VEGF signaling promotes blastema growth and proliferation of vascular and non-vascular cells during axolotl limb regeneration

Abstract

Salamanders are capable of regenerating whole limbs throughout life, a feat that is unmatched by other tetrapods. Limb regeneration is dependent upon the formation of a blastema containing progenitor cells which give rise to most tissues of the regenerated limb. Many signaling pathways, including FGF, BMP and Wnt, are required for regeneration, but the role of VEGF signaling during salamander limb regeneration is not well understood, particularly outside of angiogenesis. Here we show that VEGF signaling is essential for limb regeneration and that blastema cells and limb fibroblasts display impaired proliferation in the absence of VEGF signaling. Loss of VEGF signaling reduces expression of EMT-associated genes, suggesting VEGF signaling promotes expression of EMT-associated transcription factors, including Snai2, during axolotl limb regeneration. These findings highlight potential roles for VEGF signaling during regeneration which may extend beyond its expected pro-angiogenic function.

Graphical Abstract

Introduction

The axolotl (Ambystoma mexicanum), along with other salamanders, can perform the remarkable feat of complete regeneration of functional limbs, including the regeneration of each of the necessary tissue types. For proper limb regeneration to occur, the wound must first be healed and covered by a wound epidermis, which facilitates the coalescence of blastema cells at the site of injury, forming the blastema (McCusker et al., 2015; Tajer et al., 2023). The blastema is a transient structure composed of progenitor cells which give rise to much of a regenerated limb, including bones, muscle, and connective tissues (Currie et al., 2016; Kragl et al., 2009; Leigh et al., 2018). Understanding how some wounds in axolotls heal with simple healing, whereas wounds induced by amputation heal by blastema formation, is critical. Importantly, mammals, including humans, are not naturally capable of forming a blastema after limb loss and can only perform wound healing. Therefore, the transition between a wound-healing state and a proliferative blastema state in salamanders may represent an important aspect of limb regeneration that could be fundamental to the development of future therapeutic interventions. Candidate genetic regulatory pathways that might shift a wound-healing response to a blastema-formation and -growth response have been proposed from transcriptional analyses (Campbell et al., 2011; Knapp et al., 2013), but few experiments have addressed this process directly.

Beyond blastema formation, a hallmark of successful regeneration is its scar-free nature, which does not occur in most post-natal mammals. Unlike in mammals, axolotl limb regeneration leaves no visible trace of injury on the skin. However, repeated amputation causes failure to regenerate and differences in wound healing (Bryant et al., 2017a). This phenomenon shares some characteristics with scarring, such as a scar-like appearance to the wound epidermis, and this parallel may highlight an inability to transition between wound healing and regenerative states. In mammals, wound healing is succeeded by a burst of vascularization, which facilitates fibroblast and immune cell arrival at the wound site but often results in scar formation (Tonnesen et al., 2000). Vascular endothelial growth factor (VEGF) signaling is inhibitory to mouse digit tip regeneration (Yu et al., 2014) and VEGF overexpression in dermal fibroblasts results in increased scar formation after transplantation into rats (Shams et al., 2022), suggesting a negative role for VEGF signaling and angiogenesis in these processes in mammals. Conversely, in axolotls, inhibition of VEGF signaling was shown to be insufficient to impair tail and limb regeneration, while, similarly, increased VEGF signaling, did not block tail regeneration (Brashears et al., 2024; Ritenour and Dickie, 2017). These findings highlight how evolutionarily conserved signaling pathways may be differentially regulated following injury in different species. Here we inhibit VEGF signaling during axolotl limb regeneration and show a loss of limb regeneration but not blastema formation. We found VEGF signaling regulates proliferation in both axolotl blastemas and in cultured axolotl limb fibroblasts. Through sequencing studies, we uncover a possible role for VEGF-mediated regulation of genes previously implicated in epithelial-to-mesenchymal transition (EMT) processes in other species in axolotl limb regeneration. These results highlight previously underappreciated roles of VEGF signaling in regeneration beyond its established role in vascularization during bone regeneration (Grosso et al., 2023; Kaigler et al., 2006; Kleinheinz et al., 2005) and presumed pro-angiogenic function in salamander limb regeneration (Brashears and Dickie, 2024; Rageh et al., 2002; Ritenour and Dickie, 2017; Smith and Wolpert, 1975), illustrating the need to further characterize the intricacies of this pathway in regenerative organisms.

Results

VEGF signaling is required for limb regeneration.

VEGF signaling is most well-characterized as a regulator of angiogenesis (Astin et al., 2014; Blanco and Gerhardt, 2013; Bower et al., 2017; Cleaver and Krieg, 1998; Hogan and Schulte-Merker, 2017; Koch et al., 2011; Olsson et al., 2006; Phng and Gerhardt, 2009), but has been implicated in the development of other tissues and organ systems, such as regulation of axonal growth in the nervous system (Bellon et al., 2010), and axonal outgrowth from cultured ganglia (Sondell et al., 1999), suggesting a broader, context-specific role during tissue growth. However, how VEGF signaling functions during large-scale complex tissue regeneration is understudied. The presence of vasculature has been observed in salamander blastemas but how vascularization is regulated during blastema growth has not been established (Rageh et al., 2002; Smith and Wolpert, 1975; Whited et al., 2013). VEGF inhibition was previously performed during axolotl tail and limb regeneration, but, while reduced angiogenesis was observed, no loss of regeneration occurred (Ritenour and Dickie, 2017). This may have been because the concentration used was insufficient to inhibit VEGF receptor 3 (Vegfr3), based on other findings (Drevs et al., 2002). VEGF signaling via Vegfr3 may therefore be sufficient to induce angiogenesis in the absence of Vegfr1 and Vefgr2 function during axolotl regeneration. Given that reduced growth is a phenotype consistently observed when angiogenesis is reduced, during development (Okabe et al., 2020), regeneration (Uda et al., 2013), and diseases such as cancer (Lopes-Coelho et al., 2021), these findings suggest that more detailed analysis of VEGF function during appendage regeneration are required.

Therefore, we sought an alternative inhibitor that could be used at low doses to fully inhibit Vegfr function during axolotl limb regeneration. We used the VEGF receptor tyrosine kinase inhibitor AV951 (Tivozanib) which has been used to block angiogenesis in zebrafish (Kang et al., 2013; Nakamura et al., 2006; Savage et al., 2019). We performed mid-stylopod (upper limb) amputations on axolotls and treated them with 80nM AV951 via immersion from 3 days post amputation (dpa) until 22 dpa, to specifically examine the requirement for angiogenesis post-wound healing, through to the proliferative phase of limb regeneration (Fig. 1A). We measured blastema area every two days post amputation, beginning with 1 dpa. We found that blastema sizes in AV951-treated limbs were significantly smaller from 11 dpa onward (Fig. 1A-C), and this phenotype was consistent until 22 dpa, where control animals displayed digit formation (Fig. 1D-G). We stained limbs with Alcian blue to label cartilage in the regenerated limbs, and we observed no cartilage formation distal to the amputation plane in AV951-treated limbs (Fig. 1H, I). These data suggest a requirement for VEGF signaling following blastema formation and prior to chondrogenesis. To determine if the requirement for VEGF signaling is reversible, we treated limbs until 19 dpa and then removed treatment, since control treated limbs reached the end of blastema stage, where pallet stage begins, by 19 dpa. AV951-treated limbs were able to recover and regenerate until pallet stage, which they reached by 33 dpa (Fig. 1J), indicating that blastemas were able to regenerate as normal, once VEGF signaling was no longer blocked.

Figure 1. VEGF signaling is required for blastema growth and the effect of VEGF inhibition is reversible.

(A) schematic of experimental design; (B) representative DMSO-treated limb at 11 dpa (scale bar 100μm; arrowhead indicates blastema; dashed line indicates amputation plane); (C) representative AV951-treated limb at 11 dpa (scale bar 100μm; arrowhead indicates blastema; dashed line indicates amputation plane); (D) AV951-treated blastemas are significantly smaller at 11dpa (each data point represents the average of both forelimb blastemas from one animal; *p=<0.05); (E) representative DMSO-treated limb at 22 dpa (scale bar 100μm); (F) representative AV951-treated limb at 22 dpa (scale bar 100μm; arrowhead indicates blastema); (G) AV951-treated limbs show no regeneration at 22 dpa (each data point represents the average of both forelimb blastemas from one animal; **p=<0.01); (H) skeletal prep of representative DMSO-treated limb at 22 dpa (scale bar 250μm); (I) skeletal prep of representative AV95-treated limb at 22 dpa (scale bar 250μm); (J) AV951-treated blastemas display blocked regeneration compared to controls (n=6 animals for each group at each timepoint, average measurement of both forelimb blastemas); (K) AV951-treated blastemas can develop to pallet stage upon treatment removal. D: distal, P: proximal, dpa: days post-amputation(n=8 animals for control and n=4 animals for AV951 at each timepoint, average measurement of both forelimb blastemas).

VEGF signaling is primarily associated with angiogenesis during development and disease (Abhinand et al., 2016; Benedito et al., 2012; Bergers and Benjamin, 2003; Hogan and Schulte-Merker, 2017; Shin et al., 2016a, 2016b) and vasculature has been observed in salamander blastemas (Rageh et al., 2002; Smith and Wolpert, 1975). To determine whether angiogenesis might be impaired following treatment with AV951, as has been shown in zebrafish (Chimote et al., 2014; Savage et al., 2019), we analyzed vascular marker expression in amputated limbs. We performed in situ hybridization chain reaction (HCR) targeting Cd34 and measured the total length of contiguous endothelial cells across multiple sectioned blastemas per animal. We observed a significant reduction in Cd34⁺ vasculature in blastema tissue distal to the amputation plane (Fig. 2A-C), suggesting AV951 treatment blocks angiogenesis during axolotl limb regeneration. Small blastemas formed, even in the absence of vasculature. We sought to determine whether true blastema cells formed in these rudimentary blastemas by labeling cells for Kazald2 (Kzd2), a previously identified highly specific blastema cell marker (Bryant et al., 2017b; Leigh et al., 2018). Both treatment groups displayed Kzd2⁺ cells located within the blastema region (Fig. 2D-E). Since a blastema formed but did not grow, we wanted to determine whether proliferation was impaired in AV951-treated blastemas. We therefore administered a short pulse of EdU to label cells in S-phase, and we observed a significant reduction in EdU⁺ nuclei within the blastema of AV951-treated limbs, but no significant difference in EdU⁺ nuclei in tissue more proximal than the amputation plane (Fig. 2 F-I). This suggests that while VEGF signaling is not required for blastema cell specification nor for blastema cell coalescence at the amputation site, blastema cells do require VEGF signaling to sufficiently proliferate once the blastema forms.

Figure 2. VEGF signaling is required for blastema vascularization and blastema cell proliferation but not for blastema cell specification.

(A-A’) DMSO-treated blastemas display vascularization beyond the plane of amputation (scale bar 50μm; arrowhead indicates Cd34⁺ vasculature; dashed line indicates amputation plane); (B-B’) AV951-treated blastemas display little vascularization beyond the plane of amputation (scale bar 50μm; arrowhead indicates Cd34⁺ vasculature; dashed line indicates amputation plane) (C) AV951-treated blastemas display significantly-reduced vascularization in post-amputation area (**p=<0.01); (D-E) Blastema cell specification, highlighted by Kazald2 expression, is not affected by AV951 treatment (scale bar 50μm; dashed line indicates amputation plane); (F-G) AV951-treated blastemas display reduced proliferation; (H-I) reduced proliferation is specifically observed in the distal blastema, not in tissue proximal to the amputation (*p=<0.05).

VEGF signaling has a direct function in proliferation of non-vascular tissue during regeneration.

Our results suggested that blastema cells require VEGF signaling for proliferation. One possible explanation of this result is that the requirement is indirect: blastema cells may be dependent upon growth or survival signals supplied by nascent blood vessels, which are also forming during this time, and vascular regeneration requires VEGF signaling. An alternative, but not mutually-exclusive, possibility is a direct requirement for VEGF signaling within nascent blastema cells themselves. Many blastema cells are derived from fibroblast-like cells (Currie et al., 2016; Kragl et al., 2009; Muneoka et al., 1986). As the in vivo limb is a complicated composite of many cell types, we therefore turned to cultured AL-1 axolotl limb fibroblast cells (Roy et al., 2000) to address the possibility that axolotl limb fibroblasts may directly depend on VEGF signaling for their migration and proliferation. AL-1 cells can contribute to blastemas upon transplantation into regenerating limbs (Yu et al., 2023), indicating some blastema cell properties are conserved in these cells. We used AL-1 cells to test the hypothesis that axolotl limb fibroblasts might require VEGF signaling for migration and proliferation. We cultured AL-1 cells and performed a scratch assay, wherein cells were removed from a confluent well and regrowth was quantified, similar to previous methods (Oliveira et al., 2022). AV951-treated AL-1 cells displayed reduced recovery over the scratch compared to control treated AL-1 cells (Fig. 3 A-I), suggesting they may directly respond to VEGF signaling. However, whether this is due to migration or proliferation defects - or potentially both - was not clear. We therefore performed an EdU assay to determine whether VEGF signaling could impact AL-1 proliferation. We observed that AL-1 cells treated with AV951 displayed impaired proliferation compared to controls (reduced EdU⁺ nuclei numbers) (Fig. 3 J-L). Our data suggest that AL-1 cells, like blastema cells, respond directly to VEGF signaling via proliferation. The lack of scratch closure observed may be due to reduced proliferation, which correlates with our findings in vivo. To confirm this genetically, we transfected AL-1 cells with either a GFP-only control plasmid (CAGGs:eGFP), or a plasmid containing a dominant negative VEGFR2, which we cloned into the CAGGs:eGFP plasmid to generate CAGGs:eGFP-dnVEGFR2 (referred to as dnVEGFR2) (Millauer et al., 1996). We calculated the percentage of EdU⁺ GFP⁺ cells in each condition and found a significantly lower percentage of cells transfected with dnVEGFR2 were EdU+ compared to cells transfected with GFP-only control (Fig. 3 M-Q). Altogether, these data suggest that non-vascular cells, such as fibroblast-derived blastema cells, may respond to VEGF signaling in order to proliferate and grow, and that this may be independent of VEGF-mediated angiogenesis.

Figure 3. VEGF signaling is required for proliferation of axolotl limb fibroblast cells.

(A) representative DMSO-treated AL-1 cells at 0 days post scratch (dps) (yellow dashed line indicates periphery of scratch at 0dps; scale bar 500 μm); (B) representative DMSO-treated AL-1 cells at 1 dps (yellow dashed line indicates periphery of scratch at 0 dps; blue dashed line indicates periphery of scratch at 1 dps; scale bar 500 μm) (C) representative AV951-treated AL-1 cells at 0 dps (yellow dashed line indicates periphery of scratch at 0 dps; scale bar 500 μm); (D) representative AV951-treated AL-1 cells at 1dps (yellow dashed line indicates periphery of scratch at 0dps; blue dashed line indicates periphery of scratch at 1 dps (scale bar 500 μm); (E-H) inset images from A-D showing closure in different conditions (scale bar 250 μm); (I) AV951-treated AL-1 cells display a significantly reduced recovery from scratch (**p=<0.01); (J) representative image of EdU staining in DMSO-treated AL-1 cells (scale bar 500 μm); (K) Representative image of EdU staining in AV951-treated AL-1 cells (scale bar 500 μm); (L) AV951-teated AL-1 cells display significantly reduced proliferation compared to controls (**p=<0.01); (M) Representative image of EdU staining in AL-1 cells transfected with GFP control plasmid (yellow arrowheads represent GFP⁺ EdU⁺ cells; white arrowheads represent GFP⁺ EdU^- cells; scale bar 100μm); (N) Representative image of EdU staining in AL-1 cells transfected with dnVERGR2 plasmid (white arrowheads represent GFP⁺ EdU^- cells; scale bar 100μm); (O) AL-1 cells transfected with dnVERGR2 plasmid display significantly reduced proliferation compared to control transfected cells (*p=<0.05).

Loss of VEGF signaling maintains blastemas in a wound healing-like state and impairs expression of EMT-like genes.

Since our data indicate that VEGF signaling is required for blastema progression, we wanted to determine how loss of VEGF signaling affected regeneration at the transcriptional level. We performed Illumina bulk RNA sequencing on blastemas obtained from AV951-treated and control-treated animals at 11 dpa and we performed differential gene expression analysis. We identified 3287 differentially-expressed genes between the two conditions (Fig. 4A; extended data file 1–2). Our PCA analysis showed clustering of the two sample groups (Fig S1A), and we observed differential gene expression (Fig. 4A; Fig. S1B) highlighting a downregulation of genes associated with angiogenesis and Wnt signaling, among other pathways, which was corroborated by Kegg analysis (Fig. S2; extended data file 3).

Figure 4. Loss of VEGF signaling induces ectopic expression of early stage wound healing genes and a loss of EMT transcription factor expression.

(A) heat map displaying differentially expressed genes in AV951-treated blastemas compared to controls; (B) control treated blastemas display minimal Areg expression at 11 dpa (scale bar 100μm); (C) AV951-treated blastemas display ectopic Areg expression at 11 dpa (white arrowheads; scale bar 100μm); (D-F) Snai1 and Snai2 expression in mesenchymal cells of control blastema at 60x magnification (scale bar 50μm); (G-I) Snai1 and Snai2 expression in mesenchymal cells of AV951-treated blastema at 60x magnification (scale bar 50μm); (J-K) AV951-treated blastemas display reduced expression of Snai2, but not Snai1 (**p=<0.01).

We also observed differential expression of genes associated with epidermal differentiation and wound healing, including Amphiregulin (Areg) and Epiregulin (Ereg). We previously observed significant ectopic expression of Areg in the epidermis of limbs subjected to repeated amputation which failed to regenerate, and we demonstrated ectopic expression of Areg in naïve axolotl limbs is sufficient to block a first instance of limb regeneration (Bryant et al., 2017a). Interestingly, this overexpression phenotype was similar to that observed in AV951-treated blastemas. We used hybridization chain reaction in situ hybridization (HCR) to label Areg expression in blastemas and found that AV951-treated animals displayed ectopic expression of Areg in the wound epidermis at 11dpa (Fig. 4B-C), which is typically extinguished within a few hours after amputation in naïve axolotls regenerating a limb for the first time (Bryant et al., 2017a). This finding validates our bulk RNA-seq findings (Fig. 4B-C) and suggests that VEGF signaling activity might be required for blastemas to transition out of the wound healing phase of regeneration and into the tissue-building phase of regeneration.

We also observed significant differential expression of genes associated with epithelial-to-mesenchymal transition (EMT), including Zeb1, Zeb2, Twist2, Twist3 and Snai2 (Stemmler et al., 2019), which displayed downregulation in AV951-treated blastemas. Genes regulating EMT have been implicated in limb regeneration previously (Glotzer et al., 2022; Kawakami et al., 2006; Li et al., 2021; Sader et al., 2019; Wischin et al., 2017; Yu et al., 2023), (Fig. 4A). Our data suggest that VEGF signaling may induce proliferation, at least in part, via the induction of EMT-like processes. The role of EMT-like processes during axolotl limb regeneration is not fully clear, but they have been hypothesized to be required for the migration of blastema progenitor cells out of parent tissues and niches during blastema formation (Kim and Whited, 2024; Li et al., 2021; Sader et al., 2019). We used HCR to corroborate our transcriptomic data and observed a decrease in the number of cells expressing Snai2 in AV951-treated limbs, while Snai1 expression was unchanged (Fig. 4D-I). This might suggest Snai2 may be part of the mechanism by which blastema cells proliferate. Since the role of EMT during axolotl limb regeneration is not fully established, VEGF signaling presents an attractive target for both understanding and manipulating EMT-like processes in future studies.

Discussion

Here we present data which is consistent with a model wherein VEGF functions in two roles: firstly, the regulation of angiogenesis during regeneration, which may exert an indirect effect on blastema growth; and secondly, direct regulation of blastema cell proliferation. Multifaceted functions for VEGF signaling have been observed in development and disease, including fibroblast proliferation. Autocrine VEGF signaling from fibroblasts has been observed in some cancers (Abou Faycal et al., 2018; Goggins et al., 2023; B. Huang et al., 2019; Masood et al., 2001), where it promotes both fibroblast proliferation and angiogenesis. Moreover, cancer-associated fibroblasts (CAFs) have been shown to secrete VEGF to promote angiogenesis (Pape et al., 2020). Intriguingly, it has long been postulated that tumors co-opt molecular pathways that are used constructively by highly-regenerative species to promote blastema formation and growth (Wong and Whited, 2020), and VEGF-mediated regulation of blastema proliferation may in fact be a part of this process. In fact, senescent fibroblasts promote proliferation during axolotl limb regeneration (Yu et al., 2023) and cultured senescent fibroblasts secrete VEGF and can induce angiogenesis in mammalian cell culture (Coppe et al., 2006). It is therefore possible that blastema fibroblasts secrete VEGF to maintain proliferation in the absence of vasculature. Understanding the regulation of blastema growth in this context may therefore have translational potential for future therapies to induce blastema growth in human amputee patients.

We also showed, upon VEGF inhibition, an increase in expression of genes associated with the early stages of wound healing, which are typically extinguished after a few hours such as Areg, until 11 dpa. This is significant because ectopic expression of Areg impairs axolotl limb regeneration (Bryant et al., 2017a). Increased expression of wound-healing-related genes, which are typically associated with early stages of the regeneration process, might suggest a failure to resolve the wound healing phase of regeneration in the absence of VEGF signaling. This could be due to an absence of vasculature, or VEGF may function directly during the resolution of wound healing in salamanders. Interestingly, Areg has been implicated in the activation of VEGF signaling in cancers (Y.-W. Huang et al., 2019; Wang et al., 2017), suggesting similar requirements for angiogenesis may exist during blastema growth. Indeed, we observed decreased angiogenesis at the same time point as we observed ectopic Areg expression.

We observed a decrease in expression of genes associated with epithelial-to-mesenchymal transition (EMT). EMT-related genes have been characterized during limb regeneration in axolotls; Zeb, Twist, and Snail family gene expression has been observed at different timepoints during regeneration (Sader et al., 2019). Our data highlight a downregulation of many genes associated with EMT, including Zeb1, Zeb2, Twist2, and Snai2 in the absence of VEGF signaling. We also observed a decrease in Prdx2 expression, which has previously been associated with blastema cells that have taken on an intermediate fate between mesenchymal and epithelial states (Li et al., 2021). We did not observe a difference in Snai1 expression in our data, but we did observe a difference in Snai2 expression, suggesting the possibility that VEGF signaling may interact most strongly with Snai2. Our findings therefore identify VEGF as a promising target for future regenerative studies and further study is required to understand how VEGF signaling regulates EMT-like processes and the activity of EMT-related genes, such as Snai2.

AV951 has been shown to inhibit angiogenesis via inhibition of VEGF receptor tyrosine kinase function. AV951 inhibits VEGF receptors at a lower IC50 than previously tested inhibitors (Vegfr1 = 30 nM, Vegfr2 = 6.5 nM, and Vegfr3 =15 nM), meaning our dosage (80nM) likely completely blocks Vegfr function. PTK787, for example, inhibits VEGF receptors at 77 nM (Vegfr1), 37 nM (Vegfr2) and 664 nM (Vegfr3); previous studies used PTK787 at 100 nM, which might mean that Vegfr3 was insufficiently inhibited (Drevs et al., 2002; Ritenour and Dickie, 2017). Importantly, IC50 values are usually defined in monolayer culture, which may be inappropriate for three dimensional structures (Berrouet et al., 2020). This suggests higher doses may be required for in vivo studies. However, AV951 has potential off-target effects, including inhibition of c-Kit and Pdgfrb function (Hepgur et al., 2013). Inhibition of Pdgfr function during axolotl limb regeneration produces a similar outcome as what we observed via AV951 treatment (Currie et al., 2016). Pdgfrb expression was downregulated following AV951 treatment, suggesting some portion of the phenotype could be attributed to partial loss of Pdgfr signaling. Genetic manipulation is essential for future studies investigating the precise functions of VEGF signaling and the VEGF receptors that transduce these functions during axolotl limb regeneration. We began this process by transfecting AL-1 cells with dominant-negative VEGFR2 and observing a decrease in proliferation. However, genetic knockout for each VEGF receptor, both individually and in tandem, would strengthen these findings.

Taken together, our data support a model in which VEGF signaling serves as a critical regulator of the transition from the early wound healing phase to tissue replacement during axolotl limb regeneration. In the absence of VEGF signaling, we observed a sustained expression of early wound-healing genes such as Areg, suggesting a failure to fully resolve the initial wound-healing response. Concurrently, we observed reduced expression of EMT-related genes and Prdx2, indicative of impaired cellular plasticity and stalled progression toward a regenerative blastema. These data imply that VEGF not only facilitates angiogenesis but may also act as a molecular switch to terminate wound healing and enable blastema outgrowth and patterning. This dual role of VEGF is consistent with its multifunctional activity in development and cancer and may be essential for coordinating the dynamic shifts in cell state required for successful regeneration. Our work predicts that modulating VEGF signaling post specification of blastema cells may be a means whereby these important limb progenitor cells can be harnessed for regenerative therapies.

Materials and methods

Animal husbandry and surgical procedures.

All animal experimentation was approved by and conducted in accordance with Harvard University’s Institutional Animal Care and Use Committee (Protocol # 19–02-346). Age and sizematched (7cm ± 10%) leucistic axolotls were used for all animal experiments. Animals were housed in 40% Holtfreter’s Solution at a salinity of 3000 μS, 7.6 pH, and a room temperature of 19°C. For all experiments, animals received bilateral forelimb amputations at the mid stylopod. Animals were anaesthetized in tricaine before amputation and recovered in 19mM sulfamerizine overnight before being returned to Holtfreter’s solution (with or without pharmacological treatment/vehicle) for the remainder of the experiment.

Pharmacological treatments

For blastema AV951 treatments, animals were treated at 80nM AV951 to inhibit the function of VEGFR1, VEGFR2, and VEGFR3 without acute toxicity. AV951 was prepared as a 10mM stock in DMSO and a volume was diluted into Holtfreter’s solution to a concentration of 80nM into which animals were immersed for treatment. An equivalent volume of DMSO was added to Holtfreter’s solution for control treatments. For EdU analysis, animals were subject to a 16-hour EdU pulse. EdU was prepared as a solution of 5μg/μl and diluted to 0.1 μg/μl prior to injection. Animals received 2mg/kg EdU. For AL-1 cells, AV951 was used at a concentration of 50nM. EdU treatment was administered at 10mM for 16 hours.

Plasmid transfection

AL-1 cells were transfected with 0.5μg of plasmid per well in a 24 well plate. pCAGGs:eGFP plasmid used was used as previously (Whited et al., 2013). dnVEGFR2 was generated based on existing plasmid containing mouse dominant negative VEGFR2 (Millauer et al., 1996), obtained from Addgene (#65249) and cloned into pCAGGS:eGFP to generate a fused GFP-VEGFR2.

RNA extraction

Blastema tissue samples were collected and immediately placed in 500 μl of Trizol reagent (Thermo Fisher, 15596026) in 1.5 ml microcentrifuge tubes. Tissues were homogenized with a handheld drill homogenizer (Pro Scientific, PRO200) for 20 seconds until fully homogenous. 100 μl of Chloroform (Sigma, C2432) was added to each sample and mixed thoroughly by hand shaking the tube. After 5 minutes, tubes were spun at 12,000 rpm at 4°C in a microcentrifuge to promote phase separation. Aqueous phase was collected carefully and placed in a new 1.5 ml microcentrifuge tube, then equal volume of 100% Ethanol was added. The mixture was then added to columns from the RNA Clean & Concentrator Kit (Zymo, R1013), and the subsequent cleanup steps from the protocol were followed. Samples were eluted in 15 μl of RNAse-free water, and measured on a Nanodrop One (Thermo Fisher, ND-ONE-W) to confirm A260/A280 above 2.0 and A260/A230 values were above 1.8. Extracted total RNA was measured for quality on an Agilent Tapestation 4200 (Agilent, G2991BA) to confirm RIN scores were above 9.0. Qubit RNA BR Assay Kit (Invitrogen, Q10210) was used to measure RNA concentration on a Qubit 4 Fluoremeter (Thermo Fisher, Q33238). Samples were normalized to 5ng/ul in 60 ul and submitted to the Bauer Sequencing Core Facility at Harvard University.

RNAseq analysis

Library preparation was done using the Kapa mRNA HyperPrep kit and sequenced on an Illumina NextSeq 1000 at 50 bp read length through the Bauer Sequencing Core Facility at Harvard University. Fastq files were received from the Core and paired-end sequencing reads were quality-trimmed using Trimmomatic (v0.39). Kallisto (v22.04) was used to align reads to the AmexT_v47 transcriptome assembly (Nowoshilow et al., 2018). To find the differentially expressed genes between DMSO and AV951 conditions, DESeq2 was used (Love et al., 2014) as a package in RStudio (v2023.03.0). Genes with fewer than 10 counts in at least 3 samples were filtered out prior to analysis. Significance was determined using an adjusted p-value (padj) threshold of < 0.05.

Histology

Tissues were fixed in 4% paraformaldehyde in PBS, dehydrated into methanol in series, and stored at −20 °C. Upon use, samples rehydrated and were left overnight in 30% sucrose in PBS. Samples were embedded in optimal cutting temperature compound (O.C.T.) and stored at −80 °C. Samples were sectioned at a thickness of 16 μm.

In situ hybridization chain reaction

In situ hybridization chain reaction (HCR) was performed as previously described (Lovely et al., 2022). Briefly, sectioned blastemas were washed in 2X SSC, treated with tissue clearing solution, washed in 2X SSC, incubated at 37°C in probe hybridization buffer (Molecular Instruments) before incubating overnight at 37°C in probe solution (1:200 dilution of probe in probe hybridization buffer). The next day, slides are washed in probe wash solution (Molecular Instruments) before washing in 5X SSCT and incubating at room temperature in amplification buffer (Molecular Instruments) and incubating overnight at room temperature in hairpin solution (2μl of each hairpin in amplification buffer). Slides are then washed in 5X SSCT, stained with DAPI (1:1000 in PBS), and mounted for imaging.

Highlights.

• Inhibition of VEGF signaling blocks axolotl limb regeneration.

• Proliferation of blastema cells is impaired in the absence of VEGF signaling.

• VEGF inhibition impairs resolution of wound healing.

• Genes associated with EMT are downregulated in the absence of VEGF signaling.