UNIVERSITÉ DE LA MÉDITERRANÉE AIX-MARSEILLE II

THÈSE pour obtenir le grade de DOCTEUR DE L’UNIVERSITÉ DE LA MÉDITERRANÉE Discipline : Immunologie École Doctorale des Sciences de la Vie et de la Santé

Caractérisation de BAD-LAMP dans les cellules dendritiques plasmacyoïdes humaines

présentée et soutenue publiquement par Axel DEFAYS Le 6 décembre 2010

Directeur de thèse : Philippe PIERRE

Jury de thèse :

M. le Docteur Philippe Benaroch Rapporteur

M. le Docteur Vassili Soumélis Rapporteur

M. le Docteur Philippe Pierre Directeur de thèse

M. le Professeur Philippe Naquet Président du jury Table des Matières

Liste des abréviations...... 3

La dynamique cellulaire...... 7

I. Les voies de biosynthèse ...... 7

1. La synthèse et l’export des protéines ...... 7

2. La maturation des protéines ...... 9

3. La voie sécrétoire...... 11

II. Le processus d’endocytose ...... 12

1. Les voies d’internalisation...... 12

2. Les compartiments endocytiques ...... 13

III. La famille des protéines membranaires associées aux lysosomes...... 16

1. Caractéristiques générales ...... 16

2. Expression et fonction de LAMP1/2 ...... 16

3. Les autres LAMPs ...... 17

Les cellules dendritiques plasmacytoïdes ...... 18

IV. Description générale des pDCs...... 19

1. Phénotype des pDCs humaines...... 19

2. Origine développementale ...... 20

3. Localisation et migration des pDCs ...... 21

4. Les pDCs murines ...... 21

V. Les récepteurs de type Toll...... 22

1 1. Structure générale des TLR ...... 22

2. Spécificité de la reconnaissance ...... 23

3. Signalisation ...... 23

4. Régulation de l’adressage des TLRs...... 25

VI. Fonction des pDCs humaines ...... 26

1. Des cellules productrices professionnelles d’interféron de type-1...... 26

2. La régulation de l’activation des pDCs ...... 27

3. Les cellules dendritiques dérivées de pDCs...... 28

4. Autres fonctions des pDCs ...... 29

5. Importance clinique des pDCs in vivo ...... 30

Objectifs...... 33

VII. Contexte de l’étude ...... 33

1. Etude de la forme murine de BAD-LAMP ...... 34

2. Fonction de l’homologue chez C. elegans ...... 35

Résultats ...... 36

VIII. Résumé de l’article ...... 36

IX. Article ...... 38

Discussion ...... Erreur ! Signet non défini.

Références ...... Erreur ! Signet non défini.

Annexe 1 ...... Erreur ! Signet non défini.

Annexe 2 ...... Erreur ! Signet non défini.

2 Liste des abréviations

ACLL : motif d’adressage de type di-leucine DXXLL (pour Acidic cluster Leu-Leu)

AP : complexe adaptateur (pour Adaptor protein complex)

ARNm : acide ribonucléique messager

BAD-LAMP : Brain and dendritic cell lysosome-associated membrane protein

BDCA : Blood dendritic cells antigen bHLH : domaine hélice-boucle-hélice (pour basic helix-loop-helix)

BiP : Bi nding protein

BST2 : Bone marrow st romal cell antigen 2

CCP : puits mantelés de clathrine (pour Clathrin-coated pits)

CCV : vésicules mantelées de clathrine (pour Clathrin-coated vesicles)

CD40L : ligand du CD40 (pour CD40 ligand) cDC : cellule dendritique conventionnelle (pour Conventional Dendritic cell)

CLP : progéniteur commun lymphoïde (pour Common lymphoid progenitor)

CMH I : complexe majeur d’histocompatibilité de type I

CMH II : complexe majeur d’histocompatibilité de type II

CMKLR : récepteur de type chimiokine (pour Chemokine-like receptor)

CMP : progéniteur commun myéloïde (pour Common myeloid progenitor)

COPI et II : complexe des protéines manteau I et II (pour Co at protein complex)

CPA : cellules présentatrices de l’antigène

DC-LAMP : Dendritic-cell lysosome-associated membrane protein

DCIR : Dendritic cell immunoreceptor

[DE]XXXL[LI] : motif di-leucine Asp/Glu-X-X-X-Leu-Leu/Ile

ERES : sites de sortie du réticulum endoplasmique (pour Endoplasmic reticulum exit sites)

ESCRT : Endosomal complexes required for transport 3 Fc !RII : récepteur de faible affinité pour les IgG

!"$%" : chaîne ""#$"% écepteur aux immunoglobulines IgE à haute affinité

Fuc : fuc ose

GABA : &'(#)"" -aminobutyrique (pour "-aminobutyric acid)

Gal : gal actose

GalNAc : N-acétylgalactosamine (pour Gal actosamine N-ac etyl)

GGA : Golgi-localizing, "-adaptin ear homology domain, ARF-binding protein

Glc : gl ucose

GlcNAc : N-acétylglucosamine (pour Gl ucosamine N-ac etyl)

GM-CSF : facteur de croissance hématopoïétique granulocyte-macrophage (pour Granulocyte-monocyte colony stimulating factor)

GPI : glycophosphatidylinositol

GTP : guanosine triphosphate

HEV : veinule à endothélium élevé (pour High endothelial venules)

ICOS-L : ligand de co-stimulation inductible des cellules T (pour Inducible T-cell co - stimulator ligand)

IDO : Indoleamine 2,3-dioxygenase

IFN : interféron

IFNRA : récepteur à l’interféron-"+,-$%" Interferon Receptor )

IL : interleukine

ILT : Immunoglobulin-like transcript

IRF : facteur de régulation de l’interféron (pour Interferon regulatory factor)

JaK : Ja nus kinase

KDEL : séquence peptidique K (lysine)-D (acide aspartique)-E (acide glutamique)-L (leucine)

KDELR : récepteur KDEL (pour KDEL receptor)

KIR : Killer cell Ig-like receptor

4 LAMP : protéine membranaire associée aux lysosomes (pour Lysosome-associated membrane protein)

LDL : lipoprotéine de basse densité (pour Low density lipoprotein)

LED : lupus érythémateux disséminé

Lin : marqueurs de restriction de lignée (pour Lin eage markers)

LPS : lipopolysaccharide

LRR : domaines riches en leucine (pour Leucine-rich repeats)

Man : man nose

Man-6-P : mannose 6-phosphate

MoDC : cellule dendritique dérivée de monocyte (pour Mo nocyte-derived Dendritic cell)

MPR : récepteur mannose 6-phosphate (pour Mannose 6-phosphate receptor)

MVB : corps multi-vésiculaires (pour Multi vesicular bodies)

NCAM : Neural cell adhesion molecule

NF-'B: facteur de transcription nucléaire kappa B (pour Nuclear factor-kappa B)

NK : Natural killer

NPXY : motif tyrosine Asn-Pro-X-Tyr

ODN : oligodéoxynucléotide

OLS : organe lymphoïde secondaire

OST : complexe oligosaccharyltransférase

PACSIN : Protein kinase C and casein substrate in neurons

PAMP : motif moléculaires associés aux pathogènes (pour Pathogen-associated molecular pattern) pDC : cellule dendritique plasmacytoïde (pour Plasmacytoid Dendritic cell)

PBMC : cellule mononuclée du sang périphérique (pour Peripheral blood mononuclear cell)

PDI : protein disulfide isomerase

PRR : récepteur reconnaissant des motifs (pour Pattern-recognition receptor)

PrP : protéine prion (pour Pr ion protéin) 5 RE : réticulum endoplasmique

Sial : acide sial ique

SRP : particule de reconnaissance du signal (pour Signal recognition particle)

STAT : Signal transducer and activator of transcription

SV40 : Simian virus 40

TCR : récepteur des cellules T (pour T cell receptor)

TGN : réseau trans -Golgi (pour Trans -Golgi network)

TIR : domaine Toll/IL-1 Receptor

TLR : récepteur de type Toll (pour Toll-like receptor)

TNF : Tumor necrosis factor

TRAIL : TNF-related apoptosis-inducing ligand

VIH : virus de l’immunodéficience humaine

YXX
: motif tyrosine Tyr-X-X-

6 Introduction

La dynamique cellulaire

La première partie de ce manuscrit servira à présenter succinctement la voie de synthèse protéique puis la voie endocytique, ainsi que leurs interconnexions. Ces paragraphes apporteront ainsi toutes les informations essentielles à la compréhension de l’étude présentée ici.

I. Les voies de biosynthèse

1. La synthèse et l’export des protéines

a) Le réticulum endoplasmique

Le réticulum endoplasmique (RE) remplit de nombreuses fonctions essentielles parmi lesquelles la synthèse des lipides, la régulation du calcium intracellulaire, et surtout la synthèse des protéines, leur translocation et le contrôle de leur intégrité. Cet organite est constitué d’une membrane continue séparant son contenu, le lumen, du reste du cytoplasme 1. Le RE entoure le noyau et sépare son contenu du noyau du reste du cytoplasme en formant une structure appelée enveloppe nucléaire. Le reste du RE forme un structure tubulaire appelée RE périphérique, qui se décompose en deux domaines d’apparence caractéristique en microscopie électronique 2. La membrane du RE périphérique est en partie associée à des polysomes, qui lui donnent un aspect granuleux à l’origine de son appellation de RE rugueux. Le reste du RE périphérique est appelé RE lisse. La plupart des protéines constituantes du RE périphérique sont partagées entre les deux domaines, seule une fraction de protéines est ségrégée dans le RE par un mécanisme encore inconnu 3.

Le RE rugueux constitue un domaine spécialisé dans la traduction des ARN messagers (ARNm) associés aux polysomes en polypeptides. Les premiers acides aminés du peptide en formation constituent une séquence, dite signal, qui permet d’initier la translocation co- traductionnelle. La séquence signal est reconnue par un complexe protéique appelé particule de reconnaissance de signal (SRP) 4. Le SRP permet d’adresser le peptide vers un complexe

7 protéique nommé translocon, constitué principalement par le canal Sec61p, qui assure le passage du peptide à travers la membrane du RE 5,6. De nombreuses protéines chaperon peuvent s’arrimer au translocon et réaliser les premières modifications post-traductionnelles sur le peptide en cours d’élongation 7. Parmi ces protéines chaperon, les enzymes PDI catalysent la formation des ponts dissulfure 8, la protéine BiP se fixe aux résidus hydrophobes des protéines non-repliées 9, le complexe enzymatique oligosaccharyltransférase (OST) permet d’effectuer les glycosylations-N10 . Les glycoprotéines sont ensuite prises en charge par un processus spécifique qui sera détaillé par la suite. Les dimensions du RE et de ses domaines sont adaptées au métabolisme des différents types cellulaires, le RE rugueux est ainsi très développé dans les cellules sécrétrices. Le domaine lisse est privilégié pour assurer les autres fonctions du RE, notamment pour l’homéostasie du calcium dans les cellules musculaires.

b) Les compartiments intermédiaires RE-Golgi

Le transport des protéines néo-synthétisées hors du RE est assuré par voie vésiculaire. Les vésicules en formation se concentrent sur une portion de la membrane du RE proche de l’appareil de Golgi, formant des domaines appelés sites d’export du RE (ERES) 11 . La GTPase Sar1p initie la formation de bourgeonnements au niveau du feuillet externe de la membrane des ERES 12 . L’activation de Sar1p permet le recrutement séquentiel des autres constituants du complexe de protéines de manteau II (COPII), les complexes Sec23-Sec24 et Sec13-Sec31 13 . Le complexe Sec23-Sec24 permet d’arrimer aux vésicules COPII les différents protéines cargo qui fixent de manière spécifique les protéines néo-synthétisées à exporter 14 . Les complexes Sec13-Sec31 s’assemblent pour former une cage qui constitue la structure des vésicules 15 . Les vésicules formées se détachent par l’action de Sar1p et sont transportées vers une structure tubulaire appelée compartiment intermédiaire RE-Golgi (ERGIC). Ce compartiment, probablement formé par une fusion homotypique de vésicules COPII ayant perdu leur manteau, sert de plateforme de tri vers l’appareil de Golgi 16 .

Le transport entre l’ERGIC et l’appareil de Golgi est probablement assuré par des vésicules mantelées par le complexe COPI 17,18 . Le complexe COPI est constitué d‘un manteau heptamérique, composé des sous-unités -/"! -/"!R -/"" -/" -, #- et -COP, associé à une GTPase de la famille Ras, Arf, qui contrôle le bourgeonnement des vésicules 19 . Les vésicules COPI sont également impliquées dans un mécanisme de transport rétrograde qui permet de rapatrier à partir de l’appareil de Golgi les protéines cargo ou toute protéine résidente du RE qui aurait 8 été empaquetée dans les vésicules COPII. La spécificité du transport rétrograde est assurée par des motifs d’adressages situés sur la partie cytoplasmique des protéines résidentes du RE, le plus commun étant le motif de type Lys-Asp-Glu-Leu (KDEL)20 . Ce motif est reconnu par la famille des récepteurs KDEL (KDELR), localisés principalement dans l’appareil de Golgi, chacun des trois récepteurs ayant sa propre spécificité 21 .

2. La maturation des protéines

a) L’appareil de Golgi

L’appareil de Golgi est un organite constitué par une pile de vastes saccules membranaires aplatis, les saccules proximaux formant le cis-Golgi et les saccules distaux le trans -Golgi. L’appareil de Golgi a deux fonctions principales, assurer la maturation des protéines et permettre leur adressage spécifique. Il fonctionne comme une vaste plateforme de tri, les protéines néo-synthétisées arrivant par une route commune au niveau de la face cis , pour subir une maturation progressive jusqu’à la face trans . Une première étape de sélection a lieu au niveau des compartiments cis -Golgi, qui contiennent une concentration importante de récepteurs comme le KDELR, assurant le recyclage des protéines résidentes du RE par la voie rétrograde COPI 17 . Les compartiments trans -Golgi se prolongent en un réseau développé de tubules et de vésicules de sécrétion en formation appelé réseau trans -Golgi (TGN) 22 . Le TGN représente le point de départ de la voie sécrétoire, qui organise l’export des protéines matures vers leur destination finale.

Le cloisonnement de l’organite en saccules est essentiel pour le fonctionnement de l’appareil de Golgi, permettant d’exposer les protéines séquentiellement à différents sets d’enzymes et d’optimiser ainsi le processus de maturation. Le mécanisme à l’origine de ce cloisonnement a longtemps été sujet à controverse, le modèle basé sur la maturation des saccules est maintenant consensuel 23,24 . Les saccules proximaux du cis -Golgi se forment de novo par fusion des vésicules de transport en provenance de l’ERGIC et progressent vers le TGN au cours de la maturation. Les enzymes contenues dans les saccules sont transportées par une voie rétrograde vers des saccules plus récents dans des vésicules mantelées COPI 25,26 . Le transport vésiculaire pourrait être complété par des structures tubulaires qui se forment également transitoirement entre deux saccules 27 , mais la contribution relative des deux voies au recyclage global n’est pas encore clairement établie. La spécificité de ce transport 9 rétrograde n’est pas totalement éclaircie, mais plusieurs facteurs comme l’acidification du pH, la régulation de l’adressage ou la nature des vésicules COPI entrent probablement en jeu 28 . Une composition enzymatique spécifique est ainsi maintenue dans chaque saccule. La maturation des glycoprotéines est un processus complexe qui nécessite de nombreuses enzymes différentes agissant de manière séquentielle entre les compartiments cis -Golgi et trans -Golgi.

b) La maturation des glycoprotéines

La glycosylation est un ajout post-traductionnel d’un chaîne oligosaccharide sur une protéine. Il existe une grande diversité dans la nature des chaînes ajoutées, qui varient en fonction de la protéine, du tissu ou du stade de développement 29 . La glycosylation joue un rôle important dans de nombreux processus physiologiques, notamment l’inflammation ou l’adhérence, ou pathologiques 30 . Deux grands types de glycosylation sont distingués, classés en fonction du type de liaison entre le groupement oligosaccharide et le peptide. Le type le mieux caractérisé est la glycosylation-N, qui met en jeu une liaison entre le sucre et l’atome d’azote d’une asparagine. Il existe une séquence consensus de fixation potentielle de type Asn-X-(Ser/Thr), où X ne peut pas être une proline 29 . Pour la glycosylation-O, le sucre est lié à l’atome d’oxygène présent dans la chaîne latérale d’une sérine ou une thréonine. Contrairement à la glycosylation-N, aucun site consensus n’a pu être déterminé 31 . Les glycosylation-O et -N diffèrent également par le type de groupement oligosaccharide attaché, le mécanisme et les enzymes impliqués dans la liaison.

La glycosylation-N est initiée dans le RE, où le complexe enzymatique OST permet le transfert d’une chaîne oligosaccharide riche en mannoses commune à toutes les glycoprotéines 32 . Cette chaîne pré-assemblée est constituée de deux N-acétylglucosamines (GlcNAc) liées à un branchement de 9 mannoses (Man) et 3 glucoses (Glc). Cette chaîne est clivée séquentiellement dans le RE par les enzymes glucosidase I et II au niveau des 2 groupements Glc terminaux, permettant l’interaction transitoire entre le Glc restant et les protéines chaperon calnexine et calréticuline 29,33 . La calnexine et la calréticuline permettent un contrôle du repliement, par un cycle de clivage / association du Glc terminal 34 . La chaîne oligosaccharide subit une série de clivages successifs des groupements Man dans le RE et le cis -Golgi par des enzymes mannosidase. Le groupement, dit hybride, subit au cours de sa maturation des étapes de ramification et de clivage des Man restant, en fonction des enzymes 10 majoritaires dans le saccule golgien 29,35,36 . Les étapes de ramification successives aboutissent à un groupement oligosaccharide complexe, formé par une combinaison de groupements GlcNac, Glc, Man, galactose, fucose et acide sialique.

La glycosylation-O est moins bien caractérisée et plus hétérogène. Le processus de glycosylation est initié dans l’appareil de Golgi par la fixation d’un groupement N- acétylgalactosamine (GalNAc) à une sérine ou une thréonine par une enzyme N-acétyl-a-d- galactosaminyltransférase 31 . Le groupement est ensuite rapidement ramifié avec différents sucres par des enzymes transférases spécifiques, formant une chaîne complexe. Il existe également des glycosylation-O formées à partir d’un groupement initial GlcNac, fucose ou acide sialique 31 . La grande diversité des groupements oligosaccharides possibles, le manque de séquence consensus de glycosylation et l’absence d’enzyme clivant tous les groupements possibles constituent les principaux obstacles à l’étude des glycosylation-O. A l’inverse, de nombreuses enzymes permettant de cliver les glycosylation-N ayant été caractérisées, chacune avec une spécificité propre 37 . Parmi ces enzymes, les endoglycosidases coupent l’oligosaccharide à sa base, libérant la chaîne entière. L’enzyme endoglycosidase H a la spécificité de ne couper que les chaînes portant un Man en position terminale, soit les chaînes de type riches en mannose et hybrides. Cette particularité est régulièrement exploitée pour contrôler expérimentalement la maturation des glycoprotéines, notamment le passage à travers l’appareil de Golgi.

3. La voie sécrétoire

Les saccules golgiens matures arrivent au niveau du TGN au terme de leur maturation. Le TGN permet l’adressage des protéines matures vers l’extérieur ou vers leur compartiment de destination. Plusieurs voies de sécrétion différentes ont été caractérisées. Les protéines peuvent ainsi être adressées directement vers la membrane plasmique et l’extérieur 38 . Cette voie permet la sécrétion de protéines de toutes tailles par la formation d’extrusions de la membrane du TGN 39 . Ces extrusions peuvent former des vésicules individuelles après une étape de scission de la membrane. Une autre voie de sortie permet d’atteindre directement les endosomes, par l’intermédiaire de vésicules mantelées de clathrine (CCV) 40 . La clathrine est une protéine en forme de triskel dont les branches s’associent en polyèdre, formant une cage autour de la vésicule en bourgeonnement. La clathrine s’associe avec plusieurs protéines adaptatrices, notamment pour assurer la spécificité du chargement et de l’adressage. 11 Parmi ces protéines adaptatrices, les protéines GGA sont spécialisées dans l’adressage des CCV du TGN vers les endosomes 41 . Les GGAs interagissent avec les triskels de clathrine et possèdent un domaine VHS interagissant avec les motifs d’adressage di-leucine de type Asp-X-X-Leu-Leu (ACLL) présents sur le domaine cytoplasmique des protéines cargo 42 . Les complexes adaptateurs (AP) remplissent également ce rôle de lien entre la clathrine et les protéines cibles. Les APs reconnaissent différents motifs d’adressage, parmi lesquels les motifs di-leucine Asp/Glu-X-X-X-Leu-Leu/Ile ([DE]XXXL[LI]) et les motifs tyrosine de type Asn-Pro-X-Tyr (NPXY) ou Tyr-X-X-
"+566
7 , où
est un acide aminé hydrophobe 43 . Le transport entre le TGN et les endosomes implique plus particulièrement la protéine AP-141 .

II. Le processus d’endocytose

L’endocytose est le mécanisme qui permet l’internalisation de macromolécules en provenance du milieu extérieur, par une invagination de la membrane plasmique. C’est un processus induit, qui est finement régulé au niveau des voies d’internalisation, des mécanismes de transport intracellulaires associés et de la signalisation qui en découle. L’endocytose joue un rôle dans le maintien de l’homéostasie, mais aussi dans la communication intercellulaire, la clairance ou l’établissement d’une réponse immunitaire. Les mécanismes de l’endocytose sont également exploités par de nombreux virus et organismes microbiens pour pénétrer dans une cellule.

1. Les voies d’internalisation

Il existe plusieurs voies d’entrée dans la cellule, chacune mettant en jeu des acteurs moléculaires différents. La voie d’internalisation la mieux étudiée et caractérisée passe par des puits mantelés de clathrine (CCP). La voie clathrine est utilisée par de nombreux récepteurs membranaires, comme le récepteur de la transferrine, les récepteurs tyrosine kinase ou les récepteurs couplés aux protéines G 44 . Le recrutement de la clathrine est dépendant de complexes adaptateurs, qui assurent également la ségrégation des protéines cibles. Parmi ceux-ci, le complexe AP-2 a la capacité de se fixer à des m -8(9:"#R&#%)::&;)"#)"8<,)"566
45 . Les protéines Eps15 et Epsin possèdent des domaines d’interaction avec l’ubiquitine, et ont la capacité d’interagir avec les complexes clathrine-AP-246 . L’ubiquitine est une protéine qui peut être liée à une lysine sur une protéine cible, sous forme monomérique ou de chaîne. Cette modification post-traductionnelle, réalisée par un jeu d’enzymes ubiquitine ligases, permet 12 entre autres de promouvoir l’internalisation et de modifier l’adressage des protéines cibles 47,48 . En plus des complexes adaptateurs, l’endocytose par la voie clathrine nécessite le recrutement de la dynamine. Cette protéine possédant une activité GTPase forme un polymère hélicoïdal autour du col des CCPs et entraîne leur séparation de la membrane plasmique 49,50 .

L’inhibition de la voie clathrine ne bloque pas l’internalisation de toutes les protéines, indiquant l’existence d’autres voies d’internalisation indépendantes de la clathrine. Les caveolae, des invaginations de la membrane, ont été identifiés par microscopie électronique. Ces structures sont enrichies en oligomères de cavéoline-151 et en cholestérol, qui renforce la stabilité des structures de cavéoline-152 . Les mécanismes régissant l’internalisation des caveolae ne sont pas encore éclaircis. En effet, les caveolae forment des structures stables à la membrane plasmique et leur cinétique d’internalisation est lente en comparaison de la voie clathrine, même si son activité semble être régulée 53 . La formation des vésicules à partir des caveolae nécessite la GTPase dynamine. Le virus SV40 pénètre dans la cellule par la voie caveolae, dont la spécificité est encore mal connue 53 . La protéine flotilin-1 est également présente au niveau d’invaginations de la membrane et participe à l’internalisation des protéines à ancre glycophosphatidylinositol (GPI), associées à des microdomaines membranaires riches en cholestérol, par un mécanisme indépendant de la dynamine 54 . Une troisième voie d’internalisation en rapport avec les microdomaines membranaires riches en cholestérol, et dépendante de la GTPase de la famille Rho cdc42, a été caractérisée 55 .

2. Les compartiments endocytiques

Une fois entrés dans la cellule, les vésicules perdent leur manteau et sont adressées vers le réseau endosomal 56 . La progression dans la voie endocytique se fait ensuite par une combinaison de transport vésiculaire et de maturation des compartiments 57 . Les endosomes sont donc des compartiments très dynamiques, avec un trafic vésiculaire afférent et efférent très intense. Ils subissent également des fusions homotypiques fréquentes, jusqu’à deux par minutes pour les endosomes précoces 57 . Toutes ces propriétés rendent l’établissement d’une classification stricte difficile, voire impossible. Les compartiments endocytiques sont généralement définis selon plusieurs critères, notamment en fonction du temps nécessaire pour accéder au compartiment de la surface, et de la nature des protéines Rab associées 56 . Les Rab sont une famille de protéines GTPase membranaires qui maintiennent en contact les vésicules ou compartiments avant la fusion des membranes 58 . La progression vers la voie 13 endocytique tardive s’accompagne aussi d’un changment de pH graduel. Les membranes des endosomes contiennent des pompes à protons qui organisent l’acidification des compartiments, jusqu’à atteindre des valeurs de pH inférieures à 5 dans la lumière des lysosomes 59 .

a) Les endosomes précoces

Les premiers compartiments rencontrés, les endosomes de tri, sont des compartiments Rab5 + accessibles en moins de deux minutes. Leur fonction principale est d’orienter les molécules internalisées vers les voies de recyclage ou vers la voie endocytique tardive, le temps de résidence dans ces compartiments étant en général très court 56 . Les endosomes de tri ont un pH légèrement acide, compris entre 6,3 et 6,8. Les conditions acides permettent de séparer les récepteurs de leurs ligands 59 . Les récepteurs sont alors adressés vers un autre type d’endosomes précoce, les endosomes de recyclage, alors que les ligands sont adressés vers des compartiments endocytiques plus tardifs. L’adressage vers la voie endocytique tardive est régulé par différents signaux, dont l’ubiquitine 60 .

Les endosomes de recyclage sont des compartiments Rab4 + avec un pH légèrement plus neutre que les endosomes de tri. Ces compartiments pourraient n’être formés que de manière transitoire par l’association de vésicules Rab4 +, sans avoir de véritable fonction propre. Il existe également un type d’endosomes de recyclage, les endosomes de recyclage péri-nucléaires Rab11 +, qui nécessitent un temps de trajet plus long et pourraient remplir une fonction différente 61 . Les propriétés qui déterminent le passage dans l’un ou l’autre de ces compartiments de recyclages sont inconnues.

b) Les compartiments endocytiques tardifs

Les endosomes tardifs sont des compartiments Rab7 +, qui sont accessibles en 15 minutes à partir de la membrane plasmique. Le pH dans la lumière de ces compartiments diminue avec la progression vers les lysosomes, passant d’un pH de 6,0 environ à des valeurs inférieures à 5,0. Les endosomes tardifs ont une fonction protéolytique, et contiennent des enzymes hydrolases 56 . Une voie d’adressage permet de relier le TGN directement aux endosomes tardifs, permettant l’adressage des enzymes hydrolases lysosomales. Les enzymes hydrolases néo-synthétisées sont marquées avec un motif mannose 6-phosphate (Man-6-P) au niveau du cis -Golgi. Lors de leur passage dans le TGN, le motif Man-6-P est reconnu par des 14 récepteurs mannose 6-phosphate (MPR), qui ont une fonction de protéine cargo 62 . Les MPRs ont la capacité de se fixer aux protéines adaptateur GGAs, permettant l’empaquetage du complexe MPR-hydrolase dans des CCVs 63 . Le transport est assuré par des vésicules Rab9 + vers les endosomes tardifs. Les complexes MPR-hydrolase se dissocient sous l’action du pH acide, permettant le recyclage du MPR vers le TGN.

La maturation des endosomes est souvent accompagnée par l’apparition de structures vésiculaires visibles en microscopie électronique, qui sont à l’origine de l’autre nom donné aux endosomes tardifs, les corps multi-vésiculaires (MVBs). Ces vésicules luminaux sont créées par des invaginations de la membrane de l’organite. Ce processus nécessite l’action d’un complexe protéique appelé ESCRT et permet d’adresser des protéines membranaires vers la voie de dégradation 64 . Les MVBs prennent une importance particulière dans la fonction des cellules présentatrices de l’antigène (CPA). Les molécules du complexe majeur d’histocompatibilité de classe II (CMH II) néo-synthétisées, associées à un peptide appelé chaîne invariante, sont adressées vers la membrane plasmique, puis internalisés jusqu’aux MVBs 65 . Dans ce compartiment, la chaîne invariante est dégradée, permettant le chargement d’un peptide antigénique exogène sur la molécule du CMH II.

Les lysosomes représentent la fin de la voie endocytique et sont accessibles en 30 minutes à partir de la membrane plasmique. Ces compartiments ont une apparence dense en microscopie électronique, différente des MVBs. Ils contiennent une grande concentration en enzymes hydrolases, ayant chacune une spécificité pour les lipides, les protéines ou les acides nucléiques. Ces enzymes ne fonctionnent qu’à un pH très acide, compatible avec le pH inférieur à 5 de la lumière des lysosomes et empêchant toute dégradation inappropriée en dehors de la voie endocytique. Les MVBs déversent leur contenu luminal, contenant des enzymes hydrolases et des molécules destinées à la dégradation, vers les lysosomes par un mécanisme de fusion transitoire des deux vésicules 66 . Les produits de la dégradation sont recyclés par la cellule, faisant des lysosomes une source importante de nutriments 67 . Les protéines transmembranaires constituantes de la membrane des lysosomes sont fortement glycosylées, pour les protéger de la dégradation 68 .

15 III. La famille des protéines membranaires associées aux lysosomes

La purification et l’étude de la composition des membranes lysosomales a permis de mettre en évidence plusieurs protéines transmembranaires enrichies dans ces compartiments. Parmi celles-ci, deux glycoprotéines particulièrement abondantes de poids moléculaire compris entre 100 et 115 kDa ont été identifiées. Ces protéines, appelées protéine membranaire associées aux lysosomes (LAMP) 1 et 2 ont pu être caractérisées grâce à des anticorps générés contre les membranes lysosomales 69 . Depuis cette caractérisation initiale, trois autres membres de la famille ont été identifiés, dont le sujet de cette étude, BAD-LAMP.

1. Caractéristiques générales

Les LAMPs sont des protéines transmembranaires avec un domaine extracellulaire développé et une queue cytoplasmique courte. La partie extracellulaire de LAMP1 et LAMP2 est constituée de deux larges domaines homologues séparés par une région charnière, riche en prolines 70 . Chacun des domaines contient 4 cystéines reliées deux à deux par des ponts disulfure, formant ainsi deux boucles. L’espacement entre les cystéines est particulièrement conservé, et caractéristique du « domaine LAMP ». Les deux protéines portent de nombreuses glycosylations, les oligosaccharides formant environ 60% de la masse moléculaire totale. Le domaine cytoplasmique contient un signal d’adressage intracellulaire basé sur une tyrosine, de type YXX
68 , qui assure la localisation des LAMP néo-synthétisées dans les lysosomes en interagissant avec le complexe AP-371 . LAMP1 (CD107a) et LAMP2 (CD107b), décrites initialement comme des marqueurs des lysosomes, sont localisées de manière transitoire à la membrane plasmique lors de la dégranulation 72 .

2. Expression et fonction de LAMP1/2

LAMP1 et LAMP2 sont exprimés de manière ubiquitaire. Toutefois, plusieurs isoformes de LAMP2 existent, chacune étant exprimée selon leur propre spécificité tissulaire 73,74 . Leur fonction est longtemps restée élusive, les hypothèses se limitant à un simple rôle de glycocalyx pour protéger les autres constituants des membranes lysosomales de la dégradation 70 . Les premiers indices fonctionnels sont venus de l’étude de la maladie de Danon, une maladie génétique caractérisée par une myopathie, une cardiomyopathie et un 16 retard mental. Une mutation délétère pour LAMP2 a été identifiée comme étant responsable de la maladie, se traduisant au niveau cellulaire par une accumulation de vacuoles autophagiques dans les cellules musculaires squelettiques et cardiaques (Nishino 00). Des modèles de souris déficientes pour LAMP1 sont parfaitement viables et ont un phénotype quasiment normal 75 , alors que des souris déficientes pour LAMP2 ont un phénotype beaucoup plus sévère, dont une taille réduite et une mortalité d’environ 50% entre 20 et 40 jours 76 . L’accumulation de vacuoles autophagiques dans de nombreux tissus dont le foie, les reins, le pancréas, les muscles cardiaques et squelettiques est cohérente avec le profil d’expression spécifique de LAMP2 et les symptômes de la maladie de Danon. Ces modèles suggèrent des fonctions partiellement redondantes pour LAMP1 et LAMP2, avec une fonction plus spécifique de LAMP2 dans certains tissus. Les souris déficientes pour les deux protéines meurent à l’état embryonnaire, démontrant à la fois qu’elles ont une fonction essentielle et partiellement redondante 77 . Des études plus récentes ont démontré que LAMP1 et LAMP2 sont requis pour la fusion des phagosomes avec les lysosomes 78 . Ces données suggèrent que LAMP1 et surtout LAMP2 jouent un rôle important dans la fusion des lysosomes avec d’autres vésicules intracellulaires dont les phagosomes et les autophagosomes.

3. Les autres LAMPs

La découverte des protéines LAMP1 et LAMP2 a permis de définir une nouvelle famille de protéines transmembranaires. D’autres protéines possédant un domaine de type LAMP ont été identifiées par la suite. La protéine lysosomale CD68 est fortement glycosylée et porte également un domaine de type LAMP sur sa partie cytoplasmique 79 . CD68, connue aussi sous le nom de macrosialine, est exprimée dans de nombreux types cellulaires et enrichies dans les monocytes et macrophages. Elle est un récepteur pour les lipoprotéines de basse densité (LDL) oxydées et les liposomes riches en phosphatidylsérine, la classant parmi les récepteurs éboueurs de classe D 80,81 . Les récepteurs éboueurs sont des récepteurs membranaires reconnaissant différents ligands endogènes ou microbiens 82 .

Un autre membre de la famille des LAMP présentant un profil d’expression spécifique été identifié. La structure de cette glycoprotéine lysosomale est proche de CD68 83 . Il est intéressant de noter que, contrairement aux autres protéines de la famille, elle n’est exprimée que dans certains types de cellules, les pneumocytes de type II 84 , ainsi que dans les cellules dendritiques conventionnelles (cDCs) humaines, d’où son nom de DC-LAMP 83 . Les cDCs 17 sont des CPA professionnelles qui sont activées par en périphérie des agents microbiens avant de migrer vers les organes lymphoïdes secondaires (OLS). L’activation induit l’expression de molécules de co-stimulation CD80 et CD86 qui confèrent aux cDCs la capacité d’induire la prolifération des cellules T naïves. Cette activation est dépendante de la présentation d’un antigène en association avec les molécules du complexe majeur d’histocompatibilité de classe I (CMH I) pour les cellules T CD8 + ou du CMH II pour les cellules T CD4 +. L’expression de DC-LAMP est induite lors de la maturation des cDCs, la protéine est alors adressée vers les lysosomes et co-localise en partie avec les molécules du CMH II avant leur relocalisation en surface. Cette propriété suggère un rôle spécifique de DC-LAMP dans le fonctionnement ou la dynamique des compartiments endosomaux CMH II +, même si son rôle précis n’a toujours pas élucidé.

Les cellules dendritiques plasmacytoïdes

Ce nouveau type cellulaire n’a été identifié pour la première fois qu’en 1958 comme des cellules avec une morphologie plasmacytoïde dans les zones riches en cellules T des ganglions lymphatiques humains et nommé « plasmocytes associés aux cellules T »85 . Au fur et à mesure de l’avancée de la caractérisation, la nature et le nom de ces cellules a été remis en question plusieurs fois. L’expression du marqueur CD4 et l’absence d’immunoglobulines de surface ont amené la requalification de ce type cellulaire en « cellules T plasmacytoïdes »86 . Puis le type cellulaire a été renommé « monocytes plasmacytoïdes » pour refléter l’absence de récepteur des cellules T (TCR) et la présence de marqueurs associés à la lignée myéloïde, dont les molécules du CMH II 87 . La fonction des monocytes plasmacytoïdes a été éclaircie il y a une dizaine d’année seulement. Une stimulation avec de l’interleukine-3 (IL-3) et du ligand CD40 (CD40L) entraîne leur différenciation en cellules dendritiques matures 88 . De plus, les monocytes plasmacytoïdes ont pu être rapprochés des « cellules productrices d’interféron de type-1 professionnelles », par l’étude approfondie du phénotype des deux types cellulaires 89,90 . L’interféron (IFN) de type-1 comprend les IFN-/" -!")8" -2/">$(" se lient au récepteur à l’IFN-" +?@ABC7 91 . L’appellation de « cellules dendritiques plasmacytoïdes » (pDCs) est maintenant universellement retenue. Les paragraphes suivants permettront de récapituler les caractéristiques essentielles des pDCs ainsi que leur importance dans l’établissement de la réponse immunitaire.

18 IV. Description générale des pDCs

1. Phénotype des pDCs humaines

Les pDCs ont une morphologie proche des plasmocytes lorsqu’elles sont observées sous un microscope optique en champ clair après un marquage Giemsa. Une différence notable est le noyau, qui est en forme de haricot dans les pDCs et non rond comme dans les plasmocytes. L’espace occupé par le noyau dans la cellule est très important, et l’observation en microscopie électronique en transmission révèle que le cytoplasme est occupé essentiellement par un RE rugueux très développé, un appareil de Golgi peu développé et de nombreuses mitochondries 92 . La surface des pDCs est dépourvue des marqueurs de restriction de lignée (Lin) correspondant aux types leucocytaires principaux du sang, les cellules T (CD3), les monocytes (CD14), les granulocytes (CD16), les cellules B (CD19, CD20) et les cellules NK (CD56) 89 . Les pDCs sont également négatives pour l’expression de CD1c, CD11c et CD33 93 , qui sont des marqueurs utilisés pour les cDCs. En plus de CD4 et du CMH II, les pDCs expriment à niveau très élevé CD123, la chaîne "#$"%écepteur à l’IL -394 . Les pDCs expriment aussi spécifiquement les marqueurs BDCA-2 (CD303), BDCA-4 (CD304) et ILT7 95,96 .

Les marqueurs les plus utilisés pour l’étude des pDCs sont BDCA-2 et BDCA-495 . BDCA-2 est une lectine de type-C exprimée exclusivement par les pDCs immatures et représente jusqu’ici la seule molécule permettant d’identifier ce type cellulaire de manière univoque. BDCA-2 est régulé négativement lors de l’activation des pDCs, et son engagement avec un anticorps inhibe les capacités fonctionnelles des cellules 95 . Ces propriétés limitent fortement l’intérêt de BDCA-2 en tant que marqueur des pDCs, malgré sa grande spécificité. BDCA-4, aussi connue sous le nom de neuropiline-1, est exprimée par les pDCs immatures et matures, mais aussi pas les cellules T naïves et les mDCs activées 95,97 . La sélection des cellules BDCA-4+ parmi les cellules mononuclées du sang (PBMCs) permet d’obtenir une population cellulaire composée de plus de 95% de pDCs sans altérer les capacités fonctionnelles de ces dernières 95,98 . Cette dernière méthode est très largement utilisée pour isoler les pDCs. Lors de la mise en culture, les pDCs entrent rapidement en apoptose. Le milieu de culture doit être complété avec de l’IL-3, qui promeut la survie et l’activation des cellules 88 . 19 2. Origine développementale

Le processus de différenciation des cellules souches hématopoïétiques en pDCs est longtemps resté inconnu, les cellules ne pouvant être clairement rattachées ni à la lignée myéloïde, ni à la lignée lymphoïde. Les pDCs présentent plusieurs caractéristiques attribuées aux lignées lymphoïdes, et ont donc été considérées comme des précurseurs de cellules dendritiques d’origine lymphoïde. Parmi ces caractéristiques, l’expression de la chaîne  du pré-TCR ou de la chaîne (E , qui sont des marqueurs respectivement des cellules T et B immatures 96 . Les pDCs procèdent également à des réarrangements D-J sur les gènes chaîne lourdes des immunoglobulines. Le facteur de transcription Spi-B, qui est exprimé dans les pDCs, inhibe la différenciation des cellules souches hématopoïétiques CD34 + en cellules B 99 . La dichotomie établie entre cDCs myéloïdes et pDCs lymphoïdes repose aussi sur les différences existantes dans le développement de ces deux types cellulaires. Le facteur de croissance hématopoïétique granulocyte-macrophage (GM-CSF), indispensable pour la différenciation des cDCs, n’a aucun effet sur le destin des pDCs 100 . Les protéines Id2 et Id3, lorsqu’elles sont surexprimées dans cellules souches hématopoïétiques, inhibent la différenciation en pDCs mais pas en cDCs 101 . Id2 et Id3 sont des inhibiteurs des facteurs de transcription E, à domaine de fixation à l’ADN hélice-boucle-hélice (bHLH). Enfin, le facteur de transcription Spi-B est indispensable au développement des pDCs et pas des cDCs 102 .

La dichotomie entre cDCs et pDCs a depuis été remise en question. En effet, le facteur de croissance hématopoïétique Flt3-ligand favorise simultanément le développement des mDCs et des pDCs 103 . Les progéniteurs communs myéloïdes (CMP) et lymphoïdes (CLP) peuvent tous deux se différencier en pDCs immatures phénotypiquement identiques, remettant en cause leur origine strictement lymphoïde 104 . Le facteur de transcription E, E2-2, est exprimé à haut niveau dans les pDCs. Le facteur E2-2 régule l’expression et l’activité de Spi- B105 , l’expression des récepteurs BDCA-2 et ILT7, ainsi que le facteur de régulation de l’interféron (IRF)-7106 . L’importance du facteur E2-2 fournit un mécanisme pour l’inhibition du développement des pDCs par Id2 et Id3. Tous ces résultats suggèrent que les pDCs sont issues d’une voie de différenciation spécifique. Une analyse globale du transcriptome des principaux types leucocytaires humains et murins a toutefois permis de déterminer que les pDCs forment une lignée proche des cDCs 107 .

20 3. Localisation et migration des pDCs

Les pDCs migrent vers les OLS par voie sanguine à leur sortie de la moëlle épinière. Cette migration se fait via les veinules à endothélium élevé (HEVs), vers les zones riches en cellules T des OLS. Elle est dépendante de l’expression par les pDCs du ligand de CD62 (CD62L) et du récepteur à chimiokine à motif C-C CCR7, qui vont interagir séquentiellement avec les chimiokines CCL19 et CCL21 exprimées en grandes quantités par les HEVs et les cellules stromales des zones riches en cellules T des OLS 108,109 . Lors d’une infection virale, les pDCs quittent les OLS et infiltrent les tissus périphériques en masse. Le sang périphérique est alors pratiquement dépourvu de pDCs 110 . Les bases moléculaires de cette migration sont encore mal déterminées, mais pourraient dépendre l’expression du récepteur à chimiokine à motif C-X-C CXCR3, exprimé à haut niveau par les pDCs, et de CXCR4. En effet, les pDCs sont naturellement peu sensibles aux chimiokines CXCL9, CXCL10 et CXCL11, ligands du récepteur CXCR3. Elles adoptent un comportement migratoire normal grâce à l’action synergique du ligand du récepteur CXCR4, la chimiokine CXCL12 exprimé par les HEVs 111 . Les chimiokines CXCL9, CXCL10 et CXCL11 sont régulées positivement par les cellules endothéliales et les fibroblastes dans un contexte inflammatoire, notamment par l’action des IFN de type-1112 . Ce modèle permet d’explique le recrutement des pDCs sur le site de l’inflammation et suggère l’existence d’une boucle de régulation positive via la production d’IFN. Le récepteur de type chimiokine 1 (CMKLR1), exprimé spécifiquement par les pDCs, déclenche la migration des cellules après la fixation de son ligand, la chémérine 113 . Les données actuelles sur la chémérine lui attribuant également des propriétés pro- et anti- inflammatoires, il paraît important de mieux caractériser cette voie de communication cellulaire pour déterminer son importance dans la fonction et la migration des pDCs114 .

4. Les pDCs murines

Après l’identification des pDCs chez l’homme, un type cellulaire fonctionnellement équivalent a été recherché chez la souris. Des cellules avec un phénotype plasmacytoïde, exprimant les marqueurs de surface Ly6G/C, B220, CD4 et CD11c et capables de produire de grandes quantités d’IFN de type-1 en réponse à une stimulation virale ont été identifiées 115 . Ces cellules migrent vers la rate par les HEVs et ont la capacité à se transformer en cellules dendritiques matures capables d’activer des cellules T dans les jours qui suivent la stimulation 116 . Toutes ces caractéristiques sont clairement celles d’un équivalent fonctionnel 21 de pDCs humaines. L’étude de la population de pDCs murines a été grandement facilitée par la génération de l’anticorps monoclonal spécifique 120G8 117 . Ce clone permet le marquer de la population de pDCs in vitro et leur déplétion in vivo . Les fonctions des pDCs humaines et murines sont largement semblables, cependant il existe quelques différences entre les deux espèces, notamment pour la production d’IL-12. En effet, une partie de la population de pDC murines a la capacité de produire de l’IL-12 en réponse à une activation virale 118,119 , contrairement aux pDCs humaines 120 . Malgré ces différences, l’étude de leurs profils d’expression respectifs démontre qu’il existe une parenté forte entre les pDCs murines et humaines 107 .

V. Les récepteurs de type Toll

La découverte du rôle essentiel de la voie de signalisation Toll dans la réponse immunitaire antifongique chez la drosophile 121 a permis d’identifier une nouvelle famille de récepteurs de reconnaissance de motifs (PRR). Des récepteurs homologues de Toll, exprimés chez les mammifères ont rapidement été identifiés 122 . Les récepteurs de type Toll (TLR) sont conservés chez tous les métazoaires et sont essentiels dans l’établissement de la réponse immunitaire innée. Les TLRs sont les récepteurs principaux pour la reconnaissance des motifs viraux par les pDCs humaines et murines. Cette spécificité s’explique par les types de TLRs exprimés dans ces cellules.

1. Structure générale des TLR

Les TLRs sont une famille de récepteurs transmembranaires de type-1, comptant 13 membres caractérisés à ce jour chez les mammifères, dont 10 chez l’Homme 123,124,125,126 . Ils partagent une structure composée d’un domaine extracellulaire comportant de nombreux domaines riches en leucine dits domaines LRR et d’une partie cytoplasmique portant un domaine activateur dit TIR, commun entre les TLRs et le récepteur à l’IL-1122 . Les domaines en LRR donnent au domaine extracellulaire une forme caractéristique en fer à cheval 127 . Le domaine TIR permet de recruter des protéines adaptatrices contenant également des domaines TIR. La cascade de signalisation déclenchée aboutit à l’activation du facteur de transcription NF-'B et à l’initiation de la transcription des gènes cibles.

22 2. Spécificité de la reconnaissance

Les TLRs peuvent être séparés en deux grands groupes en fonction de leur localisation cellulaire. Les TLRs 1, 2, 4, 5, 6 et 10 sont localisés à la membrane plasmique alors que les TLRs 3, 7, 8 et 9 sont localisés dans des compartiments intracellulaires. Il est intéressant de constater que les TLRs situés à la membrane plasmique reconnaissent des motifs moléculaires associés à des pathogènes (PAMPs) situés à la surface des bactéries ou des protozoaires, i. e. des lipopeptides et lipoprotéines bactériens pour les TLR1, 2, 6 et probablement 10 128 , les lipopolysaccharides (LPS) pour TLR4 129 et la flagelline pour TLR5 130 . Les TLRs intracellulaires reconnaissent eux des acides nucléiques, i. e. des ARN double-brin pour TLR3 131 , des ARN simple-brin pour TLR7 et 8 132 et des motifs d’ADN CpG non méthylés pour TLR9 133 . La répartition intracellulaire des TLRs reflète ainsi la spécificité de leurs ligands respectifs, permettant aux cellules de détecter des bactéries à leur contact ou des virus en réplication.

L’adressage différentiel des TLRs représente également une étape de contrôle supplémentaire de l’activation des récepteurs. La localisation intracellulaire de TLR9 prévient ainsi son activation par des acides nucléiques qui pourraient être naturellement présents dans l’organisme 134 . Les pDCs répondent principalement infections virales, en accord avec leurs niveaux d’expression des TLRs 7 et 9 à haut niveau 135 . Les pDCs répondent à une stimulation par les ligands synthétiques spécifiques des TLRs 7 et 9. TLR7 peut ainsi être stimulé par l’imiquimode, une imidazoquinoline analogue de la guanosine, et ses dérivés 136 . TLR9 peut être stimulé par des oligodéoxynucléotides (ODN) de synthèse riches en CpG non-méthylés. Les séquences des ODN CpG ayant un effet activateur sont spécifiques pour chaque espèce 137 .

3. Signalisation

La reconnaissance du ligand par les différents TLRs induit une dimérisation homologue ou hétérologue des récepteurs 138 . Des protéines adaptatrices sont fixées sur le domaine cytoplasmique des TLRs par association de leurs domaines TIR respectifs. Le changement de conformation du complexe induit par la dimérisation permet la transmission du signal par un mécanisme qui n’est pas encore totalement élucidé. Cinq protéines adaptatrices contenant un domaine TIR ont été caractérisés, les protéines MyD88, Mal, TRIF, 23 TRAM et SARM, qui sont toutes impliquées dans la signalisation des TLRs 139 . Parmi celles- ci, la protéine MyD88 est indispensable à la signalisation de tous les TLRs, à l’exception de TLR3.

Les protéines adaptatrices associées aux récepteurs changent en fonction de leur localisation intracellulaire, comme il a été montré pour TLR4. Le récepteur TLR4 transite entre l’appareil de Golgi et la membrane plasmique jusqu’à la rencontre avec le LPS (Latz 02). La formation du complexe TLR4-LPS à la membrane plasmique déclenche la signalisation dépendante de Mal et MyD88, l’activation de NF-'G et l’internalisation du récepteur par la voie endosomale 140,141 . Au cours de l’internalisation de TLR4, MyD88 est décroché du complexe TLR4 et remplacé par l’adaptateur TRIF 142 . Le changement d’adaptateurs découle sur un arrêt de la signalisation NF-'G et l’initialisation d’une cascade d’activation impliquant le facteur de régulation de l’interféron (IRF)-3140 . Le complexe TLR4 arrive finalement dans les lysosomes pour être dégradé. Cette relation entre la localisation intracellulaire du complexe TLR-ligand et la nature de la voie de signalisation activée illustre à nouveau l’importance d’une régulation spatiale des TLRs.

La capacité des pDCs à produire des IFN de type-1 en masse ne peut pas s’expliquer par la seule expression des TLRs 7 et 9. Une des spécificités des pDCs humaines est l’expression constitutive et à niveau élevé des facteurs IRF7 et STAT1 143 . La stimulation des cellules par l’IFN-" induit normalement une boucle de régulation positive, impliquant l’activation de la voie de signalisation JaK/STAT et l’expression du facteur IRF7 144 . Les pDCs ont ainsi la capacité de produire de l’IFN de type-1 rapidement et en grandes quantités. Une autre spécificité des pDCs a également été identifiée. La séquence des ODN CpG utilisés pour activer les cellules exprimant TLR9 est cruciale 145 , particulièrement pour les pDCs. Certains séquences d’ODN CpG activateurs, appelés ODNs type-A, déclenchent une production importante d’IFN alors que d’autres séquences d’ODN activateurs, appelés ODNs type-B, n’induisent qu’une production faible d’IFN mais favorisent fortement la différenciation des pDCs en cellules dendritiques matures 143 . La réponse à ces deux types de CpG est initiée dans des compartiments endocytiques différents. Les ODN CpG de type-A sont retenus dans les endosomes précoces où l’association de TLR9 et de MyD88 déclenche une cascade de signalisation aboutissant à l’activation du facteur IRF7 et la transcription des gènes de l’IFN de type-1. Les ODN CpG de type-B sont au contraire adressés normalement

24 vers les endosomes tardifs et lysosomes où la voie de signalisation indépendante d’IRF7 aboutit à l’activation de NF-'G et la maturation en cellules dendritiques 146,147 .

4. Régulation de l’adressage des TLRs

a) Régulation de l’export du RE

Il existe dans le RE un mécanisme de repliement commun à tous les TLRs, à l’exception de TLR3. La protéine chaperon gp96, ubiquitaire et localisée dans le RE, est indispensable pour le repliement correct et l’export des nombreuses protéines 148 . Son rôle dans la maturation des TLRs intracellulaires et extracellulaire a été récemment mis en évidence dans des lignées de souris déficientes pour gp96, incapables de produire une réaction immunitaires aux différents ligands synthétiques des TLRs 149 . Ce phénotype s’explique par une rétention des TLRs dans le RE. Une autre protéine chaperon résidente du RE, PRAT4A, produit un phénotype similaire chez des souris déficientes 150 . Le profil d’expression de PRAT4A est cependant plus restreint et son extinction n’affecte que les TLRs et non d’autres protéines dépendantes de gp96. PRAT4A agit en tant que co-chaperon de gp96, l’association de ces deux protéines étant nécessaire pour l’interaction avec les TLRs et leur export hors du RE 151 .

b) Régulation spécifique pour les TLRs intracellulaires

Les TLRs intracellulaires ont la particularité de résider dans le RE dans les cellules et de n’être exportés vers les endosomes qu’après l’activation 152 . Le trajet suivi par les TLRs au cours de ce processus n’est pas totalement clair. Les TLRs étant des glycoprotéines, le profil de glycosylation-N a été étudié pour déterminer son adressage lors de l’activation. La présence d’oligosaccharides sensibles à la digestion par l’enzyme endoglycosidase H avant et après la relocalisation dans les endosomes suggère que les TLRs sont adressés directement du RE vers les endosomes sans traverser l’appareil de Golgi 153 . D’autres données récentes suggèrent au contraire que TLR9 traverse l’appareil de Golgi et que les oligosaccharides portés par la protéine mature sont simplement d’une forme hybride sensible à la digestion par l’endoglycosidase H 154 . Une fois TLR9 localisé dans les endosomes, son domaine extracellulaire est clivé. Ce clivage est nécessaire et augment l’affinité du récepteur pour son ligand et permettant le recrutement de l’adapteur MyD88 155 .

25 Une protéine essentielle pour la relocalisation des TLRs intracellulaires du RE vers les endosomes a été identifiée dans un modèle de souris mutantes déficientes pour la signalisation de TLR 3, 7 et 9 156 . Ces souris dites « triple déficientes » (3d) portent une mutation ponctuelle sur un des domaines transmembranaires de la protéine UNC93B1. La caractérisation d’UNC93B1 a permis d’établir que cette glycoprotéine réside dans le RE, et que la mutation identifiée dans les souris 3d prévient l’interaction directe avec les TLRs 3, 7 et 9 157 . L’activation des cellules induit ainsi la relocalisation vers les endosomes du complexe UNC93B1-TLR, entraînant le clivage du TLR, l’interaction avec son ligand et l’initiation de la signalisation 158 . Il est intéressant de noter que la relocalisation des TLRs intracellulaires vers les endosomes n’est déclenchée qu’après l’activation des cellules alors que la cascade de signalisation initiée via ces mêmes TLRs nécessite une localisation endosomale. Une fraction de la population des TLR 7 et 9 semble être constitutivement localisée dans les endosomes et clivée, fournissant alors le signal d’activation initial 155 . Alternativement, la relocalisation de TLR 7 et 9 a été observée dans une lignée macrophage après une stimulation au LPS, suggérant que tout signal activateur est suffisant 158 .

VI. Fonction des pDCs humaines

Suite à leur activation, les pDCs produisent de l’interféron (IFN) de type-1, en quantité de 100 à 1.000 fois supérieure à celle produite par tout autre type de cellule sanguine 89 . De manière surprenante, après cette production massive de cytokines, les pDCs subissent un changement radical de morphologie pour devenir des cellules dendritiques matures capables d’induire la prolifération des cellules T naïves 88 . Les pDCs représentent ainsi le seul type cellulaire spécialisé possédant la capacité de se différencier à nouveau pour remplir deux fonctions distinctes successives. Ces deux fonctions leur donnent aussi un rôle central dans l’établissement et la régulation de la réponse immunitaire innée et adaptative.

1. Des cellules productrices professionnelles d’interféron de type-1

Les pDCs sont la source de production majeure d’IFN de type-1, parmi les cellules sanguines, bien qu’elles ne représentent que 0,2 à 0,8 % des PBMCs. Le pic de production a lieu entre 6h et 12h suivant l’activation, période pendant laquelle environ 50% des ARNm

26 totaux dans la cellule codent pour de l’IFN 98 . Les pDCs sont également capables de produire de l’IL-6 et du facteur de nécrose tumoral (TNF)-, bien qu’en quantités moins importantes que l’IFN. Cette spécialisation particulièrement marquée indique que cette phase de production de cytokines est une part essentiel de la fonction de ces cellules. L’IFN permet de limiter l’infection virale par un effet direct sur les cellules infectées, mais également en activant d’autres acteurs de la réponse immunitaire 159 . Les IFN de type-1 favorisent directement la maturation des mDCs, induisant l’expression de surface des molécules du CMH de classe I et II et des molécules de co-stimulation CD80 et CD86 160 . L’IFN favorise aussi la différenciation des monocytes en cellules dendritiques 161 . Les mDCs activées par de l’IFN produisent également des cytokines en grandes quantités, notamment de l’IL-12 et de l’IL-15, non produites par les pDCs, et qui favorisent l’activation et la différentiation des cellules T CD4 + naïves en cellules effectrices Th1 162,98 . L’IFN stimule la capacité des mDCs à activer les cellules T CD8 + cytotoxiques par un mécanisme appelé présentation croisée, qui permet la présentation d’antigènes exogènes en association avec les molécules du CMH I 163 . Les pDCs induisent la différenciation des cellules B activées par CD40 en plasmocytes producteurs d’immunoglobulines, par un mécanisme dépendant de l’IFN et de l’IL-6164 . Elles sont également nécessaires pour l’activation des cellules NK lors d’une infection virale (Dalod 03). Les pDCs activées favorisent aussi le recrutement des cellules NK et des cellules T activées par leur production de cytokines 165 . La production d’IFN de type-1 par les pDCs permet ainsi d’initier et de contrôler la réponse immunitaire innée et adaptative.

2. La régulation de l’activation des pDCs

Plusieurs mécanismes limitant les capacités de production d’IFN de type-1 par les pDCs ont été mis en évidence. Le premier de ces mécanismes met en jeu la lectine BDCA-2. L’engagement de ce récepteur par un anticorps spécifique diminue fortement les capacités de production d’IFN des pDCs suite à une stimulation par des ligands TLR 166 . L’engagement du récepteur ILT7 réduit également les capacités de production d’IFN en réponse à des ligands TLR 167 . Il est intéressant de noter que la réponse des pDCs est réduite même lorsque les récepteurs BDCA-2 ou ILT7 ne sont engagés qu’après la stimulation des cellules. BDCA-2 et ILT7 s’associent avec la chaîne " du récepteur aux immunoglobulines IgE à haute affinité (@'B?" ), et initient une cascade de signalisation interférant avec la signalisation TLR et aboutissant à l’activation du facteur NF-'G 167,168 . Si le ligand naturel de BDCA-2 n’a toujours

27 pas été identifié, ILT7 se lie à la protéine membranaire BST2 169 . L’expression de BST2 est induite par l’infection virale et l’IFN, renforçant l’hypothèse d’un mécanisme limitant dans le temps l’activation des pDCs.

Le récepteur NKp44 est une molécule exprimée à la surface des cellules NK activées 170 . L’engagement de NKp44 avec des anticorps spécifiques déclenche une réaction cytolytique des NK. Le ligand naturel n’a pas encore été trouvé, mais NKp44 permet la lyse NK-dépendante de certaines cellules tumorales 170 ainsi que de cellules infectées 171 . L’activation des pDCs par CD40L induit l’expression en surface de NKp44, et son engagement diminue fortement la production d’IFN en réponse à un stimulus TLR 172 . L’expression d’un autre récepteur membranaire par les pDCs, la lectine de type-C DCIR, a récemment été démontrée 173 . Comme pour BDCA-2, les ligands naturels de DCIR sont encore inconnus. Tous ces récepteurs participent à la régulation de l’activation des pDCs au moins in vitro , et font probablement partie d’un ensemble de signaux modulant in vivo la réponse immunitaire des pDCs en fonction du contexte.

3. Les cellules dendritiques dérivées de pDCs

La production d’IFN est réalisée pendant les 24h qui suivent l’activation, les pDCs devenant alors réfractaires à toute stimulation secondaire 98 . En effet, les pDCs subissent un changement de morphologie radicale pour acquérir l’apparence et les caractéristiques des cDCs matures 88 . Les cellules dendritiques dérivées de pDCs augmentent leur niveau d’expression en surface des molécules du CMH II, expriment les marqueurs de co-stimulation CD80 et CD86 et peuvent induire in vitro l’activation et la prolifération des cellules T naïves CD4 + 174 , et CD8 + 175 . Les pDCs activées peuvent induire une différenciation des cellules T CD4 + naïves cellules effectrices avec un profil Th1 productrices d’IL-10 et d’IFN-", par un mécanisme dépendant de l’IFN-176 . Elles peuvent également induire la prolifération de cellules T avec un profil Th2 productrices d’IL-4, IL-5, IL-10 et IL-13, grâce à l’expression de la molécule co-stimulatrices ligand d’OX40 et seulement en absence d’IFN 177 .

Les capacités de présentation antigéniques des pDCs aux cellules T sont discutées, notamment en raison de leurs faibles capacités de phagocytose 88 . Les pDCs peuvent toutefois, lorsqu’elles sont infectées, présenter des antigènes d’origine virale aux cellules T CD4 + et CD8 + 178 . Des études ont récemment établi que, contrairement aux cDCs, les pDCs activées 28 ont une néo-synthèse et un recyclage continus des complexes du CMH II associés à des peptides 179,180 . La capacité des pDCs humaines de présenter des antigènes viraux associés aux molécules du CMH I par présentation croisée a été également mise en évidence 181 . Ces résultats suggèrent que la présentation antigénique des pDCs est assurée par un mécanisme spécifique, qui est adapté aux spécificités d’une réponse anti-virale. Un autre mécanisme permettant la présentation antigénique par les cellules dendritiques dérivées des pDCs a été proposé, mettant en jeu les lectines de type C BDCA-2 et DCIR. Ces lectines sont internalisées lorsqu’elles sont engagées avec un anticorps spécifique, et dirigées vers les endosomes contenant les molécules du CMH II. Lorsqu’un peptide est couplé à ces anticorps, les pDCs stimulent les cellules T naïves spécifiques pour ce peptide 182,173 . La réalité de ce mécanisme de présentation et sa robustesse doivent toutefois être confirmée, les effets qu’ont l’engagement des lectines par un anticorps sur la différenciation des pDCs en cellules dendritiques ne sont pas clairement établis 183 .

4. Autres fonctions des pDCs

Les pDCs sont aussi capables d’induire une tolérance du système immunitaire. Les pDCs expriment le ligand de co-stimulation inductible des cellules T (ICOS-L) à l’état immature, et son expression est augmentée lors de l’activation des cellules 184 . Par l’expression d’ICOS-L, les pDCs induisent la prolifération des cellules T CD4 +FoxP3 + régulatrices productrices d’IL-10 185 et l’inhibition de la réponse immunitaire. Les pDCs immatures et activées expriment également de l’indoleamine 2,3-dioxygenase (IDO) 186 . Cette enzyme impliquée dans le catabolisme du tryptophane, exerce un rôle régulateur de la réponse immunitaire, sans que les mécanismes mis en jeu ne soient totalement élucidés 187 . L’expression d’IDO par les pDCs favorise la prolifération des cellules T régulatrices CD4 +FoxP3 +.

Il existe des preuves d’une activité cytotoxique directe des pDCs. Des facteurs pro- apoptotiques tels que le granzymeB et le ligand TRAIL sont exprimés par les pDCs lors de l’activation 188 . La lyse directe dépendante des pDCs, bien que moins efficace que la lyse par les cellules T ou NK, est suffisante pour limiter la croissance de cellules tumorales in vitro . Il est intéressant de noter qu’une sous-population de pDCs a été identifiée sur la base du niveau d’expression de la molécule CD2188 . Cette fraction de la population se distingue aussi par

29 l’expression de lysozyme et semble la population majoritairement responsable de l’activité cytotoxique. Le rôle physiologique des pDCs cytotoxiques doit toutefois être approfondi.

5. Importance clinique des pDCs in vivo

Les pDCs produisent des quantités massives d’IFN de type-1 lors d’une infection virale, activant ainsi de nombreux acteurs du système immunitaire inné et adaptatif. Les pDCs ont dans le même temps une fonction immunosuppressive nécessaire pour contrôler l’intensité et la durée de la réponse, et éviter des dommages importants pour l’organisme. L’importance de ces deux fonctions et leur mise en œuvre dans des conditions pathologiques ont été étudiés de manière extensive ces dernières années.

a) Infections virales

Plusieurs études ont été effectuées sur la souris pour déterminer in vivo quelle est l’importance des pDCs dans la réponse antivirale, notamment pour une infection par le cytomégalovirus murin 189 , le virus respiratoire syncitial 190 ou le virus influenza 191 . Tous ces virus ont déclenché une production massive d’IFN par les pDCs, mais leur rôle dans le contrôle de l’infection n’est pas totalement clair et semble dépendre du type de virus concerné. Une étude a même démontré une action immunosuppressive lors d’une infection par le virus influenza 192 .

Un cas particulier est celui des virus établissant des infections persistantes. Les pDCs expriment en surface CD4, mais aussi les récepteurs à chimiokine CCR5 et CXCR4, ce qui en font une des cibles du virus de l’immunodéficience humaine (VIH) 193,194 . Le nombre de pDCs circulantes dans le sang diminue chez les patients infectés, en corrélation avec la progression de la maladie et de l’apparition d’infections opportunistes 195 . Les pDCs infectées s’accumulent dans les organes lymphoïdes secondaires et ont une capacité de production d’IFN très réduite 196 . Le VIH inhibe les voies de signalisation TLR9-dépendantes, notamment par une interaction avec le récepteur BDCA-2197 . Ces résultats suggèrent que l’inhibition de la fonction des pDCs est une étape essentielle pour l’établissement d’une infection persistante du VIH. Des mécanismes de contrôle du nombre et de l’activité des pDCs ont également été mis en évidence pour les infections par les virus de l’hépatite B 198,199 .

30 b) Maladies auto-immunes

Le lupus érythémateux disséminé (LED) est une maladie auto-immune dont les symptômes vont de simples rougeurs cutanées à une forme systémique aigüe. Les patients atteints de LED ont des taux élevés d’IFN- dans le sang 200 et des pDCs infiltées en nombre au niveau des lésions cutanées 201 . Ce sont des complexes formés d’anticorps et d’ADN issus de cellules apoptotiques qui sont à l’origine de l’activation des pDCs, par un mécanisme dépendant de TLR9 et du récepteur de faible affinité pour les IgG (Fc!RII)202 . L’activation des pDCs favorise la différenciation des cellules B en plasmocytes sécrétant des auto- anticorps 164 , créant ainsi une boucle de contrôle positive et une activation permanente. La production d’IFN induit l’activation des cDCs, qui capturent les auto-antigènes et induisent la prolifération des cellules T auto-réactives 203 .

Le psoriasis est une autre maladie auto-immune résultant d’une activation permanente des cellules T auto-réactives 204 . Cette pathologie est également caractérisée par une production soutenue d’IFN et une infiltration de pDCs au niveau des lésions cutanées 205 . Le peptide antimicrobien LL37, exprimé notamment par les kératinocytes en conditions d’inflammation 206 , est à l’origine de l’activation des pDCs 207 . LL37 forme des complexes avec de l’ADN libre, le protégeant de la dégradation endosomale dans les pDCs et permettant l’activation de TLR9. La production d’IFN par les pDCs stimule l’activation des cellules T auto-réactives et induit l’expression de LL37, qui entretient la réaction inflammatoire.

c) Cancer

Les pDCs immatures ont la capacité d’infiltrer les tumeurs dans différents types de cancers, dont le cancer spinocellulaire de la tête et du cou 208 , le carcinome du poumon « non à petites cellules »209 , le cancer du sein 210 ou le mélanome cutané 211 . Les pDCs infiltrées ne produisent que peu ou pas d’IFN en réponse à une stimulation TLR dans le microenvironnement tumoral 208 et sont associées à un taux de survie et de rémission plus faibles 210 . Les pDCs inhibent la prolifération des cellules T et induisent la production d’IL-10, créant un environnement immunosuppresseur 212 . L’identification de pDCs exprimant l’enzyme IDO dans les ganglions lymphatiques drainant la tumeur chez la souris fournit un mécanisme possible pour cette action tolérogénique, les pDCs IDO + activant les fonctions immunosuppressives des cellules T régulatrices 213 .

31 d) Tumeurs CD4 +/CD56 +

Il existe une forme rare de tumeur particulièrement agressive d’hématodermie qui est caractérisée par une infiltration de cellules Lin -CD4 +CD56 +. Les patients traités ont un taux de rechute important et le taux de survie à 24 mois est faible 214 . L’origine de ces cellules a été longtemps débattue, notamment car leur phénotype ne ressemble à aucun type cellulaire connu alors 215 . Ces néoplasmes ont été supposés d’origine plasmacytoïde rapidement après la caractérisation des pDCs. Malgré l’hétérogénéité des cellules cancéreuses prélevées chez les différents patients, de nombreuses caractéristiques des pDCs se retrouvent sur les néoplasies, comme l’expression des marqueurs CD123 et ILT3, la capacité de produire des IFN de type-1 en réponse à une stimulation par des virus, ou l’expression de marqueurs de co-stimulation en réponse à une activation par CD40L216 . Les cellules tumorales CD4 +/CD56 + expriment également les marqueurs BDCA-2 et BDCA-4 et peuvent, après une stimulation par un virus, induire la différenciation et la prolifération de cellules T naïves 217 .

32 Objectifs

VII. Contexte de l’étude

Une étude collaborative a été réalisée au sein du Centre d’Immunologie de Marseille- Luminy pour établir le profilage génétique des différents types de leucocytes humains et murins (Annexe 2). La comparaison des différents transcriptomes a permis d’identifier des ARNm exprimés spécifiquement dans chacun des types cellulaires considérés. Parmi les transcrits spécifiques des pDCs, nous avons identifié la séquence de l’ARNm correspondant à la protéine putative c20orf103. L’étude de cette protéine possède plusieurs perspectives intéressantes pour notre laboratoire. Tout d’abord, les pDCs sont un type cellulaire qui n’a été identifié que récemment et peu de marqueurs spécifiques décrits et caractérisés sont alors disponibles. De plus, la protéine c20orf103 possède toutes les caractéristiques des protéines de la famille LAMP, c’est à dire $H":(;H&I"#R&#%)::&;)"566
")H",-:(8(-H"J -terminale, un domaine transmembranaire ainsi qu’un large domaine N-terminale portant un domaine LAMP et trois sites potentiels de glycosylation-N. Parmi les autres protéines appartenant à la même famille, LAMP1 et LAMP2 sont ubiquitaires et impliquées dans la dynamique des compartiments endocytiques tardifs, CD68 est un récepteur éboueur exprimé préférentiellement dans les monocytes/macrophages et DC-LAMP est exprimée dans les compartiments contenant les molécules de CMH II des cDCs fraîchement activées. Les LAMPs semblent ainsi particulièrement impliquées dans la dynamique des endosomes, alors que le profil d’expression et les fonctions supposées de CD68 et DC-LAMP en particulier leur donne une place importante dans la réponse immunitaire. Dans ce contexte, l’étude de la protéine c20orf103 pourrait permettre de mieux connaître la biologie cellulaire des pDCs, un type cellulaire ayant un rôle régulateur majeur dans la réponse immunitaire, et de mieux caractériser les mécanismes mis en jeu lors de leur activation.

Une première analyse de comparaison de séquences a permis d’établir qu’une protéine orthologue de c20orf103 existe chez les métazoaires, dont les insectes et les nématodes. La séquence protéique est de plus particulièrement conservée chez les mammifères, avec 83% d’identités pour l’homologue murin et 91% pour le rat. L’existence d’une forme homologue chez la souris, la protéine 6330527O06Rik, nous a permis d’envisager une première approche

33 en utilisant un organisme modèle plus facile à étudier. La caractérisation d’une protéine chez l’Homme est en effet particulièrement difficile, surtout s’il n’existe aucune hypothèse de départ quant à sa fonction. Le profil d’expression de la protéine représente ici une difficulté supplémentaire, les pDCs étant un type cellulaire rare, compliqué à étudier et qui n’a pas encore été étudié dans le laboratoire. Nous avons donc choisi d’effectuer une première caractérisation de la forme homologue exprimée chez la souris, renommée depuis BAD- LAMP. Cette étude a été réalisée en collaboration avec le laboratoire du Dr Harold Cremer (IBDML, Marseille) et a donné lieu à la publication d’un article scientifique en 2007 (Annexe 1), dont les points essentiels sont résumés ici.

1. Etude de la forme murine de BAD-LAMP

L’expression de BAD-LAMP a été détectée chez la souris de manière spécifique dans le cerveau. Une analyse plus détaillée a révélé que l’expression est restreinte aux neurones corticaux, avec une spécificité particulière pour les neurones pyramidaux des couches II, III et V. L’expression de BAD-LAMP est initiée dans le cerveau seulement après la naissance, augmentant progressivement entre P2 et P15 soit après la différenciation et la migration des neurones. Les niveaux maximum d’expression de BAD-LAMP dans le cortex sont atteints lors de la synaptogénèse, et maintenus à niveaux constants à l’âge adulte.

La séquence protéique de BAD-LAMP contient toutes les caractéristiques des molécules de la famille LAMP, avec les deux boucles formant le domaine LAMP, trois sites de glycosylation-N possibles, un domaine transmembranaire et un motif d’adressage YKHM sur la queue cytoplasmique. Trois formes différentes de la protéine ont été détectées par des techniques d’analyse biochimiques, correspondant à différents niveaux de glycosylation-N. Parmi ces trois formes, la forme la plus basse porte des oligosaccharides sensibles à la digestion par l’enzyme endoglycosidase H. Le clivage de toute glycosylation-N par l’enzyme N-glycosidase F produit une protéine de 31 kDa environ, soit la taille de la chaîne peptidique seule.

Dans les neurones, BAD-LAMP est adressée dans des vésicules intracellulaires regroupées en domaines le long des neurites. L’accumulation de ces vésicules sous la membrane plasmique semble dépendante de l’organisation des microdomaines lipidiques en surface. En effet, les vésicules BAD-LAMP + semblent associées aux domaines riches en Thy- 34 1, une protéine possédant une ancre GPI associée aux microdomaines ordonnés riches en cholestérol et sphingolipides, et au contraire totalement exclues des zones riches en PrP, GM1 ou NCAM. La concentration des compartiments BAD-LAMP + en domaine est dépendante de l’organisation des microtubules. Ces compartiments ne contiennent pas la protéine lysosomale LAMP2 et ne sont pas accessibles au récepteur de la transferrine, leur nature exacte n’a pas pu être identifiée. Les compartiments BAD-LAMP + co-localisent partiellement avec des marqueurs de vésicules synaptiques comme la synaptotagmine 1 ou VAMP2 uniquement au niveau des cônes de croissance, et non dans le reste de la cellule.

Une lignée cellulaire HeLa transfectée a été utilisée pour préciser le mécanisme d’adressage de BAD-LAMP. La protéine est localisée à la membrane plasmique lorsqu’elle est surexprimée dans ces cellules, mais une portion de la population est internalisée dans la voie endocytique par un mécanisme dépendant de la clathrine et du complexe AP-2. Le signal d’adressage YKHM porté sur le domaine cytoplasmique joue un rôle essentiel dans ce processus d’interrnalisation. Toutes les données obtenues montrent que BAD-LAMP définit des domaines vésiculaires dans une sous-population de neurones corticaux. La nature des compartiments BAD-LAMP + ainsi que leur rôle précis restent à définir. L’expression de la molécule au moment de la synaptogénèse suggère que BAD-LAMP est important pour le fonctionnement des neurones, même si son rôle précis n’a pas pu être établi.

2. Fonction de l’homologue chez C. elegans

La forme homologue de BAD-LAMP chez le nématode Caenorhabditis elegans se nomme UNC-46. La dénomination UNC (pour unc oordinated), indique que l’extinction de ce gène produit un défaut de motricité chez le ver. Unc-46 a ensuite été identifiée parmi cinq gènes nécessaires pour la signalisation GABA chez C. elegans 218 . L’expression de la protéine correspondante est effectivement restreinte aux seuls neurones, rappelant le profil d’expression de BAD-LAMP chez la souris, et plus précisément au niveau de vésicules synaptiques des neurones GABA 218 . La localisation d’UNC-46 à la synapse est dépendante de l’expression du transporteur vésiculaire du GABA UNC-47, une protéine codée par un autre des cinq gènes nécessaires pour la voie GABA. De manière intéressante, l’extinction du gène unc-46 créé un défaut d’adressage de la protéine UNC-47. UNC-46 fonctionne chez C. elegans comme un partenaire d’adressage du récepteur UNC-47. Cette étude représente la

35 seule donnée fonctionnelle concernant BAD-LAMP, qui pourrait tenir un rôle de protéine chaperon similaire chez la souris et chez l’Homme.

Résultats

VIII. Résumé de l’article

La transcrit de BAD-LAMP est détecté dans le cerveau chez l’Homme, mais également dans les cellules dendritiques plasmacytoïdes (pDCs). L’expression de BAD- LAMP permet d’identifier de manière spécifique les pDCs dans des coupes de tissu provenant d’organes lymphoïdes secondaires, ainsi qu’au sein des cellules mononuclées du sang périphérique (PBMCs) isolées. BAD-LAMP est également exprimée par les cellules tumorales CD4 +/CD56 + chez une majorité de patients testés. BAD-LAMP est régulée négativement rapidement après une stimulation des cellules par des ODN CpG au niveau trancriptionnel et au niveau protéique, indiquant que sa fonction est importante dans les cellules non activées.

BAD-LAMP est adressé vers un compartiment intracellulaire d’apparence vésiculaire réparti dans l’ensemble du cytoplasme. Ces compartiments ne contiennent ni le récepteur à la transferrine, ni la protéine lysosomale LAMP1. Parmi tous les marqueurs testés, seul un marquage de l’épitope KDEL, qui est un signal de rétention dans le réticulum endoplasmique, a révélé une co-localisation partielle. Le profil de glycosylation de BAD-LAMP montre qu’une seule forme de la protéine n’est détectable et que cette forme porte des oligosaccharides de type riches en mannose ou hybrides sensibles à la digestion par l’enzyme endoglycosidase H. Ce profil supporte l’hypothèse d’une localisation de BAD-LAMP dans un domaine spécialisé du réticulum endoplasmique (RE). BAD-LAMP est aussi localisée dans le RE lorsqu’elle est exprimée dans un type cellulaire relativement proche, les cellules dendritiques dérivées de monocytes (MoDCs).

BAD-LAMP est adressé à la membrane plasmique dans un modèle de cellules HeLa transfectées, probablement à cause d’une expression de partenaires d’adressage différents entre les deux types cellulaires. De manière intéressante, la surexpression de BAD-LAMP et d’UNC93B1, une autre protéine résidente du RE exprimée en grande quantités dans les pDCs,

36 modifie l’adressage des deux molécules dans les HeLa transfectées. Les deux protéines s’accumulent alors dans un compartiment intracellulaire non caractérisé. Cette co-localisation se produit également avec des formes mutées de BAD-LAMP ayant un adressage différent. Ce résultat démontre que BAD-LAMP et UNC93B1 peuvent s’influencer mutuellement dans un système de cellules transfectées.

37 IX. Article

“BAD-LAMP represents a novel biomarker of non- activated human plasmacytoïd dendritic cells and its intracellular transport is linked to UNC93B1 expression”

Article soumis

38 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

BAD-LAMP is a novel biomarker of non-activated human

plasmacytoid dendritic cells, which

chaperons UNC93B1 for its intracellular transport

Axel Defays ¶ 1,2,3 , Alexandre David ¶ 1,2,3 , Aude de Gassart 1,2,3 ,

Francesca De Angelis Rigotti 1,2,3 , Voahirana Camossetto 1,2,3, Pierre Brousset 4,

Tony Petrella 5, Marc Dalod 1,2,3 , Evelina Gatti 1,2,3,* and Philippe Pierre 1,2,3 ,*

1Centre d’Immunologie de Marseille-Luminy, Université de la Méditerranée, Case 906, 13288

Marseille cedex 9, France

2INSERM, U631, 13288 Marseille, France

3CNRS, UMR6102, 13288 Marseille, France

4INSERM, U563, CPTP, 31024 Toulouse, France

5 Centre de Pathologie, 21000, Dijon, France

¶ both first authors contributed equally to this work

* both last authors contributed equally to this work

[email protected] or [email protected] ,

telephone: + 33 4 91 26 94 79, telefax: + 33 4 91 26 97 30

This work is supported by grants to PP from Agence Nationale de la Recherche (BADLAMP layers) and La Ligue

Nationale Contre le Cancer. ADe and ADa are supported by a bourse régionale PACA and LNCC. PP is part of the Sybaris FP7 NoE.

1 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Abstract

The b rain a nd D C associated LAMP -like molecule (BAD-LAMP/c20orf103/UNC-46) is a newly identified member of the family of l ysosome-a ssociated m embrane p roteins

(LAMPs). BAD-LAMP expression in mouse is confined to neurons. We demonstrate here that in humans, BAD-LAMP can specifically be found in the type I interferon- producing plasmacytoid DCs. Human BAD-LAMP is localized in the endoplasmic reticulum of freshly isolated CD123 + pDCs and is lost upon activation by unmethylated cytosine-phosphate-guanine (CpG) oligonucleotides. The restricted pattern of BAD-LAMP expression allows for the rapid identification of normal and leukemic human pDCs in tissues and blood. We further show that BAD-LAMP and the Toll-Like-Receptor chaperone protein UNC93B1 co-localize and influence reciprocally their intracellular trafficking in transfected cells.

Keywords: pDC, Endoplasmic reticulum,TLR9, UNC-46, c20orf103, CpG

Running title: BAD-LAMP is a novel marker of Human pDCs

2 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Introduction

Plasmacytoid dendritic cells (pDCs) represent a rare but important cell type in the hematopoietic system 1,2,3 . pDCs have been shown to be the principal cell type producing type-I interferon (IFN) in response to viruses or during autoimmune diseases 3,4 . In addition, pDCs can function as APCs during immune responses and can promote antigen specific self-tolerance 5,6 . In humans, pDCs differ from conventional dendritic cells (CD11c+ BDCA1+ myeloid DCs) as they uniquely express Toll-like receptors (TLR) 7 and 9 7,8 , which enable them to sense efficiently endocytically-captured nucleic acids (e.g. CpG oligonucleotides) 4,9,10,11 .

Upon CpG ligation to TLR9, pDCs secrete high amounts of type I IFN and/or can differentiate to acquire the ability to stimulate naïve T cells and to modulate the immune response 18,19 . During differentiation, pDCs acquire antigen presentation capacity, up-regulate MHC molecules, as well as a broad range of co-stimulatory molecules 20 . Concurrently they also loose their type-I IFN production potential and down-modulate innate immunity receptors, such as TLR-9, ILT7 or BDCA-2 3,10 . pDC activation/differentiation induces the reorganization of different intracellular compartments, including endosomes. Hence the expression of molecules participating to these changes could be specifically regulated upon pDCs activation/differentiation.

Such regulation can be observed for TLR7 and TLR 9, which reside mostly in the endoplasmic reticulum (ER) of resting pDCs and, upon microbial activation, travel to the endosomes to get proteolytically activated 11 . Several chaperone proteins are involved in controlling TLR egress from the ER 12,13 . Among these, UNC93B1, a multi-

3 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC transmembrane ER protein, specifically interacts with the transmembrane domains of

TLR3, TLR7 and TLR9 and controls their delivery to the endosomes upon activation 14,15 . Mouse Unc93B1 mutant (3d) cannot signal via their intracellular

TLRs 16 and in human patients Unc93B deficiency has been linked to the etiology of herpes simplex virus-1 encephalitis 17 .

Human pDCs are generally identified with markers such as BDCA-4 (Neuropillin-1),

BDCA-2 (C-type lectin CLEC4C) and the IL-3 receptor α chain (CD123) 21 . However, these molecules are expressed by other cell types according to the immunological context. BDCA-4 is up-regulated on activated myeloid DCs 22 and CD123 is also expressed by basophils. Thus, the characterization of new markers for human pDCs is important to improve their detection 23 . Exemplifying this situation, a rare cutaneous tumor, termed b lastic p lasmacytoid d endritic c ell n eoplasm (BPDCN), has been proposed to originate from pDCs, due to the expression of molecular markers such as CD4, CD56, CD123, TCL1 and CD2AP 23,24,25,26,27 . However, difficulties in diagnosis can arise, since these markers are not unique to pDCs and sometimes aberrantly expressed by other cell types present in tumors. There is therefore a strong need for additional and robust markers of human pDC detectable in routine biopsies of neoplastic samples.

Via an in silico search for molecules involved in the organization of the endocytic pathway, we identified a new member of the LAMP protein family: b rain a nd D C associated LAMP -like molecule (BAD-LAMP, c20orf103; UNC-46) 28 . BAD-LAMP is a transmembrane glycosylated protein, which shares sequence and structural homology with the canonical LAMP1 and LAMP2 molecules (CD107) 29,30 . BAD-LAMP

4 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC harbors an endosomal addressing signal within is cytoplasmic tail, and contains several conserved cysteine residues, which allow for the formation of particular structural loops known as “LAMP folds”. Mouse BAD-LAMP was shown to accumulate in a novel endocytic compartment in specific subtypes of cortical projection neurons 28 . At functional level, mutations in UNC-46, the Caenorhabditis elegans ortholog of BAD-LAMP cause, in the nematode, defects in most neurotransmitter GABA-mediated behaviors. UNC-46/BAD-LAMP acts as a sorting chaperone addressing the membrane-associated GABA transporter (UNC-47) to synaptic vesicles 31 .

Although human BAD-LAMP, like its murine homologue 28 , is principally expressed in the brain, we show here that it is also specifically found in primary CD123 + pDCs and

BPDCN . BAD-LAMP mRNA and protein levels are down-regulated upon CpG DNA stimulation of freshly isolated primary BDCA-4 + human blood pDCs. In these cells,

BAD-LAMP is mostly localized in the endoplasmic reticulum (ER), and like TLR9 its pattern of N-glycosylation remains endoglycosidase H-sensitive. Interestingly in HeLa cells, ectopically expressed BAD-LAMP and UNC93B1 mutually influence their intracellular localization and efficiently co-localize to a specific subset of late endosomes. Thus BAD-LAMP might be part of a specialized molecular complex chaperoning UNC93B1 and represents a novel marker of human primary and transformed pDCs.

5 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Material and Methods

Bioinformatic and gene arrays.

Cell purification and RNA preparation as well as Gene array and meta-analysis was performed as described previously 32 .

Molecular biology

Northern blot was done with FirstChoice Northern Blot Human Blot I (Ambion) using a probe corresponding to exons 4, 5 and 6 of BAD-LAMP (clone IMAGE 6044324). The cDNAs coding for BAD-LAMP were obtained from the IMAGE consortium. BAD-

LAMP mutants and tagged forms were done as described previously 28 . The human

UNC93B1-His cDNA construct was a kind gift from Dr J-L Casanova (Rockfeller

University, NY, USA). The pUNO-TLR9-HA vector was obtained from Invivogen.

Antibodies and immunocytochemistry.

Monoclonal antibody 34.2 against BAD-LAMP wasraised in rat against the peptide

“KMTANQVQIPRDRSQYKHM” corresponding to BAD-LAMP cytoplasmic tail. For

FACS analysis, 34.2 mAb was directly labeled with fluorochrome Cy5 using the Cy5

Ab Labelling kit from GE Healthcare. Anti-CD123 (AC145) and anti-BDCA-4

(AD517F6) antibodies were obtained from Miltenyi Biotec, anti-FLAG (M2) antibody was from Sigma, anti-transferrin receptor was from Dr I. Mellman (New Haven, USA).

Rabbit anti-HA tag (9110), mAb anti-LAMP1 (H4A3), anti-KDEL (10C3) and anti-PDI

(RL90) were from AbCam, anti-His from Thermo Pierce, anti-CD63 (H5C6) and anti-

GM130 (35) from BD-Transduction, rabbit anti-HLA-ABC was from Dr J. Neefjes,

(NKI, Amsterdam, NL) and anti-HLA-DR (XD5) from Dr J. Thibodeau (University of

6 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Montreal, CA). All secondary antibodies were from Molecular Probes (USA), except

Cy3-5 secondary antibodies, which were from Jackson Immunoresearch.

Immunofluorescence and confocal microscopy was performed with a Zeiss LSM 510 as described previously 33 . Briefly for IF staining, pDCs and MoDCs were incubated on 1% Alcyan blue-coated glass slides for 15 min and subsequently fixed in 3% PFA for 15 min. ICC and IF staining were done in PBS, 10mM glycine, 5% FCS, 0.05 % saponin. Human lymph node and tonsil sections were kindly provided by Dr Norbert

Vey, Institut Paoli Calmettes, Marseille. Tissue microarray (TMA) and immuhistochemical analysis was performed as described previously 34 . Spleen cells from humanized gc/RAG -/- mice were kindly provided by Dr. Sophie Ugolini (CIML,

Marseille).

Cell purification and culture

Human PBMCs were isolated from whole blood by density gradient using Ficoll-

Paque PLUS (GE Healthcare). BDCA-4 + cells were magnetically sorted by positive selection using MicroBeads kit and AutoMACS cell separator (Miltenyi Biotec). Sorted cells were >95% pDCs based on BDCA-2 staining. pDCs were cultured at 0.5 to

1.10 6 cells/mL in RPMI-1640 containing 10% FCS and complemented with IL-3 at 10 ng/mL. pDCs were stimulated with ODN 2216 (A-type), ODN 2006 (B-type) or ODN

M362 (C-type) at a concentration of 2.5 µM. CD14 + cells were magnetically sorted by positive selection using MicroBeads kit and AutoMACS cell separator (Miltenyi

Biotec). Sorted monocytes were cultured at 2.10 6 cells/mL in RPMI 1640 supplemented with 10% FCS, nonessential amino acids, penicillin/streptomycin at

100 ng/ml and complemented with GM-CSF and IL-4 for 6 days for differentiation in

7 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

MoDCs. HeLa cells were grown in DMEM containing 10% FCS supplemented with penicillin/streptomycin at 100 ng/mL.

Cell transfection

HeLa cells were seeded the day before transfection at a cell concentration of 2.10 5 cells/mL.Transfections were performed using Lipofectamine 2000 reagent. Cells were harvested 24h after transfection for lysis. HeLa cells used for IF were seeded on microscopy glass slides before transfection and fixed in 3% PFA for 15 min 24h after transfection. MoDCs were transfected at 5 days of differentiation using in vitro transcribed mRNA as described previously 35 .

RT-PCR and mRNA extraction

RNA extraction was performed with the RNeasy Mini kit (Qiagen) except for human spleen FirstChoice total RNA (Ambion). RT-PCR was performed using Superscript II enzyme (Invitrogen) for the reverse transcription and Taq polymerase (Invitrogen) for the PCR amplification. PCR amplification was performed for 30 cycles unless stated otherwise. Quantitative RT-PCR was done using SYBR Green PCR buffer (PE

Biosystems) as described previously 35 and analysis of the results were obtained with

REST software 36 .

Immunoblots and immunoprecipitation

1% Triton X-100 cell extracts complemented with protease inhibitors cocktail (Roche) and 5 mM MG132 (Sigma) were immunoblotted after separation by 12% SDS-PAGE.

Immunoprecipitation was performed with 5 µg/sample of 34.2 antibody and protein G-

8 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Agarose beads (Millipore). Endoglycosidase H (Calbiochem) treatment was performed as described 33 .

Results

Human c20orf103/BAD-LAMP/UNC-46 mRNA is expressed in pDCs

Affymetrix Human Genome U133 Plus 2.0 arrays and Mouse Genome 430 2.0 were used to generate gene expression profiles of human blood monocytes, neutrophils, B cells, NK cells, CD4 or CD8 T cells and 18 mouse leukocyte profiles 32 . Our data were complemented with public databases on human blood DC subsets (pDCs, BDCA-1 cDCs, BDCA-3 cDCs, and lin-CD16+HLA-DR+ cells). Comparing mouse and human hematopoietic cell compendia, we identified BAD-LAMP/C20orf103 as a molecule expressed specifically in human pDCs among other hematopoietic cells (Figure 1A).

At nucleotide level, the human BAD-LAMP sequence is homologous at 45% with human LAMP 1 and LAMP 2, the firstly identified members of the LAMP family. BAD-

LAMP mRNA codes for a protein of 280 aa, (PI 6.42 and MW 31.7 kDa) predicted to contain a transmembrane domain (aa 236-256) and a 24 residues cytoplasmic tail

(Figure S1A). The cytoplasmic domain contains a YKHM sequence (aa 276) corresponding to a classical YXX Φ internalization and endosomal targeting motif.

The luminal domain contains 4 highly conserved cysteine residues, separated by an amino acid stretch of a length compatible with the formation of stable di-sulfide bonds and the acquisition of a classical “LAMP fold”. Human BAD-LAMP is 85% identical at amino acid level to its murine homologues and was also predicted to contain 3 characteristic N-glycosylation sites.

9 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Northern blot analysis of different human tissues indicated that BAD-LAMP is expressed almost exclusively in the adult human brain (Figure 1B). The detected mRNA corresponded to a transcript of around 1.8 kb with no apparent alternative spliced forms. To determine, if BAD-LAMP mRNA was truly expressed in pDCs, we first carried-out a RT-PCR on total human spleen mRNA and failed to reveal the presence of any specific BAD-LAMP transcript (Figure 1C). We could however amplify successfully BAD-LAMP messenger from the same mRNA extract by nested

PCR, a result never observed when performed with mouse BAD-LAMP specific primers and mouse spleen mRNAs (not shown). The low level of detected BAD-

LAMP mRNAs in human spleen was likely to reflect the rareness of pDCs in this organ, which are certainly in insufficient numbers to reveal BAD-LAMP transcription by tissue Northern blot. These results also supported our gene expression analysis, excluding BAD-LAMP expression from mouse leukocytes and lymphoid organs

(Figure S2).

Human BAD-LAMP is expressed in CD123/BDCA-2 pDCs.

In order to detect BAD-LAMP expression as a protein, a monoclonal antibody (mAb

34.2) was raised against the last 12 amino acids of BAD-LAMP cytoplasmic tail

(Figure S1A). This antibody recognized efficiently by immunofluorescence confocal microscopy the eGFP-tagged version of BAD-LAMP ectopically expressed in HeLa cells (Figure S1B). By immunohistochemistry (IHC) performed on human spleen and tonsil sections (Figures 2A and S3A), BAD-LAMP was detected in a rare cell type also positive for the two markers CD123 and CD4 and often found in the vicinity of high endothelial venules, a characteristic localization for pDCs 1,2,3 .

10 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

We next performed FACS on human peripheral blood monocytes (PBMC) using Cy5- conjugated 34.2 mAb. A rare BDCA-4 and BDCA-2 positive population of blood cells

(0.34%), likely to represent circulating pDCs, was singled-out by 34.2 intracellular staining (Figure 2B). BAD-LAMP expression in pDCs was further confirmed by the detection of an homogenous positive labelling of magnetically immuno-purified blood

BDCA-4 + cells (>95% pDCs) (Figure 2C). Moreover, confocal microscopy experiments, performed on the same freshly isolated pDCs, indicated that BAD-

LAMP accumulates mostly in intracellular membrane compartments, explaining the importance of performing an intracellular staining to detect this molecule by FACS

(Figure 2C). The pattern of BAD-LAMP expression was exquisitely restricted to pDCs, since it was only possible to amplify BAD-LAMP mRNA by RT-PCR from magnetically purified BDCA4 +/ BDCA2 + cells and not from the remaining pDC- depleted PBMC population (Figure 2D).

BAD-LAMP mRNA being undectable in mouse leukocytes both by gene arrays and

RT-PCR, we attempted to visualize BAD-LAMP expression in human pDCs isolated from the spleen of γc/RAG -/- mouse reconstituted with human CD34+ hematopoietic stem cell 38 . Confocal microscopy performed in parallel with anti-BDCA-4 and 34.2 revealed the presence of rare double positive human splenocytes in “humanized” mouse spleen (Figure S3B). This result confirmed that human pDCs differentiation is supported efficiently in CD34 + reconstituted mice and that BAD-LAMP can be used as a marker to track this rare cell type.

11 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

IL-3- and CpG-induced maturation decrease BAD-LAMP levels.

We studied BAD-LAMP protein expression in cell extracts obtained from human pDCs, PBMCs, monocyte-derived DCs (MoDC) and HeLa cells transfected with BAD-

LAMP cDNA (Figure 3A). By immunoblot, pDCs were the only hematopoietic cell type found to express naturally a 35 kDa form of the molecule. In transfected HeLa cells, used here as control, we detected several additional glycosylated forms absent from the pDC extract 28 . We then asked if pDC activation could influence BAD-LAMP expression. Purified BDCA4 + cells were cultivated with IL-3 in presence of different types of CpG oligonucleotides, known to promote IFN-type I secretion. Upon exposure to CpG, a strong diminution in BAD-LAMP mRNA levels was observed

(Figure 3B). This decrease was progressive over 24h and independent from the type of CpG used in the experiment. Both by intracellular FACS and immunoblot quantification, BAD-LAMP protein levels were found to be steadily reduced during activation, indicating that BAD-LAMP is mostly expressed in non-activated human pDCs and lost upon nucleic acid detection (Figures 3C and 3D). Interestingly BAD-

LAMP levels were affected by IL-3 treatment alone, confirming that IL-3 is able to induce pDC activation independently of TLR signaling 3.

BAD-LAMP is expressed in blastic plasmacytoid dendritic cell neoplasm pDC neoplastic transformation gives rise to the recently described BPDCN pathology 27 . At the morphological level, skin biopsies show a monomorphous cell proliferation simulating a pleomorphic T cell cutaneous lymphoma. The diagnosis of this neoplasm is mostly based on phenotypic criteria, namely histological analysis of tissue sections. Currently, the characterizing features of BPDCN are the expression of CD4, CD56 and CD123 antigens, and the absence of lineage specific markers for

12 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

B-cell, T-cell, NK-cell and myeloid-cell lineages. Our characterization of BAD-LAMP as a non-activated human pDC-specific marker led us to explore whether the detection of this molecule could facilitate the diagnostic of this rare tumor. We examined multiple paraffin sections from CD4 +/CD56 +/CD123 + tumors and could show that almost all were strongly stained by the 34.2 monoclonal antibody (Figure

4A). Analysis by tumor protein arrays (TPA) of 33 different tumors, classified as

BPDCN, revealed that 78% stained positively for BAD-LAMP (Figure 4B).

Interestingly, BAD-LAMP was not expressed in any of the other hemato-malignancies tested, including B and T lymphomas (supplementary table 1). In different histological analysis, we could also observe some BAD-LAMP staining in different epithelia, including the supra-basal skin epithelium. However, this positive staining was easily distinguishable morphologically from the BPDCN. BAD-LAMP represents therefore a novel and relevant marker for blastic plasmacytoid dendritic cell neoplasm, improving significantly the histological characterization of these tumors by a single round of staining.

BAD-LAMP is addressed in the ER of pDCs and transfected MoDCs

In a previous report 28 , the study of BAD-LAMP intracellular localization in mouse neurons has allowed us to define a non-conventional early endosomal compartment.

We tried here to establish if in primary pDCs its sub-cellular distribution would coincide with the neuronal one (Figure 5A and S4). Confocal immunofluorescence microscopy revealed that BAD-LAMP accumulates in a vesicular pattern distinct from the staining obtained with HLA-DR, HLA-A and GM130 (Golgi) (Figure S4). Staining performed with early (transferrin receptor) and late endocytic markers (CD63 and

LAMP1) also failed to show any obvious co-localization with BAD-LAMP (Figure 5A

13 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC and S4), confirming that the molecule does not accumulate normally in late endosomes, nor at the plasma membrane. On the contrary, BAD-LAMP displayed a significant degree of co-localization with ER-retention motif/KDEL-bearing molecules

(Figure 5A), suggesting that BAD-LAMP resides at least partly in the endoplasmic reticulum of pDCs. We further evaluated biochemically BAD-LAMP distribution by establishing the type of N-glycosylation acquired by the molecule during its intracellular transport. After immunoprecipitation, exposure to glycanases and revelation by immunoblot, we found that the bulk of BAD-LAMP molecules remained endoglycosidase H-sensitive (Figure 5B), indicating that it probably resides in the ER or does not use the classical secretory pathway for its export, as previously observed for TLR receptors or UNC93B1 transport 15 . We confirmed its ER localization by transfecting BAD-LAMP mRNA in human monocyte-derived DCs and imaging a considerable overlap of the protein with the ER-resident protein disulfide isomerase

(PDI) (Figure 5C). Thus, BAD-LAMP mostly resides in the ER of APCs, such as pDCs or transfected-moDCs, but neither in neurons nor in HeLa cells 28 . This dichotomy suggests a possible interaction of BAD-LAMP with other molecules expressed specifically in DC subsets and capable of controlling its egress from or retention in the ER.

BAD-LAMP and UNC93B1 are co-localized upon transfection in HeLa cells.

In Caenorhabditis elegans, UNC-46 has been shown to interact with vesicular GABA transporter (UNC-47) and promote its co-targeting to synaptic vesicles, supporting a potential chaperone role for BAD- LAMP/UNC-46 through specific interactions with other transmembrane proteins. Interestingly, a yeast two-hybrid screen performed with C. elegans proteins has revealed a direct interaction between UNC-46 and an

UNC-93 related protein, F31D5.2 39 . Since, UNC93B1 is highly expressed in human

14 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC pDCs 2 compared to other leukocytes (Figure S5A), we decided to investigate whether BAD-LAMP could directly interact with the endosomal TLR-chaperone.

We chose HeLa cells as an experimental system, since they do not express

UNC93B1, TLR9, nor BAD-LAMP, so that we could follow the intracellular transport of different tagged-versions of these molecules expressed individually or in combination (Figure S5B). The distribution of the different molecules was examined by confocal microscopy (Figure 6). Untagged BAD-LAMP and N-terminally FLAG- tagged BAD-LAMP (flagBAD) were mostly found accumulating at the plasma membrane and more rarely in some intracellular endosomes, The mutation of the cytoplasmic tail tyrosine residue 276 to alanine restricted the flagBAD Y276A protein to a complete cell surface distribution, due to a defect in its internalization and recycling 28 (Figure 6A and C). Interestingly, a flagBAD mutant completely deleted of its cytosolic tail (flagBAD- ∆Ct), was found almost uniquely in the ER of transfected cells (Figure 6C), suggesting that BAD-LAMP cytoplasmic domain is also important for its ER export. Conversely, when an eGFP moiety was fused C-terminally to BAD-

LAMP (gpfBAD), the resulting chimera was mostly addressed to LAMP1 + late endosomes and lysosomes (Figure 7A). This abnormal sorting of gfpBAD indicates, that a profound structural modification of BAD-LAMP cytoplasmic tail or its potential dimerization induced by eGFP can enhance the capacity of BAD-LAMP to reach and to remain associated with late endosomal compartments under specific circumstances. UNC93B1 was expressed as a 6xHIS-tagged form (hisUNC).

Accordingly to what previously described, hisUNC expressed alone accumulated in the ER of transfected HeLa cells 14 (Figure 6A). Co-expression of flagBAD and hisUNC provoked a strong redistribution of the two molecules and their co-

15 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC localization in bright punctuate intracellular structures, likely to be endosomes (Figure

6B). The same phenotype was obtained for gfpBAD and hisUNC co-expression, confirming the nature of the targeted compartment as LAMP1 + endosomes (Figure

7A). This effect was BAD-LAMP-specific since co-expression of the related lysosome-associated protein, DC-LAMP, with UNC93B1 did not have any effect on the sub-cellular distribution of the TLR chaperone, which mostly remained in the ER

(Figure 6B). BAD-LAMP cytoplasmic tail, but not its YXX Φ motif, seemed important for efficient co-chaperoning, since, when co-expressed with hisUNC, flag-BAD Y276A no longer distributed to the plasma membrane and was able to support UNC93B1 endosomal targeting (Figure 6C). On the contrary, flagBAD- ∆Ct co-expression with hisUNC had a modest impact and only a small portion of the two molecules could be found in endosomal compartments, while the bulk remained in the ER (Figure 6C).

This active and efficient intracellular re-localization upon co-expression of the two molecules indicates that UNC93B1 and BAD-LAMP function as co-chaperones and have a reciprocal influence on their intracellular addressing.

We then evaluated the impact of TLR9 expression on BAD-LAMP and UNC93B1 co- chaperoning activity by expressing in HeLa cells a HA-tagged form of TLR9 (haTLR) together with flagBAD and hisUNC. ha-TLR9 accumulated together with UNC93B1 in the ER, abrogating the chaperoning effect of BAD-LAMP (Figure 7B, *). TLR9 competition with BAD-LAMP for UNC93B1 availability was demonstrated in cells expressing relatively low amounts of haTLR9 compared to higher levels of flagBAD, and in which hisUNC93B1 remained localized in endosomal compartments, away from TLR9 main ER intracellular location (Figure 7B, +). Thus the relative abundance of these molecules in the same cell is likely to govern their intracellular transport,

16 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC explaining why BAD-LAMP is mostly found in the ER of pDCs. The fact that BAD-

LAMP expression is down-regulated upon maturation is therefore likely to be an important biochemical event, which might be key to control UNC93B1 function during the detection of microbial products by TLR9 in pDCs.

Discussion

Human BAD-LAMP represents a new member of the LAMP family, based on sequence analysis. However its tissue expression pattern and intracellular distribution are quite unusual compared to other classical LAMP family members, which have a widespread expression and specifically accumulate late endosomes and in lysosomes. In mouse, BAD-LAMP is expressed exclusively in brain while in human it is also found in CD123 +/BDCA2 +/BDCA4 + plasmacytoid dendritic cells circulating in the blood or localized in secondary lymphoid organs. This situation is reminiscent of the tissue distribution of another non-conventional LAMP family member, DC-LAMP/LAMP3, which is expressed both in activated human conventional DCs and in human type II pneumocytes 40 , while its expression remains restricted to type II pneumocytes in mouse 41 .

Blastic plasmacytoid dendritic cell neoplasm, which was previously called

CD4+/CD56+ hematodermic neoplasm and blastic NK-cell lymphoma, is a hematopoietic malignancy of pDC origin. The recent discovery of CD123 and BDCA-

2 expression in BPDCN has been determinant to point towards its pDC origin 23,25,26 .

Clinically, most cases of CD4+/CD56+ leukemia show initial cutaneous involvement, although pDCs are generally absent from normal skin. Our discovery of BAD-LAMP

17 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC expression in these tumors definitely confirms their plasmacytoid origin and suggests that these neoplastic cells are in a resting state. Indeed, BAD-LAMP is only expressed abundantly in freshly isolated pDCs and its expression is lost upon activation by IL-3 or TLR ligands. A majority of leukemic pDCs are therefore phenotypically similar to their normal resting counterparts and BAD-LAMP detection offers an novel and alternative mean of identifying these aggressive tumors. 20% of the tested BPDCN being negative for BAD-LAMP, it will be of interest to determine if this lack of expression pinpoints a specific category of neoplasms, which are characterized by a different activation state or fall in a different clinical cohort.

In a previous report 28 , we showed that in neurons, BAD-LAMP was mostly addressed in a subset of endosomal structures accumulating in the growth cone. We show here that in pDCs, BAD-LAMP accumulates in the ER, prior to its disappearance upon activation by CpG nucleotides sensing. Interestingly UNC-46, the C. elegans ortholog of BAD-LAMP, has been shown to serve as a chaperone for the GABA transporter

(UNC-47) molecule and to be required to sort properly the transporter in synaptic vesicles 31 . Interestingly, UNC-47 has also been shown to influence reciprocally the traffic of UNC-46, suggesting the existence of a co-chaperoning mechanism allowing the two molecules to exit together from the ER and reach synaptic vesicles.

Interestingly, although many neuronal molecules are found in pDCs (eg. BDCA-

4/Neuropilin-1 or Pacsin 1/syndapin) 32 , no significant expression of the GABA transporter could be detected in these cells, further suggesting that BAD-LAMP could serve as a co-chaperone for other transmembrane molecules expressed in human pDCs, and potentially not in neurons.

18 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

The discovery of a direct interaction between UNC-46 and the UNC-93-related protein F31D5.2, led us to investigate if the UNC-93 ortholog, UNC93B1, which is expressed in high amount in human pDCs compared to other cells

(http://biogps.gnf.org), could be one the molecule exerting a co-chaperoning activity on BAD-LAMP. In HeLa cells, which are not expressing UNC-93B1, nor the GABA transporter, BAD-LAMP is misrouted to the cell surface and only can reach late endosomes, upon co-expression with UNC-93B1. Alternatively in MoDCs, which naturally express UNC-93B1 (Figure S5), BAD-LAMP is retained in the ER and does not display any obvious endosomal localization, even when DCs are stimulated though TLR3, which also interacts with UNC93B1 (not shown). Thus although BAD-

LAMP can be endocytosed and recycled through a tyrosine-based addressing signal within its cytoplasmic tail 28 , its transport depends on factor expressed specifically in particular cell types, such as UNC-47, UNC-93B1 and TLR9.

Upon co-expression, BAD-LAMP and UNC-93B1 have the ability to reach together a specific endocytic compartments displaying some, but not total, overlap with LAMP1- positive late endosomes or lysosomes. However this situation is artificial and several additional molecules interacting with UNC-93B1 or BAD-LAMP are likely to be present in a physiological situation. UNC-93B1 distribution and the availability of other factors in BAD-LAMP expressing cells (e.g. UNC-47 or TLR9) could influence

BAD-LAMP transport or reciprocally be influenced by BAD-LAMP. However, given the importance of UNC93B1 for TLR9 intracellular transport and endosomal activation, the presence of BAD-LAMP as a potential co-chaperone of UNC93B1 in non-activated human pDCs could be of great importance to promote or prevent the activation of these cells by nucleic acids.

19 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

20 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Authorship

A. Defays, A. David, A. de Gassart, F. De Angelis Rigotti and V. Camossetto all performed research and analyzed data. Pierre Brousset and T. Petrella performed research and analyzed TMA. M. Dalod performed research and analyzed microarrays. E. Gatti and P. Pierre designed the research, analyzed data and wrote the paper. The authors declare to have no relevant financial conflict of interest

Acknowledgements

We thank for expert technical assistance the PICsL imaging core facility and Michel

Pierres at the CIML monoclonal antibody facility. J-L Casanova for the kind gift of reagents. This work is supported by grants to PP from La Ligue Nationale Contre le

Cancer, the ANR BAD-LAMP layers and the ANRS. A. Defays and A. David are supported by fellowships from the MENRT and la Fondation pour la Recherche

Médicale.

21 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

References

1. Colonna, M., Trinchieri, G. & Liu, Y. Plasmacytoid dendritic cells in immunity. Nat. Immunol 5, 1219-1226 (2004). 2. Liu, Y. IPC: professional type 1 interferon-producing cells and plasmacytoid dendritic cell precursors. Annu. Rev. Immunol 23 , 275-306 (2005). 3. Cao, W. Molecular characterization of human plasmacytoid dendritic cells. J. Clin. Immunol 29 , 257-264 (2009). 4. Gilliet, M., Cao, W. & Liu, Y. Plasmacytoid dendritic cells: sensing nucleic acids in viral infection and autoimmune diseases. Nat. Rev. Immunol 8, 594-606 (2008). 5. Ochando, J.C. et al. Alloantigen-presenting plasmacytoid dendritic cells mediate tolerance to vascularized grafts. Nat. Immunol 7, 652-662 (2006). 6. Irla, M. et al. MHC class II-restricted antigen presentation by plasmacytoid dendritic cells inhibits T cell-mediated autoimmunity. J. Exp. Med 207 , 1891-1905 (2010). 7. Kadowaki, N. et al. Subsets of human dendritic cell precursors express different toll-like receptors and respond to different microbial antigens. J. Exp. Med 194 , 863-869 (2001). 8. Kawai, T. & Akira, S. Innate immune recognition of viral infection. Nat. Immunol 7, 131-137 (2006). 9. Bauer, S. et al. Human TLR9 confers responsiveness to bacterial DNA via species-specific CpG motif recognition. Proc. Natl. Acad. Sci. U.S.A 98 , 9237- 9242 (2001). 10. Vollmer, J. et al. Characterization of three CpG oligodeoxynucleotide classes with distinct immunostimulatory activities. Eur. J. Immunol 34 , 251-262 (2004). 11. Park, B. et al. Proteolytic cleavage in an endolysosomal compartment is required for activation of Toll-like receptor 9. Nat. Immunol 9, 1407-1414 (2008). 12. McGettrick, A.F. & O'Neill, L.A.J. Localisation and trafficking of Toll-like receptors: an important mode of regulation. Curr. Opin. Immunol 22 , 20-27 (2010). 13. Akashi-Takamura, S. & Miyake, K. TLR accessory molecules. Curr. Opin. Immunol 20 , 420-425 (2008). 14. Kim, Y., Brinkmann, M.M., Paquet, M. & Ploegh, H.L. UNC93B1 delivers nucleotide-sensing toll-like receptors to endolysosomes. Nature 452 , 234-238 (2008). 15. Brinkmann, M.M. et al. The interaction between the ER membrane protein UNC93B and TLR3, 7, and 9 is crucial for TLR signaling. J. Cell Biol 177 , 265- 275 (2007). 16. Tabeta, K. et al. The Unc93b1 mutation 3d disrupts exogenous antigen presentation and signaling via Toll-like receptors 3, 7 and 9. Nat. Immunol 7, 156- 164 (2006). 17. Casrouge, A. et al. Herpes simplex virus encephalitis in human UNC-93B deficiency. Science 314 , 308-312 (2006). 18. Grouard, G. et al. The enigmatic plasmacytoid T cells develop into dendritic cells with interleukin (IL)-3 and CD40-ligand. J. Exp. Med 185 , 1101-1111 (1997). 19. Ito, T. et al. Plasmacytoid dendritic cells prime IL-10-producing T regulatory cells by inducible costimulator ligand. J. Exp. Med 204 , 105-115 (2007). 20. Villadangos, J.A. & Young, L. Antigen-presentation properties of plasmacytoid dendritic cells. Immunity 29 , 352-361 (2008). 21. Dzionek, A. et al. Plasmacytoid dendritic cells: from specific surface markers to specific cellular functions. Hum. Immunol 63 , 1133-1148 (2002).

22 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

22. Dzionek, A. et al. BDCA-2, BDCA-3, and BDCA-4: three markers for distinct subsets of dendritic cells in human peripheral blood. J. Immunol 165 , 6037-6046 (2000). 23. Marafioti, T. et al. Novel markers of normal and neoplastic human plasmacytoid dendritic cells. Blood 111 , 3778-3792 (2008). 24. Jacob, M.C. et al. CD4+ CD56+ lineage negative malignancies: a new entity developed from malignant early plasmacytoid dendritic cells. Haematologica 88 , 941-955 (2003). 25. Petrella, T. et al. TCL1 and CLA expression in agranular CD4/CD56 hematodermic neoplasms (blastic NK-cell lymphomas) and leukemia cutis. Am. J. Clin. Pathol 122 , 307-313 (2004). 26. Petrella, T. et al. Blastic NK-cell lymphomas (agranular CD4+CD56+ hematodermic neoplasms): a review. Am. J. Clin. Pathol 123 , 662-675 (2005). 27. Swerdlow, S. & International Agency for Research on Cancer.;World Health Organization. WHO classification of tumours of haematopoietic and lymphoid tissues . (International Agency for Research on Cancer: Lyon France, 2008). 28. David, A. et al. BAD-LAMP defines a subset of early endocytic organelles in subpopulations of cortical projection neurons. J. Cell. Sci 120 , 353-365 (2007). 29. Eskelinen, E., Tanaka, Y. & Saftig, P. At the acidic edge: emerging functions for lysosomal membrane proteins. Trends Cell Biol 13 , 137-145 (2003). 30. Saftig, P., Tanaka, Y., Lüllmann-Rauch, R. & von Figura, K. Disease model: LAMP-2 enlightens Danon disease. Trends Mol Med 7, 37-39 (2001). 31. Schuske, K., Palfreyman, M.T., Watanabe, S. & Jorgensen, E.M. UNC-46 is required for trafficking of the vesicular GABA transporter. Nat. Neurosci 10 , 846- 853 (2007). 32. Robbins, S.H. et al. Novel insights into the relationships between dendritic cell subsets in human and mouse revealed by genome-wide expression profiling. Genome Biol 9, R17 (2008). 33. Cappello, F. et al. Cystatin F is secreted, but artificial modification of its C- terminus can induce its endocytic targeting. Exp. Cell Res 297 , 607-618 (2004). 34. Lima, F.P. et al. Primary diffuse large B-cell lymphoma of bone displays preferential rearrangements of the c-MYC or BCL2 gene. Am. J. Clin. Pathol 129 , 723-726 (2008). 35. De Gassart, A. et al. MHC class II stabilization at the surface of human dendritic cells is the result of maturation-dependent MARCH I down-regulation. Proc. Natl. Acad. Sci. U.S.A 105 , 3491-3496 (2008). 36. Pfaffl, M.W., Horgan, G.W. & Dempfle, L. Relative expression software tool (REST) for group-wise comparison and statistical analysis of relative expression results in real-time PCR. Nucleic Acids Res 30 , e36 (2002). 37. Grage-Griebenow, E. et al. Anti-BDCA-4 (neuropilin-1) antibody can suppress virus-induced IFN-alpha production of plasmacytoid dendritic cells. Immunol. Cell Biol 85 , 383-390 (2007). 38. Traggiai, E. et al. Development of a human adaptive immune system in cord blood cell-transplanted mice. Science 304 , 104-107 (2004). 39. Li, S. et al. A map of the interactome network of the metazoan C. elegans. Science 303 , 540-543 (2004). 40. de Saint-Vis, B. et al. A novel lysosome-associated membrane glycoprotein, DC- LAMP, induced upon DC maturation, is transiently expressed in MHC class II compartment. Immunity 9, 325-336 (1998). 41. Salaun, B. et al. Cloning and characterization of the mouse homologue of the

23 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

human dendritic cell maturation marker CD208/DC-LAMP. Eur. J. Immunol 33 , 2619-2629 (2003).

24 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Figure legends

Figure 1: BAD-LAMP mRNA expression profile in Human

A. Gene chips quantitation of BAD-LAMP mRNA expression in human leukocytes.

Results are shown as fluorescent signal intensity for Affymetrix Human Genome

U133 PLUS 2.0 ProbeSet 219463_at (expressed in arbitrary units in log scale).

Quality controls, data sources and data normalization are described in Robbins SH et al., Genome Biology. 2008. Neu: neutrophils; pM φ: peripheral blood mononuclear cell-derived macrophages; Mo-M φ: monocyte-derived macrophages; Mo-DC: monocyte-derived GM-CSF+IL-4 DC; CD16 DC: blood Lin-HLA -DR +CD16 + DC;

BDCA1 DC: blood BDCA-1 + DC; BDCA3 DC: blood BDCA-3 + DC; pDC: blood plasmacytoid DC; BL: blood B lymphocytes; CD4 TL: blood CD4 + T lymphocytes;

CD8 TL: blood CD8 + T lymphocytes; NK cells: blood natural killer cells. B. Tissue expression of BAD-LAMP assessed by Northern Blot. A signal is detected only in adult human brain. Actin mRNA levels are shown as control. C. Detection of BAD-

LAMP transcript in human spleen RNA total extracts by nested RT-PCR.

Figure 2: BAD-LAMP is detected specifically in pDCs

A. Detection of BAD-LAMP in human lymphoid tissue. Frozen human spleen sections were stained with monoclonal antibodies against CD123 (red) and BAD-LAMP

(green). Overlay show that BAD-LAMP + cells are also CD123 + (merge, yellow). Bar

20µm. Paraffin-fixed human tonsil germinative center sections were stained in IHC

(lower right). BAD-LAMP + cells display a pDCs morphology (arrows) next to HEV. B.

Intracellular FACS staining on human PBMCs. A rare cell population can be isolated based on BAD-LAMP expression (left). BAD-LAMP + cells were identified as pDCs

25 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC based on BDCA-4 expression (right). C. Detection of BAD-LAMP transcript in primary blood cells mRNA extracts by RT-PCR. A signal is detected only in pDCs and not in the pDC-depleted monocytes fractions. RT-PCR was performed for 20 or 35 cycles after mRNA extraction. Actin mRNA levels are shown as control. D. BAD-LAMP staining on purified pDCs. Cells stained in intracellular FACS (left) are homogenously

BAD-LAMP + (solid line) as compared to isotype control background (filled graph).

BAD-LAMP is localized in intracellular membranes of BDCA-4 + pDCs as shown in microscopy experiments (green,right). Nucleus (Nu) staining is shown in blue. Bar:

20µm.

Figure 3: Regulation of BAD-LAMP during pDCs activation

A. BAD-LAMP detection by immunoblot. Cell lysates from different cell types were separated by SDS-PAGE and revealed using mAb against BAD-LAMP. A single specific band is detected in pDC extracts around 35 kDa and not in immature monocyte-derived dendritic cells (MoDCs i), LPS-activated MoDCs (MoDCs m) nor in total PBMCs. HeLa cells transfected with BAD-LAMP cDNA (HeLa BAD) and control

(HeLa nt) were used as a positive control both for specificity and as a reference for the glycosylation pattern. Asterisk (*) marked lanes were loaded with a lower amount of total proteins to compensate for the high BAD-LAMP expression levels in transfected cells. Actin levels are shown as loading controls. B. BAD-LAMP mRNA levels are down-regulated upon CpG activation. Purified pDCs were cultivated for 6h or 24h in presence of IL-3 and stimulated or not with A-, B- or C-type CpG ODNs.

BAD-LAMP mRNA levels were determined using quantitative RT-PCR and results were normalized against the IL-3 only condition. Results are from one representative experiment. C. BAD-LAMP is down-regulated at the protein level upon CpG

26 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC activation. After 24h of culturing freshly isolated pDCs (filled graph) with IL-3 (solid black line) and A-type CpG ODN (dashed gray line), BAD-LAMP expression monitored by intracellular FACS staining is down-regulated pDCs. IL-3 treatment is sufficient to decrease BAD-LAMP levels. D. BAD-LAMP is no longer detectable by immunoblot in pDCs after 24h of A-type CpG ODN stimulation. Low amounts of HeLa

BAD and HeLa nt (*) were used as specificity control. Actin levels are shown as loading controls.

Figure 4: BAD-LAMP is a marker of blastic pDC neoplasms

A. IHC on paraffin sections of skin lesions from patients with BPDCN reveal a massive infiltration of BAD-LAMP + cells (arrows). B. A larger scale analysis by tissue arrays revealed that >78% of biopsies were BAD-LAMP + among 33 patients diagnosed with a CD4 +/CD56 + malignancy (left). An example of a BAD-LAMP + biopsy from the tissue array is shown (right).

Figure 5: BAD-LAMP is localized in the endoplasmic reticulum

A. Immunofluorescence staining for BAD-LAMP in purified pDCs. BAD-LAMP (green, upper panels) co-staining with early endosomal marker transferrin receptor (TfR, red) and lysosomal marker LAMP1 (blue) show no overlap. BAD-LAMP (green, lower panels) and endoplasmic reticulum marker KDEL (red) have similar intracellular distribution and display partial co-localization (arrow). Bar: 10µm. B. Analysis of BAD-

LAMP glycosylation by enzymatic treatments. Immunoprecipitation from pDC lysate and subsequent endoglycosidase H (EndoH) treatment reveals that BAD-LAMP glycosylation remains endo H-sensitive. Total lysate and antibodies alone (Ab) are shown as controls. C. Confocal microscopy of BAD-LAMP heterologous expression

27 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC in human monocyte-derived DCs. 6h after transfection, BAD-LAMP (green) and endoplasmic reticulum resident PDI (red) display extensive co-localization. Bar:

10µm.

Figure 6: BAD-LAMP co-localizes with UNC93B1 in transfected HeLa cells

A. BAD-LAMP and UNC93B1 have different cellular localization when over-

expressed together in HeLa cells. BAD-LAMP (green, top) is mainly targeted

to the plasma membrane with a small portion in endocytic compartments.

UNC93B1 (red, bottom) is localized in the ER. Bar: 20µm. B. When co-

expressed, BAD-LAMP (green) and UNC93B1 (red) co-localize in large

endosomal intracellular vesicles (upper panels, arrows). On the contrary,

upon expression of the structurally related endosomal resident DC-LAMP

(green) intracellular trafficking UNC93B1 (red) remains unchanged (lower

panels). C. Flag-tagged BAD-LAMP mutants have different sorting behaviors.

Flag-BAD-LAMP (wt) is targeted to the cell surface and partially to

endosomes (green, left panels), while the Flag-BAD-LAMP Y276A mutant is

almost exclusively localized at the plasma membrane. Flag-BAD-LAMP ∆Ct

mutant is retained in the endoplasmic reticulum. Upon co transfection with

His-UNC93B1 (red, right panels), all the different flag-tagged forms of BAD-

LAMP (green, right panels) are sorted together with His-UNC93B1 (red) in the

same intracellular endosomal compartments (arrows).

Figure 7: A. Immunofluorescence confocal microscopy of HeLa cells

transfected with an eGFP-tagged BAD-LAMP fusion. BAD-LAMP-GFP

(green) is sorted to intracellular compartments that are mostly LAMP1 + (blue,

28 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

arrow, (upper panels). In cells co-transfected with BAD-LAMP-GFP and

UNC93B1 (red, lower panels), the two molecules are sorted together in

LAMP1 + intracellular compartments. B. Immunofluorescence confocal

microscopy of HeLa cells co-transfected with BAD-LAMP (green), UNC93B1

(red) and an HA-tagged TLR9 (blue). UNC93B1 can co-localize with BAD-

LAMP in intracellular compartments (+) or with TLR9 in ER (*) depending on

the relative expression levels of the three transfected proteins.

29 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Supplementary figures legends

Figure S1: A. Human BAD-LAMP protein sequence. Cysteines in red form di-sulfide bonds, resulting in the formation of the two loops of the LAMP domain. The residues of the transmembrane domain are shown in blue and underlined. The tyrosine-based sorting signal Yxx φ is shown in green. The three putative N-glycosylation sites are squared in grey. The 34.2 mAb epitope is underlined and marked with an asterisk (*).

B. HeLa cells transfected with an eGFP-tagged version of BAD-LAMP were stained with the 34.2 monoclonal antibody against BAD-LAMP.

Figure S2: 6330527O06Rik/BAD-LAMP (Entrez Gene ID: 76161, probe set:

1423853_at) expression profile in mouse according to the Gene Expression Omnibus

(GEO) database. BAD-LAMP is not detectable in any immune tissue.

Figure S3: A. IHC on frozen human tonsil sections. Staining show BAD-LAMP+ cells (green) are also CD123 + (red) and CD4 + (blue), consistent with a pDC-restricted expression. Bar: 10µm. B. Immunofluorescence confocal microscopy of pDCs from

γc/RAG -/- mice reconstituted with CD34+ human hematopoietic stem cells. BDCA-4 + cells were sorted magnetically from splenocytes and stained for BAD-LAMP (red) and

BDCA-4 (green). Bar: 10µm.

Figure S4: Immunofluorescence confocal microscopy of purified pDCs. BAD-LAMP

(green) show no co-localization with late endocytic markers CD63 (red) and LAMP1

(blue, top panels), MHC class II molecule HLA DR (red) and MHC class I molecules

HLA A,B,C (blue middle panels), or cis -Golgi marker GM130 (red, bottom panels).

Bar: 10µm.

30 DEFAYS et al. BAD-LAMP IS A NOVEL MARKER OF HUMAN PDC

Figure S5: A. UNC93B1 (Entrez Gene ID: 81622, probe set: 225869_s_at) expression profile in human cells according to the Gene Expression Omnibus (GEO) database and Robbins SH et al., Genome Biology. 2008. UNC93B1 is predominantly expressed in DCs compared to other leukocytes, and within the different DC subsets, pDCs display the higher UNC93B1 expression levels. NK: blood natural killer cells;

CD8: blood CD8 + T lymphocytes; CD4: blood CD4 + T lymphocytes; B: blood B lymphocytes; pDC: blood plasmacytoid DC; BDCA1: blood BDCA-1 + DC; BDCA3: blood BDCA-3 + DC; CD16: blood Lin -HLA -DR +CD16 + DC; Mono: blood CD14 + monocytes; Neu: neutrophils. B. Detection of UNC93B1 in HeLa cells. No expression of UNC93B1 could be observed by RT-PCR (left) in HeLa cells, as opposed to monocyte-derived DCs used as a positive control. Actin levels are shown as control.

Tumor array Table I

1 TMA Lymphoma T (15 peripheral T lymphoma NOS, 10 angio-immunoblastic T lymphoma, 3 enteropathy T-cell lymphoma, 8 mycosis fungoides, 3 hepato-splenic gamma delta T-cell lymphoma, 5 nasal NK/T cell lymphoma, 10 ALK+ anaplastic large cell lymphoma. 1 TMA B Lymphoma : 10 Large cells B lymphoma ,10 Folicular lymphoma, 10 B-CLL, 10 mantle cell lymphoma, 10 marginal zone lymphoma

(splenic), 5 marginal zone lymphoma (Lymph node), 5 LPL. 1 TAM Hodgkin : 25 nodular lymphocyte predominance HL, 25 nodular sclerosing, 15 mixed cellularity, 5 lymphocytes depletion. TMA BPDCN (33)

31

Discussion

BAD-LAMP appartient à la famille des LAMPs

BAD-LAMP est la dernière molécule appartenant à la famille des LAMP identifiée et caractérisée à ce jour. Cette protéine transmembranaire possède en commun avec les autres LAMPs un domaine cytoplasmique contenant un motif d’adressage basé sur une tyrosine de !"# $%&''
% $!% ) $*+%, oucles formées par des ponts disulfure, caractéristiques du domaine LAMP. Elle possède en revanche plusieurs caractéristiques qui la rendent unique. Toutes les autres LAMP possèdent une partie cytoplasmique composé de deux domaines séparés par une région charnière riche en prolines et en sérines comportant de nombreuses glycosylations-O. BAD-LAMP se démarque par une taille plus réduite et une partie cytoplasmique qui ne comporte qu’un seul domaine. Cette structure atypique la rapproche ainsi plus de CD68 et DC-LAMP, dont un seul de leurs domaines cytoplasmiques est structuré en boucles LAMP, que de LAMP1 et LAMP2. De manière intéressante, BAD-LAMP possède un profil d’expression très spécifique, étant exprimé uniquement dans une sous-population de neurones et dans les pDCs. Ce profil rappelle DC-LAMP, exprimée seulement dans les pneumocytes de type II et les cDCs matures, et dans une moindre mesure de CD68, fortement enrichie dans les monocytes et macrophages, alors que LAMP1 et LAMP2 ont un profil d’expression très large. Les alignements de séquence réalisés entre les différentes LAMPs suggèrent également que BAD-LAMP est le membre de la famille qui a divergé le plus tôt au cours de l’évolution. La présence chez C. elegans de deux LAMP, l’une homologue de BAD-LAMP et l’autre de LAMP1, renforce l’hypothèse d’une divergence et d’une spécialisation ancienne.

Comparaison entre les formes murines et humaines de BAD-LAMP

Cette étude a été initiée pour réaliser la caractérisation fonctionnelle de BAD-LAMP chez l’Homme. Pour mener à bien ce projet, une étude préliminaire a été réalisée chez la souris. Il existe en effet une homologie certaine entre les formes humaine et murine de BAD- LAMP, les séquences protéiques présentant 83% d’identités. La parenté entre les deux molécules est telle que leur domaine cytoplasmique contenant le motif d’adressage basé sur la tyrosine est absolument identique et les substitutions ne concernent que des acides aminés ayant des propriétés équivalentes. Cette forte similarité suggère une fonction équivalente chez les deux espèces. Les travaux réalisés chez la souris ont mis en évidence que l’expression de

39 BAD-LAMP est restreinte à une sous-population de neurones corticaux. BAD-LAMP n’est pas détecté dans les lysosomes, mais uniquement dans un type d’endosomes encore non- caractérisés. Trois formes différentes de la protéine sont détectées, chacune ayant un profil de glycosylation-N différent et dont une forme est sensible à la digestion par l’enzyme endoglycosidase H.

Chez l’Homme, BAD-LAMP est exprimée de manière spécifique dans les pDCs en plus du cerveau, marquant une première différence avec son homologue murin. Dans les pDCs humaines, BAD-LAMP n’est pas non plus adressée vers les lysosomes, et réside dans un compartiment ne contenant aucun marqueur des endosomes classiquement utilisés. BAD- LAMP co-localise toutefois partiellement avec l’épitope « KDEL », un signal de rétention dans le RE. Ce résultat, qui tend à démontrer une localisation dans le RE, est renforcé par l’analyse du profil de glycosylation. Il n’existe qu’une seule forme de BAD-LAMP, qui porte une glycosylation-N sensible à la digestion par l’enzyme endoglycosidase H, correspondant à un oligosaccharide de type riche en mannose ou hybride, caractéristique des protéines n’ayant pas traversé l’appareil de Golgi. La localisation et le profil de glycosylation-N sont tous deux différents de ceux observés dans les neurones chez la souris, indiquant une régulation spécifique de BAD-LAMP chez l’Homme par rapport à la souris ou dans les pDCs par rapport aux neurones. Il serait intéressant d’étudier la régulation de l’adressage de BAD-LAMP dans les neurones humains et de la comparer avec les données obtenues dans les neurones murins et les pDCs humaines. Cependant, les adressages différents observés selon que la molécule est transfectée dans une lignée cellulaire HeLa ou des MoDCs humaines suggèrent qu’il s’agirait plutôt d’une spécificité des pDCs. La rétention de BAD-LAMP dans le RE pourrait être partiellement dépendante d’une interaction directe ou indirecte avec la protéine UNC93B1, exprimée dans les pDCs et les MoDCs.

Une protéine « neuronale » exprimée dans les pDCs

L’étude du profil d’expression de BAD-LAMP chez l’Homme a révélé la présence du transcrit dans le cerveau. Les formes homologues exprimées chez la souris et chez le nématode C. elegans sont, elles, exclusivement neuronales. L’expression d’une protéine « neuronale » dans les pDCs et plus largement dans des cellules du système immunitaire n’est pas un cas isolé. Une autre protéine considérée comme spécifique des neurones a été identifiée dans les pDCs humaines grâce au profilage génétique des populations leucocytaires, 40 la PACSIN1 (Annexe 2). La PACSIN1 est, dans les neurones, impliquée dans le recyclage des vésicules synaptiques à partir de la membrane plasmique, grâce à une interaction avec la GTPase dynamine 1 219 . Il est intéressant de constater qu’une protéine nécessaire pour un mécanisme spécialisé d’endocytose soit exprimée à un niveau élevé dans un type cellulaire ayant des capacités d’internalisation et de phagocytose réduites 88 . Un autre exemple vient de la protéine NCAM/CD56, qui est une molécule d’adhérence impliquée entre autres dans le développement du système nerveux et la plasticité synaptique 220 . NCAM est exprimé normalement par les cellules NK, et aussi par différents types de cancers, dont les cancers d’origine plasmacytoïde CD4 +/CD56 + 221 . Si le rôle précis de NCAM dans le processus cancéreux n’est pas encore bien établi, son expression semble être corrélée avec une forme aigüe et un taux de survie plus faible, indépendamment du type de cancer considéré.

Le cas opposé, des molécules « immunitaires » exprimées dans des neurones, est également documenté. Un exemple intéressant vient de la mise en évidence de l’expression de molécules du CMH I dans les neurones en condition physiologique 222 . Les molécules du CMH I, chargées avec des peptides, sont notamment impliquées dans la construction du réseau neuronal et dans la dynamique de formation des synapses. Les complexes du CMH I sont reconnus par plusieurs récepteurs différents, parmi lesquels les récepteurs Ly49 223 ou KIR, habituellement associés aux cellules NK 224 . Il existe encore de nombreux exemples de protéines qui sont exprimées dans deux ou plusieurs types cellulaires ou tissus différents. Ces protéines peuvent, tout en gardant la même séquence protéique, remplir des fonctions complètement différentes en fonction du contexte dans lequel elles sont exprimées. Les différences observées dans la régulation de BAD-LAMP entre tous les différents modèles que utilisés suggèrent que la fonction de BAD-LAMP pourrait n’être élucidée que par une étude réalisée uniquement sur des pDCs humaines.

BAD-LAMP est un marqueur spécifique des pDCs immatures et néoplasiques

Lors de cette étude, nous avons développé des anticorps monoclonaux dirigés contre la partie cytoplasmique de BAD-LAMP. Ces anticorps nous ont permis de marquer de manière spécifique les pDCs par des techniques d’immunohistochimie, d’immunocytochimie, d’immunofluorescence et de cytométrie de flux. Nous avons également déterminé que

41 l’expression de BAD-LAMP est régulée négativement rapidement après la stimulation des pDCs avec des ODN CpG, au niveau du transcrit et de la protéine. Cette régulation est indépendante du type d’ODN CpG, et n’est donc probablement pas directement liée aux capacités de production d’IFN des cellules. Toutes ces données nous permettent d’établir que BAD-LAMP représente un nouveau marqueur des pDCs, dans la circulation sanguine et dans les tissus.

L’expression de BAD-LAMP a également été testée dans les néoplasmes hématodermiques de type CD4 +/CD56 +. Ces cellules tumorales partagent de nombreuses caractéristiques avec les pDCs, dont un phénotype Lin -CD4 +CD123 +, l’expression de BDCA- 2 et la capacité de produire de l’IFN de type-1 démontrée chez certains patients. Sur la base de ces observation, l’origine plasmacytoïde des cellules tumorales CD4 +/CD56+ est largement acceptée. L’expression de BAD-LAMP a été détectée sur une majorité de tumeurs, renforçant l’hypothèse d’une origine plasmacytoïde de ces cellules. Ce résultat montre également que la détection de BAD-LAMP peut être envisagée à terme dans un processus de diagnostic pour cette pathologie. Il est intéressant de constater que l’expression de BDCA-2 n’est pas détectée chez tous les patients atteints de néoplasme plasmacytoïde 225 , mais également que le niveau d’expression de BDCA-2 semble corréler avec un taux de survie plus faible 226 . Le niveau de BDCA-2 pourrait ainsi être un indicateur du niveau d’activation des pDCs transformées. Il serait intéressant de déterminer si le niveau d’expression de BAD-LAMP dans les tumeurs plasmacytoïdes peut être corrélé avec le niveau de BDCA-2 ou une différence dans le pronostic vital, et ainsi permettre d’améliorer le diagnostic pour les patients.

BAD-LAMP et UNC93B1 influencent leurs adressages respectifs L’adressage de BAD-LAMP dans un système de cellules HeLa transfectées est totalement différent de celui observé dans les pDCs, la molécule s’accumulant à la membrane plasmique et recyclant dans les compartiments endocytiques précoces. La protéine UNC93B1, qui réside normalement dans le RE, reste localisée majoritairement dans le RE lorsqu’elle est surexprimée dans des cellules HeLa, avec une fraction des molécules localisée dans les endosomes. De manière surprenante, BAD-LAMP et UNC93B1 s’accumulent ensembles dans un compartiment intracellulaire, dont la nature reste inconnue, lorsque les deux molécules sont co-exprimées dans des cellules HeLa. Des formes mutantes de BAD-LAMP avec des propriétés d’adressage différentes ont alors été créées en modifiant ou retirant la queue 42 cytoplasmique de la protéine. Toutes les formes mutantes de BAD-LAMP s’accumulent dans dans des compartiments contenant UNC93B1 lorsque les molécules sont co-exprimées, indépendamment de leurs propriétés d’adressage. Ces résultats suggèrent que BAD-LAMP et UNC93B1 sont capables de s’influencer mutuellement dans un système en surexpression, même si aucune interaction directe n’a pu être mise en évidence. Il est important de déterminer par quel mécanisme BAD-LAMP et UNC93B1 s’influencent, ainsi que la nature des compartiments cellulaires dans lesquels ils s’accumulent.

43 Références

1. Terasaki, M. & Jaffe, L.A. Organization of the sea urchin egg endoplasmic reticulum and its reorganization at fertilization. J. Cell Biol 114 , 929-940 (1991). 2. Voeltz, G.K., Rolls, M.M. & Rapoport, T.A. Structural organization of the endoplasmic reticulum. EMBO Rep 3, 944-950 (2002). 3. Rolls, M.M., Hall, D.H., Victor, M., Stelzer, E.H.K. & Rapoport, T.A. Targeting of rough endoplasmic reticulum membrane proteins and ribosomes in invertebrate neurons. Mol. Biol. Cell 13 , 1778-1791 (2002). 4. Walter, P., Ibrahimi, I. & Blobel, G. Translocation of proteins across the endoplasmic reticulum. I. Signal recognition protein (SRP) binds to in-vitro-assembled polysomes synthesizing secretory protein. J. Cell Biol 91 , 545-550 (1981). 5. Matlack, K.E., Mothes, W. & Rapoport, T.A. Protein translocation: tunnel vision. Cell 92 , 381-390 (1998). 6. Johnson, A.E. & van Waes, M.A. The translocon: a dynamic gateway at the ER membrane. Annu. Rev. Cell Dev. Biol 15 , 799-842 (1999). 7. Chen, W., Helenius, J., Braakman, I. & Helenius, A. Cotranslational folding and calnexin binding during glycoprotein synthesis. Proc. Natl. Acad. Sci. U.S.A 92 , 6229- 6233 (1995). 8. Ellgaard, L. & Ruddock, L.W. The human protein disulphide isomerase family: substrate interactions and functional properties. EMBO Rep 6, 28-32 (2005). 9. Flynn, G.C., Pohl, J., Flocco, M.T. & Rothman, J.E. Peptide-binding specificity of the molecular chaperone BiP. Nature 353 , 726-730 (1991). 10. Wilson, C.M., Roebuck, Q. & High, S. Ribophorin I regulates substrate delivery to the oligosaccharyltransferase core. Proc. Natl. Acad. Sci. U.S.A 105 , 9534-9539 (2008). 11. Bannykh, S.I., Rowe, T. & Balch, W.E. The organization of endoplasmic reticulum export complexes. J. Cell Biol 135 , 19-35 (1996). 12. Lee, M.C.S. et al. Sar1p N-terminal helix initiates membrane curvature and completes the fission of a COPII vesicle. Cell 122 , 605-617 (2005). 13. Matsuoka, K. et al. COPII-coated vesicle formation reconstituted with purified coat proteins and chemically defined liposomes. Cell 93 , 263-275 (1998). 14. Springer, S. & Schekman, R. Nucleation of COPII vesicular coat complex by endoplasmic reticulum to Golgi vesicle SNAREs. Science 281 , 698-700 (1998). 15. Stagg, S.M. et al. Structure of the Sec13/31 COPII coat cage. Nature 439 , 234-238 (2006). 16. Horstmann, H., Ng, C.P., Tang, B.L. & Hong, W. Ultrastructural characterization of endoplasmic reticulum--Golgi transport containers (EGTC). J. Cell. Sci 115 , 4263-4273 (2002). 17. Orci, L. et al. Bidirectional transport by distinct populations of COPI-coated vesicles. Cell 90 , 335-349 (1997). 18. Presley, J.F. et al. Dissection of COPI and Arf1 dynamics in vivo and role in Golgi membrane transport. Nature 417 , 187-193 (2002). 19. Eugster, A., Frigerio, G., Dale, M. & Duden, R. COP I domains required for coatomer integrity, and novel interactions with ARF and ARF-GAP. EMBO J 19 , 3905-3917 (2000). 20. Lewis, M.J. & Pelham, H.R. A human homologue of the yeast HDEL receptor. Nature

44 348 , 162-163 (1990). 21. Raykhel, I. et al. A molecular specificity code for the three mammalian KDEL receptors. J. Cell Biol 179 , 1193-1204 (2007). 22. Mellman, I. & Warren, G. The road taken: past and future foundations of membrane traffic. Cell 100 , 99-112 (2000). 23. Glick, B.S. & Malhotra, V. The curious status of the Golgi apparatus. Cell 95 , 883-889 (1998). 24. Emr, S. et al. Journeys through the Golgi--taking stock in a new era. J. Cell Biol 187 , 449-453 (2009). 25. Martinez-Menárguez, J.A. et al. Peri-Golgi vesicles contain retrograde but not anterograde proteins consistent with the cisternal progression model of intra-Golgi transport. J. Cell Biol 155 , 1213-1224 (2001). 26. Gilchrist, A. et al. Quantitative proteomics analysis of the secretory pathway. Cell 127 , 1265-1281 (2006). 27. Trucco, A. et al. Secretory traffic triggers the formation of tubular continuities across Golgi sub-compartments. Nat. Cell Biol 6, 1071-1081 (2004). 28. Pelham, H.R. & Rothman, J.E. The debate about transport in the Golgi--two sides of the same coin? Cell 102 , 713-719 (2000). 29. Kornfeld, R. & Kornfeld, S. Assembly of asparagine-linked oligosaccharides. Annu. Rev. Biochem 54 , 631-664 (1985). 30. Lis, H. & Sharon, N. Protein glycosylation. Structural and functional aspects. Eur. J. Biochem 218 , 1-27 (1993). 31. Van den Steen, P., Rudd, P.M., Dwek, R.A. & Opdenakker, G. Concepts and principles of O-linked glycosylation. Crit. Rev. Biochem. Mol. Biol 33 , 151-208 (1998). 32. Silberstein, S. & Gilmore, R. Biochemistry, molecular biology, and genetics of the oligosaccharyltransferase. FASEB J 10 , 849-858 (1996). 33. Hammond, C., Braakman, I. & Helenius, A. Role of N-linked oligosaccharide recognition, glucose trimming, and calnexin in glycoprotein folding and quality control. Proc. Natl. Acad. Sci. U.S.A 91 , 913-917 (1994). 34. Rodan, A.R., Simons, J.F., Trombetta, E.S. & Helenius, A. N-linked oligosaccharides are necessary and sufficient for association of glycosylated forms of bovine RNase with calnexin and calreticulin. EMBO J 15 , 6921-6930 (1996). 35. Nilsson, T. et al. Overlapping distribution of two glycosyltransferases in the Golgi apparatus of HeLa cells. J. Cell Biol 120 , 5-13 (1993). 36. Rabouille, C. et al. Mapping the distribution of Golgi enzymes involved in the construction of complex oligosaccharides. J. Cell. Sci 108 ( Pt 4) , 1617-1627 (1995). 37. O'Neill, R.A. Enzymatic release of oligosaccharides from glycoproteins for chromatographic and electrophoretic analysis. J Chromatogr A 720 , 201-215 (1996). 38. Hirschberg, K. et al. Kinetic analysis of secretory protein traffic and characterization of golgi to plasma membrane transport intermediates in living cells. J. Cell Biol 143 , 1485- 1503 (1998). 39. Polishchuk, E.V., Di Pentima, A., Luini, A. & Polishchuk, R.S. Mechanism of constitutive export from the golgi: bulk flow via the formation, protrusion, and en bloc cleavage of large trans-golgi network tubular domains. Mol. Biol. Cell 14 , 4470-4485 (2003). 40. Tooze, J. & Tooze, S.A. Clathrin-coated vesicular transport of secretory proteins during the formation of ACTH-containing secretory granules in AtT20 cells. J. Cell Biol 103 , 839-850 (1986). 45 41. Puertollano, R. et al. Morphology and dynamics of clathrin/GGA1-coated carriers budding from the trans-Golgi network. Mol. Biol. Cell 14 , 1545-1557 (2003). 42. Shiba, T. et al. Structural basis for recognition of acidic-cluster dileucine sequence by GGA1. Nature 415 , 937-941 (2002). 43. Bonifacino, J.S. & Traub, L.M. Signals for sorting of transmembrane proteins to endosomes and lysosomes. Annu. Rev. Biochem 72 , 395-447 (2003). 44. Sorkin, A. Cargo recognition during clathrin-mediated endocytosis: a team effort. Curr. Opin. Cell Biol 16 , 392-399 (2004). 45. Traub, L.M. Sorting it out: AP-2 and alternate clathrin adaptors in endocytic cargo selection. J. Cell Biol 163 , 203-208 (2003). 46. Hofmann, K. & Falquet, L. A ubiquitin-interacting motif conserved in components of the proteasomal and lysosomal protein degradation systems. Trends Biochem. Sci 26 , 347-350 (2001). 47. Komander, D. The emerging complexity of protein ubiquitination. Biochem. Soc. Trans 37 , 937-953 (2009). 48. Acconcia, F., Sigismund, S. & Polo, S. Ubiquitin in trafficking: the network at work. Exp. Cell Res 315 , 1610-1618 (2009). 49. van der Bliek, A.M. et al. Mutations in human dynamin block an intermediate stage in coated vesicle formation. J. Cell Biol 122 , 553-563 (1993). 50. McNiven, M.A. Dynamin: a molecular motor with pinchase action. Cell 94 , 151-154 (1998). 51. Rothberg, K.G. et al. Caveolin, a protein component of caveolae membrane coats. Cell 68 , 673-682 (1992). 52. Monier, S., Dietzen, D.J., Hastings, W.R., Lublin, D.M. & Kurzchalia, T.V. Oligomerization of VIP21-caveolin in vitro is stabilized by long chain fatty acylation or cholesterol. FEBS Lett 388 , 143-149 (1996). 53. Pelkmans, L., Kartenbeck, J. & Helenius, A. Caveolar endocytosis of simian virus 40 reveals a new two-step vesicular-transport pathway to the ER. Nat. Cell Biol 3, 473-483 (2001). 54. Glebov, O.O., Bright, N.A. & Nichols, B.J. Flotillin-1 defines a clathrin-independent endocytic pathway in mammalian cells. Nat. Cell Biol 8, 46-54 (2006). 55. Sabharanjak, S., Sharma, P., Parton, R.G. & Mayor, S. GPI-anchored proteins are delivered to recycling endosomes via a distinct cdc42-regulated, clathrin-independent pinocytic pathway. Dev. Cell 2, 411-423 (2002). 56. Mellman, I. Endocytosis and molecular sorting. Annu. Rev. Cell Dev. Biol 12 , 575-625 (1996). 57. Rink, J., Ghigo, E., Kalaidzidis, Y. & Zerial, M. Rab conversion as a mechanism of progression from early to late endosomes. Cell 122 , 735-749 (2005). 58. Bock, J.B., Matern, H.T., Peden, A.A. & Scheller, R.H. A genomic perspective on membrane compartment organization. Nature 409 , 839-841 (2001). 59. Mellman, I., Fuchs, R. & Helenius, A. Acidification of the endocytic and exocytic pathways. Annu. Rev. Biochem 55 , 663-700 (1986). 60. Haglund, K., Di Fiore, P.P. & Dikic, I. Distinct monoubiquitin signals in receptor endocytosis. Trends Biochem. Sci 28 , 598-603 (2003). 61. Sheff, D.R., Daro, E.A., Hull, M. & Mellman, I. The receptor recycling pathway contains two distinct populations of early endosomes with different sorting functions. J. Cell Biol 145 , 123-139 (1999). 62. Kornfeld, S. Structure and function of the mannose 6-phosphate/insulinlike growth 46 factor II receptors. Annu. Rev. Biochem 61 , 307-330 (1992). 63. Puertollano, R., Aguilar, R.C., Gorshkova, I., Crouch, R.J. & Bonifacino, J.S. Sorting of mannose 6-phosphate receptors mediated by the GGAs. Science 292 , 1712-1716 (2001). 64. Raiborg, C., Rusten, T.E. & Stenmark, H. Protein sorting into multivesicular endosomes. Curr. Opin. Cell Biol 15 , 446-455 (2003). 65. Dugast, M., Toussaint, H., Dousset, C. & Benaroch, P. AP2 clathrin adaptor complex, but not AP1, controls the access of the major histocompatibility complex (MHC) class II to endosomes. J. Biol. Chem 280 , 19656-19664 (2005). 66. Bright, N.A., Gratian, M.J. & Luzio, J.P. Endocytic delivery to lysosomes mediated by concurrent fusion and kissing events in living cells. Curr. Biol 15 , 360-365 (2005). 67. Kolter, T. & Sandhoff, K. Lysosomal degradation of membrane lipids. FEBS Lett 584 , 1700-1712 (2010). 68. Eskelinen, E., Tanaka, Y. & Saftig, P. At the acidic edge: emerging functions for lysosomal membrane proteins. Trends Cell Biol 13 , 137-145 (2003). 69. Chen, J.W., Murphy, T.L., Willingham, M.C., Pastan, I. & August, J.T. Identification of two lysosomal membrane glycoproteins. J. Cell Biol 101 , 85-95 (1985). 70. Fukuda, M. Lysosomal membrane glycoproteins. Structure, biosynthesis, and intracellular trafficking. J. Biol. Chem 266 , 21327-21330 (1991). 71. Starcevic, M., Nazarian, R. & Dell'Angelica, E.C. The molecular machinery for the biogenesis of lysosome-related organelles: lessons from Hermansky-Pudlak syndrome. Semin. Cell Dev. Biol 13 , 271-278 (2002). 72. Betts, M.R. et al. Sensitive and viable identification of antigen-specific CD8+ T cells by a flow cytometric assay for degranulation. J. Immunol. Methods 281 , 65-78 (2003). 73. Konecki, D.S., Foetisch, K., Zimmer, K.P., Schlotter, M. & Lichter-Konecki, U. An alternatively spliced form of the human lysosome-associated membrane protein-2 gene is expressed in a tissue-specific manner. Biochem. Biophys. Res. Commun 215 , 757-767 (1995). 74. Lichter-Konecki, U. et al. Expression patterns of murine lysosome-associated membrane protein 2 (Lamp-2) transcripts during morphogenesis. Differentiation 65 , 43-58 (1999). 75. Andrejewski, N. et al. Normal lysosomal morphology and function in LAMP-1-deficient mice. J. Biol. Chem 274 , 12692-12701 (1999). 76. Tanaka, Y. et al. Accumulation of autophagic vacuoles and cardiomyopathy in LAMP-2- deficient mice. Nature 406 , 902-906 (2000). 77. Eskelinen, E. et al. Disturbed cholesterol traffic but normal proteolytic function in LAMP-1/LAMP-2 double-deficient fibroblasts. Mol. Biol. Cell 15 , 3132-3145 (2004). 78. Huynh, K.K. et al. LAMP proteins are required for fusion of lysosomes with phagosomes. EMBO J 26 , 313-324 (2007). 79. Holness, C.L. & Simmons, D.L. Molecular cloning of CD68, a human macrophage marker related to lysosomal glycoproteins. Blood 81 , 1607-1613 (1993). 80. Ramprasad, M.P. et al. The 94- to 97-kDa mouse macrophage membrane protein that recognizes oxidized low density lipoprotein and phosphatidylserine-rich liposomes is identical to macrosialin, the mouse homologue of human CD68. Proc. Natl. Acad. Sci. U.S.A 92 , 9580-9584 (1995). 81. Ramprasad, M.P., Terpstra, V., Kondratenko, N., Quehenberger, O. & Steinberg, D. Cell surface expression of mouse macrosialin and human CD68 and their role as macrophage receptors for oxidized low density lipoprotein. Proc. Natl. Acad. Sci. U.S.A 93 , 14833- 14838 (1996). 82. Plüddemann, A., Neyen, C. & Gordon, S. Macrophage scavenger receptors and host- 47 derived ligands. Methods 43 , 207-217 (2007). 83. de Saint-Vis, B. et al. A novel lysosome-associated membrane glycoprotein, DC-LAMP, induced upon DC maturation, is transiently expressed in MHC class II compartment. Immunity 9, 325-336 (1998). 84. Salaun, B. et al. CD208/dendritic cell-lysosomal associated membrane protein is a marker of normal and transformed type II pneumocytes. Am. J. Pathol 164 , 861-871 (2004). 85. Facchetti, F., Vermi, W., Mason, D. & Colonna, M. The plasmacytoid monocyte/interferon producing cells. Virchows Arch 443 , 703-717 (2003). 86. Müller-Hermelink, H.K., Stein, H., Steinmann, G. & Lennert, K. Malignant lymphoma of plasmacytoid T-cells. Morphologic and immunologic studies characterizing a special type of T-cell. Am. J. Surg. Pathol 7, 849-862 (1983). 87. Facchetti, F. et al. Plasmacytoid T cells. Immunohistochemical evidence for their monocyte/macrophage origin. Am. J. Pathol 133 , 15-21 (1988). 88. Grouard, G. et al. The enigmatic plasmacytoid T cells develop into dendritic cells with interleukin (IL)-3 and CD40-ligand. J. Exp. Med 185 , 1101-1111 (1997). 89. Siegal, F.P. et al. The nature of the principal type 1 interferon-producing cells in human blood. Science 284 , 1835-1837 (1999). 90. Cella, M. et al. Plasmacytoid monocytes migrate to inflamed lymph nodes and produce large amounts of type I interferon. Nat. Med 5, 919-923 (1999). 91. Theofilopoulos, A.N., Baccala, R., Beutler, B. & Kono, D.H. Type I interferons (alpha/beta) in immunity and autoimmunity. Annu. Rev. Immunol 23 , 307-336 (2005). 92. Liu, Y. IPC: professional type 1 interferon-producing cells and plasmacytoid dendritic cell precursors. Annu. Rev. Immunol 23 , 275-306 (2005). 93. MacDonald, K.P.A. et al. Characterization of human blood dendritic cell subsets. Blood 100 , 4512-4520 (2002). 94. Olweus, J. et al. Dendritic cell ontogeny: a human dendritic cell lineage of myeloid origin. Proc. Natl. Acad. Sci. U.S.A 94 , 12551-12556 (1997). 95. Dzionek, A. et al. BDCA-2, BDCA-3, and BDCA-4: three markers for distinct subsets of dendritic cells in human peripheral blood. J. Immunol 165 , 6037-6046 (2000). 96. Rissoan, M. et al. Subtractive hybridization reveals the expression of immunoglobulin- like transcript 7, Eph-B1, granzyme B, and 3 novel transcripts in human plasmacytoid dendritic cells. Blood 100 , 3295-3303 (2002). 97. Tordjman, R. et al. A neuronal receptor, neuropilin-1, is essential for the initiation of the primary immune response. Nat. Immunol 3, 477-482 (2002). 98. Ito, T., Kanzler, H., Duramad, O., Cao, W. & Liu, Y. Specialization, kinetics, and repertoire of type 1 interferon responses by human plasmacytoid predendritic cells. Blood 107 , 2423-2431 (2006). 99. Schotte, R. et al. The transcription factor Spi-B is expressed in plasmacytoid DC precursors and inhibits T-, B-, and NK-cell development. Blood 101 , 1015-1023 (2003). 100. Saunders, D. et al. Dendritic cell development in culture from thymic precursor cells in the absence of granulocyte/macrophage colony-stimulating factor. J. Exp. Med 184 , 2185-2196 (1996). 101. Spits, H., Couwenberg, F., Bakker, A.Q., Weijer, K. & Uittenbogaart, C.H. Id2 and Id3 inhibit development of CD34(+) stem cells into predendritic cell (pre-DC)2 but not into pre-DC1. Evidence for a lymphoid origin of pre-DC2. J. Exp. Med 192 , 1775-1784 (2000). 102. Schotte, R., Nagasawa, M., Weijer, K., Spits, H. & Blom, B. The ETS transcription 48 factor Spi-B is required for human plasmacytoid dendritic cell development. J. Exp. Med 200 , 1503-1509 (2004). 103. Maraskovsky, E. et al. In vivo generation of human dendritic cell subsets by Flt3 ligand. Blood 96 , 878-884 (2000). 104. Chicha, L., Jarrossay, D. & Manz, M.G. Clonal type I interferon-producing and dendritic cell precursors are contained in both human lymphoid and myeloid progenitor populations. J. Exp. Med 200 , 1519-1524 (2004). 105. Nagasawa, M., Schmidlin, H., Hazekamp, M.G., Schotte, R. & Blom, B. Development of human plasmacytoid dendritic cells depends on the combined action of the basic helix-loop-helix factor E2-2 and the Ets factor Spi-B. Eur. J. Immunol 38 , 2389-2400 (2008). 106. Cisse, B. et al. Transcription factor E2-2 is an essential and specific regulator of plasmacytoid dendritic cell development. Cell 135 , 37-48 (2008). 107. Robbins, S.H. et al. Novel insights into the relationships between dendritic cell subsets in human and mouse revealed by genome-wide expression profiling. Genome Biol 9, R17 (2008). 108. Sallusto, F. & Lanzavecchia, A. Understanding dendritic cell and T-lymphocyte traffic through the analysis of chemokine receptor expression. Immunol. Rev 177 , 134-140 (2000). 109. Penna, G., Sozzani, S. & Adorini, L. Cutting edge: selective usage of chemokine receptors by plasmacytoid dendritic cells. J. Immunol 167 , 1862-1866 (2001). 110. Gerlini, G., Mariotti, G., Bianchi, B. & Pimpinelli, N. Massive recruitment of type I interferon producing plasmacytoid dendritic cells in varicella skin lesions. J. Invest. Dermatol 126 , 507-509 (2006). 111. Krug, A. et al. IFN-producing cells respond to CXCR3 ligands in the presence of CXCL12 and secrete inflammatory chemokines upon activation. J. Immunol 169 , 6079- 6083 (2002). 112. Meller, S. et al. Ultraviolet radiation-induced injury, chemokines, and leukocyte recruitment: An amplification cycle triggering cutaneous lupus erythematosus. Arthritis Rheum 52 , 1504-1516 (2005). 113. Zabel, B.A., Silverio, A.M. & Butcher, E.C. Chemokine-like receptor 1 expression and chemerin-directed chemotaxis distinguish plasmacytoid from myeloid dendritic cells in human blood. J. Immunol 174 , 244-251 (2005). 114. Yoshimura, T. & Oppenheim, J.J. Chemerin reveals its chimeric nature. J. Exp. Med 205 , 2187-2190 (2008). 115. Asselin-Paturel, C. et al. Mouse type I IFN-producing cells are immature APCs with plasmacytoid morphology. Nat. Immunol 2, 1144-1150 (2001). 116. Nakano, H., Yanagita, M. & Gunn, M.D. CD11c(+)B220(+)Gr-1(+) cells in mouse lymph nodes and spleen display characteristics of plasmacytoid dendritic cells. J. Exp. Med 194 , 1171-1178 (2001). 117. Asselin-Paturel, C., Brizard, G., Pin, J., Brière, F. & Trinchieri, G. Mouse strain differences in plasmacytoid dendritic cell frequency and function revealed by a novel monoclonal antibody. J. Immunol 171 , 6466-6477 (2003). 118. Dalod, M. et al. Interferon alpha/beta and interleukin 12 responses to viral infections: pathways regulating dendritic cell cytokine expression in vivo. J. Exp. Med 195 , 517- 528 (2002). 119. Zucchini, N. et al. Individual plasmacytoid dendritic cells are major contributors to the production of multiple innate cytokines in an organ-specific manner during viral 49 infection. Int. Immunol 20 , 45-56 (2008). 120. Ito, T., Kanzler, H., Duramad, O., Cao, W. & Liu, Y. Specialization, kinetics, and repertoire of type 1 interferon responses by human plasmacytoid predendritic cells. Blood 107 , 2423-2431 (2006). 121. Lemaitre, B., Nicolas, E., Michaut, L., Reichhart, J.M. & Hoffmann, J.A. The dorsoventral regulatory gene cassette spätzle/Toll/cactus controls the potent antifungal response in Drosophila adults. Cell 86 , 973-983 (1996). 122. Medzhitov, R., Preston-Hurlburt, P. & Janeway, C.A. A human homologue of the Drosophila Toll protein signals activation of adaptive immunity. Nature 388 , 394-397 (1997). 123. Rock, F.L., Hardiman, G., Timans, J.C., Kastelein, R.A. & Bazan, J.F. A family of human receptors structurally related to Drosophila Toll. Proc. Natl. Acad. Sci. U.S.A 95 , 588-593 (1998). 124. Takeuchi, O. et al. TLR6: A novel member of an expanding toll-like receptor family. Gene 231 , 59-65 (1999). 125. Chuang, T.H. & Ulevitch, R.J. Cloning and characterization of a sub-family of human toll-like receptors: hTLR7, hTLR8 and hTLR9. Eur. Cytokine Netw 11 , 372-378 (2000). 126. Chuang, T. & Ulevitch, R.J. Identification of hTLR10: a novel human Toll-like receptor preferentially expressed in immune cells. Biochim. Biophys. Acta 1518 , 157-161 (2001). 127. Leonard, J.N., Bell, J.K. & Segal, D.M. Predicting Toll-like receptor structures and characterizing ligand binding. Methods Mol. Biol 517 , 55-67 (2009). 128. Govindaraj, R.G., Manavalan, B., Lee, G. & Choi, S. Molecular modeling-based evaluation of hTLR10 and identification of potential ligands in Toll-like receptor signaling. PLoS ONE 5, e12713 (2010). 129. Poltorak, A. et al. Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: mutations in Tlr4 gene. Science 282 , 2085-2088 (1998). 130. Mizel, S.B., West, A.P. & Hantgan, R.R. Identification of a sequence in human toll-like receptor 5 required for the binding of Gram-negative flagellin. J. Biol. Chem 278 , 23624-23629 (2003). 131. Alexopoulou, L., Holt, A.C., Medzhitov, R. & Flavell, R.A. Recognition of double- stranded RNA and activation of NF-kappaB by Toll-like receptor 3. Nature 413 , 732- 738 (2001). 132. Heil, F. et al. Species-specific recognition of single-stranded RNA via toll-like receptor 7 and 8. Science 303 , 1526-1529 (2004). 133. Hemmi, H. et al. A Toll-like receptor recognizes bacterial DNA. Nature 408 , 740-745 (2000). 134. Barton, G.M., Kagan, J.C. & Medzhitov, R. Intracellular localization of Toll-like receptor 9 prevents recognition of self DNA but facilitates access to viral DNA. Nat. Immunol 7, 49-56 (2006). 135. Jarrossay, D., Napolitani, G., Colonna, M., Sallusto, F. & Lanzavecchia, A. Specialization and complementarity in microbial molecule recognition by human myeloid and plasmacytoid dendritic cells. Eur. J. Immunol 31 , 3388-3393 (2001). 136. Lee, J. et al. Molecular basis for the immunostimulatory activity of guanine nucleoside analogs: activation of Toll-like receptor 7. Proc. Natl. Acad. Sci. U.S.A 100 , 6646-6651 (2003). 137. Bauer, S. et al. Human TLR9 confers responsiveness to bacterial DNA via species- specific CpG motif recognition. Proc. Natl. Acad. Sci. U.S.A 98 , 9237-9242 (2001). 138. Ozinsky, A. et al. The repertoire for pattern recognition of pathogens by the innate 50 immune system is defined by cooperation between toll-like receptors. Proc. Natl. Acad. Sci. U.S.A 97 , 13766-13771 (2000). 139. O'Neill, L.A.J., Fitzgerald, K.A. & Bowie, A.G. The Toll-IL-1 receptor adaptor family grows to five members. Trends Immunol 24 , 286-290 (2003). 140. Akira, S. & Takeda, K. Toll-like receptor signalling. Nat. Rev. Immunol 4, 499-511 (2004). 141. Husebye, H. et al. Endocytic pathways regulate Toll-like receptor 4 signaling and link innate and adaptive immunity. EMBO J 25 , 683-692 (2006). 142. Tanimura, N., Saitoh, S., Matsumoto, F., Akashi-Takamura, S. & Miyake, K. Roles for LPS-dependent interaction and relocation of TLR4 and TRAM in TRIF-signaling. Biochem. Biophys. Res. Commun 368 , 94-99 (2008). 143. Kerkmann, M. et al. Activation with CpG-A and CpG-B oligonucleotides reveals two distinct regulatory pathways of type I IFN synthesis in human plasmacytoid dendritic cells. J. Immunol 170 , 4465-4474 (2003). 144. Marié, I., Durbin, J.E. & Levy, D.E. Differential viral induction of distinct interferon- alpha genes by positive feedback through interferon regulatory factor-7. EMBO J 17 , 6660-6669 (1998). 145. Bauer, S. et al. Human TLR9 confers responsiveness to bacterial DNA via species- specific CpG motif recognition. Proc. Natl. Acad. Sci. U.S.A 98 , 9237-9242 (2001). 146. Honda, K. et al. Spatiotemporal regulation of MyD88-IRF-7 signalling for robust type-I interferon induction. Nature 434 , 1035-1040 (2005). 147. Guiducci, C. et al. Properties regulating the nature of the plasmacytoid dendritic cell response to Toll-like receptor 9 activation. J. Exp. Med 203 , 1999-2008 (2006). 148. Randow, F. & Seed, B. Endoplasmic reticulum chaperone gp96 is required for innate immunity but not cell viability. Nat. Cell Biol 3, 891-896 (2001). 149. Yang, Y. et al. Heat shock protein gp96 is a master chaperone for toll-like receptors and is important in the innate function of macrophages. Immunity 26 , 215-226 (2007). 150. Takahashi, K. et al. A protein associated with Toll-like receptor (TLR) 4 (PRAT4A) is required for TLR-dependent immune responses. J. Exp. Med 204 , 2963-2976 (2007). 151. Liu, B. et al. Folding of Toll-like receptors by the HSP90 paralogue gp96 requires a substrate-specific cochaperone. Nat Commun 1, doi:10.1038/ncomms1070 (2010). 152. Leifer, C.A. et al. TLR9 is localized in the endoplasmic reticulum prior to stimulation. J. Immunol 173 , 1179-1183 (2004). 153. Latz, E. et al. TLR9 signals after translocating from the ER to CpG DNA in the lysosome. Nat. Immunol 5, 190-198 (2004). 154. Chockalingam, A., Brooks, J.C., Cameron, J.L., Blum, L.K. & Leifer, C.A. TLR9 traffics through the Golgi complex to localize to endolysosomes and respond to CpG DNA. Immunol. Cell Biol 87 , 209-217 (2009). 155. Ewald, S.E. et al. The ectodomain of Toll-like receptor 9 is cleaved to generate a functional receptor. Nature 456 , 658-662 (2008). 156. Tabeta, K. et al. The Unc93b1 mutation 3d disrupts exogenous antigen presentation and signaling via Toll-like receptors 3, 7 and 9. Nat. Immunol 7, 156-164 (2006). 157. Brinkmann, M.M. et al. The interaction between the ER membrane protein UNC93B and TLR3, 7, and 9 is crucial for TLR signaling. J. Cell Biol 177 , 265-275 (2007). 158. Kim, Y., Brinkmann, M.M., Paquet, M. & Ploegh, H.L. UNC93B1 delivers nucleotide- sensing toll-like receptors to endolysosomes. Nature 452 , 234-238 (2008). 159. Boo, K. & Yang, J. Intrinsic cellular defenses against virus infection by antiviral type I interferon. Yonsei Med. J 51 , 9-17 (2010). 51 160. Luft, T. et al. Type I IFNs enhance the terminal differentiation of dendritic cells. J. Immunol 161 , 1947-1953 (1998). 161. Santodonato, L. et al. Monocyte-derived dendritic cells generated after a short-term culture with IFN-alpha and granulocyte-macrophage colony-stimulating factor stimulate a potent Epstein-Barr virus-specific CD8+ T cell response. J. Immunol 170 , 5195-5202 (2003). 162. Santini, S.M. et al. Type I interferon as a powerful adjuvant for monocyte-derived dendritic cell development and activity in vitro and in Hu-PBL-SCID mice. J. Exp. Med 191 , 1777-1788 (2000). 163. Le Bon, A. et al. Cross-priming of CD8+ T cells stimulated by virus-induced type I interferon. Nat. Immunol 4, 1009-1015 (2003). 164. Jego, G. et al. Plasmacytoid dendritic cells induce plasma cell differentiation through type I interferon and interleukin 6. Immunity 19 , 225-234 (2003). 165. Megjugorac, N.J., Young, H.A., Amrute, S.B., Olshalsky, S.L. & Fitzgerald-Bocarsly, P. Virally stimulated plasmacytoid dendritic cells produce chemokines and induce migration of T and NK cells. J. Leukoc. Biol 75 , 504-514 (2004). 166. Dzionek, A. et al. BDCA-2, a novel plasmacytoid dendritic cell-specific type II C-type lectin, mediates antigen capture and is a potent inhibitor of interferon alpha/beta induction. J. Exp. Med 194 , 1823-1834 (2001). 167. Cao, W. et al. Plasmacytoid dendritic cell-specific receptor ILT7-Fc epsilonRI gamma inhibits Toll-like receptor-induced interferon production. J. Exp. Med 203 , 1399-1405 (2006). 168. Cao, W. et al. BDCA2/Fc epsilon RI gamma complex signals through a novel BCR-like pathway in human plasmacytoid dendritic cells. PLoS Biol 5, e248 (2007). 169. Cao, W. et al. Regulation of TLR7/9 responses in plasmacytoid dendritic cells by BST2 and ILT7 receptor interaction. J. Exp. Med 206 , 1603-1614 (2009). 170. Vitale, M. et al. NKp44, a novel triggering surface molecule specifically expressed by activated natural killer cells, is involved in non-major histocompatibility complex- restricted tumor cell lysis. J. Exp. Med 187 , 2065-2072 (1998). 171. Arnon, T.I. et al. Recognition of viral hemagglutinins by NKp44 but not by NKp30. Eur. J. Immunol 31 , 2680-2689 (2001). 172. Fuchs, A., Cella, M., Kondo, T. & Colonna, M. Paradoxic inhibition of human natural interferon-producing cells by the activating receptor NKp44. Blood 106 , 2076-2082 (2005). 173. Meyer-Wentrup, F. et al. Targeting DCIR on human plasmacytoid dendritic cells results in antigen presentation and inhibits IFN-alpha production. Blood 111 , 4245-4253 (2008). 174. Rissoan, M.C. et al. Reciprocal control of T helper cell and dendritic cell differentiation. Science 283 , 1183-1186 (1999). 175. Gilliet, M. & Liu, Y. Generation of human CD8 T regulatory cells by CD40 ligand- activated plasmacytoid dendritic cells. J. Exp. Med 195 , 695-704 (2002). 176. Kadowaki, N., Antonenko, S., Lau, J.Y. & Liu, Y.J. Natural interferon alpha/beta- producing cells link innate and adaptive immunity. J. Exp. Med 192 , 219-226 (2000). 177. Ito, T. et al. Plasmacytoid dendritic cells regulate Th cell responses through OX40 ligand and type I IFNs. J. Immunol 172 , 4253-4259 (2004). 178. Fonteneau, J. et al. Activation of influenza virus-specific CD4+ and CD8+ T cells: a new role for plasmacytoid dendritic cells in adaptive immunity. Blood 101 , 3520-3526 (2003). 179. Young, L.J. et al. Differential MHC class II synthesis and ubiquitination confers distinct 52 antigen-presenting properties on conventional and plasmacytoid dendritic cells. Nat. Immunol 9, 1244-1252 (2008). 180. Sadaka, C., Marloie-Provost, M., Soumelis, V. & Benaroch, P. Developmental regulation of MHC II expression and transport in human plasmacytoid-derived dendritic cells. Blood 113 , 2127-2135 (2009). 181. Di Pucchio, T. et al. Direct proteasome-independent cross-presentation of viral antigen by plasmacytoid dendritic cells on major histocompatibility complex class I. Nat. Immunol 9, 551-557 (2008). 182. Jaehn, P.S., Zaenker, K.S., Schmitz, J. & Dzionek, A. Functional dichotomy of plasmacytoid dendritic cells: antigen-specific activation of T cells versus production of type I interferon. Eur. J. Immunol 38 , 1822-1832 (2008). 183. Jähn, P.S., Zänker, K.S., Schmitz, J. & Dzionek, A. BDCA-2 signaling inhibits TLR-9- agonist-induced plasmacytoid dendritic cell activation and antigen presentation. Cell. Immunol 265 , 15-22 (2010). 184. Ito, T. et al. Plasmacytoid dendritic cells prime IL-10-producing T regulatory cells by inducible costimulator ligand. J. Exp. Med 204 , 105-115 (2007). 185. Ito, T. et al. Two functional subsets of FOXP3+ regulatory T cells in human thymus and periphery. Immunity 28 , 870-880 (2008). 186. Chen, W., Liang, X., Peterson, A.J., Munn, D.H. & Blazar, B.R. The indoleamine 2,3- dioxygenase pathway is essential for human plasmacytoid dendritic cell-induced adaptive T regulatory cell generation. J. Immunol 181 , 5396-5404 (2008). 187. Mellor, A.L. & Munn, D.H. IDO expression by dendritic cells: tolerance and tryptophan catabolism. Nat. Rev. Immunol 4, 762-774 (2004). 188. Matsui, T. et al. CD2 distinguishes two subsets of human plasmacytoid dendritic cells with distinct phenotype and functions. J. Immunol 182 , 6815-6823 (2009). 189. Krug, A. et al. TLR9-dependent recognition of MCMV by IPC and DC generates coordinated cytokine responses that activate antiviral NK cell function. Immunity 21 , 107-119 (2004). 190. Wang, H., Peters, N. & Schwarze, J. Plasmacytoid dendritic cells limit viral replication, pulmonary inflammation, and airway hyperresponsiveness in respiratory syncytial virus infection. J. Immunol 177 , 6263-6270 (2006). 191. Wolf, A.I. et al. Plasmacytoid dendritic cells are dispensable during primary influenza virus infection. J. Immunol 182 , 871-879 (2009). 192. Langlois, R.A. & Legge, K.L. Plasmacytoid dendritic cells enhance mortality during lethal influenza infections by eliminating virus-specific CD8 T cells. J. Immunol 184 , 4440-4446 (2010). 193. Penna, G., Sozzani, S. & Adorini, L. Cutting edge: selective usage of chemokine receptors by plasmacytoid dendritic cells. J. Immunol 167 , 1862-1866 (2001). 194. Fong, L., Mengozzi, M., Abbey, N.W., Herndier, B.G. & Engleman, E.G. Productive infection of plasmacytoid dendritic cells with human immunodeficiency virus type 1 is triggered by CD40 ligation. J. Virol 76 , 11033-11041 (2002). 195. Soumelis, V. et al. Depletion of circulating natural type 1 interferon-producing cells in HIV-infected AIDS patients. Blood 98 , 906-912 (2001). 196. Nascimbeni, M. et al. Plasmacytoid dendritic cells accumulate in spleens from chronically HIV-infected patients but barely participate in interferon-alpha expression. Blood 113 , 6112-6119 (2009). 197. Martinelli, E. et al. HIV-1 gp120 inhibits TLR9-mediated activation and IFN-{alpha} secretion in plasmacytoid dendritic cells. Proc. Natl. Acad. Sci. U.S.A 104 , 3396-3401 53 (2007). 198. Xie, Q. et al. Patients with chronic hepatitis B infection display deficiency of plasmacytoid dendritic cells with reduced expression of TLR9. Microbes Infect 11 , 515- 523 (2009). 199. Xu, Y. et al. HBsAg inhibits TLR9-mediated activation and IFN-alpha production in plasmacytoid dendritic cells. Mol. Immunol 46 , 2640-2646 (2009). 200. Kim, T. et al. Serum levels of interferons in patients with systemic lupus erythematosus. Clin. Exp. Immunol 70 , 562-569 (1987). 201. Farkas, L., Beiske, K., Lund-Johansen, F., Brandtzaeg, P. & Jahnsen, F.L. Plasmacytoid dendritic cells (natural interferon- alpha/beta-producing cells) accumulate in cutaneous lupus erythematosus lesions. Am. J. Pathol 159 , 237-243 (2001). 202. Means, T.K. et al. Human lupus autoantibody-DNA complexes activate DCs through cooperation of CD32 and TLR9. J. Clin. Invest 115 , 407-417 (2005). 203. Blanco, P., Palucka, A.K., Gill, M., Pascual, V. & Banchereau, J. Induction of dendritic cell differentiation by IFN-alpha in systemic lupus erythematosus. Science 294 , 1540- 1543 (2001). 204. Lew, W., Bowcock, A.M. & Krueger, J.G. Psoriasis vulgaris: cutaneous lymphoid tissue supports T-cell activation and "Type 1" inflammatory gene expression. Trends Immunol 25 , 295-305 (2004). 205. Nestle, F.O. et al. Plasmacytoid predendritic cells initiate psoriasis through interferon- alpha production. J. Exp. Med 202 , 135-143 (2005). 206. Frohm, M. et al. The expression of the gene coding for the antibacterial peptide LL-37 is induced in human keratinocytes during inflammatory disorders. J. Biol. Chem 272 , 15258-15263 (1997). 207. Lande, R. et al. Plasmacytoid dendritic cells sense self-DNA coupled with antimicrobial peptide. Nature 449 , 564-569 (2007). 208. Hartmann, E. et al. Identification and functional analysis of tumor-infiltrating plasmacytoid dendritic cells in head and neck cancer. Cancer Res 63 , 6478-6487 (2003). 209. Perrot, I. et al. Dendritic cells infiltrating human non-small cell lung cancer are blocked at immature stage. J. Immunol 178 , 2763-2769 (2007). 210. Treilleux, I. et al. Dendritic cell infiltration and prognosis of early stage breast cancer. Clin. Cancer Res 10 , 7466-7474 (2004). 211. Vermi, W. et al. Recruitment of immature plasmacytoid dendritic cells (plasmacytoid monocytes) and myeloid dendritic cells in primary cutaneous melanomas. J. Pathol 200 , 255-268 (2003). 212. Zou, W. et al. Stromal-derived factor-1 in human tumors recruits and alters the function of plasmacytoid precursor dendritic cells. Nat. Med 7, 1339-1346 (2001). 213. Sharma, M.D. et al. Plasmacytoid dendritic cells from mouse tumor-draining lymph nodes directly activate mature Tregs via indoleamine 2,3-dioxygenase. J. Clin. Invest 117 , 2570-2582 (2007). 214. Feuillard, J. et al. Clinical and biologic features of CD4(+)CD56(+) malignancies. Blood 99 , 1556-1563 (2002). 215. Petrella, T. et al. CD4+ CD56+ cutaneous neoplasms: a distinct hematological entity? Groupe Français d'Etude des Lymphomes Cutanés (GFELC). Am. J. Surg. Pathol 23 , 137-146 (1999). 216. Chaperot, L. et al. Identification of a leukemic counterpart of the plasmacytoid dendritic cells. Blood 97 , 3210-3217 (2001). 217. Chaperot, L. et al. Leukemic plasmacytoid dendritic cells share phenotypic and 54 functional features with their normal counterparts. Eur. J. Immunol 34 , 418-426 (2004). 218. Schuske, K., Palfreyman, M.T., Watanabe, S. & Jorgensen, E.M. UNC-46 is required for trafficking of the vesicular GABA transporter. Nat. Neurosci 10 , 846-853 (2007). 219. Anggono, V. & Robinson, P.J. Syndapin I and endophilin I bind overlapping proline-rich regions of dynamin I: role in synaptic vesicle endocytosis. J. Neurochem 102 , 931-943 (2007). 220. Kleene, R. et al. NCAM-induced neurite outgrowth depends on binding of calmodulin to NCAM and on nuclear import of NCAM and fak fragments. J. Neurosci 30 , 10784- 10798 (2010). 221. Gniadecki, R. et al. CD56+ lymphoma with skin involvement: clinicopathologic features and classification. Arch Dermatol 140 , 427-436 (2004). 222. Huh, G.S. et al. Functional requirement for class I MHC in CNS development and plasticity. Science 290 , 2155-2159 (2000). 223. Zohar, O. et al. Cutting edge: MHC class I-Ly49 interaction regulates neuronal function. J. Immunol 180 , 6447-6451 (2008). 224. Bryceson, Y.T., Foster, J.A., Kuppusamy, S.P., Herkenham, M. & Long, E.O. Expression of a killer cell receptor-like gene in plastic regions of the central nervous system. J. Neuroimmunol 161 , 177-182 (2005). 225. Garnache-Ottou, F. et al. Extended diagnostic criteria for plasmacytoid dendritic cell leukaemia. Br. J. Haematol 145 , 624-636 (2009). 226. Jaye, D.L. et al. Expression of the plasmacytoid dendritic cell marker BDCA-2 supports a spectrum of maturation among CD4+ CD56+ hematodermic neoplasms. Mod. Pathol 19 , 1555-1562 (2006).

55 Annexe 1

“BAD-LAMP defines a subset of early endocytic organelles in subpopulations of cortical projection neurons”

Article publié dans la revue Journal of Cell Science, 2007 Jan 15;120(Pt 2):353-65

56 Research Article 353 BAD-LAMP defines a subset of early endocytic organelles in subpopulations of cortical projection neurons

Alexandre David 1,2,3, *, Marie-Catherine Tiveron 4, *, Axel Defays 1,2,3 , Christophe Beclin 4, Voahirana Camosseto 1,2,3 , Evelina Gatti 1,2,3 , Harold Cremer 4,‡,§ and Philippe Pierre 1,2,3,‡,§ 1Centre d’Immunologie de Marseille-Luminy, Université de la Méditerranée, Case 906, 13288 Marseille cedex 9, France 2INSERM, U631 and 3CNRS, UMR6102,13288 Marseille, France 4Institut de Biologie du Développement de Marseille-Luminy, CNRS UMR 6216, Université de la Méditerranée, Case 907, 13288 Marseille cedex 9, France *Both first authors contributed equally to this work ‡Both last authors contributed equally to this work §Authors for correspondence (e-mail: [email protected]; [email protected])

Accepted 25 October 2006 Journal of Cell Science 120, 353-365 Published by The Company of Biologists 2007 doi:10.1242/jcs.03316

Summary The brain-associated LAMP-like molecule (BAD-LAMP) is promote its sorting away from lysosomes. Analysis of BAD- a new member of the family of lysosome associated LAMP endocytosis in transfected HeLa cells provided membrane proteins (LAMPs). In contrast to other LAMPs, evidence that the protein recycles to the plasma membrane which show a widespread expression, BAD-LAMP through a dynamin/AP2-dependent mechanism. Thus, expression in mice is confined to the postnatal brain and BAD-LAMP is an unconventional LAMP-like molecule and therein to neuronal subpopulations in layers II/III and V defines a new endocytic compartment in specific subtypes of the neocortex. Onset of expression strictly parallels of cortical projection neurons. The striking correlation cortical synaptogenesis. In cortical neurons, the protein is between the appearance of BAD-LAMP and cortical found in defined clustered vesicles, which accumulate along synatogenesis points towards a physiological role of this neurites where it localizes with phosphorylated epitopes of vesicular determinant for neuronal function. neurofilament H. In primary neurons, BAD-LAMP is endocytosed, but is not found in classical lysosomal/ Supplementary material available online at endosomal compartments. Modification of BAD-LAMP by http://jcs.biologists.org/cgi/content/full/120/2/353/DC1 Journal of Cell Science addition of GFP revealed a cryptic lysosomal retention motif, suggesting that the cytoplasmic tail of BAD-LAMP Key words: Corticogenesis, Endocytosis, Synaptogenesis, LAMP, is actively interacting with, or modified by, molecules that Lipid microdomains, Cortex

Introduction inhibits the activity of Src family kinases, whereas PrP is Neurons are polarized cells specialized to carry out regulated rapidly endocytosed and induces axonal outgrowth via the secretion of storage vesicles when an appropriate stimulus is activation of fyn-related kinases (Santuccione et al., 2005). applied. Furthermore, synapse formation, stabilization and Vesicular transport and lipid microdomain organization, maintenance require the delivery of transport vesicles to the therefore, play key roles in neuronal development and function. site of initial contact between axons and dendrites. These The LAMP family is composed of proteins bearing sequence vesicles, containing the different proteins necessary for proper and structural homology with the canonical LAMP1 and establishment and function of synapses, are the results of LAMP2 molecules. LAMP molecules harbor an endosomal complex interplay between the secretion and the endocytic and lysosomal addressing signal within their short cytoplasmic membrane transport pathways (Kennedy and Ehlers, 2006). tail, and contain several conserved cysteine residues, which Another layer of complexity is introduced by the existence allow the formation of particular structural loops known as of ordered lipid domains in the plasma membrane (Maxfield ‘LAMP folds’. Although the structure, subcellular localization and Tabas, 2005). In neurons, several types of microdomains and interaction partners of LAMP1 and LAMP2 have been have been shown to be distinguishable by the partitioning of extensively characterized, their physiological function is still different membrane-associated proteins such as thymus cell elusive (Eskelinen et al., 2003). Lamp1 -deficient mice are antigen 1 (THY1) or the prion protein (PrP) (Sunyach et al., viable and show a mild astrogliosis in the brain (Andrejewski 2003), which are found in different, albeit often closely et al., 1999), whereas Lamp2 mutants show increased postnatal adjacent domains (Madore et al., 1999). These differences in lethality and massive accumulation of autophagic vesicles in surface localization are reflected in the different trafficking and different tissues (Tanaka et al., 2000; Eskelinen et al., 2002). functions of these proteins. THY1 is slowly internalized and Interestingly, LAMP2 deficiency in humans induces Danon 354 Journal of Cell Science 120 (2)

disease, a lysosomal glycogen storage disorder characterized a protein of 280 aa (PI 6.42 and molecular mass of 31.7 kDa) by cardio- and skeletal myopathy and a variable degree of predicted to contain a transmembrane domain (aa 236-256) and mental retardation (Saftig et al., 2001). a cytoplasmic tail of 24 residues (Fig. 1A). This cytoplasmic We identified a new member of the LAMP protein family in domain contains a YKHM (aa 276) motif corresponding to a mice. B rain-a ssociated LAMP-like molecule (BAD-LAMP) is classical Yxx ⌽ internalization and endosomal targeting signal. expressed after birth in cortical neurons of particular layers, The sequence also contains four highly conserved cysteine where it is enriched in defined zones along the neuronal residues separated by a fixed number of amino acids and is projections. BAD-LAMP mainly accumulates in distinct likely to form characteristic internal di-sulfide bonds required intracellular vesicles, which do not contain any known markers for a classical ‘LAMP fold’. The protein was also predicted to of classical intracellular transport pathways. BAD-LAMP- contain three consensus N-glycosylation sites. The nucleotide containing vesicles have a remarkable clustered organization sequence shares 45% identity with LAMP1 and LAMP2, the mirroring at the neuronal surface the presence of THY1- founding members of the family, whereas alignments at the containing microdomains, but not of N-CAM and the protein level displayed 25% similarity (19% identity) (see ganglioside GM1-enriched microdomains. Interestingly the supplementary material Fig. S1). Thus, the protein was placed phosphoepitopes present on microtubule-associated protein 1B on an evolutionary classification tree between LAMP1 and and neurofilament H also define BAD-LAMP-containing DC-HIL sequences, clearly identifying it as a new member of vesicle positioning in neurons. BAD-LAMP has the ability to the LAMP family (Fig. 1B). The tree indicates that DC-HIL be endocytosed, but is not targeted to the late (15.5% of similarity), a dendritic cell specific molecule endosomal/lysosome compartments (Gruenberg and Stenmark, functioning as an integrin ligand (Shikano et al., 2001), shared 2004). The spatiotemporal specificity of BAD-LAMP a common ancestor molecule after diverging away from the expression and its distribution reveal therefore a new level of LAMP1/CD68 evolutionary axis. The molecule is extremely interplay involving unconventional endocytic compartments conserved, since it is found in worm, fly, fish, chicken, rodent and membrane microdomains in specific cortical neurons. and human (see supplementary material Fig. S1). The degree of identity at the amino acid level is close to 85% among Results mammals and 45% between mouse and fugu. This very high BAD-LAMP is a new member of the LAMP family level of conservation across species suggests that the molecule expressed specifically in the post-natal mouse brain performs a conserved cellular function, not accommodating During a bioinformatics search to identify lysosomal- many variations of its tertiary structure. associated molecules, several overlapping nucleotide Northern blot analysis of the identified mRNA using sequences were identified. After PCR cloning of the full-length different mouse tissues indicated that it is expressed almost cDNA from mouse cortex and extensive sequencing, we exclusively in the adult brain, with a close to background signal identified a potential open reading frame coding for a new in the E14 embryo (Fig. 1C). The detected mRNA corresponds putative member of the LAMP family. The new ORF codes for to a unique transcript of around 1.8 kb with no apparent Journal of Cell Science

Fig. 1. Characterization of a new LAMP molecule. (A) BAD-LAMP protein sequence showing predicted glycosylation sites (bold and blue), putative di-sulfide bridges between two pairs of cysteins (red line), transmembrane domain (aa 236-256 in blue and underlined) and tyrosine endocytic sorting motif (YKHM, aa 276 in red). (B) Phylogram representation of all the LAMP family members in human and mouse. (C) Tissue distribution of BAD-LAMP by Northern blot. Among all mouse tissues tested BAD-LAMP appears to be expressed specifically in brain as a 2 kb mRNA transcript. Actin mRNA levels are shown as control. BAD-LAMP in cortical neurons 355

Fig. 2. BAD-LAMP is heavily glycosylated and is expressed after birth. (A) Immunoblot performed with a polyclonal antibody raised against the predicted peptide of the BAD-LAMP cytoplasmic tail. Lysates of mouse adult cortex and BAD-LAMP-transfected HeLa cells produced several bands migrating at above 35 kDa, whereas no reactivity is observed in control untransfected cells. The lowest form in the transfected Journal of Cell Science cells is probably due to ER accumulation. (B) HeLa cells transfected with FLAG-BAD-LAMP cDNA were lysed and immunoprecipitated with anti-FLAG antibody. Immunoprecipitated material was treated with endoglycosidase H (endo H) or N-glycosidase F (N-gly F) prior to immunoblotting with anti-BAD-LAMP. In untransfected cells (NC), just the anti-FLAG IgG band (arrow) is observed, whereas several isoforms of BAD-LAMP were detected in transfected cells (Wt). Endo H treatment shows that the major band is endo H sensitive (gH) thus probably accumulating in the ER. The higher molecular mass bands were all N-gly F sensitive (gF). N-gly F treatment also demonstrated that all isoforms of BAD-LAMP are glycosylated and that the native molecular mass of the molecule is around 32 kDa (g0). The anti-FLAG IgG bands are also N-gly F sensitive, arrows. (C) After fractionation and isolation of cortical membranes, BAD-LAMP was found to be present exclusively in the membrane pellet (MbP) and not in the supernatant (Sup). Control syntaxin 6 and syntaxin 13 were also found in the membrane pellet, whereas RAB3, as expected, had a shared distribution due to its shuttling nature. (D) BAD-LAMP expression after birth. Mouse cortex lysates of different ages were immunoblotted with BAD-LAMP polyclonal antibody. BAD-LAMP expression levels are increased from birth to adulthood. (E) In situ hybridization for Bad-lamp on coronal post-natal brain sections from P2 to P12. Hemisections are presented. Bad-lamp expression appears at P2 in the cingulate cortex (arrowhead) and extends ventrally during the first post-natal weeks as a superficial and a deep band in the cortex. Subcortically, Bad-lamp is expressed transiently in the caudate putamen (cp). Bar, 500 ␮m.

alternative spliced forms. Based on its relationship to the remained non-reactive. Brain extracts also displayed several LAMP family and its restricted pattern of expression, the bands, mostly corresponding to those observed in transfected molecule was named BAD-LAMP, for b rain-a ssociated - cells, confirming the existence of several isoforms of BAD- LAMP. LAMP. The detected proteins had a significantly higher molecular mass than the one predicted from the primary BAD-LAMP is a glycosylated membrane-associated sequence of BAD-LAMP (31.7 kDa). In order to define the protein nature of these post-translational modifications and in absence To investigate further BAD-LAMP distribution and function, of immunoprecipitating antibodies, we transfected an N- we raised antibodies against peptide epitopes present in its terminally tagged form of BAD-LAMP (FLAG-BAD-LAMP), cytoplasmic tail. These antibodies were characterized by allowing efficient immunoprecipitation and treatment with immunoblot of HeLa cells transfected with the cDNA coding endoglycosidase H (Endo H) and N-glycosidase F (N-gly F). for mouse BAD-LAMP (Fig. 2A). Several bands were detected Immunoprecipitated FLAG-BAD-LAMP was shown to be in extracts from transfected cells, whereas control extracts heavily glycosylated (Fig. 2B). The major form of the protein 356 Journal of Cell Science 120 (2)

glycosylated on at least two of its three acceptor sites, a situation likely to be shared with endogenous BAD-LAMP detected in brain. Although glycosylation was in support of BAD-LAMP membrane association, we demonstrated the membrane-bound nature of BAD-LAMP by submitting mouse cortex post- nuclear supernatants to high speed ultracentrifugation, in which BAD-LAMP was found associated with the membranes pellets similar to other membrane-associated molecules such as RAB3a, syntaxin 6 and syntaxin 13 (Fig. 2C). Thus BAD- LAMP, is a glycosylated LAMP-like molecule associated with cortical membranes.

BAD-LAMP is expressed in neurons of specific cortical layers after birth Analysis of mouse brain extracts by immunoblotting revealed that levels of BAD-LAMP increased strongly after birth (P0) reaching its maximum level at adulthood, but being already strongly expressed at P10-P12 (Fig. 2D). We used in situ hybridization to investigate in detail the expression pattern of BAD-LAMP in the developing mouse forebrain. The first expression of BAD-LAMP was found at P2 in the cingulate cortex, in a thin band of intermediate cells (Fig. 2E). At P5, expression extended ventrally into the cortical plate. Furthermore, the caudate putamen showed a punctuate expression of Bad-lamp transcripts. This expression pattern was maintained at P7, when an additional broad band of large and strongly Bad-lamp -positive cells appeared in superficial parts of the cortical plate. Although the cortical expression intensified until P9, no major regional changes in Bad-lamp expression were obvious during this period. At P12, expression of Bad-lamp in the striatum ceased, while expression further intensified in the cortex. This expression pattern was stable until adulthood. Altogether, this expression pattern indicates

Journal of Cell Science Fig. 3. BAD-LAMP is specifically expressed in neurons of the that BAD-LAMP does not function in the early steps of brain cortical layers II, III and V. (A) In situ hybridization for Bad-lamp development, such as neurogenesis and cell migration, but (top panel), Cux2 (middle) and ER81 (bottom) on adult mouse brain potentially during terminal steps of neuronal differentiation coronal hemisections. (B) High magnification views of in situ and neuronal function. hybridization on wild-type cortex shown in A. Bad-lamp is expressed Within the adult cortex, the homogenous staining in outer in layers II, II and V, but excluded from layers IV and VI. In the regions of the cortical plate, as well as in a more restricted band Scrambler cortex, the entire region appears disorganized. However, the typical inversion of cortical layers is reflected by the altered of cells localized centrally, was suggestive of an expression in BAD-LAMP staining, demonstrating that projection neurons express neurons of specific cortical layers. We used well-known the protein. (C) Combined in situ hybridization for Bad-lamp (in markers for cortical layers to further characterize the respective blue) with immunohistochemistry for the specific neuronal marker populations. Comparison of the expression of Bad-lamp to that NeuN (in brown). The left panel is a higher magnification of the of Cux2 , a marker for layers II-IV, showed that the BAD- boxed area. Bad-lamp is co-expressed with this pan-neuronal marker LAMP domain is included in the CUX2 domain and confined in many, although not all, neurons. (D) Immunohistochemistry of to its outer part (Fig. 3A,B). Thus, BAD-LAMP is expressed BAD-LAMP in the indicated cortex layers. (E) Immunofluorescence in the upper layers II and III of the neocortex, but is excluded staining on adult cortex using anti-MAP2 (green) and anti-BAD- from layer IV. Furthermore, there was a perfect overlap with LAMP (red) antibodies; the merged image is on the left. Whereas the layer V marker ER81 (Fig. 3A,B) demonstrating that the MAP2 is present along the entire dendrites, BAD-LAMP accumulates in defined domains (arrows). Bars, 500 ␮m in A; 200 deeply positioned Bad-lamp -positive population is located in ␮m in B; 100 ␮m (left panel) and 10 ␮m (right panel) in C; 10 ␮m layer V. in D. The size of the Bad-lamp -positive cells in the respective cortical layers was suggestive of neuronal cells. To confirm this observation we investigated the expression of Bad-lamp in (38 kDa, gH) remained Endo H sensitive, thus reflecting Scrambler mice. These animals show a well described endoplasmic reticulum (ER) retention due to over-expression. inversion of the layers of cortical projection neurons, with The two higher additional bands (47 and 53 kDa, gF) were upper layer neurons (layers II-IV) positioned deeply whereas Endo H resistant but remained N-gly F sensitive, as indicated deep layer neurons (layers V and VI) are positioned by the accumulation of a fully trimmed 31 kDa protein (g0) superficially (Rice and Curran, 1999). The organization of B after treatment. Transfected BAD-LAMP is therefore heavily Bad-lamp -positive cells in the Scrambler cortex was strikingly BAD-LAMP in cortical neurons 357

Fig. 4. BAD-LAMP is present in vesicles clustering in specific areas of the neurons. Immunofluorescence confocal microscopy of cortical neurons. (A) Staining for BAD-LAMP (red) and cholera toxin (GM1, green). (B) Staining for BAD-LAMP (red) and PrP (green). (C) Staining for BAD- LAMP (red) and N-CAM (green). BAD-LAMP is expressed in small vesicles clustered in neurites and accumulates in areas lacking surface semi-ordered lipid microdomain resident proteins (arrowheads). (D) THY1 labelling (green) defines the zones in which BAD-LAMP vesicles accumulate (red). However, THY1 (red) is present at the cell surface and does not co-localize with BAD-LAMP as seen at higher magnification (bottom panels, arrowheads). (E) Cholesterol depletion disrupts cluster organization and induces BAD-LAMP (red) and cholera toxin (GM1, green) co- localization. Bars, 20 ␮m; 10 ␮m for THY1 high magnification.

defined domains along cellular projections (Fig. 3E arrowheads). It could be detected at the plasma membrane and in vesicles present in the cell bodies, but was also enriched in vesicles clustered in defined domains of dendrites. BAD-LAMP mostly accumulated within the boundaries of specific neuronal areas in vivo. In order to confirm the relevance of these observations,

Journal of Cell Science embryonic cortical neurons were explanted and BAD- LAMP sub-cellular distribution was investigated after 3 days in culture. Owing to the particular clustered distribution of BAD-LAMP, we also investigated the distribution of proteins known to partition in different cellular domains, such as lipid microdomain-associated proteins (Madore et al., 1999). Using confocal microscopy we found that semi-ordered lipid microdomain residents such as PrP, N-CAM, as well as the ganglioside GM1 (stained with cholera toxin, CT) altered (Fig. 3B). Small, lighter stained cell bodies were were enriched in zones excluding BAD-LAMP vesicles (Fig. displaced towards the ventricular side, whereas larger and more 4A-C). This observation was particularly striking with CT strongly labelled cells were merely found at the pial side of the staining and N-CAM, which accumulated almost exclusively cortex. This pattern was in agreement with an inversion of the in areas negative for BAD-LAMP (Fig. 4A,C). By contrast, at position of Bad-lamp -positive cells and suggestive of a this low magnification, THY1, a molecule representing ordered projection neurone identity. Furthermore, all Bad-lamp - lipid microdomain-associated proteins, displayed an positive cells in the cortex were co-expressing the neuronal overlapping distribution with BAD-LAMP (Fig. 4D). However, marker NeuN, again confirming the neuronal identity of at higher magnification, no direct co-localization of THY1 and labelled cells (Fig. 3C), which were found also to express BAD-LAMP molecules could be observed. Instead, BAD-LAMP protein (Fig. 3D). accumulation of BAD-LAMP-containing vesicles was revealed directly underneath THY1-enriched areas at the plasma BAD-LAMP is distributed in specific domains of cortical membrane (Fig. 4D, arrowheads). BAD-LAMP-containing neurons vesicles therefore accumulate in cellular zones, defined by the BAD-LAMP distribution in cortical brain sections was presence of THY1 at the plasma membrane, whereas they are monitored by confocal microscopy (Fig. 3E). BAD-LAMP was segregated from the detergent-resistant microdomains found in vesicles mostly located in cell bodies as delineated by containing most of the PrP, GM1 and N-CAM (Madore et al., the MAP2 staining. In addition, BAD-LAMP accumulated in 1999). 358 Journal of Cell Science 120 (2)

Fig. 5. BAD-LAMP organization is defined by microtubules. Immunofluorescence confocal microscopy of cortical neurons. (A) BAD-LAMP vesicles (red) accumulate in areas of the cell that are strongly enriched in the phospho- epitope detected by the Smi31 antibody (blue), whereas the L1 molecule (green) is distributed throughout the neuronal plasma membrane. (B) (top) BAD-LAMP vesicle accumulation (white) coincides with microtubule bundling (␤-tubulin, red) and weak GM1 staining (green). Higher magnification (Z1) shows that BAD- LAMP vesicles align along microtubules. (Bottom) Nocodazole treatment induces microtubule destabilization and disorganization of BAD-LAMP vesicle clusters. Bars, 20 ␮m; 10 ␮m for Z1. Journal of Cell Science

BAD-LAMP distribution and microdomain organization epitopes present in neurofilament H and mostly in the seemed to be closely linked. Cholesterol depletion efficiently microtubule-associated molecule 1B (MAP1B) (Fischer and affects lipid microdomains and was therefore tested for its Romano-Clarke, 1990). Although the precise function of ability to influence BAD-LAMP distribution. Cortical neurons MAP1B phosphorylation is still debated, experimental were treated for 2 hours with cholesterol-esterase prior to evidence suggests a role in regulating microtubules and actin immunostaining and confocal microscopy visualization (Fig. dynamics as well as being necessary for axonal growth 4E). As expected, cholesterol depletion had a potent effect on (Dehmelt and Halpain, 2004; Del Rio et al., 2004). The perfect GM1 distribution at the plasma membrane. Moreover, BAD- overlapping distribution of BAD-LAMP and Smi31 strongly LAMP vesicular staining was also deeply affected, displaying suggested that microtubules or actin are likely to play in an extensive co-localization with GM1, which was never important role in the organization and the clustering of BAD- observed in normal conditions. Thus, microdomain LAMP-positive vesicles, however BAD-LAMP distribution is organization at the membrane and clustering not dependent on the neuronal polarity. of BAD-LAMP-positive vesicles appeared to be directly BAD-LAMP-positive vesicles were found in the close linked. vicinity of the microtubule network, mirroring, by their accumulation, the intensity of the tubules bundling (Fig. 5B). The cytoskeleton controls the distribution of BAD-LAMP A treatment with the microtubule depolymerizing agent vesicles nocodazole was thus carried out (Fig. 5B). Nocodazole induced This particular organization was likely to be maintained with a strong redistribution of BAD-LAMP-containing vesicles and the active participation of the cytoskeleton and/or associated a loss of BAD-LAMP staining intensity in cortical neurons. proteins. In order to test this hypothesis, several candidate Thus the microtubule network influences the positioning of molecules were followed by confocal microscopy in cortical BAD-LAMP vesicles. Lipid microdomain organization and neurons. Surprisingly, BAD-LAMP containing-vesicles BAD-LAMP distribution in cortical neurons are therefore clustered within punctate zones delimited by staining with the linked, use the microtubule network and possibly depend on Smi31 antibody (Fig. 5A). Smi31 detects phosphorylated MAP1B phosphorylation for their regulation. BAD-LAMP in cortical neurons 359

BAD-LAMP defines a specific subset of early endosomes The unusual distribution of BAD-LAMP vesicles led us to investigate their relationship with other types of sub-cellular compartments. We focused primarily on endocytic organelles, likely to be relevant to a transmembrane molecule bearing a Yxx ⌽ motif in its cytoplasmic tail. BAD-LAMP could not be detected in classical endosomal compartments as judged from its lack of co-localization with LAMP2 (late endosomes and lysosomes) and internalized transferrin-FITC (sorting and recycling endosomes) (Fig. 6A,B) or syntaxin 13 (Hirling et al., 2000). BAD-LAMP was not found in more specialized endocytic compartment such as TI-VAMP-positive vesicles (Coco et al., 1999) (Fig. 6C). Co-labeling with synaptic vesicle proteins such as synaptotagmin 1, RAB3a, VAMP2 revealed some level of co-localization with BAD-LAMP in the growth cones (Fig. 6D-F). Interestingly, co-localization was not observed in other cellular areas and a similar overlapping distribution in the growth cone was observed with TI-VAMP, which is not found enriched in synaptic vesicles (Coco et al., 1999). Thus, this co-distribution in the growth cone probably reflects the difficulty of segregating, at this optical resolution, individual carrier vesicles congregating in the same area of the cone, rather than a true co-localization in the same vesicles. Pre-and post-synaptic transport carriers are derived from trans- Golgi network (TGN) vesicles, which aggregate at initial contacts between axons and dendrites (Sytnyk et al., 2002). We, therefore, examined the possible association of BAD- LAMP with other known vesicular markers of these pathways, such as syntaxin 6 or N-CAM (Sytnyk et al., 2004) (Fig. 6E). We failed to detect co-localization of BAD-LAMP with any of these markers (see supplementary material Fig. S2), suggesting that the molecule is sorted in an uncharacterized type of vesicles, which can accumulate in the growth cone of

Journal of Cell Science developing axon, as well as in defined and organized domains along the cellular processes. BAD-LAMP distribution at the plasma membrane as well as in localized intracellular vesicles suggested a possible shuttling of the molecule between the cell surface and the vesicles. The co-localization of GM1 and BAD-LAMP upon cholesterol depletion suggests that BAD-LAMP vesicles are accessible to plasma membrane constituents under specific conditions. To address this issue, cortical neurons were surface biotinylated at 4°C prior to incubation at 37°C. Biotinylated surface proteins could either diffuse, or be internalized, and their intermixing Fig. 6. Confocal immunofluorescence microscopy analysis of BAD- with BAD-LAMP-positive compartments was evaluated at LAMP transport in cortical neurons. (A) Staining for LAMP2 different time points by confocal microscopy (Fig. 7). (green) and BAD-LAMP (red). (B) Internalized transferrin-FITC Biotinylated proteins were detected rapidly co-localizing with (green) in early and recycling endosomes and BAD-LAMP (red). BAD-LAMP after 5 minutes of internalization. This significant (C) Staining for Ti-VAMP (green) and BAD-LAMP (red). overlapping distribution decreased after 45 minutes, suggesting (D) Staining of a growth cone for synaptotagmin 1 (SYT1, green) that BAD-LAMP-containing organelles could represent a and BAD-LAMP (red). (E) Staining of a growth cone for VAMP2 subset of early endocytic vesicles, rapidly accessible from the (green) and BAD-LAMP (red). (F) Staining for syntaxin 6 (green) neuronal surface and serving as an intermediate step for the and BAD-LAMP (red). BAD-LAMP does not display any significant intracellular sorting of specific surface molecules present in co-localization with LAMP1 and internalized transferrin. Bars, 20 developing neurons. ␮m in A,B,C,F, 10 ␮m in D,E.

BAD-LAMP sorting in transfected neurons neurons (Fig. 8). Surprisingly, endogenous BAD-LAMP To further investigate the distribution of BAD-LAMP, we expression and domain organization were strongly inhibited in generated N- terminally FLAG-tagged and C-terminally GFP- electroporated neurons. Nevertheless, transfected FLAG- tagged BAD-LAMP constructs and monitored their behavior tagged BAD-LAMP was found enriched in vesicles clustered by microscopy in co-transfection experiments of cortical in specific zones along the neurites. Clearly, the tagged protein 360 Journal of Cell Science 120 (2)

Fig. 7. Surface biotinylation reveals the endocytic nature of BAD- LAMP-containing vesicles. Cortical neurons were surface biotinylated for 15 minutes at 4°C, prior to warming at 37°C for different times, fixation and visualization by confocal microscopy. (A) Prior to warming, no significant co-localization of biotinylated proteins with BAD- LAMP was observed. Co- localization was evaluated using the Image J image analysis software. A low Pearson’s coefficient and strong negative pixel shift are both indicative of the absence of staining overlap (right). (B) After 5 minutes of endocytosis at 37°C, extensive co-localization of biotinylated proteins (green) was observed with BAD-LAMP (red) in neurites (arrowheads), as also shown by a higher Pearson’s coefficient and the absence of pixel shift (right). (C) After 45 minutes of endocytosis co-localization of BAD-LAMP with biotin is decreased as shown by a decreased Pearson’s coefficient and negative Journal of Cell Science pixel shift (right). Bar, 10 ␮m.

is not addressed in conventional endo/lysosomes as judged by 37°C. Inside the cell, it was detected in a different compartment its lack of co-localization with LAMP2 (Fig. 8A), internalized from conventional endo/lysosomes as shown by the absence of transferrin (not shown) or cholera toxin (see supplementary co-localization with co-transfected BAD-GFP, LAMP2 (Z3 material Fig. S3). The exact location of FLAG-BAD-LAMP in and arrows) and internalized cholera toxin (supplementary the cell body was difficult to establish since its over-expression material Fig. S3A). After 30 minutes of synchronous uptake induced an accumulation of the molecule in the ER and Golgi (Z4 and arrowheads), co-localization of the antibodies with network. Surprisingly, the C-terminally GFP-tagged, BAD- BAD-LAMP-GFP and LAMP 2 indicated that BAD-LAMP LAMP (BAD-GFP) was found accumulating in LAMP2- can reach conventional endocytic compartments, after being positive lysosomal compartments (Fig. 8A, arrowheads in Z1). internalized from the surface. Surprisingly, this co-localization Therefore, the BAD-LAMP cytoplasmic tail contains a cryptic was more evident in the more discrete LAMP2-positive lysosomal retention motif, which is revealed by the addition of organelles present in the neurite (late endosomes, arrowheads) the GFP moiety. This observation also suggests that the than in the large lysosomes observed in the cell body (Fig. 8B). cytoplasmic tail of BAD-LAMP is actively interacting with, or We next investigated the contribution of tyrosine 276 to modified by, molecules that promote its sorting away from BAD-LAMP trafficking by introducing a mutational change to traditional endocytic compartments. Co-localization of BAD- alanine at this position (Tyr276Ala). The FLAG-tagged mutant GFP and FLAG-tagged BAD-LAMP was observed in discrete was also found accumulating in the ER and Golgi network of vesicles in neurites (Fig. 8A, Z2 arrowheads), despite the fact transfected neurons. However, the fraction of the mutant that that BAD-GFP was found mostly accumulating in large exited these organelles accumulated at the surface of the lysosomes in the cell body. This demonstrates that a small neurites in a manner very distinct from the normal molecule fraction of BAD-GFP can be sorted normally. (wild type), which was mostly found in intracellular vesicles We next evaluated the internalization dynamics of BAD- (supplementary material Fig. S3B). Similar results were LAMP by using the N-terminally FLAG-tagged construct and obtained with a construct lacking the entire cytoplasmic tail of by monitoring FLAG antibody uptake after cold binding (Fig. BAD-LAMP (not shown). Thus, tyrosine 276 is directly 8B). The antibody was rapidly endocytosed after 5 minutes at involved in intracellular addressing of BAD-LAMP and allows BAD-LAMP in cortical neurons 361

its internalization from the surface. FLAG antibody uptake after binding in the cold was performed in neurons expressing FLAG-BAD-LAMP Tyr276Ala. Transfected cells remained mostly antibody-decorated at the surface 30 minutes after warming at 37°C (supplementary material Fig. S3C). Thus, BAD-LAMP is probably cycling between the plasma membrane and a subset of endocytic vesicles.

BAD-LAMP recycles in HeLa cells In order to further dissect the molecular mechanisms governing BAD-LAMP endocytosis, we studied the distribution and transport of transfected BAD-LAMP in a cell type easy to manipulate, such as HeLa and mouse NIH 3T3 cells. In HeLa cells, FLAG-BAD-LAMP was found at the cell surface and in internalized transferrin- containing vesicles distributed in the vicinity of the plasma membrane, whereas the Tyr276Ala mutant accumulated only at the cell surface (Fig. 9A). No co- localization was found in LAMP1-positive late endosomes or lysosomes, nor with co- transfected DC-LAMP tagged with GFP (Fig. 9B), another lysosomal resident of the LAMP family (de Saint-Vis et al., 1998). These observations were confirmed after Percoll density gradient subcellular fractionation of transfected HeLa cells (supplementary material Fig. S4). BAD-

Journal of Cell Science LAMP was mostly detected in the low density fractions of the gradient containing plasma membrane, ER and early endosomes, but it was absent from the high density fractions containing lysosomes, as indicated by ␤-hexosaminidase activity. Thus, most of transfected BAD-LAMP was found on the cell surface contrasting with transfected neurons in which BAD-LAMP mostly accumulated intracellularly, underlining the specificity of its sorting even when over-expressed. Anti-FLAG antibody uptake in Fig. 8. Localization of FLAG-tagged BAD-LAMP in transfected cortical neurons. transfected cells saturated with FITC- Cortical neurons co-transfected with BAD-GFP and FLAG-BAD-LAMP were visualized by confocal microscopy. (A) Staining for FLAG antibody (red), BAD-GFP (green) and transferrin (FITC-TF), confirmed that LAMP2 (white). FLAG-BAD-LAMP does not co-localize with LAMP 2. BAD-GFP is BAD-LAMP could be internalized rapidly targeted to lysosomes upon addition of the GFP moiety at the C-terminal end of BAD- in sorting endosomes (Fig. 9C). LAMP. Bar, 20 ␮m; 10 ␮m for high magnification of Z1 and 5 ␮m for Z2. (B) Interestingly, 15 minutes after uptake Internalization of FLAG antibody (red) in transfected neurons for indicated times and BAD-LAMP was found present in staining for LAMP2 (white). High magnifications reveal a late accessibility of BAD- transferrin-positive recycling endosomes LAMP into LAMP2 positive compartments in neurites. Bars, 20 ␮m; 5 ␮m for high clustered around the microtubule magnification. organizing center, suggesting that BAD- LAMP could recycle to the plasma membrane after internalization. This hypothesis was supported again a difference with neurons, in which the antibodies could by the poor co-localization of the antibody with LAMP1 after be detected in discrete LAMP1-positive compartment 45 45 minutes of uptake, indicating that the molecule does not minutes after uptake. efficiently reach late endocytic compartments. This underlines We next investigated the molecular mechanisms involved in 362 Journal of Cell Science 120 (2)

Fig. 9. BAD-LAMP is targeted to early endosomes and recycles in HeLa cells. HeLa cells transfected with FLAG-BAD- LAMP were submitted to immunofluorescent staining and confocal microscopy visualization. (A) FLAG-tagged BAD-LAMP (anti-flag antibody, red) was found at the cell surface and in internalized transferrin-FITC-containing endosomes. Cytoplasmic tail tyrosine 276 mutant (Tyr276-Ala) was found accumulating at the surface of transfected cells (anti-flag antibody, red) with little intracellular distribution (transferrin-FITC, green). (B) Transfected FLAG-BAD-LAMP is not detected in LAMP1- (blue) and DC-LAMP (green)-positive late endosomes and lysososomes. (C) Kinetics of FLAG antibody uptake after cold binding on the surface of transfected HeLa cells. Only transfected cells accumulate the antibody (red) on their surface, which upon warming reaches rapidly sorting (5 minutes) and recycling (15 minutes) endosomes containing transferrin-FITC (green). No co- localization with LAMP1 (white, 45 minutes) could be observed, suggesting that BAD-LAMP and associated antibodies do not access the late endocytic pathway. (D) Co-expression of dynamin dominant negative mutant A44K (right panel, green) in FLAG-BAD-LAMP-transfected HeLa cells prevents the internalization of associated flag antibodies (right panel, red), whereas expression of wild-type dynamin has no effect (green, left panel). (E) Co- transfection of HeLa cells with FLAG- BAD-LAMP (anti-BAD-LAMP, red) and Journal of Cell Science pSuper control plasmid (left) has no effect on the internalization of associated flag antibodies (green). Conversely RNAi inhibition of the clathrin adaptor AP2 blocks flag antibodies uptake (green). Bars, 20 ␮m.

BAD-LAMP endocytosis. Experiments performed in cells co- shown), suggesting that BAD-LAMP is internalized transfected with wild-type GTPase dynamin II or dominant- constantly through a dynamin/AP2-dependent endocytic negative mutant A44K, indicated that BAD-LAMP pathway. internalization is mediated in a dynamin-dependent manner, Interestingly, monitoring of surface anti-FLAG antibody by since antibody internalization was abolished in cells FACS also indicated that the molecule was rapidly internalized expressing dynamin A44K (Fig. 9D and control, between 5 and 7.5 minutes after warming (supplementary supplementary material Fig. S4C). In order to further define material Fig. S4B). Surface levels of antibodies then re- the endocytic pathway used by BAD-LAMP to enter the cell, increased after 10 minutes, to be diminished again but with a we used an RNA inhibition approach to reduce the expression relatively slower internalization rate. These observations of molecules involved in protein triage from the surface, such confirm that BAD-LAMP and associated antibodies constantly as the clathrin adaptor AP2 (Dugast et al., 2005; McCormick recycle to the plasma membrane with a relatively high et al., 2005). Antibody uptake was monitored by efficiency. immunostaining and FACS detection after binding at 4°C and internalization at 37°C. Cells co-transfected with FLAG- Discussion BAD-LAMP and control RNAi plasmid showed rapid BAD-LAMP sequence analysis clearly indicates that it internalization of the antibody (Fig. 9E), whereas RNAi represents a new member of the LAMP family. However, its depletion of AP2 clearly inhibited BAD-LAMP expression pattern and intracellular distribution are internalization as well as transferin uptake (supplementary unconventional compared to other LAMP family members, material Fig. S4C). In cells depleted for AP2, higher surface which show a widespread expression and specifically levels of BAD-LAMP were also consistently detected (not accumulate in the lysosomes. BAD-LAMP in cortical neurons 363

Our observations on BAD-LAMP intracellular distribution increase in functional synapses in the cortex is strikingly are clearly indicative of a strong regulation of its trafficking in mirrored by the expression of BAD-LAMP during a subset of early endosomes. Although we have not been able corticogenesis. Thus, it appears very possible that BAD- to identify molecular markers able to identify these organelles, LAMP, together with MAP1B, is involved in the terminal the absence of transferrin or synaptotagmin 1, as well as late maturation steps and/or function of defined cortical neurone endosomal markers such as LAMP1 suggests that these populations. vesicles represent a distinct class of neuronal endosomes. The Most of our observations point towards a link between BAD- kinetics of biotinylated proteins and antibody uptake indicate LAMP and endocytosis. The transformation of a transient that they can serve as sorting platforms, prior to transport to contact between two neurons into a stable and functional other organelles, which are positive for LAMP1, but only synapse requires major changes in the membrane composition represent a minor fraction of the neuronal organelles of the respective neuronal surface areas. Endocytic processes containing LAMP1. have been implicated in the regulation of synaptic function and We have shown that BAD-LAMP, through an interaction plasticity in vertebrates (Vissel et al., 2001) and in Drosophila with its YKHM domain, requires dynamin and AP2 to be (Dickman et al., 2006). For example, NMDA receptors are internalized and sorted towards the early endocytic recycling subject to constitutive (Roche et al., 2001) as well as agonist- pathway of transfected HeLa cells. LAMP1 has also been induced (Vissel et al., 2001) internalization through clathrin- shown to require the AP2 adaptor, but its sorting is directed mediated endocytosis. Interestingly, in situ hybridization for towards lysosomes (Janvier and Bonifacino, 2005). NMDAR1 resulted in strong cellular labeling in neurons of Interestingly, modification of the BAD-LAMP C-terminal layers II/III, V and VI (Rudolf et al., 1996), resembling the domain by GFP deeply affects its transport in neurons and pattern we found for BAD-LAMP in the postnatal cortex. The demonstrates the existence of an active sorting pathway in BAD-LAMP-containing endocytic compartment could these cells, which normally prevents the accumulation of BAD- therefore play a regulatory role in these events by maintaining LAMP in the lysosomes. The YKHM domain is a relatively specific zones in the neuronal projections. weak consensus endosomal/lysosomal addressing signal (Bonifacino and Traub, 2003), although it is also found in Materials and Methods CTLA-4, a molecule known to recycle upon activation of T Bioinformatics cells (Linsley et al., 1996). As suggested by its early The BAD-LAMP protein sequence ID in Ensembl database is endosomes distribution, we could show that BAD-LAMP also ENSMUSP00000061180. All LAMPs sequences were aligned using CLUSTALW package (EBI) and results were treated with TreeView for phylogeny. Image recycles in transfected HeLa cells. Whether this is the case in analysis was performed with the Image J software and the plugin JacoP. neurons remains to be further investigated, although it clearly indicates that the ‘YKHM’ domain is not normally used as a Animals and tissues lysosomal addressing signal. All animals were treated according to protocols approved by the French Ethical One of the features of BAD-LAMP-containing organelles is Committee. CD1 mice (Iffa-Credo, Town?, France) were used to determine the Bad- lamp expression pattern. Disabled 1 deficient Scrambler mice were purchased from their clustered distribution. This distribution mirrors the Jackson Laboratories. The day of the vaginal plug appearance was considered as

Journal of Cell Science organization of the different microdomains at the cell surface. embryonic day (E)0.5 and the day of the birth as postnatal day (P)0. For in situ Whether BAD-LAMP-containing organelles participate in the hybridization and immunohistochemistry, postnatal and adult brains were collected after the animals were anaesthetized with a lethal dose of Rompun/Imalgen 500 and maintenance of this organization within the neuritic plasma intracardially perfused with 4% paraformaldehyde (PFA). Brains were further fixed membrane remains to be proved. Nevertheless their sensitivity in 4% PFA overnight. Adult brains were sectioned at 80 ␮m on a vibratome whereas to cholesterol-depleting drugs suggests that microdomains and P2-P12 brains were cryoprotected in 20% sucrose/PBS, frozen in OCT compound and sectioned at 16 ␮m on a cryostat. Sections collected on Superfrost slides were BAD-LAMP-containing vesicles are functionally linked. treated as described below. Strikingly, the clustering the BAD-LAMP-containing vesicles is also defined by the distribution of the phosphorylated Molecular biology epitopes (SMI31) found on the microtubule-associated protein Northern blot analysis was done with FirstChoice Northern Blot Mouse Blot I MAP1B or neurofilament H (Fischer and Romano-Clarke, (Ambion) using a probe corresponding to exons 4, 5 and 6 of BAD-LAMP (clone IMAGE 2588577). 2 mg of Trizol extracted total mouse cortex RNA was used for 1990). MAP1B in the cortex has been strongly implicated in reverse transcription with oligo(dT) primers. The cDNAs coding for BAD-LAMP synapse formation and function (Kawakami et al., 2003). Such were amplified after 30 cycles of PCR using Taq polymerase. Sense primer was a role has been recently functionally demonstrated through the ACC GGC CAC TTT GAG GGA and antisense GGG GCG GCC TTT GCA GCA (1.5 kb). PCR products were cloned into pGEM-Teasy plasmid (Promega). BAD- observation that mice lacking the phosphorylated form of LAMP-GFP fusion construct was constructed using pEGFP-NI vector (Clontech). MAP1B specifically in the hippocampus, show deficits in long- FLAG-BAD-LAMP was constructed using pTEJ-8-HA- FLAG plasmid (Didier term potentiation in the Schaeffer collaterals pathway (Zervas Marguet, Marseille, France). A tyrosine mutant of BAD, FLAG -BAD-Tyr-276-Ala et al., 2005). Therefore, it is conceivable that MAP1B is was produced by targeted PCR mutagenesis. FLAG-BAD-LAMP cDNA were transferred into pCX-MCS2 plasmid, a pCAAGS derived plasmid with an extended implicated in the positioning and transport of BAD-LAMP cloning site (a kind gift from Xavier Morin, Marseille, France). Dynamin-GFP wt vesicles at sites of postsynaptic densities on the dendrites of plasmid and dynamin-GFP A44K were kindly given by M. McNiven, Rochester, cortical neurons, and that this process could be essential for MN. RNAi constructs pSUPER AP2 ␮2 and pSUPER control were a gift from stabilization, function and plasticity of cortical synapses. Philippe Benaroch, Paris, France. Indeed, BAD-LAMP expression is temporally and spatially In situ hybridization and immunohistochemistry restricted in cortical neurons of layers II, III and V. Whereas IMAGE clone 2588577 was used to make an antisense RNA probe. Antisense RNA the generation and migration of cortical neurons in rodents is probes for Bad-lamp, Cux2 (Zimmer et al., 2004) and ER81 (Lin et al., 1998) were an embryonic process, synaptogenesis in the cortex occurs in generated using the Dig-RNA labelling kit (Roche). Single in situ hybridization and combined in situ hybridization with immunohistochemistry were described the postnatal animal with a peak between P10 and P15 to previously (Tiveron et al., 1996; Zimmer et al., 2004) for all probes and the NeuN approach adult values (Micheva and Beaulieu, 1996). This monoclonal mouse IgG (MAB377; Chemicon). 364 Journal of Cell Science 120 (2)

Antibodies and immunocytochemistry R., Matteoli, M., Louvard, D. and Galli, T. (1999). Subcellular localization of tetanus A polyclonal rabbit anti-BAD-LAMP was raised in rabbit against two peptides neurotoxin-insensitive vesicle-associated membrane protein (VAMP)/VAMP7 in of the BAD-LAMP cytoplasmic tail, KMTANQVQIPRDRSQC and neuronal cells: evidence for a novel membrane compartment. J. Neurosci. 19 , 9803- KQIPRDRSQYKHMC. Anti-synaptotagmin 1 and anti-RAB3a/b antibodies were 9812. obtained from P. Di Camilli, New Haven, CT, anti-FLAG M2 antibody and anti- ␤- de Saint-Vis, B., Vincent, J., Vandenabeele, S., Vanbervliet, B., Pin, J. J., Ait-Yahia, tubulin-Cy3 were obtained from Sigma, anti-VAMP2 from SYSY, anti-syntaxin 6 S., Patel, S., Mattei, M. G., Banchereau, J., Zurawski, S. et al. (1998). A novel from BD Transduction Laboratories; Aanti-PrP (6H4) was from Prionics (Schlieren, lysosome-associated membrane glycoprotein, DC-LAMP, induced upon DC Switzerland), anti-syntaxin 13 from Stressgen (Ann Arbor, MI); anti-Thy1 from maturation, is transiently expressed in MHC class II compartment. Immunity 9, 325- 336. Michel Pierres, Marseille, France; human Alexa Fluor 568-Tf from Molecular Dehmelt, L. and Halpain, S. (2004). Actin and microtubules in neurite initiation: are Probes; mouse FITC-Tf from Rockland (Gilbertsville, PA), Cy3- ␤-tubulin from MAPs the missing link? J. Neurobiol. 58 , 18-33. Sigma, FITC-cholera toxin B subunit (GM1 staining) from Sigma; anti-NCAM H28 Del Rio, J. A., Gonzalez-Billault, C., Urena, J. M., Jimenez, E. M., Barallobre, M. from C. Goridis, Paris, France; Anti-Ti-VAMP from T. Galli, Paris, France; Rat anti- J., Pascual, M., Pujadas, L., Simo, S., La Torre, A., Wandosell, F. et al. (2004). mouse LAMP2 from I. Mellman (New Haven, CT) and anti-human LAMP1 from MAP1B is required for Netrin 1 signaling in neuronal migration and axonal guidance. Abcam. All FITC and Cy3-5 secondary antibodies were from Jackson Curr. Biol. 14 , 840-850. ImmunoResearch. All Alexa secondary antibodies were from Molecular Probes. Dickman, D. K., Lu, Z., Meinertzhagen, I. A. and Schwarz, T. L. (2006). Altered Immunofluorescence and confocal microscopy was performed with a Zeiss LSM synaptic development and active zone spacing in endocytosis mutants. Curr. Biol. 16 , 510 microscope as described previously (Cappello et al., 2004). Vibratome adult 591-598. brain sections were immunostained with rabbit anti-BAD-LAMP and mouse anti- Dugast, M., Toussaint, H., Dousset, C. and Benaroch, P. (2005). AP2 clathrin adaptor MAP2. complex, but not AP1, controls the access of the major histocompatibility complex (MHC) class II to endosomes. J. Biol. Chem. 280 , 19656-19664. Cell culture Eskelinen, E. L., Illert, A. L., Tanaka, Y., Schwarzmann, G., Blanz, J., von Figura, HeLa cells were grown in DMEM containing 10% FCS. Cortical neurons were K. and Saftig, P. (2002). Role of LAMP-2 in lysosome biogenesis and autophagy. Mol. prepared from E15.5 embryonic cortices. Cortices were dissected out in HBSS, Biol. Cell 13 , 3355-3368. Eskelinen, E. L., Tanaka, Y. and Saftig, P. (2003). At the acidic edge: emerging treated for 15 minutes at 37°C in Trypsin/EDTA-HBSS (Invitrogen), washed once functions for lysosomal membrane proteins. Trends Cell Biol. 13 , 137-145. in NeuroBasal medium (NB; Invitrogen) complemented with 10% horse serum to Fischer, I. and Romano-Clarke, G. (1990). Changes in microtubule-associated protein block trypsin activity and washed once more in NB alone. Cortical neurons were MAP1B phosphorylation during rat brain development. J. Neurochem. 55 , 328-333. dissociated, plated on glass coverslips in NB with B27 complement, 2 mM L- Gruenberg, J. and Stenmark, H. (2004). The biogenesis of multivesicular endosomes. glutamine and 50 ␮g/ml penicillin/streptomycin (Invitrogen) and cultured for 3 days Nat. Rev. Mol. Cell Biol. 5, 317-323. at 37°C, 5% CO 2. Coverslips were coated overnight with poly-L-lysine (10 ␮g/ml). Hirling, H., Steiner, P., Chaperon, C., Marsault, R., Regazzi, R. and Catsicas, S. (2000). Syntaxin 13 is a developmentally regulated SNARE involved in neurite Transfection and internalization experiments outgrowth and endosomal trafficking. Eur. J. Neurosci. 12 , 1913-1923. Neurons were electroporated using Amaxa Nucleofactor Kit according to the Janvier, K. and Bonifacino, J. S. (2005). Role of the endocytic machinery in the sorting manufacturer’s instructions. HeLa cells were grown on coverslips and transfected of lysosome-associated membrane proteins. Mol. Biol. Cell 16 , 4231-4242. using Lipofectamine 2000 (Invitrogen) using the manufacturer’s protocol. After 8- Kawakami, S., Muramoto, K., Ichikawa, M. and Kuroda, Y. (2003). Localization of 24 hours of transfection the HeLa cells were processed to study internalization microtubule-associated protein (MAP) 1B in the postsynaptic densities of the rat kinetics or fixed using 3% paraformaldehyde. Internalization assays were performed cerebral cortex. Cell. Mol. Neurobiol. 23 , 887-894. using FITC-conjugated transferrin or unconjugated antibodies. The cells were first Kennedy, M. J. and Ehlers, M. D. (2006). Organelles and trafficking machinery for postsynaptic plasticity. Annu. Rev. Neurosci. 29 , 325-362. incubated for 20 minutes at 37°C in DMEM/100 mM HEPES to eliminate Lin, J. H., Saito, T., Anderson, D. J., Lance-Jones, C., Jessell, T. M. and Arber, S. endogenous transferrin. Cells were incubated for 15 minutes at 4°C with ligand (1998). Functionally related motor neuron pool and muscle sensory afferent subtypes and/or antibody and washed twice in ice-cold PBS before incubation with DMEM, defined by coordinate ETS gene expression. Cell 95 , 393-407. 1% BSA, 100 mM Hepes at 37°C, for different times prior to fixation and Linsley, P. S., Bradshaw, J., Greene, J., Peach, R., Bennett, K. L. and Mittler, R. S. immunocytochemistry. Neurons were processed identically in NB medium. Cortical (1996). Intracellular trafficking of CTLA-4 and focal localization towards sites of TCR neurone biotinylation was performed using EZ-Link Sulfo-NHS-Biotin kit (Pierce) engagement. Immunity 4, 535-543. with a 15-minute reaction time at 4°C, followed by three washes with ice-cold PBS Madore, N., Smith, K. L., Graham, C. H., Jen, A., Brady, K., Hall, S. and Morris, Journal of Cell Science containing 10 mM glycine. Cells were incubated for 5 and 45 minutes at 37°C to R. (1999). Functionally different GPI proteins are organized in different domains on allow endocytosis of biotinylated membrane proteins, prior to fixation and the neuronal surface. EMBO J. 18 , 6917-6926. immunostaining. Maxf ield, F. R. and Tabas, I. (2005). Role of cholesterol and lipid organization in disease. Nature 438 , 612-621. Immunoblots and immunoprecipitation McCormick, P. J., Martina, J. A. and Bonifacino, J. S. (2005). Involvement of clathrin 1% Triton X-100 cell extracts complemented with protease inhibitors cocktail and AP-2 in the trafficking of MHC class II molecules to antigen-processing (Roche) were immunoblotted after separation by 12% or a 7-17% gradient SDS- compartments. Proc. Natl. Acad. Sci. USA 102 , 7910-7915. PAGE. Immunoprecipitation with anti-FLAG antibody and N-glycosidase F or Micheva, K. D. and Beaulieu, C. (1996). Quantitative aspects of synaptogenesis in the rat barrel field cortex with special reference to GABA circuitry. J. Comp. Neurol. 373 , endoglycosidase H (Calbiochem) treatment were performed as described previously 340-354. (Cappello et al., 2004). Rice, D. S. and Curran, T. (1999). Mutant mice with scrambled brains: understanding the signaling pathways that control cell positioning in the CNS. Genes Dev. 13 , 2758- This work was supported by grants to P.P. from CNRS-INSERM, 2773. the Ministère de la Recherche et de la Technologie (ACI), La Ligue Roche, K. W., Standley, S., McCallum, J., Dune Ly, C., Ehlers, M. D. and Wenthold, R. J. (2001). Molecular determinants of NMDA receptor internalization. Nat. Neurosci. Nationale Contre le Cancer and the Human Frontier of Science 4, 794-802. Program. A.D. is supported by the MRT and ARC. P.P. is part of the Rudolf, G. D., Cronin, C. A., Landwehrmeyer, G. B., Standaert, D. G., Penney, J. B., EMBO Young Investigator Program. H.C. was supported by the Jr and Young, A. B. (1996). Expression of N-methyl-D-aspartate glutamate receptor French Fondation pour la Recherche sur le Cerveau (FRC), the subunits in the prefrontal cortex of the rat. Neuroscience 73 , 417-427. Saftig, P., Tanaka, Y., Lullmann-Rauch, R. and von Figura, K. (2001). Disease model: Association Francaise contre le Myopathies (AFM) and the European LAMP-2 enlightens Danon disease. Trends Mol. Med. 7, 37-39. Community through the NOE NeuroNE. We thank the PICsL imaging Santuccione, A., Sytnyk, V., Leshchyns’ka, I. and Schachner, M. (2005). Prion protein core facility for expert technical assistance. We are grateful to Vilma recruits its neuronal receptor NCAM to lipid rafts to activate p59fyn and to enhance Arce for expert technical advice. neurite outgrowth. J. Cell Biol. 169 , 341-354. Shikano, S., Bonkobara, M., Zukas, P. K. and Ariizumi, K. (2001). Molecular cloning of a dendritic cell-associated transmembrane protein, DC-HIL, that promotes RGD- References dependent adhesion of endothelial cells through recognition of heparan sulfate Andrejewski, N., Punnonen, E. L., Guhde, G., Tanaka, Y., Lullmann-Rauch, R., proteoglycans. J. Biol. Chem. 276 , 8125-8134. Hartmann, D., von Figura, K. and Saftig, P. (1999). Normal lysosomal morphology Sunyach, C., Jen, A., Deng, J., Fitzgerald, K. T., Frobert, Y., Grassi, J., McCaffrey, and function in LAMP-1-deficient mice. J. Biol. Chem. 274 , 12692-12701. M. W. and Morris, R. (2003). The mechanism of internalization of Bonifacino, J. S. and Traub, L. M. (2003). Signals for sorting of transmembrane proteins glycosylphosphatidylinositol-anchored prion protein. EMBO J. 22 , 3591-3601. to endosomes and lysosomes. Annu. Rev. Biochem. 72 , 395-447. Sytnyk, V., Leshchyns’ka, I., Delling, M., Dityateva, G., Dityatev, A. and Schachner, Cappello, F., Gatti, E., Camossetto, V., David, A., Lelouard, H. and Pierre, P. (2004). M. (2002). Neural cell adhesion molecule promotes accumulation of TGN organelles Cystatin F is secreted, but artificial modification of its C-terminus can induce its at sites of neuron-to-neuron contacts. J. Cell Biol. 159 , 649-661. endocytic targeting. Exp. Cell Res. 297 , 607-618. Sytnyk, V., Leshchyns’ka, I., Dityatev, A. and Schachner, M. (2004). Trans-Golgi Coco, S., Raposo, G., Martinez, S., Fontaine, J. J., Takamori, S., Zahraoui, A., Jahn, network delivery of synaptic proteins in synaptogenesis. J. Cell Sci. 117 , 381-388. BAD-LAMP in cortical neurons 365

Tanaka, Y., Guhde, G., Suter, A., Eskelinen, E. L., Hartmann, D., Lullmann-Rauch, dependent tyrosine dephosphorylation of NMDA receptors is independent of ion flux. R., Janssen, P. M., Blanz, J., von Figura, K. and Saftig, P. (2000). Accumulation of Nat. Neurosci. 4, 587-596. autophagic vacuoles and cardiomyopathy in LAMP-2-deficient mice. Nature 406 , 902- Zervas, M., Opitz, T., Edelmann, W., Wainer, B., Kucherlapati, R. and Stanton, P. 906. K. (2005). Impaired hippocampal long-term potentiation in microtubule-associated Tiveron, M. C., Hirsch, M. R. and Brunet, J. F. (1996). The expression pattern of the protein 1B-deficient mice. J. Neurosci. Res. 82 , 83-92. transcription factor Phox2 delineates synaptic pathways of the autonomic nervous Zimmer, C., Tiveron, M. C., Bodmer, R. and Cremer, H. (2004). Dynamics of Cux2 system. J. Neurosci. 16 , 7649-7660. expression suggests that an early pool of SVZ precursors is fated to become upper Vissel, B., Krupp, J. J., Heinemann, S. F. and Westbrook, G. L. (2001). A use- cortical layer neurons. Cereb. Cortex 14 , 1408-1420. Journal of Cell Science Fig. S1. Homology of BAD-LAMP with other LAMP family members. (A) Comparison of BAD-LAMP sequences with others members of LAMP family and DC-HIL. All alignments were done using EMBOSS needle (Needleman-Wunsch global alignment) to obtain a percentage of similarity. (B) BAD-LAMP conservation among species. All alignments were done using SSEARCHp able to calculate identity percentage. Drosophila melanogaster peptide ID: CG32225-PA; Takifugu rubripes peptide ID: NEWSINFRUP00000148857; Tetraodon nigroviridis peptide ID: GSTENP00020285001; Ratus norvegicus peptide ID: ENSRNOP00000007274; Anopheles gambiae peptide ID: ENSANGP00000022286; Homo sapiens peptide ID: ENSP00000246070; Mus musculus peptide ID: ENSMUSP00000061180; Gallus gallus : ENSGALP00000014483; Bos taurus: ENSBTAP00000010487

Fig. S2. Summary of molecular markers and BAD-LAMP distribution in cortical neurones. A list of the different molecules imaged by confocal microscopy in cortical neurones is given along with their level of co-localization with BAD-LAMP and a representative staining (BAD-LAMP in red, markers in green).

Fig. S3. Tyrosine 276 is necessary for BAD-LAMP endocytosis in neurones. (A) Internalization of FLAG antibody (red) and cholera toxin (green) in FLAG-BAD-LAMP-transfected neurones for indicated time. !:`5%%-I8%^!_%-Q` 1H:C%JV%`QJV6%HQ -transfected with BAD-GFP and FLAG-BAD-LAMP cytoplasmic tail tyrosine 276 mutant (Tyr-276-Ala) were visualized by confocal microscopy. Staining for FLAG antibody (red), BAD-GFP (green) and LAMP2 (white). B:`5%%-I8%^-_%7J V`J:C1<: 1QJ%Q`%:;<=%:J 1GQR7% (red) in Tyr-276-Ala-transfected neurones and staining for LAMP2 (white). Tyrosine 276 is necessary for proper BAD-;

Fig. S4. BAD-LAMP recycles in transfected HeLa cells. (A) Subcellular Percoll gradient fractionation showing that in transfected HeLa cells BAD-LAMP is mostly localized in light density fractions and is :G6VJ %``QI%.V:07%RVJ61 7%``:H 1QJ6%HQJ :1J1J$% .V%C76Q6QI:C%VJ<7IV% -hexosaminidase. (B) Representative FACS staining experiment (of three), showing the level of internalized Flag antibody against time at the surface of FLAG-BAD-LAMP-transfected HeLa cells. The timing of internalization and the recovery of surface staining at 10 minutes indicates the BAD-LAMP recycles in HeLa cells. (C) Control experiment demonstrating inhibition of transferrin uptake (red) after dynamin A44K expression (green). (D) Control experiment demonstrating inhibition of transferrin uptake (red) by AP2 RNAi transfection (green).

Annexe 2

“Novel insights into the relationships between dendritic cell subsets in human and mouse revealed by genome- wide expression profiling”

Article publié dans la revue Genome Biology, 2008 Jan 24;9(1):R17. Research2008Robbinset al.Volume 9, Issue 1, Article R17 Open Access Novel insights into the relationships between dendritic cell subsets in human and mouse revealed by genome-wide expression profiling Scott H Robbins *†‡‡‡ , Thierry Walzer *†‡ , Doulaye Dembélé §¶¥# , Christelle Thibault §¶¥# , Axel Defays *†‡ , Gilles Bessou *†‡ , Huichun Xu ** , Eric Vivier *†‡†† , MacLean Sellars §¶¥# , Philippe Pierre *†‡ , Franck R Sharp ** , Susan Chan §¶¥# , Philippe Kastner §¶¥# and Marc Dalod *†‡

Addresses: *CIML (Centre d'Immunologie de Marseille-Luminy), Université de la Méditerranée, Parc scientifique de Luminy case 906, Marseille F-13288, France. †U631, INSERM (Institut National de la Santé et de la Recherche Médicale), Parc scientifique de Luminy case 906, Marseille F-13288, France. ‡UMR6102, CNRS (Centre National de la Recherche Scientifique), Parc scientifique de Luminy case 906, Marseille F-13288, France. §Hematopoiesis and leukemogenesis in the mouse, IGBMC (Institut de Génétique et de Biologie Moléculaire et Cellulaire), rue Laurent Fries, ILLKIRCH F-67400, France. ¶U596, INSERM, rue Laurent Fries, ILLKIRCH F-67400, France. ¥UMR7104, CNRS, rue Laurent Fries, ILLKIRCH F-67400, France. #UM41, Université Louis Pasteur, rue Laurent Fries, Strasbourg F-67400, France. ** The Medical Investigation of Neurodevelopmental Disorders Institute, University of California at Davis Medical Center, Sacramento, CA 95817, USA. †† Hôpital de la Conception, Assistance Publique-Hôpitaux de Marseille, Boulevard Baille, Marseille F-13385, France. ‡‡ Current address: Genomics Institute of the Novartis Research Foundation, John Jay Hopkins Drive, San Diego, CA 92121, USA.

Correspondence: Marc Dalod. Email: [email protected]

Published: 24 January 2008 Received: 28 August 2007 Revised: 19 December 2007 Genome Biology 2008, 9: R17 (doi:10.1186/gb-2008-9-1-r17) Accepted: 24 January 2008 The electronic version of this article is the complete one and can be found online at http://genomebiology.com/2008/9/1/R17

© 2008 Robbins et al .; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Profiling

Genome-wideventional dendritic dendritic expressioncell cell subsets subsets.

profiling of mouse and human leukocytes reveal conserved transcriptional programs of plasmacytoid or con-

Abstract

Background: Dendritic cells (DCs) are a complex group of cells that play a critical role in vertebrate immunity. Lymph-node resident DCs (LN-DCs) are subdivided into conventional DC (cDC) subsets (CD11b and CD8 ! in mouse; BDCA1 and BDCA3 in human) and plasmacytoid DCs (pDCs). It is currently unclear if these various DC populations belong to a unique hematopoietic lineage and if the subsets identified in the mouse and human systems are evolutionary homologs. To gain novel insights into these questions, we sought conserved genetic signatures for LN-DCs and in vitro derived granulocyte-macrophage colony stimulating factor (GM-CSF) DCs through the analysis of a compendium of genome-wide expression profiles of mouse or human leukocytes.

Results: We show through clustering analysis that all LN-DC subsets form a distinct branch within the leukocyte family tree, and reveal a transcriptomal signature evolutionarily conserved in all LN- DC subsets. Moreover, we identify a large gene expression program shared between mouse and human pDCs, and smaller conserved profiles shared between mouse and human LN-cDC subsets. Importantly, most of these genes have not been previously associated with DC function and many have unknown functions. Finally, we use compendium analysis to re-evaluate the classification of interferon-producing killer DCs, lin -CD16 +HLA-DR + cells and in vitro derived GM-CSF DCs, and show that these cells are more closely linked to natural killer and myeloid cells, respectively.

Conclusion: Our study provides a unique database resource for future investigation of the evolutionarily conserved molecular pathways governing the ontogeny and functions of leukocyte subsets, especially DCs.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.2

Background rately classifying pDCs within leukocyte subsets are recent Dendritic cells (DCs) were initially identified by their unique reports describing cell types bearing mixed phenotypic and ability to present antigen for the priming of naïve CD4 and functional characteristics of NK cells and pDCs in the mouse CD8 T lymphocytes [1]. DCs have more recently been shown [15,16]. Collectively, these findings raise the question of how to be key sentinel immune cells able to sense, and respond to, closely related human and mouse pDCs are to one another or danger very early in the course of an infection due to their to cDCs as compared to other leukocyte populations. expression of a broad array of pattern recognition receptors [2]. Indeed, DCs have been shown to play a major role in the Global transcriptomic analysis has recently been shown to be early production of effector antimicrobial molecules such as a powerful approach to yield new insights into the biology of interferon (IFN)- ! and IFN- " [3] or inducible nitric oxide specific cellular subsets or tissues through their specific gene synthase [4] and it has been demonstrated that DCs can also expression programs [17-21]. Likewise, genome-wide com- activate other innate effector cells such as natural killer (NK) parative gene expression profiling between mouse and man cells [5]. In light of these properties, it has been clearly estab- has recently been demonstrated as a powerful approach to lished that DCs are critical for defense against infections, as uncover conserved molecular pathways involved in the devel- they are specially suited for the early detection of pathogens, opment of various cancers [22-27]. However, to the best of the rapid development of effector functions, and the trigger- our knowledge, this approach has not yet been applied to ing of downstream responses in other innate and adaptive study normal leukocyte subsets. Moreover, DC subsets have immune cells. not yet been scrutinized through the prism of gene expression patterns within the context of other leukocyte populations. In DCs can be divided into several subsets that differ in their tis- this report, we assembled compendia comprising various DC sue distribution, their phenotype, their functions and their and other leukocyte subtypes, both from mouse and man. ontogeny [6]. Lymph node-resident DCs (LN-DCs) encom- Using intra- and inter-species comparisons, we define the pass conventional DCs (cDCs) and plasmacytoid DCs (pDCs) common and specific core genetic programs of DC subsets. in both humans and mice. LN-cDCs can be subdivided into two populations in both mouse (CD8 ! and CD11b cDCs) [6] and in human (BDCA1 and BDCA3 cDCs) [7]. In mouse, Results CD8 ! cDCs express many scavenger receptors and may be Generation/assembly and validation of the datasets for especially efficient for cross-presenting antigen to CD8 T cells the gene expression profiling of LN-DC subsets [8] whereas CD11b cDCs have been suggested [9,10], and We used pan-genomic Affymetrix Mouse Genome 430 2.0 recently shown [11], to be specialized in the activation of CD4 arrays to generate gene expression profiles of murine splenic T cells. As human cDC functions are generally studied with CD8 ! (n = 2) and CD11b (n = 2) cDCs, pDCs (n = 2), B cells (n cells derived in vitro from monocytes or from CD34 + hemat- = 3), NK cells (n = 2), and CD8 T cells (n = 2). To generate a opoietic progenitors, which may differ considerably from the compendium of 18 mouse leukocyte profiles, these data were naturally occurring DCs present in vivo , much less is known complemented with published data retrieved from public of the eventual functional specialization of human cDC sub- databases, for conventional CD4 T cells (n = 2) [28] and sets. Due to differences in the markers used for identifying DC splenic macrophages (n = 3) [29]. We used Affymetrix subsets between human and mouse and to differences in the Human Genome U133 Plus 2.0 arrays to generate gene expression of pattern recognition receptors between DC sub- expression profiles of blood monocytes, neutrophils, B cells, sets, it has been extremely difficult to address whether there NK cells, and CD4 or CD8 T cells [30]. These data were com- are functional equivalences between mouse and human cDC plemented with published data on human blood DC subsets subsets [6]. (pDCs, BDCA1 cDCs, BDCA3 cDCs, and lin -CD16 +HLA-DR + cells) retrieved from public databases [31]. All of the human pDCs, a cell type discovered recently in both human and samples were done in independent triplicates. Information mouse, appear broadly different from the other DC subsets to regarding the original sources and the public accessibility of the point that their place within the DC family is debated [3]. the datasets analyzed in the paper are given in Table 1. Some common characteristics between human and mouse pDCs that distinguish them from cDCs [3] include: their abil- To verify the quality of the datasets mentioned above, we ana- ity to produce very large amounts of IFN- !/" upon activation, lyzed signal intensities for control genes whose expression their limited ability to prime naïve CD4 and CD8 T cells under profiles are well documented across the cell populations steady state conditions, and their expression of several genes under consideration. Expression of signature markers were generally associated with the lymphocyte lineage and not confirmed to be detected only in each corresponding popula- found in cDCs [12]. Several differences have also been tion (see Table 2 for mouse data and Table 3 for human data). reported between human and mouse pDCs, which include the For example, Cd3 genes were detected primarily in T cells and unique ability of mouse pDCs to produce high levels of IL-12 often to a lower extent in NK cells; the mouse Klrb1c (nk1.1 ) upon triggering of various toll-like receptors (TLRs) or stim- gene or the human KIR genes in NK cells; Cd19 in B cells; the ulation with viruses [13,14]. Adding to the complexity of accu- mouse Siglech and Bst2 genes or the human LILRA4 (ILT7 )

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.3

Table 1

Information on the sources and public access for the datasets analyzed in the paper

Figures ‡

Dataset Population* Laboratory † Public repository Accession number 1a,c; 2a 1b,d; 2b 1e 3 4a 4b 5a 5b

Affymetrix Mouse Genome 430 2.0 data Spleen CD8 DCs (2) MD/SCPK GEO [95] GSE9810 X X X X X Spleen CD11b DCs (2) MD/SCPK GEO GSE9810 X X X X X Spleen pDCs (2) MD/SCPK GEO GSE9810 X X X X X Spleen NK cells (2) MD/SCPK GEO GSE9810 X X X Spleen CD8 T cells (2) MD/SCPK GEO GSE9810 X X Spleen B cells (3) MD/SCPK GEO GSE9810 X X X Spleen CD4 T cells (2) AYR GEO GSM44979; GSM44982 X X X Spleen monocytes (3) SB NCI caArray [96] NA X X X Spleen monocytes (2) BP GEO GSM224733; X GSM224735 Peritoneal M # (1) SA GEO GSM218300 X BM-M # (2) RM GEO GSM177078; X GSM177081 BM-M # (1) CK GEO GSM232005 X BM-DCs (2) RM GEO GSM40053; GSM40056 X BM-DCs (2) MH GEO GSM101418; X GSM101419 Affymetrix Mouse U74Av2 data Spleen CD4 T cells (3) CB/DM GEO GSM66901; X GSM66902; GSM66903 Spleen B2 cells (2) CB/DM GEO GSM66913; GSM66914 X Spleen B1 cells (2) CB/DM GEO GSM66915; GSM66916 X Spleen NK cells (2) FT EBI ArrayExpress E-MEXP-354 X [97] Spleen CD4 DCs (2) CRES GEO GSM4697; GSM4707 X Spleen CD8 DCs (2) CRES GEO GSM4708; GSM4709 X Spleen DN DCs (2) CRES GEO GSM4710; GSM4711 X Spleen IKDCs (2) FH GEO GSM85329; GSM85330 X Spleen cDCs (2) FH GEO GSM85331; GSM85332 X Spleen pDCs (2) FH GEO GSM85333; GSM85334 X Affymetrix Human Genome U133 Plus 2.0 data Blood monocytes (3) FRS Authors' webpage NA X X X X [86] Blood CD4 T cells (3) FRS Authors' webpage NA X X X Blood CD8 T cells (3) FRS Authors' webpage NA X X X Blood B cells (3) FRS Authors' webpage NA X X X Blood NK cells (3) FRS Authors' webpage NA X X X

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.4

Table 1 (Continued)

Information on the sources and public access for the datasets analyzed in the paper

Blood neutrophils (3) FRS Authors' NA X X X webpage Blood pDCs (3) CAKB EBI ArrayExpress E-TABM-34 X X X X X Blood BDCA1 DCs (3) CAKB EBI ArrayExpress E-TABM-34 X X X X X Blood BDCA3 DCs (3) CAKB EBI ArrayExpress E-TABM-34 X X X X X Blood CD16 DCs (3) CAKB EBI ArrayExpress E-TABM-34 X X PBMC-derived M # (2) SYH GEO GSM109788; X GSM109789 Monocyte-derived M # LZH GEO GSM213500 X Monocyte-derived MVD GEO GSM181931; X DCs (3) GSM181933; GSM181971

*The number of replicates is shown in parentheses. †MD/SCPK, M Dalod, S Chan, P Kastner; AYR, AY Rudensky; SB, S Bondada; BP, B Pulendran; SA, S Akira; RM, R Medzhitov; CK, C Kim; MH, M Hikida; CB/DM, C Benoist, D Mathis; FT, F Takei; CRES, C Reis e Sousa; FH, F Housseau; FRS, FR Sharp; CAKB, CAK Borrebaeck; SYH, S Yla-Herttuala; LZH, L Ziegler-Heitbrock; MVD, MV Dhodapkar. ‡Shown in the indicated figure in this study. BM-DC, mouse bone-marrow derived GM-CSF DCs; BM-M #, mouse bone marrow-derived M-CSF macrophages; monocyte-derived M #, monocyte-derived M-CSF macrophages; NA, not applicable; PBMC-derived M #, human peripheral blood mononuclear cell-derived M-CSF macrophages; peritoneal M #, peritoneal mouse macrophages. and IL3RA (CD123 ) genes in pDCs; and Cd14 in myeloid cells. cific expression levels in all DC subtypes (Figure 2, 'pan DC' As expected, many markers were expressed in more than a clusters). These analyses, which are based on very different single cell population. For example, in the mouse, Itgax mathematical methods, thus highlight the unity of the LN-DC (Cd11c ) was found expressed to high levels in NK cells and all family. To investigate the existence of a core genetic program DC subsets; Itgam (Cd11b ) in myeloid cells, NK cells, and common to the LN-DC subsets and conserved in mammals, CD11b cDCs; Ly6c at the highest level in pDCs but also clustering of mouse and human data together was next per- strongly in many other leukocyte populations; and Cd8a in formed. We identified 2,227 orthologous genes that showed pDCs and CD8 ! cDCs. However, the analysis of combinations significant variation of expression in both the mouse and of these markers confirmed the lack of detectable cross-con- human datasets. After normalization (as described in Materi- taminations between DC subsets: only pDCs expressed high als and methods), the two datasets were pooled and a com- levels of Klra17 (Ly49q ) and Ly6c together, while Cd8a , ly75 plete linkage clustering was performed. As shown in Figure (Dec205 , Cd205 ), and Tlr3 were expressed together at high 1e, the three major cell clusters, lymphocytes, LN-DCs, and levels only in CD8 ! cDCs, and Itgam (Cd11b ) with Tlr1 and myeloid cells, were obtained as observed above when cluster- high levels of Itgax (Cd11c ) only in CD11b cDCs. Thus, each ing the mouse or human data alone. Thus, this analysis shows cell sample studied harbors the expected pattern of expres- that DC subsets constitute a specific cell family distinct from sion of control genes and our data will truly reflect the gene the classic lymphoid and myeloid cells and that pDCs belong expression profile of each population analyzed, without any to this family in both mice and humans. All the LN-DC sub- detectable cross-contamination. sets studied therefore share a common and conserved genetic signature, which must determine their ontogenic and func- LN-DCs constitute a specific leukocyte family that tional specificities as compared to other leukocytes, including includes pDCs in both the human and the mouse other antigen-presenting cells. To determine whether LN-DCs may constitute a specific leu- kocyte family, we first evaluated the overall proximity Identification and functional annotation of the between LN-DC subsets as compared to lymphoid or myeloid conserved transcriptional signatures of mouse and cell types, based on the analysis of their global gene expres- human leukocyte subsets sion program. For this, we used hierarchical clustering with Genes that are selectively expressed in a given subset of leu- complete linkage [32], principal component analysis (PCA) kocytes in a conserved manner between mouse and human [33], as well as fuzzy c-means (FCM) partitional clustering were identified and are presented in Table 4. Our data analy- approaches [34]. Hierarchical clustering clearly showed that sis is validated by the recovery of all the genes already known the three LN-DC subsets studied clustered together, both in to contribute to the characteristic pathways of development mouse (7,298 genes analyzed; Figure 1a) and human (11,507 or to the specific functions for the leukocyte subsets studied, genes analyzed; Figure 1b), apart from lymphocytes and mye- as indicated in bold in Table 4. These include, for example, loid cells. The close relationship between all the DC subsets in Cd19 and Pax5 for B cells [35], Cd3e-g and Lat for T cells [36], each species was also revealed by PCA for mouse (Figure 1c) as well as Ncr1 [37] and Tbx21 (T-bet ) [38] for NK cells. Sim- and human (Figure 1d). Finally, FCM clustering also allowed ilarly, all the main molecules involved in major histocompat- clear visualization of a large group of genes with high and spe- ibility (MHC) class II antigen processing and presentation are

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.5

Table 2

Expression of control genes in mouse cells

Dendritic cells Lymphocytes

Probe set ID Gene Myeloid cells pDC CD8 ! DC CD11b DC NK CD8 T CD4 T B

1419178_at Cd3g 40 ± 10 <20 <20 <20 97 ± 31 2,074 ± 287 1,974 ± 478 22 ± 3 1422828_at Cd3d 111 ± 14 <20 <20 <20 214 ± 16 2,815 ± 11 4,520 ± 1,414 21 ± 2 1422105_at Cd3e 115 ± 30 27 ± 10 22 ± 2 23 ± 5 26 ± 9 387 ± 58 522 ± 210 26 ± 10 1426396_at Cd3z <20 <20 <20 <20 1,147 ± 81 1,545 ± 10 2,117 ± 482 25 ± 9 1426113_x_at Tcra 83 ± 8 <20 23 ± 4 <20 116 ± 39 2,517 ± 42 5,601 ± 1,818 34 ± 13 1419696_at Cd4 24 ± 2 1,233 ± 144 <20 369 ± 49 <20 <20 1,052 ± 73 <20 1450570_a_at Cd19 190 ± 44 <20 <20 <20 <20 <20 23 ± 5 2,259 ± 292 1449570_at Klrb1c (NK1.1) <20 <20 <20 <20 2,328 ± 112 <20 25 ± 7 <20 1425436_x_at Klra3 (Ly49C) 130 ± 11 24 ± 3 156 ± 0 242 ± 31 9,186 ± 479 170 ± 61 70 ± 42 <20 1450648_s_at H2-Ab1 6,887 ± 84 7,339 ± 5 9,101 ± 100 9,056 ± 277 81 ± 6 83 ± 56 978 ± 11 7,028 ± 239 1419128_at Itgax (CD11c) 454 ± 5 1,928 ± 169 2,827 ± 454 4,701 ± 56 3,403 ± 45 108 ± 44 22 ± 2 <20 1457786_at Siglech 31 ± 4 3,454 ± 536 24 ± 5 <20 <20 <20 33 ± 13 <20 1425888_at Klra17 (Ly49Q) 98 ± 4 3,413 ± 116 30 ± 14 163 ± 2 28 ± 11 24 ± 6 38 ± 10 <20 1424921_at Bst2 (120G8) 2,364 ± 149 5,571 ± 718 237 ± 30 196 ± 44 61 ± 24 162 ± 12 90 ± 3 88 ± 32 1421571_a_at Ly6c 4,420 ± 261 8,255 ± 151 98 ± 5 30 ± 8 2,082 ± 365 4,530 ± 229 1,789 ± 1,242 302 ± 303 1422010_at Tlr7 439 ± 13 846 ± 40 <20 322 ± 45 <20 <20 22 ± 2 118 ± 83 1440811_x_at Cd8a <20 337 ± 134 825 ± 44 <20 <20 1,235 ± 227 22 ± 2 <20 1449328_at Ly75 (Dec205) 249 ± 27 <20 159 ± 4 22 ± 3 24 ± 6 170 ± 29 79 ± 1 21 ± 1 1422782_s_at Tlr3 27 ± 2 25 ± 3 3,376 ± 159 287 ± 14 <20 <20 <20 52 ± 45 1422046_at Itgam (CD11b) 956 ± 57 <20 <20 162 ± 1 188 ± 38 <20 <20 21 ± 1 1449049_at Tlr1 1,218 ± 54 31 ± 15 101 ± 4 1,601 ± 92 <20 889 ± 109 498 ± 103 1,141 ± 484 1417268_at Cd14 7,649 ± 169 187 ± 52 107 ± 0 115 ± 34 <20 <20 31 ± 8 27 ± 12 1449498_at Marco 174 ± 19 <20 <20 <20 <20 <20 <20 <20 1460282_at Trem1 415 ± 19 <20 <20 <20 <20 <20 <20 <20 found selectively expressed in antigen-presenting cells scripts present in all the other leukocytes studied here, (APCs). Indeed, a relatively high proportion of the genes including members of the gimap family, especially gimap4 , selectively expressed in lymphocytes or in APCs has been which have been very recently shown to be expressed to high known for a long time to be involved in the biology of these levels in T cells and to regulate their development and sur- cells. However, we also found genes identified only recently as vival [49-51]. important in these cells, such as March1 [39] or Unc93b1 [40,41] for APCs, and Edg8 for NK cells [42]. Interestingly, Thus, the identity of the gene signatures specific for the vari- we also identified genes that were not yet known to be ous leukocyte subsets studied highlights the sharp contrast involved in the biology of these cells, to the best of our between our advanced understanding of the molecular bases knowledge, such as the E430004N04Rik expressed sequence that govern the biology of lymphocytes or the function of tag in T cells, the Klhl14 gene in B cells, or the Osbpl5 gene in antigen presentation and our overall ignorance of the genetic NK cells. programs that specifically regulate DC biology. This contrast is enforced upon annotation of each of the gene signatures In contrast to the high proportion of documented genes selec- found with Gene Ontology terms for biological processes, tively expressed in the cell types mentioned above, most of the molecular functions, or cellular components, and with path- genes specifically expressed in LN-DCs have not been previ- ways, or with interprotein domain names, using DAVID bio- ously associated with these cells and many have unknown informatics tools [52,53] (Table 5). Indeed, many significant functions. Noticeable exceptions are Flt3 , which has been annotations pertaining directly to the specific function of recently shown to drive the differentiation of all mouse [43- myeloid cells, lymphocyte subsets or APCs are recovered, as 45] and human [46] LN-DC subsets [47], and Ciita (C2ta ), indicated in bold in Table 5. In contrast, only very few signif- which is known to specifically regulate the transcription of icant annotations are found for LN-DCs, most of which may MHC class II molecules in cDCs [48]. Interestingly, mouse or not appear to yield informative knowledge regarding the spe- human LN-DCs were found to lack expression of several tran- cific functions of these cells.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.6

Table 3

Expression of control genes in human cells

Lymphocytes Dendritic cells Myeloid cells

Probe setID Genes NK CD8 T CD4 T B pDC BDCA1 BDCA3 Mono Neu

206804_at CD3G 858 ± 71 1,760 ±241 1,975 ±132 53 ± 6 <50 <50 <50 <50 52 ± 4 213539_at CD3D 5,413 ±238 7,134 ±635 6,291 ±285 276 ± 24 <50 <50 51 ± 2 112 ± 9 276 ± 4 205456_at CD3E 247 ± 21 569 ± 67 679 ± 91 <50 <50 <50 <50 <50 <50 210031_at CD3Z 8,688 ±181 5,223 ±218 4,749 ±123 2,996 ±217 56 ± 10 60 ± 17 54 ± 7 914 ± 96 132 ± 15 209671_x_at TCR@ 147 ± 16 3,127 ±260 3,462 ±170 71 ± 7 <50 <50 <50 <50 111 ± 16 205758_at CD8A 911 ± 26 5,259 ±217 67 ± 10 79 ± 16 <50 <50 <50 <50 99 ± 7 207979_s_at CD8B 77 ± 9 3,596 ±299 <50 <50 <50 <50 <50 <50 53 ± 5 203547_at CD4 <50 <50 391 ± 20 83 ± 20 1,301 ±119 1,004 ± 74 278 ± 61 205 ± 34 <50 206398_s_at CD19 <50 51 ± 1 <50 1,726 ±115 <50 <50 <50 57 ± 12 <50 212843_at NCAM1 ( CD56) 2,074 ± 96 144 ± 14 65 ± 2 135 ± 9 <50 <50 82 ± 17 52 ± 3 <50 207314_x_at KIR3DL2 3,131 ±172 454 ± 14 227 ± 18 265 ± 16 <50 <50 <50 59 ± 8 <50 208203_x_at KIR2DS5 3,472 ±140 444 ± 7 236 ± 10 284 ± 14 <50 <50 <50 <50 <50 239975_at HLA-DPB2 <50 <50 <50 63 ± 22 777 ± 701 1,565 ±519 2,056 ±577 <50 <50 210184_at ITGAX (CD11c) 1,017 ± 50 112 ± 37 166 ± 17 752 ± 45 74 ± 21 2,151 ±43 0 729 ± 98 1,284 ±115 2,133 ±196 210313_at LILRA4 (ILT7) 226 ± 10 117 ± 13 346 ± 42 1,109 ± 76 7,916 ±612 230 ± 16 1,659 ±1,183 524 ± 41 <50 206148_at IL3RA (CD123) 84 ± 3 59 ± 8 91 ± 2 324 ± 9 4,728 ±365 61 ± 10 116 ± 110 120 ± 3 74 ± 12 1552552_s_at CLEC4C (BDCA2) 93 ± 6 61 ± 5 99 ± 4 408 ± 9 6,789 ±737 76 ± 39 859 ± 434 217 ± 8 175 ± 25 205987_at CD1C (BDCA1) 76 ± 8 61 ± 12 159 ± 8 1,715 ± 85 64 ± 23 8,313 ±272 722 ± 845 560 ± 59 <50 204007_at FCGR3B (CD16) 459 ± 54 115 ± 24 65 ± 5 322 ± 46 63 ± 23 <50 51 ± 1 160 ± 11 5,554 ± 57 201743_at CD14 94 ± 3 139 ± 5 343 ± 5 1,274 ±113 <50 202 ± 183 <50 7,638 ±446 4,621 ±374 205786_s_at ITGAM (CD11b) 5,688 ±116 1,980 ±147 1,161 ± 71 2,513 ±117 360 ± 184 703 ± 28 86 ± 63 5,541 ±193 5,232 ±576 208982_at PECAM1 (CD31) 2,232 ± 48 2,144 ± 91 1,487 ± 58 4,644 ±102 3,834 ±601 2,825 ±290 2,680 ±363 5,479 ±219 7,699 ±853 205898_at CX3CR1 10,056 ±53 6,633 ±232 4,351 ±170 6,055 ±263 262 ± 45 1,296 ± 84 362 ± 419 5,717 ±451 616 ± 21 39402_at IL1B 69 ± 6 72 ± 7 52 ± 3 209 ± 27 <50 195 ± 131 69 ± 27 198 ± 9 2,920 ±183 202859_x_at IL8 95 ± 7 77 ± 6 72 ± 5 385 ± 26 218 ± 185 90 ± 9 680 ± 561 310 ± 17 8,685 ±776 207094_at IL8RA 199 ± 30 74 ± 8 81 ± 12 82 ± 2 <50 61 ± 9 67 ± 1 90 ± 1 4,784 ±521

Mono, monocyte; neu, neutrophil. Thus, when taken together, our data show that LN-DC sub- number of genes comparable to those of the transcriptional sets constitute a specific family of leukocytes, sharing selec- signature of NK or T cells. To the best of our knowledge, most tive expression of several genes, most of which are still of of these genes had not been reported to be selectively unknown function. We believe that the identification of these expressed in pDCs, with the exception of Tlr7 [31,54] and genes selectively expressed in LN-DC subsets in a conserved Plac8 (C15 ) [55]. Second, although mouse and human cDCs manner between mouse and human will be very helpful for clustered together, the two cDC subsets of each species future investigation of the mechanisms regulating LN-DC appeared closer to one another than to the subsets of the biology by the generation and study of novel genetically other species. Thus, no clear homology could be drawn manipulated animal models. between human and mouse cDC subsets in this analysis. However, it should be noted that known homologous human Search for a genetic equivalence between mouse and and mouse lymphoid cell types also failed to cluster together human LN-DC subsets in this analysis and were closer to the other cell populations To search for equivalence between mouse and human LN-DC from the same species within the same leukocyte family. This subsets, we examined their genetic relationships in the hier- is clearly illustrated for the T cell populations as mouse CD4 archical clustering depicted in Figure 1e. Two observations and CD8 T cells cluster together and not with their human can be made. First and remarkably, mouse and human pDCs CD4 or CD8 T cell counterparts (Figure 1e). Therefore, to fur- clustered together. This result indicates a high conservation ther address the issue of the relationships between human in their genetic program and establishes these two cell types and mouse cDC subsets, we used a second approach. We per- as homologs. Indeed, human and mouse pDCs share a large formed hierarchical clustering with complete linkage on the and specific transcriptional signature (Table 4), with a mouse and human LN-DC datasets alone (1,295 orthologous

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.7

(a) (c) 0.3 T cells 0.2 Myeloid cells NK cells 0.1

0 B cells

m. B -0.1 m. NK m. pDC m. CD8 -0.2 pDCs m. CD8 T m. CD4 T m. CD11b m. CD11b

Principal component 3 -0.3 cDCs Myeloid DCs Lymphocytes cells -0.4 -0.6-0.4 -0.2 00.2 0.4 Principal component 2 (b) (d) 0.8 0.6 Neutrophils 0.4 Monocytes 0.2 NK cells

h. B B cells 0 h. NK CD4 T h. neu. h. pDC CD8 T BDCA1 cDCs

h. mono. cells h. CD8 T h. CD4 T -0.2 cells h. BDCA1 h. BDCA3 pDCs Lymphocytes DCs Myeloid Principal component 3 -0.4 BDCA3 cDCs cells -0.4 -0.2 0 0.2 0.4 0.6 Principal component 2 (e) h. B m. B h. NK m. NK h. neu. h. pDC m. pDC m. CD8 h. mono. h. CD8 T h. CD4 T m. CD8 T m. CD4 T h. BDCA1 h. BDCA3 m. CD11b m. CD11b

Lymphocytes DCs Myeloid cells

ClusteringFigure 1 of mouse and human leukocyte subsets Clustering of mouse and human leukocyte subsets. Hierarchical clustering with complete linkage was performed on the indicated cell populations isolated from: (a) mouse, (b) human, and (e) mouse and human. PCA was performed on the indicated cell populations isolated from: (c) mouse and (d) human. Mono, monocytes; neu, neutrophils.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.8

(a) sCD T sllec (b) sCD T sllec s il s d s h te s i ll p y 1 3 ll lo b s ls e T T o c s ls e T T 1 l c r o A A l c e ls 8 1 C e 8 4 t n C C C e 8 4 y l D c K D u D c K D M e D D D e D D D c C C p B N C C N Mo B B p B N C C

Neutrophils Myeloid cells

Monocytes

BDCA1 DCs BDCA3 DCs pan DCs cDCs cDCs pan DCs CD8 DCs CD11b DCs

pDCs pDCs

B cells

NK cells

B cells pan T CD8 T NK cells CD4 T

pan T -4-2 0 2 4 -4-2 0 2 4

FCMFigure partitional 2 clustering FCM partitional clustering. FCM partitional clustering was performed on the mouse and human gene chip datasets. (a) FCM partitional clustering for mouse data. (b) FCM partitional clustering for human data. The color scale for relative expression values as obtained after log 10 transformation and median centering of the values across cell samples for each gene is given below the heat map.

LN-DC genes), without taking into account the pattern of cDCs versus mouse CD11b and human BDCA1 cDCs have expression of each gene in the other leukocyte subsets as it unknown functions and have not been previously described to may have hidden some degree of similarity between subsets exhibit a conserved pattern of expression between these clustering in the same branch. The results of the analysis of mouse and human cell types. Notable exceptions are Tlr3 gene expression focused on DCs confirmed that mouse and [31,56] and the adhesion molecule Nectin-like protein 2 human pDCs cluster together and apart from cDCs (Figure 3). (Cadm1 , also called Igsf4 ) [57], which have been previously Importantly, when analyzing the DC datasets alone, mouse described to be conserved between mouse CD8 ! and human CD8 ! and human BDCA3 cDCs on the one hand, and mouse BDCA3 cDCs. When comparing cDC to pDCs, a few genes CD11b and human BDCA1 cDCs on the other hand, clustered already known to reflect certain functional specificities of together and shared a conserved genetic signature (Figure 3 these cells when compared to one another are identified. Tlr7 and Table 6). Thus, although a higher genetic distance is and Irf7 are found preferentially expressed in pDCs over observed between mouse and human conventional DC cDCs, consistent with previous reports that have documented subsets as opposed to pDCs, a partial functional equivalence their implication in the exquisite ability of these cells to pro- is suggested between these cell types. The majority of the duce high levels of IFN- !/" in response to viruses [58-60]. genes conserved between mouse CD8 ! and human BDCA3 Ciita , H2-Ob , Cd83 and Cd86 are found preferentially

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.9

Table 4

Specific transcriptomic signatures identified in the leukocyte populations studied

Expression ratio (log 2) of specific genes* Cell type 3-4 2-3 1-2 0,4-1

Myeloid cells - Steap4; Clec4d; Clec4e; Fpr1 Nfe2; Mpp1; Snca; Ccr1; Slc40a1; Sepx1; Ninj1; Hp; Sdcbp; Bst1; Ifit1; S100a9; Cd14 ; Tlr4 ; F5; Fcgr3; Fpr-rs2; S100a8; Adipor1; Bach1; Marcks; Tlr2 ; Abhd5; Gca; Atp6v1b2; Ier3; Sod2; Pira2; Wdfy3; Ifrd1; Fcho2; Csf3r ; Pilra; Slc11a1 C5ar1; Cd93; Snap23; Cebpb; Clec7a; Yipf4; Hmgcr; Slc31a2; Fbxl5 Pan-DC Flt3 Sh3tc1 Trit1; Bri3bp; Prkra; Etv6; Tmed3; Bahcc1; Scarb1 cDC - - Arhgap22; Btbd4; Slamf8; C2ta ; Avpi1; Spint1; Cs 9130211I03Rik; Nav1 pDC Epha2 ; Pacsin1; Zfp521; Sh3bgr Tex2; Runx2; Atp13a2; Maged1; Nucb2; Alg2; Pcyox1; LOC637870; Tm7sf2; Tcf4; Gpm6b; Cybasc3 Scarb2; Dnajc7; Trp53i13; Plac8 ; Pls3; Tlr7 ; Ptprs; Bcl11a B cells Ebf1 ; Cd19 ; Klhl14 Bank1 ; Pax5 Blr1 ; Ralgps2; Cd79b ; Pou2af1 ; Ms4a1; Blk ; Cd72 ; Syvn1; Fcer2a; Cr2; Cd79a ; Fcrla BC065085; Fcrl1; Phtf2; Tmed8; Grap; Pip5k3; Pou2f2 NK cells - Ncr1 Tbx21 ; Osbpl5 Rgs3; 1700025G04Rik; Plekhf1; Fasl ; Zfpm1; Edg8 ; Cd160 ; Klrd1 ; Il2rb ; Il18rap ; Ctsw; Ifng ; Prf1 ; Sh2d2a; Llgl2; Gpr178; Prkx; Gab3; Nkg7 ; Cst7; Sntb2; Runx3; Myo6; F2r; Vps37b; Dnajc1; Gfi1 Pan-T cells - Camk4 ; E430004N04Rik; Trat1 Cxcr6 ; Tnfrsf25; Ccdc64; Plcg1 Cd3e ; Cd5 ; Lrig1; Cd3g ; Ubash3a; Cd6 ; Lat ; Bcl11b; Tcf7 ; Ic os CD8 T cells - - - Gzmk CD4 T cells - Ctla4 - Icos ; Tnfrsf25; Cd5 ; Cd28 ; Trat1 Lymphocytes - - Ablim1; Lax1; D230007K08Rik; Spnb2; Cdc25b; Ets1 ; Sh2d2a; Rasgrp1; Bcl2 Ppp3cc; Cnot6l Myeloid, B, DC - H2-DMb2 ; H2-DMb1 C2ta ; March1 ; Aldh2; Bcl11a; Btk Ctsh ; H2-Eb1 ; Cd74 ; Ctsz ; Clic4; Kynu ; 5031439G07Rik; Nfkbie; Unc93b1 Non-DC Gimap4 - Vps37b Lck ; Pde3b

*Ratio expressed as Minimum expression among the cell types selected/Maximum expression among all other cell types. Genes already known to be preferentially expressed in the cell types selected are shown in boldface. expressed in cDCs over pDCs, which is consistent with their most significant clusters of functional annotations in cDCs higher efficiency for MHC class II antigen presentation and T concerns the response to pest, pathogens or parasites and the cell priming [61]. activation of lymphocytes, which include genes encoding TLR2, costimulatory molecules (CD83, CD86), proinflamma- The functional annotations associated with the genes selec- tory cytokines (IL15, IL18), and chemokines (CXCL9, tively expressed in specific DC subsets when compared to the CXCL16), consistent with the specialization of cDCs in T cell others are listed in Table 7. The most significant clusters of priming and recruitment. Clusters of genes involved in functional annotations in pDCs point to the specific expres- inflammatory responses are found in both pDCs and cDCs. sion in these cells of many genes expressed at the cell surface However, their precise analysis highlights the differences in or in intracellular compartments, including the endoplasmic the class of pathogens recognized, and in the nature of the reticulum, the Golgi stack, and the lysosome. A cluster of cytokines produced, by these two cell types: IFN- !/" produc- genes involved in endocytosis/vesicle-mediated transport is tion in response to viruses by pDCs through mechanisms also observed. This suggests that pDCs have developed an involving IRF7 and eventually TLR7; and recognition and exquisitely complex set of molecules to sense, and interact killing of bacteria and production of IL15 or IL18 by cDCs with, their environment and to regulate the intracellular through mechanisms eventually involving TLR2 or lys- trafficking of endocytosed molecules, which may be ozymes. Many genes selectively expressed in cDCs are consistent with the recent reports describing different intrac- involved in cell organization and biogenesis, cell motility, or ellular localization and retention time of endocytosed CpG cytoskeleton/actin binding, consistent with the particular oligonucleotides in pDCs compared to cDCs [62,63]. The morphology of DCs linked to the development of a high mem-

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.10

Table 5

Selected annotations for the conserved transcriptomic signatures identified for the cell types studied

Cell type* Annotation Genes

Myeloid cells Defense response/response to pest, pathogen or C5ar1, Sod2, Fcgr3, Tlr2, Ccr1, Ifrd1, Csf3r, Clec7a, Bst1, Ifit1, parasite/inflammatory response Clec4e, Tlr4, Clec4d, Cd14, Cebpb, Hp Response to bacteria or fungi/pattern recognition SLC11A1, TLR2, TLR4, CLEC7A, Clec4e, Clec4d receptor activity/C-type lectin H_tollpathway: Toll-like receptor pathway CD14, TLR2, TLR4 Regulation of cytokine biosynthesis/positive regulation Fcgr3, Tlr2, Tlr4, Cebpb, Clec7a of TNF- ! or IL-6 biosynthesis Macrophage activation/mast cell activation/neutrophil CD93, TLR4, Fcgr3, Csf3r chemotaxis Pan-DC Binding ETV6, PRKRA, FLT3, SCARB1, TRIT1, BAHCC1, SH3TC1 cDC Nucleobase, nucleoside, nucleotide and nucleic acid metabolism NAV1, BTBD4, CIITA, SNFT Molecular function unknown Btbd4, Avpi1, Arhgap22 pDC Transcription cofactor activity Maged1, Bcl11a, Tcf4 Integral to membrane TLR7, EPHA2, TMEPAI, SCARB2, ATP13A2, ALG2, CYBASC3, TM7SF2, GPM6B, PTPRS Cellular component unknown Maged1, Sh3bgr, Cybasc3, Alg2, Plac8 B cells MMU04662: B cell receptor signaling pathway/B cell Cr2, Cd79a, Cd79b, Cd72, Cd19, Blr1, Ms4a1 activation MMU04640: hematopoietic cell lineage Cr2, Fcer2a, Ms4a1, Cd19 Defense response/response to pest, pathogen or PAX5, POU2F2, CR2, MS4A1, CD72, CD19, POU2AF1, BLR1, parasite/humoral immune response CD79A, CD79B, FCER2 NK cells MMU04650: natural killer cell mediated cytotoxicity/ Klrd1, Ifng, Ncr1, Fasl, Prf1, Prf1, Plekhf1 apotosis Defense response IL18RAP, CTSW, IFNG, FASLG, CD160, NCR1, PRF1, KLRD1, CST7 Pan-T cells HSA04660: T cell receptor signaling pathway/ CD3E, ICOS, PLCG1, LAT, CD3G, Trat1 immunological synapse Defense response/immune response Cd5, Icos, Cd3e, Ubash3a, Lat, Trat1, Cd3g HSA04640: hematopoietic cell lineage CD3E, CD3G, CD5 CD8 T cells No annotations - CD4 T cells Defense response/immune response Cd28, Icos, Cd5, Ctla4, Trat1 M_ctla4pathway: the co-stimulatory signal during T-cell Cd28, Icos, Ctla4 activation Lymphocytes Immune response BCL2, LAX1, ETS1 Myeloid, B, DC Antigen presentation, exogenous antigen via MHC class H2-Eb1, H2-DMb2, H2-DMb1, Cd74 II HSA04612: antigen processing and presentation HLA-DRB1, CIITA, CD74, HLA-DMB Defense response/immune response H2-Eb1, H2-DMb2, H2-DMb1, Bcl11a, Cd74 Non-DC Phosphoric ester hydrolase activity LCK, PDE3B

*The annotations recovered are written in boldface when they correspond to known specificities of the cell subset studied and are thus confirmatory of the type of analysis performed. brane surface for sampling of their antigenic environment tistically significant association with asthma also highlights and for the establishment of interactions with lymphocytes. the proinflammatory potential of this cell type. Recently, it pDCs and cDCs also appear to express different arrays of has been reported that the mouse CD11b cDC subset is spe- genes involved in signal transduction/cell communication, cialized in MHC class II mediated antigen presentation in transcription regulation and apotosis. A statistically signifi- vivo [11]. In support of our findings here that mouse CD11b cant association with lupus erythematosus highlights the pro- cDCs are equivalent to human BDCA1 cDCs, we found that posed harmful role of pDCs in this autoimmune disease [64]. many of the genes involved in the MHC class II antigen pres- entation pathway that were reported to be expressed to higher The mCD11b/hBDCA1 cDC cluster of genes comprises many levels in mouse CD11b cDCs over CD8 ! cDCs [11] are also genes involved in inflammatory responses and the positive preferentially expressed in the human BDCA1 cDC subset regulation of the I-kappaB kinase/NF-kappaB cascade. A sta- over the BDCA3 one. These genes include five members of the

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.11

CD11b cDC subset, human BDCA1 cDCs serve as a subset of DCs that are specialized in presenting antigen via MHC class II molecules. It is also noteworthy that mCD11b and hBDCA1 cDCs express high constitutive levels of genes that are known to be induced by IFN- !/" and that can contribute to cellular antiviral defense ( Oas2 , Oas3 , Ifitm1 , Ifitm2 , Ifitm3 ).

No significant informative functional annotations are found h. pDC m. pDC m. CD8 for the mCD8 !/hBDCA3 cDC gene cluster. However, groups h. BDCA3 h. BDCA1 m. CD11b of genes involved in cell organization and biogenesis or in small GTPase regulator activity are found and the study of these genes may increase our understanding of the specific m CD11b functions of these cells. Mouse CD8 ! cDCs have been pro- huBDCA1 posed to be specialized for a default tolerogenic function but (111) to be endowed with the unique ability to cross-present anti- gen for the activation of naïve CD8 T cells within the context of viral infection [65]. It will be important to determine whether this is also the case for hBDCA3 cDCs. From this point of view, it is noteworthy that hBDCA3 cDCs selectively express TLR3 , lack TLR7 and TLR9 , and exhibit the highest ratio of IRF8 (ICSBP )/ TYROBP (DAP12 ) expression, all of which have been shown to participate in the regulation of the balance between tolerance and cross-presentation by mouse CD8 ! cDCs [65,66].

Use of leukocyte gene expression compendia to classify cell types of ambiguous phenotype or function (53) mCD8 α Interferon-producing killer dendritic cells huBDCA3 A novel cell type has been recently reported in the mouse that (21) presents mixed phenotypic and functional characteristics of pDCs and NK cells, IKDCs [15,16]. A strong genetic relationship between IKDCs and other DC populations was suggested. However, this analysis was based solely on com- parison of the transcriptional profile of IKDCs to DCs and not pDC to other cell populations [15]. As IKDCs were also reported to (228) be endowed with antigen presentation capabilities [15] and to be present in mice deficient for the expression of RAG2 and the common $ chain of the cytokine receptors [16], they have been proposed to belong to the DC family rather than to be a subset of NK cells in a particular state of differentiation or activation. However, IKDCs have been reported to express many mRNA specific for NK cells and many of their pheno- typic characteristics that were claimed to discriminate IKDCs from NK cells [16] are in fact consistent with classical NK cell features as recently reviewed [67], including the expression of -2-1 0 1 2 B220 [68] and CD11c [69,70] (BD/Pharmingen technical datasheet of the CD11c antibody) [71]. To clarify the genetic nature of IKDCs, we reanalyzed the published gene chip data ConservedFigure 3 genetic signatures between mouse and human DC subsets on the comparison of these cells with other DC subsets [15], Conserved genetic signatures between mouse and human DC subsets. together with available datasets on other leukocyte popula- Hierarchical with complete linkage clustering was performed on the indicated DC populations isolated from mouse and human. tions. We thus assembled published data generated on the same type of microarrays (Affymetrix U74Av2 chips) to build a second mouse compendium, allowing us to compare the cathepsin family ( Ctsb , Ctsd , Ctsh , Ctss , and Ctsw ) as well as transcriptomic profile published for the IKDCs (n = 2) with Ifi30 and Lamp1 and Lamp2 (see Additional data file 2 for that of pDCs (n = 2), cDCs (n = 2) [15], CD8 !+ (n = 2), CD4 + expression values). Thus, it is possible that, like the mouse (n = 2) or double-negative (n = 2) cDC subsets [56], NK cells

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.12

Table 6

Conserved specific transcriptomic signatures of DC subsets compared to one another

Expression ratio (log 2) of specific genes* Cell type >4 3-4 2-3 1-2 0,4-1 pDC Pacsin1; Sla2; - Epha2; Sh3bgr; Ets1; Runx2; LOC637870; Ifnar2; Ugcg; Kmo; Tspan31; Xbp1; Alg2; 2210020M01Rik Cobll1; Blnk; Myb; Sit1; Hs3st1; Asph; L3mbtl3; Txndc5; Abca5; Carhsp1; Ptp4a3; Lypla3; Zfp521; Nucb2; Igj; Tex2; Nrp1; Npc1; Cxxc5; Sema4c; Vamp1; Klhl9; BC031353; Stambpl1; Ptprcap; Maged1; Tm7sf2; Igh-6 ; Cybb; Scarb2; Card11; Cdkn2d; Spib; Glcci1; Syne2; Csf2rb2; Ccr2; Cdk5r1; 4931406C07Rik; Gimap8; Plxdc1; Lman1; Ahi1; Atp13a2; Tcf4; Fcrla; Rnasel; Arid3a; 4631426J05Rik; Tcta; Mgat5; Ern1; Atp8b2; Lair1 Rassf8; Tgfbr3; Tlr7 ; Lrrc16; Cln5; Rexo2; Atp2a3; Tspyl4; Anks3; Trp53i11; Ltb4dh; Slc23a2; Gata2; Trp53i13; Slc44a2; Arhgap24; Creb3l2; Itpr2; Tmem63a; Dnajc7; Rhoh; Daam1; Lancl1; Bcl11a; Usp11; Gpm6b; Aff3; Chst12; Unc5cl; Rwdd2; Armcx3; Snx9; Hivep1; Irf7 ; Cnp1; Vps13a; Mcoln2; Tm7sf3; Stch; Glt8d1; Pscd4; Cybasc3; Pcyox1; Aacs Ormdl3; 1110028C15Rik; Snag1; Prkcbp1; Klhl6; Cbx4; Pcmtd1; Bet1; Ccs; Tceal8; Dpy19l3; Pcnx; LOC672274; Sec11l3; Ctsb; Slc38a1; Ostm1; Acad11; Zbtb20; 1110032A03Rik; Ralgps2; Dtx3; Pls3; Ptprs; Zdhhc8; Rdh11; Bcl7a; Tbc1d2b cDC - 9130211I03Rik; Chn2; Ddef1; Havcr2; Arrb1; H2-Ob ; Arhgap22; Il18; Vasp; Ppfibp2; Itfg3; Wdfy3; Atad2; Hck; Hnrpll; Fgl2; Id2; A530088I07Rik; Aytl1; 2810417H13Rik; Cnn2; BC039210; Lima1; Fhod1; Klhl5; Flna; Slamf8 Rab32; Adam8; Pik3cb; Nav1; Acp2; Egr1; Mrps27; Gas2l3; Atp2b1; Gypc; Lst1; 2610034B18Rik; Tnfaip2; Tspan33; Ralb; 8430427H17Rik; Lmnb1; Junb; Irf2; Soat1; Dusp2; Btbd4; Pak1; Marcks; Epb4.1l2; Rab31; Cd83 ; Spg21; Nab2; Rbpsuh; Tiam1; Spfh1; Bzrap1; Anpep; Aim1; Cias1; Cd86 ; Cdca7; Gemin6; Entpd1; Lzp-s; Lyzs; Slc8a1; Dusp16; Apob48r; Aif1 Rin3; Hk2; Actn1; Snx8; Plscr1; Ptcd2; Slc19a2; Mthfd1l; Copg2; Dym; Cd1d1; Cxcl9; Sestd1; Limd2; Bag3; Csrp1; Ppa1; Nr4a2; Snx10; Anxa1; Il15; Ahr; Myo1f; Hmgb3; Plekhq1; Oat; Rgs12; Numb; Hars2; Avpi1; Pde8a; Stom; Spint1; Pacs1; Gtdc1; Ezh2; Swap70; Rasgrp4; Asahl; Kit; 1100001H23Rik; Susd3; Lrrk2; Sec14l1; Asb2; Txnrd2; Specc1; Bcl6; Tpi1; Kcnk6; E330036I19Rik; Sla; Fscn1; Nr4a1; Inpp1; Efhd2; Cxcl16; Ddb2; Tdrd7; 4933406E20Rik; Usp6nl C2ta ; Tgif; Pfkfb3; Ptpn12; Pitpnm1; Rtn1; Maff; Sgk; BB220380; Tes; Elmo1; Tm6sf1; Mast2; Stx11; Dhrs3; Tlr2 mCD8 and - Clnk Gcet2; BC028528; sept3; Sema4f; Fkbp1b; Rasgrp3; Btla; Asahl; 4930506M07Rik; Lrrc1; hBDCA3 Igsf4a Tlr3 ; Lima1; Dbn1; 1700025G04Rik; Tspan33; Fnbp1; Itga6; Plekha5; Fuca1; Fgd6; Zbed3; 9030625A04Rik; Rab32; Ptcd2; Snx22; Gfod1 Gas2l3; Rab11a; Ptplb; Cbr3; Pqlc2; Slamf8; St3gal5; 4930431B09Rik; Dock7; Stx3; Csrp1; Nbeal2; Gnpnat1; Slc9a9; Ncoa7 mCD11b - - Il1rn; Papss2; Pram1 Il1r2; Oas3; Rin2; Ptgs2; Gbp2; Oas2; Ccl5; Pilra; Sirpa; Pla2g7; Ifitm2; and Csf1r; Tlr5; Centa1; Pygl; Ms4a7; Cdcp1; Nfam1; BC013672; Slc7a7; hBDCA1 Igsf6; Csf3r; Tesc; Ncf2; Ripk2; Map3k3; Ripk5; Lactb; Rsad2; Parp14; S100a4; Rtn1; Cst7; Car2; D930015E06Rik; Gyk; Ank; Atp8b4; Emilin2; Ifitm1; 1810033B17Rik; Arrdc2; Slc16a3; Fcgr3; Clec4a2; Ksr1; Itgax; Lrp1; Dennd3; Ifitm3 Sqrdl; Hdac4; Rel; Pou2f2; Chka; Lyst; Ubxd5; Jak2; Cd300a; Lst1; Ssh1; Casp1; D12Ertd553e; Ogfrl1; Rin3; Cd302; Pira2

*Ratio expressed as Minimum expression among the cell types selected/Maximum expression among all other cell types. Genes already known to be preferentially expressed in the cell types selected are shown in boldface.

[72], CD4 T cells (n = 2), and B1 (n = 2) and B2 (n = 2) cells Indeed, IKDCs express the conserved genetic signature of NK [18]. Information regarding the original sources and the pub- cells but not of DCs (Table 8 and Additional data file 4). Thus, lic accessibility of the corresponding datasets are given in these results strongly support the hypothesis that the cells Table 1. As depicted in Figure 4a, the hierarchical clustering described as IKDCs feature a specific subset of mouse NK with complete linkage results of these data sets, together with cells that are in a particular differentiation or activation sta- our novel 430 2.0 data, clearly show that IKDCs cluster with tus, rather than a new DC subset. NK cells, close to other lymphocytes, and not with DCs.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.13

Table 7

Selected annotations for the conserved transcriptomic signatures identified for DC subsets when compared to one another

Cell type Annotation Genes pDC Endoplasmic reticulum Ern1, Lman1, Txndc5, Rdh11, Tm7sf2, Asph, Ormdl3, Stch, Nucb2, Ugcg, Itpr2, Bet1, Sec11l3, Atp2a3 Golgi stack BET1, HS3ST1, CHST12, SNAG1, LMAN1, MGAT5, GLCCI1, Pacsin1 Lysosome Lypla3, Npc1, Scarb2, Ctsb, Pcyox1, Cln5 Endocytosis/vesicle-mediated transport Bet1; Gata2; Igh-6; Lman1; Npc1; Pacsin1; Vamp1 Integral to plasma membrane EPHA2, SCARB2, CSF2RB, SIT1, ATP2A3, IFNAR2, VAMP1, PTPRS, SLC23A2, PTPRCAP, LANCL1, TM7SF2, CCR2, TSPAN31 Inflammatory response TLR7, CYBB, IRF7, CCR2, BLNK Intracellular signaling cascade/I- %B kinase/NF- %B cascade SNAG1, SLC44A2, TMEPAI, CARD11, ERN1, SLA2, IFNAR2, CARHSP1, SNX9, RALGPS2, CXXC5, CCR2, BLNK, RHOH Regulation of transcription, DNA-dependent/DNA binding/ 1110028C15Rik; Aff3; Anks3; Arid3a; Bcl11a; Carhsp1; Cbx4; Cdkn2d; transcription regulator activity/RNA polymerase II transcription Creb3l2; Cxxc5; Ern1; Ets1; Gata2; Hivep1; Ifnar2; Irf7; Maged1; Myb; factor activity/IPR004827: Basic-leucine zipper (bzip) transcription Nucb2; Prkcbp1; Runx2; Sla2; Spib; Tcf4; Tspyl4; Xbp1; Zbtb20 factor Systemic lupus erythematosus LMAN1, CCR2, ETS1 Regulation of apoptosis CDK5R1, CARD11, ERN1, CBX4, TXNDC5, CTSB cDC Response to pest, pathogen or parasite/defense response/immune ANXA1; NR4A2; CIAS1; TLR2; CD83; CD86; IL18; CXCL16; MAST2; response/response to stress/inflammatory response/cytokine AIF1; CIITA; SNFT; Lzp-s, Lyzs; ENTPD1; CXCL9; PLSCR1; BCL6; SGK; biosynthesis/response to bacteria/lymphocyte activation TXNRD2; DDB2; AHR; IRF2; LST1; SOAT1; HLA-DOB; CD1D; IL15; Rbpsuh; Swap70; Hmgb3; Egr1 Cytoskeleton/actin binding/filopodium/cell motility FLNA; FHOD1; CNN2; MYO1F; ACTN1; VASP; EPB41L2; FSCN1; KLHL5; MARCKS; Epb4,1l2; Mast2; Aif1; Csrp1; Elmo1; LIMA1; LMNB1; STOM; Nav1, CXCL16, ANXA1 Morphogenesis/cell organization and biogenesis/neurogenesis Rasgrp4; Myo1f; Aif1; Pak1; Pacs1; Vasp; Tiam1; Lst1; Cnn2; Numb; Csrp1; Fhod1; Nav1; Rab32; Stx11; Ezh2; Epb4,1l2; Flna; Acp2; Elmo1; Ralb; Rab31; Id2; Tnfaip2; Txnrd2; Anpep; Il18; Rbpsuh, Nr4a2; Spint1 Signal transduction/cell communication/MMU04010:MAPK signaling ADAM8; AHR; ANXA1; ARRB1; Asb2; Avpi1; CD83; CD86; Chn2; pathway/regulation of MAPK activity/GTPase regulator activity/ CIAS1; CXCL9; Dusp16; DUSP2; Elmo1; ENTPD1; FLNA; Hck; IL15; small GTPase mediated signal transduction/IPR003579:Ras small IL18; INPP1; Kit; Lrrk2; Mast2; NR4A1; NR4A2; PAK1; PDE8A; PIK3CB; GTPase, Rab type PPFIBP2; Rab31; Rab32; Ralb; Rasgrp4; RBPSUH; RGS12; Rin3; RTN1; Sla; SLC8A1; Snx10; Snx8; Tiam1; TLR2; Arhgap22; Ddef1; Rgs12; Usp6nl Transcription regulator activity Junb, Id2, Asb2, Ddef1, Irf2, Nr4a2, C2ta, Nab2, Egr1, Nr4a1, Ahr, 9130211I03Rik, Tgif, Rbpsuh, Bcl6 Apoptosis Ahr, Nr4a1, Il18, Bag3, Cias1, Elmo1, Cd1d1, Sgk, Bcl6 mCD8 Cell organization and biogenesis DBN1, RAB32, ITGA6, FGD6, RAB11A, SEMA4F and hBDCA3 Intracellular signaling cascade/small GTPase mediated signal MIST, TLR3, SNX22; DOCK7; FGD6; RAB11A; RAB32; RASGRP3; sep3 mCD11b Immune response/defense response/inflammatory response/positive IFITM3, PTGS2, POU2F2, LST1, GBP2, CCL5, OAS2, FCGR2A, NCF2, and regulation of cytokine production/response to pest, pathogen or CSF1R, TLR5, CSF3R, IL1R2, CST7, IL1RN, NFAM1, IFITM2, IFITM1, hBDCA1 parasite/antimicrobial humoral response/IPR006117:2-5- LILRB2, OAS3, LYST, CLEC4A, IGSF6, HDAC4, PLA2G7, RIPK2, OAS2, oligoadenylate synthetase OAS3; Rel; Fcgr3 Signal transduction/cell communication/signal transducer activity/ CASP1; CCL5; CD300A; CD302; CENTA1; CHKA; CLEC4A; CSF1R; positive regulation of I- %B kinase/NF- %B cascade/protein-tyrosine CSF3R; FCGR2A; IFITM1; IGSF6; IL1R2; IL1RN; ITGAX; JAK2; KSR1; kinase activity/IPR003123:Vacuolar sorting protein 9; LILRB2; LRP1; LYST; MAP3K3; MS4A7; NFAM1; OGFRL1; REL; RIN2; vesicle-mediated transport; endocytosis RIN3; RIPK2; RIPK5; RTN1; TLR5; Fcgr3 Chemotaxis/cell adhesion ITGAX, CD300A, CSF3R, EMILIN2, CLEC4A, CCL5, Fcgr3 HSA04640:hematopoietic cell lineage CSF1R, CSF3R, IL1R2 Asthma. Atopy PLA2G7, CCL5,

Lineage -CD16 +HLA-DR + cells based on their antigen-presentation capabilities. This subset A subset of leukocytes characterized as lineage -CD16 +HLA- segregates apart from BDCA1 and BDCA3 DCs and pDCs DR + (hereafter referred to as CD16 cells) has been reported in upon gene expression profiling [31]. It is not found in signifi- human blood, and claimed to be a subpopulation of DCs cant amounts in secondary lymphoid organs of healthy

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.14

donors, contrary to pDCs and BDCA1 or BDCA3 cDCs. It that of macrophages from hematopoeitic precursors [74], is expresses specific pattern recognition receptors, such as expressed to much higher levels in CD16 cells and monocytes TLR4 and TLR8, and chemokine receptors, such as CX3CR1 compared to DCs (average signal intensity of 6,263 in CD16 and CMKOR1 [31], which were initially described to be pref- cells compared to 3,479 in monocytes, 65 in pDCs, 309 in erentially expressed by monocytes in humans [73]. As the BDCA1 DCs and <50 in BDCA3 DCs). CD16 cells also express transcriptional relationship of CD16 cells with other known to high levels many genes that are absent or only expressed to DC populations was originally established based solely on the very low levels in LN-DCs compared to both lymphoid and myeloid cells, in particular many members of the gimap fam- ily. Reciprocally, many of the genes characterized above as (a) specifically expressed in human and mouse LN-DCs are absent or expressed only to low levels in CD16 cells, in partic- ular FLT3 and SCARB1 . Thus, CD16 cells likely differentiate along the canonical myeloid lineage rather than belong to the LN-DC family. However, many genes are also specifically expressed to much higher levels in LN-DC subsets and CD16 cells than in monocytes, neutrophils and lymphocytes, attest- m. B m. B2 m. B1 m. NK m. NK m. DN ing to the existence of biological functions common, and spe- m. cDC m. CD4 m. pDC m. pDC m. CD8 m. CD8 m. IKDC m. CD4 T m. CD4 T cific, to DC subsets and CD16 cells. Thus, these results m. CD11b DCs Lymphocytes strongly suggest that CD16 cells represent a particular subset NK of monocytes endowed with DC-like properties. One possibility is that CD16 cells are the naturally occurring equiv- alents of the 'monocyte-derived DCs' generated in vitro .

(b) In vitro GM-CSF derived DCs In vitro derived GM-CSF DCs are the most commonly used model to analyze DC biology. They are often used to investi- gate the interaction between DCs and other cell types or with pathogens, both in mouse (bone marrow (BM)-derived GM- CSF DCs) and human (monocyte-derived GM-CSF DCs). However, the relationship between these in vitro GM-CSF- h. B

h. NK derived DCs and the LN-DC subsets present in vivo in the h. neu. h. pDC h. CD16 h. mono. h. CD8 T h. CD4 T steady state is not clear. A very recent publication suggests h. BDCA1 h. BDCA3 that in vitro derived GM-CSF mouse DCs may correspond to Lymphocytes DCs Myeloid the DCs that differentiate from Ly6C + monocytes in vivo only cells under inflammatory conditions and appear specialized in the production of high levels of tumor necrosis factor-! and inducible nitric oxide synthase in response to intracellular bacteria, therefore differing from LN-DCs according to both ClusteringFigure 4 of mouse IKDCs and human CD16 cells ontogenic and functional criteria [75]. To gain further Clustering of mouse IKDCs and human CD16 cells. Hierarchical clustering insights into the relationship between monocytes, with complete linkage was performed on the indicated cell populations macrophages, LN-DCs, and in vitro derived GM-CSF DCs, we isolated from: (a) mouse and (b) human. Mono, monocytes; neu, neutrophils. thus compared their global gene expression profiling in both human and mouse, using publicly available gene chip data. Information regarding the original sources and the public accessibility of the corresponding datasets are given in Table transcriptional profile of DCs, we sought to better understand 1. The results depicted in Figure 5 clearly show that the in the nature of these cells. For this, we reanalyzed the global vitro derived GM-CSF DCs cluster with monocytes and mac- gene expression profile of CD16 cells in comparison to not rophages and not with the LN-DCs. This result was further only DC subsets but also to monocytes, neutrophils, and lym- confirmed by PCA, which also showed that both mouse and phocytes. The results depicted in Figure 4b clearly show that human GM-CSF DCs are close to macrophages, and distant the CD16 cells cluster with neutrophils and monocytes and from LN-DCs (Additional data file 6). Indeed, we found many not with LN-DCs. Indeed, we find many genes that are genes that are expressed to much higher levels in monocytes, expressed to much higher levels in monocytes or neutrophils macrophages and in vitro derived GM-CSF DCs than in LN- and CD16 cells than in LN-DC subsets (Table 9 and DC subsets (Tables 10 and 11). As for human CD16 cells, these Additional data file 2). Interestingly, MAFB , which has been genes include the transcription factor Mafb . Reciprocally, described to inhibit the differentiation of DCs but to promote some of the genes identified in this study as specific to LN-

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.15

Table 8

Expression of APC, DC and NK signature genes in IKDCs

Ratio

Probe set ID Gene CD8 DC DN DC CD4 DC pDC cDC IKDC NK IKDC/DC NK/DC IKDC/NK

APC signature genes 98035_g_at H2-DMb1 2,701* 3,416 4,281 1,105 2,722 179 36 0.2 <0.1 5 92668_at Btk 454 259 331 252 277 91 20 0.4 <0.1 5 94834_at Ctsh 1,606 2,650 2,862 2,993 1,653 129 20 0.1 <0.1 6 94285_at H2-Eb1 8,183 7,761 7,201 5,285 14,120 1,018 74 0.2 <0.1 14 101054_at Cd74 9,094 7,810 7,313 5,158 12,258 1,031 55 0.2 <0.1 19 92633_at Ctsz 520 1,246 1,171 887 750 117 44 0.2 <0.1 3 94256_at Clic4 1,668 1,067 1,234 739 717 440 295 0.6 0.4 1 160781_r_at Unc93b1 683 710 789 301 138 36 22 0.3 0.2 2

Pan-DC signature genes 95295_s_at Flt3 2,769 2,004 2,231 2,069 2,547 270 45 0.1 <0.1 6 100095_at Scarb1 716 405 333 297 398 125 73 0.4 0.2 2

Non-DC signature genes 96172_at Gimap4 29 62 20 314 319 5,274 982 263 49 5 92398_at Vps37b 111 139 44 76 56 462 159 11 4 3 161265_f_at Lck 99 80 105 235 199 1,991 366 25 5 5

NK signature genes 97781_at Ncr1 20 20 20 73 39 1,483 120 20 2 12 97113_at Fasl 20 28 20 22 30 440 263 15 9 2 102272_at Cd160 75 107 62 82 58 780 246 7 2 3 100764_at Il2rb 26 45 40 50 65 84 501 1 8 0.2 99334_at Ifng 20 20 20 29 38 203 109 5 3 2 93931_at Prf1 33 21 35 94 86 839 1,287 9 14 1 92398_at Vps37b 111 139 44 76 56 462 159 11 4 3 cDCs are expressed only to much lower levels in GM-CSF papers have recently established that in vitro derived FLT3-L DCs. However and interestingly, compared to monocytes, in DCs constitute the true equivalent of LN-DCs and constitute vitro derived GM-CSF DCs harbor stronger levels of other the only proper surrogate model currently available for their lymph node resident cDC-specific genes, including scarb1 , study [75-77]. snft/9130211l03Rik , spint1 , ctsh , C22ORF9/ 5031439G07Rik , and bri3bp . Thus, in vitro derived GM-CSF DCs seem to harbor a strong myeloid gene signature but also Discussion express some of the LN-DC-specific genes, consistent with By performing meta-analyses of various datasets describing their myeloid ontogeny and their ability to exert myeloid-type global gene expression of mouse spleen and human blood functions but also with their acquisition of DC functional leukocyte subsets, we have been able to identify for the first properties. In conclusion, our gene chip data analysis is time conserved genetic programs common to human and consistent with a very recent report suggesting that in vitro mouse LN-DC subsets. All the LN-DC subsets examined here derived GM-CSF mouse DCs correspond to inflammatory are shown to share selective expression of several genes, while DCs and differ greatly from LN-DCs [75]. Indeed, several

Table 9

Expression of APC, DC and myeloid signature genes in CD16 cells

Dendritic cells Myeloid cells Ratio to DC

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.16

Table 9 (Continued)

Expression of APC, DC and myeloid signature genes in CD16 cells

Probe set ID Gene BDCA1 BDCA3 pDC Mono Neu CD16 cells CD16 Mono Neu

APC signature genes 203932_at HLA-DMB 8,636* 7,929 5,894 5,194 173 2,581 0.3 0.6 <0.1 205101_at CIITA 2,803 2,354 724 531 50 226 <0.1 0.2 <0.1 219574_at MARCH1 587 777 544 1,214 58 810 1 2 <0.1 201425_at ALDH2 9,279 7,841 6,034 8,504 706 1,760 0.2 0.9 <0.1 222891_s_at BCL11A 569 747 4,502 310 50 213 <0.1 <0.1 <0.1 205504_at BTK 1,120 822 1,132 1,409 281 1,786 2 1 0.3 202295_s_at CTSH 6,197 2,528 1,211 3,949 75 2,440 0.39 0.6 <0.1 213831_at DQA1 11,535 7,503 5,919 4,701 50 252 <0.1 0.4 <0.1 215536_at DQB2 432 391 157 180 81 52 0.1 0.4 0.2 209312_x_at DRB1 14,608 14,477 13,250 11,915 228 14,007 1 0.8 <0.1 209619_at CD74 12,533 12,210 10,498 9,020 867 7,383 0.6 0.7 <0.1 210042_s_at CTSZ 906 848 692 370 153 673 0.7 0.4 0.2 201560_at CLIC4 920 305 663 3,023 165 354 0.4 3 0.2 217388_s_at KYNU 2,414 1,059 2,204 3,516 50 3,738 2 1 <0.1 203927_at NFKBIE 529 272 232 197 63 290 0.5 0.4 0.1 220998_s_at UNC93B1 966 850 1,938 862 449 1,235 0.6 0.4 0.2

Pan-DC signature genes 206674_at FLT3 3,032 5,883 2,169 208 <50 <50 <0.1 <0.1 <0.1 219256_s_at SH3TC1 1,263 899 1,128 392 166 858 0.7 0.3 0.1 218617_at TRIT1 1,159 1,246 1,851 509 <50 339 0.2 0.3 <0.1 231810_at BRI3BP 691 735 836 298 146 279 0.3 0.4 0.2 209139_s_at PRKRA 846 1,067 1,440 316 74 497 0.3 0.2 <0.1 225764_at ETV6 2,172 2,432 1,726 1,143 938 941 0.4 0.5 0.4 208837_at TMED3 1,317 1,852 1,859 665 <50 1,022 0.6 0.4 <0.1 219218_at BAHCC1 87 86 250 <50 <50 <50 0.2 0.2 0.2 1552256_a_at SCARB1 325 425 942 165 128 59 <0.1 0.2 0.1

Non-DC signature genes 219243_at GIMAP4 68 <50 <50 4,404 3,504 1,334 20 65 52 221704_s_at VPS37B 54 <50 <50 593 962 487 9 11 18 204891_s_at LCK <50 <50 <50 92 181 65 - - - 214582_at PDE3B 78 <50 <50 129 625 114 1 2 8

Myeloid signature genes 225987_at STEAP4 <50 <50 <50 877 6,090 <50 - - - 1552773_at CLEC4D <50 <50 <50 452 520 <50 - - - 222934_s_at CLEC4E 214 124 133 2,837 5,885 229 1 13 28 202974_at MPP1 591 281 377 3,721 2,408 1,341 2 6 4 205098_at CCR1 93 <50 115 3,712 3,627 106 1 32 31 223044_at SLC40A1 769 276 321 5,018 3,444 <50 - 6 4 224341_x_at TLR4 94 <50 <50 1,411 2,869 540 6 15 31 204714_s_at F5 <50 <50 <50 1,392 2,313 <50 - - - 203561_at FCGR2A 1,010 44 51 2,985 7,151 2,857 3 3 7 210772_at FPRL1 <50 <50 <50 389 3,454 70 3 - - 204924_at TLR2 904 211 57 2,870 5,548 1,606 2 3 6 215223_s_at SOD2 1,474 946 528 3,528 7,599 4,236 3 2 5 222218_s_at PILRA 1,168 150 136 2,899 4,035 3,982 3 2 3 210423_s_at SLC11A1 81 60 38 1,767 2,930 3,334 41 22 36 203045_at NINJ1 357 66 71 1,104 3,129 1,934 5 3 9 201669_s_at MARCKS 521 389 <50 2,449 3,224 1,730 3 5 6 207697_x_at LILRB2 1,271 78 774 3,353 3,711 4,903 4 3 3 1553297_a_at CSF3R 1,902 409 156 3,433 6,687 282 0.2 2 4 220088_at C5AR1 56 34 93 2,316 5,099 3,824 41 25 55 221698_s_at CLEC7A 3,229 4,295 79 6,642 7,061 5,680 1 2 2 204204_at SLC31A2 442 187 <50 1,579 2,047 1,671 4 4 5

*Average expression across replicates. Mono, monocyte; neu, neutrophil.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.17

Table 10

Comparison of the transcriptome of human GM-CSF monocyte-derived DCs to that of blood DCs

Ratio to monocytes

Probe set ID Name Mono PBMC-M # mo-M # mo-DC CD16 BDCA3 BDCA1 pDCs

Myeloid signature genes 222934_s_at CLEC4E 2,358 0.20 0.19 0.04 - - 0.05 - 209930_s_at NFE2 823 0.06 0.06 0.89 0.10 - 0.06 - 202974_at MPP1 3,622 0.40 1.25 0.68 0.33 0.08 0.15 0.11 205098_at CCR1 3,528 0.76 1.63 1.83 0.03 - 0.03 0.03 203535_at S100A9 11,192 0.05 0.37 0.01 0.12 0.02 0.17 0.01 201743_at CD14 8,096 0.44 1.13 0.34 0.01 - 0.02 0.01 224341_x_at TLR4 1,417 0.13 1.10 0.35 0.34 - 0.06 - 203561_at FCGR2A 2,946 0.18 0.80 1.36 0.85 - 0.33 0.02 204924_at TLR2 3,220 0.14 0.80 0.32 0.54 0.08 0.31 0.02 218739_at ABHD5 285 0.35 0.99 0.67 0.33 - - - 201089_at ATP6V1B2 3,178 2.05 2.46 1.70 0.66 0.12 0.34 0.21 201631_s_at IER3 2,042 0.42 1.74 0.82 0.10 0.06 0.14 0.12 222218_s_at PILRA 2,709 0.73 1.24 1.23 1.25 0.05 0.39 0.05 210423_s_at SLC11A1 1,713 0.47 0.82 0.25 1.75 0.04 0.05 - 203045_at NINJ1 1,190 1.69 3.59 3.41 1.59 0.27 0.44 0.26 200958_s_at SDCBP 11,323 0.87 1.16 0.90 0.61 0.33 0.40 0.26 202917_s_at S100A8 15,661 0.02 0.41 0.01 0.11 0.01 0.27 0.03 217748_at ADIPOR1 2,229 0.57 0.48 1.16 0.30 0.30 0.36 0.28 201669_s_at MARCKS 2,340 0.84 2.57 1.57 0.65 0.16 0.20 - 207697_x_at LILRB2 3,260 0.29 0.64 0.76 1.36 0.02 0.39 0.24 228220_at FCHO2 619 4.50 4.04 3.62 0.76 0.35 0.26 0.23 1553297_a_at CSF3R 3,121 0.42 0.69 0.37 0.08 0.11 0.52 0.04 220088_at C5AR1 2,059 2.56 3.63 1.30 1.60 - 0.03 0.04 212501_at CEBPB 3,490 3.26 3.23 3.30 1.26 0.06 0.49 0.06 221698_s_at CLEC7A 6,596 0.24 0.55 0.63 0.74 0.62 0.46 0.01 209551_at YIPF4 526 0.85 1.65 1.91 0.41 0.37 0.44 0.37 204204_at SLC31A2 1,933 0.94 1.14 0.69 0.76 0.10 0.22 0.03

Pan-DC signature genes 206674_at FLT3 221 - - - - 24.01 12.76 9.26 219256_s_at SH3TC1 395 1.02 2.73 1.12 2.01 2.22 3.01 2.86 218617_at TRIT1 498 0.49 0.58 0.86 0.71 2.46 2.15 3.61 231810_at BRI3BP 301 0.98 1.42 1.99 0.98 2.35 2.10 2.70 209139_s_at PRKRA 325 1.12 1.77 1.47 1.57 3.17 2.42 4.37 225764_at ETV6 1,097 0.43 1.13 2.00 0.75 2.04 1.78 1.48 208837_at TMED3 595 1.50 2.81 1.64 1.46 2.91 1.98 2.94 219218_at BAHCC1 - - - - - >1.7 >1.5 >4.7 1552256_a_at SCARB1 151 8.98 6.58 7.21 - 2.33 1.70 5.30 cDC signature genes 206298_at ARHGAP22 - >5.8 >6.5 >3.1 - >6.2 >4.6 - 227329_at BTBD4 - >1.6 >2.8 >5.8 - >9.3 >8.7 - 219386_s_at SLAMF8 98 24.75 38.66 23.99 0.51 15.48 5.30 0.51 220358_at SNFT 148 0.62 0.34 8.62 5.66 16.01 4.82 0.34 224772_at NAV1 64 2.01 3.25 1.40 2.00 23.87 10.50 1.62 205101_at CIITA 481 0.29 0.12 1.09 0.48 4.51 5.28 1.43

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.18

Table 10 (Continued)

Comparison of the transcriptome of human GM-CSF monocyte-derived DCs to that of blood DCs

218631_at AVPI1 - >18.7 >31.3 >64.8 >1.6 >3.2 >7.0 - 202826_at SPINT1 84 4.65 7.15 8.79 0.90 2.59 2.92 0.68 208660_at CS 1,848 1.24 0.99 1.04 0.84 1.70 1.63 0.89

APC signature genes 203932_at HLA-DMB 5,137 1.28 0.64 1.37 0.44 1.45 1.62 1.14 219574_at MARCH1 1,133 0.42 0.89 0.73 0.62 0.64 0.44 0.46 201425_at ALDH2 8,782 0.51 0.54 0.34 0.18 0.84 1.01 0.69 222891_s_at BCL11A 310 0.98 0.34 0.50 0.74 2.40 1.73 14.23 205504_at BTK 1,372 0.29 0.47 0.64 1.13 0.58 0.75 0.81 202295_s_at CTSH 3,755 1.76 2.37 2.09 0.56 0.63 1.57 0.31 209312_x_at HLA-DRB1 12,737 1.02 0.57 1.34 1.11 1.12 1.11 1.00 209619_at CD74 8,540 1.49 0.86 2.12 0.73 1.33 1.34 1.11 210042_s_at CTSZ 369 0.76 1.13 17.00 1.66 2.13 2.17 1.83 201560_at CLIC4 2,828 0.87 0.88 1.00 0.12 0.10 0.28 0.22 217388_s_at KYNU 3,429 1.50 1.95 0.90 0.94 0.30 0.65 0.63 217118_s_at C22orf9 1,617 3.33 3.46 2.77 1.43 1.85 1.79 1.04 203927_at NFKBIE 173 3.30 9.96 3.13 1.45 1.39 2.60 1.25 220998_s_at UNC93B1 847 0.60 1.31 0.97 1.31 0.99 1.06 2.27

Non-DC signature genes 219243_at GIMAP4 4,384 0.15 0.11 0.19 0.27 - - - 221704_s_at VPS37B 559 0.26 0.90 0.47 0.80 - - - 204891_s_at LCK 96 1.48 0.52 0.52 0.59 - - - 214582_at PDE3B 144 2.82 2.99 2.43 0.76 - 0.51 -

*Average expression across replicates. Genes for which expression between monocyte-derived DCs and blood DCs or blood cDCs varies more than two-fold are shown in bold. mo-DC, monocyte-derived GM-CSF DC; mo-M #, monocyte-derived M-CSF macrophages; mono, monocyte; PBMC-M #, human peripheral blood mononuclear cell-derived M-CSF macrophages. harboring only low levels of other transcripts present in all cytes/macrophages and to cDCs, but not to polymorphonu- other leukocytes. These analyses indicate that LN-DCs, clear cells or to lymphoïd cells [80,81]. Under the including pDCs, constitute a specific family of leukocytes, experimental conditions used in the corresponding report, distinct from those of classic lymphoid or myeloid cells. Fur- pDCs were not detected in the progeny of MDPs. Here, we thermore, we demonstrate a striking genetic proximity show that the transcriptome programs of mouse spleen and between mouse and human pDCs, which are shown for the human blood cDCs exhibit only a very limited overlap with first time to harbor a very distinct transcriptional signature as that of monocytes/macrophages (Figure 2). This is consistent large and specific as that observed for NK cells or T cells. In with the recent observation that monocytes can give rise to contrast, a higher genetic distance is observed between mucosal, but not splenic, cDCs, suggesting that splenic cDCs mouse and human conventional DC subsets, although a par- develop from MDPs without a monocytic intermediate [81]. tial functional equivalence is suggested between mCD8 ! and While mouse pDCs have been argued to arise from both lym- hBDCA3 cDCs on the one hand versus mCD11b and hBDCA1 phoid or myeloid progenitors, their gene expression overlaps cDCs on the other hand. with lymphoid or myeloid cells are limited. Interestingly, a murine progenitor cell line that exhibits both cDC and pDC Our finding that LN-DCs constitute a distinct entity within differentiation potential has been described recently [82], immune cells raises the question of whether these cells form suggesting that putative pan-DC progenitors might also exist a distinct lineage in terms of ontogeny, or whether their in vivo , which would be consistent with the gene profiling shared gene expression profile (notably that between cDCs analyses presented here. and pDCs) reflects a functional rather than a developmental similarity. To date, the place of both cDCs and pDCs in the Our study identifies transcriptional signatures conserved hematopoietic tree is not clear [78,79]. A BM progenitor, between mouse and human, common to all LN-DC subsets named macrophage and dendritic cell progenitor (MDP), has examined, or specific to pDCs, cDCs, or individual cDC been recently identified that specifically gives rise to mono- subsets. A genetic equivalence is suggested between mouse

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.19

Table 11

Comparison of the transcriptome of mouse GM-CSF BM-derived DCs to that of spleen DCs

Ratio to monocytes

Probe set ID Name Mono Mono(2) M # BM-M # BM-DC pDC CD8 DC CD11b DC

Myeloid signature genes 1420804_s_at Clec4d 4,934 0.65 0.49 0.75 0.41 - - - 1420330_at Clec4e 5,511 0.11 0.22 0.23 0.11 - - - 1450808_at Fpr1 119 1.91 - 5.55 2.41 - - - 1452001_at Nfe2 139 1.44 - - 3.31 - - - 1450919_at Mpp1 1,888 0.15 2.05 1.75 0.52 0.23 0.09 0.07 1419609_at Ccr1 403 1.27 4.04 0.53 3.98 0.2 - - 1417061_at Slc40a1 2,588 0.68 - 0.56 0.07 0.01 0.01 0.02 1448756_at S100a9 8,664 1.2 - 0.01 0.99 0 0 - 1417268_at Cd14 6,745 0.1 0.3 0.6 0.19 0.02 0.01 0.01 1418163_at Tlr4 464 0.1 0.36 0.93 0.66 - 0.07 0.06 1448620_at Fcgr3 1,471 2.02 3.56 2.15 2.46 - 0.02 0.07 1422953_at Fpr-rs2 839 2.04 0.12 0.85 1.58 - - 0.05 1419132_at Tlr2 1,763 0.11 0.42 0.24 0.48 0.04 0.1 0.14 1417566_at Abhd5 170 0.19 0.72 0.86 2.2 0.18 0.45 0.25 1415814_at Atp6v1b2 1,556 0.22 2.75 1.57 1.43 0.18 0.27 0.24 1427327_at Pilra 434 1.53 0.16 0.47 2.29 0.1 - 0.21 1418888_a_at Sepx1 4,416 0.48 0.34 0.31 0.56 0.03 0.04 0.05 1438928_x_at Ninj1 5,574 0.03 1.3 0.46 0.36 0.03 0.02 0.02 1448881_at Hp 400 3.19 0.14 0.06 3.09 - - - 1449453_at Bst1 340 1.08 4.97 0.58 1.61 0.21 0.51 - 1419394_s_at S100a 8 10,190 1.37 0.01 0.01 0.66 - 0 - 1437200_at Fcho2 311 1.09 1.32 1.06 0.76 0.28 0.2 0.33 1418806_at Csf3r 2,598 0.2 0.14 0.19 0.11 - - 0.03 1439902_at C5ar1 317 8.21 0.19 1.63 0.37 - - - 1456046_at Cd93 1,559 0.1 0.49 1.18 0.33 0.02 - - 1418901_at Cebpb 3,797 0.14 0.7 0.22 0.42 0.02 0.01 0.02 1420699_at Clec7a 2,748 0.83 2.62 0.44 1.71 0.08 0.06 0.54

Pan-DC signature genes 1419538_at Flt3 51 0.74 - - 0.7 16.2 25.32 17.78 1427619_a_at Sh3tc1 - >1.1 >6.8 >2.8 >4.9 >5.2 >6.5 >4.6 1424489_a_at Trit1 54 7.28 0.44 0.76 1.23 9.03 11.53 8.63 1428744_s_at Bri3bp 161 0.84 0.6 1.44 3.28 6.09 7.24 5.98 1448923_at Prkra 72 1.28 0.77 2.89 2.57 4.45 7.88 3.63 1434880_at Etv6 140 5.39 1.52 0.74 1.75 5.79 6.02 7.78 1416108_a_at Tmed3 154 0.81 3.74 2.63 4.65 10.17 4.48 3 1436633_at Bahcc1 41 1.77 - 0.83 - 1.8 3.88 2.35 1437378_x_at Scarb1 97 5.02 1.25 2.61 3.17 7.41 8.27 4.05 cDC signature genes 1435108_at Arhgap22 63 - - 2.37 0.57 0.59 10.65 4.43 1429168_at Btbd4 129 0.19 0.27 - 0.47 0.81 3.89 3.8 1425294_at Slamf8 146 1.06 39.89 1.83 1.77 0.39 8.48 5.27 1453076_at 9130211I03Rik 36 1.61 2.85 1.03 13.11 0.62 30.94 25.64 1436907_at Nav1 102 1.59 0.74 2.63 1.96 1.21 6.08 13.14 1421210_at C2ta 125 0.17 1.79 0.19 0.93 1.46 5.94 5.43

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.20

Table 11 (Continued)

Comparison of the transcriptome of mouse GM-CSF BM-derived DCs to that of spleen DCs

1423122_at Avpi1 150 0.32 - 0.2 0.86 0.61 2.47 7.62 1416627_at Spint1 - >1.5 >1.1 - >22.9 >1.6 >25.7 >30.6 1450667_a_at Cs 396 2.47 0.9 1.19 3.54 2.83 4.64 4.5

APC signature genes 1419744_at H2-DMb2 451 0.12 0.1 0.08 1.47 0.45 0.48 1.69 1443687_x_at H2-DMb1 547 0.56 0.13 0.11 1.56 1.06 0.82 3.13 1434955_at March1 80 32.64 0.83 1.51 3.48 3.73 13.4 8.57 1448143_at Aldh2 867 0.47 2.14 2.07 1.32 0.95 0.65 0.45 1419406_a_at Bcl11a 60 1.47 0.34 - 0.71 20.41 7.63 9.19 1422755_at Btk 416 0.56 0.76 1.3 1.15 0.88 1.45 1.17 1418365_at Ctsh 1,393 0.81 3.9 2.19 2.15 3.69 1.24 2.16 1417025_at H2-Eb1 6,385 0.13 0.39 0.04 0.8 0.9 1.31 1.33 1425519_a_at Cd74 8,377 0.36 0.95 0.2 0.9 0.83 0.97 0.98 1417868_a_at Ctsz 7,061 0.05 1.16 0.95 0.85 0.5 0.3 0.49 1423393_at Clic4 2,807 0.07 2.04 0.84 0.57 0.69 0.72 0.67 1430570_at Kynu 31 1.23 - - 3.21 12.87 5.16 11.56 1435745_at 5031439G07Rik 356 0.95 0.73 2.76 2.51 3.23 3.14 4.28 1458299_s_at Nfkbie 767 0.4 0.62 0.1 0.44 1.25 0.65 1.27 1423768_at Unc93b1 663 0.1 2.27 2.69 1.46 1.2 0.93 0.91

Non-DC signature genes 1424375_s_at Gimap4 362 0.14 0.29 - 0.1 0.11 - 0.11 1424380_at Vps37b 313 0.44 0.46 0.45 0.26 0.28 0.28 0.27 1425396_a_at Lck 118 - 0.57 0.2 0.32 0.21 - 0.17 1433694_at Pde3b 352 0.69 0.15 0.16 0.42 - 0.65 0.35

*Average expression across replicates. Genes for which expression between mouse bone-marrow derived GM-CSF DCs (BM-DCs) and spleen DCs or spleen cDCs varies more than two-fold are shown in bold. BM-M #, mouse bone marrow-derived M-CSF macrophages; M #, peritoneal mouse macrophages; mono, mouse spleen monocytes from the SB laboratory; mono(2), mouse spleen monocytes from the BP laboratory, as listed in Table 1.

CD8 ! cDCs and human BDCA3 cDCs, and between mouse species. This could be due to a true absence of orthologous CD11b cDCs and human BDCA1 cDCs. In contrast to the genes genes between these two vertebrate species, or to a lack of selectively expressed in subsets of myeloid or lymphoid cells identification of an existing orthology relationship. It is also in a conserved manner between mouse and human, most of possible that some of the genes expressed only in mouse DCs the genes specifically increased in all LN-DC subsets or in or only in human DCs, and not conserved between the two individual LN-DC subsets are currently uncharacterized. As a species, might represent functional homologs, similar to what consequence, the functional annotations of the LN-DC is observed for human KIR and mouse Ly49 NK cell transcriptional signatures appear much less informative than receptors. This may be the case for the human LILRA4 (ILT7) those for myeloid cells, lymphocytes or APCs. This highlights and the mouse SIGLECH molecules, as both of them signal how much has already been deciphered regarding the molec- through immunoreceptor tyrosine-based activation motif ular regulation of antigen presentation or lymphocyte biol- (ITAM)-bearing adaptors to downmodulate IFN- !/" produc- ogy, as opposed to how little we know about the genetic tion by human and mouse pDCs, respectively, upon triggering programs that determine the specific features of LN-DCs. We of TLRs [83,84]. Thus, understanding the role in LN-DCs of believe that our study provides a unique database resource for genes identified here only in mouse or human might be future investigation of the evolutionarily conserved molecular important. The transcriptional signatures identified for pathways governing specific aspects of the ontogeny and mouse LN-DC subsets in this study have been confirmed by functions of leukocyte subsets, especially DCs. analyses of independent data recently published by others on mouse cDC subsets, B cells and T cells [11] or on cDCs and It should be noted that many genes are found to be expressed pDCs [15]. Most of the data for the mouse 430 2.0 compen- to very high levels in specific subsets of either mouse or man dium were generated in-house, with the exceptions being while no orthologous gene has been identified in the other CD4 T cells and myeloid cells. In humans, we generated the

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.21

data for non-DC populations, whereas data for DC subsets previously mentioned studies were analyzed. Thus, the and CD16 cells were all generated by another group and results of our analysis of the transcriptional signature of both retrieved from a public database. It is well known that data- IKDCs and CD16 cells emphasize the need to study the tran- sets for the same cell type can vary considerably between lab- scriptional signatures of individual cell populations in the oratories. However, many of the genes identified as specific context of multiple cell types of various phenotypes and for each mouse LN-DC subset using our own data were con- functions. Finally, this approach also allowed us to confirm a firmed by the analysis of other data independently generated very recent report that demonstrated that in vitro derived by the groups of M Nussenzweig and R Steinman [11]. These GM-CSF mouse DCs likely correspond to inflammatory DCs data are given in Additional data file 5. and greatly differ from LN-DCs, based on ontogenic and func- tional studies [75]. Thus, extrapolation to LN-DCs of the Our clustering analyses and PCA also showed relatively little results of the cell biology and functional studies performed dataset-dependent biases, and generally grouped related cell with in vitro derived GM-CSF DCs should only be made with populations together, even if they were from different origins extreme caution. (see, for instance, the PCA clustering of in vitro derived GM- CSF DC samples, which originated from two independent datasets in Additional data file 6). In addition, we analyzed by Conclusion real-time PCR the expression profile of 27 genes across mouse This study comparing whole genome expression profiling of leukocyte subsets from biological samples independent of human and mouse leukocytes has identified for the first time those used in the gene chips analysis. All the results were conserved genetic programs common to all LN-DCs or spe- consistent with the gene chip data (Additional data file 7). We cific to the plasmacytoid versus conventional subsets. In also confirmed specific expression of PACSIN1 in human depth studies of these genetic signatures should provide novel pDCs at both the mRNA and protein levels (Additional data insights on the developmental program and the specific func- file 8). Finally, we believe that our approach validates the tions of LN-DC subsets. The study in the mouse of the novel, gene expression profile identified for leukocyte subsets in the cDC-specific genes identified here should accelerate the strongest way possible, by demonstrating the evolutionary understanding of the mysteries of the biology of these cells in conservation between mouse and human. Indeed, the gene both mouse and human. This should help to more effectively signatures that we describe here are based on genes found translate fundamental immunological discoveries in the specifically expressed in putatively homologous subsets of mouse to applied immunology research aimed at improving mouse and human leukocytes compared to several other human health in multiple disease settings. types of leukocytes. This approach does not rely solely on the use of independent biological samples of similar origin and on different techniques for measurement of the expression of Materials and methods mRNA. It actually shows that orthologous genes share the Sorting of cell subsets same specific expression pattern in putatively homologous Duplicates of pDCs (Lin -CD11c +120G8 high ), CD8 ! cDCs (Lin - immune cell subsets from two different species, under condi- CD11c high CD8 !+120G8 -/low ), CD11b cDCs (Lin - tions where the markers used to purify the human and mouse CD11c high CD11b +120G8 -/low ) and NK cells (NK1.1 +TCR "-) cell populations, and the probes used to check the expression were sorted during two independent experiments from of the orthologous genes, differ considerably. Thus, we pooled spleens of untreated C57BL/6 mice. Splenic CD19 + B believe that the analyses presented here are extremely robust lymphocytes, CD4 T cells and CD8 T cells were sorted in other even though they were, in part, performed by creating com- independent experiments. Purity of sorted cell populations pendia regrouping data generated by different laboratories was over 98% as checked by flow cytometry (not shown). for different cell type. Processing of cell samples for the Affymetrix GeneChip In addition to our discovery of transcriptional signatures spe- assays cific to all LN-DCs or to LN-DC subsets, we demonstrate that, RNA was extracted from between 7.5 × 10 5 and 1.5 × 10 6 cells once identified, the transcriptional signatures of multiple cell for each leukocyte subset with the Qiagen (Courtaboeuf, types can be effectively used to help determine the nature of France) micro RNAeasy kit, yielding between 200 and 700 ng newly identified cell types of ambiguous phenotype or func- of total RNA for each sample. Quality and absence of genomic tions. In our attempt to appropriately place IKDCs and CD16 DNA contamination were assessed with a Bioanalyser (Agi- cells within the leukocyte family, we used the microarray data lent, Massy, France). RNA (100 ng) from each sample was from the original reports aimed at characterizing these cells used to synthesize probes, using two successive rounds of and compared them to the data from several other leukocyte cRNA amplification with appropriate quality control to populations. The conclusions of this analysis are in sharp con- ensure full length synthesis according to standard Affymetrix trast to those originally reported [15,31]. We believe that protocols, and hybridized to mouse 430 2.0 chips (Affyme- these opposing conclusions arise from the difference in the trix, Santa Clara, CA, USA). Raw data were transformed with contextual framework within which our data and that of the the Mas5 algorithm, which yields a normalized expression

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.22

value, and 'absent' and 'present' calls. Target intensity was set membership indexes, each comprised between 0 and 1 [34]. to 100 for all chips. For both mouse and human datasets, we heuristically set the number of clusters to 30, and the fuzziness parameter m was Individual analysis of the mouse 430 2.0 or human U133 taken as 1.2 (see [34] for the determination of m). Ten inde- Plus 2.0 compendia pendent runs of the algorithm were performed, and the one For each compendium, all datasets were normalized with the minimizing the inertia criterion was selected [34]. A thresh- invariant rank method and only one representative dataset old value of 0.9 was taken to select probe sets most closely was kept for redundant ProbeSets targeting the same gene. associated with a given cluster. This selection retained 4,062 The datasets were further filtered to eliminate genes with and 4,751 probe sets from mouse and human datasets, respec- similar expression in all samples, by selecting only the genes tively. Probe set clusters were then manually ordered to pro- expressed above 50 (respectively 100) in all the replicates of vide coherent pictures, which were visualized with Treeview. at least one population for the mouse (respectively human) datasets and whose expression across all samples harbored a Meta-analysis of aggregated mouse and human coefficient of variation above the median of the coefficient of datasets variation of all ProbeSets. The final dataset consisted of 7,298 We identified 2,227 orthologous genes that showed signifi- (respectively 11,507) ProbeSets for the mouse 430 2.0 com- cant variation of expression in both the mouse 430 2.0 and pendium (respectively human U133 Plus 2.0), representing U133 Plus 2.0 human datasets. This dataset is accessible as an individual genes with differential expression between ex vivo Excel workbook in Additional data file 3. In order to compare isolated cell subsets. The final dataset consisted of 12,857 the expression patterns of these genes between human and (respectively 6,724) ProbeSets for the mouse 430 2.0 com- mouse, the log signal values for each of these genes were first pendium (respectively human U133 Plus 2.0), representing normalized to a mean equal to zero and a variance equal to 1, individual genes with differential expression between LN- independently in the mouse and human datasets, as previ- DCs, monocytes/macrophages and in vitro derived GM-CSF ously described for comparing the gene expression program DCs. These datasets for ex vivo isolated cells are accessible as of human and mouse tumors [22,27]. The two normalized Excel workbooks in Additional data files 1 and 2. The software datasets were then pooled and a hierarchical clustering with Cluster and Treeview were used to classify cell subsets complete linkage was performed. A similar analysis was per- according to the proximity of their gene expression pattern as formed for the comparison of human and mouse LN-DCs, assessed by hierarchical clustering with complete linkage. monocytes, macrophages and in vitro derived GM-CSF DCs.

We implemented a function in the Matlab software to per- Meta-analysis of mouse 430 2.0 and U74Av2 datasets form PCA. This function computes the eigenvalues and eigen- In order to classify the IKDCs based on the optimal gene sig- vectors of the dataset using the correlation matrix. The natures of the different cell subsets examined, with only min- eigenvalues were then ordered from highest to lowest, indi- imal impact of differences in the experimental protocols used cating their relative contribution to the structure of the data. to prepare the cells and to perform the gene chips assays, the For both mouse and human datasets, the first principal com- clustering of the cell populations was performed as a meta- ponent accounted for most of the information (54% and 68% analysis of our own mouse 430 2.0 dataset together with the for mouse and human, respectively) and was associated with published U74Av2 datasets. The Array Comparison support a similar coordinate for all samples. This component thus information of the NetAffyx™ analysis center (Affymetrix) reflected the common gene expression among the samples. was used to identify matched ProbSets between the two types Second and third components together represented 24% and of microarrays. Only one representative dataset was kept for 21%, respectively, of the information for mouse and human redundant ProbeSets targeting the same gene. This yielded a datasets, and thus accounted for a large part of the variability. set of 2,251 genes whose expression could be compared The projection of each sample on the planes defined by these between the two datasets, using the same normalization components was represented as a dot plot to generate the method as described above. This dataset is accessible as Excel PCA figures. workbooks in Additional data file 4. As expected, this meta- analysis led to co-clustering of all the samples derived from Partitional clustering was performed using the FCM algo- identical cell types whether their gene expression had been rithm, which links each gene to all clusters via a vector of

ClusteringFigure 5 (see of in following vitro GM-CSF page) derived DCs with monocytes, macrophages and LN-resident DCs Clustering of in vitro GM-CSF derived DCs with monocytes, macrophages and LN-resident DCs. Hierarchical clustering with complete linkage was performed on the indicated cell populations isolated from: (a) mouse, (b) human, and (c) both. The heat maps used for illustration were selected as the two clusters of genes encompassing either Flt3 or Mafb , with a correlation cut-off for similarity of gene expression within each cluster at 0.8.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.23

(a) (b) Φ Φ Φ Φ Φ M pDCs mono CD16 pDC mo-DC mono mono mo-M BM-DC BM-DC BM-M BM-M CD8 DC PBMC-M BDCA1 DC BDCA3 DC CD11b DC - - - 0610010K14Rik ZNF532 - - - MAGEF1 Fkbp3 - - - LOC677143 CIITA Surf6 NFATC2 Slc4a1ap - - - Ermp1 ITGB7 Apbb3 RASGEF1A - - - Setd6 CCNB1IP1 Wdr92 FLT3 Rab11fip2 SH3BP4 Mmaa TMEM120A Rnf215 RRAS - - - LRRC25 Dffb PAPSS2 Armcx2 MAFB Crtc2 MTSS1 Hspa8 ------CHST7 Per1 FCGR3A Zdhhc23 PIK3IP1 Tmem170 GPR109B E230012J19Rik PYGB Flt3 Color scale Zfp566 -3 0 +3 Nsmaf ------Tmem161b Ddx10 Ofd1 A030009H04Rik mo-DCs = human monocyte-derived GM-CSF DCs - - - mo-M Φ = human monocyte-derived M-CSF macrophages 6330509M05Rik Φ - - - PBMC-M = human peripheral blood mononuclear cell-derived M-CSF macrophages Rp137 LOC100042986 Gp49a Plk2 mono = human blood monocytes or mouse spleen monocytes Mitf Clec4d Mast1 Soat1k BM-M Φ = mouse bone marrow-derived M-CSF macrophages Mafb MΦ = mouse peritoneal macrophages Tlk1 BM-DC = mouse bone marrow-derived GM-CSF DCs Tmem65 Color scale -3 0 +3 (c) Φ Φ Φ Φ Φ h. pDC m. CD8 m. pDC m. h. mono h. CD16 h. mo-DC h. mo-M h. BDCA1 h. BDCA3 m. CD11b m. BM-DC m. BM-DC m. m. BM-M m. m. BM-M m. m. mono(2) m. mono(1) m. perit.-M h. PBMC-M

m Bxdc1; h BXDC1 m Nsmaf; h NSMAF m Flt3; h FLT3 m Itgb7; h ITGB7 m Ptk2; h PTK2 m Cd14; h CD14 m Mpp1; h MPP1 m Sdcbp; h SDCBP m Mafb; h MAFB m Dok3; h DOK3 m Tlr4; h TLR4

Figure 5 (see legend on previous page)

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.24

measured by us on 430 2.0 microarrays or by others on Authors' contributions U74Av2 microarrays, with the exception of the cDC popula- SHR, TW, SC, PK, and MD designed the research; SHR, TW, tion from [15], which segregated with pDCs rather than with CT, HX, MS, GB, AD and MD performed the research; EV and the cDC subsets from the other datasets. PP contributed new reagents/analytical tools; SHR, TW, CT, HX, DD, MS, FRS, SC, PK, and MD analyzed data; and SHR, Data mining TW, and MD wrote the paper. Gene lists were analyzed using the DAVID 'functional annota- tion chart' tool accessible on the NIAID website [52,53]. Dif- ferent databases were used for these annotations: gene Note added in proof ontology (Amigo), knowledge pathways (KEGG), interactions During the review process of this paper, two reports were (BIND), interprotein domains (INTERPRO), and disease published in Nature Immunology that identified a common (OMIM/OMIA). The annotations shown in Tables 5 and 7 progenitor characterized as FLT3 +M-CSF + for mouse LN-DCs were selected as the most highly significant terms retrieved by (pDCs, CD8 ! cDCs and CD11b cDCs), devoid of any capability performing an over-representation study. To this end, a to generate lymphoid cells or monocytes/macrophages, and modified Fisher exact P value called the 'EASE score' was cal- named common dendritic progenitor (CDP) [87,88]. This culated to measure the enrichment in gene-annotation terms observation is thus consistent with our gene profiling analysis between the gene signature specific to the leukocyte of human and mouse leukocytes. The question whether this subpopulation examined ('List') and the complete set of all pathway for LN-DCs is the major one, or just one possibility the genes selected for the compendium analyzed ('Back- among others, including differentiation from monocytes, has ground'). The significance threshold was set at an EASE score been raised [89]. Our gene profiling data would suggest that below 0.05 in most instances, or below 0.1 for DC signatures most mouse LN-DCs derive from the recently identified CDP that did not yield many highly significant terms as discussed or MDP in vivo , without a monocytic intermediate, consistent in Results. Individual significant annotations encompassing with a recent report [81]. It also implies that a similar path- many common genes or similar biological processes were way must exist in humans. The relationship between the CDP regrouped using the 'Functional annotation clustering' tool of and the MDP still remains to be established. Three reports the DAVID software. More information on this type of analy- have been published very recently in the Journal of Experi- sis is available on the DAVID website [85]. mental Medicine that showed that IKDCs are a specific subset of NK cells, based on functional and ontogenic approaches Public access to the raw data for the datasets analyzed comparing these cells to DCs and NK cells [90-92]. This is in the paper consistent with the results of our clustering analysis of IKDCs Our datasets for mouse DC subsets, NK cells, CD8 T cells, and with other leukocyte subsets. Finally, two recent reports have B lymphocytes have been deposited in the Gene Expression identified a new transduction pathway in human pDCs Omnibus (GEO) database under reference number GSE9810. involving a B cell receptor-like ITAM-signaling pathway The references for download of the public data used from the [93,94]. This pathway involves the BLNK transduction original websites where they were first made available are molecule, which we have identified here as expressed to very given in Table 1. In addition, all raw transcriptomic data ana- high levels in mouse and human pDCs compared to the other lyzed here have been regrouped on our website [86] and are LN-DCs (Table 6) and many other leukocytes. We believe that available for public download. the conserved transcriptional signatures identified here for mouse and human LN-DC subsets will lead to many more dis- coveries for the understanding of the specialized functions of Abbreviations these cells. APC, antigen-presenting cell; BM, bone marrow; cDC, con- ventional dendritic cell; CDP, common dendritic progenitor; DC, dendritic cell; FCM, fuzzy c-means; GEO, Gene Expres- Additional data files sion Omnibus; GM-CSF, granulocyte-macrophage colony The following additional data are available. Additional data stimulating factor; IFN, interferon; IKDC, interferon-produc- file 1 is a Microsoft Excel workbook with raw data for the ing killer dendritic cell; ITAM, immunoreceptor tyrosine- mouse gene chip compendium. Additional data file 2 is a based activation motif; LN-DC, lymph node-resident DC; M- Microsoft Excel workbook with raw data for the human gene CSF, macrophage colony-stimulating factor; MDP, macro- chip compendium. Additional data file 3 is a Microsoft Excel phage and dendritic cell progenitor; MHC, major histocom- workbook with raw data for the human/mouse gene chip patibility; NK, natural killer; PCA, principal component compendium. Additional data file 4 is a Microsoft Excel work- analysis; pDC, plasmacytoid dendritic cell; TLR, toll-like book with raw data for the IKDC gene chip compendium. receptor. Additional data file 5 is a Microsoft Excel workbook giving the mouse DC subset gene signatures according to our datasets with confirmation from two other independent datasets (one for pDCs and one for cDC subsets). Additional data file 6 is a

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.25

figure showing the results of PCA for investigation of the rela- lating dendritic cell cytokine expression in vivo. J Exp Med tionships between in vitro derived GM-CSF DCs and LN-DCs 2002, 195: 517-528. 15. Chan CW, Crafton E, Fan HN, Flook J, Yoshimura K, Skarica M, in mouse and human. Additional data file 7 is a table giving Brockstedt D, Dubensky TW, Stins MF, Lanier LL, Pardoll DM, Hous- real-time PCR data for the pattern of expression of 27 genes seau F: Interferon-producing killer dendritic cells provide a link between innate and adaptive immunity. Nat Med 2006, across mouse leukocyte subsets. Additional data file 8 is a fig- 12: 207-213. ure illustrating PACSIN1 expression in human pDCs versus 16. Taieb J, Chaput N, Menard C, Apetoh L, Ullrich E, Bonmort M, PBMCs by RT-PCR and western blotting. Pequignot M, Casares N, Terme M, Flament C, Opolon P, Lecluse Y, Metivier D, Tomasello E, Vivier E, Ghiringhelli F, Martin F, Klatzmann andClickvitroAdditionalReal-timemousePACSIN1ResultsRawMouseconfirmation onewestern data derivedhere leukocyteDC of for expressionfor PCR PCAforsubsetfile cDC blotting. the GM-CSFfromfile 78234516 datafor subsets).humanIKDChuman/mousemousesubsetssubsets. gene investigationtwo for in DCs signaturesgenetheotherhuman genegene pattern and chip independent chipchip pDCs LN-DCsofgene compendiumcompendium. according compendiumcompendium.ofcompendiumcompendium.the expressionversuschip relationshi in datasets compendium.compendiummouse PBMCsto our ofps and 27da(o betweenby netasetsgenes human.human RT for-PCR acrosspDCswith in D, Poynard T, Tursz T, Raposo G, Yagita H, Ryffel B, Kroemer G, Zitvogel L: A novel dendritic cell subset involved in tumor immunosurveillance. Nat Med 2006, 12: 214-219. Acknowledgements 17. Hyatt G, Melamed R, Park R, Seguritan R, Laplace C, Poirot L, Zuc- The authors are indebted to Bertrand Nadel and Jean-Marc Navarro for chelli S, Obst R, Matos M, Venanzi E, Goldrath A, Nguyen L, Luckey help with the real-time PCR experiments and to Markus Plomann for the J, Yamagata T, Herman A, Jacobs J, Mathis D, Benoist C: Gene generous gift of the anti-PACSIN1 antibody. The authors also thank the expression microarrays: glimpses of the immunological staff of the animal care facilities and the flow cytometry core facility of the genome. Nat Immunol 2006, 7: 686-691. CIML for excellent assistance. This work was supported by an ATIP grant 18. Yamagata T, Benoist C, Mathis D: A shared gene-expression sig- from the CNRS, a grant from the Association pour la Recherche sur le Can- nature in innate-like lymphocytes. Immunol Rev 2006, cer (ARC) and a grant from the Réseau National des Génopoles (RNG) to 210: 52-66. MD. SHR was supported by the CNRS, the Fondation pour la Recherche 19. Benoist C, Germain RN, Mathis D: A plaidoyer for 'systems Médicale, and the Philippe Foundation. The CIML is supported by institu- immunology'. Immunol Rev 2006, 210: 229-234. tional grants from the INSERM, the CNRS, and the Université de la 20. Shaffer AL, Rosenwald A, Hurt EM, Giltnane JM, Lam LT, Pickeral OK, Méditerranée. We thank the IPSOGEN company for their advice on the Staudt LM: Signatures of the immune response. Immunity 2001, analysis of the data. The authors declare no conflict of interest. 15: 375-385. 21. Hutton JJ, Jegga AG, Kong S, Gupta A, Eber t C, Williams S, Katz JD, Aronow BJ: Microarray and comparative genomics-based identification of genes and gene regulatory regions of the References mouse immune system. BMC Genomics 2004, 5: 82. 1. Mellman I, Steinman RM: Dendritic cells: specialized and 22. Ellwood-Yen K, Graeber TG, Wongvipat J, Iruela-Arispe ML, Zhang J, regulated antigen processing machines. Cell 2001, Matusik R, Thomas GV, Sawyers CL: Myc-driven murine prostate 106: 255-258. cancer shares molecular features with human prostate 2. Iwasaki A, Medzhitov R: Toll-like receptor control of the adap- tumors. Cancer Cell 2003, 4: 223-238. tive immune responses. Nat Immunol 2004, 5: 987-995. 23. Zeller KI, Jegga AG, Aronow BJ, O'Donnell KA, Dang CV: An inte- 3. Liu YJ: IPC: professional type 1 interferon-producing cells and grated database of genes responsive to the Myc oncogenic plasmacytoid dendritic cell precursors. Annu Rev Immunol 2005, transcription factor: identification of direct genomic targets. 23: 275-306. Genome Biol 2003, 4: R69. 4. Serbina NV, Salazar-Mather TP, Biron CA, Kuziel WA, Pamer EG: 24. Reichling T, Goss KH, Carson DJ, Holdcraft RW, Ley-Ebert C, Witte TNF/iNOS-Producing Dendritic Cells Mediate Innate D, Aronow BJ, Groden J: Transcriptional profiles of intestinal Immune Defense against Bacterial Infection. Immunity 2003, tumors in Apc(Min) mice are unique from those of embry- 19: 59-70. onic intestine and identify novel gene targets dysregulated in 5. Walzer T, Dalod M, Robbins SH, Zitvogel L, Vivier E: Natural-killer human colorectal tumors. Cancer Res 2005, 65: 166-176. cells and dendritic cells: "l'union fait la force". Blood 2005, 25. Kaiser S, Park YK, Franklin JL, Halberg RB, Yu M, Jessen WJ, Freuden- 106: 2252-2258. berg J, Chen X, Haigis K, Jegga AG, Kong S, Sakthivel B, Xu H, Reich- 6. Shortman K, Liu YJ: Mouse and human dendritic cell subtypes. ling T, Azhar M, Boivin GP, Roberts RB, Bissahoyo A, Gonzales F, Nature Rev Immunol 2002, 2: 151-161. Bloom GL, Eschrich S, Carter SL, Aronow JE, Kleimeyer J, Kleimeyer 7. Narbutt J, Lesiak A, Zak-Prelich M, Wozniacka A, Sysa-Jedrzejowska M, Ramaswamy V, Settle SH, Boone B, Levy S, Graff JM, et al. : Tran- A, Tybura M, Robak T, Smolewski P: The distribution of periph- scriptional recapitulation and subversion of embryonic colon eral blood dendritic cells assayed by a new panel of anti- development by mouse colon tumor models and human BDCA monoclonal antibodies in healthy representatives of colon cancer. Genome Biol 2007, 8: R131. the Polish population. Cell Mol Biol Lett 2004, 9: 497-509. 26. Graeber TG, Sawyers CL: Cross-species comparisons of cancer 8. Heath WR, Belz GT, Behrens GM, Smith CM, Forehan SP, Parish IA, signaling. Nat Genet 2005, 37: 7-8. Davey GM, Wilson NS, Carbone FR, Villadangos JA: Cross-presen- 27. Lee JS, Chu IS, Mikaelyan A, Calvisi DF, Heo J, Reddy JK, Thorgeirsson tation, dendritic cell subsets, and the generation of immu- SS: Application of comparative functional genomics to iden- nity to cellular antigens. Immunol Rev 2004, 199: 9-26. tify best-fit mouse models to study human cancer. Nat Genet 9. Kronin V, Fitzmaurice CJ, Caminschi I, Shortman K, Jackson DC, 2004, 36: 1306-1311. Brown LE: Differential effect of CD8(+) and CD8(-) dendritic 28. Fontenot JD, Rasmussen JP, Williams LM, Dooley JL, Farr AG, Ruden- cells in the stimulation of secondary CD4(+) T cells. Int sky AY: Regulatory T cell lineage specification by the fork- Immunol 2001, 13: 465-473. head transcription factor foxp3. Immunity 2005, 22: 329-341. 10. Pooley JL, Heath WR, Shortman K: Cutting edge: intravenous 29. Chelvarajan RL, Liu Y, Popa D, Getchell ML, Getchell TV, Stromberg soluble antigen is presented to CD4 T cells by CD8- dendritic AJ, Bondada S: Molecular basis of age-associated cytokine dys- cells, but cross-presented to CD8 T cells by CD8+ dendritic regulation in LPS-stimulated macrophages. J Leukoc Biol 2006, cells. J Immunol 2001, 166: 5327-5330. 79: 1314-1327. 11. Dudziak D, Kamphorst AO, Heidkamp GF, Buchholz VR, Trumpfhel- 30. Du X, Tang Y, Xu H, Lit L, Walker W, Ashwood P, Gregg JP, Sharp ler C, Yamazaki S, Cheong C, Liu K, Lee HW, Park CG, Steinman RM, FR: Genomic profiles for human peripheral blood T cells, B Nussenzweig MC: Differential antigen processing by dendritic cells, natural killer cells, monocytes, and polymorphonuclear cell subsets in vivo. Science 2007, 315: 107-111. cells: comparisons to ischemic stroke, migraine, and 12. Shigematsu H, Reizis B, Iwasaki H, Mizuno S, Hu D, Traver D, Leder Tourette syndrome. Genomics 2006, 87: 693-703. P, Sakaguchi N , Akashi K: Plasmacytoid dendritic cells activate 31. Lindstedt M, Lundberg K, Borrebaeck CA: Gene family clustering lymphoid-specific genetic programs irrespective of their cel- identifies functionally associated subsets of human in vivo lular origin. Immunity 2004, 21: 43-53. blood and tonsillar dendritic cells. J Immunol 2005, 13. Ito T, Kanzler H, Duramad O, Cao W, Liu YJ: Specialization, kinet- 175: 4839-4846. ics, and repertoire of type 1 interferon responses by human 32. Eisen MB, Spellman PT, Brown PO, Botstein D: Cluster analysis plasmacytoid predendritic cells. Blood 2006, 107: 2423-2431. and display of genome-wide expression patterns. Proc Natl 14. Dalod M, Salazar-Mather TP, Malmgaard L, Lewis C, Asselin-Paturel Acad Sci USA 1998, 95: 14863-14868. C, Briere F, Trinchieri G, Biron CA: Interferon alpha/beta and 33. Alter O, Brown PO, Botstein D: Singular value decomposition interleukin 12 responses to viral infections: pathways regu- for genome-wide expression data processing and modeling.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.26

Proc Natl Acad Sci USA 2000, 97: 10101-10106. Genome Biol 2003, 4: R70. 34. Dembélé D, Kastner P: Fuzzy C-means method for clustering 54. Edwards AD, Diebold SS, Slack EM, Tomizawa H, Hemmi H, Kaisho microarray data. Bioinformatics 2003, 19: 973-980. T, Akira S, Reis e Sousa C: Toll-like receptor expression in 35. Nutt SL, Eberhard D, Horcher M, Rolink AG, Busslinger M: Pax5 murine DC subsets: lack of TLR7 expression by CD8 alpha+ determines the identity of B cells from the beginning to the DC correlates with unresponsiveness to imidazoquinolines. end of B-lymphopoiesis. Int Rev Immunol 2001, 20: 65-82. Eur J Immunol 2003, 33: 827-833. 36. Malissen B, Aguado E, Malissen M: Role of the LAT adaptor in T- 55. Rissoan MC, Duhen T, Bridon JM, Bendriss-Vermare N, Peronne C, cell development and Th2 differentiation. Adv Immunol 2005, de Saint Vis B, Briere F, Bates EE: Subtractive hybridization 87: 1-25. reveals the expression of immunoglobulin-like transcript 7, 37. Walzer T, Jaeger S, Chaix J, Vivier E: Natural killer cells: from Eph-B1, granzyme B, and 3 novel transcripts in human plas- CD3(-)NKp46(+) to post-genomics meta-analyses. Curr Opin macytoid dendritic cells. Blood 2002, 100: 3295-3303. Immunol 2007, 19: 365-372. 56. Edwards AD, Chaussabel D, Tomlinson S, Schulz O, Sher A, Reis e 38. Townsend MJ, Weinmann AS, Matsuda JL, Salomon R, Farnham PJ, Sousa C: Relationships among murine CD11c(high) dendritic Biron CA, Gapin L, Glimcher LH: T-bet regulates the terminal cell subsets as revealed by baseline gene expression maturation and homeostasis of NK and Valpha14i NKT cells. patterns. J Immunol 2003, 171: 47-60. Immunity 2004, 20: 477-494. 57. Galibert L, Diemer GS, Liu Z, Johnson RS, Smith JL, Walzer T, 39. Matsuki Y, Ohmura-Hoshino M, Goto E, Aoki M, Mito-Yoshida M, Comeau MR, Rauch CT, Wolfson MF, Sorensen RA, Van der Vuurst Uematsu M, Hasegawa T, Koseki H, Ohara O, Nakayama M, Toyooka de Vries AR, Branstetter DG, Koelling RM, Scholler J, Fanslow WC, K, Matsuoka K, Hotta H, Yamamoto A, Ishido S: Novel regulation Baum PR, Derry JM, Yan W: Nectin-like protein 2 defines a sub- of MHC class II function in B cells. Embo J 2007, 26: 846-854. set of T-cell zone dendritic cells and is a ligand for class-I- 40. Tabeta K, Hoebe K, Janssen EM, Du X, Georgel P, Crozat K, Mudd S, restricted T-cell-associated molecule. J Biol Chem 2005, Mann N, Sovath S, Goode J, Shamel L, Herskovits AA, Portnoy DA, 280: 21955-21964. Cooke M, Tarantino LM, Wiltshire T, Steinberg BE, Grinstein S, Beut- 58. Honda K, Yanai H, Negishi H, Asagiri M, Sato M, Mizutani T, Shimada ler B: The Unc93b1 mutation 3d disrupts exogenous antigen N, Ohba Y, Takaoka A, Yoshida N, Taniguchi T: IRF-7 is the master presentation and signaling via Toll-like receptors 3, 7 and 9. regulator of type-I interferon-dependent immune responses. Nat Immunol 2006, 7: 156-164. Nature 2005, 434: 772-777. 41. Casrouge A, Zhang SY, Eidenschenk C, Jouanguy E, Puel A, Yang K, 59. Diebold SS, Kaisho T, Hemmi H, Akira S, Reis e Sousa C: Innate Alcais A, Picard C, Mahfoufi N, Nicolas N, Lorenzo L, Plancoulaine S, antiviral responses by means of TLR7-mediated recognition Senechal B, Geissmann F, Tabeta K, Hoebe K, Du X, Miller RL, Heron of single-stranded RNA. Science 2004, 303: 1529-1531. B, Mignot C, de Villemeur TB, Lebon P, Dulac O, Rozenberg F, Beut- 60. Lund JM, Alexopoulou L, Sato A, Karow M, Adams NC, Gale NW, ler B, Tardieu M, Abel L, Casanova JL: Herpes simplex virus Iwasaki A, Flavell RA: Recognition of single-stranded RNA encephalitis in human UNC-93B deficiency. Science 2006, viruses by Toll-like receptor 7. Proc Natl Acad Sci USA 2004, 314: 308-312. 101: 5598-5603. 42. Walzer T, Chiossone L, Chaix J, Calver A, Carozzo C, Garrigue-Antar 61. Trombetta ES, Mellman I: Cell biology of antigen processing in L, Jacques Y, Baratin M, Tomasello E, Vivier E: Natural killer cell vitro and in vivo. Annu Rev Immunol 2005, 23: 97 5-1028. trafficking in vivo requires a dedicated sphingosine 1-phos- 62. Honda K, Ohba Y, Yanai H, Negishi H, Mizutani T, Takaoka A, Taya phate receptor. Nat Immunol 2007, 8: 1337-1344. C, Taniguchi T: Spatiotemporal regulation of MyD88-IRF-7 sig- 43. McKenna HJ, Stocking KL, Miller RE, Brasel K, De Smedt T, Marask- nalling for robust type-I interferon induction. Nature 2005, ovsky E, Maliszewski CR, Lynch DH, Smith J, Pulendran B, Roux ER, 434: 1035-1040. Teepe M, Lyman SD, Peschon JJ: Mice lacking flt3 ligand have 63. Guiducci C, Ott G, Chan JH, Damon E, Calacsan C, Matray T, Lee KD, deficient hematopoiesis affecting hematopoietic progenitor Coffman RL, Barrat FJ: Properties regulating the nature of the cells, dendritic cells, and natural killer cells. Blood 2000, plasmacytoid dendritic cell response to Toll-like receptor 9 95: 3489-3497. activation. J Exp Med 2006, 203: 1999-2008. 44. Onai N, Obata-Onai A, Tussiwand R, Lanzavecchia A, Manz MG: 64. Banchereau J, Pascual V: Type I interferon in systemic lupus ery- Activation of the Flt3 signal transduction cascade rescues thematosus and other autoimmune diseases. Immunity 2006, and enhances type I interferon-producing and dendritic cell 25: 383-392. development. J Exp Med 2006, 203: 227-238. 65. Schulz O, Diebold SS, Chen M, Naslund TI, Nolte MA, Alexopoulou 45. Tussiwand R, Onai N, Mazzucchelli L, Manz MG: Inhibition of nat- L, Azuma YT, Flavell RA, Liljestrom P, Reis e Sousa C: Toll-like ural type I IFN-producing and dendritic cell development by receptor 3 promotes cross-priming to virus-infected cells. a small molecule receptor tyrosine kinase inhibitor with Flt3 Nature 2005, 433: 887-892. affinity. J Immunol 2005, 175: 3674-3680. 66. Orabona C, Puccetti P, Vacca C, Bicciato S, Luchini A, Fallarino F, 46. Maraskovsky E, Daro E, Roux E, Teepe M, Maliszewski CR, Hoek J, Bianchi R, Velardi E, Perruccio K, Velardi A, Bronte V, Fioretti MC, Caron D, Lebsack ME, McKenna HJ: In vivo generation of human Grohmann U: Toward the identification of a tolerogenic signa- dendritic cell subsets by Flt3 ligand. Blood 2000, 96: 878-884. ture in IDO-competent dendritic cells. Blood 2006, 47. Bonasio R, von Andrian UH: Generation, migration and function 107: 2846-2854. of circulating dendritic cells. Curr Opin Immunol 2006, 67. Spits H, Lanier LL: Natural Killer or Dendritic: What's in a 18: 503-511. Name? Immunity 2007, 26: 11-16. 48. LeibundGut-Landmann S, Waldburger JM, Reis e Sousa C, Acha- 68. Rolink A, ten Boekel E, Melchers F, Fearon DT, Krop I, Andersson J: Orbea H, Reith W: MHC class II expression is differentially reg- A subpopulation of B220+ cells in murine bone marrow does ulated in plasmacytoid and conventional dendritic cells. Nat not express CD19 and contains natural killer cell Immunol 2004, 5: 899-908. progenitors. J Exp Med 1996, 183: 187-194. 49. Chtanova T, Newton R, Liu SM, Weininger L, Young TR, Silva DG, 69. Kamath AT, Sheasby CE, Tough DF: Dendritic cells and NK cells Bertoni F, Rinaldi A, Chappaz S, Sallusto F, Rolph MS, Mackay CR: stimulate bystander T cell activation in response to TLR ago- Identification of T cell-restricted genes, and signatures for nists through secretion of IFN-alpha beta and IFN-gamma. J different T cell responses, using a comprehensive collection Immunol 2005, 174: 767-776. of microarray datasets. J Immunol 2005, 175: 7837-7847. 70. Laouar Y, Sutterwala FS, Gorelik L, Flavell RA: Transforming 50. Krücken J, Schroetel RM, Müller IU, Saïdani N, Marinovski P, Benten growth factor-beta controls T helper type 1 cell develop- WP, Stamm O, Wunderlich F: Comparative analysis of the ment through regulation of natural killer cell interferon- human gimap gene cluster encoding a novel GTPase family. gamma. Nat Immunol 2005, 6: 600-607. Gene 2004, 341: 291-304. 71. BD/Pharmingen technical datasheet of the CD11c antibody 51. Nitta T, Nasreen M, Seike T, Goji A, Ohigashi I, Miyazaki T, Ohta T, [http://www.bdbiosciences.com/external_files/pm/doc/tds/mouse/ Kanno M, Takahama Y: IAN family critically regulates survival live/web_enabled/09709A_550261.pdf] and development of T lymphocytes. PLoS Biol 2006, 4: e103. 72. Veinotte LL, Greenwood CP, Mohammadi N, Parachoniak CA, Takei 52. Dennis G Jr, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, Lem- F: Expression of rearranged TCRgamma genes in natural picki RA: DAVID: Database for Annotation, Visualization, and killer cells suggests a minor thymus-dependent pathway of Integrated Discovery. Genome Biol 2003, 4: P3. lineage commitment. Blood 2006, 107: 2673-2679. 53. Hosack DA, Dennis G Jr, Sherman BT, Lane HC, Lempicki RA: Iden- 73. Gordon S, Taylor PR: Monocyte and macrophage tifying biological themes within lists of genes with EASE. heterogeneity. Nat Rev Immunol 2005, 5: 953-964.

Genome Biology 2008, 9: R17 http://genomebiology.com/2008/9/1/R17 Genome Biology 2008, Volume 9, Issue 1, Article R17 Robbins et al. R17.27

74. Bakri Y, Sarrazin S, Mayer UP, Tillmanns S, Nerlov C, Boned A, a novel BCR-like pathway in human plasmacytoid dendritic Sieweke MH: Balance of MafB and PU.1 specifies alternative cells. PLoS Biol 2007, 5: e248. macrophage or dendritic cell fate. Blood 2005, 105: 2707-2716. 95. Gene Expression Omnibus (GEO) Main page [http:// 75. Xu Y, Zhan Y, Lew AM, Naik SH, Kershaw MH: Differential Devel- www.ncbi.nlm.nih.gov/projects/geo/] opment of Murine Dendritic Cells by GM-CSF versus Flt3 96. National Cancer Institute caArray Ligand Has Implications for Inflammation and Trafficking. J [https:caarydb.nci.nih.gocaarrapublicSearch.do;jses=8933FE3445EEB Immunol 2007, 179: 7577-7584. E52CD30?de=experi ments] 76. Naik SH, Proietto AI, Wilson NS, Dakic A, Schnorrer P, Fuchsberger 97. ArrayExpress [http://www.ebi.ac.uk/microarray-as/aer/?#ae- M, Lahoud MH, O'Keeffe M, Shao QX, Chen WF, Villadangos JA, main[0]] Shortman K, Wu L: Cutting edge: generation of splenic CD8+ and CD8- dendritic cell equivalents in Fms-like tyrosine kinase 3 ligand bone marrow cultures. J Immunol 2005, 174: 6592-6597. 77. Lahoud MH, Proietto AI, Gartlan KH, Kitsoulis S, Curtis J, Wettenhall J, Sofi M, Daunt C, O'Keeffe M, Caminschi I, Satterley K, Rizzitelli A, Schnorrer P, Hinohara A, Yamaguchi Y, Wu L, Smyth G, Handman E, Shortman K, Wright MD: Signal regulatory protein molecules are differentially expressed by CD8- dendritic cells. J Immunol 2006, 177: 372-382. 78. Zenke M, Hieronymus T: Towards an understanding of the tran- scription factor network of dendritic cell development. Trends Immunol 2006, 27: 140-145. 79. Blom B, Spits H: Development of human lymphoid cells. Annu Rev Immunol 2006, 24: 287-320. 80. Fogg DK, Sibon C, Miled C, Jung S, Aucouturier P, Littman DR, Cumano A, Geissmann F: A clonogenic bone marrow progeni- tor specific for macrophages and dendritic cells. Science 2006, 311: 83-87. 81. Varol C, Landsman L, Fogg DK, Greenshtein L, Gildor B, Margalit R, Kalchenko V, Geissmann F, Jung S: Monocytes give rise to mucosal, but not splenic, conventional dendritic cells. J Exp Med 2007, 204: 171-180. 82. Rathinam C, Sauer M, Ghosh A, Rudolph C, Hegazy A, Schlegelberger B, Welte K, Klein C: Generation and characterization of a novel hematopoietic progenitor cell line with DC differenti- ation potential. Le ukemia 2006, 20: 870-876. 83. Cao W, Rosen DB, Ito T, Bover L, Bao M, Watanabe G, Yao Z, Zhang L, Lanier LL, Liu YJ: Plasmacytoid dendritic cell-specific recep- tor ILT7-Fc epsilonRI gamma inhibits Toll-like receptor- induced interferon production. J Exp Med 2006, 203: 1399-1405. 84. Blasius AL, Cella M, Maldonado J, Takai T, Colonna M: Siglec-H is an IPC-specific receptor that modulates type I IFN secretion through DAP12. Blood 2006, 107: 2474-2476. 85. DAVID Bioinformatics Resources National Institute of Allergy and Infectious Diseases (NIAID), NIH [http:// david.abcc.ncifcrf.gov/home.jsp] 86. Microarray platform of the IGBMC and Génopole Alsace- Lorraine [http://www-microarrays.u-strasbg.fr/files/datasetsE.php] 87. Onai N, Obata-Onai A, Schmid MA, Ohteki T, Jarrossay D, Manz MG: Identification of clonogenic common Flt3 +M-CSFR + plasmacytoid and conventional dendritic cell progenitors in mouse bone marrow. Nat Immunol 2007, 8: 1207-1216. 88. Naik SH, Sathe P, Park HY, Metcalf D, Proietto AI, Dakic A, Carotta S, O'Keeffe M, Bahlo M, Papenfuss A, Kwak JY, Wu L, Shortman K: Development of plasmacytoid and conventional dendritic cell subtypes from single precursor cells derived in vitro and in vivo. Nat Immunol 2007, 8: 1217-1226. 89. Merad M, Ginhoux F: Dendritic cell genealogy: a new stem or just another branch? Nat Immunol 2007, 8: 1199-1201. 90. Blasius AL, Barchet W, Cella M, Colonna M: Development and function of murine B220+CD11c+NK1.1+ cells identify them as a subset of NK cells. J Exp Med 2007, 204: 2561-2568. 91. Caminschi I, Ahmet F, Heger K, Brady J, Nutt SL, Vremec D, Pietersz S, Lahoud MH, Schofield L, Hansen DS, O'Keeffe M, Smyth MJ, Bedoui S, Davey GM, Villadangos JA, Heath WR, Shortman K: Putative IKDCs are functionally and developmentally similar to natural killer cells, but not to dendritic cells. J Exp Med 2007, 204: 2579-2590. 92. Vosshenrich CA, Lesjean-Pottier S, Hasan M, Richard-Le Goff O, Corcuff E, Mandelboim O, Di Santo JP: CD11cloB220+ interferon- producing killer dendritic cells are activated natural killer cells. J Exp Med 2007, 204: 2569-2578. 93. Röck J, Schneider E, Grün JR, Grützkau A, Küppers R, Schmitz J, Win- kels G: CD303 (BDCA-2) signals in plasmacytoid dendritic cells via a BCR-like signalosome involving Syk, Slp65 and PLCgamma2. Eur J Immunol 2007, 37: 3564-3575. 94. Cao W, Zhang L, Rosen DB, Bover L, Watanabe G, Bao M, Lanier LL, Liu YJ: BDCA2/Fc epsilon RI gamma complex signals through

Genome Biology 2008, 9: R17 a c

b

d e

Fig.1 – Représentation schématique de la voie de synthèse Les protéines synthétisées dans le réticulum endoplasmique (RE) sont transportées par voie vésiculaire vers les compartiment intermédiaires ER Golgi (ERGIC) (a). Les vésicules formées en sortie des ERGIC fusionnent ensuite en formant les saccules du cis -Golgi (b). Les saccules subissent une maturation progressive qui se termine dans le réseau trans -Golgi (c). En parallèle, une voie rétrograde permet de rapatrier les protéines cargo résidentes du RE à partir du cis-Golgi (d). La voie rétrograde est également nécessaire pour le rapatriement des enzymes golgiennes au fur et à mesure de la maturation des saccules (e). Les protéines de manteau des complexes COPI et COPII sont essentielles pour assurer la sélectivité du transport vésiculaire. Les complexes COPII sont impliqués dans le transport antérograde entre le RE et les ERGIC. Les complexes COPI participent au transport antérograde post-ERGIC et dans le transport rétrograde. Fig.2 – Processus de maturation des saccules de l’appareil de Golgi Les vésicules contenant les protéines néo-synthétisées fusionnent pour former les saccules cis de l’appareil de Golgi (a). Au cours de la maturation des saccules, un transport rétrograde permet d’exposer les protéines à différentes enzymes. Les enzymes du cis -Golgi sont rapatriées vers les nouveaux saccules formés en début de maturation (b). Le recyclage des enzymes se poursuit de manière sélective jusqu’à la face trans -Golgi (c). Fig.3 – Exemples de glycosylation-N et -O Les oligosaccharides associé à la glycosylation-N partage une structure de base commune avec 2 N-acétylamineglucose (GlcNAc) et 3 mannoses (Man). Les oligosaccharides sont qualifiées d’hybride lorsqu’elles portent des Man en position terminale. Le nombre de ramifications possibles est immense. La glycosylation-O est plus hétérogène. Les chaînes sont classées en fonction des sucres attachés avant la ramification et la diversification de l’oligosaccharide. Les deux chaîne présentées ici sont du type mucine. RE b

c a d

cis -Golgi e

Golgi intermédiaire

f g trans -Golgi

i h

Fig.4 – Représentation schématique de la maturation des glycosylations-N Une chaîne pré-assemblée commune est fixée sur les protéines en cours d’élongation (a). Les deux Glc terminaux sont coupés séquentiellement (b), permettant à la glycoprotéine d’entrer dans la boucle de contrôle du repliement par la calnexine/calréticuline (c). La glycoprotéine est transférée vers le ci-Golgi après une dernière étape de clivage (d). Plusieurs Man sont clivés avant la progression vers le Golgi intermédiaire (e). La chaîne est ensuite successivement ramifiée (f), puis clivée au niveau des Man restants (g). La ramification se poursuit dans le trans -Golgi (h) jusqu’à obtenir une chaîne complexe composée de plusieurs sucres différents (i). La taille et la nature de la chaîne finale dépendent de la glycoprotéines, des enzymes golgiennes présentes et de leur disponibilité. 6,8 4,5 pH

a b

Endosomes de recyclage

Endosomes de tri

Lysosomes Appareil de Golgi MVBs

Fig.5 – Les différents types d’endosomes Les macromolécules internalisées par la voie clathrine (a) ou par des voies indépendantes de la clathrine (b) sont adressées par voie vésiculaire vers les endosomes de tri. Les molécules peuvent ensuit être adressés vers la membrane plasmique via les endosomes de recyclage Rab4+ ou Rab11 +, ou progresser vers les endosomes tardifs (MVB). Le passage des endosomes précoces aux MVBs s’accompagne d’un changement progressif de compartiments Rab5+ vers des compartiments Rab7+. Les MVBs déversent finalement leur contenu dans les lysosomes. Une voie de communication entre les MVBs et le TGN permet le passage des hydrolases vers les lysosomes et le rappatriement des nutriments vers l’appareil de Golgi. LAMP1 DC-LAMP Domaine LAMP

Glycosylation-N

Glycosylation-O

Fig.6 – Représentation schématique de deux LAMPs Les LAMPs sont des protéines transmembranaires caractérisés par un vaste domaine cytoplasmique comportant un domaine LAMP, un fort niveau de glycosylation et un signal d’adressage YXX
sur la queue cytoplasmique. Le domaine LAMP est constitué par deux boucles formées par des ponts disulfure. Les molécules LAMP1 et LAMP2 possèdent 2 domaines LAMP séparés par une région charnière, alors que DC-LAMP et CD68 n’ont qu’un seul domaine LAMP. TLR9 ADN CpG

UNC93B1

b +gp96 PRAT4A a

c

d RE

Fig.7 – Régulation de la localisation de TLR9 TLR9 est localisé dans le RE dans les cellules non activées (a). Lors d’une infection virale, de l’ADN se retrouve libéré dans les endosomes (b). L’activation des cellules déclenche la relocalisation des récepteurs TLRs associés à la protéine chaperon UNC93B1 (c). La relocalisation de TLR9 dépend des protéines chaperon gp96 et PRAT4A. TRL9 est adressé vers les endosomes tardifs où il subit un clivage de son domaine cytoplasmique (d). Sa présence dans ce compartiment permet aussi sa rencontre avec son ligand, l’ADN CpG. La fixation du ligand entraîne une dimérisation du récepteur et la cascade de signalisation activatrice.