• Aucun résultat trouvé

 

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