Hif1α and Wnt are required for posterior gene expression during Xenopus tropicalis tail regeneration

Abstract

Regeneration of complex tissues is initiated by an injury-induced stress response, eventually leading to activation of developmental signaling pathways such as Wnt signaling. How early injury cues are interpreted and coupled to activation of these developmental signals and their targets is not well understood. Here, we show that Hif1α, a stress induced transcription factor, is required for tail regeneration in Xenopus tropicalis. We find that Hif1α is required for regeneration of differentiated axial tissues, including axons and muscle. Using RNA-sequencing, we find that Hif1α and Wnt converge on a broad set of genes required for posterior specification and differentiation, including the posterior hox genes. We further show that Hif1α is required for transcription via a Wnt-responsive element, a function that is conserved in both regeneration and early neural patterning. Our findings indicate that Hif1α has regulatory roles in Wnt target gene expression across multiple tissue contexts.

Graphical Abstract

Introduction

Following severe injury, all organisms initiate a wound healing program, though regenerative outcomes vary across species (Kakebeen & Wills, 2019). Organisms such as planarians, axolotls, and zebrafish exhibit complete and robust tissue regeneration, being able to replace lost limbs and organs with a variety of cell types that have proper structural organization (Erickson & Echeverri, 2018; Ivankovic et al., 2019; Kakebeen & Wills, 2019; Marques et al., 2019). Conserved injury-induced stresses including immune cell activation, inflammation, and reactive oxygen signaling occur in both regenerative and non-regenerative species but are essential for proper regeneration in species with this capability (Erickson & Echeverri, 2018; Kakebeen & Wills, 2019; Phipps et al., 2020). The downstream events that selectively link universal stress signals to regenerative growth in some animals but not others remain a persistent enigma in regeneration biology.

A specific gap in our understanding of stress signaling is how injury cues are coupled to changes in gene expression that direct patterning in regenerating tissues. Macrophages, reactive oxygen species (ROS), and oxygen flux at the wound edge are all injury induced stresses that are essential for regeneration in multiple species (Ferreira et al., 2018; Godwin et al., 2013; Love et al., 2013; Pirotte et al., 2015; Romero et al., 2018; Simkin et al., 2017), suggesting that transcription factors activated by these signals represent a possible link between injury and downstream gene activation. One such stress-activated transcriptional regulator is Hypoxia Inducible Factor 1α (Hif1α), which acts downstream of ROS, inflammatory signaling, and oxygen sensing, and is poised to couple cell extrinsic stressors with regenerative gene expression (Movafagh et al., 2015). Previous work has shown that Hif1α is necessary and sufficient for regeneration of Xenopus laevis tails (Ferreira et al., 2018). However, the downstream effects of Hif1α, in particular its transcriptional targets, are not known in a regenerative context. Moreover, while Hif1α canonically regulates apoptosis, cell proliferation, and cellular metabolism, it has also been shown to have roles in patterning and differentiation which have not been explored in appendage regeneration. Hif1α is required for proper neural tube development in mice (E. Y. Chen et al., 1999; Iyer et al., 1998) and neural crest chemotaxis in both Xenopus and chicks (Barriga et al., 2013). Hif1α also modulates Wnt signaling to direct gene expression and establishment of neural and skeletal muscle fates (Kaidi et al., 2007; Majmundar et al., 2015; Rohwer et al., 2019; Večeřa et al., 2017).

Wnt signaling is a deeply conserved factor in regeneration and is regarded as a key component of a complete regenerative response. It has been shown to be sufficient to drive regeneration of Xenopus tails, posterior structures in Planaria, and limb buds in chick embryos (Gurley et al., 2008; Kawakami et al., 2006; Lin & Slack, 2008). In Planaria, knockdown of wnt1 results in ectopic anterior regeneration, suggesting that Wnt is critical in regulating posterior identity in new tissues (Petersen & Reddien, 2009). Other regeneration models, including zebrafish fin and heart, as well as Xenopus and axolotl limbs, also require Wnt signaling to promote growth and cell type specification (Kawakami et al., 2006; Strand et al., 2016; Wehner et al., 2014). Work in Xenopus suggests that Wnt acts upstream of other developmental signaling pathways such as BMP, Notch, and FGF in establishing muscle and neural cell fates in regeneration (Beck et al., 2003; Slack et al., 2004). Wnt signaling has been shown to be downstream of injury responses in Drosophila imaginal discs, Hofstenia, and Hydra (Cazet et al., 2021; Harris et al., 2016; Ramirez et al., 2020), though the mechanisms coupling injury stresses to Wnt activation and regional interpretation of these cues has yet to be fully articulated.

Here we set out to define the downstream functions of Hif1α in Xenopus tropicalis tail regeneration, a model for rapid and complex appendage regeneration that benefits from a well annotated diploid genome (Kakebeen & Wills, 2019; Li et al., 2016). We show that pharmacological inhibition of Hif1α with two orthogonal reagents inhibits tail regeneration in X. tropicalis, as in X. laevis, and that Hif1α is required for muscle and axon growth. We find that genes sensitive to Hif1α perturbation include posteriorizing factors known to be targeted by canonical Wnt signaling. To broadly address whether these genes are regulated by both Hif1α and Wnt, we perform RNA-sequencing on Hif1α and Wnt antagonized tails during regeneration to show that Hif1α and Wnt are required for activation of largely overlapping groups of genes. Notably, both Hif1α and Wnt are required for posterior hox gene expression in the regeneration bud, suggesting a unique role for Hif1α in spatial patterning in this context. We find that Hif1α is required for activation of gene expression via Wnt-responsive promoter elements (WREs) during regeneration, as well as during early neural development, using a transgenic reporter line. Our findings suggest a model in which Hif1α is upstream of expression of Wnt target genes and highlights multiple in vivo contexts in which Hif1α activates developmental posterior patterning genes.

METHODS

Xenopus tropicalis husbandry and use

Use of Xenopus tropicalis was carried out under the approval and oversight of the IACUC committee at UW, an AALAC-accredited institution, under animal protocol 4374–01. Ovulation of adult Xenopus tropicalis and generation of embryos by natural matings were performed according to published methods (Khokha et al., 2002; Sive et al., 2000). Embryos were reared as described in (Khokha et al., 2002). Staging was assessed by the Nieuwkoop and Faber (NF) staging series (Nieuwkoop & Faber, 1994).

Xenopus tropicalis amputation assay

NF stage 41 tadpoles were anesthetized with 0.05% ms-222 in 1/9x MR and tested for response to touch prior to amputation surgery. Once fully anesthetized, a sterilized scalpel was used to amputate the posterior third of the tail. Amputated tadpoles were removed from anesthetic media within 10 minutes of amputation into new 1/9x MR. Tadpoles were kept at a density of no more than 2.5 tadpoles per mL.

Pharmacological inhibition

2-methoxyestradiol (Sigma m6383–5mg) was resuspended to a 10mM stock in DMSO, Echinomycin (Calbiochem 512–64-1) to a 0.5mM stock in DMSO, and IWR (Sigma I0161) 10mM stock in DMSO. Uninjured and injured tadpoles were reared with 0.1% DMSO, 5μM 2ME, 0.5μM Echinomycin, or 10μM IWR diluted in 1/9x MR until collection at 24- or 72-hours following treatment. Other concentrations were used in establishing doses and are reported in Supp. Fig. S1.

Morpholino injections

hif1α morpholino (MO) was ordered from GeneTools (sequence: CTCGCTACTACAGATCCCTCCATGC). Fertilized eggs from wild type and pbin7:GFP matings were de-jellied in 3% cysteine in H₂O for 10–15 minutes. 10ng of MO were injected into 1 cell at the 2-cell stage to generate unilateral morphants. Embryos were reared to NF stage 18 prior to fixation. 5 or 10ng of morpholino were injected at the 1 cell stage to generate uniform morphants which were reared to stage 41 for tail amputation assays.

Morpholino tail vein injections

hif1α vivo-morpholino (vMO) was ordered from GeneTools (sequence: CTCGCTACTACAGATCCCTCCATGC). Wild type tadpoles at stage 35–36 were anaesthetized with MS-222 and moved from culture dish to a thickly coasted agarose coated dish cover in one drop of media. Excess media was removed. Using a microinjector, a pulled needle containing vMO and labeled dextran tracer was inserted into the ventral tail vein in the tail and 2×2 nL (10 ng) or 4×2 nL injections were delivered (20 ng). Controls were injected with equal volumes of tracer. Embryos were returned to fresh media and screened for tracer fluorescence in the blood stream. Injected tadpoles were grown 24 hours to stage 41 and amputated. At 24hpa, tadpoles were re-injected into the tail vein as before and fixed for 1 hour in 1x MEM with 3.7% formaldehyde at room temperature.

Regeneration length measurement

Stereoscope imaging was performed on a Leica M205 FA with a color camera. Fixed tadpoles were imaged in PBS on a 1% agarose pads and measurements were recorded using the LAS software.

Immunohistochemistry

Xenopus tropicalis tadpoles were fixed for 1 hour in 1x MEM with 3.7% formaldehyde at room temperature. Tadpoles were permeabilized by washing 3×15 minutes in PBS + 0.01% Triton x-100 (PBT). Tadpoles were blocked for 1 hour at room temperature in 10% CAS-block (Invitrogen #00–8120) in PBT. Then tadpoles were incubated in primary antibody [1:1 mouse anti-12/101, gift from Richard Harland (Kintner & Brockes, 1984); 1:50 mouse anti-neurofilament associated protein (DSHB 3A10) (Dodd & Jessell, 1988)] diluted in 100% CAS-block overnight at 4°C. Tadpoles were then washed 3×10 minutes at room temperature in PBT and blocked for 30 minutes in 10% CAS-block in PBT. Secondary antibody (goat anti-mouse 488, ThermoFisher A11001) were diluted 1:500 in 100% CAS-block and incubated for 2 hours at room temperature. Tadpoles were then washed 3×10 minutes in PBT followed by a 10-minute incubation in 1:2000 DAPI (Sigma D9542) before being washed with 1xPBS for 10 more minutes. Isolated tails were mounted on slides in ProLong Gold (ThermoFisher P36930). Images were acquired using a Leica DM 5500 B microscope using a 10X objective and processed using FIJI image analysis software (Schindelin et al., 2012).

Whole mount in situ hybridization

Embryos and tadpoles were fixed overnight in 1x MEM with 3.7% formaldehyde at 4°C. Xenopus tropicalis multibasket in situ hybridization protocols were followed as described in (Khokha et al., 2002), with the notable change that pre-hybridization was always performed overnight. Isolated tails were mounted on slides in ProLong Gold (ThermoFisher P36930). Mounted tails and whole tadpoles were imaged on a Leica M205 FA with a color camera. Plasmids for sox2, otx2, en2, cdx4, and snai2 probes were gifted by the Harland lab. Other probes were synthesized using the following primer pairs designed against a single exon of each mRNA: fgf20 (forward - CTTTTGGGGATTTTGGGACT, reverse - taatacgactcactatagggGGCAGTATCTGCAGGTGGA), pbin7:GFP (forward - CACATGAAGCAGCACGACTT, reverse - taatacgactcactatagggTGCTCAGGTAGTGGTTGTCG), hoxc9 (forward - CCAGCTACTGCCAGACCTTC, reverse - taatacgactcactatagggTCCAATTCCGACTTGTCCTC), hoxc10 (forward - TCAATGGAGAAGACCCCAAG, reverse - taatacgactcactatagggTTGCTTCAGCGTCAGAATTG), hoxd11 (forward - TGCTGTCCAAGCTCTCTTGA, reverse - taatacgactcactatagggCTCTGTGCATCACCTCCTCA).

RNA extraction and sequencing

Tadpoles were anesthetized with 0.05% ms-222 in 1/9x MR and tested for response to touch prior to tissue collection. For 0hpa collections, a sterilized scalpel was used to amputate the posterior third of the tail followed by amputation 250μm anterior to the wound. For 24hpa tails, 250μm of tissue, including the regenerating tissue, was collected. 25 tails were collected for each condition and experiments were carried out in triplicate. DNA/RNA were purified by lysing tissues followed by phenol:chloroform extraction with ethanol precipitation. DNAseI treatment (Invitrogen, 18068015) was used to remove DNA. Total RNA samples were quantified using Qubit 2.0 Fluorometer (Life Technologies) and RNA integrity was checked using Agilent TapeStation 4200 (Agilent Technologies). RNA sequencing libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina following manufacturer’s instructions (NEB). The sequencing libraries were validated on the Agilent TapeStation (Agilent Technologies), and quantified by using Qubit 2.0 Fluorometer (Invitrogen) as well as by quantitative PCR (KAPA Biosystems). The sequencing libraries were clustered on a single lane of a flowcell. After clustering, the flowcell was loaded on the Illumina HiSeq instrument (4000 or equivalent) according to manufacturer’s instructions. The samples were sequenced using a 2×150bp Paired End (PE) configuration. Image analysis and base calling were conducted by the HiSeq Control Software (HCS). Raw sequence data (.bcl files) generated from Illumina HiSeq was converted into fastq files and de-multiplexed using Illumina’s bcl2fastq 2.17 software. One mismatch was allowed for index sequence identification.

Trimming, alignment, and counts

TrimGalore-0.6.5 was used to remove low quality reads (Phred33) and trim adapter sequences (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Pseudoalignment and count quantification was performed by Kallisto using an index built from the Xenopus tropicalis 9.1 transcriptome (Bray et al., 2016), returning .tsv files. These .tsv files were read into R, estimated counts for each gene were converted back to raw counts to generate a counts table suitable for processing with EdgeR.

Multidimensional Scaling and differential expression analysis

Analysis was performed using EdgeR (Robinson et al., 2010). The counts table was made into a DGEList object and filtered for transcripts with low counts and scaled with the calcNormFactors command. MDS plots were generated using plotMDS. Differential expression between 0hpa and DMSO (24hpa), and DMSO and each treatment group – 2ME, Ech, and IWR – were performed using glmQLFTest. To determine significance, p-values for each gene generated for the DMSO vs 2ME, Ech, and IWR conditions were ordered and corrected using the Benjamini-Hochberg procedure to determine a false discovery rate (FDR). Genes considered differentially expressed between DMSO and each condition has FDR < 0.05 (as shown in Supp. Figure 2). Significantly downregulated genes were called using filters of FDR < 0.05 and log₂FC < −0.2.

Heatmap generation

Heatmaps were generated by pheatmap (Kolde, 2019). Sequencing counts were converted to counts per million (cpm) and averaged across triplicates. The average cpm was normalized to DMSO to visualize fold change between DMSO and each treatment.

PANTHER Gene Ontology

Gene ontology was performed on the list of genes IDs called as significantly downregulated. This list was supplied to the PANTHER (Mi et al., 2019; Thomas, 2003) online portal using the reference genome for Xenopus tropicalis and a statistical overrepresentation test for GO biological processes was performed.

Plotting and statistical analysis

Boxplots and stacked bar plots were generated using the R package ggplot2 (Wickham, 2009). Venn diagrams were generated using eulerr (Larsson, 2020). Length measurements were compared using ANOVA and post hoc Tukey HSD to identify differences between groups. Difference in distribution of phenotypes across multiple treatments was assessed using a chi-square test. Statistical analysis was performed in R (R Core Team, 2020).

Analysis of previously published datasets

The heatmap in Figure 3C was generated as above using counts per million calculated using the supplementary counts table from GSE88975 (Chang et al., 2017).

Figure 3: Posterior hox gene expression in regeneration requires Hif1α and Wnt.

A) Pie chart of hox gene expression changes following 2ME, Ech, or IWR treatment. B) Schematic representation of hox gene organization in Xenopus. Color coding is as in (A). C) Schematic depicting hox gene expression in tailbud stage tadpoles. Color coding is as in (A). Boundaries were determined by lining up in situ expression patterns from Xenbase at the noted stages and using structural landmarks to determine the anterior and posterior domains. D) Heatmap displaying all detected hox genes across a regeneration timecourse. E) Heatmap displaying all detected hox genes clustered by expression. Genes significantly differentially expressed in all treatments are indicated by *. F) in situ hybridization for selected hox genes at 24hpa following treatment with DMSO, 2ME, Ech, or IWR. Arrowheads indicate amputation plane. Scale bars in E are 150μm.

Results

Hif1α is necessary for muscle and axon regeneration in X. tropicalis

Previously, Hif1α inhibition via Echinomycin (Ech), has been shown to reduce tail regeneration in X. laevis tadpoles (Ferreira et al., 2018; Kong et al., 2005). We first set out to determine if Hif1α antagonism inhibited tail regeneration in X. tropicalis tadpoles using Ech, which inhibits Hif1α binding to DNA, as well as 2-methoxyestradiol (2ME), which inhibits Hif1α nuclear localization and transcriptional activity (Mabjeesh et al., 2003). We identified effective doses of both inhibitors which reduced tail regeneration but did not cause lethality or impair health (Fig. 1A,B, Supp. Fig. 1). Compared to DMSO treated tails, tails treated with 5μM 2ME or 0.5μM Ech had significantly reduced tail lengths compared to control clutchmates 72 hours post amputation (hpa) (Fig. 1C,D). We scored regeneration by binning tails into 4 categories ranging from no regeneration to complete tail regeneration and found that Hif1α inhibited tails have significantly reduced regenerative outcomes relative to DMSO controls (Fig. 1C,E) (Beck, 2012). These data confirm that Hif1α is required for tail regeneration in X. tropicalis.

Figure 1: Hif1α is required for regeneration of muscle and axons.

A-B) Quantification of regeneration length normalized to DMSO clutchmates across dose curves for 2ME (A) and Ech (B). C) Dapi counterstained tails at 24 and 72hpa following treatment with DMSO, 5μM 2ME, or 0.5μM Ech. D) Quantification of regeneration length normalized to DMSO clutchmates. Statistical significance between groups was determined by ANOVA (p<2e-16) followed by Tukey’s posttest. * indicates p < 0.001. E) Regeneration scores binned to complete (full tail regeneration), strong (incomplete fin regeneration), poor (very little regeneration), or none. The treatments have statistically significant distribution of phenotypes (chi-square test, p < 2.2e-16). F) Immunohistochemical stains for 12/101 (muscle) and neurofilament (axons) at 72hpa after treatment with DMSO, 2ME, or Ech. G-H) in situ hybridization for fgf20 (G) or cdx4 (H) at 24hpa following DMSO, 2ME, or Ech treatment. Indicated numbers in F-H represent number of tails with prevented phenotype over total number assayed. Scale bars in C and F are 100μm, scale bars in G-H are 75μm. Arrowheads in C and F-H indicate amputation plane.

We next asked if axial tissues were specifically sensitive to loss of Hif1α function. Neither 2ME nor Ech completely abrogated regeneration in all treated tadpoles, but axial tissues always appeared to be reduced (Fig. 1E). The variation in regeneration outcomes could be attributed to differences in how much inhibitor each animal effectively gets from the media, though offers an opportunity to interrogate more nuanced consequences of Hif1α perturbation than completely abrogated regeneration would permit. To ask if axial tissues were specifically sensitive to loss of Hif1α function, we performed immunohistochemical staining for axons and muscle at 72hpa in tadpoles treated with DMSO, 2ME, or Ech from the time of amputation to collection and found that Hif1α antagonism resulted in tails with little to no muscle (12/101) or axon (neurofilament) regeneration, despite a large degree of fin regeneration (Fig. 1F). We then asked whether Hif1α was required for the activation of neural or mesodermal fate specification genes during regeneration. To this end, we performed in situ hybridization for fgf20 and cdx4, posterior mesoderm and neuronal markers, respectively (Lea et al., 2009; Northrop & Kimelman, 1994). We found that at 24hpa, both fgf20 and cdx4 are strongly induced in the regeneration bud in controls but not if Hif1α is inhibited (Fig. 1G,H). These results suggested that Hif1α might be specifically required for the activation of posterior fate genes during regeneration.

Hif1α and Wnt are necessary for expression of similar genes during regeneration

Having found that Hif1α is required to activate fgf20 and cdx4, known targets of canonical Wnt signaling, we hypothesized that Hif1α might broadly target the same genes as Wnt signaling (Chamorro et al., 2005; Haremaki et al., 2003). To test the requirement of Wnt signaling in regeneration, we used the Wnt inhibitor IWR-1 (IWR), which stabilizes the axin destruction complex (Borday et al., 2018; Chen et al., 2009). We find that treatment with 10μM IWR greatly reduces regeneration length and quality, in agreement with previous studies inhibiting Wnt in Xenopus tail regeneration (Supp. Fig. 1) (Lin & Slack, 2008). We then performed RNA-sequencing to identify regeneration-induced changes in gene expression dependent on either Hif1α or Wnt. We collected tail tissue 250μm anterior to the amputation plane at either 0 or 24hpa following treatment with DMSO, 2ME, Ech, or IWR (Fig. 2A). RNA-Seq libraries were prepared and analyzed in triplicate. When multidimensional scaling (MDS) was applied to visualize the relationship between groups (Robinson et al., 2010) we found that biological replicate libraries clustered tightly together, demonstrating their reproducibility (Fig. 2B, Supp. Fig. 2A). We also found that 0hpa tails are largely distinct from all four 24hpa groups, suggesting that there is a gene expression signature specific to the tissue following injury regardless of regenerative outcome (Supp. Fig. 2A). To visualize gene expression across treatments, we normalized all groups to DMSO and generated a heatmap. This revealed that the directionality of changes in expression relative to DMSO were largely shared following inhibition of Hif1α or Wnt (Fig. 2C). PANTHER Gene Ontology (GO) Enrichment analysis on genes differentially expressed by all three treatments called terms associated with several known Hif1α regulated processes, including erythrocyte cell homeostasis, as well as several not commonly associated with Hif1α, including macromolecule biosynthesis and mRNA splicing (Table S1). Notably, developmental pattern specification terms which are not traditionally affiliated with Hif1α but are often associated with Wnt signaling, among other key developmental pathways, were enriched (Supp. Fig. 2B). We focused our downstream analysis on the 1443 genes downregulated by all treatments, as these represent genes which require Hif1α or Wnt for expression in this context (Fig. 2D). We then asked if the majority of these Hif1α and Wnt responsive genes were normally upregulated following injury or if they were normally downregulated. To this end, we analyzed only genes which increased from the 0hpa control to 24hpa DMSO, or injury-induced genes. We found that 1106 injury-induced genes fail to be upregulated when Hif1α or Wnt are inhibited, indicating that most of the 1443 Hif1α and Wnt regulated genes are normally upregulated during regeneration (Supp. Fig. 2C). These results suggest that Hif1α and Wnt direct similar gene expression programs during regeneration and identify those genes which are sensitive to perturbation of each pathway. GO analysis of genes downregulated in Hif1α and Wnt inhibited tadpoles found anterior/posterior pattern specification genes were enriched (Fig. 2E, Table S2). This result, together with our observed loss of posterior markers fgf20 and cdx4 (Fig. 1G,H), suggests that Hif1α may be a transcriptional activator of posterior patterning genes in the regenerating tail.

Figure 2: RNA-sequencing reveals shared gene regulatory roles for Hif1α and Wnt.

A) Schematic depicting experimental setup for RNA-sequencing. Tails were amputated and treated with DMSO, 2ME, Ech, or IWR and collected at 0 or 24hpa for sequencing. Purple chevrons in tails represent somites, green line indicates spinal cord, and grey box marks tissues collected for sequencing. Amputation plane is marked with an orange dotted line. B) MDS plot depicting relationship between samples. C) Heatmap of gene expression for genes differentially expressed in all 3 treatment groups as log₂FC relative to DMSO clustered by expression. *Lower bound of color scale represented values from −6 to −15. D) Venn diagram showing overlap in downregulated genes (FDR < 0.05 and log₂FC < −0.2). E) Selected GO terms from PANTHER gene ontology enrichment of 1443 downregulated genes plotted by -log₁₀FDR.

Posterior hox gene expression in the regeneration bud requires Hif1α and Wnt

Because Hif1α and Wnt inhibition resulted in a downregulation of genes involved in patterning, we examined the genes under these GO terms and noticed that more than half of all hox genes were downregulated under all 3 treatments (Fig. 3A,B). To determine the normal expression for relevant hox genes during development, we examined in situ expression patterns at stages 30–35 on Xenbase for hox genes that were differentially expressed in all 3 of our treatment groups to determine the normal expression of this family of patterning related genes (Bowes et al., 2010). Notably, all but one of the hox genes that were downregulated following Hif1α and Wnt inhibition are expressed in the most posterior region of tailbud stage tadpoles. The only gene that was upregulated in each group was hoxb4, which we noted had a distinct domain in the anterior of the tadpole (Fig. 3C, Table S3). Using previously published RNA-sequencing data over a regeneration timecourse in X. tropicalis (Chang et al., 2017), we queried the normal expression dynamics of the hox genes were across a regenerative timecourse. Relative to uninjured tadpoles, the expression of most hox genes decreased at 0 and 6hpa but then increased by 15 and 24hpa (Fig. 3D). Of those genes that were activated during regeneration, most failed to increase in expression when Hif1α and Wnt were inhibited (Fig. 3E). Examining expression of several representatives of this trend (hoxc9, hoxc10, and hoxd11), we found that at 24hpa these transcripts are normally abundant in the regenerating axial tissue but that they fail to be induced when tadpoles are treated with 2ME, Ech, or IWR (Fig. 3F). These results suggest that hox genes are activated in response to injury and that this activation depends on both Hif1α and Wnt.

Most Wnt signaling ligands and receptors are not regulated by Hif1α

Because the expression of posterior patterning genes was decreased if either Hif1α or Wnt signaling was blocked, we asked if Hif1α might be acting upstream of Wnt signaling, such that inhibition of Hif1α caused an indirect downregulation of Wnt target genes. Specifically, we asked whether inhibition of Hif1α resulted in downregulation of Wnt ligands or receptors. To test this, we examined expression of Wnt ligands and receptors and found that the majority were not differentially expressed following Hif1α or Wnt inhibition, several had an increase in expression under these treatments, and only wnt5a and wnt5b are weakly downregulated (Fig. 4A). This suggested that it is unlikely that the primary role of Hif1α in patterning is to transcriptionally upregulate Wnt ligands or receptors. We also considered that Hif1α might be required to repress expression of components of the β-catenin degradation complex, but found that dsh2 and axin2 expression are decreased upon Hif1α inhibition, while other components are not significantly affected (Fig. 4A). These results suggest that the sensitivity of Wnt target genes to Hif1α perturbation is not likely due to Hif1α being a direct activator of Wnt signaling components or repressor of factors that destabilize β-catenin. The upregulation of several Wnt signaling components that we observed following 2ME, Ech, and IWR treatments does suggest that there may be compensatory upregulation of these factors when Wnt or Hif1α is inhibited, but it is not clear whether this regulation is direct.

Figure 4: Hif1α directs WRE activity in regeneration.

A) Heatmap displaying genes of the Wnt signaling pathway ordered by expression in IWR and sorted by position in pathway. Genes differentially expressed in all treatments are indicated by *. B) Timecourse of WRE-driven GFP transcript in pbin7:GFP tadpoles via in situ hybridization. C) Visualization of GFP transcripts at 24hpa following treatment with DMSO, 2ME, Ech, or IWR. D) Heatmap displaying posterior neural patterning Wnt target genes. Genes differentially expressed in all treatments are indicated by *. Indicated numbers in C represent number of tails with prevented phenotype over total number assayed. Scale bars in B-C are 150μm. Arrowheads indicate amputation plane.

Hif1α is required for expression of Wnt responsive elements and direct target genes of canonical Wnt signaling

Our data to this point suggested that Hif1α is required for expression of many of the same patterning genes targeted by canonical Wnt signaling, but that it is not likely to act by directly upregulating Wnt ligands or receptors. We next asked more specifically whether Hif1α acts on Wnt-dependent gene expression by determining if Hif1α perturbation affects transcription via Wnt-responsive promoter elements (WREs). To visualize when and where WRE-dependent transcription is active during regeneration, we utilized the pbin7:GFP transgenic line, in which GFP expression serves as a readout of Wnt activity via WREs (Tran & Vleminckx, 2014). Although GFP fluorescence in the regenerating tail was relatively faint and variable, by using in situ hybridization to visualize GFP transcripts directly, we found pbin7:GFP transcripts throughout the axial tissues in uninjured tails which appeared to be strongly upregulated at the posterior extreme of the tail (Fig. 4B, Supp. Fig. 3). Following amputation, we find GFP transcript localized to regenerating tissue 24hpa and find that this transcription is sustained until 72hpa while signal anterior to the wound site is reduced (Fig. 4B). To confirm that pbin7:GFP transcripts are a reliable readout of Wnt activity, we treated regenerating tadpoles from this line with IWR, and found that IWR treatment reduced GFP expression at 24hpa (Fig. 4C). Inhibition of Hif1α with either 2ME or Ech also reduced GFP expression at 24hpa, at a comparable degree to IWR (Fig. 4C). As an independent test of whether Hif1α is required for Wnt-dependent gene expression, we examined expression of established direct target genes of canonical Wnt signaling established at gastrula and neurula stages (Kjolby & Harland, 2017; Young et al., 2014). We find that, while not all of these targets are sensitive to Wnt inhibition in regeneration, those that are downregulated by IWR, including well-established targets such as axin2, cdx2, cdx4, prickle, sall1 and sall4, are also downregulated by Hif1α inhibition (Fig. 4D). These results suggest that Hif1α is acting upstream of the activation of regeneration specific Wnt targets and WRE-mediated transcription, though they do not distinguish between a direct or indirect interaction at WREs.

Hif1α regulates anteroposterior patterning during neurula stages

Having seen that Hif1α is required during regeneration to activate expression of posterior neural Wnt targets, we asked if Hif1α modulated Wnt-mediated gene expression in other contexts. First, we confirmed that inhibiting translation of Hif1α resulted in the same inhibition of regeneration as pharmacological Hif1α inhibition. To do this, we first examined regeneration in tadpoles that were injected with a hif1α translation blocking morpholino (MO). Using a unilateral injection strategy, we confirmed that our MO caused the same developmental effects as previously reported for Hif1α, specifically a reduction in the migration of cranial neural crest on the injected side relative to the contralateral control (Fig. 5A) (Barriga et al., 2013). At stage 41, amputation of the tail resulted in minimal regeneration compared to controls (Fig. 5B, D, E). We noted Hif1α morphants had shortened anterior-posterior axes at tailbud and tadpole stages. To bypass the significant early developmental effects of Hif1α depletion, we blocked Hif1α translation using tail vein injection of a vivo-morpholino (vivo-hif1α MO), which we utilized to inhibit Hif1α during regeneration while circumventing any developmental consequences (Kakebeen et al., 2020; Liu et al., 2012). This method of Hif1α inhibition also reduced regeneration length and score (Fig. 5C, F, G).

Figure 5: Hif1α regulates WRE expression and AP patterning at neurula stages.

A) Dorsal and lateral views of in situ hybridization for snai2 in a stage 23 embryo injected unilaterally with hif1α morpholino. Miniruby fluorescent tracer was used to identify the injected side (dorsal view). snai2 expression in the migrating neural crest streams is reduced on the injected side, noted with white arrowheads. B) Control and 5ng hif1α morpholino injected tails at 72hpa. C) Tracer and 10ng vivo-hif1α morpholino injected tails at 72hpa. D) Length of regenerated tissue normalized to average control length in control, 5ng, and 10ng hif1α morpholino injected tails at 72hpa. E) Regeneration score for control, 5ng, and 10ng hif1α morpholino injected tails at 72hpa. F) Length of regenerated tissue normalized to average tracer-only length in tracer and 10ng vivo-hif1α morpholino injected tails at 72hpa. G) Regeneration score for tracer-only and 10ng vivo-hif1α morpholino injected tails at 72hpa. Legend as in (E). H) Dorsal views of in situ hybridization for pbin7:GFP, sox2, otx2, and en2 in stage 18–20 embryos. Uninjected controls (upper panels) and stage matched, unilaterally injected Hif1α morphants (lower panels) are shown. Morpholino injected side is marked with * in each lower panel. Scale bars in A, H are 150μm and in C,D are 100μm.

Because Hif1α morphants had shortened axes, we hypothesized that Hif1α might be required for proper anterior-posterior patterning in the early embryo, and specifically, might modulate Wnt-dependent gene expression in this context, as in regeneration. We therefore used unilaterally-injected Hif1α morphants to assay pbin7:GFP expression during embryonic neural patterning. GFP expression at stage 18 spans the posterior of the neural folds with a clear anterior boundary in agreement with previous descriptions and with the well-known function of Wnt signaling in posteriorization of the neural tube (Fig. 5H) (Borday et al., 2018; Kiecker & Niehrs, 2001). Embryos in which hif1α MO was injected unilaterally showed posteriorly-shifted pbin7:GFP expression on the injected side. This reflects an anteriorization of that side, as would be expected for loss of Wnt-dependent transcriptional activation. To confirm this, we assayed expression of regional neural tube markers (sox2, otx2, and en2) in unilaterally injected Hif1α morphants at stage 20 (Fig. 5H) (Brivanlou & Harland, 1989; Mancilla & Mayor, 1996; Pannese et al., 1995). Sox2, a pan-neural marker, is largely unaffected by hif1α MO, suggesting that Hif1α is not required for neural induction. However, the forebrain marker otx2 has an expanded domain on the MO injected side, while the hindbrain marker en2 shifted towards the posterior following hif1α MO injection. Both these outcomes are consistent with an anteriorization of the neural plate on the injected side. These results suggest that Hif1α is necessary for establishing posterior identity of the embryo during neurulation as well as during tail regeneration.

Discussion

Regenerative healing requires that initial wounding signals of injury be coupled not to scarring, but to growth and patterning of new tissues. Injury-induced stresses, including ROS, bioelectrical signals, and innate immune cell recruitment, are critical to regeneration across taxa (Kakebeen & Wills, 2019; Tseng & Levin, 2008). Similarly, developmental patterning signals including Wnt, FGF, BMP, and TGF-β are required for full tail regeneration in Xenopus (Beck et al., 2003; Christen et al., 2003; Lin & Slack, 2008). In some cases, injury stresses are known to act upstream of specific growth factor pathways, as in the case of ROS, which acts upstream of FGF in X. laevis tail regeneration (Love et al., 2013). Still, the mechanistic link between these aspects of regeneration is not clear. Recent work has shown that Hif1α is required for regeneration in X. laevis, acting in the first hour after injury and responding to injury-induced hypoxia and, at least indirectly, ROS (Ferreira et al., 2018). Here we propose that Hif1α bridges injury to transcriptional remodeling during regeneration by activating a broad suite of posterior patterning factors and regulatory elements that are also targeted by Wnt signaling.

Interactions between Hif1α and Wnt signaling have been described in several in vitro or ex vivo models. In prostate cancer cell lines and embryoid bodies, Hif1α has been shown to increase β-catenin target gene expression to drive differentiation (Luo et al., 2018; Večeřa et al., 2017). Hif1α has also been reported to be essential for transcription via Wnt-responsive elements in colon cancer cell lines (Rohwer et al., 2019). An in vivo model of murine muscle regeneration found that Hif1α actually represses Wnt signaling to inhibit regeneration (Majmundar et al., 2015). Our study demonstrates that Hif1α is upstream of Wnt target gene expression during both axial regeneration and early embryonic patterning.

One of the principal findings of our analysis is that the majority of hox genes transcriptionally upregulated during regeneration, and that this expression is reduced when either Hif1α or Wnt signaling are inhibited. Of note, hox gene expression in Xenopus does not follow typical temporal co-linearity (Kondo et al., 2017) and we find that posterior hox expression domains are largely overlapping in the tail (Fig. 3E). Previous work has found that several hox genes, including hoxc10 and hoxd13, are activated in regenerating limbs and tails in Xenopus, and Hoxc13 orthologs are also required for regeneration in the zebrafish tail fin (Thummel et al., 2007). It has also been proposed that ectopic hox gene expression many contribute to homeotic transformation of regenerating tails into limbs in Rana temporaria via retinyl palmitate (Christen et al., 2003; Maden, 1993). These observations could point to a mechanism in which regenerating tissues re-establish positional identity through hox gene expression, and our results suggest that Hif1α and Wnt are both required for this positional mapping in the regenerating tail. Our data further support that Hif1α activates other posteriorizing genes in regeneration, including FGF, cdx, and sal family members, which are known to be regulated by Wnt signaling. We also find that Hif1α depletion anteriorizes the neural tube, suggesting that tail regeneration could be recapitulating the developmental gene expression programs used during embryonic regionalization. Our study does not preclude the possibility that other developmental pathways, such as FGF and retinoic acid signaling, are also involved in establishing posterior identify, but we propose that interactions between Hif1α and Wnt signaling help drive this process.

The specific mechanism by which Hif1α and Wnt signaling converge on shared gene targets is not yet clear, and our work is consistent with several possibilities. While our RNASeq data suggest that Hif1α inhibition does not lead to transcriptional downregulation of most Wnt ligands, receptors, or effector proteins, it remains possible that Hif1α could be contributing to Wnt target regulation through alterations in the stability, localization, or longevity of any of these components. Our finding that Hif1α is required for TCF/LEF reporter expression suggests that Hif1α does have a specific effect on transcriptional regulation at Wnt reporter elements. However, this could be mediated by direct binding of Hif1α at these sites, as is the case in colon cancer (Rohwer et al., 2019), or by Hif1α forming a complex with β-catenin or TCF, or by indirectly enhancing the ability of β-catenin to activate these sites. We are eager to pursue these possibilities further.

Our study expands the range of processes and genes known to be targeted by Hif1α in Xenopus, where it was previously known to regulate heart development, neural crest migration, and redox balance after injury (Barriga et al., 2013; Ferreira et al., 2018; Nagao et al., 2008). That Hif1α is capable of interacting with Wnt signaling is not a new premise, but here we show that Hif1α can activate canonically Wnt-dependent gene expression during vertebrate regeneration as well as embryonic neural pattering. Further, we demonstrate that Hif1α is essential for establishing muscle and neural tissue in the regenerating tail. Finally, our work introduces the possibility that other Wnt dependent processes may be coregulated by Hif1α, providing another level of regulation to the establishment of embryonic gradients and cell fate branches.

Highlights.

• Hif1α is required for axon and muscle regeneration in Xenopus tropicalis tails.

• Hif1α and Wnt signaling are upstream of a shared set of injury-induced transcripts.

• hox gene activation is downstream of Hif1α and Wnt in the injury response.

• Hif1α directs anteroposterior identity during neural development.

• Transcription via Wnt responsive elements relies on Hif1α in development and regeneration.