Aerobic glycolysis is important for zebrafish larval wound closure and tail regeneration

Abstract The underlying mechanisms of appendage regeneration remain largely unknown and uncovering these mechanisms in capable organisms has far‐reaching implications for potential treatments in humans. Recent studies implicate a requirement for metabolic reprogramming reminiscent of the Warburg effect during successful appendage and organ regeneration. As changes are thus predicted to be highly dynamic, methods permitting direct, real‐time visualisation of metabolites at the tissue and organismal level would offer a significant advance in defining the influence of metabolism on regeneration and healing. We sought to examine whether glycolytic activity was altered during larval fin regeneration, utilising the genetically encoded biosensor, Laconic, enabling the spatiotemporal assessment of lactate levels in living zebrafish. We present evidence for a rapid increase in lactate levels within min following injury, with a role of aerobic glycolysis in actomyosin contraction and wound closure. We also find a second wave of lactate production, associated with overall larval tail regeneration. Chemical inhibition of glycolysis attenuates both the contraction of the wound and regrowth of tissue following tail amputation, suggesting aerobic glycolysis is necessary at two distinct stages of regeneration.


| INTRODUCTION
While some organisms have the ability to heal scarlessly and regenerate fully functional tissues as adults, others possess this ability only in the early developmental stages. Understanding the underlying cellular and molecular processes responsible for successful regeneration may provide essential clues for the development of novel clinical therapies that will promote a better healing and regenerative outcome in humans.
Accumulating evidence indicates metabolism influences complex tissue and cellular processes, including cell differentiation and cell behaviour, and an interest in the role of cell metabolism in regeneration is undergoing a revival, driven largely by the development of new techniques that facilitate addressing the link between metabolism and tissue repair and regeneration. Metabolites, such as lactate, have been reported to act as second messengers in cell signalling, 1 and a switch from oxidative phosphorylation (OXPHOS) to glycolysis is involved in epithelial to mesenchymal transitions (EMTs), which are important for blastema formation in gecko limb regeneration 2 and in cancer metastasis. 3,4 Work in C. elegans has shown that reduction of mitochondrial activity has positive effects on ageing, 5 and multiple studies have linked a switch to glycolytic metabolism to the proliferative potential of stem cells. [6][7][8] Metabolism further plays an important part in cell identity and differentiation in a variety of settings, including immune cells and neurons. 9,10 Thus, metabolism plays a wider role in physiology than simply energy production. Given that EMT, proliferation, and differentiation are all processes important for regeneration and wound healing, investigating the potential roles for metabolic reprogramming during regeneration and how these are regulated, may provide insight into how cellular metabolism could be hijacked to facilitate regeneration in humans.
The Warburg effect describes the phenomenon of aerobic glycolysis, in which cells preferentially up-regulate processing of glucose through the conventionally anaerobic pathway of glycolysis and fermentation while decreasing their mitochondrial activity, regardless of the lower energy yield and availability of oxygen. 11 This strategy was originally discovered in cancer cells, but has since been implicated in multiple highly proliferative systems, putatively allowing glycolytic and pentose phosphate pathway (PPP) intermediates to support macromolecule synthesis for new cells. 12 Since regeneration is highly dependent on cell proliferation and growth, one might expect regenerating cells to employ the Warburg effect to provide for the requirements of forming the new tissues of the regenerate. This appears to be the case in multiple regeneration models. The gene profiles of regenerating Xenopus tails and adult zebrafish hearts show an upregulation of glycolytic genes and a corresponding down-regulation of mitochondrial genes. 13,14 This switch to glycolysis has been linked to cell proliferation during cardiomyocyte regeneration. 14 While varied transcriptomic analyses have suggested that metabolic reprogramming plays a critical role during tissue and appendage regeneration, there is a critical need for improved methods that will facilitate the direct assessment of Warburg-like metabolism during regeneration with temporal and spatial resolution. Recent developments in genetically encoded sensors for various metabolites have advanced the field of metabolic research and have great potential for exploitation in the zebrafish, due to its impressive regenerative capacity, 15,16 combined with the transparency of the embryos. Here we aimed to test the potential of a genetic ratiometric Förster resonance energy transfer (FRET)-based genetic sensor, named Laconic, which is responsive to varying lactate levels. 17 Given that rising lactate levels can be used as a measure of aerobic glycolysis/ Warburg-like metabolism, we aimed to determine whether this sensor could be used to assess metabolic reprogramming during two models of zebrafish larval fin regeneration, namely after fin fold and tail amputations.
We also aimed to ask whether altering metabolic reprogramming, using chemical inhibitors targeting glycolysis or lactate dehydrogenase, affected the speed or efficiency of wound closure and/or fin and tail regeneration.

| Cloning
Laconic/pcDNA3.1(À) was a gift from Luis Felipe Barros (Addgene plasmid #44238; http://n2t.net/addgene:44238; RRID:Addgene_44238). 17 The Laconic genetic sensor in Laconic/pcDNA3.1(À) was cloned into the pCS2+ vector and the p3 vector from the pTransgenesis system, 18 using standard restriction digest and sticky end recombination methods. In some cases, complimentary primers were designed and annealed to produce short sticky end fragments that were inserted into constructs in order to generate additional complementary restriction sites. For specific restriction enzymes, inserts, and buffers used, see Table S1.
For the transgene cassettes, the modular cloning system pTransgenesis 18 based on the Gateway system of cloning 19

| mRNA microinjections
Wild-type strain AB zebrafish embryos were injected at the one-cell stage into the cell cytoplasm with 1 ng laconic sensor mRNA in nuclease-free water with phenol red. Laconic sensor mRNA was synthesised from pCS2 plasmids linearised with NotI (NEB), with mMESSAGE mMACHINE SP6 Transcription Kit (Ambion) and purified with lithium chloride (LiCl) extraction.

| Generation of transgenic lines
Wild-type strain AB zebrafish embryos were injected at the one-cell stage into the cell cytoplasm with 25pg tol2 mRNA and 25 pg circular plasmid in 1 nL. Tol2 mRNA was synthesised from pT3-Tol2 linearised with SmaI (NEB) with mMESSAGE mMACHINE T3 Transcription Kit (Ambion) and purified with LiCl extraction. Injected embryos with the strongest expression of mosaic GFP were grown into adults and outcrossed to screen for germline transmission.

| Biochemical lactate assay
A commercially available colorimetric lactate assay kit (MAK064, Sigma Aldrich) was used and protocol adapted for embryonic samples. Lactate in the sample reacts with the enzyme mix provided in the kit, the product of which interacts with the supplied lactate probe to produce colour (A 570 ) and fluorescence (excitation/emission = 535/587 nm). We chose to identify lactate concentration by measuring the colorimetric product of the enzymatic reaction with lactate at an absorbance of 570 nm.
Samples were prepared by flash freezing on dry ice and macerating 25 dechorionated eggs or embryos with a plastic micropestle in 45 μl 2:2:1 acetonitrile:methanol:dH 2 O at À20 C or pre-chilled on dry ice. Samples were then centrifuged at 4 C at 15000rcf for 10 min, the supernatant collected into a new tube and stored at À20 C until use in the assay. 5 μl of the embryo supernatant was used per reaction.
A standard curve was set up using known concentrations of a lactate standard (0, 2, 4, 6, 8 and 10 nM per reaction) with the addition of 5 μl 2:2:1 to each reaction in order to control for any background or change in enzyme activity caused by the buffer.
Triplicate reactions were set up otherwise according to manufacturer instructions, with a minimum of three biological repeats. Reaction incubation time was extended to 3 h, and absorbance at 570 nm (A 570 ) was read on a microplate reader (BioTek Synergy H1) in triplicate to give a total of nine readings per sample.

| Fin fold and tail amputations
2 days post fertilisation (dpf) embryos were mounted in 1% low melting agarose (Invitrogen 16520100) supplemented with 0.04% MS-222 (tricaine, Sigma Aldrich E10521) on a glass microscope slide for imaging with an upright microscope or in a 35 mm glass-bottomed dish (Thermo Scientific Nunc) for imaging with an inverted microscope, and imaged pre-amputation. Amputations were made while embryos were mounted in agarose with either a size 10 or 15 scalpel blade, the agarose surrounding the fins excavated, and the embryos covered with media. Fin fold amputations were performed just distal to the tip of the notochord, and tail amputations were oriented using the pigment gap and transected just distal to the circulatory loop of the caudal vein. Images were then taken at various timepoints post amputation with the embryos being de-mounted from the agarose and kept in 1X egg water or 1X E3 at 28 C between imaging timepoints of longer than one hour. For experiments where imaging immediately post amputation was not required, 2dpf embryos were not mounted in agarose, and instead amputated in a droplet of 1X egg water supplemented with 0.04% MS-222 on a glass microscope slide, transferred to 1X egg water with or without drug treatment within 5 min of amputation, and maintained at 28 C.

| Microscope sample preparation
Embryos were visually screened using a fluorescent dissecting microscope for transgenic expression and dechorionated manually with forceps. Embryos were mounted in 1% low melting agarose (Invitrogen 16520100) supplemented with 0.04% MS-222 (tricaine, Sigma Aldrich E10521) on a microscope slide for imaging with an upright microscope, or in a 35 mm glass-bottomed dish (Thermo Scientific Nunc) for imaging with an inverted microscope.

| Microscopy
Images were acquired on an AxioImager.M2 upright microscope

| Image analysis
All processing of images for calculating ratio and measuring fluorescence or ratio was conducted in Fiji (version 2.0.0). Average background was subtracted and threshold applied to remove the remaining background, then Laconic ratio was calculated by dividing the 428 nm emission channel by the 485 nm emission channel using the Image Calculator function.
Pseudo-colouring was applied using Lookup Table "

| Statistical analysis
GraphPad Prism 8 was used for statistical testing, with sample numbers exceeding 6 in all experiments, and each experiment was replicated three or more times. Column or grouped statistics and analyses of differences between means were implemented for all data sets. For column statistics, two-tailed unpaired t-tests with assumed Gaussian distribution were used. Two-way ANOVA was used with Sidak's multiple comparisons test to compare means between groups. All data are presented as mean ± s.d., and differences were considered significant to * at P < 0.05, ** at P < 0.01, *** at P < 0.001, and **** at P < 0.0001. Not significant (ns) was considered P ≥ 0.05, 95% confidence interval.

| Laconic can be used to monitor lactate levels in zebrafish embryos and larvae
The genetically encoded biosensor, Laconic, was originally developed and tested on cells in culture 17 and has since been used in mouse brains, 23 but has not been used in whole organisms, such as the zebrafish. The FRET-based sensor is composed of a lactate binding region, the bacterial transcription factor LldR, linked with the fluorescent proteins mTFP and Venus. 17 Upon binding of lactate, a conformational change decreases the FRET efficiency of energy transfer from the donor chromophore, mTFP, to the acceptor chromophore, Venus ( Figure 1A). By exciting mTFP and measuring the emission from both mTFP and Venus, one can form a ratio to depict the changes in lactate with temporal and spatial resolution. The mTFP/Venus ratio (the Laconic ratio) increases with lactate levels ( Figure 1B).
We devised a positive control, whereby we treated embryos 17 h post fertilisation (hpf) with antimycin A (AA), a mitochondrial OXPHOS inhibitor, which acts to drive glucose into aerobic glycolysis and thus conversion into lactate. We first established the action of AA was indeed increasing lactate levels by measuring the concentration of lactate in treated embryos using a commercial biochemical assay kit. When comparing lactate concentrations in embryos before and after 10 min of treatment with either AA or DMSO, we found that AA-treated samples gave significantly higher readings of lactate concentrations ( Figure S1). From this, we were confident that AA treatment would provide a satisfactory method to test the efficacy of the Laconic sensor.
In order to confirm Laconic reports lactate dynamics in the zebrafish embryo, we injected in vitro transcribed laconic mRNA into onecell stage zebrafish embryos and imaged them before treatment and after one hour of treatment, either with AA or DMSO vehicle control.
The Laconic ratio increased significantly in response to AA but not DMSO ( Figure S1). Thus, we concluded that Laconic can successfully report lactate levels in zebrafish embryos.
We then generated a transgenic line (Tg[ubb:laconic] lkc1 ) expressing Laconic under the control of the ubiquitin B promotor (ubb) 24 in order to visualise lactate levels over the course of larval fin regeneration. This Laconic transgenic line showed a higher Laconic FRET ratio following treatment with AA, indicating it reliably reports lactate levels in embryos and larvae ( Figure S2). As further confirmation, we performed a second positive control using an alternate mitochondrial inhibitor, sodium azide (NaN 3 ), which acts on complex IV of the electron transport chain, 25 similarly blocking OXPHOS, and driving the cell toward glycolysis. As with AA, one hour of treatment with NaN 3 increased the Laconic ratio significantly compared to PBS control treatment. Furthermore, as NaN 3 is a reversible inhibitor, we also measured Laconic ratio after 24 h of recovery following washout of the drug and observed lactate levels had returned to those of controls ( Figure S2).

| Lactate levels increase transiently immediately post-larval fin fold amputation
We next assessed whether lactate levels change following two types of larval fin injury, namely following distal fin fold amputation, where only distal epidermal fin tissue was excised, and following tail amputations, where many additional tissues are transected, including the notochord and spinal cord ( Figure 2A). Both types of injury induce three similar phases of tissue response and regeneration (wound healing, proliferation, and outgrowth and differentiation, Figure 2B) that are mostly comparable to adult fin regeneration, albeit on a much faster time scale. [26][27][28] Upon amputation, most regenerative responses form a highly proliferative structure termed the blastema. 29 Since regeneration involves the regrowth of lost tissue, it would stand to reason that larval tail regeneration would be dependent on the generation of significant amounts of new biomass. As such, appendage regeneration is a good candidate for the Warburg effect. 13 Thus, we aimed to image lactate levels, as a proxy for the Warburg effect, during regeneration following larval zebrafish fin fold and tail amputations utilising the Tg[ubb:laconic] lkc1 line.
We found that the Laconic ratio increased immediately following distal fin fold amputation, peaking within the first five minutes post amputation (mpa) and returning to control levels by 50 mpa

| Inhibition of lactate production impairs wound contraction following fin fold amputation
Given that the elevated lactate levels we measured are coincident with the rapid wound healing phase following injury, we decided to test whether aerobic glycolysis might play a crucial role during the actomyosin-driven wound healing phase, which takes place within the first ten minutes after injury 27 (also see: Supplementary Movie 1).
Glycolysis is able to produce ATP more rapidly than OXPHOS, which is why the fastest contracting muscle fibres, which are also actomyosin based, are largely glycolysis-based in their metabolism. 30 Potentially this is an explanation for the transient burst of lactate and glycolysis activity after injury, as a strategy for swiftly producing large quantities of ATP to fuel the energy-intensive contraction of the actomyosin cable during wound closure.
We tested this hypothesis by inhibiting aerobic glycolysis and assessing the effects on the actomyosin cable contraction at the wound. To do this, we transiently inhibited the activity of lactate dehydrogenase (LDH) using the competitive inhibitor, sodium oxamate, 31 during the wound healing phase and up to an hour post fin fold amputation, and took images and movies during the wound healing process ( Figure 4A). LDH converts pyruvate to lactate and regenerates NAD+, permitting continued glycolysis independent of mitochondrial activity. We reasoned that using chemical inhibitors These data suggest that rapid glycolysis activity immediately following amputation is required for the rapid contraction of the wound margin To test the importance of rapid glycolysis for actomyosin activity, we stained oxamate-treated and control amputated embryos for phosphorylated non-muscle myosin (pNM) and actin using immunohistochemistry and phalloidin, respectively. Phosphorylation of non-muscle myosin was unaffected by treatment with lower levels of oxamate, and only slightly reduced with higher concentrations (Figure 5A, B). However, actin at the wound border was significantly diminished with both high and low concentrations of oxamate treatment. Myosin phosphorylation requires only a single ATP to donate the phosphate group for each myosin, and therefore is not the most energetically demanding process, whereas active contraction requires one molecule of ATP for each myosin stroke cycle. 34 We propose that it is this contraction that requires the use of glycolysis. It may be that actin stabilisation and condensation at the site of action requires activity of myosin, and the lack of actin remodelling may indicate a loss of active contraction. Differences were considered significant to * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, and ns P ≥ 0.05.

| Wound contraction in fin fold amputation is not dependent on oxidative phosphorylation
Our findings suggest that wound contraction is highly reliant on glycolysis, but this does not preclude a similar requirement for OXPHOS during wound contraction. Therefore, we endeavoured to determine whether inhibition of OXPHOS using NaN 3 similarly affected wound contraction following larval tail fin amputation.
We performed fin fold amputations on 2dpf Tg[ubb:laconic] lkc1 embryos and immersed them immediately in NaN 3 treatment, and then measured Laconic ratios after one-hour of treatment ( Figure 6A).
We found no significant difference in treated versus control embryos, although the NaN 3 embryos trended toward having higher levels of lactate ( Figure 6B,C). Additionally, there was no difference in fin widths or wound contraction ( Figure 6D), thus we concluded that a reduction of mitochondrial activity does not affect the rapid wound contraction phase following injury.
By 1hpa control embryos have reduced in Laconic ratio, while NaN 3 treated embryos continued producing elevated lactate levels ( Figure 6B,C). After washing out the drug at this point and allowing embryos to regenerate their fin folds, we imaged both conditions and found that there was no difference in Laconic ratio or fin regrowth at five days post amputation (dpa) (Figure 6B-D).
This suggests an overabundance of lactate and a transient loss in mitochondrial OXPHOS activity during the early wound healing phase has no consequence on either wound closure or overall regeneration, while a reduction in lactate and glycolysis activity negatively impacts the rapid wound healing phase. Moreover, the apparent lack of requirement for OXPHOS in wound contraction suggests that this process is predominantly dependent on glycolysis.

| Lactate is elevated in the notochord bead following tail amputations
After seeing that lactate levels rise dramatically, but only transiently, during the wound healing phase following fin fold amputation, we asked whether there was any evidence for metabolic reprogramming during the later regeneration phase following tail amputation ( Figure 2B). During both larval fin fold and tail regeneration, a proliferative second phase occurs; however, tail amputation additionally involves the formation of a "notochord bead", which arises from extruding notochord sheath cells and displays high rates of proliferation, 27 which could be linked to changes in metabolism. Overall, early lactate dynamics were similar following both fin fold and tail amputations, with a near instant and rapid increase in Laconic ratio ( Figure 7A,B). However, the initial elevated lactate persisted following tail amputations (24 h post amputation (hpa)) when compared to distal fin fold amputations. In particular, we found sustained higher Laconic  (i)). Students' t-test to calculate significance, n = 24. Scale bar represents 200 μm. Differences were considered significant to * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, and ns P ≥ 0.05. ratios in the notochord bead until 48 hpa ( Figure 7C). This blastemalike structure begins formation at 12 h post amputation and blastema genetic markers begin to be lost after 48 hpa 25,27 ; therefore, higher lactate levels correlated with the presence of the proliferative notochord bead and blastema-like structure.
3.6 | Inhibition of glycolysis prevents successful tail regeneration, but not fin fold regeneration Differences were considered significant to * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, and ns P ≥ 0.05. regenerating zebrafish hearts 14 and genes involved in glycolysis and the PPP are significantly up-regulated during Xenopus tadpole tail regeneration. 13 Along with the elevation of aerobic glycolysis in the tail notochord bead noted previously, this prompted us to assess the potential role for aerobic glycolysis during zebrafish larval tail regeneration versus fin fold regeneration.
We thus treated 2 dpf amputated Tg[ubb:laconic] lkc1 embryos over the course of regeneration with 2DG, a competitive inhibitor of hexokinase and glucose-6-phosphate isomerase, two critical enzymes in the glycolytic pathway 35 and analysed the resulting tail lengths at 5 dpa ( Figure 8A). To confirm the efficacy of 2DG on lactate production, we measured lactate levels at 120 hpa after treatment for the first 72 hpa and verified that 2DG caused a significant decrease in Laconic ratio, and therefore lactate level, in both distal fin fold and tail amputation conditions at 120 hpa and in 7 dpf unamputated controls ( Figure 8B,C). Given that 2DG can potently reduce lactate levels in both our regeneration assays, we asked if regeneration of either the fin fold or tail was compromised in its presence. Tail amputated embryos did not regenerate when treated with 2DG ( Figure 8D), resulting in a significantly shorter tail length at 120 hpa ( Figure 8E). In contrast, distal fin fold amputations, though averaging a shorter length of regrowth than controls, were not significantly affected Finally, to support the findings that attenuating LDH activity affected tail regeneration, but not distal fin regeneration, we also treated embryos with the alternative LDH inhibitor, GNE-140. As with oxamate, a lower concentration of the inhibitor was required (40 μM) to sustain viability over the five days of treatment. Also like oxamate, GNE-140 treatment significantly reduced the regenerative ability of the larvae following tail amputation ( Figure S4). Again, though in general regenerative length was shorter than controls, GNE-140 did not significantly affect regeneration following distal fin fold amputation ( Figure S4).
Thus, we hypothesise that aerobic glycolysis plays an essential role during larval tail regeneration, which involves the regeneration of many tissues, including the spinal cord and notochord, but aerobic glycolysis is not similarly required for distal fin fold regeneration, which is not associated with long-term elevation of lactate levels and is not dependent on a blastema-like structure. Whilst effective, the Laconic sensor suffered from low fluorescence intensity in our transgenic line, which limited its sensitivity.
There may also be additional interference from autofluorescence, especially in cells with high pigment or yolk content, and the multicellular nature of organisms, which additionally affects the sensitivity of the sensor. We were able to confirm the ability of Laconic to report lactate levels with a biochemical lactate assay, and, in the future, combining genetically encoded metabolite sensors with biochemical assays, whole embryo metabolomics and/or MALDI mass spectrometry imaging will produce a complementary array of data, with the sensors providing a broader depiction of the temporal and spatial changes in metabolism while metabolomic approaches supplying a more comprehensive dataset of information of a large range of metabolites at given timepoints.
Other in vivo studies have also demonstrated the applicability of genetically encoded biosensors for measuring metabolite dynamics in zebrafish. The iNap1 sensor was used to show that NADPH levels decrease following embryonic fin amputation and co-localises with hydrogen peroxide (H₂O₂). This was interpreted as a result of dual oxidase (DUOX) activity, which consumes NADPH while generating H₂O₂. 36 Activity of the pentose PPP, specifically the enzyme glucose-6-phosphate dehydrogenase, is the main contributor to NADPH production, 37 and thus iNap sensors could in future studies be also utilised as an indicator of the Warburg effect alongside Laconic.

| A role for aerobic glycolysis in early wound healing and formation of a blastema-like structure
In both distal larval fin fold and tail amputations the rapid increase in lactate levels within minutes following amputation occurs prior to the proliferative phase of regeneration, and correlates with the actomyosin contraction of the wound margin. Our chemical inhibitor experiments suggest that this rapid rise in lactate levels is necessary for wound contraction. More specifically, inhibition of LDH results in failure of the wound to contract and an attenuation of actin re-organisation and concentration at the wound border through purse-string action of myosin on actin. This reduction was not seen upon inhibition of mitochondrial OXPHOS activity, suggesting this process is primarily dependent on glycolysis. However, we did not directly measure OXPHOS activity nor did we measure the speed of wound closure at timepoints prior to 10 mpa, thus we are unable to rule out the possibility that OXPHOS inhibition does not result in a brief acceleration or delay in wound contraction prior to 10mpa.
One might ask whether this rapid rise in lactate levels immediately after amputation is the result of aerobic glycolysis or anaerobic glycolysis? Two lines of evidence point toward aerobic glycolysis. The first is that, as mentioned above, inhibition of OXPHOS has little effect on wound closure, thus rapid oxygen consumption due to OXPHOS is unlikely to be occurring. The second is that, based on evidence in Xenopus tadpoles, where there is a rapid rise in oxygen levels at the wound immediately following tadpole tail amputations, 38 it is unlikely that the wound margin in our zebrafish embryos is becoming anoxic or hypoxic immediately after injury. However, given we did not specifically assess oxygen levels, we cannot conclude this with certainty.
A similar role for aerobic glycolysis has been shown in the brains of mice, whereby upon neuronal excitation, glycolysis temporarily exceeds the rate of oxidative metabolism needed to provide for the rapid increase in energy demand. 39 Furthermore, the process of enucleation in erythrocytes requires contraction of an actomyosin ring and is prevented when aerobic glycolysis is blocked by inhibition of the glycolytic enzymes glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or LDH. 40 Thus, we propose that aerobic glycolysis provides a means of rapid ATP production necessary for driving the energy consuming process of actomyosin-mediated wound contraction minutes after amputation, showcasing the growing importance of considering the impact of metabolism on the regulation of biomechanics.
Further elucidation of this hypothesis would benefit from the use of another biosensor to measure ATP levels, such as PercevalHR, 41 to confirm whether ATP levels drop dramatically specifically at the wound margin when aerobic glycolysis is inhibited during the rapid wound healing phase. An alternative role for lactate may be for stabilisation of actomyosin cable itself, as previous work has shown that sodium lactate is able to stabilise actomyosin in vitro against temperature perturbations. 42,43 While we have found that both larval fin amputations and tail amputations lead to a rapid metabolic shift toward aerobic glycolysis, which is necessary for the rapid wound healing phase, it remains unknown whether other forms of injury are also associated with a similar shift in metabolism. However, given the conserved involvement of actomyosin / purse-string mediated contraction during rapid wound healing events in both single cell and multicellular injury models, 44 we would predict that such a metabolic shift may be a general hallmark of rapid wound healing mechanisms that are energy demanding, but evidence to support this assertion will require further investigation.
During the subsequent regeneration phases following wound healing, aerobic glycolysis, as indicated by localised elevated lactate levels, is once again implicated in the notochord bead/blastema during tail regeneration. In contrast, we saw no significant increase in lactate levels at these later stages during fin fold regeneration. The larval fin fold "blastema" does not play a specific role in proliferation as it does in adult fin regeneration, and instead proliferation occurs in a more spatially distributed manner. 27 Tail amputation, however, is more akin to a canonical appendage regenerative response, in that multiple tissues must be replenished, including the notochord, spinal cord and skeletal muscle. The blastema-like structure formed following tail amputations expresses genes typically associated with the blastema in other regenerative organisms and is partly made up of extruded notochord cells, creating the "notochord bead". 28 We show that the blastema-like notochord bead has elevated lactate levels as early as 3 hpa and continues until 48 hpa. The raised lactate levels correlate temporally with the blastema, returning to control levels after 48 hpa as the regenerant enters the third phase of regeneration, characterised by differentiation and progressive scaling back of proliferation. 26,27 Other work has also shown elevated levels of glycolysis gene expression and decreased mitochondrial activity during zebrafish heart regeneration. Inhibition of glycolysis with 2DG resulted in a reduction of proliferating cardiomyocytes, 14 indicating this metabolic switch to glycolysis is required for regrowth. Increased expression of glycolysis genes has also been observed in zebrafish following larval tail amputation, with inhibition with 2DG resulting in abnormal blastema formation, 45 and we additionally find that activity of the glycolytic enzymes hexokinase and LDH are required for larval tail regeneration. Thus, aerobic glycolysis appears to be required for successful regeneration through the formation or output of the blastemalike notochord bead. As 2DG acts very early in glycolysis, it is able to impact other glucose metabolic pathways, including the pentose PPP and the hexosamine biosynthetic pathway (HBP). Though we demonstrate a reduction of lactate levels with 2DG and oxamate treatment, suggesting the importance of glycolysis specifically during regeneration, it is possible that other pathways may also be involved in zebrafish tail regeneration or that these inhibitors are impacting on pathways unrelated to glycolysis, which then impact on regeneration.
Indeed, requirement of the HBP during larval appendage regeneration has been proposed. 45 The regrowth of the epidermis following fin fold amputations, however, does not require the function of these glycolytic enzymes and achieved regrowth comparable to controls despite glycolytic inhibition. It is unclear at this point if this different reliance on aerobic glycolysis between the two amputation models reflects diversity in the constituent cell types being regenerated or the differing anabolic needs for regeneration of the fin fold, versus overall regeneration of many tissue types. Intriguingly, recent findings suggest that a similar metabolic switch also occurs following adult fin regeneration in zebrafish, and inhibition of this switch results in failure in blastema formation in the adult fin as well. 46 Thus, our work suggests that aerobic glycolysis is important at two distinct points following injury: the first being within minutes following injury, during the rapid wound healing phase and the second during the tail regeneration phase. Though a blastema is typically highly proliferative, there is an absence of raised lactate levels in any region aside from the notochord bead during the proliferative phase of fin and tail regeneration. Aside from being a product of the Warburg effect, lactate may also have a direct effect on blastema formation and function, such as acting as a second messenger. For example, lactate has recently been shown to mediate magnesium uptake into the mitochondria, 1 which in turn has been reported to have a stimulatory effect on oxidative metabolism and may affect mitochondrial calcium flux. 47 The downstream targets and signalling stimulated by lactate in this instance remain unknown, their elucidation a possible direction for future studies. Other future work could also look into whether proliferative cells are reduced in glycolysis-inhibited tail amputations, as is the case in zebrafish heart regeneration. 14 The underlying mechanisms governing metabolic reprogramming during tail regeneration remain unknown. Both hypoxia-inducible factor-1α (HIF1α) signalling and the embryonic form of pyruvate kinase (PKM2) have been implicated in the switch of induced pluripotent stem cells to glycolytic metabolism, leading to their de-differentiation. 48,49 More broadly, there is increasing evidence that hypoxic conditions and reactive oxygen species (ROS) influence glycolytic switching. HIF1α signalling is also sufficient for inducing reprogramming to glycolytic metabolism in mouse embryonic stem cells 50 and is known to have a positive effect on glycolysis, such as in cancer 51 and macrophages. 52 Further, H₂O₂ has been shown to positively regulate glycolysis in cancer cells. 53 Illuminating the molecular pathways involved in successful regeneration, such as the relationship between H₂O₂ and glycolysis, will assist in determining the logic of metabolic reprogramming in different phases of regeneration. Zebrafish imaging approaches, combining an expanding genetically encoded biosensor toolbox with high regenerative capacity, offer a unique system to determine principles of metabolic programming in regeneration.