomography (μCT) imaging to track the regeneration response (Fernando et al., 2011; Sammarco et al., 2014). Representative 3D-μCT renderings of regenerating denervated and sham operated digits show that digit specific denervation delays but does not inhibit the regeneration response (Fig. 2A). Regeneration of sham control digits is completed within 28 days post-amputation (DPA) whereas regeneration of digit specific denervated digits is not complete until 42 DPA. This delay in regeneration is validated by quantitative measurements of bone volume (BV) and bone length (BL) at different stages of the regeneration response (Fig. 2B and C). Digit specific denervated and sham control digits are statistically different at most stages of bone regrowth (14 DPA and on) but reach similar levels by both BV (P = 0.1094) and BL (P = 0.1900) at 42 DPA (Fig. 2B and C). Analysis of regeneration rate from 14 to 42 DPA indicates that denervated digits regenerate BV (P = 0.7793) and BL (P = 0.4512) at a rate comparable to sham control digits (Fig. 2D and E). The outcome of the regeneration response (42 DPA) is the replacement of a digit tip that is indistinguishable between denervated and sham operated controls.

Fig. 2.

Digit specific denervation does not inhibit digit tip regeneration

(A) Representative μCT renderings of a sham control (top) and a digit specific denervated (bottom) digit during a regeneration period of 42 days. (B, C) Quantification of bone volume (B) and bone length (C) measurements for sham control and digit specific denervated digits (n = 14–22 digits/group; mean ± SD). DPA = day post amputation; * = P < 0.05; **** = P < 0.0001. (E, F) Simple linear regression of bone volume and bone length regeneration for sham control and digit specific denervated digits (n = 14–22 digits/group).

3.3. Digit denervation delays wound healing leading to increased bone resorption

The delay in regeneration is not attributed to re-innervation of the digit since regenerates at 42 DPA contain sparse β3T⁺ axons and S100β⁺ Schwann cells (Fig. 3). β3T alone was used to confirm denervation as we have previously shown that it is expressed more prominently than other axonal markers in the digit tip (Dolan et al., 2019), and it is robustly regarded as a pan-axonal marker in uninjured and regenerating axons (Latremoliere et al., 2018; Moskowitz and Oblinger, 1995). Instead, the regenerative delay appears to be linked a nerve dependent regulation of bone resorption between 7 and 10 DPA (P < 0.0001; Fig. 2B) resulting in enhanced stump bone loss during blastema formation. Deregulation of the bone resorptive phase of regeneration increases in the volume of bone needed to complete the regeneration response. Bone resorption is mediated by osteoclasts at the amputation wound (Dolan et al., 2018; Fernando et al., 2011; Sammarco et al., 2014, 2015; Simkin et al., 2015b) and quantification of cathepsin K (CK)-immunopositive multinucleated bone resorbing osteoclasts at 6 DPA indicates that the initial recruitment of osteoclasts is not modified by denervation (Fig. 4A, B, E). However, at 12 DPA, when osteoclasts are depleted from the regenerate in sham control digits, osteoclasts are retained in denervated regenerates (Fig. 4C, D, E). This extends the bone resorptive phase and correlates with the enhanced level of bone resorption indicated by BV changes (Fig. 2B). Thus, innervation appears to be required for terminating osteoclast retention and important for the transition from catabolic to anabolic bone growth during the regeneration response. The data support the conclusion that the nerve dependent delay in regeneration is caused by conditions that enhance osteoclast retention, which leads to excessive stump bone degradation.

Fig. 3.

Axons and Schwann cells in digit specific denervated digits 6 weeks post-amputation

(A, B) Serial sections of β3T⁺ axons (green) in the BM of sham control digits (A-A″) and denervated digits (B–B″) at 42 DPA. (C) Quantification of β3T⁺ axons in the P3 BM of sham control and digit specific denervated digits at 42 DPA (n = 7–8 digits/group; mean ± SD; **** = P < 0.001). (D, E) S100β⁺ Schwann cells (green) are shown in the connective tissue of sham control (D) and digit specific denervated (E) digits at 42 DPA. (F) Quantification of S100β-immunopositive (S100β⁺) Schwann cells in the connective tissue of sham control and digit specific denervated digits at 42 DPA (n = 6–8 digits/group; mean ± SD). (A, B, D,E) White dashed line identifies the boundary of bone (b); Yellow dashed lines outline the dermis; Blue = DAPI stained nuclei; Scale bars = 100 μm.

Fig. 4.

Digit specific denervation delays wound healing

(A-D) Cathepsin K-immunopositive (CK⁺) osteoclasts (red) in the bone marrow (bm) of sham controls (A, C) and digit specific denervated digits (B, D) at 6- and 12-days post amputation (DPA). White dashed lines indicate bone (b) boundary. Blue = DAPI stained nuclei. Scale bars = 100 μm. (E) Quantification of CK⁺ multinucleated-osteoclasts at 6 and 12 DPA. (n = 6–8 digits/group; mean ± SD; ****P < 0.0001; *P < 0.05). (F) Mallory Trichrome staining showing wound closure is complete in all sham control digits by 10 DPA and a blastema (B) of undifferentiated cells is formed. (G) Mallory Trichrome staining showing incomplete wound closure in a denervated digit at 10DPA although a blastema (B) is present. Arrow indicates ejected bone. (H) Table showing the number of digits with wound closure in sham and denervated digits at 6, 8, 10, 12, and 14DPA.

The timing of wound closure correlates with the duration and extent of bone resorption in digit regeneration: delaying wound closure leads to excessive bone resorption, whereas accelerating wound closure results in reduced bone degradation (Sammarco et al., 2014, 2015; Simkin et al., 2015b). These studies support a model in which wound closure regulates oxygen availability which serves as a signal to terminate the bone degradation phase of blastema formation and transition to the anabolic bone growth phase (Simkin et al., 2015a). Because denervation is associated with enhanced osteoclast-mediated bone resorption, we hypothesized that denervation may be delaying wound closure. To test this hypothesis, denervated and sham control regenerates were analyzed at different stages of regeneration (6, 8, 10, 12, and 14 DPA) to determine the timing of wound closure. Wound closure occurred in 100% (8/8) of sham control regenerates by 10 DPA consistent with previous reports (Fernando et al., 2011; Simkin et al., 2015b) (Fig. 4F, H). In denervated regenerates, only 33% (n = 3/8) of amputation wounds were closed by 10 DPA and wound closure was not observed in the majority of denervated digits (n = 7/8) until 14 DPA (Fig. 4G and H). The data demonstrate that innervation is required for normal wound closure during digit tip regeneration.

3.4. Blastema formation is peripheral nerve independent

The blastema is a transient structure that distinguishes epimorphic regeneration from other regenerative responses and is characterized by a heterogeneous accumulation of proliferating cells (Fernando et al., 2011). In digit tip regeneration the blastema is distinct by 10 DPA and histological analysis of sham control and digit specific denervated regenerates both show an accumulation of a distinct blastemal cell population which is not dependent on the completion of wound closure (Fig. 4F and G). Sciatic denervation decreases proliferation of blastema cells and nail stem cells associated with regenerative inhibition (Johnston et al., 2016; Rinkevich et al., 2014; Takeo et al., 2013). Similarly, innervated limbs that are hindlimb suspended to reduce mechanical load display decreased blastema cell proliferation associated with regenerative inhibition (Dolan et al., 2022). Digit specific denervation effects on cell proliferation during blastema formation was examined by EdU incorporation at 10 and 12 DPA. Blastema cell proliferation was not inhibited by denervation at either time point and surprisingly, at 12 DPA when proliferation of sham control regenerates declines with re-differentiation, blastema cell proliferation is increased by denervation (P = 0.0179; Fig. 5A–C). Increased proliferation at 12 DPA correlates with the delay in the initiation of the re-differentiation of regenerated bone (see Fig. 2A). We next investigated cell proliferation within the nail matrix at 12 DPA and found no difference between sham control and digit specific denervated digits (P = 0.6795; Fig. 5D–F). Additionally, denervated digits express established blastema cell markers Runx2 and CXCR4 (Dawson et al., 2018; Lee et al., 2013), with many of Runx2⁺-osteoblasts are also EdU-positive in double immunostained preparations (Fig. 5J-M). These studies demonstrate that blastema formation and cell proliferation during blastema formation is not nerve dependent. Finally, sciatic denervation studies show that Schwann cells present in the blastema mediate the neurotrophic effect (Johnston et al., 2016) although quantitative studies demonstrate that Schwann cells are much reduced during digit tip regeneration (Dolan et al., 2019). Quantitative assessment of S100β⁺ Schwann cells based on immunostaining at 12 DPA shows: 1) Schwann cells are much reduced in both sham control and denervated blastemas compared to unamputated digits (compare Figs. 5I to 1F) and 2) there is no statistical difference between sham control and digit specific denervated blastemas (P = 0.5728; Fig. 5G–I). Taken together, these results support the conclusion that in the mouse, blastema formation is peripheral nerve independent.

Fig. 5.

Blastema formation is peripheral nerve independent

(A, B) EdU⁺ (pink) proliferating blastema cells in sham control (A) and digit specific denervated digits (B) at 12 DPA. (C) Quantification of EdU⁺ cells in the blastema at 10- and 12- DPA (n = 6–8 digits/group; mean ± SD; * = P < 0.05). (D, E) EdU⁺ (pink) proliferating cells in the nail matrix of sham control (D) and digit specific denervated digits (E) at 12 DPA. The solid white lines denotes the boundary between the nail (n) and nail matrix. The solid yellow lines denote the boundary between the nail matrix and ventral dermis. The white dashed line denotes the boundary between the P3 bone (b) and dermis. (F) Quantification of EdU⁺ cells in the nail matrix at 12 DPA (n = 5–6 digits/group; mean ± SD). (G, H) S100β + (green) Schwann cells in the blastema of sham control (D) and digit specific denervated digits (E) at 12 DPA. (I) Quantification of S100β + cells in the blastema at 12 DPA (n = 5–6 digits/group; mean ± SD). (J, K) Representative images of Runx2⁺ (green) osteoblasts and EdU⁺ (pink) proliferating cells in the blastema of sham control (J) and digit specific denervated digits (K) at 10 DPA. White arrowheads indicate cells co-labeled for Runx2 and EdU. (L, M) Representative images of CXCR4⁺ cells in the blastema of sham control (L) and digit specific denervated digits (M) at 10 DPA. (A, B, D, E, G, H, J, K, L, M) Blue = DAPI stained nuclei; Scale bars = 100 μm.

3.5. Dermabond partially rescues digit tip regeneration after digit specific denervation

Digit specific denervation delays wound closure causing increased bone resorption therefore leading to a longer regeneration time. We hypothesized that rescuing wound healing in denervated digits could prevent increased bone resorption and thereby enable digit tip regeneration comparable to sham control digits. To test this hypothesis, we applied Dermabond®, which has previously been shown to enhance wound healing after digit tip amputation (Simkin et al., 2015b), immediately after amputation, and monitored regeneration over the course of 4 weeks using μCT. At 28 DPA, sham control and digit specific denervated digits treated with Dermabond® appear morphologically similar (Fig. 6A, D, E). While sham and denervated digits treated with Dermabond® were not different (P = 0.5531) by BL at 28 DPA (Fig. 6C), denervated digits treated with Dermabond® did display a statistically significant reduction in BV (P = 0.0147; Fig. 6B). Taken together, these results suggest that enhancing wound closure with Dermabond® partially rescues digit tip regeneration after digit specific denervation.

Fig. 6.

Digit specific denervation is partially rescued by treating with Dermabond®

(A) Representative μCT renderings of a sham control (top) and a digit specific denervated (bottom) digit treated with Dermabond® during a regeneration period of 28 days. (B, C) Quantification of bone volume (B) and bone length (C) measurements for sham control and digit specific denervated digits treated with Dermabond® (n = 10 digits/group; mean ± SD). DPA = day post amputation; * = P < 0.05; *** = P < 0.001; **** = P < 0.0001. (D, E) Mallory Trichrome staining of Sham + Dermabond (D) and Denervated + Dermabond (E) digits at 28 DPA. Scale bars = 200 μm.

4. Discussion

There are few experimental models of epimorphic, or blastema-mediated, regeneration in mammals and the regenerating mouse digit tip, which parallels an established human response, is best characterized (Seifert and Muneoka, 2018). The role that nerves play in digit tip regeneration was first investigated by Neufeld and colleagues who established the experimental model of transection of the sciatic nerve to denervate the rat digit tip and reported that the regeneration response was delayed but not inhibited (Mohammad and Neufeld, 2000). Subsequent amputation studies on the mouse digit tip found that sciatic denervation slowed blastema cell proliferation and inhibited the regeneration response (Johnston et al., 2016; Takeo et al., 2013) or demonstrated gross morphological patterning defects (Rinkevich et al., 2014). While sciatic denervation effectively prevents innervation to the amputated digit tip, it also causes limb paralysis and a measurable reduction in mechanical load to the limb (Gross et al., 2010). Regeneration studies in which mechanical loading of the mouse limb was reduced by hindlimb suspension without modifying digit innervation, demonstrated that blastema formation was completely inhibited and the digit tip failed to regenerate (Dolan et al., 2022). Thus, the role that nerves play in regulating the regeneration response may be the result of direct re-innervation by axonal re-growth in combination with the indirect effect on mechanical loading caused by paralysis from sciatic denervation. To dissociate between these two possible effects, digit specific denervation was used to determine the regenerative outcome of locally removing innervation to the digit tip without influencing mechanical loading to the limb. Consistent with the earlier rat studies (Mohammad and Neufeld, 2000), we demonstrate that denervation of the digit tip in mice delays but does not inhibit the regeneration response. In combination with mechanical loading studies (Dolan et al., 2022), these findings provide strong evidence that epimorphic regeneration in mammals is not dependent on direct innervation by regrowing axons from the amputated stump in contrast to the role that innervation plays in poikilotherm regeneration models (Farkas and Monaghan, 2017; Pirotte et al., 2016). This conclusion is supported by studies showing that axonal regrowth following peripheral axon transection in mammals is severely impaired (Stankovic et al., 1996), and, as well, axonal regrowth associated with digit tip amputation is similarly impaired (Dolan et al., 2019). Thus, the endogenous regenerative response of mouse digit tips occurs with minimal involvement of regrowing axons from the stump.

Digit specific denervation results in two distinct modifications of the regeneration response: 1) enhancement of bone degradation, and 2) a delay in the completion of epidermal wound closure during blastema formation. Osteoclast-mediated bone degradation is known to be stimulated by sympathetic innervation (Togari et al., 2005) suggesting that enhanced degradation associated with denervation of the digit tip is not caused by inhibiting innervation. Alternatively, peripheral nerves are known to be actively involved in cutaneous wound healing and the absence of innervation results in a decreased rate of wound closure (Ashrafi et al., 2016; Barker et al., 2006). Recently, innervation of Lgr6 expressing epidermal stem cells was found to be responsible for the nerve dependent delay in cutaneous wound closure (Huang et al., 2021). Lgr6 was previously found to be expressed in nail stem cells of the digit and is required for digit tip regeneration (Lehoczky and Tabin, 2015). The role that innervation plays in wound healing is likely mediated by nerve-associated Schwann cells that remain following denervation, dedifferentiate and enhance the healing response by promoting wound contraction (Johnston et al., 2013; Parfejevs et al., 2018). Nerve-derived Schwann cells have also been shown to play a paracrine role in digit tip regeneration by stimulating blastema cell proliferation (Johnston et al., 2016). Denervation studies during rat digit tip regeneration was also found to be associated with a delay in wound closure (Mohammad and Neufeld, 2000). These studies support the conclusion that wound closure rate is regulated in part by re-growing axons and/or axon associated Schwann cells.

Schwann cells have been shown to dedifferentiate and participate in digit tip regeneration as a source of blastema cell specific mitogens (Johnston et al., 2016). The current study validates previous findings that Schwann cell survival is not nerve dependent however their participation in blastema formation is nerve dependent. There is an apparent discrepancy with respect to the participation of Schwann cells during blastema formation when nerves are present. Quantitative studies show that Schwann cells are rarely observed in the blastema proper (Dolan et al., 2019) so how can these cells play a significant role in the regeneration response? The digit blastema is a heterogeneous population of cells and can be partitioned into a central region of densely packed pre-osteogenic cells that differentiates into the replacement skeletal element and a peripheral population of loosely packed connective tissue cells contiguous with the stump dermis (Dawson et al., 2018). The dense central region is avascular, hypoxic, devoid of axons and Schwann cells, and is distinct from the peripheral connective tissues that is vascularized and contains non-myelinating Schwann cells (Dolan et al., 2019; Fernando et al., 2011; Sammarco et al., 2014; Yu et al., 2014). Thus, while the central blastema does not contain Schwann cells, the surrounding connective tissue can function as a paracrine source of growth mediators for the central blastema. Recent single cell RNAseq analyses of digit tip regeneration identified Schwann cells in relatively low numbers by comparison to other cell types identified and while this technology is not quantitative for all cell types, the data are consistent with minimal participation to the regenerating blastema (Johnson et al., 2020; Storer et al., 2020). Nevertheless, the current studies show that blastema formation and regeneration does occur following digit specific denervation with minimal involvement of re-growing axons and Schwann cells.

In mice, monocyte-derived osteoclasts are recruited to the amputated stump where they actively degrade the stump bone during blastema formation (Fernando et al., 2011; Simkin et al., 2017b). Regeneration studies of fish, salamanders and mice have shown that monocyte depletion is inhibitory for epimorphic regeneration demonstrating the importance of inflammation for initiating the regeneration (Godwin et al., 2013; Petrie et al., 2014; Simkin et al., 2017a, 2017b). Quantitative microCT analysis of digit specific denervated regenerates indicates that the bone degradation phase is extended compared to sham controls and this causes excessive stump bone degradation. Quantitation of osteoclasts in denervated regenerates demonstrate normal osteoclast recruitment but deregulation of osteoclast termination resulting in excessive bone degradation. A similar extension of the bone degradation phase is observed during digit regeneration of aged mice (Brunauer et al., 2021) and in young mice treated with hyperbaric oxygen (Sammarco et al., 2015). In all of these studies, the deregulation of osteoclast termination is associated with a delay in the completion of wound closure and a temporal shift in the transition of the blastema from catabolic to anabolic bone growth (Muneoka, 2017). This catabolic to anabolic transition is intricately linked to dynamic changes in oxygen tension: blastema avascularity (Fernando et al., 2011; Yu et al., 2014) coupled with the completion of wound closure initiates a transient hypoxic state that is later resolved with enhanced angiogenesis associated with anabolic bone growth (reviewed in (Simkin et al., 2015a)). Dermabond® treatment of the amputation wound induces precocious hypoxia, stimulates wound closure, and reduces osteoclast-mediated bone degradation (Simkin et al., 2015b), whereas hyperbaric oxygen treatment inhibits blastema hypoxia, delays wound closure, and increases osteoclast-mediated bone degradation (Sammarco et al., 2015). These studies point to dynamic changes in oxygen tension that is controlled in part by wound closure as playing a key role in terminating the bone degradation phase of regeneration, and studies showing that osteoclast termination is directly regulated by hypoxia support this conclusion (Wu et al., 2015). We conclude that the rate of cutaneous wound closure influences the overall rate of skeletal regeneration by shifting the timing of the catabolic to anabolic transition in bone regrowth during regeneration. Since the rate of new bone formation in denervated versus sham control digits is not different, excessive bone degradation results in an increase in the time required to complete the regeneration response.

Nerve dependent regeneration was first documented in 1823 (Todd, 1823) and, while several vertebrate models require nerves for successful regeneration, the regenerating salamander limb is the best characterized (Tanaka, 2016). Denervation is accomplished by nerve transection, and this causes paralysis in conjunction with axon depletion, but most denervation studies fail to distinguish between a direct role of innervation from an indirect role of innervation dependent mechanical loading. The clearest example of direct innervation dependent effects on regeneration involves the induction of limb regeneration by deviation of a transected nerve to a non-regenerative skin wound (Endo et al., 2004; Satoh et al., 2007). Since such regenerates are ectopic to the limb skeleton, they are not subjected to mechanical loading, suggesting that mechanical load is not required for salamander limb regeneration. This conclusion is supported by documentation of successful regeneration in salamanders that were flown in space (Mitashov et al., 1996). Alternatively, epimorphic regeneration of the fish fin has been shown to be both nerve dependent (Geraudie and Singer, 1985; Simoes et al., 2014) and mechanical load dependent (Recidoro et al., 2014). From an evolutionary perspective it is curious that regeneration in fish is both innervation and load dependent, regeneration in salamanders is innervation dependent and load independent, and regeneration in mammals is innervation independent and load dependent. In salamanders, there is evidence that innervation dependent limb regeneration is a response that evolved specifically among the Salamandroidea suggesting that specific regeneration-specific signaling pathways have evolved in some phyla but are absent among other vertebrate groups (Garza-Garcia et al., 2009, 2010). On the other hand, the regenerating mouse digit tip displays characteristics suggestive of a gain of function response that evolved from modification of the fracture healing response (Muneoka and Dawson, 2021), and this conclusion is supported by studies showing that fracture healing is both nerve independent (Miura et al., 2015) and mechanical load dependent (Betts and Muller, 2014). These data support a view in which regeneration strategies among vertebrates involve evolutionary gain of function (e.g. salamander limb regeneration, mouse digit tip regeneration) in addition to the classical view of regenerative loss (Muneoka and Dawson, 2021).

Data availability

All data is available in the main text or the supplementary materials.