Hif1α is required for Wnt regulated gene expression during Xenopus tropicalis tail regeneration

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 a regulatory role for Hif1α in Wnt mediated gene expression across multiple tissue contexts.


Introduction 23
Following severe injury, all organisms initiate a wound healing program, though 24 regenerative outcomes vary across species (Kakebeen & Wills, 2019). Organisms such 25 as planaria, axolotls, and zebrafish exhibit complete and robust tissue regeneration, 26 being able to replace lost limbs and organs with a variety of cell types and proper 27 structural organization ( which acts downstream of ROS, inflammatory signaling, and oxygen sensing, and is 43 poised to couple cell extrinsic stressors with regenerative gene expression (Movafagh et 44 al., 2015). Previous work has shown that Hif1α is necessary and sufficient for 45 regeneration of Xenopus laevis tails (Ferreira et al., 2018). However, the downstream 46 7 several not commonly associated with Hif1α, including macromolecule biosynthesis and 139 mRNA splicing (Table S1). Notably, developmental pattern specification terms affiliated 140 with Wnt signaling were called (Fig. 2E). This result, together with our observed loss of 141 posterior markers fgf20 and cdx4 (Fig. 1E,F), suggested that Hif1α may be a 142 transcriptional activator of posterior patterning genes in the regenerating tail. 143

Posterior hox gene expression in the regeneration bud requires Hif1α and Wnt 144
Noting that Hif1α and Wnt inhibition resulted in a downregulation of genes 145 involved in patterning, we examined the genes under these GO terms and noticed that 146 more than half of all hox genes were downregulated under these treatments (Fig. 3A,B). 147 We examined in situ expression patterns at stages 32-34 on Xenbase of hox genes 148 which were differentially expressed in all 3 of our treatment groups to determine the 149 normal expression of this family of patterning related genes (Bowes et al., 2010). 150 Notably, all but 1 of the hox genes that were downregulated following Hif1α and Wnt 151 inhibition have expression in the most posterior region of tailbud stage tadpoles. The 152 only gene that was upregulated in each group was hoxb3, which we noted had a distinct 153 domain in the anterior of the tadpole (Fig. 3C). Using previously published RNA-154 sequencing data over a regeneration timecourse in X. tropicalis (Chang et al., 2017), we 155 asked what the normal expression dynamics of the hox genes were across a 156 regenerative timecourse. Relative to uninjured tadpoles, the expression of most hox 157 genes is decreased at 0 and 6hpa but then increased by 15 and 24hpa (Fig. 3D). Of 158 those genes that are activated during regeneration, most fail to increase in expression 159 when Hif1α and Wnt are inhibited (Fig. 3E). Examining expression hoxc10, hoxd11, and 160 hoxa13 at 24hpa, we find that normally these transcripts are abundant in the 161 8 regenerating axial tissue but that they fail to be induced when tadpoles are treated with 162 2ME, Ech, or IWR (Fig. 3F). These results suggest that hox genes are activated in 163 response to injury and that this activation depends on both Hif1α and Wnt. 164

Wnt signaling ligands and receptors are not regulated by Hif1α 165
Because the expression of posterior patterning genes was decreased if either 166 Hif1α or Wnt signaling was blocked, we asked if Hif1α might be acting upstream of Wnt 167 signaling, such that inhibition of Hif1α caused an indirect downregulation of Wnt 168 signaling. Specifically, we asked whether inhibition of Hif1α resulted in downregulation 169 of Wnt ligands or receptors. To test this, we examined expression of Wnt ligands and 170 receptors and found that the majority were not differentially expressed following Hif1α or 171 Wnt inhibition, several had an increase in expression under these treatments, and only 172 wnt5a and wnt5b are weakly downregulated (Fig. 4A). This suggests to us that it is 173 unlikely that the primary role of Hif1α in patterning is to transcriptionally upregulate Wnt 174 ligands or receptors. We also considered that Hif1α might be required to repress 175 expression of components of the β-catenin degradation complex but found that dsh2 176 and axin2 expression are decreased upon Hif1α inhibition, while other components are 177 not significantly affected (Fig. 4A). These results suggest that the sensitivity of Wnt 178 target genes to Hif1α perturbation is less likely due to Hif1α being a direct activator of 179 Wnt signaling components or repressor of factors that destabilize β-catenin. The 180 upregulation of several Wnt signaling components that we observe following 2ME, Ech, 181 and IWR treatments does suggest to us that there may be compensatory upregulation 182 of these factors when Wnt or Hif1α is inhibited, but it is not clear that this regulation is 183 direct. 184

HIf1a is required for expression of Wnt responsive elements and direct target 185 genes of canonical Wnt signaling 186
Our data to this point suggested that Hif1α is required for expression of many of 187 the same patterning genes targeted by canonical Wnt signaling, but that it is not likely to 188 act by directly upregulating Wnt ligands or receptors. We next asked more specifically visualize GFP transcripts directly, we found pbin7:GFP transcripts throughout the axial 196 tissues in uninjured tails which appears to be strongly upregulated in the posterior 197 extreme of the tail (Fig. 4B, Supp. Fig. 3). Following amputation, we find GFP transcript 198 localized to regenerating tissue 24hpa and find that this transcription is sustained until 199 72hpa (Fig. 4B). To confirm that pbin7:GFP transcripts are a reliable readout of Wnt 200 activity, we treated regenerating tadpoles from this line with IWR, and found that IWR 201 treatment reduced GFP expression at 24hpa (Fig. 4C). Inhibition of Hif1α with either 202 2ME or Ech also reduced GFP expression at 24hpa, at a comparable degree to IWR 203 Young et al., 2014). We find that, while not all of these targets are still sensitive to Wnt 207 inhibition in regeneration, those that are downregulated by Wnt inhibition, including well-208 established targets such as axin2, cdx2, cdx4, prickle, sall1 and sall4, are also 209 downregulated by Hif1a inhibition (Fig. 4D). These results suggest that activation of 210 regeneration specific Wnt target requires Hif1α as well as Wnt. 211

Hif1α regulates anteroposterior patterning during neurula stages 212
Having seen that Hif1α is required during regeneration to activate expression of indirectly enhancing the ability of β-catenin to activate these sites. We are eager to 292 pursue these possibilities further. 293 Our study expands the range of processes and genes known to be targeted by 294

Competing Interests 316
The authors declare no competing interests. 317

Data Availability 318
All sequencing data associated with this manuscript will be made publicly available on 319 GEO prior to final publication. 320

Xenopus tropicalis husbandry and use 322
Use of Xenopus tropicalis was carried out under the approval and oversight of the 323 IACUC committee at UW, an AALAC-accredited institution, under animal protocol 4374-324

Xenopus tropicalis amputation assay 329
NF stage 41 tadpoles were anesthetized with 0.05% ms-222 in 1/9x MR and tested 330 for response to touch prior to amputation surgery. Once fully anesthetized, a 331 sterilized scalpel was used to amputate the posterior third of the tail. Amputated 332 tadpoles were removed from anesthetic media within 10 minutes of amputation into 333 new 1/9x MR. Tadpoles were kept at a density of no more than 2.5 tadpoles per mL. 334

Tadpole size and regeneration length measurement 349
Stereoscope imaging was performed on a Leica M205 FA with a color camera. 350 Fixed tadpoles were imaged in PBS on a 1% agarose pads and measurements 351 were recorded using the LAS software. Leica DM 5500 B microscope using a 10X objective and processed using FIJI 368 image analysis software (Schindelin et al., 2012). 369

Whole mount in situ hybridization 370
Embryos and tadpoles were fixed overnight in 1x MEM with 3.7% formaldehyde at 4°C. 371 Xenopus tropicalis multibasket in situ hybridization protocols were followed as described 372 in (Khokha et al., 2002), with the notable change that pre-hybridization was always 373 performed overnight. Isolated tails were mounted on slides in ProLong Gold 374 (ThermoFisher P36930). Mounted tails and whole tadpoles were imaged on a Leica 375 M205 FA with a color camera . Plasmids for sox2, otx2, en2, egr2, cdx4, and snai2  Illumina's bcl2fastq 2.17 software. One mismatch was allowed for index sequence 409 identification. 410

Trimming, alignment, and counts 411
TrimGalore-0.6.5 was used to remove low quality reads (Phred33) and trim adapter 412 sequences (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). files. These .tsv files were read into R, estimated counts for each gene were converted 416 back to raw counts to generate a counts table suitable for processing with EdgeR. 417

Multidimensional Scaling and differential expression analysis 418
Analysis was performed using EdgeR (Robinson et al., 2010). The counts table was 419 made into a DGEList object and filtered for transcripts with low counts and scaled with 420 the calcNormFactors command. MDS plots were generated using plotMDS. Differential 421 expression between 0hpa and DMSO (24hpa), and DMSO and each treatment group -422 2ME, Ech, and IWR -were performed using glmQLFTest. To determine significance, p-423 values for each gene generated for the DMSO vs 2ME, Ech, and IWR conditions were 424 ordered and corrected using the Benjamini-Hochberg procedure to determine a false 425 discovery rate (FDR). Genes considered differentially expressed between DMSO and 426 each condition has FDR < 0.05 (as shown in Supp. Figure 2). Significantly 427 downregulated genes were called using filters of FDR < 0.05 and log2FC < -0.2. 428

Heatmap generation 429
Heatmaps were generated by pheatmap (Kolde, 2019). Sequencing counts were 430 converted to counts per million (cpm) and averaged across triplicates. The average cpm 431 was normalized to DMSO to visualize fold change between DMSO and each treatment. 432

PANTHER Gene Ontology 433
Gene ontology was performed on the list of genes IDs called as significantly 434 downregulated. This list was supplied to the PANTHER (Mi et al., 2019;Thomas, 2003) 435 online portal using the reference genome for Xenopus tropicalis and a statistical 436 overrepresentation test for GO biological processes was performed. 437

Plotting and statistical analysis 438
Boxplots and stacked bar plots were generated using the R package ggplot2 439 (Wickham, 2009). Venn diagrams were generated using eulerr (Larsson, 2020). 440 Length measurements were compared using ANOVA and post hoc Tukey HSD to 441 identify differences between groups. Difference in distribution of phenotypes across 442 multiple treatments was assessed using a chi-square test. Statistical analysis was 443 performed in R (R Core Team, 2020). 444

Analysis of previously published datasets 445
The heatmap in Figure 3C

508
Indicated numbers in C represent number of tails with prevented phenotype over total number assayed.