The Hydra regeneration model was used in this PhD project. Hydra was shown to have a unique asymmetric cellular and molecular response upon bisection. As shown in 1995, 2004, 2009 and 2011 by the Galliot lab, head-‐regenerating tips are characterized by an immediate activation of the MAPK/RSK/CREB signaling pathway, an activation that is necessary to induce cell death and i-‐cell proliferation. However, the immediate injury signals remained unidentified. A previous PhD student had started to investigate the production of ROS signals but could not identify an asymmetrical regulation, leaving their putative role on apical versus basal regeneration unclear (Reiter, 2014). This project was designed to clarify what are the immediate injury-‐induced signaling molecules that play a role in the early discrimination between the two programs driving apical versus basal regeneration.
The AIM of this project was to study the earliest phases of regeneration and wound healing in Hydra in order to identify signaling molecules that lead to injury-‐induced cell death on one side of the cut and not on the other.
In order to tackle this biological question I focused on:
1. Studying the cellular processes that are occurring after two types of injury in Hydra. This was used to deduce which type of ROS signaling molecule is regeneration-‐specific.
2. Following the role of different stem cell lineages on ROS production, to tackle the putative ROS – cell death crosstalk that may occur in i-‐cells.
3. Dissecting the impact of different types of ROS on wound healing and regeneration.
4. Analyzing the regulation of redox signaling pathway during the regeneration in order to explore further the possible key regulator proteins in Hydra.
RESULTS
CHAPTER-1 INJURY-INDUCED ROS SIGNALING
Results presented in this chapter are the main part of my work during this doctoral thesis. The data presented here led to our current working model on ROS signaling in Hydra in the context of wound healing and regeneration.
We characterize H2O2 as regeneration-‐specific, asymmetrically produced signaling molecule highly present during apical but not basal regeneration. Next, we show that in absence of ROS, necessary molecular events such as phosphorylation of CREB protein and injury-‐induced cell death are not present during apical regeneration.
Furthermore we analyze the metabolic enzymes that are active during oxidative eustress, catalase and SOD, and find that different ratios of their activity contribute to the different levels of H2O2 during apical and basal regeneration. As a conclusion, ROS are important signaling molecules in Hydra, with mitochondrial superoxide being the key player for wound healing and hydrogen peroxide (H2O2) orchestrating molecular events that launch apical regeneration. While there are still many questions to answer, this work presents a significant effort to elucidate the immediate signaling in regenerating Hydra.
This project was started with the previous PhD student in the lab, Silke Reiter, who contributed significantly in setting up the methods for ROS detection, and in obtaining data that lead us to further explore the putative ROS – cell death crosstalk (Reiter et al., 2012; Reiter, 2014). Dr. Osvaldo Chara (Dresden) contributed with his work on mathematical signaling that predicted the existence of an early-‐injury asymmetrically produced signaling molecule, as presented in Figure 1 and Figure S1. This model is based on the quantitative data previously obtained by the Galliot lab on injury-‐
induced cell death (Chera et al., 2009b). Chemical ROS scavengers, such as Tiron were proposed by Dr. Denis Martinvalet, a pharmacological agent of the utter importance to dissect the ROS signaling pathway. Finally, my contribution to this project was in developing the imaging conditions as such as they can be used for producing quantitative data, together with the optimization of the other ROS detecting methods, mainly biochemical. The hypotheses that led to the experimental planning, the analysis of the obtained data and their integration into existing knowledge were the result of a fruitful collaboration between my mentor Brigitte Galliot and myself.
Asymmetric regulation of injury-‐induced ROS signaling in regenerating Hydra
Nenad SUKNOVIC1, Silke REITER1, Osvaldo CHARA2, Wanda BUZGARIU1, Denis MARTINVALET3 * and Brigitte GALLIOT1
1 Department of Genetics and Evolution, Faculty of Sciences, University of Geneva, Switzerland;
2 Center for Information Services and High Performance Computing, Technische Universität Dresden, Dresden, Germany;
3 Department of Cell Physiology and Metabolism, Faculty of Medicine, University of Geneva, Switzerland; *present address: Department of Cancer Biology, Venetian Institute of Molecular Medicine, Padova, Italy
Corresponding author: brigitte.galliot@unige.ch
Keywords: paracrine signals, injury-‐induced ROS signals, mitochondrial superoxide, asymmetrical regulation of hydrogen peroxide, injury-‐induced cell death
ABSTRACT
What signals elicit two distinct regenerative responses on each side of the cut in bisected Hydra remains unknown. A mathematical modeling approach based on quantitative data linked to MAPK activation and injury-‐induced cell death predicts an immediate release of a locally restricted short-‐lived signal in head-‐regenerating tips. Reactive oxygen species (ROS) are obvious candidates and here we monitored their production, role and regulation. We show that mitochondrial superoxide (mitoO2.-‐) and hydrogen peroxide (H2O2) are produced within minutes following bisection, MitoO2.-‐ predominantly by the gastrodermal epithelial stem cells, symmetrically on each side of the cut but also after lateral nick suggesting a regeneration-‐independent role. Upon mitoO2.-‐ scavenging, animals do not heal properly, while in sod1(RNAi) animals that accumulate mitoO2.-‐, wound healing is enhanced proving that mitoO2.-‐ contributes to it. Apical regeneration is only transiently delayed in the absence of mitoO2.-‐. By contrast, H2O2 levels are highly asymmetrical, three-‐fold higher in apical-‐ than in basal-‐regenerating tips, while undetectable after lateral nick, indicating that H2O2 is specifically enhanced upon apical regeneration. High levels of H2O2 are necessary to trigger interstitial-‐derived cell death and H2O2 levels get significantly lower in the absence of interstitial-‐derived cells, suggesting they get amplified upon cell death. Activities of the ROS-‐
processing enzymes, super oxide dismutase (SOD) and catalase are asymmetrical, in agreement with asymmetrical H2O2 levels. This study shows that H2O2 acts as an immediate paracrine signal to trigger apical regeneration, and its asymmetrical regulation appears crucial to activate two distinct regenerative responses in Hydra.
INTRODUCTION
Organisms that elicit regeneration respond to damage and injury by achieving a complex cellular remodeling that relies on the combination of several cellular processes as cell death, cell dedifferentiation, cell proliferation and cell differentiation (Bergmann and Steller, 2010;
Vriz et al., 2014; Perez-‐Garijo and Steller, 2015). This integrative process, tightly controlled in time and space, leads to a perfect 3D reconstruction of the missing structures. In this study, we make use of the freshwater cnidarian Hydra polyp to investigate the mechanisms launched at the time of injury that drive this cellular remodeling. Indeed, the tube-‐shaped Hydra possesses the amazing ability to regenerate any missing part after bisection of its body column, regenerating the apical part (head) on one side and the basal disc (foot) on the other side . Hydra is made of two cell-‐layers, epidermis and gastrodermis, and three stem cell populations that cannot replace each other, the unipotent gastrodermal and epidermal epithelial stem cells (eESCs, gESCs), and the multipotent interstitial stem cells (ISCs) (Bosch, 2008; Bosch et al., 2009; Galliot, 2013b; Buzgariu et al., 2015). Our laboratory showed that upon mid-‐gastric bisection, head regenerating tips exhibit an immediate wave of cell death that predominantly affects all cells of the interstitial lineage. The dying cells transiently release signals such as Wnt3 that trigger the mitotic division of the surrounding progenitors and the up-‐regulation of Wnt3 in the gESCs (Chera et al., 2009). This cascade of events seems sufficient to rapidly launch the apical regeneration program after mid-‐gastric bisection. Upstream of this cellular remodeling, the asymmetric activation of the MAPK/RSK/CREB pathway provides the signaling triggering injury-‐induced cell death (Galliot et al., 1995; Kaloulis et al., 2004; Chera et al., 2011). Indeed, a short exposure to the MEK inhibitor UO126 at the time of bisection suffices to prevent MAPK activation and CREB phosphorylation, inhibit injury-‐induced cell death and significantly delay apical regeneration (Kaloulis et al., 2004; Chera et al., 2011). The aim of this study is to identify the injury-‐
induced signals that translate into an asymmetric activation of MAPK signaling and induce cell death in apical-‐regenerating (AR) tips.
Reactive Oxygen Species (ROS) appeared as suitable candidate molecules as hydrogen peroxide (H2O2) is immediately produced upon injury in the absence of any transcriptional response as demonstrated in the zebrafish larvae (Niethammer et al., 2009) and more
generally in most wound healing contexts (Rojkind et al., 2002; Moreira et al., 2010; Xu and Chisholm, 2014). ROS signaling is not only involved in the wound healing process but also implicated in the wound closure of the Drosophila embryo (Moreira et al., 2010; Razzell et al., 2013), in the regeneration of the Drosophila gut (Buchon et al., 2009), the Xenopus tadpole tail (Love et al., 2013), the adult zebrafish fin (Gauron et al., 2013), the Drosophila wing (Santabarbara-‐Ruiz et al., 2015; Santabarbara-‐Ruiz et al., 2019).
ROS metabolism arose several billion years ago with H2O2 produced directly from O2.-‐
through dismutation and decomposed to H2O and oxygen by catalases in bacteria, plants and metazoans (Inupakutika et al., 2016). There are two major sources of superoxide (O2-‐-‐) production in the cell, either the enzymes of the mitochondrial Electron Chain Transport (Murphy, 2009) or the membrane NADPH Oxidase (NOX) enzymes (Bedard and Krause, 2007; Jiang et al., 2011). O2— is a highly toxic and unstable molecule that upon dismutation by the Super Oxide Dismutase (SOD) is transformed into H2O2, a relatively stable ROS molecule that shows a longer half-‐life time when extra-‐cellular and can thus function as a second messenger, in a cell-‐autonomous or non-‐cell-‐autonomous fashion. The biological impact of ROS metabolism shares numerous similarities in plants and animals (Sies, 2017;
Noctor et al., 2018). At high concentrations, H2O2 is able to trigger cell death either by activating the apoptosis signal-‐regulated kinase ASK (Saitoh et al., 1998; Furuhata et al., 2009) and/or through MAPK phosphatase inactivation and subsequent JNK activation (Kamata et al., 2005; Chen et al., 2009).
A series of pharmacological inhibitors are available to test the biological impact of these two sources of injury-‐induced superoxide. Among them, Tiron is a cell-‐permeable iron chelator that is commonly used as a superoxide scavenger (Yamada et al., 2003; Taiwo, 2008; Han and Park, 2009), while diphenyleneiodonium (DPI) and apocynin (APO) act as unspecific antioxidants. DPI is a flavoprotein inhibitor that prevents the activity of several oxidases including the NOX enzymes as well as the enzymes of the mitochondrial Electron Transport Chain (ETC) (Altenhofer et al., 2015), while Apocynin exhibits ROS-‐scavenging properties and Rho kinases inhibition. By contrast VAS2870 specifically inhibits the NOX enzymes. DPI, Apocynin and VAS-‐2870 proved to negatively affect wound healing and regeneration in zebrafish and Xenopus (Niethammer et al., 2009; Gauron et al., 2013; Love et al., 2013).
Given the tools available to monitor ROS signaling, we decided to investigate what ROS molecules are produced in Hydra bisected at mid-‐gastric position, and what role they might play in regenerating Hydra. We focused on three aspects: (i) the spatial and temporal production of ROS signals during Hydra regeneration; (ii) the role of ROS signals on wound healing as well as regeneration; (iii) the role of ROS signals in the asymmetric activation of the MAPK/RSK/CREB pathway and injury-‐induced cell death.
RESULTS
Mathematical prediction of an immediate injury-‐induced signal produced in Hydra regenerating tips
We hypothesize that mid-‐gastric bisection releases or produces a diffusing signal, yet unidentified and notated U, which would activate the MAPK pathway. We tested the hypothesis by developing a mathematical model comprising three components: the cells, the extracellular signaling and the intracellular signaling (Fig. 1C). The peculiar space response of the interstitial cells following the amputation-‐induced apoptosis indicates that the mathematical model should incorporate the space dimension. Since the only relevant direction is the distance perpendicular to the amputation plane (Fig. 1A) the mathematical model involves not just ordinary differential equations (ODE) but also one-‐dimensional partial differential equations (PDE). In the absence of quantitative information on the interaction between cells and signaling, the model is constrained by a number of plausible assumptions. According to the model, mid-‐gastric bisection releases or produces the signal U, which diffuses and undergoes lytic degradation (Eq. 1: !!"! = 𝑒!! !
!!! −𝑓 𝑈 ). For simplicity it is assumed that the signal is linearly degraded, which means that the effective concentration of the signal would be lower than the Km of a putative enzyme responsible for its degradation.
The link between the signal U and the MAPK pathway is proposed as follows: the signal U activates the phosphorylation of the inactive forms of the MAPK pathway (Mi). By simplicity it is considered that the reaction of activation is bilinear with the concentration of the substrate, the non-‐phosphorylated or inactive form of the enzymes (Mi) and the signal (U)
(Eq. 2 in the Materials and Methods section). The backward reaction rate is assumed linear in the phosphorylated or activated signal (Ma, Eq. 2: !!"!! =𝑗 𝑈 𝑘− 𝑘 𝑈 +𝑙 𝑀! ).
The activated form of MAPK (Ma) induces apoptosis by triggering the apoptotic cascade (via caspases) in the interstitial lineage. The simplest way to model this process is to assume that the rate of change of the apoptotic cell density is proportional to the interstitial cell density and the concentration of activated MAPK. It was previously described that three stages of apoptotic cells appear sequentially after mid-‐gastric bisection in Hydra (Chera et al., 2009).
However, the first two stages (formerly called early and advanced apoptotic cells) are kinetically equivalent (data not shown) suggesting that both stages could be modeled as a single stage hereafter constituted by the early apoptotic cells (Ae). Hence, in the model Ma induces apoptosis by linearly decreasing the density of interstitial cells (I), which is the source of the density of the early apoptotic cells (Ae, Eqs. 3 and 4). That is, the number of interstitial cells (I) is reduced and the number of early apoptotic cells (Ae) is augmented in the same proportion (Eq. 3: !"!"= 𝑎 𝑊 −𝑏 𝑀! 𝐼 and Eq. 4: !!!"! =𝑏 𝑀! 𝐼−𝑐𝐴!). These cells are in turn linearly transformed in late apoptotic cells, which are also linearly depleted (Al, Eq. 5: !!!"! =𝑐𝐴!−𝑑𝐴!). The early apoptotic cells release Wnt3 (W), which diffuses and is linearly degraded (W, Eq. 6: !!"! = 𝑔!!!!!! −ℎ 𝑊 +𝑖𝐴!) while it promotes the mitotic division of the neighbor interstitial cells (Eq. 3).
The model was fitted to previous experimental results of cell dynamics after mid-‐gastric bisection (Chera et al., 2009) as shown in (Fig. 1D). The density of interstitial, early apoptotic and late apoptotic cells was calculated at different time points in two space regions, Region-‐
1, i.e. 0 -‐ 100 µm close to the amputation plane, and, Region-‐2 corresponding to the 100 -‐
200 µm area underneath. By using the procedure detailed in Supplementary Methods, the model was successfully fitted to the experimental data (Fig. 1D-‐E). The predicted pattern of the hypothetical signal U is localized within approximately 150 µm from the amputation plane and it vanishes after one hour (Fig. 1E). Although more diffuse, the model-‐predicted pattern of Wnt3 (Fig. 1E) is also in agreement with the previously reported distribution in Hydra after amputation (Chera et al., 2009). The increase of interstitial cell number far away from the amputation plane (region 2, Fig. 1D) is successfully reproduced by the model. Other mechanisms assuming chemo-‐attracted migration of interstitial cells driven by either U or
Instead of the Wnt3-‐induced compensatory proliferation of interstitial cells, alternative mechanisms were considered: interstitial cell diffusion, interstitial cell migration driven by a chemo-‐attractant process guided by the U gradient or the Wnt3 gradient or combination of all of them. All of these mechanisms were implemented in mathematical models (see Supplementary information). None of them successfully fitted to the previous experimental data (data not shown).
Immediate production of mitochondrial superoxide upon injury in Hydra
To evidence injury-‐induced ROS production in Hydra, we applied two different live imaging methods (Fig. 2A): the free permeable radical sensor H2DCFDA previously used in hydrozoans including Hydra (Blackstone, 2001; Murugadas et al., 2016), which generates fluorescent signals upon any kind of total ROS catalyzed reaction (Fig. 2B, left). As a second method, we used the mitochondrial localized probe MitoSOX that emits a fluorescent signal when oxidized by mitochondrial superoxide (mitoO2.-‐) (Mukhopadhyay et al., 2007) (Fig. 2B, right). Both methods could be adapted to detect levels of ROS on whole mounts of intact, regenerating and wounded Hydra. Since the specificity of H2DCFDA is not sufficient to discriminated between the different ROS molecules, we utilized mostly MitoSOX to characterize the mitochondrial superoxide production in two different contexts: mid-‐gastric bisected animals and animals submitted to a lateral, non-‐regeneration inducing, cut.
To identify the cells that produce mitoO2.-‐, we used two transgenic lines produced by the Hobmayer lab that express the LifeAct-‐GFP reporter in their epidermal and gastrodermal epithelial cells respectively (Aufschnaiter et al., 2017; Livshits et al., 2017). We detected mitoO2.-‐ signals in both apical-‐ and basal-‐regenerating (AR, BR) tips, mostly in the gastrodermal epithelial layer (Fig. 2C). To quantify the percentage of total mitoO2.-‐produced in the gastrodermis, we quantified the MitoSOX signal dots that co-‐localize with the GFP-‐
expressing cells and found around 80% of total mitoO2.-‐ produced by the gastrodermal ESCs (Fig. 2D).
Symmetrical levels of mitoO2.-‐ in apical and basal regenerating tips
To assess the putative role of injury-‐induced mitoO2.-‐, we first characterized the temporal and spatial variations in mitoO2.-‐ levels during the immediate phase of apical and basal regeneration (Fig. 2E-‐F). MitoO2.-‐ signals were initially detected at 10 minutes post-‐
amputation (mpa), forming a ring-‐like structure at the bisected planes of both upper and
lower halves (Fig. 2E). Then, we observed an accumulation of mitoO2.-‐ in the AR-‐ and BR-‐tips, up to 60 mpa. Since MitoSOX is a cumulative type of cellular dye, we could not evidence any fluctuation of mitoO2.-‐ levels, but a rather sustained increase. To compare the mitoO2.-‐ levels in AR and BR tips, we counted the total number of MitoSOX dots every five minutes from 5 to to 60 mpa. We recorded similar levels of mitoO2.-‐ in apical and basal regenerating tips, suggesting a similar regulation in both contexts (Fig. 2F).
Next, we monitored the mitoO2.-‐ levels in wounded non-‐regenerating tissues. We used polyps from the epidermal and gastrodermal LifeAct-‐GFP strains as above. We injured them laterally, making a notch with a scalpel at mid-‐body position. As observed in the AR and BR tips, we detected mitoO2.-‐ dots located outside the epidermal GFP+ cells and the epidermal layer, consistent with a source in the gastrodermal epithelial cells (Fig. 2G).
Detection of asymmetrical levels of H2O2 in regenerating Hydra
To monitor the modulations of H2O2 levels, we applied to Hydra tissues the Amplex Red quantification method, which is widely-‐used in human leukocytes (Mohanty et al., 1997), human cancer cells (Fu et al., 2006), plants (Xiong et al., 2007), D. melanogaster (Venkatachalam et al., 2008), mice (Wang et al., 2015) and E. coli (Zhang et al., 2015). We quantified the H2O2 levels in the upper and lower halves during the first 60 mpa (Fig. 2H, left). We did not record any significant difference in H2O2 production between apical and basal regeneration during the first 10 minutes, while soon after, the H2O2 levels increase much faster in AR than in BR halves, reaching a three-‐fold difference at 60 mpa. By contrast, in a non-‐regenerative injury such as lateral nick, the H2O2 levels remain stable, indicating that, unlike mitoO2.-‐, the up-‐regulation of H2O2 is regeneration-‐specific (Fig. 2H, right). These results indicate that at least two distinct types of ROS molecules are produced in bisected Hydra, mitoO2.-‐ as an immediate injury-‐induced signal equally produced by the gastrodermal epithelial cells on each side of the bisection plane, and H2O2, a molecule that shows a highly asymmetrical regulation.
Superoxide scavenging leads to lower H2O2 production
As Hydra cells express three superoxide dismutase (SODs) genes and one catalase gene that all encode evolutionarily-‐conserved proteins (Fig. S2, Fig. S3, Fig. S4), we anticipated that the regulation of ROS turnover in Hydra relies on mechanisms that are similar to those active in plant and animal cells (Sies, 2017; Noctor et al., 2018). To further elucidate the respective
roles of mitoO2.-‐ and H2O2 in regeneration-‐related biological processes, we first modulated their levels pharmacologically preventing or scavenging superoxide production and thus negatively impact the production of H2O2 (Fig. 3A). We tested a series of anti-‐oxidant drugs such as Tiron and the NADPH-‐oxidase inhibitors DPI, apocynin and VAS2870 drugs (Yamada et al., 2003; Altenhofer et al., 2015). For all these drugs, we performed toxicological tests to identify the non-‐toxic working concentration (Fig. S5). We never detected any toxic or
roles of mitoO2.-‐ and H2O2 in regeneration-‐related biological processes, we first modulated their levels pharmacologically preventing or scavenging superoxide production and thus negatively impact the production of H2O2 (Fig. 3A). We tested a series of anti-‐oxidant drugs such as Tiron and the NADPH-‐oxidase inhibitors DPI, apocynin and VAS2870 drugs (Yamada et al., 2003; Altenhofer et al., 2015). For all these drugs, we performed toxicological tests to identify the non-‐toxic working concentration (Fig. S5). We never detected any toxic or