Chemical Genetics of Regeneration: Contrasting Temporal Effects of CoCl₂ on Axolotl Tail Regeneration

Abstract

Histone deacetylases (HDACs) regulate transcriptional responses to injury stimuli that are critical for successful tissue regeneration. Previously we showed that HDAC inhibitor romidepsin potently inhibits axolotl tail regeneration when applied for only 1-minute post-amputation (MPA). Here we tested CoCl₂, a chemical that induces hypoxia and cellular stress, for potential to reverse romidepsin inhibition of tail regeneration. Partial rescue of regeneration was observed among embryos co-treated with romidepsin and CoCl₂ for 1 MPA, however, extending the CoCl₂ dosage window either inhibited regeneration (CoCl₂:0-30 MPA) or was lethal (CoCl₂:0-24 hours post amputation; HPA). CoCl₂:0-30 MPA caused tissue damage, tissue loss, and cell death at the distal tail tip, while CoCl₂ treatment of non-amputated embryos or CoCl₂:60-90 MPA treatment after re-epithelialization did not inhibit tail regeneration. CoCl₂-romidepsin:1 MPA treatment partially restored expression of transcription factors that are typical of appendage regeneration, while CoCl₂:0-30 MPA significantly increased expression of genes associated with cell stress and inflammation. Additional experiments showed that CoCl₂:0-1 MPA and CoCl₂:0-30 MPA significantly increased levels of glutathione and reactive oxygen species, respectively. Our study identifies a temporal window from tail amputation to re-epithelialization, within which injury activated cells are highly sensitive to CoCl₂ perturbation of redox homeostasis.

Introduction

Amphibians and fish are well-suited for studies of tissue regeneration, not only because they regenerate damaged tissues, but also because they are amenable to chemical screening. Chemicals can be readily delivered to aquatic fish, frogs, and salamanders for the purpose of altering developmental, cellular, and molecular mechanisms that regulate tissue regeneration^1-6. A variety of chemicals with relatively well-characterized targets and mechanisms of action have been shown to block regeneration. These include chemicals that inhibit primary developmental signaling pathways, including retinoic acid, BMP, TGFα, FGF, Wnt, and HSP90^4,7-12. When a chemical with a known biological activity is shown to block regeneration, it provides strong evidence that the biological activity is required for regeneration. Further resolution of regeneration processes can be achieved by precisely varying the timing of chemical delivery and dose, and by coupling chemical screening with transcriptional analyses^6,13.

Histone deacetylases (HDACs) remove ε-N-lysine acetyl groups from proteins and this changes biophysical properties and protein function (reviewed by ^14,15). HDACs are known to deacetylate several hundred different proteins, including proteins that directly or indirectly function in the regulation of transcription. HDACs are integral components of multiprotein corepressor complexes that are recruited to specific genomic sites by transcription factors. Deacetylation of histone lysine residues is generally associated with chromatin compaction and transcriptional repression. Inhibitors of HDACs (HDACi) block histone deacetylation and this causes chromatin to attain a more relaxed state that is generally associated with transcriptional activation. However, HDACi can block transcription through acetyl modifications of transcriptional factors and by blocking RNA synthesis at enhancer sites and Pol II mediated-transcriptional elongation¹⁶.

HDACi have been shown to block or alter regeneration in several organisms, including planeria¹⁷, zebrafish¹⁸, and Xenopus¹⁹. A recent chemical genetic screen of epigenetic modulators found that the HDAC inhibitor romidepsin, administered for only 1-minute post amputation (1 MPA), blocks axolotl tail regeneration⁶. Also, romidepsin treatment up-regulated over 100 genes that positively and negatively regulate gene expression, representing approximately 36% of the total number of genes that were expressed more highly in romidepsin-treated embryos. These results suggest a requirement for HDAC activity to regulate initial transcriptional responses at the time of injury. In support of this hypothesis, cited2, a negative regulator of Hif1α transcriptional activation^20-23 and inflammation^24-26 was significantly up-regulated at 3 hpa in romidepsin-treated embryos while leptin, a gene transcribed in response to hypoxia in zebrafish embryos and a marker of axolotl limb progenitor cells was significantly down-regulated^27,28. Also, romidepsin significantly up-regulated txnip, a negative regulator of thioredoxin and cellular redox state; over-expression of txnip by romidepsin is associated with catastrophic oxidative stress and cell death in tumor cells²⁹. These patterns of gene expression implicate HDAC activity in regulating critical transcriptional responses that are associated with hypoxia and oxidative stress stimuli at the time of injury.

In this study, we further investigated romidepsin-inhibition of axolotl tail regeneration. To determine if romidepsin attenuates responses to injury stimuli, we pursued studies using cobalt chloride (CoCl₂), a chemical that is often used in cell culture studies to induce hypoxia and oxidative cellular stress ^30-32. We found that acute, romidepsin-CoCl₂ co-treatment for 1 MPA partially rescued tail regeneration but longer dosage times were inhibitory to both wound healing and tail regeneration. Interestingly, CoCl₂ negatively affected wound healing and regeneration if applied during the first 30-60 minutes of regeneration, but not after 60 minutes. These results suggest that acute 1 MPA CoCl₂ can partially counter romidepsin inhibition of tail regeneration, but continuous CoCl₂ treatment up until wound closure (re-epithelialization) is inhibitory to regeneration. Our results show that strategic application of chemical inhibitors can reveal critical windows and biological processes that are essential for wound healing and regeneration.

Results

CoCl₂ can partially rescue romidepsin-inhibited tail regeneration

Previously we reported that romidepsin inhibited axolotl tail regeneration when delivered within a narrow treatment window⁶. Specifically, when developmental stage 42 embryos were treated with 10 μM romidepsin for 1 MPA, regeneration was inhibited at 7-days post amputation (DPA). After this time, some regeneration was observed with variable outgrowth of tailfin tissue and abnormal patterning. Also, romidepsin altered the expression of many transcription factors and cited2 was identified as a highly, differentially expressed gene. We reasoned that romidepsin-mediated upregulation of cited2 might attenuate a necessary hypoxia mediated gene expression response. This is because CITED2 can effectively outcompete Hif1α for a protein binding domain on CBP/p300 that activates transcription in response to hypoxia^20-23. To test this hypothesis, we performed amputations and co-treated embryos for 1 MPA in solutions containing romidepsin and chemicals (CoCl₂, DFO, or 1,4-DPAC) that stabilize Hif1α and promote cellular responses to hypoxia. We first validated that embryos pre-treated for 24 hours with romidepsin prior to amputation regenerated normally, and embryos treated for 1 MPA with romidepsin showed inhibited tail regeneration at 7 DPA (Table 1; Figure 1). We then co-treated embryos with romidepsin (10 μM) and either CoCl₂, DFO, or 1,4-DPAC for 1 MPA. While romidepsin treated embryos showed the same non-regenerative outcome at 7 DPA, romidepsin- CoCl₂ (≥ 5 mM) co-treated embryos, but not DFO or 1,4-DPAC co-treated embryos, regenerated significantly more tail tissue (Table 1; Figure 1). In contrast, 1 MPA CoCl₂ treatment alone did not significantly affect regenerative outcome at 7 DPA. These results show that CoCl₂ can partially rescue the inhibitory effect of romidepsin on tail regeneration, possibly through a hypoxia-independent mechanism.

Table 1.

Drugs that were tested for effect on tail regeneration.

Single Drug Treatments
Drug	Dose	Delivery	Result at 7 DPA
Romidepsin	10 μM	24 HPT	No effect
Romidepsin	10 μM	0-1 MPA	Inhibition
CoCl2	0.1, 0.5, 1, 5, 7.5, 10 mM	0-1 MPA	No effect
CoCl2	0.01, 0.1, 0.5 mM	0-3 HPA	No effect
CoCl2	1 mM	0-3 HPA	Inhibition
CoCl2	7.5 mM	0-3 HPA	Inhibition
CoCl2	7.5 mM	30 MNA	No effect
CoCl2	7.5 mM	30 MPA	Inhibition
CoCl2	7.5 mM	60-90 MPA	No effect
Combinatorial Drug Treatments with 10 uM Romidepsin
Drug	Dose	Delivery	Result at 7 DPA
CoCl2	0.1, 0.5, 1 mM	0-1 MPA	Inhibition
CoCl2	0.1, 0.5, 1 mM	0-3 HPA	Inhibition
CoCl2	7.5 mM	0-15 MPA	Inhibition
CoCl2	7.5 mM	0-30 MPA	Inhibition
CoCl2	7.5 mM	0-24 HPA	Lethal
CoCl2	5, 7.5, 10 mM	0-1 MPA	Partial Regeneration
CoCl2	7.5 mM	0-5 MPA	Partial Regeneration
DFO	1 mM	0-1 MPA	No effect
DFO	5 mM	0-1 MPA	No effect
1,4 DAPC	10 μM	0-1 MPA	No effect
1,4 DAPC	5 μM	0-1 MPA	No effect
1,4 DAPC	1 μM	0-1 MPA	No effect
1,4 DAPC	0.5 μM	0-1 MPA	No effect

HPT = hours pre-treatment before amputation, HPA = hours post amputation, MPA = minutes post amputation, MNA = minutes no amputation.

Figure 1. CoCl₂ partially rescues axolotl tail regeneration.

A) 7 DPA Embryos treated continuously with 0.01% DMS0. B) 7 DPA embryos treated continuously with 10 μM romidepsin. C) 7 DPA embryos treated with 10 μM romidepsin and 10 mM CoCl₂ for 1-minute post amputation (MPA). D) 7 DPA embryos treated with 10 μM romidepsin for 1 MPA. E) Higher magnification image of embryo treated with 10 μM romidepsin for 1 MPA. F) Higher magnification image of embryo treated with 10 mM romidepsin and 10 mM CoCl₂ for 1 MPA. G) 7 DPA embryos treated with 10 μM romidepsin and 10 mM CoCl₂ for 1 MPA. H) Quantification of regenerated tail tissue among embryos treated with romidepsin and or CoCl₂. Embryos that were treated with 10 μM romidepsin and CoCl₂ (5, 7.5, and 10 mM) regenerated significantly (* ANOVA, p < 0.05) more tissue than embryos that were only treated with 10 μM romidepsin.

Microarray analysis of CoCl₂-romidepsin co-treatment

Transcriptional changes associated with CoCl₂-romidepsin 1 MPA treatment were characterized using a custom microarray³³. Embryos were administered tail amputations and treated in a solution with both 7.5mM CoCl₂ and 10 μM romidepsin for 1 MPA. These concentrations of CoCl₂ and romidepsin were used in all subsequent experiments. At 3 HPA, one mm of distal tail tip tissue was collected for RNA isolation and microarray analysis. The resulting data were analyzed along with microarray data from Voss et al⁶, who generated gene expression estimates for 0 and 3 HPA embryos that were treated with 10 μM romidepsin for 1 MPA. Only 28 of over 20,000 probesets (genes) on the microarray were identified as significantly differentially expressed (FDR adjusted prob. < 0.05; > 1.5 fold average difference), with all showing higher expression in the co-chemical treatment than the romidepsin treatment (Supplemental Table 1). Strikingly, almost all of these genes, the majority of which are transcription factors, were previously reported to be significantly upregulated at 3 and 12 HPA during axolotl tail⁶ and limb regeneration³⁴ (Figure 2). Twelve of these genes encode protein interactants of the HIF1α subunit (cited2, dnajb1, fos, hspa1l, hspa8, hsp90aa1, isg15, jun, plk3, sqstm1, nr4a1, txnip). Other genes in this list encode proteins that function in regulation of transcription (egr1, klf2, klf10, sik1, zfanda2, zfp36, cyr61), cell signaling (adm, pim3, errfi1, dusp5, f2rl1), and cell growth (areg). Relative to the 1 MPA romidepsin treatment, CoCl₂-romidepsin altered expression of the majority of these genes in the direction of controls, with several genes exceeding the values of controls by > 2 fold (e.g. hsp90aa1, hspa8, hspa1l). The expression of heat shock proteins is a well characterized adaptive transcriptional response to various forms of cellular stress. Also, CoCl₂-romidepsin co-treatment yielded expression values for several genes (cited2, isg15, nr4a1, zfand2a) implicated in regulation of inflammation that deviated more from controls than was observed for romidepsin-treated embryos. Thus, CoCl₂ augmented romidepsin-dampened transcriptional responses that associate with appendage regeneration, but also increase transcriptional responses associated with hypoxia, cellular stress and inflammation.

Figure 2. Genes identified as significantly differentially expressed between romidepsin and CoCl2/romidepsin treated embryos at 3 hours post amputation.

Fold change is relative to expression level measured at the time of amputation. Control refers to non-chemically treated axolotls that were previously administered tail⁶ and limb amputations³⁴.

Prolonged CoCl₂ treatment inhibits regeneration

Because CoCl₂ was shown to partially rescue regeneration in 1 MPA co-treatment assays, we reasoned that longer post-amputation treatment times would further enhance regeneration. Instead, we found that longer CoCl₂ treatment times were either lethal or inhibited tail regeneration at 7 DPA. For example, extending the CoCl₂-romidepsin treatment window by 15 and 30 MPA did not rescue regeneration and a 24 HPA extension was lethal (Table 1). To determine if longer CoCl₂ treatment times were inhibitory to regeneration, we assayed for the effect of CoCl₂ alone, without romidepsin. Surprisingly, CoCl₂ (≥ 1 mM) inhibited regeneration at 7 DPA if treatment times were 0-30 MPA, but non-inhibitory if treatments were applied after 60 minutes (Figure 3A, B). Embryos that were treated for 0-30 minutes showed evidence of tissue loss, hemorrhage, and cell death at the distal tail tip, as early as 3 HPA (Figure 3C, E-H). In contrast, embryos that were treated with CoCl₂ from 60-90 MPA (Figure 3D), as well as non-amputated embryos that were treated for 0-30 MPA, did not show tissue disruption or cell death, and amputated embryos regenerated normally. Histologically, re-epithelization in embryos treated for 0-30 minutes was severely perturbed, with clumps of red blood cells in the injury area (Figure 4). Thus, CoCl₂ only affected wound healing and regeneration if it was applied after injury and during the first 30 MPA.

Figure 3. Contrasting effects of CoCl₂ on axolotl tail regeneration.

A) 7 DPA embryos treated with 7.5 mM CoCl₂ for 0-30 MPA presented blunt tail tips, consistent with inhibition of tail regeneration. B) 7 DPA embryos treated with 7.5 mM CoCl₂ for 60–90 MPA regenerated their tail tips. C) Methylene blue staining of damaged tissue in a 3 HPA embryo treated with 7.5 mM CoCl₂ for 0-30 MPA. Arrows indicate areas of tail tissue loss. D) Methylene blue staining of a 3 HPA embryo treated with 7.5 mM CoCl₂ for 60-90 MPA. E) 16 HPA embryos treated with 7.5 mM CoCl₂ for 0-30 MPA presented abnormal tail tips. F) Comparison of 0-30 MPA vs 60-90 MPA CoCl₂ treated embryos at 16 HPA. G) Higher magnification image of 7.5 mM CoCl₂ 0-30 MPA treated embryo at 16 HPA. H) Propidium iodide staining reveals dead cells in 0-30 MPA CoCl₂ treated embryo tail tips at 16 HPA.

Figure 4. Histological evaluation of the effect of CoCl2 treatment on regenerating tails at 3 hpa.

Compared to control embryos (A), CoCl2 treatment (B) disrupted wound epidermis formation. Clumps of blood cells (yellow arrow heads) indicate hemorrhage in CoCl2-treated embryo tails. SC: spinal cord, NC: notochord, WE: wound epidermis.

Microarray analysis of CoCl₂ treated embryos

Microarray analysis was performed to investigate transcriptional changes associated with CoCl₂ treatment. Embryos were administered tail amputations and one group was treated in CoCl₂ for 0-30 MPA (CoCl₂:0-30 MPA) while a second group was treated in CoCl₂ for 60-90 MPA (CoCl₂:60-90 MPA). One mm of distal tail tip tissue was collected at 3 HPA for RNA isolation and microarray analysis. A total of 210 microarray probesets (Figure 5; Supplemental Table 2) were identified as significantly differentially expressed (FDR adjusted prob. < 0.05; > 1.5 fold average difference). A relatively large number of probesets that were expressed more highly in CoCl₂:0-30 MPA treated embryos annotate to genes that encode humoral proteins associated with red blood cells (hba, hba2, hbd, hbe, hbg, hbz, alas2, slc10a4, ca2, prdx2) and platelets/thrombocytes (gp1bb, adm, cyr61, serpinb10)²⁸. Other significant genes included cited2 and genes that were discovered in the CoCl₂-romidepsin microarray analysis above, including genes that encode protein interactants of the HIF1α subunit and heat shock proteins (txnip, dnajb1, jun, isg15, sqstm1, fos, nr4a1, hspa1a1, hsp90aa1). Genes known to be expressed in response to hypoxia were also identified, including adm, alas2, cdnk1b, ctgf, ddit4, egr1, and nr4a1. Additionally, significantly higher expression of sqstm1, nef2, and txnip, and significantly lower expression of keap1 further suggest that CoCl₂:0-30 MPA induced a reactive oxygen state as proteins encoded by these proteins act as molecular sensors to maintain cellular redox homeostasis³⁵. Higher expression of genes that regulate pH (ca2, slc26a4, slc4a1, atp6voa4), redox state (prdx2, ddit3), and positively regulate apoptosis (dusp1, mapt, nr4a1, tspo, sqstm1) were also observed. We also note that higher expression of heat shock proteins was inversely correlated with the expression of mRNA splicing factors; this is consistent with stress-dependent inhibition of splicing³⁶. Together, these data suggest that CoCl₂:0-30 MPA induced transcriptional changes consistent with hypoxia, heat shock response, redox homeostasis, and abnormal wound healing.

Figure 5. CoCl₂ treatment regime differentially affected gene expression.

CoCl₂ treatment for 0-30 MPA inhibited regeneration and was associated with increased transcription of genes that encode protein interactants of HIF1α, and genes that are typically expressed by red blood cells and platelets. CoCl₂ treatment for 60-90 MPA did not inhibit regeneration and was associated with increased transcription of matrix metalloproteinases, cell signaling pathway components, and genes that are typically upregulated during axolotl tail and limb regeneration. Rep = replicate.

Probesets expressed more highly in CoCl₂:60-90 MPA treated embryos, which presented normal tail tip architecture and regenerated successfully, annotate to loci that typically are upregulated during successful tail and limb regeneration^4,28,34,37 (Figure 5; Supplemental Table 2). These include multiple matrix metalloproteinases (adamts1, mmp19, mmp1a, mmp1b, mmp2, mmp3b, mmp3c, mmp3d) that function in extracellular matrix (ECM) remodeling, genes that encode or synthesize ECM components (fn1, has2), genes expressed in the wound epithelium (cyp26b1, dsc2, ereg, fosl2, krt17, tgm1), and markers of axolotl blastemal cells (has2, lep). Also, genes previously identified as upregulated during limb and tail regeneration were identified (e.g. ereg, fgfbp1, fn1, klf4, mas1, rrad, timp1), along with genes that mark limb blastema progenitor cells (lep, has2). Finally, genes associated with activation (TGFβ: inhbb, pmepa1, ugdh; FGF: fgfbp1) and repression (FGF: spry1; WNT: dact1) of key signaling pathways were expressed more highly and lowly, respectively. These gene expression changes are consistent with the non-inhibitory effect of CoCl₂:60-90 MPA on tail regeneration.

Reactive oxygen species (ROS) and reduced glutathione (GSH)

Cellular redox state describes the relative balance of systems that regulate reactive oxygen species (ROS) and antioxidant molecules³⁸. The transcription results above suggest that continuous CoCl₂ treatment before re-epithelialization tips the balance toward a more reactive oxidative state. To test this hypothesis, we quantified ROS in embryos for the following treatments: romidepsin:0-1 MPA, CoCl₂:0-1 MPA, romidepsin and CoCl₂:0-1 MPA, CoCl₂:0-30 MPA, CoCl₂:0-60 MPA, CoCl2:60-90 MPA, and non-treated controls. Relative to controls, ROS production was significantly higher for CoCl₂:0-30 MPA and CoCl₂:0-60 MPA treatments (Student’s T-test p < 0.05)(Figure 6). ROS production in embryos treated for only 1 MPA with romidepsin and/or CoCl₂, or embryos treated with CoCl₂ from 60-90 MPA, did not differ from controls (Student’s T-test p > 0.05). These results further suggest that continuous CoCl₂ treatment engenders an oxidative cellular state that is not conducive for proper wound healing or successful regeneration.

Figure 6. Effect of Romidepsin-CoCl2 co-treatment on ROS and GSH at 3 HPA.

(A) Representative images of ROS (top row) and GSH production (bottom row) at 3 HPA for control, Romidepsin:0-1 MPA, CoCl₂:0-1 MPA, Romidepsin and CoCl2:0-1 MPA, CoCl₂:0-30 MPA, CoCl₂:0-60 MPA, and CoCl₂:60-90 MPA. (B) Quantification of ROS and GSH production. Error bars represent standard deviations of the mean (N=7-8 embryos/group). *: indicates statistical significance (Student’s T-test P value < 0.01) compared to control.

Romidepsin is a pro-drug that is activated by intracellular antioxidants like GSH, which are highly abundant in cells. We reasoned that romidepsin and CoCl₂ treatment might alter levels of GSH and thus romidepsin availability. Given romidepsin’s apparent dose-dependent effect on transcription, this would provide an explanation for CoCl₂ partial rescue of regeneration. To investigate this possibility, we quantified GSH at 3 HPA for the same treatments that were tested for ROS production (Figure 6). Romidepsin:0-1 MPA significantly decreased GSH levels relative to controls. In contrast, GSH levels were significantly higher than controls in embryos co-treated with romidepsin and CoCl₂, and also embryos treated with CoCl₂ for 0-1 MPA, 0-30 MPA, and 0-60 MPA. GSH levels were not significantly increased in embryos that were treated from 60-90 MPA. These results show that CoCl₂ elevates GSH, a possible antioxidant response to CoCl₂-mediated oxidative stress.

Discussion

We followed up on results of a previous study⁶ that established HDAC activity as essential for axolotl tail regeneration. In that study, HDAC inhibitor romidepsin blocked regeneration at 7 DPA when embryos were only treated for only 1 MPA. Many differentially expressed transcription factors were identified in romidepsin-treated embryos at 3 HPA, consistent with alteration of HDAC-mediated transcriptional regulation. In particular, romidepsin significantly upregulated cited2, a gene implicated in regulation of hypoxia and inflammation associated transcriptional responses. We found that CoCl₂ treatment, which promotes hypoxia and oxidative cellular stress, can both rescue and inhibit axolotl tail regeneration. If CoCl₂ is administered along with romidepsin for 1 MPA, the inhibitory effect of romidepsin on tail regeneration is partially reversed, and regeneration is partially recovered at 7 DPA. However, if CoCl₂ is administered for a longer period of time post-amputation, it is lethal (0-24 HPA) or inhibits (0-30 MPA) tail regeneration. In contrast, CoCl₂ is not lethal and does not affect tail regeneration if it is administered after 60 MPA, or if it is administered to embryos that do not receive an amputation. Interestingly, CoCl₂-romidepsin co-treatment rescued the expression of some transcription factors that are associated with appendage regeneration but increased the expression of cited2 and heat-shock proteins (hsp90aa1, hspa8, hspa1l) well-beyond what was observed for romidepsin-treated embryos. Below we discuss these primary findings and our results that suggest hypotheses for CoCl₂-romidepsin rescue of regeneration and CoCl₂ disruption of cellular redox homeostasis. Also, we highlight a significant finding of our study- the identification of a temporal window from tail amputation to re-epithelialization, within which cells are highly sensitive to CoCl₂.

Possible mechanism of CoCl₂ rescue of regeneration

HDACs deacetylate several hundred different proteins, including proteins that directly or indirectly function in regulation of transcription¹⁵. Romidepsin is a prodrug that is activated within cells by antioxidants that act on an internal disulfide bond, releasing a thiol that binds to a zinc atom in the binding pocket of Zn-dependent HDACs^39,40. Antioxidants like glutathione are highly abundant in cells and thus romidepsin could potentially disrupt transcription in a diversity of cells that express class I HDACs. Microarray data⁶ indicate that hdac1 is one of the most highly expressed genes within axolotl tail samples (94^th percentile) and a recent single cell analysis found hdac1 expression to be more highly enriched in monocytes than any other cell type during adult axolotl limb regeneration. These observations, coupled with previous transcriptional results (and results of this study), predict that HDAC activity is essential for proper regulation of immune and inflammatory responses to injury stimuli.

We observed many significant transcriptional changes in romidepsin-treated embryos in our previous study, including genes that regulate cell cycle arrest and cell signaling through primary signal transduction pathways (FGF, BMP, WNT, etc), but we were especially struck by the significant up-regulation of cited2. CITED2 is a transcription cofactor that can act as an activator or repressor of gene expression, depending upon the transcriptional complex that it associates. CITED2 can potently interfere with the binding of HIF1α to the p300/CBP transcriptional coactivator complex, thus attenuating transcriptional responses to hypoxia^22,23. CITED2 also regulates pro and anti-inflammatory gene expression programs in murine macrophages²⁶. We note that cited2, like hdac1, is most highly expressed in macrophages during adult axolotl limb regeneration²⁸. It remains to be determined if cited2 and hdac1 are similarly co-expressed in cells that function in immunity and inflammation in axolotl embryos.

Under the assumption that cited2 up-regulation disrupts hypoxia and inflammatory transcriptional responses (and ultimately regenerative outcome), we treated embryos with CoCl₂ in attempt to bring these responses in line with transcriptional responses associated with successful regeneration. A very small number of transcription factors were identified as significantly differentially expressed between the romidepsin and romidepsin-CoCl₂ cotreatment, with 40% encoding HIF1α protein-interactants that function in the response to hypoxia as well as other cellular stressors. The majority of the transcriptional responses were moderated in the direction of untreated control embryos, consistent with rescue of at least some components of the initial transcriptional response to injury. Also, the majority of these genes are similarly up-regulated at the outset of axolotl limb regeneration, suggesting they comprise a HDAC-regulated, regeneration-specific transcription network that is responsive to injury stimuli, including hypoxia and oxidative stress (Figure 7A). It is most parsimonious to propose that CoCl₂ rescued transcriptional responses by decreasing the availability of antioxidants that activate romidepsin, as this is consistent with CoCl₂ generation of ROS³² and the dose-dependent effect of HDACi on transcriptional output⁶. However, we showed that 1 MPA CoCl₂ increased GSH levels at 3 HPA, which would be expected to increase activated romidepsin and the dampening effect of romidepsin on regeneration. Thus, it seems more likely that CoCl₂ altered the expression of specific HDAC-associated transcription factors, although it is not clear if CoCl₂ stabilized Hif1α mediated transcription or independently affected other transcription factors. Failure to replicate the regeneration inhibiting effect of CoCl2 with two additional hypoxia inducing factors tentatively supports the later mechanism. For example, CoCl2 has been shown to stimulate HSP90 expression in cardiomyocytes and this confers protection to cells after CoCl₂-induced chemical injury⁴¹. The significant upregulation of heat shock proteins by CoCl₂ in this study, well beyond that of control embryos, may have increased survival of a sufficient number of regeneration competent cells to mount a partial regenerative response. Clearly, more in depth studies of Hif1α stability and transcriptional regulation, and Hif1α-Heat Shock Factor 1 interaction are needed. Sensitive assays of cellular redox homeostasis and hypoxia will also be needed, along with finer temporal sampling, to better understand how CoCl₂ moderates redox stimuli and how these in turn alter HDAC-regulated transcriptional responses.

Figure 7. Models for transcriptional regulation and windows of CoCl₂ sensitivity after tail amputation.

(A) HDAC activity at the time of amputation is informed by injury stimuli, including hypoxia and oxidative stress, to regulate a regeneration specific, transcriptional network. (B) CoCl₂ inhibits tail regeneration if administered continuously before wound closure (re-epithelialization), but not after.

Our results clearly show that continuous CoCl₂ treatment for 0-30 MPA or 0-60 MPA affects ROS and GSH levels, and also genes expressed in response to cellular oxidative stress. Reactive oxygen species (ROS) are generated within the first 10 minutes of axolotl tail amputation and inhibition of ROS signaling inhibits regeneration⁴². However, CoCl₂ increased ROS beyond levels that mediate cell signaling pathways essential for regeneration, instead yielding levels of a highly reactive oxidative state. While we interpret the robust up-regulation of GSH by CoCl₂ as a compensatory response to oxidative stress, this proposed mechanism deserves further study.

A temporal window of CoCl₂ perturbation

After amputation injury, basal epithelial cells are rapidly activated to cover the exposed wound surface. Baddar et al⁴² reported that epithelial cells cover the wound surface of amputated embryo tail tips by 40 MPA. In light of our findings, CoCl₂ only increased ROS and GSH, and inhibited regeneration, when applied prior to re-epithelialization. Thus, the period of time from amputation till wound closure defines an injury state within which cells are highly sensitive to CoCl₂ perturbation (Figure 7B). This suggest that CoCl₂ specifically targets injury activated cells during the initial wound healing phase. While macrophages are known to be activated by injury stimuli, and they are essential for adult axolotl limb regeneration⁴³, immune cells in axolotl embryos are poorly described and it is possible that immune cellular responses are performed by injury activated, non-immune cells, at least prior to maturation of the immune system. However, we did find evidence of an impaired hemostatic response that is typically mediated by activated platelets. Significantly more transcripts for platelet-specific proteins were quantified from tails of CoCl₂ treated embryos and CoCl₂ induced a persistent wound pathology with embryos exhibiting abnormal tail tips at 7 DPA. CoCl₂ is known to impair hemostasis, possibly via chemical modifications of fibrinogen that disrupt platelet aggregation⁴⁴. Again, the effect of CoCl₂ was not only temporally precise, perhaps reflecting a narrow window of time when platelets and initial immune responses are activated, but also spatially precise as CoCl₂ only affected the amputated distal tail.

Temporal delivery of chemicals to investigate regeneration mechanisms

Our study shows that by manipulating dose and delivery time of chemicals to axolotl embryos, chemical effects can be detailed in ways that reveal novel biological insights and the temporal ordering of regeneration processes. It is interesting that both romidepsin and CoCl₂ do not systematically affect homeostatic processes, instead, their effects are specific to the injury state and to a very narrow, post-injury window of time. In this respect, our studies of romidepsin and CoCl₂ show that amputation injury induces cellular susceptibility to chemical modulation, thus revealing rapid changes in cellular phenotypes that are coincident with injury. Understanding the signals and mechanisms that rapidly induce regeneration-permissive phenotypes in highly regenerative models will be key to translating information to non-regenerative models.

Methods

Animal procedures

The use of pre-feeding stage axolotls does not require a protocol approved by the Institutional Animal Care and Use Committee (IACUC) at University of Kentucky, however embryos used in this study were treated according to the same ethical standards that apply to feeding axolotls under IACUC protocol 2017-2580. Embryos (RRID:AGSC_100E, AGSC_101E, AGSC_102E, ) were obtained from the Ambystoma Genetic Stock Center (RRID:SCR_006372).

Chemical Dosing Experiments

Developmental stage 42⁴⁵ axolotl embryos were manually hatched by removing the egg jelly and membrane, anesthetized in 0.02% benzocaine, and tail amputations were performed with a sterile razor blade to remove 2 mm (~20% of the body length) of the distal tail tip. Axolotl embryos were then distributed into 12-well microtiter plates containing chemicals and rearing water⁴⁶. Chemicals were dissolved in water (CoCl₂) or DMSO (romidepsin) and diluted to a stock concentration of 10 mM (0.1% DMSO). Chemical dosing experiments (Table 1) were replicated 2 or more times using 6-12 embryos per experiment. Romidepsin was purchased from Selleckchem (Catalog No. S3020), and CoCl₂ (Product No. 15862) and propidium iodide (Product No. P4864) were purchased from Sigma-Millipore. Embryos were imaged at various times post-amputation using an Olympus dissecting microscope with 0.5X objective lens and DP400 camera. Embryo survival and distal tail shape were used to classify chemicals as lethal, inhibitory, or having no effect on tail regeneration. Tailfin area was quantified using Olympus CellSens imaging software and these measurements were analyzed statistically using analysis of variance to determine if chemical treatments significantly affected tailfin regeneration. After performing these tests, pair-wise contrasts were performed to test for significant differences between each chemically treated group. Significance of F statistics was evaluated at p < 0.05.

Microarray analysis of CoCl₂

A microarray experiment was performed to determine the effect of 7.5 mM CoCl₂ treatment on axolotl embryo gene expression at 3 hours post amputation. For each treatment tested (0-30 MPA CoCl₂, 60-90 MPA CoCl₂, 10 μM romidepsin and 7.5 mM CoCl₂ 1 MPA, Control: 0.1% DMSO), tail amputations were performed on 36 embryos (2 mm removed from distal tip with a razor blade) and then embryos were placed into microtiter plates containing rearing water (Control) or rearing water with chemicals. Exactly 1 mm of the distal tail tip was removed from embryos at 3 HPA. Tissues from 12 embryos were pooled into a 1.5 ml tubes with 0.5 ml of RNA-later (Qiagen) to obtain three replicate pools per treatment. Tissue samples were maintained at 4° C in RNA later prior to RNA isolation using first the Trizol method and then a Qiagen minikit with on-the-column DNAse treatment of DNA. Microarray hybridization using an Ambystoma Affymetrix array³² was performed by the University of Kentucky Microarray Core Facility. The raw microarray data (.CEL files) were deposited in the GEO database (accession number GSE150947, reviewer access token: ghetkewkbpijvil). GeneChips were normalized using the affy R package⁴⁷ to accomplish robust multichip averaging (RMA)⁴⁸. Batch effects for the CoCl₂ and Romidepsin analyses were adjusted using Combat, available in the svaR package; batches were determined by date of microarray data generation. Differential expression analysis was conducted using the limma R package⁴⁹. RMA normalized signal intensity values were fit to a linear model and empirical Bayes smoothing applied to standard errors. Moderated t-tests were performed to identify probe sets that yielded significantly different average expression values as a function of treatment. These lists were further filtered using a false discovery rate of α = 0.05 and by requiring a 1.5-fold difference between treatment and control means. Heat maps were generated using Heatmapper⁵⁰ Genes identified as significantly differentially regulated were queried using PubMed, PMC, Panther⁵¹, and the NCBI Gene database to obtain functional information.

H&E Histology, in vivo ROS and GSH detection

Histological assessment of tissues and in vivo ROS imaging was performed as described previously⁴². Briefly, following anesthesia with 0.02% benzocaine, the distal 2mm of the tail tip was amputated (0 hpa) from Stage 42 embryos. Embryos were divided into seven groups (N = 7-8 embryos/group): control, Romidepsin:0-1 MPA, CoCl₂:0-1 MPA, Romidepsin and CoCl2:0-1 MPA, CoCl₂:0-30 MPA, CoCl₂:0-60 MPA, and CoCl₂:60-90 MPA. Romidepsin was tested at 10 μM and CoCl₂ was tested at 7.5 mM. ROS production was detected at 3 HPA in all groups by adding Dihydroethidium (DHE) (final concentration= 5 μM) to the exposure solution for two hours in darkness prior to imaging using an SZX16 Olympus microscope with 2X objective lens and DP400 camera. For detection of GSH, monocholorobimane was used at (final concentration= 10 μM). Embryos were divided into seven groups as described above. All embryos were imaged at 3 HPA.

ROS and GSH production quantification

Image analysis was performed using Imagej⁵² as described previously ⁴². An exact region of interest was drawn around the distal tail for each embryo and the gray pixel intensity was measured and background subtracted. The measurements were then averaged for embryos within each group (N= 7-8 embryos/ group) and the results for each treatment were compared to the control using Student’s T-test. The procedure was replicated to ensure precision of the approach.

Supplementary Material

Supplemental Table 1

Genes identified as significantly differentially expressed between Romidepsin and CoCl2/Romidepsin treated embryos at 3 hours post amputation. Average normalized log2 expression values are shown.

Supplemental Table 2

Genes identified as significantly differentially expressed between CoCl2:0-30 MPA and CoCl2:60-90 MP treated embryos at 3 hours post amputation. Normalized log2 expression values are shown for three biological replicate samples for each CoCl2 treatment.

Data Availability Statement.

The raw microarray data (.CEL files) were deposited in the GEO database (accession number GSE150947, reviewer access token: ghetkewkbpijvil).