The reinnervation and revascularisation pattern of scarless murine fetal wounds


Prof Mark Ferguson, Renovo Ltd, The Manchester Incubator Building, 48 Grafton Street, Manchester M13 9XX, UK. T: + 44 (0)161 276 7121; F: + 44 (0)161 276 7240;


Fetal wounds can heal without scarring. There is evidence that the sensory nervous system plays a role in mediating inflammation and healing, and that the reinnervation pattern of adult wounds differs from that of unwounded skin. Ectoderm is required for development of the cutaneous nerve plexus in early gestation. It was hypothesised that scarless fetal wounds might completely regenerate their neural and vascular architecture. Wounds were made on mouse fetuses at embryonic day 16.5 of a 19.5-day gestation, which healed without visible scars. Immunohistochemical analysis of wound sites was performed to assess reinnervation, using antibodies to the pan neuronal marker PGP9.5 as well as to the neuropeptides calcitonin gene-related peptide (CGRP) and substance P (SP). Staining for the endothelial marker von Willebrand factor (VWF) allowed comparison of reinnervation and revascularisation. Wounds were harvested at timepoints from day 1 after wounding to postnatal day 6. Quantification of wound reinnervation and revascularisation was performed for timepoints up to 6 days post-wounding. Hypervascularisation of the wounds occurred within 24 h, and blood vessel density within the wounds remained significantly elevated until postnatal day 2 (4 days post- wounding), after which VWF immunoreactivity was similar between wound and control groups. Wound nerve density returned to a level similar to that of unwounded skin within 48 h of wounding, and PGP9.5 immunoreactive nerve fibre density remained similar to control skin thereafter. CGRP and SP immunoreactivity followed a similar pattern to that of PGP9.5, although wound levels did not return to those of control skin until postnatal day 1. Scarless fetal wounds appeared to regenerate their nerve and blood vessel microanatomy perfectly after a period of hypervascularisation.


Cutaneous wounds made on fetuses before the third trimester are reported to heal without scarring (Ferguson & O’Kane, 2004). The age of gestation at which scarless healing occurs varies, even between individuals, and the extent of injury is important in determining whether a wound will scar. Even in the adult, a pinprick usually heals with no scar (Ferguson et al. 1996). Investigation into fetal scar-free healing has led to the development of therapeutic scar-reducing strategies for adult wounds (Occleston et al. 2008), but there has been no investigation of the reinnervation pattern of scar-free fetal wounds, or adult wounds treated to reduce scarring. The MRL/MpJ mouse heals ear punch wounds without scarring, and nerve regeneration in these wounds preceded vascularisation, in contradistinction to dorsal skin wounds in the same animal, which heal with a scar (Buckley et al. 2011). Scarring adult wounds are abnormally reinnervated and hypervascular, so it is important to clarify the reinnervation and revascularisation pattern of these scar-free wounds as this may lead to future strategies for improving wound reinnervation, as well as a deeper understanding of the nature of scar-free healing.

Fetal wounds show less early inflammation compared to adult wounds. The fetus is thought to be significantly neutropenic and may not have developed self–nonself immunological identity (Longaker et al. 1990). Fewer polymorphonuclear leucocytes, macrophages and lymphocytes migrate into fetal wounds, although these cells are capable of responding to inflammatory signals in the same manner as in adult wounds (Cowin et al. 1998).

Experimental reduction of transforming growth factor beta (TGF-β)1 levels in adult wounds using a neutralising antibody markedly improves scarring (Shah et al. 1994). By contrast, fetal wounds contain high levels of TGF-β3 (Ferguson & O’Kane, 2004) and addition of human recombinant TGF-β3 reduces scarring in experimental animals (Shah et al. 1995) and man (Ferguson et al. 2009; Bush et al. 2010).

Cutaneous nerve fibres are seen in dorsal mouse skin at embryonic day 15 (e15) of a 19.5-day gestation, and by day e16, some fibres appear to be associated with developing hair follicles (Peters et al. 2002). These may be the precursor of the follicular neural network seen in adult skin. At e18, some nerves were seen in the epidermis. Calcitonin gene-related peptide (CGRP) and substance P (SP) were detected only at postnatal day 1 (p1) in subcutaneous and dermal nerve fibres of the dorsal skin, although SP immunoreactivity has been found in mouse cranial nerve nuclei at day e13, and in facial skin and mucosa at days e16–17 (Mohamed & Atkinson, 1982). A similar pattern of feather innervation was seen in chicks (Saxod et al. 1996). Ablation of chick ectoderm at embryonic day 4 prevented cutaneous nerve plexus formation. It is suggested that embryonic skin may trigger divergence of nerve branches and plexus development by secretion of trophic factors (Lumsden & Davies, 1986; Martin et al. 1989).

Human fetal cutaneous innervation follows a similar sequence of events to that found in mice; nerve plexuses were detected with antibodies to PGP 9.5 from 10 weeks whilst the unequivocal presence of CGRP and small amounts of SP were not detected until 17 weeks (Terenghi et al. 1993).

Adult murine wounds become reinnervated (Rajan et al. 2003) and hypervascular (Henderson et al. 2006) during the healing process, and the pattern of cutaneous innervation of adult wounds is altered (Zhang & Laato, 2001; Liang et al. 2004; Henderson et al. 2006). Neonatal wounds have been found to be hyperinnervated by capsaicin-sensitive (C and Aδ) nerve fibres (Reynolds & Fitzgerald, 1995), but the reinnervation and revascularisation pattern of nonscarring fetal wounds are unknown.

We hypothesised that fetal wounds healing without scarring would completely regenerate their cutaneous nerve and vascular plexuses, possibly after transient hyperinnervation and hypervascularisation. We aimed to assess quantitatively wound revascularisation and reinnervation to test this hypothesis.

Materials and methods

All procedures were performed under Home Office licence and in accordance with the UK Animal Act (1986). Fetal wounds were performed using an operating microscope and microsurgical instruments. A pregnant female CD1 mouse of 16.5 days’ gestation (the day of finding a vaginal plug being taken as day 0) was anaesthetised with oxygen, nitrous oxide and isofluorane. The mouse was placed supine, and the lower abdomen shaved and cleaned with 70% ethanol. A sterile drape was used and the procedure was performed using aseptic techniques. A low midline laparotomy was performed. The skin and abdominal wall musculature were divided. Fetuses were identified through the translucent wall of the elongated bicornate uterus and were orientated to place an incision over the body wall of the fetus. A uterotomy incision allowed the fetal body wall (flank) to be wounded with a small cup-shaped sharp forceps (Aesculap). Wound sites were marked to allow wounds to be identified postnatally with two 9/0 sutures. Control skin was harvested from the contralateral flank of each animal at the time of wound harvest.

The gestation period of the CD1 mice was a consistent 19.5 days. Fetal wounds were harvested at embryonic days e17.5 and e18.5, as well as postnatal days p1, p2, p3 and p6. Day p1 is the day of birth, equivalent to e19.5. Six embryos were wounded for each timepoint (n = 36 in total), but not all were available for analysis, due to loss of some marker sutures and maternal cannibalism. Table 1 shows the final numbers of wounds available for analysis. Data from the second postnatal day and the sixth postnatal day, where the numbers of wounds analysed were small (one or two tissue samples in most cases) should be regarded as provisional findings. The data are included because they are consistent with the findings from other timepoints, and add to the overall picture. For each wound, three sections were analysed.

Table 1.   Numbers of wounds analysed at each timepoint after fetal wounding. Three sections from each wound were analysed.
e19.5 = p166666665

Tissue processing

The harvested tissue from all experiments was immediately placed into cold Zamboni’s solution and stored at 4 °C for 24 h before being transferred to 15% sucrose in phosphate-buffered saline (PBS), which was changed daily until the tissue was saturated. Whole tissue samples were frozen in optimum cutting temperature (OCT) embedding matrix (Cellpath, Powys, UK) over liquid nitrogen. Samples were then stored at −80 °C until analysis. A cryostat (CM3050; Leica, Nussloch, Germany) was used to cut 14-μm sections, which were collected sequentially onto slides that had been coated with poly-l-lysine. Toluidine blue stain (Sigma) along with the marker sutures was used to confirm wound location in the sections. The slides were dried overnight at 37 °C before immunohistochemical or simple staining.


Sections were permeabilised in 0.2% Triton detergent for 60 min, followed by washes (2 × 3 min) in PBS at pH 7.4. To decrease background autofluorescence, the sections were placed in a PBS solution containing 10% pontamine sky blue (BDH; Poole, Dorset, UK) and 10% dimethylsulphoxide (DMSO). After two more washes in PBS, sections were incubated with primary antibodies for 20 h at 4 °C in PBS with 1% sodium azide preservative and 5% goat serum as a blocking agent. Primary antibodies were rabbit anti-human protein gene product 9.5 (PGP9.5) (diluted 1 : 500; Affiniti, Exeter, UK), which stains all nerve tissue; rabbit anti-human von Willebrand factor (VWF) (diluted 1 : 2000; Abcam, Cambridge, UK), or rabbit anti-rat calcitonin gene-related peptide (CGRP) (diluted 1 : 3000; Affiniti) or rabbit anti-cow Substance P (SP) antibodies (diluted 1 : 5000; Affiniti).

Following more washes in PBS (2 × 6 min) the sections were incubated with a fluorescein conjugated polyclonal goat anti-rabbit secondary antibody (diluted 1 : 100; Vector Laboratories, Burlingame, CA, USA) at room temperature for 1 h. After final washes in PBS (2 × 6 min) sections were mounted with Vectashield™ (Vector). The slides were stored in the dark at 4 °C to avoid fading of fluorescence, and analysed within 48 h.


A Leica DMRB microscope was used to view the images under fluorescent light. Images were captured at 20× magnification using a digital camera (Diagnostic Instruments, Sterling Heights, MI, USA) from three adjacent sections of each wound. Images were analysed using an automated method of quantifying the area of positive staining in each field of view (Image Pro-Plus; Media Cybernetics, Silver Spring, MD, USA).

Masson’s trichrome staining was carried out on separate sections of all wounds, allowing wound architecture to be compared with wound area measurements and immunohistochemical findings.

Statistical analysis was performed between timepoints using analysis of variance, and between wounded and unwounded tissue at the same timepoint with a non-parametric t-test, assuming unequal variance between groups.


Fetal wounds healed extremely well, and by the day of birth (e19.5, also called p1) 3 days’ post-wounding, the wounds were invisible to the naked eye except for some residual erythema. Masson’s stained sections of the wounds showed how the histological architecture of the healing wounds changed over time, becoming virtually indistinguishable by postnatal day 6, by which time the wound has become almost impossible to distinguish from the surrounding skin. Although an increase in cellularity is apparent, it is difficult to be certain of the exact wound margins (Fig. 1).

Figure 1.

 Masson’s stained section through a wound at embryonic days 17.5 (A), postnatal day 1 (B), postnatal day 3 (C), and postnatal day 6 (D). The scale bar is 100 μm in each, and the red arrows show the edges of the wound. By postnatal day 6, the wound has become almost impossible to distinguish from the surrounding skin. Although an increase in cellularity is apparent, it is difficult to be certain of the exact wound margins.

Not all of the wounds created could be used for analysis, either because the marker sutures had come off or been removed by the mother, or due to maternal cannibalism. Without marker sutures being present, it was not possible reliably to identify the wound sites in cut sections. The result of this problem is that fewer wounds were available for analysis, particularly at later timepoints. The number of wounds and control skin specimens finally used for analysis are shown in Table 1. Some of the timepoints are only represented by one wound, in which case analysis was based on three sections through the one available wound.

Unwounded fetal skin showed a changing pattern of innervation as the fetuses matured. PGP9.5 immunostaining was present from day e16.5, showing the presence of a developing nerve plexus (Fig. 2A) although SP was only present in very small amounts, and CGRP was not detected until e17.5 and only in small amounts until day p1 (Fig. 3). VWF immunoreactivity demonstrated the presence of a vascular plexus from day e16.5 (Fig. 4A).

Figure 2.

 Immunostaining for the pan neuronal marker PGP9.5 in green. (A) Unwounded fetal skin from day e16.5. A cutaneous nervous plexus is present (white arrows). SP and CGRP immunoreactivity were not present at this stage of development. (B) A fetal wound at postnatal day 1, 3 days after wounding. PGP9.5 immunostaining shows nerve fibres in green (white arrow). In this wound they appear to be regenerating from the adjacent skin. The blue arrows indicate the wound edges. Scale bar: 100 μm.

Figure 3.

 Unwounded mouse skin from the day of birth (e19.5, which is the same as p1) showing immunostaining for CGRP (White arrows) that was only detected in very small amounts before this time. Scale bar: 100 μm.

Figure 4.

 Immunostaining for the vascular marker von Willebrand factor (VWF). (A) Unwounded fetal skin at day e16.5. A vascular plexus is seen (white arrows). (B) VWF immunostaining (green) in a fetal wound at postnatal day 1 (white arrows). The epithelium is counterstained red. The blue arrows show the wound margins, one of which is right at the edge of the field of view. Vessels appear to be regenerating from the wound edge at the right-hand side. Scale bar: 200 μm.

Wound reinnervation during healing appeared to be by both collateral sprouting from intact nerves in the base of the wound and by regeneration of divided axons at the wound peripheries (Fig. 2B) (classified by Griffin et al. 2010). Revascularisation was also seen to occur from the wound edge (Fig. 4B).

A reduction in cutaneous nerve fibre density was seen in unwounded fetal skin at day e18.5 (P < 0.05 compared to e17.5) (Fig. 5). The overall reinnervation pattern demonstrated by PGP9.5 immunofluorescence showed that reinnervation of the wounds made at day 16.5 occurred over 2 days (Figs 2 and 5). Although the innervation density of wounds 1 day after wounding (day e17.5) was significantly less than in the control skin, nerve fibre density was similar in control and wound groups by day e18.5, and remained so for the duration of the study period.

Figure 5.

 Histogram showing mean ± SEM reinnervation density of fetal wounds and control skin as indicated by PGP9.5 immunostaining at times after wounding. Wounds were made at day e16.5. A significant reduction in innervation density in control skin was seen between days e17.5 and e18.5 (*P < 0.05). Nerve density was significantly less in wounds 1 day after wounding (e17.5), but at all other timepoints, the density of nerve fibres was similar between wounds and control skin. Nerve density was measured as percentage immunofluorescence per high-power field after artefacts were excluded.

The reinnervation of the wound by nerve fibres immunopositive for CGRP occurred over 4 days after wounding. The levels of CGRP in wounds were significantly lower than in unwounded skin at days e17.5 and e18.5. By day p1, CGRP levels in healing wounds were lower than in controls (P = 0.12) and there was no significant difference between wounded and unwounded skin CGRP density thereafter (Fig. 6).

Figure 6.

 Histogram showing mean ± SEM reinnervation density of fetal wounds and control skin as indicated by CGRP immunostaining at times after wounding. Wounds were made at day e16.5, at which time no CGRP was detected in the fetal skin. A reduction in innervation density in control skin was seen between days p1 and p2. The CGRP innervation density of wounds was significantly lower than that of control skin for the first 2 days after wound creation; thereafter, CGRP density between control and wounded skin was similar. Nerve fibre density was measured as percentage immunofluorescence per high-power field after artefacts were excluded.

Very little substance P could be detected in wounds until day p1, at which point the levels were higher than in control skin, although not significantly so. At all subsequent timepoints, there was no difference between the density of control skin and wound SP innervation (Fig. 7).

Figure 7.

 Histogram showing mean ± SEM reinnervation density of fetal wounds and control skin as indicated by SP immunostaining at times after wounding. Wounds were made at day e16.5. A significant reduction in innervation density in control skin was seen between day e17.5 and day e18.5 (*P < 0.05. Very little SP was detected in the fetal skin until the first postnatal day. SP levels were decreased in wounds on the first day. Otherwise, there was no difference in the density of SP in the wounds and control tissue at any timepoint. Nerve fibre density was measured as percentage immunofluorescence per high-power field after artefacts were excluded.

Fetal wounds showed dramatic hypervascularisation in the 3 days after wounding (Figs 4 and 8), although the density of VWF staining returned to levels similar to those in control wounds by day p2 (4 days after wounding) and remained so thereafter. Revascularisation preceded reinnervation.

Figure 8.

 Histogram showing mean ± SEM for fetal wound revascularisation and control skin as indicated by density of VWF immunoreactivity at times after wounding. Wounds were made at day e16.5. The hypervascularisation of wounds seen at days e17.5, e18.5 and p1 is statistically significant (*P < 0.05). VWF density was measured as percentage immunofluorescence per high-power field after artefacts were excluded.


Murine fetal wounds made at e16.5 healed without scarring and also appeared to regenerate completely their cutaneous neurovascular structures along with the rest of the cutaneous architecture by postnatal day 2. Wounds were hypervascularised during the healing process, but no hyperinnervation occurred.


We found that unwounded fetal skin contains nerve fibres from e16.5, CGRP from day e17.5 and SP from day p1, the same sequence and similar time course as found by others (Peters et al. 2002). The levels of SP and CGRP immunoreactivity are low in comparison with that of PGP9.5. Wounds made at e16.5 on CD1 mouse fetuses were found to regenerate their cutaneous nerve and vascular plexuses after a period of hypervascularisation.

Given that nerve fibres were only just becoming detectable in the murine skin at the time of wounding in these experiments, it might be interesting to wound fetuses at earlier times in gestation to see whether there is a time before which damage to the surface prevents innervation. Developing skin provides trophic factors for its own innervation, and removal of ectoderm from chick hindlimb prevents normal cutaneous innervation of that limb (Honig et al. 2004). It was suggested that embryonic skin may trigger divergence of nerve branches and plexus development by secretion of trophic factors (Martin et al. 1989), and so the effects of fetal wounding on wound reinnervation may be dependent on the timing and nature of the injury inflicted.

Fetal skin reinnervation after wounding differed from adult wound reinnervation. Although adult skin nerve density was not elevated overall (Rajan et al. 2003), SP and GGRP levels were found to be elevated during the healing process and SP levels remained elevated (Henderson et al. 2006). Initial adult wound hyperinnervation followed by a return in nerve density to that of unwounded skin has been found in guinea pig burn wounds (Kishimoto, 1984) and superficial wounds in the rat (Aldskogius et al. 1987). This study found no hyperinnervation during or after healing of scarless fetal wounds. Nerves have been identified growing into adult wounds either from the adjacent skin edges (regeneration) or as branches from intact deeper nerves (collateral sprouting; Rajan et al. 2003). Our results suggest that wound reinnervation in this model also appeared to be by mechanisms of both collateral sprouting from intact nerves in the base of the wound as well as by regeneration of divided axons at the wound peripheries (Fig. 2D). However, the fetal model differs from those of adult murine nerve regeneration (Griffin et al. 2010) because the nerve plexuses are also developing de novo in the fetal skin at and around the time of experimental wounding. This may make the distinction between repair by regeneration of injured axons or collateral sprouting from uninjured deep nerves difficult to differentiate from axonal growth and sprouting that would occur in an uninjured fetus.

Cytokines acting in wound healing have actions on nerve growth and regeneration, particularly nerve growth factor (NGF). NGF is produced by keratinocytes (Yaar et al. 1991), possibly more so after they have been injured (Taherzadeh et al. 2003). Keratinocytes cause hyperexcitability and chronic pain when interacting with peripheral nerves in an injury model (Radtke et al. 2010). NGF mRNA and then protein are increased after cutaneous injury in adult rats. NGF preferentially stimulates the growth of sensory neurons that express CGRP and SP (Terenghi, 1999; Micera et al. 2003), leading to the selective survival of C and Aδ fibres (Hari et al. 2004), which mediate pain, temperature and pruritis. NGF is produced by keratinocytes (Yaar et al. 1991), Mast cells (Artuc et al. 1999) and by injured tissue (Hasan et al. 2000; Cruise et al. 2004) and denervated skin (Diamond et al. 1992). Dorsal root ganglia in vitro have been found to grow preferentially towards injured rather than uninjured skin (Taherzadeh et al. 2003).

It would be interesting to compare NGF mRNA and peptide levels in adult and fetal wounds to see if these correlate with reinnervation patterns.

We found transient marked hypervascularisation of the mouse fetal scarless wounds. This is consistent with the finding of a two-fold increase in vessel counts in scar-free rat fetal wounds (Colwell et al. 2005), which was associated with an increase in vascular endothelial growth factor (VEGF) expression. Our finding that revascularisation preceded reinnervation is different to the scarless healing seen following punch wounds to the MRL/MpJ mouse ear, in which nerve regeneration preceded vascularisation (Buckley et al. 2011). Increased vessel counts were found in scarring fetal wounds, but without the same transient increase in VEGF (Colwell et al. 2005). VEGF is expressed in normal human fetal endothelial cells and is likely to have a role in normal developmental differentiation and angiogenesis, which in the context of fetal injury is part of the mechanism by which tissue regeneration occurs instead of scarring (Peters et al. 1993). It would be of interest to quantify VEGF.

Human adult scars appear to have variable innervation patterns reflecting the wide range of sensory symptoms experienced by patients. Electron microscopic examination of human punch biopsy scars found small unmyelinated fibres growing into the neodermis and epidermis of the scar but almost no myelinated epidermal fibres (Mihara, 1984). Hypertrophic human scars have been found to contain greater levels of SP, CGRP and NPY compared with normal skin, although normotrophic scars were not found to contain any neuropeptides (Crowe et al. 1994). Only SP and CGRP immunopositive fibres were found to penetrate into painful human hypertrophic scars and it was suggested that SP antagonism might reduce scar hypertrophy (Parkhouse et al. 1992). Normotrophic scars in humans were found to contain less SP, CGRP, vasoactive intestinal polypeptide (VIP) and neuropeptide Y (NPY) than control skin at 7 months (Altun et al. 2001).

Clinical evaluation of keloid scars found that 86% were pruritic and 46% painful, suggesting abnormalities of small fibre innervation within keloid scars (Lee et al. 2004). Limited immunohistochemical studies of keloid innervation have confirmed abnormal nerve morphology within the scar (Zhang & Laato, 2001).

Our findings that fetal wounds that heal without scarring also appear to regenerate their cutaneous innervation is reassuring for the development of scar-reducing treatments for human use. It might be extrapolated that scar reduction would encourage more normal cutaneous reinnervation than the limited unmyelinated pattern observed in adult wounds, although clearly this will need to be investigated.


We are grateful to Kelly Middleton for tuition in the fetal wounding technique, and to Renovo Ltd and The Royal College of Surgeons of Edinburgh for financial assistance.

Authors’ contributions

The experiment was conceived by all three authors. J.H. carried out the experimental work, which was supervised by G.T. and M.W.J.F. All authors were involved in writing the paper.