UNIVERSITE DE LA MEDITERRANEE AIX-MARSEILLE II École Doctorale des Sciences de la Vie et de la Santé

Année 2009 N° attribué par la bibliothèque

THESE

Pour l’obtention du grade de DOCTEUR DE L’UNIVERSITE DE LA MEDITERRANEE Spécialité : Microbiologie Présentée et soutenue publiquement par : Rémi DULERMO Le 15 décembre 2009

Étude des mécanismes de l’extrême tolérance aux radiations de la bactérie deserti par une approche de génomique fonctionnelle

JURY :

Mr Cheng-Cai ZHANG (Professeur de l’Université de la Méditerranée) Président Mme Marie-Agnès PETIT (CR1, INRA Jouy en Josas) Rapporteur Mr Pablo RADICELLA (DR, CEA Fontenay-aux-Roses) Rapporteur Mr Alain DOLLA (DR, CNRS Marseille) Examinateur Mme Pascale SERVANT (Maître de conférence à l’Université Paris-Sud) Examinateur Mr Arjan de GROOT (Chercheur CEA à Cadarache) Directeur de Thèse

Laboratoire d’Ecologie Microbienne de la Rhizosphère et d’Environnements Extrêmes DSV/IBEB/SBVME/LEMiRE, UMR 6191 CNRS/CEA/Univ. Aix-Marseille CEA Cadarache, F-13108 Saint Paul Lez Durance

Remerciements

J’ai commencé ma thèse le 02 octobre 2006 et après près de 3 ans au sein du LEMIRE, c’est avec regret que je quitte ce laboratoire. Je souhaite donc remercier les différentes personnes qui m’auront permis d’effectuer cette thèse et qui m’auront soutenu tout au long de ces 3 ans.

Je souhaite tout d’abord remercier l’équipe Deinococcus dans laquelle j’ai travaillé durant un peu plus de 3 ans. Merci à Arjan De Groot et Laurence Blanchard ainsi qu’à Boris Mirabella qui avait rejoint l’équipe pendant 1 an. Je tiens à dire qu’Arjan est quelqu’un de très compétent, de très rigoureux dans son travail et j’espère que j’ai pu acquérir au moins un peu de ses qualités scientifiques (je sais qu’il me reste encore un long chemin à parcourir pour arriver à son niveau). Ses conseils ont été précieux pour la réussite de ce travail. Merci aussi à Arjan pour avoir été très patient avec moi pendant les nombreuses corrections de ce manuscrit. Je pense que sans toi, je n’aurais pas réussi à finir dans les temps et à rendre une si bonne thèse, donc merci encore pour tes nombreuses suggestions et corrections. Je tiens aussi à remercier Arjan, Laurence, Thierry et Wafa de m’avoir choisi comme candidat à présenter au CEA pour la bourse de thèse car en venant à l’entretien, je ne pensais pas forcément être retenu. Merci aussi à Laurence, très compétente également et avec qui j’ai apprécié discuter de tout et rien et qui a aussi pu me donner des conseils dans les moments difficiles. J’ai été très content de travailler avec Boris, c’est quelqu’un de compétent et de très sympa, avec qui j’ai beaucoup apprécié de discuter. Je voudrais dire à Arjan, Laurence et Boris, ne changez pas. Merci également au CEA d’avoir financer ma thèse.

Je voudrais remercier toutes les autres personnes du LEMIRE que j’ai côtoyé durant ces 3 ans, Bruno Allainmat, Mohamed Barakat, Odile Berge, Géraldine Brandelet, Mélanie Bressan, Nathalie Byrne, Sylvain Fochesato, Agnès Gastaud, Zahar Haichar, David Lalaouna, Gilles de Luca, Philippe Ortet, Camille Profizi, Marie-Anne Roncato, Philippe Roumagnac, Lisa Sanchez, Catherine Santaella, Mathieu Schue et Françoise Tellier. Je souhaite encore plus remercier Sylvain, Agnès, Gilles, David et Marie-Anne pour leur bonne humeur et nos déconnades durant les repas ou les pots. Merci à Laurence qui me ramenait sur Terre lorsque je prenais volontairement la grosse tête ;-). Je remercie aussi tous les participants de l’équipe de foot (Arjan, Cyril, David (Lalalouna et Lemaire), Gilles, Miguel, Momo, Nicolas, Sylvain et tous les autres) avec qui j’ai passé de très bons moments et avec qui j’ai eu l’honneur de jouer 3 tournois au CEA Cadarache. C’est une équipe qui a un très bon état d’esprit (l’équipe a remporté 2 coupes de fair-play ; aller, l’année prochaine, vous gagnerez la coupe du CEA 3 Cadarache). Merci aussi aux supportrices qui venaient nous voir jouer et nous encourager (mention spéciale à Laure). Je remercie aussi Wafa et Zahar avec qui j’ai beaucoup apprécié de discuter au labo ou sur le chemin en rentrant avec le bus. Pour mon colocataire de bureau David, bon courage pour la suite de ta thèse, il te reste un peu moins de 2 ans, tu vas voir ça passe vite ;-). Surtout, reste comme tu es et prend ma relève pour les bétises à dire au labo ou à la cantine. Je souhaite remercier également d’autres personnes qui ne sont pas au LEMIRE, Christine Marol avec qui j’ai très souvent pris le bus entre la gare de triage et le labo et avec qui j’ai beaucoup discuté, Colette Jungas que j’ai connu grâce à Laurence et avec qui j’ai bien aimé discuter et déjeuner, Dorothée Jannaud qui devrait finir sa thèse peu de temps après moi et que j’apprécie beaucoup ainsi que Clémence Bonnot avec qui j’ai aussi beaucoup rigolé et discuté. Je te souhaite, Clémence, du courage pour la fin de ta thèse. Je souhaite encore remercier Miguel Costa, c’est quelqu’un de très très sympas. J’espère Miguel que tu pourras trouver rapidement un emploi fixe en France ou au Portugal.

Maintenant, je souhaiterais remercier les membres de ma famille, ma maman, Didier, mon grand frère, Thierry, mon frère jumeau, mes belles-sœurs Vanessa et Hui. Merci aussi à mon petit chat Lulu :-). Thierry, je te remercie infiniment pour l’aide que tu m’as apporté, notamment durant la rédaction de ma thèse, ainsi que pour nos discussions scientifiques sur ton sujet et le mien, même si parfois, il n’était pas facile de se comprendre puisque nous avions des sujets très différents. Je remercie énormément ma femme Ting ainsi que ses parents et toute sa famille pour leur grande gentillesse, leur amour et leur soutien. Merci aussi à Ting pour avoir réussi à me supporter dans les moments de stress. Je remercie aussi mes amis et plus particulièrement John et David pour leur amitié et leur soutien.

Merci aussi au personnel administratif du service SBVME, et entre autres à Elisabeth Dussauze, Carol Fortuno et Joséphine Ramon.

Enfin j’aimerais remercier l’ensemble des professeurs que j’ai eu tout au long de mon cursus et qui ont fait que j’ai réussi à en arriver là.

4

Résumé

5 6 Résumé

Le génome de Deinococcus deserti, une bactérie très radiotolérante, a été analysé et comparé à ceux de D. radiodurans et D. geothermalis. Environ 230 protéines sont spécifiquement conservées chez ces 3 espèces, dont IrrE, un régulateur essentiel pour la radiotolérance. D. deserti possède plusieurs gènes supplémentaires liés à la réparation de l’ADN, dont imuY et dnaE2 (ADN polymérases translesionnelles). En plus, D. deserti a 3 recA qui codent pour 2 protéines RecA différentes (RecAC et RecAP). Pour étudier ces gènes, des outils génétiques ont été mis au point. Différents résultats suggèrent qu’IrrE, nécessaire pour l’induction de plusieurs gènes après irradiation, a une activité peptidase. Les 2 RecA sont fonctionnelles pour la réparation de l’ADN. D. deserti est mutable par UV, ce qui nécessite ImuY, DnaE2 et RecAC, mais pas RecAP.

Abstract

Title: The study of the extreme radiation tolerance mechanisms of the bacterium Deinococcus deserti by a functional genomics approach

The of Deinococcus deserti, a highly radiation-tolerant bacterium, was analyzed and compared to those of D. radiodurans and D. geothermalis. About 230 are specifically conserved in these 3 , including IrrE, a regulator essential for radiotolerance. D.deserti has several supplementary DNA repair genes, like imuY and dnaE2 (translesion DNA polymerases). Moreover, D. deserti has 3 recA that code for 2 different

RecA proteins (RecAC et RecAP). To study these genes, genetic tools were developed for D. deserti. Different results suggest that IrrE, required for the induction of several genes after irradiation, has peptidase activity. The 2 RecA proteins are functional for DNA repair.

D. deserti is mutable by UV, which requires ImuY, DnaE2 and RecAC, but not RecAP.

7 8

SOMMAIRE

9 10 Sommaire

Sommaire...... 11 I) Introduction générale...... 17 1) Tolérance aux radiations et à la dessiccation de ...... 18 1.1) Le genre Deinococcus ...... 18 1.2) Le génome de D. radiodurans...... 23 1.2.1) Quelques caractéristiques ...... 23 1.2.2) Les gènes de la réparation de l’ADN...... 25 La réversion de bases endommagées ...... 25 Le système NER ...... 26 Le système BER...... 26 Le mismatch repair ...... 27 Le système RER...... 28 1.3) Transcriptomique et protéomique de D. radiodurans après irradiation ou dessiccation...... 34 1.3.1) Protéome après irradiation gamma ...... 34 1.3.2) Transcriptome après irradiation gamma/dessiccation...... 34 1.3.3) Les gènes ddrA, ddrB, ddrC, ddrD et pprA sont impliqués dans la radiotolérance...... 36 1.4) La régulation des gènes de la radiotolérance chez D. radiodurans...... 38 1.4.1) La régulation de recA chez D. radiodurans...... 38 1.4.2) Régulation globale des gènes de la radiotolérance ...... 41 1.5) Les mécanismes de la radiotolérance chez D. radiodurans ...... 46 1.5.1) Les mécanismes passifs...... 46 Le nombre de copies de génome...... 46 Le nucléoide de D. radiodurans...... 46 La protection contre le stress oxydant...... 48 1.5.2) Les mécanismes actifs ...... 49 Le mécanisme de ESDSA...... 50 Un système NHEJ-like chez D. radiodurans ? ...... 51 Autres protéines intervenant dans la réparation des cassures double brin de l’ADN...... 52 La lutte contre le stress oxydant...... 53 Dégradation et réparation des protéines endommagées après irradiation ...... 54 1.6) Conclusion sur les connaissances de la radiotolérance chez D. radiodurans...... 56 2) Deinococcus deserti ...... 58 3) Plan de la thèse...... 60 II) Matériels et Méthodes...... 63 Souches, plasmides, milieux et conditions de culture...... 63 Extraction d’ADN chromosomique ...... 63 Extraction d’ADN plasmidique...... 66 PCR...... 66 Clonage dans pCR4®BluntTopo® ...... 70 Digestion des plasmides...... 70 Déphosphorylation...... 70 Ligation...... 71 Constructions des plasmides ...... 71 Transformation des bactéries ...... 74 Cellules compétentes et transformation par choc thermique...... 74 Cellules compétentes et transformation par électroporation ...... 75 Conjugaison ...... 75 Survie bactérienne aux UV ...... 75 Mutagenèse induite aux UV...... 76 Survie bactérienne après dessiccation...... 76 Survie bactérienne après irradiation gamma ...... 76 III) Le génome de Deinococcus deserti...... 79 1) Annotation du génome de Deinococcus deserti...... 79

11 2) Exemples d’annotation de gènes...... 82 2.1) Gène DeideP101880 (imuY)...... 82 2.2) Gène Deide19010 (rpsL)...... 83 3) Gènes potentiellement impliqués dans la tolérance aux radiations et à la dessiccation...... 86 4) Article 1 : Alliance of proteomics and genomics to unravel the specificities of Sahara bacterium Deinococcus deserti ...... 89 IV) Mise au point des outils génétiques pour Deinococcus deserti...... 105 1) Electroporation ...... 105 1.1) Obtention d’un plasmide réplicatif chez D. deserti...... 105 1.2) Construction des cassettes de résistance et tests de mutagenèse par recombinaison après électroporation...... 107 2) La transformation par chimio-compétence...... 108 3) La conjugaison ...... 109 4) Intégration d’un plasmide par simple recombinaison homologue à l’aide de cellules chimio- compétentes ...... 110 5) Construction de mutants par double recombinaison homologue...... 112 5.1) Utilisation d’un fragment PCR/plasmidique ...... 113 5.2) Utilisation d’un plasmide circulaire contenant rpsL...... 113 6) Conclusions...... 116 V) Deux protéines RecA fonctionnellement différentes et mutagenèse induite par les UV chez Deinococcus deserti...... 119 1) Introduction ...... 119 2) Résumé de l’article ...... 121 3) Données supplémentaires...... 122 4) Article 2 : Mutagenic lesion bypass and two functionally different RecA proteins in Deinococcus deserti...... 125 VI) Les domaines fonctionnels de la protéine régulatrice IrrE...... 147 1) Introduction ...... 147 2) Résumé de l’article ...... 147 3) Conclusion ...... 149 4) Article 3 : Crystal structure of the IrrE protein, a central regulator of DNA damage repair in deinococcaceae ...... 151 VII) Discussion et perspectives ...... 169 1) Deinococcus deserti, un nouveau modèle pour étudier la radiotolérance des Deinococcus 169 2) Les autres bactéries radiotolérantes ...... 174 VIII) Bibliographie...... 179 IX) Annexes ...... 195

12

13 14

I

Introduction générale

15 16 I) Introduction générale

Maintenir l’intégrité du génome est essentiel pour tous les organismes. En effet, des mutations ou des réarrangements chromosomiques peuvent avoir des conséquences très graves pour les cellules. Les dommages de l’ADN peuvent provenir du stress oxydatif lié au métabolisme aérobie (facteurs endogènes) ou de facteurs extérieurs tels que les UV, la dessiccation ou d’autres agents génotoxiques (facteurs exogènes). Pour faire face à ces dommages, les organismes ont développé des systèmes de réparation de l’ADN. Cependant, tous les organismes ne sont pas égaux dans leur capacité à réparer les dommages de l’ADN. La bactérie Deinococcus radiodurans, découverte en 1956 (Anderson, 1956), est un des organismes les plus tolérant aux rayonnements UV et gamma connu à ce jour. Elle tolère des doses de rayonnement gamma entre 200 et 5000 fois plus élevées que celles tolérées respectivement par Escherichia coli et les cellules humaines (Minton, 1994). L’extrême tolérance de cette bactérie s’explique surtout par le fait qu’elle est capable de reconstituer un génome intact à partir d’un génome fragmenté par l’irradiation ou la dessiccation. De nombreux travaux ont été réalisés sur cette bactérie, mais les mécanismes exacts de la réparation du génome de D. radiodurans ne sont pas encore complètement élucidés (Blasius et al., 2008 ; Cox & Battista, 2005). De nombreuses espèces du genre Deinococcus ont été isolées, souvent dans des régions désertiques. Notre laboratoire, le LEMiRE (Laboratoire d’Ecologie Microbienne de la Rhizosphère et d’Environnements Extrêmes), s’intéresse à l’adaptation des bactéries à ces environnements extrêmes, et récemment Deinococcus deserti a été isolée du désert saharien et caractérisée au laboratoire (de Groot et al., 2005). Ma thèse porte sur l’étude de la radiotolérance de cette bactérie par une approche de génomique fonctionnelle. Le but est de mieux comprendre l’extraordinaire résistance aux radiations et à la dessiccation des bactéries du genre Deinococcus en général et de D. deserti en particulier.

17 1) Tolérance aux radiations et à la dessiccation de Deinococcus radiodurans

1.1) Le genre Deinococcus

D. radiodurans est l’espèce type de Deinococcus, un genre connu pour être extrêmement tolérant aux radiations (Cox & Battista, 2005). Ce genre est le seul de la famille Deinococcaceae et il contient plus de 40 espèces isolées d’environnements très variables (Tableau 1). Récemment quelques espèces de Deinococcus moins radiotolérantes que D. radiodurans ont été décrites (Tableau 1). La liste de l’ensemble des espèces de Deinococcus est régulièrement mise à jour à l’adresse suivante : http ://www.bacterio.cict.fr/d/deinococcus.html. La plupart des Deinococcus sont mésophiles (croissance optimale autour de 30°C) alors que 2 espèces, D. geothermalis et D. murrayi, sont des thermophiles modérés (croissance optimale autour de 50°C) (Ferreira et al., 1997) et que quelques-unes comme D. alpinitundrae ou D. saxicola sont psychrophiles (température optimale de croissance autour de 9°C) (Callegan et al., 2008 ; Hirsch et al., 2004). Le génome des Deinococcus présente généralement un GC % élevé, entre 60 et 71 %, sauf pour D. misasensis (54 %) et D. roseus (57 %). Toutes forment des colonies de couleur rose pâle à rouge, en raison de la présence de caroténoïdes. Ce sont des bactéries en forme de coque (comme D. radiodurans) ou de bâtonnet (comme D. deserti) et qui sont positifs (par exemple D. radiodurans) ou négatifs (par exemple D. deserti) pour la coloration de Gram. Enfin, toutes les espèces de Deinococcus ont besoin d’oxygène pour croître et seulement 2 espèces, D. frigens et D. gobiensis, sont décrites comme aérobie et anaérobie facultative (Yuan et al., 2009 ; Hirsch et al., 2004). D’un point de vue phylogénétique, les Deinococcaceae sont proches des Thermaceae et les deux familles appartiennent au phylum Deinococcus/Thermus (Henne et al., 2004). Contrairement aux Deinococcus, les bactéries du genre Thermus sont thermophiles (croissance optimale entre 50-70°C) (Williams et al., 1995) et ne sont pas connues pour être radiotolérantes. Effectivement, il a été montré qu’au moins Thermus aquaticus (Zimmerman & Battista, 2005) et Thermus thermophilus (Omelchenko et al., 2005) sont radiosensibles (Fig. 1). La comparaison des génomes de D. radiodurans et T. thermophilus suggère que lors de la séparation des familles, chacune a acquis des compétences particulières, liées à la thermophilie et/ou à la radiorésistance (Omelchenko et al., 2005 ; Makarova et al., 2007). Récemment, Truepera (Trueperaceae), un autre genre et famille appartenant au phylum Deinococcus/Thermus, a été décrit. L’espèce Truepera radiovictrix est la seule représentante

18 de ce genre à l’heure actuelle (Albuquerque et al., 2005). Sa température optimale de croissance est de 50°C et elle est radiotolérante. Il serait intéressant que le génome de cette bactérie soit séquencé afin de déterminer les gènes susceptibles d’être impliqués dans la radiotolérance des genres Deinococcus et Truepera.

Tableau 1 : Les Deinococcaceae. Espèces Source / Tolérance aux UV, γ, et Références Environnement à la dessiccation Deinococcus aerius Air DR, UVR, γR (Yang et al., 2009a) Deinococcus aetherius Air DR, UVR, γR (Yang et al., 2009b) Deinococcus alpinitundrae Environnement alpin DS, UVR, γS (Callegan et al., 2008) Deinococcus altitudinis Environnement alpin DS, UVR, γS (Callegan et al., 2008) Deinococcus apachensis Désert γR (Rainey et al., 2005) Deinococcus aquaticus Aquatique γS (Im et al., 2008) Deinococcus aquatilis Aquatique UVR (Kampfer et al., 2008) Deinococcus aquiradiocola Aquatique/ « Radioactif » UVR, γR (Asker et al., 2009) Deinococcus caeni Boues activées γS (Im et al., 2008) Deinococcus cellulosilyticus Air UVR (Weon et al., 2007) Deinococcus claudionis Environnement alpin DS, UVS, γS (Callegan et al., 2008) Deinococcus deserti Désert DR, UVR, γR (de Groot et al., 2005) Deinococcus ficus * Rhizosphère UVR (Lai et al., 2006) Deinococcus frigens Sol aride de l’Antarctique UVR (Hirsch et al., 2004) Deinococcus geothermalis Sources thermales DR, UVR, γR (Ferreira et al., 1997) (Makarova et al., 2007) (Daly et al., 2004) Deinococcus gobiensis Désert UVR, γR (Yuan et al., 2009) Deinococcus grandis Intestin de Cyprinus carpio UVR, γR (Rainey et al., 1997, Oyaizu, 1987) Deinococcus hohokamensis Désert γR (Rainey et al., 2005) Deinococcus hopiensis Désert γR (Rainey et al., 2005) Deinococcus indicus Eau d’un sol aquifère contaminé UVR (Suresh et al., 2004) Deinococcus maricopensis Désert γR (Rainey et al., 2005) Deinococcus marmoris Sol de l’Antarctique γR (Hirsch et al., 2004) Deinococcus misasensis Aquatique/ « Radioactif » UVR, γR (Asker et al., 2008) Deinococcus mumbaiensis * Gélose contaminée UVR (Shashidhar & Bandekar, 2006) Deinococcus murrayi Sources thermales γR (Ferreira et al., 1997) Deinococcus navajonensis Désert γR (Rainey et al., 2005) Deinococcus papagonensis Désert γR (Rainey et al., 2005) Deinococcus peraridilitoris Désert γR (Rainey et al., 2007) Deinococcus piscis Poisson γR (Shashidhar & Bandekar, 2009) Deinococcus pimensis Désert γR (Rainey et al., 2005) Deinococcus proteolyticus Fèces de Lama glama γR (Brooks, 1981) Deinococcus radiodurans Boîte de corned-beef irradiée DR, UVR, γR (Anderson, 1956) Deinococcus radiomollis Environnement alpin DS, UVS, γS (Callegan et al., 2008) Deinococcus radiophilus Canard irradié γR (Brooks, 1981) Deinococcus radiopugnans Haddock irradié γR (Brooks, 1981) Deinococcus roseus Aquatique/ « Radioactif » UVR, γR (Asker et al., 2008) Deinococcus saxicola Sol aride de l’Antarctique UVR (Hirsch et al., 2004) Deinococcus sonorensis Désert γR (Rainey et al., 2005) Deinococcus wulumuqiensis Sol pollué/radioactif UVR, γR (Wang et al., 2009) Deinococcus yavapaiensis Désert γR (Rainey et al., 2005) Deinococcus xibeiensis Sol pollué/radioactif UVR, γR (Wang et al., 2009) Deinococcus xinjiangensis Désert DS, UVR, γS (Peng et al., 2009) Deinococcus yunweiensis Gélose contaminée UVR, γR (Zhang et al., 2007b)

*Les espèces Deinococcus ficus et Deinococcus mumbaiensis ne forment en fait qu’une seule et même espèce, Deinococcus ficus (Kampfer, 2009). Les indications de tolérance aux UV, γ et à la dessiccation sont relatives aux tolérances de D. radiodurans. R, résistante ; S, sensible.

19 A B

D. radiodurans

D. radiodurans E. coli T. thermophilus E. coli T. thermophilus

C

Figure 1 : Tolérance aux radiations et à la dessiccation de D. radiodurans, T. thermophilus et E. coli. (A) Survie D. radiodurans aux rayons gamma. (B) Survie à la dessiccation. (C) Survie aux UV. Adapté de Omelchenko et al., (2005) et de Battista, (1997). E. coli

Lorsque le travail de cette thèse a commencé, D. radiodurans était la seule espèce parmi les Deinococcus a être étudiée extensivement. La Figure 1 montre qu’elle est beaucoup plus résistante aux irradiations UV, gamma et à la dessiccation que T. thermophilus ou qu’E. coli. Son extraordinaire capacité à résister aux rayons UV et gamma n’est pas dû au fait qu’elle est capable de protéger son ADN des dommages. En effet, une dose d’UV de 500 J/m2 génère chez D. radiodurans des milliers de dimères de pyrimidine par génome (Pogoda de la Vega et al., 2005). Un exemple de ces dimères est donné en Figure 2.

Figure 2 : Formation de dimère de thymine (T< >T) après irradiation UV. D’après Friedberg et al. (2006).

L’irradiation par de fortes doses de rayons gamma entraîne également de nombreuses lésions de l’ADN, incluant des cassures double brin qui sont particulièrement difficiles à réparer. Les radiations, en plus de leurs effets directs, engendrent également beaucoup de radicaux oxygénés par la radiolyse de l’eau, qui vont entraîner l’oxydation des protéines, des

20 lipides et de l’ADN. Chez la plupart des organismes, tous ces dommages entraînent la mort des cellules or ce n’est pas le cas chez D. radiodurans et autres Deinococcaceae. Après exposition à une forte dose d’irradiation gamma, D. radiodurans est capable de reconstituer son génome à partir d’ADN cassé en centaines de fragments (Fig. 3) (Blasius et al., 2008).

Figure 3 : Pulsed field gel electrophoresis (PFGE) montrant la reconstitution du génome de D. radiodurans après irradiation gamma. Les bactéries ont été irradiées à 6,8 kGy puis incubées dans le milieu de culture et des extraits ont été prélevés à différents temps (de 0 à 4 h). L’ADN génomique a été digéré par NotI, ce qui donne une dizaine de fragments de grandes tailles pour les bactéries non irradiées (pre-irradiation). Ceux- ci disparaissent après irradiation, et sont remplacés par une trainée, due aux nombreuses cassures de l’ADN, causée par l’irradiation gamma. Le génome est reconstitué au bout de 2- 3 h. D’après Blasius et al. (2008).

Étant donné que sur Terre, il n’existe aucun écosystème où de fortes doses de rayonnement gamma sont produites, comment expliquer l’extrême tolérance de D. radiodurans à ce type de radiation ? Une publication émet l’hypothèse étonnante selon laquelle D. radiodurans serait résistante aux rayons gamma parce qu’elle ou son ancêtre) aurait subit plusieurs allers-retours entre la Terre et la planète Mars (les astéroïdes jouant le rôle de vecteur de « transport ») (Pavlov A.K. ; 2006). Une explication plus plausible est que cette extrême tolérance est liée à des conditions environnementales naturelles qui induisent d’importants dommages de l’ADN dont des cassures double brin. Il a été montré que la dessiccation provoque, comme les rayons gamma, des cassures double brin chez D. radiodurans (Mattimore & Battista, 1996). De plus, plusieurs mutants radiosensibles de D. radiodurans sont également sensibles à la dessiccation (Mattimore & Battista, 1996). Une étude a montré que des Deinococcus radiotolérants sont retrouvés plus fréquemment dans un sol naturellement aride plutôt que dans un sol naturellement non aride (Rainey et al., 2005). D’ailleurs, comme le montre le tableau 1, beaucoup de Deinococcus décrits ont été trouvés dans des environnements secs (désert, sol aride de l’Antarctique). Cependant, toutes les bactéries résistantes à la dessiccation ne sont pas forcément radiotolérantes. En effet, des bactéries du genre Azotobacter et Ramlibacter par exemple sont très résistantes à la dessiccation (Socolofsky & Wyss, 1962 ; Heulin et al., 2003), mais ne sont pas radiotolérantes. Cette résistance serait due à la formation d’un polymère extracellulaire capable de limiter la perte d’eau (Billi & Potts, 2002 ; Potts, 1994).

21 Plusieurs études sur D. radiodurans ont été effectuées dont le séquençage et l’analyse de son génome. Les résultats de ces analyses et quelques hypothèses pouvant expliquer la radiotolérance de cette bactérie sont décrits dans les chapitres suivants.

22 1.2) Le génome de D. radiodurans

1.2.1) Quelques caractéristiques

Le génome de D. radiodurans a été séquencé en 1999 (White et al., 1999) et durant ma thèse, celui de D. geothermalis a été publié (Makarova et al., 2007). Le génome de D. radiodurans se compose de 2 (2,65 Mb et 412 kb), d’un mégaplasmide (177 kb) et d’un plasmide (46 kb), et comporte environ 3200 gènes. Celui de D. geothermalis comprend un (2,46 Mb) et 2 mégaplasmides (574 kb et 206 kb) et il comporte un peu moins de 3100 gènes. L’étude des génomes de D. radiodurans et de D. geothermalis montre qu’elles partagent un grand nombre de gènes. Il ne sera donc décrit que les gènes de D. radiodurans sauf dans les cas où il y a des différences avec D. geothermalis. D. radiodurans possède un arsenal de gènes de ménage similaire mais moins diversifié que celui d’E. coli. Les voies de la glycolyse, du cycle de Krebbs, des pentoses phosphates et du glyoxylate sont présentes et semblent fonctionnelles (Makarova et al., 2001). Toutefois certaines du métabolisme sont absentes (enzymes clés de la synthèse de la lysine, cystéine, méthionine, certaines enzymes produisant différents co-facteurs tel que le NAD, la phosphatidylglycerophosphate synthase impliquée dans la synthèse de phospholipides acides). D. radiodurans n’a pourtant pas besoin de certains des produits finaux de ces réactions métaboliques (lysine, co-facteurs) pour pousser, donc il est probable qu’elle possède d’autres enzymes qui les remplacent (Makarova et al., 2001). En ce qui concerne la réponse aux stress, D. radiodurans présente des protéines conservées chez les bactéries, comme par exemple, des chaperonnes, des catalases et superoxydes dismutases et des protéines de choc thermique (Makarova et al., 2001). Cependant, une des 3 catalases de D. radiodurans, CatA, codée par DR_A0146, est plus proche des catalases de plantes que de celles des bactéries. De plus, DR_A0146 est en opéron avec un gène codant une peroxydase (DR_A0145) proche de celles qui sont retrouvées chez les champignons. Ainsi cet opéron a probablement été acquit par transfert horizontal (Makarova et al., 2001). Cet opéron n’est pas présent chez D. geothermalis. D. radiodurans possède également 4 gènes potentiellement impliqués dans la résistance à la dessiccation. DR_B0118 code un homologue de PCC13-62, une protéine impliquée dans la résistance à la dessiccation de Craterostigma plantagineum et exprimée durant la déshydratation de cette plante (Piatkowski et al., 1990). DR_1372, DR_0105 et DR_1172 codent respectivement des protéines de la famille LEA14, LEA76 et LEA76/LEA29 (LEA = late embryogenesis abundant). Ces types de protéines sont impliqués dans la résistance à la dessiccation des graines par exemple (Caramelo & Iusem, 2009). Deux de ces gènes, DR_1172 et DR_B0118,

23 ont été montrés comme impliqués dans la résistance à la dessiccation mais pas dans la résistance aux rayons gamma (Battista et al., 2001). L’irradiation ne produit pas exactement les mêmes effets que la dessiccation et ces types de protéines sont probablement impliqués dans la protection/réparation de dommages spécifiques à la dessiccation. À noter toutefois que ces gènes ne semblent pas induits après irradiation gamma ou après dessiccation (Tanaka et al., 2004 ; Liu et al., 2003), ce qui suggère qu’une expression constitutive permet une réponse ou une protection immédiate à la dessiccation. Une des particularités de D. radiodurans est le fait qu’elle présente 26 gènes codant pour des protéines ayant un domaine Nudix (Nudix = nucleoside diphosphates linked to some other moiety, x) alors qu’E. coli n’en possède que 13 (McLennan, 2006 ; Makarova et al., 2000). Ces Nudix sont parfois impliquées dans la dégradation des nucléotides oxydés. Par exemple, MutT chez E. coli hydrolyse le 8-oxodGTP, ce qui empêche l’incorporation de cette base oxydée et limite ainsi les mutations spontanées (transversion AT→CG) lors de la réplication de l’ADN (Fowler & Schaaper, 1997). Cependant, une activité similaire à celle de MutT n’a pas été montrée pour les protéines Nudix de D. radiodurans (Xu et al., 2001). Certaines hydrolases Nudix peuvent enlever des substances potentiellement toxiques et d’autres peuvent moduler les quantités de composés du métabolisme et de la signalisation (Xu et al., 2001). Ces activités peuvent être importantes après un stress comme l’irradiation. Cependant, la radiotolérance d’un organisme ne dépend pas du nombre de gènes codant pour une protéine à domaine Nudix puisque d’autres bactéries radiorésistantes comme Kineococcus radiotolerans et Rubrobacter xylophilus n’en possède respectivement que 14 et 8 (McLennan, 2006). D. radiodurans possède tous les gènes typiques pour les machineries de traduction et de réplication. Elle possède également un arsenal de gènes de réparation de l’ADN, en général très similaire à celui des bactéries radiosensibles comme E. coli. Cependant, avec une radiotolérance si forte, il est étonnant que quelques enzymes ne soient pas présentes chez D. radiodurans. C’est le cas notamment de RecB et RecC, qui sont impliquées dans la réparation par recombinaison chez E. coli (Cox, 2007b). Sont également manquants, des gènes codant une photolyase et des polymérases translésionnelles (ADN polymérases capables de passer des lésions, mais qui sont peu fidèles et qui peuvent donc générer des mutations) (Makarova et al., 2001 ; Makarova et al., 2007). Comme D. radiodurans peut subir des dommages très importants, il a été proposé qu’il serait trop dangereux pour elle d’avoir des polymérases translésionnelles (Sale, 2007). De plus, un homologue de Ku, une protéine appartenant au système de non-homologous end joining (NHEJ), n’a pas été retrouvé chez D. radiodurans. Ce mécanisme de réparation est peu fréquent chez les bactéries, mais

24 existe par exemple chez Bacillus subtilis (Moeller et al., 2007 ; Weller et al., 2002). Cependant, D. radiodurans possède quelques enzymes rares chez les bactéries, telles qu’une ADN polymérase de la famille X et la topoisomérase 1B (Makarova et al., 2001). Avec un set de gènes de réparation similaire, comment D. radiodurans peut tolérer de très fortes doses de radiations alors qu’E. coli ne peut pas ? Il est possible que la réparation par ces enzymes conservées soit plus efficace chez D. radiodurans (grâce à la propriété des enzymes et/ou à des conditions intracellulaires particulières) et/ou qu’elle possède de nouvelles enzymes de réparation de l’ADN. Des gènes supplémentaires de réparation peuvent être présents parmi les centaines de gènes de fonction inconnue de D. radiodurans, dont environ 150 sont retrouvés exclusivement chez D. geothermalis (Makarova et al., 2007). D. radiodurans possèdent des gènes homologues codant pour les enzymes impliquées dans le système NER (nucleotide excision repair), BER (base excison repair), MMR (mismatch repair), RER (recombination repair) ainsi que la réversion de certaines bases modifiées. Dans la partie suivante, ces différents systèmes et les gènes présents chez D. radiodurans seront très succinctement décrits.

1.2.2) Les gènes de la réparation de l’ADN

La réversion de bases endommagées Une manière « simple » et peu coûteuse de réparer les dommages à l’ADN est de faire une réversion de la lésion. Certaines enzymes comme les photolyases (retrouvées chez E. coli par exemple) ou les spore photoproduct- (SP-lyases ; retrouvées par exemple chez B. subtilis) sont spécialisées dans la réversion des dommages liés aux UV (Goosen & Moolenaar, 2008). Les photolyases utilisent l’énergie lumineuse pour faire leur réaction alors que les SP-lyases utilisent le S-Adénosyl-Méthionine (Goosen & Moolenaar, 2008 ; Sancar, 2008 ; Rebeil et al., 1998). Au vu de la résistance des Deinococcus aux UV, il est étonnant de ne pas trouver de gènes homologues codant une photolyase ou une SP- chez D. radiodurans et D. geothermalis. Il est également surprenant que D. radiodurans ne possède pas d’homologues à Ada et AlkB, des protéines nécessaires à la réparation des bases alkylées chez E. coli (Nieminuszczy & Grzesiuk, 2007). Toutefois D. radiodurans possède des homologues pour les protéines Ogt et YggV qui sont impliquées respectivement dans la réparation de guanines alkylées et dans la prévention d’incorporation d’hypoxanthine/xanthine dans l’ADN chez E. coli (Bradshaw & Kuzminov, 2003 ; Margison et al., 2007).

25 Ainsi, contrairement à E. coli, il semblerait que D. radiodurans n’ait pas beaucoup de systèmes de réparation directe des bases modifiées, suggérant ainsi l’existence d’autres systèmes de réparation très efficaces.

Le système NER Le système NER est capable de reconnaître une déformation au niveau de l’hélice d’ADN et d’induire sa réparation (Friedberg, 2006). Ce système comprend notamment les enzymes UvrAB (reconnaissance de la lésion), UvrC (clive en 5’ et 3’ de la lésion) et UvrD (hélicase) (Friedberg, 2006). En plus de ces protéines, D. radiodurans possède le gène uvsE, qui code une endonucléase β, appartenant à la voie de réparation alternative UVDE (UV- damage ). Cette voie est plutôt rencontrée chez les Eucaryotes (Yasui & McCready, 1998). Ces différentes enzymes sont impliquées dans la réparation des dommages induits par les UV chez D. radiodurans (Minton, 1994 ; Moseley & Evans, 1983 ; Earl et al., 2002b ; Kitayama et al., 2003 ; Tanaka et al., 2005). Contrairement au mutant uvrA d’E. coli, celui de D. radiodurans n’est que peu affecté dans la survie aux UV. Toutefois, ce mutant est sensible à la mitomycine C, un agent endommageant l’ADN (Moseley & Copland, 1978 ; Earl et al., 2002b ; Tanaka et al., 2005). La délétion du gène uvsE ne réduit pas non plus beaucoup sa capacité à résister aux UV alors que la suppression des 2 gènes uvrA et uvsE la diminue très fortement (Tanaka et al., 2005 ; Earl et al., 2002b). Ces résultats montrent que ces 2 voies travaillent parallèlement afin de réparer les dommages liés aux UV (Tanaka et al., 2005). Le gène uvrA d’E. coli est capable de complémenter la sensibilité à la mitomycine C du mutant uvrA de D. radiodurans, ce qui prouve que le système UvrABC de D. radiodurans est comparable à celui d’E. coli (Agostini et al., 1996). À noter toutefois que D. radiodurans possède 2 gènes uvrA, uvrA1 (DR_1771) et uvrA2 (DR_A0188), alors que D. geothermalis n’en a qu’un (Dgeo_0694, l’homologue d’uvrA1). Cependant, le gène uvrA2 ne semble pas avoir un rôle important dans le système NER (Tanaka et al., 2005).

Le système BER Le système BER permet de retirer et de réparer des bases modifiées grâce à l’action combinée d’ADN glycosylases et d’endonucléases AP (apurinic/apyrimidinic) (Friedberg, 2006 ; Lindahl, 1976). Certaines glycosylases, les glycosylases bifonctionnelles, possèdent également l’activité endonucléase AP alors que les glycosylases monofonctionnelles ne possèdent que la fonction glycosylase. D. radiodurans possède plusieurs glycosylases (mono ou bi-fonctionnelles) et endonucléases AP. Au moins une de ces glycosylases, MUG (mismatch-specific uracil-DNA glycosylase), présente une plus large spécificité de

26 reconnaissance de substrat que son homologue chez E. coli, ce qui peut contribuer à la radiotolérance (Moe et al., 2006). En plus de ces enzymes, D. radiodurans possède une ADN polymérase X, PolX (Makarova et al., 2001). Elle est peu commune chez les Procaryotes et a probablement été acquise par transfert de gène horizontal. PolX est similaire à l’ADN polymérase β des Eucaryotes, qui intervient dans le système BER. Un travail récent suggère que PolX de D. radiodurans est également impliqué dans le BER (Khairnar & Misra, 2009). De plus, cette protéine a un rôle dans la réparation des cassures double brin chez D. radiodurans, peut-être via le NHEJ (voir chapitre 1.1.5.2) (Lecointe et al., 2004b ; Blasius et al., 2006). B. subtilis possède également une ADN polymérase X qui serait impliquée dans le BER et le NHEJ (Banos et al., 2008a ; Banos et al., 2008b).

Le mismatch repair Le système de MMR est capable de repérer les mésappariements de bases arrivant lors de la réplication de l’ADN, lors de la recombinaison homologue s’il y a des différences dans les séquences homologues, ou lorsque le 5-méthyl-cytosine se désamine en thymine (Friedberg, 2006). Les 4 principales protéines constituant ce système chez E. coli sont MutS, qui reconnaît le mésappariement (Parker & Marinus, 1992), MutL, qui permet l’assemblage du complexe MutSLH (Galio et al., 1999 ; Robertson et al., 2006), MutH, une endonucléase capable de reconnaître le brin parental (Langle-Rouault et al., 1987 ; Welsh et al., 1987) et UvrD, une hélicase qui intervient aussi dans le NER (Brosh & Matson, 1997). Chez E. coli, ce système de réparation ne fonctionne qu’en présence de séquences GATC méthylées. Cette méthylation, dépendante de la méthylase Dam, permet de discriminer le brin parental et le brin nouvellement synthétiser ou étranger (Lahue et al., 1987 ; Langle-Rouault et al., 1987). D. radiodurans possède un système MMR quelque peu différent de celui d’E. coli. Elle n’a pas d’homologue de mutH, un gène peu conservé chez les bactéries, et elle présente 2 gènes homologues à mutS, mutS1 et mutS2 (Makarova et al., 2001 ; Mennecier et al., 2004). Le MMR de D. radiodurans ne semble pas être essentiel pour la réparation des dommages à l’ADN dû aux UV ou aux rayons gamma puisque les mutants mutS1 et mutL sont aussi radiotolérants que la souche sauvage (Mennecier et al., 2004). Son rôle serait le même que celui d’E. coli, à savoir, repérer et réparer les mésappariements lors de la réplication et cette fonction implique les gènes mutL, mutS1 et uvrD mais pas mutS2 (Mennecier et al., 2004 ; Mennecier et al., 2006). Contrairement à E. coli et à quelques autres bactéries, D. radiodurans n’a pas d’homologue de la protéine Dam. Il est possible que le marquage du brin parental soit réalisé par d’autres méthylases.

27 Le système RER La recombinaison homologue, extensivement étudiée chez E. coli, implique l’échange d’ADN simple brin entre 2 molécules d’ADN, l’une simple brin et l’autre double brin,

Figure 4 : Formation du Nucléation (lent) nucléofilament RecA. La formation de ce filament se fait en 2 phases, la nucléation et l’extension. Extension (rapide) L’extension du nucléofilament se fait sur un ADN simple brin de 5’ vers 3’ et peut se propager jusqu’à un ADN double brin. Le Désassemblage désassemblage du filament utilise de l’ATP. Adapté de Cox, (2007b).

comportant des régions identiques (homologues) ou quasiment identiques. RecA (recombinase A) est la protéine qui permet aux cellules de réaliser la recombinaison homologue (Cox, 2007a). RecA d’E. coli est capable de fixer l’ADN simple brin lorsqu’il y en a dans la cellule, par exemple après la maturation d’une cassure double brin ou lors de l’arrêt de la fourche de réplication. Une fois un monomère de RecA fixé à l’ADN, un nucléofilament composé de plusieurs molécules de RecA sur l’ADN va se former (Fig. 4) (Cox, 2007a). Le filament de RecA recherche une séquence complémentaire au simple brin d’ADN où elle est fixée. Cette recherche va impliquer un ADN double brin et va donc conduire à la formation d’un ADN triplex (3 brins d’ADN, aussi nommé D-loop) (Cox, 2007a ; McGrew & Knight, 2003). Il y a ensuite formation d’une jonction de Holliday (structure en croix qui se

Figure 5 : Modèle de la réparation de cassures double brin par la voie RecBCD. L’extrémité double brin est reconnue par le complexe RecBCD (1). Ce complexe possède une activité hélicase bipolaire couplée à une activité exonucléase bipolaire. Ceci entraîne la dégradation des deux brins (2-3). La rencontre avec un site Chi permet d’atténuer l’activité nucléase 3’-5’, ce qui conduit à la formation d’un ADN 3’ simple brin sur lequel RecBCD charge RecA (4). Ensuite, ce brin d’ADN recouvert de RecA envahit une région d’ADN double brin qui lui est homologue. A ce moment-là, il y a formation de la D loop puis de la jonction de Holliday. Adapté de Kowalczykowski (2000).

28 forme lors de l'enjambement du brin d’ADN), puis le complexe RuvABC catalyse la migration des branches et la résolution de ces jonctions (Cox, 2007a). La formation du nucléofilament et l’activité de RecA sont modulées par un grand nombre de protéines, qui sont entre autres, SSB (single strand binding protein) qui rentre en compétition avec RecA pour la fixation de l’ADN simple brin et dont le rôle serait d’une part d’éviter la formation de structures secondaires au niveau de l’ADN simple brin et d’autre part de recruter des protéines du métabolisme de l’ADN (Kowalczykowski et al., 1987 ; Cox, 2007b) ; RecBCD (Lucarelli et al., 2009 ; Cox, 2007b) et RecF (comprenant les protéines RecQJFOR) (Handa et al., 2009 ; Sakai & Cox, 2009), respectivement des protéines de la première et de la deuxième voie de chargement de RecA sur l’ADN simple brin chez E. coli (Fig. 5 et 6) ; l’hélicase UvrD, capable de désassembler le filament RecA (Veaute et al., 2005) ; DinI et RecX qui respectivement stabilise et déstabilise le filament RecA (Renzette et al., 2007 ; Cox, 2007b). Outre sa fonction dans la recombinaison homologue, le filament de RecA est capable de stimuler le clivage du répresseur LexA et de quelques autres protéines (cette fonction est appelée l’activité co-protéase de RecA). Chez E. coli et plusieurs autres bactéries, le clivage de LexA entraîne l’expression de la réponse SOS, c’est-à-dire l’induction de plusieurs dizaines de gènes impliqués dans la réparation de l’ADN (Schlacher & Goodman, 2007 ; Butala et al., 2009).

Figure 6 : Modèle de la réparation de cassures double brin par la voie RecF. RecJ, une exonucléase, dégrade un des 2 brins de l’ADN de 5’ vers 3’. Son activité est stimulée par RecQ et SSB. SSB se fixe alors sur l’ADN simple brin en formation. A gauche : Les complexes RecFOR se fixent sur l’ADN et ceux étant proches de la jonction simple/double brin chargent RecA sur l’ADN simple brin. A droite : les complexes RecOR fixés à SSB vont échanger SSB contre RecA lorsqu’ils sont proches de la jonction simple/double brin. Ensuite, dans les deux cas, le brin d’ADN recouvert de RecA envahit une région d’ADN double brin qui lui est homologue. A ce moment-là, il y a formation de la D loop puis de la jonction de Holliday. D’après Handa et al. (2009).

29 D. radiodurans est capable de réparer les cassures double brin et pour cela il semble naturel que la recombinaison homologue joue un grand rôle. À l’exception de RecB, RecC et DinI, les homologues des protéines du RER décrites ci-dessus sont trouvés chez

WT recA Figure 7 : PFGE montrant la reconstitution du génome de D. radiodurans après irradiation gamma chez une souche sauvage et un mutant recA. Les bactéries ont été irradiées à 7 kGy puis incubées dans le milieu de culture et des extraits ont été prélevés à différents temps (de 0 à 24 h). L’ADN génomique a été digéré par NotI, ce qui donne un profil de digestion présentant de grands fragments pour les bactéries non irradiées (C). La souche sauvage reconstitue son génome en 3h alors que le mutant recA ne parvient pas à le reconstituer même après 24h. C = contrôle avant irradiation, S = marqueur de taille. D’après Zahradka et al. (2006).

D. radiodurans (Makarova et al., 2001). Comme chez E. coli, un mutant recA de D. radiodurans est hypersensible aux irradiations (Gutman et al., 1994) et est incapable de reconstituer son génome (Fig. 7) (Daly et al., 1994). Une réparation partielle est visible dans le mutant recA, mais elle n’est pas suffisante pour reconstruire un chromosome entier. Cette réparation indépendante de RecA serait due au single strand annealing (SSA ; Fig. 8), où des fragments ayant des extrémités 3’ sortantes s’assemblent (Slade et al., 2009 ; Zahradka et al., 2006 ; Daly & Minton, 1996).

Cassure double brin dans 2 copies de l’ADN

Maturation des extrémités double brin formant des extrémités 3’ sortants simples brins

Appariement d’une région d’ADN complémentaire

Des nucléases coupent les fragments non appariés Figure 8: Le single strand annealing. Lorsqu’une ou des cassure(s) double brin sont générées, deux fragments d’ADN complémentaires vont pouvoir s’apparier et permettre la réparation de la cassure double brin. Adapté de Zahradka et al. 2006.

Il est important de noter que le gène recA de Shigella flexnerii (gène identique à celui d’E. coli) ne permet pas de restaurer la résistance aux rayons gamma du mutant recA de D. radiodurans (Carroll et al., 1996). Plus récemment, il a été montré que recA d’E. coli pouvait restaurer partiellement la résistance aux UV d’un mutant recA de D. radiodurans

30 (Schlesinger, 2007). recA de D. radiodurans complémente entièrement la sensibilité aux rayons gamma d’un mutant recA d’E. coli (Narumi et al., 1999). L’ensemble suggère que RecA de D. radiodurans possède des propriétés partiellement différentes de celles de RecA d’E. coli ou qu’elle est capable d’interagir avec des protéines, spécifiques à D. radiodurans, impliquées dans la radiotolérance. D’ailleurs, une des propriétés de RecA de D. radiodurans est sa capacité à se fixer préférentiellement sur de l’ADN double brin in vitro (Kim & Cox, 2002 ; Kim et al., 2002) alors que RecA d’E. coli fixe préférentiellement l’ADN simple brin. Un résultat surprenant de Satoh et al. (2002) montre que la souche de D. radiodurans comportant l’allèle recA424, codant une protéine (RecA G142L) qui ne possède plus d’activité pour l’échange de brin in vitro en présence d’ATP mais qui conserve toujours une activité co-protéase (aide au clivage de LexA), est à peine plus sensible aux rayons gamma que la souche sauvage. En comparaison, une mutation rendant les fonctions de co-protéase et

Figure 9: Sensibilité aux rayons gamma de D. radiodurans et de souches mutantes. Souche sauvage de D. radiodurans ({), recA424 ( ), recA424 complémenté („), souche rec30 (recA670) (Δ), souche rec30 complémenté (▲). D’après Satoh et al. (2002).

d’échange de brin nulles (recA670), donne une souche hypersensible aux rayons gamma (Fig. 9) (Satoh et al., 2002). Les auteurs proposent que c’est l’activité co-protéase qui est importante pour la radiorésistance et non pas l’activité d’échange de brin. Cependant, il est bon de noter que la protéine RecA G142L présente une activité résiduelle d’échange de brin dans d’autres conditions in vitro (en présence de dATP). Il est possible qu’une activité résiduelle soit également présente dans des conditions in vivo, ce qui pourrait expliquer le phénotype de résistance partielle aux rayons gamma. Contrairement à E. coli qui possède 2 voies de chargement, RecBCD et RecF, D. radiodurans ne posséderait que la voie RecF car les homologues de RecBC sont absents. D’autres bactéries comme celles du genre Bacillus ne possèdent pas non plus d’homologues de RecBC, mais contrairement à D. radiodurans, elles possèdent les analogues AddAB (Rocha et al., 2005). Chez E. coli, le mutant recB/recC est très sensible aux irradiations sauf lorsque SbcB (exonucléase 3’ vers 5’) est rendue non fonctionnelle. En fait cela laisse la possibilité à la voie RecF de charger RecA sur l’extrémité 3’ d’un ADN simple brin chez

31 E. coli (Clark, 1991). Ces données sont en accord avec le fait que D. radiodurans ne possède pas non plus d’homologue de SbcB. De plus, l’expression de sbcB (Misra et al., 2006) ou de recB et recC d’E. coli (Khairnar et al., 2008) chez D. radiodurans la rend plus sensible à l’irradiation, suggérant un rôle important de la voie RecF dans la réparation de l’ADN chez cet organisme. En plus, un mutant recO de D. radiodurans a un phénotype similaire à celui d’un mutant recA, c’est-à-dire une très grande sensibilité aux irradiations UV et gamma ainsi qu’une croissance très ralentie (Xu et al., 2008). recO d’E. coli est capable de complémenter entièrement la sensibilité aux UV du mutant recO de D. radiodurans mais ne complémente que partiellement sa sensibilité aux rayons gamma (Xu et al., 2008). Ceci suggère que RecO de D. radiodurans a des propriétés différentes de celle d’E. coli. L’ensemble des résultats suggère que D. radiodurans ne possède qu’une seule voie de chargement de RecA, la voie RecF. Chez E. coli, la voie RecF fait appel à plusieurs protéines afin de permettre le chargement de RecA sur l’ADN. Une de ces protéines est l’hélicase RecQ (Fig. 6). D. radiodurans possède 2 gènes, DR_1289 et DR_2444, codant pour des homologues de RecQ. Cependant, DR_2444, n’est pas réellement une protéine RecQ car elle ne présente pas de domaine hélicase (Huang et al., 2006b). DR_1289 et DR_2444 comportent, cependant, respectivement 3 et 1 domaine HRDC (Helicase RNAse D C-terminal), un domaine normalement propre aux hélicases RecQ (Fig. 10) (Huang et al., 2006a). Il est très rare de voir une hélicase avec 3 domaines HRDC ainsi qu’un domaine HRDC sur une protéine autre qu’une hélicase ou une nucléase. Contrairement à la délétion du gène DR_2444, la délétion de DR_1289 entraîne une forte sensibilité au péroxyde d’hydrogène, aux rayons gamma, aux UV

Figure 10: Représentation de quelques protéines RecQ. Les domaines Helicases, RQC (RecQ C-terminal ; domaine de fixation à l’ADN, propre aux protéines de la famille RecQ) et HRDC sont très conservés dans les protéines de la famille RecQ. Adapté de Huang et al. (2006). et à la mitomycine C (Hua et al., 2008 ; Huang et al., 2007 ; Huang et al., 2006b). L’effet de la délétion des domaines HRDC sur l’activité de DR_1289 a été analysé et montre que les 3 domaines sont importants pour son fonctionnement (Huang et al., 2007 ; Killoran & Keck, 2006) et qu’ils modulent son activité (Killoran & Keck, 2006). Le troisième domaine HRDC a été cristallisé récemment et la structure conforte le fait que les domaines HRDC modulent l’activité de DR_1289 (Killoran & Keck, 2008). À noter que l’homologue de DR_1289 chez

32 D. geothermalis code pour une protéine de seulement 195 résidus (Makarova et al., 2007). Peut-être est-ce dû à une erreur de séquençage ou alors il est possible que D. geothermalis possède un analogue fonctionnel à RecQ. Bien que D. radiodurans ne possède pas de voie RecBCD, elle présente un homologue de RecD. Cependant, cette protéine apparaît comme différente de celle qui est présente avec RecB et RecC chez d’autres bactéries (Makarova et al., 2001 ; Rocha et al., 2005). RecD de D. radiodurans fait partie de la famille RecD2 et possède une extension N-terminale, comportant un domaine RuvA domain 2–like (Rocha et al., 2005). Cette protéine est une hélicase 5’→3’ peu processive (Wang & Julin, 2004). L’effet de la délétion du gène recD sur la radiotolérance de D. radiodurans n’est pas très clair. En effet, Servinsky et al. (2007) montrent que ce mutant est sensible au péroxyde d’hydrogène, aux rayons gamma et aux UV mais pas à la mitomycine C (Servinsky & Julin, 2007), alors que Zhou et al. (2007) montrent qu’il est très sensible au péroxyde d’hydrogène, peu sensible aux UV mais qu’il est aussi résistant aux radiations gamma que la souche sauvage (Zhou et al., 2007). La raison pour laquelle les 2 équipes n’ont pas les mêmes résultats est inconnue. Ces deux études s’accordent sur le fait que RecD jouerait un rôle dans la résistance aux dommages d’oxydation (Servinsky & Julin, 2007 ; Zhou et al., 2007). RecD stimulerait l’activité de la catalase B (Zhou et al., 2007). Le gène recD d’E. coli ne complémente pas la sensibilité au péroxyde d’hydrogène du mutant recD de D. radiodurans (Zhou et al., 2007), suggérant ainsi que les protéines RecD de ces 2 organismes ont un rôle distinct. La protéine SSB de D. radiodurans, comme celles de D. geothermalis et de Thermus, est particulière car contrairement à celle d’E. coli, elle présente 2 domaines OB (Oligonucleotide/oligosaccharide-Binding) au lieu d’un. Ce domaine très conservé chez les protéines SSB sert à la fixation non spécifique de l’ADN. Alors que les protéines SSB chez les bactéries et les Eucaryotes forment des tétramères, celles de D. radiodurans forment des dimères (Bernstein et al., 2004). Une étude montre que chez D. radiodurans, la quantité de SSB augmente d’un facteur 3 après irradiation (Bernstein et al., 2004), mais une autre étude montre que cette quantité est stable après irradiation (Eggington et al., 2006). Néanmoins, SSB favorise l’échange de brin par RecA lors de la recombinaison homologue (Eggington et al., 2004) et présente un mécanisme commun à celle d’E. coli (Kim et al., 2002 ; Eggington et al., 2004).

33 1.3) Transcriptomique et protéomique de D. radiodurans après irradiation ou dessiccation

L’étude du génome de D. radiodurans n’a pas apporté les réponses qui permettent d’expliquer l’extrême radiotolérance de cet organisme (Makarova et al., 2001). Il est possible qu’une partie de ces réponses se cache dans les gènes spécifiques à Deinococcus. Ces gènes étant de fonction inconnue, il est impossible, sur la base des séquences, de savoir lesquels sont importants pour la radiotolérance. Pour trouver des gènes impliqués dans la radiotolérance, des expériences de protéomique (Hansen, 1980 ; Tanaka et al., 1996 ; Lipton et al., 2002 ; Joshi et al., 2004 ; Zhang et al., 2005 ; Lu et al., 2009) et de transcriptomique (Liu et al., 2003 ; Tanaka et al., 2004) avant et après irradiation gamma ou dessiccation ont été menées.

1.3.1) Protéome après irradiation gamma

Plusieurs études de protéomique ont été faites et la plupart d’entre elles montrent qu’après irradiation UV ou gamma, la quantité totale de protéines diminue, dû très vraisemblablement à une dégradation (Hansen, 1980 ; Tanaka et al., 1996 ; Joshi et al., 2004). Parmi ces protéines dégradées, il y a des protéines essentielles au fonctionnement du cycle de Krebbs (CitA, la citrate synthase ; Acn, l’aconitate hydratase) et des chaperonnes (GroEL, DnaK) (Joshi et al., 2004 ; Lu et al., 2009). Cependant, la quantité de certaines protéines telles que le facteur d’élongation EF-Tu, une protéine nécessaire à la synthèse protéique, et une sérine protéase, reste inchangée après irradiation (Joshi et al., 2004). Enfin, ces études ont permis de déterminer que certaines protéines étaient induites après irradiation (Hansen, 1980 ; Tanaka et al., 1996 ; Lipton et al., 2002 ; Joshi et al., 2004 ; Lu et al., 2009), notamment RecA, SSB et PprA. PprA est une nouvelle protéine impliquée dans la radiotolérance (voir chapitre 1.3.2 et 1.3.3) (Lipton et al., 2002 ; Zhang et al., 2005). Une fois que la réparation de l’ADN a eu lieu, les protéines dégradées sont de nouveau synthétisées (Joshi et al., 2004).

1.3.2) Transcriptome après irradiation gamma/dessiccation

La protéomique n’a permis de déterminer que quelques protéines potentiellement impliquées dans la radiotolérance. Ceci est en parti dû au fait que l’analyse par gel 2D ne permet de voir qu’un nombre restreint de protéines. Ainsi, des expériences de

34 transcriptomique ont été réalisées et ont permis d’identifier de nouveaux gènes potentiellement impliqués dans la radiotolérance (Liu et al., 2003 ; Tanaka et al., 2004). La première étude de transcriptomique réalisée après irradiation gamma (à 15 kGy) montre que, durant les 9 premières heures après irradiation, 832 et 451 gènes sont respectivement surexprimés et réprimés d’un facteur supérieur ou égal à 2. D’une manière générale, les plus fortes réponses ont lieu aux alentours de 3h après irradiation (Liu et al., 2003). La seconde étude montre que, dans la première heure qui suit l’irradiation gamma (3 kGy) ou la dessiccation (2 semaines), 72 et 73 gènes sont respectivement surexprimés d’un facteur supérieur ou égal à 3. Trente deux de ces gènes sont communs à la réponse à la dessiccation et à l’irradiation gamma (Tanaka et al., 2004), ce qui montre le chevauchement dans la réponse à ces deux stress. En plus des différences dans les doses, les points après irradiation et les facteurs d’induction, une raison pour lesquelles ces 2 études ne conduisent pas au même nombre de gènes induits est due au fait que Liu et al. ont irradié des cellules en début de phase stationnaire de croissance et qu’ils les ont remises dans un milieu frais pour que les bactéries puissent se régénérer. Ensuite, ils ont comparé les ARNs de cette culture avec ceux d’une culture non irradiée en phase stationnaire de croissance. Ainsi, ils détectent la réponse à l’irradiation mais aussi la réponse induite par la dilution des cellules dans un milieu frais. Il est important de souligner que le nombre de gènes induits est surévalué car chez les Procaryotes les gènes sont souvent sous forme d’opéron. Ainsi, un opéron induit après irradiation/dessiccation va donner plusieurs gènes induits même si un seul de ces gènes est réellement lié à la tolérance à ces types de stress. Dans l’étude menée par Liu et al. (2003), un tiers des gènes connus pour être impliqués dans le métabolisme de l’ADN sont induits après irradiation (Liu et al., 2003). En effet, recA, uvrA, uvrB, uvrC, ssb, gyrA ou gyrB (les gyrases sont impliquées dans le surenroulement négatif de l’hélice de l’ADN) sont induits mais recF, recO et recR, mutS ou mutL ne le sont pas par exemple. Ce nombre est encore plus restreint dans l’étude de Tanaka et al. (2004) puisque seulement 6 de ces gènes, recA, ruvB, uvrA, uvrB, gyrA et gyrB, sont fortement induits (Tanaka et al., 2004). Ceci suggère que chez D. radiodurans, le taux basal d’expression, qui peut déjà être presque maximal, de beaucoup des gènes du métabolisme de l’ADN est suffisant pour contribuer à la forte radiotolérance. Un des points important révélé par la transcriptomique est que D. radiodurans limite la quantité de radicaux oxygénés, d’une part en surexprimant des catalases et des superoxydes dismutases et d’autre part en limitant leur production par un changement des voies métaboliques produisant l’énergie (Liu et al., 2003 ; Tanaka et al., 2004). Ainsi durant la

35 réparation, les gènes codant les enzymes du cycle de Krebbs sont réprimés et remplacés par des gènes codant les enzymes de la voie du glyoxylate, moins productrices de radicaux libres. Il semblerait donc qu’une partie de la radiotolérance de D. radiodurans passe par la diminution du stress oxydant provenant du métabolisme afin d’éviter d’endommager encore plus les constituants cellulaires en cours de réparation. Lors de la reprise de la croissance, une fois que les dommages sont réparés, les gènes du cycle de Krebbs sont à nouveau ré-exprimés (Liu et al., 2003), ce qui confirme les analyses protéomiques (Joshi et al., 2004). Un autre point intéressant est que la majeure partie des gènes surexprimés après irradiation ou dessiccation concerne des gènes dont la fonction est inconnue. Seize de ces gènes ont été nommés ddrA à ddrP (ddr = DNA damage response) (Tanaka et al., 2004). Un autre, pprA (ppr = pleiotropic protein promoting DNA repair), a aussi été identifié dans une autre étude (Narumi et al., 2004). Cependant, tous ne sont pas présents chez D. geothermalis (Makarova et al., 2007). Cinq d’entre eux, ddrA, ddrB, ddrC, ddrD et pprA, sont les gènes les plus induits après dessiccation et irradiation gamma, et tous sont présents chez D. geothermalis (Makarova et al., 2007).

1.3.3) Les gènes ddrA, ddrB, ddrC, ddrD et pprA sont impliqués dans la radiotolérance

Afin de voir si les gènes ddrA, ddrB, ddrC, ddrD et pprA sont impliqués dans la radiotolérance, des mutants ont été construits. La délétion de ddrA, ddrB ou de pprA entraîne, à partir de 5 kGy pour ddrA et ddrB et dès 1 kGy pour pprA, une diminution de la résistance à l’irradiation gamma (Fig. 11) (Tanaka et al., 2004 ; Liu et al., 2003 ; Harris, 2004 #208 ; Narumi et al., 2004). Le mutant ddrA est également sensible à la mitomycine C (Harris et al., 2004). La délétion de ddrC, ddrD ou de ddrC et ddrD, n’affecte pas la survie aux irradiations gamma. Le mutant pprA présente un retard de croissance (temps de génération de 2h25 au lieu de 1h30) et le mutant ddrD présente des cellules environ 1,5 fois plus grosses que celles du sauvage (Tanaka et al., 2004), suggérant que PprA et DdrD ont également un rôle en conditions standard de croissance (non-induits). La délétion de ddrA ou ddrB dans un mutant pprA augmente encore plus sa sensibilité aux rayons gamma alors que celle de ddrC ou ddrD la réduit. La délétion de ddrA et ddrC dans un mutant ddrB augmente sa sensibilité aux rayons gamma (Tableau 2) (Tanaka et al., 2004). Lorsque la délétion de ces différents gènes se fait dans un contexte ΔrecA, les doubles mutants recA ddrA, recA ddrB, recA ddrC et recA ddrD sont encore plus sensibles aux rayons gamma que le mutant recA alors que l’inactivation de pprA ne l’affecte pas plus (Tanaka et al., 2004). Les résultats de ces analyses génétiques suggèrent que PprA peut contribuer à la réparation dépendante de RecA. Deux

36 Souche sauvage ΔddrD ΔddrC

ΔddrA Figure 11 : Courbe de survie aux rayons gamma de la souche sauvage et des mutants recA, ddrA, ddrB,

ΔrecA ddrC, ddrD et pprA. Adapté de ΔddrB Tanaka et al. (2004).

ΔpprA

autres voies, au moins en partie indépendantes de RecA, l’une nécessitant DdrA et l’autre nécessitant DdrB, contribuent à la radiotolérance de D. radiodurans. Daly et Minton avaient déjà proposé qu’il existe un mécanisme de réparation indépendant de RecA, capable de réassembler environ 30 % du génome (Daly & Minton, 1996). Il est possible que DdrA et DdrB participe à ce mécanisme. Il a d’ailleurs été montré que DdrA, DdrB et PprA se fixent à l’ADN et qu’elles seraient impliquées dans la réparation de l’ADN (voir chapitre 1.5.2) (Norais et al., 2009 ; Murakami et al., 2006 ; Narumi et al., 2004 ; Harris et al., 2004). Enfin, bien que DdrC et DdrD soient impliquées dans la radiotolérance, le mécanisme dans lesquels DdrC et DdrD fonctionnent est peu clair.

Tableau 2 : Effet de simples et doubles mutations sur la survie aux rayons gamma de D. radiodurans ddrA ddrB ddrC ddrD pprA

ddrA ------

ddrB ------

ddrC + + - -

ddrD + - -

pprA - - - Les moins traduisent une plus faible résistance aux rayons gamma par rapport à la souche sauvage et leur nombre traduit le degré de sensibilité (« - : un peu plus sensible » à « ---- : hypersensible »). Les plus traduisent une résistance normale. D’après Tanaka et al. (2004).

37 1.4) La régulation des gènes de la radiotolérance chez D. radiodurans

Si la synthèse et l’action de RecA ne sont pas régulées finement (voir chapitre 1.2.2), cela peut engendrer de graves instabilités génomiques (Cox, 2007b). Chez E. coli, l’expression de recA et de plusieurs autres gènes de réparation de l’ADN est sous le contrôle du répresseur LexA du système SOS (Courcelle et al., 2001 ; Fernandez De Henestrosa et al., 2000 ; Wade et al., 2005). Lorsque l’ADN est endommagé, RecA va former un nucléofilament sur l’ADN simple brin. Ce nucléofilament active l’autoclivage de LexA, ce qui déréprime les gènes du système SOS (Schlacher & Goodman, 2007 ; Butala et al., 2009). Comme décrits ci-dessous, D. radiodurans possède 2 lexA, mais recA et d’autres gènes nécessaires à la radiotolérance sont régulés différemment comparé à E. coli.

1.4.1) La régulation de recA chez D. radiodurans

Chez D. radiodurans, RecA participe à sa propre régulation. La souche rec30 (mutation recA670), produit une protéine RecA (RecA Gly224Ser) déficiente pour les activités de recombinaison et co-protéase de LexA. Cette souche présente une quantité de RecA plus importante que la souche sauvage en condition normale de culture (Fig. 12) (Bonacossa de Almeida et al., 2002). Afin de vérifier que la surexpression de RecA670 n’est

Allèle recA

Plasmide + 670 + 670 RecA (ng) Chromosome + 670 + + + + 80 160

RecA

- MMC +MMC Figure 12 : Western Blot sur les protéines RecA dans divers contextes génétiques. Une copie sauvage ou mutante de recA sur un plasmide est apportée à une souche sauvage. + = recA de la souche sauvage ; 670 = recA de la souche rec30 ; -MMC = sans mitomycine C ; + MMC = exposition à de la mitomycine C (0,5 μg/mL). Adapté de Bonacossa de Almeida, (2002). pas due au fait que la souche n’est pas capable de réparer les dommages de l’ADN, les auteurs ont combiné l’allèle recA670 avec une copie sauvage de recA dans la cellule. La souche diploïde recA+/recA670 est radiotolérante mais le niveau d’expression de RecA est toujours élevé en condition normale de culture (Fig. 12), ce qui signifie que RecA régule sa propre expression (Bonacossa de Almeida et al., 2002). Le mécanisme par lequel cette régulation se

38 fait est inconnu. L’expression de recA dans le diploïde recA+/recA670 n’est pas complètement déréprimé, car RecA est davantage induit lorsque les cellules sont exposées à de la mitomycine C (Fig. 12) (Bonacossa de Almeida et al., 2002). D. radiodurans possède 2 gènes homologues à lexA, DR_A0344 (LexA1) et DR_A0074 (LexA2) alors que D. geothermalis n’en possède qu’un, assez éloigné des 2 gènes lexA de D. radiodurans (Makarova et al., 2001 ; Makarova et al., 2007). Comme chez E. coli, les protéines LexA de D. radiodurans s’autoclivent avec l’aide de RecA (Narumi et al., 2001, Satoh et al., 2006 ; Sheng et al., 2004) et au moins pour LexA1, cet auto-clivage requiert une certaine quantité de RecA dans la cellule (Jolivet et al., 2006). Cependant, contrairement à recA d’E. coli, l’expression de recA chez D. radiodurans n’est pas contrôlée par LexA (Satoh et al., 2006 ; Bonacossa de Almeida et al., 2002 ; Narumi et al., 2001 ; Sheng et al., 2004). LexA1 n’intervient pas dans la radiotolérance (Narumi et al., 2001 ; Satoh et al., 2006) et agit plutôt sur la composition de la membrane externe car les cellules du mutant DR_A0344 s’agrègent entre elles (Bonacossa de Almeida et al., 2002). Le mutant DR_A0074 est plus résistant que la souche sauvage pour des irradiations gamma supérieures à 6-8 kGy et ceci serait dû en partie à une plus forte expression de pprA après irradiation gamma (Satoh et al., 2006) et/ou une modification du métabolisme (Sheng et al., 2004). L’expression de recA chez D. radiodurans est régulée, mais de façon inconnue, par au moins 2 protéines, IrrE et RecX (Earl et al., 2002a ; Hua et al., 2003 ; Sheng et al., 2005b). Le gène irrE, essentiel pour la radiotolérance, a été identifié après caractérisation d’un mutant

A B

Figure 13: Effet de la délétion d’irrE sur la survie aux irradiations et sur l’expression de RecA et PprA. (A, B) C Survie de D. radiodurans (○) et du mutant irrE (▲) aux rayons gamma et UV. (C) Niveau d’expression de recA D WT irrE dans une souche sauvage (en noir) et un mutant irrE (en gris) après irradiation gamma (3 kGy). Adapté d’Earl et al. (2002). (D) Western blot après irradiation gamma (2 kGy) montrant l’expression de RecA et PprA. Adapté d’Hua et al. (2003).

recA

39 très radiosensible (Fig. 13A et B) (Mattimore et al., 1995 ; Earl et al., 2002a). La délétion d’irrE (DR_0167, aussi appelé pprI) diminue fortement l’induction de la transcription de recA (Fig. 13C) et de pprA après irradiation gamma (Earl et al., 2002a ; Hua et al., 2003 ; Ohba et al., 2005). La figure 13D montre la différence entre les quantités de protéines RecA et PprA d’une souche sauvage et d’un mutant irrE avant et après irradiation gamma. IrrE est une protéine unique à D. radiodurans et D. geothermalis. Elle comporte un domaine métallo- protéase putatif et un domaine HTH potentiellement impliqué dans la fixation à l’ADN (Earl et al., 2002a ; Hua et al., 2003 ; Makarova et al., 2007). Contrairement à E. coli et à d’autres bactéries, le gène recX de D. radiodurans n’est pas organisé en opéron avec le gène recA (White et al., 1999 ; De Mot et al., 1994). La délétion de recX n’affecte pas la radiotolérance de D. radiodurans alors que sa surexpression l’affecte légèrement (Fig. 14A) (Sheng et al., 2005a ; Sheng et al., 2005b). Comme chez E.

WT recX recX++ A B

C WT recX recX++

Figure 14 : Effet de la délétion et de la surexpression de recX sur la survie aux rayons gamma et sur l’expression de RecA chez D. radiodurans. (A) Effet de la délétion et de la surexpression de recX sur la survie aux rayons gamma de D. radiodurans. Souche sauvage ( ), mutant recX ( ), surexpression de recX (S). (B) Activité du promoteur recA mesurée par l’activité de la β-galactosidase à l’aide d’une fusion entre le promoteur de recA et le gène lacZ. (C) Western blot montrant RecA et RecX dans la souche sauvage (WT), le mutant recX (recX)et la souche surexprimant recX (recX++). En B et C, les cellules sont (γ+) ou non (γ-) irradiées à 2 kGy. Adapté de Sheng et al. DNA repair (2005). coli, RecX de D. radiodurans est capable d’interagir in vitro avec RecA pour inhiber son activité recombinase, ATPase et co-protéase (Sheng et al., 2005b). De plus, le mutant recX présente une expression constitutive de recA (pas plus d’induction par l’irradiation) alors qu’une souche surexprimant RecX présente une faible expression de recA (Fig. 14B) (Sheng et al., 2005a). La perte d’un gène de résistance au chloramphénicol inséré dans le chromosome a été observée dans le mutant recX après quelques générations, suggérant que l’inactivation de recX provoque une instabilité du génome, probablement parce que l’expression (Fig. 14B et C) et l’activité de RecA ne sont plus inhibées (Sheng et al., 2005a).

40

1.4.2) Régulation globale des gènes de la radiotolérance

IrrE et RecX ne sont pas seulement impliqués dans la régulation de RecA. Une analyse de protéomique comparative a montré qu’IrrE contrôle positivement l’induction d’au moins 31 protéines après irradiation. Parmi ces protéines se trouvent RecA, PprA, SSB, mais aussi des protéines du métabolisme énergétique, de la régulation de la transcription, de la transduction de signaux, des chaperonnes et des protéines de fonctions inconnues (Lu et al., 2009). IrrE régule également positivement l’activité de catalases (Hua et al., 2003). Ces résultats suggèrent qu’IrrE est un régulateur général de la réponse aux dommages de l’ADN chez D. radiodurans, mais le mécanisme d’action d’IrrE est inconnu. RecX de D. radiodurans présente quant à elle des activités supplémentaires par rapport à son homologue chez E. coli. En effet, en condition normale de culture, la délétion de recX entraîne une plus forte expression des gènes DR_A0146 (codant la catalase CatA) et DR_1279 (codant la superoxyde dismutase SodA), impliqué dans la détoxification des radicaux oxygénés alors qu’une souche surexprimant recX présente une faible expression de ces gènes. Cependant chez la souche surexprimant recX, leur expression reste toujours induite après irradiation (Fig. 15). La fixation de RecX au niveau des promoteurs des gènes codant DR_A0146 et DR_1279 n’ayant pu être démontrée, le mécanisme de répression de ces gènes par RecX reste inconnu. Récemment, il a été montré par protéomique et par analyse de transcription que RecX régule, en condition standard de culture, au moins 35 gènes (Sheng et

Figure 15: Effet de la délétion et de la surexpression de recX sur la transcription des gènes catA (DR_A0146) et sodA (DR_1279) chez D. radiodurans. (1) souche sauvage non irradié,e (2) souche sauvage irradiée à 2 kGy, (3) mutant recX non irradié, (4) mutant recX irradié à 2 kGy, (5) souche surexprimant recX non irradiée, (6) souche surexprimant recX irradiée à 2 kGy. L’activité transcriptionnelle est mesurée grâce à la fusion du gène codant la β-galactosidase au promoteur du gène DR_A0146 (en noir), du gène DR_1279 (en gris). Le contrôle consiste en la mesure de l’activité β-galactosidase pour une construction sans promoteur. Adapté de Sheng et al. (2005) (FEMS Microbiology Letter). al., 2009). Ces gènes appartiennent à la réparation de l’ADN (recA, pprA et ssb), au métabolisme et à la réponse aux stress environnementaux (ex : sodA) (Sheng et al., 2009). La

41 transcription de douze de ces gènes est régulée négativement par RecX et la majorité d’entre eux sont radio-induits. Les 23 autres, plutôt des gènes liés au métabolisme en général, sont régulés positivement par RecX. Les auteurs suggèrent que RecX intervient dans le checkpoint régulant la réparation de l’ADN et la reprise de la croissance (Sheng et al., 2009). Très récemment, 3 nouveaux gènes importants pour la radiotolérance de D. radiodurans ont été découverts : drRRA (DR_2418), pqqE (DR_C0034), et pprM (DR_0907), (Fig. 16). Le gène drRRA a été identifié lors de la recherche de systèmes à 2 composants impliqués dans la radiotolérance (Wang et al., 2008). Les systèmes à 2

A B C

Dose (kGy) Dose (kGy) Dose (kGy)

Figure 16 : Survie à l’irradiation gamma des mutants drRRA, pqqE et pprM. (A) Survie du mutant drRRA ( ), de la souche sauvage ( ) et du mutant complémenté par drRRA (S). (B) Survie du mutant pqqE ( ) et de la souche sauvage ( ). (C) Survie du mutant pprM ( ), de la souche sauvage ({) et des mutants pprA (S), pprA pprM ( ), irrE (†). D’après Wang et al. (2008) pour drRRA, Rajpurohit et al. (2008) pour pqqE et Ohba et al. (2009) pour pprM. composants servent à la transduction d’un signal perçu par la cellule et sont composés d’un senseur (histidine kinase) et d’un régulateur de réponse. drRRA code un régulateur de réponse, conservé chez les bactéries, et dont la délétion entraîne chez D. radiodurans une forte diminution de la survie aux rayons gamma (Fig. 16A) et une légère diminution de survie aux UV, au péroxyde d’hydrogène et à la dessiccation (Wang et al., 2008). Dans ce mutant, des gènes liés au stress général (ex : groE, dnaK, dnaJ, lon, dps), à la résistance à la dessiccation (DR_B0118, DR_0105 et DR_1172), aux dommages à l’ADN (ex : recA, pprA, gyrB, ddrI, ddrC) et à la détoxification des radicaux oxygénés (katE, sodA, sodC, DR_A0145) sont moins exprimés en condition normale de croissance ou après irradiation gamma comparé à la souche sauvage (Wang et al., 2008). Il est dommage que les auteurs n’aient pas indiqué pour la souche mutante le taux d’induction des gènes après irradiation par rapport aux conditions standards. En effet, l’examination des données montre que chez le mutant drRRA, l’induction de la plupart des gènes tels que recA, pprA ou ddrC est similaire à celle trouvée dans la souche sauvage. Ceci suggère que le phénotype de sensibilité aux dommages à l’ADN de ce mutant est en partie dû à la diminution générale de l’expression d’un ensemble de gènes

42 et non pas à une non-induction de ces gènes. DrRRA se fixe à au moins un promoteur in vitro, celui de ddrI, un gène radio-induit qui coderait un régulateur transcriptionnel de la famille Crp/Fnr (Dgeo_1015 est l’homologue chez D. geothermalis). D’autres promoteurs semblent avoir été testés par ces auteurs mais la liste n’a pas été précisée. Un régulateur de réponse agit généralement sous l’action d’un senseur histidine kinase, mais le(s) senseur(s) qui régule(nt) DrRRA n’est(ne sont) pas connu(s) (Wang et al., 2008). D. radiodurans possède le gène pqqE (DR_C0034), codant pour une qui produirait du PQQ (pyrroloquinoline-quinone). Le mutant pqqE, ne produisant plus de PQQ, est très sensible aux rayons gamma (Fig. 16B) et est affecté dans la réparation de l’ADN (Rajpurohit et al., 2008). Le PQQ est un composé ayant un rôle d’antioxydant, de vitamine, et de co-facteur pour les quinoprotéines comme la glucose déhydrogénase. Le motif de fixation du PQQ dans ces quinoprotéines est également présent dans des Ser/Thr protéines kinases chez plusieurs organismes. D. radiodurans possède 5 Ser/Thr kinases comportant le motif de fixation du PQQ et le mutant pqqE présente une baisse d’un facteur 10 de la phosphorylation des protéines (Rajpurohit et al., 2008). Les résultats suggèrent un rôle du PQQ dans la régulation de l’activité de protéines kinases potentiellement impliquées dans la radiotolérance de D. radiodurans. À noter que le gène pqqE n’a pas été trouvé chez D. geothermalis et pourtant, cette bactérie est aussi radiotolérante que D. radiodurans. PprM a été trouvé lors de la comparaison d’un profil de protéines en gel 2D entre une souche sauvage et un mutant irrE en conditions standard de croissance (Ohba et al., 2009). C’est une protéine qui présente peut-être une modification post-traductionnelle qui varie selon que la souche soit sauvage ou délétée d’irrE (Ohba et al., 2009). PprM est une « cold shock protein » et elle comporte de nombreux homologues chez les bactéries dont les plus proches sont les 2 protéines de D. geothermalis, Dgeo_0638 et Dgeo_1006 (Ohba et al., 2009). Une étude montre cependant que PprM est induite après un heat shock (3h à 42°C) (Airo et al., 2004), contrairement à une autre étude qui montre que pprM est plutôt réprimé par un heat shock (1h à 48°C) (Schmid et al., 2005). Le mutant pprM est aussi sensible à l’irradiation gamma qu’un mutant pprA (Fig. 16C). D’autres résultats suggèrent que PprM est impliqué dans la répression de PprA, mais pas de RecA, dans des conditions standards de culture (Ohba et al., 2009). Le fait que le double mutant pprM pprA soit plus sensible que les simples mutants pprM ou pprA (Fig. 16C) peut suggérer que PprM régule l’expression d’autres protéines importantes pour la radiotolérance. L’effet de l’inactivation de pprM sur la survie ou la réponse après un heat shock ou un cold shock n’a pas été testé. La régulation des gènes permettant la radiotolérance chez D. radiodurans est très complexe car elle semble dépendre de plusieurs régulateurs, incluant IrrE, DrRRA, RecX,

43 PprM et LexA2 (les 2 derniers régulant au moins pprA). Cependant d’autres études devront être menées afin de clarifier le rôle de chacun et/ou de l’ensemble de ces régulateurs.

Tableau 3 : Gènes présentant un RDRM chez D. radiodurans et D. geothermalis Gène Description Induction* DR_0070 DdrB, une protéine SSB-Like OUI DR_0099 SSB, Single Strand DNA Binding protein OUI DR_0171 IrrI, régulateur transcriptionnel, impliqué dans la radiotolérance ; Absent chez D. geothermalis OUI DR_0219 DdrF, une protéine de fonction inconnue ; Absent chez D. geothermalis OUI DR_0326 DdrD, impliquée dans la radiotolérance OUI DR_0423 DdrA, appartient à la famille Rad52/22, nécessaire à la réparation des cassures double brin OUI DR_0596 RuvB, hélicase ADN pour la résolution des jonctions de Holliday OUI DR_0659 FrnE, DsbA-like thioredoxin OUI DR_0906 GyrB, la sous-unité B de la DNA-gyrase OUI DR_1039 MutS, une des protéines du mismatch repair NON DR_1143 Similaire à DR_1142 mais avec un frameshift ; Absent chez D. geothermalis OUI DR_1262 Rsr, protéine se fixant à l’ARN de la famille Ro-like OUI DR_1289 RecQ, hélicase ADN comportant 3 domaines HRDC NON DR_1696 MutL, une des protéines du mismatch repair NON DR_1771 UvrA, Excinuclease du NER OUI DR_1775 UvrD, hélicase ADN OUI DR_1913 GyrA, la sous-unité A de la DNA-gyrase OUI DR_1921 SbcD, exonucléase NON DR_2256 Tkt, translokase OUI DR_2275 UvrB, hélicase ADN du NER OUI DR_2338 CinA, LigT, RecA, opéron ou se situe recA OUI DR_2574 DdrO, regulateur transcriptionel OUI DR_A0151 HutU, Urocanate hydratase (3 gènes supplémentaires en opéron pour la dégradation de l’histidine) OUI DR_A0346 PprA, impliquée dans la radiotolérance OUI Adapté d e Makarova et al. (2007). * D’après le tableau 4 de Marakova et al. (2007) (Induction des gènes observés d ans les travaux de Liu et al. (2003) ou de Tanaka et al. (2004)).

Lors de la comparaison des génomes de D. radiodurans et D. geothermalis (Makarova et al., 2007), les auteurs ont trouvé une séquence palindromique spécifique de 17 pb (Fig. 17) devant une vingtaine de gènes/opérons surexprimés après dessiccation/irradiation ou impliqués dans la réparation de l’ADN (Tableau 3). Cette séquence porte le nom de radiation desiccation response motif (RDRM). Il est probable que ce motif soit reconnu par une/des

Figure 17: Séquence du Radiation Desiccation Response Motif. La taille de chacune des lettres (A, T, G, C) est fonction de leur abondance au niveau des différentes séquences RDRM dans les génomes de D. radiodurans et D. geothermalis. D’après Makarova et al. (2007).

44 protéine(s) régulatrice(s) mais qui est/sont inconnue(s) à ce jour. Par exemple, aucune fixation d’IrrE au niveau du promoteur de pprA n’a pu être mis en évidence (Ohba et al., 2005). À noter qu’irrE ne présente pas de séquence RDRM et qu’il n’est pas radio-induit (Gao et al., 2006). D’autres candidats possibles pour la fixation au RDRM pourraient être DdrO et/ou IrrI (Liu et al., 2003 ; Udupa et al., 1994). Ces protéines présentent des domaines prédits de fixation à l’ADN, et elles sont toutes les deux induites après irradiation. Le mutant irrI est très affecté dans la survie aux radiations UV et gamma (Udupa et al., 1994). Cependant, irrI n’est pas présent chez D. geothermalis. Le gène ddrO a été trouvé exclusivement chez les 2 Deinococcus séquencées, mais aucun travail n’a été réalisé sur ce gène.

45 1.5) Les mécanismes de la radiotolérance chez D. radiodurans

Différentes études suggèrent que des mécanismes passifs (protecteur) et actifs (réparateur) contribuent à la radiotolérance de D. radiodurans. Ces mécanismes et hypothèses sont décrits ci-dessous.

1.5.1) Les mécanismes passifs

Le nombre de copies de génome Chaque cellule de D. radiodurans possède entre 4 et 10 copies de son génome et sa radiotolérance est la même quel que soit ce nombre (Hansen, 1978, Harsojo et al., 1981). La recombinaison homologue étant le système de réparation des cassures double brin le plus répandu chez les bactéries, le nombre de copies du génome est donc important. Plus le nombre de copies est important et moins il y a de risque qu’un gène soit inactivé. Ainsi pour 100 lésions introduites aléatoirement dans le génome de D. radiodurans, la probabilité d’inactiver un gène en particulier est de 3,4 , 0,12 et 0,004 % avec respectivement 1, 2 et 3 copies du génome (Cox & Battista, 2005). Par ailleurs, il a été montré que lorsque la levure Saccharomyces cerevisiae a 2 ou 4 copies de son génome, elle est plus résistante aux rayons gamma que lorsqu’elle en a 1 (Mortimer, 1958). Toutefois, la radiotolérance ne se mesure pas toujours au nombre d’exemplaires de génome. E. coli par exemple, qui contient plusieurs copies de son génome en phase exponentielle de croissance (Akerlund et al., 1995) et Azotobacter vinelandii qui en possède jusqu’à 100 copies, ne sont pas radiotolérantes (Maldonado et al., 1994). Ainsi la polyploïdie est probablement importante mais pas suffisante pour expliquer la radiotolérance de D. radiodurans.

Le nucléoide de D. radiodurans Un autre aspect concernant les molécules d’ADN chez D. radiodurans est leurs organisations structurales. Comparé à E. coli et T. aquaticus, le nucléoide de D. radiodurans est hyper-compact et prend la forme d’un anneau (Fig. 18) (Zimmerman & Battista, 2005 ; Englander et al., 2004 ; Levin-Zaidman et al., 2003). Cet anneau n’est pas altéré par de fortes doses de rayons gamma (Zimmerman & Battista, 2005). Il est proposé que lorsque des cassures double brin sont générées chez D. radiodurans, cette structure permet de maintenir les fragments et les extrémités d’ADN à proximité et évite leur diffusion dans le cytoplasme, ce qui faciliterait la réparation (Levin-Zaidman et al., 2003). Un nucléoide hyper-compact, pas forcément en anneau, est également retrouvé chez les autres Deinococcus observées (D. murayi, D. radiophilus, D. grandis, D. geothermalis et D. radiopugnans) ainsi que chez

46 A B C

Figure 18: Images en épifluorescence de bactéries. (A) D. radiodurans, (B) E. coli, (C) T. aquaticus, (D) D. grandis et (E) R. radiotolerans. L’ADN est marqué au DAPI et apparaît en bleu alors que la E membrane est colorée au FM4-64 et apparaît en rouge. D’après D Zimmerman et Battista (2005).

Rubrobacter radiotolerans, une autre bactérie radiotolérante (Fig. 18) (Zimmerman & Battista, 2005). Le fait que la forme du nucléoide ne soit pas standard suggère que c’est l’ADN hyper-compacté et non pas la forme qui serait important pour la radiotolérance. Cependant, une organisation particulière de l’ADN dans le nucléoide n’est pas observée dans certaines études et est très discutée (Eltsov & Dubochet, 2005 ; Eltsov & Dubochet, 2006). Pour une recombinaison très efficace, Minton et Daly (1995) avaient proposé que les copies de chromosomes soient alignées mais cela n’a jamais été démontré (Minton & Daly, 1995). Toutefois, Eltsov et al. (2005) ont montré que par endroits, le chromosome de D. radiodurans semblait présenter des arrangements parallèles. La condensation du nucléoide chez les bactéries fait intervenir des protéines associées au nucléoide (NAP = nucleoid-associated protein). Ces protéines sont généralement de petite taille et sont chargées positivement. Douze de ces NAPs (HU, DnaA, Fis, H-NS, IHF, Lrp, CbpA, CbpB, Dps, Hfq, IciA et StpA) ont été étudiées au niveau de leurs affinités et de leurs spécificités à reconnaître des séquences de l’ADN chez E. coli (Azam & Ishihama, 1999). Plusieurs de ces protéines (HU, Fis, IHF, H-NS, Lrp) sont capables de courber et de condenser l’ADN (Dame, 2005). Des recherches par homologie de séquence montrent que seulement 4 de ces NAPs sont conservées chez D. radiodurans (HU, Dps, DnaA et Lrp), ce qui suggère que d’autres nouvelles protéines pourraient être impliquées dans la très forte compaction de son génome. Un autre type de protéines, de la famille SMC (SMC = structural maintenance of chromosomes), est important pour l’organisation du chromosome (Thanbichler et al., 2005). D. radiodurans possède un homologue à cette famille, la protéine SMC. Parmi les 5 protéines citées ci-dessus, 3 ont été inactivées et étudiées chez D. radiodurans, HU (DR_A0065), SMC (DR_1471) et Dps (DR_2263 et DR_B0092). Parmi elles, seule HU a été montrée comme étant essentielle à la survie et pour la compaction du 47 nucléoide de D. radiodurans (Bouthier de la Tour et al., 2009 ; Nguyen et al., 2009). À noter également que les gyrases (GyrA et GyrB), qui sont radio-induites (voir chapitre 1.1.3), ainsi que les topoisomérases (dont Topo 1B, une topoisomérase plutôt retrouvée chez les Eucaryotes) pourraient jouer un rôle dans la structuration du nucléoide et dans la radiotolérance (Blasius et al., 2008). De plus, le manganèse (Mn2+) pourrait aider à la compaction du nucléoide. En effet, D. radiodurans possède une forte concentration en manganèse au niveau du cytosol (Daly, 2009) et de l’ADN (Leibowitz et al., 1976). La présence du Mn2+ au niveau de l’ADN pourrait diminuer les forces de répulsion dues aux charges négatives de l’ADN (Cox & Battista, 2005).

La protection contre le stress oxydant Des espèces réactives de l’oxygène (ROS : reactive oxygen species), comme le •- • peroxyde d’hydrogène (H2O2), le superoxyde (O2 ) et le radical hydroxyle (OH ), sont générés par le métabolisme aérobie. En conditions de stress, comme par exemple lors d’irradiations, les quantités de ROS seront plus élevées et pourront endommager l’ADN, l’ARN, les protéines et les lipides. Les cellules se protègent contre les ROS en utilisant certaines enzymes (Voir 1.5.2) et des composés antioxydants. Chez D. radiodurans, la protection passive contre le stress oxydant implique au moins 2 composés, les caroténoïdes (Carbonneau et al., 1989) et le manganèse (Daly, 2009). La deinoxanthine, le principal caroténoïde de D. radiodurans, donne la couleur rouge à cette bactérie et est un puissant antioxydant (Tian et al., 2007). La délétion de gènes servant à la synthèse de caroténoïdes tels que crtI ou crtB, montre que ces mutants sont plus sensibles aux irradiations, au stress oxydant et à la dessiccation que la souche sauvage (Xu et al., 2007 ; Zhang et al., 2007a). Toutefois, certains mutants non colorés de D. radiodurans sont toujours radiotolérants (Blasius et al., 2008). Ainsi, la pigmentation ne semble pas nécessaire à la radiotolérance de D. radiodurans. Le Mn2+ est un très bon antioxydant qui, en concentration élevée de l’ordre du millimolaire, peut avoir un rôle similaire aux catalases et superoxydes dismutases. En effet, Lactobacillus plantarum ne possède pas ces enzymes, mais utilise le Mn2+ pour se protéger des radicaux libres (Archibald & Fridovich, 1981). Contrairement au Mn2+, le fer libre (Fe2+) dans la cellule réagit avec le peroxyde d’hydrogène pour former des radicaux hydroxyles (OH•), augmentant ainsi la quantité de ce radical très réactif. Comme dit précédemment, le Mn2+ est fortement présent chez D. radiodurans (1-2 mM) et semble être réparti dans toute la cellule (Fig. 19) (Daly, 2009). Daly propose que D. radiodurans accumule du Mn2+ sous forme de complexes, peut-être avec des phosphates (Daly, 2009). D. radiodurans présente une faible quantité de fer (Fe) et celui-ci est plutôt présent au niveau du septum de division

48 (Fig. 19) (Daly et al., 2004 ; Daly, 2009). La raison de cette localisation est inconnue. D’autres bactéries radiotolérantes (D. radiodurans, D. geothermalis, K. radiotolerans, Enterococcus faecium …) possèdent également un ratio Mn/Fe élevé alors que ce ratio est faible chez des bactéries radiosensibles (E. coli, Shewanella oneidensis) (Daly et al., 2007 ; Fredrickson et al., 2008 ; Ghosal et al., 2005 ; Bagwell et al., 2008), ce qui suggère qu’un ratio élevé Mn/Fe est un élément commun aux bactéries radiotolérantes. Le ratio élevé de Mn/Fe a été corrélé à une limitation considérable de l’oxydation des protéines après irradiation ou dessiccation (Daly et al., 2007 ; Fredrickson et al., 2008). Par contre, l’ADN n’est pas protégé. Ainsi, la protection des protéines par le Mn2+ permet à la cellule de conserver des protéines fonctionnelles qui restent donc capables de réparer les dommages de l’ADN. En effet, en condition limitante de Mn2+, D. radiodurans est moins résistante aux rayons gamma (Daly et al., 2004), mais ceci peut également être lié au fait que certaines

Figure 19 : Carte de fluorescence au rayons X de la distribution du fer et du manganèse chez D. radiodurans. À gauche et à droite, sont montrés respectivement la répartition et les gradients de concentration du fer (Fe) et du manganèse (Mn). Au milieu figure une tétrade observée au microscope électronique à transmission. L’échelle en bas indique les quantités (qualitatives) des métaux observés, En bleu, une faible quantité et en rouge, une forte quantité. Adapté de Daly, (2009). enzymes ont besoin de Mn2+ pour fonctionner. Cependant, une dégradation de protéines a lieu après irradiation gamma (Joshi et al., 2004), ce qui suggère qu’une partie des protéines n’est pas protégée par le Mn2+. De plus, bien que L. plantarum contienne de l’ordre de 20 mM de 2+ Mn et pas de fer (Archibald & Fridovich, 1981), la D10 (dose tuant 90 % des bactéries) observée après irradiation gamma dans un milieu riche est de l’ordre de 0,5 à 0,8 kGy

(Hastings et al., 1986). Cette D10 est similaire à celle observée pour E. coli et est beaucoup moins élevée que celle de D. radiodurans (10-12 kGy). Ainsi, même si le Mn2+ est important, il n’est pas suffisant pour expliquer la forte radiotolérance de D. radiodurans.

1.5.2) Les mécanismes actifs

Les mécanismes actifs permettant la radiotolérance incluent bien sûr la réparation de l’ADN. Après irradiation gamma, la première étape conduisant à la réparation passe par une dégradation contrôlée de l’ADN, avec une exportation des nucléotides endommagés (Boling & Setlow, 1966 ; Hariharan & Cerutti, 1972). Il est possible que cette étape permette de

49 réduire les erreurs d’incorporation de nucléotides endommagés lors de la réparation. De plus, la dégradation de l’ADN génère des régions simple brin de l’ADN qui sont des substrats pour les mécanismes de réparation comme la recombinaison homologue et le SSA. D. radiodurans possède des homologues aux différents gènes impliqués dans la réparation de l’ADN (BER, NER, MMR, RER) comme décrit dans le chapitre 1.2.2. Pour réparer les nombreuses cassures double brin de l’ADN, 4 mécanismes ont été proposés chez D. radiodurans (Fig. 20). En plus du système RER, une réparation partielle des cassures double brin est observée dans un mutant recA (Fig. 7) (Daly & Minton, 1996). Un des mécanismes qui pourrait être responsable de cette réparation est le SSA (Fig. 8 et 20). Plus récemment, un mécanisme proche du SSA mais dépendant de RecA, le ESDSA (Zahradka et al., 2006 ; Slade et al., 2009) a été découvert (Fig. 20 et 21). Un quatrième mécanisme, le NHEJ (Fig. 20) est également proposé. Les deux derniers mécanismes sont décrits ci-dessous.

Le mécanisme de ESDSA

Cassure double brin de l’ADN

Recombinaison Extended Synthesis-Dependent Single Strand Non-Homologous homologue (RER) Strand Annealing (ESDSA) Annealing (SSA) End Joining (NHEJ)

Figure 20 : Schémas simplifiés des différentes voies de réparation des cassures double brin chez D. radiodurans. Adapté de Blasius et al. (2008).

Les mutants recA et polA (codant une ADN polymérase de la famille A impliquée dans la réparation de l’ADN et la synthèse de l’ADN) sont hypersensibles aux radiations gamma (Gutman et al., 1994 ; Gutman et al., 1993). Zahradka et al. (2006) ont mis en évidence un mécanisme dépendant à la fois de RecA et de PolA (aussi appelée PolI) et ce modèle est le ESDSA (Fig. 21) (Zahradka et al., 2006). Slade et al. (2009) ont complété ce modèle récemment en montrant que RadA (une protéine impliqué dans la recombinaison, et 50 plus particulièrement dans la maturation des jonctions de Holliday chez E. coli (Beam et al., 2002)) ainsi que PolIII, l’ADN polymérase réplicative, étaient importantes pour le fonctionnement de ce système de réparation (Slade et al., 2009). Ainsi, RecA et RadA permettent aux extrémités 3’ sortantes d’envahir de l’ADN complémentaire, puis PolIII et PolA synthétisent de grands fragments d’ADN simple brin. Après appariement de ceux-ci, la reconstruction du génome circulaire est assurée par la recombinaison homologue (Fig. 21) (Zahradka et al., 2006 ; Slade et al., 2009).

Cassure double brin Figure 21 : Processus de reconstitution de chromosomes cassés en morceaux. Deux modèles expliquent par de la complémentarité de brin d’ADN le réassemblage de petits 1 fragments de chromosome en plus grands fragments. L’un est le SSA et l’autre est le ESDSA. Plusieurs copies du génome de D. radiodurans subissent des cassures double SSA ESDSA 2 brin aléatoires, produisant ainsi de nombreux fragments. Après la formation d’extrémités 3’ sortantes (dégradation de l’ADN de 5’ vers 3’) (Etape 1), les extrémités d’ADN 3 4 peuvent se rejoindre directement grâce à une complémentarité de séquence existant entre les différents fragments qui proviennent de plusieurs copies 4 chromosomiques (Etape 4, le SSA). Dans le modèle du ESDSA, les extrémités 3’ vont envahir des régions 5 5 complémentaires (Etape 2, dépendant de RecA et RadA) et une synthèse importante d’ADN aura lieu, produisant alors de longues extrémités 3’ (Etape 3, en rouge, dépendant de 6 6 PolA et PolIII). Ces longues extrémités 3’ vont ensuite pouvoir s’apparier avec des régions d’ADN complémentaires (Etape 4 et 5, ces étapes sont identiques entre le SSA et le ESDSA). Les longs fragments ainsi Recombinaison homologue formés vont pouvoir reformer un ADN circulaire grâce à la recombinaison homologue (Etape 6). Adapté de Zahradka et al. (2006).

Un système NHEJ-like chez D. radiodurans ? Plusieurs auteurs dans la littérature proposent qu’un autre système de réparation, le NHEJ, pourrait exister chez D. radiodurans (Narumi et al., 2004 ; Bentchikou et al., 2007 ; Blasius et al., 2006 ; Kobayashi et al., 2004 ; Lecointe et al., 2004b). Ce système est peu fréquent chez les bactéries, mais existe par exemple chez B. subtilis (Moeller et al., 2007 ; Weller et al., 2002). Ku, une protéine essentielle au NHEJ, reconnaît les cassures double brin et permet le recrutement d’une (ATP-dependante) permettant de liguer les 2 fragments d’ADN (Yano et al., 2009). Cependant, aucune protéine homologue à Ku n’est présente chez D. radiodurans (Blasius et al., 2008). Toutefois, D. radiodurans possède d’autres protéines qui pourraient être des composants du NHEJ. Le premier composant pourrait être PprA car elle se fixe à l’ADN et stimule la ligation de l’ADN in vitro (Murakami et al., 2006 ; Narumi et al., 2004). Ainsi, elle interviendrait dans 2 voies différentes, le NHEJ et la réparation

51 dépendante de RecA (Voir chapitre 1.3.3). Le deuxième composant pourrait être DdrP (DR_B0100), une protéine radio-induite (Tanaka et al., 2004), faiblement homologue aux ADN ATP-dépendant (Blasius et al., 2007). Cependant, le mutant ΔddrP n’est pas affecté dans la radiotolérance (Makarova et al., 2007), aucune activité ligase n’a pu être montrée in vitro pour DdrP (Blasius et al., 2007), et D. geothermalis ne possède pas d’homologue de ddrP (Makarova et al., 2007). Ainsi, si ddrP a un rôle dans le NHEJ chez D. radiodurans, celui-ci doit être minoritaire. PolX et SbcCD qui auraient un rôle dans la maturation des extrémités d’ADN endommagées (voir le prochain chapitre pour ces protéines), pourraient également avoir un rôle dans le NHEJ. Cependant, le NHEJ n’est généralement pas un système de réparation fidèle et cela n’est pas compatible avec le faible nombre de mutations observées après irradiation chez D. radiodurans. Bien que non démontré expérimentalement, un mécanisme de réparation par NHEJ ou NHEJ-like chez D. radiodurans ne peut être exclu.

Autres protéines intervenant dans la réparation des cassures double brin de l’ADN La délétion de polX chez D. radiodurans entraîne une diminution de la survie aux rayons gamma surtout à forte dose (diminution de la survie d’un facteur 3 à 13,6 kGy et de 300 fois à 20,4 kGy) ainsi qu’un retard de réparation des cassures double brin (Bentchikou et al., 2007 ; Lecointe et al., 2004b). En plus de son activité polymérase, PolX de D. radiodurans possède une activité nucléase qui est modulée par des structures en tige boucle de l’ADN (Blasius et al., 2006). La délétion de sbcC et/ou de sbcD donne le même type de phénotype que celui du mutant polX (Bentchikou et al., 2007). Le triple mutant polX/sbcCD montre une plus forte diminution de la résistance aux rayons gamma et un plus fort retard dans la réparation des cassures double brin (Bentchikou et al., 2007). PolX et SbcCD interviendraient dans la maturation des extrémités d’ADN endommagées, qui peuvent être utilisées comme substrat pour les mécanismes de réparation de l’ADN (Bentchikou et al., 2007 ; Blasius et al., 2006). Le fait que le triple mutant présente une diminution de la radiotolérance lors de l’application de très hautes doses de rayons gamma suggère que PolX et SbcCD sont nécessaires pour aider la cellule à réparer de l’ADN extrêmement endommagé et/ou des lésions qui sont très difficiles à réparer. Les nouvelles protéines DdrA et DdrB, impliquées dans l’extrême tolérance aux radiations de D. radiodurans, ont été caractérisées. Il a été montré in vitro que DdrA (23 kDa) se fixe aux extrémités 3’ de l’ADN simple brin et qu’il les protège de la dégradation. De plus, DdrA forme des oligomères composés de 7-10 sous-unités (Gutsche et al., 2008 ; Harris et al., 2004). In vivo, son rôle est particulièrement important en absence de nutriments (Harris et al., 2004). Une souche de D. radiodurans produisant une quantité limitée de RecA n’est pas

52 affectée dans sa radiotolérance quand elle est étalée sur milieu gélosé directement après irradiation (Jolivet et al., 2006). Cependant, lorsqu’un mutant ddrA produit une quantité limitée de RecA, cela conduit à une forte diminution de la survie du mutant ddrA aux radiations gamma (Jolivet et al., 2006), suggèrant que DdrA permet de protéger les extrémités 3’ jusqu’à ce qu’elles soient utilisées dans les divers mécanismes réparant les cassures double brin. In vitro, une protéine tronquée de DdrA, à laquelle il manque les 40 derniers résidus, est toujours capable de se fixer aux extrémités 3’ de l’ADN, de les protéger de la dégradation et de s’oligomériser. Toutefois, in vivo, cette protéine tronquée n’est pas fonctionnelle. Ces résultats montrent que c’est le domaine N-terminal de la protéine qui est responsable de la fixation à l’ADN et suggère que la partie C-terminale serait importante pour l’interaction avec d’autres protéines nécessaires pour la réparation de l’ADN (Gutsche et al., 2008 ; Harris et al., 2008). DdrA est légèrement homologue à la protéine Rad52, une protéine qui intervient dans le SSA chez les Eucaryotes. Ainsi, il a été proposé que DdrA intervienne dans le SSA chez D. radiodurans (Blasius et al., 2008 ; Harris et al., 2004). DdrB est une protéine de petite taille (20,8 kDa) comportant un domaine OB-fold putatif (Norais et al., 2009), ce qui la rapproche de SSB. La comparaison des séquences montre que l’homologie entre DdrB et SSB est très faible. Par contre les 6 derniers résidus de DdrB et SSB sont similaires (Norais et al., 2009). Chez SSB, ces 6 résidus sont nécessaires pour l’interaction avec différentes protéines du métabolisme de l’ADN (Lu & Keck, 2008, Shereda et al., 2008). Excepté le fait que DdrB forme des pentamères en solution (tétramère pour SSB), DdrB se comporte comme SSB dans les études in vitro (elle se fixe à l’ADN simple brin et module la fixation de RecA sur celui-ci) (Norais et al., 2009). Il est proposé que DdrB puisse servir de plateforme pour le recrutement de protéines particulières de réparation de l’ADN (Norais et al., 2009). Compte tenu des données connues au sujet de PolX, SbcCD, DdrA, DdrB et de PprA, il est possible que l’ensemble de ces protéines interviennent dans une ou plusieurs des voies de réparation des cassures double brin (ESDSA, SSA, NHEJ et recombinaison homologue) de D. radiodurans.

La lutte contre le stress oxydant La diminution du stress oxydant chez D. radiodurans passe par un mécanisme passif (voir plus haut) mais également par un mécanisme actif faisant intervenir plusieurs enzymes. D. radiodurans possède 3 superoxydes dismutases (DR_1279, DR_1546, et DR_A0202), 3 catalases (DR_1998, DR_A0259 et DR_A0146), ainsi qu’une péroxydase (DR_A0145) (Makarova et al., 2001). La délétion de DR_1998 et de DR_1279 ne diminue que légèrement la radiorésistance (Markillie et al., 1999), ce qui suggère que ces 2 gènes sont peu importants

53 pour la radiotolérance. Cependant, il est probable qu’avec la redondance de systèmes et de protéines qu’utilise D. radiodurans pour se protéger du stress oxydant, il est difficile de voir un effet en supprimant un seul gène. D. geothermalis présente seulement des homologues pour DR_1998 (catalase) et DR_1279 (super oxyde dismutase). D. geothermalis étant aussi radiotolérante que D. radiodurans, il semblerait donc que ces 2 protéines soient suffisantes pour lutter contre le stress oxydant en combinaison avec des antioxydants comme le Mn2+. Il est intéressant de noter que l’expression de PprA ou d’IrrE stimule des activités catalases chez E. coli (Kota & Misra, 2006 ; Gao et al., 2003). PprA stimule l’activité de KatE probablement en interagissant avec elle (Kota & Misra, 2006). IrrE stimule l’activité de KatG par un mécanisme inconnu (augmentation de la transcription, de la traduction, de l’activité ?) (Gao et al., 2003). Chez D. radiodurans, en condition normale de culture, un mutant irrE présente des activités catalases 2-3 fois plus faible que la souche sauvage, ce qui suggère qu’IrrE stimule ces activités (Hua et al., 2003). Il est possible que chez D. radiodurans, PprA stimule également les activités catalases comme chez E. coli.

Dégradation et réparation des protéines endommagées après irradiation De nombreuses protéases sont induites après irradiation (Tanaka et al., 2004) et sont probablement responsables de la forte dégradation de protéines qui a lieu après irradiation gamma (voir chapitre 1.3.1). Il est possible que la dégradation des protéines endommagées joue un rôle dans la radiotolérance de D. radiodurans. Afin de tester cette hypothèse, Servant et al. (2007) ont délété les protéases appartenant à la famille Clp et Lon chez D. radiodurans (Servant et al., 2007). L’inactivation de lon1 ou lon2 n’a pas d’effet sur la radiotolérance mais diminue la résistance à la puromycine, un antibiotique provoquant la synthèse de protéines malformées, suggérant ainsi que Lon1 et Lon2 ont un rôle dans la dégradation de ce type de protéines. De plus, ce résultat suggère que la dégradation des protéines endommagées après irradiation gamma n’est pas vraiment essentielle pour l’extrême radiotolérance. L’inactivation de clpX, de clpP ou de clpXP ne diminue pas la résistance à la puromycine, ce qui suggère que ClpP et ClpX n’ont qu’un rôle mineur dans la dégradation de protéines malformées. En revanche, chez ces 3 mutants, la tolérance aux rayons gamma est diminuée et la réparation des cassures double brin est retardée par rapport à la souche sauvage (Servant et al., 2007). De plus, ces mutants présentent un retard de croissance ainsi que des défauts morphologiques après irradiation gamma (Fig. 22A, B, C), ce qui suggère que les protéines ClpX et ClpP sont impliqu ées dans le contrôle du cycle cellulaire après irradiation gamma chez D. radiodurans (Servant et al., 2007). D. radiodurans serait aussi capable de réparer les protéines endommagées. En effet, de nombreuses chaperonnes, induites après irradiation gamma pourraient restaurer la

54 conformation et donc l’activité de protéines endommagées et d’autres protéines induites pourraient directement réparer certaines altérations comme le système MsrA/B qui réverse l’oxydation des résidus méthionines (Blasius et al., 2008).

A B C

Fig 22 : Retard de croissance et morphologie des cellules clpXP après irradiation gamma. (A) Croissance de la souche sauvage („ ; ) et du mutant clpXP (z ; { ) non-irradié („ ; z) et après irradiation gamma à 6,8 kGy ( ; {). (B) et (C) Images en épifluorescence d’une souche sauvage (B) et d’une souche clpXP (C), respectivement 5 et 7h après irradiation gamma à 6,8 kGy. 1- Cellules géantes ; 2- Nucléoides décondensés ; 3- Cellules anucléés ou nucléoides coupés ; 4- Septation anormales. D’après Servant et al. (2007).

55 1.6) Conclusion sur les connaissances de la radiotolérance chez D. radiodurans

En plus d’un arsenal de gènes de réparation de l’ADN commun aux bactéries radiosensibles, D. radiodurans possède différents gènes supplémentaires nécessaires à la radiotolérance tels que ddrA, ddrB, ddrC, ddrD, irrE et pprA. Un système de réparation unique, le ESDSA, a été décrit pour D. radiodurans, ainsi qu’un nucléoide compact et une polyploïdie. Le tout serait renforcé par un système de protection contre le stress oxydant (Mn2+, caroténoïdes, enzymes détoxifiantes). L’ensemble de ces propriétés donnerait à D. radiodurans son extraordinaire capacité à résister à de très fortes doses d’irradiation et à de longues périodes de dessiccation. Bien que de nombreuses avancées dans la compréhension du mécanisme de radiotolérance de D. radiodurans aient été faites durant cette dernière décennie, de nombreuses questions restent en suspens. Le rôle de DdrA, de DdrB et de PprA commence à être élucidé, mais la fonction exacte et le mécanisme d’action de ces protéines in vivo ne sont toujours pas compris. De plus, le fonctionnement du régulateur d’IrrE et les rôles de DdrC et DdrD restent mystérieux. Il en va de même pour toutes les autres protéines, de fonctions inconnues, fortement induites après irradiations et qui sont conservées uniquement chez les Deinococcus (Ex : DdrO). Comment le Mn2+ protège-t-il les protéines des irradiations gamma/des radicaux oxygénés? Si le Mn2+ est vraiment complexé, quel est la nature de ces complexes ? Pourquoi et comment le fer est séquestré au niveau du septum de division ? Le mécanisme de réparation par le ESDSA n’explique que la réparation des cassures double brin or de nombreuses autres lésions (sites abasiques, bases oxydées ou modifiées, lésions en tandem) sont générées par les irradiations. La réparation de ces lésions est cruciale car elles peuvent bloquer l’action de PolA et PolIII lors du ESDSA et/ou provoquer des mutations. La formation et la cinétique de réparation des différentes bases modifiées n’ont pas vraiment été étudiées chez D. radiodurans. Au moment où le projet « Deinococcus » a débuté au laboratoire, en 2004, le génome d’une seule bactérie du genre Deinococcus, D. radiodurans, avait été séquencée (White et al., 1999). L’étude du génome de D. radiodurans avait conclu qu’une analyse comparative impliquant une autre espèce de Deinococcus permettrait de mieux comprendre les mécanismes de radiotolérance de ces bactéries (Makarova et al., 2001). La conception du projet « Deinococcus » a été réalisée par T. Heulin et sa mise en place et son développement effectué au laboratoire par A. de Groot. Le but du projet est de comparer une nouvelle espèce de Deinococcus avec D. radiodurans à l’aide d’approches globales (génomique et

56 protéomique) et ciblées (génétiques, biochimiques et structurales), afin de déterminer les points communs et les différences entre ces deux espèces, puis de les étudier pour nous aider à comprendre la radiotolérance de ces bactéries. L’espèce choisie pour cette étude de génomique comparative a été Deinococcus deserti souche VCD115.

57 2) Deinococcus deserti

D. deserti a été isolée d’un mélange d’échantillons de sable du Sahara irradié à 15 kGy puis caractérisée au laboratoire (de Groot et al., 2005). D. deserti est une bactérie en forme de bâtonnet, formant de petites colonies lisses de couleur blanches ou roses en fonction du milieu utilisé (D. deserti ne pousse pas sur milieu riche), et est négative lors de la coloration de Gram (de Groot et al., 2005). Au moment de sa caractérisation, seulement 8 espèces de Deinococcus étaient décrites. La figure 23 montre les espèces de Deinococcus décrites à ce jour (juillet 2009). D. deserti VCD115 est tolérante à de fortes doses d’irradiation (UV et gamma) et à de longues périodes de dessiccation (Fig. 24A, B, C) (de Groot et al., 2005) (A. de Groot, communication personnelle). Elle est capable, comme D. radiodurans, de reconstituer son génome en quelques heures après irradiation gamma ou dessiccation (Fig. 24D, E). Le

D. aetherius ST0316T (AB087287)

Figure 23 : Arbre phylogénétique des bactéries du genre Deinococcus. Sont encadrés en rouge D. radiodurans, D. geothermalis et D. deserti. Adapté de Yang et al. (2009).

58 séquençage du génome de D. deserti a été réalisé en 2005/2006 et la séquence complète (3,8 Mb) fût disponible en octobre 2006, au moment de mon arrivée au laboratoire. Au cours de ma thèse, le génome de D. geothermalis a été publié (Makarova et al., 2007) et a également été inclus dans les analyses de génomique comparative, impliquant donc 3 Deinococcus.

A B C

D E

Figure 24 : Survie aux radiations et à la dessiccation de D. deserti et cinétique de réparation de l’ADN après stress. (A-C) Survie de D. deserti ( ) et de D. radiodurans („) aux UV (A), aux rayons gamma (B) et à la dessiccation (C) (A. de Groot, communication personnelle). (D-E) Gel PFGE montrant la reconstitution du génome de D. deserti après irradiation gamma à 6,8 kGy (D) et 27 jours de dessiccation (E). Après irradiation gamma ou dessiccation, les cellules ont été ressuspendues dans du TSB/10 et des échantillons ont été prélevés à différent temps. L’ADN a été digéré par PmeI et SwaI. Les images D et E proviennent de de Groot et al. (2009).

59 3) Plan de la thèse

Le sujet de ma thèse est l’étude des mécanismes de l’extrême tolérance aux radiations et à la dessiccation de la bactérie Deinococcus deserti par une approche de génomique fonctionnelle. Pour pouvoir réaliser ce travail, il fallait dans un premier temps annoter et analyser le génome de D. deserti pour pouvoir trouver des cibles intéressantes à étudier, puis dans un second temps, mettre au point des outils génétiques afin de les étudier. Ainsi, le plan de ma thèse suit les différentes étapes nous ayant conduit à l’étude de cibles particulières à D. deserti ou spécifiques au genre Deinococcus. Outre le Matériels et Méthodes, la première partie de ma thèse présente ma participation à l’annotation du génome de D. deserti et contient l’article consacré au génome de D. deserti ainsi qu’une liste de plusieurs gènes potentiellement impliqués dans la radiotolérance. La deuxième partie explique la mise au point des outils génétiques. La troisième et la quatrième partie portent respectivement sur (1) l’étude de gènes supplémentaires de réparation de l’ADN trouvés chez D. deserti (codant des polymérases translésionnelles et des RecA) et (2) sur l’étude structure/fonction du régulateur IrrE et elles comportent chacune un article. Enfin, une discussion générale et les perspectives sont décrites dans le dernier chapitre.

60

II Matériels et Méthodes

61 62 II) Matériels et Méthodes

Souches, plasmides, milieux et conditions de culture Les souches et les plasmides utilisés sont listés respectivement dans les tableaux 4 et 5. E. coli est cultivée en milieu riche LB, composé de tryptone (10 g/L), d’extrait de levure (5 g/L), de NaCl (5 g/L), pH 7,3. Toutes les cultures en milieu liquide sont réalisées sous agitation vigoureuse, à 150 rpm, en aérobiose et à une température de 37°C durant 2 à 16 h suivant le stade de croissance voulu. D. deserti est cultivée en milieu trypticase soy broth dilué 10 fois (TSB/10) additionné de 5 % d’une solution d’oligoéléments (sol II) constitué pour 1 L d’eau : EDTA (acide éthylène diamine tétra acétique) (Titriplex II), 400 mg ;

MgSO4.7H2O, 2 g ; CaCl2.2H2O, 2 g ; FeSO4.7H2O, 140 mg ; 20 mL d’une solution d’oligoéléments (composée pour 1 L d’eau : ZnSO4.7H2O, 430 mg ; MnSO4.H2O, 1,3 g ;

Na2MoO4.2H2O, 740 mg ; H3BO3, 2,8 g ; CoSO4, 70 mg ; CuSO4.5H2O, 26 mg). Toutes les cultures en milieu liquide sont réalisées sous agitation vigoureuse à 150 rpm, en aérobiose et à une température de 30°C durant 6 à 64 h suivant le stade de croissance voulu et le type d’expérience. Quand nécessaire, les antibiotiques sont ajoutés selon les concentrations suivantes pour E. coli : ampicilline, 100 μg/mL ; chloramphénicol (Cm), 25 μg/mL ; kanamycine (Kan), 50 μg/mL ; tétracycline, 10 μg/mL. Pour les Deinococcus : acide nalidixique (Nx) 15 μg/mL ; chloramphénicol 2 à 3 μg/mL ; kanamycine 3 à 20 μg/mL ; rifampicine (Rif) 10 μg/mL ; streptomycine (Sm) 10 μg/mL. Les milieux solides sont obtenus en ajoutant 16 g/L d’agar à chacun des milieux liquides. La mesure de la turbidité des suspensions à 600 nm (DO600) 8 permet d’estimer la concentration bactérienne. À une DO600 de 0,4 correspond 3.10 CFU/mL pour E. coli et 9.107 CFU/mL pour D. deserti.

Extraction d’ADN chromosomique 5-10 mL de culture en phase stationnaire de D. deserti sont centrifugés à 10000 g pendant 10 min. Le culot est re-suspendu dans 500 μL de Tris 50 mM, EDTA 50 mM, NaCl 0.15 M, pH 8. 30 μL de lysozyme 10 mg/mL sont ajoutés à la suspension, le mélange est incubé 1 h à 37°C. Après addition de 100 μL de SDS 10 %, 15 μL de protéinase K à 20 mg/mL et 15 μL de pronase E à 20 mg/mL, la suspension est incubée 1h à 37°C. 1 volume de phénol-chloroforme-isoamyl alcool (25 : 24 : 1) est ajouté et vortexé. Le mélange est centrifugé 5 min à 16100 g puis une seconde étape phénol-chloroforme-isoamyl alcool est réalisée. L’ADN est précipité par l’ajout de 1 volume d’isopropanol et centrifugé 30 min à 16100 g. L’ADN est ensuite lavé avec de l’éthanol 70 % puis re-centrifugé 10 min à 16100 g.

63 L’ADN est repris dans 100-200 μL d’eau ultra pure et la concentration est mesurée à DO260 (Biophotometer d’Eppendorf ou au Nanovue de GE Healthcare).

Tableau 4 : Souches utilisées Souche Génotype et caractéristiques Source Deinococcus deserti VCD115 Sauvage (de Groot et al., 2005) VCD115 RifR Mutant spontané RifR de VCD115 A. de Groot RD9 VCD115/pRD9 Ce travail RD12 VCD115 avec insertion de pRD10 au niveau d’amyE Ce travail RD18 VCD115 avec insertion de pRD18 au niveau d’amyE Ce travail RD19 Mutant spontané SmR de VCD115 Ce travail RD20 VCD115 avec insertion de pRD20 au niveau d’amyE Ce travail RD22 RD19 Ptuf::kan inséré au niveau d’amyE Ce travail RD31 RD19 ΔimuYΩkan Ce travail RD38 RD19 ΔpolBΩkan Ce travail RD40 RD19 ΔrecACΩkan Ce travail RD41 RD19 Δ(imuY-dnaE2)Ωkan Ce travail RD42 RD19 ΔirrEΩkan Ce travail RD44 RD19 ΔrecAP3Ωcat Ce travail RD45 RD19 ΔdnaE2Ωkan Ce travail RD47 RD19 ΔrecAP1Ωkan Ce travail RD4046 RD19 ΔrecACΩkan ΔrecAP3Ωcat Ce travail RD4076 RD40/pRD76 Ce travail RD4746 RD19 ΔrecAP1Ωkan ΔrecAP3Ωcat Ce travail RD4753 RD19 ΔrecAP1Ωkan ΔrecACΩcat Ce travail RD51 RD42/pRD51 Ce travail RD52 RD31/pRD52 Ce travail RD55 RD42/pRD55 Ce travail RD56 RD42/pRD56 Ce travail RD57 RD42/pRD57 Ce travail RD58 RD42/pRD58 Ce travail RD59 RD42/pRD59 Ce travail RD60 RD42/pRD60 Ce travail RD61 RD19 ΔDeideP302150Ωkan Ce travail Deinococcus radiodurans R1 (Anderson, 1956) Deinococcus grandis DSM 3963 DSMZ Escherichia coli Top10 F- mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZΔM15 ΔlacX74 recA1 Invitrogen araΔ139 Δ(ara-leu)7697 galU galK rpsL (SmR) endA1 nupG SCS110 rpsL (SmR) thr leu endA thi-1 lacY galK galT ara tonA tsx dam dcm Stratagene supE44 Δ(lac-proAB) [F' traD36 proAB lacIqZΔM15] cat : gène de résistance au chloramphénicol, kan : gène de résistance à la kanamycine, RifR : résistant à la rifampicine, SmR : résistant à la streptomycine.

64 Tableau 5 : Plasmides utilisés Plasmide Description Référence pBBR1MCS Vecteur conjugatif (CmR) (Kovach et al., 1994) pBR322 Vecteur de clonage utilisé chez E. coli (TetR ; ApR) Invitrogen pCR4®BluntTopo® Vecteur pour le clonage des fragments PCR (KmR, ApR) Invitrogen pET-TEV Vecteur d’expression (KmR) (Houben et al., 2007) pGTC101 Source de DrPtuf::cat (Earl et al., 2002b) pI3 Vecteur navette entre E. coli (ApR) et Deinococcus (CmR) (Meima & Lidstrom, 2000) pI8 Vecteur navette entre E. coli (ApR) et Deinococcus (CmR) (Meima & Lidstrom, 2000) pRK2013 Vecteur helper pour la conjugaison (KmR) (Figurski & Helinski, 1979) pUC19 Vecteur de clonage utilisé chez E. coli (ApR) Invitrogen pZT29 Vecteur navette entre E. coli et D. grandis (CmR) (Satoh et al., 2009) pRD0 pUC19 Ptuf::kan Ce travail pRD1 pBR322 Pgro::tet A. de Groot pRD2 pBR322 Pkat::tet A. de Groot pRD3 pBR322 Ptuf::tet A. de Groot pRD4 pZT29 Ptuf::repD Pgro::cat A. de Groot pRD5 pUC19 Ptuf::kan amyE* A. de Groot pRD6 pBR322 Pgro::tet amyE* Ce travail pRD7 pBR322 Pkat::tet amyE* Ce travail pRD8 pBR322 Ptuf::tet amyE* Ce travail pRD6a pCR4®BluntTopo® ardU ; Le fragment cloné contient les sites BamHI et EcoRI Ce travail pRD7a pI8 ardU Ce travail pRD8a pCR4®BluntTopo® ardU ; Le fragment cloné contient les sites HindIII et SalI Ce travail pRD9 pI3 Ptuf::kan Ce travail pRD10 pRD5 ardU Ce travail pRD11 pRD5 amyEUp Ce travail pRD13 pI3 Pkat::tet Ce travail pRD14 pI3 Pgro::tet Ce travail pRD15 pI3 Ptuf::tet Ce travail pRD18 pRD11 ardU Ce travail pRD20 pBBR1MCS Ptuf::kan amyE* Ce travail pRD21 pCR4®BluntTopo® rpsL ; Le fragment cloné contient les sites HindIII Ce travail pRD22 pRD18 rpsL Ce travail pRD25 pRD5 ΔamyE* + Fragment Deide16950 Ce travail pRD26 pRD5 ΔamyE* + Fragment Deide16750 Ce travail pRD27 pRD5 ΔamyE* + Fragment DeideP101880 Ce travail pRD28 pRD20 ΔamyE* + Fragment Deide16750 Ce travail pRD29 pRD20 ΔamyE* + Fragment Deide16950 Ce travail pRD30 pRD20 ΔamyE* + Fragment DeideP101880 Ce travail pRD31 pRD0 avec rpsL et fragments Up/Dn de DeideP101880 Ce travail pRD36 pUC19 DrPtuf::cat Ce travail pRD37 pRD0 avec fragments Up/Dn de polB Ce travail pRD38 pRD0 avec rpsL et fragments Up/Dn de polB Ce travail pRD39 pRD0 avec fragments Up/Dn de recAC Ce travail pRD40 pRD0 avec rpsL et fragments Up/Dn de recAC Ce travail pRD41 pRD0 avec fragments Up d’imuY et Dn de dnaE2 Ce travail pRD42 pRD0 avec fragments Up/Dn d’irrE Ce travail pRD44 pRD36 avec fragments Up/Dn de recAP3 Ce travail pRD45 pRD0 avec fragments Up/Dn de dnaE2 Ce travail pRD46 pRD44 rpsL Ce travail pRD47 pRD0 avec fragments Up/Dn de recAP1 Ce travail pRD48 pCR4®BluntTopo® irrE ; Le fragment cloné contient les sites XbaI et HindIII Ce travail pRD48A pCR4®BluntTopo® irrE (H85A, F48A) L. Serre et A. Vujicic-Zagar pRD48B pCR4®BluntTopo® irrE (H86S) L. Serre et A. Vujicic-Zagar pRD48C pCR4®BluntTopo® irrE (Y160A) L. Serre et A. Vujicic-Zagar pRD48D pCR4®BluntTopo® irrE (E83Q) L. Serre et A. Vujicic-Zagar pRD48E pCR4®BluntTopo® irrE (C175A) L. Serre et A. Vujicic-Zagar pRD48F pCR4®BluntTopo® irrE (H217L) L. Serre et A. Vujicic-Zagar pRD49 pUC19 Ptuf::kan (Cassette clonée en EcoRI/SacI) Ce travail pRD51 pI3 irrE Ce travail pRD52 pI3 lexA-imuY Ce travail pRD53 pRD39 Δ(Ptuf::kan) + DrPtuf::cat Ce travail pRD54 pRD49 avec rpsL et fragments Up/Dn de DeideP100132 Ce travail pRD55 pI3 irrE H35A, F48A Ce travail pRD56 pI3 irrE Y160A Ce travail pRD57 pI3 irrE H217L Ce travail pRD58 pI3 irrE H86S Ce travail pRD59 pI3 irrE E83Q Ce travail pRD60 pI3 irrE C175A Ce travail pRD61 pRD0 avec fragments Up/Dn de DeideP302150 Ce travail pRD76 pI3 recAC avec promoteur de l’opéron Deide19430-Deide19450 Ce travail pET-TEV/LexA lexA cloné dans pET-TEV Ce travail pET-TEV/RecAc recAC cloné dans pET-TEV Ce travail pET-TEV/RecAp recAP1 cloné dans pET-TEV Ce travail Les termes Up et Dn désignent des fragments d’ADN situé respectivement en amont et en aval d’un gène en particulier, par exemple, pour un fragment Up/Dn de DeideP101880, cela signifie que des fragments en amont et en aval du gène DeideP101880 ont été cloné. ApR : résistant à l’ampicilline, CmR : résistant au chloramphénicol, KmR : résistant à la kanamycine, cat : gène de résistance au chloramphénicol, TetR : Résistant à la tétracycline, kan : gène de résistance à la kanamycine, tet : gène de résistance à la tétracycline. aSignifie une erreur dans la numérotation de ces plasmides.

65 Extraction d’ADN plasmidique Le kit QIAprep spin miniprep de QIAGEN est utilisé selon les instructions du fabriquant pour extraire les plasmides construits dans E. coli. Pour D. deserti, le même protocole est utilisé, mais avec les modifications suivantes : le volume de culture utilisé pour l’extraction des plasmides est de 10 mL ; lors de la première étape (re-suspension des bactéries), 20 μL de lysozyme 10 mg/mL sont ajoutés et le mélange est incubé 20-30 min à 37°C ; les volumes, avant de passer le mélange sur colonne, sont doublés.

PCR Pour le clonage de fragments d’ADN, la Pfx polymerase (Invitrogen) et la Phusion (Finnzyme), des enzymes avec une activité proof-reading ont été utilisées. Les protocoles fournis par les fabriquants ont été suivis, sauf pour la Phusion où 5% de DMSO ont été ajoutés afin d’améliorer l’amplification de l’ADN haut GC (D. deserti : 63%). Pour les PCRs de contrôle, les réactions sont réalisées avec la HotGoldstar (Eurogentec) afin de vérifier les constructions des mutants de D. deserti ou l’insertion d’un fragment d’ADN dans un plasmide lorsqu’un seul site de restriction est utilisé. Ces PCRs sont réalisées en suivant le protocole du fabricant avec une modification : l’ajout de 5% de DMSO dans le mix PCR. Les amorces sont listées dans les tableaux 6 à 9. L’ADN génomique de D. deserti a été utilisé comme matrice, sauf indiqué autrement.

66 Tableau 6 : Amorces pour l’amplification des fragments upstream et downstream des gènes à déléter. Amorce Séquence Fragment amplifié AmyUL4 GAGGAATTCCGTGCCATCAGTGAGGTGT Fragment upstream d’amyE (amyEUp) AmyUR3 GCTTGGATCCGTGTTGCGTCAGCCGGTG P101880UpBamHI5’ CTTGGATCCCGTTTCGGCCCCGTGTCTGG Fragment upstream d’imuY P101880UpBamHI3’ CTTGGATCCGCTCAGACCCTCCGGCTCCT P101880DnHindIII5’ GTGAAGCTTCCATGAAGGCCGTACAGCAGC Fragment downstream d’imuY P101880DnHindIII3’ GTGAAGCTTCTGCATTCAATAGGGCGTCCGG PolBUpEcoRI GAGGAATTCGGTGGTCGATGCTGCGAGTGG Fragment upstream de polB PolBUpKpnI CTTGGTACCGGGGAGGTCAGCAGGCCG PolBDnHindIII5’ GTGAAGCTTGGTTCAACCCACTCGCTGCTGTG Fragment downstream de polB PolBDnHindIII3’ GTGAAGCTTGGCAACTGGTGCAGGACCTGC RecAUp3EcoRI GAGGAATTCCTGACCCGGCAGGCCAATCC Fragment upstream de recAC RecAUp3BamHI GCCGGATCCCGGTGGGAGCGGAGAGTTCC RecADnPstI GTGCTGCAGCCAGTCTTCCTTACCCTGGCTTTCG Fragment downstream de recAC RecADnHindIII’ GTGAAGCTTGCTGGTTGTTCGCATTGCCAGGC RecA2UpKpnI CTTGGTACCGGAGCTCCTCGTGACCCCTAC Fragment upstream de recAP1 RecA2UpBamHI CTTGGATCCCAGCCCCAGAACAGGG RecA2DnPstI GTGCTGCAGCGCCTTAATATCAATCCCGCTGCC Fragment downstream de recAP1 RecA2DnHindIII GTGAAGCTTCGTGCCTTTGAGGGAGGCTATGG RecA3DnEcoRI GAGGAATTCCTGGCAGGCTTGCCGATGCG Fragment upstream de recAP3 RecA3DnSacI TACGAGCTCCATGGGGTCTCCGGGTGGG RecA3UpHindIII5’ GTGAAGCTTGGCGCCTTCGGTTACGGCAC Fragment downstream de recAP3 RecA3UpHindIII3’ GTGAAGCTTCGCACCTCCGGGGTCACTG IrrEUp3KpnI CTTGGTACCGCGCAGGATGCTGTGCTTGGC Fragment upstream d’irrE IrrEUp3BamHI CTTGGATCCGCAGAAACAGTGGGTTATCCCGG IrrE3DnPstI AGTCTGCAGCGGAGCGAGCCCGATGCAG Fragment downstream d’irrE IrrE3DnHindIII GTGAAGCTTCTGGCCTCCTCCTGCTCCAG DnaE2Up3EcoRI GAGGAATTCCCGAGCAGCTCCCCAGGTATG Fragment upstream de dnaE2 DnaE2Up3BamHI CTTGGATCCGCTGCACGTACAGGCTGAAGGG DnaE2Dn2HindIII5’ GTGAAGCTTCCAGAGCCTGACGTTGAGGCG Fragment downstream de dnaE2 DnaE2Dn2HindIII3’ GTGAAGCTTGGTGTACCGCGAGGAGGAGC PhotoUpBam CGGGATCCTCCGGCCACTGAGGTGGCTT Fragment upstream de DeideP302150 PhotoUpEco TAGAATTCCCACCATCCGCAAGGGGCAG PhotoDnPst GAACTGCAGCAGGCCGAAGGGCGCTTTCA Fragment downstream de DeideP302150 PhotoDnHdIII TAAAGCTTCGGCAAAAGCACCACGGCAC ExoUp3SacI CTTGAGCTCGGTGGTGTCGAGAGTGTTG Fragment upstream de DeideP100132 ExoUp3KpnI CTTGGTACCAACAGGGCAGTCAAGAAACG ExoDn3 BglII GTTAGATCTGGCTCACTGGCTTCTTGTTC Fragment downstream de DeideP100132 ExoDn3HindIII GTGAAGCTTTCGGCCAGAAACGTAAAGTC

67 Tableaux 7 : Amorces pour l’amplification de gènes entiers, des fragments et des cassettes de résistances aux antibiotiques. Amorce Séquence Utilisation GroEco CTGGAATTCGATATGGGATATGGGGCAC Promoteur Pgro GroPvu CTTCGATCGGAAAATGGACCTGGGACG KatEco CTTGAATTCTCGGAGAGACGGCACTCAG Promoteur Pkat KatPvu GAACGATCGAAACGGCGATTGTCATGCTG TufEco GGTGAATTCACACGCAAAATCGCCCCTT Promoteur Ptuf TufPvu GCGCGATCGTTGTAATTGCCGTGCTGAG AmyPst GAACTGCAGCTTTACTGCTGGCTGCCTT AmyPvu TATACGATCGAACACGTCGCGCTCTG Fragment amyE (amyE*) AmyHd3 GGGAAGCTTCGAACACGTCGCGCTC AmyUL3 GCAGCGTGCCATCAGTGAGGTGT Amplification du fragment amyEUp- AmyDnR CCGCTTTCCGTGACTACAGC Ptuf::kan-amyEDn TufXho GAACTCGAGACACGCAAAATCGCCCCTT TufGro GGTCCATTTTCTTGTAATTGCCGTGCTGAG Fragment avec Ptuf et Pgro pour pRD4 GroTuf GGCAATTACAAGAAAATGGACCTGGGACG GroERV CTGGATATCGATATGGGATATGGGGCAC ArdEco TAGAATTCAGCAAGACCCCGCCGCCTAC ArdBam CGGGATCCTCATGCGTGACCGTTGAAG ardU ArdHd3 TAAAGCTTAGCAAGACCCCGCCGCCTAC ArdSal CGGTCGACTCATGCGTGACCGTTGAAG RpsLleftHdIII GAGAAGCTTCACCTCGGGCGTTTCTCGGCA rpsL RpsLrightHdIII GAGAAGCTTGACCCGACAGGGGTCAGGCCA DeideP101880I5’ GTGAAGCTTCGAGTCTGATCGCCTGCGTCC Fragment pour l’insertion d’un plasmide au DeideP101880I3’ AGTCTGCAGAGCGGACCGTCAGGTACGCGG niveau du gène DeideP101880 Deide16750I5’ GTGAAGCTTCCGGGAACTGCTGCCGGTGC Fragment pour l’insertion d’un plasmide au Deide16750I3’ AGTCTGCAGGGGCCGGAAGGCATCAAGCGC niveau du gène Deide16750. Deide16950I5’ GTGAAGCTTGTTGCAGCGCGACGCTCAGG Fragment pour l’insertion d’un plasmide au Deide16950I3’ AGTCTGCAGGGATCAGGTCATGGCCCGGATG niveau du gène Deide16950. GroXba CTTCTAGAGAAAATGGACCTGGGACG TufXba CTTCTAGATTGTAATTGCCGTGCTGAG KatXba CTTCTAGAAAACGGCGATTGTCATGCTG Clonage des cassettes de sélection dans le pI3 TetSac CTGAGCTCTTCAGGTCGAGGTGGCC KanHd3 GTTAAGCTTTCAGAAGAACTCGTCAAG TufBam GAAGGATCCTTGTAATTGCCGTGCTGAG TufKan CTTGTTCAATCATGTTCTTACTCCCTCCAAG KanTuf GGAGTAAGAACATGATTGAACAAGATGGATTG Construction et amplification de Ptuf::kan KanPst GTTCTGCAGTCAGAAGAACTCGTCAAG TufEcoRI GAGGAATTCTTGTAATTGCCGTGCTGAG KanSacI TACGAGCTCTCAGAAGAACTCGTCAAG DRTuf1Bam TAAGGATCCTCTGACGGCGCTGCTGATTC Amplification de DrPtuf::cat CatPst ATACTGCAGTTACGCCCCGCCCTGCCA Lexpol-XbaFW2 GAAATCTAGAGCACGGAAGGCCTGATGC Amplification de lexA-imuY pour le clonage Lexpol-Hd3RV2 GATAAAGCTTCCTGCTGCTGTACGGCCT dans pI3 IrrEXbaFW GAATCTAGAGCTCAGCGGCAGTAAACC Amplification de irrE pour le clonage dans pI3 IrrEHd3RV TATAAAGCTTTCAGGACTGGTCCCCTGG

RecAPromoXbaI CATCTAGAACGCCGCGCGGCTCTGTGC Amplification du promoteur de recAC pour le RecAPromoEcoR1 TAGAATTCCACGGGCCAGGAAGGCAGC clonage dans pI3

RecAG4EcoRI TAGAATTCGTGCTGTACCGCAGCATCCT Amplification de recAC pour le clonage dans RecAGene2Hd3 TAAAGCTTTGGCACGAAAGCCAGGGTAAGG pI3

5DdrecACNdeI CAGCCATATGAGCAAAGACAGCACCAAG Amplification de recAC pour le clonage dans 3DdrecACNew TATACTCGAGTTACTCGGCCAGGGCCGGT pET-TEV

5DdrecAPNdeI CAGCCATATGGATACCGACCAGCACAAAG Amplification de recAP1 pour le clonage dans 3DdrecAPXhoI TATACTCGAGTCAGTCGGCAGAGAGGGC pET-TEV

5DdlexAP1NdeI TATACATATGCCCCCCGAACTGACCCGCA Amplification de DeideP101870 (lexAP1) pour 3DdlexAP1XhoI TATACTCGAGTCAGACCCTCCGGCTCCT le clonage dans pET-TEV

68 Tableau 8 : Amorces utilisées pour les vérifications des mutants construits. Amorce Séquence Gène ou locus Pour vérifier la présence/absence des gènes TufBam GAAGGATCCTTGTAATTGCCGTGCTGAG Ptuf::kan KanPst GTTCTGCAGTCAGAAGAACTCGTCAAG DRtuf1Bam TAAGGATCCTCTGACGGCGCTGCTGATTC DrPtuf::cat CatPst ATACTGCAGTTACGCCCCGCCCTGCCA DeideP101880I5 GTGAAGCTTCGAGTCTGATCGCCTGCGTCC imuY DeideP101880I3 AGTCTGCAGAGCGGACCGTCAGGTACGCGG PolB5’ CGGACGCGACACTCTCTGGAC polB PolB3’ GAACAGGTGAACCGCTCCGCC RecABegin CCCTTGCCGAAGGCCTTCTCG recAC RecAEnd GCGTCCCGAAATGGAGCAGGAG RecA2Begin GTCGAGGCTGAGGCTGCCG recAP1 RecA2End GAGCGTGTGCTGGCCCAGATG RecA3Begin CTGAACCCGCCCTCTCTGCC recAP3 RecA3End GGTGCTGATGGCCTGGACGTC IrrE5’ GCGCTGGCCGCAGCAAAAGC irrE IrrE3’ GCCGCTCCGGGAAAGCGTC DnaE25’ GGATTACGGGCACCTGCGCC dnaE2 DnaE23’ GGTTCAAACCGCCGAGCGCC Photo5’ ATAGGCGGCCGTCCAACCCT DeideP302150 Photo3’ ACGGCCGCAGTGCAGACTTC Exo5’ ACACCCGGAATTTCAGACAG DeideP100132 Exo3’ GTACACCGGGGATATTGTCG Pour vérifier la double recombinaison homologue à un locus donnéa KanTuf GGAGTAAGAACATGATTGAACAAGATGGATTG TufBamComp2 CTCAGCACGGCAATTACAAGGATCC KanPstComp CTTGACGAGTTCTTCTGACTGC Servent à plusieurs PCR DRtuf1BamComp GAATCAGCAGCGCCGTCAGAGGATCC CatPstComp2 TGGCAGGGCGGGGCGTAACTGCAG AmyDnLComp CACAGCAGGACGCTGGTTTGG AmyUR3Comp CACCGGCTGACGCAACACCTG amyE ; Servent à plusieurs PCR AmyEUpUp GGAAGCCCGCGTCCTTGTTGC Vérification de la présence/absence de pUC19 Puc19A CACACAGGAAACAGCTATGACC lors de la mise au point des outils génétiques P101880UpUp CCAGCCAGGCAGGAATGAGGG imuY P101880DnDn CTCGAAGGTGCACCAGTCTGCG PolBUpUp GCCATCCTGCCTCCAGAAGAGG polB PolBDnDn GCGCACGGTCACGAGCTATACC RecAUpUp GGTGATCGTGGCGCTGCCAG recAC RecADnDn CATGCTGGTCGGGGCTGCTG RecA2UpUp GCTTGCGCATCACCGGCCC recAP1 RecA2DnDn CATCGCTCAGGAGCGGGGTG RecA3UpUp2 CGTTCCCGCAGCCCTAACGG recAP3 RecA3DnDn2 GCGTGATGGCCCCTTCTCTCC IrrEUpUp GCTGAGGCGCGAAAGTGTGGG irrE IrrEDnDn CCTGGTCCTCCAGACGGTGC DnaE2UpUp CGTGAAGGTGGGCATGCGGG dnaE2 DnaE2DnDn GGCTGACCCGCTTCGTGAGC PhotoUpUp2 GCTTGGCGTCTTCAGGGGACG DeideP302150 PhotoDnDn2 GGGCTGGCATGGATGGGAAGG ExoDnDn AGACCTCGCGCACCTTACT Exo5’compl CTGTCTGAAATTCCGGGTGT Exo15’ GCTTGGCCTGGGAGTCTCGG DeideP100132 Exo3’compl CGACAATATCCCCGGTGTAC Exo13’ GACTCCACGTACGCGCACGG aPour vérifier la double recombinaison à un locus donné, une amorce UpUp a été utilisée avec TufBamComp2 ou DRtuf1BamComp, et une amorce DnDn a été utilisé avec KanTuf, KanPstComp ou CatPstComp2.

69 Tableau 9 : Amorces pour les RT-PCR Gène Amorce Séquence Dd06120FW TTTGAGCGCACGAAGCCCCA Deide06120 (tufA) Dd06120RV TGACGGGTCGGCGTGTTGTA recAC-FW2 AGCTACGGTGACGAGCGCAT Deid19450 (recAC) recAC-DN AGCCAGGGTAAGGAAGACTGGA recAP-FW2 AGCTACGGGGACGAACGCAT DeideP101260 (recAP1) recAP1-DN GGGCAGCGGGATTGATATTAAGG recAP-FW2 AGCTACGGGGACGAACGCAT DeideP300210 (recAP3) recAP3-DN GCCATACCAGTGACCCCGGA DdP100180FW2 ACCTGCCGTTTCTGCATTAC DeideP100180 (polB) DdP100180RV2 CCAAATAGACCCGTCCTTCA DdP101880FW2 TGCGTGCCAAAGCGGGAGAA DeideP101880 (imuY) DdP101880RV2 TCGGTGCGGTGCCCTTTCAA DdP101900FW TTGCGGTCCATTACGCCGGA DeideP101900 (dnaE2) DdP101900RV AAATGCGTGGGCGTGCGACT DeideP302150 (photoproduct-lyase- DdP302150FW TTTTCCTGACGCCCAGCGCA like) DdP302150RV AGGACAGCGCGCACCCATTT DdP100132FW TGTGGTCATTGCAGCACCGCT DeideP100132 () DdP100132RV TTCAGCGGCCTGTGTCAGCA Dd20570FW GCAGCCCAGGAGGTGAATAC Deide20570 (ddrOC) Dd20570RV CAGAGCACCTTCGGTGGAGG DdP302170FW GAAGGGTTTAAGCTGAGGACAT DeideP302170 (ddrOP3) DdP302170RV AAGGGCTCCGGCGGTCTGAT Dd18730FW TTGCTGCACCTGCTCTGCGA Deide18730 Dd18730RV ACGCCCGGCCACAGCAAAAT Dd02990FW TAAATGGACCAGCCCCGACGGT Deide02990 (ddrB) Dd02990RV AACGCTCCTGGCGTTTTCCG DdP101280FW TCGCAGACGCCAGGCATTGT DeideP101280 (RecQ-like) DdP101280RV AGCGCACGTTGGGCTTGTCA

Clonage dans pCR4®BluntTopo® Tous les fragments PCR, ont été clonés dans le vecteur pCR4®BluntTopo® du « Zero Blunt® TOPO® PCR Cloning Kit for sequencing » d’Invitrogen, en respectant le protocole fourni par le fabriquant.

Digestion des plasmides Dans un volume final de 20-30 μL, 0,5-1 μg de plasmide est digéré par 1-2 U d’enzyme de restriction adéquate (Biolabs) à la température adéquate durant 3-16 h. L’enzyme est inactivée à 65°C pendant 20 min.

Déphosphorylation 0,5-1 μg de plasmide digéré par 1 ou 2 enzyme(s) est déposé sur gel puis purifié grâce au « QIAquick® PCR purification kit » de QIAGEN selon les instructions fournies par le fabricant. L’ADN est ensuite déphosphorylé en présence de 1 unité « d’alkaline shrimp » (Roche) durant 10-20 min à 37°C. Le mélange réactionnel est ensuite placé pendant 15-20 min à 65°C pour inactiver la phosphatase.

70 Ligation En respectant le rapport 1 : 3 à 1 : 4 (pour vecteur : insert), 10-20 ng de vecteur sont ajoutés à la quantité adéquate d’inserts avec 2,5 U de T4 DNA ligase (Roche ou Biolabs) et 2 μL de tampon T4 DNA ligase 10X (Roche) pour un volume final de 20 μL. Le mélange de ligation est incubé 1h à température ambiante puis la nuit à 4°C.

Constructions des plasmides - Plasmides pRD1 à pRD3 : Les fragments d’ADN Ptuf (344 pb), Pgro (275 pb) et Pkat (296 pb) contiennent respectivement les promoteurs des gènes codant EF-Tu (Deide18970), GroES (Deide22580) et Catalase (DeideP200330). Les fragments obtenus par PCR avec les amorces TufPvu et TufEco pour Ptuf, GroPvu et GroEco pour Pgro, et KatPvu, et KatEco pour Pkat, ont été introduits dans le vecteur pBR322 au niveau des sites PvuI- EcoRI. - Plasmide pRD4 : Un fragment d’ADN contenant Pgro et Ptuf en orientation inverse a été construit par PCR en 2 étapes. D’abord, avec les amorces GroERV et GroTuf pour Pgro, et TufXho et TufGro pour Ptuf. Les parties 5’ des amorces GroTuf et TufGro sont complémentaires, et une seconde PCR a pu être réalisée avec les amorces GroERV et TufXho en utilisant comme matrice (et amorces) les 2 produits PCR précédemment construits. Le produit final a été cloné comme fragment XhoI-PmeI (PmeI du vecteur pCR4®BluntTopo®) dans les sites XhoI-EcoRV du plasmide pZT29. - Plasmides pRD6 à pRD8 : Un fragment (1120 pb) d’amyE (amyE*), synthétisé par PCR avec les amorces AmyPst et AmyPvu a été introduit au niveau des sites PstI-PvuI des plasmides pRD1 à 3 donnant respectivement pRD6, pRD7 et pRD8. - Plasmides pRD0, pRD5, pRD10, pRD11, pRD18 et pRD22 : Un fragment d’ADN contenant la fusion traductionnelle Ptuf::kan a été construit par A. de Groot en 2 étapes. D’abord, avec les amorces TufBam et TufKan pour Ptuf, et KanTuf et KanPst pour kan (la matrice utilisée pour kan est le plasmide pCR4®BluntTopo®). Les parties 5’ des amorces TufKan et KanTuf sont complémentaires, et une seconde PCR a pu être réalisée avec les amorces TufBam et KanPst en utilisant comme matrice (et amorces) les 2 produits PCR précédemment construits. Le produit final (Ptuf::kan) a été introduit au niveau des sites BamHI-PstI du plasmide pUC19 pour former pRD0. Ensuite le fragment amyE* obtenu par PCR avec les amorces AmyPst et AmyHd3 a été introduit au niveau des sites PstI-HindIII. Ceci a permis d’obtenir pRD5. À ce plasmide ont été ajoutés, les fragments d’ADN ardU ou amyEUp, obtenus par PCR avec les amorces ArdEco et ArdBam pour ardU, et AmyUL4 et AmyUR3 pour amyEUp, au niveau des sites EcoRI-BamHI pour donner respectivement les plasmides pRD10 et pRD11. Après digestion de pRD8* (pCR4®BluntTopo® ardU) par

71 EcoRI, le fragment ardU, est cloné dans pRD11 au niveau du site EcoRI pour donner pRD18. Enfin, rpsL synthétisé par PCR avec les amorces RpsLLeftHdIII et RpsLRightHdIII est introduit au niveau du site HindIII de pRD18 pour former pRD22. Le gène rpsL est également cloné dans le vecteur pCR4®BluntTopo® pour donner le plasmide pRD21. - Plasmides pRD9 et pRD13 à pRD15 : les cassettes Ptuf::kan, Ptuf::tet, Pgro::tet et Pkat::tet ont été synthétisés par PCR avec les amorces TufXba et KanHd3 pour Ptuf::kan, TufXba et TetSac pour Ptuf::tet, GroXba et TetSac pour Pgro::tet, et KatXba et TetSac pour Pkat::tet. Une fois cloné dans le plasmide pCR4®BluntTopo®, une digestion par XbaI-SacI pour Ptuf::tet, Pgro::tet et Pkat::tet et XbaI-HindIII pour Ptuf::kan a été fait pour ressortir les fragments et les cloner dans pI3 digéré également par XbaI-SacI ou XbaI-HindIII. - Plasmide pRD20 : Afin de sortir le fragment Ptuf::kan-amyE* de pRD5, ce plasmide est digéré par HindIII-BamHI. Ptuf::kan-amyE* est ensuite cloné dans le plasmide pBBR1MCS au niveau des sites HindIII-BamHI pour former pRD20. - Plasmides pRD25 à pRD27 : Le plasmide pRD5 est digéré par PstI-HindIII afin d’enlever le fragment amyE* et de le remplacer par les fragments des gènes Deide16950 (synthétisé par PCR avec les amorces Deide16950I5’ et Deide16950I3’), Deide16750 (synthétisé par PCR avec les amorces Deide16750I5’ et Deide16750I3’) ou DeideP101880 (synthétisé par PCR avec les amorces DeideP101880I5’ et DeideP101880I3’). La taille du fragment de DeideP101880 est de seulement 505 pb car le fragment amplifié d’une taille de 825 pb contenait un site PstI. - Plasmides pRD28 à pRD30 : Les fragments Deide16950, Deide16750 ou DeideP101880 des plasmides pRD25, pRD26, pRD27 sont clonés dans le plasmide pRD20 au niveau des sites HindIII-PstI. - Plasmide pRD31 : pRD0 est digéré par HindIII pour cloner le fragment downstream à DeideP101880, DeideP101880Dn (amorces P101880DnHind5’ et P101880DnHind3’). Le plasmide qui en découle est digéré par BamHI pour y cloner le fragment upstream à DeideP101880, DeideP101880Up (amorces P101880UpBam5’ et P101880UpBam3’). Enfin, rpsL, ressorti du plasmide pRD21 par EcoRI, est cloné en EcoRI. - Plasmide pRD36 : La cassette DrPtuf::cat est amplifiée (amorces CatPst et DRtuf1Bam) en utilisant comme matrice le plasmide pGTC101, et est cloné dans le plasmide pUC19 au niveau des sites BamHI-PstI. - Plasmides pRD37 et pRD38 : pRD37 est construit en clonant au niveau des sites EcoRI-KpnI le fragment polBUp (amorces PolBUpEcoRI5’ et PolBUpKpnI3’) puis au niveau de HindIII, le fragment polBDn (amorces PolBDnHindIII5’ et PolBDnHindIII3’). pRD38 est construit en ajoutant rpsL au niveau du site EcoRI de pRD37.

72 - Plasmides pRD39 et pRD40 : pRD39 est construit à partir de pRD0 digéré par

EcoRI-BamHI où est cloné le fragment recACUp (amorces RecAUp3EcoRI5’ et

RecAUp3BamHI3’) puis par PstI-HindIII où est cloné le fragment recACDn (amorces RecADnPstI5’et RecADnHindIII3’). pRD40 est construit en ajoutant rpsL à pRD39 au niveau du site EcoRI. - Plasmide pRD41 : Le fragment dnaE2Dn, synthétisé par PCR avec les amorces DnaE2Dn2HindIII3’ et DnaE2Dn2HindIII5’ a été cloné au niveau du site HindIII du vecteur pRD0. Puis le plasmide a été digéré par BamHI pour y cloner le fragment DeideP101880Up (amorces P101880UpBam5’ et P101880UpBam3’). - Plasmide pRD42 : Le plasmide pRD0 est digéré par HindIII-PstI afin de cloner le fragment IrrEDn, synthétisé par PCR avec les amorces IrrE3DnPstI5’ et IrrE3DnHindIII3’. Ensuite, le plasmide nouvellement construit est digéré par BamHI-KpnI afin de cloner le fragment IrrE3Up, synthétisé par PCR avec les amorces IrrEUp3BamHI et IrrEUp3KpnI. - Plasmides pRD44 et pRD46: Le plasmide pRD36 est digéré par EcoRI-SacI afin de cloner le fragment RecAP3Up, synthétisé par PCR avec les amorces RecA3DnEcoRI et

RecA3DnSacI. Ensuite, le plasmide est digéré par HindIII pour cloner le fragment RecAP3Dn, synthétisé par les amorces RecA3UpHindIII3’ et RecA3UpHindIII 5’. Le plasmide pRD46 est construit en clonant au niveau du site EcoRI de pRD44, le gène rpsL. - Plasmide pRD45 : Le plasmide pRD0 est digéré par EcoRI-BamHI afin de permettre le clonage du fragment dnaE2Up, synthétisé par PCR avec les amorces DnaE2Up3EcoRI et DnaE2Up3BamHI. Ensuite, le plasmide est digéré par HindIII afin de cloner dnaE2Dn, synthétisé par les amorces DnaE2Dn2HindIII3’ et DnaE2Dn2HindIII5’. - Plasmide pRD47 : Le plasmide pRD0 est digéré par PstI-HindIII pour cloner le fragment recAP1Dn, synthétisé par PCR avec les amorces RecA2DnPstI et RecA2HindIII.

Ensuite le plasmide est digéré par BamHI-KpnI afin de cloner le fragment recAP1Up, synthétisé par PCR avec les amorces RecA2UpBamHI et RecA2UpKpnI. - Plasmide pRD48 : Le gène irrE avec son promoteur est synthétisé par PCR avec les amorces IrrEXbaFW et IrrEHd3RV et le fragment d’ADN est cloné dans le plasmide pCR4®BluntTopo®. Les plasmides pRD48A à pRD48F ont été réalisés par l’équipe de L. Serre à l’IBS Grenoble et comporte différentes mutations ponctuelles dans irrE (H35A et F48A ; H86S ; Y160A ; E83Q ; C175A ; H217L). - Plasmide pRD49 : Le plasmide pUC19 est digéré par EcoRI-SacI pour y cloner la cassette Ptuf::kan amplifiée avec les amorces TufEcoRI et KanSacI. - Plasmide pRD51 : Le fragment XbaI-HindIII de pRD48 comportant irrE et son promoteur est cloné dans les sites XbaI-HindIII de pI3.

73 - Plasmide pRD52 : Le fragment comportant le promoteur de l’opéron lexAP1-imuY- dnaE2 et les gènes lexA-imuY, synthétisé par PCR avec les amorces lexpol-XbaFW2 et lexpol-Hd3RV2, est cloné dans les sites XbaI-HindIII du plasmide pI3. - Plasmide pRD53 : La cassette DrPtuf::cat sortie du plasmide pRD36 par digestion avec les enzymes BamHI-PstI est clonée dans pRD40. - Plasmide pRD54 : Le fragment d’ADN correspondant à DeideP100132Up a été synthétisé par PCR avec les amorces ExoUp3SacI et ExoUp3KpnI puis cloné en SacI-KpnI dans le plasmide pRD49. Ensuite le vecteur a été digéré par BamHI-HindIII afin d’introduire le fragment DeideP100132Dn, synthétisé par PCR avec les amorces ExoDn3BglII (les sites BglII et BamHI sont compatibles) et ExoDn3HindIII. Enfin, rpsL a été cloné en HindIII. - Plasmides pRD55 à pRD60 : Les fragments XbaI-HindIII des plasmides pRD48A à pRD48F avec les différentes mutations d’irrE ont été clonés dans le plasmide pI3. - Plasmide pRD61 : Le plasmide pRD0 est digéré par HindIII-PstI afin de permettre le clonage du fragment DeideP302150Dn, synthétisé par PCR avec les amorces PhotoDnHdIII et PhotoDnPstI. Ensuite, le plasmide est digéré par EcoRI-BamHI afin de cloner DeideP302150Up, synthétisé par les amorces PhotoUpBamHI et PhotoUpEcoRI. - Plasmide pRD76 : Deux fragments PCR comportant respectivement le promoteur de l’opéron Deide19430-Deide19450 avec les 85 premières paires de base de Deide19430 (amorces RecAPromoXbaI et RecAPromoEcoRI ; digéré avec XbaI-EcoRI) et les 96 dernières paires de base de Deide19440 avec l’ensemble du gène recA (Deide19450) (amorces RecAG4EcoRI et RecAG3Hd3 ; digéré par EcoRI-HindIII), sont clonés dans les sites XbaI et HindIII du plasmide pI3. -Plasmides dérivés de pET-TEV : pour toutes les constructions dans ce vecteur, les gènes (recAC, recAP1 et lexA) sont clonés dans les sites NdeI et XhoI. Pour chaque construction, lorsqu’un fragment d’ADN est cloné dans un seul site de restriction, le sens de celui-ci est vérifié, sauf lorsque cela concerne rpsL puisque ce gène n’est pas conservé après double recombinaison homologue.

Transformation des bactéries

Cellules compétentes et transformation par choc thermique

100 mL de milieux de culture sont ensemencés à DO600= 0,05 à partir d’une préculture en phase stationnaire. Les cellules sont prélevées à DO600 = 0,3-0,4, puis centrifugées à 4°C durant 10 min à 7000 g. Le culot est repris par 50 mL de CaCl2 100 mM froid. La suspension est incubée dans la glace pendant une nuit puis centrifugée à 4°C durant 10 min à 7000 g. Enfin, le culot est repris par 1/15 du volume initial par une solution froide de glycérol 20 %,

CaCl2 100 mM puis aliquoté par 150 μL et conservé à -80°C. 74 L’ADN (>500 ng pour D. deserti) à introduire est incubé avec 150 μL de cellules compétentes, décongelées sur la glace, durant 30 min à 4°C, puis un choc thermique est effectué à 42°C durant 2 min. 0,5 mL de LB ou de TSB/10 sol II est ajouté à la suspension cellulaire. Le mélange est incubé ensuite 1 h à 37°C (E. coli) ou 6 h à 30°C (D. deserti). Les suspensions bactériennes sont étalées sur milieu de sélection et incubées une nuit à 37°C (E. coli) ou 3 jours à 30°C (D. deserti).

Cellules compétentes et transformation par électroporation

100 mL de milieux de culture sont ensemencés à DO600= 0,05 à partir d’une préculture en phase stationnaire. Les cellules sont prélevées à DO = 0,3-0,4, puis centrifugées à 4°C durant 10 min à 7000 g. Le culot est successivement lavé par 50 mL, 25 mL et 10 mL de glycérol 10% froid. Le culot est repris par 250 μL d’une solution froide de glycérol 10% puis aliquoté par 40 μL et conservé à -80°C. Les bactéries sont décongelées sur la glace puis l’ADN est ajouté. Le mélange est introduit dans une cuve à électroporation de 1 mm et électroporé avec les paramètres suivant : 18 kV/cm, 25 μF, 600Ω. Les cellules électroporées sont ensuite reprises par 1 mL de TSB/10 sol II et incubées 20 h à 30°C. Les suspensions bactériennes sont étalées sur milieu de sélection et incubées 3 jours à 30°C.

Conjugaison Trois cultures en phase exponentielle de croissance de D. deserti, E. coli comportant pRK2013 (helper) et d’E. coli Top10 comportant le plasmide à conjuguer (donneuse) sont mélangées puis déposées sur une boîte TSA/10 sol II avec une quantité de cellules respectivement égale à 80, 10 et 10 μL. Les cellules sont incubées 1 nuit à 30°C puis récoltées dans 500 μL de TSB/10 sol II. 50 μL sont étalés sur boîte TSA/10 sol II avec Nx 15 μg/mL + kan 3 μg/mL et incubé 3 jours à 30°C. D. deserti est naturellement résistante à l’acide nalidixique, ainsi son utilisation permet de sélectionner uniquement D. deserti et d’empêcher la croissance des 2 souches d’E. coli.

Survie bactérienne aux UV Les différentes souches de D. deserti sont précultivées durant 3 jours à 30°C. Ces précultures sont utilisées pour lancer les cultures (50 mL) qui seront irradiées aux UV. La culture est lancée à DO = 0,05 et est cultivée jusqu’à DO = 0,2-0,5. Les cultures sont ensuite diluées par cascade de 10 en 10 jusqu’à 10-5. Pour les irradiations en liquide, 4 mL des dilutions sont irradiés, dans des boîtes de Petri vides et sans couvercle, de 200 à 3000 J/m2 d’UV-C (254 nm). Puis 100 μL de chaque dilution sont étalés sur boîtes TSA/10 sol II et ceci est fait en duplicat. Pour les irradiations sur boîtes, les différentes dilutions sont déposées en

75 gouttes de 20 μL sur milieu TSA/10 + sol II et irradiées aux UV-C (254 nm) ou UV-A (365 nm) entre respectivement 100-750 J/m2 et 2-100 kJ/m2. Le taux de survie est ensuite calculé par le rapport « nombre de CFU/mL après irradiation » / « nombre de CFU/mL avant irradiation ».

Mutagenèse induite aux UV Les mêmes cultures que lors des tests de survie aux UV sont utilisées pour le test de mutagenèse induite par les UV. 4 mL de culture non dilués sont irradiés à 0, 500 et 1000 J/m2 d’UV-C (254 nm). Ensuite les 4 mL sont récupérés et placés à 30°C sous agitation durant 20 h afin de permettre aux mutations de se stabiliser. 1 mL de cette culture est étalé sur milieu gélosé avec Rif 10 μg/mL. En parallèle, sont étalées sur milieux de culture sans rifampicine (mais avec chloramphénicol si nécessaire), des bactéries issues de cette culture afin de déterminer le nombre total de CFU/mL. L’ensemble est fait en duplicat. Le nombre de mutants RifR est observé après une incubation de 3-4 jours à 30°C. La fréquence de mutation est ensuite calculée par le rapport « nombre de CFU RifR/mL » / « nombre total de CFU/mL ».

Survie bactérienne après dessiccation Des portions de 100 μL de cellules en phase stationnaire de croissance sont déposées sur une boîte de Petri stérile et vide, séchées et maintenues à température ambiante dans un dessiccateur scellé contenant du CaSO4. Le produit séchant est imprégné de CoCl2, un indicateur coloré. Le CoCl2 est bleu lorsqu’il est anhydre alors qu’il est rose en présence d’eau. Les cellules sont resuspendues et le nombre de CFU/mL est comptabilisé entre 0 et 130 jours. La survie est calculée par le rapport CFU/mL au temps tx / CFU/mL au temps t0.

Survie bactérienne après irradiation gamma Des cellules en phases exponentielles de croissance ou en phase stationnaire sont concentrées 10 fois pour être irradiées sur glace entre 3 et 15 kGy (Irradiation au CEA Cadarache par une source de 60Co à 40 Gy/min). A chaque irradiation, un aliquot est prélevé et dilué en cascade de 10 en 10 afin d’être déposé en goutte de 20 μL sur milieu gélosé. La survie est calculée par le rapport « CFU/mL après irradiation » / « CFU/mL avant irradiation ».

76

III Le génome de Deinococcus deserti

77 78 III) Le génome de Deinococcus deserti

1) Annotation du génome de Deinococcus deserti

Le but de l’analyse du génome de D. deserti était de trouver de nouveaux gènes candidats (cibles) qui pourraient avoir un lien avec la radiotolérance. La comparaison des séquences des génomes de D. deserti, D. radiodurans et D. geothermalis devait nous aider à en trouver. L’article “Alliance of proteomics and genomics to unravel the specificities of Sahara bacterium Deinococcus deserti” inséré dans la partie III-4 présente les résultats concernants le génome et le protéome de D. deserti ainsi que la comparaison des séquences entre les 3 Deinococcus. Ma contribution à cet article a été l’annotation d’environ 33 % du génome. Le séquençage du génome de D. deserti a été réalisé au Génoscope et la séquence brute fut disponible en octobre 2006. Le génome de D. deserti se compose d’un chromosome d’environ 2,82 Mb et de 3 plasmides, nommé plasmide P1, P2 et P3 dont les tailles respectives sont de 325, 314 et 396 Kb. La prédiction des ORFs (Open Reading Frame) a été réalisée par P. Ortet et M. Barakat en utilisant des logiciels appropriés, et en considérant les 3 codons start les plus fréquents (ATG, GTG et TTG). La séquence a ensuite été mise à disposition pour une annotation manuelle sur GenoBrowser (une interface d’annotation créée par P. Ortet et M. Barakat) à la fin de l’année 2006. Le chromosome a été annoté par 6 personnes du laboratoire (W. Achouak, M. Barakat, L. Blanchard, A. de Groot, P. Ortet, et moi-même). S. Sommer a également participé à l’annotation des gènes de réparation de l’ADN. Les 3 plasmides ont été annotés par L. Blanchard, A. de Groot et moi-même. En parallèle A. de Groot et moi avons analysé « manuellement », par blastX et autres analyses, la présence de gènes non-prédits dans les régions inter-géniques. Une analyse protéomique réalisée par le laboratoire de J. Armengaud a permis d’améliorer l’annotation du génome. 1348 protéines ont ainsi été identifiées et 15 d’entre elles n’avaient pas été prédites ou trouvées lors de l’annotation. De plus, la protéomique a permis de déterminer de manière sure, le codon start de 112 protéines (validation de 79 start et correction de 33 start). L’annotation complète a pris environ 1 an et a permis d’identifier 3455 gènes. L’annotation manuelle a consisté à vérifier pour chaque ORF le codon start et, en parallèle, à chercher sa fonction potentielle, son homologie avec d’autres protéines, la présence d’une séquence signal (qui sert à l’exportation des protéines) et de domaines transmembranaires. Cette analyse a été réalisée à l’aide de BLAST (basic local alignment

79 search tool) sur des banques de données de protéines (Swiss-Prot, TrEMBL, nr), de logiciels qui prédisent les séquences signal (SignalP, LipoP, TatP), les domaines transmembranaires (TMHMM), et grâce à différents outils tels que SMART (simple modular architecture reseach tool), CDD (conserved domain database) et PRIAM (profil pour l’identification automatique du métabolisme) qui recherchent des domaines conservés, des familles de protéines ou des enzymes. M. Barakat et P. Ortet ont défini au préalable différents critères afin d’annoter le génome d’une manière précise et identique pour tous les annotateurs (Tableau 10). Plusieurs cas de figure existent : - Pour une protéine présentant un taux de recouvrement de plus de 80 % avec une protéine ayant une fonction connue, 3 cas peuvent être envisagés : (1) Elle présente une identité de plus de 40 % avec cette protéine, elle sera alors annotée « Candidate X ». X étant la fonction de la protéine avec lequelle elle se rapproche le plus. (2) Elle présente une identité comprise entre 30 et 40 %, elle sera alors annotée « Related to X». (3) Elle présente une identité inférieure à 30%, elle sera alors annotée « Distantly related to X». - Si la protéine répond au même critère qu’au-dessus, mais qu’elle est homologue à une protéine de fonction inconnue, elle sera alors annotée « Conserved hypothetical protein ». - Enfin si elle ne présente pas d’homologie suffisante (inférieure à 30% d’identité) et un recouvrement plus faible (inférieur à 80%), alors cette protéine sera annotée « Hypothetical protein ». La seconde partie du tableau 10 donne quelques exemples et cas de figure. Les protéines prédites présentant au moins 2 régions transmembranaires comportent l’annotation « Candidate membrane protein ». De même si une séquence signal est retrouvée, apparaîtra au niveau de l’annotation, « precursor ». Lors de la parution du génome de D. deserti sur Genbank, les annotations « Candidate/related to/distantly related to » présentes sur notre site (http://rb.afmb.univ- mrs.fr/DEINOCOCCUS-DESERTI/) ont été remplacées par « putative ». De même, un gène chromosomique nommé par exemple Deide03030 est nommé sur Genbank Deide_03030 et un gène plasmidique, par exemple DeideP101880 est nommé Deide_1p01880.

80 Tableau 10 : Tableaux des critères ayant servit à l’annotation du génome de D. deserti Cas de figure %Id Recouvrement Segments TM Description Nom Gène Enzymes Nom Fonction démontrée expérimentalement renseigné, à créer - Fonction EC Number chez l’organisme ou récupérer dans la littérature Nom Fonction d montr e exp rimentalement "Candidate" - renseign si aucune - chez un autre organismeÊ Fonction ambigu•t Fonction établie sur des critères de similarité structurale avec ou sans "Candidate" Nom Fortes présence de motifs conservés, ou - - >40% >80% Fonction renseign si synt nie similarités contexte génétique comparable (synténie)

- Conserved hypothetical protein --

Homologie avec des g nes de fonction inconnue Conserved hypothetical protein; >=2 -- Candidate membrane protein

Homologie avec des g nes de fonction "Related to" - -- connue Fonction Similarités >30% & <40% >80% - Conserved hypothetical protein -- significatives Homologie avec des g nes de fonction Conserved hypothetical protein; inconnue >=2 Candidate membrane protein Homologie avec des g nes de fonction "Distantly related to" - -- connue <30% + FonctionÊ motif ou Similarités domaine - Conserved hypothetical protein -- non conservé, >80% Homologie avec des g nes de fonction significatives contexte inconnue génétique Conserved hypothetical protein; >=2 -- comparable… Candidate membrane protein

- Hypothetical protein -- Aucune <30% <80% Hypothetical protein; similarité >=2 -- Candidate membrane protein

Cas de figure Modèle Exemple candidate dephospho-CoA kinase (Dephosphocoenzyme A nomenclature PRIAM Enzyme kinase), c andidat e b- glyc osidase, Glyc oside Family 1 (synonyme) Candidate methylcrotonyl-CoA carboxylase, beta subunit candidate transcriptional regulator EpsA, LuxR family Fac teur transc riptionnel transc riptional regulator, family relat ed t o t ransc riptional regulators, LysR family candidate histidine kinase, hybrid histidine kinase, type candidate histidine kinase, unorthodox Syst. à 2 composants candidate histidine kinase, classic candidate response regulator, OmpR response regulator, type c andidate response regulator, CheY candidate ABC transporter, ATP-binding component Transporteur ABC ABC transporter, type candidate ABC transporter, permease component candidate ABC transporter, periplasmic component candidate bifunctional protein: chorismate mutase; Protéine bifonctionnelle Bifunctional protein : enzyme1; enzyme2 prephenate dehydratase Paralogues A annoter en même temps -

Gène à modifier position1-position2 (+ prb prédiction) 1120233-1121122 Gène à supprimer CDS à remplac er ou supprimer artefac t

Gène à rajouter position1-position2,frame 1120233-1121122,-3 Commentaire libre - Le premier tableau, en couleur, explique les annotations générales en fonction des pourcentages de recouvrement, d’identité et de la fonction des protéines homologues. Le second tableau, en blanc, indique des exemples qui sont rencontrés lors de l’annotation d’un génome.

81 2) Exemples d’annotation de gènes

2.1) Gène DeideP101880 (imuY)

Un des gènes que j’avais annoté sur le plasmide P1 est DeideP101880 (Fig. 25). La protéine DeideP101880 présente un domaine conservé entier Pol_Y_like (Fig. 26). Le blast montre que DeideP101880 n’est pas très conservée (Fig. 27). Le start prédit semble bon car le RBS (GGAGG) est très proche de la séquence consensus (AGGAGG) (RBS = ribosome , aussi appelé Shine-Dalgarno sequence). Le codon start est localisé à 3

Figure 25: Partie du plasmide P1 de D. deserti où se situe le gène DeideP101880. Les nombres 318209 et 319480 correspondent respectivement au début et à la fin de l’ORF. En vert, sont indiqués tous les codons start potentiels et en DeideP101880 DeideP101890 rouge, tous les codons stop potentiels. Capture d’écran de GenoBrowser. nucléotides en aval du codon stop du gène précédent (DeideP101870), suggérant la présence d’un opéron. De plus, l’organisation des gènes DeideP101870 (lexA), DeideP101880 (imuY) et DeideP101900 (dnaE2) est similaire à la « mutagenesis cassette » retrouvée chez différentes bactéries (Erill et al., 2006) (Voir aussi chapitre V). L’ensemble de ces résultats permet de donner un nom à la protéine codée par DeideP101880, Related to Y-family DNA polymerase.

Figure 26: Capture d’écran des domaines de la protéine DeideP101880 obtenus en utilisant le programme CDD (http://www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml).

82

Figure 27 : Capture d’écran d’une partie du blast (sur TrEMBL) obtenue à partir de la protéine DeideP101880.

2.2) Gèneène Deide19010 (rpsL)

Le gène Deide19010 est situé sur le chromosome (Fig. 28). La protéine Deide19010 présente un domaine tronqué (Fig. 29), Ribosomal_S12_like superfamily. Le fait que le domaine soit tronqué peut indiquer que le start prédit est faux. D’ailleurs, le blast montre que la protéine est plus courte que ses homologues (Fig. 30). Le logiciel qui a prédit les gènes ne donne pas de start alternatif en amont du start prédit. Ainsi, j’ai regardé manuellement s’il y avait un autre start possible pour ce gène. Pour cela, j’ai pris une séquence plus longue du coté 5’ de Deide19010 (Fig. 31) et j’ai regardé si cela améliorait l’homologie avec le domaine

Figure 28 : Partie du chromosome de D. deserti où se situe le gène Deide19010. Les nombres 2291296 et 2291535 correspondent respectivement la fin et au début de l’ORF prédit. En vert, sont indiqués tous les codons start potentiels et en rouge, tous les codons stop potentiels.

83 Figure 29 : Domaines de la protéine Deide19010 en utilisant le programme CDD.

« Ribosomal_S12_like » ainsi qu’avec les protéines homologues. Le start potentiel serait le codon CTG, un codon start rare chez les bactéries. Grâce à ce codon start, le domaine « Ribosomal_S12_like» est entier, et le blast montre une parfaite homologie entre Deide19010 et d’autres protéines (Fig. 32 et 33). De plus, en amont de ce codon se trouve une séquence Shine-Dalgarno (GAGGAG) bien meilleure que celle qui se trouverait en amont du start prédit initialement (AAGA?) (Fig. 31). Après correction du start, ce gène est situé à la position 2291296-2291697. Cette ORF a été annotée comme Candidate 30S ribosomal protein S12, RpsL.

Figure 30 : Capture d’écran d’une partie du blast (sur la base TrEMBL) obtenue à partir de la protéine Deide19010.

84

Figure 31 : Traduction de l’ORF Deide19010 et d’une séquence en amont. En bleu apparaissent le codon start initialement prédit (GTG) et le codon start très probable (CTG) de Deide19010 ainsi que les possibles séquences Shine-Dalgarno (AAGA et GAGGAG).

Figure 32 : Domaine de la protéine Deide19010 après correction de son codon start.

Figure 33 : Capture d’écran d’une partie du blast (sur la base TrEMBL) obtenue à partir de la protéine Deide19010 après correction du codon start.

85 3) Gènes potentiellement impliqués dans la tolérance aux radiations et à la dessiccation

L’analyse du génome de D. deserti a permis l’identification de nombreux gènes potentiellement liés à la radiotolérance. La comparaison avec les génomes de D. radiodurans et D. geothermalis montre que beaucoup de gènes sont conservés (dont les gènes de réparation de l’ADN ou de réponse aux stress ; Annexes 1 et 2), essentiellement situés sur le chromosome et le plasmide P2. Ces 3 bactéries partagent 230 protéines uniques au genre Deinococcus. Parmi ces protéines se trouvent un homologue du régulateur IrrE, des protéines très exprimées mais de fonction inconnue comme Deide11730, Deide14700 et Deide16110, et des protéines comme DdrC et DdrH qui avaient été mal prédites chez D. radiodurans et/ou D. geothermalis mais corrigées grâce à la génomique comparative avec le génome de D. deserti. D. deserti possède plusieurs gènes supplémentaires potentiellement impliqués dans la réparation de l’ADN, par rapport à D. radiodurans et D. geothermalis (ex : 2 gènes recA plasmidiques, 3 gènes codant des polymérases translésionnelles putatives). Les gènes supplémentaires sont principalement portés par les plasmides P1 et P3 (Tableau 11A). Chez D. radiodurans, une vingtaine de gènes radio-induits et impliqués dans la radiotolérance tels que recA, ddrA, ddrB, ddrD, pprA, gyrA présentent un « radiation desiccation response motif « (RDRM) (Makarova et al., 2007). D. deserti possède également un RDRM au niveau des mêmes gènes que ceux de D. radiodurans. Ceci montre que le régulon de réponse aux dommages de l’ADN est conservé entre ces deux espèces. En plus, 7 autres gènes, ddrC,

Deide04721, Deide08700, Deide18730, Deide22180, DeideP101260 (recAP1) et

DeideP300210 (recAP3) présentent ce motif (Tableau 11A et B). Le gène ddrC de D. radiodurans et D. geothermalis comporte également un RDRM, mais celui-ci n’avait pas été décrit car ddrC était mal prédit (dans la mauvaise orientation) chez ces bactéries. Cinq de ces autres gènes, ddrC, Deide04721, Deide18730, DeideP101260 et DeideP300210 sont radio-induits (Tableau 11A et B ; A. de Groot, communication personnelle). À noter que certains gènes (DeideP101880, DeideP101900, DeideP302150) sont radio-induits alors que le RDRM n’a pas été identifié (Tableau 11A). Concernant la compaction du nucléoide, D. deserti présente 4 gènes codant des homologues à HU (Deide00200, DeideP300060, DeideP300832 et DeideP201940) alors que D. radiodurans n’en possède qu’un (DR_A0065). Les gènes en gras dans le Tableau 11 ainsi que irrE ont été sélectionnés en premier pour que leurs rôles dans la radiotolérance soient étudiés. Pour cela nous avons d¹abord mis au point des outils génétiques pour D. deserti.

86

Tableau 11 : Quelques gènes liés ou potentiellement liés à la radiotolérance de D. deserti et induction de ces gènes après irradiation A. Principaux gènes supplémentaires liés à la réparation de l’ADN chez D. deserti Gène RDRM Inductiona Nom de la protéine/fonction putative Deide05970 NON NON DnaQ Deide06510 NON NT Protéine comportant plusieurs domaines : DnaQ/DinG/ RecQ Deide06520 NON NT UvrD/REP hélicase DeideP100132 NON NON 5’3’ exonucléase DeideP100180 NON NON PolB, ADN polymérase II DeideP100290 NON NT DNA ligase, NAD-dépendent DeideP101260 OUI OUI RecAP1, recombinase A DeideP101280 NON NON Hélicase ADN de la famille RecQ (pas de domaine HRDC) DeideP101880 NON OUI ImuY, ADN polymerase de la famille Y DeideP101900 NON OUI DnaE2, ADN polymérase peu fidèle DeideP300210 OUI OUI RecAP3, recombinase A DeideP302150 NON OUI DNA repair Photoproduct-lyase DeideP302170 NON NON DdrOP3, régulateur transcriptionnel

B. Autres gènes présentant un RDRM chez D. deserti Gène Inductiona Nom de la protéine/fonction putative Deide00120 NT Ssb Deide00600 NT Transketolase Deide01160 NT DdrD Deide02990 OUI DdrB Deide03120 OUI UvrB Deide04721 OUI Protéine de fonction inconnue ; Homologue chez D. geothermalis Deide07131 NON Protéine membranaire ; en opéron avec recJ Deide08700 NON Protéine conservée ; présente chez D. geothermalis et D. radiodurans Deide09150 NT DdrA Deide11320 NT Hélicase ADN RecQ (2 domaines HRDC) Deide12100 NT UvrD Deide12520 NT GyrA Deide12760 NT UvrA Deide15490 NT GyrB Deide16210 NT RecD Deide16610 NON Protéine membranaire Deide18730 OUI Protéine conservée, codée par le premier gène de l’opéron Deide18730-Deide18690 ; absent chez D. geothermalis et D. radiodurans Deide19450 OUI RecAC, recombinase A Deide20570 OUI DdrOC, régulateur transcriptionnel Deide20312 NT Glycosyl Deide22810 NON Protéine spécifique à Deinococcus Deide23280 OUI DdrC DeideP201380 NT PprA En gras, sont représentés les gènes qui ont été étudiés durant ma thèse. RDRM : radiation desiccation response motif. a Induction comme observée dans les conditions expérimentales : l’expression des différents gènes a été testée par RT-PCR, 30 min après irradiation UV à 0 ou 250 J/m2 et a été analysée sur gel d’agarose (les RT- PCRs ont été réalisés par S. Fochesato et A. de Groot). NT = Non testé.

87

88 4) Article 1 : Alliance of proteomics and genomics to unravel the specificities of Sahara bacterium Deinococcus deserti

Auteurs : de Groot A, Dulermo R, Ortet P, Blanchard L, Guérin P, Fernandez B, Vacherie B, Dossat C, Jolivet E, Siguier P, Chandler M, Barakat M, Dedieu A, Barbe V, Heulin T, Sommer S, Achouak W, Armengaud J.

Journal : PLoS genetics

Année : 2009

Volume : 5

Pages : e1000434

89

90 Alliance of Proteomics and Genomics to Unravel the Specificities of Sahara Bacterium Deinococcus deserti

Arjan de Groot1,2,3*,Re´mi Dulermo1,2,3, Philippe Ortet1,2,3, Laurence Blanchard1,2,3, Philippe Gue´rin4, Bernard Fernandez4, Benoit Vacherie5, Carole Dossat5, Edmond Jolivet6, Patricia Siguier7, Michael Chandler7, Mohamed Barakat1,2,3, Alain Dedieu4, Vale´rie Barbe5, Thierry Heulin1,2,3, Suzanne Sommer6, Wafa Achouak1,2,3, Jean Armengaud4 1 CEA, DSV, IBEB, Laboratory of Microbial Ecology of the Rhizosphere and Extreme Environments (LEMiRE), Saint-Paul-lez-Durance, France, 2 CNRS, UMR 6191, Biologie Vegetale et Microbiologie Environnementales, Saint-Paul-lez-Durance, France, 3 Aix-Marseille Universite´, Saint-Paul-lez-Durance, France, 4 CEA, DSV, IBEB, Lab Biochim System Perturb, Bagnols-sur-Ce`ze, France, 5 CEA, Institut de Ge´nomique, Ge´noscope, Evry Cedex, France, 6 Universite´ Paris-Sud 11, CNRS UMR 8621, CEA LRC42V, Institut de Ge´ne´tique et Microbiologie, Universite´ Paris Sud, Orsay Cedex, France, 7 Laboratoire de Microbiologie et Ge´ne´tique Mole´culaire, CNRS UMR5100, Toulouse Cedex, France

Abstract To better understand adaptation to harsh conditions encountered in hot arid deserts, we report the first complete genome sequence and analysis of a bacterium, Deinococcus deserti VCD115, isolated from Sahara surface sand. Its genome consists of a 2.8-Mb chromosome and three large plasmids of 324 kb, 314 kb, and 396 kb. Accurate primary genome annotation of its 3,455 genes was guided by extensive proteome shotgun analysis. From the large corpus of MS/MS spectra recorded, 1,348 proteins were uncovered and semiquantified by spectral counting. Among the highly detected proteins are several orphans and Deinococcus-specific proteins of unknown function. The alliance of proteomics and genomics high- throughput techniques allowed identification of 15 unpredicted genes and, surprisingly, reversal of incorrectly predicted orientation of 11 genes. Reversal of orientation of two Deinococcus-specific radiation-induced genes, ddrC and ddrH,and identification in D. deserti of supplementary genes involved in manganese import extend our knowledge of the radiotolerance toolbox of Deinococcaceae. Additional genes involved in nutrient import and in DNA repair (i.e., two extra recA, three translesion DNA polymerases, a photolyase) were also identified and found to be expressed under standard growth conditions, and, for these DNA repair genes, after exposure of the cells to UV. The supplementary nutrient import and DNA repair genes are likely important for survival and adaptation of D. deserti to its nutrient-poor, dry, and UV-exposed extreme environment.

Citation: de Groot A, Dulermo R, Ortet P, Blanchard L, Gue´rin P, et al. (2009) Alliance of Proteomics and Genomics to Unravel the Specificities of Sahara Bacterium Deinococcus deserti. PLoS Genet 5(3): e1000434. doi:10.1371/journal.pgen.1000434 Editor: Paul M. Richardson, Progentech, United States of America Received November 25, 2008; Accepted February 23, 2009; Published March 27, 2009 Copyright: ß 2009 de Groot et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: Work in our laboratories is supported by the Commissariat a` l’Energie Atomique (CEA LRC42V for SS), the Centre National de la Recherche Scientifique and the Agence Nationale de la Recherche (ANR-07-BLAN-0106-02). SS is further supported by the University Paris-Sud XI and Electricite´ de France. EJ gratefully acknowledges the Centre National de la Recherche Scientifique for its support in this work (CNRS post-doctoral fellowship Nu1002121). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. This work was carried out in compliance with the current laws governing genetic experimentation in France. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected]

Introduction to a high dose of ionizing radiation [5]. Its genome sequence was published in 1999 [6]. More recently, the genome sequence of the The surface sands of hot arid deserts are exposed to intense slightly thermophilic Deinococcus geothermalis, isolated from a hot ultraviolet (UV) radiation, cycles of extreme temperatures, and spring in Italy [7], was determined [8]. desiccation. Nevertheless, an extensive diversity of bacterial species Heavy UV- and desiccation-induced damage to membranes, has been identified in such extreme and nutrient-poor environ- proteins and nucleic acids is lethal to most organisms. Vegetative ments [1,2]. To better understand how life is adapted to these that survive these stresses must therefore either protect specific environmental conditions, we are characterizing Deinococ- vital components from damage and/or repair them efficiently, cus deserti strain VCD115, recently isolated from upper sand layers especially upon rehydration [9]. For D. radiodurans, it was suggested of the Sahara [3]. that its tolerance to high doses of ionizing radiation is a D. deserti belongs to the Deinococcaceae, a family of extremely consequence of its response to natural DNA damaging conditions radiation tolerant bacteria that comprises more than 30 species in such as desiccation [10,11]. Its extreme resistance phenotype has a single genus. Deinococci have been isolated from a wide range of not been fully explained, but has been proposed to result from a environments, such as soil, water, air, faeces, hot springs and combination of different molecular mechanisms and physiological irradiated food [4]. Fourteen of the currently recognized determinants (for reviews, see [4,12]). Repair of massive DNA Deinococcus species were isolated from arid environments, like damage in D. radiodurans involves widespread DNA repair proteins, desert soil and antarctic rock. Among the Deinococci, Deinococcus such as RecA and PolA [13]. In addition, the use of microarrays radiodurans is by far the best characterized and was first isolated resulted in the identification of various hypothetical genes highly more than 50 years ago from canned meat that had been exposed induced following gamma irradiation or desiccation [11]. Of these, 91 PLoS Genetics | www.plosgenetics.org 1 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

Author Summary addition, the proteome analysis allowed correction of unexpected gene prediction errors. Comparative genomics combined with D. deserti belongs to the Deinococcaceae, a family of proteomics data revealed novel Deinococcus-specific proteins that bacteria characterized by an exceptional ability to with- could be involved in stress tolerance mechanisms, as well as stand the lethal effects of DNA-damaging agents, includ- specific characteristics of D. deserti VCD115 that may contribute to ing ionizing radiation, UV light, and desiccation. It was survival and adaptation of this unique bacterium to dry desert soil. isolated from Sahara surface sands, an extreme and nutrient-poor environment, regularly exposed to intense Results UV radiation, cycles of extreme temperatures, and desiccation. The evolution of organisms that are able to Resistance of D. deserti to desiccation, UV and gamma survive acute irradiation doses of 15,000 Gy is difficult to explain given the apparent absence of highly radioactive radiation habitats on Earth over geologic time. Thus, it seems more Besides its resistance to high doses of gamma and UV radiation likely that the natural selection pressure for the evolution [3], D. deserti also tolerated prolonged desiccation, with about 50% of radiation-resistant bacteria was chronic exposure to survival after 40 days of desiccation. Genome repair of D. nonradioactive forms of DNA damage, in particular those radiodurans has been analyzed with cells grown and recovered in promoted by desiccation. Here, we report the first rich medium. As D. deserti is unable to grow in rich media, the fate complete genome sequence of a bacterium, D. deserti of its DNA after exposure to a high dose of gamma radiation VCD115, isolated from hot, arid desert surface sand. (6.8 kGy) or after 27 days of desiccation was analyzed with cells Accurate genome annotation of its 3,455 genes was pre-grown and recovered in tenfold diluted tryptic soy broth guided by extensive proteome analysis in which 1,348 (TSB/10). Like gamma-irradiation, desiccation generated numer- proteins were uncovered after growth in standard ous double-strand DNA breaks in D. deserti. An intact genome was conditions. Supplementary genes involved in manganese reconstituted within six to eight hours under these conditions import, in nutrient import, and in DNA repair were (Figure 1). Therefore, D. deserti’s tolerance to desiccation is related identified and are likely important for survival and to efficient DNA repair as found in D. radiodurans [10] rather than adaptation of D. deserti to its hostile environment. DNA protection mechanisms as observed in Nostoc commune [26]. ddrA, ddrB, ddrC, ddrD and pprA genes were shown to be involved in Genome sequence and structure: general features DNA repair or radiation tolerance [11], and the proteins DdrA The genome of D. deserti VCD115 is composed of four replicons: (DNA extremity protection) [14] and PprA (stimulation of ligase a main chromosome (2.82 Mb) and three plasmids, P1 (325 kb), activity) [15] were characterized in vitro. Tolerance to desiccation P2 (314 kb) and P3 (396 kb) (Genbank accession numbers and radiation does not only imply DNA repair but also protection CP001114, CP001115, CP001116 and CP001117, respectively). of proteins from oxidative damage by a high intracellular Mn/Fe Table 1 presents their main characteristics. Counting of sequence concentration ratio [16]. Moreover, D. radiodurans encodes plant reads gave the relative abundance of each replicon: 1:1:1:1 (62%). protein homologs involved more exclusively in its tolerance to Genome size and equimolarity of the four DNA entities were desiccation [17]. consistent with PFGE results (data not shown). The main D. radiodurans is one of the first bacteria whose were chromosome contains three 16S rRNA genes at different locations, completely sequenced [6,18]. Since this pioneering work, more than and three clusters located elsewhere that each contain one 23S and 600 bacterial genomes have been sequenced. However, very few of these organisms have been thoroughly analyzed at both the genome and proteome levels. Over recent years, various proteomic strategies based on liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS) technology were developed as an aid to genomic annotation [19]. Validation of the existence of orphan genes, prediction of short genes, annotation of genes with unusual codon usage, accurate determination of start codons, and post-translational modifications can be deduced through detection and sequencing of peptides [20,21]. Such proteogenomic strategies were initially used to re-analyse previously annotated and published genomes such as the relatively small bacterial genome of Mycoplasma pneumoniae [22], and then larger genomes such as that of Shewanella oneidensis [23]. They were further applied to the primary annotation of a newly sequenced but related bacterium, Mycoplasma mobile [24]. Transcrip- tomics may also be helpful to validate gene predictions, and may lead to gene discovery when tiling arrays that span the entire genome are used [25]. Figure 1. Kinetics of genome reconstitution in D. deserti after To learn more about the mechanisms of adaptation of D. deserti gamma radiation and after desiccation. D. deserti cells were to the harsh environmental conditions found in the Sahara desert, subjected to 6.8 kGy gamma rays (Panel A) or 27 days of dessication we determined and accurately annotated its entire 3.86 Mb (Panel B). Genomic DNA, purified from cells before irradiation or genome sequence in parallel with an extensive proteome analysis. desiccation and at different times after irradiation or rehydration, was After the 777 kb genome of M. mobile, this is the second bacterial digested with PmeI and SwaI, resulting in 8 DNA fragments for an intact genome, and separated by pulsed-field gel electrophoresis (PFGE). genome whose primary annotation was guided by proteomics. We Lanes M, Yeast Chromosome PFG Marker (New England Biolabs). developed a new strategy that allowed 40% coverage of the whole Lengths (in kb) of several marker fragments are indicated on the left. theoretical proteome for a standard cultivation condition only. In doi:10.1371/journal.pgen.1000434.g001 92 PLoS Genetics | www.plosgenetics.org 2 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

Table 1. General characteristics of the D. deserti genome.

Molecule Chromosome Plasmid P1 Plasmid P2 Plasmid P3 All

General characteristics Size (bp) 2,819,842 324,711 314,317 396,459 3,855,329 GC content (%) 63.38 60.7 63.53 61.41 62.96 Coding density (%) 85.53 72.53 83.74 78.56 84.31 rRNAs 9 3 — — 12 tRNAs 48 ———48 miscRNAsa 5 ———5 Repeat content (%) ,1 ,1 ,1 ,1 ,1 IS contentb (%) 0.3 2.05 1.46 1.97 0.71 Proteins Protein-coding genes 2,594 262 250 349 3455 Detected by MS 1,155 44 83 66 1,348 With assigned function 1,785 160 191 249 2,385 Detected by MS 918 34 65 53 1,070 Conserved hypothetical 611 32 32 45 720 Detected by MS 210 4 11 9 234 Hypothetical 198 70 27 55 350 Detected by MS 27 67444

aOther non-coding RNAs: one RNase P RNA, one SRP RNA, one tmRNA, two THI elements (TPP riboswitch). bComplete and partial IS elements. doi:10.1371/journal.pgen.1000434.t001 one 5S rRNA gene. A single cluster containing one 16S, one 23S therefore identified these systematically using appropriate SDS- and one 5S rRNA gene is also present on plasmid P1. The rRNA PAGE conditions. After 369 nanoLC-MS/MS runs on ESI-ion trap genes on P1 are identical to the corresponding genes on the mass spectrometers, a large corpus of MS/MS spectra (264,251) was chromosome. acquired. The MASCOT search engine was then used to identify Nine complete and different insertion sequences (IS), designated tryptic peptides using various in-house D. deserti polypeptide ISDds1 to ISDds9 according to the standard IS nomenclature with a databases that will be detailed below. Up to 11,129 unique peptides total of 13 copies and belonging to 6 distinct IS families, were were identified when stringent search parameters were applied, identified in D. deserti (Table S1; see also www-is.biotoul.fr). One of corresponding to 1,348 proteins (Tables S2 and S3). Such 3D- these, ISDds1 (an IS3 family member), is present in 5 copies, whereas fractionation resulted in an increased number of protein identifica- each of the remaining 8 ISs is present in only one copy. Remarkably, tions compared to a single 1D SDS-PAGE shotgun, as well as a D. deserti contains many fewer IS elements than D. radiodurans and D. better protein sequence coverage with an average of 27% of the geothermalis (Table S1): 13, 46 and 72, respectively. The IS family protein sequences determined by MS/MS. composition also differs in these species; for example, complete IS200/IS605 elements are absent in D. deserti but present in the other Genome annotation guided by proteome data: a two Deinococci, while complete IS3 elements, present in D. deserti, continuous back and forth strategy are absent in D. radiodurans and D. geothermalis. The complete bacterial genome was automatically searched for CDSs with FrameD, software well suited for gene prediction in Proteome fractionation and extra-large MS/MS shotgun GC-rich bacterial genomes [28]. For proteogenomic annotation analysis (Figure 2A), two polypeptide databases were constructed: a first We investigated several strategies based on 1D SDS-PAGE and version of CDS list (CDS1) containing 3,051 genes predicted by shotgun nanoLC-MS/MS to fractionate the D. deserti proteome from FrameD, and an open reading frame (ORF) list, ORF0. The latter cells collected either in exponential phase or stationary phase: longer comprised all six reading frames (65,801) with at least 100 migration, increase of the percentage of acrylamide for covering low nucleotides of length that were translated. We searched the two molecular weight proteins, additional separation by chromatography polypeptide databases with a preliminary corpus containing prior to SDS-PAGE. Ammonium sulphate precipitation followed by 33,480 MS/MS spectra. A large difference between the two lists phenyl sepharose chromatography was a good means to consider- of identified proteins was observed, i.e. 344 proteins found in ably expand the proteome coverage, as measured by the number of CDS1 and 423 in the ORF0 database, indicating that approxi- proteins detected and their peptide coverage. Small coding mately one fifth of the genes were not correctly predicted with the sequences (CDSs) for non-conserved proteins are difficult to predict. settings chosen for FrameD inspection. These results led us to Moreover, many small prokaryotic proteins are likely to show low further annotate intergenic regions with the recently developed gene expression levels as suggested by their low average codon Med 2.0 algorithm [29]. A second version of CDS list (CDS2) was adaptation index [27]. These low molecular weight proteins (below established and contained 486 additional polypeptides. The CDS2 50 kDa) represent two thirds of the theoretical proteome. We and ORF0 polypeptide databases were searched with the final 93 PLoS Genetics | www.plosgenetics.org 3 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

prediction error in D. deserti (Figure 2A). Deide_19980 was first predicted on the forward strand. As four identified peptides correspond exactly to the polypeptide encoded on the reverse strand at the same locus, we corrected the orientation of this gene and replaced Deide_19980 with Deide_19972 in the opposite orientation. Remarkably, comparative genomics with D. radio- durans and D. geothermalis suggests various prediction errors in these species as well (Table 2). For example, we found four different peptides substantiating the existence of Deide_15980, initially thought to be a D. deserti-specific protein (Figure S1). However, protein sequences highly homologous to Deide_15980 are found when DR_0869 (D. radiodurans) and Dgeo_0511 (D. geothermalis) are translated in the reverse direction. Therefore, DR_0869 and Dgeo_0511 should be reassessed, possibly reversing their orienta- tion. D. deserti’s genome was further manually scrutinized for such errors. For 35 D. deserti genes, previously unpredicted homologs were identified in D. radiodurans and/or D. geothermalis, and these include nine additional instances of reversal of gene orientation (Table 2). Twenty-six of the conserved but differently annotated loci are present in each of the three sequenced Deinococcus Figure 2. Gene prediction corrections revealed by proteomics genomes. Interestingly, two of these correspond to DNAdamage and comparative genomics. (A) Schematic representation of the response genes ddrC (DR_0003; Dgeo_0047) and ddrH (DR_0438). proteogenomic annotation strategy of D. deserti genome. First, genes Their gene products have not been characterized, but DR_0003 were automatically predicted (upper part). Peptides detected by MS, has been inactivated, resulting in decreased radiation resistance represented by vertical black bars, were then located on the [11]. Homologs of the reported DdrC and DdrH proteins were corresponding nucleic acid loci (lower part). Manual validation revealed various cases: (a) validation of predicted genes, (b) detection of non- not found in the D. deserti protein database. However, when the predicted genes, (c) reversal of gene orientation, and (d, e) correction of orientations of DR_0003 and Dgeo_0047 are reversed, proteins start codons. (B) Reversal of ddrC gene orientation in D. radiodurans and highly homologous to Deide_23280 are found (Figures 2B and S2). D. geothermalis. Several data revealed that the orientation of DR_0003 The data strongly suggest that DR_0003 and Dgeo_0047 were and Dgeo_0047 (red broken arrows) has to be reversed, resulting in incorrectly annotated and that Deide_23280 and its homologs are genes highly homologous to correctly predicted Deide_23280 (green the correct genes. Indeed, Deide_23280 and homologs are much arrows). Previous and corrected annotations are labeled 1 and 2, respectively. Green boxes represent the radiation response motif more similar to each other than the DR_0003 homologs (Figure upstream of correct ddrC genes. Flanking genes Deide_23270 and S2); specific RT-PCR experiments showed that Deide_23280 is Dgeo_0048 are homologs. transcribed in D. deserti (Figure S3); and a palindromic motif doi:10.1371/journal.pgen.1000434.g002 identified in the upstream regions of genes that are induced upon radiation/desiccation [8] is present upstream of Deide_23280 and corpus of MS/MS spectra that was collected in the meantime. its two homologs (Figure 2B). For ddrH, a homolog of DR_0438 Again, some differences between the two lists of identified proteins was not found in D. geothermalis [8]. However, if DR_0438 is were observed, leading to the discovery and manual validation of transcribed from the opposite DNA strand, homologs are found in 15 new genes. We further manually inspected intergenic regions D. geothermalis (Dgeo_0322) and D. deserti (Deide_20641). These and found 21 additional genes. Finally, we manually checked the protein similarities suggest that Deide_20641 and Dgeo_0322 are the annotation and translational start codons of the entire set of genes correctly annotated ddrH (Figure S4). The sequences of the bona fide to create a refined CDS database, CDS3, containing 3,455 CDSs DdrC, DdrH and the other newly recognized Deinococcus-specific (Table 1). Using CDS3, 1,348 proteins were validated with proteins (Table 2) do not contain any described domain or motif. stringent search parameters (Tables S2 and S3). These shotgun Characterization of their function and structure requires further data indicate that at least 40% of the genes were expressed in the experimental analysis. standard growth conditions at a sufficient level to be detected by MS. Among the 3,455 genes found in the D. deserti genome, 720 Genome comparisons and Deinococcus-specific genes are predicted to code for conserved hypothetical proteins and 350 Entire genome comparisons showed that 2,046 of the predicted for orphans (Table 1). Our MS data definitively validate the gene products of D. deserti are homologous (at least 30% identity and expression of 234 of the former and 44 of the latter (Table S2). 70% coverage) to proteins from both D. radiodurans and D. geothermalis We checked more specifically for N-terminal peptides in the (Figure 3). These correspond to 1,844 and 1,938 proteins in D. CDS3 and ORF0 lists and verified their adequacy with predicted radiodurans and D. geothermalis, respectively, showing that D. deserti translational start codons. We found 212 distinct peptide globally has more paralogous CDSs. Multiple paralogs are found for signatures corresponding to 145 N termini of proteins (Table exported S8 peptidases, exported serine proteinases, cold shock S3). These data confirmed the starts of 112 proteins but also proteins, and several DNA repair proteins (see below). For example, corrected the starts of 33 polypeptides that were incorrectly each of the three cold shock proteins Deide_09930, Deide_2p00490 predicted even after manual inspection. and Deide_3p00840 is homologous to only one protein of D. radiodurans (DR_0907) and two of D. geothermalis (Dgeo_0638 and Reversal of direction of several ORFs upgrades DNA Dgeo_1006). For 876 D. deserti proteins, no homolog was detected in damage response knowledge D. radiodurans or D. geothermalis (Figure 3). Besides detecting non-predicted coding sequences and correct- Pairwise comparisons of the different replicons of D. deserti with ing start codons, comparing theoretical annotation and MS data those of the other two Deinococcus species revealed strong led to the discovery of another unexpected computational homologous relationships (Table S4) and significant levels of gene 94 PLoS Genetics | www.plosgenetics.org 4 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

Table 2. Differently annotated conserved loci in D. deserti, D. radiodurans, and D. geothermalis.

D. deserti D. radiodurans D. geothermalisa Comments

Deide_06040 (262) reverse DR_0908 (262) Dgeo_0637 (266) conserved Deide_11910 (249) reverse DR_1566 (251) Dgeo_1299 (236) 8 conserved cysteines; Fe-S cluster Deide_13820 (136) reverse DR_0818 (130) Dgeo_0854 (133) Deinococcus-specific Deide_14940 (331) reverse DR_1660 (327–368) — conserved Deide_15980 (192) reverse DR_0869 (192) reverse Dgeo_0511 (185) Deinococcus-specific Deide_16650 (73) reverse DR_0371 (73) Dgeo_1881 (73) Deinococcus-specific Deide_18240 (199) reverse DR_0931 (frameshift?) downstream Dgeo_0497 (199) Deinococcus-specific membrane protein Deide_20641 (82) reverse DR_0438 (92) Dgeo_0322 (82) ddrH reverse Deide_23061 (61) upstream DR_2438 (59) Dgeo_2289 (89–171?) hypothetical Deide_23280 (232) reverse DR_0003 (231) reverse Dgeo_0047 (229) ddrC Deide_1p00840 (283) reverse DR_A0170 (325) Dgeo_2841 (304) intradiol dioxygenase Deide_00870 (136) DR_2373 (136) upstream Dgeo_2056 (136) conserved Deide_01520 (60) upstream DR_2353 (60–130) upstream Dgeo_0094 (62) Deinococcus-specific membrane protein Deide_02110 (236) DR_2585 (227) Dgeo_0028 (695nt); not in Genbank; conserved frameshift? Deide_02380 (106) downstream DR_2376 (.42?) Dgeo_0116 (94) Deinococcus-specific Deide_03200 (56) downstream DR_1936 (56) Dgeo_0456 (56) Deinococcus-specific membrane protein Deide_03861 (126) DR_1855 (126) upstream Dgeo_1580 (125) comEA Deide_04640 (120) upstream DR_2629 (109) — membrane protein Deide_05260 (62) — downstream Dgeo_0866 (66) Deinococcus-specific Deide_07560 (51) between DR_1662/DR_1663 (46) upstream Dgeo_0448 (50) Deinococcus-specific Deide_09630 (147) downstream DR_1062 (218?) Dgeo_0746 (144) conserved membrane protein Deide_10900 (164) DR_0846 (175) upstream Dgeo_1265 (167–178) peroxiredoxin Deide_11240 (103–162?) between DR_2165/DR_2166 (103–346?) Dgeo_1280 (170?) rhodanese-like Deide_11250 (92) DR_0570 (93) upstream Dgeo_1280 (101) rhodanese-like Deide_13440 (57) upstream DR_2194 (57) Dgeo_1152 (57) lysW Deide_14300 (122) DR_1990 (129) downstream Dgeo_0594 (123) Deinococcus-specific; signal peptide Deide_14920 (163) — downstream Dgeo_0948 (162) signal peptide; SCP domain Deide_17020 (290) upstream DR_2201 (286–303) Dgeo_1405 (285) conserved Deide_17971 (53) DR_1463 (97) (too long? 61 aa?) downstream Dgeo_1389 (49) Deinococcus-specific Deide_19480 (133) upstream DR_2342 (99–289) Dgeo_2141 (137) fur Deide_19965 (63) — upstream Dgeo_0366 (63) Deinococcus-specific Deide_20690 (83) upstream DR_0433 (84) upstream Dgeo_0316 (82–105) Deinococcus-specific Deide_1p01660 (346) — Dgeo_2813 (920nt); not in Genbank; mannonate dehydratase frameshift? Deide_2p00970 (81) — downstream Dgeo_1505 (78) Deinococcus-specific Deide_2p00990 (444) — downstream Dgeo_0929 (450) erythromycin Deide_2p01000 (227) — downstream Dgeo_0931 (220) phosphoribosyltransferase

aFor the first 11 loci in this table, the orientation of one or two of the predicted genes has to be reversed. For the other loci, a gene was not predicted in one or two species, but found near the indicated locus. The products of the 18 D. deserti genes indicated in bold face were identified in the proteome analysis after standard cultivation. Lengths of gene products are indicated between parentheses. doi:10.1371/journal.pgen.1000434.t002 order conservation (Figure S5) between the main chromosomes. geothermalis plasmid pDGEO02. Overall, a reciprocal best hit with As determined by reciprocal best hit analysis, the chromosome of D. radiodurans and/or D. geothermalis was found with 80, 35, 65 and D. deserti shares 1,686 and 1,804 gene homologs respectively with 40% of the genes present on the D. deserti chromosome and those of D. radiodurans and D. geothermalis. Although each of the plasmids P1, P2, and P3, respectively. Among those without three D. deserti plasmids contain genes and gene clusters that are reciprocal best hits are genes related to functions such as signal conserved in D. radiodurans and D. geothermalis, D. deserti plasmid P2 transduction and regulation, cell envelope biogenesis, DNA repair, appeared to have the highest levels of homology with D. geothermalis nutrient metabolism and transport. plasmid pDSM11300 (DG574) and with D. radiodurans chromo- The distinctive characteristics of Deinococcus bacteria are likely in some 2 (Table S4). No or very little homology was observed part determined by proteins that are unique in this genus. We have between D. deserti, and D. radiodurans plasmid pCP1 or D. identified 230 proteins, mostly of unknown function and including 95 PLoS Genetics | www.plosgenetics.org 5 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

response (Table S8). Four SoxR-related transcriptional regulators occur in D. deserti while only one and two are found in D. geothermalis and D. radiodurans, respectively. Like D. radiodurans and D. geothermalis, D. deserti contains four close homologs of plant desiccation resistance-associated proteins (Table S8). Interestingly, three of these were detected in D. deserti after standard cultivation, suggesting their importance for swift adaptation to rapid environmental changes occurring in the hot desert. The products of many other stress response-related genes were also detected in the proteome analysis (Table S8). Besides having more Mn ABC transporter proteins, D. deserti is also rich in ABC transporters for oligopeptides, amino acids and sugars in comparison to D. radiodurans and D. geothermalis (Table S9). Thirty-one of 90 amino acid and peptide transporter proteins are specific to D. deserti, as are 17 of 54 proteins for sugar transport. Several components of these ABC transporters were detected by the proteome analysis (Table S9), showing that they are used by Figure 3. Orthologous gene comparison between the three the bacterium in standard cultivation conditions. A large diversity sequenced Deinococcus strains. Orthologous genes are defined by in ABC transporters is likely important for D. deserti’s adaptation to BLAST search when their gene products shared a minimum of 30% identity and 70% coverage. The circle intersections give the number of the desert where nutrients are limiting, and for osmoprotection in genes found in two or three of the compared species, including the case of glycine/betaine transporters. paralogous CDSs. Numbers of genes specific to each species are represented outside these areas. D. deserti gene numbers are in red and D. deserti genome contains three recA and three TLS underlined, those of D. radiodurans are in black with dotted underlining, and those of D. geothermalis are in blue and not underlined. polymerase genes doi:10.1371/journal.pgen.1000434.g003 D. deserti has many DNA repair genes in common with D. radiodurans and D. geothermalis (Table S10), but several interesting six identified after reversal of gene orientation, that are specifically specific D. deserti features were also observed (Table 3). Surpris- conserved in the three sequenced Deinococcus genomes (Table S5). ingly, three different recA genes were found. RecA is a key A role in radiotolerance has been demonstrated for a few of these, component in DNA repair, and required for extreme radiation e.g. IrrE (also called PprI) [30,31] and PprA [15]. Predicted tolerance in D. radiodurans. Most other known bacterial species membrane and exported proteins, accounting for 39% of the possess a single recA. The presence of multiple recA genes has been Deinococcus-specific proteins, may contribute to cell envelope reported for only three species: two in Myxococcus xanthus [33] and integrity, stress tolerance and viability. Indeed, two desiccation Bacillus megaterium [34], and seven in the cyanobacterium resistance-associated proteins (Deide_07540 and Deide_09710) Acaryochloris marina [35]. The recA genes on plasmid P1 and P3 have a predicted signal-peptide and are likely functional outside code for identical proteins. They share 81% identity with the the cytoplasm. Of the 230 Deinococcus-specific proteins, 92 were chromosome-encoded RecA. As in D. radiodurans and D. detected in our proteome analysis (Table S5), many among the geothermalis, the chromosomal recA of D. deserti is the last gene in highly expressed proteins, strongly suggesting their importance in an operon that also contains cinA and ligT, encoding a DNA general metabolism and perhaps cell viability and stress resistance damage/competence-inducible CinA-like protein and a 29-59 in D. deserti. RNA ligase, respectively. For both the chromosome-encoded RecA (90% identical residues) and the plasmid-encoded RecA Adaptation to extreme environment (80% identical residues), the best hits in BLAST analyses were the D. radiodurans has very high intracellular manganese and low RecA proteins from D. radiodurans and D. geothermalis. Moreover, iron levels [32], which are correlated with protection of proteins unlike the other two Deinococci, D. deserti possesses three putative from oxidative modifications [16]. D. deserti was also found to translesion synthesis (TLS) DNA polymerases: a Pol II homolog accumulate Mn, with Mn/Fe ratios of 0.54 and 1.05 in two (PolB); a protein distantly related to members of the DinB different growth media (Table S6). These ratios were even higher subfamily of Y-family DNA polymerases; and DnaE2, a potential than found with D. radiodurans in the same experimental error-prone DNA polymerase homologous to the alpha subunit conditions: 0.16 and 0.47, respectively. We checked for the DnaE of the major replicative DNA polymerase III (Table 3). In presence of Mn- and Fe-transport and regulation related-systems addition, the D. deserti genome encodes a protein related to DNA in D. deserti (Table S7). While only one ABC-type Mn/Zn repair photolyases, a 59-39 exonuclease distantly related to transport system, composed of an ATPase, a permease and a Escherichia coli exonuclease IX, and a large multidomain helicase periplasmic component, is present in D. radiodurans and D. (Deide_06510) also found in Thermus thermophilus HB27. There are geothermalis, several homologs of these components are encoded also two DnlJ DNA ligases sharing 57% identity while the other in D. deserti: three ATPases, five permeases and three periplasmic Deinococci contain only one dnlJ gene. D. deserti also possesses two components. A distantly related homolog of the Nramp family of genes encoding proteins highly homologous to DdrO (DR_2574), Mn(II) transporters was also found. The Mn-dependent transcrip- a proposed regulator of the radiation response regulon in D. tional regulator TroR/DtxR is triplicated in D. deserti. For Fe- radiodurans [8], Deide_20570 and Deide_3p02170 sharing 95% homeostasis, some differences were observed between the three and 85% identity respectively with DR_2574. On the contrary, D. Deinococcus strains (Table S7). The specific presence in D. deserti deserti has only one ung/udg homolog for uracil-DNA glycosylase, genome of an operon comprising two homologous genes specifying whereas D. radiodurans has three and D. geothermalis two (Table 3). siderophore synthetase components is remarkable. Several differ- Four ssb gene homologs were detected in D. geothermalis, while only ences were also observed for proteins related to oxidative stress one is present in the other two Deinococci. 96 PLoS Genetics | www.plosgenetics.org 6 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

Table 3. Main differences among Deinococci regarding DNA repair proteins.

D. deserti D. radiodurans D. geothermalis Protein name and description

Deide_19450 (355), Deide_1p01260 (344), DR_2340 (363) Dgeo_2138 (358) RecA Deide_3p00210 (344) Deide_1p00180 (765) —— DNA polymerase II Deide_1p01880 (423) —— Y-family DNA polymerase Deide_1p01900 (1058) —— error-prone DNA polymerase DnaE2 Deide_3p02150 (354) —— DNA repair photolyase Deide_20570 (129), Deide_3p02170 (129) DR_2574 (131) Dgeo_0336 (140) DdrO; transcriptional regulator Deide_11320a (731) DR_1289b (824) Dgeo_1226b (195) DNA helicase RecQ Deide_1p01280b (548) —— DNA helicase, RecQ family Deide_06510b (1683) —— Multidomain protein: DnaQ/DinG/RecQ Deide_06520 (849) —— UvrD/REP helicase Deide_17790 (202), Deide_05970 (204) DR_0856 (197) Dgeo_1818 (180) DnaQ Deide_12290 (686), Deide_1p00290 (686) DR_2069 (700) Dgeo_0696 (684) DNA ligase, NAD-dependent Deide_1p00132 (244) —— 59-39 exonuclease Deide_00830 (248) DR_0689 (247), DR_1751 (237), Dgeo_2059 (245), Dgeo_1556 (230) Uracil-DNA glycosylase (Ung/Udg) DR_0022 (199) Deide_00120 (297) DR_0099 (301) Dgeo_0165 (301), Dgeo_2964 (283), Single-stranded DNA-binding protein Dgeo_2969 (352), Dgeo_3087 (200)

Polypeptide lengths are indicated between parentheses. aTwo HRDC domains were found in Deide_11320, three in DR_1289, and one in Dgeo_1226. bNo detectable HRDC domain. doi:10.1371/journal.pgen.1000434.t003

Among the 92 DNA repair and radiation tolerance-associated general trait of Deinococcus species. This is further supported by the proteins of D. deserti listed in Table S10, 36 were identified in the observation that D. deserti IrrE fully restored radiation resistance proteome analysis, indicating detectable basal levels of these when expressed in a D. radiodurans irrE deletion mutant [36]. The proteins in the absence of exposure to exogenous DNA damaging motif was not found near the other additional D. deserti DNA repair agents. Among these are DNA glycosylases, nucleotide excision genes listed in Table 3, including those specifying the photolyase- and recombinational repair proteins, several ‘‘house-cleaning’’ related protein and TLS DNA polymerases. These may be Nudix hydrolases, both DNA ligases, and the DNA repair regulated in another manner. Interestingly, the palindromic motif regulator and putative sensor protein IrrE (PprI). Such basal was also found near several genes that have no obvious relation cellular levels of DNA repair proteins might be expected if DNA with DNA repair, notably near Deide_18730, the first of five damage occurs frequently due to generation of high levels of uncharacterized genes in a putative operon that is conserved in metabolism-derived reactive oxygen species or if cells must be various species but absent from D. radiodurans and D. geothermalis. constantly ready to quickly respond to external stress. The chromosome-encoded RecA was detected by the proteome Discussion analysis, whereas peptides that specifically demonstrate the presence of plasmid-encoded RecA were not identified. Similarly, D. deserti was isolated from the surface sand of the Sahara, an none of the TLS polymerases was detected after standard extreme environment, where it is exposed to harsh conditions with cultivation. However, RT-PCR experiments showed that the concomitant UV irradiation, desiccation and nutrient limitation. three recA, the three TLS polymerase genes and the photolyase- To learn more about the adaptation of D. deserti to arid desert as related gene were transcribed. Moreover, except for polB, well as the remarkable tolerance to radiation and desiccation of transcription of each of these genes was induced after exposure Deinococcaceae in general, we determined the sequence of its entire to UV (Figure 4), indicating a role in the DNA damage response in genome, composed of a chromosome and three large plasmids. D. deserti. Moreover, accurate gene annotation was guided by extensive proteome analysis after growth of D. deserti under standard The Deinococcus radiation response regulon is conserved conditions. Besides validation of about 40% of the predicted A potential common radiation response regulon in D. radiodurans genes, peptide data allowed correction of several initiation codons, and D. geothermalis identified by a palindromic motif has been identification of unpredicted genes and, surprisingly, reversal of reported [8]. This radiation/desiccation response motif is found in incorrectly predicted gene orientation. Comparative genomics and the upstream regions of radiation-induced genes such as recA, ddrA, proteomics showed that the orientation of several annotated D. ddrO, pprA, uvrA, and gyrA. Induction of transcription of at least recA radiodurans and D. geothermalis genes should also be reversed. This is and pprA is IrrE (PprI)-dependent in D. radiodurans [30,31]. We an important matter and to our knowledge has never been searched for such motif in D. deserti and detected its signature in the reported previously at the genomic scale. Interestingly, transcrip- upstream regions of the same genes and, in addition, upstream of tion of two of these, ddrC and ddrH, was induced in D. radiodurans Deide_23280 (the bona fide ddrC) and the two plasmid-encoded recA after exposure to radiation and desiccation [11]. These transcrip- genes (Table S11). Therefore, the radiation response regulon is a tion data may seem contradictory to our results. However, we 97 PLoS Genetics | www.plosgenetics.org 7 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

spectra per protein were reached taking into consideration the global set of data. The redundancy is even higher for abundant proteins. Although these data are only semi-quantitative and extracted from the entire set of MS/MS spectra recorded in this study, several interesting features should be noted. In our MS/MS corpus, 50% of the spectra were attributed to 106 proteins, 75% to 281 proteins, and 90% to 547 proteins (Table S2). Interestingly, among the highly detected proteins are several orphans (Dei- de_3p02433, Deide_15630, and Deide_11191), Deinococcus-specific proteins (Deide_06180, Deide_1p01253, Deide_21340, Deide_11730, Deide_16050, Deide_16110, Deide_09460, Deide_01434, and Deide_15100) and more widely conserved proteins (Deide_3p01280, Deide_21050, Deide_13500, Deide_04930, and Deide_14540) of unknown function, suggesting an important role in D. deserti. The proteome data indicate that plasmids P1 and P3 are underused in our standard growth conditions in comparison to P2 and the chromosome. The products of 17–19% of the genes present on P1 and P3 were detected, compared to 34–45% of those on the chromosome and P2. These data suggest that most of the P1 and P3 genes may be used under more specific conditions (such as in the desert and/or after stress) than the laboratory conditions used here. Extreme tolerance to radiation and desiccation in Deinococci requires efficient DNA repair. In addition to most of the classical DNA repair genes, the previously identified novel, Deinococcus- specific genes involved in DNA repair and radiotolerance, i.e. ddrA-D, pprA and irrE (pprI), are all conserved in D. deserti.In addition, using epifluorescence microscopy, we have observed that D. deserti (results not shown), like other radioresistant bacteria [39], has a highly condensed nucleoid, which may restrict diffusion of radiation/desiccation-generated DNA fragments. Furthermore, in Figure 4. RT-PCR analysis of recA and putative translesion DNA common with D. radiodurans and D. geothermalis, D. deserti has a high polymerase and photolyase genes. RNA was isolated 30 min after 22 Mn/Fe ratio, important for protein protection. exposure to 0 (2) or 250 (+) J.m UV. The constitutively expressed tuf Besides the common set of genes and characteristics involved in gene was included as control. doi:10.1371/journal.pgen.1000434.g004 protection or repair, each species has probably also evolved specific functions to adapt to its environment. This is supported by believe that Tanaka et al. [11], without awareness, measured the identification of several genes for which a role in stress transcription of the correct ddrC and ddrH genes. Their resistance and DNA repair has been shown or can be expected, transcription analyses could not distinguish between one DNA but that are not shared by the Deinococci. For example, homologs strand and its complementary strand since the DNA spotted on the of D. radiodurans irrI, associated with radiation resistance [40], and microarrays were obtained by PCR, and thus included both DNA several radiation-induced ddr genes [11], are absent in D. deserti strands. Moreover, random hexamers were used for initiation of and D. geothermalis (Table S10). As another example, the conserved cDNA probe synthesis (A. Earl, personal communication), and these radiation response motif was found upstream of two D. deserti genes hexamers annealed with any RNA, including the correct ddrC and of unknown function with a homolog only in D. geothermalis ddrH mRNA. Several other previously unannotated genes were (Deide_04721) or without homologs in the other Deinococci found in D. geothermalis and/or D. radiodurans, including homologs of (Deide_18730). The latter is the first gene of a putative operon Deide_17971 and Deide_19965, genes that were not predicted but encoding five uncharacterized proteins, of which Deide_18710 whose products were detected by proteomics. These and 5 other (related to MoxR-like AAA+ ATPases) could function with genes that were discovered by proteomics, code for very small Deide_18690 (containing a Von Willebrand Factor Type A polypeptides of 4.5–9.6 kDa. Our work shows that a combination of domain) as a chaperone system for the folding/activation of high-throughput proteomic and genomic techniques allows the most specific substrate proteins [41]. accurate genome annotation presently obtainable. By extension, the Another interesting feature is the presence in D. deserti of TLS annotation of the whole Deinococcus-Thermus group may be revisited DNA polymerase genes, as well as a photolyase-related gene and taking into account evolutionary constraints, a concept already two supplementary recA, all present on plasmid P1 or P3. All of applied to the genera Shewanella and Mycobacterium [20,21]. these genes could be involved in tolerance to UV-light. E. coli Besides proteogenomic annotation, shotgun proteomics allows RecA plays a major role in recombinational repair of UV-lesions, semi-quantitation of the major proteins by spectral counting. Since has a regulatory role in lesion bypass through its coprotease the demonstration by Liu et al. [37] that spectral counts of MS/MS activity, which includes stimulation of self-cleavage of repressor spectra for a given protein in a mixture correlated linearly with protein LexA, and participates directly in lesion bypass by protein abundance within over two orders of magnitude, interacting with TLS Pol V [42,43]. As the three TLS DNA numerous semi-quantitative studies have been based on this polymerase genes in D. deserti are located on plasmid P1, it is concept [38]. Among the 264,251 MS/MS spectra that were tempting to speculate that the plasmid-encoded RecA is involved acquired, 50,461 spectra were confidently assigned to a peptide in regulation and/or activation of one or more of these TLS sequence. Averages of 4.5 spectra per unique peptide and 37 polymerases. Moreover, a lexA homolog, Deide_1p01870,is 98 PLoS Genetics | www.plosgenetics.org 8 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti adjacent and possibly in an operon with the TLS DNA polymerase comparisons, were used to determine the presence of homologs genes Deide_1p01880 (Y-family DNA polymerase) and Dei- in different species. de_1p01900 (DnaE2). Similar so-called adaptive mutagenesis gene cassettes have been recently described and shown to be under Proteome fractionation by phenyl sepharose and SDS- RecA/LexA regulation in several bacterial species [44], and a role PAGE for the Y-family DNA polymerase and/or DnaE2 in survival and D. deserti cells grown in tenfold diluted tryptic soy broth (TSB/ induced-mutagenesis after exposure to DNA-damaging conditions 10) supplemented with trace elements [36] and harvested in Mycobacterium tuberculosis Caulobacter has been demonstrated in [45], exponential growth and in pre-stationary phase (4 g of wet crescentus [46] and Pseudomonas putida [47]. material) were resuspended in 20 mL of cold 100 mM TRIS/HCl Taken together, the supplementary nutrient import and DNA buffer (pH 8.0 at 20uC) containing 5 mM EDTA, disrupted and repair genes in D. deserti may contribute to survival in nutrient- centrifuged. Proteins from 8 mL of both resulting supernatants poor extreme conditions. The additional DNA repair genes likely were subjected to ammonium sulphate precipitation. Proteins still provide an advantage in an environment where many DNA soluble after addition of 50% final (NH ) SO were precipitated lesions can be generated due to intense UV irradiation and 4 2 4 with 67% final (NH ) SO . The resulting four equivalent pellets desiccation. Moreover, TLS polymerases generate mutations that 4 2 4 were resuspended in 50 mM TRIS/HCl buffer pH 8.0, contain- will increase genetic diversity, which may lead to better adaptation ing 2.5 mM EDTA and 1.5 M (NH ) SO (Buffer A). Proteins to harsh conditions, such as those encountered in the desert [48]. 4 2 4 from each of the four samples were applied at a flow rate of 1.0 mL/min onto a 5 mL HiTrap Phenyl HP column (Amersham Materials and Methods Biosciences) previously equilibrated with Buffer A. After wash with Genome sequencing Buffer A, proteins were eluted over a 60 mL linear gradient The complete sequence of D. deserti VCD115 was determined comprising 1.5–0 M (NH4)2SO4. Eluted proteins (24 fractions per after genomic DNA fragmentation by mechanical shearing or phenyl chromatography) were precipitated by trichloroacetic acid BamHI partial digest for the construction of plasmid and large (10% final) and collected by centrifugation. Proteins were dissolved insert libraries, respectively. The 3 kb (A), 10 kb (B) and 25 kb (C) in LDS sample buffer (Invitrogen) and then analyzed by SDS- fragments were cloned into pcDNA2.1 (INVITROGEN), pCNS PAGE on 4–12% gradient NuPAGE (Invitrogen) gels. The gels (pSU18 derived) and pBeloBac11, respectively. Vector were were stained with Coomassie Safe Blue stain (Invitrogen) and then purified and end-sequenced (49536 (A), 16128 (B) and 7680 (C)) each relevant lane was excised into 4–10 mm-thick pieces from the using dye-terminator chemistry on ABI3730 sequencers. A pre- top to the bottom polypeptide bands. assembly was made without repeat sequences as described [49] using Phred/Phrap/Consed software package (www.phrap.org). In-gel proteolysis The finishing step was achieved by primer walking and PCR. A Protein bands were washed with MilliQ water, treated with complement of 665 sequences was needed for gap closure and CH3CN, and then with 100 mM NH4HCO3. Gel pieces were quality assessment. dehydrated with 100% CH3CN and dried for 20 min under vacuum. For in-gel digestion, dry gel pieces were rehydrated for Gene prediction and annotation 45 min at 56uC with 100 mM NH4HCO3 containing 10 mM Protein-coding regions in the assembled genome sequence were DTT. Gel pieces were rinsed with CH3CN, treated with 100 mM identified using FrameD [28] and MED [29]. Predicted proteins NH4HCO3 containing 55 mM iodoacetamide, then dehydrated larger than 10 amino acids were analysed for sequence similarity with 100% CH3CN and dried for 20 min under vacuum. For in- against protein databases (SWISSPROT, TREMBL and non- gel digestion, dry gel pieces were rehydrated with 12 ng/mL redundant GenBank proteins). Similarity searches were carried out trypsin in 25 mM NH4HCO3 (pH 8.5) containing 1% CaCl2. using BLAST programs [50]. Annotation of the complete genome After overnight proteolysis at 37uC, digests were extracted first was performed using GenoBrowser, an in-house bioinformatic tool with 100% HCOOH, then with 56% CH3CN/1% HCOOH, and for data management (Ortet et al., submitted). Our tool allows an finally with 100% CH3CN. The resulting pools were dried expert annotation by manual verification and curation of completely under vacuum and stored at 220uC until MS analysis. functional protein categories after automatic assignment. Genome regions without match were re-evaluated by using BLASTX as LC-MS/MS analysis initial search, and ORFs were extrapolated from regions of LC-MS/MS experiments were performed on two equipments: alignment. rRNA and tRNA genes were identified with BLASTN (i) a Esquire 3000 plus ion trap (Bruker Daltonics) equipped with a and tRNA-Scan [51]. miscRNA were identified using INFERNAL nanoelectrospray online ion source, and coupled to a UltiMate- software suite with RFAM database [52] version 8.1 containing Switchos-Famos LC system (Dionex-LC Packings), and (ii) a LTQ- 607 families. Repeats were identified using RepSeek [53]. Orbitrap XL hybrid mass spectrometer (ThermoFisher) coupled to a UltiMate 3000 LC system (Dionex-LC Packings). Peptide Genbank submission mixtures (0.5–5 pmol) were loaded and desalted online in a The annotated genome sequence of D. deserti VCD115 was reverse phase precolumn (C18 Pepmap column, LC Packings), deposited in the GenBank database with accession numbers and resolved on a nanoscale C18 Pepmap TM capillary column CP001114, CP001115, CP001116 and CP001117 for the main (LC Packings) at a flow rate of 0.2–0.3 mL/min with a gradient of chromosome, plasmids P1, P2 and P3, respectively. CH3CN/0.1% formic acid prior injection in the ion trap mass spectrometer. Peptides were separated using a 90 min-gradient Comparative genomics from 5 to 95% solvent B (0.1% HCOOH/80% CH3CN). Solvent All predicted D. deserti proteins were aligned using BLASTP A was 0.1% HCOOH/5% CH3CN for the Esquire nanoLC against the available of published complete bacterial system, and 0.1% HCOOH/0% CH3CN for the LTQ-orbitrap genomes (625 genomes). The results of these searches, requiring nanoLC system. The full-scan mass spectra were measured from bi-directional best hit (BBH) among all possible pairwise m/z 50 to 2000 with the Esquire ion trap mass spectrometer, and 99 PLoS Genetics | www.plosgenetics.org 9 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti m/z 300 to 1700 with the LTQ-orbitrap XL mass spectrometer. dishes), and then further incubated. Samples for RNA isolation The latter was operated in the data-dependent mode using the were taken after 30 min. Cells were treated with RNAprotect TOP7 strategy. In brief, a scan cycle was initiated with a full scan Bacteria Reagent (Qiagen) prior to RNA isolation using the of high mass accuracy in the orbitrap, which was followed by MS/ RNeasy Mini Kit (Qiagen) according to the manufacturer’s MS scans in the linear ion trap on the 7 most abundant precursor instructions. RNA samples were treated twice with DNase. For ions with dynamic exclusion of previously selected ions. A total of RT-PCR, cDNA was synthesized in 20 mL reactions using 1 mgof 326 and 40 samples were analyzed on the mass spectrometers, RNA and the Transcriptor First Strand cDNA Synthesis Kit resulting in 163,779 and 100,472 MS/MS spectra recorded, (Roche). DNA fragments of 150–250 bp were then amplified in respectively. 25 mL reactions using 1 mL of cDNA from the first step, Taq polymerase (Sigma) and gene-specific primers (Table S12). These Polypeptide database MS/MS search amplifications were carried out by incubating reactions at 94uC for Using the MASCOT search engine (Matrix Science), we 10 min prior to 30 cycles of 1 min at 94uC, 30 sec at 56uC and searched all MS/MS spectra against home made polypeptide 30 sec at 72uC, followed by a final step at 72uC for 2 min, with sequence database. Searches for trypsic peptides were performed modifications for tuf (25 cycles), recA-P3 and dnaE2 (both 35 cycles). with the following parameters: full-trypsin specificity, a mass Controls for DNA contamination were performed with reactions tolerance of 5 ppm on the parent ion and 0.5 Da on the MS/MS lacking reverse transcriptase. (LTQ-orbitrap mass spectra) or 0.4 Da for the parent ion and 0.5 Da for the MS/MS (Esquire ion trap mass spectra), static Supporting Information modifications of carboxyamidomethylated Cys (+57.0215), and Figure S1 Evidence for correct annotation of Deide_15980. dynamic modifications of oxidized Met (+15.9949). The maximum Deide_15980 proteogenomic annotation with four peptides detect- number of missed cleavage was set at 1. All peptide matches with a ed by mass spectrometry (A). Multiple sequence alignments of Peptide Score of at least 31 (average threshold for p,0.007 with Deide_15980 and homologs found encoded in the reverse the final database search using the Esquire ion trap data) and rank orientation of DR_0869 and Dgeo_0511 (B). Multiple sequence 1 were filtered by the IRMa 1.16.0 software (J. Garin&C. Bruley, alignments of incorrectly predicted DR_0869 and Dgeo_0511 CEA, DSV, iRTSV, Grenoble). The criteria adopted for protein with a protein found when the orientation of Deide_15980 is identification were very conservative with either 1) that at least two reversed (C). Identified peptides for Deide_15980 are indicated peptides with ion score 31 or higher match or 2) that at least one with black and blue bars in (A). The peptide sequences can be peptide with ion score 50 (average threshold for p,0.0001) or found in Table S3. higher matches. A False-Positive rate of 0.2% was estimated using Found at: doi:10.1371/journal.pgen.1000434.s001 (0.07 MB PDF) the corresponding decoy database. Further data analysis was performed for semi-trypsin specificity (p,0.001). Spectral count Figure S2 Evidence for correct annotation of ddrC in Deinococcus (number of spectra recorded per protein) was performed on the species. Schematic representation of the ddrC loci and flanking whole set of MS/MS spectra recorded over the entire study. genes in D. deserti, D. radiodurans and D. geothermalis (A). Multi- Proteins were then evaluated for their respective abundance after alignment of polypeptide sequences for correct DdrC proteins (B) molecular weight normalization. and wrongly annotated DdrC proteins (Panel C). In Panel A, the predicted genes are labeled with the locus tag number. DR_0003 Kinetics of DNA repair after gamma-irradiation or and Dgeo_0047 are incorrectly predicted, and their orientation has to be reversed. Correct ddrC genes are shown in black arrows, desiccation-rehydration wrong ddrC ORFs in open arrows. The black boxes upstream the For gamma irradiation, cells were grown at 30uCinTSB/10toan correct ddrC genes indicate the radiation response motif. Flanking OD600 of 0.5, concentrated to an OD600 of 25, and irradiated on ice 137 genes Deide_23270 and Dgeo_0048 are homologs. to 6,800 Gy at 56.6 Gy/min dose rate ( Cs source). After Found at: doi:10.1371/journal.pgen.1000434.s002 (0.05 MB PDF) irradiation, cultures were diluted to an OD600 of 0.2 and incubated at 30uC. At different post-irradiation incubation times, 5 ml aliquots Figure S3 Specific RT-PCR showing transcription of correct were removed to prepare agarose plugs as described by Harris et al. ddrC (Deide_23280). Genome region of Deide_23280 (A). [14]. The DNA in the plugs was digested with PmeI and SwaI Deide_23280 is present on the reverse DNA strand. The opposite restriction enzymes, and then subjected to pulsed-field gel DNA strand could encode a protein related to DR_0003 and electrophoresis for 24 h at 12uCusingaCHEFMAPPER Dgeo_0047 (indicated by the blue ellipse). Green and red bar electrophoresis system (Biorad) with the following conditions: indicate start and stop codons, respectively. Black bar indicates the 6.0 V/cm, linear pulse ramp of 60–120 s, and a switching angle of radiation/desiccation response motif upstream Deide_23280.RV 120u (260u to +60u). For desiccation, stationary phase cells were and FW indicate schematically the orientation of the primers used in the RT-PCR. Specific two-step RT-PCR (B). Either purified concentrated to an OD600 of 20 and placed on sterile glass slides (100 ml cells per slide), dried and stored at room temperature in a primer FW (Dd23280FW) or RV (Dd23280RV) was used to synthesize cDNA in the RT reaction (first step). The PCR in the sealed desiccator over anhydrous CaSO4. The CaSO4 desiccant is second step was performed with both primers. When Deide_23280 impregnated with CoCl2. This latter chemical compound is a visual is the correctly predicted gene, cDNA synthesis and thus an RT- moisture indicator: anhydrous CoCl2 is blue while pink in presence of water. After 27 days of desiccation, cells from 5 glass slides were PCR product is expected with primer RV in the RT reaction. RNA was isolated 30 min after the cells were irradiated (+) or not resuspended in 50 ml TSB/10, incubated at 30uC and 5 ml aliquots 2 were taken at different times and treated as described above. (2) with UV (250 J/m ). N and P indicated negative (no template) and positive (genomic DNA) control, respectively. RNA isolation and RT-PCR Found at: doi:10.1371/journal.pgen.1000434.s003 (0.11 MB PDF) For analysis of gene expression after UV exposure, samples Figure S4 Evidence for correct annotation of ddrH. Multi- from the same exponential phase culture (OD600 0.2–0.4) were alignment of correct DdrH protein sequences (A) and wrongly exposed to either 0 or 250 J.m22 of UV-C (4 mL volumes in Petri annotated DdrH protein sequences (B). The D. radiodurans protein 100 PLoS Genetics | www.plosgenetics.org 10 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti highly homologous to Deide_20641 and Dgeo_0322 is found Table S8 Stress response-related proteins in D. deserti, D. when the orientation of DR_0438 (ddrH) is reversed. radiodurans and D. geothermalis. Found at: doi:10.1371/journal.pgen.1000434.s004 (0.03 MB PDF) Found at: doi:10.1371/journal.pgen.1000434.s013 (0.16 MB PDF) Figure S5 Genome dot plots for the chromosomes of sequenced Table S9 ABC-transporters identified in D. deserti. Deinococcus species. Genome dot plots for the chromosome of D. Found at: doi:10.1371/journal.pgen.1000434.s014 (0.15 MB PDF) deserti (horizontal axis) vs. D. geothermalis (vertical axis) (A) and vs. D. radiodurans (vertical axis) (B). Each dot represents the location of a Table S10 DNA repair genes in D. deserti, D. radiodurans and D. pair of Bidirectional Best Hits (with a minimum of 30% identity geothermalis. and 70% coverage) between the two chromosomes. Found at: doi:10.1371/journal.pgen.1000434.s015 (0.20 MB PDF) Found at: doi:10.1371/journal.pgen.1000434.s005 (0.22 MB PDF) Table S11 Radiation/desiccation response motif identified in D. Table S1 Insertion sequences identified in the genome of D. deserti. deserti and comparison with other Deinococcus/Thermus species. Found at: doi:10.1371/journal.pgen.1000434.s016 (0.11 MB PDF) Found at: doi:10.1371/journal.pgen.1000434.s006 (0.09 MB PDF) Table S12 Primers for RT-PCR. Table S2 Proteins identified by LC-MS/MS shotgun. Found at: doi:10.1371/journal.pgen.1000434.s017 (0.06 MB PDF) Found at: doi:10.1371/journal.pgen.1000434.s007 (0.26 MB XLS) Acknowledgments Table S3 Peptides identified by LC-MS/MS shotgun. Found at: doi:10.1371/journal.pgen.1000434.s008 (2.67 MB XLS) We gratefully acknowledge Je´roˆme Garin and Christophe Bruley (IRMa software and discussions), the Institut Curie (137Cs irradiation facilities), Table S4 Homology between the DNA molecules of D. deserti, D. Ashlee Earl (for communicating methodological details), as well as Sylvain radiodurans and D. geothermalis. Fochesato (RT-PCR), Paul Soreau (ICP analysis), Charles Marchetti Found at: doi:10.1371/journal.pgen.1000434.s009 (0.07 MB PDF) (fermentor facilities), and V. Favaudon (c-irradiation) for their technical Table S5 Deinococcus-specific proteins. assistance. Found at: doi:10.1371/journal.pgen.1000434.s010 (0.16 MB PDF) Author Contributions Table S6 Mn/Fe ratio in D. deserti and D. radiodurans. Found at: doi:10.1371/journal.pgen.1000434.s011 (0.08 MB PDF) Conceived and designed the experiments: AdG TH SS JA. Performed the experiments: AdG RD PO LB PG BF BV CD EJ AD VB JA. Analyzed the Table S7 Mn- and Fe- transport and regulation related systems data: AdG RD PO LB PG PS MC MB AD SS WA JA. Contributed in D. deserti. reagents/materials/analysis tools: PO MB. Wrote the paper: AdG JA. Found at: doi:10.1371/journal.pgen.1000434.s012 (0.08 MB PDF) Contributed to writing the paper: SS.

References 1. Chanal A, Chapon V, Benzerara K, Barakat M, Christen R, et al. (2006) The 14. Harris DR, Tanaka M, Saveliev SV, Jolivet E, Earl AM, et al. (2004) Preserving desert of Tataouine: an extreme environment that hosts a wide diversity of genome integrity: the DdrA protein of Deinococcus radiodurans R1. PLoS Biol 2: microorganisms and radiotolerant bacteria. Environ Microbiol 8: 514–525. e304. doi:10.1371/journal.pbio.0020304. 2. Rainey FA, Ray K, Ferreira M, Gatz BZ, Nobre MF, et al. (2005) Extensive 15. Narumi I, Satoh K, Cui S, Funayama T, Kitayama S, et al. (2004) PprA: a novel diversity of ionizing-radiation-resistant bacteria recovered from Sonoran Desert protein from Deinococcus radiodurans that stimulates DNA ligation. Mol Microbiol soil and description of nine new species of the genus Deinococcus obtained from a 54: 278–285. single soil sample. Appl Environ Microbiol 71: 5225–5235. 16. Daly MJ, Gaidamakova EK, Matrosova VY, Vasilenko A, Zhai M, et al. (2007) 3. de Groot A, Chapon V, Servant P, Christen R, Fischer-Le Saux M, et al. (2005) Protein oxidation implicated as the primary determinant of bacterial Deinococcus deserti sp. nov., a gamma-radiation-tolerant bacterium isolated from radioresistance. PLoS Biol 5: e92. doi:10.1371/journal.pbio.0050092. the Sahara Desert. Int J Syst Evol Microbiol 55: 2441–2446. 17. Battista JR, Park MJ, McLemore AE (2001) Inactivation of two homologues of 4. Blasius M, Hubscher U, Sommer S (2008) Deinococcus radiodurans: what belongs to proteins presumed to be involved in the desiccation tolerance of plants sensitizes the survival kit? Crit Rev Biochem Mol Biol 43: 221–238. Deinococcus radiodurans R1 to desiccation. Cryobiology 43: 133–139. 5. Anderson AW, Nordan HC, Cain RF, Parrish G, Duggan D (1956) Studies on a 18. Makarova KS, Aravind L, Wolf YI, Tatusov RL, Minton KW, et al. (2001) radio-resistant micrococcus. I. Isolation, morphology, cultural characteristics, Genome of the extremely radiation-resistant bacterium Deinococcus radiodurans and resistance to gamma radiation. Food Technol 10: 575–578. viewed from the perspective of comparative genomics. Microbiol Mol Biol Rev 6. White O, Eisen JA, Heidelberg JF, Hickey EK, Peterson JD, et al. (1999) 65: 44–79. Genome sequence of the radioresistant bacterium Deinococcus radiodurans R1. 19. Ansong C, Purvine SO, Adkins JN, Lipton MS, Smith RD (2008) Science 286: 1571–1577. Proteogenomics: needs and roles to be filled by proteomics in genome 7. Ferreira AC, Nobre MF, Rainey FA, Silva MT, Wait R, et al. (1997) Deinococcus annotation. Brief Funct Genomic Proteomic 7: 50–62. geothermalis sp. nov. and Deinococcus murrayi sp. nov., two extremely radiation- 20. Gallien S, Perrodou E, Carapito C, Deshayes C, Reyrat JM, et al. (2009) Ortho- resistant and slightly thermophilic species from hot springs. Int J Syst Bacteriol proteogenomics: multiple proteomes investigation through orthology and a new 47: 939–947. MS-based protocol. Genome Res 19: 128–135. 8. Makarova KS, Omelchenko MV, Gaidamakova EK, Matrosova VY, Vasilenko A, 21. Gupta N, Benhamida J, Bhargava V, Goodman D, Kain E, et al. (2008) et al. (2007) Deinococcus geothermalis: the pool of extreme radiation resistance genes Comparative proteogenomics: combining mass spectrometry and comparative shrinks. PLoS ONE 2: e955. doi:10.1371/journal.pone.0000955. genomics to analyze multiple genomes. Genome Res 18: 1133–1142. 9. Potts M (2001) Desiccation tolerance: a simple process? Trends Microbiol 9: 22. Dandekar T, Huynen M, Regula JT, Ueberle B, Zimmermann CU, et al. (2000) 553–559. Re-annotating the Mycoplasma pneumoniae genome sequence: adding value, 10. Mattimore V, Battista JR (1996) Radioresistance of Deinococcus radiodurans: function and reading frames. Nucleic Acids Res 28: 3278–3288. functions necessary to survive ionizing radiation are also necessary to survive 23. Kolker E, Picone AF, Galperin MY, Romine MF, Higdon R, et al. (2005) Global prolonged desiccation. J Bacteriol 178: 633–637. profiling of Shewanella oneidensis MR-1: expression of hypothetical genes and 11. Tanaka M, Earl AM, Howell HA, Park MJ, Eisen JA, et al. (2004) Analysis of improved functional annotations. Proc Natl Acad Sci USA 102: 2099–2104. Deinococcus radiodurans’s transcriptional response to ionizing radiation and 24. Jaffe JD, Stange-Thomann N, Smith C, DeCaprio D, Fisher S, et al. (2004) The desiccation reveals novel proteins that contribute to extreme radioresistance. complete genome and proteome of Mycoplasma mobile. Genome Res 14: Genetics 168: 21–33. 1447–1461. 12. Cox MM, Battista JR (2005) Deinococcus radiodurans—the consummate survivor. 25. Mockler TC, Chan S, Sundaresan A, Chen H, Jacobsen SE, et al. (2005) Nat Rev Microbiol 3: 882–892. Applications of DNA tiling arrays for whole-genome analysis. Genomics 85: 1–15. 13. Zahradka K, Slade D, Bailone A, Sommer S, Averbeck D, et al. (2006) Reassembly 26. Shirkey B, McMaster NJ, Smith SC, Wright DJ, Rodriguez H, et al. (2003) of shattered chromosomes in Deinococcus radiodurans. Nature 443: 569–573. Genomic DNA of Nostoc commune (Cyanobacteria) becomes covalently modified 101 PLoS Genetics | www.plosgenetics.org 11 March 2009 | Volume 5 | Issue 3 | e1000434 Genome and Proteome of Deinococcus deserti

during long-term (decades) desiccation but is protected from oxidative damage 40. Udupa KS, O’Cain PA, Mattimore V, Battista JR (1994) Novel ionizing and degradation. Nucleic Acids Res 31: 2995–3005. radiation-sensitive mutants of Deinococcus radiodurans. J Bacteriol 176: 7439–7446. 27. Klein C, Aivaliotis M, Olsen JV, Falb M, Besir H, et al. (2007) The low 41. Snider J, Houry WA (2006) MoxR AAA+ ATPases: a novel family of molecular molecular weight proteome of Halobacterium salinarum. J Proteome Res 6: chaperones? J Struct Biol 156: 200–209. 1510–1518. 42. Sommer S, Boudsocq F, Devoret R, Bailone A (1998) Specific RecA amino acid 28. Schiex T, Gouzy J, Moisan A, de Oliveira Y (2003) FrameD: a flexible program changes affect RecA-UmuD’C interaction. Mol Microbiol 28: 281–291. for quality check and gene prediction in prokaryotic genomes and noisy matured 43. Sutton MD, Smith BT, Godoy VG, Walker GC (2000) The SOS response: eukaryotic sequences. Nucleic Acids Res 31: 3738–3741. recent insights into umuDC-dependent mutagenesis and DNA damage tolerance. 29. Zhu H, Hu GQ, Yang YF, Wang J, She ZS (2007) MED: a new non-supervised Annu Rev Genet 34: 479–497. gene prediction algorithm for bacterial and archaeal genomes. BMC Bioinfor- 44. Erill I, Campoy S, Mazon G, Barbe J (2006) Dispersal and regulation of an matics 8: 97. adaptive mutagenesis cassette in the bacteria domain. Nucleic Acids Res 34: 30. Earl AM, Mohundro MM, Mian IS, Battista JR (2002) The IrrE protein of 66–77. Deinococcus radiodurans R1 is a novel regulator of recA expression. J Bacteriol 184: 45. Boshoff HI, Reed MB, Barry CE III, Mizrahi V (2003) DnaE2 polymerase 6216–6224. contributes to in vivo survival and the emergence of drug resistance in 31. Hua Y, Narumi I, Gao G, Tian B, Satoh K, et al. (2003) PprI: a general switch Mycobacterium tuberculosis. Cell 113: 183–193. responsible for extreme radioresistance of Deinococcus radiodurans. Biochem 46. Galhardo RS, Rocha RP, Marques MV, Menck CF (2005) An SOS-regulated Biophys Res Commun 306: 354–360. operon involved in damage-inducible mutagenesis in Caulobacter crescentus. 32. Daly MJ, Gaidamakova EK, Matrosova VY, Vasilenko A, Zhai M, et al. (2004) Nucleic Acids Res 33: 2603–2614. Accumulation of Mn(II) in Deinococcus radiodurans facilitates gamma-radiation 47. Koorits L, Tegova R, Tark M, Tarassova K, Tover A, et al. (2007) Study of resistance. Science 306: 1025–1028. involvement of ImuB and DnaE2 in stationary-phase mutagenesis in Pseudomonas 33. Norioka N, Hsu MY, Inouye S, Inouye M (1995) Two recA genes in Myxococcus putida. DNA Repair (Amst) 6: 863–868. xanthus. J Bacteriol 177: 4179–4182. 34. Nahrstedt H, Schroder C, Meinhardt F (2005) Evidence for two recA genes 48. Nohmi T (2006) Environmental stress and lesion-bypass DNA polymerases. mediating DNA repair in Bacillus megaterium. Microbiology 151: 775–787. Annu Rev Microbiol 60: 231–253. 35. Swingley WD, Chen M, Cheung PC, Conrad AL, Dejesa LC, et al. (2008) Niche 49. Vallenet D, Nordmann P, Barbe V, Poirel L, Mangenot S, et al. (2008) adaptation and genome expansion in the chlorophyll d-producing cyanobacte- Comparative analysis of Acinetobacters: three genomes for three lifestyles. PLoS rium Acaryochloris marina. Proc Natl Acad Sci USA 105: 2005–2010. ONE 3: e1805. doi:10.1371/journal.pone.0001805. 36. Vujicic-Zagar A, Dulermo R, Le Gorrec M, Vannier F, Servant P, et al. (2009) 50. Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, et al. (1997) Gapped Crystal structure of the IrrE protein, a central regulator of DNA damage repair BLAST and PSI-BLAST: a new generation of protein database search in Deinococcaceae. J Mol Biol 386: 704–716. programs. Nucleic Acids Res 25: 3389–3402. 37. Liu H, Sadygov RG, Yates JR III (2004) A model for random sampling and 51. Lowe TM, Eddy SR (1997) tRNAscan-SE: a program for improved detection of estimation of relative protein abundance in shotgun proteomics. Anal Chem 76: transfer RNA genes in genomic sequence. Nucleic Acids Res 25: 955–964. 4193–4201. 52. Griffiths-Jones S, Moxon S, Marshall M, Khanna A, Eddy SR, et al. (2005) 38. Mueller LN, Brusniak MY, Mani DR, Aebersold R (2008) An assessment of Rfam: annotating non-coding RNAs in complete genomes. Nucleic Acids Res software solutions for the analysis of mass spectrometry based quantitative 33: D121–124. proteomics data. J Proteome Res 7: 51–61. 53. Achaz G, Boyer F, Rocha EP, Viari A, Coissac E (2007) Repseek, a tool to 39. Zimmerman JM, Battista JR (2005) A ring-like nucleoid is not necessary for retrieve approximate repeats from large DNA sequences. Bioinformatics 23: radioresistance in the Deinococcaceae. BMC Microbiol 5: 17. 119–121.

102 PLoS Genetics | www.plosgenetics.org 12 March 2009 | Volume 5 | Issue 3 | e1000434

IV Mise au point des outils génétiques pour Deinococcus deserti

103 104 IV) Mise au point des outils génétiques pour Deinococcus deserti

Une étape nécessaire à l’étude de la radiotolérance de D. deserti est d’avoir la possibilité de modifier son patrimoine génétique. Pour réussir à transformer D. deserti, il faut que l’ADN à transformer ne soit pas dégradé par cette dernière (avant ou après être entré dans la cellule) et qu’il puisse pénétrer dans la cellule. De plus, si l’ADN rentre, il faut pouvoir sélectionner les bactéries transformées par cet ADN. Ainsi, les gènes de sélection doivent pouvoir s’exprimer. Dans le cas d’un plasmide, il faut qu’il soit transmissible à la descendance. Deux possibilités s’offre à nous. Soit le plasmide est répliquable au sein de la cellule (grâce à une origine de réplication fonctionnellement reconnue par des protéines nécessaires à la réplication) soit il est capable de s’insérer au niveau de l’ADN génomique. Pour travailler avec D. deserti, nous devons disposer de plasmides réplicatifs, utilisables à la fois chez D. deserti et E. coli (qui servira à la construction de plasmides) et de plasmides capables de s’insérer dans le génome afin d’inactiver un gène cible, ou mieux, un plasmide ou une autre construction permettant de faire un knock-out par double recombinaison homologue (remplacement d’un gène cible par une cassette de résistance à un antibiotique). Pour obtenir ces outils génétiques, plusieurs méthodes de transformation et plusieurs plasmides ou autres constructions ont été testés.

1) Electroporation

Contrairement à D. radiodurans, D. deserti n’est pas naturellement transformable dans les conditions testées. Des études préliminaires au laboratoire (réalisées par A. de Groot) ont suggéré que D. deserti était transformable par électroporation. En effet, l’ADN génomique d’une souche de D. deserti rendue résistante à la rifampicine peut être utilisée pour transformer une souche de D. deserti sauvage afin d’obtenir des bactéries résistantes à la rifampicine.

1.1) Obtention d’un plasmide réplicatif chez D. deserti

Nous avons d’abord testé si pI3 et pI8, des plasmides réplicatifs chez D. radiodurans et E. coli, et pZT29, un plasmide réplicatif chez D. grandis et E. coli, pouvaient servir à transformer D. deserti par électroporation. L’expression du gène cat qui confère la résistance au chloramphénicol sur ces plasmides est permise grâce à un promoteur de D. radiodurans. Les plasmides pI8 et pZT29 n’ont pas permis d’obtenir de transformants de D. deserti. 105 Différentes hypothèses peuvent expliquer cela : (1) la transformation par électroporation n’est pas très efficace, (2) les plasmides sont dégradés avant ou après être entrés dans la cellule, (3) les plasmides ne se répliquent pas, et/ou (4) le gène cat n’est pas (assez) exprimé pour conférer une résistance au chloramphénicol chez D. deserti. Meima and Lidstrom (Meima & Lidstrom, 2000) ont montré que chez D. radiodurans, le plasmide pI3, dont est dérivé le

Figure 34 : Schéma des plasmides pI3 et pI8 (Meima & Lidstrom, 2000). pI3, un plasmide réplicatif chez D. radiodurans et E. coli, a été obtenu par sous clonage de pUE10, le petit plasmide naturel de la souche SARK de D. radiodurans, dans le plasmide pKK232-8 utilisable chez E. coli. Seul le fragment d’ADN issu de pUE30 est représenté. pI8 correspond au plasmide pI3 tronqué du fragment KpnI-KpnI. La flèche représente l’emplacement d’un promoteur (P) permettant l’expression du gène cat (résistance au chloramphénicol) situé en amont du fragment cloné. Ori : Origine de réplication, E : EcoRI, K : KpnI, H : HindIII. orfA, orfB, htrU, orfC, orfD, orfE, orfF, ardU, orfG, orfH, repU et resU sont les gènes présents dans le fragment cloné.

plasmide pI8 (Fig. 34), permettait d’obtenir 100 fois plus de transformants que pI8. Nous avons alors essayé d’introduire pI3 dans D. deserti par électroporation. Nous avons extrait ce plasmide d’E. coli Top10, de D. radiodurans et également de D. grandis où nous avons réussi a l’introduire en suivant un protocole d’électroporation fournit par I. Narumi. Après un premier essai, seul le plasmide issu de D. grandis (quantité d’ADN indéterminée) a permis l’obtention d’une centaine de colonies, après environ 20h de culture d’expression à 30°C. Un second essai a permis d’obtenir une 50aine de transformants avec 1 μg de pI3 issu d’E. coli Top10 ou d’un mutant dam dcm (cette souche, affectée dans la méthylation de l’ADN, est utilisée au cas où D. deserti reconnait l’ADN méthylé comme étranger et le dégrade). Il est possible que le plasmide issu de D. radiodurans n’ait pas permis d’obtenir de transformants car sa concentration était plus faible ou parce qu’il a été dégradé dans D. deserti. Après ré- isolement sur milieu contenant 2 ou 3 μg/mL de chloramphénicol (D. deserti est sensible à 1- 2 μg/mL de chloramphénicol ; 2 ou 3 μg/mL de chloramphénicol est utilisée couramment pour sélectionner des transformants de D. radiodurans), et après extraction du plasmide, nous avons confirmé que les colonies CmR étaient bien porteuses du plasmide pI3. En parallèle, le plasmide pRD4 (pZT29 modifié pour que repD et cat soit exprimés chez D. deserti) a été construit afin d’avoir un second plasmide réplicatif chez D. deserti. Malheureusement aucun transformant n’a été obtenu. Ainsi, seul le plasmide pI3 a permis d’obtenir des transformants chez D. deserti. Toutefois, la transformation est beaucoup moins efficace chez D. deserti que 106 chez D. radiodurans où la transformation de pI3 issu d’E. coli donne des milliers de transformants.

1.2) Construction des cassettes de résistance et tests de mutagenèse par recombinaison après électroporation

L’obtention des transformants de D. deserti par électroporation permet d’envisager la mise au point de plasmides capables d’inactiver des gènes. Des études préliminaires avaient été menées au laboratoire pour tester la construction des mutants par l’insertion de plasmides au niveau du génome de D. deserti. Pour cela nous avions choisi des plasmides utilisables chez E. coli, tels que pCRXL-Topo, pGEM-T, pUC19 et pBR322, mais ne se répliquant pas chez Deinococcus. L’utilisation de tels plasmides dans lesquels avaient été clonés, pour permettre la recombinaison homologue, des fragments d’ADN de 400 et 800 pb de D. deserti n’avait pas permis d’obtenir de transformants. Il est possible que cela soit dû à une absence d’expression des gènes de sélection ou au fait que les plasmides ne soient pas insérés par recombinaison. À mon arrivée au laboratoire, nous avons choisi de modifier les plasmides pUC19 et pBR322 pour que ces plasmides permettent (1) l’expression d’un gène de résistance à un antibiotique chez D. deserti et (2) l’insertion au niveau du gène amyE (Deide08810) grâce au clonage d’un fragment d’un kilobase de ce gène (amyE*). Lors de l’élaboration des outils génétiques chez D. radiodurans, la cible choisie avait été le gène amyE, un gène non essentiel chez D. radiodurans (Lecointe et al., 2004a) d’où son utilisation lors de cette étude. Les plasmides pDR5, dérivé de pUC19 et pRD6, pRD7 et pRD8, dérivés de pBR322 ont été construits (Fig. 35). Pour permettre l’expression du gène conférant la résistance à un antibiotique, le promoteur putatif du gène tufA (Deide06120, un gène codant pour le facteur d’élongation EF-Tu), de katE (DeideP200330, le gène codant la catalase E) ou de groESL (Deide22580-Deide22590, des gènes codants pour des chaperonnes), ont été clonés directement (fusion traductionnelle) en amont du codon start du gène de résistance à la kanamycine (pRD5) ou en amont du promoteur naturel du gène de résistance à la tétracycline (pRD6, pRD7, pRD8). Les promoteurs des gènes tufA, katE, groESL ont été choisi parce que

BamHI PvuI EcoRI Figure 35 : Schémas des plasmides pRD5 à Ptuf::kan PstI tet P pRD8. Les promoteurs de pRD5, pRD6, pRD7, EcoRI amyE* amyE* pRD8 sont respectivement ceux des gènes tufA, HindIII groES, katE et tufA. amyE* signifie un pRD5 PstI pRD6, 7, 8 fragment du gène amyE.

107 ce sont des promoteurs forts et constitutifs, et que des cassettes similaires sont fonctionnelles chez D. radiodurans (Lecointe et al., 2004a). Les fragments clonés correspondent à des séquences d’environ 300 pb, situé en amont des gènes groESL, katE et tufA, car généralement les promoteurs se situent dans les 300 pb en amont des gènes. Aucun des plasmides n’a permis d’obtenir de transformants et ce quelles que soient les différences apportées aux paramètres de l’électroporation (18 ou 25 kV/cm, électroporation à température ambiante ou à 4°C). Pour vérifier la fonctionnalité des cassettes Ptuf::kan, Ptuf::tet, Pgro::tet et Pkat::tet chez D. deserti, celles-ci ont été introduites dans pI3. Les plasmides pRD9 (pI3 Ptuf::kan), pRD13 (pI3 Pkat::tet), pRD14 (pI3 Pgro::tet) et pRD15 (pI3 Ptuf::tet) ont été introduits dans la souche sauvage de D. deserti (sélection sur chloramphénicol) puis les transformants ont été testés pour leurs résistances à la kanamycine ou à la tétracycline. pRD9 permet à D. deserti de devenir résistante à la kanamycine (jusqu’à au moins 20 µg/mL), alors que pRD13, pRD14 et pRD15 ne lui permettent pas de devenir résistante à la tétracycline. Ainsi, seule la cassette kanamycine est utilisable et la raison pour laquelle nous n’avions pas obtenu de transformants avec pRD5 lors des expériences précédentes est peut-être liée à la dégradation du plasmide ou parce que l’électroporation de D. deserti n’est pas assez efficace.

2) La transformation par chimio-compétence

Afin d’améliorer la transformation de D. deserti, 2 autres protocoles ont été essayés.

Le premier est une méthode utilisant le CaCl2 (cellules chimio-compétentes ; ce chapitre) et l’autre est basé sur la conjugaison (Chapitre 3). La mise au point du protocole de transformation par la méthode de Cacl2 a été menée avec l’utilisation du plasmide pI3. Différents paramètres comme la température (température ambiante, 37 et 42°C) et le temps d’incubation pour le choc thermique (30 sec ou 2 min), la durée de la culture d’expression (3, 6, 9 ou 20 h) et le type de milieux (TSB/10 sol II ou TSB/10 sol II + 1% glucose) ont été testés. Ces différents tests ont tous permis d’obtenir des transformants, mais le meilleur résultat a été obtenu avec les paramètres indiqués dans le Matériels et Méthodes (100-1000 transformants/μg de pI3 extrait de E. coli). Les résultats obtenus avec cette méthode de transformation ont été meilleurs qu’avec le protocole d’électroporation utilisé. Ceci peut-être parce qu’il est possible d’utiliser une plus grande quantité d’ADN, sans risquer d’avoir un arc électrique comme c’est parfois le cas avec l’électroporation. Par ailleurs la transformation réalisée avec des cellules chimio-compétentes et pI3 issue d’E. coli Top10 ou dam dcm est 10 fois moins efficace qu’avec pI3 extrait de

108 D. deserti (1000-10000 transformants/μg d’ADN). Ces résultats suggèrent que des enzymes de restrictions de D. deserti dégradent l’ADN étranger. Nous avons aussi testé la transformation de pI8 dans D. deserti par cette méthode. Des transformants ont été obtenus, mais ceux-ci sont 10-30 fois moins nombreux, pour une même quantité d’ADN, qu’avec le plasmide pI3, pourtant plus grand de 6 Kb. Ainsi, D. deserti est plus facile à transformer avec pI3 qu’avec pI8, comme chez D. radiodurans. Meima et Lidstrom supposent que cela serait dû au gène ardU présent dans pI3 mais absent de pI8 (Fig. 34). ardU coderait une « anti-restriction protein » et permettrait de protéger pI3 de l’action de certaines enzymes de restriction (Meima & Lidstrom, 2000 ; Thomas et al., 2003).

3) La conjugaison

La transformation par conjugaison n’a pas été observée pour D. radiodurans (J. Battista, communication personnelle). Pour savoir s’il est possible de conjuguer D. deserti, le plasmide conjugatif à large spectre d’hôte pBBR1MCS a été modifié. Son dérivé, le plasmide pRD20 (Fig. 36), comporte la cassette de résistance Ptuf::kan ainsi que le fragment

mob

cat rep

pRD20 BamHI HindIII amyE* Ptuf::kan Figure 36 : Schéma du plasmide pRD20. amyE* pour permettre l’insertion du plasmide dans le chromosome au cas où ce plasmide ne se répliquerait pas chez D. deserti. Différents paramètres ont été testés pour réaliser la conjugaison triparentale, comme l’utilisation de bactéries dans différentes phases de croissance (phase exponentielle ou stationnaire des 3 bactéries, E. coli contenant le plasmide helper pRK2013, E. coli « donneuse » et D. deserti), le milieu sur lequel a été testé la conjugaison (LA ou TSA/10 sol II) ainsi que les quantités de bactéries mises à conjuguer (10 μL de chaque souche d’E. coli avec 80 μL de D. deserti ou 50 μL de chaque bactérie). Après une nuit à 30°C, les cellules ont été ressuspendues dans 500 μL de TSB/10 sol II et 50 μL ont été étalées sur milieu sélectif Nx 15 μg/mL + Kan 3 μg/mL. Environ 25 transformants ont été obtenus avec les conditions décrites dans le Matériels et Méthodes. Sept transformants ont été testés par PCR sur la cassette Ptuf::kan et 5 se sont révélés positifs. Un transformant a été sélectionné et les PCRs montraient l’insertion du plasmide dans le

109 chromosome au niveau du gène amyE (données non montrées). La conjugaison est, comme la transformation par choc thermique, un moyen pour transformer D. deserti. De plus ces expériences ont également permis de montrer qu’il est possible d’insérer un plasmide au niveau de l’ADN génomique de D. deserti.

4) Intégration d’un plasmide par simple recombinaison homologue à l’aide de cellules chimio-compétentes

Un protocole de transformation plus efficace ayant été mis au point pour D. deserti et l’insertion d’un plasmide au niveau de l’ADN génomique par conjugaison ayant réussi, la mutagenèse par insertion d’un plasmide a été retentée, mais cette fois-ci par transformation avec des cellules chimio-compétentes. Pour essayer d’inhiber l’éventuelle dégradation du plasmide à transformer, le gène ardU ainsi que son promoteur putatif (∼100 pb), provenant du plasmide pI3, ont été introduits dans le plasmide pRD5, pour former pRD10 (Fig. 37). ardU est un gène qui coderait pour une « anti-restriction protein », capable d’inhiber certaines enzymes de restriction (voir Chapitre 2).

BamHI Ptuf::kan

ardU amyE* Figure 37 : Schéma du plasmide pRD10. EcoRI pRD10

La transformation avec le plasmide pRD10 (environ 500 ng) a permis d’obtenir une quinzaine de colonies résistantes à la kanamycine. Deux temps de culture d’expression (6 et 22 h) avaient été choisis lors de la transformation. Six colonies (5 provenant des 22 h d’incubation et 1 provenant des 6 h d’incubation) ont été ré-isolées et testées par PCR. Seule la colonie ayant été incubée 6 h était un vrai transformant car elle était positive pour la cassette Ptuf::kan. Nous avons pu montrer que pRD10 s’était intégré au niveau du gène amyE (Fig. 38A, B, C). Afin de vérifier que l’insertion a été possible grâce à ardU, le plasmide pI8 ardU (pRD7a) a été construit. Le nombre de transformants obtenu avec une même quantité de pI8 ou de pRD7a s’est avéré similaire, suggérant que le gène ardU cloné, n’est pas exprimé ou, que seul, il n’est pas suffisant pour protéger la molécule d’ADN à introduire.

110 Il a été montré que D. radiodurans possède plusieurs copies de son génome. Pour D. deserti, ce nombre de copies est inconnu. Nous avons alors analysé, par PCR, si le mutant d’insertion est hétérozygote (au moins une copie du génome présente l’intégration de pRD10 et au moins une copie du génome est sauvage) ou homozygote (toutes les copies du génome présente l’intégration de pRD10) (Fig. 38D et E). Si la souche est homozygote, avec les amorces AmyDnLComp et AmyUR3Comp, nous devrions voir un fragment d’environ 7 Kb,

A B amyE Figure 38 : Insertion du plasmide Chr Ptuf::kan ardU pRD10 dans le chromosome de D. deserti au niveau du gène amyE. (A) amyE* 1,5 1 Schéma de l’insertion par recombinaison Ptuf::kan homologue au niveau du gène amyE. pRD10 Les différentes amorces sont indiquées par des flèches (B) L’insertion de ardU C pRD10 dans le chromosome vu par AB PCRs sur Ptuf::kan (TufBam et KanPst) et ardU (ArdUBamHI et ArdUEcoRI). ardU (C) Gel d’agarose montrant que pRD10 B 1,5 est inséré au niveau d’amyE puisque les 1 A fragments A (Puc19A et Ptuf::kan AmyUR3Compl) et B (KanPstComp et AmyDnLComp) sont amplifiés. (D) E Schéma montrant soit une souche D homozygote, soit une souche Si homozygote hétérozygote pour l’insertion de pRD10 8 au niveau d’amyE. (E) PCR 1 bande 6 (AmyDnLComp et AmyUR3Comp) ≈ 7 kb essayant de déterminer si la souche est Si hétérozygote 2 hétérozygote ou homozygote. La bande 1,5 de la taille d’amyE indique que la 2 bandes souche est hétérozygote. Entre ≈ 1,5 kb 1 parenthèses sont indiquées les amorces. ≈ 7 kb correspondant à l’insertion de pRD10, alors que dans le cas d’une souche hétérozygote, nous devrions avoir 2 fragments, un de 7 Kb et un de 1,5 Kb, ce dernier correspondant au gène amyE. La figure 37E ne montre qu’un seul fragment, correspondant au gène amyE, or, nous venons de montrer que le plasmide était inséré à l’intérieur de ce gène. Ainsi ce mutant présente au moins une copie du gène entier amyE et au moins une copie où le plasmide pRD10 a été inséré. Nous pouvons donc conclure que ce mutant est hétérozygote et que D. deserti possède au moins 2 copies de son génome. Il est possible que le fragment de 7 kb soit difficile à obtenir compte tenu du GC % de D. deserti (63 %). L’insertion d’un plasmide se fait en tandem chez D. radiodurans (Smith et al., 1988). Il est possible que cela soit également le cas chez D. deserti et donc ce ne serait pas un fragment de 7 kb que nous devrions amplifier mais un fragment de plusieurs dizaines de kilobases. Un Southern blot

111 permettrait de montrer ce qui s’est réellement passé lors de cette insertion. Afin d’obtenir une souche homozygote, le transformant a été repiqué 2 fois sur kanamycine 10 et 20 μg/mL et nous avons observé la disparition de la bande à 1,5 Kb alors que la bande correspondant à la cassette de sélection était toujours présente. Ainsi, le transformant est devenu homozygote. Vu que l’intégration de pRD10 dans amyE a fonctionné, des plasmides similaires ont été testés pour les gènes Deide16750, Deide16950 et DeideP101880. Ces gènes ont été choisis car ils pourraient intervenir dans différents aspects de la radiotolérance. Deide16950 coderait une glutathionylspermidine synthase, le glutathionylspermidine pourrait avoir un rôle dans la protection contre le stress oxydant. Deide16750 coderait une protéine de la famille des histones déacétylases, une protéine qui pourrait jouer un rôle dans l’organisation du nucleoïde. DeideP101880 coderait une polymérase translésionnelle de la famille Y (Voir Chapitre III). Les plasmides pRD25 (Deide16950), pRD26 (Deide16750) et pRD27 (DeideP101880) contiennent des fragments d’une taille de 500 à 700 pb du gène à inactiver. Malheureusement, la transformation avec ces plasmides n’a pas permis d’obtenir de transformants. L’inactivation des 3 gènes cités ci-dessus a également été tentée par conjugaison. Pour se faire, des plasmides similaires à pRD20 mais contenant non pas amyE* mais des fragments de Deide16750 (pRD28), Deide16950 (pRD29) et DeideP101880 (pRD30) ont été construits. Malheureusement là non plus, la conjugaison n’a pas permis d’obtenir de transformants. Il est possible que la taille des fragments clonés ne permette pas l’insertion des plasmides au niveau des gènes cibles. Par la suite, tous les fragments clonés servant à la délétion des gènes ont une taille supérieure à 1000 pb. Par la suite seule l’étude de DeideP101880 a été poursuivie. En conclusion, la transformation par choc thermique comme la conjugaison a permis d’insérer un plasmide au sein du génome de D. deserti. Cependant, d’un point de vue pratique, la méthode par choc thermique est plus simple à mettre en œuvre. Ainsi, par la suite, toutes les transformations de D. deserti ont été réalisées par la méthode utilisant les cellules chimio-compétentes.

5) Construction de mutants par double recombinaison homologue

Puisque l’insertion d’un plasmide dans le génome de D. deserti est possible, nous avons essayé de remplacer un gène par une cassette de résistance à un antibiotique grâce à une double recombinaison homologue. Pour cela, la cassette Ptuf::kan est flanquée de 2 fragments d’ADN d’environ 1Kb de D. deserti. Différentes méthodes ont été testées. Pour ne pas reconstruire des plasmides entièrement nouveaux, un fragment d’ADN, amyEUp (un

112 fragment d’ADN situé en amont du gène amyE) a été inséré dans les plasmides pRD5 et pRD10 pour donner les plasmides pRD11 et pRD18 (Fig. 39).

5.1) Utilisation d’un fragment PCR/plasmidique

Le fragment amyEUp-Ptuf::kan-amyE* a été obtenu par (1) amplification PCR à partir du plasmide pRD11 ou (2) digestion du plasmide pRD11 par EcoRI-HindIII. Malheureusement, la transformation avec ces 2 fragments d’ADN n’a pas permis d’obtenir de transformants, peut-être parce que l’ADN linéaire est dégradé dans la cellule.

BamHI BamHI Ptuf::kan Ptuf::kan

amyEUp amyE* amyEUp amyE*

EcoRI HindIII EcoRI HindIII pRD11 pRD18 ardU

EcoRI

Figure 39 : Schéma des plasmides pRD11 et pRD18.

5.2) Utilisation d’un plasmide circulaire contenant rpsL

Un premier test de transformation avec un plasmide circulaire (pRD11 et pRD18) n’avait pas permis de conclure sur le succès d’une double recombinaison. Le problème de l’utilisation de plasmide circulaire est que la souche peut être résistante au marqueur de sélection si le plasmide s’insert dans le génome par une simple recombinaison. Ainsi, pour

Ptuf::kan amyEUp amyE* HindIII Figure 40 : Schéma du plasmide pRD22. pRD22 ardU rpsL HindIII augmenter les chances d’obtenir une double recombinaison, nous avons construit le plasmide pRD22, qui contient le gène sauvage rpsL (Fig. 40). Il est connu que chez E. coli par exemple, un mutant rpsL qui est streptomycine résistant (SmR) redevient sensible à la streptomycine lorsqu’une copie sauvage du gène rpsL est apporté à ce mutant (Reyrat et al., 1998). Ainsi en utilisant un mutant rpsL, il est théoriquement possible d’obtenir, grâce à une contre sélection sur streptomycine, une double recombinaison en introduisant rpsL sauvage dans un plasmide comportant une cassette de sélection flanquée par 2 fragments d’ADN

113 servant à la recombinaison homologue. Il suffit de sélectionner, dans un premier temps, les transformants sur le premier marqueur de sélection (ex : kanamycine), puis dans un second temps de contre-sélectionner sur kanamycine + streptomycine. Ainsi seules seront sélectionnées les bactéries ayant perdu le gène sauvage rpsL sur le plasmide et ayant conservé la cassette de résistance à la kanamycine. Les bactéries sélectionnées auront donc subit une double recombinaison homologue. Afin d’essayer d’adapter cette méthode à D. deserti, des mutants résistants à la streptomycine ont été obtenus par étalement de 1 mL de culture de cellules en phase stationnaire sur TSA/10 sol II contenant 10 μg/mL de streptomycine. Une de ces colonies SmR, la souche RD19, a été sélectionnée (l’analyse de la séquence de rpsL a montré que

A amyEUp amyE amyEDn amyEUp amyE** Chr Chr Ptuf::kan amyEDn A ardU Ptuf::kan rpsL Double recombinaison B C amyEUp amyE* pRD22 Simple L recombinaison

amyEUp rpsL amyEUp amyE** Chr amyE*** ardU Ptuf::kan amyEDn D E A B K C F B RD22 WT pRD22 RD22 WT RD22 pRD22 WT MT A B C A MT D E F D E F D MTKFKK

Fig 41 : Schémas et PCR de l’insertion de pRD22. (A) Schémas de l’insertion de pRD22 après simple (dans amyE) ou double recombinaison. Les flèches colorées représentent les différentes amorces utilisées lors de la PCR. (B) PCR sur un transformant RD22 après réisolement sur kan 10 + Sm 10. Des contrôles PCR ont été réalisés avec de l’ADN provenant de la souche sauvage (WT) et le plasmide pRD22. A : AmyE (amorces AmyUR3Comp + AmyDnLComp) ; B : Ptuf::kan (TufBam + KanPst) ; C : recombinaison par amyE* (KanPstComp + AmyDnLComp) ; D : ardU (ArdHd3 + ArdSal) ; E : rpsL-amyE (AmyPst + RpsLleftHdIII) ; F : kan (KanTuf + KanPst) ; K : ardU-amyEUp (ArdHd3 + amyUR3). L : recombinaison par amyEUp (TufBamComp2 + AmyEUpUp), MT : marqueur de taille. amyE** et amyE*** sont les différents fragments du gène amyE qui peuvent se former lors de l’insertion du plasmide pRD22. Les PCRs D (ardU) et K (ardU-amyUp) présentent des bandes aspécifiques lorsqu’elles sont réalisées sur RD22 ou la souche sauvage. Les bons fragments correspondent à ceux qui sont obtenus avec la matrice pRD22 (pour la PCR K, la bande la plus intense avec pRD22 est le fragment de la taille attendue). La PCR L ne donne pas de produit PCR et ne figure pas sur les gels.

114 RD19 contient une mutation qui entraîne le changement du résidu K88 en arginine). Le gène rpsL de D. deserti ainsi que 200 pb en amont et 50 pb en aval de ce gène ont été cloné dans le plasmide pRD18 pour obtenir le plasmide pRD22 (Fig. 40). Ce plasmide a été introduit dans RD19 et les transformants ont été ré-isolées sur kanamycine 10 µg/mL puis repiqués sur kanamycine 10 µg/mL et streptomycine 10 µg/mL. Lors du ré-isolement sur kanamycine 10 µg/mL, les PCRs effectuées sur un transformant semblaient indiquer qu’il y avait eu double recombinaison mais cela n’était pas très clair. La figure 41 présente le résultat des PCRs effectuées sur un transformant ré-isolé sur kanamycine 10 µg/mL et streptomycine 10 µg/mL. Si le transformant a subi une seule recombinaison (dans amyE), les PCRs amplifiant Ptuf::kan (Fig. 41B, piste B), un fragment de la cassette de sélection avec une partie du chromosome (Fig. 41B, piste C), ardU (Fig. 41B, piste D), et un fragment du plasmide avec rpsL ou ardU (Fig. 41B pistes E et K) devraient être positives. À l’inverse, si les transformants ont subi une double recombinaison homologue, alors seules les PCRs amplifiant Ptuf::kan (Fig. 41B, piste B), un fragment de la cassette de sélection avec une partie du chromosome (Fig. 41B, piste C) devraient être positives. Les PCRs montrent que seuls sont amplifiés les fragments Ptuf::kan et kan-AmyEDn dans le transformant (Fig. 41B). Malheureusement, l’amorce AmyEUpUp ne fonctionne pas, ce qui fait que l’insertion du coté AmyEUp n’a pas pu être vérifiée (Fig. 41A, PCR L). De plus, la PCR A (AmyDnLComp + AmyUR3Comp) est négative chez le transformant testé, probablement parce que le fragment à amplifier est grand (∼2,3 Kb). Ce résultat ne permet pas de conclure définitivement sur l’amélioration du protocole permettant d’obtenir des doubles recombinants à l’aide de la contre sélection par la streptomycine. Cependant, cela montre que la double recombinaison est possible chez D. deserti. La souche nouvellement créée a été nommé RD22. Afin de vérifier que rpsL aide à la construction de mutant par double recombinaison homologue et également pour mettre au point une technique de délétion de gène qui ne laisse plus de cassette de sélection, nous avons construit pRD54 (pour obtenir une délétion de DeideP100132 codant l’exonucléase 5’ˆ3’ chez RD19) (Fig. 42). Comme dit précédemment, la contre sélection par la streptomycine peut aider à réaliser des mutants par double recombinaison. Malheureusement, cette contre-sélection n’a pas fonctionné car contrairement à E. coli l’ajout du gène sauvage rpsL ne rend pas RD19 sensible à la streptomycine. En effet, la souche ayant inséré le plasmide pRD54 entier par simple recombinaison est toujours

DeideP100132Up Ptuf::kan DeideP100132Dn

pRD54 rpsL

Figure 42 : Schéma du plasmide115 pRD54. capable de croître sur kanamycine + streptomycine. D’autres constructions ont montré que rpsL n’est pas nécessaire pour obtenir une double recombinaison. La figure 43 donne un exemple d’un plasmide (pRD42) ne contenant pas la copie rpsL sauvage et où la construction d’un mutant de délétion (ici d’irrE) a réussi. Ceci confirme que le knock-out d’un gène est réalisable chez D. deserti. À noter également que nous avons pu utiliser la cassette de résistance au chloramphénicol du plasmide pGTC101 pour construire des doubles mutants (KmR, CmR). Cette cassette contient le promoteur tuf de D. radiodurans fusionné au gène cat qui confère la résistance au chloramphénicol.

Fig. 43 : Délétion du gène irrE chez D. deserti. (A) Représentation schématique du remplacement d’irrE. En gris apparaissent les fragments upstream et downstream d’irrE utilisés pour permettre la double recombinaison homologue. La ligne épaisse représente le chromosome de D. deserti alors que la ligne fine représente le reste du plasmide pRD42. Les flèches représentent les amorces utilisées durant la PCR pour analyser le mutant RD42 (ΔirrE) : 1, TufBam; 2, KanPst; 3, IrrE5’; 4, IrrE3’; 5, KanPstComp; 6, IrrEDnDn; 7, TufBamComp2; 8, IrrEUpUp. (B) PCRs servant à confirmer l’absence d’irrE et la présence de la cassette de sélection au bon locus chez le mutant RD42. L’ADN du mutant RD42 et de la souche sauvage (WT) ont été utilisés comme matrice pour les réactions PCR. Ces réactions PCR ont été réalisées à l’aide des primers forward et reverse indiqués (les numéros des amorces correspondent à ceux de la partie A de cette figure).

6) Conclusions

Pour étudier génétiquement D. deserti, nous avons mis au point des outils génétiques.

Pour D. deserti la transformation par chimio-compétence (CaCl2) s’avère être la méthode la plus appropriée. A l’heure actuelle nous disposons d’un plasmide réplicatif, pI3. La délétion de gènes est réalisable chez D. deserti et pour cela, 2 cassettes de sélection (résistance à la kanamycine et au chloramphénicol) sont utilisables. Ainsi, D. deserti est la 2ième Deinococcus pour laquelle des outils génétiques permettant l’étude de gènes cibles ont été développés.

116

V Deux protéines RecA fonctionnellement différentes et mutagenèse induite par les

UV chez Deinococcus deserti

117

118 V) Deux protéines RecA fonctionnellement différentes et mutagenèse induite par les UV chez Deinococcus deserti

1) Introduction

En plus des gènes communs aux Deinococcus, D. deserti possède des gènes supplémentaires potentiellement liés à la réparation de l’ADN (voir chapitre III). Parmi ces gènes, 3 codent des polymérases translésionnelles putatives (imuY, dnaE2 et polB) et un code pour une photoproduct-lyase putative (DeideP302150). De plus, D. deserti possède 3 gènes recA alors que la majorité des bactéries n’en a qu’un. Les protéines homologues aux polymérases translésionnelles ou à la photoproduct- lyase peuvent aider D. deserti, une bactérie du Sahara, à réparer des dommages de l’ADN induits par les UV. Les polymérases translésionnelles aident à passer les lésions de l’ADN lors de la réplication, mais peuvent induire des erreurs (Tippin et al., 2004), alors que les photoproduct-lyases réparent les lésions par réversion directe. Les gènes imuY et dnaE2 de D. deserti sont localisés directement après un gène codant un homologue du répresseur transcriptionnel LexA (lexAP1). Cette organisation est très similaire à celle de la « mutagenesis cassette » trouvée chez différents organismes (Fig. 44) (Erill et al., 2006). Ce type de cassette

lexAP1 imuY hyp dnaE2 D. deserti

lexA imuA imuB dnaE2 P. putida

lexA imuA imuB dnaE2 C. crescentus

lexA imuA’ imuB dnaE2 M. tuberculosis

Figure 44 : La « mutagenesis cassette » de D. deserti. La « mutagenesis cassette » est trouvée dans plusieurs organismes comme P. putida, C. crescentus ou M. tuberculosis, et est organisée de différente manière. ImuY de D. deserti est faiblement homologue (< 30 % d’identité) aux polymérases de la famille Y, DinB/ImuB-like. Aucun homologue à ImuA n’a été trouvé chez D. deserti. Le produit putatif (80 aa) du gène hypothéthique (hyp) entre imuY et dnaE2 de D. deserti ne montre aucune homologie dans les bases de données. est impliqué dans la mutagenèse induite aux UV et/ou la survie aux UV de certaines bactéries (Koorits et al., 2007 ; Galhardo et al., 2005 ; Boshoff et al., 2003). De plus, des études montrent ou suggèrent que cette cassette est régulée par LexA et RecA (Abella et al., 2007 ; Galhardo et al., 2005 ; Erill et al., 2006). La protéine PolB de D. deserti appartient à la

119 famille des polymérases B. Chez E. coli, cette polymérase est impliquée dans la mutagenèse et provoque, entre autres, des mutations par décalage de cadre de lecture (Al Mamun, 2007 ; Fuchs et al., 2001). Enfin, DeideP302150 n’est que faiblement homologue à la spore photoproduct-lyase de B. subtilis (26 % d’identité), mais est beaucoup plus homologue (64 %) à des protéines annotées comme « radical SAM domain proteins », présentes chez plusieurs bactéries.

Les gènes recA de D. deserti sont localisés sur le chromosome (recAC) et les plasmides

P1 (recAP1) et P3 (recAP3). Les recA plasmidiques codent pour des protéines identiques

(RecAP), très homologue à RecAC (81% d’identité). Chez E. coli, RecA est essentielle à la réparation de l’ADN et, en stimulant l’auto-clivage du répresseur LexA, à l’induction des gènes de réparation de l’ADN, qui incluent recA et ceux codants les polymérases translésionnelles (Cox, 2007a). Chez D. radiodurans, RecA stimule également l’autoclivage de LexA. Cependant, l’induction de recA n’est pas contrôlée par LexA (Bonacossa de Almeida et al., 2002 ; Narumi et al., 2001 ; Satoh et al., 2006 ; Sheng et al., 2004), mais est dépendante d’IrrE (Earl et al., 2002a ; Hua et al., 2003), une protéine indispensable à la radiotolérance et à l’induction, après irradiation, d’une trentaine de protéines chez D. radiodurans (Lu et al., 2009). D. deserti possède aussi une protéine homologue à IrrE. Plusieurs questions se posent quant aux rôles et à la fonctionnalité de ces différents gènes supplémentaires chez D. deserti. Concernant les polymérases translésionnelles, il a été proposé que si D. radiodurans ne possédait pas ce type de polymérases, c’est probablement parce que leurs présences seraient imprudentes chez une bactérie capable de subir de très nombreux dommages de l’ADN (Sale, 2007). Ainsi, la présence de polymérases translésionnelles chez D. deserti peut sembler étonnante. Les polymérases translésionnelles putatives de D. deserti interviennent-elles dans la radiotolérance et/ou la mutagenèse ? Comment sont-elles régulées ? Par IrrE ? Par RecA ? Les bactéries ayant plus d’un gène recA sont extrêmement rares, ainsi, pourquoi D. deserti possède 3 gènes recA ? Les protéines

RecAC et RecAP ont-elles un rôle différent dans la cellule ? Les 2 gènes recA supplémentaires sont situés sur des plasmides et la majorité des autres gènes supplémentaires liés à la réparation de l’ADN chez D. deserti sont également situés sur des plasmides. Ainsi, existe-t-il un lien entre les recA plasmidiques et ces gènes supplémentaires plasmidiques ?

120 2) Résumé de l’article

Les travaux présentés dans cette partie ont donné lieu à une publication (voir la partie 4 de ce chapitre). Ma contribution à cet article concerne la construction et l’analyse

(radiotolérance et induction de la mutagenèse aux UV) des mutants des gènes recA (ΔrecAC,

ΔrecAP1, ΔrecAP3, ΔrecAC ΔrecAP1, ΔrecAC ΔrecAP3, ΔrecAP1 ΔrecAP3), des gènes codant les polymérases translésionnelles (ΔimuY, ΔdnaE2, Δ(imuY-dnaE2), ΔpolB), du gène codant la photoproduct-lyase putative (ΔDeideP302150) et du gène codant IrrE (ΔirrE). Nous avons tout d’abord montré que RecAC et RecAP étaient fonctionnelles pour la réparation de l’ADN.

Tous les mutants recA, sauf ΔrecAC ΔrecAP1 qui présente une faible quantité de RecAP, sont aussi radiotolérants que la souche sauvage. Le double mutant ΔrecAC ΔrecAP1 s’avère très sensible aux irradiations. De plus, il filamente (Fig. 45) et présente un retard de croissance important comparé à la souche sauvage (donnée non montrée). Ces résultats montrent que la faible quantité de RecAP présente dans le mutant ΔrecAC ΔrecAP1 n’est pas suffisante pour assurer une réparation efficace des dommages de l’ADN et le développement normal de la bactérie.

A B

Figure 45 : Morphologie du mutant ΔrecAC ΔrecAP1. Observation, au grossissement 1000 X, au microscope optique (Olympus BX50), en nomarski, du mutant ΔrecAC ΔrecAP1 (A) et de la souche sauvage (B) en phase exponentielle de croissance. Les images ont été prises avec la caméra Olympus U-PMTVC.

L’analyse des mutants ΔimuY, ΔdnaE2, Δ(imuY-dnaE2), ΔpolB et ΔDeideP302150 indiquent qu’ils ne sont pas affectés dans la survie aux irradiations gamma et UV. Ces résultats suggèrent que ces gènes ne sont pas importants dans la radiotolérance, au moins dans les conditions testées, ou que la seule délétion de ces gènes n’est pas suffisante pour affecter la radiotolérance de D. deserti. Cependant, nous avons montré que les gènes imuY et dnaE2, mais aussi recAC, sont impliqués dans l’induction de la mutagenèse aux UV chez D. deserti.

En effet, RecAC (et non RecAP) est nécessaire à l’induction après irradiation UV d’imuY et de

121 dnaE2. RecAC (et pas RecAP) induirait l’auto-clivage de LexAP1 (DeideP101870), la protéine qui serait impliquée dans la répression de l’opéron lexAP1-imuY-dnaE2. La délétion de lexAP1, bien qu’ayant un effet polaire sur les gènes imuY et dnaE2, a été tentée chez D. deserti, mais malheureusement le mutant ΔlexAP1 n’a pas pu être obtenu. Alors que les 2 protéines RecA sont impliquées dans la réparation de l’ADN, seule RecAC joue un rôle dans l’induction de la mutagenèse par les UV. C’est la première fois qu’est mise en évidence une différence de fonction entre 2 protéines RecA différentes au sein d’un même organisme. Enfin, nous avons montré qu’irrE est nécessaire, après irradiation, à la survie de D. deserti et à l’induction des 3 gènes recA et de DeideP302150. De plus, l’induction d’imuY et dnaE2 est conservée dans un mutant délété d’irrE, ce qui signifie qu’une quantité basale de

RecAC est suffisante pour cette induction.

3) Données supplémentaires

Les polymérases translésionnelles pourraient permettre à D. deserti de s’adapter aux conditions extrêmes dans lesquelles elle vit. Pour vérifier cette hypothèse, des tests de fitness ont été réalisés. Ces expériences servent à comparer la compétitivité des souches entre elles. Des tests de fitness ont, par exemple, permis de montrer que chez E. coli, les polymérases translésionnelles PolB, PolIV et PolV sont nécéssaires à la survie au long terme et donc à l’adaptation en condition limitante de nourriture (Yeiser et al., 2002). En effet, lorsque les mutants de ces gènes sont cultivés en présence de la souche sauvage, ils sont incapables de s’adapter au manque de nourriture et meurent très rapidement alors que la souche sauvage persiste. Les différentes souches à tester de D. deserti (ΔimuY, ΔdnaE2, Δ(imuY-dnaE2)

(testés 4 fois chacune), ΔpolB (testé 3 fois), ΔrecAC (testé 2 fois) et ΔirrE (testé 1 fois)) ont été cultivées jusqu’en phase stationnaire, puis diluées 1000 fois et des co-cultures (bactérie sauvage/bactérie mutante) ont été réalisées. Le nombre de CFU dans les co-cultures a été dénombré par étalement sur milieu non-sélectif (quantité totale de bactéries) et milieu sélectif avec antibiotique (quantité de bactéries mutantes). Malheureusement, les tests de fitness ont été peu reproductibles (voir exemples donnés dans la figure 46) et sont donc inexploitables. La souche délétée de DeideP302150 reste résistante aux rayons UV et gamma. Dans la nature, il n’y a pas de forts rayonnements gamma ni d’UV-C, par contre, il y a des UV-A et des UV-B. Peut-être que DeideP302150 sert à la protection contre les dommages engendrés par ces types d’UV. Ainsi, nous avons testé, une fois, la résistance aux UV-A de la souche ΔDeideP302150 (la résistance aux UV-B n’a pas été testée). Aucune différence de survie avec la souche sauvage n’a pu être mise en évidence, suggérant que DeideP302150 n’est pas

122 importante, dans les conditions testées, pour la résistance aux UV-A de D.deserti. Il est possible que DeideP302150 soit nécessaire à la réparation de lésions spécifiques formées lorsque la bactérie est soumise à la fois aux UV et à la dessiccation.

A B

WT WT ΔimuY ΔimuY

C Temps (jours) D Temps (jours)

WT WT ΔpolB ΔpolB

Temps (jours) Temps (jours) Figure 46 : Exemples des résultats des compétitions entre la souche sauvage et des souches délétées d’imuY ou de polB. (A et B) Compétition entre la souche sauvage et le mutant ΔimuY. (C et D) Compétition entre la souche sauvage et le mutant ΔpolB. Ces figures représentent 2 expériences indépendantes.

123 124 4) Article 2 : Mutagenic lesion bypass and two functionally different RecA proteins in Deinococcus deserti

Auteurs : Dulermo R., Fochesato S., Blanchard L. & de Groot A.

Journal : Molecular Microbiology

Année : 2009

Volume : 74

Pages : 194-208

À noter que les figures supplémentaires S1 et S2 se trouvent après l’article « Mutagenic lesion bypass and two functionally different RecA proteins in Deinococcus deserti ». Les autres données supplémentaires (les tableaux d’amorces S1 à S4, les outils génétiques et la figure S3) se trouvent dans les parties précédentes (Matériels et Méthodes et le chapitre IV).

125 126 Molecular Microbiology (2009) 74(1), 194–208 ᭿ doi:10.1111/j.1365-2958.2009.06861.x First published online 2 September 2009 Mutagenic lesion bypass and two functionally different RecA

proteins in Deinococcus desertimmi_6861 194..208

Rémi Dulermo,1,2,3 Sylvain Fochesato,1,2,3 Introduction Laurence Blanchard1,2,3 and Arjan de Groot1,2,3* 1CEA, DSV, IBEB, Lab Ecol Microb Rhizosphere & To preserve genome integrity, cells have developed strat- Environ Extrem (LEMiRE), Saint-Paul-lez-Durance, egies to protect and repair their genome. The bacterium F-13108, France. Deinococcus radiodurans and other members of the 2CNRS, UMR 6191 Biol Veget & Microbiol Environ, Deinococcaceae are famous for their extraordinary toler- Saint-Paul-lez-Durance, F-13108, France. ance towards high doses of gamma and UV radiation or 3Aix-Marseille Université, Saint-Paul-lez-Durance, long periods of desiccation. This extreme tolerance is F-13108, France. linked to the ability of these bacteria to accurately repair massive DNA damage, including hundreds of double- strand DNA breaks and thousands of other lesions (Cox Summary and Battista, 2005; Blasius et al., 2008). It has been pro- posed that DNA repair in Deinococcus could be facilitated RecA is essential for extreme radiation tolerance in by a high intracellular Mn/Fe ratio that prevents oxidation Deinococcus radiodurans. Interestingly, Sahara bac- of proteins (Daly et al., 2007) and by a highly condensed terium Deinococcus deserti has three recA genes nucleoid that may restrict diffusion of DNA fragments gen- (recA , recA , recA ) that code for two different C P1 P3 erated by irradiation (Zimmerman and Battista, 2005). RecA proteins (RecA , RecA ). Moreover, and in con- C P The genomes of D. radiodurans and Deinococcus geo- trast to other sequenced Deinococcus species, thermalis contain homologues of most Escherichia coli D. deserti possesses homologues of translesion syn- DNA repair genes (Makarova et al., 2001; 2007), and thesis (TLS) DNA polymerases, including ImuY and several of these contribute to various extents to gamma DnaE2. Together with a lexA homologue, imuY and and/or UV radiation tolerance (Blasius et al., 2008). The dnaE2 form a gene cluster similar to a widespread recombinational repair protein RecA is absolutely required RecA/LexA-controlled mutagenesis cassette. After for extreme radiation tolerance in D. radiodurans (Gutman having developed genetic tools, we have constructed et al., 1994; Bonacossa de Almeida et al., 2002). The mutant strains to characterize these recA and TLS products of additional genes specifically conserved in polymerase genes in D. deserti. Both RecA and C Deinococcus are also involved in DNA repair and radia- RecA are functional and allow D. deserti to survive, P tion tolerance. These include PprA (reported to stimulate and thus repair massive DNA damage, after exposure ligase activity) (Narumi et al., 2004), DdrA (an oligomeric to high doses of radiation. D. deserti is mutable by ring-forming protein involved in DNA extremity protection) UV, which requires ImuY, DnaE2 and RecA , but not C (Harris et al., 2004; Gutsche et al., 2008), DdrB, DdrC and RecA . RecA , but not RecA , facilitates induced P C P DdrD (functions unknown) (Tanaka et al., 2004; de Groot expression of imuY and dnaE2 following UV et al., 2009) and IrrE (regulator of recA and pprA expres- exposure. We propose that the extra recA and P1 sion) (Earl et al., 2002a; Hua et al., 2003; Lu et al., 2009; recA genes may provide higher levels of RecA P3 Vujicic-Zagar et al., 2009). Homologues of DNA repair protein for efficient error-free repair of DNA damage, photolyases and specialized translesion synthesis (TLS) without further increasing error-prone lesion bypass DNA polymerases have not been identified in by ImuY and DnaE2, whereas limited TLS may con- D. radiodurans and D. geothermalis. As lesion bypass by tribute to adaptation to harsh conditions by generat- TLS polymerases is potentially mutagenic, such activity ing genetic variability. might be imprudent in case of thousands of DNA lesions (Sale, 2007). Recently, the genome of Deinococcus deserti, a bacte- rium isolated from surface sand of the Sahara desert Accepted 14 August, 2009. *For correspondence. E-mail nicolaas. (de Groot et al., 2005), was sequenced and analysed (de [email protected]; Tel. (+33) 4 42 25 37 86; Fax (+33)442256648. Groot et al., 2009). Its genome consists of a 2.8 Mb

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd 127 Lesion bypass and two RecA in Deinococcus deserti 195 chromosome and three large plasmids of 324, 314 and our knowledge, this is the first time that different functions 396 kb. Besides many common genes, D. deserti has for different RecA proteins within a bacterial species are several additional DNA repair genes compared with demonstrated. D. radiodurans and D. geothermalis. Interestingly, among these are genes encoding three putative TLS poly- merases and a potential photoproduct lyase. Moreover, Results and highly uncommon, D. deserti has three different recA Three recA, three TLS polymerases and a putative genes, one located on the chromosome and the other two photoproduct lyase in D. deserti on different plasmids. The plasmid recA genes code for identical proteins that have 81% identity with RecA In contrast to D. radiodurans and D. geothermalis, encoded by the chromosome. RecA is nearly ubiquitous D. deserti has genes encoding putative TLS DNA poly- (Rocha et al., 2005), and most known bacterial species, merases (Deide_1p00180, Deide_1p01880 and including D. radiodurans and D. geothermalis, possess Deide_1p01900) and a putative photoproduct lyase only one recA.TworecA were found in Bacillus megate- (Deide_3p02150), located on plasmid P1 and plasmid P3 rium (Nahrstedt et al., 2005), Myxococcus xanthus respectively (Fig. 1). It has also three different recA genes (Norioka et al., 1995), and seven in Acaryochloris marina (Deide_19450, Deide_1p01260 and Deide_3p00210, (Swingley et al., 2008). A recA was also found on a 65 kb present on the chromosome, plasmid P1 and plasmid P3 conjugative lactococcal plasmid (Garvey et al., 1997) and respectively). Except for Deide_1p00180, transcription of on a 53 kb plasmid from an environmental Vibrio strain all these genes was induced after exposure of D. deserti (Hazen et al., 2007). RecA from E. coli has been studied to UV, suggesting a role in the DNA damage response (de most extensively. It is a multifunctional protein with a key Groot et al., 2009). role in DNA repair and genetic recombination (Cox, Deide_1p00180 (PolB) shares up to 48% identity with 2007a,b). Through its coprotease activity, it also has a DNA polymerases of the B-family (DNA polymerase II). regulatory role in induction of the SOS response. Follow- Deide_1p01880 encodes a distantly related homologue of ing DNA damage, activated RecA nucleoprotein filaments the DinB subfamily of Y-family DNA polymerases. After are formed that facilitate the autocleavage of repressor BLAST analysis with Deide_1p01880, several proteins LexA, resulting in induction of many DNA repair genes annotated as DNA polymerase IV were found, but with including recA itself and TLS polymerase genes (Sutton less than 30% identity. As experiments in this work dem- et al., 2000; Erill et al., 2007; Butala et al., 2009). The onstrated that Deide_1p01880 is involved in induced RecA coprotease activity also facilitates cleavage of mutagenesis (see below), it will be named imuY. phage repressor proteins and of UmuD to UmuD′, a com- Deide_1p01900 (DnaE2) is a potentially error-prone ponent of TLS polymerase V (Pol V). In D. radiodurans, polymerase homologous to the alpha subunit DnaE RecA stimulates cleavage of its two LexA homologues, (Deide_21950) of DNA polymerase III. The genes imuY but contrary to E. coli, recA induction is not LexA- and dnaE2 are located close to each other, separated by dependent (Narumi et al., 2001; Bonacossa de Almeida a small hypothetical orphan gene of 243 bp (Fig. 1). The et al., 2002; Sheng et al., 2004; Satoh et al., 2006). gene immediately upstream of imuY encodes a LexA Instead, recA induction is dependent on a novel regula- homologue (Deide_1p01870), with the highest similarity tory protein IrrE, also called PprI (Earl et al., 2002a; Hua to D. radiodurans LexA (DR_A0344, 50% identity). et al., 2003). RT-PCR analysis indicated that lexA, imuY and dnaE2 are In this study, we characterized the three recA genes, as part of a single operon (Fig. S1). Similar so-called well as the genes encoding the potential TLS poly- mutagenesis cassettes, in different configurations, have merases and photoproduct lyase in D. deserti. This work recently been described in various groups of bacteria, but included the development of tools allowing the construc- not previously in members of the Deinococcus–Thermus tion of D. deserti gene deletion mutants. Survival after phylum, and shown to be under RecA/LexA control (Erill exposure to radiation, UV-induced mutagenesis and regu- et al., 2006; 2007). In most of these cassettes, a gene lation of gene expression were investigated in wild-type called imuA is present upstream of the Y polymerase and mutant strains. Analysis of various recA mutant gene (imuB), but an imuA homologue was not found in the strains showed that both RecA proteins are functional for genome of D. deserti. Furthermore, similarity between DNA repair. Induced expression of each recA was depen- D. deserti ImuY and ImuB proteins is very low, e.g. only dent on irrE. UV-induced mutagenesis was observed and 15% identity with ImuB of Caulobacter crescentus and shown to depend on two TLS polymerases, DnaE2 and a Pseudomonas putida. Y-family DNA polymerase, ImuY. Their radiation-induced Using the Conserved Domain Database (Marchler- expression was dependent on the chromosome-encoded Bauer et al., 2007), analysis of Deide_3p02150 revealed RecA, but not on IrrE and the plasmid-encoded RecA. To a COG1533 domain, defined as SplB DNA repair

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 128 196 R. Dulermo, S. Fochesato, L. Blanchard and A. de Groot ᭿

Fig. 1. Schematic representation of the three recA, the three TLS polymerase and the putative photoproduct lyase genes in D. deserti. The recA genes are shown in black, the TLS polymerase and photoproduct lyase genes in grey, and flanking genes in white. Also indicated is the replicon on which these genes are located. The black filled circles upstream each recA indicate the position of the conserved deinococcal radiation/desiccation response motif (RDRM) (see Discussion). Gene numbers and products are indicated above and below the genes respectively. HP, hypothetical protein; CHP, conserved hypothetical protein.

photolyase. However, Deide_3p02150 has only limited residues is present at the N-terminus of RecAC and RecA homology with spore photoproduct lyases such as Bacil- from D. radiodurans and D. geothermalis. This region is lus subtilis SplB (26% identity). It should be noted that flexible in the crystal structure of D. radiodurans RecA SplB is not a photolyase as it does not use light for activity (Rajan and Bell, 2004). The C-terminal 15–25 residues but a radical mechanism. BLAST analysis with are variable in sequence in RecA proteins and are disor- Deide_3p02150 revealed higher levels of identity (up to dered in published crystal structures. E. coli RecA activity 64%) with proteins annotated as radical SAM domain is autoregulated by its C-terminus; deletion of the last 17 proteins (SAM, S-adenosylmethionine). D. deserti and residues enhances most RecA activities (Cox, 2007a,b). several of the species that contain homologues of For the region of E. coli RecA that corresponds to resi-

Deide_3p02150 do not form spores. Like Deide_1p00180 dues 240–255 of D. deserti RecAC, analysis of point muta- (polB), Deide_3p02150 is not adjacent to other known tions and structural data have established that residues in DNA repair genes (Fig. 1). this region are implicated in coprotease substrate binding The genetic environments of the three recA genes in and cleavage (Dutreix et al., 1989; Mustard and Little, D. deserti are different (Fig. 1). As in D. radiodurans and 2000; McGrew and Knight, 2003; VanLoock et al., 2003). D. geothermalis, the chromosomal recA of D. deserti is the last gene in an operon that also contains cinA and Development of genetic tools for D. deserti ligT, encoding a DNA damage/competence-inducible CinA-like protein and a 2′-5′ RNA ligase respectively. The To investigate the role in D. deserti of the genes described two recA genes on the plasmids share 97% identity at the in the previous section, it was necessary to obtain genetic

DNA level and code for 100% identical proteins (RecAP), tools for this bacterium. Unlike D. radiodurans, D. deserti highly homologous to the chromosome-encoded RecAC appeared not to be naturally transformable under the con-

(81% identity). Both for RecAC (about 90% identity) and ditions tested. Therefore, other methods were sought to for RecAP (about 80% identity), the next best hits in BLAST genetically modify D. deserti. The aim was to obtain a analyses were the RecA proteins from D. radiodurans and transformation method, a plasmid that can replicate in

D. geothermalis, suggesting that RecAP is of deinococcal D. deserti and a system to inactivate/delete genes. We origin. Figure 2A permits a direct comparison of RecAC tested different transformation techniques (electropora- and RecAP. A multiple alignment of these proteins with tion, transformation using a CaCl2 method and conjuga- RecA from other Deinococcaceae and E. coli is shown in tion) while varying the parameters in each method, as well

Fig. 2B. The major differences between RecAC and RecAP as different DNA molecules (various available and newly are present at the N- and C-termini, and between residues constructed plasmids, purified from various strains,

240 and 255 (RecAC numbering). Compared with RecAP PCR fragments). Details and results of these assays and RecA from E. coli, a conserved extension of 11–13 are described in Supporting information. Although

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 129 Lesion bypass and two RecA in Deinococcus deserti 197

Fig. 2. Comparison of RecA protein sequences. A. Alignment of the two D. deserti RecA proteins (RecA-C and RecA-P). B. Multiple alignment of the RecA proteins from D. deserti (DdRecA-C, DdRecA-P), D. radiodurans (DrRecA), D. geothermalis (DgRecA) and E. coli (EcRecA). Point mutations of the residues underlined in E. coli RecA differentially affect cleavage of coprotease substrates in E. coli (see main text).

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 130 198 R. Dulermo, S. Fochesato, L. Blanchard and A. de Groot ᭿ transformants were obtained with each method, a classic and B. subtilis (Sciochetti et al., 2001). Moreover,

CaCl2 transformation method appeared to be most appro- whereas deletion of recA in D. radiodurans results in priate for D. deserti. Moreover, the D. radiodurans–E. coli extreme sensitivity to radiation (Bonacossa de Almeida shuttle vector pI3 could be introduced in D. deserti as a et al., 2002), the D. deserti DrecAC strain is as resistant to replicating plasmid. About 100–1000 transformants per UV and gamma irradiation as the wild type (Fig. 3). These microgram of pI3 plasmid were obtained by this technique results strongly suggest that the supplementary recA when pI3 was extracted from E. coli, against 10000 trans- genes on the plasmids code for a functional RecAP formants per microgram when pI3 was extracted from protein, allowing normal growth and radiation resistance

D. deserti. Gene deletion mutants of D. deserti could be of the DrecAC strain. The two other single mutants, obtained by allelic replacement via double homologous DrecAP1 and DrecAP3, also presented a normal growth and recombination after transformation with plasmid con- radiation resistance (data not shown). structs derived from E. coli vector pUC19, which does not In addition to the kanamycin resistance cassette replicate in D. deserti. In these plasmids, fragments cor- described above, a second antibiotic resistance cassette responding to DNA upstream and downstream of the was needed to construct D. deserti double mutants. For gene of interest are cloned in the correct orientation, this, it appeared to be possible to use the DrPtuf::cat respectively, upstream and downstream of a kanamycin cassette, containing the chloramphenicol resistance gene resistance cassette. Expression of the kanamycin resis- fused to the promoter of the D. radiodurans tuf gene tance gene in this cassette is driven by the constitutive (Earl et al., 2002b). The DrecAP1 DrecAP3 double mutant promoter of the tuf gene (Deide_18970, encoding elonga- showed growth (data not shown) and radiation resistance tion factor EF-Tu). Only 10–20 kanamycin-resistant trans- characteristics as the wild-type strain (Fig. 3), indicating formants were usually obtained, which was sufficient for that RecAC is also functional. Two other double mutants, the work described here, i.e. the construction and char- in which either recAP1 or recAP3 remains intact, were also acterization of D. deserti gene deletion mutants. Double constructed. Note that the DrecAC strain RD40 was used homologous recombination at the correct locus and com- as the parental strain for the construction of the DrecAC plete absence of the gene of interest was systematically DrecAP3 double mutant, showing that RecAC is not specifi- confirmed by diagnostic PCR. After D. radiodurans, cally required for gene replacement by homologous

D. deserti is the second Deinococcus species for which a recombination. The DrecAC DrecAP3 strain grew normally system for gene disruption has been developed. and was radiation resistant (data not shown). However,

the DrecAC DrecAP1 mutant showed a severely affected growth rate and formed long filaments (results not Each single recA mutant is as UV and gamma shown). Therefore, in this double mutant the remaining radiation-resistant as the wild type recAP3 did not permit normal growth. As recAP1 and recAP3 To analyse the role of each recA, three mutant strains code for identical proteins, these results indicate that were constructed in which one of the recA genes was recAP3 expression is weak compared with that of recAP1,at deleted. The chromosomal recA mutant, DrecAC, showed least under the experimental conditions used. This is in growth characteristics as the wild type (data not shown). agreement with our previously reported RT-PCR data that

For several other bacteria, various growth defects upon indicated that recAP3 transcription levels are lower than recA inactivation have been described, such as reduced that of recAP1 and recAC (de Groot et al., 2009). The growth rate and plating efficiency for D. radiodurans promoter activities of recAP1 and recAP3 may be different: (Bonacossa de Almeida et al., 2002) and filamentous the first 72 bp upstream of their start codons is nearly growth for Thermus thermophilus (Castan et al., 2003) identical (only one difference), but the sequences further

Fig. 3. Both RecAC and RecAP are functional and allow survival of D. deserti after exposure to radiation. (A) and (B) Survival curves after exposure to UV and gamma radiation respectively. Strains in (B) are indicated with the same symbols as in (A).

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 131 Lesion bypass and two RecA in Deinococcus deserti 199

Fig. 4. Induction of RecAC and RecAP after exposure of D. deserti to UV. Cell extracts of the indicated strains were prepared 60 min after exposure to 0 (-)or500Jm-2 (+)ofUV. Three different immunoblots are shown.

RecAC (37.8 kDa) and RecAP (36.6 kDa) were visualized using antiserum raised against E. coli RecA. Cell extract aliquots equivalent to 15 mg of protein were loaded in each lane, except for the boxed lanes where 25 mg was loaded. Levels of RecAP specifically encoded by recAP3 (arrows) were at the limit of

detection. Fifty nanograms of purified RecAC was loaded in the lanes at the right (the N-terminal His-tag used for purification was removed by TEV protease).

upstream are different. Radiation resistance was strongly type strain. Analysis of the various mutants showed that reduced in the DrecAC DrecAP1 strain (Fig. 3). However, the large majority of RecAP in the cells was encoded by survivors were still observed at the highest doses tested recAP1, in agreement with the observed weak transcription

(0.001% at 15 kGy g rays), in contrast to D. radiodurans of recAP3. Nevertheless, the immunoblot results indicate

DrecA mutants that are extremely sensitive to radiation that low levels of RecAP3 were induced, which probably (< 0.001% at 2 kGy) (Bonacossa de Almeida et al., 2002; contribute to DNA repair and the observed residual radia-

Tanaka et al., 2004), suggesting that weak expression of tion tolerance of the DrecAC DrecAP1 double mutant. The recAP3 allows repair of some DNA damage. A triple recA results further showed that irrE is required for induction of mutant may provide evidence for the contribution of both RecAC and RecAP (Fig. 4). recAP3 to DNA repair and survival. However, as a third antibiotic resistance cassette functional in D. deserti is TLS DNA polymerases ImuY and DnaE2 are involved currently unavailable, this triple recA mutant, if viable, in UV-induced mutagenesis in D. deserti could not be constructed. In D. radiodurans, radiation tolerance and DNA To investigate the role of the putative TLS DNA damage-induced expression of recA requires the pres- polymerase and photoproduct lyase genes in D. deserti, ence of irrE (Earl et al., 2002a; Hua et al., 2003), encod- four mutant strains with deletions of the corresponding ing a Deinococcus-specific regulator protein (de Groot genes were constructed, DpolB, DimuY, DdnaE2 and et al., 2009; Vujicic-Zagar et al., 2009). A D. deserti DirrE DDeide_3p02150. A double mutant, D(imuY-dnaE2), in mutant, which grows normally under standard conditions, which the DNA fragment encompassing both imuY and is extremely sensitive to radiation (Fig. 3). The DirrE strain dnaE2 was deleted, was also constructed. The mutants is less tolerant to radiation than the DrecAC DrecAP1 grew normally and showed the same survival rates as the double mutant, probably because IrrE regulates multiple wild-type strain after exposure to UV or gamma irradiation DNA repair and protection pathways (Hua et al., 2003; Lu (data not shown). et al., 2009). Moreover, the DrecAC DrecAP1 strain still Translesion synthesis polymerases often generate contains intact recAP3. mutations due to their low fidelity. To determine whether To get further insight in the expression levels of the D. deserti’s TLS polymerases are involved in error-prone three different recA genes, Western blotting experiments lesion bypass, rifampicin assays were performed after UV were performed with cell extracts of the different mutant irradiation. This test detects point mutations in the rpoB strains that were irradiated or not with UV. Due to the gene that confer a rifampicin-resistant phenotype (RifR). small difference in migration, both RecAC and RecAP First, we analysed whether induced mutagenesis could be could be detected, confirming that D. deserti produces demonstrated in the wild-type strain. This was indeed the two different RecA proteins (Fig. 4). Only RecAC was case, since an approximately 60-fold increase in the visible in un-irradiated cell extracts, whereas both RecA number of RifR mutants was observed after exposure to proteins were clearly induced after irradiation of the wild- UV (Fig. 5A). For the DpolB strain, the amount of RifR

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 132 200 R. Dulermo, S. Fochesato, L. Blanchard and A. de Groot ᭿

dnaE2) showed the same decrease of induction of RifR mutants as the DimuY and DdnaE2 single mutants, sug- gesting that ImuY and DnaE2 act in the same pathway. Complementation of the DimuY mutant by plasmid pRD52, carrying lexA-imuY, indicated that the decreased mutagenesis in this mutant was not due to a potential negative effect on dnaE2 expression (Fig. 5C). RT-PCR analysis indicated that dnaE2 expression in the DimuY mutant occurred by read-through transcription from the kanamycin resistance cassette (results not shown). If the putative photoproduct lyase Deide_3p02150 is involved in removal of DNA lesions, its inactivation might result in increased mutagenesis. However, the rifampicin assay revealed a similar amount of RifR mutants for the DDeide_3p02150 strain as for the wild type (data not shown). In conclusion, the results indicate that ImuY and DnaE2 are functional and involved in error-prone lesion bypass in D. deserti, whereas our experiments did not reveal a role for polB and Deide_3p02150.

Chromosome-encoded RecA, but not plasmid-encoded RecA, is required for induced-mutagenesis In E. coli and other bacteria, RecA is involved in induced mutagenesis. Following DNA damage, RecA stimulates autocleavage of repressor protein LexA, resulting in induced expression of genes encoding TLS polymerases and other DNA repair proteins. E. coli RecA also facilitates cleavage of UmuD to UmuD′, a component of TLS Pol V. It also participates directly in lesion bypass by interacting with Pol V (Fuchs and Fujii, 2007; Schlacher and Goodman, 2007). In D. deserti,alexA homologue is present immediately upstream of imuY and dnaE2, like in similar, RecA/LexA-regulated mutagenesis cassettes identified in several Proteobacteria (Abella et al., 2004; Erill et al., 2006). To see if one or several recA genes are involved in UV-induced mutagenesis in D. deserti, the rifampicin assay was performed with the various recA mutant strains. A strong decrease in number of RifR mutants was

observed in the mutants lacking recAC (Fig. 5B). This reduction was comparable with that found in the DimuY and DdnaE2 mutant strains. In contrast, high amounts of Fig. 5. D. deserti is mutable by UV and this requires imuY, dnaE2 R Rif mutants were still obtained with the DrecAP1, DrecAP3 and recAC. A. UV-induced mutagenesis in wild-type and TLS polymerase and DrecAP1 DrecAP3 strains. Introduction of plasmid mutant strains. pRD76 carrying recAC in the DrecAC mutant restored B. UV-induced mutagenesis in wild-type and RecA mutant strains. induced mutagenesis (Fig. 5C). Therefore, chromosome- C. Plasmids carrying imuY and recAC allow restoration of UV-induced mutagenesis in DimuY and DrecAC mutant strains encoded RecAC, but not plasmid-encoded RecAP,is respectively. required for UV-induced mutagenesis.

RecA is required for induction of imuY and dnaE2 mutants was similar as for the wild type (Fig. 5A). C expression However, an about 10-fold decrease in the number of R UV-induced Rif mutants was observed with the DimuY Our results show that recAC, imuY and dnaE2 are required and DdnaE2 strains (Fig. 5A). The double mutant D(imuY- for UV-induced mutagenesis in D. deserti. To investigate if

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 133 Lesion bypass and two RecA in Deinococcus deserti 201

Fig. 6. UV-induced expression of imuY and dnaE2 is dependent on recAC but not on

recAP and irrE. Expression of the eight genes indicated at the left was analysed by RT-PCR, using RNA isolated from the four strains indicated at the top. RNA was isolated 30 min after exposure to 0 (-) or 250 (+)Jm-2 UV.

recAC is involved in transcriptional regulation of imuY and lexA-imuY-dnaE2 in D. deserti is likely repressed by its dnaE2, RT-PCR experiments were performed. Whereas own encoded LexA, and derepressed by RecA-mediated the expression of imuY and dnaE2 is induced upon UV cleavage of LexA. Since RecAC, but not RecAP,is irradiation in the wild-type strain, induction of these genes required for induced expression of imuY and dnaE2, only was not visible in the DrecAC strain (Fig. 6). In contrast, RecAC might be able to stimulate cleavage of LexA. This

UV-induced expression of imuY and dnaE2 was still was tested in vitro, using RecAC, RecAP and LexA proteins present in the DrecAP1 DrecAP3 double mutant (Fig. 6). that were purified after expression in E. coli. Unlike RecAC

Therefore, the chromosome-encoded RecAC protein, but (and LexA), purified RecAP started to precipitate rapidly not plasmid-encoded RecAP, is required for induction of after purification, and this could not be prevented by the TLS polymerase genes imuY and dnaE2 that are immediate dialysis against another buffer. Experiments present on plasmid P1. These results are in agreement were therefore performed immediately after purification of with those obtained in the rifampicin assays. the proteins. Purified LexA, RecAC and RecAP migrated Transcription of the various genes characterized in the on SDS polyacrylamide gel with their expected molecular present study was also analysed in the D. deserti DirrE masses (Fig. 7). Similar to E. coli and D. radiodurans mutant (Fig. 6). Induction of each recA gene was depen- LexA (Narumi et al., 2001), D. deserti LexA undergoes dent on irrE (note that 35 PCR cycles were used for autodigestion when subjected to alkaline conditions amplification of the recAP3 fragment, and 30 for recAP1 and recAC). However, UV-induced transcription of imuY and dnaE2 was still observed in the DirrE strain, indicating that un-induced levels of RecAC are sufficient to facilitate induction of imuY and dnaE2 following UV damage. When UV-induced mutagenesis was analysed in the DirrE strain, a 6- to 10-fold increase in the number of RifR mutants was observed in comparison with the wild type (data not shown), most likely because error-free repair pathways are no longer induced in the absence of IrrE. Expression of the polymerase II (polB) and the putative photoproduct lyase (Deide_P302150) genes was also analysed in the various strains (Fig. 6). As in the wild type, exposure to UV did not result in induced transcription of polB in the DirrE and DrecA mutant strains (if anything, polB expression is rather slightly diminished after UV exposure). For the putative photoproduct lyase, the observed induction was dependent on irrE. Fig. 7. RecAC- and RecAP-mediated cleavage of LexA. Purified proteins and other components were present in the reaction Both RecAC and RecAP promote LexA cleavage in vitro mixtures as indicated at the top. LexA cleavage reactions were performed as described in Experimental procedures. The lane at As described for similar mutagenesis cassettes in other the left contains molecular weight marker proteins (masses, in kDa, bacteria, e.g. P. putida (Abella et al., 2004), expression of are indicated).

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 134 202 R. Dulermo, S. Fochesato, L. Blanchard and A. de Groot ᭿ leading to two breakdown products (Fig. S2). When Of the genes characterized in this work, the RDRM was testing RecA-mediated digestion at pH 7.4, LexA was found upstream of each recA (de Groot et al., 2009) cleaved by incubation with either RecAC or RecAP to (Fig. 1). Together with data on D. radiodurans, their IrrE- produce two breakdown products of the same size as dependent radiation-induced expression suggests a link those observed in the autodigestion (Fig. 7). The between IrrE and this motif. Indeed, IrrE-dependent observed cleavage was less efficient with RecAP than with induction of several other genes with the RDRM in their

RecAC. However, we cannot exclude that the concentra- upstream region, such as ddrB and ddrC, was also tion of available RecAP in the reaction mixture was observed in D. deserti (S. Fochesato and A. de Groot, decreased due to protein precipitation (see above). Both unpubl. data). IrrE possibly regulates the putative photo- for RecAC and for RecAP, this cleavage required the pres- product lyase gene in a different manner, because the ence of the oligonucleotide, ATPgS and MgCl2 (Fig. 7 and RDRM was not found upstream of Deide_3p02150.It Fig. S2). should be noted that direct binding of IrrE to the RDRM has not been demonstrated (Ohba et al., 2005; Vujicic- Discussion Zagar et al., 2009). The absence of TLS polymerase genes in the genome Bacterial species belonging to the Deinococcaceae are of D. radiodurans (and D. geothermalis) is in agreement famous for their extreme tolerance to ionizing and with previous studies that reported that D. radiodurans UV radiation and other stresses that provoke massive and other Deinococcus species are immutable by UV and DNA damage. Genome sequences are available for that they repair UV-induced DNA damage accurately D. radiodurans (White et al., 1999), D. geothermalis (Sweet and Moseley, 1974; Tempest and Moseley, 1982; (Makarova et al., 2007) and D. deserti (de Groot et al., Tanaka et al., 2005). Moreover, it was thought that error- 2009), species that were isolated from irradiated canned prone TLS might be imprudent in Deinococcus (Sale, meat, a hot spring and Sahara surface sand respectively. 2007). However, our results with D. deserti demonstrate A common radiation/desiccation response motif (RDRM) that an error-prone pathway for the repair of such damage in these species is present upstream of the same DNA is not necessarily absent from all Deinococcaceae. As repair genes, such as recA, pprA, ddrA, uvrA, uvrB most generated mutations will be deleterious, expression (Makarova et al., 2007; de Groot et al., 2009). A con- and/or activity of ImuY and DnaE2 should be tightly regu- served radiation response regulon is further supported by lated to allow the majority of the many lesions generated the observation that D. deserti irrE is able to restore radia- by intense radiation to be repaired by error-free tion resistance in a D. radiodurans DirrE mutant (Vujicic- mechanisms. Various bacterial proteins are involved in Zagar et al., 2009). D. deserti possesses several DNA accurate repair of UV-induced lesions (Goosen and repair genes that are absent from the two other Moolenaar, 2008), and D. deserti expresses several of sequenced Deinococcus species (de Groot et al., 2009), these, including UV-damage endonuclease (UVDE/UvsE) including a putative photoproduct lyase (Deide_3p02150) and UvrABC proteins (de Groot et al., 2009). and three TLS polymerase genes (polB, imuY, dnaE2). Deletion of polB or the photoproduct lyase-like gene

Moreover, D. deserti has three recA genes (recAC, recAP1, (Deide_3p02150) did not result in decrease or increase of recAP3), which code for two different RecA pro- induced mutagenesis. For polB, this might be because teins (RecAC, RecAP), whereas D. radiodurans and PolB activity is error-free, PolB is expressed under con- D. geothermalis have only one recA. The aims of this work ditions different from those applied here, or PolB gener- were (i) the development of genetic tools for D. deserti, (ii) ates frameshift mutations (Becherel and Fuchs, 2001), to investigate the functional expression of the three recA, which cannot be detected in the rifampicin assay that the three TLS polymerases and the putative photoproduct detects point mutations. Deide_3p02150 has only low lyase, and (iii) to see if the presence of the extra recA similarity with known spore photoproduct lyases, and genes is correlated with the other supplementary genes. there is no evidence that Deide_3p02150 is indeed a After construction and characterization of various single protein involved in DNA repair. However, its UV-induced and double mutant strains, we conclude that both RecAC and IrrE-dependent expression suggests a role in the and RecAP are functional proteins that allow repair of DNA damage response. Deide_3p02150 might be massive DNA damage after exposure of D. deserti to high involved in repair of specific lesion(s) formed during con- doses of gamma and UV radiation. ImuY and DnaE2 are comitant exposure to UV and desiccation. Although dele- involved in UV-induced point mutagenesis. Not the tion of the TLS polymerase genes and Deide_3p02150 supplementary RecAP but RecAC facilitates induction of did not affect tolerance to irradiation under the experimen- imuY and dnaE2 expression. IrrE is involved in induction tal conditions, i.e. in the presence of water and nutrients, of each recA and of the putative photoproduct lyase gene, their role in survival in harsh environmental conditions but it is not required for imuY and dnaE2 induction. such as the hot and arid desert may be more important.

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 135 Lesion bypass and two RecA in Deinococcus deserti 203

In addition, mutations generated by TLS will increase nucleoprotein filaments, or are they targeted to different genetic variability, which may lead to better adaptation of DNA substrates? What is the advantage to possess two D. deserti to its hostile habitat (Nohmi, 2006). different RecA proteins (three recA genes)? Under the

In this article we report, for the first time, the presence of experimental conditions, either RecAC or RecAP is suffi- two different RecA proteins with partly different functions in cient for extreme radiation resistance. However, as a single bacterial species. Only RecAC facilitates described for the TLS polymerases, the presence of mul- UV-induced mutagenesis, most likely by promoting cleav- tiple recA genes may contribute to survival under nutrient- age of LexA encoded by the lexA-imuY-dnaE2 cassette. poor, dry and UV-exposed conditions in the desert. In Moreover, as observed in the DirrE mutant, only basal such an environment, growth will be slow and frequently levels of RecAC are sufficient for induction of imuY and arrested while DNA damage will accumulate due to des- dnaE2. In contrast, induced levels of RecAP, although iccation and UV. The extra recAP genes may allow syn- functional for repair of massive DNA damage in the DrecAC thesis of higher levels of RecA for a rapid error-free repair mutant, do not allow induction of imuY and dnaE2, possibly of the accumulated DNA damage, without further inducing resulting from its inability to promote (sufficient) LexA error-prone lesion bypass. cleavage in vivo. Indeed, in vitro experiments showed that The RecA/LexA-regulated SOS system, a response

LexAcleavage is less efficient with RecAP than with RecAC. mechanism to address DNA damage in bacteria, has

Interestingly, one of the regions that are different in RecAC been studied most extensively in E. coli where it regulates and RecAP, i.e. residues 240–255 (RecAC numbering), about 40 genes. Studies in other bacteria have revealed corresponds to a region in E. coli RecA that has been substantial heterogeneity in LexA regulon contents. In implicated in coprotease substrate binding and cleavage D. radiodurans, induction of recA and other repair genes (Dutreix et al., 1989; Mustard and Little, 2000; McGrew is LexA-independent and requires IrrE. The presence in and Knight, 2003; VanLoock et al., 2003). Therefore, the D. deserti of two different RecA and both IrrE- and RecA- difference in this region of the RecA proteins might be an dependent regulation of DNA repair genes illustrates explanation for the inability of RecAP to induce TLS. Nev- further diversity of DNA damage response regulons ertheless, coprotease activity of RecAP is not abolished, among different bacterial species. and TLS induction may be controlled by another mecha- nism in addition to LexA binding and cleavage efficiency. Experimental procedures This is supported by observations that not all RecA-DNA filaments induce the SOS response in E. coli. It has been Bacterial strains, plasmids and growth conditions suggested that initially formed RecA nucleoprotein fila- The strains and plasmids used in this study are listed in ments are competent for recombination, but that SOS Table 1. D. deserti was grown at 30°C in 10-fold diluted tryptic induction requires a change in property of the filament or soy broth (TSB/10) supplemented with trace elements formation of extended filaments (Gruenig et al., 2008; (Vujicic-Zagar et al., 2009). Antibiotics were used at the Long et al., 2008). In D. deserti, only RecAC filaments following concentrations for D. deserti: streptomycin, might adopt this special conformation, whereas both 10 mgml-1; kanamycin, 10 mgml-1; chloramphenicol, 3 mgml-1; rifampicin, 10 mgml-1. E. coli was grown in Luria– RecAC and RecAP filaments are competent for recombina- tion. Like for E. coli RecA, not all RecA filaments may Bertani medium (LB), when necessary supplemented with C 50 mgml-1 kanamycin or 100 mgml-1 ampicillin. adopt the conformation necessary for LexA cleavage. Pre- vious studies have shown a competition between second- ary DNA and LexA for binding to a RecA nucleofilament Transformation of D. deserti (Harmon et al., 1996; Rehrauer et al., 1996). Proteins that An exponential culture of D. deserti was centrifuged (5000 g) modulate RecA function (Cox, 2007b) may play a role in and cells were re-suspended in 0.1 M CaCl2 (half of the initial determining which process is undertaken, recombination culture volume) and placed overnight at 4°C. Cells were or LexA cleavage (Gruenig et al., 2008; Long et al., 2008). centrifuged again and re-suspended in 1/15 of initial culture Several interesting questions remain. In addition to TLS volume of 0.1 M CaCl2 with 20% glycerol. The cells were induction, are there other functional differences between stored at -80°C until use. The mixture of cells and DNA was RecA and RecA ? They likely have largely overlapping placed on ice for at least 30 min, followed by a heat shock at C P 42°C for 2 min, and then again on ice for 1–2 min. Five functions, but RecA and RecA may be optimized for C P hundred microlitres of TSB/10 plus trace elements was added different activities in for example homologous recombina- and cells were incubated for 6 h at 30°C before plating. tion, repair of stalled replication forks or RecA-mediated excision repair (Ishimori et al., 1996; Namsaraev et al., DNA isolation 1998; Bichara et al., 2007; Cox, 2007a,b). Furthermore,

RecAC and RecAP may have yet unidentified different Deinococcus deserti chromosomal DNA was extracted from coprotease targets. Do RecAC and RecAP form mixed 10 ml of stationary-phase cells that were harvested by cen-

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 136 204 R. Dulermo, S. Fochesato, L. Blanchard and A. de Groot ᭿

Table 1. List of strains and plasmids used in this study.

Strain or plasmid Genotype or other relevant characteristics Source or reference

Strain Escherichia coli Top10 F′ mcrA D(mrr-hsdRMS-mcrBC) f80lacZDM15 Invitrogen DlacX74 deoR recA1 araD139 D(ara-leu)7697 galU galK rpsL (StrR) endA1 nupG BL21(DE3) F′ ompT hsd SB(rB-,mB-) gal dcm (DE3) Invitrogen BL21(AI) F′ ompT hsdSB(rB-,mB-) gal dcm Invitrogen araB::T7RNAP-tetA Deinococcus deserti VCD115 Wild type de Groot et al. (2005) RD19 As VCD115, but streptomycin-resistant Vujicic-Zagar et al. (2009) RD31 RD19 DimuYWkan This work RD38 RD19 DpolBWkan This work RD40 RD19 DrecACWkan This work RD41 RD19 D(imuY-dnaE2)Wkan This work RD42 RD19 DirrEWkan Vujicic-Zagar et al. (2009) RD44 RD19 DrecAP3Wcat This work RD45 RD19 DdnaE2Wkan This work RD47 RD19 DrecAP1Wkan This work RD4046 RD19 DrecACWkan DrecAP3Wcat This work RD4746 RD19 DrecAP1Wkan DrecAP3Wcat This work RD4753 RD19 DrecAP1Wkan DrecACWcat This work RD4076 RD40/pRD76 This work RD52 RD31/pRD52 This work RD61 RD19 DDeide_3p02150Wkan This work Plasmid pCR4Blunt-TOPO Cloning vector Invitrogen pUC19 Cloning vector Invitrogen pET-TEV pET28a derivative; Houben et al. (2007) for protein overproduction in E. coli pI3 Shuttle vector that replicates in E. coli, Masters and Minton (1992); D. radiodurans and D. deserti;ApR in E. coli, Meima and Lidstrom (2000) CmR in Deinococcus pGTC101 Source of DrPtuf::cat Earl et al. (2002b) pRD0 pUC19 with Ptuf::kan This work pRD36 pUC19 with DrPtuf::cat This work pRD52 pI3 with lexA-imuY This work pRD76 pI3 with recAC This work pET-TEV/LexA pET-TEV with lexA This work pET-TEV/RecAC pET-TEV with recAC This work pET-TEV/RecAP pET-TEV with recAP1 This work

trifugation (10 min, 5000 g) and re-suspended in 500 mlof PCR and plasmid construction 50 mM Tris-HCl, pH 8, 50 mM EDTA, 0.15 M NaCl. Thirty microlitres of lysozyme (10 mg ml-1) was added and the For PCR amplification of DNA fragments for cloning pur- mixture was incubated for 1 h at 37°C. Then 100 mlof10% poses, Pfx polymerase (Invitrogen) was used. For diagnostic SDS, 15 ml of proteinase K (20 mg ml-1) and 15 mlof PCR, HotGoldstar (Eurogentec) was used following the pronase E (20 mg ml-1) were added and the mixture was instructions of the manufacturer but with addition of DMSO incubated for 1 h at 37°C. One volume of phenol– (5% final concentration) in the PCR mix. Primers are listed in chloroform–isoamyl alcohol (25:24:1) was added and the Tables S1–S4. solution was vigorously mixed and centrifuged at For cloning, the blunt-end PCR products were purified with 13200 r.p.m. for 5 min in a table centrifuge (Eppendorf). the QIAquick gel extraction kit (Qiagen) and first inserted into After a second phenol–chloroform–isoamyl alcohol extrac- the pCR®4Blunt-TOPO® vector (Invitrogen). DNA fragments tion, the DNA was precipitated by addition of 1 vol. of iso- were recloned in other plasmids using appropriate restriction propanol and centrifugation at 13200 r.p.m. for 30 min. DNA enzymes. Derivatives of plasmids pRD0 and pRD36, which was washed with 70% ethanol, and centrifuged again at do not replicate in D. deserti, were used to construct 13200 r.p.m. DNA was re-suspended in 100–200 mlof D. deserti gene deletion mutants by double homologous water. DNA concentration was measured at OD260 (Eppen- recombination. Plasmid pRD0 contains a BamHI–PstI frag- dorf Biophotometer). The QIAprep spin miniprep Kit ment with the kanamycin resistance gene translationally (Qiagen) was used for plasmid extraction from E. coli. fused to the 370 bp promoter region of D. deserti gene tuf

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 137 Lesion bypass and two RecA in Deinococcus deserti 205

(Deide_18970; encoding elongation factor EF-Tu), and was ally diluted until 10-5 and 4 ml of dilutions in Petri dishes were constructed as described in Supporting information. Plasmid exposed to UV-C (254 nm) for the desired dose, and then pRD36 contains a BamHI–PstI fragment with a fusion of the 100 ml were spread on plates. Colonies were counted after D. radiodurans tuf promoter to the chloramphenicol resis- 3–4 days of incubation. For analysis of gene or protein tance gene. This DrPtuf::cat cassette was amplified from expression after UV irradiation, two samples from the same pGTC101 with primers DRtuf1Bam and CatPst (Table S1). culture in exponential phase (OD600 0.2–0.4) were taken. One Fragments of 1–1.5 kb corresponding to DNA upstream and sample was exposed to UV-C (4 ml portions in Petri dishes) downstream of genes to be deleted were amplified from and the other was not irradiated. Then both were further D. deserti chromosomal DNA (with primers listed in incubated, and samples for RNA isolation or total-cell extract Table S2), and cloned in the correct orientation, respectively, preparation were taken at the desired time point. For upstream and downstream of the cassettes in pRD0 mutagenesis experiments, the same cultures as those for (Ptuf::kan) or pRD36 (DrPtuf::cat). The resulting plasmids testing survival to UV were used. Four millilitres of non- were used to transform D. deserti. Diagnostic PCR to confirm diluted cultures were UV-irradiated for the desired dose and double homologous recombination at the correct locus and then incubated for 20 h at 30°C with shaking to permit fixation complete absence of the gene of interest was performed of mutations. In case of strains containing plasmids (comple- using primers listed in Table S3. mentation), irradiated cultures were first centrifuged and cells For pRD52, a 2.1 kb XbaI–HindIII DNA fragment carrying were re-suspended in fresh medium with chloramphenicol lexA-imuY (Deide_1p01870-Deide_1p01880) and 185 bp prior to incubation for 20 h. Then, 1 ml of culture was spread upstream of lexA with its probable promoter was cloned in on plates supplemented with rifampicin. Serial dilutions of pI3. Plasmid pRD76 contains a 385 bp XbaI–EcoRI fragment, these cultures were also spread on plates without antibiotic. corresponding to the probable promoter of the cinA-ligT-recA The number of rifampicin-resistant mutants was observed operon and 85 bp of the beginning of cinA, fused at the EcoRI after 3–4 days at 30°C. The frequency of mutation per 109 site to a 1.2 kb EcoRI–HindIII DNA fragment carrying the last colony-forming units (cfu) was calculated by (number of -1 -1 9 96 bp of ligT and entire recAC (Deide_19450). The pET-TEV rifampicin-resistant cfu ml /total number of cfu ml ) ¥ 10 . derivatives pET-TEV/LexA, pET-TEV/RecAC and pET-TEV/ RecAP contain NdeI–XhoI fragments with the genes Preparation of cell extracts for immunoblotting lexA (Deide_1p01870), recAC (Deide_19450) and recAP1

(Deide_1p01260) respectively. The cloned fragments in the Cells from 10 ml of culture (OD600 0.3–0.6) were collected by pI3 and pET-TEV derivatives were amplified with the primers centrifugation, washed with 1 ml of 10 mM Tris-HCl pH 8.0, listed in Table S1. Cloned DNA fragments were verified by and stored at -80°C until use. Cells were re-suspended in nucleotide sequencing (Genome Express, France). 0.5 ml of 50 mM Tris-HCl pH 8.0, 5 mM EDTA, supplemented with protease inhibitor cocktail as recommended by the RNA isolation and RT-PCR manufacturer (Sigma, P8849), and disrupted by sonication on ice (six times for 10 s with pauses of 1 min; microtip, Vibra For RNA isolation from D. deserti, cells were treated with Cell 72408, Bioblock Scientific). After centrifugation for 5 min RNAprotect Bacteria Reagent (Qiagen) and RNA was iso- at 13200 r.p.m. at 4°C (Eppendorf table centrifuge) to remove lated using the RNeasy Mini Kit (Qiagen). RNA samples were cell debris, the supernatant was kept as total-cell extract. treated twice with DNase. RT-PCR was performed in two Protein concentrations were measured using CooAssay steps. First, cDNA was synthesized in 20 ml of reactions using Protein Dosage Reagent UPF86420 (Uptima/Interchim). 1 mg of RNA and the Transcriptor First Strand cDNA Synthe- After addition of SDS sample buffer, extracts were heated for sis Kit (Roche). In the second step, DNA fragments of 150– 10 min at 95°C. 250 bp were amplified in 25 ml reactions using 1 ml of cDNA from the first step, Taq polymerase (Sigma) and gene-specific Protein separation and immunoblot analysis primers (Table S4). These amplifications were carried out by incubating reactions at 94°C for 10 min prior to 30 cycles of Fractions of cell extracts were subjected to electrophoresis 1 min at 94°C, 30 s at 56°C and 30 s at 72°C, followed by a through a 10% Bis-Tris NuPAGE gel (Invitrogen). The final step at 72°C for 2 min, with modifications for tuf (25 proteins were transferred to a nitrocellulose membrane cycles), Deide_3p02150 (32 cycles), recAP3 and dnaE2 (both (Whatman, 10401 196) using a semi-dry transfer system 35 cycles). Controls for DNA contamination were performed (Bio-Rad) and then treated with a 1:1000 dilution of poly- with reactions lacking reverse transcriptase. clonal antibodies raised against RecA protein from E. coli (generous gifts of S. Sommer and S. Marsin). The goat anti- Gamma and UV irradiation and induced mutagenesis rabbit peroxidase (Sigma, A6154) was used as a second antibody. The antigen–antibody complexes were visualized For gamma irradiation, cells were grown overnight to expo- with G:BOX (SynGene) using the SuperSignal West Pico nential phase (OD600 0.1–0.3), concentrated 10-fold in growth Chemiluminescent Substrate (Thermo Scientific). medium and exposed on ice to gamma rays for the desired -1 60 dose (39.6 Gy min , Co source, CEA-Cadarache, France). Expression and purification of LexA and RecA proteins Then, 20 ml of serial dilutions were spotted on solidified medium to determine survival. In plasmids pET-TEV/LexA, pET-TEV/RecAC and pET-TEV/

To determine survival after UV irradiation, cells were grown RecAP, the lexA, recAC and recAP genes are under the control to exponential phase (OD600 0.2–0.4). The cultures were seri- of the T7 promoter. Each protein produced contains an

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 138 206 R. Dulermo, S. Fochesato, L. Blanchard and A. de Groot ᭿

N-terminal His tag of 21 residues (MGSSHHHHHHSSGEN- We thank S. Cuiné and C. Brutesco for help with Western LYFQGH) including a TEV protease cleavage site blotting and detection, J. Vicente for access to gamma irra- (underlined). For expression of LexA, E. coli BL21(DE3) cells diation facilities, G. Brandelet for help with protein purification (Invitrogen) freshly transformed with pET-TEV/LexA were and D. Lemaire for mass spectrometry. This work was grown at 37°C overnight to saturation in 10 ml of LB medium supported by the Commissariat à l’Energie Atomique and containing 50 mgml-1 kanamycin. This pre-culture was then the Agence Nationale de la Recherche (ANR-07-BLAN- diluted in 1 l of LB medium, 50 mgml-1 kanamycin (in a 3 l 0106-02).

flask) and grown at 37°C, 160 r.p.m. At OD600 of 0.6–0.7, isopropyl-b-D-thiogalactopyranoside (IPTG) was added to a final concentration of 0.1 mM, and the cells were grown at References

17°C for 17 h. RecAC and RecAP were expressed similarly, except that E. coli BL21 (AI) cells (Invitrogen) were used, and Abella, M., Erill, I., Jara, M., Mazon, G., Campoy, S., and Barbe, J. (2004) Widespread distribution of a lexA- not only IPTG but also L-arabinose (0.2% final concentration) regulated DNA damage-inducible multiple gene cassette in was added to induce gene expression. The induced cells the Proteobacteria phylum. Mol Microbiol 54: 212–222. were harvested by centrifugation and cell pellets were frozen Becherel, O.J., and Fuchs, R.P. (2001) Mechanism of DNA at -20°C. polymerase II-mediated frameshift mutagenesis. Proc Natl For purification, cell pellets containing the recombinant pro- Acad Sci USA 98: 8566–8571. teins were re-suspended in 30 ml of sodium phosphate buffer Bichara, M., Pinet, I., Lambert, I.B., and Fuchs, R.P. (2007) (20 mM) pH 8, 500 mM NaCl, supplemented with 25 mlof RecA-mediated excision repair: a novel mechanism for protease inhibitor cocktail (Sigma, P8849) and 1 mg of repairing DNA lesions at sites of arrested DNA synthesis. DNase (Sigma, DN25). Cells were disrupted by French press Mol Microbiol 65: 218–229. (two cycles) and the soluble extracts were then recovered Blasius, M., Hubscher, U., and Sommer, S. (2008) Deinococ- after centrifugation at 10 000 g for 10 min at 4°C and ultra- cus radiodurans: what belongs to the survival kit? Crit Rev centrifugation at 150 000 g for 45 min at 4°C. The superna- Biochem Mol Biol 43: 221–238. tants were injected onto HisTrapTM HP columns (1 ml) (GE Bonacossa de Almeida, C., Coste, G., Sommer, S., and Healthcare), previously equilibrated in sodium phosphate Bailone, A. (2002) Quantification of RecA protein in Deino- buffer (20 mM) pH 8, 500 mM NaCl, containing 20 mM coccus radiodurans reveals involvement of RecA, but not imidazole. A step gradient of imidazole (20, 40, 100 and LexA, in its regulation. Mol Genet Genomics 268: 28–41. 500 mM) was used for elution (20 ml was used for each Butala, M., Zgur-Bertok, D., and Busby, S.J. (2009) The fraction except for the 500 mM fraction where 4 ml was used bacterial LexA transcriptional repressor. Cell Mol Life Sci for RecA and RecA and 2.5 ml for LexA). Eluted fractions C P 66: 82–93. were analysed by SDS-PAGE for the presence of each Castan, P., Casares, L., Barbe, J., and Berenguer, J. (2003) protein and verification of the protein purity. Protein concen- Temperature-dependent hypermutational phenotype in tration was determined by the CooAssay Protein Dosage recA mutants of Thermus thermophilus HB27. J Bacteriol Reagent UPF86420 (Uptima/Interchim). The mass of each 185: 4901–4907. purified protein was verified on a micrOTOF-Q mass spec- Cox, M.M. (2007a) Motoring along with the bacterial RecA trometer (Bruker, Wissembourg, France). protein. Nat Rev Mol Cell Biol 8: 127–138. Cox, M.M. (2007b) Regulation of bacterial RecA protein RecA-mediated LexA cleavage reaction function. Crit Rev Biochem Mol Biol 42: 41–63. Cox, M.M., and Battista, J.R. (2005) Deinococcus radiodu- The protocol was adapted from Satoh et al. (2006). The reac- rans – the consummate survivor. Nat Rev Microbiol 3: tion mixture contained 20 mM Tris-HCl pH 7.4, 10 mM MgCl2, 882–892. ′ ′ 10 mM oligonucleotide (5 -GCGTGTGTGGTGGTGTGC-3 ) Daly, M.J., Gaidamakova, E.K., Matrosova, V.Y., Vasilenko, (Giese et al., 2008), 1 mM ATPgS, 8 mM RecA (RecAC or A., Zhai, M., Leapman, R.D., et al. (2007) Protein oxidation RecAP) and 7 mM LexA. To activate RecAC and RecAP by implicated as the primary determinant of bacterial binding with ATPgS and oligonucleotide, the mixture without radioresistance. PLoS Biol 5: e92. LexA was first pre-incubated at 37°C for 20 min. The reaction Dutreix, M., Moreau, P.L., Bailone, A., Galibert, F., Battista, was then initiated by addition of LexA, and the reaction J.R., Walker, G.C., and Devoret, R. (1989) New recA muta- mixture was incubated at 37°C for 2 h. After quenching by tions that dissociate the various RecA protein activities in ® addition of NuPAGE LDS sample buffer (Invitrogen) and Escherichia coli provide evidence for an additional role for heating at 95°C for 10 min, the reaction products were sub- RecA protein in UV mutagenesis. J Bacteriol 171: 2415– jected to a 10% Bis-Tris NuPAGE gel (Invitrogen). LexA and 2423. breakdown products were visualized by staining with Imperi- Earl, A.M., Mohundro, M.M., Mian, I.S., and Battista, J.R. TM al protein stain (Pierce). The LexA autodigestion reaction in (2002a) The IrrE protein of Deinococcus radiodurans R1 is alkaline conditions was performed by incubating 7 mM LexA a novel regulator of recA expression. J Bacteriol 184: in 50 mM Tris-HCl pH 10.5 at 37°C for 20 h. 6216–6224. Earl, A.M., Rankin, S.K., Kim, K.P., Lamendola, O.N., and Acknowledgements Battista, J.R. (2002b) Genetic evidence that the uvsE gene product of Deinococcus radiodurans R1 is a UV damage We are grateful to S. Sommer and S. Marsin for generous endonuclease. J Bacteriol 184: 1003–1009. gifts of anti-RecA antiserum, and to P. Servant for plasmids. Erill, I., Campoy, S., Mazon, G., and Barbe, J. (2006) Dis-

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 139 Lesion bypass and two RecA in Deinococcus deserti 207

persal and regulation of an adaptive mutagenesis cassette Hua, Y., Narumi, I., Gao, G., Tian, B., Satoh, K., Kitayama, S., in the bacteria domain. Nucleic Acids Res 34: 66–77. and Shen, B. (2003) PprI: a general switch responsible for Erill, I., Campoy, S., and Barbe, J. (2007) Aeons of distress: extreme radioresistance of Deinococcus radiodurans. an evolutionary perspective on the bacterial SOS Biochem Biophys Res Commun 306: 354–360. response. FEMS Microbiol Rev 31: 637–656. Ishimori, K., Sommer, S., Bailone, A., Takahashi, M., Cox, Fuchs, R.P., and Fujii, S. (2007) Translesion synthesis in M.M., and Devoret, R. (1996) Characterization of a mutant Escherichia coli: lessons from the NarI mutation hot spot. RecA protein that facilitates homologous genetic recombi- DNA Repair (Amst) 6: 1032–1041. nation but not recombinational DNA repair: RecA423. J Mol Garvey, P., Rince, A., Hill, C., and Fitzgerald, G.F. (1997) Biol 264: 696–712. Identification of a RecA homolog (RecALP) on the conju- Long, J.E., Renzette, N., Centore, R.C., and Sandler, S.J. gative lactococcal phage resistance plasmid pNP40: (2008) Differential requirements of two recA mutants for evidence of a role for chromosomally encoded RecAL constitutive SOS expression in Escherichia coli K-12. PLoS in abortive infection. Appl Environ Microbiol 63: 1244– ONE 3: e4100. 1251. Lu, H., Gao, G., Xu, G., Fan, L., Yin, L., Shen, B., and Hua, Giese, K.C., Michalowski, C.B., and Little, J.W. (2008) RecA- Y. (2009) Deinococcus radiodurans PprI switches on DNA dependent cleavage of LexA dimers. J Mol Biol 377: 148– damage-response and cellular survival networks after 161. radiation damage. Mol Cell Proteomics 8: 481–494. Goosen, N., and Moolenaar, G.F. (2008) Repair of UV McGrew, D.A., and Knight, K.L. (2003) Molecular design and damage in bacteria. DNA Repair (Amst) 7: 353–379. functional organization of the RecA protein. Crit Rev de Groot, A., Chapon, V., Servant, P., Christen, R., Biochem Mol Biol 38: 385–432. Fischer-Le Saux, M., Sommer, S., and Heulin, T. (2005) Makarova, K.S., Aravind, L., Wolf, Y.I., Tatusov, R.L., Minton, Deinococcus deserti sp. nov., a gamma-radiation-tolerant K.W., Koonin, E.V., and Daly, M.J. (2001) Genome of bacterium isolated from the Sahara Desert. Int J Syst Evol the extremely radiation-resistant bacterium Deinococcus Microbiol 55: 2441–2446. radiodurans viewed from the perspective of comparative de Groot, A., Dulermo, R., Ortet, P., Blanchard, L., Guérin, P., genomics. Microbiol Mol Biol Rev 65: 44–79. Fernandez, B., et al. (2009) Alliance of proteomics and Makarova, K.S., Omelchenko, M.V., Gaidamakova, E.K., genomics to unravel the specificities of Sahara bacterium Matrosova, V.Y., Vasilenko, A., Zhai, M., et al. (2007) Deinococcus deserti. PLoS Genet 5: e1000434. Deinococcus geothermalis: the pool of extreme radiation Gruenig, M.C., Renzette, N., Long, E., Chitteni-Pattu, S., resistance genes shrinks. PLoS ONE 2: e955. Inman, R.B., Cox, M.M., and Sandler, S.J. (2008) RecA- Marchler-Bauer, A., Anderson, J.B., Derbyshire, M.K., mediated SOS induction requires an extended filament DeWeese-Scott, C., Gonzales, N.R., Gwadz, M., et al. conformation but no ATP hydrolysis. Mol Microbiol 69: (2007) CDD: a conserved domain database for interactive 1165–1179. domain family analysis. Nucleic Acids Res 35: D237– Gutman, P.D., Carroll, J.D., Masters, C.I., and Minton, K.W. D240. (1994) Sequencing, targeted mutagenesis and expression Masters, C.I., and Minton, K.W. (1992) Promoter probe and of a recA gene required for the extreme radioresistance of shuttle plasmids for Deinococcus radiodurans. Plasmid 28: Deinococcus radiodurans. Gene 141: 31–37. 258–261. Gutsche, I., Vujicic-Zagar, A., Siebert, X., Servant, P., Meima, R., and Lidstrom, M.E. (2000) Characterization of the Vannier, F., Castaing, B., et al. (2008) Complex oligomeric minimal replicon of a cryptic Deinococcus radiodurans structure of a truncated form of DdrA: a protein required SARK plasmid and development of versatile Escherichia for the extreme radiotolerance of Deinococcus. Biochim coli–D. radiodurans shuttle vectors. Appl Environ Microbiol Biophys Acta 1784: 1050–1058. 66: 3856–3867. Harmon, F.G., Rehrauer, W.M., and Kowalczykowski, S.C. Mustard, J.A., and Little, J.W. (2000) Analysis of Escherichia (1996) Interaction of Escherichia coli RecA protein with coli RecA interactions with LexA, lambda CI, and UmuD by LexA repressor. II. Inhibition of DNA strand exchange by site-directed mutagenesis of recA. J Bacteriol 182: 1659– the uncleavable LexA S119A repressor argues that recom- 1670. bination and SOS induction are competitive processes. Nahrstedt, H., Schroder, C., and Meinhardt, F. (2005) Evi- J Biol Chem 271: 23874–23883. dence for two recA genes mediating DNA repair in Bacillus Harris, D.R., Tanaka, M., Saveliev, S.V., Jolivet, E., Earl, megaterium. Microbiology 151: 775–787. A.M., Cox, M.M., and Battista, J.R. (2004) Preserving Namsaraev, E.A., Baitin, D., Bakhlanova, I.V., Alexseyev, genome integrity: the DdrA protein of Deinococcus radio- A.A., Ogawa, H., and Lanzov, V.A. (1998) Biochemical durans R1. PLoS Biol 2: e304. basis of hyper-recombinogenic activity of Pseudomonas Hazen, T.H., Wu, D., Eisen, J.A., and Sobecky, P.A. (2007) aeruginosa RecA protein in Escherichia coli cells. Mol Sequence characterization and comparative analysis of Microbiol 27: 727–738. three plasmids isolated from environmental Vibrio spp. Narumi, I., Satoh, K., Kikuchi, M., Funayama, T., Yana- Appl Environ Microbiol 73: 7703–7710. gisawa, T., Kobayashi, Y., et al. (2001) The LexA protein Houben, K., Marion, D., Tarbouriech, N., Ruigrok, R.W., and from Deinococcus radiodurans is not involved in RecA Blanchard, L. (2007) Interaction of the C-terminal domains induction following gamma irradiation. J Bacteriol 183: of sendai virus N and P proteins: comparison of 6951–6956. polymerase–nucleocapsid interactions within the paramyx- Narumi, I., Satoh, K., Cui, S., Funayama, T., Kitayama, S., ovirus family. J Virol 81: 6807–6816. and Watanabe, H. (2004) PprA: a novel protein from Deino-

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 140 208 R. Dulermo, S. Fochesato, L. Blanchard and A. de Groot ᭿

coccus radiodurans that stimulates DNA ligation. Mol Swingley, W.D., Chen, M., Cheung, P.C., Conrad, A.L., Microbiol 54: 278–285. Dejesa, L.C., Hao, J., et al. (2008) Niche adaptation and Nohmi, T. (2006) Environmental stress and lesion-bypass genome expansion in the chlorophyll d-producing cyano- DNA polymerases. Annu Rev Microbiol 60: 231–253. bacterium Acaryochloris marina. Proc Natl Acad Sci USA Norioka, N., Hsu, M.Y., Inouye, S., and Inouye, M. (1995) 105: 2005–2010. Two recA genes in Myxococcus xanthus. J Bacteriol 177: Tanaka, M., Earl, A.M., Howell, H.A., Park, M.J., Eisen, J.A., 4179–4182. Peterson, S.N., and Battista, J.R. (2004) Analysis of Deino- Ohba, H., Satoh, K., Yanagisawa, T., and Narumi, I. (2005) coccus radiodurans’s transcriptional response to ionizing The radiation responsive promoter of the Deinococcus radiation and desiccation reveals novel proteins that con- radiodurans pprA gene. Gene 363: 133–141. tribute to extreme radioresistance. Genetics 168: 21–33. Rajan, R., and Bell, C.E. (2004) Crystal structure of RecA Tanaka, M., Narumi, I., Funayama, T., Kikuchi, M., from Deinococcus radiodurans: insights into the structural Watanabe, H., Matsunaga, T., et al. (2005) Characteriza- basis of extreme radioresistance. J Mol Biol 344: 951–963. tion of pathways dependent on the uvsE, uvrA1,oruvrA2 Rehrauer, W.M., Lavery, P.E., Palmer, E.L., Singh, R.N., and gene product for UV resistance in Deinococcus Kowalczykowski, S.C. (1996) Interaction of Escherichia radiodurans. J Bacteriol 187: 3693–3697. coli RecA protein with LexA repressor. I. LexA repressor Tempest, P.R., and Moseley, B.E. (1982) Lack of ultraviolet cleavage is competitive with binding of a secondary DNA mutagenesis in radiation-resistant bacteria. Mutat Res molecule. J Biol Chem 271: 23865–23873. 104: 275–280. Rocha, E.P., Cornet, E., and Michel, B. (2005) Comparative VanLoock, M.S., Yu, X., Yang, S., Galkin, V.E., Huang, H., and evolutionary analysis of the bacterial homologous Rajan, S.S., et al. (2003) Complexes of RecA with LexA recombination systems. PLoS Genet 1: e15. and RecX differentiate between active and inactive RecA Sale, J.E. (2007) Radiation resistance: resurrection by nucleoprotein filaments. J Mol Biol 333: 345–354. recombination. Curr Biol 17: R12–R14. Vujicic-Zagar, A., Dulermo, R., Le Gorrec, M., Vannier, F., Satoh, K., Ohba, H., Sghaier, H., and Narumi, I. (2006) Servant, P., Sommer, S., et al. (2009) Crystal structure of Down-regulation of radioresistance by LexA2 in Deinococ- the IrrE protein, a central regulator of DNA damage repair cus radiodurans. Microbiology 152: 3217–3226. in Deinococcaceae. J Mol Biol 386: 704–716. Schlacher, K., and Goodman, M.F. (2007) Lessons from 50 White, O., Eisen, J.A., Heidelberg, J.F., Hickey, E.K., Peter- years of SOS DNA-damage-induced mutagenesis. Nat son, J.D., Dodson, R.J., et al. (1999) Genome sequence of Rev Mol Cell Biol 8: 587–594. the radioresistant bacterium Deinococcus radiodurans R1. Sciochetti, S.A., Blakely, G.W., and Piggot, P.J. (2001) Science 286: 1571–1577. Growth phase variation in cell and nucleoid morphology in Zimmerman, J.M., and Battista, J.R. (2005) A ring-like nucle- a Bacillus subtilis recA mutant. J Bacteriol 183: 2963– oid is not necessary for radioresistance in the 2968. Deinococcaceae. BMC Microbiol 5: 17. Sheng, D., Zheng, Z., Tian, B., Shen, B., and Hua, Y. (2004) LexA analog (dra0074) is a regulatory protein that is irrel- evant to recA induction. J Biochem 136: 787–793. Supporting information Sutton, M.D., Smith, B.T., Godoy, V.G., and Walker, G.C. (2000) The SOS response: recent insights into umuDC- Additional supporting information may be found in the online dependent mutagenesis and DNA damage tolerance. Annu version of this article. Rev Genet 34: 479–497. Please note: Wiley-Blackwell are not responsible for the Sweet, D.M., and Moseley, B.E. (1974) Accurate repair of content or functionality of any supporting materials supplied ultraviolet-induced damage in Micrococcus radiodurans. by the authors. Any queries (other than missing material) Mutat Res 23: 311–318. should be directed to the corresponding author for the article.

© 2009 The Authors Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 194–208 141 142

Fig. S1. RT-PCR demonstrating that lexA, imuY and dnaE2 genes are part of a single operon. RNA was isolated from wild-type D. deserti 30 min after exposure to 0 (-) or 250 (+) J.m-2 UV. RT-PCR was performed as described in the main text. Small arrows indicate the primers used for amplification (35 cycles): 1, DdP101870FWop; 2, DdP101880RVop; 3, DdP101880FWop; 4, DdP101900RVop (for primer sequences, see Table S4). HP, hypothetical protein.

Fig. S2. LexA autodigestion and RecAP-mediated LexA cleavage. The lanes at the left in panels A and B contain molecular weight marker proteins (masses, in kD, are indicated). A. Lane 1, purified LexA. Lane 2, LexA after autodigestion under alkaline conditions.

B. RecAP-mediated LexA cleavage. Purified proteins and other components were present in the reaction mixtures as indicated at the top. LexA cleavage reactions were performed as described in Experimental procedures.

143

144

VI Les domaines fonctionnels de la protéine régulatrice IrrE

145 146 VI) Les domaines fonctionnels de la protéine régulatrice IrrE

1) Introduction

IrrE, une protéine spécifique aux Deinococcus, est un régulateur essentiel à la radiotolérance chez D. radiodurans (Mattimore et al., 1995 ; Earl et al., 2002a) et D. deserti (voir chapitre V). Elle est nécessaire à l’induction de recA, pprA et à celle d’au moins 31 protéines de D. radiodurans dont plusieurs sont impliquées dans la radiotolérance (Earl et al., 2002a ; Hua et al., 2003 ; Lu et al., 2009). IrrE est une protéine exprimée constitutivement (Gao et al., 2006). Une analyse bioinformatique prédit qu’IrrE possède un domaine HTH et un domaine Zinc peptidase putatif (Earl et al., 2002a ; Hua et al., 2003). L’homologue d’irrE de D. deserti se situe, comme chez D. radiodurans, en amont des gènes codant les protéines FolP, FolB et FolK (Fig. 47). Chez D. radiodurans, irrE n’est pas en opéron avec les gènes folP, folB et folK car le promoteur de ces gènes se situe dans la séquence codante d’irrE (Gao et al., 2006). Plusieurs questions se posent quant au fonctionnement d’IrrE. Quel signal permet d’activer IrrE après irradiation ? Comment agit IrrE ?

Deide03013 Deide03030 Deide03050 Deide03060 16S Ribosomal IrrE Dihydroneopterin CHP RNA aldolase (FolB)

Deide03010 Deide03020 Deide03040 Deide03051 Mandelate Acetyl-CoA- Dihydropteroate HPKK racemase binding protein synthase (FolK) (FolP) 1000 bp

Fig 47 : Contexte génétique d’irrE de D. deserti. HP = hypothetical protein, CHP = conserved hypothetical protein, HPKK = 2-amino-4-hydroxy-6-hydroxymethyldihydropteridine diphosphokinase.

2) Résumé de l’article

Pour mieux comprendre le fonctionnement d’IrrE, la structure tridimensionnelle de la protéine IrrE de D. deserti a été résolue par le laboratoire de L. Serre à l’IBS Grenoble (voir l’article « Crystal structure of the IrrE protein, a central regulator of DNA damage repair in deinococcaceae » dans la partie 4 de ce chapitre). En plus, des études de génétique (groupe de Suzanne Sommer à Orsay pour D. radiodurans et du LEMIRE pour D. deserti) ont été associées pour compléter l’étude. Ma contribution dans cet article concerne l’étude de la radiotolérance conférée par différents allèles d’irrE chez D. deserti. Nous avons montré

147 qu’irrE de D. deserti est capable de complémenter la radiosensibilité d’un mutant irrE de D. radiodurans. Le mécanisme de régulation par IrrE est par conséquent conservé entre ces 2 espèces. La structure 3D d’IrrE confirme la présence d’un domaine Zinc peptidase-like (8- 125, en bleu) et d’un domaine HTH (helix-turn-helix, en rouge) similaire aux domaines liant l’ADN (126-167), et montre également la présence d’un domaine « senseur-like » de type GAF (168-263, en vert) (Fig. 48). Les domaines GAF sont des domaines capables de fixer de petites molécules (ex : citrate, cGMP), retrouvés dans des protéines de transduction de signaux. De plus, des expériences de DLS (dynamic light scattering, qui permet d’estimer la taille d’une protéine en solution) et la cristallographie suggère qu’IrrE puisse former un dimère. Cependant, l’analyse par gel filtration du profil d’élution d’IrrE ne permet pas d’affirmer ou d’infirmer la formation de dimère. Pour tester la capacité putative d’IrrE à fixer l’ADN, une modélisation de l’interaction d’IrrE, par son domaine HTH, avec de l’ADN double brin a été réalisée. Le résultat montre que de gros changements dans la structure d’IrrE

Figure 48 : Structure 3D d’IrrE de D. deserti. En bleu, le domaine Zn peptidase-like ; en rouge, le domaine HTH ; et en vert, le domaine « senseur » GAF.

seraient nécessaires pour qu’une telle interaction puisse avoir lieu. Ce résultat est en accord avec les expériences in vitro de fixation d’IrrE sur le promoteur de pprA où aucune interaction n’avait pu être mis en évidence (Ohba et al., 2005). L’analyse de la structure de chacun des domaines d’IrrE suggère que certains acides aminés sont particulièrement cruciaux pour le fonctionnement de cette protéine. Ainsi, des mutations ponctuelles ont été introduites dans les différents domaines (E83Q, H86S, Y160A, C175A, H217L). De plus, afin de vérifier si l’éventuelle dimérisation d’IrrE est importante pour son fonctionnement, la double mutation H35A-F48A, a été introduite dans la protéine. Cette dernière devrait empêcher la dimérisation d’IrrE. Les mutations E83Q et H86S devraient inactiver respectivement le site actif et la fixation du zinc au niveau du domaine Zinc peptidase-like. Y160, un résidu aromatique situé dans le domaine HTH, pourrait avoir un rôle dans l’interaction avec l’ADN. C175 et H217 sont les seuls résidus polaires de la cavité hydrophobe du domaine « senseur-like » GAF et pourraient avoir un rôle dans la fixation du ligand. La complémentation du mutant irrE de D. deserti par les diverses

148 constructions a donc été testée (Fig. 49). La double mutation H35A-F48A n’empêche pas la protéine IrrE de complémenter le mutant irrE, suggérant qu’IrrE n’agit pas sous forme de dimère. La souche avec les mutations affectants le site actif putatif du domaine Zinc peptidase-like (E83Q et H86S) est radiosensible comme le mutant délété d’irrE, ce qui montre que ces résidus et ce domaine sont essentiels au fonctionnement d’IrrE. Lorsque IrrE comporte la mutation C175A, elle reste capable de complémenter la sensibilité du mutant ΔirrE, alors que lorsqu’elle présente la mutation H217L, elle ne permet qu’une complémentation partielle. Ainsi, le domaine GAF est important dans le fonctionnement d’IrrE et le résidu H217 joue un rôle prépondérant dans ce domaine. La protéine mutée dans le domaine HTH (Y160A) restaure presque complétement la radiotolérance du mutant ΔirrE, ce qui suggère que le résidu Y160 joue un rôle mineur dans le fonctionnement d’IrrE.

Dose (J/m2) Dose (kGy) Figure 49 : Survie aux UV et aux rayons gamma de D. deserti et de divers mutants d’irrE. Les souches RD19 pI3, RD42 pI3 et RD51 sont respectivement la souche sauvage comportant le vecteur de clonage pI3, la souche ΔirrE portant pI3 et la souche ΔirrE complémenté avec le gène sauvage irrE cloné dans pI3. Les autres souches (H35A F48A ; Y160A ; H217L ; H86S ; E83Q et C175A) correspondent au mutant ΔirrE complémenté par les divers allèles d’irrE.

3) Conclusion

La structure d’IrrE a permis de mettre en évidence l’existance de 3 domaines, un domaine Zinc peptidase-like, un domaine HTH, et un domaine « senseur-like » de type GAF. Cette analyse a également montré qu’IrrE ne pouvait pas se fixer à l’ADN double brin. Les résidus E83 et H86, appartenant au domaine Zinc peptidase-like sont essentiels à l’activité d’IrrE, suggérant fortement qu’IrrE peut avoir une activité protéase. Plusieurs questions restent en suspens, et d’autres travaux seront nécessaires pour comprendre la fonction précise d’IrrE. Il est possible qu’un intermédiaire existe entre IrrE et les gènes radio-induits, mais quel est-il ? Le substrat du domaine Zinc peptidase-like, peut-être cet intermédiaire, n’est toujours pas connu. DdrO pourrait être un candidat intéressant puisque c’est un facteur de

149 transcription unique aux Deinococcus et que le gène ddrO présente un « radiation desiccation response motif » et est radio-induit après irradiation chez D. radiodurans et D. deserti (Voir Tableaux 3 et 11). Le domaine GAF est également très important mais la molécule, si elle existe, intéragissant avec ce domaine reste inconnue.

150 4) Article 3 : Crystal structure of the IrrE protein, a central regulator of DNA damage repair in deinococcaceae

Auteurs : Vujicić-Zagar A, Dulermo R, Le Gorrec M, Vannier F, Servant P, Sommer S, de Groot A, Serre L.

Journal : Journal of molecular biology

Année : 2009

Volume : 386

Pages : 704-716

151 152 doi:10.1016/j.jmb.2008.12.062 J. Mol. Biol. (2009) 386, 704–716

Available online at www.sciencedirect.com

Crystal Structure of the IrrE Protein, a Central Regulator of DNA Damage Repair in Deinococcaceae

Andreja Vujičić-Žagar1, Remi Dulermo2, Madalen Le Gorrec1, Françoise Vannier3, Pascale Servant3, Suzanne Sommer3, Arjan de Groot2 and Laurence Serre1⁎

1Laboratoire des Protéines Deinococcaceae are famous for their extreme radioresistance. Transcriptome Membranaires, Institut de analysis in Deinococcus radiodurans revealed a group of genes up-regulated Biologie Structurale UMR5075 in response to desiccation and ionizing radiation. IrrE, a novel protein (CEA/CNRS/Université Joseph initially found in D. radiodurans, was shown to be a positive regulator of Fourier), 41 rue Jules Horowitz, some of these genes. Deinococcus deserti irrE is able to restore radioresistance 38027 Grenoble Cedex 01, France in a D. radiodurans ΔirrE mutant. The D. deserti IrrE crystal structure reveals a unique combination of three domains: one zinc peptidase-like domain, 2Laboratoire d'Ecologie one helix-turn-helix motif and one GAF-like domain. Mutant analysis Microbienne de la Rhizosphère indicates that the first and third domains are critical regions for radio- et d'Environnements Extrêmes, tolerance. In particular, mutants affected in the putative zinc-binding site CEA/DSV/IBEB/SBVME, are as sensitive to gamma and UV irradiation as the ΔirrE bacteria, and UMR 6191, F-13108 radioresistance is strongly decreased with the H217L mutation present in Saint-Paul-lez-Durance, France the C-terminal domain. In addition, modeling of IrrE–DNA interaction 3Université Paris Sud 11, CNRS suggests that the observed IrrE structure may not bind double-stranded UMR 8621, CEA LRC42V, DNA through its central helix-turn-helix motif and that IrrE is not a classic Institut de Génétique et transcriptional factor that activates gene expression by its direct binding to Microbiologie, Bâtiment 409, DNA. We propose that the putative protease activity of IrrE could be a key 91405 Orsay Cedex, France element of transcription enhancement and that a more classic transcription factor, possibly an IrrE substrate, would link IrrE to transcription of genes Received 1 November 2008; specifically involved in radioresistance. received in revised form 16 December 2008; © 2009 Elsevier Ltd. All rights reserved. accepted 22 December 2008 Available online 3 January 2009 Edited by I. Wilson Keywords: IrrE; Deinococcus; gene regulation; radiotolerance; zinc

Introduction ionizing radiation or long periods of desiccation. The exceptional radioresistance of these bacteria is To preserve the integrity of their genome, living linked to their ability to reconstruct a functional cells have developed strategies to protect, control genome from hundreds of radiation-induced chro- and repair their genome. The non-spore-forming mosomal fragments. Active and passive mechan- bacterium Deinococcus radiodurans and other mem- isms, such as DNA repair, nucleoid organization bers of the Deinococcaceae family are famous for and protein protection against oxidation, are prob- ably intimately combined to enable survival from their extraordinary tolerance toward high doses of – various stresses.1 4 Analysis of D. radiodurans transcriptome revealed a *Corresponding author. Partnership for Structural set of genes that are up-regulated in response to Biology, Carl-Ivàr Bränden Building, BP181, 6 rue Jules desiccation or ionizing radiation, including several Horowitz, 38042 Grenoble, France. E-mail address: hypothetical genes of unknown function.5 For some [email protected]. of these original genes (i.e., ddrA, ddrB, ddrC, ddrD Abbreviations used: HTH, helix-turn-helix; Dcp, and pprA), their contribution to the extreme radio- dipeptidyl carboxypeptidase; PDE, ; tolerance of D. radiodurans had been clearly – DLS, dynamic light scattering; RDRM, radiation/ established.5 7 IrrE, a novel regulatory protein also desiccation response motif. called PprI, was shown to be a positive effector that

0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.153 Crystal Structure of Deinococcus deserti IrrE 705 enhances in an unelucidated manner the expression key element of its transcriptional enhancement and – of recA and pprA following exposure to radiation.7 9 that a more classic transcription factor, possibly an The irrE gene is constitutively expressed,10 and its IrrE substrate, links IrrE to gene transcription. inactivation results in a dramatic decrease in resis- tance to ionizing radiation, UV light and mitomycin Results C, suggesting that IrrE could be a key regulator of the DNA damage response in D. radiodurans,regulating multiple genome protection/repair pathways and a D. deserti irrE restores radiation resistance in a putative signal transduction pathway relative to D. radiodurans irrE mutant genome damage.8,9 To gain more insights on this central DNA repair We have amplified from D. deserti genomic DNA a regulator, we solved the crystal structure of IrrE 1630-bp fragment containing the 846-bp IrrE coding from Deinococcus deserti, a new species that is as sequence, 270 bp upstream of the initiation GTG resistant as D. radiodurans to ionizing and UV codon and 514 bp downstream of the TGA stop radiation.11 IrrE is a well-conserved Deinococcus- codon. Sequence alignment of D. deserti, D. radio- specific protein. The crystal structure shows that durans and Deinococcus geothermalis IrrE proteins IrrE is composed of three structural domains, shows that this protein is highly conserved in including a zinc peptidase-like domain and a Deinococci (Fig. 1). The functionality of D. deserti putative DNA-binding domain as suggested by IrrE was tested in a D. radiodurans host devoid of its previous predictions8,9 and a GAF-type domain own IrrE. Whereas cells deficient in IrrE were not predicted by bioinformatics. These crystallo- radiosensitive, the expression of D. deserti IrrE in graphic data, combined with a site-directed muta- D. radiodurans ΔirrE bacteria was sufficient to fully genesis study, reveal several protein regions critical restore radioresistance (Fig. 2a). As D. deserti IrrE for IrrE function and radiotolerance. This analysis shares 64% and 73% sequence identity with its D. suggests that IrrE has a putative proteolytic activity radiodurans and D. geothermalis homologs, respec- essential for the radiotolerance phenotype. We tively, these results suggest a high conservation of postulate that this putative enzymatic activity is a the regulation mechanisms involving IrrE, targeted

Fig. 1. Sequence alignment of IrrE proteins from D. deserti (DeiDe), D. geothermalis (DeiGe) and D. radiodurans (DeiRa). Identical residues are highlighted in black boxes. The mutated residues studied in this work are marked by white triangles. The secondary structures correspond to DSSP definition.12 This alignment was performed using the program ESPript.13

154 706 Crystal Structure of Deinococcus deserti IrrE

Fig. 2. Radiotolerance phenotype study. (a) D. deserti IrrE can complement D. radiodurans ΔirrE for survival after gamma irradiation. Bacterial strains were exposed to gamma irradiation and cell survival was measured as described in Materials and Methods. (b) D. deserti survival curves after exposure to UV and gamma radiation. The mutants are described in the text.

DNA sequences and/or factors interacting with IrrE superpose with a corresponding r.m.s.d. of 0.6 Å for in these processes. 242 aligned Cα atoms, showing that no major overall structural rearrangement occurs upon zinc binding. Overall crystal structure No electron density was present in these models for the 6His tag, the first 7 and the last 17 residues as The D. deserti IrrE crystal structure was deter- well as the loop constituted by residues 180 to 188. mined at 2.6-Å resolution by the single anomalous The IrrE crystal structure reveals that the protein is dispersion method using crystals of selenomethio- folded into three distinct domains: an N-terminal nine-substituted protein. In addition, one crystal domain (residues 8–125) that consists of five α- structure of zinc-bound IrrE was refined to 3.5-Å helices and a three-stranded antiparallel β-sheet; a resolution. The apo and zinc-bound IrrE models middle helix-turn-helix (HTH)-containing domain

Fig. 3. Overall crystal structure of D. deserti IrrE. (a) Stereoview of the overall structure: N-terminal, HTH and C- terminal domains are shown in blue, salmon and green, respectively. The disordered loop is schematized by a dotted line. Residues composing the metal-binding site are in ball-and-stick representation. (b) IrrE surface colored by electrostatic surface potential (calculated by APBS15). Figures 3–8 were generated with PyMOL.14

155 Crystal Structure of Deinococcus deserti IrrE 707

(residues 126–167); and a C-terminal domain (resi- family classified in the MEROPS database.18 Metal- dues 168–263) folded into a highly twisted six- lopeptidases of the MA clan bind a single zinc ion stranded antiparallel β-sheet with a short α-helix essential for their catalysis. Thermolysin, the best (Fig. 3a). Additional information regarding a puta- characterized member of this clan, is a 30- to 40-kDa tive role of each of the three domains was obtained monomeric protein consisting of two domains: one by comparing them separately to structurally similar N-terminal α/β domain and one C-terminal α-helical protein domains present in the Research Collabora- domain.19 The zinc-binding active-site cleft of ther- tory for Structural Bioinformatics (RCSB) Protein molysin is located between the two domains. A Data Bank (PDB). Surface charge distribution indi- similar structural pattern is observed for the IrrE N- cates that the surface of D. deserti IrrE is mainly acidic terminal domain (residues 8–125). Superposition of (APBS program15), apart from a region present in the the IrrE N-terminal domain and the zinc-binding center of the protein that corresponds to a large region of E. coli Dcp (residues 362–510) or Bacillus positive patch (Fig. 3b). thermolysin complexed with dipeptides highlights a putative substrate-binding cleft in the IrrE model The N-terminal domain of IrrE exhibits a (Fig. 4b). The N-terminal domain of IrrE has a wide- mono-zinc metallopeptidase fold open and long active-site cleft (N20 Å) that bears negative charges (Fig. 3b). The IrrE metallopeptidase- According to DALI server analysis,16 the N- like N-terminal domain (Fig. 4a) looks like a terminal domain of IrrE shows a high level of “minimal thermolysin” structure (r.m.s.d. of 2.0 Å structural similarity to the zinc-binding region of on 67 Cα atoms and 12% sequence identity). Escherichia coli dipeptidyl carboxypeptidase (Dcp).17 The metallopeptidase MA clan is characterized by This 78-kDa cytosolic enzyme belongs to the metal- one zinc ion coordinated by two histidine residues lopeptidase MA clan, or, more precisely, to the M3A present in the conserved HEXXH sequence and a

Fig. 4. N-terminal domain of D. deserti IrrE. (a) N-terminal domain (residues 8–125; blue ribbons) and Bacillus thermolysin (yellow ribbons; RCSB PDB code 8TLN) are superposed. (b) Shown are the cleft detected at the IrrE accessibility surface (white) in ball-and-stick representation, the peptides found in the E. coli Dcp complex (black; RCSB PDB code 1Y79) and those in the Bacillus thermolysin complex (white; RCSB PDB code 8TLN).

156 708 Crystal Structure of Deinococcus deserti IrrE conserved glutamate residue generally located 20–30 mentioned above and involved in zinc binding and/ residues downstream of this signature. In addition, or catalysis (H82, E83, H86 and E113) are structu- the conserved glutamate present inside the HEXXH rally conserved in both apo form and zinc-bound pentapeptide is believed to orient and polarize a form. In the apo form, however, the side chains of water molecule that acts as a nucleophile in the H86 and E113 are oriented away from the zinc- reaction mechanism proposed for this family of binding site, and they move in upon zinc binding enzymes.19 All these structural features are present (Fig. 5a and b). in the D. deserti N-terminal domain of IrrE (Fig. 5a), In the IrrE-1Zn crystal structure, the predicted and they superpose onto the zinc-binding site of the zinc-binding site shows a single positive peak in the − above mentioned metalloenzymes (Fig. 5c). difference (Fo Fc) electron density map. The zinc- bound structure was refined with one zinc ion The N-terminal domain of IrrE can bind one coordinated by the Nɛ2 atoms of H82 and H86 zinc ion present inside the HEXXH sequence as well as the carboxylate group of the conserved glutamate In the Se-Met-IrrE structure, no zinc ion was residue E113 located 30 residues downstream of observed in the N-terminal domain. To investigate theHEXXHglutamateinaccordancewiththe the putative zinc-binding ability of IrrE, we soaked scheme of the MA clan metallopeptidases. The the apo-protein crystals in the mother liquor modeled zinc ion and the neighboring protein solution supplemented with 10 mM ZnSO4 for atoms have similar B-factor values, and after 2 min. As a result, one zinc ion was found in the refinement, no residual peak is observed in the α − predicted zinc-binding site (Fig. 5b and d). The C difference (Fo Fc) electron density map in the zinc- atom positions of the four amino acid residues binding region. The coordination of the zinc ion in

Fig. 5. Metal-binding site of D. deserti IrrE: (a) apo form, (b) zinc-bound form and (c) superposition of the metal-binding − site in IrrE (blue/purple) and Bacillus thermolysin (yellow; RCSB PDB code 8TLN). (d) (2mFo DFc) electron density map calculated around the putative zinc-binding site in the IrrE-1Zn model at 3.5-Å resolution (map contoured at 1σ).

157 Crystal Structure of Deinococcus deserti IrrE 709 metallopeptidases of the M3A family is in general age P1 ParB protein complexed with DNA.24 This completed by a water molecule.20 Because of the superposition shows that in such an IrrE–DNA data resolution limit of 3.5 Å, we cannot exclude that complex, severe clashes would occur between the a similar solvent molecule is present in the zinc- double-stranded DNA molecule and the N- and C- bound IrrE structure. In this structure, H82, H86 and terminal domains of IrrE (Fig. 6). Large structural E113 adopt the same conformations as described for rearrangements within IrrE and/or the DNA double the MA clan enzymes. Like in the other structures, helix itself would be required to allow a similar IrrE– H82 is maintained in the correct orientation by a DNA interaction. hydrogen bond with a vicinal residue, N117 in IrrE (Fig. 5), which corresponds to a well-conserved The C-terminal domain has a “sensor” aspartate in the MA enzymes.21 In contrast, the domain fold position of the H86 side chain, the second conserved histidine, is not stabilized in IrrE, whereas it is A previous study demonstrated that deletion of restrained by hydrogen bonds in zinc peptidases the 117 C-terminal residues of D. radiodurans IrrE (Fig. 5c). resulted in a non-functional protein.25 This deleted fragment corresponds to the whole third domain in The HTH domain has a peculiar location the structure of D. deserti IrrE. In the IrrE crystal structure, the last 96 residues (168–263) are folded The IrrE central domain (residues 126–167) is into a six-stranded antiparallel β-sheet with an α- folded in a three-helix bundle, with the second and helix inserted between the third and fourth β- third helices forming the HTH motif. HTH domains strands. The C-terminal domain shows a high level are widespread structures found in many transcrip- of structural similarity to the GAF domain of the tional regulators,22 and, in general, the C-terminal Thermotoga maritima IclR transcription factor.26 GAF helix, called the recognition helix, penetrates the domains are ubiquitous small-molecule binding major groove in protein–DNA complexes. The IrrE units present in signal-transducing proteins (e.g., HTH domain shows a high level of structural Klebsiella pneumoniae sensor histidine kinase CitA), similarity to HTH domains of the Lrp/AsnC bacterial response regulators (e.g., Lrp/AsnC and transcriptional regulatory family.23 However, in IclR families), cGMP-regulated cyclic nucleotide contrast to the proteins of this family, the HTH (PDEs) and other sensor domain of D. deserti IrrE is located neither at the N- proteins.27 Structural comparisons of the C-term- terminal end nor at the C-terminal end of the inal domain of IrrE with the GAF domains of polypeptide chain, which might endow it with K. pneumoniae CitA28 and E. coli cyclic nucleotide some specific properties. A putative interaction of PDE2A29 in complex with citrate and cGMP, IrrE with DNA is in agreement with the surface respectively, show that their ligand-binding sites distribution of positive charge in the area of the N- and their access channels superpose (Fig. 7). A terminal and C-terminal domains adjacent to the small internal hydrophobic cavity (about 61 Å3 recognition helix (Fig. 3b). In the absence of calculated with a 1.4-Å radius probe by CASTp30) structural data regarding DNA binding by the is present and overlaps partially with the ligand- Lrp/AsnC family members and to investigate binding sites of PDE2A and CitA. In the IrrE crystal putative DNA binding by IrrE, we superposed the structure, a putative entrance to this cavity seems HTH domains of D. deserti IrrE onto the bacterioph- blocked by crystal contacts. A depression at the

Fig. 6. HTH domain of D. deserti IrrE. (a) Superposition of the IrrE domain (salmon; residues 126–167) and the bacteriophage P1 ParB HTH domain (green; RCSB PDB code 1ZX4). (b) IrrE–DNA interaction modeled from ParB–DNA interaction. Y160 is in ball-and-stick representation.

158 710 Crystal Structure of Deinococcus deserti IrrE

Fig. 7. The putative sensor domain of D. deserti IrrE (green) compared with the GAF domains (yellow) of (a) PDE2A (RCSB PDB code 1MC0) and (b) CitA (RCSB PDB code 2J80). cGMP and citrate bound to these structures are shown in ball-and-stick representation. C175 and H217 from IrrE are shown in magenta. The IrrE disordered loop is schematized by a dotted line. surface, which, based on structural comparison with To investigate the oligomeric state of IrrE in the GAF domains of CitA and PDE2A, might solution, we performed dynamic light-scattering correspond to the entry channel, is delineated by (DLS) and gel-filtration experiments on a concen- the S206 and L207 residues that are in contact with a trated protein sample. DLS gives an apparent symmetry-related molecule. The long loop (201–211) molecular mass of 78-kDa with a polydispersity of that connects the second and third strands in the IrrE about 20%, suggesting that IrrE in solution is at least GAF domain is directed toward the core of the a dimer. However, under these conditions, gel- twisted β-sheet, as opposed to the PDE2A model, filtration chromatography shows that the elution of closing the access to the interior of the structure (Fig. 7). Thus, blocked access to the hydrophobic pocket of the C-terminal domain might be an artifact related to IrrE crystal packing.

Oligomeric state of IrrE

Transcription factors often recognize palindromic DNA sequences, and for this purpose, most of them act as homodimers. In the IrrE crystals, two vicinal monomers form a dimer related by a 2-fold crystal- lographic symmetry axis (Fig. 8). The interface formed between these two molecules buries approximately 1340 Å2 of surface.31 Based on PISA criteria, this interface would correspond to a stable dimer.32 It is formed by polar and hydrophobic interactions, including P30, T34, H35, M38, H39, G43, I44, L46, T47, F48, M49, P50, S69, V71, R72, P73, Q76 and R77. Numerous backbone–backbone hydrogen bonds participate in the contacts. All amino acid residues that are involved in the interaction with the symmetry-related molecule are conserved in Deinococcus IrrE, apart from T47, which Fig. 8. IrrE crystallographic homodimer. (a) N-terminal is a lysine in D. radiodurans, and H39, which is domains of the two symmetry-related monomers form a replaced by a serine and an alanine in D. geothermalis dimer interface. (b) Interactions of H35 and F48 and their and D. radiodurans IrrE, respectively. symmetry equivalents present at the dimer interface.

159 Crystal Structure of Deinococcus deserti IrrE 711

IrrE corresponds to a molecular mass of 40–50 kDa, identical with that of the wild type, suggesting that which does not fit either a monomer or a dimer. these mutations do not interfere with the radio- tolerance phenotype (Fig. 2b). Radiotolerance phenotype of IrrE mutants

The chromosomal irrE gene was deleted and then Discussion wild-type irrE and irrE genes containing point mutations were expressed from a plasmid that can When bacteria are exposed to genotoxic agents, replicate in D. deserti to study the importance of the specific genes are induced in response to DNA different structural domains of IrrE in D. deserti.As damage. In E. coli, the genes of the well-known SOS in D. radiodurans, the absence of the IrrE protein system are under the control of the LexA repressor sensitizes the D. deserti cells to UV and gamma that is regulated by a RecA-mediated self-cleavage. radiation (Fig. 2b), further supporting the notion Interestingly, D. radiodurans possesses two LexA that the role of IrrE in radiotolerance is conserved in homologs: LexA1 and LexA2. As in E. coli, D. Deinococci. radiodurans LexA1 and LexA2 undergo RecA-stimu- To explore the impact of zinc-binding site on IrrE lated self-cleavage in vitro, but in contrast, they are – function, we constructed H86S and E83Q single irrelevant to recA induction.34 36 As in E. coli, many mutants. The H86S mutation was designed to affect genes, including recA, are induced in D. radiodurans mostly the metal binding, whereas the E83Q mutant after treatment by genotoxic agents, which impli- was designed to target a putative proteolytic activity cates the existence of precise genetic control in these as this mutation should not disturb metal binding. bacteria. IrrE, first identified as an orphan gene in D. Both mutations lead to bacteria as radiosensitive as radiodurans and later found in D. geothermalis4 and the ΔirrE bacteria (Fig. 2b). The metal-binding site of D. deserti (this work) as well, is related to the IrrE indeed seems essential for the tolerance toward extraordinary radiotolerance of D. radiodurans and UV light and gamma radiation. From these results, it to the positive regulation of recA and pprA, both appears that the N-terminal domain of IrrE could involved in DNA repair.5,7 The ability of D. deserti display a proteolytic activity similar to the one IrrE to be functional in D. radiodurans indicates a described for the MA clan zinc peptidases. conservation of IrrE function in Deinococci. The The Y160 residue is present in the consensus presence of a common radiation/desiccation sequence defined by the LacI-type DNA-binding response motif (RDRM) found in the upstream HTH domain signature (PROSITE33). This is the region of recA, pprA and other highly induced single aromatic residue located on the surface of the genes in D. radiodurans and their homologs in D. predicted recognition helix, and we supposed that geothermalis4 suggests that IrrE could be the central its aromatic character could be important in the deinococcal transcription enhancer that would potential interaction with DNA. Y160 was therefore target, either directly or indirectly, this common mutated to alanine. Partial loss of viability of D. DNA sequence. deserti cells carrying the IrrE-Y160A mutant follow- To gain more insights on this central DNA repair ing exposure to ionizing radiation and UV light regulator, we determined the three-dimensional suggests a role of this residue for overall IrrE structure of D. deserti IrrE. Our crystallographic function, but the effect is not as drastic as the work reveals that IrrE has a unique combination of mutations of the residues involved in zinc binding three domains: one small thermolysin-like domain, and/or putative catalysis (Fig. 2b). one HTH motif and one GAF domain typically In the GAF domain, C175 and H217 are the only found in sensor/signaling proteins (Fig. 3). From polar residues that delineate the otherwise hydro- our mutant analysis, the first and third domains phobic internal cavity (Fig. 7). If this small cavity is would correspond to the most critical regions for indeed a ligand-binding site, then these two amino radiotolerance (Fig. 2b). Although we cannot acids might be involved in specific interactions with exclude that enzymatic posttranslational modifica- the unknown ligand. We changed C175 and H217 tions in Deinococcus could modify the intriguing for hydrophobic residues with an equivalent steric structural organization of IrrE or modulate its hindrance. The C175A mutant behaves exactly as biological activity, we propose that IrrE would not the wild-type bacteria. In contrast, tolerance toward be a classic transcriptional factor that activates gene both UV and γ-rays is strongly decreased for the expression by its direct binding to double-stranded H217L mutant (Fig. 2b). DNA. Finally, since the DLS and gel-filtration analyses Although a large basic patch surrounding the were inconclusive regarding the IrrE oligomeric HTH domain of IrrE suggests that this region may state, a double mutant (H35A/F48A) was con- interact with DNA, the IrrE DNA-binding motif has structed to disrupt the hypothetical dimer in vivo. a very peculiar location in the center of the structure H35 and F48 are present in the core region of the between the putative zinc peptidase and sensor dimer interface and are strictly conserved in domains. A model based on the bacteriophage P1 Deinococcus IrrE (Figs. 1 and 8). An analysis by ParB–DNA structure shows that large conforma- PISA predicted that these mutations would disturb tional rearrangements of the DNA and the first and the dimer's stability. Following irradiation, the third IrrE domains would be necessary to avoid survival of the D. deserti H35A/F48A mutant is numerous clashes between the DNA and the

160 712 Crystal Structure of Deinococcus deserti IrrE protein. It is however difficult to conclude from our Materials and Methods model if such large changes of the protein structure are possible. We cannot exclude that more flexible molecules, such as small single-stranded DNA or Bacterial strains, cultures, media and transformation RNA, might be better IrrE substrates. Because of its central position in the protein structure, the HTH Bacterial strains are listed in Table 1. All the data related to D. radiodurans cultures, growth media and cell domain may influence the cross-talk between the N- 35 transformation were previously described. terminal and C-terminal domains. As demonstrated 28 D. deserti strains were grown at 30 °C in 10-fold diluted by the recent CitA study, ligand binding induces tryptic soy broth (TSB/10) supplemented with 5% of a some conformational changes of the GAF domain solution with the following composition (per liter of water): structure that are responsible for the signal trans- FeSO4.7H2O, 140 mg; MgSO4.7H2O, 2 g; CaCl2.2H2O, 2 g; duction through the membrane. These sensor EDTA (ethylenediaminetetraacetic acid) (Titriplex II), 0.4 g; domains exhibit a very conserved fold, and super- and trace element solution [per liter of water: ZnSO4.7H2O, position of ligand complexes of CitA or PDE2A28,29 430 mg; MnSO4.H2O, 1.3 g; NaMoO4.2H2O, 750 mg; H3BO3, on the D. deserti IrrE model indicates that the H217L 2.8 g; CoSO4.7H2O, 70 mg; CuSO4.7H2O, 26 mg], 20 mL, or mutation would affect the biological function of IrrE on plates containing the same medium solidified with 1.6% (Fig. 7). The radiotolerance of the H217L mutant is agar. When required, antibiotics were used at the following concentrations for D. deserti: streptomycin, 10 μg/mL; indeed greatly diminished (Fig. 2b). Although we kanamycin, 10 μg/mL; and chloramphenicol, 3 μg/mL. suppose that in our model the GAF domain is maintained in a closed configuration, the opening of the entrance channel could be easily caused by DNA, PCR and plasmids movement of loop regions upon binding of a putative effector in the vicinity of H217 (Fig. 7). Oligonucleotides used in this study are listed in Supple- This putative natural ligand is unknown, however. mental Material. Chromosomal DNA of D. deserti was We hypothesize that, like CitA, the C-terminal isolated from stationary-phase cells in TSB/10 medium as domain of IrrE could adopt different conformations previously described.38 Amplification of genomic DNA by and would be able to transfer a signal to the rest of PCR was performed with HiFi polymerase (ABgene). the structure. In that hypothesis, the HTH domain would act in concert with the other structural Inverse PCR domains to relay stimuli between the peripheral IrrE domains. The irrEf1 and irrEr2 primers designed from the D. Our phenotypic study supports the hypothesis radiodurans irrE gene were used to amplify a 350-bp PCR D. deserti that IrrE would be a putative metallopeptidase and fragment from genomic DNA. After sequencing of this fragment, FV7 and FV8 primers were used to that this function could be crucial for radiotolerance. perform inverse PCR as previously described38 except that The N-terminal domain of D. deserti IrrE exhibits a chromosomal DNA (5 μg) was digested with NarI. Inverse large negatively charged cleft accessible to solvent, PCR was carried out on the ligated circular DNA (200 ng) suggesting that IrrE may cleave large substrates previously denaturated for 15 min at 100 °C using HiFi (Figs. 1 and 4). From these observations, we propose DNA polymerase (ABgene) in the presence of FV7 and that IrrE would enhance transcription by the FV8 primers. The PCR products contained a major 800-bp proteolytic cleavage of a second, so far unidentified fragment that was purified, cloned in pGEM-TEasy TA “transcriptional messenger” that is probably a rather cloning vector (Promega) and sequenced using universal basic protein. The conserved palindromic RDRM sequencing primers (T7 and SP6 promoters). The amplified irrE found in the upstream region of radiation-induced fragment contains 396 bp upstream of the encoding 4 region and 400 bp of the encoding region itself. The FV36 genes in Deinococcus could be targeted by this primer located upstream of this irrE sequence and the unknown transcriptional messenger, as direct bind- FV21 primer located downstream of D. radiodurans irrE ing of IrrE to the RDRM has not, to our knowledge, inside the folP gene were used to amplify from genomic been demonstrated. Indeed, preliminary experiments DNA of D. deserti VCD115 a 1630-bp DNA fragment indicated that IrrE did not bind to the promoter containing the irrE gene. The nucleotide sequence of this region of the radiation-induced pprA gene.37 fragment was verified by DNA sequencing. Deinococcal radiotolerance is based on multiple gene induction, and IrrE is a major player in this Plasmid construction process. The constitutive expression of IrrE certainly enables the bacteria to face stress conditions very p12720, the plasmid designed for irrE complementation rapidly. At an early step of the reaction, IrrE would in D. radiodurans, was obtained by inserting a DNA detect a molecular signal related to cellular damages, fragment containing D. deserti irrE (generated by PCR with FV45 and FV48 primers and D. deserti genomic DNA an event that would trigger DNA repair response. In 39 this context, the complex multidomain architecture as template) between the NdeI and DraI sites of p11559. of D. deserti IrrE could be a key element, as the IrrE was overproduced in E. coli using p12714. This plasmid was constructed by cloning an NdeI–EcoRI DNA transfer of environmental stimuli could involve fragment corresponding to D. deserti irrE (generated by concerted activities of each of the IrrE domains. PCR with FV45/FV47 primers and D. deserti genomic Understanding the precise role of IrrE in regulation DNA as template) between the NdeI and EcoRI sites awaits further studies to identify its potential of pET-TEV.40 All constructions were verified by DNA ligands, including other critical target proteins. sequencing.

161 Crystal Structure of Deinococcus deserti IrrE 713

Table 1. Bacterial strains and plasmids

Genotype or other relevant characteristics Source or reference E. coli DH5α supE44 ΔlacU(ϕ80lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1 Laboratory stock SCS110 endA dam dcm supE44 Δ(lac-proAB)(F'traD36 proAB lacIqZΔM15) Laboratory stock − − BL21(DE3) F' ompT hsdSB (r BmB ) gal dcm (λDE3) Laboratory stock

D. radiodurans R1 ATCC13939 Laboratory stock GY11921 R1/p11559 38 GY12415 ΔirrEΩcat Laboratory stock GY12801 ΔirrEΩcat/p11559 This work GY12805 ΔirrEΩcat/p12720 This work

D. deserti VCD115 Isolated from Sahara Desert sand 11 RD19 As VCD115 but streptomycin resistant (SmR) This work RD42 As RD19 but ΔirrEΩkan This work RD51 RD42/pRD51 (irrE+) This work RD55 RD42/pRD55 (irrE H35A F48A) This work RD56 RD42/pRD56 (irrE Y160A) This work RD57 RD42/pRD57 (irrE H217L) This work RD58 RD42/pRD58 (irrE H86S) This work RD59 RD42/pRD59 (irrE E83Q) This work RD60 RD42/pRD60 (irrE C175A) This work

Plasmids p11559 Expression vector in D. radiodurans;Pspac,PtufAΩlacI, SpcR in E. coli 39 and in D. radiodurans p12720 p11559 with a fragment encoding D. deserti IrrE This work pET-TEV pET28a derivative; for protein overproduction in E. coli 40 p12714 D. deserti irrE in pET-TEV This work pCR4Blunt-TOPO Cloning vector Invitrogen pI3 Shuttle vector that replicates in E. coli, D. radiodurans and D. deserti; 41,42 ApR in E. coli,CmR in Deinococcus pRD48 D. deserti irrE in pCR4Blunt-TOPO This work pRD51 D. deserti irrE in pI3 This work pRD55 As pRD51 but encoding IrrE with mutations H35A and & F48A This work pRD56 As pRD51 but encoding IrrE with mutation Y160A This work pRD57 As pRD51 but encoding IrrE with mutation H217L This work pRD58 As pRD51 but encoding IrrE with mutation H86S This work pRD59 As pRD51 but encoding IrrE with mutation E83Q This work pRD60 As pRD51 but encoding IrrE with mutation C175A This work

Construction and complementation of a D. deserti construct, which does not replicate in D. deserti,was ΔirrE mutant introduced into D. deserti RD19, a spontaneous strepto- mycin resistant derivative of strain VCD115, by a classic E. Inthecourseofthisstudy,sequencingofthegenomeof coli CaCl2 transformation method. Replacement of the irrE D. deserti was initiated. The availability of a first draft gene by the kanamycin resistance gene by double cross- sequence allowed the construction of an irrE mutant in over and total absence of irrE were confirmed by which irrE was replaced by a kanamycin resistance gene. diagnostic PCRs. The resulting irrE deletion mutant was The promoter region until the initiation codon of the D. called RD42. deserti tuf gene encoding elongation factor EF-Tu was Among several plasmids tested, it was observed that the translationally fused to the kanamycin resistance gene to D. radiodurans–E. coli shuttle vector pI341,42 could be used express the kanamycin resistance gene in D. deserti.Thetuf as a replicating plasmid in D. deserti using chloramphe- promoter region (Ptuf) was first amplified with TufBam nicol resistance as a selection marker. Therefore, wild-type and TufKan primers (D. deserti genomic DNA as template); irrE and point mutation derivatives were cloned in pI3 for kanamycin resistance gene, with KanTuf and KanPst complementation experiments. First, a 1.2-kb fragment primers [pCR4Blunt-TOPO (Invitrogen) as template]. carrying irrE and 350 bp of upstream DNA expected to TufKan and KanTuf primers are partially complementary, contain its promoter was amplified with Pfx polymerase allowing use of the products of the first PCR as primers (Invitrogen) and irrE-XbaFWand irrE-Hd3RV primers and and templates in a second PCR (together with TufBam and cloned in pCR4Blunt-TOPO (Invitrogen). KanPst primers). The resulting Ptuf::kan fusion was cloned irrE point mutations were introduced in the resulting as a BamHI–PstI fragment in E. coli cloning vector pUC19. plasmid using a site-directed mutagenesis kit (Stratagene) One-kilobase fragments corresponding to DNA upstream and the oligonucleotides indicated in Supplemental (amplified with IrrE-Up3-KpnI and IrrE-Up3-BamHI Material. The correctness of the sequences was confirmed primers) and downstream (IrrE3-Dn-PstI and IrrE3-Dn- by DNA sequencing. Then, irrE and derivatives were HindIII primers) of D. deserti irrE were cloned in the cloned as XbaI–HindIII fragments in pI3, and the resulting correct orientation upstream and downstream of the plasmids were called pRD51 (irrE+) and pRD55 to pRD60 kanamycin resistance cassette, respectively. The resulting (six irrE mutants).

162 714 Crystal Structure of Deinococcus deserti IrrE

Radiotolerance phenotype analysis (polyethylene glycol) 3350 solution. Crystals grew to final dimensions of about 100 μm×80 μm×10 μmwithin3– Gamma irradiation of D. radiodurans 4 days and belong to the P21212 space group with one protein molecule per asymmetric unit and 47% of solvent Bacteria transformed by p11559 derivatives were grown (cell parameters: a=86.05 Å; b=52.81 Å; c=64.91 Å). To μ in TGY2X media supplemented with 75 g/mL of study IrrE zinc binding, we soaked the native protein spectinomycin and 10 mM IPTG. Exponential-phase crystals in mother liquor solution supplemented with cultures (A650 =1.1) were concentrated 10 times in TGY2X 10 mM ZnSO for 2 min. Prior to flash freezing in liquid 137 4 and irradiated on ice with a Cs irradiation system nitrogen, the crystals were transferred to the respective (Institut Curie, Orsay, France) at a dose rate of 56.6 Gy/min. solutions supplemented with 30% glycerol. All X-ray Following irradiation, diluted samples were plated on diffraction data were collected at 100 K at the European TGY plates supplemented with 1 mM IPTG and incubated Synchrotron Radiation Facility (Grenoble, France). Data – 44 45 3 4 days at 30 °C before the colonies were counted. sets were processed using MOSFLM (Se-Met) or XDS (IrrE-1Zn) and scaled with SCALA from the CCP4 suite.46 UV and gamma irradiation of D. deserti Data collection and processing statistics are listed in Table 2. The D. deserti IrrE crystal structure was solved using the For UV irradiation, 20-μL serial dilutions of D. deserti single anomalous dispersion method, with data collected strains grown to exponential phase were spotted on at λ=0.9793 Å and processed to 2.6-Å resolution. The solidified medium. Immediately after drying, the cells on structure was automatically determined using the auto- the plates were exposed to UV-C (254 nm) for the desired SHARP protocol.47,48 The program SHELXD49 found five dose and subsequently incubated in the dark. For gamma of seven selenium sites predicted by the IrrE primary irradiation, cells grown to late exponential phase were structure. The two missing sites corresponded to methio- concentrated in growth medium and exposed on ice to nines located at the N-terminal region, which turned out gamma rays for the desired dose (39.6 Gy/min, 60Co to be disordered in the crystal structure. An interpretable source; CEA/Cadarache, France). Then, 20 μL of serial electron density map obtained after solvent flattening dilutions was spotted on solidified medium. Colonies with 51% solvent (figure of merit=0.63) enabled the were counted after 3–4 days of incubation. program ARP/wARP50 to automatically build 175 of 247 residues present in the final IrrE model. Further model Protein expression and purification building and refinement were done using the programs COOT51 and REFMAC5,52 respectively. Five percent of the reflections were set aside to monitor refinement progress Ten milliliters of an overnight culture of E. coli BL21 with the R factor. This way, 45 additional residues were (DE3) cells harboring the p12714 plasmid was used to free inoculate 1 L of LB medium supplemented by 25 μg/mL Table 2. of kanamycin at 37 °C. Protein overexpression was Data collection and refinement statistics induced by the addition of 0.1 mM IPTG at A =0.5. 650 IrrE Se-Met After 3 h, cells were harvested by centrifugation (5500g, apo form IrrE zinc-bound form 4 °C) and resuspended in 50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 10 mM imidazole and a cocktail of EDTA- Beamline ID23-1 ID29 Wavelength (Å) 0.97930 0.97618 free anti-proteases (Roche). Cells broken by sonication – – were centrifuged (16,000g, 4 °C), and clear lysate was Resolution range (Å) 40.00 2.60 40.00 3.50 (2.74–2.60) (3.69–3.50) incubated overnight with Ni-nitrilotriacetic acid beads No. of unique observations 9581 4071 (Qiagen) at 4 °C. After extensive washing with 50 mM Average redundancy 11.2 (11.7) 3.5 (3.6) – Tris HCl, pH 7.5, 150 mM NaCl and 20 mM imidazole, the Rsym (%) 12.9 (43.4) 7.7 (36.4) protein was eluted with a linear gradient of 0.02–0.3 M I/σI 18.6 (7.6) 8.1 (2.0) imidazole in 50 mM Tris–HCl, pH 7.5, and 150 mM NaCl. Completeness (%) 100.0 (100.0) 99.4 (99.6) The purest fractions were collected and dialyzed over- Wilson B-factor (Å2)4463 – Resolution range (Å) 20.00–2.60 20.00–3.50 night against 50 mM Tris HCl, pH 7.5, and 150 mM NaCl – – at 4 °C. The dialyzed sample was concentrated and loaded (2.67 2.60) (3.59 3.50) on a Superdex-200 HR 16/60 column pre-equilibrated Rfactor/Rfree 0.22/0.25 0.24/0.31 – (0.27/0.31) (0.23/0.30) with 50 mM Tris HCl, pH 7.5, and 150 mM NaCl. Purified No. of residues 245 243 protein was concentrated to 5.6 mg/mL. Se-methiony- No. of atoms lated protein was produced following a described Protein 1875 1841 43 protocol and purified as described above. The purified Solvent molecules 49 — recombinant 32-kDa IrrE exhibits the additional peptide Zinc — 1 MGSSHHHHHHSSGENLYFQG that includes a TEV Average B-factors (Å2) protease cleavage site. The DLS experiment using con- Protein 43.5 78 – Solvent molecules 47 — centrated protein sample (5.6 mg/mL, 50 mM Tris HCl, — pH 7.5, and 150 mM NaCl) was performed on a Zinc 63 ™ r.m.s.d. from ideality DynaPro Dynamic Light Scattering Instrument from Bond lengths (Å2) 0.016 0.010 Wyatt Technology. Bond angles (°) 1.45 1.07 Ramachandran statistics (%) Crystallographic study Most favored 89.3 83.8 Additionally allowed 10.7 13.8 Generously allowed — 1.0 The crystallization of D. deserti IrrE was performed using Disallowed — 1.4a the hanging-drop vapor-diffusion method at 281 K. Crys- Values in parentheses refer to the highest-resolution shell. tals were obtained by mixing equal volumes of protein a (5.6 mg/mL) and precipitant solutions over wells contain- A201, R226 and S233 are the residues present in the Ramachandran disallowed region. ing 0.2 M potassium fluoride and 20% (w/v) PEG

163 Crystal Structure of Deinococcus deserti IrrE 715

built (Rfactor/Rfree =0.34/0.38); however, two loop regions 4. Makarova, K. S., Omelchenko, M. V., Gaidamakova, (residues 101–106 and 176–196) were still missing in the E. K., Matrosova, V. Y., Vasilenko, A., Zhai, M. et al. model. An improved electron density map was obtained (2007). Deinococcus geothermalis: the pool of extreme after simulated annealing (at 2500 K), energy minimization radiation resistance genes shrinks. PLoS ONE, 2, e955. and individual B-factor refinement using CNS,31 which 5. Tanaka, M., Earl, A. M., Howell, H. A., Park, M. J., Eisen, enabled building of 27 extra residues. Water molecules were J. A., Peterson, S. N. & Battista, J. R. (2004). Analysis of added manually at the final stages of the refinement (final Deinococcus radiodurans's transcriptional response to Rfactor/Rfree=0.22/0.25), and the quality of the models was ionizing radiation and desiccation reveals novel pro- checked by PROCHECK implemented in the CCP4 suite.46 teins that contribute to extreme radioresistance. Genetics, In the zinc-bound IrrE crystals, positive difference electron 168,21–33. − σ density in the (Fo Fc) difference maps contoured at the 4 6. Harris, D. R., Tanaka, M., Saveliev, S. V., Jolivet, E., level revealed the presence of a bound heavy atom, Earl, A. M., Cox, M. M. & Battista, J. R. (2004). supposedly a zinc ion, which was included in the model. Preserving genome integrity: the DdrA protein of Refinement of the zinc-bound structures was performed Deinococcus radiodurans R1. PLoS Biol. 2, e304. using the REFMAC5 and CNS programs. Diffraction data 7. Narumi, I., Satoh, K., Cui, S., Funayama, T., Kitayama, and refinement statistics are listed in Table 2. S. & Watanabe, H. (2004). PprA: a novel protein from Deinococcus radiodurans that stimulates DNA ligation. Mol. Microbiol. 54, 278–285. Accession codes 8. Earl, A. M., Mohundro, M. M., Mian, I. S. & Battista, J. R. (2002). The IrrE protein of Deinococcus radiodurans The D. deserti irrE gene has been deposited to the R1 is a novel regulator of recA expression. J. Bacteriol. European Molecular Biology Laboratory data bank with 184, 6216–6224. code FM200036. Diffraction and coordinate data sets have 9. Hua, Y., Narumi, I., Gao, G., Tian, B., Satoh, K., been deposited to the RCSB PDB with accession codes Kitayama, S. & Shen, B. (2003). PprI: a general 3DTE and 3DTI. switch responsible for extreme radioresistance of Deinococcus radiodurans. Biochem. Biophys. Res. Commun. 306,354–360. 10. Gao, G., Le, D., Huang, L., Lu, H., Narumi, I. & Hua, Y. (2006). Internal promoter characterization and Acknowledgements expression of the Deinococcus radiodurans pprI–folP gene cluster. FEMS Microbiol. Lett. 257, 195–201. This work was supported by the Centre National 11. de Groot, A., Chapon, V., Servant, P., Christen, R., Fischer-Le Saux, M., Sommer, S. & Heulin, T. (2005). de la Recherche Scientifique, the Commissariat à Deinococcus deserti sp. nov., a gamma-radiation-tolerant l'Energie Atomique, Electricité de France and the bacterium isolated from the Sahara Desert. Int. J. Syst. Agence Nationale de la Recherche (ANR-07-BLAN- Evol. Microbiol. 55, 2441–2446. 0106-02). We thank the Institut Curie for allowing 12. Kabsch, W. & Sander, C. (1983). Dictionary of protein use of the 137Cs irradiation system; V. Favaudon for secondary structure: pattern recognition of hydrogen- helping with gamma irradiation; B. Gallet and the bonded and geometrical features. Biopolymers, 22, ROBIOMOL platform for helping with the initial 2577–2637. protein expression tests; A. Bailone and Y. Timsit for 13. Gouet, P., Courcelle, E., Stuart, D. I. & Metoz, F. (1999). ESPript: analysis of multiple sequence alignments in providing helpful discussions; and M. DuBow for 15 – his careful reading of the manuscript. We are also PostScript. Bioinformatics, , 305 308. 14. DeLano, W. L. (2002). The PyMOL User's Manual. very grateful to the European Synchrotron Radia- DeLano Scientific, San Carlos, CA. tion Facility (Grenoble, France) for allowing access 15. Baker, N., Sept, D., Joseph, S., Holst, M. & McCammon, to the MX beamlines. J. (2001). Electrostatics of nanosystems: application to microtubules and the ribosome. Proc. Natl Acad. Sci. USA, 98, 10037–10041. Supplementary Data 16. Dietmann, S. & Holm, L. (2001). Identification of homology in protein structure classification. Nat. Struct. 8 – Supplementary data associated with this article Biol. ,953 957. can be found, in the online version, at doi:10.1016/ 17. Comellas-Bigler, M., Lang, R., Bode, W. & Maskos, K. j.jmb.2008.12.062 (2005). Crystal structure of the E. coli dipeptidyl carboxypeptidase Dcp: further indication of a ligand- dependent hinge movement mechanism. J. Mol. Biol. 349,99–112. References 18. Rawlings, N. D., Morton, F. R., Kok, C. Y., Kong, J. & Barrett, A. J. (2008). MEROPS: the peptidase database. 1. Blasius, M., Hübscher, U. & Sommer, S. (2008). Nucleic Acids Res. 36, D320–D325. Deinococcus radiodurans: what belongs to the survival 19. Matthews, B. W. (1988). Structural basis of the action of kit? Crit. Rev. Biochem. Mol. Biol. 43, 221–238. thermolysin and related zinc peptidases. Acc. Chem. Res. 2. Cox, M. M. & Battista, J. R. (2005). Deinococcus radio- 21,333–340. durans: the consummate survivor. Nat. Rev. Microbiol. 3, 20. Roe, R. R. & Pang, Y. P. (1999). Zinc's exclusive 882–892. tetrahedral coordination governed by its electronic 3. Daly, M. J., Gaidamakova, E. K., Matrosova, V. Y., structure. J. Mol. Model. 5, 134–140. Vasilenko, A., Zhai, M., Leapman, R. D. et al. (2007). 21. Rawlings, N. D. & Barrett, A. J. (1995). Evolutionary Protein oxidation implicated as the primary determi- families of metallopeptidase. Methods Enzymol. 248, nant of bacterial radioresistance. PLoS Biol. 5, e92. 105–120.

164 716 Crystal Structure of Deinococcus deserti IrrE

22. Aravind, L., Anantharaman, V., Santhanam, B., Mohan 36. Sheng, D., Zheng, Z., Tian, B., Shen, B. & Hua, Y. Babu, M. & Iyer, L. M. (2005). The many faces of the (2004). LexA analog (dra0074) is a regulatory protein helix-turn-helix domain: transcription regulation and that is irrelevant to recA induction. J. Biochem. 136, beyond. FEMS Microbiol. Rev. 29,231–262. 787–793. 23. Thaw, P., Sedelnikova, S. E., Muranova, T., Wiese, S., 37. Ohba, H., Satoh, K., Yanagisawa, T. & Narumi, I. (2005). Ayora, S., Alonso, J. C. et al. (2006). Structural insight The radiation responsive promoter of the Deinococcus into gene transcriptional regulation and effector radiodurans pprA gene. Gene, 363,133–141. binding by the Lrp/AsnC family. Nucleic Acids Res. 38. Gutsche, I., Vujicic-Zagar, A., Siebert, X., Servant, P., 34, 1439–1449. Vannier, F., Castaing, B. et al. (2008). Complex 24. Schumacher, M. A. & Funnell, B. E. (2005). Structures oligomeric structure of a truncated form of DdrA: a of ParB bound to DNA reveal mechanism of partition protein required for the extreme radiotolerance of complex formation. Nature, 438, 516–519. Deinococcus. Biochim. Biophys. Acta, 1784, 1050–1058. 25. Gao, G. J., Lu, H. M., Huang, L. F. & Hua, Y. J. (2005). 39. Mennecier, S., Coste, G., Servant, P., Bailone, A. & Construction of DNA damage response gene pprI Sommer, S. (2004). Mismatch repair ensures fidelity of function-deficient and function-complementary replication and recombination in the radioresistant mutants in Deinococcus radiodurans. Chin. Sci. Bull. organism Deinococcus radiodurans. Mol. Genet. Genomics, 50,311–316. 272, 460–469. 26. Zhang, R. G., Kim, Y., Skarina, T., Beasley, S., 40. Houben, K., Marion, D., Tarbouriech, N., Ruigrok, Laskowski, R., Arrowsmith, C. et al. (2002). Crystal R. W. H. & Blanchard, L. (2007). Interaction of the C- structure of Thermotoga maritima 0065, a member of terminal domains of Sendai virus N and P proteins: the IclR transcriptional factor family. J. Biol. Chem. 277, comparison of polymerase–nucleocapsid interactions 19183–19190. within the paramyxovirus family. J. Virol. 81,6807–6816. 27. Ho, Y. S., Burden, L. M. & Hurley, J. H. (2000). 41. Masters, C. I. & Minton, K. W. (1992). Promoter probe Structure of the GAF domain, a ubiquitous signaling and shuttle plasmids for Deinococcus radiodurans. motif and a new class of cyclic GMP receptor. EMBO J. Plasmid, 28,258–261. 19, 5288–5299. 42. Meima, R. & Lidstrom, M. E. (2000). Characterization 28. Sevvana, M., Vijayan, V., Zweckstetter, M., Reinelt, S., of the minimal replicon of a cryptic Deinococcus Madden, D. R., Herbst-Irmer, R. et al. (2008). A ligand- radiodurans SARK plasmid and development of induced switch in the periplasmic domain of sensor versatile Escherichia coli–D. radiodurans shuttle vectors. histidine kinase CitA. J. Mol. Biol. 377, 512–523. Appl. Environ. Microbiol. 66, 3856–3867. 29. Martinez, S. E., Wu, A. Y., Glavas, N. A., Tang, X. B., 43. Doublié, S. (1997). Production of selenomethionyl Turley, S., Hol, W. G. & Beavo, J. A. (2002). The two proteins for phase determination. Methods Enzymol. GAF domains in phosphodiesterase 2A have distinct 276, 523–530. roles in dimerization and in cGMP binding. Proc. Natl 44. Leslie, A. G. W. (1999). Integration of macromolecular Acad. Sci. USA, 99, 13260–13265. diffraction data. Acta Crystallogr. D55, 1696–1999. 30. Dundas, J., Ouyang, Z., Tseng, J., Binkowski, A., 45. Kabsch, W. (1993). Automatic processing of rotation Turpaz, Y. & Liang, J. (2006). CASTp: computed atlas diffraction data from crystals of initially unknown of surface topography of proteins with structural and symmetry and cell constants. J. Appl. Crystallogr. 26, topographical mapping of functionally annotated 795–800. residues. Nucleic Acids Res. 34,W116–W118. 46. Collaborative Computational Project, Number 4. 31. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano, (1994). The CCP4 suite: programs for protein crystal- W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). lography. Acta Crystallogr. D50, 760–763. Crystallography & NMR System: a new software 47. De la Fortelle, E. & Bricogne, G. (1997). Maximum- suite for macromolecular structure determination. likelihood parameter refinement program for MIR Acta Crystallogr., Sect. D: Biol. Crystallogr. 54, 905–921. and MAD phasing. Methods Enzymol. 276, 472–494. 32. Krissinel, E. & Henrick, K. (2007). Inference of macro- 48. Vonrhein, C., Blanc, E., Roversi, P. & Bricogne, G. molecular assemblies from crystalline state. J. Mol. Biol. (2007). Automated structure solution with autoSHARP. 372,774–797. Methods Mol. Biol. 364,215–230. 33. Hulo, N., Bairoch, A., Bulliard, V., Cerutti, L., De 49. Sheldrick, G. M. (2008). A short history of SHELX. Castro, E., Langendijk-Genevaux, P. S. et al. (2006). The Acta Crystallogr. A64,112–122. PROSITE database. Nucleic Acids Res. 34,D227–D230. 50. Perrakis, A., Morris, R. & Lamzin, V. S. (1999). 34. Narumi, I., Satoh, K., Kikuchi, M., Funayama, T., Automated protein model building combined with Yanagisawa, Y., Watanabe, H. & Yamamoto, K. (2001). iterative structure refinement. Nat. Struct. Biol. 6, The LexA protein from Deinococcus radiodurans is not 458–463. involved in RecA induction following gamma irradia- 51. Emsley, P. & Cowtan, K. (2004). Coot: model-building tion. J. Bacteriol. 183, 6951–6956. tools for molecular graphics. Acta Crystallogr. D60, 35. Bonacossa de Almeida, C., Coste, G., Sommer, S. & 2126–2132. Bailone, A. (2002). In Deinococcus radiodurans, RecA 52. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). protein quantification reveals involvement of RecA, Refinement of macromolecular structures by the but not LexA, in its regulation. Mol. Genet. Genomics, maximum-likelihood method. Acta Crystallogr. D53, 268,28–41. 240–255.

165 166

VII Discussion et perspectives

167

168 VII) Discussion et perspectives

Deinococcus radiodurans est un des organismes les plus radiotolérants, dû en grande partie à sa capacité de réparer des dommages massifs de l’ADN. Différents travaux ont été réalisés et hypothèses/mécanismes proposés au cours des dernières années pour expliquer cette radiotolérance. Pour réparer son génome, cette bactérie utilise des protéines conservées chez de nombreux organismes (ex : RecA, PolA), et y associe des protéines spécifiques aux Deinococcus (DdrA, DdrB, DdrC, DdrD, PprA). De plus, D. radiodurans possède un régulateur spécifique aux Deinococcus, IrrE, nécessaire à la radiotolérance et à l’induction d’au moins une trentaine de protéines dont RecA et PprA. En amont d’une vingtaine de gènes radio-induits (ex : recA, ddrA, ddrO, pprA) se trouve un motif de 17 pb nommé le « radiation desiccation response motif » (RDRM). DdrO est un autre régulateur transcriptionnel spécifique aux Deinococcus qui pourrait être important pour la radiotolérance. Contrairement à E. coli ou à d’autres bactéries radiosensibles, D. radiodurans ne possède pas d’ADN polymérases translésionnelles et n’est pas mutable par les UV. Enfin, il est proposé que la réparation de l’ADN chez D. radiodurans est aidée par un rapport Mn/Fe élevé (qui protège les protéines de l’oxydation) ainsi que par un nucléoide hyper-compact (qui limiterait la diffusion des fragments d’ADN et faciliterait la réparation). Cependant, malgré ces résultats/hypothèses, le mécanisme de radiotolérance chez D. radiodurans n’est pas encore complètement élucidé, c’est pourquoi l’étude de Deinococcus deserti, une bactérie isolée du Sahara, a été entreprise. Afin de mieux comprendre l’extrême radiotolérance de D. deserti en particulier et des bactéries du genre Deinococcus en général, le génome de D. deserti a été séquencé, analysé, et comparé à celui de D. radiodurans et D. geothermalis (dont le génome a été publié pendant ma thèse), afin de voir les points communs et les différences entre ces bactéries. Le but de ma thèse était également de mettre au point des outils génétiques pour D. deserti et d’étudier des gènes (potentiellement) impliqués dans la radiotolérance de D. deserti.

1) Deinococcus deserti, un nouveau modèle pour étudier la radiotolérance des Deinococcus

L’étude du génome de D. deserti ainsi que d’autres analyses ont montré qu’elle possède des homologues à plusieurs gènes (potentiellement) impliqués dans la radiotolérance comme irrE, ddrA, ddrB, ddrC (après correction du sens du gène ddrC de D. radiodurans et

169 D. geothermalis), ddrD, ddrO et pprA, un ratio Mn/Fe élevé et un nucléoide hyper-compact. D. deserti présente le RDRM au niveau des mêmes gènes que D. radiodurans (ex : ddrA, ddrB, ddrO et pprA) et également en amont de quelques gènes supplémentaires. Par rapport à D. radiodurans et D. geothermalis, D. deserti possède des gènes supplémentaires ainsi que de multiples homologues de gènes (potentiellement) liés à la réparation de l’ADN (ex : imuY et dnaE2 (codant pour des ADN polymérases translésionnelles), 3 recA, 2 ddrO, 4 gènes codants HU (protéine associée au nucléoide)) (voir Chapitre III). Plusieurs de ces gènes ont

été étudiés chez D. deserti. Huit gènes ont été caractérisés (irrE, recAC, recAP1, recAP3, polB, imuY, dnaE2, DeideP302150) alors que d’autres ne sont qu’au début de leur caractérisation. Les informations que nous possédons sur certains de ces gènes seront toutefois discutées dans cette partie. Pour pouvoir étudier des gènes chez D. deserti, des outils génétiques (plasmide réplicatif et délétion de gènes) ont été mis au point avec succès (Chapitre IV). Ce travail important fait de D. deserti la 2ième Deinococcus décrite pour laquelle un système d’inactivation de gènes est développé. Pour l’instant, seulement 2 cassettes de sélection (chloramphénicol et kanamycine) sont disponibles pour D. deserti. L’obtention de souches délétées pour plus de 2 gènes (à des loci différents) dépendra de la possibilité de (1) construire de nouvelles cassettes fonctionnelles chez D. deserti ou (2) de construire des mutants de délétion où la cassette de sélection est excisable. Cette 2ième solution semble être la plus séduisante (si réalisable) car elle permet, avec une seule cassette de sélection, de déléter plusieurs gènes. Nous avons essayé de mettre au point cette technique en utilisant rpsL comme gène de contre sélection, mais malheureusement cela n’a pas fonctionné (Chapitre IV). L’utilisation de cette technique de contre sélection pourrait être rendue possible par le développement d’autres cassettes de contre sélection. Nous pourrions peut-être utiliser sacB (code pour la levansucrase qui convertit le sucrose en un polymère de fructose de haut poids moléculaire, toxique pour la cellule), ccdB (un gène codant un puissant inhibiteur de gyrase, à utiliser sous le contrôle d’un promoteur inductible) ou encore lacY (code pour la perméase permettant l’entrée de lactose et rend les bactéries sensibles au t-o-nitrophenyl-β-D- galactopyranoside) (Reyrat et al., 1998). Ainsi, pour obtenir un mutant, il suffirait de sélectionner tout d’abord des transformants sur milieu sélectif (antibiotique) puis de contre- sélectionner dans les conditions adéquates. Après contre sélection, les bactéries seront redevenues sauvages (le plasmide s’excisant au niveau du site ayant subit la première recombinaison) ou seront délétées du gène choisi car elles auront réalisé la seconde recombinaison homologue (le plasmide étant perdu grâce à cette deuxième recombinaison).

170 Chez D. deserti et D. radiodurans, la radio-induction de plusieurs gènes qui possèdent le RDRM en amont de leurs régions codantes est dépendante du régulateur IrrE (Chapitre V). La structure cristallographique d’IrrE de D. deserti a été résolue (Chapitre VI). IrrE possède un domaine N-terminal Zinc peptidase-like, un domaine central HTH (souvent impliqué dans l’interaction avec l’ADN) et un domaine C-terminal senseur-like de type GAF. L’étude de mutations ponctuelles introduites dans ces 3 domaines a montré que les domaines Zinc peptidase-like et senseur-like de type GAF sont très importants pour le fonctionnement d’IrrE (Chapitre VI). L’étude structure/fonction d’IrrE a permis une avancée dans la compréhension du fonctionnement de ce régulateur mais n’a pas permis de répondre à toutes les questions concernant cette protéine. Après irradiation quel est le signal qui active IrrE ? Il est possible qu’une petite molécule soit reconnue par le domaine GAF. De la même manière, le mode d’action d’IrrE pour induire ses gènes cibles reste inconnu. Le domaine HTH d’IrrE ne semblant pas se fixer à l’ADN double brin et donc pas directement au motif RDRM, peut-être se fixe-t-il à de l’ADN simple brin ou, comme cela a été suggéré pour RodZ (Gerdes, 2009), interagit-il avec une ou des protéines ? Enfin, la cible du domaine Zinc peptidase-like n’est pas connue. L’utilisation d’une protéine tagguée fonctionnelle (ou d’anticorps anti-IrrE) pourrait permettre de (co-)purifier la protéine IrrE et ses partenaires éventuels chez D. deserti et/ou de voir si IrrE est modifiée après irradiation. Il a été proposé que DdrO soit le régulateur transcriptionnel des gènes présentant un RDRM (Makarova et al., 2007). DdrO pourrait être le lien entre IrrE et les gènes radio- induits. Chez D. radiodurans, ddrO présente lui-même le RDRM et est radio-induit. De façon

étonnante, D. deserti présente 2 homologues de DdrO, un codé par le chromosome (DdrOC) et un par le plasmide P3 (DdrOP3) (84 % d’identité entre eux). Ces protéines présentent une très forte similarité avec DdrO de D. radiodurans (95 et 85 % d’identité respective ; Fig. 50). Les protéines DdrO comportent un domaine N-terminal de type HTH, et un domaine C-terminal

DdrOC DdrO (D. rad.) DdrO (D. geo.) DdrOP3

DdrOC DdrO (D. rad.) DdrO (D. geo.) DdrOP3

DdrOC DdrO (D. rad.) DdrO (D. geo.) DdrOP3

Figure 50 : Alignement des séquences des protéines DdrO des Deinococcus. En rouge, le domaine HTH prédit. D. geo. = D. geothermalis ; D. rad. = D. radiodurans. Les astérisques (*) indiquent les résidus conservés et les points (. :) indiquent les résidus similaires.

171 spécifique aux deinococci. Le RDRM a été identifié en amont de ddrOC, mais pas en amont de ddrOP3. La caractérisation des DdrO de D. deserti a été entreprise. Des RT-PCR indiquent que ddrOC est radio-induit de manière IrrE-dépendante, et que ddrOP3 est exprimé faiblement comparé à ddrOC et non radio-induit. Les simples mutants ΔddrOC et ΔddrOP3 ont pu être construits. Par contre, le double mutant n’a pas pu être obtenu pour l’instant, suggérant que DdrO est une protéine essentielle. Alors que l’induction de recA et d’autres gènes est affectée dans un mutant ΔirrE, celle-ci est toujours observée dans le mutant ΔddrOC. Ceci suggère que

DdrOC n’est pas l’intermédiaire entre IrrE et ces gènes radio-induits, mais nous ne pouvons pas exclure que DdrOP3 et DdrOC soient impliquées toutes les deux dans l’expression de ces gènes. Cependant, des expériences d’irradiation indiquent que les simples mutants ddrO sont légèrement moins radiotolérants que la souche sauvage. Des études supplémentaires devront

être menées afin de comprendre le rôle de DdrOC et DdrOP3. D. deserti a la particularité de posséder des ADN polymérases translésionnelles fonctionnelles (ImuY et DnaE2) impliquées dans la mutagenèse induite aux UV, ainsi que 2 protéines RecA différentes (RecAC et RecAP) qui sont fonctionnelles pour la réparation de l’ADN après irradiation (Chapitre V). Ainsi, l’hypothèse selon laquelle D. radiodurans ne possède pas de polymérases translésionnelles infidèles car ce serait imprudent en cas de nombreuses lésions de l’ADN n’est pas valable pour toutes les Deinococcus. Il est intéressant de noter que Kineococcus radiotolerans, une bactérie radiotolérante, possède également un gène dnaE2 et aussi 2 homologues de PolIV, une autre polymérase translésionnelle (Erill et al., 2006 ; Bagwell et al., 2008). Ainsi, D. deserti n’est pas la seule bactérie radiotolérante à posséder des polymérases translésionnelles (putatives). Vu leur activité mutagène, les polymérases translésionnelles de D. deserti doivent être bien régulées. Nous avons montré que RecAC et non RecAP est nécessaire à l’induction de l’opéron lexAP1-imuY-dnaE2 et donc

à la mutagenèse induite aux UV. Cela est dû à une meilleure activité co-protéase de RecAC par rapport à RecAP. Une des régions différenciant RecAC de RecAP se situe au niveau des résidus 240-255 de RecAC. La région correspondante de RecA d’E. coli est impliquée dans la fixation et le clivage de LexA (répresseur transcriptionnel). La présence de plusieurs gènes recA est extrêmement rare et n’est connue que chez 3 autres espèces, Bacillus megaterium (2 recA), Myxococcus xanthus (2 recA) et Acaryochloris marina (7 recA). L’alignement des séquences des 2 RecA de B. megaterium, des 2 RecA de M. xanthus ou des 7 RecA d’A. marina ne montre pas de telles différences au niveau de la région correspondant aux résidus 240-255 de RecAC (données non montrées). Des résultats préliminaires indiquent que d’autres gènes sont impliqués dans la mutagenèse. Alors que le mutant ΔddrOP3 conserve l’induction de la mutagenèse aux UV, cette induction est absente dans le mutant ΔddrOC.

172 Dans ce dernier, imuY est toujours induit après irradiation. Par contre, une surexpression de

DeideP101280 a été observée dans le mutant ΔddrOC. DeideP101280, qui ne présente pas d’homologue chez les 2 autres Deinococcus séquencées, code pour une hélicase de la famille RecQ. Des expériences supplémentaires seront nécessaires afin de comprendre le rôle de

DdrOC et DeideP101280 dans la mutagenèse de D. deserti. Comme évoqué précédemment, en plus des gènes ddrO, l’analyse de quelques autres gènes/protéines identifiés par des approches globales a commencé. L’opéron putatif Deide18730-Deide18690, absent chez D. radiodurans et D. geothermalis, en est un exemple. La présence du RDRM en amont de cet opéron constitué de cinq gènes, codant pour des protéines de fonction inconnue, nous a interpelés. Nous avons montré par des expériences de RT-PCR qu’au moins le gène Deide18730 était induit après irradiation et que cette induction était IrrE dépendante. Quel est donc le rôle de cet opéron putatif dans la réponse au stress chez D. deserti ? La construction et la caractérisation des souches mutantes, dans lesquelles les gènes de cet opéron sont délétés, sont en cours.

D’autre part, nos résultats montrent (pour IrrE et RecAC) ou indiquent (pour DdrO) que ces protéines ont un rôle dans la régulation de gènes impliqués dans la radiotolérance et la réparation de l’ADN. En comparant avec la souche sauvage, il serait très intéressant d’étudier l’expression globale des gènes et/ou protéines dans différents mutants afin de trouver d’autres gènes candidats régulés par IrrE, DdrOC, DdrOP3 et RecAC chez D. deserti. La comparaison du génome de D. deserti avec ceux de D. radiodurans et D. geothermalis a montré que ces trois bactéries possèdent des points communs liés à la radiotolérance (ex : gènes de réparation et régulon de réponse conservé). Cependant, chaque espèce présente des gènes spécifiques qui ont un rôle (ou un rôle supposé) dans la radiotolérance ou la réparation de l’ADN (ex : DR_1262, irrI pour D. radiodurans (voir aussi le Tableau 13) ; plusieurs gènes recA, des polymérases translésionnelles pour D. deserti). La présence de ces gènes spécifiques s’explique peut-être par des spécificités de l’environnement naturel où vit chacune de ces espèces. Par exemple, dans un environnement aride et pauvre en nutriments comme le Sahara, les ADN polymérases translésionnelles de D. deserti sont probablement importants pour tolérer des dommages d’ADN difficiles à réparer.

173 2) Les autres bactéries radiotolérantes

Les bactéries du genre Deinococcus sont les bactéries radiotolérantes les plus étudiées, mais quelques autres bactéries radiotolérantes sont connues (Tableau 12). Ainsi, y a-t-il des points communs entre les Deinococcus et les autres bactéries radiotolérantes ? Ces bactéries étant relativement dispersées dans l’arbre phylogénétique (Fig. 51), cela suggère que soit (1) la radiotolérance est un vestige de l’ancêtre commun à l’ensemble de ces bactéries, qui a été conservé chez des bactéries ayant besoin de ce phénotype (cela signifie également que la majorité des descendants de cet ancêtre a perdu ce phénotype et que les bactéries

Tableau 12: Exemple d’espèces de bactéries radiotolérantes

*La D10 est la dose tuant 90 % des bactéries. Adapté de Cox et Battista (2005). radiotolérantes possèdent un mécanisme de radiotolérance similaire), soit (2) qu’il y a eu émergence de la radiotolérance dans plusieurs espèces (par convergence évolutive et/ou transfert horizontal de gènes). Les génomes de K. radiotolerans (Bagwell et al., 2008) et de Methylobacterium radiotolerans (non publié) ont été séquencés, mais aucun ne comporte d’homologue à ddrB, ddrC, ddrD, ddrO ou irrE par exemple, suggérant qu’il y a eu convergence évolutive. Cependant, certains points communs existent entre certaines bactéries radiotolérantes. En effet, des bactéries radiotolérantes ont un ratio Mn/Fe élevé (deinococci, Enterococcus faecium, K. radiotolerans) (Daly, 2009) et, même si son implication dans la radiotolérance n’est que supposé, les deinococci et Rubrobacter radiotolerans ont un nucléoide hyper-compact (Zimmerman & Battista, 2005). Il est possible que les différentes espèces bactériennes radiotolérantes possèdent chacune des particularités qui leur permettent d’être radiotolérantes mais qu’une partie d’entre elles possède aussi des points communs. Très récemment, les génomes de Meiothermus ruber et de Meiothermus silvanus (appartiennent à la famille des Thermaceae et au phylum Deinococcus/Thermus) sont parus dans les banques de séquences. Ces bactéries ne sont pas connues dans la littérature pour être radiotolérantes, mais certaines espèces de Meiothermus pourraient l’être (Masurat et al.,

174 2005). Les 2 Meiothermus séquencées comportent les homologues (lointains pour certains) des protéines radio-induites DdrA, DdrE, DdrH, DdrI, DdrN et DdrO, et du régulateur IrrE de Deinococcus (Tableau 13). Ainsi, il semblerait que l’ancêtre commun aux Deinococcus, Thermus et Meiothermus possédait déjà des homologues à ces protéines. Les Deinococcus et les Meiothermus auraient gardé ces gènes alors que les Thermus les auraient perdus. Le séquençage du génome de Truepera radiovictrix, qui appartient également au phylum

Figure 51: Arbre phylogénétique basé sur la séquence du gène codant l’ARNr 16S des principales lignées de bactéries. Les branches en rouges sont celles des taxons des bactéries radiotolérantes. L’échelle représente 10 substitutions de base sur 100 nucléotides. D’après Cox et Battista (2005).

Deinococcus/Thermus et qui est radiotolérante, permettrait de voir si cette bactérie possède également des gènes de radiotolérance communs aux Deinococcus. Par ailleurs, si M. silvanus et M. ruber sont vraiment radiotolérantes, la comparaison des génomes de D. radiodurans, D. geothermalis, D. deserti, M. ruber et M. silvanus permettrait de trouver un set de gènes conservés entre ces 5 espèces, ce qui pourrait réduire le nombre de gènes communs (à ces 5 espèces) impliqués dans la radiotolérance. Plusieurs études ont montré que des bactéries radiosensibles, comme E. coli, pouvaient devenir plus radiotolérantes si elles étaient soumises à des cycles d’irradiation/récupération, en utilisant des doses permettant 1% de survie pour chaque cycle (Battista et al., 1999 ; Harris et al., 2009 ; Levanduski & Jaczynski, 2008). Une étude très récente a montré que des E. coli devenues très tolérantes aux radiations gamma (comparable à la tolérance de D. radiodurans) présentaient entre 40 et 71 mutations. Ces mutations touchent différents gènes selon les souches (ex : recA, ruvB, clpP, clpX, dnaT (ce dernier est impliqué dans le redémarrage de la réplication)), suggérant que différents mécanismes peuvent conduire à la radiotolérance (Harris et al., 2009). Contrairement aux Deinococcus, ces souches d’E. coli ne présentent pas de nucléoide hyper-compact ou de ratio élevé de Mn/Fe. De plus, elles présentent une aussi forte oxydation des protéines après irradiation gamma que la souche d’E. coli de départ (Harris et al., 2009). Ainsi, il semblerait que la radiotolérance de ces bactéries pourrait, en partie, être expliquée par une modification de leurs systèmes de réparation de l’ADN afin de les rendre plus efficace. Évidemment ces bactéries n’ont pas acquis d’homologues aux gènes ddrA, ddrB, pprA ou irrE par exemple, ce qui suggère encore

175 une fois qu’il existe plusieurs voies pour devenir résistant aux rayons gamma. Toutefois, ces bactéries devenues très tolérantes aux radiations gamma ne sont pas devenues aussi tolérantes aux UV ou à la mitomycine C que D. radiodurans (Harris et al., 2009). Cela signifie qu’elles ont surtout évolué pour résister au stress auquel elles ont été soumises, les rayons gamma.

Tableaux 13: Quelques protéines liées à la radiotolérance et uniquement présentes chez les Deinococcus et les Meiothermus D. deserti D. radiodurans D. geothermalis M. ruber M. silvanus Description des protéines Deide01160 (158 aa) DR_0326 (198 aa) ; Dgeo_2186 (160 aa) - - DdrD 56 % ; 71 % Deide02930 (180 aa) DR_2441 (170 aa) ; Dgeo_0078 (168 aa) MrubDRAFT_19870 MesilDRAFT_13840 DdrN (a) 62 % ; 78 % (146 aa) ; 42 % (162 aa) ; 45 % Deide02990 (178 aa) DR_0070 (199 aa) Dgeo_0295 (178 aa) - - DdrB 73 % 82 % Deide03030 (281 aa) DR_0167 (328 aa) ; Dgeo_0395 (289 aa) MrubDRAFT_08400 MesilDRAFT_22520 IrrE 65 % ; 79 % (241 aa) ; 33 % (235 aa) ; 34 % Deide09150 (201 aa) DR_0423 (208 aa) ; Dgeo_0977 (201 aa) MrubDRAFT_24800 MesilDRAFT_00510 DdrA (a) 78 % ; 80 % (211 aa) ; 35 % (225 aa) ; 43 % Deide11220 (228 aa) DR_0194 (227 aa) ; Dgeo_1282 (226 aa) MrubDRAFT_13570 MesilDRAFT_35620 DdrE (a) ; faiblement 73 % ; 73 % (217 aa) ; 47 % (227 aa) ; 48 % homologue à une métallopeptidase à zinc Deide12530 (205 aa) DR_0997 (260 aa) ; Dgeo_1015 (204 aa) MrubDRAFT_22520 MesilDRAFT_01810 DdrI (a) ; régulateur 96 % ; 91 % (202 aa) ; 52 % (202 aa) ; 54 % transcriptionnel de la famile Crp/Fnr Deide20570 (129 aa) DR_2574 (131 aa) ; Dgeo_0336 (140 aa) MrubDRAFT_08440 MesilDRAFT_22540 DdrO DeideP302170 95 % ; 93 % (130 aa) ; 70 % (130 aa) ; 70 % (129 aa) ; 84 % Deide20641 (82 aa) DR_0438 reversed Dgeo_0322 (82 aa) ; MrubDRAFT_06610 MesilDRAFT_34870 DdrH (92 aa) ; 72 % 76 % (74 aa) ; 32 % (90 aa) ; 28 % Deide23280 (232 aa) DR_0003 (reversed Dgeo_0047 reversed - - DdrC (231 aa) ; 69 % (229 aa) ; 76 % DeideP201380 DR_A0346 (300 aa) ; Dgeo_2628 (302 aa) - - PprA (294 aa) 53 % (plasmide) ; 63 % - DR_0171 (230 aa) - - - IrrI ; impliqué dans la radiotolérance de D. radiodurans - DR_0219 (444 aa) - - - DdrF - DR_0227 (242 aa) - - - DdrG - DR_1262 (531 aa) - - - Faiblement homoloque à la ribonucleoprotéine Ro/SS-A ; impliqué dans la résistance UV de D. radiodurans - DR_1263 (222 aa) - - - DdrJ - DR_1264 (140 aa) - - - DdrK - DR_1439 (175 aa) - - - DdrL - DR_B0100 (210 aa) - - - DdrP - DR_B0141 (168 aa) - - - Faiblement homologue à HicB (a) Pas uniquement présent chez ces 5 bactéries. Les pourcentages d’identité sont donnés en fonction des protéines de D. deserti. Pour DdrOP3, 84 % est le pourcentage d’identité par rapport à DdrOC. Les gènes codant DdrA, DdrB, DdrC, DdrD, DdrE, DdrF, DdrG, DdrH, DdrI, DdrJ, DdrK, DdrL, DdrN, DdrO, DdrP et PprA sont radio-induits chez D. radiodurans.

Il est intéressant de noter que quelques organismes, autres que des bactéries, sont résistants aux rayons gamma. C’est le cas par exemple de l’archeae Thermococcus gammatolerans (Zivanovic et al., 2009), du champignon Ustilago maydis (Holloman et al., 2007) ou de l’animal microscopique Adineta vaga (Phylum Rotifera, classe Bdelloidea) (Gladyshev & Meselson, 2008). Chez ces organismes comme chez les Deinococcus, cette radiotolérance est probablement le résultat de l’adaptation à des conditions extrêmes qui endommagent l’ADN. Cependant, chez ces organismes, les mécanismes permettant cette radiotolérance ne sont pas connus.

176

VIII Bibliographie

177

178 VIII) Bibliographie

Abella, M., S. Campoy, I. Erill, F. Rojo & J. Barbe, (2007) Cohabitation of two different lexA regulons in Pseudomonas putida. J Bacteriol 189: 8855-8862. Agostini, H. J., J. D. Carroll & K. W. Minton, (1996) Identification and characterization of uvrA, a DNA repair gene of Deinococcus radiodurans. J Bacteriol 178: 6759-6765. Airo, A., S. L. Chan, Z. Martinez, M. O. Platt & J. D. Trent, (2004) Heat shock and cold shock in Deinococcus radiodurans. Cell Biochem Biophys 40: 277-288. Akerlund, T., K. Nordstrom & R. Bernander, (1995) Analysis of cell size and DNA content in exponentially growing and stationary-phase batch cultures of Escherichia coli. J Bacteriol 177: 6791-6797. Al Mamun, A. A., (2007) Elevated expression of DNA polymerase II increases spontaneous mutagenesis in Escherichia coli. Mutat Res 625: 29-39. Albuquerque, L., C. Simoes, M. F. Nobre, N. M. Pino, J. R. Battista, M. T. Silva, F. A. Rainey & M. S. da Costa, (2005) Truepera radiovictrix gen. nov., sp. nov., a new radiation resistant species and the proposal of Trueperaceae fam. nov. FEMS Microbiol Lett 247: 161-169. Anderson, A., et al., (1956) Studies on a radio-resistant micrococcus. Isolation, morphology, cultural characteristics, and resistance to gamma radiation. Food Technology 10: 575- 582. Archibald, F. S. & I. Fridovich, (1981) Manganese and defenses against oxygen toxicity in Lactobacillus plantarum. J Bacteriol 145: 442-451. Asker, D., T. S. Awad, T. Beppu & K. Ueda, (2008) Deinococcus misasensis and Deinococcus roseus, novel members of the genus Deinococcus, isolated from a radioactive site in Japan. Syst Appl Microbiol 31: 43-49. Asker, D., T. S. Awad, T. Beppu & K. Ueda, (2009) Deinococcus aquiradiocola sp. nov., isolated from a radioactive site in Japan. Int J Syst Evol Microbiol 59: 144-149. Azam, T. A. & A. Ishihama, (1999) Twelve species of the nucleoid-associated protein from Escherichia coli. Sequence recognition specificity and DNA binding affinity. J Biol Chem 274: 33105-33113. Bagwell, C. E., S. Bhat, G. M. Hawkins, B. W. Smith, T. Biswas, T. R. Hoover, E. Saunders, C. S. Han, O. V. Tsodikov & L. J. Shimkets, (2008) Survival in nuclear waste, extreme resistance, and potential applications gleaned from the genome sequence of Kineococcus radiotolerans SRS30216. PLoS One 3: e3878. Banos, B., J. M. Lazaro, L. Villar, M. Salas & M. de Vega, (2008a) Characterization of a Bacillus subtilis 64-kDa DNA polymerase X potentially involved in DNA repair. J Mol Biol 384: 1019-1028. Banos, B., J. M. Lazaro, L. Villar, M. Salas & M. de Vega, (2008b) Editing of misaligned 3'- termini by an intrinsic 3'-5' exonuclease activity residing in the PHP domain of a family X DNA polymerase. Nucleic Acids Res 36: 5736-5749. Battista, J. R., A. M. Earl & M. J. Park, (1999) Why is Deinococcus radiodurans so resistant to ionizing radiation? Trends Microbiol 7: 362-365. Battista, J. R., M. J. Park & A. E. McLemore, (2001) Inactivation of two homologues of proteins presumed to be involved in the desiccation tolerance of plants sensitizes Deinococcus radiodurans R1 to desiccation. Cryobiology 43: 133-139. Beam, C. E., C. J. Saveson & S. T. Lovett, (2002) Role for radA/sms in recombination intermediate processing in Escherichia coli. J Bacteriol 184: 6836-6844. Bentchikou, E., P. Servant, G. Coste & S. Sommer, (2007) Additive effects of SbcCD and PolX deficiencies in the in vivo repair of DNA double-strand breaks in Deinococcus radiodurans. J Bacteriol 189: 4784-4790.

179 Bernstein, D. A., J. M. Eggington, M. P. Killoran, A. M. Misic, M. M. Cox & J. L. Keck, (2004) Crystal structure of the Deinococcus radiodurans single-stranded DNA- binding protein suggests a mechanism for coping with DNA damage. Proc Natl Acad Sci U S A 101: 8575-8580. Billi, D. & M. Potts, (2002) Life and death of dried prokaryotes. Res Microbiol 153: 7-12. Blasius, M., R. Buob, I. V. Shevelev & U. Hubscher, (2007) Enzymes involved in DNA ligation and end-healing in the radioresistant bacterium Deinococcus radiodurans. BMC Mol Biol 8: 69. Blasius, M., I. Shevelev, E. Jolivet, S. Sommer & U. Hubscher, (2006) DNA polymerase X from Deinococcus radiodurans possesses a structure-modulated 3'-->5' exonuclease activity involved in radioresistance. Mol Microbiol 60: 165-176. Blasius, M., S. Sommer & U. Hubscher, (2008) Deinococcus radiodurans: what belongs to the survival kit? Crit Rev Biochem Mol Biol 43: 221-238. Boling, M. E. & J. K. Setlow, (1966) The resistance of Micrococcus radiodurans to ultraviolet radiation. 3. A repair mechanism. Biochim Biophys Acta 123: 26-33. Bonacossa de Almeida, C., G. Coste, S. Sommer & A. Bailone, (2002) Quantification of RecA protein in Deinococcus radiodurans reveals involvement of RecA, but not LexA, in its regulation. Mol Genet Genomics 268: 28-41. Boshoff, H. I., M. B. Reed, C. E. Barry, 3rd & V. Mizrahi, (2003) DnaE2 polymerase contributes to in vivo survival and the emergence of drug resistance in Mycobacterium tuberculosis. Cell 113: 183-193. Bouthier de la Tour, C., M. Toueille, E. Jolivet, H. H. Nguyen, P. Servant, F. Vannier & S. Sommer, (2009) The Deinococcus radiodurans SMC protein is dispensable for cell viability yet plays a role in DNA folding. Extremophiles. Bradshaw, J. S. & A. Kuzminov, (2003) RdgB acts to avoid chromosome fragmentation in Escherichia coli. Mol Microbiol 48: 1711-1725. Brooks, M. R. G. E. B. W., (1981) Nomenclature for “Micrococcus radiodurans” and Other Radiation-Resistant Cocci: Deinococcaceae fam. nov. and Deinococcus gen. nov., Including Five Species Int. J. Syst. Bacteriol. 31: 353-360. Brosh, R. M., Jr. & S. W. Matson, (1997) A point mutation in Escherichia coli DNA helicase II renders the enzyme nonfunctional in two DNA repair pathways. Evidence for initiation of unwinding from a nick in vivo. J Biol Chem 272: 572-579. Butala, M., D. Zgur-Bertok & S. J. Busby, (2009) The bacterial LexA transcriptional repressor. Cell Mol Life Sci 66: 82-93. Callegan, R. P., M. F. Nobre, P. M. McTernan, J. R. Battista, R. Navarro-Gonzalez, C. P. McKay, M. S. da Costa & F. A. Rainey, (2008) Description of four novel psychrophilic, ionizing radiation-sensitive Deinococcus species from alpine environments. Int J Syst Evol Microbiol 58: 1252-1258. Caramelo, J. J. & N. D. Iusem, (2009) When cells lose water: Lessons from biophysics and molecular biology. Prog Biophys Mol Biol 99: 1-6. Carbonneau, M. A., A. M. Melin, A. Perromat & M. Clerc, (1989) The action of free radicals on Deinococcus radiodurans carotenoids. Arch Biochem Biophys 275: 244-251. Carroll, J. D., M. J. Daly & K. W. Minton, (1996) Expression of recA in Deinococcus radiodurans. J Bacteriol 178: 130-135. Clark, A. J., (1991) rec genes and homologous recombination proteins in Escherichia coli. Biochimie 73: 523-532. Courcelle, J., A. Khodursky, B. Peter, P. O. Brown & P. C. Hanawalt, (2001) Comparative gene expression profiles following UV exposure in wild-type and SOS-deficient Escherichia coli. Genetics 158: 41-64. Cox, M. M., (2007a) Motoring along with the bacterial RecA protein. Nat Rev Mol Cell Biol 8: 127-138.

180 Cox, M. M., (2007b) Regulation of bacterial RecA protein function. Crit Rev Biochem Mol Biol 42: 41-63. Cox, M. M. & J. R. Battista, (2005) Deinococcus radiodurans - the consummate survivor. Nat Rev Microbiol 3: 882-892. Daly, M. J., (2009) A new perspective on radiation resistance based on Deinococcus radiodurans. Nat Rev Microbiol 7: 237-245. Daly, M. J., E. K. Gaidamakova, V. Y. Matrosova, A. Vasilenko, M. Zhai, R. D. Leapman, B. Lai, B. Ravel, S. M. Li, K. M. Kemner & J. K. Fredrickson, (2007) Protein oxidation implicated as the primary determinant of bacterial radioresistance. PLoS Biol 5: e92. Daly, M. J., E. K. Gaidamakova, V. Y. Matrosova, A. Vasilenko, M. Zhai, A. Venkateswaran, M. Hess, M. V. Omelchenko, H. M. Kostandarithes, K. S. Makarova, L. P. Wackett, J. K. Fredrickson & D. Ghosal, (2004) Accumulation of Mn(II) in Deinococcus radiodurans facilitates gamma-radiation resistance. Science 306: 1025-1028. Daly, M. J., O. Ling & K. W. Minton, (1994) Interplasmidic recombination following irradiation of the radioresistant bacterium Deinococcus radiodurans. J Bacteriol 176: 7506-7515. Daly, M. J. & K. W. Minton, (1996) An alternative pathway of recombination of chromosomal fragments precedes recA-dependent recombination in the radioresistant bacterium Deinococcus radiodurans. J Bacteriol 178: 4461-4471. Dame, R. T., (2005) The role of nucleoid-associated proteins in the organization and compaction of bacterial chromatin. Mol Microbiol 56: 858-870. de Groot, A., V. Chapon, P. Servant, R. Christen, M. F. Saux, S. Sommer & T. Heulin, (2005) Deinococcus deserti sp. nov., a gamma-radiation-tolerant bacterium isolated from the Sahara Desert. Int J Syst Evol Microbiol 55: 2441-2446. De Mot, R., G. Schoofs & J. Vanderleyden, (1994) A putative regulatory gene downstream of RecA is conserved in Gram-negative and Gram-postive bacteria. Nucleic Acids Res 22: 1313-1314. Earl, A. M., M. M. Mohundro, I. S. Mian & J. R. Battista, (2002a) The IrrE protein of Deinococcus radiodurans R1 is a novel regulator of recA expression. J Bacteriol 184: 6216-6224. Earl, A. M., S. K. Rankin, K. P. Kim, O. N. Lamendola & J. R. Battista, (2002b) Genetic evidence that the uvsE gene product of Deinococcus radiodurans R1 is a UV damage endonuclease. J Bacteriol 184: 1003-1009. Eggington, J. M., N. Haruta, E. A. Wood & M. M. Cox, (2004) The single-stranded DNA- binding protein of Deinococcus radiodurans. BMC Microbiol 4: 2. Eggington, J. M., A. G. Kozlov, M. M. Cox & T. M. Lohman, (2006) Polar destabilization of DNA duplexes with single-stranded overhangs by the Deinococcus radiodurans SSB protein. Biochemistry 45: 14490-14502. Eltsov, M. & J. Dubochet, (2005) Fine structure of the Deinococcus radiodurans nucleoid revealed by cryoelectron microscopy of vitreous sections. J Bacteriol 187: 8047-8054. Eltsov, M. & J. Dubochet, (2006) Study of the Deinococcus radiodurans nucleoid by cryoelectron microscopy of vitreous sections: Supplementary comments. J Bacteriol 188: 6053-6058; discussion 6059. Enderle, P. J. & M. A. Farwell, (1998) Electroporation of freshly plated Escherichia coli and Pseudomonas aeruginosa cells. Biotechniques 25: 954-956, 958. Englander, J., E. Klein, V. Brumfeld, A. K. Sharma, A. J. Doherty & A. Minsky, (2004) DNA toroids: framework for DNA repair in Deinococcus radiodurans and in germinating bacterial spores. J Bacteriol 186: 5973-5977. Erill, I., S. Campoy, G. Mazon & J. Barbe, (2006) Dispersal and regulation of an adaptive mutagenesis cassette in the bacteria domain. Nucleic Acids Res 34: 66-77.

181 Fernandez De Henestrosa, A. R., T. Ogi, S. Aoyagi, D. Chafin, J. J. Hayes, H. Ohmori & R. Woodgate, (2000) Identification of additional genes belonging to the LexA regulon in Escherichia coli. Mol Microbiol 35: 1560-1572. Ferreira, A. C., M. F. Nobre, F. A. Rainey, M. T. Silva, R. Wait, J. Burghardt, A. P. Chung & M. S. da Costa, (1997) Deinococcus geothermalis sp. nov. and Deinococcus murrayi sp. nov., two extremely radiation-resistant and slightly thermophilic species from hot springs. Int J Syst Bacteriol 47: 939-947. Figurski, D. H. & D. R. Helinski, (1979) Replication of an origin-containing derivative of plasmid RK2 dependent on a plasmid function provided in trans. Proc Natl Acad Sci U S A 76: 1648-1652. Fowler, R. G. & R. M. Schaaper, (1997) The role of the mutT gene of Escherichia coli in maintaining replication fidelity. FEMS Microbiol Rev 21: 43-54. Fredrickson, J. K., S. M. Li, E. K. Gaidamakova, V. Y. Matrosova, M. Zhai, H. M. Sulloway, J. C. Scholten, M. G. Brown, D. L. Balkwill & M. J. Daly, (2008) Protein oxidation: key to bacterial desiccation resistance? ISME J 2: 393-403. Friedberg, E. C., (2006) DNA repair and mutagenesis. ASM press Second edition. Fuchs, R. P., N. Koffel-Schwartz, S. Pelet, R. Janel-Bintz, R. Napolitano, O. J. Becherel, T. H. Broschard, D. Y. Burnouf & J. Wagner, (2001) DNA polymerases II and V mediate respectively mutagenic (-2 frameshift) and error-free bypass of a single N-2- acetylaminofluorene adduct. Biochem Soc Trans 29: 191-195. Galhardo, R. S., R. P. Rocha, M. V. Marques & C. F. Menck, (2005) An SOS-regulated operon involved in damage-inducible mutagenesis in Caulobacter crescentus. Nucleic Acids Res 33: 2603-2614. Galio, L., C. Bouquet & P. Brooks, (1999) ATP hydrolysis-dependent formation of a dynamic ternary nucleoprotein complex with MutS and MutL. Nucleic Acids Res 27: 2325- 2331. Gao, G., D. Le, L. Huang, H. Lu, I. Narumi & Y. Hua, (2006) Internal promoter characterization and expression of the Deinococcus radiodurans pprI-folP gene cluster. FEMS Microbiol Lett 257: 195-201. Gao, G., B. Tian, L. Liu, D. Sheng, B. Shen & Y. Hua, (2003) Expression of Deinococcus radiodurans PprI enhances the radioresistance of Escherichia coli. DNA Repair (Amst) 2: 1419-1427. Gerdes, K., (2009) RodZ, a new player in bacterial cell morphogenesis. EMBO J 28: 171-172. Ghosal, D., M. V. Omelchenko, E. K. Gaidamakova, V. Y. Matrosova, A. Vasilenko, A. Venkateswaran, M. Zhai, H. M. Kostandarithes, H. Brim, K. S. Makarova, L. P. Wackett, J. K. Fredrickson & M. J. Daly, (2005) How radiation kills cells: survival of Deinococcus radiodurans and Shewanella oneidensis under oxidative stress. FEMS Microbiol Rev 29: 361-375. Gladyshev, E. & M. Meselson, (2008) Extreme resistance of bdelloid rotifers to ionizing radiation. Proc Natl Acad Sci U S A 105: 5139-5144. Goosen, N. & G. F. Moolenaar, (2008) Repair of UV damage in bacteria. DNA Repair (Amst) 7: 353-379. Gutman, P. D., J. D. Carroll, C. I. Masters & K. W. Minton, (1994) Sequencing, targeted mutagenesis and expression of a recA gene required for the extreme radioresistance of Deinococcus radiodurans. Gene 141: 31-37. Gutman, P. D., P. Fuchs, L. Ouyang & K. W. Minton, (1993) Identification, sequencing, and targeted mutagenesis of a DNA polymerase gene required for the extreme radioresistance of Deinococcus radiodurans. J Bacteriol 175: 3581-3590. Gutsche, I., A. Vujicic-Zagar, X. Siebert, P. Servant, F. Vannier, B. Castaing, B. Gallet, T. Heulin, A. de Groot, S. Sommer & L. Serre, (2008) Complex oligomeric structure of a truncated form of DdrA: a protein required for the extreme radiotolerance of Deinococcus. Biochim Biophys Acta 1784: 1050-1058.

182 Handa, N., K. Morimatsu, S. T. Lovett & S. C. Kowalczykowski, (2009) Reconstitution of initial steps of dsDNA break repair by the RecF pathway of E. coli. Genes Dev 23: 1234-1245. Hansen, M. T., (1978) Multiplicity of genome equivalents in the radiation-resistant bacterium Micrococcus radiodurans. J Bacteriol 134: 71-75. Hansen, M. T., (1980) Four proteins synthesized in response to deoxyribonucleic acid damage in Micrococcus radiodurans. J Bacteriol 141: 81-86. Hariharan, P. V. & P. A. Cerutti, (1972) Formation and repair of gamma-ray induced thymine damage in Micrococcus radiodurans. J Mol Biol 66: 65-81. Harris, D. R., K. V. Ngo & M. M. Cox, (2008) The stable, functional core of DdrA from Deinococcus radiodurans R1 does not restore radioresistance in vivo. J Bacteriol 190: 6475-6482. Harris, D. R., S. V. Pollock, E. A. Wood, R. J. Goiffon, A. J. Klingele, E. L. Cabot, W. Schackwitz, J. Martin, J. Eggington, T. J. Durfee, C. M. Middle, J. E. Norton, M. C. Popelars, H. Li, S. A. Klugman, L. L. Hamilton, L. B. Bane, L. A. Pennacchio, T. J. Albert, N. T. Perna, M. M. Cox & J. R. Battista, (2009) Directed evolution of ionizing radiation resistance in Escherichia coli. J Bacteriol 191: 5240-5252. Harris, D. R., M. Tanaka, S. V. Saveliev, E. Jolivet, A. M. Earl, M. M. Cox & J. R. Battista, (2004) Preserving genome integrity: the DdrA protein of Deinococcus radiodurans R1. PLoS Biol 2: e304. Harsojo, S. Kitayama & A. Matsuyama, (1981) Genome multiplicity and radiation resistance in Micrococcus radiodurans. J Biochem 90: 877-880. Hastings, J. W., W. H. Holzapfel & J. G. Niemand, (1986) Radiation resistance of lactobacilli isolated from radurized meat relative to growth and environment. Appl Environ Microbiol 52: 898-901. Henne, A., H. Bruggemann, C. Raasch, A. Wiezer, T. Hartsch, H. Liesegang, A. Johann, T. Lienard, O. Gohl, R. Martinez-Arias, C. Jacobi, V. Starkuviene, S. Schlenczeck, S. Dencker, R. Huber, H. P. Klenk, W. Kramer, R. Merkl, G. Gottschalk & H. J. Fritz, (2004) The genome sequence of the extreme thermophile Thermus thermophilus. Nat Biotechnol 22: 547-553. Heulin, T., M. Barakat, R. Christen, M. Lesourd, L. Sutra, G. De Luca & W. Achouak, (2003) Ramlibacter tataouinensis gen. nov., sp. nov., and Ramlibacter henchirensis sp. nov., cyst-producing bacteria isolated from subdesert soil in Tunisia. Int J Syst Evol Microbiol 53: 589-594. Hirsch, P., C. A. Gallikowski, J. Siebert, K. Peissl, R. Kroppenstedt, P. Schumann, E. Stackebrandt & R. Anderson, (2004) Deinococcus frigens sp. nov., Deinococcus saxicola sp. nov., and Deinococcus marmoris sp. nov., low temperature and draught- tolerating, UV-resistant bacteria from continental Antarctica. Syst Appl Microbiol 27: 636-645. Holloman, W. K., J. Schirawski & R. Holliday, (2007) Towards understanding the extreme radiation resistance of Ustilago maydis. Trends Microbiol 15: 525-529. Houben, K., D. Marion, N. Tarbouriech, R. W. Ruigrok & L. Blanchard, (2007) Interaction of the C-terminal domains of sendai virus N and P proteins: comparison of polymerase- nucleocapsid interactions within the paramyxovirus family. J Virol 81: 6807-6816. Hua, X., L. Huang, B. Tian & Y. Hua, (2008) Involvement of recQ in the ultraviolet damage repair pathway in Deinococcus radiodurans. Mutat Res 641: 48-53. Hua, Y., I. Narumi, G. Gao, B. Tian, K. Satoh, S. Kitayama & B. Shen, (2003) PprI: a general switch responsible for extreme radioresistance of Deinococcus radiodurans. Biochem Biophys Res Commun 306: 354-360. Huang, L., X. Hua, H. Lu, G. Gao, B. Tian, B. Shen & Y. Hua, (2007) Three tandem HRDC domains have synergistic effect on the RecQ functions in Deinococcus radiodurans. DNA Repair (Amst) 6: 167-176.

183 Huang, L. F., X. T. Hua, H. M. Lu, G. J. Gao, B. Tian, B. H. Shen & Y. J. Hua, (2006a) Functional analysis of helicase and three tandem HRDC domains of RecQ in Deinococcus radiodurans. J Zhejiang Univ Sci B 7: 373-376. Huang, L. F., S. W. Zhang, X. T. Hua, G. J. Gao & Y. J. Hua, (2006b) [Construction of the recQ double mutants and analysis of adversity in Deinococcus radiodurans]. Wei Sheng Wu Xue Bao 46: 205-209. Im, W. T., H. M. Jung, L. N. Ten, M. K. Kim, N. Bora, M. Goodfellow, S. Lim, J. Jung & S. T. Lee, (2008) Deinococcus aquaticus sp. nov., isolated from fresh water, and Deinococcus caeni sp. nov., isolated from activated sludge. Int J Syst Evol Microbiol 58: 2348-2353. Jolivet, E., F. Lecointe, G. Coste, K. Satoh, I. Narumi, A. Bailone & S. Sommer, (2006) Limited concentration of RecA delays DNA double-strand break repair in Deinococcus radiodurans R1. Mol Microbiol 59: 338-349. Joshi, B., R. Schmid, K. Altendorf & S. K. Apte, (2004) Protein recycling is a major component of post-irradiation recovery in Deinococcus radiodurans strain R1. Biochem Biophys Res Commun 320: 1112-1117. Kampfer, P., (2009) Deinococcus mumbaiensis Shashidhar and Bandekar 2006 is a later heterotypic synonym of Deinococcus ficus Lai et al. 2006. Int J Syst Evol Microbiol 59: 365-366. Kampfer, P., N. Lodders, B. Huber, E. Falsen & H. J. Busse, (2008) Deinococcus aquatilis sp. nov., isolated from water. Int J Syst Evol Microbiol 58: 2803-2806. Khairnar, N. P., V. A. Kamble & H. S. Misra, (2008) RecBC enzyme overproduction affects UV and gamma radiation survival of Deinococcus radiodurans. DNA Repair (Amst) 7: 40-47. Khairnar, N. P. & H. S. Misra, (2009) DNA polymerase X from Deinococcus radiodurans implicated in bacterial tolerance to DNA damage is characterized as a short patch base excision repair polymerase. Microbiology. Killoran, M. P. & J. L. Keck, (2006) Three HRDC domains differentially modulate Deinococcus radiodurans RecQ DNA helicase biochemical activity. J Biol Chem 281: 12849-12857. Killoran, M. P. & J. L. Keck, (2008) Structure and function of the regulatory C-terminal HRDC domain from Deinococcus radiodurans RecQ. Nucleic Acids Res 36: 3139- 3149. Kim, J. I. & M. M. Cox, (2002) The RecA proteins of Deinococcus radiodurans and Escherichia coli promote DNA strand exchange via inverse pathways. Proc Natl Acad Sci U S A 99: 7917-7921. Kim, J. I., A. K. Sharma, S. N. Abbott, E. A. Wood, D. W. Dwyer, A. Jambura, K. W. Minton, R. B. Inman, M. J. Daly & M. M. Cox, (2002) RecA Protein from the extremely radioresistant bacterium Deinococcus radiodurans: expression, purification, and characterization. J Bacteriol 184: 1649-1660. Kitayama, S., I. Narumi, T. Funayama & H. Watanabe, (2003) Cloning of structural gene of Deinococcus radiodurans UV-endonuclease beta. Biosci Biotechnol Biochem 67: 613- 616. Kobayashi, Y., I. Narumi, K. Satoh, T. Funayama, M. Kikuchi, S. Kitayama & H. Watanabe, (2004) Radiation response mechanisms of the extremely radioresistant bacterium Deinococcus radiodurans. Biol Sci Space 18: 134-135. Koorits, L., R. Tegova, M. Tark, K. Tarassova, A. Tover & M. Kivisaar, (2007) Study of involvement of ImuB and DnaE2 in stationary-phase mutagenesis in Pseudomonas putida. DNA Repair (Amst) 6: 863-868. Kota, S. & H. S. Misra, (2006) PprA: A protein implicated in radioresistance of Deinococcus radiodurans stimulates catalase activity in Escherichia coli. Appl Microbiol Biotechnol 72: 790-796.

184 Kovach, M. E., R. W. Phillips, P. H. Elzer, R. M. Roop, 2nd & K. M. Peterson, (1994) pBBR1MCS: a broad-host-range cloning vector. Biotechniques 16: 800-802. Kowalczykowski, S. C., J. Clow, R. Somani & A. Varghese, (1987) Effects of the Escherichia coli SSB protein on the binding of Escherichia coli RecA protein to single-stranded DNA. Demonstration of competitive binding and the lack of a specific protein-protein interaction. J Mol Biol 193: 81-95. Kowalczykowski, S. C., (2000) Initiation of genetic recombination and recombination- dependent replication. Trends Biochem Sci 25, 156-165. Lahue, R. S., S. S. Su & P. Modrich, (1987) Requirement for d(GATC) sequences in Escherichia coli mutHLS mismatch correction. Proc Natl Acad Sci U S A 84: 1482- 1486. Lai, W. A., P. Kampfer, A. B. Arun, F. T. Shen, B. Huber, P. D. Rekha & C. C. Young, (2006) Deinococcus ficus sp. nov., isolated from the rhizosphere of Ficus religiosa L. Int J Syst Evol Microbiol 56: 787-791. Langle-Rouault, F., G. Maenhaut-Michel & M. Radman, (1987) GATC sequences, DNA nicks and the MutH function in Escherichia coli mismatch repair. EMBO J 6: 1121- 1127. Lecointe, F., G. Coste, S. Sommer & A. Bailone, (2004a) Vectors for regulated gene expression in the radioresistant bacterium Deinococcus radiodurans. Gene 336: 25-35. Lecointe, F., I. V. Shevelev, A. Bailone, S. Sommer & U. Hubscher, (2004b) Involvement of an X family DNA polymerase in double-stranded break repair in the radioresistant organism Deinococcus radiodurans. Mol Microbiol 53: 1721-1730. Leibowitz, P. J., L. S. Schwartzberg & A. K. Bruce, (1976) The in vivo association of manganese with the chromosome of Micrococcus radiodurans. Photochem Photobiol 23: 45-50. Levanduski, L. & J. Jaczynski, (2008) Increased resistance of Escherichia coli O157:H7 to electron beam following repetitive irradiation at sub-lethal doses. Int J Food Microbiol 121: 328-334. Levin-Zaidman, S., J. Englander, E. Shimoni, A. K. Sharma, K. W. Minton & A. Minsky, (2003) Ringlike structure of the Deinococcus radiodurans genome: a key to radioresistance? Science 299: 254-256. Lindahl, T., (1976) New class of enzymes acting on damaged DNA. Nature 259: 64-66. Lipton, M. S., L. Pasa-Tolic, G. A. Anderson, D. J. Anderson, D. L. Auberry, J. R. Battista, M. J. Daly, J. Fredrickson, K. K. Hixson, H. Kostandarithes, C. Masselon, L. M. Markillie, R. J. Moore, M. F. Romine, Y. Shen, E. Stritmatter, N. Tolic, H. R. Udseth, A. Venkateswaran, K. K. Wong, R. Zhao & R. D. Smith, (2002) Global analysis of the Deinococcus radiodurans proteome by using accurate mass tags. Proc Natl Acad Sci U S A 99: 11049-11054. Liu, Y., J. Zhou, M. V. Omelchenko, A. S. Beliaev, A. Venkateswaran, J. Stair, L. Wu, D. K. Thompson, D. Xu, I. B. Rogozin, E. K. Gaidamakova, M. Zhai, K. S. Makarova, E. V. Koonin & M. J. Daly, (2003) Transcriptome dynamics of Deinococcus radiodurans recovering from ionizing radiation. Proc Natl Acad Sci U S A 100: 4191-4196. Lu, D. & J. L. Keck, (2008) Structural basis of Escherichia coli single-stranded DNA-binding protein stimulation of exonuclease I. Proc Natl Acad Sci U S A 105: 9169-9174. Lu, H., G. Gao, G. Xu, L. Fan, L. Yin, B. Shen & Y. Hua, (2009) Deinococcus radiodurans PprI switches on DNA damage-response and cellular survival networks after radiation damage. Mol Cell Proteomics. Lucarelli, D., Y. A. Wang, V. E. Galkin, X. Yu, D. B. Wigley & E. H. Egelman, (2009) The RecB Domain Binds to RecA-DNA Filaments: Implications for Filament Loading. J Mol Biol.

185 Makarova, K. S., L. Aravind, M. J. Daly & E. V. Koonin, (2000) Specific expansion of protein families in the radioresistant bacterium Deinococcus radiodurans. Genetica 108: 25-34. Makarova, K. S., L. Aravind, Y. I. Wolf, R. L. Tatusov, K. W. Minton, E. V. Koonin & M. J. Daly, (2001) Genome of the extremely radiation-resistant bacterium Deinococcus radiodurans viewed from the perspective of comparative genomics. Microbiol Mol Biol Rev 65: 44-79. Makarova, K. S., M. V. Omelchenko, E. K. Gaidamakova, V. Y. Matrosova, A. Vasilenko, M. Zhai, A. Lapidus, A. Copeland, E. Kim, M. Land, K. Mavrommatis, S. Pitluck, P. M. Richardson, C. Detter, T. Brettin, E. Saunders, B. Lai, B. Ravel, K. M. Kemner, Y. I. Wolf, A. Sorokin, A. V. Gerasimova, M. S. Gelfand, J. K. Fredrickson, E. V. Koonin & M. J. Daly, (2007) Deinococcus geothermalis: the pool of extreme radiation resistance genes shrinks. PLoS ONE 2: e955. Maldonado, R., J. Jimenez & J. Casadesus, (1994) Changes of ploidy during the Azotobacter vinelandii growth cycle. J Bacteriol 176: 3911-3919. Margison, G. P., A. Butt, S. J. Pearson, S. Wharton, A. J. Watson, A. Marriott, C. M. Caetano, J. J. Hollins, N. Rukazenkova, G. Begum & M. F. Santibanez-Koref, (2007) Alkyltransferase-like proteins. DNA Repair (Amst) 6: 1222-1228. Markillie, L. M., S. M. Varnum, P. Hradecky & K. K. Wong, (1999) Targeted mutagenesis by duplication insertion in the radioresistant bacterium Deinococcus radiodurans: radiation sensitivities of catalase (katA) and superoxide dismutase (sodA) mutants. J Bacteriol 181: 666-669. Masurat, P., E. C. Fru & K. Pedersen, (2005) Identification of Meiothermus as the dominant genus in a storage system for spent nuclear fuel. J Appl Microbiol 98: 727-740. Mattimore, V. & J. R. Battista, (1996) Radioresistance of Deinococcus radiodurans: functions necessary to survive ionizing radiation are also necessary to survive prolonged desiccation. J Bacteriol 178: 633-637. Mattimore, V., K. S. Udupa, G. A. Berne & J. R. Battista, (1995) Genetic characterization of forty ionizing radiation-sensitive strains of Deinococcus radiodurans: linkage information from transformation. J Bacteriol 177: 5232-5237. McGrew, D. A. & K. L. Knight, (2003) Molecular design and functional organization of the RecA protein. Crit Rev Biochem Mol Biol 38: 385-432. McLennan, A. G., (2006) The Nudix hydrolase superfamily. Cell Mol Life Sci 63: 123-143. Meima, R. & M. E. Lidstrom, (2000) Characterization of the minimal replicon of a cryptic Deinococcus radiodurans SARK plasmid and development of versatile Escherichia coli-D. radiodurans shuttle vectors. Appl Environ Microbiol 66: 3856-3867. Mennecier, S., G. Coste, P. Servant, A. Bailone & S. Sommer, (2004) Mismatch repair ensures fidelity of replication and recombination in the radioresistant organism Deinococcus radiodurans. Mol Genet Genomics 272: 460-469. Mennecier, S., P. Servant, G. Coste, A. Bailone & S. Sommer, (2006) Mutagenesis via IS transposition in Deinococcus radiodurans. Mol Microbiol 59: 317-325. Minton, K. W., (1994) DNA repair in the extremely radioresistant bacterium Deinococcus radiodurans. Mol Microbiol 13: 9-15. Minton, K. W. & M. J. Daly, (1995) A model for repair of radiation-induced DNA double- strand breaks in the extreme radiophile Deinococcus radiodurans. Bioessays 17: 457- 464. Misra, H. S., N. P. Khairnar, S. Kota, S. Shrivastava, V. P. Joshi & S. K. Apte, (2006) An exonuclease I-sensitive DNA repair pathway in Deinococcus radiodurans: a major determinant of radiation resistance. Mol Microbiol 59: 1308-1316. Moe, E., I. Leiros, A. O. Smalas & S. McSweeney, (2006) The crystal structure of mismatch- specific uracil-DNA glycosylase (MUG) from Deinococcus radiodurans reveals a novel catalytic residue and broad substrate specificity. J Biol Chem 281: 569-577.

186 Moeller, R., E. Stackebrandt, G. Reitz, T. Berger, P. Rettberg, A. J. Doherty, G. Horneck & W. L. Nicholson, (2007) Role of DNA repair by nonhomologous-end joining in Bacillus subtilis spore resistance to extreme dryness, mono- and polychromatic UV, and ionizing radiation. J Bacteriol 189: 3306-3311. Mortimer, R. K., (1958) Radiobiological and genetic studies on a polyploid series (haploid to hexaploid) of Saccharomyces cerevisiae. Radiat Res 9: 312-326. Moseley, B. E. & H. F. Copland, (1978) Four mutants of Micrococcus radiodurans defective in the ability to repair DNA damaged by mitomycin-C, two of which have wild-type resistance to ultraviolet radiation. Mol Gen Genet 160: 331-337. Moseley, B. E. & D. M. Evans, (1983) Isolation and properties of strains of Micrococcus (Deinococcus) radiodurans unable to excise ultraviolet light-induced pyrimidine dimers from DNA: evidence for two excision pathways. J Gen Microbiol 129: 2437- 2445. Murakami, M., I. Narumi, K. Satoh, A. Furukawa & I. Hayata, (2006) Analysis of interaction between DNA and Deinococcus radiodurans PprA protein by atomic force microscopy. Biochim Biophys Acta 1764: 20-23. Narumi, I., K. Satoh, S. Cui, T. Funayama, S. Kitayama & H. Watanabe, (2004) PprA: a novel protein from Deinococcus radiodurans that stimulates DNA ligation. Mol Microbiol 54: 278-285. Narumi, I., K. Satoh, M. Kikuchi, T. Funayama, S. Kitayama, T. Yanagisawa, H. Watanabe & K. Yamamoto, (1999) Molecular analysis of the Deinococcus radiodurans recA locus and identification of a mutation site in a DNA repair-deficient mutant, rec30. Mutat Res 435: 233-243. Narumi, I., K. Satoh, M. Kikuchi, T. Funayama, T. Yanagisawa, Y. Kobayashi, H. Watanabe & K. Yamamoto, (2001) The LexA protein from Deinococcus radiodurans is not involved in RecA induction following gamma irradiation. J Bacteriol 183: 6951-6956. Nguyen, H. H., C. B. de la Tour, M. Toueille, F. Vannier, S. Sommer & P. Servant, (2009) The essential histone-like protein HU plays a major role in Deinococcus radiodurans nucleoid compaction. Mol Microbiol 73: 240-252. Nieminuszczy, J. & E. Grzesiuk, (2007) Bacterial DNA repair genes and their eukaryotic homologues: 3. AlkB dioxygenase and Ada methyltransferase in the direct repair of alkylated DNA. Acta Biochim Pol 54: 459-468. Norais, C. A., S. Chitteni-Pattu, E. A. Wood, R. B. Inman & M. M. Cox, (2009) An alternative Deinococcus radiodurans SSB induced by ionizing radiation: The DdrB protein. J Biol Chem. Ohba, H., K. Satoh, H. Sghaier, T. Yanagisawa & I. Narumi, (2009) Identification of PprM: a modulator of the PprI-dependent DNA damage response in Deinococcus radiodurans. Extremophiles 13: 471-479. Ohba, H., K. Satoh, T. Yanagisawa & I. Narumi, (2005) The radiation responsive promoter of the Deinococcus radiodurans pprA gene. Gene 363: 133-141. Omelchenko, M. V., Y. I. Wolf, E. K. Gaidamakova, V. Y. Matrosova, A. Vasilenko, M. Zhai, M. J. Daly, E. V. Koonin & K. S. Makarova, (2005) Comparative genomics of Thermus thermophilus and Deinococcus radiodurans: divergent routes of adaptation to thermophily and radiation resistance. BMC Evol Biol 5: 57. Oyaizu, H., E. Stackebrandt, K. H. Schleifer, W. Ludwig, H. Pohla, H. Ito, A. Hirata, Y. Oyaizu & K. Komagata, (1987) A Radiation-Resistant Rod-Shaped Bacterium, Deinobacter grandis gen. nov., sp. nov., with Peptidoglycan Containing Ornithine Int. J. Syst. Bacteriol. 37: 62-67. Parker, B. O. & M. G. Marinus, (1992) Repair of DNA heteroduplexes containing small heterologous sequences in Escherichia coli. Proc Natl Acad Sci U S A 89: 1730-1734.

187 Pavlov A.K. , K. V. L., Konstantinov A.N., Shelegedin V.N., Pavlov A.A., (2006) Was Earth Ever Infected by Martian Biota? Clues from Radioresistant Bacteria. Astrobiology 6: 911-918. Peng, F., L. Zhang, X. Luo, J. Dai, H. An, Y. Tang & C. Fang, (2009) Deinococcus xinjiangensis sp. nov., isolated from desert soil. Int J Syst Evol Microbiol 59: 709-713. Piatkowski, D., K. Schneider, F. Salamini & D. Bartels, (1990) Characterization of Five Abscisic Acid-Responsive cDNA Clones Isolated from the Desiccation-Tolerant Plant Craterostigma plantagineum and Their Relationship to Other Water-Stress Genes. Plant Physiol 94: 1682-1688. Pogoda de la Vega, U., P. Rettberg, T. Douki, J. Cadet & G. Horneck, (2005) Sensitivity to polychromatic UV-radiation of strains of Deinococcus radiodurans differing in their DNA repair capacity. Int J Radiat Biol 81: 601-611. Potts, M., (1994) Desiccation tolerance of prokaryotes. Microbiol Rev 58: 755-805. Rainey, F. A., M. Ferreira, M. F. Nobre, K. Ray, D. Bagaley, A. M. Earl, J. R. Battista, B. Gomez-Silva, C. P. McKay & M. S. da Costa, (2007) Deinococcus peraridilitoris sp. nov., isolated from a coastal desert. Int J Syst Evol Microbiol 57: 1408-1412. Rainey, F. A., M. F. Nobre, P. Schumann, E. Stackebrandt & M. S. da Costa, (1997) Phylogenetic diversity of the deinococci as determined by 16S ribosomal DNA sequence comparison. Int J Syst Bacteriol 47: 510-514. Rainey, F. A., K. Ray, M. Ferreira, B. Z. Gatz, M. F. Nobre, D. Bagaley, B. A. Rash, M. J. Park, A. M. Earl, N. C. Shank, A. M. Small, M. C. Henk, J. R. Battista, P. Kampfer & M. S. da Costa, (2005) Extensive diversity of ionizing-radiation-resistant bacteria recovered from Sonoran Desert soil and description of nine new species of the genus Deinococcus obtained from a single soil sample. Appl Environ Microbiol 71: 5225- 5235. Rajpurohit, Y. S., R. Gopalakrishnan & H. S. Misra, (2008) Involvement of a protein kinase activity inducer in DNA double strand break repair and radioresistance of Deinococcus radiodurans. J Bacteriol 190: 3948-3954. Rebeil, R., Y. Sun, L. Chooback, M. Pedraza-Reyes, C. Kinsland, T. P. Begley & W. L. Nicholson, (1998) Spore photoproduct lyase from Bacillus subtilis spores is a novel iron-sulfur DNA repair enzyme which shares features with proteins such as class III anaerobic ribonucleotide reductases and pyruvate-formate lyases. J Bacteriol 180: 4879-4885. Renzette, N., N. Gumlaw & S. J. Sandler, (2007) DinI and RecX modulate RecA-DNA structures in Escherichia coli K-12. Mol Microbiol 63: 103-115. Reyrat, J. M., V. Pelicic, B. Gicquel & R. Rappuoli, (1998) Counterselectable markers: untapped tools for bacterial genetics and pathogenesis. Infect Immun 66: 4011-4017. Robertson, A. B., S. R. Pattishall, E. A. Gibbons & S. W. Matson, (2006) MutL-catalyzed ATP hydrolysis is required at a post-UvrD loading step in methyl-directed mismatch repair. J Biol Chem 281: 19949-19959. Rocha, E. P., E. Cornet & B. Michel, (2005) Comparative and evolutionary analysis of the bacterial homologous recombination systems. PLoS Genet 1: e15. Sakai, A. & M. M. Cox, (2009) RecFOR and RecOR as distinct RecA loading pathways. J Biol Chem 284: 3264-3272. Sale, J. E., (2007) Radiation resistance: resurrection by recombination. Curr Biol 17: R12-14. Sancar, A., (2008) Structure and function of photolyase, and in vivo enzymology - 50th anniversary. J Biol Chem. Satoh, K., I. Narumi, M. Kikuchi, S. Kitayama, T. Yanagisawa, K. Yamamoto & H. Watanabe, (2002) Characterization of RecA424 and RecA670 proteins from Deinococcus radiodurans. J Biochem 131: 121-129. Satoh, K., H. Ohba, H. Sghaier & I. Narumi, (2006) Down-regulation of radioresistance by LexA2 in Deinococcus radiodurans. Microbiology 152: 3217-3226.

188 Satoh, K., Z. Tu, H. Ohba & I. Narumi, (2009) Development of versatile shuttle vectors for Deinococcus grandis. Plasmid 62: 1-9. Schlacher, K. & M. F. Goodman, (2007) Lessons from 50 years of SOS DNA-damage- induced mutagenesis. Nat Rev Mol Cell Biol 8: 587-594. Schlesinger, D. J., (2007) Role of RecA in DNA damage repair in Deinococcus radiodurans. FEMS Microbiol Lett 274: 342-347. Schmid, A. K., H. A. Howell, J. R. Battista, S. N. Peterson & M. E. Lidstrom, (2005) Global transcriptional and proteomic analysis of the Sig1 heat shock regulon of Deinococcus radiodurans. J Bacteriol 187: 3339-3351. Servant, P., E. Jolivet, E. Bentchikou, S. Mennecier, A. Bailone & S. Sommer, (2007) The ClpPX protease is required for radioresistance and regulates cell division after gamma-irradiation in Deinococcus radiodurans. Mol Microbiol 66: 1231-1239. Servinsky, M. D. & D. A. Julin, (2007) Effect of a recD mutation on DNA damage resistance and transformation in Deinococcus radiodurans. J Bacteriol 189: 5101-5107. Shashidhar, R. & J. R. Bandekar, (2006) Deinococcus mumbaiensis sp. nov., a radiation- resistant pleomorphic bacterium isolated from Mumbai, India. FEMS Microbiol Lett 254: 275-280. Shashidhar, R. & J. R. Bandekar, (2009) Deinococcus piscis sp. nov., a radiation-resistant bacterium isolated from marine fish. Int J Syst Evol Microbiol. Sheng, D., G. Gao, B. Tian, Z. Xu, Z. Zheng & Y. Hua, (2005a) RecX is involved in antioxidant mechanisms of the radioresistant bacterium Deinococcus radiodurans. FEMS Microbiol Lett 244: 251-257. Sheng, D., J. Jao, M. Li, P. Xu & J. Zhang, (2009) RecX is involved in the switch between DNA damage response and normal metabolism in D. radiodurans. J Biochem. Sheng, D., R. Liu, Z. Xu, P. Singh, B. Shen & Y. Hua, (2005b) Dual negative regulatory mechanisms of RecX on RecA functions in radiation resistance, DNA recombination and consequent genome instability in Deinococcus radiodurans. DNA Repair (Amst) 4: 671-678. Sheng, D., Z. Zheng, B. Tian, B. Shen & Y. Hua, (2004) LexA analog (dra0074) is a regulatory protein that is irrelevant to recA induction. J Biochem 136: 787-793. Shereda, R. D., A. G. Kozlov, T. M. Lohman, M. M. Cox & J. L. Keck, (2008) SSB as an organizer/mobilizer of genome maintenance complexes. Crit Rev Biochem Mol Biol 43: 289-318. Slade, D., A. B. Lindner, G. Paul & M. Radman, (2009) Recombination and replication in DNA repair of heavily irradiated Deinococcus radiodurans. Cell 136: 1044-1055. Smith, M. D., E. Lennon, L. B. McNeil & K. W. Minton, (1988) Duplication insertion of drug resistance determinants in the radioresistant bacterium Deinococcus radiodurans. J Bacteriol 170: 2126-2135. Socolofsky, M. D. & O. Wyss, (1962) Resistance of the Azotobacter cyst. J Bacteriol 84: 119-124. Suresh, K., G. S. Reddy, S. Sengupta & S. Shivaji, (2004) Deinococcus indicus sp. nov., an arsenic-resistant bacterium from an aquifer in West Bengal, India. Int J Syst Evol Microbiol 54: 457-461. Tanaka, A., H. Hirano, M. Kikuchi, S. Kitayama & H. Watanabe, (1996) Changes in cellular proteins of Deinococcus radiodurans following gamma-irradiation. Radiat Environ Biophys 35: 95-99. Tanaka, M., A. M. Earl, H. A. Howell, M. J. Park, J. A. Eisen, S. N. Peterson & J. R. Battista, (2004) Analysis of Deinococcus radiodurans's transcriptional response to ionizing radiation and desiccation reveals novel proteins that contribute to extreme radioresistance. Genetics 168: 21-33. Tanaka, M., I. Narumi, T. Funayama, M. Kikuchi, H. Watanabe, T. Matsunaga, O. Nikaido & K. Yamamoto, (2005) Characterization of pathways dependent on the uvsE, uvrA1, or

189 uvrA2 gene product for UV resistance in Deinococcus radiodurans. J Bacteriol 187: 3693-3697. Thanbichler, M., S. C. Wang & L. Shapiro, (2005) The bacterial nucleoid: a highly organized and dynamic structure. J Cell Biochem 96: 506-521. Thomas, A. T., W. J. Brammar & B. M. Wilkins, (2003) Plasmid R16 ArdA protein preferentially targets restriction activity of the type I restriction-modification system EcoKI. J Bacteriol 185: 2022-2025. Tian, B., Z. Xu, Z. Sun, J. Lin & Y. Hua, (2007) Evaluation of the antioxidant effects of carotenoids from Deinococcus radiodurans through targeted mutagenesis, chemiluminescence, and DNA damage analyses. Biochim Biophys Acta 1770: 902- 911. Tippin, B., P. Pham & M. F. Goodman, (2004) Error-prone replication for better or worse. Trends Microbiol 12: 288-295. Udupa, K. S., P. A. O'Cain, V. Mattimore & J. R. Battista, (1994) Novel ionizing radiation- sensitive mutants of Deinococcus radiodurans. J Bacteriol 176: 7439-7446. Veaute, X., S. Delmas, M. Selva, J. Jeusset, E. Le Cam, I. Matic, F. Fabre & M. A. Petit, (2005) UvrD helicase, unlike Rep helicase, dismantles RecA nucleoprotein filaments in Escherichia coli. EMBO J 24: 180-189. Wade, J. T., N. B. Reppas, G. M. Church & K. Struhl, (2005) Genomic analysis of LexA binding reveals the permissive nature of the Escherichia coli genome and identifies unconventional target sites. Genes Dev 19: 2619-2630. Wang, J. & D. A. Julin, (2004) DNA helicase activity of the RecD protein from Deinococcus radiodurans. J Biol Chem 279: 52024-52032. Wang, L., G. Xu, H. Chen, Y. Zhao, N. Xu, B. Tian & Y. Hua, (2008) DrRRA: a novel response regulator essential for the extreme radioresistance of Deinococcus radiodurans. Mol Microbiol 67: 1211-1222. Wang, W., J. Mao, Z. Zhang, Q. Tang, Y. Xie, J. Zhu, L. Zhang, Z. Liu, Y. Shi & M. Goodfellow, (2009) Deinococcus wulumuqiensis sp. nov., and Deinococcus xibeiensis sp. nov., isolated from radiation-polluted soil. Int J Syst Evol Microbiol. Weller, G. R., B. Kysela, R. Roy, L. M. Tonkin, E. Scanlan, M. Della, S. K. Devine, J. P. Day, A. Wilkinson, F. d'Adda di Fagagna, K. M. Devine, R. P. Bowater, P. A. Jeggo, S. P. Jackson & A. J. Doherty, (2002) Identification of a DNA nonhomologous end- joining complex in bacteria. Science 297: 1686-1689. Welsh, K. M., A. L. Lu, S. Clark & P. Modrich, (1987) Isolation and characterization of the Escherichia coli mutH gene product. J Biol Chem 262: 15624-15629. Weon, H. Y., B. Y. Kim, P. Schumann, J. A. Son, J. Jang, S. J. Go & S. W. Kwon, (2007) Deinococcus cellulosilyticus sp. nov., isolated from air. Int J Syst Evol Microbiol 57: 1685-1688. White, O., J. A. Eisen, J. F. Heidelberg, E. K. Hickey, J. D. Peterson, R. J. Dodson, D. H. Haft, M. L. Gwinn, W. C. Nelson, D. L. Richardson, K. S. Moffat, H. Qin, L. Jiang, W. Pamphile, M. Crosby, M. Shen, J. J. Vamathevan, P. Lam, L. McDonald, T. Utterback, C. Zalewski, K. S. Makarova, L. Aravind, M. J. Daly, K. W. Minton, R. D. Fleischmann, K. A. Ketchum, K. E. Nelson, S. Salzberg, H. O. Smith, J. C. Venter & C. M. Fraser, (1999) Genome sequence of the radioresistant bacterium Deinococcus radiodurans R1. Science 286: 1571-1577. Williams, R. A., K. E. Smith, S. G. Welch, J. Micallef & R. J. Sharp, (1995) DNA relatedness of Thermus strains, description of Thermus brockianus sp. nov., and proposal to reestablish Thermus thermophilus (Oshima and Imahori). Int J Syst Bacteriol 45: 495- 499. Xu, G., L. Wang, H. Chen, H. Lu, N. Ying, B. Tian & Y. Hua, (2008) RecO is essential for DNA damage repair in Deinococcus radiodurans. J Bacteriol 190: 2624-2628.

190 Xu, W., J. Shen, C. A. Dunn, S. Desai & M. J. Bessman, (2001) The Nudix hydrolases of Deinococcus radiodurans. Mol Microbiol 39: 286-290. Xu, Z., B. Tian, Z. Sun, J. Lin & Y. Hua, (2007) Identification and functional analysis of a phytoene desaturase gene from the extremely radioresistant bacterium Deinococcus radiodurans. Microbiology 153: 1642-1652. Yang, Y., T. Itoh, S. Yokobori, S. Itahashi, H. Shimada, K. Satoh, H. Ohba, I. Narumi & A. Yamagishi, (2009a) Deinococcus aerius sp. nov., isolated from the high atmosphere. Int J Syst Evol Microbiol. Yang, Y., T. Itoh, S. I. Yokobori, H. Shimada, S. Itahashi, K. Satoh, H. Ohba, I. Narumi & A. Yamagishi, (2009b) Deinococcus aetherius sp. nov., isolated from the stratosphere. Int J Syst Evol Microbiol. Yano, K., K. Morotomi-Yano, N. Adachi & H. Akiyama, (2009) Molecular mechanism of protein assembly on DNA double-strand breaks in the non-homologous end-joining pathway. J Radiat Res (Tokyo) 50: 97-108. Yasui, A. & S. J. McCready, (1998) Alternative repair pathways for UV-induced DNA damage. Bioessays 20: 291-297. Yeiser, B., E. D. Pepper, M. F. Goodman & S. E. Finkel, (2002) SOS-induced DNA polymerases enhance long-term survival and evolutionary fitness. Proc Natl Acad Sci U S A 99: 8737-8741. Yuan, M., W. Zhang, S. Dai, J. Wu, Y. Wang, T. Tao, M. Chen & M. Lin, (2009) Deinococcus gobiensis sp. nov., an extremely radiation-resistant bacterium. Int J Syst Evol Microbiol 59: 1513-1517. Zahradka, K., D. Slade, A. Bailone, S. Sommer, D. Averbeck, M. Petranovic, A. B. Lindner & M. Radman, (2006) Reassembly of shattered chromosomes in Deinococcus radiodurans. Nature 443: 569-573. Zhang, C., J. Wei, Z. Zheng, N. Ying, D. Sheng & Y. Hua, (2005) Proteomic analysis of Deinococcus radiodurans recovering from gamma-irradiation. Proteomics 5: 138-143. Zhang, L., Q. Yang, X. Luo, C. Fang, Q. Zhang & Y. Tang, (2007a) Knockout of crtB or crtI gene blocks the carotenoid biosynthetic pathway in Deinococcus radiodurans R1 and influences its resistance to oxidative DNA-damaging agents due to change of free radicals scavenging ability. Arch Microbiol 188: 411-419. Zhang, Y. Q., C. H. Sun, W. J. Li, L. Y. Yu, J. Q. Zhou, L. H. Xu & C. L. Jiang, (2007b) Deinococcus yunweiensis sp. nov., a gamma- and UV-radiation-resistant bacterium from China. Int J Syst Evol Microbiol 57: 370-375. Zhou, Q., X. Zhang, H. Xu, B. Xu & Y. Hua, (2007) A new role of Deinococcus radiodurans RecD in antioxidant pathway. FEMS Microbiol Lett 271: 118-125. Zimmerman, J. M. & J. R. Battista, (2005) A ring-like nucleoid is not necessary for radioresistance in the Deinococcaceae. BMC Microbiol 5: 17. Zivanovic, Y., J. Armengaud, A. Lagorce, C. Leplat, P. Guerin, M. Dutertre, V. Anthouard, P. Forterre, P. Wincker & F. Confalonieri, (2009) Genome analysis and genome-wide proteomics of Thermococcus gammatolerans, the most radioresistant organism known amongst the Archaea. Genome Biol 10: R70.

191

192

IX Les annexes

193

194 IX) Annexes

Les annexes 1 et 2 proviennent respectivement des tableaux S10 (les gènes de réparation de l’ADN chez D. deserti, D. radiodurans et D. geothermalis) et S8 (les gènes impliqués dans la réponse aux stress chez D. deserti, D. radiodurans et D. geothermalis) de l’article « Alliance of proteomics and genomics to unravel the specificities of Sahara bacterium Deinococcus deserti ».

195

196

Annexe 1. DNA repair genes in D. deserti, D. radiodurans and D. geothermalis

A. Main DNA repair genes for Base Excision Repair (BER) a Gene name D. deserti D. radiodurans D. geothermalis Putative function of the predicted gene product

Monofunctional DNA glycosylases alkA (282 aa) Deide_08050 (204 aa) DR_2074 (190 aa) Dgeo_1660 (209 aa) 3-methyladenine-DNA glycosylase II Deide_02320 (216 aa) DR_2584 (225 aa) Dgeo_0107 (216 aa) tag (187 aa) - - - 3-methyladenine-DNA glycosylase I mutY (350 aa) Deide_01970 (353aa) DR_2285 (363 aa) Dgeo_0019 (343 aa) A/8oxoG adenine glycosylase mug (168 aa) Deide_17530 (185 aa) DR_0715 (199 aa) Dgeo_1718 (195 aa) Removes uracil, thymine or ethenocytosine opposite guanine - - Dgeo_2568 (177 aa) (plasmid) ung or udg (229 aa) Deide_00830 (248 aa) DR_0689 (247 aa) Dgeo_2059 (245 aa) Uracil-DNA glycosylase - DR_1751 (237 aa) Dgeo_1556 (230 aa) - DR_0022 (199 aa)b -

Bifunctional DNA glycosylases (displaying also a AP lyase activity) fpg (mutM) (269 aa) Deide_16240 (276 aa) DR_0493 (280 aa) Dgeo_0442 (289 aa) Formamidopyrimidine-DNA glycosylase nth (211 aa) Deide_01790 (222 aa) DR_0289 (nth1)(225 aa) Dgeo_0248 (233 aa) Endonuclease III; removes ring-saturated or Deide_23070 (247 aa) DR_2438 (nth2)(259 aa) Dgeo_2290 (242 aa) fragmented pyrimidines Deide_12880 (237 aa) DR_0928 (338 aa) Dgeo_0785 (269 aa) nei (263 aa) - - - Endonuclease VIII, DNA N-glycosylase/AP lyase

AP nfi (223 aa) - DR_2162 (181 aa) - Endonuclease V nfo (285 aa) - - - Endonuclease IV xthA (268 aa) Deide_03250 (260 aa) DR_0354 (283 aa) Dgeo_0461 (267 aa) Exodeoxyribonulease III Deide_2p01470 (270 aa) - Dgeo_2484 (254 aa) (plasmid)

B. Main DNA repair genes for Nucleotide Excision Repair (NER) a Gene name D. deserti D. radiodurans D. geothermalis Protein description mfd (1148 aa) Deide_06920 (1041 aa) DR_1532 (1054 aa) Dgeo_0545 (1041 aa) transcription-repair coupling factor; helicase uvrA (940 aa) Deide_12760 (1004 aa) DR_1771 (1016 aa) Dgeo_0694 (1004 aa) DNA damage recognition protein UvrA; DNA Deide_2p02060 (831 aa) DR_A0188 (922 aa) - independent ATPase and DNA binding protein uvrB (673 aa) Deide_03120 (671 aa) DR_2275 (730 aa) Dgeo_1890 (676 aa) DNA damage binding protein UvrB ; helicase uvrC (610 aa) Deide_11450 (617 aa) DR_1354 (617 aa) Dgeo_1124 (616 aa) Excision nuclease uvrD (720 aa) Deide_12100 (744 aa) DR_1775 (745 aa) Dgeo_0868 (741 aa) DNA helicase II BS_ywjD (320 aa) Deide_17800 (301 aa) DR_1819 (305 aa) Dgeo_1819 (311 aa) UV DNA damage endonuclease UvsE yejH (586 aa) Deide_17320 (467 aa) DR_A0131 (895 aa) - DNA or RNA helicase of superfamily II (COG1061); also predicted nuclease cho (295 aa) - - - UvrC homolog

C. Main DNA repair genes for Mismatch Repair a Gene name D. deserti D. radiodurans D. geothermalis Protein description mutL (615 aa) Deide_15600 (550 aa) DR_1696 (547 aa) Dgeo_1538 (549 aa) DNA mismatch repair protein; ATPase mutS (853 aa) Deide_15540 (mutS1) DR_1039 (mutS1) Dgeo_1537 (mutS1) DNA mismatch repair protein; ATPase (849 aa) (850 aa) (880 aa) Deide_05000 (mutS2) DR_1976 (mutS2) Dgeo_0899 (mutS2) MutS2 family protein; ATPase (767 aa) (766 aa) (789 aa) xseA (456aa) Deide_22980 (402aa) DR_0186 (416aa) Dgeo_0148 (413aa) Exonuclease VII, large subunit xseB (80 aa) Deide_02120 (87 aa)c DR_2586 (85 aa)c Dgeo_0027 (83 aa)c Exonuclease VII, small subunit dam (278 aa) - - - Methylase GATC mutH (229 aa) - - - Endonuclease GATC dcm (472 aa) - - - C-5 cytosine methyltransferase vsr (156 aa) - - - Very short patch repair protein; endonuclease

D. Main DNA repair genes for Direct Reversal of DNA damage (DR) a Gene name D. deserti D. radiodurans D. geothermalis Protein description ada (354 aa) - - - O-6-methylguanine-DNA alkyltransferase; Transcription activator ogt/ybaZ (171 aa Deide_22770 (118 aa) DR_0428 (139 aa) Dgeo_2101 (124 aa) O-6-alkylguanine transferase /129 aa) alkB (216 aa) - - - N1-methyladenine and N3-methylcytocine repair protein phrB (472 aa) - - - DNA photolyase BS_splB (342 aa) Deide_3p02150 (354 aa) - - DNA repair photolyase dut (151 aa) - - - dUTPase dcd (193 aa) Deide_16640 (186 aa) - Dgeo_1880 (186 aa) Deoxycytidine triphosphate deaminase yggV (197 aa) Deide_19360 (202 aa) DR_0179 (200 aa) Dgeo_2209 (199 aa) Xanthosine triphosphate pyrophosphatase

E. Main DNA repair genes for recombinational repair (RER) a Gene name D. deserti D. radiodurans D. geothermalis Protein description recA (353 aa) Deide_19450 (355 aa) DR_2340 (363 aa) Dgeo_2138 (358 aa) DNA strand exchange and renaturation, DNA- dependent ATPase, DNA- and ATP-dependent Deide_1p01260 (344 aa) coprotease Deide_3p00210 (344 aa) recB (1180 aa) -d - - DNA helicase, ATP-dependent dsDNA/ssDNA exonuclease V subunit, ss DNA endonuclease recC (1122 aa) - - - DNA helicase, ATP-dependent dsDNA/ssDNA exonuclease V subunit, ss DNA endonuclease recD (608 aa) Deide_16210 (710 aa) DR_1902 (715 aa) Dgeo_0826 (726 aa) DNA helicase, ATP-dependent dsDNA/ssDNA exonuclease V subunit, ss DNA endonuclease

197 recE (866 aa) - - - Exonuclease VIII, dsDNA exonuclease, 5’=>3’ specific recF (357 aa) Deide_14250 (363 aa) DR_1089 (359 aa) Dgeo_1620 (358 aa) Recombinational repair protein recG (693 aa) Deide_09960 (780 aa) DR_1916 (784 aa) Dgeo_1139 (776 aa) Holliday junction-specific DNA helicase; branch migration inducer recJ (577 aa) Deide_07130 (714 aa) DR_1126 (684 aa) Dgeo_1599 (702 aa) ssDNA exonuclease, 5’=>3’ specific recN (553 aa) Deide_12310 (534 aa) DR_1477 (564 aa) Dgeo_1194 (555 aa) Recombination and repair protein recO (242 aa) Deide_13810 (247 aa) DR_0819 (244 aa) Dgeo_0855 (243 aa) Bacterial recombinational repair protein recQ (609 aa) Deide_11320 (731 aa) DR_1289 (824 aa)e Dgeo_1226 (195 aa !) ATP-dependent DNA helicase RecQ (one HRDC (two HRDC domains) (three HRDC domains) frameshift? (one HRDC domain) domain) (no RQC domain) Deide_02180 (591 aa) DR_2444 (603 aa)e Dgeo_0021 (653 aa) Nucleic acid-binding protein, HRDC family (one HRDC domain) (one HRDC domain) (one HRDC domain) Deide_1p01280 (548 aa) - - ATP-dependent DNA helicase, RecQ family; (no RQC and HRDC partial COG0514 domains) Deide_06510_3 (1683 aa) - - Multidomain protein: 1 is DnaQ, 2 is DinG, 3 (no RQC and HRDC is RecQf domains) recR (201 aa) Deide_06340 (198 aa) DR_0198 (220 aa) Dgeo_1513 (222 aa) Recombination and repair protein recT (269 aa) - - - Recombinase, DNA renaturation recX (166 aa) Deide_12350 (195 aa) DR_1310 (204 aa) Dgeo_1433 (191 aa) Regulatory protein radA (sms) (460 aa) Deide_12660 (449 aa) DR_1105 (503 aa) Dgeo_1212 (449 aa) DNA repair protein RadA rusA (ybcP) (120 - - - Endonuclease/Holliday junction resolvase aa) ruvA (203 aa) Deide_09360 (202 aa) DR_1274 (201 aa) Dgeo_0726 (198 aa) Holliday junction helicase subunit A; branch migration ruvB (336 aa) Deide_18350 (334 aa) DR_0596 (333 aa) Dgeo_0404 (331 aa) Holliday junction helicase subunit B; branch migration ruvC (173 aa) Deide_20630 (169 aa) DR_0440 (179 aa) Dgeo_0327 (168 aa) Holliday junction endonuclease yqgF (138 aa) Deide_04280 (139 aa) DR_2509 (136 aa) Dgeo_0425 (232 aa) Putative Holliday junction resolvase sbcB (475 aa) - - - I sbcC (1048 aa) Deide_16170 (911 aa) DR_1922 (909 aa) Dgeo_0823 (910 aa) ATP dependent dsDNA exonuclease sbcD (400 aa) Deide_16180 (393 aa) DR_1921 (416 aa) Dgeo_0824 (395 aa) ATP dependent dsDNA exonuclease ssb (178 aa) Deide_00120 (297 aa) DR_0099 (301 aa) Dgeo_0165 (301 aa) Single-stranded DNA-binding protein (+ Dgeo_2964 (283 aa),

Dgeo_2969 (352 aa), Dgeo_3087 (200 aa) on plasmid)

F. Other DNA repair related genes a Gene name D. deserti D. radiodurans D. geothermalis Protein description mutT and other 18 proteins containing 23 proteins containing 14 proteins containing Nudix hydrolases Nudix hydrolases Nudix box Nudix boxg Nudix box dnlJ (671 aa) Deide_12290 (686 aa)h DR_2069 (700 aa) Dgeo_0696 (684 aa) DNA ligase, NAD-dependent Deide_1p00290 (686 aa)h gyrA (875 aa) Deide_12520 (811 aa) DR_1913 (812 aa) Dgeo_1016 (809 aa) DNA gyrase subunit A gyrB (804 aa) Deide_15490 (674 aa) DR_0906 (663 aa) Dgeo_0546 (711 aa) DNA gyrase subunit B topA (865 aa) Deide_07410 (966 aa) DR_1374 (1021 aa) Dgeo_2001 (964 aa) DNA topoisomerase I (Topo IA) - Deide_00840 (344 aa) DR_0690 (346 aa) Dgeo_2058 (349 aa) DNA topoisomerase IB radC (222 aa) - - - DNA repair protein xni (281 aa) Deide_1p00132 (244 aa) - - 5’-3’ exonuclease; exonuclease IX - - DR_1757 (328 aa) - nuclease-related domain (NERD) protein lexA (202 aa) Deide_1p01870 (211 aa) DR_A0344 (lexA1)(210 Dgeo_1366 (236 aa) Transcriptional regulator, repressor of the SOS aa) regulon, autoprotease Deide_01180 (240 aa) DR_A0074 (lexA2)(248 aa) polA (928 aa) Deide_15130 (926 aa) DR_1707 (956 aa) Dgeo_1666 (907 aa) DNA polymerase I polB (783 aa) Deide_1p00180 (765 aa) - - DNA polymerase II dnaE (1160 aa) Deide_21950 (1341 aa) DR_0507 (1335 aa) Dgeo_0255 (1322 aa) DNA polymerase III alpha subunit Deide_1p01900 (1058 aa) - - Putative error-prone DNA polymerase DnaE2 dnaQ (243 aa) Deide_17790 (202 aa) DR_0856 (197 aa) Dgeo_1818 (180 aa) DNA polymerase III epsilon subunit (3’-5’ Deide_05970 (204 aa) - - exonuclease subunit) Deide_06510_1 (1683 aa) - - multidomain proteinf: 1 is DnaQ, 2 is DinG, 3 is RecQ holE (76 aa) - - - DNA polymerase III theta subunit dnaN (366 aa) Deide_00020 (360 aa) DR_0001 (393 aa) Dgeo_0003 (360 aa) DNA polymerase III beta subunit Dgeo_3063 (367 aa)

(plasmid) dnaX (643 aa) Deide_01610 (762 aa) DR_2410 (615 aa) Dgeo_2135 (737 aa) DNA polymerase III tau/gamma subunit holA (343 aa) Deide_10170 (300 aa) DR_1244 (300 aa) Dgeo_0745 (312 aa) DNA polymerase III delta subunit holB (334 aa) Deide_21710 (318 aa) DR_2332 (321 aa) Dgeo_2262 (316 aa) DNA polymerase III delta prime subunit holC (147 aa) - - - DNA polymerase III chi subunit holD (137 aa) - - - DNA polymerase III psi subunit BS_yshC (570 aa) Deide_07030 (568 aa) DR_0467 (572 aa) Dgeo_1609 (572 aa) DNA polymerase, family X umuC (422 aa) - - - DNA polymerase type-Y family; error-prone DNA polymerase; in conjunction with UmuD and RecA catalyses translesion DNA synthesis umuD (139 aa) - - - In conjunction with UmuC and RecA facilitates

198 translesion DNA synthesis; autoprotease dinB (dinP) (351 - - - Pol IV, DNA polymerase type-Y family, aa) contains 1 UmuC domain - Deide_1p01880 (423 aa) - - Related to putative Y-family DNA polymerase dinG (716 aa) Deide_06510_2 (1683 aa) - - Predicted helicase; SOS inducer in E. coli multidomain proteinf: 1 is DnaQ, 2 is DinG, 3 is RecQ - Deide_06520 (849 aa) - - UvrD/REP helicase; COG0210, UvrD, Superfamily I DNA and RNA helicases BS_ywqA (922 aa) Deide_08980 (1132 aa) DR_1259 (600 aa) + Dgeo_1491 (1126 aa) DNA helicase SNF2/Rad54 family DR_1258 (916 aa)

(2 genes or frameshift?)

- Deide_17480 (689 aa) DR_1572 (frameshift) - Superfamily I DNA and RNA helicases (COG3973) dnaA (467 aa) Deide_00010 (459 aa) DR_0002 (466 aa) Dgeo_0001 (470 aa) Chromosomal replication initiator protein dnaB (471 aa) Deide_04710 (448 aa) DR_0549 (448 aa) Dgeo_2037 (849 aa; Replicative DNA helicase intein-mediated protein splicing) dnaC (245 aa) - - - DNA replication protein dnaG (581 aa) Deide_04900 (572 aa) DR_0601 (571 aa) Dgeo_1910 (587 aa) DNA primase dnaT (179 aa) - - - Primosomal protein 1 priA (732 aa) Deide_00480 (857 aa) DR_2606 (925 aa) Dgeo_0271 (841 aa) Primosomal protein N’ priB (104 aa) - - - Primosomal replication protein n priC (175 aa) - - - Primosomal replication protein N’’ rarA (mgsA, ycaJ) Deide_04980 (451 aa) DR_1898 (434 aa) Dgeo_1401 (451 aa) Replication-associated recombination protein (447 aa)

G. Other radiation tolerance-associated genes in D. deserti, D. radiodurans and D. geothermalis a,i D. deserti D. radiodurans D. geothermalis Protein description Deide_03030 (281 aa) DR_0167 (328 aa) Dgeo_0395 (289 aa) IrrE, regulator of recA expression in D. radiodurans Deide_2p01380 (294 DR_A0346 (300 aa) Dgeo_2628 (302 aa) DNA damage repair protein PprA aa) (plasmid) Deide_09150 (201 aa) DR_0423 (208 aa) Dgeo_0977 (201 aa) DNA damage response protein DdrA; Rad52/22 double-strand break repair protein Deide_02990 (178 aa) DR_0070 (199 aa) Dgeo_0295 (178 aa) DdrB Deide_23280 (232 aa) DR_0003 reversed (231 Dgeo_0047 reversed DdrC (DR_0003 and Dgeo_0047 probably incorrectly annotated, see main aa) (229 aa) text) Deide_01160 (158 aa) DR_0326 (198 aa) Dgeo_2186 (160 aa) DdrD Deide_11220 (228 aa) DR_0194 (227 aa) Dgeo_1282 (226 aa) DdrE; Related to zinc metallopeptidase - DR_0219 (444 aa) - DdrF - DR_0227 (242 aa) - DdrG Deide_20641 (82 aa) DR_0438 reversed (92 Dgeo_0322 (82 aa) DdrH (DR_0438 probably incorrectly annotated, see main text) aa) Deide_12530 (205 aa) DR_0997 (260 aa) Dgeo_1015 (204 aa) DdrI; Transcriptional regulator, Crp/Fnr family - DR_1263 (222 aa) - DdrJ - DR_1264 (140 aa) - DdrK - DR_1439 (175 aa) - DdrL Deide_02930 (180 aa) DR_2441 (170aa) Dgeo_0078 (168aa) DdrN Deide_20570 (129 aa)j DR_2574 (131aa) Dgeo_0336 (140aa) DdrO; transcriptional regulator, XRE family Deide_3p02170 (129 aa)j - DR_B0100 (210 aa) - DdrP - DR_B0141 (168 aa) - HicB-related protein - DR_1262 (531 aa) - ribonucleoprotein Ro/SS-A-related protein; involved in UV resistance in D. radiodurans - DR_0171 (230 aa) - IrrI a The gene names are from E. coli whenever an E. coli ortholog exists, or from B. subtilis (with the prefix BS_). The sizes of the predicted gene products (in number of amino acid residues) are indicated. D. deserti genes in blue and bold face indicate supplementary genes when compared to D. radiodurans and D. geothermalis. Grey indicates gene products identified in proteome analysis after standard cultivation of D. deserti. b DR_0022 did not demonstrate any appreciable uracil-DNA glycosylase activity in a study by Sandigursky et al (1) c DR_2586 contains domain COG1722; this domain is not found in the DR_2586 homologs Deide_02120 and Dgeo_0027 d D. deserti plasmid P2 contains a “pseudogene” (Deide_2p00840-Deide_2p00850-Deide_2p00860, interrupted by 2 stop codons), whose putative product has 27% identity with BS_AddA, and which is probably in operon with Deide_2p00830 encoding a protein with 25% identity with BS_AddB e In D. radiodurans, DR_1289 is required for resistance to DNA damaging agents, whereas DR_2444, which does not contain helicase and RQC domains, is not (2). All three HRDC domains are involved in DR_1289 RecQ function (3, 4) f A homolog of this multidomain protein Deide_06510 is found in Thermus thermophilus HB27: TT_P0128 (1649 aa) g Most, if not all, of the D. radiodurans Nudix hydrolases are probably not functional homologs of MutT (5) h Deide_12290 and Deide_1p00290 have 57% identity i DdrA, DdrE, DdrI and DdrN homologs with more than 40% identity are present in T. thermophilus. Other common genes/proteins are specific to Deinococcus. j Deide_20570 and Deide_3p02170 have 84% identity

1. Sandigursky M, Sandigursky S, Sonati P, Daly MJ, Franklin WA (2004) Multiple uracil-DNA glycosylase activities in Deinococcus radiodurans. DNA Repair (Amst) 3: 163-169. 2. Huang LF, Zhang SW, Hua XT, Gao GJ, Hua YJ (2006) Construction of the recQ double mutants and analysis of adversity in Deinococcus radiodurans. Wei Sheng Wu Xue Bao 46: 205-209. 3. Killoran MP, Keck JL (2006) Three HRDC domains differentially modulate Deinococcus radiodurans RecQ DNA helicase biochemical activity. J Biol Chem 281: 12849-12857. 4. Huang L, Hua X, Lu H, Gao G, Tian B, et al. (2007) Three tandem HRDC domains have synergistic effect on the RecQ functions in Deinococcus radiodurans. DNA Repair (Amst) 6: 167-176. 5. Xu W, Shen J, Dunn CA, Desai S, Bessman MJ (2001) The Nudix hydrolases of Deinococcus radiodurans. Mol Microbiol 39: 286-290.

199

200 Annexe 2. Stress response-related proteins in D. deserti, D. radiodurans and D. geothermalis Gene namea D. deserti D. radiodurans D. geothermalis Protein description General/heat groL Deide_22590 (549 aa) DR_0607 (548 aa) Dgeo_2231 (545 aa) Hsp10, molecular chaperone grpE Deide_21980 (216 aa) DR_0128 (221aa) Dgeo_2077 (218 aa) Hsp20, molecular chaperone groS Deide_22580 (95 aa) DR_0606 (120 aa) Dgeo_2230 (95 aa) Hsp60, molecular chaperone dnaK Deide_21970 (628 aa) DR_0129 (628 aa) Dgeo_2076 (629 aa) Hsp70, molecular chaperone dnaJ Deide_22000 (306 aa) DR_0126 (313 aa) Dgeo_2078 (302 aa) Hsp70, chaperone DR_1424 (420 aa) Dgeo_0451 (371 aa) Deide_07590 (373 aa) ibpA/ibpB Deide_18090 (164 aa) DR_1114 (182 aa) Dgeo_0505 (158 aa) Small heat shock protein Deide_11290 (136 aa) DR_1691 (166 aa) Dgeo_0859 (134 aa) Deide_1p01285 (129 aa) hslJ Deide_07610 (221 aa) DR_2056 (147 aa) - Related to heat shock protein DR_1940 (403 aa) - htpG - - - Hsp90, molecular chaperone dksA - - - DnaK suppressor protein clpA/clpB/clpC Deide_08490 (739 aa) DR_0588 (741 aa) Dgeo_1475 (737 aa) ATPase subunits of Clp protease DR_1046 (875 aa) Dgeo_1210 (861 aa) Deide_12640 (852 aa) DR_1117 (747 aa) Dgeo_1215 (745 aa) Deide_12680 (747 aa) clpX Deide_19580 (401 aa) DR_1973 (403 aa) Dgeo_2152 (406 aa) ATPase subunit of Clp protease clpP Deide_19570 (184 aa) DR_1972 (204 aa) Dgeo_2151 (200 aa) ATP-dependent protease with chaperone activity lon Deide_19590 (808 aa) DR_1974 (813 aa) Dgeo_2153 (813 aa) ATP-dependent Lon Serine protease DR_0349 (821 aa) Dgeo_0427 (820 aa) Deide_05670 (820aa) BS_hslU (clpY) - - - ATPase subunit of Clp photolytic system BS_hslV - - - ATP-dependent protease HslV (clpQ) sms (radA) Deide_12660 (449 aa) DR_1105 (503 aa) Dgeo_121 (449 aa) ATP-dependent serine protease COG1066 / DNA repair protein htrA Deide_21350 (391 aa) DR_0327 (364 aa) Dgeo_2185 (384 aa) Do serine protease, with PDZ domain Deide_01170 (389 aa) DR_0745 (366 aa) Dgeo_2025 (375 aa) Deide_04820 (387 aa) DR_1599 (470 aa) Dgeo_1495 (373 aa) DR_1756 (441 aa) Dgeo_0676 (426 aa) Deide_1p01286 (378 aa) DR_0984 frameshift Dgeo_0552 (423 aa) Deide_06250 (417aa) DR_0300 frameshift Dgeo_0103 (367 aa) Deide_10420 (434 aa) Deide_21550 (366 aa) prc Deide_21230 (445 aa) DR_1308 (447 aa) Dgeo_0277 (447 aa) Tail-specific periplasmic serine protease DR_1491 (445 aa) Dgeo_1216 (438 aa) Deide_12780 (418 aa) DR_1551 (447 aa) Dgeo_1479 (449 aa) Deide_10700 (438 aa) yaeL Deide_13240 (376 aa) DR_1507 (377 aa) Dgeo_1043 (372 aa) Membrane-associated Zn-dependent protease I ftsH Deide_23310 (626 aa) DR_0583 (618 aa) Dgeo_2075 (621 aa) ATP-dependent Zn protease DR_1020 (655 aa) Dgeo_1832 (623 aa) Deide_18550 (618 aa) DR_A0290 (655 aa) Dgeo_2182 (621 aa) Deide_01120 (622 aa) htpX Deide_21060 (354 aa) Predicted Zn-dependent proteases (possible chaperone) Deide_11220 (228 aa) DR_0194 (227 aa) Dgeo_1282 (226 aa) sugE Deide_20500 (106 aa) DR_1004 (103 aa) Dgeo_2170 (106 aa) Small multidrug resistance membrane protein DR_1005 (113 aa) Dgeo_1957 (112 aa) Dgeo_1956 (111 aa) hit Deide_15190 (115 aa) DR_1621 (118 aa) Dgeo_1329 (114 aa) Diadenosine tetraphosphate (Ap4A) hydrolase, HIT Dgeo_2064 (164 aa) family, cell cycle regulation Deide_12650 (137 aa) yebL Deide_07330 (299 aa) DR_2523 frameshift Dgeo_0534 (293 aa) Zn-binding (lipo)protein of the ABC type Zn transport system (surface adhesin A) Deide_3p02720 (322 aa) hflX Deide_19040 (569 aa) DR_0139 (525 aa) Dgeo_2006 (569 aa) GTPase, protease modulator DR_0646 (311 aa) Dgeo_1825 (309 aa) Deide_18500 (307 aa) BS_ytxJ Deide_14700 (217 aa) DR_1832 (211 aa) Dgeo_1464 (219 aa) Thioredoxin-like thiJ Deide_04910 (189 aa) DR_1199 (190 aa) Dgeo_0863 (188 aa) Protease I, related to general stress protein 18, ThiJ DR_0491 (261 aa) superfamily protein uspA Deide_22300 (177 aa) DR_2363 (160 aa) Dgeo_0155 (177 aa) Universal stress protein, nucleotide-binding DR_2132 (150 aa) Dgeo_1279 (158 aa) Deide_08840 (164 aa) Dgeo_2742 (157 aa)

Deide_3p00273 (301 aa) Deide_1p01770 (230 aa) Starvation spoT Deide_14760 (760 aa) DR_1838 (787 aa) Dgeo_1308 (766 aa) Guanosine polyphosphate (ppGpp) pyrophosphohydrolase/synthetase mazF - DR_0417 (117 aa) Dgeo_1937 (110 aa) ppGpp regulated growth inhibitor - DR_0662 (115 aa) mazE - DR_0416 (80 aa ) Dgeo_1936 (88 aa) Regulatory protein, MazF antagonist ppx Deide_01820 (500 aa) DR_A0185 (515 aa) - Phosphatase of ppGpp dps Deide_21200 (203 aa) DR_2263 (207 aa) Dgeo_0281 (222 aa) Starvation inducible DNA-binding protein DR_B0092 (241 aa)

201 cstA - - - Carbon starvation-induced protein, membrane Osmotic mscL Deide_02220 (131 aa) DR_2422 (128 aa) Dgeo_0305 (131 aa) Large conductance mechano-sensitive channel yggB Deide_18560 (417 aa) DR_1995 (426 aa) Dgeo_1833 (394 aa) Membrane protein - DR_0211 (368 aa) kdpD - DR_B0088 (365aa) Dgeo_0389 (557aa) Osmosensitive K+ channel histidine kinase sensor Dgeo_2852 (365 aa) domain trkA Deide_03820 (219 aa) DR_1666 (226 aa) Dgeo_1584 (219 aa) Potassium uptake system, NAD-binding component trkH/trkG Deide_03830 (425 aa) DR_1667 (458 aa) Dgeo_1583 (458 aa) Potassium uptake component DR_1668 (512 aa) Deide_3p00460 (451 aa) proW Deide_00160 (239 aa) DR_A0138 (247 aa) Dgeo_0171 (249 aa) Proline/glycine betaine ABC-type transport, permease Deide_00180 (393 aa) DR_A0136 (484 aa) Dgeo_0173 (397 aa) subunit proV Deide_00170 (308 aa) DR_A0137 (309 aa) Dgeo_0172 (310 aa) Proline/glycine betaine ABC-type transport, ATPase subunit yehZ Deide_00190 (301 aa) DR_A0135 (369 aa) Dgeo_0174 (300 aa) Proline/glycine betaine ABC-type transport, periplasmic subunit otsA Deide_21370 (457 aa) - Dgeo_0059 (457 aa) Trehalose-6-phosphate synthase otsB Deide_21360 (239 aa) - Dgeo_0060 (238 aa) Trehalose-6-phosphatase Deide_16570 (954 aa) DR_0463 (978 aa) Dgeo_0539 (944 aa) Maltooligosyltrehalose synthase TreY Deide_2p01290 (597 aa) DR_0464 (600 aa) Dgeo_0540 (603 aa) Maltooligosyltrehalose trehalohydrolase TreZ Deide_07350 (552 aa) DR_2036 (552 aa) Dgeo_0537 (556 aa) Trehalose synthase-like TreS aqpZ Deide_1p01582 (245 aa) - Dgeo_0516 (259 aa) Major intrinsic protein (aquaporin Z and glycerol Deide_3p02450 (232 aa) - Dgeo_1799 (221 aa) uptake facilitator) glpF DR_1929 (274 aa) Dgeo_2523 (274 aa) Deide_2p00230 (276 aa) Oxidative / detoxication oxyR Deide_03130 (301 aa) DR_0615 (317 aa) Dgeo_1888 (297 aa) Transcriptional regulator, LysR family Deide_16400 (318 aa) - Dgeo_1692 (329 aa) Deide_3p01240 (298 aa) - Dgeo_2711 (292 aa) Dgeo_2840 (299 aa) soxR Deide_2p01800 (285 aa) DR_2448 (280 aa) Dgeo_1424 (279 aa) Transcriptional regulator, MerR family - DR_2519 (127 aa) - Deide_3p02530 (131 aa) - - Deide_3p02771 (139 aa) - - Deide_04091 (157 aa) - - katE - DR_A0259 (772 aa) - Catalase DR_1998 (536 aa) Dgeo_2728 (543 aa) Deide_2p00330 (540 aa) katA S. pombe - DR_A0146 (361 aa) - Catalase; Eukaryotic type - - DR_A0145 (460 aa) - Fe dependent peroxidase katG - - - Catalase (peroxidase) - - - - Mn-containing catalase sodA Deide_07760 (207 aa) DR_1279 (211 aa) Dgeo_0830 (212 aa) Superoxide dismutase Mn or Fe dependent sodC - DR_1546 (182 aa) - Superoxide dismutase Cu/Zn dependent Deide_1p00740 (468 aa) DR_A0202 (462 aa) - Deide_19880 (182 aa) - - fur Deide_16041 (116 aa) DR_0865 (132 aa) Dgeo_0519 (135 aa) Ferric uptake regulation protein Deide_19480 (133 aa) Upstr DR_2342 Dgeo_2141 (137 aa) Deide_2p00340 (142 aa) Dgeo_2727 (135 aa) bcp Deide_10900 (164 aa) DR_0846 (175 aa) Upstr Dgeo_1265 (167- Peroxiredoxin, bacterioferritin comigratory protein, 178 aa) antioxidant protein

DR_1209 (159 aa) Dgeo_0990 (152 aa) Deide_09051 (170 aa) DR_1208 (163 aa) - - - Dgeo_2729 (159 aa) - tlp-like Deide_08290 (192 aa) DR_0189 (185 aa) Dgeo_1248 (192 aa) Peroxiredoxin, disulfide reductase DR_0345 (188 aa) - - - - Deide_2p00420 (174 aa) osmC Deide_16090 (147 aa) DR_1538 (172 aa) Dgeo_0526 (145 aa) Protein involved in alkylperoxide and oxidative stress DR_1857 (139 aa) Dgeo_0446 (146 aa) response, osmotically induced protein yhfA Deide_10790 (132 aa) DR_1177 (140 aa) Dgeo_1268 (132 aa) Protein involved in alkylperoxide and oxidative stress response, osmotically induced protein msrA Deide_10980 (195 aa) DR_1849 (206 aa) Dgeo_0843 (209 aa) Peptide methionine sulfoxide reductase A msrB Deide_04050 (150 aa) DR_1378 (157 aa) Dgeo_2072 (152 aa) Peptide methionine sulfoxide reductase B ahpC/ahpE- Deide_02430 (151 aa) DR_2242 (151 aa) Dgeo_0122 (151 aa) Thioredoxin reductase/alkyl hydroperoxide reductase like ahpf/trxB Deide_05800 (325 aa) DR_1982 (325 aa) Dgeo_1576 (327 aa) Thiol-alkyl hydroperoxide reductase DR_2623 (332 aa) Dgeo_2772 (321 aa) Deide_23360 (336 aa) DR_0412 (324 aa) Dgeo_1013 (331 aa) Deide_12541 (356 aa) DR_B0033 (833 aa) Dgeo_2331 (338 aa) Dgeo_0975 (320 aa) Deide_09090 (321 aa) grxA Deide_06390 (81 aa) DR_2085 (81 aa) Dgeo_1508 (81 aa) Glutaredoxin - DR_A0072 (91 aa) Dgeo_2583 (84 aa) nrdH Deide_13741 (92 aa) DR_0057 (84 aa) Dgeo_0729 (87 aa) trx Deide_01140 (138 aa) DR_A0164 (142 aa) Dgeo_2518 (141 aa) Thioredoxin Deide_18600 (121 aa) DR_0944 (141 aa) Dgeo_1837 (110 aa) btuE /BS_bsaA - - - Glutathione peroxidase 202 BS_cyp Deide_22920 (408 aa) DR_2538 (409 aa) Cytochrome P450 Deide_01550 (421 aa) Dgeo_0143 (407 aa) Deide_08170 (405 aa) Dgeo_0944 (405 aa) DR_2473 (381 aa) DR_1723 (445 aa) DR_C0001 (103 aa) DR_A0186 (fr.shift)

BS_yceH Deide_05270 (405 aa) DR_1127 (416 aa) Dgeo_0931 (406 aa) Toxic anion resistance protein, possibly tellurite resistance Desiccation - Deide_09710 (166 aa) DR_1372 (164 aa) Dgeo_1551 (166 aa) Desiccation-related protein, LEA 14 family - Deide_07540 (320 aa) DR_B0118 (337 aa) Dgeo_0097 (311 aa) Desiccation-related protein Dgeo_1323 (307 aa) Deide_08080 (312 aa) - Deide_08510 (200 aa) DR_1172 (298 aa) Dgeo_1473 (236 aa) LEA76 family desiccation resistance protein DR_0105 (163 aa)

Other hupA Deide_2p01940 (125 aa) DR_A0065 (122 aa) Dgeo_2501 (149 aa) Histone-like nucleoid DNA-binding protein Dgeo_0175 (116 aa) Deide_00200 (115 aa)

Deide_3p00060 (121 aa) Deide_3p00832 (166 aa) pspA Deide_09450 (226 aa) DR_1473 (223 aa) Dgeo_0996 (227 aa) Phage shock protein A, controls membrane integrity BS_yloU/ Deide_12280 (116 aa) DR_2068 (135 aa) Dgeo_0697 (152 aa) Alkaline shock protein, function unknown BS_yqhY Deide_17570 (110 aa) DR_0389 (110 aa) Dgeo_1721 (110 aa) cps Deide_3p00840 (87 aa) DR_0907 (86 aa) Dgeo_0638 (87 aa) Cold shock protein, OB fold nucleic acid binding Dgeo_1006 (89 aa) protein Deide_2p00490 (87 aa) Deide_09930 (94 aa) arsC Deide_3p02440 (153 aa) DR_A0123 (180 aa) Dgeo_0756 (153 aa) Arsenate (ArsC-like Rodanese protein) DR_0136 (127 aa) Dgeo_0403 (118 aa) Deide_18400 (118 aa) Dgeo_2548 (153 aa) Dgeo_2768 (153 aa) Dgeo_2770 (143 aa) aThe gene names are from E. coli, whenever an E. coli ortholog exists, or from B. subtilis (with the prefix BS_). Proteins in grey were identified in the proteome analysis after standard cultivation.

203

204