1. Developmental Biology
  2. Stem Cells and Regenerative Medicine
Download icon

Generic injuries are sufficient to induce ectopic Wnt organizers in Hydra

  1. Jack F Cazet
  2. Adrienne Cho
  3. Celina E Juliano  Is a corresponding author
  1. Department of Molecular and Cellular Biology, University of California, Davis, United States
Research Article
  • Cited 1
  • Views 1,946
  • Annotations
Cite this article as: eLife 2021;10:e60562 doi: 10.7554/eLife.60562

Abstract

During whole-body regeneration, a bisection injury can trigger two different types of regeneration. To understand the transcriptional regulation underlying this adaptive response, we characterized transcript abundance and chromatin accessibility during oral and aboral regeneration in the cnidarian Hydra vulgaris. We found that the initial response to amputation at both wound sites is identical and includes widespread apoptosis and the activation of the oral-specifying Wnt signaling pathway. By 8 hr post amputation, Wnt signaling became restricted to oral regeneration. Wnt pathway genes were also upregulated in puncture wounds, and these wounds induced the formation of ectopic oral structures if pre-existing organizers were simultaneously amputated. Our work suggests that oral patterning is activated as part of a generic injury response in Hydra, and that alternative injury outcomes are dependent on signals from the surrounding tissue. Furthermore, Wnt signaling is likely part of a conserved wound response predating the split of cnidarians and bilaterians.

Introduction

Regeneration is an injury-induced morphogenetic process that enables the restoration of lost or damaged body parts. Although nearly all animals are capable of some form of regeneration, the greatest regenerative capacity is found in the invertebrate species capable of rebuilding their entire body from small tissue fragments through a process called whole-body regeneration. In these highly regenerative systems, amputation injuries trigger morphogenesis on both sides of the amputation plane, leading to the reconstruction of all missing body parts. As a result, the two tissue fragments generated by an amputation will give rise to two morphologically identical individuals, even in cases where the two fragments need to regenerate entirely different structures. This raises the question of how the appropriate morphogenetic programs are activated in response to injury during whole-body regeneration.

Divergent morphogenetic outcomes during regeneration are driven by the differential activation of deeply conserved signaling pathways on either side of the amputation plane (Cary et al., 2019; Holstein et al., 2003; Owlarn and Bartscherer, 2016; Srivastava et al., 2014). This results in patterning asymmetries that drive a divergence in gene expression in the two injury sites, leading to the regeneration of different structures. Currently, little is known about how these signaling pathways are reactivated during regeneration or how that reactivation is restricted to only one side of the amputation plane.

The canonical Wnt signaling pathway plays a central role in the positional specification of tissue regenerating along the primary body axis in several distantly related phyla, including cnidarians, acoels, and planarians (Gurley et al., 2008; Hobmayer et al., 2000; Lengfeld et al., 2009; Petersen and Reddien, 2008; Srivastava et al., 2014; Stückemann et al., 2017). In these systems, amputations that transect the primary body axis (i.e., the oral-aboral axis in cnidarians or the anterior-posterior axis in planarians and acoels) induce new Wnt organizer formation on only one side of the amputation plane; in cnidarians, Wnt organizers form only at oral-facing wounds (Hobmayer et al., 2000), and in planarians and acoels, Wnt organizers form only at posterior-facing wounds (Gurley et al., 2008; Petersen and Reddien, 2008; Srivastava et al., 2014). These similarities raise the possibility that injury-induced Wnt organizer formation is an ancestral feature of the metazoan wound response. However, establishing the plausibility of this hypothesis requires a thorough understanding of the regulation and function of injury-induced Wnt signaling in multiple distantly related animal regeneration models.

Wnt signaling has a similar function in specifying posterior tissue during regeneration in both acoels and planarians, but there appear to be marked differences in how the pathway is activated. During acoel regeneration, Wnt signaling is induced via the upregulation of the Wnt ligand wnt-3 within 3 hr post amputation (hpa; Ramirez et al., 2020). This occurs only in posterior-facing injuries and requires the injury-responsive transcription factor (TF) Egr. In contrast, during planarian regeneration several positive regulators of canonical Wnt signaling, including the Wnt ligand wntP-1, are initially upregulated as part of an early generic response to all injuries and do not become restricted to posterior regeneration until between 24 and 48 hpa (Gurley et al., 2010; Petersen and Reddien, 2009; Stückemann et al., 2017; Wenemoser et al., 2012; Wurtzel et al., 2015); however, the secreted Wnt inhibitor notum, while also upregulated after all injuries, shows higher levels of upregulation in anterior-facing wounds by 6 hpa (Petersen and Reddien, 2011; Wurtzel et al., 2015). The TFs involved in upregulating wntP-1 and notum in response to injury in planarians are not known. The differences between acoels and planarians suggest a lack of conservation in the gene regulatory networks that reactivate canonical Wnt signaling during bilaterian regeneration; however, it is unclear if this lack of conservation is the result of divergence from a shared ancestral state or because planarian and acoel regeneration evolved independently.

Cnidaria is the sister clade to bilateria and as such provides an informative context for questions related to the evolution of the metazoan wound response. Hydra vulgaris is a well-established cnidarian model that possesses an exceptional regenerative capacity, capable of fully rebuilding its body from tissue pieces containing as few as 300 cells, or ~1% of an adult polyp (Shimizu et al., 1993). It has a simple body plan defined by a single oral-aboral axis, with tentacles and a hypostome at the oral end, referred to as the head, and adhesive cells at the aboral end, referred to as the foot (Figure 1A). In intact Hydra, tissue polarity is dictated by two stable self-maintaining organizers, a head organizer and a foot organizer, that are found at the poles of the oral-aboral axis (Webster, 1971). Although the molecular basis for the foot organizer is unknown, the Hydra head organizer is based on canonical Wnt signaling; the head organizer expresses multiple Wnt ligands (Hobmayer et al., 2000; Lengfeld et al., 2009), and Wnt signaling is necessary and sufficient for the specification of oral tissue (Broun et al., 2005; Gee et al., 2010; Gufler et al., 2018). Consistently, regenerating Hydra form new Wnt organizers at oral-facing, but not aboral-facing injuries within the first 12 hpa (MacWilliams, 1983a; Petersen et al., 2015). Currently, there are conflicting data on the function and dynamics of Wnt signaling during cnidarian whole-body regeneration. In addition, the mechanism by which Wnt signaling is activated during regeneration is not well understood. These factors make it difficult to meaningfully compare the injury-induced Wnt signaling activity observed during Hydra regeneration to other whole-body regeneration models.

Figure 1 with 4 supplements see all
The initial transcriptional response to mid-gastric bisection is structure independent but diverges by 8 hr post amputation (hpa).

(A) A diagram of the Hydra body plan. Red coloration indicates a head-specific molecular signature. Blue coloration indicates a foot-specific molecular signature. (B) Experimental design for ATAC-seq and RNA-seq library generation. (C–H) Comparison of average log2 fold change (log2FC) in transcript abundance (C, E, G) or chromatin accessibility (D, F, H) during head and foot regeneration at 3, 8, and 12 hpa. The transcriptional response during head and foot regeneration is identical at 3 hpa but becomes distinct by 8 hpa. Features that showed significant differences in the injury response during head and foot regeneration when compared to 0 hpa controls are highlighted in red for head regeneration-specific features or blue for foot regeneration-specific features (false discovery rate [FDR] ≤ 1e-3 for RNA-seq and FDR ≤ 1e-4 for ATAC-seq). The dotted line indicates perfect correlation between regeneration types. All significance values were calculated with edgeR using quasi-likelihood tests of count data fitted to a negative binomial generalized log-linear model. Information on RNA-seq and ATAC-seq biological replicates is provided in Tables 1 and 2. Individual FDR values for all genewise/peakwise tests can be found in Supplementary files 1 and 2. (I) Heatmap of the average relative accessibility of transcription factor binding motifs during regeneration in head and foot regenerates calculated using chromVAR. Transcription factor activity during the first 12 hr of regeneration was highly dynamic and largely similar between head and foot regeneration. (J–M) Late-stage apoptotic cells labeled in Hydra undergoing head or foot regeneration using acridine orange. Widespread apoptosis was observed during head and foot regeneration at 3 hpa but was largely absent at 1 hpa. Additional representative images are presented in Figure 1—figure supplement 4. Scale bars indicate 250 µm.

Figure 1—source data 1

Excel workbook containing all tables that were used to generate plots for Figure 1C–I.

The first sheet in the workbook (‘Descriptions’) contains individual descriptions for all tables included in the workbook.

https://cdn.elifesciences.org/articles/60562/elife-60562-fig1-data1-v2.xlsx
Figure 1—source data 2

Transcription factor binding motifs included in this study.

The sequence, shortened name, and full name for all transcription factor binding motifs (TFBMs) that were associated with significant changes in chromatin accessibility during the first 12 hr of head and foot regeneration. All TFBMs were pulled from the list of HOMER motifs provided by the chromVARmotifs package.

https://cdn.elifesciences.org/articles/60562/elife-60562-fig1-data2-v2.docx

One of the primary points of uncertainty regarding injury-induced Wnt activity in the Hydra literature relates to the timing and location of Wnt signaling activation during regeneration. Early grafting experiments in Hydra found that donor tissue fragments taken from recent head and foot amputations had an increased capacity to induce ectopic heads in host tissue, suggesting that head-specifying pathways are initially activated regardless of the structure being regenerated (MacWilliams, 1983a; Müller, 1996). More recently, Gufler et al., 2018 found that several head-specific transcripts, including β-catenin, were initially upregulated during both head and foot regeneration in Hydra. They also reported that inhibiting TCF—the downstream transcriptional effector of Wnt signaling—prevented head and foot regeneration, raising the possibility that Wnt signaling may play an active role in both types of regeneration. In a pre-print manuscript, Wenger et al., 2019 compared transcriptomic data collected during head and foot regeneration and found that transcriptional differences did not arise until between 4 and 8 hpa, indicating that the onset of head-specific transcription occurs several hours after wounding. However, these findings are contrasted by data from Chera et al., 2009, who reported that injury-induced apoptosis—which they found occurred only at oral-facing amputations—leads to the activation of Wnt signaling during head but not foot regeneration by 1.5 hpa. These results appear inconsistent with reports of symmetric transcription during early head and foot regeneration as this would require that the dramatic asymmetries between the two types of regeneration at 1.5 hpa are not associated with any concomitant transcriptional changes.

The available data on Hydra regeneration leave significant gaps in knowledge that make it difficult to effectively understand the mechanisms underlying patterning during regeneration. In particular, the role of canonical Wnt signaling during Hydra regeneration remains incompletely characterized. Furthermore, it is unclear when and how canonical Wnt signaling genes, particularly Wnt ligands and negative regulators of Wnt signaling, are upregulated during head and foot regeneration following injury. We therefore sought to gain insight into these questions by comprehensively characterizing changes in chromatin accessibility and transcript abundance during the first 12 hr of head and foot regeneration—the period when new organizers are being formed. We found that the initial transcriptional response was indistinguishable during head and foot regeneration and included the symmetric upregulation of apoptotic markers and canonical Wnt signaling components. This was correlated with a symmetric increase in chromatin accessibility near predicted TCF binding sites, suggesting that Wnt signaling was initially activated during both head and foot regeneration. By 8 hpa, head and foot regeneration became transcriptionally distinct, which coincided with Wnt signaling transcripts becoming restricted to head regeneration. Inhibiting TCF delayed this transcriptional divergence genome-wide, suggesting a central role for canonical Wnt signaling during oral and aboral regeneration. A systematic analysis of chromatin accessibility data found that injury-induced Wnt component upregulation was likely driven by conserved injury-responsive basic leucine zipper (bZIP) TFs. In support of a link between the generic injury response and canonical Wnt signaling, we found that non-amputation injuries also triggered the upregulation of Wnt pathway components and could induce ectopic head formation when pre-existing organizers were removed. In addition, prolonged aboral-facing amputation injuries also induced ectopic head formation, but only after inhibitory signals from head-regenerating tissue were removed. Overall, these findings suggest that injuries initiate an oral patterning cascade regardless of the surrounding tissue context through the activation of canonical Wnt signaling, and that the outcome of this activation is regulated by long-range signals generated by organizers. These results also suggest that the upregulation of canonical Wnt signaling components may be part of an ancestral metazoan injury response.

Results

The initial transcriptional response to injury does not depend on the structure being regenerated

To better understand the transcriptional regulatory mechanisms underlying positional divergence during head and foot regeneration in Hydra, we generated ATAC-seq (Buenrostro et al., 2013; Corces et al., 2017) and RNA-seq libraries from polyps undergoing head and foot regeneration at 0, 3, 8, and 12 hr after mid-gastric bisection (Figure 1B, Figure 1—figure supplements 13, Tables 1 and 2). In designing our experiments, we focused on the first 12 hr of regeneration because new organizers are established by approximately 12 hpa (Hicklin and Wolpert, 1973; MacWilliams, 1983a). ATAC-seq reveals accessible regions of chromatin, allowing us to identify active cis-regulatory elements genome-wide. In addition, comparing ATAC-seq data from different timepoints allows us to characterize changes in cis-regulatory element activity over the course of regeneration. The inclusion of RNA-seq data allows us to correlate changes in chromatin accessibility to changes in transcript abundance in nearby genes. Thus, these data enable the characterization of gene regulatory network activity during head and foot regeneration that started from the same initial injury.

Table 1
ATAC-seq library statistics.

A total of 71 ATAC-seq libraries were generated, with between 3 and 5 biological replicates per treatment. HPA refers to hours post amputation. + in the iCRT column indicates regenerating animals were pre-incubated in 5 µM iCRT14 for 2 hr prior to amputation and then left in the iCRT14 solution until tissue was collected for library preparation; – in the iCRT column indicates animals were left untreated. Total read pairs refers to the number of raw read pairs generated for each library. Final mapped read pairs refers to the number of read pairs remaining after mitochondrial, duplicated, and unmapped or ambiguously mapped reads were removed. Transcription start sites (TSS) enrichment refers to the fold enrichment in ATAC-seq signal at the TSS of 2000 highly expressed genes relative to regions ±1 kb from the TSS. Reproducible peaks refers to the number of peaks within each treatment group that were biologically reproducible in at least three pairwise comparisons using an irreproducible discovery rate cutoff of 0.1. Benchmarks for both the TSS score and the number of reproducible peaks are highly dependent on the model system and the quality of the reference annotation, but in well-studied systems such as mice, ENCODE considers TSS scores > 5 and >50,000,000 reproducible peaks to be acceptable. The self-consistency ratio refers to the largest fold difference in the number of reproducible peaks recovered from pseudo-replicates (see Materials and methods) when comparing biological replicates within a single treatment group. The rescue ratio refers to the fold difference in the number of reproducible peaks recovered from a dataset with perfect reproducibility (generated by pooling and then randomly splitting reads from a given treatment group) compared to the number of true biologically reproducible peaks. High-quality datasets should have self-consistency ratios and rescue ratios <2.

Sample IDHPAiCRTTotal read pairsFinal mapped read pairsTSS enrichmentReproducible peaksSelf-consistency ratioRescue ratio
0F1068865654318450476.15781581.171.00
0F2065301503219709637.16
0F3080642153258439286.61
0F4076300368259255956.79
0F5081363105207147656.79
0H1072424905316766796.31823961.171.04
0H2068870186273769676.81
0H3081121398310849666.50
0H4067378757260761246.58
0H5081380944262473906.44
3F13120749030357472016.44579801.111.20
3F23133453367375511246.85
3F33128451210448070826.76
3H13127697432332011247.45582591.091.15
3H23124784335385171607.09
3H33130420684367034977.05
8F1871642160202709177.30842481.291.05
8F2877325666179682457.83
8F3875894343173913397.15
8F4871117490256538376.98
8F5874203082306144247.17
8H1872411934187146497.18839741.181.04
8H2876210411166187197.78
8H3874495918222079726.53
8H4868874107283970856.78
8H5885121623263334876.31
12F11235989738163713647.81719491.091.00
12F21237815306163337327.26
12F31231696505101755187.47
12F41235835498103837728.43
12F5122775339294588387.78
12H11246530322172252906.98714631.161.05
12H21233411133162313896.68
12H3123812807690531488.13
12H4123703878580697337.94
12H5123550814890772957.10
0iF10+29450456139335916.99683931.131.03
0iF20+47282308239058226.18
0iF30+54230058253443686.02
0iF40+40244798189543946.43
0iH10+64726838269529965.79707711.081.05
0iH20+46838471222469786.8
0iH30+40909454189033706.87
0iH40+54067924251193466.35
3iF13+54065239108869195.96673911.161.06
3iF23+48714606144706866.55
3iF33+44688328162948106.71
3iF43+36970695157378345.81
3iF53+32077041120655425.3
3iH13+4690354890574277.19625791.121.00
3iH23+41743618156545876.79
3iH33+38888634124840176.85
3iH43+45598782162424666.54
8iF18+50536148134301126.78611391.261.05
8iF28+54337061126484407.11
8iF38+52282170118529485.21
8iF48+42891267121204235.43
8iH18+42290670117963876.72650431.571.10
8iH28+39596909149989185.20
8iH38+39883594125634375.27
8iH48+2824929283936014.57
8iH58+43953185107994517.85
12iF112+39257829138794177.14705051.051.01
12iF212+42214218162597217.39
12iF312+40921039105365617.64
12iF412+45775901125129027.54
12iF512+36779064102691917.39
12iH112+43772456188069816.49654921.141.01
12iH212+37690946152704946.95
12iH312+42788391114355787.54
12iH412+42138119150162986.48
Table 2
RNA-seq library statistics.

A total of 42 RNA-seq libraries were generated, with three biological replicates per treatment. HPA refers to hours post amputation. + in the iCRT column indicates regenerating animals were pre-incubated in 5 µM iCRT14 for 2 hr prior to amputation and then left in the iCRT14 solution until tissue was collected for library preparation; – in the iCRT column indicates animals were left untreated. Total reads refers to the number of raw reads generated for each library. Final mapped reads refers to the number of reads that were successfully mapped to the Hydra 2.0 genome gene models.

Sample IDHPAiCRTTotal readsFinal mapped reads
0F102928400421541719
0F202459736017966439
0F302392634117350620
0H102583310118858726
0H202724125819701074
0H302680309019487287
3F132482446818479637
3F232384050517694124
3F332390053217743737
3H132587457019207371
3H232417231017898155
3H332339793717670977
8F182304034417160851
8F282601955719216264
8F382937812322033550
8H182554880119003624
8H282160115016154188
8H382390401117656019
12F1122699520418138493
12F2122370352715445837
12F3122548824516491319
12H1123487815422910100
12H2123166257121545505
12H3122819228818744249
0iF102663548517851347
0iF20+2805216918904815
0iF30+3071837020708864
0iH10+3632807623784082
0iH20+2604443917240456
0iH30+3138407320624671
8iF18+3797973224369941
8iF28+3172206120432162
8iF38+3825215525503172
8iH18+3452212022339316
8iH28+3444270521537194
8iH38+2972702219082280
12iF112+2235428913460197
12iF212+3514163721948916
12iF312+3524087922241372
12iH112+2588330516660844
12iH212+3223779020843729
12iH312+7311372844132834

We first sought to identify the earliest transcriptional differences that arose between head and foot regeneration. To do this, we identified transcripts and ATAC-seq peaks that changed in significantly different ways during the two types of regeneration (full differential gene expression results are available in Supplementary files 1 and 2). At 3 hpa, although we detected extensive changes in chromatin accessibility and transcript abundance relative to uninjured controls, there were no significant differences in those changes when comparing head and foot regeneration (Figure 1C, D). However, by 8 hpa there was clear evidence of head and foot regeneration-specific transcriptional responses, with 63 transcripts and 464 peaks showing significant differences when comparing head and foot regeneration (Figure 1E, F). The number of transcriptional differences increased at 12 hpa, with 139 transcripts and 631 peaks exhibiting differential activation during the two types of regeneration (Figure 1G, H). We postulate that we observe a greater number of differentially activated peaks relative to transcripts because gene loci are often associated with multiple peaks. In addition, changes in certain individual cis-regulatory elements may not be sufficient to drive detectable differences in transcript abundance. Overall, our analysis demonstrates that the initial transcriptional response to amputation in Hydra does not depend on the structure being regenerated, and that transcriptional differences between head and foot regeneration begin to arise between 3 and 8 hpa.

To evaluate the congruence of our ATAC-seq and RNA-seq data, we used a hypergeometric enrichment test to determine if ATAC-seq peaks that were identified as head or foot regeneration-specific were significantly enriched near head or foot regeneration-specific gene loci. We found that structure-specific transcripts were significantly enriched in nearby structure-specific peaks (14.1-fold enrichment at 8 hpa, p=3.6e-25; 9.8-fold enrichment at 12 hpa, p=3.7e-40), demonstrating a meaningful correspondence between our ATAC-seq and RNA-seq datasets. We also compared our RNA-seq data with a previously released transcriptomic dataset of Hydra head and foot regeneration and found that there was a significant overlap in the head and foot regeneration-specific transcripts identified in the two studies (507-fold enrichment at 8 hpa, p~0; 35-fold enrichment at 12 hpa, p=1.11e-93) (Wenger et al., 2019). We therefore found that the identification of transcriptional asymmetries during regeneration was highly reproducible across independent studies and across orthogonal next-generation sequencing methodologies.

We next used our chromatin accessibility data to identify candidate TFs that could plausibly underlie the observed transcriptional changes during regeneration. To do this, we used chromVAR (Schep et al., 2017) to systematically identify transcription factor binding motifs (TFBMs) throughout the genome and characterize how the aggregate chromatin accessibility near those motifs changed over the course of regeneration. Using this approach, we identified 32 TFBMs that were associated with significant variability in chromatin accessibility during the first 12 hr of regeneration (Figure 1I; TFBM sequences are provided in Figure 1—source data 2). Consistent with our observation that the majority of transcriptional changes during head and foot regeneration were highly similar, we found that the large majority of injury responsive TFBMs had similar accessibility dynamics during oral and aboral regeneration. These data suggest that injury-responsive gene regulatory networks behave in a largely symmetric manner during early head and foot regeneration. Notably, we found that the TFBMs associated with increases in accessibility from 0 to 3 hpa included motifs associated with TFs known to play a role in the early injury response in bilaterians such as AP-1, Atf, and Egr (Schäfer and Werner, 2007). The likely involvement of these TFs in Hydra regeneration suggests that they have an ancestral role in the metazoan wound response that predates the split of cnidaria and bilateria.

Mid-gastric bisection induces extensive apoptosis in oral and aboral-facing wounds by 3 hpa

Although our finding of an initially symmetric transcriptional response during head and foot regeneration is consistent with other transcriptional studies in Hydra (Gufler et al., 2018; Wenger et al., 2019), it appears to be inconsistent with a previous report of extensive apoptosis occurring only at oral-facing amputations by 1 hpa (Chera et al., 2009). Apoptosis induces dramatic changes in gene expression (Sun et al., 2017), and asymmetric apoptosis would therefore be expected to result in dramatic transcriptional asymmetries during early head and foot regeneration. To reconcile these seemingly discrepant observations, we sought to characterize apoptosis during early head and foot regeneration using acridine orange, a previously validated stain for late-stage apoptotic cells in Hydra (Cikala et al., 1999; Kuznetsov et al., 2002; Technau et al., 2003). In contrast to previous findings, we found no evidence of head regeneration-specific apoptosis at 1 hpa (Figure 1J, L, Figure 1—figure supplement 4A). Rather, at this timepoint both head and foot regeneration exhibited limited apoptosis and appeared largely similar to uninjured controls. However, by 3 hpa, we observed extensive apoptosis throughout the body in both head and foot regenerating tissue fragments (Figure 1K, M, Figure 1—figure supplement 4A). Consistent with this observation, our transcriptomic data showed that numerous apoptotic markers such as caspase-3c, bcl-like-6 (Lasi et al., 2010), p53, and bax were upregulated in a similar fashion during head and foot regeneration at 3 hpa (Figure 1—figure supplement 4B–I; the Hydra 2.0 genome gene model IDs for the gene names used throughout this study can be found in Table 3). These data indicate that amputations induce apoptosis in a symmetric manner at both oral and aboral-facing injuries by 3 hpa, consistent with a generic initial response to injury. This conclusion is also consistent with the findings reported by Tursch et al., 2020, who found no evidence of head regeneration-specific apoptosis at 2 hpa following mid-gastric bisection.

Table 3
Genome gene model IDs for in-text gene names.

The Hydra 2.0 genome gene model IDs that correspond to the gene names used in this study. The Hydra 2.0 genome gene models can be found at arusha.nhgri.nih.gov/hydra/download/?dl=n.

Gene nameGenome gene model ID
axing14938
baxg7775
bcl-like-6g22655
brachyury1g24952
budheadg24126
caspase3-likeg664
caspase-3cg31854
cr3lg32585
crebg16491
dapk2-likeg14048
dff40g11021
dishevelledg29781
distal-lessg18245
fosg23720
foxd2-likeg28449
fzd4/9/10g27891
gata-3-likeg20911
gremlin-likeg14229
hsp70-likeg16168
jung1450
naked cuticleg30442
nk-2g31954
notumg26902
prdl-ag15226
p53g21693
sFRPg12274
sp5g33422
tcfg27364
wnt3g28064
wnt7g29750
wnt9/10cg33373
wntlessg18842

Wnt signaling is likely activated during early head and foot regeneration, but becomes head specific by 8 hpa

For regeneration to rebuild oral and aboral structures, part of the injury response must include the reactivation of the pathways that specify and maintain these structures under steady-state conditions. Therefore, to better understand how transcriptional symmetry is broken during regeneration, we looked for evidence of the reactivation of head and foot-specific pathways to identify the genes involved in the early stages of new head or foot organizer formation. This required that we first identify genes associated with uninjured head and foot tissue. To do this, we used a previously published single-cell RNA-seq atlas (Siebert et al., 2019) to identify genes that showed significantly different expression when comparing epithelial head cell clusters to epithelial foot cell clusters. We then characterized changes in the transcript abundance of these genes along with changes in chromatin accessibility in nearby peaks during regeneration.

At 8 hpa, the earliest time point in our dataset with transcriptional differences between head and foot regeneration, there were 55 transcripts and 426 peaks that showed significantly higher activation during head regeneration relative to foot regeneration. We found that 23/55 of the head regeneration-specific transcripts also exhibited head-specific expression in uninjured polyps, and 65/426 of the head regeneration-specific peaks were located near genes that exhibited head-specific expression in uninjured polyps (Figure 2A, B). These early head-specific factors included previously characterized genes associated with the Hydra head such as wnt9/10c (Lengfeld et al., 2009), wnt3 (Hobmayer et al., 2000), brachyury1 (Technau and Bode, 1999), naked cuticle (Petersen et al., 2015), and sp5 (Vogg et al., 2019). In contrast, we did not observe a similar activation of foot-specific transcription at 8 hpa, although there was some evidence of foot-specific chromatin remodeling (Figure 2A, B). At 12 hpa, however, several genes enriched in homeostatic foot tissue, including the known foot-associated TFs distal-less (Hemmrich et al., 2012) and nk-2 (Grens et al., 1996; Figure 2C, D), were upregulated in a foot regeneration-specific manner. We also found two previously undescribed foot-specific TFs, gata-3-like and foxd2-like, that were upregulated during foot regeneration by 12 hpa. We hypothesize that these early context-specific genes reside at the top of the regulatory hierarchy directing the specification of head or foot tissue during regeneration.

Figure 2 with 1 supplement see all
Wnt signaling is likely initially activated in a structure-independent manner but becomes restricted to regenerating head tissue by 8 hr post amputation (hpa).

(A–D) Plots comparing enrichment in head and foot tissue in uninjured Hydra and during regeneration. Genes associated with the homeostatic maintenance of head and foot organizers are upregulated within 12 hpa. Enrichment in uninjured head and foot tissue (i.e., single-cell [SC] Atlas Structural Enrichment) was calculated by comparing the average relative expression levels in epithelial head cells isolated from the Hydra SC atlas to expression in epithelial foot cells. A positive fold change indicates enrichment in head tissue, and a negative fold change indicates enrichment in foot tissue. Head or foot-specific activation during regeneration (i.e., Regeneration Context Effect) was calculated by subtracting the log2FC in regenerating head tissue from the log2FC in regenerating foot tissue using the 0 hpa timepoint as the control. A positive fold change indicates enrichment in head-regenerating tissue, and a negative fold change indicates enrichment in foot-regenerating tissue. Transcripts (A, C) or ATAC-seq peaks (B, D) that were enriched in head-regenerating tissue and were also enriched in uninjured head tissue are highlighted in red. ATAC-seq peaks or transcripts that were enriched in foot-regenerating tissue and were also enriched in uninjured foot tissue are highlighted in blue. (E) Average relative chromatin accessibility plot of ATAC-seq peaks containing the TCF transcription factor binding motif (TFBM) during head and foot regeneration. The TCF TFBM is associated with increases in accessibility during both head and foot regeneration at 3 hpa, but subsequent increases in accessibility are restricted to head regeneration. Relative changes in chromatin accessibility were calculated using chromVAR using the HOMER TCFL2 TFBM sequence. (F–L) RNA expression plots showing average normalized RNA-seq read counts for Wnt signaling components in log2 counts per million (log2CPM) during head and foot regeneration. Wnt signaling components are initially upregulated during head and foot regeneration but become head regeneration-specific by 12 hpa. HR: head regeneration; FR: foot regeneration.

Figure 2—source data 1

Excel workbook containing all tables that were used to generate plots for Figure 2A–D.

The first sheet in the workbook (‘Descriptions’) contains individual descriptions for all tables included in the workbook.

https://cdn.elifesciences.org/articles/60562/elife-60562-fig2-data1-v2.xlsx

We then used chromVAR to identify candidate TFBMs that could plausibly be involved in the onset of head and foot regeneration-specific transcription by looking for TFBMs associated with asymmetric changes in chromatin accessibility. We found that the TFBM for TCF, the downstream transcriptional effector of canonical Wnt signaling, was among the small number of motifs associated with asymmetric accessibility, exhibiting head regeneration-specific increases by 8 hpa onward (Figure 1I, Figure 2E). Because TCF constitutively binds to target loci and activates transcription in part by recruiting chromatin remodeling enzymes that increase chromatin accessibility (Cadigan, 2012), we interpret these data as indicative of a head regeneration-specific increase in TCF transcriptional activation. Thus, our analysis suggests that canonical Wnt signaling is activated in a head regeneration-specific manner by 8 hpa and is among the earliest asymmetries in gene regulatory network activity during regeneration, consistent with its important role in specifying oral tissue (Broun et al., 2005; Gee et al., 2010; Hobmayer et al., 2000; Lengfeld et al., 2009).

While examining the chromatin accessibility dynamics associated with TCF, we noted that although chromatin accessibility near TCF TFBMs increased in a head regeneration-specific manner from 8 hpa onward, at 3 hpa the TCF TFBM was associated with symmetric accessibility increases during both head and foot regeneration (Figure 2E). This raised the possibility that canonical Wnt signaling may be activated as part of the initial symmetric response to injury. To further explore the plausibility of this hypothesis, we used KEGG pathway annotations to systematically identify and characterize the expression dynamics of canonical Wnt pathway components during regeneration (Kanehisa and Goto, 2000). We found that several canonical Wnt signaling components—including wnt9/10c, wnt3, wnt7, wntless, dishevelled, β-catenin, and sp5—were first upregulated during the initial symmetric phase of regeneration and did not become head regeneration specific until 8 or 12 hpa (Figure 2F–L, Figure 2—figure supplement 1). These observations demonstrate that, similar to the planarian wound response (Gurley et al., 2010; Petersen and Reddien, 2009; Stückemann et al., 2017; Wurtzel et al., 2015), canonical Wnt signaling components are upregulated as part of the early generic injury response in Hydra. In addition, it suggests that the upregulation of these genes initially results in the activation of Wnt signaling during both head and foot regeneration, but that Wnt signaling becomes restricted to head regeneration by 8 hpa.

TCF inhibition delays the onset of head and foot-specific transcription during regeneration transcriptome-wide

The possible activation of Wnt signaling during both head and foot regeneration raises questions about the pathway’s function during regeneration. Wnt signaling has a well-established function specifying oral tissue during regeneration (Broun et al., 2005; Gufler et al., 2018; Lengfeld et al., 2009), but the role of Wnt signaling during aboral regeneration is poorly understood. Recent work by Gufler et al., 2018 found that the small molecule inhibitor iCRT14, which both inhibits the interaction between β-catenin and TCF and disrupts TCF binding to DNA (Gonsalves et al., 2011), blocked head and foot regeneration and prevented the downregulation of several head-specific transcripts during foot regeneration; however, because this study only characterized a small number of genes, the broader function of TCF in regulating transcription during regeneration remains incompletely understood. We therefore sought to characterize the impact of TCF inhibition during regeneration across the entire transcriptome by generating RNA-seq and ATAC-seq libraries from head and foot regenerating tissue fragments treated with iCRT14.

Before performing the regeneration timecourse experiments, we first validated that our iCRT14 treatment conditions yielded results consistent with previously published findings (Gufler et al., 2018). To do this, we tested the ability of iCRT14 to inhibit head and foot regeneration. We found that 5 µM iCRT14 significantly inhibited both head and foot regeneration relative to dimethyl sulfoxide (DMSO)-treated controls (Figure 3—figure supplement 1A, B), thus recapitulating previously reported results. We therefore proceeded with generating ATAC-seq and RNA-seq libraries from bisected animals undergoing either head or foot regeneration in the presence of 5 µM iCRT14, repeating the injury conditions used for our untreated sequencing libraries. We collected 0, 8, and 12 hpa timepoints for the RNA-seq timecourse, and 0, 3, 8, and 12 hpa timepoints for the ATAC-seq dataset.

We first determined if our RNA-seq data recapitulated the previously reported finding that TCF inhibition prevents the downregulation of head-specific genes during foot regeneration. Indeed, as reported by Gufler et al., 2018, we found that iCRT14 inhibited the downregulation of the head-specific genes brachyury1 and β-catenin in foot regenerating tissue (Figure 3—figure supplement 1C, D). We next validated our ATAC-seq data by assessing the effects of iCRT14 on chromatin accessibility near TCF TFBMs to determine if we could detect perturbations in TCF-dependent chromatin remodeling. Our chromVAR analysis revealed that iCRT14 treatment resulted in a diminished and delayed increase in average TCF TFBM accessibility during head regeneration (Figure 3A), thus demonstrating our ability to capture changes in chromatin state induced by TCF inhibition.

Figure 3 with 2 supplements see all
TCF is required for the initiation of head and foot-specific transcription during regeneration.

(A) Average relative chromatin accessibility plot of peaks containing the TCF transcription factor binding motif (TFBM) during head and foot regeneration in both untreated animals and animals incubated in 5 µM iCRT14. iCRT14 treatment significantly diminished and delayed increases in chromatin accessibility associated with the TCF TFBM during head regeneration. The HOMER TCFL2 motif sequence was used to identify putative TCF binding sites. (B, C) Scatter plots depicting the effect of 5 µM iCRT14 during foot regeneration on the injury induced expression of genes that were specific to head or foot regeneration in untreated animals. TCF inhibition prevents the upregulation of foot regeneration-specific genes and prolongs the expression of head-specific genes during foot regeneration. Negative values on the x-axis indicate a decrease in the injury-induced expression of a gene in iCRT14-treated animals relative to untreated controls, while positive values indicate an increase in injury-induced expression. Negative values on the y-axis indicate that the gene was downregulated relative to 0 hr post amputation (hpa) controls in iCRT14-treated animals, while positive values indicate that the gene was upregulated. Green coloration denotes transcripts that showed diminished injury-induced expression in iCRT14-treated animals (i.e., expression was TCF-dependent). Purple coloration denotes transcripts that showed enhanced injury-induced expression in iCRT14-treated animals (i.e., expression was inhibited by TCF). TCF-dependent transcripts (green) were defined as transcripts that were not significantly upregulated following injury in iCRT14-treated animals and that showed a significant (false discovery rate [FDR] ≤ 1e-3) reduction in injury-induced expression relative to untreated animals. TCF-inhibited transcripts (purple) were defined as transcripts that were significantly upregulated following injury in iCRT14-treated animals and that showed a significant increase in injury-induced expression relative to untreated animals. (D) Average relative chromatin accessibility plot of peaks containing the AP-1 TFBM during head and foot regeneration in both untreated and iCRT14-treated samples. iCRT14 treatment prolonged the injury-induced increase in chromatin accessibility associated with the AP-1 TFBM (i.e., 12-O-tetradecanoylphorbol 13-acetate[TPA] response element [TRE]) during regeneration. (E, F) Scatter plots depicting the effect of 5 µM iCRT14 during head regeneration on the injury-induced expression of genes that were specific to head regeneration in untreated animals. TCF inhibition blocks the upregulation of head-specific genes during head regeneration. (G, H) Comparison of the average log2FC in (G) transcript abundance or in (H) chromatin accessibility between head and foot regenerates treated with 5 µM iCRT14 at 8 hpa. iCRT14 virtually abolished all context-specific transcription at 8 hpa. Features that did not show a significant (FDR ≤ 1e-3) difference in the injury response between head and foot regenerates when compared to 0 hpa controls are highlighted in gray. Features that did show a significant difference are highlighted in red. The dotted line indicates perfect correlation between regeneration types.

Figure 3—source data 1

Excel workbook containing all tables that were used to generate plots for Figure 3C–H.

The first sheet in the workbook (‘Descriptions’) contains individual descriptions for all tables included in the workbook.

https://cdn.elifesciences.org/articles/60562/elife-60562-fig3-data1-v2.xlsx

After validating the data in our iCRT14-treated dataset, we next sought to determine if other head-specific genes beyond those reported by Gufler et al., 2018 were also inappropriately expressed during foot regeneration when TCF was inhibited. Indeed, we found that several additional head-specific genes—most notably the canonical Wnt signaling components wnt9/10c, wnt3, and sp5—were not appropriately downregulated during foot regeneration by 8 hpa (Figure 3B). Our data therefore support the conclusion that TCF is required for the downregulation of a subset of head-specific genes during foot regeneration.

In addition, we found that TCF was also required for the downregulation of transcripts associated with the initial wound response observed during both head and foot regeneration. Several TFs that were transiently upregulated at 3 hpa in untreated animals—including the bZIP-domain containing TFs creb, jun, and fos—showed prolonged upregulation when TCF was inhibited (Figure 3—figure supplement 2A–C). In addition, we found that iCRT14 caused an increase in chromatin accessibility near the injury-induced TPA response element (TRE), which is bound by bZIP TFs, in both head and foot regenerates at 8 and 12 hpa (Figure 3D). We also observed similar increases for the injury-responsive Egr TFBM (Figure 3—figure supplement 2D). These data indicate that TCF is required for the downregulation of early wound-responsive TFs during the transition from the generic wound response to head and foot-specific transcription.

Although Wnt signaling is known to be both necessary and sufficient for head specification in Hydra, it is not known which of the transcripts expressed during head regeneration are TCF-dependent. To address this question, we identified head regeneration-specific transcripts that showed reduced upregulation in the presence of iCRT14. We found that TCF was required for the upregulation of numerous head-specific transcripts, including several Wnt signaling components such as sFRP, frizzled4/9/10, axin, and tcf itself (Figure 3E, F). This potential autoregulatory relationship is consistent with the existence of a Wnt-based feedback loop in Hydra, which is thought to provide the molecular basis for the self-maintaining properties of the head organizer (Gee et al., 2010; Nakamura et al., 2011). Other notable TCF-dependent head-specific transcripts included the TFs budhead (Martinez et al., 1997) and prdl-a (Gauchat et al., 1998) and the bmp inhibitor gremlin-like (Watanabe et al., 2014). The diminished expression of these transcripts highlights potential interactions between Wnt signaling and other regulatory genes implicated in Hydra patterning.

We found that TCF inhibition also blocked foot regeneration-specific transcription. Most significantly, all TFs that were enriched in foot regenerating tissue at 12 hpa in untreated animals showed decreased expression in the presence of iCRT14 (Figure 3C, D). This suggests a previously unknown role for TCF in the activation of foot-specific transcriptional programs during regeneration and provides a possible molecular basis for the inhibition of foot regeneration by iCRT14 (Gufler et al., 2018).

Given the apparent requirement for TCF in the activation of both head and foot-specific transcriptional programs, we next examined how TCF inhibition influenced transcriptional divergence transcriptome-wide. At 8 hpa, we found that iCRT14 blocked virtually all structure-specific transcription, with transcript abundance and chromatin accessibility in head and foot regenerates appearing nearly indistinguishable (Figure 3G, H). At 12 hpa, however, we began to see indications of context-dependent transcription in both our chromatin accessibility and transcriptomic data (Figure 3—figure supplement 2E, F). These data therefore demonstrate a global delay in transcriptional divergence and suggest that TCF plays a critical role in the transition from the generic wound response to head and foot-specific transcription during regeneration. This may also indicate that the transient activation of Wnt signaling during early foot regeneration is required to potentiate aboral patterning.

A conserved injury-responsive gene regulatory network may directly activate Wnt signaling during regeneration

Our observation that iCRT14 inhibited the expression of some Wnt signaling components during regeneration but not others suggested the existence of two distinct regulatory paradigms for injury-induced Wnt component expression: a TCF-dependent mode of activation and a TCF-independent mode of activation. We also noted that the genes exhibiting TCF-independent upregulation were primarily those that were upregulated as part of the early generic injury response, such as wnt9/10c, wnt3, wntless, sp5, and β-catenin. We therefore hypothesized that these transcripts may be upregulated by injury-responsive TFs, which could account for their expression when TCF was inhibited.

To identify injury-responsive TFs that could plausibly regulate Wnt signaling components during the generic injury response, we performed a systematic candidate screen using our ATAC-seq and RNA-seq datasets. A TF was considered to be a likely candidate if it matched the following criteria when comparing 0 and 3 hpa timepoints: (1) its binding motif was associated with an injury-induced increase in chromatin accessibility, (2) it was transcriptionally upregulated following injury, (3) its binding motif was significantly enriched in peaks that increased in accessibility following injury, and (4) its binding motif was found in predicted cis-regulatory elements near injury-responsive Wnt signaling components. Although these criteria do not definitively eliminate any given TF as a possible regulator, they can provide a basis for evaluating the relative plausibility of different candidates.

Our screen identified ETS, c/EBP, AP-1, and ATF TFBMs as candidate motifs driving injury-induced expression of at least one Wnt signaling component (Figure 4A). Among our list of candidates, the ATF TFBM (also known as the cAMP response element [CRE])—which can be bound by the bZIP TFs Jun, Fos, CREB, and ATF (Eferl and Wagner, 2003)—best matched our criteria. Of all TFBMs included in our analysis, the CRE TFBM was associated with the largest increase in accessibility from 0 to 3 hpa and showed the strongest enrichment in injury-responsive peaks (Supplementary file 3). These chromatin changes were also correlated with the upregulation of several bZIP TFs that are predicted to bind the CRE, including jun, fos, and two CREB-like TFs (Figure 4—figure supplement 1A–D). Furthermore, we identified instances of the CRE in the putative regulatory sequences of wntless, wnt9/10 c, wnt3, and sp5. In addition, the CRE-containing peaks near wntless, wnt9/10c, and wnt3 increased in accessibility from 0 to 3 hpa during head and foot regeneration, indicating increased activity within those cis-regulatory elements (Figure 4B–D). Overall, this analysis indicates that injury-responsive bZIP TFs are the most plausible regulators of canonical Wnt signaling components during the early generic wound response. This potential association between injury-responsive TFs and positive regulators of canonical Wnt signaling provides a mechanistic hypothesis for the link between injury and the reactivation of patterning pathways during Hydra regeneration.

Figure 4 with 1 supplement see all
Canonical Wnt signaling genes may be directly activated by injury-responsive transcription factors.

(A) Plot depicting which injury-responsive transcription factor binding motifs (TFBMs) matched the criteria of a plausible regulator of injury-induced Wnt component expression. The AP-1, Atf7, c/EBP, and Elf1 TFBMs are plausible candidate regulators of injury-induced Wnt component expression. ‘Accessibility ↑’ refers to TFBMs that were associated with a significant increase in chromatin accessibility at 3 hr post amputation (hpa) as calculated by chromVAR. ‘TF ↑’ refers to TFBMs with at least one transcription factor that could plausibly bind that TFBM showing significant upregulation during head and foot regeneration at 3 hpa. ‘Enriched’ refers to TFBMs that were significantly enriched in peaks that showed significant increases in accessibility during head and foot regeneration at 3 hpa when compared to peaks that did not increase in accessibility using HOMER. The remaining rows indicate whether or not a given TFBM was found in ATAC-seq peaks located near each respective injury-responsive Wnt signaling component. (B–D) ATAC-seq accessibility data for Wnt signaling gene loci during regeneration. The presumptive promoters of injury-induced Wnt signaling genes are likely directly regulated by basic leucine zipper transcription factors. Black tick marks indicate predicted cAMP response element (CRE) hits in putative promoter regions. CRE hits were identified using HOMER. CPM: average ATAC-seq counts per million calculated using a 10 bp bin size.

Puncture wounds induce ectopic head formation when pre-existing organizers are absent

Our transcriptomic analysis of head and foot regeneration suggested that canonical Wnt signaling was activated as part of a generic response to injury during early head and foot regeneration, and that the onset of head regeneration-specific Wnt activity did not occur until after the initial generic injury response had ended. We therefore hypothesized that experimentally preventing wound closure could lead to a sustained generic injury response, thus prolonging the generic phase of injury-induced Wnt signaling. Because ectopic Wnt signaling activity is sufficient to induce the formation of ectopic head structures in Hydra (Broun et al., 2005; Gee et al., 2010), we predicted that prolonged injuries could result in ectopic head formation.

To test this prediction, we first established a method for generating a constitutively active injury signal, which we accomplished by transversely impaling Hydra on fishing line at the midpoint of their oral-aboral axis (Figure 5A). This created two large puncture wounds that could not close while the fishing line remained in place. To test the sufficiency of prolonged injury to induce ectopic head organizers, we impaled animals and then either removed the fishing line immediately (0 hr timepoint) or left it in place for 12 hr. Wound outcomes were assessed 4 days post injury. We found that neither the 0 hr nor the 12 hr impalement injuries resulted in aberrant patterning phenotypes (Figure 5B, C), indicating that prolonged injuries are not sufficient to induce ectopic organizers in intact polyps.

Figure 5 with 1 supplement see all
Prolonged injuries induce ectopic head formation.

(A) Experimental design to test for the sufficiency of injury to induce new organizer formation in tissue with or without pre-existing organizers. (B) Prolonged injuries induce ectopic head regeneration when pre-existing organizers are removed. Scale bars indicate 1 mm. (C) Quantification of the number of ectopic tentacles induced by various impalement injury conditions. Both the presence of pre-existing organizers and the duration of the injury signal significantly contributed to the ectopic head regeneration phenotype. –Org indicates that organizers were removed just prior to impalement, and +Org indicates that organizers were left intact. (D) RT-qPCR results for head-specific transcripts in tissue that had been impaled for 12 hr. The Wnt signaling genes wnt3, wntless, and wnt9/10c and the head-specific transcription factor brachyury1 (bra1) are significantly upregulated after 12 hr of impalement. Three biological replicates, each consisting of tissue originating from 15 polyps, were generated for each treatment. Error bars indicate standard deviation. All statistical analyses were performed on 2-∆Cq values using Tukey’s HSD in R. Numbers in parenthesis indicate p-values. (E) Experimental design to test both the sufficiency of injury to induce ectopic heads during foot regeneration and the inhibitory capacity of oral-facing amputations. (F) Prolonged injuries induce ectopic heads at aboral-facing amputations, but only if prolonged oral-facing amputations are removed. Scale bars indicate 0.5 mm. The white arrowhead indicates a regenerated basal disk. (G) Quantification of the number of ectopic tentacles induced by various impalement injury conditions. +reamp indicates that the oral half of the regenerating tissue fragment was removed after 12 hr of impalement. HR: head regeneration; FR: foot regeneration, * indicates p-value≤0.05, ** indicates p-value≤0.01, *** indicates p-value≤0.001. All animals were imaged 4 days post injury, giving sufficient time for any removed structures to regenerate.

Extensive research has shown that Hydra organizer tissue generates long-range inhibitory signals that prevent ectopic organizers from forming elsewhere in the body, and that amputating organizers transiently increases the capacity of tissue throughout the body column to form secondary axes (Cohen and MacWilliams, 1975; MacWilliams, 1983b; MacWilliams and Kafatos, 1968; Wilby and Webster, 1970). This led us to hypothesize that inhibitory signals from pre-existing organizers were preventing ectopic head formation in impaled animals. We therefore modified our approach by amputating head and foot tissue just prior to impalement. In the absence of pre-existing organizers, both 0 and 12 hr impalement injuries resulted in the formation of ectopic oral structures—ranging from single tentacles to fully formed heads—at one or both of the presumptive impalement sites (Figure 5B, C). Importantly, the extent of this phenotype was dependent on the duration of the injury signal as animals impaled for 12 hr grew significantly more ectopic tentacles than animals that had been impaled and then immediately removed from the fishing line (Figure 5C). These findings demonstrate that puncture wounds can induce ectopic head organizer formation, and that pre-existing organizers prevent injury-induced ectopic head formation in intact polyps. In addition, the finding that prolonged injuries only resulted in ectopic oral structures and never ectopic aboral structures—despite the fact that Hydra body column tissue is equally capable of both oral and aboral regeneration—indicates that oral regeneration is the default outcome promoted by injury.

Prolonged puncture wounds induce the expression of Wnt pathway components

Because Wnt signaling is both necessary and sufficient for head organizer formation in cnidarians (Broun et al., 2005; Gee et al., 2010; Gufler et al., 2018), we interpret ectopic head formation following impalement as evidence that puncture wounds can induce Wnt signaling. Consistent with this conclusion, we found that treating animals with 5 µM iCRT14 following transverse impalement significantly reduced the number of ectopic tentacles at 4 days post injury, indicating that the phenotype was TCF dependent (Figure 5—figure supplement 1A). Furthermore, using RT-qPCR, we found that the Wnt signaling components wnt9/10c, wnt3, and wntless as well as the head-specific TF brachyury1 were significantly upregulated in tissue that had been transversely impaled for 12 hr (Figure 5D).

While performing transverse impalement experiments, we noted that ectopic head formation was strain-dependent as the Basel and 105 strains formed ectopic heads, but the AEP strain did not (Figure 5—figure supplement 1B–E). We nonetheless found that Wnt signaling components were significantly upregulated following 12 hr of impalement in all three strains (Figure 5D, Figure 5—figure supplement 1F, G). We therefore speculate that the differences in injury outcomes among these strains may be caused by genetic variation affecting inhibition levels in body column tissue.

Despite the dramatic effects that pre-existing organizers had on patterning outcomes following impalement, we found that the presence or absence of head and foot tissue did not significantly affect the expression of Wnt signaling components in impaled tissue after wound closure was blocked for 12 hr (Figure 5D). However, the ectopic heads induced by impalement expressed Wnt pathway-associated transcripts at significantly higher levels than impaled tissue that did not form ectopic heads (Figure 5—figure supplement 1H). Thus, the presence of pre-existing organizers inhibited the expression of injury-induced Wnt pathway genes, but it remains unclear when and how this inhibition occurred.

Regenerating heads can inhibit ectopic head formation at aboral-facing amputations

The role that pre-existing head and foot tissue played in preventing ectopic head formation following transverse impalement raised the question of why ectopic heads do not form at aboral-facing injuries when both the head and foot are amputated. We hypothesized that in such cases heads do not form at aboral-facing amputations because—although head patterning pathways are initially activated in both injuries—the oral-facing amputation recovers its ability to inhibit ectopic heads faster than the aboral-facing amputation, thus preventing ectopic head formation at the aboral pole. This hypothesis is consistent with the observations that regenerating heads gradually regain their inhibitory capacity over time and that head regeneration is faster when head amputations are located closer to the oral pole (MacWilliams, 1983a). We therefore sought to test if regenerating head tissue can prevent ectopic head formation at aboral-facing amputations.

To address this question, we first sought to increase the capacity for aboral-facing amputations to form ectopic heads by amputating both the head and aboral third of Hydra polyps and impaling the regenerating animals through the two resulting amputation injuries for 12 hr (Figure 5E). After impalement, we either let the animals recover unperturbed or we re-amputated the regenerating head promptly after the animals were removed from the fishing line. Thus, the former set of injury conditions prolonged both the oral and the aboral-facing amputations, while the latter set of conditions only prolonged the aboral-facing amputation. We found that when both amputations were prolonged there was no ectopic head formation (Figure 5F, G). However, when only the aboral-facing amputation was prolonged, head structures formed at both poles of the oral-aboral axis in ~61% of cases (13/23 animals). These findings demonstrate that prolonged aboral-facing amputation injuries can induce head formation, and that regenerating heads can inhibit ectopic head formation at aboral-facing amputations. Notably, foot regeneration occurred in all cases, such that one pole of the regenerated animal had a head while the other pole had both a foot and a full or partial head (Figure 5F). This therefore suggests that the specification of foot tissue during regeneration occurs independently from head specification.

Discussion

Using RNA-seq and ATAC-seq, we have comprehensively characterized transcription during the first 12 hr of head and foot regeneration in Hydra. We found that the initial response to amputation at 3 hpa did not depend on the structure being regenerated and included widespread apoptosis and the upregulation of canonical Wnt signaling components at both oral- and aboral-facing wounds. During this initial response, we also observed symmetric increases in chromatin accessibility near predicted TCF binding sites, suggesting that Wnt signaling is activated during early head and foot regeneration. By 8 hpa, however, Wnt signaling component expression and chromatin accessibility near TCF TFBMs became specific to head regeneration. Inhibiting TCF delayed the onset of head and foot regeneration-specific transcription transcriptome-wide, demonstrating a central role for the Wnt-responsive TF during the two types of regeneration. We found that Wnt pathway components were also upregulated in non-regenerative puncture wounds, supporting the conclusion that these genes are part of the generic transcriptional response to injury. Furthermore, removing pre-existing organizers induced puncture wounds to form ectopic head organizers, and experimentally blocking wound closure for 12 hr increased the penetrance of this phenotype. This demonstrates that injury signals are capable of initiating head patterning, even in the context of non-amputation injuries. We also found that prolonged foot amputation injuries could induce ectopic head formation, and that this phenotype could be suppressed by regenerating head tissue. To our knowledge, our work is the first demonstration in any animal that injuries alone are sufficient to induce secondary axes.

Based on our findings, we propose the following model of Wnt signaling activity during Hydra regeneration: injuries activate canonical Wnt signaling regardless of the injury type, which has the potential to give rise to a new head organizer through an autocatalytic feed-forward loop; however, in the case of foot regeneration or non-amputation injuries, inhibitory signals from pre-existing head tissue or oral-facing amputations block the Wnt feed-forward loop and thus prevent ectopic head formation (Figure 6). A version of this model was originally proposed over 30 years ago based on an extensive body of research that used classical tissue manipulation techniques (Gierer and Meinhardt, 1972; Kobatake and Sugiyama, 1989; MacWilliams, 1983a; Newman, 1974); however, this original model lacked an underlying molecular mechanism, relying instead on abstract theoretical actors to account for the injury-induced activation of head patterning—a phenomenon referred to as the ‘injury effect’ (Kobatake and Sugiyama, 1989; MacWilliams, 1983a). Our work suggests that the injury effect documented in previous research is the result of injury-induced Wnt signaling, thus providing an updated molecular framework that is highly consistent with the pre-existing theoretical model.

Figure 6 with 1 supplement see all
Proposed model of canonical Wnt signaling activity following mid-gastric bisection in Hydra.

Injury triggers a Wnt signaling cascade regardless of tissue context through the direct upregulation of the Wnt ligands wnt9/10c and wnt3 via injury-responsive basic leucine zipper transcription factors. An increase in Wnt signaling then represses injury-responsive transcription factors in a TCF-dependent manner. While Wnt signaling has the capacity for autocatalytic amplification, this only occurs during head regeneration as the autocatalytic amplification is blocked during foot regeneration by an uncharacterized inhibitor originating from pre-existing oral tissue. HR: head regeneration; FR: foot regeneration. Red coloration denotes high Wnt signaling activity. Blue coloration denotes low Wnt signaling activity.

An alternative model of oral patterning during Hydra regeneration proposed that canonical Wnt signaling is rapidly activated within 1.5 hpa exclusively at oral-facing amputation injuries as a result of head regeneration-specific apoptosis (Chera et al., 2009). However, we found no evidence of these early asymmetries. Instead, we found that the apoptotic and transcriptional response to amputation was indistinguishable between head and foot regeneration at 3 hpa. The recently reported finding that MAP kinases—highly conserved upstream regulators of wound repair and apoptosis—showed identical levels of activation during head and foot regeneration is also consistent with the existence of an initial generic response to injury (Tursch et al., 2020). Finally, our finding that TCF is required for the activation of foot-specific transcription during regeneration appears to be inconsistent with injury-induced Wnt signaling only occurring during head regeneration.

Our results highlight the existence of notable similarities in the transcriptional regulation of Wnt signaling components across distantly related regeneration models. In addition to Hydra, Wnt signaling components are upregulated as part of an early generic response to injury in the cnidarian species Nematostella vectensis and Clytia hemisphaerica as well as in planarians (DuBuc et al., 2014; Gurley et al., 2010; Petersen and Reddien, 2009; Schaffer et al., 2016; Sinigaglia et al., 2020; Stückemann et al., 2017; Wurtzel et al., 2015). In addition, recent research has demonstrated that MAP kinase (MAPK) signaling is required for the upregulation of Wnt pathway components during the generic wound response in both planarians and Hydra (Owlarn et al., 2017; Tursch et al., 2020). Although the TFs that regulate Wnt components during regeneration in planarians are unknown, our systematic analysis of injury-responsive Wnt component regulatory sequences in Hydra found that all of the most plausible regulators of Wnt pathway genes during Hydra regeneration are TFs that are directly regulated by MAPK signaling (Whitmarsh, 2007). This raises the possibility that MAPK-induced TFs are conserved regulators of Wnt pathway transcripts during regeneration. Both Nematostella and bilaterians diverged from Hydra over 500 million years ago (Erwin et al., 2011; Khalturin et al., 2019). Thus, these similarities span sizable evolutionary timescales and suggest the existence of deeply conserved aspects of the gene regulatory networks involved in the response to injury during regeneration.

There nonetheless exist important differences in the regulation of Wnt signaling during regeneration in bilaterians and Hydra. In planarians, Wnt signaling is restricted to tail regeneration through the head regeneration-specific upregulation of the Wnt inhibitor notum (Petersen and Reddien, 2011). In cnidarians, notum and canonical Wnt ligands are co-expressed in both uninjured and regenerating head tissue (Petersen et al., 2015; Siebert et al., 2019, Figure 6—figure supplement 1). Furthermore, in contrast to Hydra, pre-existing organizers do not appear to influence patterning during planarian regeneration (Petersen and Reddien, 2011). In acoels, Wnt pathway components are not part of the generic transcriptional response to injury and are instead rapidly upregulated specifically at aboral-facing amputations (Ramirez et al., 2020). In addition, the TF Egr is responsible for the injury-induced expression of wnt-3 in acoels, but Egr does not appear to be a plausible regulator of injury-responsive Wnt pathway genes in Hydra. Thus, there are likely significant differences in the regulation of Wnt signaling during regeneration in Hydra, planarians, and acoels.

There also appear to be differences in the downstream effects of Wnt pathway gene upregulation during whole-body regeneration in Hydra and bilaterians. In planarians, there are conflicting reports on whether canonical Wnt signaling is activated as part of the generic response to injury (Stückemann et al., 2017; Sureda-Gómez et al., 2016); however, knocking down Wnt signaling genes causes head regeneration to occur at both anterior and posterior-facing amputations (Gurley et al., 2008; Petersen and Reddien, 2008), demonstrating that the upregulation of Wnt transcripts is not required for the specification of head tissue. Similar knockdown phenotypes have also been reported in acoels (Srivastava et al., 2014). This is in contrast to Hydra, where TCF inhibition blocks regeneration of both the oral and aboral ends. Our findings that TCF TFBM accessibility increased during both early head and foot regeneration and that non-amputation injuries could induce ectopic heads strongly suggest that the transcriptional upregulation of Wnt signaling components during the generic wound response is accompanied by the activation of Wnt signaling in Hydra. Furthermore, the finding that inhibiting TCF blocks head and foot regeneration indicates that this activation likely plays an important role in both types of regeneration (Gufler et al., 2018). However, further characterization of the Wnt pathway in Hydra, in particular the identification of direct Wnt pathway targets and the characterization of β-catenin protein dynamics during the early wound response, will be required to fully understand the role of Wnt signaling during Hydra regeneration.

Materials and methods

Key resources table
Reagent type (species)
or resource
DesignationSource or referenceIdentifiersAdditional information
Strain, strain background
(Hydra vulgaris)
AEPMartin et al., 1997;
Wittlieb et al., 2006
Strain, strain background
(Hydra vulgaris)
105Chapman et al., 2010
Strain, strain background
(Hydra vulgaris)
BaselTechnau and Holstein, 1995
Commercial assay or kitTagment DNA Enzyme
and Buffer Small Kit
Illumina20034197
Commercial assay or kitKapa mRNA-seq Hyper kitRocheKK8581
Commercial assay or kitMinElute PCR Purification KitQIAGEN28004
Commercial assay or kitDNase SetQIAGEN79254
Commercial assay or kitRNA Clean and Concentrator kitZymogenR1017
Chemical compound, drugiCRT14Sigma-AldrichSML0203
Software, algorithmRRRRID:SCR_001905
OtherAgencourt AMPure XP beadsBeckman CoulterA63881
Other2X NEBNext master mixNEBM0541S
OtherM-MLV RNase H Minus
Point Mutant Reverse
Transcriptase
PromegaM3682
OtherSsoAdvanced universal
SYBR green master mix
Bio-Rad1725271

Code and data availability

Request a detailed protocol

All code used in this study is available both as a git repository at https://github.com/cejuliano/jcazet_regeneration_patterning (Cazet and Juliano, 2021; copy archived at swh:1:rev:6e9b9067f0adf1bd223c5ee0194d75847fa18321) and on Dryad at https://doi.org/10.25338/B8S612. FASTQ files of raw ATAC-seq and RNA-seq reads, expression matrices for ATAC-seq and RNA-seq reads mapped to the Hydra 2.0 genome reference, consensus peak files, and bigwig genome tracks of individual and pooled ATAC-seq replicates are available through the Gene Expression Omnibus under the accession GSE152994. The Hydra 2.0 genome gene model IDs associated with the gene names used throughout this study are provided in Table 3.

ATAC-seq library preparation

Request a detailed protocol

For each ATAC-seq replicate, ~15 whole, bud-free H. vulgaris (strain 105) polyps that had been fed once weekly were starved for 2 days and then transversely bisected at the midpoint of their oral-aboral axis. Regeneration was then allowed to proceed for 0, 3, 8, or 12 hr. Regenerating tips corresponding to ~1/3 of the total regenerate length were then isolated from head and foot regenerates and used for generating ATAC-seq libraries. For iCRT14-treated samples, 10 mM iCRT14 (SML0203; Sigma-Aldrich, St. Louis, MO) dissolved in dimethyl sulfoxide (DMSO) was diluted in Hydra medium to a final concentration of 5 µM. iCRT14 solution was added to the media 2 hr before amputation and was left in place until the tissue was collected for library preparation. For each treatment, 3–5 biological replicates were prepared (replicates are listed in Table 1). The number of replicates was chosen based on the widely used standard of three biological replicates per treatment for high-throughput sequencing experiments, although we slightly increased the number of replicates to increase the statistical power of differential tests.

To generate ATAC-seq libraries, we made use of a modified version of the OMNI-ATAC protocol as described previously (Corces et al., 2017; Siebert et al., 2019). Briefly, regenerating tips were washed once with 1 ml of chilled Hydra dissociation medium (DM) (Gierer et al., 1972) and then homogenized in 1 ml fresh DM using ~30 strokes of a tight-fitting dounce. Cells were then pelleted at 500 RCF for 5 min at 4°C in a benchtop centrifuge and subsequently lysed in 50 µl of chilled resuspension buffer (RSB; 10 mM Tris-HCl, pH 7.4, 10 mM NaCl, 3 mM MgCl2) with 0.1% Tween-20, 0.1% NP-40, and 0.01% digitonin for 3 min on ice. Lysis was halted by adding 1 ml of RSB with 0.1% Tween-20. The lysate’s nuclear concentration was then quantified by loading 19 µl of lysate and 1 µl of 0.2 mg/ml Hoechst 33342 onto a Fuchs-Rosenthal hemocytometer. A volume of lysate corresponding to ~50,000,000 nuclei was then aliquoted and spun at 500 RCF for 10 min at 4°C in a benchtop centrifuge. The resulting crude nuclear pellet was then resuspended in 50 µl of tagmentation solution (1× TD buffer [20034197; Illumina, San Diego, CA], 33% phosphate-buffered saline [PBS], 0.01% digitonin, 0.1% Tween-20, 5 µl TDE1 [Illumina 20034197]) and shaken at 1000 rpm for 30 min at 37°C. Tagmentation was halted by adding 250 µl of PB buffer from a QIAGEN MinElute PCR Purification Kit (28004; QIAGEN, Hilden, Germany).

Tagmented DNA was purified using a QIAGEN MinElute PCR Purification Kit following the standard manufacturer’s protocol. Libraries were then amplified with 2X NEBNext master mix (M0541S; NEB, Ipswitch, MA) using cycle numbers determined by quantitative reverse transcription PCR (qPCR) as described in the standard ATAC-seq protocol (Buenrostro et al., 2015; Buenrostro et al., 2013). Agencourt AMPure XP beads (A63881; Beckman Coulter, Pasadena, CA) were then used to purify libraries and restrict fragment sizes to between 100 and 700 base pairs. Libraries were then pooled and sequenced on an Illumina HiSeq4000 using 2 × 150 bp reads.

ATAC-seq data processing

Request a detailed protocol

Sequencing adapters, stretches of low-quality base calls, and unpaired reads were removed from the raw sequencing data using Trimmomatic (Bolger et al., 2014). Filtered reads were then mapped to the Hydra vulgaris 2.0 genome (arusha.nhgri.nih.gov/hydra/) using Bowtie2 (Langmead and Salzberg, 2012). Mitochondrial reads were removed by independently mapping filtered reads to the Hydra mitochondrial genome (Voigt et al., 2008) and removing mitochondrial reads from the genome-mapped data using Picard Tools (broadinstitute.github.io/picard). Ambiguously mapped reads (defined as having a mapping quality (MAPQ) value of ≤3) and discordantly mapped read pairs were removed using SAMtools (Li et al., 2009). PCR duplicates were labeled using Picard Tools and removed using SAMtools.

Peak calling was performed using a modified version of the ENCODE consortium’s ATAC-seq analysis pipeline (encodeproject.org/atac-seq) (Landt et al., 2012). First, unambiguously mapped deduplicated non-mitochondrial reads were centered over the transposase binding site by shifting + strand reads +4 bp and – strand reads −5 bp using deepTools2 (Ramírez et al., 2016). Peaks were then called using Macs2 (Zhang et al., 2008) with a permissive p-value cutoff of 0.1. We then generated consensus lists of biologically reproducible peaks using the irreproducible discovery rate (IDR) framework (Li et al., 2011) by identifying peaks that were reproducible (IDR score ≤0.1) across at least three pairwise comparisons of biological replicates in at least one treatment group.

To assess the quality and reproducibility of our ATAC-seq data, we made use of several metrics used by the ENCODE consortium. Because core promoters are expected to be highly accessible, ATAC-seq reads should be strongly enriched near transcription start sites (TSS). The current Hydra genome gene models lack UTR annotations, so we determined the TSS enrichment score for each biological replicate by measuring read enrichment near the first start codon of the 2000 most highly expressed genes in the Hydra single-cell RNA-seq atlas relative to the average read density ±1 kb from the start codon. We observed between a five- and eightfold TSS enrichment across the individual samples in our dataset. We also calculated the self-consistency and rescue ratios for each treatment group to evaluate consistency across replicates. We found that all treatment groups had self-consistency and rescue ratios less than 2, indicating good overall reproducibility across replicates. A full table of ATAC-seq library metrics is provided in Table 1.

Differential accessibility and chromVAR analysis

Request a detailed protocol

Read counts for peaks in our consensus peakset were calculated using the R Diffbind package (Ross-Innes et al., 2012). edgeR was then used for downstream differential accessibility analyses (Robinson et al., 2010). Peaks that did not have at least 10 counts per million in at least three replicates were excluded from further analyses. Differentially accessible peaks were then identified using quasi-likelihood tests of count data fitted to a negative binomial generalized log-linear model with a fFalse discovery rate (FDR) cutoff of 1e-4. The full results tables for all ATAC-seq pairwise comparisons performed as part of the edgeR analysis are included in Supplementary file 1.

For the chromVAR TFBM accessibility analyses, the width of all peaks that showed a significant change during regeneration was fixed at 250 bp and new read counts were generated. Changes in accessibility were then calculated for all TFBMs provided in the custom list of HOMER motifs found in the chromVARmotifs package. Because our unfiltered enrichment results often included redundant TFBMs with highly similar sequence composition and accessibility dynamics, TFBM redundancy was reduced by performing hierarchical clustering on a matrix of pairwise TFBM similarity scores generated by the HOMER compareMotifs function. The TFBM within a cluster that showed the greatest change in accessibility across treatments was then used as the representative motif for that cluster. To identify motifs enriched in injury-responsive peaks at 3 hpa, peaks that showed a significant increase in accessibility during either head or foot regeneration at 3 hpa were compared to all other peaks in our consensus peakset using the HOMER findMotifsGenome function.

RNA-seq library preparation

Request a detailed protocol

30 whole, bud-free H. vulgaris (strain 105) polyps that had been fed once weekly were starved for 2 days and then transversely bisected at the midpoint of their oral-aboral axis in batches of 10 to reduce variability in amputation and collection times. iCRT14 treatment conditions followed the same protocol as was described above for ATAC-seq library preparation. Regeneration was then allowed to proceed for 0, 3, 8, or 12 hr. Regenerating tips corresponding to ~1/3 of the total regenerate length were then isolated from head and foot regenerates and frozen in Trizol (15596018; Thermo Fisher Scientific, Waltham, MA) at −80°C. The three individual batches of 10 regenerating tips were then pooled into a single biological replicate, and RNA was subsequently extracted using a standard Trizol RNA purification protocol. For each treatment, three biological replicates of 30 animals each were prepared (replicates are listed in Table 2). The number of replicates was chosen based on the widely used standard of three biological replicates per treatment for high-throughput sequencing experiments. DNA contamination was then removed using the QIAGEN DNase Set (79254; QIAGEN) following the manufacturer’s protocol. A final cleanup was then performed using a Zymogen RNA Clean and Concentrator kit (R1017; Zymogen, Irvine, CA) following the standard manufacturer’s protocol. Strand-specific polyA-enriched libraries were prepared using the Kapa mRNA-seq Hyper kit (KK8581; Kapa Biosystems, Cape Town, South Africa). Untreated samples were sequenced on an Illumina HiSeq4000 with 1 × 100 bp reads. 12 hpa and all iCRT14-treated samples were sequenced on an Illumina NovaSeq using 2 × 150 bp reads.

RNA-seq data processing and differential gene expression analysis

Request a detailed protocol

Sequencing adapters and stretches of low-quality base calls were removed from the raw sequencing data using Trimmomatic. Filtered reads were then mapped to the Hydra 2.0 gene models, and read counts per gene were calculated using RSEM (Li and Dewey, 2011). A full table of RNA-seq library metrics is provided in Table 2. Using edgeR, read counts were normalized and genes that did not have at least two counts per million in at least three replicates were excluded from further analyses. Differentially expressed genes were then identified using quasi-likelihood tests of count data fitted to a negative binomial generalized log-linear model with a FDR cutoff of 1e-3. The full results tables for all RNA-seq pairwise comparisons performed as part of the edgeR analysis are included in Supplementary file 2.

To identify genes that were enriched in head or foot tissue in homeostatic Hydra, single-cell read counts for epithelial head and foot cells from a previously published Hydra single-cell RNA-seq atlas (Siebert et al., 2019) were compared using a Wilcoxon rank-sum test. Enrichment was determined using an adjusted p-value cutoff of 1e-6.

To systematically identify and characterize the expression of canonical Wnt signaling genes during regeneration, KEGG-annotated Wnt pathway components were isolated from the Nematostella vectensis, Exaiptasia pallida, Hydra vulgaris 1.0, and Homo sapiens genomes. We queried multiple species because many of the cnidarian references used to generate KEGG annotations are highly fragmented and likely incomplete. Putative orthologs of these KEGG references were then identified in the Hydra vulgaris 2.0 genome reference using reciprocal blast searches. BLAST hits were then manually evaluated to exclude false positives by examining protein domain composition as determined by InterPro (Blum et al., 2021) and performing additional BLAST searches against the NCBI NR reference. Wnt pathway components that had been previously characterized in Hydra and were not recovered using KEGG annotations (naked cuticle, wntless, and sp5) were added to the candidate list manually. Additionally, Wnt signaling genes that were not directly associated with the canonical pathway (e.g., non-canonical Wnt signaling) were excluded.

Vital staining of apoptotic cells using acridine orange

Request a detailed protocol

Staining of apoptotic cells using acridine orange was performed using a modified version of a previously described protocol (Cikala et al., 1999). Briefly, animals were collected in batches of 15 and incubated in 1 ml 1.6 µM acridine orange diluted in Hydra medium for 2 min in the dark. The Hydra were then quickly washed twice with 1 ml Hydra medium and immediately documented. Staining was performed on whole uninjured Hydra and on head or foot regenerating animals following mid-gastric bisection at 1 or 3 hpa.

Validation of iCRT14 treatment conditions

Request a detailed protocol

To validate the efficacy of iCRT14 in inhibiting head regeneration and foot regeneration,three biological replicates consisting of 30 polyps each were pre-treated for 2 hr in either 5 µM iCRT14 or 0.05% DMSO and then bisected at the midpoint of their oral aboral axis. The animals were then left to recover in either 5 µM iCRT14 or 0.05% DMSO that was regularly refreshed every 12 hr. Head regeneration was assessed by quantifying the number of regenerated tentacles at 60 hpa. Foot regeneration was assessed at 36 hpa by performing a foot peroxidase staining assay as previously described (Hoffmeister and Schaller, 1985). Briefly, Hydra were relaxed in 2% urethane in Hydra medium for 1 min and then fixed for 1 hr at room temperature in 4% paraformaldehyde in Hydra medium. Fixative was removed using three quick PBS washes. The samples were then incubated in 5% sucrose in PBS for 24 hr at 4°C. All subsequent steps were performed at room temperature. Excess sucrose was removed using three 5 min washes in PBT (0.1% Tween in PBS). Basal disks were then stained by incubating Hydra in a solution of 0.02% diaminobenzidine and 0.003% H2O2 in PBT for 15 min. Excess stain was removed using three 5 min washes in PBT, and the number of stained disks per biological replicate was quantified. Student’s t-test function in R was used to test for significant differences in the number of regenerated tentacles and the proportion of animals with regenerated basal disks.

Hydra tissue manipulation experiments

Request a detailed protocol

For transverse impalement experiments, H. vulgaris (either Basel, 105, or AEP strain) polyps that had been fed twice weekly were starved for at least 24 hr and then impaled at the midpoint of their oral-aboral axis in batches of ~15 on a single piece of 0.3 mm diameter fishing line. Each treatment had a total of ~30 samples. The sample size was chosen to maximize statistical power while minimizing temporal variability in injury duration. To remove pre-existing organizer tissue before impalement, heads and feet were removed by performing transverse amputations just below the tentacle ring and just above the peduncle, respectively. The Hydra were then either removed from the fishing line immediately after being impaled for the 0 hr timepoint or were left on the fishing line for 12 hr and then removed for the 12 hr timepoint. Phenotypes were then documented after 4 days of recovery post impalement. Occasionally, during the 12 hr impalement experiments, the epithelia would heal around the fishing line, thus prematurely terminating the injury signal. Animals from the 12 hr timepoint that underwent premature wound healing were excluded from subsequent analyses. Significant differences in the number of ectopic tentacles at 4 days post injury were identified using an ANOVA followed by Tukey’s HSD test in R using the agricolae package.

To test the role of TCF in ectopic head formation induced by transverse impalement, the heads and feet of Basel strain Hydra were first removed as described above. The animals were then transversely impaled for 12 hr and subsequently allowed to recover in either 0.05% DMSO or 5 µM iCRT14 in Hydra medium that was refreshed every 12 hr for a total of 4 days. Each treatment group consisted of ~30 animals. The number of ectopic tentacles was quantified at 4 days post injury, and the treatment groups were compared using the R Student’s t-test function.

For prolonged aboral amputation experiments, transverse cuts were used to remove the head and aboral-most third of Basel strain H. vulgaris polyps in batches of ~10–15 animals each. Animals were then impaled through the two resulting amputation wounds using 0.3 mm fishing line. After 12 hr of impalement, Hydra were either left to recover undisturbed or their oral-most half was amputated and discarded. To track the original orientation of the oral-aboral axis of impaled tissue fragments, one end of the fishing line was marked with a permanent marker and all Hydra were impaled such that their aboral pole pointed towards the marked end of the fishing line. The number of ectopic tentacles was quantified 4 days post injury, and significant differences in the number of ectopic tentacles were determined using Student’s t-test in R.

Quantitative reverse transcription PCR

Request a detailed protocol

Transversely impaled Basel, AEP, or 105 strain Hydra were prepared as described above, with 15 polyps per biological replicate and a total of three biological replicates per treatment. The number of replicates was chosen based on the widely used standard of three biological replicates per treatment for qPCR experiments. For samples collected immediately after impalement, tissue surrounding the impalement site, corresponding to ~1/4 of the total polyp length, was collected for downstream RNA isolation.

To quantify gene expression in ectopic heads, Basel strain animals lacking pre-existing organizers were impaled for 12 hr and then left to recover. As a control, body column tissue was collected from animals that were impaled for 12 hr with their organizers intact after 4 days of recovery. Both the ectopic heads and the control tissue were collected 4 days post injury.

RNA purification was performed using the same protocol as described above for RNA-seq library preparation. cDNA was synthesized using 1 µg of purified RNA and Promega M-MLV RNase H Minus Point Mutant Reverse Transcriptase (M3682; Promega, Madison, WI) using the manufacturer’s recommended protocol for oligo dT-primed synthesis. cDNA was then diluted 1:3 in nuclease-free water for use in qPCR experiments.

Three technical replicates of 10 μl qPCR reactions per sample were prepared using Bio-Rad SsoAdvanced universal SYBR green master mix (1725271; Bio-Rad, Hercules, CA) and were run on a CFX96 Touch Real-Time PCR Detection System (1855195; Bio-Rad). Cq values from technical replicates were pooled for subsequent analyses. rp49 was used as an internal control to calculate ∆Cq values after first being found to give similar results across all treatments when compared to a second housekeeping gene, actin. Statistically significant differences in mRNA expression were calculated by performing an ANOVA on 2-∆Cq values followed by Tukey’s HSD test in R using the agricolae package.

Data availability

All code used in this study is available both as a git repository at https://github.com/cejuliano/jcazet_regeneration_patterning (copy archived at https://archive.softwareheritage.org/swh:1:rev:6e9b9067f0adf1bd223c5ee0194d75847fa18321/) and on Dryad at https://doi.org/10.25338/B8S612. FASTQ files of raw ATAC-seq and RNA-seq reads, expression matrices for ATAC-seq and RNA-seq reads mapped to the Hydra 2.0 genome reference, consensus peak files, and bigwig genome tracks of individual and pooled ATAC-seq replicates are available through the Gene Expression Omnibus under the accession GSE152994. The Hydra 2.0 genome gene model IDs associated with the gene names used throughout this study are provided in Figure 1—figure supplement 8. Full differential gene expression results are available in Supplementary files 1 and 2. Motif enrichment and variability results are available in Supplementary file 3.

The following data sets were generated
    1. Cazet JF
    2. Juliano CJ
    (2020) NCBI Gene Expression Omnibus
    ID GSE152994. Oral Regeneration Is the Default Pathway Triggered by Injury in Hydra.
    1. Cazet JF
    2. Juliano CJ
    (2020) Dryad Digital Repository
    Oral Regeneration Is the Default Pathway Triggered by Injury in Hydra.
    https://doi.org/10.25338/B8S612
The following previously published data sets were used
    1. Siebert S
    2. Cazet JF
    3. Farrell JA
    (2018) NCBI Gene Expression Omnibus
    ID GSE121617. Stem cell differentiation trajectories in Hydra resolved at single cell resolution.
    1. Wenger Y
    2. Galliot B
    (2019) NCBI Gene Expression Omnibus
    ID GSE111534. Generic and context-dependent gene modulations during Hydra whole body regeneration.

References

    1. Gauchat D
    2. Kreger S
    3. Holstein T
    4. Galliot B
    (1998)
    Prdl-a, a gene marker for Hydra apical differentiation related to triploblastic Paired-Like Head-Specific genes
    Development 125:1637–1645.
    1. Hicklin J
    2. Wolpert L
    (1973)
    Positional information and pattern regulation in Hydra: formation of the foot end
    Journal of Embryology and Experimental Morphology 30:727–740.
    1. Kobatake E
    2. Sugiyama T
    (1989)
    Genetic analysis of developmental mechanisms in Hydra. XIX. Stimulation of regeneration by injury in the regeneration-deficient mutant strain, reg-16
    Development 105:521–528.
    1. Kuznetsov SG
    2. Anton-Erxleben F
    3. Bosch TC
    (2002)
    Epithelial interactions in Hydra: apoptosis in interspecies grafts is induced by detachment from the extracellular matrix
    The Journal of Experimental Biology 205:3809–3817.
    1. Müller WA
    (1996)
    Head formation at the basal end and mirror-image pattern duplication in Hydra vulgaris
    The International Journal of Developmental Biology 40:1119–1131.
    1. Newman SA
    (1974)
    The interaction of the organizing regions in Hydra and its possible relation to the role of the cut end in regeneration
    Journal of Embryology and Experimental Morphology 31:541–555.
    1. Technau U
    2. Bode HR
    (1999)
    HyBra1, a brachyury homologue, acts during head formation in Hydra
    Development 126:999–1010.
    1. Technau U
    2. Holstein TW
    (1995)
    Head formation in Hydra is different at apical and basal levels
    Development 121:1273–1282.
    1. Wilby OK
    2. Webster G
    (1970)
    Studies on the transmission of hypostome inhibition in Hydra
    Journal of Embryology and Experimental Morphology 24:583–593.

Decision letter

  1. Tatjana Piotrowski
    Reviewing Editor; Stowers Institute for Medical Research, United States
  2. Edward E Morrisey
    Senior Editor; University of Pennsylvania, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This clearly written manuscript from Cazet et al. sets out to define the molecular logic of positional specification in Hydra using RNAseq and ATACseq to profile transcriptome/chromatin accessibility changes that occur within the first 12 hours after injury. Although the initial response appears similar, the oral/aboral injuries begin to diverge by 8 hrs after injury, with several components of the Wnt pathway more highly upregulated in oral vs aboral regeneration. As a potential mechanism for this divergence, TCF binding sites appear more open in oral vs aboral regeneration. A TCF inhibitor previously shown to block regeneration prolongs high expression of transcription factors fos, jun, creb, and AP-1, and there are CRE response elements upstream of these genes. In the final, very clever experiment of the paper, the authors prolong wound signaling in the presence/absence of tissue organizers and find that increased wound signaling increases the prevalence of head regeneration at wound sites. Overall, this paper makes an important contribution to our understanding of regeneration by providing a clear molecular explanation for an older descriptive model and confirms a general theme in metazoan regeneration that initial signaling is general, but that it gets refined by signals from the rest of the tissue.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Oral Regeneration Is the Default Pathway Triggered by Injury in Hydra" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The three reviewers agreed that this manuscript reports interesting observations and is a nice gene expression resource with some associated hypotheses but lacks the level of experimental insight that would provide a major advance in the field. The descriptive genomics/transcriptomics data sets are valuable but should be experimentally tested. The reviewers concluded that without further experiments to test specific requirements or regulatory logic, the present work is primarily adding correlative gene expression evidence to a previously known phenotypic result which dampened their enthusiasm. In addition, the conceptual interpretation of the results clings unnecessarily closely to the Gierer/Meinhardt framework and important findings since the publication of this framework have not been taking into consideration.

Reviewer #1:

This very clearly written manuscript from Cazet et al. sets out to define the molecular logic of positional specification in Hydra using RNAseq and ATACseq to profile transcriptome/chromatin accessibility changes that occur within the first 12 hours after injury. Although the initial response appears similar, the oral/aboral injuries begin to diverge by 8 hrs after injury, with several components of the Wnt pathway more highly upregulated in oral vs aboral regeneration. As a potential mechanism for this divergence, TCF binding sites appear more open in oral vs aboral regeneration. A TCF inhibitor previously shown to block regeneration prolongs high expression of transcription factors fos, jun, creb, and AP-1, and there are CRE response elements upstream of these genes. In the final, very clever experiment of the paper, the authors prolong wound signaling in the presence/absence of tissue organizers, and find that increased wound signaling increases the prevalence of head regeneration at wound sites. Overall, this paper makes an important contribution to our understanding of regeneration by providing a clear molecular explanation for an older descriptive model, and confirms a general theme in metazoan regeneration that initial signaling is general, but that it gets refined by signals from the rest of the tissue.

1. The mechanisms are nicely described with bioinformatics, but functional experiments would really add significantly to the manuscript. For example, if the impaling experiment in Figure 5 is done in the presence of the TCF inhibitor, does it blunt the aberrant head regeneration? The increase in Wnt expression stimulated regardless of the presence of head/foot organizer tissue is an unsatisfying explanation, because the Wnts obviously increase dramatically in response to the injury. Perhaps this single timepoint is not the appropriate one to test; RT-qPCR timecourses may provide more resolution and identify some differences.

2. In the other Hydra strains that don't regenerate heads at the wound site, is there less Wnt/wntless/brachyury expression? Characterizing Wnt expression in these strains in response to injury would be both interesting, potentially useful, and may further strengthen the model.

3. The increased accessibility of TCF binding sites in head regeneration (Figure 2G) is beautiful, but would be more informative and compelling if placed in context. For example, Gehrke et al. included a comprehensive analysis of motifs in Hofstenia.

4. Use of the TCF inhibitor throughout assumes that the effect shown in the Gonsalves paper causes a similar effect in this lab, and is of similar penetrance. It is important to verify that the TCF inhibitor prevents both head and foot regeneration to provide confidence in the inhibition.

Reviewer #2:

Owing to its spectacular ability to regenerate entire polyps from small tissue fragments, Hydra is a historically important model for animal regeneration. When polyps are bisected along their oral-aboral axis, an initial wound response occurs and is followed by either oral or aboral regeneration. In "Oral Regeneration Is the Default Pathway Triggered by Injury in Hydra," Cazet and Juliano investigate the transcriptional mechanisms underlying this differential oral-aboral response to injury in Hydra. The authors utilize RNA-seq and ATAC-seq libraries to look at gene expression changes on a global scale, as well as drug inhibition studies to investigate the role of canonical Wnt signaling in the regenerative response. Their key finding is that Wnt signaling is activated as a component of the generalized injury response but only maintained in regenerating oral fragments, implying that aboral regeneration must entail a context specific repressive effect.

Overall, the genomic aspects of the work (RNA-seq and ATAC-seq) appear technically strong. My chief concerns, while not an expert in the field, surround the conceptual novelty and related to that, the degree of advance provided by the manuscript.

1. Conceptual novelty/degree of advance. The authors should make a stronger effort to differentiate this study from others, particularly Wegner et al. 2019 which really covers much of the same ground, defines the same oral-aboral transcriptional divergence time, and comes to some conceptually similar conclusions regarding the general injury response. It would also help general readers to point more explicitly toward the aspects of this work that are unique or different from what was previously known (eg Petersen et al. MBE 2015; Gufler et al. 2018 and the extensive literature on Wnt signaling in Hydra oral patterning/regeneration). As it reads now, the paper is well executed and may reflect a useful resource of -omics data for Hydra specialists, but the results are largely incremental and confirmatory of the long-known role of Wnt signaling in injury responses, regeneration, and body patterning throughout animals (eg Whyte et al. 2012).

2. Functional requirements for Wnt signaling. Gufler et al. (2018) utilized iCRT14 treatments to directly demonstrate a requirement for Wnt pathway activity in aboral regeneration. Here the authors use a similar paradigm to characterize the transcriptional changes that result from inhibiting Wnt signaling in oral and aboral fragments. Without further experiments to test specific requirements or regulatory logic, the present work is primarily adding correlative gene expression evidence to a previously known phenotypic result.

3. Wnt activation. To better understand how Wnt is regulated during regeneration, the authors looked for binding sites for known injury responsive TFs is cis regulatory regions of Wnt pathway components, and then correlated these sites with changes in chromatin accessibility during regeneration. A large number of such sites are identified and used to implicate bZIP TFs as "likely playing a role", but no validation is provided (either loss of function studies or even transgenic reporter constructs with and without the elements proposed to drive injury-responsive expression). Without this level of experimentation, the present work is a great resource of gene expression data and possible new hypothesis but lacks the experimental component that would represent a major advance in the field.

Reviewer #3:

This manuscript examines the interplay between wound polarity and gene expression during Hydra regeneration. Key points include i) the initial activation of Wnt pathway component expression irrespective of wound polarity, ii) the enrichment of bZIP binding sites in the promoters of multiple Wnt genes; iii) prolonged expression of early wound response transcription factors upon interference with Wnt signaling and iv) the induction of hypostome-like features in the Hydra foot upon experimental prevention of wound closure. Overall, the authors interpret their data as confirmation of a Gierer/Meinhardt model that was proposed 40 years ago.

Although the combination of RNAseq and ATACseq provides interesting glimpses of the underlying gene regulatory mechanisms, much of the data and analyses appear somewhat premature. This is especially the case for the latter parts of the manuscript that go beyond what is already known from published RNAseq studies (e.g., common wound response; upregulation of Wnt pathway components early during regeneration, timing of head organizer specification). Moreover, the conceptual interpretation of the results and the model the authors arrive at is overly narrow in parts. Altogether, this manuscript does report interesting observations, but would require extensive revisions to be suitable for publication.

1. The authors need to make their data more accessible.

Figures 1 and 3 require heat maps to illustrate i) the specific expression changes of all the IRTs and IRPs and ii) the identity of the underlying genes. This is essential for making the data broadly accessible to the community. Further, broadening the text perspective beyond Wnt & usual suspects would make the manuscript a lot more engaging, i.e., by naming other genes with interesting expression dynamics or enrichment analyses of Wnt pathway components versus all differentially expressed genes or the genes that become upregulated upon Wnt inhibition in 4.

2. The manuscript takes an overly simplistic view of Wnt signaling.

a. Expression of Wnt signaling components does not equate pathway activation- this assumption is made at multiple points in the text and needs to be corrected.

b. Changes in the accessibility of TCF binding sites does not equate Wnt signaling. The authors disregard that TCF can bind either bind βCatenin or transcriptional repressors (e.g., Groucho). Hence the "opening" of TCF binding sites alone does not signify the activation of Wnt signaling, as it could merely reflect the establishment of Wnt sensitivity or even Wnt-independent gene repression. The authors also cite the molecular mechanism of action of iCRT14 in this context, which is relevant to the interpretation of their results.

c. The authors appear to entirely disregard the many different branches of Wnt signaling that could conceivably result in very different signals in spite of common expression of a subset of transduction components.

Alltogether, these considerations make the "default activation of the head program" a very tentative premise at best. Without a read-out of canonical Wnt signaling activity and comparison of signaling activity between tail and foot regenerating wounds, it is equally possible that the wound type independent expression of Wnt only establishes the potential for future pathway activity, with the actual initiation of signaling requiring a context-dependent trigger.

3. The conceptual interpretation of the results clings unnecessarily closely to the Gierer/Meinhardt framework and remains insufficiently tested.

a. Important findings since Gierer/Meinhardt are not cited or introduced, including for example the suggestion of wound-type specific responses already within the first hour after injury (Chera et al., 2009) or the recent report of dynamic, actin-mediated intrinsic tissue polarization (Livshits et al., 2017). These papers and their key findings need to be cited in the introduction and they need to be discussed in the context of an evaluation of the author's model in the discussion. Moreover, the authors missed that a wound-context independent activation of Wnt pathway transcription also occurs in planarians. This paper needs to be cited (Stuckemann et al., 2017) and the interpretation of wound type independent Wnt signaling in this study might also be of interest in contemplating the Hydra data.

b. Necessary tests of the present model include examination of the assumption that the inhibition of Wnt signaling is really sufficient for the conversion of an established head organizer into a foot organizer. This could be easily done by treating regenerating Hydra with iCRT14 post 8 h or when the putative Wnt signaling requirement in the initial wound response has passed. Further interesting validations/tests could include the examination of position-dependence of impalement-induced head formation along the body axis (does the efficiency increase with the distance to existing or former "inhibitor sources"? Are any injuries capable of inducing feet?) or qPCR comparisons of Wnt induction efficiency between the two Hydra strains.

c. The model disregards the organizing influence of the Hydra foot. As is, one might expect that pieces resulting from bisection and additional foot amputation might regenerate double heads (due to absence of the "source" of the long-range inhibitor). This is not observed. Also, the close spacing of the injury-induced double heads in Figure 5 is a problem for the Gierer/Meinhardt framework, as the inhibitor emanating from each head would be expected to interfere with the formation/maintenance of the other head organizer due to their close spacing.

4. Validation of the importance of bZIP binding motifs in Wnt promoter regions.

As is, this section reads very much like a "just so" story. First, the authors have to analyze the probability of bZIP binding motif occurrence per kb in the hydra genome. Second, they should examine the occurrence of the motif in non wound-sensitive promoters. Third, they have to demonstrate significant enrichment of the motif in wound-induced genes. The strong emphasis on the importance of the bZip binding motifs in the present manuscript further necessitates experimental validation- either in the manuscript itself or by citing published data in a peer-reviewed journal.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Generic Injuries Are Sufficient to Induce Ectopic Wnt Organizers in Hydra" for consideration by eLife. Your article has been reviewed by 1 peer reviewer, and the evaluation has been overseen by a Reviewing Editor and Edward Morrisey as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

Summary:

This clearly written manuscript from Cazet et al. sets out to define the molecular logic of positional specification in Hydra using RNAseq and ATACseq to profile transcriptome/chromatin accessibility changes that occur within the first 12 hours after injury. Although the initial response appears similar, the oral/aboral injuries begin to diverge by 8 hrs after injury, with several components of the Wnt pathway more highly upregulated in oral vs aboral regeneration. As a potential mechanism for this divergence, TCF binding sites appear more open in oral vs aboral regeneration. A TCF inhibitor previously shown to block regeneration prolongs high expression of transcription factors fos, jun, creb, and AP-1, and there are CRE response elements upstream of these genes. In the final, very clever experiment of the paper, the authors prolong wound signaling in the presence/absence of tissue organizers and find that increased wound signaling increases the prevalence of head regeneration at wound sites. Overall, this paper makes an important contribution to our understanding of regeneration by providing a clear molecular explanation for an older descriptive model and confirms a general theme in metazoan regeneration that initial signaling is general, but that it gets refined by signals from the rest of the tissue.

The authors addressed many of the previous critiques. The revised manuscript is substantially rewritten and several additional experiments and figures were added.

Revisions:

1. The authors now added acridine orange stainings to demonstrate that apoptosis is induced in head and foot regenerates, which is in contrast to what Chera et al. published in 2009. However, the acridine orange stainings have quite a bit of background and variability (and they are not quantified). As the Chera et al., data also looks convincing, the authors are encouraged to repeat a TUNEL staining that Chera et al., used. This would allow a better comparison between the differing results.

2. The literature describing the generic injury response that gradually polarizes that has been extensively described in planarians should be better cited. For example, Gurley et al., Dev Biol 2010 should be cited.

3. Please arrange panels in Figure 3 according to how they are cited to facilitate easy reading of the manuscript.

https://doi.org/10.7554/eLife.60562.sa1

Author response

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Reviewer #1:

This very clearly written manuscript from Cazet et al. sets out to define the molecular logic of positional specification in Hydra using RNAseq and ATACseq to profile transcriptome/chromatin accessibility changes that occur within the first 12 hours after injury. Although the initial response appears similar, the oral/aboral injuries begin to diverge by 8 hrs after injury, with several components of the Wnt pathway more highly upregulated in oral vs aboral regeneration. As a potential mechanism for this divergence, TCF binding sites appear more open in oral vs aboral regeneration. A TCF inhibitor previously shown to block regeneration prolongs high expression of transcription factors fos, jun, creb, and AP-1, and there are CRE response elements upstream of these genes. In the final, very clever experiment of the paper, the authors prolong wound signaling in the presence/absence of tissue organizers, and find that increased wound signaling increases the prevalence of head regeneration at wound sites. Overall, this paper makes an important contribution to our understanding of regeneration by providing a clear molecular explanation for an older descriptive model, and confirms a general theme in metazoan regeneration that initial signaling is general, but that it gets refined by signals from the rest of the tissue.

We thank the reviewer for recognizing the important contribution of our work. One misunderstanding we would like to point out: the CRE elements are found in the regulatory regions of the Wnt genes, not in the regulatory regions of the bZIP TFs (see Figure 4). The manuscript has been substantially re-written, so we hope that this point is now clearer.

1. The mechanisms are nicely described with bioinformatics, but functional experiments would really add significantly to the manuscript. For example, if the impaling experiment in Figure 5 is done in the presence of the TCF inhibitor, does it blunt the aberrant head regeneration?

We agree that the inclusion of functional experiments can greatly enhance the value of a descriptive dataset. As such, we’d like to highlight the functional data that we included in our original submission, including the iCRT14-treated RNA-seq and ATAC-seq datasets presented in figure 3 and the tissue manipulation experiments presented in figure 5. These data allow us to functionally validate our use of chromatin accessibility data to quantify TCF transcriptional activity (figure 3A), to identify novel regulatory functions for TCF during regeneration (Figure 3B-H), and to demonstrate the sufficiency of injury to induce ectopic Wnt organizers (Figure 5).

We also agree that the addition of an experiment to determine if injury-induced ectopic head formation is TCF-dependent would improve the manuscript. Therefore we performed the experiment suggested by the reviewer and present the results in Figure 5—figure supplement 1A. We found that iCRT14 significantly reduced the number of ectopic tentacles formed after prolonged transverse impalement, thus demonstrating that the phenotype is TCF-dependent.

The increase in Wnt expression stimulated regardless of the presence of head/foot organizer tissue is an unsatisfying explanation, because the Wnts obviously increase dramatically in response to the injury. Perhaps this single timepoint is not the appropriate one to test; RT-qPCR timecourses may provide more resolution and identify some differences.

It is a valid and useful point that the qPCR data we present from impaled tissue does not shed light on why the presence or absence of pre-existing organizers influences the outcome of prolonged impalement experiments. However, in our original manuscript, it was not our intention to use these data to address that particular question. Rather, we were aiming to demonstrate that despite the different outcomes, the initial response was nonetheless the same. This was in service of our broader argument that injury-induced expression of Wnt pathway components is not sensitive to signals from the surrounding tissue context during the generic phase of the injury response (even if that generic phase is prolonged to last for 12 hours).

In interpreting the results of the impalement experiments, we worked under the assumption that the observation that one treatment condition resulted in ectopic head organizers and the other did not was sufficient evidence that the two conditions resulted in differences in Wnt signaling activity. Essentially, our position is that the extensive research on the role of Wnt signaling in head specification in Hydra​ ​ renders the alternative hypothesis (that head organizer formation did not result from differential Wnt signaling activity) implausible. We believe that the fact that aspects of our rationale were not made fully explicit in the original text contributed to the lack of clarity on this point. We hope that the extensive changes we made to our discussion of the impalement experiments help to make understanding our rationale more straightforward.

Although we still maintain that head formation can be taken as a reliable proxy for Wnt signaling activity and Wnt pathway gene expression, we nonetheless recognize that investigating later time points using qPCR could serve as a valuable validation of our underlying assumptions. We therefore characterized Wnt pathway gene expression four days after impalement in animals that did form ectopic heads and those that did not (Figure 5—figure supplement 1H). We found that ectopic heads expressed Wnt pathway genes at significantly higher levels than tissue lacking ectopic heads. This therefore demonstrates that pre-existing organizers downregulated Wnt pathway genes at some point following impalement, although it is unclear when or how.

2. In the other Hydra strains that don't regenerate heads at the wound site, is there less Wnt/wntless/brachyury expression? Characterizing Wnt expression in these strains in response to injury would be both interesting, potentially useful, and may further strengthen the model.

We agree that this is an important question to address. In our revised manuscript, we have included qPCR data showing that 12 hours of impalement resulted in the significant upregulation of Wnt pathway genes in both 105 and AEP strain animals (Figure 5—figure supplement 1F,G). Thus, while ectopic head formation is strain specific (we observe it in Basel and 105 animals but not AEP animals) the injury induced expression of Wnt pathway components following puncture wounds is not.

3. The increased accessibility of TCF binding sites in head regeneration (Figure 2G) is beautiful, but would be more informative and compelling if placed in context. For example, Gehrke et al. included a comprehensive analysis of motifs in Hofstenia.

We think that presenting a systematic analysis of binding site accessibility would indeed provide a useful context for our discussion of the accessibility dynamics of TCF binding sites specifically. We wish to point out that we did present such a systematic analysis in Figure 4A of our original manuscript. However, because we presented these results quite a bit after our discussion of TCF binding site accessibility, it may not have effectively provided the context requested by the reviewer. In our revised manuscript, we now present our systematic analysis of motif accessibility in Figure 1I and explicitly discuss TCF in the context of the broader trends in transcription factor binding motif accessibility during regeneration.

4. Use of the TCF inhibitor throughout assumes that the effect shown in the Gonsalves paper causes a similar effect in this lab, and is of similar penetrance. It is important to verify that the TCF inhibitor prevents both head and foot regeneration to provide confidence in the inhibition.

We recognize the importance of demonstrating the reproducibility previous findings in the literature and we appreciate the suggestion that we include such data in our manuscript (Note: We assume the reviewer intended to refer to the paper by Gufler et al. which first reported the block of Hydra​ ​ head and foot regeneration, not the paper by Gonsalves et al. which initially developed and biochemically characterized iCRT14 in cultured cells). Before using iCRT14 in our study, we did do these validations, but we did not present them in our original manuscript. We have rectified this oversight and added these experiments to our revised manuscript demonstrating that iCRT14 significantly inhibited both head and foot regeneration in our hands (Figure 3—figure supplement 1A,B), thus recapitulating previously reported findings.

Reviewer #2:

Owing to its spectacular ability to regenerate entire polyps from small tissue fragments, Hydra is a historically important model for animal regeneration. When polyps are bisected along their oral-aboral axis, an initial wound response occurs and is followed by either oral or aboral regeneration. In "Oral Regeneration Is the Default Pathway Triggered by Injury in Hydra," Cazet and Juliano investigate the transcriptional mechanisms underlying this differential oral-aboral response to injury in Hydra. The authors utilize RNA-seq and ATAC-seq libraries to look at gene expression changes on a global scale, as well as drug inhibition studies to investigate the role of canonical Wnt signaling in the regenerative response. Their key finding is that Wnt signaling is activated as a component of the generalized injury response but only maintained in regenerating oral fragments, implying that aboral regeneration must entail a context specific repressive effect.

Overall, the genomic aspects of the work (RNA-seq and ATAC-seq) appear technically strong. My chief concerns, while not an expert in the field, surround the conceptual novelty and related to that, the degree of advance provided by the manuscript.

1. Conceptual novelty/degree of advance. The authors should make a stronger effort to differentiate this study from others, particularly Wegner et al. 2019 which really covers much of the same ground, defines the same oral-aboral transcriptional divergence time, and comes to some conceptually similar conclusions regarding the general injury response. It would also help general readers to point more explicitly toward the aspects of this work that are unique or different from what was previously known (eg Petersen et al. MBE 2015; Gufler et al. 2018 and the extensive literature on Wnt signaling in Hydra oral patterning/regeneration). As it reads now, the paper is well executed and may reflect a useful resource of -omics data for Hydra specialists, but the results are largely incremental and confirmatory of the long-known role of Wnt signaling in injury responses, regeneration, and body patterning throughout animals (eg Whyte et al. 2012).

We recognize that the way our manuscript was written may not have provided readers with sufficient context to understand our work’s contribution to the regeneration field. In our revised manuscript, we have entirely changed the way in which we frame our data. The introduction in the revised text now states more clearly the current gaps in knowledge in the Hydra regeneration literature, and the Results section now includes more specific explanations of the novelty of our findings. We hope that these changes will provide the reader with a better understanding of this manuscript’s contribution to regeneration biology.

A critical point that we did not sufficiently stress in our original manuscript is that a generic role for Wnt signaling during an initial context-independent injury response in Hydra is not long known or well established. Indeed, the best known molecular model of Hydra regeneration claims that Wnt signaling is exclusively activated during head regeneration within 1 hour post amputation (Chera et al., 2009). In addition, recent work by Ramirez et al. (2020) reports that Wnt signaling is rapidly activated during tail, but not head regeneration in aceols. Our work is the first to document the upregulation of Wnt ligands and other Wnt signaling components in non-regenerative injuries and during foot regeneration in Hydra. We are also the first to characterize TCF activity during Hydra regeneration. These findings significantly impact not only our understanding of cnidarian regeneration, but also our understanding of the evolution of the metazoan wound response. In our revised manuscript, we now more clearly explain the importance and novelty of our findings on Wnt signaling during regeneration. In particular, we now extensively discuss both the Chera et al. and Ramirez et al. publications in our introduction and discussion. Furthermore, we provide new data that directly refutes the claim made in the Chera et al. paper that injury apoptosis occurs in oral-facing but not aboral-facing amputation injuries (Figure 1J-M, Figure 1—figure supplement 5).

The recently released pre-print from Wenger et al. (2019) does contain an RNA-seq dataset that overlaps with our own. Based on this RNA-seq data, the authors report the existence of an initial context independent transcriptional response to injury and estimate that transcription during head and foot regeneration diverge between 4 and 8hpa. These findings are highly consistent with the conclusions we draw in figure 1(C-H) of our manuscript. However, we wish to emphasize that our study goes far beyond these two observations and makes numerous additional contributions that are wholly distinct from the work of Wenger et al.

1. Wenger et al. do not report the initial context-independent upregulation of Wnt signaling components during the context-independent injury response in their paper. However, we did find that this phenomenon is supported by their data after performing our own analysis.

2. They do not characterize chromatin remodeling and are thus far more limited in their ability to shed light on transcriptional regulation during head and foot regeneration.

3. They do not explore potential mechanisms that drive the transcriptional divergence of head and foot regeneration, nor do they explore potential mechanisms driving injury-responsive Wnt expression.

4. Their study does not include the functional experiments included in our study (figures 3 and 5)

The Gufler et al. (2018) paper was the first to report that TCF is required for both head and foot regeneration in Hydra using the TCF/β-catenin inhibitor iCRT14. In addition, they show that TCF inhibition prevents the downregulation of 8 head regeneration-specific genes during foot regeneration. Based on these findings, they proposed that Wnt signaling may play a role during foot regeneration. In our work, we extend these initial findings by comprehensively characterizing chromatin accessibility and transcript abundance during head and foot regeneration in the presence of iCRT14. With these data we find that TCF is required for the divergence in expression of all transcripts during head and foot regeneration, not just the 8 head-specific genes reported in Gufler et al. In addition, we find that TCF is required for the proper downregulation of the injury response and for the upregulation of head and foot specific transcripts. These inhibitor-dependent changes in transcription demonstrate a causal relationship, and allow us to better understand TCF function during regeneration. Importantly, these findings are wholly novel and are not foregone conclusions that can be deduced from the work of Gufler et al. Finally, by using our chromatin accessibility data to characterize TCF transcriptional activation during regeneration, we are able to get a readout of Wnt signaling activity that is comparable to popular TCF-based reporter assays (e.g. TOPFLASH). Such a readout, in conjunction with our transcriptomic data, gives us significant insight into Wnt signaling activity during regeneration that was not possible given the limited molecular data presented in Gufler et al. In our revised manuscript, we now more clearly delineate which of the results from our iCRT14-treated RNA-seq and ATAC-seq data are recapitulations of previous findings (Figure 3 – —figure supplement 1) and which are novel to this study (Figure 3, Figure 3—figure supplement 2).

The Petersen et al. (2015) paper included a comprehensive transcriptomic characterization of head regeneration in Hydra. Because this study exclusively focused on head regeneration, it does not shed light on how head and foot regeneration diverge, which is the primary concern of our paper. In addition, all the points outlined above for Wenger et al. also apply.

While there are numerous other studies covering Wnt signaling’s role in Hydra patterning, they do not characterize the pathway outside the context of head specification during regeneration and homeostasis. Therefore, our work looking at Wnt signaling during foot regeneration and in response to non-amputation injuries is novel.

2. Functional requirements for Wnt signaling. Gufler et al. (2018) utilized iCRT14 treatments to directly demonstrate a requirement for Wnt pathway activity in aboral regeneration. Here the authors use a similar paradigm to characterize the transcriptional changes that result from inhibiting Wnt signaling in oral and aboral fragments. Without further experiments to test specific requirements or regulatory logic, the present work is primarily adding correlative gene expression evidence to a previously known phenotypic result.

Please refer to our response to the first comment for the ways in which our study is distinct from the work presented by Gufler et al. Briefly, these are the novel functional results arising from our iCRT14 experiments:

1. TCF is required for the downregulation of only a small subset of head specific factors during foot regeneration, which includes the ligands wnt3 and wnt9/10c.

2. TCF is required to downregulate aspects of the generic wound response

3. TCF is required to upregulate foot-specific transcripts during foot regeneration

4. TCF is required for the upregulation of a subset of head specific transcripts

5. TCF is required for all context-specific expression during head and foot regeneration

In addition, we would like to note that the functional data presented in figure 5 demonstrate that non-amputation injuries are sufficient to activate Wnt signaling and induce the formation of secondary axes. These findings are entirely unique to our work.

3. Wnt activation. To better understand how Wnt is regulated during regeneration, the authors looked for binding sites for known injury responsive TFs is cis regulatory regions of Wnt pathway components, and then correlated these sites with changes in chromatin accessibility during regeneration. A large number of such sites are identified and used to implicate bZIP TFs as "likely playing a role", but no validation is provided (either loss of function studies or even transgenic reporter constructs with and without the elements proposed to drive injury-responsive expression). Without this level of experimentation, the present work is a great resource of gene expression data and possible new hypothesis but lacks the experimental component that would represent a major advance in the field.

We wholeheartedly agree that loss of function and reporter experiments would be ideal approaches for directly testing our hypothesis that Wnt signaling components are directly regulated by injury-responsive bZIP transcription factors. We have either attempted, or are currently attempting both types of experiments; however, we face significant technical hurdles presented by our model system. The Hydra research community is small, and lacks many of the tools that are readily available in more widely used model systems. Our lab is working on improving those tools, but we have only limited resources to address these significant challenges.

Currently, the best established protocol for gene knockdown in Hydra is to create a transgenic line that constitutively expresses a RNAi hairpin directed against the gene of interest. Although we have attempted to establish such knockdown lines for multiple injury-responsive bZIP transcription factors, we have failed to generate any KD positive transgenic lines. This can occur when a gene of interest is essential. Therefore perturbing the function of these transcription factors will likely first require the development of an inducible knockdown system in Hydra.

The only available method for transgenesis in Hydra (random insertion of injected circular DNA) makes performing reporter experiments difficult. We are currently unable to control where the injected plasmid breaks or where it is inserted in the genome. Reporter experiments therefore require a large number of injections for each construct being tested. We are currently attempting to generate lines to test the function of putative bZIP binding sites in the Wnt9/10c promoter, but this work is slow and has been hindered by the pandemic. Our plan is to present such data in a follow-up study.

Importantly, while the above experiments are necessary to test the hypothesis that bZIPs directly regulate Wnt components, they are not necessary for the broader conclusion of the paper, which is that injuries induce Wnt signaling outside the context of head amputation. Functional data supporting this conclusion is presented in figure 5.

Reviewer #3:

This manuscript examines the interplay between wound polarity and gene expression during Hydra regeneration. Key points include i) the initial activation of Wnt pathway component expression irrespective of wound polarity, ii) the enrichment of bZIP binding sites in the promoters of multiple Wnt genes; iii) prolonged expression of early wound response transcription factors upon interference with Wnt signaling and iv) the induction of hypostome-like features in the Hydra foot upon experimental prevention of wound closure. Overall, the authors interpret their data as confirmation of a Gierer/Meinhardt model that was proposed 40 years ago.

Although the combination of RNAseq and ATACseq provides interesting glimpses of the underlying gene regulatory mechanisms, much of the data and analyses appear somewhat premature. This is especially the case for the latter parts of the manuscript that go beyond what is already known from published RNAseq studies (e.g., common wound response; upregulation of Wnt pathway components early during regeneration, timing of head organizer specification). Moreover, the conceptual interpretation of the results and the model the authors arrive at is overly narrow in parts. Altogether, this manuscript does report interesting observations, but would require extensive revisions to be suitable for publication.

We first wish to clarify a few apparent misunderstandings. Regarding point iv, we did not show the induction of hypostome-like features in the foot following prolonged injuries. Rather, we demonstrated that prolonged injuries could induce ectopic heads in body column tissue. We would also like to point out that, with the exception of β-catenin, our paper is the first to describe the upregulation of Wnt pathway genes during aboral regeneration and in response to non-amputation injuries. While the RNA-seq dataset presented in Wenger et al. does indeed show the upregulation of Wnt pathway genes during foot regeneration, we were only able to make that conclusion by performing our own analysis on their data, as they do not report this finding anywhere in their manuscript. We would also point out that the Wenger et al. data set is from a pre-print manuscript and has not undergone peer review.

We also wish to clarify that we are not claiming to confirm the original Gierer-Meinhardt model as it was proposed in the initial 1972 publication. Rather, our data provides a molecular basis for the observation made after the Gierer-Meinhardt model was proposed that injuring Hydra tissue promotes the formation of a new head organizer, which we refer to as the “oral induction model.” We do invoke a Gierer-Meinhardt type of reaction-diffusion system when presenting our model of patterning during regeneration, but our model incorporates numerous observations made after the original Gierer-Meinhardt publication.

Upon reflection, we believe that our decision to focus on the “oral induction model” as a way to frame our paper led to several issues in our manuscript that were highlighted by the reviewer’s comments. In particular, we did not adequately address the ambiguities and gaps in knowledge currently present in the Hydra regeneration literature on Wnt activity and Wnt pathway gene expression during head and foot regeneration. We also did not properly convey that resolving the question of Wnt pathway regulation during Hydra regeneration has important implications regarding the evolution of regeneration in metazoans, especially given the findings on Wnt regulation in acoels and planarians. We have made a significant effort to rework the text to address these issues. We now extensively discuss both the planarian and acoel regeneration literature in our introduction and discussion. We also directly address the model presented by Chera et al. (2000), both through modifications to the text and through the inclusion of additional experiments (Figure 1J-M, Figure 1—figure supplement 5). Our Results section has also been reworked to more clearly explain the novelty of our findings in the context of the pre-existing literature. In addition, please refer to our response to comment 1 from reviewer 2 for our in-depth explanation of the ways in which our study can be differentiated from other related works.

1. The authors need to make their data more accessible.

Figures 1 and 3 require heat maps to illustrate i) the specific expression changes of all the IRTs and IRPs and ii) the identity of the underlying genes. This is essential for making the data broadly accessible to the community.

We agree that transparency is crucial when presenting scientific findings. We would like to highlight that we did make a meaningful effort to make our data accessible to readers. In particular, we included full results tables for the differential analyses we performed in Supplementary files 1 and 2. In addition, we made all code used in this study publicly available upon submission both on GitHub and Dryad. We also deposited both raw and processed data, including fastq files, peak files, read count matrices, and bigwig files on GSE and made these data publicly available upon submission.

We did inadvertently omit the results tables from the single cell structural enrichment analysis that were used to generate the plots in Figure 2A-D. We have now added these tables to Supplementary files 1 and 2. We have also added the tables that were directly used to generate all scatter plots and heatmaps throughout the manuscript as ‘source data’ files in our revised submission. We have also added Supplementary file 3, which includes results from our HOMER motif enrichment analysis and our chromVAR chromatin accessibility analysis.

We do not believe that the inclusion of heatmaps would be an effective way to make our data more accessible, as the number of peaks/transcripts being considered (hundreds to thousands in many cases) would either require that the heatmaps be excessively large or would render the gene labels too small to be legible.

Further, broadening the text perspective beyond Wnt & usual suspects would make the manuscript a lot more engaging, i.e., by naming other genes with interesting expression dynamics or enrichment analyses of Wnt pathway components versus all differentially expressed genes or the genes that become upregulated upon Wnt inhibition in 4.

In writing this paper, we chose to primarily focus our attention on genes that had either been previously characterized in Hydra or were directly relevant to the paper’s main focus—the regulation of Wnt signaling during regeneration. Our intention was that readers with a broader curiosity in our data could easily identify other genes of interest using the supplementary tables we provided. This choice was based on the fact that we did not intend to present our work as a resource paper, but rather we sought to focus on a specific set of questions.

However, we would note that figures 1 and 3 are not exclusively focused on Wnt signaling. In figure 1 of our original manuscript, we highlighted the foot-specific transcription factors (distall-less and nk-2). In figure 3 we highlighted multiple genes that aren’t Wnt pathway components including prdl-a, gremlin-like, budhead, creb, fos, and jun. Additionally, we highlighted TCF-dependent changes in the accessibility of the injury responsive TRE and Erg transcription factor binding motifs. In our revised manuscript we additionally highlight two novel foot specific TFs that are among the first to be upregulated in a foot regeneration-specific fashion (gata-3-like and foxd2-like; lines 219-221). We also note that these foot TFs exhibit TCF-dependent expression during regeneration (Figure 3E,F).

2. The manuscript takes an overly simplistic view of Wnt signaling.

a. Expression of Wnt signaling components does not equate pathway activation- this assumption is made at multiple points in the text and needs to be corrected.

b. Changes in the accessibility of TCF binding sites does not equate Wnt signaling. The authors disregard that TCF can bind either bind βCatenin or transcriptional repressors (e.g., Groucho). Hence the "opening" of TCF binding sites alone does not signify the activation of Wnt signaling, as it could merely reflect the establishment of Wnt sensitivity or even Wnt-independent gene repression. The authors also cite the molecular mechanism of action of iCRT14 in this context, which is relevant to the interpretation of their results.

c. The authors appear to entirely disregard the many different branches of Wnt signaling that could conceivably result in very different signals in spite of common expression of a subset of transduction components.

Alltogether, these considerations make the "default activation of the head program" a very tentative premise at best. Without a read-out of canonical Wnt signaling activity and comparison of signaling activity between tail and foot regenerating wounds, it is equally possible that the wound type independent expression of Wnt only establishes the potential for future pathway activity, with the actual initiation of signaling requiring a context-dependent trigger.

We agree that it is important to acknowledge that expression of wnt signaling components does not equate to an increase in Wnt signaling activity, which is why we did so in our original manuscript: “Although canonical Wnt signaling components were upregulated during both head and foot regeneration, it was not clear if this resulted in the downstream activation of Wnt responsive transcription.”

We would also like to note that the language at line 163 of our original manuscript did not claim that the upregulation of Wnt pathway transcripts was evidence that the pathway was active, merely that it suggested that such a hypothesis was plausible.

However, we disagree with the reviewer that changes in TCF binding site accessibility can’t be used to characterize Wnt signaling activity. β-catenin/TCF transcriptional activation is driven in part through the recruitment of histone acetyl-transferases (ncbi.nlm.nih.gov/pmc/articles/PMC302022/), which increase chromatin accessibility. Further, transcriptional repressors such as groucho inhibit transcription by decreasing chromatin accessibility through the recruitment of histone de-acetylases

(ncbi.nlm.nih.gov/pmc/articles/PMC316998/). Therefore, an increase in the accessibility of TCF binding sites would be an expected consequence of an increase in TCF transcriptional activation, not repression. Our updated manuscript clarifies this rationale. Furthermore, our iCRT14 data demonstrates that chromatin accessibility changes near TCF binding motifs during regeneration are TCF dependent (Figure 3A), providing validation for this approach.

While we recognize that Wnt signaling is highly complex, we do not believe that this fact prevents us from using our data to characterize canonical Wnt signaling. TOPFLASH—a TCF-based reporter construct—is a well-established and widely used method for assaying Wnt signaling activity. Our ATAC-seq analysis provides information that is comparable to TOPFLASH, as it measures TCF activity via TCF-dependent changes in chromatin accessibility. Therefore, despite the complexities of Wnt signaling, we argue that TCF transcriptional activation can provide a meaningful readout of Wnt signaling activity.

Although we maintain that the above rationale justifies our conclusions regarding Wnt signaling during regeneration, we nonetheless recognize that stating that Wnt signaling is activated during normal foot regeneration is a strong claim that would ideally be validated through multiple sources of evidence (such as characterizing β-catenin protein abundance and localization). We have therefore softened the language we use in discussing our ATAC-seq and RNA-seq data throughout our updated manuscript to be less definitive.

However, our primary conclusion that head formation is the default outcome promoted by injury does not hinge on how we interpret our sequencing data. Rather, the primary evidence for this conclusion is presented in Figure 5, where we demonstrate that prolonged injuries in the body column exclusively leads to ectopic head formation—despite the fact that mid-body column tissue is equally capable of regenerating either a head or a foot. In addition, because Wnt signaling is necessary and sufficient for head formation, our finding that prolonged puncture wounds and aboral amputations induce ectopic heads demonstrates that the generic injury response can induce Wnt signaling. We provide further support for this conclusion with the addition of data showing that iCRT14 significantly reduces the number of ectopic tentacles formed after prolonged injury, demonstrating that the phenotype is TCF-dependent.

3. The conceptual interpretation of the results clings unnecessarily closely to the Gierer/Meinhardt framework and remains insufficiently tested.

a. Important findings since Gierer/Meinhardt are not cited or introduced, including for example the suggestion of wound-type specific responses already within the first hour after injury (Chera et al., 2009) or the recent report of dynamic, actin-mediated intrinsic tissue polarization (Livshits et al., 2017). These papers and their key findings need to be cited in the introduction and they need to be discussed in the context of an evaluation of the author's model in the discussion. Moreover, the authors missed that a wound-context independent activation of Wnt pathway transcription also occurs in planarians. This paper needs to be cited (Stuckemann et al., 2017) and the interpretation of wound type independent Wnt signaling in this study might also be of interest in contemplating the Hydra data.

As discussed above, we agree we should have included discussion of Chera et al., 2009 in our manuscript. We have addressed this issue in the revised text and now discuss the findings from Chera et al. throughout the manuscript. We have also added additional experimental evidence that directly addresses aspects of the Chera model (Figure 1J-M, Figure 1—figure supplement 5).

Regarding the data from Livshits et al. (2017), while their findings are interesting, we do not believe their work influences the interpretation of our data. Livshits et al. exclusively focused on how axis orientation is re-established in injury contexts when pre-existing positional information has presumably been lost (i.e. in very small tissue fragments). In contexts where a defined Wnt gradient is present, it is well established that Wnt signaling overrides cytoskeletal orientation. Furthermore, Livshits et al. do not perform any experiments that shed light on how the direction of polarity along the rebuilt axis is established. Their work therefore does not provide insight into the question of head vs. foot decisions in their chosen injury context. Our injury context does not cause the loss of the pre-existing axis and is primarily concerned with head vs. foot specification. Additionally, more recent research has called into question important aspects of the findings presented in the Livshits et al. study (pubmed.ncbi.nlm.nih.gov/32871156). We therefore argue that omission of Livshits et al. from our study does not meaningfully detract from the interpretation of our data.

We agree that findings from the planarian literature have interesting and important implications for the interpretation of our data. While we did cite the planarian literature (see list below), we did omit two important studies (Stuckmann et al., 2017 and Petersen and Reddien, 2009). We have rectified these omissions in our updated manuscript. More broadly, we have modified the text to more extensively discuss findings from bilaterian whole-body regeneration research in both the introduction and discussion.

Planarian papers cited in the original manuscript:

Owlarn and Bartscherer, 2016

Wurtzel et al., 2015

Wenemoser et al., 2012

Owlarn et al., 2017

b. Necessary tests of the present model include examination of the assumption that the inhibition of Wnt signaling is really sufficient for the conversion of an established head organizer into a foot organizer. This could be easily done by treating regenerating Hydra with iCRT14 post 8 h or when the putative Wnt signaling requirement in the initial wound response has passed. Further interesting validations/tests could include the examination of position-dependence of impalement-induced head formation along the body axis (does the efficiency increase with the distance to existing or former "inhibitor sources"? Are any injuries capable of inducing feet?) or qPCR comparisons of Wnt induction efficiency between the two Hydra strains.

We wish to clarify an apparent misunderstanding mentioned by the reviewer. We are not claiming that inhibition of Wnt signaling is sufficient for conversion of an established head organizer into a foot organizer. We do claim that Wnt signaling is transiently activated during early foot regeneration, but in our discussion of this hypothesis we do not claim that this transient activation temporarily results in head organizer formation. Head and foot organizers are self-maintaining and insensitive to inhibitory signals from the surrounding tissue. If a true head organizer formed during foot regeneration it would always trigger ectopic head formation.

We believe that treating regenerating Hydra after 8 hours post amputation would likely not be informative, because our molecular data indicates that Wnt signaling has already been repressed in foot regenerates by this time point. In addition, we believe that exploring the effect of position along the oral-aboral axis on ectopic head formation would be largely redundant with results already published in the literature (in particular the MacWilliams studies from 1983), which extensively tested the influence of position on secondary axis formation in Hydra using grafting experiments.

To provide additional evidence on the induction of canonical Wnt signaling following aboral-facing amputations, we have added new experimental data in Figure 5E-G of the updated manuscript. In these experiments, we demonstrate that prolonging aboral-facing amputation injuries can induce ectopic head formation at the aboral pole, and that this can be inhibited by tissue undergoing head regeneration at the oral pole. Because Wnt signaling is necessary and sufficient for head organizer formation in Hydra, these data demonstrate that aboral injuries can induce Wnt signaling.

c. The model disregards the organizing influence of the Hydra foot. As is, one might expect that pieces resulting from bisection and additional foot amputation might regenerate double heads (due to absence of the "source" of the long-range inhibitor). This is not observed. Also, the close spacing of the injury-induced double heads in Figure 5 is a problem for the Gierer/Meinhardt framework, as the inhibitor emanating from each head would be expected to interfere with the formation/maintenance of the other head organizer due to their close spacing.

The question of why ectopic heads don’t form when both organizers are amputated is indeed interesting and highly relevant to our model. Our inclusion of the experiments in Figure 5E-G addresses this question by looking at this precise injury context. Within the context of our model, the lack of ectopic head formation can be accounted for through two observations: 1) regenerating head tissue gradually increases the amount of inhibition it produces throughout the course of regeneration, 2) head regeneration is faster the closer the amputation site was to the original head organizer. Thus, in cases where both organizers are amputated, the oral-facing amputation injury will recover its inhibitory capacity first, and will subsequently block head formation at the aboral-facing amputation (these points are discussed in the results). The data we present in Figure 5E-G are consistent with this model, as they demonstrate that aboral amputations can induce ectopic heads, but that this can be blocked by tissue undergoing head regeneration at the oral pole.

The close spacing of the injury induced ectopic heads in some transversely animals can also be readily incorporated into our model based on the observation that regenerating heads do not become an effective source of inhibition for some time after they have become committed to head formation (see the discussion in MacWilliams et al., 1983, doi.org/10.1016/0012-1606(83)90325-1). In transverse impalement experiments, both injury sites are at the same position along the oral-aboral axis, and so head organizer formation would be expected to have highly similar kinetics at the two injuries. Essentially, the difference in how long it takes the two organizers to form does not provide enough time for either nascent head to produce sufficient inhibition to block the other. As previously noted, once established, head organizers are self-sufficient and insensitive to inhibitory signals from the surrounding tissue.

4. Validation of the importance of bZIP binding motifs in Wnt promoter regions.

As is, this section reads very much like a "just so" story. First, the authors have to analyze the probability of bZIP binding motif occurrence per kb in the hydra genome. Second, they should examine the occurrence of the motif in non wound-sensitive promoters. Third, they have to demonstrate significant enrichment of the motif in wound-induced genes. The strong emphasis on the importance of the bZip binding motifs in the present manuscript further necessitates experimental validation- either in the manuscript itself or by citing published data in a peer-reviewed journal.

We agree that motif enrichment analyses would be an informative addition to the paper. We have in fact performed such analyses and found that CRE sequences are significantly enriched in wound responsive elements. We omitted that work in favor of using the chromVAR analytical approach, but we have now integrated it into our updated manuscript (Figure 4A; Supplementary file 3).

We also agree that, as originally presented, the consideration of only a single injury-responsive motif appears rather arbitrary. In our revised manuscript, we present a systematic analysis of all motifs that exhibited an injury induced increase in chromatin accessibility from 0 to 3 hpa (Figure 4A). Using this systematic approach, we find that while there are other motifs that could plausibly play a role in injury-induced Wnt upregulation, CRE-binding bZIPs remain the most likely candidate regulators.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Revisions:

1. The authors now added acridine orange stainings to demonstrate that apoptosis is induced in head and foot regenerates, which is in contrast to what Chera et al. published in 2009. However, the acridine orange stainings have quite a bit of background and variability (and they are not quantified). As the Chera et al., data also looks convincing, the authors are encouraged to repeat a TUNEL staining that Chera et al., used. This would allow a better comparison between the differing results.

We agree that repeating the published TUNEL staining experiment would be a helpful clarification regarding the discrepancy between our manuscript and Chera et al., 2009. Before using acridine orange, we did try TUNEL staining and found that the results were inconsistent and gave high background at early regeneration time points. For this reason, we turned to acridine orange as an alternative approach because it has been previously validated as specifically staining apoptotic cells based on nuclear morphology and was shown to mirror TUNEL staining in Hydra (Cikala et al., 1999; Kuznetsov et al., 2002). With acridine orange, we obtain very clear results so this is what we included in the manuscript. In preparing this revision, we did try the TUNEL staining again, but we still could not obtain reasonable results. It should be noted that we saw no evidence of asymmetric TUNEL staining in the experiments that we performed. Rather, the results showed high background and were inconsistent from animal to animal (see Author response image 1). It is important to point out that our RNA-seq results mirror our acridine orange results in that we see symmetric upregulation of genes involved in apoptosis in both head and foot regeneration. Therefore, we have two different lines of evidence that refute the published TUNEL results in Chera et al. In addition, we have added a citation to our manuscript for a recent pre-print article that included TUNEL staining at 2 hpa that also found no evidence of asymmetric apoptosis.

Author response image 1
TUNEL staining gives inconsistent results in uninjured and regenerating tissue.

Whole uninjured animals, or animals that were bisected and allowed to regenerate for 1 or 3 hours, were fixed and then stained using the ApopTag TUNEL kit (Σ-Aldrich, S7110). All timepoints and regeneration types, including uninjured controls, exhibited highly variable TUNEL signal..

2. The literature describing the generic injury response that gradually polarizes that has been extensively described in planarians should be better cited. For example, Gurley et al., Dev Biol 2010 should be cited.

We have added citations for Gurley et al. in the introduction, results and discussion. We thank the reviewer for pointing out this omission.

3. Please arrange panels in Figure 3 according to how they are cited to facilitate easy reading of the manuscript.

The panels in Figure 3 are now arranged in the order that they are cited in the text.

https://doi.org/10.7554/eLife.60562.sa2

Article and author information

Author details

  1. Jack F Cazet

    Department of Molecular and Cellular Biology, University of California, Davis, Davis, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7331-5631
  2. Adrienne Cho

    Department of Molecular and Cellular Biology, University of California, Davis, Davis, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6837-420X
  3. Celina E Juliano

    Department of Molecular and Cellular Biology, University of California, Davis, Davis, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration, Writing - review and editing
    For correspondence
    cejuliano@ucdavis.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4222-0987

Funding

National Institutes of Health (R35 GM133689)

  • Celina Juliano

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Bruce Draper, Gary Wessel, Rob Steele, Bryan Teefy, Sergio Campos, Ben Cox, Thomas Holstein, Anja Tursch, Prashanth Rangan, Jeffrey Farrell, and Stefan Materna for critical reading of the manuscript; the members of the Juliano lab for their thoughtful input on this study; Vanessa Rashbrook, Emily Kumimoto, Siranoosh Ashtari, and Lutz Froenicke from the DNA Technologies and Expression Analysis Core at the UC Davis Genome Center (supported by NIH Shared Instrumentation Grant 1S10OD010786-01) for technical advice and assistance with library preparation and sequencing; and the Vincent J. Coates Genomics Sequencing Laboratory at UC Berkeley for additional library preparation and sequencing.

Senior Editor

  1. Edward E Morrisey, University of Pennsylvania, United States

Reviewing Editor

  1. Tatjana Piotrowski, Stowers Institute for Medical Research, United States

Publication history

  1. Received: June 30, 2020
  2. Accepted: March 28, 2021
  3. Accepted Manuscript published: March 29, 2021 (version 1)
  4. Version of Record published: April 15, 2021 (version 2)

Copyright

© 2021, Cazet et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,946
    Page views
  • 194
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Fabio Da Silva et al.
    Research Article Updated

    Retinoic acid (RA) is an essential signaling molecule for cardiac development and plays a protective role in the heart after myocardial infarction (MI). In both cases, the effect of RA signaling on cardiomyocytes, the principle cell type of the heart, has been reported to be indirect. Here we have developed an inducible murine transgenic RA-reporter line using CreERT2 technology that permits lineage tracing of RA-responsive cells and faithfully recapitulates endogenous RA activity in multiple organs during embryonic development. Strikingly, we have observed a direct RA response in cardiomyocytes during mid-late gestation and after MI. Ablation of RA signaling through deletion of the Aldh1a1/a2/a3 genes encoding RA-synthesizing enzymes leads to increased cardiomyocyte apoptosis in adults subjected to MI. RNA sequencing analysis reveals Tgm2 and Ace1, two genes with well-established links to cardiac repair, as potential targets of RA signaling in primary cardiomyocytes, thereby providing novel links between the RA pathway and heart disease.

    1. Cell Biology
    2. Developmental Biology
    Shaun Abrams, Jeremy F Reiter
    Research Article

    Craniofacial defects are among the most common phenotypes caused by ciliopathies, yet the developmental and molecular etiology of these defects is poorly understood. We investigated multiple mouse models of human ciliopathies (including Tctn2, Cc2d2a and Tmem231 mutants) and discovered that each displays hypotelorism, a narrowing of the midface. As early in development as the end of gastrulation, Tctn2 mutants displayed reduced activation of the Hedgehog (HH) pathway in the prechordal plate, the head organizer. This prechordal plate defect preceded a reduction of HH pathway activation and Shh expression in the adjacent neurectoderm. Concomitant with the reduction of HH pathway activity, Tctn2 mutants exhibited increased cell death in the neurectoderm and facial ectoderm, culminating in a collapse of the facial midline. Enhancing HH signaling by decreasing the gene dosage of a negative regulator of the pathway, Ptch1, decreased cell death and rescued the midface defect in both Tctn2 and Cc2d2a mutants. These results reveal that ciliary HH signaling mediates communication between the prechordal plate and the neurectoderm to provide cellular survival cues essential for development of the facial midline.