The design of successful regenerative medicine treatments requires leveraging the key factors that have pivotal roles in tissue healing. Modulating the immune system to promote tissue healing has demonstrated to be remarkably effective, since immune components coordinate a complex series of events that are critical for proper healing1,2. Unsurprisingly, the immune dysregulation that occurs as a consequence of ageing or conditions such as diabetes has a substantial negative effect on tissue healing and is often characterized by persistent inflammation at the injured site2,3. Another potentially important factor in tissue repair and regeneration is the nervous system. For instance, studies using neurotomy models have shown that peripheral nerves have an essential role in some animal species that are capable of regenerating tissues13,14,15,16. Other studies using nerve depletion in mouse, which are more specific to sensory neurons, have shown that nociceptive neurons and non-peptidergic C-fibres that express Gαi-interacting protein promote adipose tissue regeneration17 and skin repair after UV-induced damage18, respectively. Nociceptive sensory neurons (also referred to as nociceptors) are specialized primary sensory neurons that have nerve endings in tissues such as skin, muscles and joints. They detect and respond to noxious stimuli including temperature, chemicals and inflammatory mediators19. Nociceptors have been shown to be critical immunoregulators with both protective and harmful effects4,20,21. For example, nociceptors either diminish or exacerbate inflammation, enhance resistance to pathogens, or impede their clearance5,6,7,8,9,10,11,12. Therefore, given the immunomodulatory properties of nociceptors and the central role of the immune system in tissue repair and regeneration, we investigated the importance of peptidergic sensory neurons in tissue healing following acute injury, and examined whether neuro–immune interactions could be harnessed to promote tissue healing. Specifically, we used skin and muscle as relevant tissue models that are known to be innervated by nociceptors and in which repair and regeneration outcomes are considerably modulated by the immune system1,2,3,22,23.

Tissue healing in the absence of sensory neurons

To determine the importance of sensory neurons during tissue healing after acute injury, we used the Nav1.8cre/Rosa26DTA mouse. In this mouse, NaV1.8+ dorsal root ganglion (DRG) neurons, which mainly represent nociceptors involved in mechanical, cold and inflammatory pain, are ablated by the expression of diphtheria toxin fragment A5,7,8,10,12,24 (DTA). Rosa26DTA littermates with intact NaV1.8-expressing sensory neurons were used as controls. As acute injury models, we selected full-thickness wounds in the dorsal skin25,26 and volumetric muscle loss injuries in quadriceps23. The absence of NaV1.8-expressing sensory neurons resulted in a significant delay in skin wound closure, which was evident through decreased epithelial migration, a reduced number of proliferative keratinocytes and wounds remaining largely open after six days (Fig. 1a,b and Extended Data Fig. 1a–g). Similarly, muscle regeneration was impaired in Nav1.8cre/Rosa26DTA mice, which were characterized by increased fibrotic tissue and reduced muscle tissue formation (Fig. 1c,d and Extended Data Fig. 1h).

Fig. 1: NaV1.8+ nociceptors that express CGRP mediate tissue healing via myeloid cells.
figure 1

a,b, Full-thickness skin wounds were created in denervated (Nav1.8cre/Rosa26DTA) mice and littermate controls (Rosa26DTA). a, Wound closure evaluated by histomorphometric analysis on day 6 and day 10 post-injury (Rosa26DTA day 6, n = 14; Nav1.8cre/Rosa26DTA day 6, n = 12; day 10, n = 8). b, Representative histology at day 6 post-injury. Black arrows indicate wound edges and red arrows indicate tips of epithelium tongue. Scale bar, 1 mm. c,d, Volumetric muscle loss was evaluated on quadriceps of Nav1.8cre/Rosa26DTA and Rosa26DTA littermate controls. c, Extent of muscle regeneration was evaluated by histomorphometric analysis on day 8 and day 12 post-injury (day 8, n = 7; day 12, n = 10). d, Representative histology at day 12 post-injury (fibrotic tissue is stained blue; muscle tissue is stained red). Scale bar, 500 μm. e, Distribution of NaV1.8+ sensory neurons (red) in skin and muscle of Nav1.8cre/Rosa26tdT mice before and after injury. White lines indicate keratinocyte layers and nuclei are in blue. GT, granulation tissue area. Scale bars, 500 μm. The experiment was repeated independently four times. f, Expression of CGRP in skin and muscle before and after injury detected by immunohistochemistry. Scale bars: 500 μm (skin), 100 μm (muscle). The experiment was repeated independently four times. gj, Ramp1 deletion in myeloid cells using LysMcre+/−/Ramp1fl/fl mice. LysMcre+/− mice were used as controls. g, Wound closure was quantified on day 6 post-injury (n = 10). h, Representative skin histology. i, Muscle regeneration was quantified on day 12 post-injury (n = 10). j, Representative muscle histology. Boxes show median (centre line) and interquartile range (edges), whiskers show the range of values and dots represent individual data points. a,c, Two-way ANOVA with Bonferroni post hoc test for pairwise comparisons. g,i, Two-tailed Student’s t-test. P values are indicated.

Source Data

Nociceptor terminals release neuropeptides in response to danger signals that include inflammatory cytokines commonly present in tissues after acute injury27,28. Thus, we investigated the distribution of NaV1.8+ sensory neurons during skin and muscle healing following acute injury. To visualize NaV1.8+ sensory neuron distribution and the neuropeptides that they express, we used Nav1.8cre/Rosa26tdT mice where tdTomato fluorescent protein expression is restricted to NaV1.8+ sensory neurons. Tissue sections of Nav1.8cre/Rosa26tdT uninjured skin and muscle showed NaV1.8+ sensory neuron distribution across the epidermis and dermis in skin and nearby connective tissues in muscle (Fig. 1e). Following skin and muscle injury, we observed that NaV1.8+ sensory neuron endings growing in clusters into the granulation tissue, thus establishing innervation within the injured area during the healing process (Fig. 1e and Extended Data Fig. 2a). NaV1.8+ neurons exhibited expression of calcitonin gene-related peptide (CGRP) and substance P, and no detectable expression of vasoactive intestinal peptide (VIP) or galanin (GAL) (Fig. 1f and Extended Data Fig. 2b–d). Similarly, in the DRG following both skin and muscle injuries, these neurons showed expression of CGRP and substance P in cell bodies (Extended Data Fig. 2e,f). Additionally, the absence of CGRP signal in Nav1.8cre/Rosa26DTA mice pre- and post-tissue injury underscores the role of NaV1.8+ nociceptors as primary sources of CGRP during skin and muscle healing and suggests that nociceptor-derived CGRP has a pivotal role in these processes (Extended Data Fig. 2g,h).

CGRP effect on injury immune cells

To investigate whether CGRP mediates neuro–immune interactions that drive tissue healing, we generated mice in which immune cells or non-immune cells were unable to respond to CGRP. Bone marrow cells deficient for RAMP129, a co-receptor that is essential for CGRP signalling, were transplanted into γ-irradiated wild-type or Ramp1−/− mice. Wild-type mice that received wild-type bone marrow cells or Ramp1−/− mice that received Ramp1−/− bone marrow cells were used as controls. Skin and muscle healing were severely impaired in wild-type mice reconstituted with Ramp1−/− cells compared with those with wild-type cells. Similarly, transfer of Ramp1−/− cells into Ramp1−/− mice impaired healing, whereas wild-type cells rescued skin repair and muscle regeneration (Extended Data Fig. 3a–d). Considering that myeloid cells such as neutrophils and monocytes/macrophages constitute the majority of immune cells in injured tissues undergoing repair or regeneration and can represent up to 50% of the total wound cells1,2,23,26, we hypothesized that CGRP promotes tissue healing by modulating myeloid cells. To investigate this, we produced mice that lacked CGRP signalling in myeloid cells by crossing LysMcre (LyzM is also known as Lyz2) mice with Ramp1 floxed (Ramp1fl/fl) mice30. The LysMcre+//Ramp1fl/fl mice displayed a significant reduction in skin wound closure and muscle regeneration, compared with LysMcre+/− control mice (Fig. 1g–j). The extent of healing impairment, characterized by a significant delay in skin wound closure and compromised muscle regeneration with high level of fibrotic area, was indeed very similar to that observed in Nav1.8cre/Rosa26tdT mice. Together, these results suggested that CGRP from NaV1.8+ sensory neurons promotes tissue healing via myeloid cells. In addition, CGRP treatment in vitro had no significant effect on the proliferation of key cell types involved in skin and muscle healing, including fibroblasts, keratinocytes, myoblasts and endothelial cells (Extended Data Fig. 3e). Further supporting this observation, we found that keratinocytes and myoblasts have low expression levels of either Ramp1 or calcitonin receptor-like receptor (Calcrl), which together form the CGRP receptor complex (Extended Data Fig. 3f). Finally, NaV1.8+ nociceptors in granulation tissue were found to be surrounded by CD11b+ myeloid cells, which consist mainly of neutrophils and monocytes/macrophages in the context of tissue healing after acute injury (Extended Data Fig. 3g).

To gain insights into the effect of NaV1.8+ sensory neurons on immune cells during tissue healing, we analysed immune cell dynamics in injured tissues by flow cytometry. We focused on neutrophils, monocytes/macrophages, dendritic cells and T cells, because they constitute the predominant immune populations during tissue healing1,2,23,26 (Extended Data Fig. 4a,b). Compared with control mice, Nav1.8cre/Rosa26DTA mice exhibited an increased number of neutrophils and pro-inflammatory Ly6Chi monocytes/macrophages in skin and muscle 3 days post-injury (Fig. 2a). Similarly, macrophages in Nav1.8cre/Rosa26DTA mice showed a delayed polarization towards an anti-inflammatory and pro-repair (M2-like) phenotype in both tissues, characterized by a lower expression of CD206, a well-established M2-like macrophage marker, at later stages of the healing process (Fig. 2a). Although the total number of monocytes/macrophages observed 3 days post-injury in the skin was higher in Nav1.8cre/Rosa26DTA mice compared with controls (Fig. 2a), this phenomenon was not evident in muscle, a difference that may be attributed to either the lower density of NaV1.8+ nociceptors in injured muscle or variations in the dynamic accumulation of monocytes/macrophages between the two tissues. In both skin and muscle tissues, no major differences were observed in dendritic cells except for a lower number in the skin of Nav1.8cre/Rosa26DTA at a late time point (day 10 post-injury). However, some variations were noted in the counts of CD4 T cells, γδ T cells and cytotoxic (CD8) T cells 3 days post-injury (Extended Data Fig. 4c). Nevertheless, although these cell populations are recognized for their role in modulating tissue healing, their proportions were considerably lower in both Rosa26DTA and Nav1.8cre/Rosa26DTA tissues. They constituted 20–50 times fewer cells compared with neutrophils and monocytes/macrophages in skin and muscle, respectively (Extended Data Fig. 4d). Thus, although CGRP signalling on T cells may affect tissue healing to some extent, the profound impairment of tissue healing upon conditional Ramp1 knockout in myeloid cells suggests nociceptor-derived CGRP promotes tissue healing after acute injury primarily by influencing myeloid cells (Fig. 1g–j). Overall, the data demonstrated that the absence of nociceptors resulted in an increase in neutrophils and inflammatory monocytes/macrophages. This in turn delayed the transition towards an anti-inflammatory and pro-repair phase1,2,26, leading to impaired tissue healing. Of note, these findings align with previous reports showing that nociceptor ablation leads to an increased local accumulation of neutrophils after skin5,8 and lung9 infection, as well as neutrophils and inflammatory macrophages in colitis10 and brain infection12.

Fig. 2: CGRP regulates myeloid cell function during tissue healing.
figure 2

a, Analysis of neutrophil and monocyte/macrophage (Mo/Mϕ) populations by flow cytometry during tissue healing. Geometric mean of fluorescence intensity (MFI) of CD206 in macrophages was used to assess M2-like polarization. Data are plotted in kinetic line plots showing mean ± s.e.m. Skin: day 0, day 6 and day 10, n = 20; day 3, n = 22; and day 14, n = 12. Muscle: day 0, n = 16; day 3, n = 20; day 6, day 10, n = 18; and day 14, n = 10. be, Neutrophils and macrophages were treated with saline (PBS, 0 nM CGRP) or CGRP (1 or 20 nM). Results are expressed as fold change over the PBS (0 nM CGRP) group. b, Transwell migration towards CXCL1 or CCL2 with or without CGRP (n = 6–8). c, Cell death in response to CGRP and TNF plus IL-1 (neutrophils, n = 7; macrophages, n = 6). d, Macrophage efferocytosis of neutrophils after CGRP treatment with or without TNF/IL-1 (n = 4). e, Macrophage polarization determined via CD206 and arginase-1 protein expression following CGRP and IL-4/IL-13 or IL-10 treatment (n = 6). f, tdTomato+ bone marrow cells were administered systemically on day 2 post-injury. Fold change of tdTomato+ neutrophils and monocytes/macrophages in injured tissues on day 3 post-injury was assessed by flow cytometry. For efferocytosis, dead or dying tdTomato+ neutrophils were injected in skin wound borders on day 3 post-injury. Fold change of tdTomato+ endogenous monocytes/macrophages was assessed by flow cytometry 30 min post-injection (n = 6). g, Death of CD11b+ cells 3 days post-injury, assessed by TUNEL assay on injured tissue sections (n = 6). Boxes show median (centre line) and interquartile range (edges), whiskers show the range of values. Dots represent independent experiments. ae, Two-way ANOVA with Bonferroni post hoc test for pairwise comparisons. f,g, Two-tailed Student’s t-test. P values are indicated. NS, not significant.

Source Data

The increased number of neutrophils and inflammatory monocytes/macrophages observed in the absence of NaV1.8+ sensory neurons, along with the delayed transition of macrophages towards an anti-inflammatory and pro-repair phenotype, suggested that CGRP may regulate these cells through multiple mechanisms. Thus, we investigated the effect of CGRP on neutrophils and macrophages in vitro. First, we verified that these cells strongly express CGRP receptor subunits (Extended Data Fig. 5a,b). Then, we conducted migration assays and observed that CGRP severely inhibited neutrophil and macrophage migration towards common chemokines found in wounds (CXCL1 for neutrophils and CCL2 for macrophages) (Fig. 2b). Notably, cell migration was also inhibited to some extent in the absence of chemokines. However, CGRP did not inhibit cell migration in Ramp1−/− neutrophils and macrophages, confirming that CGRP signals via RAMP1–CALCRL to mediate its effects on these cells (Extended Data Fig. 5c). Next, we investigated the effect of CGRP on neutrophil and macrophage viability. CGRP triggered increased neutrophil and macrophage death in the presence of the inflammatory cytokines IL-1β and TNF, which are typically found in acute injuries. Notably, CGRP did not promote macrophage death when cells were cultured without inflammatory cytokines or with cytokines that induce an M2-like polarization, suggesting that CGRP elicits its effect on macrophage death during the inflammatory phase of tissue healing (Fig. 2c and Extended Data Fig. 5d). Subsequently, we explored whether CGRP influences neutrophil clearance by regulating macrophage efferocytosis. Stimulation with CGRP resulted in a marked increase of macrophage efferocytosis, both in the absence and presence of inflammatory cytokines (IL-1β and TNF) (Fig. 2d). Additionally, CGRP had no direct impact on macrophage Ly6C expression in vitro despite our earlier observation of an increased number of pro-inflammatory monocytes/macrophages (Ly6Chi) in injured tissues of mice lacking nociceptors (Extended Data Fig. 5e). This suggested that the high number of Ly6Chi monocytes/macrophages was probably a consequence of delayed neutrophil clearance and/or macrophage polarization. Finally, we tested whether CGRP accelerates macrophage polarization towards an M2-like phenotype. We found that in the presence of the typical anti-inflammatory cytokines IL-4 and IL-13 (IL-4/IL-13) or IL-10, CGRP treatment increased levels of the M2-like markers CD206 and arginase-1, respectively, indicating that CGRP accelerates polarization into an anti-inflammatory and pro-repair phenotype (Fig. 2e).

To validate inhibitory migration effects of CGRP in vivo, we used an adoptive transfer model in which tdTomato+ bone marrow cells were systemically administered into Rosa26DTA control and Nav1.8cre/Rosa26DTA mice following skin and muscle injuries. One day after the transfer, relative accumulation of tdTomato+ neutrophils and monocytes/macrophages in the injured tissues was assessed by flow cytometry. Compared with Rosa26DTA mice, Nav1.8cre/Rosa26DTA mice showed a significant increase in neutrophils and monocytes/macrophages into injured skin and muscle, suggesting that migration of these cell types into injured tissues was considerably increased (Fig. 2f). For validation of efferocytosis in vivo, we injected dead or dying tdTomato+ neutrophils intradermally at the border of skin wounds. The relative number of monocytes/macrophages positive for tdTomato was then assessed by flow cytometry and was found be significantly lower in Nav1.8cre/Rosa26DTA mice, demonstrating an impairment of efferocytosis (Fig. 2f). Finally, to assess neutrophil and macrophage cell death in vivo, we performed a TUNEL assay on injured tissue sections 3 days post-injury. Slightly fewer CD11b+ cells were TUNEL-positive in Nav1.8cre/Rosa26DTA mice (Fig. 2g and Extended Data Fig. 5f). However, as observed in vitro, the effect on cell death was relatively modest. Thus, CGRP-induced cell death is probably not a major mechanism in vivo. Together, these data support a model in which nociceptor-derived CGRP promotes tissue healing by tightly regulating neutrophil and monocyte/macrophage dynamics, functions and phenotypes in injured tissues, resulting in a faster transition from a pro-inflammatory to a pro-healing phase. CGRP inhibits neutrophil and monocyte/macrophage migration into injured tissues and may further enhance neutrophil and inflammatory macrophage death in the presence of inflammatory cytokines. Meanwhile, CGRP increases neutrophil clearance by stimulating macrophage efferocytosis, which, together with a direct effect of CGRP on macrophage polarization, supports macrophages switching towards an anti-inflammatory and pro-repair phenotype. This model aligns well with the function of neutrophils and macrophages during tissue healing. For instance, excessive mobilization of neutrophils and inflammatory monocytes/macrophages is generally associated with impaired wound healing1,2,26,31,32. Enhanced clearance rate of pro-inflammatory cells within injured tissues is also well-known to prevent excessive inflammation and facilitate the transition towards the pro-repair phase. Indeed, impaired tissue healing has been linked to delayed inflammatory cell death and a decrease in macrophage efferocytosis capability22,33,34.

Mediation of the CGRP effect by TSP-1

To further understand the molecular mechanisms by which CGRP modulates neutrophils and macrophages, we analysed the transcriptome of bone marrow-derived neutrophils and macrophages after CGRP stimulation by RNA sequencing (RNA-seq). For both cell types, gene ontology (GO) analysis of differentially expressed genes (DEGs) between CGRP-treated and control groups identified biological processes that were congruent with the in vitro effects of CGRP. In addition, the biological processes reflected the differences in neutrophil and macrophage dynamics observed in tissue injuries of control and nociceptor-depleted mice. For instance, for both neutrophils and macrophages, pathways enriched in the upregulated DEGs included cell death, whereas those among the downregulated DEGs included cell migration (for example, Ccr2, Cx3cr1, Itgav and Itgb3) (Fig. 3a and Extended Data Fig. 6a). Furthermore, pathways enriched in the upregulated DEGs were associated with tissue remodelling and differentiation in macrophages, featuring genes typically linked to an anti-inflammatory and pro-repair phenotype such as Arg1, Cebpb, Stat3, Sgk1, Tgfb3, Tgm2 and Vegfa35. Finally, in response to CGRP, macrophages upregulated genes that are strongly associated with efferocytosis such as Cd14, Clu, S1pr1, Rarg and Tgm236,37 (Fig. 3a and Extended Data Fig. 6a). Crucially, we found that thrombospondin-1 (Thbs1)—which encodes TSP-1, a multifunctional extracellular matrix (ECM) protein that regulates many biological processes including tissue healing38—was the most upregulated gene in both neutrophils and macrophages (Fig. 3b). Further confirming that TSP-1 is expressed in response to CGRP via RAMP1–CALCRL, Ramp1−/− neutrophils and macrophages did not upregulate Thbs1 after CGRP stimulation (Extended Data Fig. 6b). Moreover, TSP-1 levels after skin and muscle injuries were lower in Nav1.8cre/Rosa26DTA mice (Extended Data Fig. 6c).

Fig. 3: CGRP upregulates TSP-1 in neutrophils and macrophages to mediate its activity.
figure 3

a,b, RNA-seq analysis of CGRP-treated neutrophils and macrophages (1 nM or saline for 4 h). Differential gene expression was performed with limma-voom (false discovery rate (FDR) < 0.05). Fold change values returned by limma were used for pathway analysis with FDR < 0.05 to correct for multiple comparisons. a, GO enrichment analysis of significantly upregulated (red) and downregulated (blue) genes in CGRP-treated and saline-treated groups. b, Volcano plots showing DEGs with fold change > |1.5| between CGRP-treated and saline-treated groups. Significantly upregulated (red) and downregulated (blue) genes in CGRP-treated neutrophils and macrophages are shown (n = 3). cf, Neutrophils and macrophages were treated with saline (PBS, 0 nM TSP-1) or TSP-1 (1, 10 or 100 nM). Results are expressed as fold change over the PBS (0 nM TSP-1) control group. Boxes show median (centre line) and interquartile range (edges), whiskers show the range of values. Dots represent independent experiments. c, Transwell migration towards a chemoattractant (CXCL1 or CCL2) after TSP-1 treatment (n = 3 for neutrophils, n = 4 for macrophages). d, Cell death in response to TSP-1 with or without TNF plus IL-1 (n = 4). e, Macrophage efferocytosis of neutrophils after TSP-1 treatment (n = 4). f, Macrophage M2-like polarization determined via CD206 and arginase-1 expression in response to TSP-1 and IL-4/IL-13 or IL-10 treatments (CD206, n = 8; arginase-1, n = 4). cf, One-way ANOVA with Tukey post hoc test for pairwise comparisons. P values are indicated.

Source Data

Notably, it has been reported that macrophages are the primary source of TSP-1 in wounds, and the absence of TSP-1 leads to impaired wound healing and prolonged inflammation39,40. Furthermore, increasing evidence suggests that TSP-1 is an important regulator in immune responses41. Thus, similar to the assays performed with CGRP, we tested whether TSP-1 affects neutrophil and macrophage migration and death, as well as efferocytosis and macrophage polarization. Although some reports have suggested that TSP-1 promotes neutrophil and macrophage migration41, we clearly observed an inhibition of migration when cells were treated with TSP-1, similarly to our observations with CGRP (Fig. 3c). Moreover, TSP-1 accelerated death of neutrophils and macrophages in a dose-dependent manner in the presence of inflammatory cytokines (Fig. 3d). Efferocytosis of neutrophils by macrophages was also greatly enhanced following macrophage treatment with TSP-1 (Fig. 3e), in line with a previous study suggesting that TSP-1 acts as a bridge between neutrophils and macrophages to facilitate efferocytosis41. Finally, TSP-1 enhanced expression of CD206 and arginase-1 in the presence of anti-inflammatory cytokines (IL-4/IL-13 or IL-10), demonstrating that TSP-1 accelerates macrophage polarization towards an anti-inflammatory and pro-repair phenotype (Fig. 3f). To further confirm that CGRP exerted its effects primarily via an autocrine or paracrine action of TSP-1, we tested the effect of CGRP upon short interfering RNA (siRNA)-mediated knockdown of Thbs1 (Extended Data Fig. 6d). Only macrophages were used, as transfection of neutrophils is very challenging. Knockdown of Thbs1 abolished the effects of CGRP on cell migration, death, efferocytosis and polarization, whereas scramble siRNA showed no effect on all these cellular processes (Extended Data Fig. 6e–h). To validate these effects in vivo, we administered TSP-1 in Nav1.8cre/Rosa26DTA mice following skin and muscle injuries. TSP-1 delivery resulted in reduced numbers of neutrophils, monocytes/macrophages and inflammatory monocytes/macrophages (Ly6Chi) at day 3 post-injury. Additionally, TSP-1 led to an increase in CD206 expression at day 14 post-injury (Extended Data Fig. 6i). Together, these results provide clear support for the role of TSP-1 in mediating the immunomodulatory effect of CGRP on neutrophils and macrophages.

eCGRP promotes diabetic tissue healing

There is evidence suggesting that CGRP has a role in wound healing. For example, mice lacking CGRP exhibit impaired skin wound healing42. Additionally, CGRP application is likely to promote corneal repair43. Therefore, since we found that nociceptor-derived CGRP regulates neutrophils and macrophages to facilitate tissue healing, we investigated whether local delivery of CGRP could restore the impaired healing observed in mice ablated of nociceptors. CGRP is a small peptide, which poses a challenge in achieving sustained effects without immediate burst signalling when delivered locally into tissue, as it can rapidly signal to cells, diffuse away from the delivery site and undergo degradation. Moreover, for clinical application, a high concentration of CGRP circulating in the body is undesirable, owing to possible off-target effects44. Thus, we engineered CGRP to enhance retention and protection at delivery sites by fusing it to a sequence with a very high affinity for ECM components25,26,45. The ECM-binding sequence was fused to the N terminus of CGRP followed by a plasmin-sensitive sequence to allow the release of CGRP from ECM via proteolytic activity (Extended Data Fig. 7a,b). These modifications did not impair CGRP activity, as demonstrated by the preserved capacity of the engineered CGRP (eCGRP) to inhibit neutrophil and macrophage migration and induce cAMP in these cells via RAMP1–CALCRL (Extended Data Fig. 7c,d). As demonstrated previously25,26,45, the ECM binding enabled better retention of CGRP after delivery into tissues (Extended Data Fig. 7e,f). We then tested whether CGRP variants (wild-type CGRP and eCGRP) could promote closure of splinted skin wounds and the regeneration of quadriceps after volumetric muscle loss in Nav1.8cre+/−/Rosa26DTA mice. For skin, CGRP variants were delivered topically, whereas for muscle, they were delivered via a fibrin hydrogel. Both CGRP variants enhanced the extent, compared with saline control, of wound closure and muscle regeneration upon delivery of 1 μg. Moreover, at a lower dose (250 ng) eCGRP promoted greater wound closure and muscle regeneration compared with wild-type CGRP (Extended Data Fig. 7g–j). Notably, delivering a relatively high dose of CGRP (10 μg) into mice has been shown to contribute to peripheral nociceptive sensitization46. Thus we tested whether injection of CGRP variants induced pain (Extended Data Fig. 7k,l). Administration of wild-type CGRP or eCGRP into mouse hind paws did not elicit significant pain behaviours over the 48-h experiment. However, a transient and slight increase in thermal sensitivity was observed only for wild-type CGRP. This difference is probably owing the binding of eCGRP to the ECM, which prevents an immediate burst of signalling after delivery25,26,45.

We next searched for a model with more clinical relevance compared with the Nav1.8cre+/−/Rosa26DTA mice. Indeed, more than half of patients with diabetes develop peripheral neuropathy, characterized by the presence of dysfunctional peripheral nerves and a decrease in intraepidermal nerve fibres47,48. Consequently, the decrease in neuropeptide levels, including CGRP, may disrupt neuro–immune interactions crucial for tissue healing49,50. Indeed, patients with diabetes commonly experience chronic non-healing wounds, which represent the most prevalent and severe complication of the condition, alongside the development of muscle atrophy49,50,51. The diabetic Leprdb/db mouse is a model for type 2 diabetes that is commonly used to study impaired tissue healing, since it mimics some aspects of human chronic wounds, including immune dysregulation and peripheral neuropathies25,26,52. Additionally, it has been shown that Leprdb/db have impaired muscle regeneration53. Thus, we investigated the distribution of CGRP in Leprdb/db mice skin and muscle to confirm that these mice exhibited peripheral neuropathy. We observed a significant reduction in neuron-like structures expressing CGRP in both skin and muscle of Leprdb/db mice, mirroring the pattern seen in Nav1.8cre+/−/Rosa26DTA mice (Fig. 4a and Extended Data Fig. 8a). This observation supported the use of diabetic mice to assess the regenerative potential of local CGRP delivery. Notably, wound closure and muscle regeneration was greatly improved when injuries were treated with eCGRP, compared with saline control and wild-type CGRP (Fig. 4b–e). We also examined TSP-1 expression in granulation tissue post-eCGRP delivery, given the induction of TSP-1 expression by CGRP in neutrophils and macrophages. Immunostaining revealed a significant increase in TSP-1 deposition, colocalizing to some extent with myeloid (CD11b+) cells in skin and muscle granulation tissues (Extended Data Fig. 8b). Next, since neutrophil and macrophage dynamics were disrupted in Nav1.8cre+/−/Rosa26DTA mouse injuries, we investigated whether eCGRP delivery modulated those cells in injured tissues (Extended Data Fig. 9). eCGRP delivery reduced the number of neutrophils and monocytes/macrophages in both skin and muscle injures at early timepoints post-injury. Moreover, eCGRP delivery led to reduced levels of the pro-inflammatory marker Ly6C and increased the level of the anti-inflammatory marker CD206 in skin wounds (Fig. 4f). Indeed, high numbers of neutrophils and inflammatory macrophages in diabetic injured tissues are known to delay healing2,31,32. An abnormally high number of neutrophils in chronic wounds leads to an over-production of pro-inflammatory cytokines, reactive oxygen species and proteases54, and increased NETosis55. Similarly, failure to convert macrophages to an anti-inflammatory phenotype leads to high levels of pro-inflammatory cytokines and proteases32. Thus, we measured concentrations of inflammatory cytokines/chemokines and proteases in response to saline and eCGRP treatment. eCGRP resulted in a significant reduction of IL-1β, CCL2, CXCL2, MMP-2 and MMP-9 (Extended Data Fig. 9b)—factors that are known to impair tissue healing when their levels are increased26,56,57. Thus, the results overall support the idea that the local delivery of eCGRP accelerates the transition of diabetic injured tissues towards an anti-inflammatory and pro-repair phase.

Fig. 4: Delivery of eCGRP promotes tissue healing in diabetic mice.
figure 4

a, CGRP expression in skin and muscle of Nav1.8cre/Rosa26DTA and diabetic (Leprdb/db) mice detected by immunostaining. CGRP, green; nuclei, blue. White lines indicate the keratinocyte layer. Scale bars, 300 μm. The experiment was repeated independently 6 times. b,c, Saline, CGRP (500 ng) or equimolar eCGRP was delivered on Leprdb/db skin wounds at day 1 and day 3 post-injury. b, Wound closure was evaluated by histomorphometric analysis on day 10 post-injury (saline and eCGRP, n = 14; CGRP, n = 12). Boxes show median (centre line) and interquartile range (edges), whiskers show the range of values. c, Representative skin histology. Black arrows indicate wound edges and red arrows point to tips of epithelium tongue. Scale bar, 1 mm. d,e, Saline, CGRP (1 μg) or equimolar eCGRP was delivered in Leprdb/db quadricep volumetric muscle loss defect via a fibrin hydrogel. d, Muscle regeneration (represented by the percentage of fibrotic tissue and muscle area) was evaluated by histomorphometric analysis on day 8 post-injury. Boxes show median (centre line) and interquartile range (edges), whiskers show the range of values (n = 10). e, Representative histology (fibrotic tissue is stained blue; muscle tissue is stained red). Scale bar, 500 μm. f, Numbers of neutrophils (CD11b+Ly6G+F4/80) and monocytes/macrophages (CD11b+F4/80+Ly6G), and expression of Ly6C and CD206 in macrophages (represented by MFI) in Leprdb/db skin wounds and muscle injuries treated with saline or eCGRP were quantified by flow cytometry. Skin: n = 6. Muscle: day 3, n = 8 for saline, n = 7 for eCGRP; day 7, n = 9. Data are plotted as kinetic line plots showing mean ± s.e.m. b,d, One-way ANOVA with Tukey post hoc test for pairwise comparisons. f, Two-way ANOVA with Bonferroni post hoc test for pairwise comparisons. P values are indicated.

Source Data


CGRP and nociceptors have been shown to have immunomodulatory effects in various contexts7,8,9,10,11,12. For example, CGRP has a key role as an immunosuppressive mediator during sepsis58. Additionally, during bacterial infection, nociceptor-derived CGRP has been shown to induce an anti-inflammatory transcriptional programme in macrophages and suppress neutrophil recruitment12. However, the role of nociceptors during tissue healing after acute injury was unclear. This study reveals the existence of an important neuro–immune regenerative axis following acute tissue injury and highlights the intricate interplay between nociceptors, immune cells and tissue healing processes. We found that NaV1.8+ sensory neurons expressing CGRP, which mostly represent nociceptors, extend into the granulation tissue formed after skin and muscle injuries and profoundly modulate neutrophil and macrophage dynamics during tissue healing. This neuro–immune modulation promotes the transition towards an anti-inflammatory and pro-repair phase, which ultimately facilitates the healing process. Mechanistically, we demonstrate that CGRP signalling in neutrophils and macrophages induces the expression of TSP-1, which in turn limits their accumulation and accelerates cell death in the presence of inflammatory cytokines. In addition, CGRP promotes neutrophil clearance via efferocytosis and boosts macrophage polarization into an anti-inflammatory and pro-repair phenotype via autocrine and paracrine effects of TSP-1 (Extended Data Fig. 10). Collectively, the mechanisms uncover important implications for advancing our understanding of the tissue healing process after acute injury. These findings also have significant implications for advancing regenerative medicine, particularly for patients with peripheral neuropathies, including those associated with conditions such as diabetes. Harnessing the potential of this neuro–immune regenerative axis opens new avenues for effective therapies, whether as standalone treatments or in combination with existing therapeutic approaches. Utilizing or mimicking neuro–immune interactions holds great promise for addressing chronic wounds and other non-healing tissues, in which dysregulated neuro–immune interactions have a pathologic role and impair tissue healing.


Ethical statement for animal experiments

Animal experiments were approved by the Monash Animal Research Platform ethics committee and the Animal Research Committee of the Research Institute for Microbial Diseases of Osaka University (approval numbers 13294, 13335, 17075, 14013 and 23006).


Wild-type C57BL/6 J mice were from the Monash Animal Research Platform. Sperm from Nav1.8cre+/+ mice (B6.129-Scn10atm2(cre)Jnw/H B6, stock ID EM:04582, European Mouse Mutant Archive) were used for in vitro fertilization to generate Nav1.8cre+/− mice on a C57BL/6 J background. Rosa26DTA+/+ mice (B6.129-Gt(ROSA)26Sortm1(DTA)Mrc/J), strain 010527, Jackson Laboratory) were maintained on a C57BL/6 J background. To delete sensory neurons expressing NaV1.8, Nav1.8cre+/− mice were bred with Rosa26DTA+/+ mice to generate Nav1.8cre+/−/Rosa26DTA+/− mice. Nav1.8cre−/−/Rosa26DTA+/− littermates were used as controls. For visualizing NaV1.8+ neurons, Rosa26tdT reporter mice (B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J, strain 007914, Jackson Laboratory) were bred with Nav1.8cre+/+ mice to generate Nav1.8cre+/−/Rosa26tdT+/−. Leprdb/db mice (BKS.Cg-Dock7m + /+Leprdb/J, strain 000642) were obtained from Jackson Laboratory. Mice were bred as heterozygotes to generate Leprdb/db and Leprdb/+ littermates. B6.129S2-Ramp1<tm1.2Tsuj> mouse sperm was kindly provided K. Tsujikawa and used for in vitro fertilization to generate Ramp1fl/+ mice. Ramp1−/− mice were generated by crossing Ramp1fl/fl mice with CAGcre mice (C57BL/6-Tg(CAG-cre)13Miya, RIKEN BioResource Research Center, strain 09807). Specific deletion of Ramp1 in myeloid cells (LysMcre+/−/Ramp1fl/fl mouse) was done by crossing Ramp1fl/fl mice with LysMcre+/+ mice (B6.129P2-Lyzs<tm1(cre)Ifo>, RIKEN BioResource Research Center, strain 02302). LysMcre+/− littermates were used as controls. To obtain mice constitutively expressing tdTomato, Rosa26tdT mice were crossed with B6.C-Tg(CMV-cre)1Cgn/J mice from Jackson Laboratory (strain 006054).

Full-thickness skin wound model

Male 10- to 12-week-old mice were used for most experiments, except for experiment with CGRP variant delivery in which female (10- to 12-week-old non-diabetic or 12- to 14-week-old diabetic) mice were used. Full-thickness punch-biopsy wounds (5 mm in diameter) were created while mice were under isoflurane anaesthesia as described25,26. For analgesia, mice received subcutaneous administration of 0.1 mg kg−1 buprenorphine. In non-diabetic mice, wounds were covered with a round seal spot plaster (22.5 mm, Livingstone International, Australia) secured with 3M surgical tape. In experiments involving CGRP delivery, a nylon ring (Zenith 5/16 inch and M8 Nylon Washer) was attached with superglue (UHU) to prevent wound contraction. Wounds received topical treatment with either 10 μl of saline (PBS) or a CGRP variant in PBS. Solutions were applied in two different dosages: 250 ng of CGRP or equimolar eCGRP on day 1 post-injury for the low dose, and 500 ng of CGRP or equimolar eCGRP on day 1 and day 4 post-injury for the high dose. For diabetic mice, 4 wounds were created and treated with PBS or CGRP variant in PBS (500 ng CGRP or equimolar eCGRP) on day 1 and day 3.

Volumetric muscle loss model

Non-diabetic (10- to 12-week-old) and diabetic (12- to 14-week-old) male mice underwent isoflurane anaesthesia. For analgesia, mice received subcutaneous administration of 0.1 mg kg−1 buprenorphine. A 1-cm unilateral incision was made, exposing the fascia. Muscle injuries were created either with a 3-mm biopsy punch or by excising a 3 mm × 5 mm segment of the quadriceps, including the rectus femoris muscle. In experiments involving CGRP delivery, muscle defects were covered with a fibrin matrix (40 μl total, 8 mg ml−1 fibrinogen (Enzyme Research Laboratories), 12 U ml−1 bovine thrombin (Sigma), 5 mM CaCl2, and 17 μg ml−1 aprotinin (Roche, Sigma)) containing CGRP (250 ng or 1 μg for non-diabetic mice and 1 μg for diabetic mice) or equimolar eCGRP. The incision site was sutured with non-absorbable sutures.

Adoptive transfer of bone marrow cells

Bone marrow cells (1 × 107) from 6-week-old Ramp1−/− or wild-type C57BL/6 J mice were intravenously injected into lethally irradiated 6-week-old recipient wild-type or Ramp1−/− mice that received 100 mg l−1 neomycin sulfate for 2 weeks post-irradiation. Skin or muscle defect surgeries were performed 6 weeks after transplantation.

Histological analysis

Skin wounds were collected using an 8-mm biopsy punch, fixed in 10% formalin at room temperature for 24 h, cut at the edge of the wounds, embedded in paraffin and sectioned at 4 μm until the centre of the wound was passed. Re-epithelialization was measured by histomorphometric analysis. Slides were stained with haematoxylin and eosin, and the centre of the wound was determined by measuring the distance between the panniculus carnosus muscle gap using Aperio ImageScope Viewer (Leica Biosystems). Closure was calculated as the ratio of epidermis closure to the length of the panniculus carnosus gap. Muscle injury sites, including the proximal and distal quadriceps segments, were collected, fixed in 10% formalin solution for 24 h, embedded in paraffin, and sectioned at 4 μm thickness for 5 depths, starting from the edge of the patella, passing the centre of the wound, up to the proximal end of the defect site. Cross-sections were stained with Masson’s Trichrome. Muscle regeneration was determined by averaging the percentage of blue-stained fibrotic area (normalized to the total area) and the remaining non-fibrotic muscle area across five tissue section depths, using Aperio ImageScope.

Immunohistochemistry for neuropeptides, TSP-1 and myeloid cells

Tissues and DRGs (L1–L6 vertebrae) were fixed in 4% paraformaldehyde, cryoprotected in 30% sucrose, and embedded in OCT compound for 10 μm cryosections. Sections were stored at −20 °C, thawed, permeabilized and blocked with 1% bovine serum albumin (BSA), 10% normal goat serum (NGS) or normal donkey serum in PBS for 1 h. Sudan Black B solution (0.1% in 70% ethanol) was applied for 10 min. For neuropeptide detection, primary antibodies were added in staining buffer (0.5% BSA, 5% NGS or 0.5% BSA, 5% normal donkey serum in PBS) overnight at 4 °C. The primary antibodies included rabbit anti-CGRP (66.7 μg ml−1, Sigma, C8198), rabbit anti-substance P (1:500, Thermo Fisher Scientific, 20064), rabbit anti-VIP (1:500, Thermo Fisher Scientific, 20077), and goat anti-galanin (1 μg ml−1, Abcam, 99452). For TSP-1 detection, sections were incubated with AffiniPure Fab Fragment goat anti-mouse IgG (H + L) at 100 μg ml−1 (Jackson ImmunoResearch Labs, 115-007-003) in PBS for 2 h at room temperature, followed by mouse anti-thrombospondin-1 (5 μg ml−1, Thermo Fisher Scientific, 14-9756-82). For myeloid cell detection, slides were incubated with rat anti-mouse CD11b (5 μg ml−1, Thermo Fisher Scientific, 14-0112-82). Sections were washed and incubated with respective secondary antibodies for 1 h at room temperature. The secondary antibodies included F(ab′)2-Goat anti-Rabbit IgG Alexa Fluor 488 (2.6 μg ml−1, Thermo Fisher Scientific, A-11070), donkey anti-goat IgG Alexa Fluor 488 (2.6 μg ml−1, Thermo Fisher Scientific, A-11055), goat anti-mouse IgG Alexa Fluor Plus 488 (2.6 μg ml−1, Thermo Fisher Scientific, A48286TR), and goat anti-rat IgG Alexa Fluor Plus 594 (2.6 μg ml−1, Thermo Fisher Scientific, A48264). Counterstaining with DAPI for 10 min and mounting with Fluoroshield followed. Imaging was done using Leica DMi8 fluorescent microscope and Leica SP8 inverted confocal microscope.

Evaluation of neuropeptide expression

Skin and muscle samples from male Nav1.8cre+/−/Rosa26tdT+/− mice (10- to 12-week-old) were immunostained as detailed above, imaged on a Leica DMi8 fluorescent microscope and processed using Fiji59. Binary images were created with an optimal threshold, and overlapping areas were determined by combining region of interest binary images. Area fraction values, indicating neuropeptide expression in NaV1.8+ nerves, were calculated based on pixel ratios and converted using a built-in scale bar60,61.

Immunofluorescence for Ki-67 and KRT14

Paraffin sections underwent 20-minute antigen retrieval in 10 mM sodium citrate buffer (pH 6.0), followed by PBS washes and 5-minute permeabilization (0.2% Triton X-100 in PBS). Blocking with 10% NGS in 1% BSA/PBS occurred for 2 h, and endogenous IgG was blocked with unconjugated affinity-purified F(ab) fragment anti-mouse IgG (H + L) (10 μg ml−1, Jackson ImmunoResearch, AB_2338476) for 1 h at room temperature. Staining overnight at 4 °C utilized rat anti-mouse Ki-67 (5 μg ml−1, Thermo Fisher Scientific, 5698-82) and mouse anti-mouse cytokeratin 14 (4 μg ml−1, Thermo Fisher Scientific, MA5-11599) in 1% NGS in PBS with 0.1% BSA. After PBS-T washes, incubation with secondary antibodies occurred: goat anti-mouse Alexa Fluor 647 (2 μg ml−1, Thermo Fisher Scientific, A-21235) and goat anti-rat Alexa Fluor 488 (2.67 μg ml−1, Thermo Fisher Scientific, A48262TR) for 1 h at room temperature, followed by PBS-T wash. Counterstaining with DAPI (1 μg ml−1) for 10 min at room temperature preceded mounting with Fluoroshield.

TUNEL assay

The In Situ Cell Death Detection Kit, TMR red (Roche, 12156792910) was used, following the manufacturer’s instructions on muscle and skin tissue cryosections. To detect CD11b+ cells, sections were incubated overnight at 4 °C with rat anti-mouse CD11b (5 μg ml−1, M1/70, Thermo Fisher Scientific, 14-0112-82) in staining buffer. After PBS-T washes, sections were incubated with Alexa Fluor 488 goat anti-rat antibody (2.67 µg ml−1, Thermo Fisher Scientific, A48262TR), washed with PBS-T, and counterstained with DAPI (1 μg ml−1) before mounting with Fluoroshield. Two tissue section levels were evaluated per sample to determine the percentage of TUNEL+ apoptotic cells over total CD11b+ cells, examining three fields per section within the injury site.

Fibroblast, keratinocytes, myoblasts and endothelial cell maintenance

Human umbilical vein endothelial cells (HUVECs; Sigma, 200P-05N) cultured in EGM-2 medium (Lonza, CC-4176) up to 3 passages, and primary mouse fibroblasts from C57BL/6 J mouse tails26 (passages 2–3) were used. MCDB-131 medium (Thermo Fisher Scientific) with 100 mg ml−1 penicillin/streptomycin and 2 mM glutamine was employed for proliferation assays. C2C12 mouse myoblasts (CellBank Australia) were cultured in a 1:1 ratio of DMEM to F10 medium (2 mM glutamine, 10% FBS, 100 units ml−1 penicillin/streptomycin). HaCaT keratinocytes (a gift from R. Boyd) were cultured in DMEM without Ca2+ and Mg2+ (2 nM glutamine, 10% Chelex-treated FBS, 0.03 nM calcium chloride, 100 units ml−1 penicillin/streptomycin) for at least 3 passages. Cells obtained from vendors were authenticated and certified negative for Mycoplasma contamination. For proliferation assays, FBS was reduced to 2% or kept at 10% for 24 h. Detached with TrypLE, cells were seeded (2,000 cells per well for HUVECs, fibroblasts, HaCaTs; 1,000 cells per well for C2C12) and treated with CGRP (1 or 20 nM) or 10–20% FBS. Incubation for 48 h (fibroblasts, C2C12) or 72 h (HUVECs, HaCaTs) at 37 °C with 5% CO2 followed. Proliferation was determined using the CyQUANT Cell Proliferation Assay (Invitrogen), presented as fold change over basal proliferation (medium only). PerkinElmer EnVision multi-mode plate reader with EnSpire Manager software was used.

Flow cytometry with tissue samples

Skin wounds were collected using an 8-mm biopsy punch, and muscle defects were dissected to isolate the quadriceps. Samples were minced with scissors and subjected to two serial digestions with collagenase XI (1 mg ml−1) at 37 °C (two times 20 min for skin, two times 15 min for muscle). After the first digestion, the supernatant was collected and mixed with neutralization buffer (DMEM/F12 with 10% FBS and 5 mM EDTA). The first collection was kept on ice and fresh collagenase XI was added to the undigested tissue for the second digestion. Digestion mixtures were passed through a 70-μm cell strainer and stained with LIVE/DEAD Fixable Aqua dye (Thermo Fisher Scientific, 1:400 dilution in PBS) for 20 min on ice. Cells were incubated with TruStain FcX anti-CD16/32 (10 μg ml−1; clone 93, BioLegend) diluted in staining buffer (5% FBS and 2 mM EDTA in PBS) for 20 min and subsequently incubated with primary antibodies in staining buffer for a further 30 min on ice. The following anti-mouse antibodies from BioLegend were used: FITC anti-CD11b (clone M1/70, 6.6 μg ml−1) or BV711 anti-CD11b (clone M1/70, 2 μg ml−1); PE anti-F4/80 (clone BM8, 4 μg ml−1); BV421 anti-Ly6G (clone 1A8, 2 μg ml−1); BV711 anti-Ly6C (clone HK1.4, 1 μg ml−1) or FITC anti-Ly6C (clone HK1.4, 5 μg ml−1); PE-Cyanine7 anti-CD206 (clone C068C2, 2.6 μg ml−1); APC anti-CD206 (clone C068C2, 2 μg ml1); PE-Cyanine7 anti-CD3 (clone 17A2, 4 μg ml−1); BV711 anti-CD3 (clone 17A2, 4 μg ml−1);APC anti-CD4 (clone GK1.5, 2 μg ml−1); BV421 anti-CD8 (clone 53-6.7, 2 μg ml−1); PE anti-TCR β (clone H57-597, 2 μg ml−1); APC/Fire 750 anti-TCR γ/δ (clone GL3, 2 μg ml−1); PE-Cyanine anti-CD11c (clone N418, 2 μg ml−1); APC/Fire 750 anti-MHC Class II (clone M5/114.15.2, 2 μg ml−1). Cells were washed once with a large volume of staining buffer before analysis with BD LSR Fortessa X-20 and FlowJo software (BD Biosciences).

Mouse bone marrow neutrophil and monocyte isolation

Bone marrow cells were flushed from femora and tibiae of C57BL/6 J mice (8- to 12-week-old) with HBSS without Ca2+ and Mg2+ containing 2% FBS and 1 mM EDTA. Cell suspension was passed through a 70-μm strainer. Next, EasySep Mouse Neutrophil Enrichment Kit or EasySep Mouse Monocyte Isolation Kit (STEMCELL Technologies) was used to isolate neutrophils or monocytes according to the manufacturer’s instructions. Neutrophils were resuspended in RPMI containing 100 units ml−1 penicillin/streptomycin and 10% FBS for cell migration assay and cell death assay or 2% FBS for efferocytosis. Monocytes were cultured in DMEM/F12 (Thermo Fisher Scientific) containing 10% FBS, 2–10 ng ml−1 M-CSF (PeproTech) and 100 units ml−1 penicillin/streptomycin for subsequent experiments. RAMP1 and CALCRL were detected on neutrophils and monocytes using rabbit anti-RAMP1 (8.5 μg ml−1, Alomone Lab, ARR-021) and rabbit anti-calcitonin receptor-like receptor (5 μg ml−1, Biorbyt, orb526584).

Neutrophil cell death

Bone marrow-isolated neutrophils were cultured in RPMI 1640 medium (10% FBS). Cells were incubated with CGRP (1-20 nM, Tocris Bioscience, 83651-90-5) for 10 min, followed by treatment with IL-1 (5 ng ml−1) and TNF (50 ng ml−1) for 12 h at 37 °C with 5% CO2 to induce cell death. After 12 h, cells were washed with PBS and incubated with LIVE/DEAD Fixable Aqua dye (Thermo Fisher Scientific, 1:400 dilution) in PBS on ice for 20 min. Cell death was assessed using BD LSR Fortessa X-20 and FlowJo software (BD Biosciences).

Macrophage cell death and polarization marker expression

Bone marrow cells from 8- to 12-week-old C57BL/6 J mice were flushed, filtered and cultured in conditioned medium (DMEM/F12 with 10% heat-inactivated FBS, 100 units ml−1 penicillin/streptomycin, and 20% L929 fibroblasts-conditioned medium) at 37 °C with 5% CO2. After 7–9 days, differentiated macrophages were collected and seeded in 12-well or 6-well plates. The next day, cells were treated with CGRP (1 or 20 nM, Tocris Bioscience, 83651-90-5) for 20 min before exposure to mouse IL-1 (5 ng ml−1) and TNF (50 ng ml−1), IL-4 (2 ng ml−1) and IL-13 (2 ng ml−1), or IL-10 (2 ng ml−1) (PeproTech Inc) for 24 or 72 h. Macrophages were detached with TrypLE (Gibco) containing 3 mM EDTA, stained with LIVE/DEAD Aqua dye for 20 min on ice, and incubated with blocking solution (10 μg ml−1 TruStain FcX anti-CD16/32 (clone 93, BioLegend)) for 20 min before staining with antibodies for 30 min on ice. Antibodies from BioLegend included PE anti-CD11b (clone M1/70, 1 μg ml−1), BV711 anti-F4/80 (clone BM8, 2 μg ml−1), APC anti-CD80 (clone 16-10A1, 0.5 μg ml−1) and PE-Cyanine7 anti-CD206 (clone C068C2, 1 μg ml−1). For intracellular staining, cells were fixed and permeabilized using FluoroFix Buffer and Intracellular Staining Permeabilization Wash Buffer (Perm buffer, BioLegend). APC anti-mouse arginase-1 (Thermo Fisher, Clone AlexF5, 1 μg ml−1) was added to the Perm buffer and incubated with the cells for 30 min on ice. After washing with Perm buffer and staining buffer, cells were analysed using BD LSR Fortessa X-20 and FlowJo software (BD Biosciences).

Neutrophil and macrophage migration

Assays were conducted using 6.5-mm-diameter culture plate inserts (Corning) with 5-μm and 3-μm pore sizes for macrophages and neutrophils, respectively. Macrophages (1 × 105) or neutrophils (3 × 105) in migration media (DMEM/F12 with 0.25% BSA) were added to the inserts. The lower chambers contained migration buffer alone or chemoattractant (mouse CCL2 10 ng ml−1 for macrophages or mouse CXCL1/KC 150 ng ml−1 for neutrophils, PeproTech) with or without CGRP. Cells were allowed to migrate through the insert membrane for 3-4 h at 37 °C with 5% CO2. For macrophages, the inserts were then fixed with 4% paraformaldehyde, and cells on the upper side were removed. DAPI (1 μg ml−1) was used to stain cells on the bottom side, and they were counted using a fluorescent microscope. For neutrophils, cells that migrated into the lower chamber were collected and counted using a haemocytometer. The data are presented as the fold change, calculated by dividing the number of cells that migrated in response to treatments by the number of cells that migrated spontaneously (migration media only).


An efferocytosis assay kit (Cayman, 601770) was used following the manufacturer’s instructions. Neutrophils were labelled with CFSE and cultured in RPMI with 2% serum for 12 h to induce cell death. Bone marrow-derived macrophages cultured for 7 days were seeded at a density of 4 × 105 cells per well in a 6-well plate with DMEM/F12 containing 10% FBS and 100 units ml−1 penicillin/streptomycin. Prior to the assay, macrophages were pre-treated with CGRP (1 or 20 nM) for 24 h. Macrophages were collected, labelled with CytoTell Blue, and then incubated with CFSE-labelled dead/dying neutrophils at different ratios (1:1, 1:2, and 1:4) at 37 °C for 15 min. The reaction was stopped by washing cells with ice-cold PBS containing 5% FBS and 1 mM EDTA. Cells were analysed with BD LSR Fortessa X-20 and FlowJo software (BD Biosciences). Macrophages were identified by CytoTell Blue-positive staining, and the efferocytosis index was calculated as the percentage of CFSE-positive cells in CytoTell Blue-labelled macrophages.

Adoptive transfer of tdTomato+ cells for in vivo migration and efferocytosis

tdTomato+ bone marrow cells from CMV-cre/Rosa26tdTomato male mice (8- to 12-week-old) were adoptively transferred into Nav1.8cre/Rosa26DTA and Rosa26DTA mice either directly after red blood cell lysis (migration assay) or following neutrophil isolation (efferocytosis assay). In the migration assay, 1 × 107 cells were intravenously delivered on day 2 after skin or muscle injury. On day 3, collected tissues were analysed via flow cytometry to detect tdTomato+ cells. For the efferocytosis assay, neutrophils were cultured in low serum (2%) for 24 h to induce cell death, and 2 × 106 dead or dying neutrophils were intradermally injected at the skin wound border on day 3 post-injury. After 30 min, collected tissues were assessed via flow cytometry to quantify efferocytosis as the number of monocytes or macrophages that had taken up tdTomato+ apoptotic neutrophils. Results were presented as fold change relative to Rosa26DTA control mice.

RT–PCR, qPCR and RNA-seq

Isolated neutrophils were treated with CGRP (1 nM) in RPMI with 10% FBS and 100 units ml−1 penicillin/streptomycin for 4 h at 37 °C with 5% CO2. Isolated monocytes cultured in DMEM/F12 with 10% FBS, 100 units ml−1 penicillin/streptomycin, and M-CSF (10 ng ml−1) for 3 days, had their medium replaced with CGRP (1 nM) for 4 h at 37 °C with 5% CO2. After collection, RNA extraction used the RNeasy Plus Micro Kit (Qiagen). For PCR with reverse transcription (RT–PCR) and quantitative PCR (qPCR), reverse transcription used ReverTra Ace (Toyobo). RT–PCR primers were: Human_Calcrl 5′-CATGCACATCCTTATGCAC-3′ and 5′-CCATCACTGATTGTTGACAC-3′; Human_Ramp1 5′-GCCAGGAGGCTAACTACG-3′ and 5′-GAAGAACCTGTCCACCTCTG-3′; Mouse_Calcrl 5′-GGTACCACTACTTGGCATTG-3′ and 5′-GTCACTGATTGTTGACACTG-3′; Mouse_Ramp1 5′-GACGCTATGGTGTGACT-3′ and 5′-GAGTGCAGTCATGAGCAG-3′. Human or mouse GAPDH primers were from Integrated DNA Technologies (51-01-07-12 and 51-01-07-13, respectively). PCR products were analysed by gel electrophoresis. qPCR was performed using LightCycle96 with software LightCycle96 (Roche Diagnostics) and TaqMan Assay primers from Thermo Fisher Scientific (Thbs1, Mm00449032_g1; Gapdh, Mm99999915_g1). For RNA-seq, RNA quantity and quality assessment, library preparation and sequencing were performed at the Medical Genomics Facility, Monash Health Translation Precinct (MHTP). RNA quantity was assessed using Qubit. RNA samples (20 ng) with RNA integrity number (RIN) value ≥ 7 were used for library preparation. First strand synthesis was performed using a dT primer which adds the Illumina P7 (5′-CAAGCAGAAGACGGCATACGAGAT-3′), 8-bp i7 index for each sample and a 10-bp unique molecular identifier. The modified reverse transcriptase reaction also adds a template switching sequence at the 5′ end of the RNA during the generation of indexed cDNA. These first stand indexed cDNA were pooled and amplified using primers to P7 and the template switch sequence. Illumina P5 was added by tagmentation by Nextera transposase during amplification. Standard Illumina R1 primer was used (main cDNA read), followed by standard i7 primer for index or unique molecular identifier. R2 primer was present but not used as it will read into poly-A tail. Sequencing was performed on the NextSeq2000 (Illumina), using NextSeq 1000/2000 P2 Reagents (100 cycles) v3 (Illumina) in accordance with the Illumina Protocol 1000000109376 v3 Nov2020.

Demultiplexing and mapping

Fastq files were processed using the nfCore/RNAseq (v3.2) pipeline using the umi function62. Reads were aligned to the Mus musculus GRCm38 reference using STAR aligner63. Reads were quantified using featureCounts producing the raw genes count matrix and various quality control metrics which were summarized in a multiQC report64,65. Raw counts were analysed with Degust66, a web tool which performs normalization using trimmed mean of M values (TMM)67. Differential gene expression analysis was performed using limma/voom68 in Degust and genes with a FDR-adjusted P value < 0.05 were considered significantly upregulated or downregulated. Volcano plots were made using the web tool, VolcaNoseR69. Gene ontology enrichment analysis for biological processes was performed with the web tool, ShinyGO 0.77, by providing all upregulated or downregulated DEGs separately as the input for each experimental group70.

siRNA-mediated knockdown

Macrophages (4 × 105 cells per well in a 6-well plate) were transfected with 10 nM scrambled siRNA (Silencer Select Negative Control No. 1 siRNA, Thermo Fisher, 4390843) or Silencer Select Pre-Designed siRNA against mouse TSP-1 (Thermo Fisher, s124596) using Reduced-Serum Medium (Opti-MEM, Gibco) and Lipofectamine RNAiMAX (Invitrogen, 51985034) for 6 h. The medium was then replaced with fresh culture medium (DMEM/F12 with 10% FBS). After 24 h, cells were collected for the migration assay. For the efferocytosis assay, cells were cultured with 1 nM CGRP immediately after transfection. After 24 h, cells were collected and co-cultured with dead or dying neutrophils. The evaluation of cell death and polarization used the same methods as those for assessing macrophage death and polarization marker expression.

CGRP variants

CGRP and eCGRP were synthesized by ProteoGenix. eCGRP was designed to contain PlGF residues 123–141 at the N terminus followed by a plasmin-sensitive sequence from vitronectin (KGYR)71. For both variants, a disulfide bond was formed between the two cysteine residues and the C-terminal phenylalanine was amidated. Peptide purity, determined by high performance liquid chromatography, was 89.63% for CGRP and 87.33% for eCGRP.

Cleavage of eCGRP by plasmin

CGRP (4 μg) and equimolar eCGRP in 20 μl of PBS (pH 7.2) were incubated with plasmin (0.0005 U μg−1, Sigma) at 37 °C for 60 min. Aprotinin (25 μg ml−1, Sigma) was added for 5 min at 37 °C to stop plasmin activity. Samples were analysed by SDS–PAGE.

Retention of CGRP and eCGRP into skin and muscle

CGRP (1 μg) or an equal molar amount of eCGRP was intradermally administered to the shaved dorsal skin of male 10- to 12-week-old Nav1.8cre+/−/Rosa26tdT+/− mice, with injection sites marked using a marker. For muscle, CGRP variants were injected into the quadriceps. After 24 h, collected injection sites underwent cryosectioning and immunostaining. Fiji59 was used for analysis, excluding the co-localization area of CGRP with tdTomato fluorescence, indicating endogenous CGRP expression.

cAMP quantification

Freshly isolated neutrophils or bone marrow-derived macrophages (1 million cells) were treated with CGRP (1 nM) in RPMI with 10% FBS for 30 min at 37 °C with 5% CO2. cAMP levels were quantified using a cAMP ELISA kit from Cayman Chemical (581001) according to the manufacturer’s instructions.

Spontaneous pain behaviour assessment

Eight mice per group (4 males, 4 females, C57BL/6 J, 10- to 12-week-old) were acclimatized for 1 h in empty cages. The right hind paw received an intraplantar injection 1 μg of wild-type CGRP, equimolar amount of eCGRP, 0.05% capsaicin (Sigma, M2028), or 20 μl saline. Mice were immediately placed in the cage, and their behaviour was recorded. The number of episodes and the time spent licking, shaking, flinching and lifting the paw were recorded for first 5 min and for 5 min after 1, 6, 24 and 48 h.

Hot plate test

Eight mice per group (4 males, 4 females, C57BL/6 J, 10- to 12-week-old) received an intraplantar injection of 1 μg of wild-type CGRP, equimolar amount of eCGRP, or 20 μl saline in the right hind paw. After 30 min, mice were individually placed on a metal hot plate set to 52 °C. The latency, from mouse placement on the surface to the first behavioural sign of nociception (for example, lifting, shaking, licking the hind paw or jumping), was measured. Mice were immediately removed from the hot plate after responding or after a 30 s cut-off. The test was repeated after 1, 6, 24 and 48 h.

ELISAs for cytokines and MMPs

Homogenized skin wound and muscle tissues were incubated for 30 min on ice in T-PER Tissue Protein Extraction Reagent (10 ml per g of tissue, Thermo Fisher Scientific) containing 1 tablet of protease inhibitor for 7 ml (Roche). Samples were then centrifuged at 10,000g for 5 min and supernatants were stored at −80 °C. Total protein concentration was measured with a Bradford assay (Millipore). Cytokines and MMPs were detected by ELISA from R&D Systems; Mouse IL-1 beta/IL-1F2 DuoSet ELISA; Mouse CCL2/JE/MCP-1 DuoSet ELISA, Mouse CXCL2/MIP-2 DuoSet ELISA; Total MMP-2 Quantikine ELISA Kit; Mouse Total MMP-9 DuoSet ELISA.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 10 (GraphPad). Significant differences were calculated with Student’s t-test, one-sample t-test, and by ANOVA when performing multiple comparisons between groups. P < 0.05 was considered as a statistically significant difference.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.