A HISTOCHEMICAL STUDY OF ROOT NODULE DEVELOPMENT

ONTVANGEN

\ 1 JUNI 1QQ1

GêHMABiX

Clemensva n deWie l

CENTRALE LANDBOUW/CATALOGUS

0000 0429128 8 Promotor : Dr.A .va n Kammen,hoogleraa ri nd emoleculair e biologie

Co-promotoren : Dr.A.H.J . Bisseling, universitair hoofddocent moleculaire biologie Dr.A.A.M ,va nLammeren , universitairdocen tplantencytologi e en -morfologie jUAlOpzol, WIS

Clemensva n deWie l

A HISTOCHEMICAL STUDY OF ROOT NODULE DEVELOPMENT

Proefschrift terverkrijgin g vand egraa dva n doctori n delandbouw -e n milieuwetenschappen, op gezagva nd erecto r magnificus, dr. H.C. van der Plas, inhe topenbaa r teverdeTlige n opdinsda g 11 juni 1991 desnamiddag st evie ruu ri nd eaul a vand eLandbouwuniversitei t teWageninge n

\ T'r^y- 1A( O o 5b BIBLIOTHEEK CAKDBOUWUNIVERSITJ33 KAGENINGE8

Theinvestigation s described inthi sthesi swer ecarrie dou t atth edepartmen t ofMolecula r Biology andth edepartmen t ofPlan t Cytology andMorphology , Agricultural University, Wageningen, TheNetherlands .

The front cover shows adar k field micrograph of a transection through aroo t of a six- day-old pea (Pisum sativum L.) root. Expression of the PsENOD12 gene in an early stageo froo tnodul eformatio n isvisualize d byth epresenc e of white silvergrains . fJMO#E£)ft /y&s

STELLINGEN

1 De integratie van moleculair-biologische en microscopische technieken leidt bij het onderzoek aan wortelknollen tot een verdieping van het inzicht in het ontwikkelingsproces.

Dit proefschrift

2 Nu immunolocalisatiewaarnemingen hebben laten zien dat het voorkomen van leghemoglobine beperkt is tot de laat-symbiontische zone van de erwtewortelknol, is de waarneming van Newcomb dat de "region of thread invasion" in een "pink zone" van de wortelknol ligt, uiterst onwaarschijnlijk geworden.

Newcomb(1976 )Can . J.Bot .54 : 2163. Ditproefschrif t

3 Het strijdig zijn van de immunolocalisatie van xanthinedehydrogenase (XDH) in bepaalde organellen van de ongeïnfecteerde cellen van sojawortelknollen door Nguyen et al. met observaties uit voorgaande rapporten is te verklaren door aspecifieke hechting van hun antiserum tegen XDH aan zetmeelkorrels bevattende plastiden van deongeïnfecteerd e cellen.

Nguyene tal .(1986 )Plant a 167: 190.

4 Het maltraiteren van rijstwortels met celwand degraderende enzymen en poly- ethyleenglycol door Al-Mallah et al. mag dan wel leiden tot het terechtkomen van rhizobia in afgestorven cellen van een opgezwollen wortelcortex, maar heeft als zodanig weinig te maken met het aanzetten van rijst tot symbiontische wortelknolvorming.

Al-Mallahe tal .(1989 )J.Exp.Bot .40 : 473.

5 Het in een onderzoek combineren van deexpertis e van tweeverschillend e vakgroepen kan vergeleken worden met het tot stand komen van een /?/jizofci«m/waardpIant- symbiose: in beide gevallen isee n uitgebalanceerde communicatiestructuur tussen de tweepartner s nodig, wil er geen voortijdige abortie plaatsvinden. Gezien de aard van de beschikbare gegevens en depraktij k van de theorievorming is het paleontologisch onderzoek naar de afstammingsgeschiedenis van de mens veel interessanter vanuit een wetenschapssociologisch dan vanuit een evolutiebiologisch oogpunt.

R. Lewin (1987) Bones of contention, Simons & Schuster, Inc. D.C.Johanso n &M.A .Ede y (1981) Lucy, The beginningso f humankind,Granad aPublishing ,Ltd .

De discussie overd e verwerpelijkheid van hetvogelsoortenjage n zoals gepraktizeerd door de "Dutch Birding Association" als te weinig nuttig voor natuurbescherming en/of -wetenschap getuigt van een misplaatst soort Hollands calvinisme, aangezien er uitd e aard van de bezigheid, namelijk vrije-tijdsbesteding, geen enkele aanleiding is enig nut te vooronderstellen.

De discussie tussen jagers en verklaarde tegenstanders van dejach t heeft veel weg van een competentiestrijd over wie van beidepartije n de meest adequate natuurkennis ten aanzien van hetnatuurbehee r bezit.

9 In al hun diversiteit komen de plannen voor een creatief natuurbeheer, zoals het uitzetten van bevers in de Biesbosch en het plan "Ooievaar", overeen in een voorspiegeling van een terugkeer naar een soort aards paradijs, waarvan het de vraag iso f hetech t bestaan heeft en, zoja , of hetoo k ind e gewenste vorm zal terugkeren.

10 De Nederlandse Spoorwegen N.V. zouden de door hen geproclameerde aansluiting bij het huidige milieubewustzijn het beste kunnen waarmaken door de eerste- klascoupés te vervangen door ruimte voor een efficiënt en betaalbaar vervoer van fietsen ten behoeve van de verplaatsing naar en van het station.

Stellingen bij het proefschrift "A histochemical study of root nodule development" door Clemens van deWiel , teverdedige n op 11 juni 1991 te Wageningen. Aan mijn ouders CONTENTS

Chapter 1: Outline. 9

Chapter 2: General introduction. 13

Chapter 3: Adefenc e responseo f thehos tplan t mightinterfer e withnoduli n 35 geneexpressio n in Viciasativa roo tnodule sinduce d bya n Agrobacteriwn transconjugant.

Chapter 4: Therelationshi p between nodulin geneexpressio n andth e 49 Rhizobiwn nodgene si n Viciasativa roo tnodul e development.

Chapter 5: Theearl y nodulin transcriptENOD 2i slocate di nth enodul e 73 parenchyma (innercortex )o fpe aan d soybean rootnodule s

Chapter 6: Nodulin geneexpressio n andENOD 2localizatio n ineffective , 91 nitrogen-fixing andineffective , bacteria-free noduleso f alfalfa (Medicagosativa)

Chapter 7: TheENOD1 2gen eproduc t isinvolve d inth einfectio n process 107 during thepea-Rhizobium interaction

Chapter 8: Amolecula rmarke rfo r theuninfecte d cellsi n thecentra ltissu eo f 139 pea (Piswnsativum) root nodules

Chapter 9: Generaldiscussio n 155

Chapter 10: Samenvatting 177

Account 185

Nawoord 187

Curriculumvita e 189 CHAPTER 1

Outline In cooperation with soil bacteria of the genera Rhizobium, Bradyrhizobium or Azorhizobium,man y members of the legume family are ablet o form specialized organs on their roots, called root nodules. The bacteria, wrapped up inside a plant membrane, are accomodated in largeparenchymati c cells located centrally in theseroo t nodules.Fo r this,the yrewar d their host byconvertin g atmospheric nitrogen into afor m usable for the plant. The central infected tissue of the nodule is surrounded by a peripheral tissue provided with vascular bundles through which metabolites areexchange d with theothe r partso f theplant . Inth einteractio n with thebacteria , thehos tplan texpresse s specific genes thatar e nottranscribe d at adetectabl e leveli nothe rpart so f theplant .Th eproduct s of severalo f these genesar emad edurin gth eformatio n of thenodul ean dar ename dearl ynodulins . The present study aims at elucidating the role of these early nodulins in the formation and infection of theroo t nodules.Fo r that purpose, we set out to combine the molecular approach of studying gene expression with the microscopical approach of studyingth estructura ldevelopmen to fth enodule . To provide a background to these studies, chapter II summarizes existing knowledge about nodule development from an anatomical/cytological point of view, supplemented with data on already described nodulins and with brief excursions into physiological phenomenarelevan t toth eres to f our study. In chapter III and IV, nodulin gene expression is analysed in common vetch (Vicia sativa) noduleselicite d by apane lo f bacterial strains with variousdefine d genetic changes. Such nodules were blocked atdifferen t stages in thedevelopmen t of thecentra l tissue depending on the bacterium involved; the precise stage at which the blockade occurred wasdetermine d bylight - andelectron-microscopica l observations.I ntha t way, insight could begaine d in the diverse genetic information supplied by the bacterium for nodule development to proceed through the successive developmental stages and the induction of the appropriate nodulin genes going with it. Furthermore, the start of the expression of individual nodulin genes,fo r instance theearl y nodulin Nps-40',coul d be related tocertai n stageso f central tissuedevelopment . In the caseo f the leghemoglobin genes, such acorrelatio n between nodulin gene expression and specific developmental stagescoul db econfirme d by thedirec t localization ofth eleghemoglobi n proteins inpe a (Pisum sativum) nodule sectionscomprisin g different consecutivedevelopmenta l stages, by immunolabeling. Suchdirec t approach of studying nodulin geneexpressio n innodul e sectionswa s further pursued in the chapters V, VI, VII and VIII. In chapters V, VI and VII early nodulin gene transcripts for which sequenced cDNA clones had become available were localized by in situ hybridization: in chapter V, ENOD2 in soybean (Glycinemax) and

1 1 pea nodules, respectively; in chapter VI, ENOD2i n alfalfa (Medicago sativa) nodules; andi nchapte r VII,PsENOD1 2 inpea . Inchapte r VIII anattemp t tolocaliz e theNps-40 ' protein by immunolabeling in pea nodules is described. By these in situ localization methods,differen t temporal and spatialpattern so f geneexpressio n for eachearl y nodulin were determined. Speculations about the functions of the individual nodulins are made based upon the geneexpressio n patterns andth e amino acid sequences of thenodulin s as deduced from thenucleotid e sequenceo f thecorrespondin g cDNAclones . In addition, in chapter VI, in situ localization of MsENOD2 transcripts was performed on alfalfa nodules induced by certain engineered bacterial strainso rb y auxin transport inhibitors. Such nodules do not have bacteria in their central tissue and also differ in other structural details from effective nodules, but nevertheless were shown to exhibit a tissue-specific expression pattern of the MsENOD2 gene similar to effective nodules.I n chapter VII theresult s of further experiments arereporte d pertaining to the influence of the bacterium on nodulin gene expression, particularly the involvement of bacterial factors and the bacterial nod genes in the induction of the expression of the PsENOD12genes . Finally, chapter IX summarizes the results of the in situ localization of early nodulin geneproducts .I n the light of theseresults ,th e significance of our histochemical approach toelucidatin g therol eo f nodulinsi nroo tnodul edevelopmen t isdiscussed .

12 CHAPTER 2

General introduction The Leguminosae are well known for the very intriguing property, found in the majority ofit smembers ,o f developingroo t nodules in symbiosiswit h soilbacteri ao f the genera Rhizobium, Bradyrhizobium and/or Azorhizobium. The interaction between bacterium andplan t is specific in terms ofrecognitio n between thepartners , asmostl y a distinct plant species establishes aneffectiv e symbiosis with adefine d bacterial species. For instance,Rhizobium meliloti establishes aneffectiv e symbiosis only withMedicago, Melilotus andTrigonella species.However , there areals ointeraction s with ales s specific character, exemplified by the so-called broad host range strains that are accepted by a whole serieso f unrelated tropical legumes (Lim& Burton , 1982). After inoculation with ahost-specifi c bacterium, theplan t forms awel l organized and specialized outgrowth, the nodule. Cells in the root cortex are induced to resume meristematic activity andth enewl y formed cellsdevelo pint oth e nodule.Th ebacteri aar e eventually accomodated inside special cellsi n acentra ltissu eo f thenodule , after which they become large, often pleiomorphic-shaped bacteroids. They are separated from the plant cell cytoplasm by a membrane of host origin. The intracellular state enables the bacteria tofi x atmospheric dinitrogen andt oprovid e thehos t withth eprimar yproduc to f thefixation , ammonia,whic h isfurthe r assimilated byth ehos tplant . The development of root nodules and their functioning have been extensively studied from an anatomical and physiological point of view since the discovery of its nitrogen-fixing capability more than a century ago (Hellriegel, 1886). With the rise of molecular-biological techniques to study the structure and organization of genes and the regulation of their expression, the possibility was created to tackle also the molecular aspectso f theinfectio n ofth eroo tan dth edevelopmen to f theroo t nodule. Onth ebacteria l sidethi sle dt oth eidentificatio n ofgene sinvolve d inth einductio n of the nodulation process, the nodgenes ,an d of genes involved in the nitrogen fixation process, the niflfix genes. These genes are located on a large Sym(biotic) plasmid in Rhizobiumstrains ,bu t on the chromosome inBradyrhizobium strains.I n addition, genes involved in the infection process, such as the exo genes that direct the synthesis of extracellular polysaccharides, and the ndv (nodule development) genes that are homologous toth echromosoma l virulence genes, chvA and c/ivB,o f theplan t pathogen Agrobacteriumtumefaciens, havesimilarl y beenfoun d tob eessentia l for establishmento f the symbiosis (for review of thebacteria l genes:Appelbaum , 1990). On theplan t side theeffort s were directed towards theidentificatio n of genes that areonl y transcribed inroo t nodules.Thes e genes weredesignate d asnoduli n genes (Van Kammen, 1984).Som eo f thesegene s arewel lcharacterize d andthei rproduct shav eeve n been assigned clearly circumscribed functions, e.g. the leghemoglobins regulating the oxygen supply toth e bacteroids (Appleby, 1984; Sheehy &Bergersen , 1986).Th eearl y

15 nodulin genes,whic h are expressed during development of the nodule well ahead of the onset of nitrogen fixation, have been implicated in processes of nodule development or infection (Nap &Bisseling , 1990).I norde r toestablis h theirrole si n theseprocesse s iti s necessary todetermin e inwhic h tissueo f thenodul e and atwhic h moment during nodule development, these genes are expressed. To that end, in situ localization studies of nodulin gene products have been performed at both the mRNA and the protein level. Thesestudie sfor m themajo r themeo f thisthesis . As abasi sfo r the interpretation of ourlocalizatio n studies, adetaile d description of thedevelopmen to f theroo tnodul efro m amorphological/anatomica l pointo fvie wwil l be given in the following account. Where relevant, information about aspectso f nodule physiology and therol eo f nodulins, as far asknown , will beprovided . Before this ,th e nodulin concept will first bebriefl y outlined.

NODULINS

By definition, nodulins are plant proteins that are only present in root nodules (Van Kammen, 1984). For practical reasons, nodulins are frequently identified by comparing geneexpressio n in nodulesan duninfecte d roots.Nodulin s thusidentifie d may then stillprov et ob epresen t inothe rpart so f theplant .I ntha tcase ,i twil lb enecessar y to adapt thedesignatio n of theseprotein s asnodulin s asne wobservation s become available. One way of comparing gene expression in nodules and roots is the analysis of the proteins produced upon in vitro translation of RNA in a eukaryotic protein synthesis system by two-dimensional gel electrophoresis. The nodulins identified in that way are indicated by the letter N followed by the molecular weight in kD as determined by SDS/polyacrylamide gel electrophoresis (SDS-PAGE). In addition, the initials of the plant species' scientific name are added in lower case to the "N"i n order to be able to discriminate between nodulins of different plant species. For instance, Nvs-40 is a nodulin with an apparent molecular weight of 40 kD in common vetch (Vicia sauva). Should the nodulin be identified as a nodule-specific form of a protein with a known function occurring elsewhere inth eplant ,e.g . glutaminesynthetase ,th eprefi x "n"ca n be added toth e nameo f theknow n protein if the nodulin isinvolved ; thus,i n ourexample , n-glutaminesynthetase . Another approach to identify nodulins has been the isolation of nodulin cDNA clones by differential screening of a cDNA library from nodule poly(A+) RNA with

16 cDNA from nodule and uninfected rootRNA ,respectively . Thesenoduli ncDN A clones aredesignate d byth eprefi x NOD,precede d byth eprope rplan t speciesinitials .Whe nth e protein product encoded by the cDNA clone has been identified by hybrid-released translation, the nodulin can also beindicate d by the letter N,followe d by the molecular weight in kD as determined by SDS/PAGE of the hybrid-released translation product. Thus, the Ngm-75 nodulin corresponds with the GmENOD2 cDNA clone in soybean (Glycinemax) (Franssen et al., 1987). The nodulin genes have been divided into two classes on the basis of the time course of appearance of their products during nodule development. The early nodulin genes are expressed well in advance of the onset of nitrogen fixation. cDNA clones representing these genes are designated by ENOD after the plant species initials. This classo f nodulin genes isthough t tob einvolve d in theformatio n of thenodule ,includin g the infection process. The major part of the investigations described in this thesis is directed towardsth edeterminatio n of thespatia lan dtempora lpatter n of theexpressio no f early nodulin genes. The majority of described nodulin genes are expressed around the onset of nitrogenfixation an dar ename d late nodulin genes.The y areexemplifie d byth e well-studied leghemoglobin genes.Th elat e nodulin genes arethough t to beinvolve d in establishing a proper environment for nitrogen fixation and metabolizing the fixed nitrogen for transport toth eres to f theplan t (Nap& Bisseling , 1990).

LEGUME ROOT NODULE DEVELOPMENT

Legume root nodule ontogeny and organization has been extensively reviewed, a.o. by Dart (1977), Goodchild (1977), Newcomb (1981) and Bergersen (1982). Basically, two nodule types are distinguished based on the presence or absence of a persistent meristem: an indeterminate anda determinat e type,respectively . Exampleso f thedeterminat e typear eth enodule so f soybean (Glycinemax), bea n (Phaseolus vulgaris) and birdsfoot trefoil (Lotusspec) . The indeterminate type isexemplifie d by noduleso f pea (Pisum sativum),vetc h (Viciaspec) , clover (Trifolium spec.) and alfalfa(Medicago sativa). In the following, the ontogeny of the two nodule types will be described separately, based mainly on theroo t noduleso f soybean andpea ,respectively . Thelatte r two species have been the main subject of this thesis.Th edevelopmen t of both typeso f nodules is represented schematically in Fig. 1; the end result is shown in Fig. 2.Ligh t micrographs of nodule development are shown in chapter 5fo r soybean and in chapters 3, 5 and 7fo r pea.

17 DETERMINATE NODULE INDETERMINATE NODULE

^infection thread ^infection thread infected root hair nodule primordium

root protoxytem pole root phloem root endodermis root pericycle root cortex parenchyma root epidermis-

infection thread infection thread

nodule persistent meristem nodule provascular strand

Fig. 1. Schematic representation of the ontogeny of the determinate (left part of the figure) and the indeterminate (rightpar t of the figure) nodule. In the upper part of the figure infection thread formation in a root hair and concomitant initiation of a nodule primordium is shown. In the lower part of the figure a later stage is shown, in which the infection of the nodule primordium and the development of the nodule vasculature, including the linking up with the root vasculature, are taking place. In the indeterminate noduleth edelimitatio n of thepersisten t meristem is indicated.

18 Indeterminate nodules

Specific tricyclic phenolic compounds,flavonoids , present in theroo t exudateo f the host plant induce Rhizobium in the root vicinity to transcribe its nod genes.Thi s is thought tooccu rvi ath einteractio n of theseflavonoid s with theproduc t of theregulator y gene nodD. When activated by the proper -i.e. host-specific- flavonoid, this nodD product is thought to switch on the transcription of the other nod genes. Of these, the nodABC genes,whic h form one operon, are absolutely necessary for theproductio n of one or more signal(s) that induce(s) the deformation of root hairs and the initiation of nodule primordia in the plant's root cortex. Besides nodABC, other nod genes are involved incertai n aspectso f noduledevelopment , such ashos trecognitio n (Longe tal. , 1989).A rhizobial root hair deformation factor produced byRhizobium meliloti, called NodRm-1, has recently been identified as a sulfated and acylated glucosamine tetrasaccharide by Lerouge et al. (1990). There is also evidence that perturbing the auxin/cytokininbalanc ei nth eroo tplay sa rol ei nelicitin g noduledevelopmen t (Hirsche t al, 1989). Rhizobia areabl e toadher et oth eroo thai r surface by abacteria l Ca2+-dependent protein, the so-called rhicadhesin. Subsequently^hey can anchor themselves to theroo t hair surface by producing a network of cellulose fibrils (Smit et al., 1987). Lectins provided by the host plant have been implicated in the host-specific attachment of the bacteria,bu tthei rrol ei scontroversia l andma ydepen do nenvironmenta l conditions (Diaz et al., 1989a). The most compelling recent piece of evidence for the involvement of lectins in determining host-specificity is provided by Diaz et al. (1989b), who transformed whiteclove r with ape alecti n gene.Transformatio n with thispe alecti n gene enabled the white clover plants to become nodulated by the normally pea-specific bv. viciae of Rhizobium leguminosarum. The oligosaccharide character of the root hair- deforming factor NodRm-1 also suggests the involvement of lectins in the interaction with therhizobi a (Lerougee tal , 1990). Thecurlin g of theroo t hairs,frequentl y described as the formation of "sheperd's crooks",lead s toth eentrapmen t of thebacteria .Afte r entrapment, thebacteri a areabl et o induceloca llesion s inth eroo thai rcel lwall .Th ehos tplan treact sb ydepositin gne wcel l wall material around the lesion in the form of an inwardly growing tube (Callaham & Torrey, 1981).Th e tube is filled with proliferating bacteria surrounded by a matrix and becomesa ninfectio n thread.Th einfectio n thread growstoward sth einne rtangentia l wall of the root hair cell by a process of tip growth (Bakhuizen et al., 1988). During this growth, andlate ron ,th einfectio n threadremain s at'all timesenvelope d byth eplan tcel l membrane.Th egrowin gti pi softe n accompanied byth ehos tcel lnucleus .I nth e majority

19 of legumesi n which theinfectio n process hasbee n studied,jus t emergingroo thair shav e been shown to beth emai n siteo f infection (Bauer, 1981; Bhuvaneswari et al., 1981).I n some species, such asclove r and alfalfa, also old hairs can be infected, after they have been induced toresum e growth, manifested by branching or other signs of deformation (Bhuvaneswari &Solheim , 1985; Wood &Newcomb , 1989). Whileinfectio n isproceedin g inth eroo t hairs,severa l cells of theinne r layerso f theroo tcortex ,mostl y thoseadjacen t toa protoxyle mpol eopposit e theinfectio n site (cf. Libbenga &Bogers , 1974),becom e activated (Fig. 1).Th e nuclei of thesecell s migrate to themiddl e of the largecentra lvacuol evi a cytoplasmic strands.Next , thecell sdivide , at first in a transversal plane, later on also in radial and tangential planes (Libbenga & Harkes, 1973). This group of dividing cells forms the nodule primordium, also named noduleinitia l (Dart, 1977).A sth ecell scontinu e with dividing, they become smalleran d richer in cytoplasmic content and eventually the group takes on ahemispherica l shape. Meanwhile,th einfectio n thread growsthroug h thecorte x towards thisdevelopin g nodule primordium. Infection thread growth from cell tocel lproceed svi aa mechanis m of local dissolution of the host cell wall and formation of ane w wall in thefor m of an inwardly growing tube,jus t as in the original infection of the root hair. Ahead of the growing infection thread, cells in the outer layers of the root cortex also become activated in preparation for infection. These cells deposit a thin additional wall layer and form cytoplasmic strands through theircentra lvacuoles .Th ecytoplasmi c strands subsequently servea spathway sfo r thegrowin g infection thread (Bakhuizen et al., 1988). Theinfectio n threadpenetrate s thenodul eprimordiu m mostofte n atth emiddl eo f the distal end (Fig. 1).Fro m there it ramifies into thinner threads that grow intocell si n thecentra l parto f the primordium. Here,th e release of rhizobia intoth e host cells takes place from swollen, unwalled tips of side branches of the infection thread, via an endocytotic process. The swollen tips can apparently also detach from the infection thread, before the rhizobia are released, and are then called "unwalled droplets" (Newcomb, 1976). After release into the plant cell, the rhizobia are called bacteroids (Bergersen, 1982). Within thehos tcytoplas m theyremai n surrounded bya membran eo f host plant origin, called the peribacteroid membrane (Robertson et al., 1978). The bacteroidsmultiply ,increas ei n sizeconcomitantl y with anincreas eo f theirDN A content (Bisseling et al., 1977),an d finally almost completely fill the host cell.Durin g bacterial proliferation, the host cellproduce s large amounts of peribacteroid membranes. Several late nodulins have been shown to be incorporated into the peribacteroid membrane, emphasizing the specific role of this membrane in the symbiosis (Verma et al., 1986). Especially in indeterminate nodules the bacteroids adopt apleomorphi c shape (Sprent, 1980); they can, for instance, become club-shaped or branch dichotomously. The

21 nucleus,whic h hasbecom emulti-lobed ,an da relativel y smallvacuol eremai ni n acentra l position, but other organelles, like mitochondria and plastids, become confined to a peripheral position in the infected cells. The infected cell as a whole enlarges considerably. Not allth ecell s of thecentra l part of theprimordiu m become infected cells.I na considerable number of cells, rhizobia are not released. These cells differentiate into uninfected cells,als oname d interstitial cells (Goodchild, 1977),characterize d by alarg e centralvacuol ean dplastid s often containingprominen t starchgrains . After the infection of the central part of the nodule primordium, the persistent meristem isestablishe d atth edista l endo f thenodul eprimordiu m next toth e site where theinfectio n thread originally penetrated theprimordiu m (Fig. 1).Thi smeriste m enables the nodule to follow an indeterminate growth pattern. The persistent meristem is also called apical (e.g. Newcomb, 1981),dista l (e.g.Hirsc h et al., 1984) ornodul e meristem (e.g. Bergersen, 1982). The meristem proximally adds new cells to the nodule, which leads to the emergence of the nodule through the outer layers of theroo t cortex. In the central tissueinfectio n thread growth isno wredirecte d distally andit sbranche spenetrat e apar to f thecell stha tar eforme d byth e meristem.Thes ecell s subsequently developint o infected cells next to the cells also arising from the meristem that remain uninfected. Thus, a zonation of consecutive stages of development arises in the central tissue (Newcomb, 1976; seeFig .2) .Goin g in aproxima l direction from the apically localized persistent meristem, one first encounters an invasion zone in which infection threads penetrateplan tcell s andreleas erhizobi a intoth ecells .The n anearl y symbiotic zoneca n be discerned in which the rhizobia proliferate untill the infected cells are almost completely filled with bacteroids, while the infected and uninfected cells themselves enlargeconsiderably . Inth elat e symbiotic zoneth erhizobi a continuethei r differentiation culminatingi nth eonse to f nitrogen fixation. In the late symbiotic zone several late nodulins, among which leghemoglobin accumulatei nth ecytoplas m of theinfecte d cells (Robertsone tal. , 1984).Leghemoglobi n transports oxygen to the bacteroids. Free oxygen diffuses only slowly into the central tissuedu et oa ga sbarrie rpose db yth erestricte d intercellular spacei nth einne rpar to f the peripheral tissue, i.e. the peripheral tissue inside the nodule endodermis (see Fig. 2).A verylo wleve lo ffre e oxygeni nth einfecte d cellsi sa prerequisit e toprotec tth eextremel y oxygen-sensitive nitrogen-fixing enzyme nitrogenase against damage.A t the sametime, the bacteroids need large amounts of oxygen to generate the energy necessary for the nitrogenfixation process .Th eleghemoglobi n apparently serves adua lrole :i tenable sth e infected cells to efficiently take up the oxygen, which so slowly diffuses through the innerpar to f theperiphera ltissue, an dt otranspor ti t toth ebacteroids ,while ,a tth e same

22 time, keeping the concentration of free oxygen sufficiently low to prevent damage to nitrogenase (Wittye tal. , 1986). In older nodules the late symbiotic zone is followed by a senescent zone at the base of the nodule,i n which plant cells,togethe r with the bacteroids, are degraded. The central tissue is delimited from the peripheral tissue by several layers of cells that morphologically resemble theuninfecte d cellso f theres to f thecentra l tissue (Goodchild, 1977),name d boundary layerb yGresshof f andDelve s (1986; seeFig .2) . Returning toth enodul eprimordium ,whil e theinfectio n of thecentra l part of the nodule primordium is going on, the peripheral tissue starts to be formed at the lateral sideso f theprimordium .Provascula r strandsar einitiate d by additionaldivisions , starting in the root pericycle opposite one of the protoxylem poles. From these pericyclic derivatives, division activity spreads out on the one hand into the residual procambial tissuebetwee n thexyle man dth ephloe m of theroo t stele tobrin g about aconnectio n at oneo rtw o sites with both thexyle m stranditsel f and thephloe m strands ateithe r sideo f the xylem strand (Bond, 1948).O n the other hand,divisio n activity spreadsou tint o the nodule peripheral tissue (Fig. 1).Thes eprovascula r strands develop into dichotomously branching nodule vascular bundles. The nodule vascular bundles have a collateral organization, but, contrary toothe r plant organs,th ephloe m is located at theinne r side, i.e.mor eclosel y toth ecentra linfecte d tissuetha n thexyle m (Pate, 1976;se eFig . 2).Th e xylem contains tracheids, which first develop a ring/spiral and later on a scalariform secondary wall thickening pattern (Dart, 1977).Aroun d the vascular tissue, a pericycle develops thati npe aconsist so f asingl elaye ro f transfer cells, i.e.cell sric h incytoplas m with anextensiv e network ofwal lprotuberance s intoth ecel l (Patee tal. , 1969). Transfer cells also develop in the xylem parenchyma and pericycle of the root stele near the nodules (Newcomb & Peterson, 1979). Each of the vascular bundles will also be surrounded bya bundleendodermi sexhibitin g Casparian strips.Thi sbundl eendodermi s iscontinuou swit h theroo tendodermi s (Bond, 1948;se eFig .2) . Liaddition , anodul eendodermis ,als ocalle d common endodermis,i sforme d that dividesth eperiphera l tissueint oa ninne rpart ,containin g thevascula rtissue ,an da noute r part (Fig.2) .Fraze r (1942)di dno tfind Casparia n stripsi n thenodul eendodermi so fpe a and clover, among others. On the contrary, Casparian strips have been reported to develop in the nodule endodermis of pea by Bond (1948). At a later stage in development, the noduleendodermi scell sdeposi t an additional suberin-containingwal l layeral laroun dth ecell .Th ecell so fth eoute rpar to f theperiphera l tissuelin ku pwit hth e parenchymatic cellso f theroo tcorte x andals oresembl e thelatte rcell si nmorphology ,i n that they are relatively large, and loosely packed. No epidermal layer is discernable around the peripheral tissue. The parenchyma cells of the inner part of the peripheral

23 tissuediffe r from thoseo f theoute rpar ti ntha tthe yusuall y are smaller andmor edensel y packed, while their cell walls stain more heavily with toluidine blue (Goodchild, 1977). The limited intercellular space enables this tissue to restrict the rate of free oxygen diffusion into the central tissue, which helps toprotec t nitrogenase (see above). Several authors,e.g . Newcomb, 1981; Bergersen, 1982,hav e simply named the outer and inner parts of the peripheral tissue "outer" and "inner" nodule cortex, respectively. Furthermore,Newcom b (1981) stated that thisnodul ecortex ,a sa whole ,i sdeposite db y the apical (persistent) meristem. A different opinion is held by Bond (1948), which is supported by Libbenga and Harkes (1973). In their view, only the inner part of the peripheral tissue, together with the nodule endodermis, is deposited by the distal meristem and therefore belongs to the nodule proper. This inner part of the peripheral tissuethe n isname d nodulecortex .Th eoute rpar to f theperiphera ltissu ei sregarde d asa derivative of a few layers of root cortex cells around the nodule primordium, in which some divisions and considerable stretching must have occurred to accomodate the enlarging nodule. This outer part then is called root cortex (Bond, 1948), or host tissue (Libbenga and Harkes, 1973) surrounding the nodule. In this view, the nodule endodermis should beregarde d asth eoute r boundary of thenodul e proper. Inchapte r 5 of this thesis (= Van de Wiel et al., 1990), a new term has been proposed for the inner part of the peripheral tissue to avoid the potentially misleading term "cortex". The problemswit h terminology arefurthe r discussed in thegenera ldiscussio n (chapter9) . Theindeterminat enodul eha sa potentia l tobranc h dichotomously, apparently bya splitting of the persistent meristem. Thus, older nodules can develop a coralloid appearance. In perennial species the persistent meristem can resume activity each new growing season,leadin g tonodule swit h annularconstriction s indicatingth econtribution s from each growing season. In such species a periderm develops in the outer part of the peripheral tissue (Dart, 1977).

Determinatenodules

Many aspects of nodule development in determinate nodules are comparable to those of indeterminate nodules. The following section about determinate nodule development therefore forms an addition to the last section about indeterminate nodule development,wit h emphasiso nth edifference s between thetw otype so fnodules . Likei nth e afore-described development of theindeterminat e nodule,th eprimar y infection indeterminat e noduledevelopmen t occurs through adherence of thebacteri a to root hairs, which are induced to curl and initiate infection threads (Turgeon & Bauer,

24 1982).I n contrast, the initiation of noduledevelopmen t occurs in a group of cells in the outer cell layers of theroo t cortex (Fig. 1).Her e a nodule primordium is formed by the subdivision of highlyvacuolat e cortical cells.Thi s subdivision leads to the formation of relatively smallcell swit h ahig hcytoplasmi c density, afe w scattered vacuoles and large nuclei withprominen t nucleoli.Th eprimordiu m subsequently enlargesb yth econtinuin g division activity of thesecells .Fro m the primordium, cell division activity alsoradiate s outtoward s theinne rcel l layers of theroo tcorte x (Newcomb et al., 1979;Calver t etal. , 1984).Here ,betwee n thenodul eprimordiu m andon eo f theprotoxyle mpole so fth eroo t stele,a provascula r strand isforme d that differentiates intoa vascula rbundl e connecting the nodule vasculature to the root vascular system (Fig. 1). The connecting vascular bundlebecome s surrounded by abundl e endodermisan d anewl y formed parenchymatic tissuetha t becomes continuous with theinne rpar t of theperiphera l tissueo f the nodule. This newly formed parenchymatic tissue becomes surrounded by an endodermal layer, i.e. thepar to f the nodule endodermis that links up the part surrounding the nodule with theroo t endodermis (Fig.2) . Whileth ecel ldivisio n activity of thenodul eprimordiu m continues,th e infection thread grows out of the root hair cell and ramifies into the enlarging globular nodule primordium (Fig. 1).Afte r acertai n period, meristematic activity ceases, starting in the centreo f theprimordiu m (Vancee tal. , 1982).Her e theinfectio n threadsbegi n toreleas e rhizobia into the cytoplasm of the plant host cells, via an endocytotic process,jus t as described for indeterminate nodules. In soybean on the other hand, cells with released bacteria can stillprocee d with dividingfo r some time (Newcomb, 1981).Th e bacteroids proliferate, enlarge andcom et ofil l thehos tcel lcytoplasm . Theinfecte d cells asa whol e increasei nsiz eenormously . Asi nindeterminat e nodules,th ecentra l tissueals ocontain s uninfected cells.Th e uninfected cells constitute a more or less continuous network throughout the whole central tissue (Selker, 1988).I n addition toa larg ecentra l vacuole and amyloplasts these cells develop prominent peroxisomes. These peroxisomes contain large amounts of the late nodulin, n-uricase (Bergmann et al., 1983; Van den Bosch & Newcomb, 1986). Uricase catalyzes one of the final steps in the assimilation of the ammonia produced in nitrogen fixation, leading to the production of the ureides, allantoin and allantoic acid. Thesecompound s areth emai n form inwhic h nitrogen istransporte d toothe rpart so f the planti ndeterminat enodule s (Schubert, 1986). Around the central tissue, a peripheral tissue develops. The peripheral tissue becomes divided into an inner and an outer part by an endodermal layer, the nodule endodermis (Fig. 2). No Casparian strips were found in the nodule endodermis of soybean by Frazer (1942). The vast majority of nodule endodermis cells develop into

25 sclereids in soybean (Bergersen, 1982). In the inner part of the peripheral tissue, dichotomously branching vascular bundles develop that meet atth e base in thevascula r bundle connecting the nodule to theroo t vascular system. It was originally thought that these vascular bundles form a closed system around the whole circumference of the nodule.However , recent work by Walsh et al. (1989) shows that these vascular bundles end blindly close to each other at the nodule apex. In the nodule vascular bundles, the phloem completely surrounds the xylem , i.e. the bundles show an amphicribral organization.Ther ei sa pericycl eo f severallayer so fcytoplasmic-ric hcells ,whic h dono t exhibit thecel l wallcharacteristic s of transfer cells.Eac hvascula r bundlei s surrounded by abundl eendodermi swit h Casparian strips (Walsh et al., 1989) (Fig.2) . The peripheral tissue becomes separated from the central tissue by a boundary layer of uninfected cells (Gresshoff &Delves , 1986; seeFig . 2).Als o in the uninfected cellso f theboundar y layer,peroxisome s containinglarg e amountso fn-uricas e havebee n shown to be present (Newcomb et al., 1989). Later on, the outer part of the peripheral tissue gives rise to lenticels, which are positioned in line with the vascular bundles, givingth eoutsid eo f thenodul e astriate d appearance (Corby, 1981). Meristematic activity persists for the longest time in the apical periphery of the central tissue, but finally it comes to an end; thus, the primordium differentiates completely intoa nitrogen-fixin g nodule:a determinat e growthpattern .

Several deviations from the above-described sequence of events in nodule development areknown .Thes ewil lb ebriefl y reviewed in thenex t section.

Alternativemodes of nodule formation andl or infection in legumes

In peanut (Arachis hypogea)(Chandler , 1978) and Stylosanthessp . (Chandler et al., 1982) an alternative mode of infection has been described. Here the bradyrhizobia penetrate the main root intercellularly at thejunctio n of epidermal cells and the large, septate root hairs that only occur at the sites where lateral roots haveemerged . Further penetrationo f theroo tcorte x alsotake splac eintercellularly . Celldivision s areinduce di n the cortex of the lateral root, near to the attachment point to the main root, and these newly divided cells are infected from irregular wall ingrowths containing the rhizobia. After thereleas e of therhizobia , theinfecte d cellsdivid erepeatedl y witheac h daughter cell receiving apar t of the number of bacteroids. In this way acentra l nodular tissue is formed that does notcontai n uninfected cells.Sometime s stretches of non-infected cells are discerned, but these are probably partitions between separate infection zones

26 (Chandler, 1978). In their growth pattern the nodules resemble the determinate type. Although they are formed in thecorte x of the lateral root, theirvascula r tissue becomes connected to the vascular system of the main root. An additional deviation from the soybean noduledescribe d abovei sth e absenceo flenticel si n thematur e nodules (Corby, 1981). Like inArachis, the nodules of Lupinus spec, lack uninfected cells. Most of the central tissue apparently arises by repeated divisions of a few primarily infected cells, with the multiplying bacteria being distributed over the daughter cells (Dart, 1977). Infection of cellsfro m intracellularly locatedinfectio n threadsha sals obee ndescribe d for nodules of Lupinus, i.e. forL. angustifolius, by Robertson et al. (1978). In addition, Lupinusnodule shav ea peculia r growthpatter n aspersisten tdivisio n activity occursbasi - laterally in the nodule. As a consequence, the nodules adopt a collar-shape around the rootdurin g theirdevelopment . In somecase sthi speculia r shape might,however , alsob e duet oth emergin go f severalnodule slyin gclos et oeac h other (Dart, 1977). In a survey of "primitive"legum e trees,D e Faria et al. (1987) described that all memberso f the subfamily Caesalpinioideae studied, and someo f thePapilionoideae ,ha d indeterminate nodules in which bacteria were apparently notrelease d from the infection threadsi nth ecentra ltissue. Infectio n of thecell sappeare d tooccu rb yintercellula r spread rather than by infection threads. This phenomenon of so-called persistent infection threadsi sreminiscen to f thesituatio n inth eBradyrhizobium-induced noduleso nth enon - legumeParasponia (Smith etal. , 1986)(se efurthe r below). In Sesbania rostrata, root nodule formation does not only occur on the subterranean root, but also on the stem. Here so-called mamillae, which are in fact adventitious roots, are the targets for infection by Azorhizobium sesbaniae (Duhoux, 1984). Infection at first proceeds by the dissolution and subsequent penetration of epidermal andoute r cortical cellso f theadventitiou s rootnea rit s siteo femergenc e from thestem .Infectio n pocketsar eformed , surrounded byradia lrow so f tangentially divided host cells. Several nodule primordia are formed in inner layers of the cortex of the adventitious root. Later on, these nodule primordia merge and form a single nodule between thealread ydifferentiate d parto fth e stelea tth ebas eo fth eadventitiou s rootan d the root's apical meristem. From the primary infection pockets the bacteria penetrate intercellularlyfurthe r intoth ecorte x andsubsequentl y intoth enodul eprimordi a (Tsiene t al., 1983). Next, infection threads grow into cells of the nodule primordia and bacteria arerelease d from these into the host cells.Aroun d the infected tissue vascular strands develop that merge at the base with the vascular system of the adventitious root. This vascular systemo f theroo t intur n isconnecte d toth erin go f vascular tissueo f the stem. Atrac eo f theadventitiou sroo tapica lmeriste mremain sdiscernibl e atth edista len do fth e

27 mature nodule.Th enodul e itself showsa determinat e growth pattern (Duhoux, 1984). In another Sesbania species,S. grandiflora, nodule rootlets have been reported. These appear to arise from within the bundle endodermis of nodule vascular traces (Harrise t al., 1949).O nth e noduleso f other species,e.g . inth e genusTrifolium, nodul e rootlets can be induced by high temperature treatments. These also appeart o bederive d from thenodul evascula rbundle s (Nutman, 1956;Da y &Dart , 1971). On the whole, these deviations appear to be only minor variations of the afore- described modes of nodule development. Intercellular infection is often interpreted as intercellular infection thread formation (e.g.Duhoux , 1984),sinc eth ebacteri a are often encased in a structure strongly resembling an infection thread as the bacteria are embedded inon eo r severalrow si na matri x surrounded bya plan tcel l wall.A sfa r asw e know,i ti sno texactl ydescribe d howthi sinfectio n thread-like morphology is generated. Sinceth epenetratin g bacteria areno t immediately surrounded by aplan t membrane,th e infection thread-like morphology will probably not arise by aproces s of tip growth as described for intracellular infection threadformatio n (seesectio n "indeterminatenodules " and Bakhuizen et al., 1988). A more precise study of the mechanisms that cause the infection thread-like morphology and the role of the plant cell membrane in these processes isclearl y required. Eventh e "stem"nodule so fSesbania rostrata areactuall yforme d inth ecorte xo fa root, if an adventitious one, and the resulting nodule can be clearly recognized as belongingt oth edeterminat etype . Thedevelopmen t of rootlets on nodules, such asi n Sesbania grandifloraan d in Trifolium spec, is more reminiscent of the development of adventitious roots in an existing nodule than a fundamental change in the developmental program of a whole nodule,fo r example by achang ei n thepatter n ofdifferentiatio n ofcell sdeposite d by the persistent meristem. In nodules induced by the actinomycete Frankia on, e.g. Alnus spec, nodule rootlets can actually be formed by the nodule meristem (Becking, 1977) (seefurthe r below). Fromth edescription s of noduleformatio n onth eroot so fLeguminosae ,i twil lb e apparent that the nodule can beconsidere d asa tru e plant organ that shows a consistent anatomical organization and a well integrated physiological machinery specialized for symbioticnitroge nfixation . Thelatte ri sexemplifie d byth etigh tregulatio n ofth eoxyge n supply towards the bacteroids, in which the infected cells of the central tissue and the inner part of the nodule peripheral tissue cooperatively participate (Witty et al., 1986), and by the distribution of the nitrogen assimilation pathway over the infected and uninfected cellsi nth edeterminat enodul e(cf . Dilworth &Glenn , 1984;Schubert , 1986). In its anatomical organization, the leguminous root nodule system appears to

28 occupy arathe r uniquepositio n amongplant/microb einteractions .Thi sca n beillustrate d by acompariso n with other root-based nitrogen-fixing systems. In such systems, either no specialized root outgrowth is formed, as is the case for the intercellularly located Azospirillumspec . (Patriquin et al., 1983),o r a structure is formed with the appearance of a modified lateral root, as is apparent in the nodules induced by the actinomycete Frankia spec, on species of several angiosperm families (Becking, 1977), and in the nodules induced by the cyanobacterium Nostoc spec, on cycads (Pate et al. 1988). Regarding thesenodule sa slatera lroot sca nb ejustifie d bythei rroot-pericycli corigi nan d their centrally located vascular tissue.Nevertheless , deviations from normal lateralroo t development occur: for instance, in Comptonia peregrina root cortical cell derivatives contribute to theformatio n of theFrankia-induced nodule,wherea s theroo t cortex does not contribute to normal lateral root development (Callaham & Torrey, 1977). The Frankiasymbion t is harboured ininfecte d cells locatedi n thecentra l layerso f thecorte x (Becking, 1977).Th eNostoc symbion t islocate d intercellularly in aspecialize d cell layer in the middle of the cortex (Pate et al., 1988). Nevertheless, examples of striking similarities between leguminous nodules andactinomycete-induce d nodulesca n be found in theinfectio n process, such as theinitiatio n of infection threads in curled root hairsi n manyLeguminosa e aswel la si n manyhos tspecie so fFrankia (Callaha m et al., 1979),o r inth ephysiology , like theregulatio n of theoxyge n supply byhemoglobi n in the infected cells and/or restrictions in the amount of intercellular space around the infected tissue (Tjepkema etal. , 1988). In contrast, it is hardly possible to consider the legume nodule as a modified lateralroot . Itsmod eo f origin (inth ecorte x instead of thepericycle) ,th ecentra l location of its infected tissue, and the peculiar organization of its peripheral tissue with the vascular tissue distributed over several bundles, each surrounded by a bundle endodermis,an d theadditiona l noduleendodermi s dividing theperiphera l tissueint otw o morphologically different parts,al largu eagains t such aproposition . This becomes the more striking, as one realizes that a same strain of Bradyrhizobium (e.g. NGR234) is able to elicit not only a leguminous type of root nodule on, for instance, the legume siratro (Macroptiliwn atropurpureum), but also a lateral root-like typeo f root noduleo n the non-legumeParasponia (Ulmaceae) (Pricee t al., 1984; see further: Lancelle & Torrey, 1984a, b, and Scott & Bender, 1990). Since thenod gene so f theBradyrhizobium axe, involved in eliciting anodul e inParasponia a s well as in compatible legumes (Scott &Bender , 1990), apparently acomparabl e seto f signals,produce d under thedirectio n of thenod genes ,ca n leadt odifferen t responses,i n terms of nodule structure, in legumes and in Parasponia. Indeed, within the legume family, the typeo f nodule is also largely determined by the host plant (Dart, 1977).O n

29 the other hand, for example, aR. melilotimutan t with aTn 5 insertion 3k b downstream from nodCelicit s on alfalfa apparently bacteria-free nodules resembling lateral roots in thatthe y havea centrall y located vascular bundle.However , in contrast tonorma l lateral roots, thispseudonodul e exhibits anodul eendodermis-lik elaye r in its "cortex" (Dudley et al., 1987).I t will be of considerable interest to determine the mode of origin of such aberrant nodules.Fro m such studiesi tma y become apparent whether thenorma l typeo f legume nodule and the lateral root-like type are in fact minor variations of the same developmental program. On the otherhand ,i tmigh tb eth etha t smalldifference s ina se t of bacterial signals can result, with the same plant, in the initiation of two alternative, basically different, developmental pathways, i.e. a legume nodule or a lateral root, respectively.

In summary, the nodulin genes must play a specific and significant role in the development of symbiotically nitrogen-fixing root nodules. The elucidation of the functions of the nodulins and theregulatio n of the expression of the nodulin genes will contribute toth eunderstandin g of themolecula rmechanism sresultin g in theformatio n of the root nodule and in meeting the requirements for the functioning of this plant organ. Among the root nodules formed on different leguminous plant species there is rather a variation in anatomy andmorphology .A comparativ e studyo f nodulin geneexpressio ni n different types of nodules might reveal to what extent similarities and differences in the varioustype so fnodule sar eth eresul to fvariatio n in thebasi cmechanism s underlying the respective developmental programs. At the same time, such a study of legume nodule formation might offer opportunities to gain insight into the mechanism of plant organ developmenti ngeneral .

REFERENCES

Appelbaum, E„ 1990,Th e Rhizobium/Bradyrhizobium-legume symbiosis, in: Molecular biology of symbiotic nitrogen fixation, P.M.Gresshoff , ed.,CR CPress ,Boc aRaton ,Florida . Appleby, CA., 1984,Annu. Rev.Plant Physiol., 35, 443. Bakhuizen, R., Van Spronsen, P.C., Diaz, C.L., and Kijne, J.W., 1988,Th e plant cytoskeleton in the Rhizobium-legume symbiosis,Ph.D .thesis ,Leide nUniversity , TheNetherlands . Bauer,W.D. , 1981,Ann. Rev.Plant Physiol., 32: 407.

30 Becking,J.H. , 1977,Dinitrogen-fixin gassociation s inhighe rplant sothe rtha nlegumes ,in: A treatiseo n dinitrogenfixation, vol . HI,R.W.F .Hard yan dW.S .Silver ,eds. ,Wiley ,Ne wYork . Bergersen,F.J. , 1982,Roo tnodule s of legumes: structure and functions, Research studies PressWiley , Chichester. Bergmann,H. ,Freddie ,E. ,an d Verma,D.P.S. , 1983,EMBO J.,2 :2333 . Bhuvaneswari,T.V. , andSolheim ,B. , 1985,Physiol. Plant., 63: 25. Bhuvaneswari,T.V. ,Bhagwat ,A.A. ,an d Bauer,W.D. , 1981,Plant Physiol., 68:1144 . Bisseling, T., Van den Bos,R.C. , Van Kammen,A. ,Va nde n Ploeg,M. ,Va n Duijm,P. ,an dHouwers , A., 1977,/ . Gen. Microbiol., 101:79 . Bond,L., 1948,Bot. Gaz., 109: 411. Callaham,D. ,an d Torrey,J.G. , 1977,Can J.Bot., 55:2306 . Callaham,D. ,Newcomb ,W. ,Torrey ,J.G. ,an dPeterson ,R.L. , 1979,Bot. Gaz. Suppl., 140:SI . Callaham,D.A. ,an dTorrey ,J.G. , 1981,Can. J.Bot., 59:1647 . Calvert,H.E. ,Pence ,M.K. ,Pierce ,M. ,Malik ,N.S.A. ,an d Bauer,W.D. , 1984,Can J. Bot., 62:2375 . Chandler, M.R., 1978, /. Exp. Bot.,29 : 749. Chandler, M.R., Date,R.E. ,an d Roughly, R.J., 1982,J. Exp. Bot., 33:47 . Corby, H.D.L., 1981,Th e systematic value of leguminous root nodules, in: Advances in legume systematics,R.M .Polhil lan dP.H .Raven ,eds. , Royal BotanicGardens , Kew. Dart,P. , 1977,Infectio n and development of leguminous nodules,in: A treatis eo n dinitrogen fixation, vol.DI ,R.W.F .Hard y andW.S .Silver ,eds. ,Wiley ,Ne wYork . Day,J.M. ,an d Dart,P.J. , 1971,Rep. Rothamsted Exp.Stat., 1970: 85. DeFaria , S.M., Mclnroy, S.G., and Sprent,J.I. , 1987,Can. J.Bot., 65:553 . Diaz,C.L. , Hengstmengel, M.E., Smit, G., Lugtenberg, BJ.J. , and Kijne, J.W., 1989a, in:Roo t lectin asa determinan t of host-plant specificity in theRhizobium-\tgume symbiosis,C.L . Diaz, Ph.D. thesis,Leide nUniversity ,Th eNetherlands . Diaz,C.L. , Melchers,L.S. ,Hooykaas , P.J.J.,Lugtenberg , BJ.J. , and Kijne, J.W., 1989b, Nature, 338: 579. Dilworth,M. ,an d Glenn, A., 1984,TIBS, 9:519 . Dudley,M.E. ,Jacobs ,T.W. , andLong ,S.R.,1987 ,Planta 171: 289. Duhoux,E. , 1984,Can J. Bot., 62:982 . Franssen,HJ. , Nap,J.-P. ,Gloudemans ,T. , Stiekema,W. , Van Dam,H. ,Govers ,F. ,Louwerse ,J. ,Va n Kammen,A. ,an d Bisseling,T. , 1987,Proc. Natl. Acad. Sei.USA, 84: 4495. Frazer, H.L., 1942,Proc. Roy. Soc. (Edinburgh), Ser. B, 61:328 . Goodchild, D.J., 1977,Int .Rev .Cytol .Suppl. ,6 :235 . Gresshoff, P.M.,an dDelves ,A.C. , 1986,Plan tgeneti capproache s tosymbioti c nodulation andnitroge n fixation in legumes,in: Ageneti c approach toplan t biochemistry, A.D.Blonstei n andP.J .King , eds.,Springer ,Heidelberg .

31 Harris,J.O. , Allen,E.K. ,an d Allen,O.N. , 1949,Am. J. Bot., 36 : 651. Hellriegel, H., 1886,Welch e stickstoffquellen stehen der Pflanze zu Gebote?, Z. Ver. Rübenzucker- IndustrieDeutschen Reichs, 36: 863. Hirsch, A.M., Bhuvaneswari, T.V., Torrey, J.G., and Bisseling, T., 1989,Proc. Natl. Acad. Sei. USA 86: 1244. Hirsch, A.M., Wilson, K.J., Jones, J.D.G., Bang, M., Walker, V.V., and Ausubel, F.M., 1984, J. Bacterid, 158: 1133. Lancelle,S.A. , andTorrey ,J.G. , 1984a,Protoplasma, 123: 26. Lancelle,S.A. , andTorrey ,J.G. , 1984b,Can. J.Bot., 63:25 . Lerouge, P., Roche, P., Faucher, C, Maillet, F., Truchet, G., Promé, J.C., and Dénarié, J., 1990, Nature, 344: 781. Libbenga, K.R., and Bogers, R.J., 1974, Root-nodule morphogenesis, in: The biology of nitrogen fixation, A.Quispel ,ed. ,North-Hollan dPublishin g Company,Amsterdam . Libbenga,K.R. ,an d Harkes,P.A.A. , 1973,Planta, 114:17. Lim,G. ,an d Burton,J.C. , 1982,Nodulatio n status of theLeguminosae ,in: Nitroge n fixation, Volume 2:Rhizobium, W.J.Broughton ,ed. ,Clarendo n Press,Oxford . Long, S.R., (1989),Cell, 56: 203. Nap, J.-P., and Bisseling, T., 1990, Nodulin function and nodulin gene regulation in root nodule development, in:Molecula r biology of symbiotic nitrogen fixation, P.M. Gresshoff, ed., CRC Press,Boc aRaton . Nap,J.P. , Van deWiel ,C , Spaink, H.P., Moerman, M.,Va n den Heuvel,M. , Djordjevic, M.A.,Va n Lammeren, A.A.M.,Va nKammen , A.,an d Bisseling,T. , 1989,Mol. Plant-Microbe Interac, 2: 53. Newcomb,E.H. ,Kaneko ,Y. ,an dVa nde nBosch ,K.A. , 1989,Protoplasma, 150:150 . Newcomb,W. , 1976,Can. J.Bot., 54: 2163. Newcomb,W. , 1981, Int.Rev. Cytol. Suppi, 13:247 . Newcomb,W. , andPeterson ,RI. , 1979,Can. J.Bot., 57: 2583. Newcomb,W. , Sippell,D. ,an dPeterson ,R.L. , 1979,Can. J.Bot., 57:2603 . Nutman,P.S. , 1956,Biol Rev.Cambridge Phil. Soc, 31: 109. Pate,J.S. , 1976,Transpor t in symbiotic systems fixing nitrogen, in:Transpor t in plants,Encyclopedi a of plant physiology, N.S.,vol . 2, pt. B,Tissue s and organs, U. Liittge and M.G. Pitman,eds. , Springer,Heidelberg . Pate,J.S. ,Gunning , B.E.S.,an d Briarty,L.G. , 1969,Planta , 85:11. Pate,J.S. ,Lindblad , P.,an d Atkins,CA. , 1988,Planta, 176: 461. Patriquin,D.G. ,Döbereiner ,J . andJain ,D.K. , 1983,Can. J.Microbiol., 29: 900. Price,G.D. , Mohapatra, S.S.,an dGesshoff , P.M., 1984,Bot. Gaz., 145: 444. Robertson,J.G. , Lyttleton, P.,Bullivant , S., and Grayston, G.F., 1978, /. Cell Sei., 30:129 .

32 Robertson, J.G., Wells, B.,Bisseling , T., Farnden, K.J.F., and Johnston, A.W.B., 1984, Nature, 311: 254. Scheres, B., Van de Wiel, C, Zalensky, A., Horvath, B., Spaink, H., Van Eck, H., Zwartkruis, F., Wolters,A.-M. ,Gloudemans ,T. ,Va n Kammen,A. ,an d Bisseling,T. , 1990,Cell, 60, 281. Schubert, KU., 1986,Ann. Rev. Plant Physiol., 37:539 . Scott,K.F . andBender ,G.L. , 1990, TheParasponia-Bradyrhizobium symbiosis, in: Molecularbiolog yo f symbiotic nitrogen fixation, P.M.Gresshoff , ed.,CR CPress ,Boc aRaton . Selker,J.M.L. , 1988,Protoplasma, 147: 178. Sheehy,J.E. ,an d Bergersen,FJ. , 1986, Ann.Bot., 58, 121. Smit,G. , Kijne J.W., andLugtenberg ,B J J. , 1987,J. Bacteriol., 169: 4294. Smith, CA., Skvirsky, R.C.,an d Hirsch, A.M., 1986,Can. J.Bot., 64: 1474. Sprent,J.I. , 1980,Plant Cell Env., 3: 35. Tjepkema, J.D., Pathirana, M.S., and Zeng, S., 1988, The gas diffusion pathway and hemoglobin contenti nactinorhiza l nodules,in: Nitrogen fixation: hundredyear safter , H.Bothe ,FJ . deBruij n andW.E .Newton ,eds. , GustavFischer ,Stuttgart ,NY . Tsien,H.C. ,Dreyfus , B.L.,an d Schmidt,E.L. , 1983,/ . Bacteriol., 156: 888. Turgeon,B.G. ,an dBauer ,W.D. , 1982,Can. J.Bot., 60: 152. Vance,C.P. ,Johnson , L.E.B.,Stade ,S ,an d Groat,R.G. , 1982,Can. J.Bol., 60: 505. Vande nBosch ,K.A. ,an dNewcomb , E.H., 1986,Planta, 167: 425. Van de Wiel,C , Nap,J.P. , Van Lammeren, A.A.M., and Bisseling, T., 1988, J. PlantPhysiol., 132: 446. Van de Wiel, C, Norris, J.H., Bochenek, B., Dickstein, R., Bisseling, T., and Hirsch, A.M., 1990a, Plant Cell, 2:1009 . Van deWiel ,C , Scheres, B.,Franssen , H.,Va n Lierop, M.-J.,Va n Lammeren, A., Van Kammen,A. , and Bisseling, T., 1990b,EMBO J.,9 :1 . Van Kammen, A., 1984,Plant Mol. Biol. Rep.,2: 43. Verma, D.P.S., Fortin, M.G., Stanley, J., Mauro, V.P., Purohit, S., and Morrison, M., 1986,Plant Mol.Biol., 7: 51. Walsh, K.B.,McCully , M.E.,an d Canny, MJ., 1989,Plant Cell Env.,12 : 395. Witty, J.F., Minchin, F.R., Sk0t, L., and Sheehy, J.E., 1986, Oxford Surv. PlantMol. CellBiol., 3: 275. Wood, S.M., andNewcomb ,W. , 1989,Can. J.Bot., 67:3108 .

33 CHAPTER 3

A defence response of the host plant might interfere with nodulin gene expression in Vicia saliva root nodules induced by an Agrobacterium transconjugant.

Clemensva n deWiel ,Jan-Pete r Nap,Andr éva nLammeren , andTo nBisselin g SUMMARY

The Agrobacterium transconjugant LBA2712 harbours a complete R. leguminosarum symplasmi d in anAgrobacterium chromosoma l background. OnVicia sativa (common vetch) this strain induces nodules in which a number of plant cells is filled with bacteria. Nevertheless the leghemoglobin (Lb) genes were not found to be expressed in thesenodules , suggesting that theRhizobium sy mplasmid-born egene s are not sufficient for inducing Lb gene expression. Detailed structural analyses on the nodules induced by LBA2712 indicated that the bacteria are subject to degradation as soon as they are released from the infection thread. An immunocytological analysis of developing wild-type pea root nodules using a leghemoglobin antiserum showed that leghemoglobin isonl y detectable asfro m thethir d tofift h celllaye r ofth e late symbiotic zone.Thus ,th edifferentiatio n intoinfecte d anduninfecte d cellsprecede s leghemoglobin accumulation. In combination with the structural analysis of the nodules formed by the Agrobacterium transconjugant, iti sconclude d thatth eAgrobacterium-induced nodule sd o not reach the developmental stage in which the leghemoglobin genes normally are induced, despite the release of the bacteria from the infection thread. As the results presentedindicat e thatapparentl y ahos tdefenc e mechanism interferes withnoduli n gene expression, it follows that anAgrobacterium transconjugant does not allow conclusions with respect to the involvement of sym plasmid-borne genes in the induction of the expression of latenoduli n genes.

INTRODUCTION

Duringnodul eformatio n onth eroot so fleguminou splant sa tleas t twenty nodulin genes are expressed (Legocki & Verma, 1980, Govers et al., 1985, Lang-Unnasch & Ausubel, 1985). We showed for pea (Pisum sativum) (Govers et al., 1985), soybean (Glycinemax) (Gloudeman s et al., 1987)an dcommo n vetch (Vicia sativa) (Moerman et al., 1987), that most nodulin genes are expressed shortly before or concomitantly with the onset of nitrogen fixation. This classo f nodulin genes,o f which the leghemoglobin (Lb) genesar eth e best studied sofar , willb ecalle d latenoduli n genes.I n allthre eplan t species a number of nodulin genes were expressed about a week before the onset of nitrogenfixation . This secondclas so f nodulin genesi sname dearl y nodulin genes.

37 Asa n approach toidentif y theRhizobium genes that are involved inth einductio n of nodulin genes, we analysed nodules induced by a strain harbouring the Rhizobium sym plasmid separated from the Rhizobium chromosome. The Agrobacterium transconjugant LBA2712 harbours the complete R. leguminosarum sym plasmid in an Agrobacterium chromosomal background (Hooykaas et al., 1982). On V. sativa this strain forms nodulesi n which somecell sar e filled with bacteria (Hooykaase t al., 1984, Moerman et al., 1987). Only early nodulin genes were found to be expressed and no expression of late nodulin genes could bedetecte d in these nodules (Moerman et al.,i n press), suggesting that theAgrobacterium transconjugant lacks genetic information for inducinglat enoduli n geneexpression .Th eapparen t contradiction between thelac ko f late nodulin gene expression and the presence of infected cells prompted us to study the infection processi nth e nodules induced by theAgrobacterium transconjugant LBA2712 atth eelectro n microscopical level.T oobtai n additional information about thecorrelatio n between the infection process and Lb gene expression we determined by an immunocytochemical method in whichdevelopmenta l zone of awildtyp e peanodul eL b wasdetectable .

RESULTS

Histologyof nodules inducedby the Agrobacterium transconjugantLBA27'12

In nodules induced on V.sativa b y anAgrobacterium cured from its Ti plasmid and carrying acomplet e symplasmi d from R. leguminosarum instead, designated strain LBA2712 (Hooykaas et al., 1982), two early nodulin genes are expressed, whereas no expression of late nodulin genes could be detected. Full details of these studies are presentedi na separat epape r (Moermane tal .i npress) .A sa n approach tounderstan d this pattern of nodulin gene expression, we studied the histology of the nodules induced by LBA2712a tth eelectro nmicroscopica llevel . The overall morphology of the nodules induced by the Agrobacterium transconjugant is shown in transection inFig . 1A . These nodules have many features in common witha wild-typ e nodule (Newcomb, 1981).Tha t is,a napica lmeristem , anearl y symbiotic zonewit hpenetratin g infection threads,an da lat e symbiotic zonecharacterize d byinfecte d anduninfecte d cells,periphericall y locatedvascula rbundle s andendodermis . In the early symbiotic zone bacteria are released from unwalled regions of the infection

38 Fig. 1. Light (A) and electron (B - E) micrographs of a 16-day-old Vicia nodule induced by the Agrobacterium transconjugant LBA2712. A) Low magnification micrograph montage of a section showing the apical meristem (M),vascula r bundles (VB) and endodermis (E).Ba r = 100(im . B) Detail of an unwalled part of an infection thread (IT) showingreleas eo f bacteria (arrow). The "empty spaces" in the outer region of the cytoplasm of the bacteria are indicated by arrowheads. Bar = 0.3 |im. C) Detail of an infected celli n thelat esymbioti c zone.Bacteria insideth eremnant so f an infection thread (left sideo f the micrograph) are degenerating, whereas the bacteria at the right side of the micrograph are apparently not within vesicles and look like vegetative bacteria. Bar = 1.0 (im.D ) Detail of the late symbiotic zone. The contents of the infected cells (IC) are deteriorated. No organelles,excep t for a very dark staining nucleus (N),ca n bediscerned . Incontrast , the uninfected cells(UC ) appear normal.The y havea prominen t central vacuole and large starch granules (S) inside plasties. Bar = 5 um. E) Detail of an infected cell in the late symbiotic zone.Th ebacteri a inside the vesicles in these cells aredegenerating . Bar = 0.3 (tm.

39 thread. Byenterin g the cytoplasmo f thehos tcells ,th e bacteria become surrounded bya host-derivedmembrane ,th eperibacteroi d membrane (Fig. IB). However, immediately upon release from the infection thread the bacteria are characterized by "empty spaces" in the outer regions of their cytoplasm (Fig. IB). In addition, aggregatedelectro n dense material canb eobserve d inth ecytoplas mo f bacteria withinvesicle s inth elat e symbiotic zone (Fig. IE).Thes e ultrastructural changes canb e considered asfeature s of bacterial degeneration (Kijne, 1975).I nthi slat esymbioti czone , not only the bacteria that occur inside vesicles degenerate (Fig. IE), but also bacteria within theremnant s of infection threads (Fig. 1C)deteriorate , andth ehos tcell s (Fig. ID) disintegrate. The nucleus and thecytoplas m of the hostcel l have become electron dense andal lothe rcel lorganelle shav edisappeare dcompletel y (Fig. ID).Th eoveral lpatter no f breakdown bears resemblance to the premature senescence described for ineffective nodulesfro m severalplan t species (Newcombe t al., 1977, Wernere t al., 1980,Vanc ee t al., 1980). However, some bacteria appear much less deteriorated than other bacteria observed in the infected cells of the late symbiotic zone (Fig. 1C). These bacteria are apparently notfoun d within vesicles.The yloo k likevegetativ e bacteria thatpossibl y feed saprophytically on products of infected cell autolysis (Bergersen, 1974). In contrast to infected cells,th euninfecte d cellsresembl eth e uninfected cells inwild-typ e nodules,a s judged by cytological criteria. These cells have aprominen t central vacuole, and in the cytoplasm plastids containing large starch granules are present. No indications of disintegration can beobserve d (Fig. ID).

Localizationof leghemoglobin in wild-type pea rootnodules

To obtain further information about the correlation between late nodulin gene expression androo t noduleorganogenesis ,w euse dL b asa marke rprotein . Bymean so f immunocytochemistry wehav edetermine di n whichdevelopmenta l zoneo f thepe aroo t noduleL bi spresent . A toluidine blue stained sectiono f anodul efro m a 15-day-oldpe aplan t is shown inFig .2 .A t 15day safte r sowingan dinoculation , thenodul e hasth edistinc t zonationo f the indeterminate type of nodule (Newcomb, 1981). Details of the zones indicated by capitalsi nFig .2 ar eshow n inth ephas e contrast micrographs inth erigh tpane lo f Fig.3 . Thedetail sinclud eth eapica lmeriste m (Fig.3A) ,th eearl y symbiotic zonei nwhic hcell s become infected by Rhizobium (Fig. 3B) and progressively older cells of the late symbioticzon ei nwhic h thecell sar efull y differentiated intoinfecte d anduninfecte d cells. Thecytoplas mo f theinfecte d cellsi stightl ypacke d with thecharacteristicall y Y-shaped

40 bacteroids (Fig.3C-E) . Theresults o fth eindirec timmunofluorescenc e methoduse dt odetermin ei nwhic h cells Lb was located are presented in the left panel of Fig. 3. The fluorescence micrographscorrespon d toth ephas econstras t micrographs in therigh t panel.Neithe ri n the apical meristem (Fig. 3A), nor in the early symbiotic zone (Fig. 3B) fluorescence intensities could be discerned above the background of fluorescence in cortex or uninfected root tissue(no t shown).I nth eyounges t celllayer so f thelat e symbiotic zone the nodule cells can beconsidere d fully differentiated into infected and uninfected cells (Fig. 3C), because Y-shaped bacteroids fill the cells as in older cells of the same developmental zone (Fig. 3E). Despite this differentiation, fluorescence cannot be detected inthes eyounges tcell so f thelat e symbiotic zone (Fig. 3C).Fluorescenc e above background becomesdetectabl ei nth ethir dt oth efifth cel llaye ro f thelat e symbioticzon e (Fig. 3E) and fluorescense intensities increase more and more towards the older part of thenodul e (Fig.3E) .Fluorescenc e isfoun d in thecytoplas m and nucleuso f the infected cells, but isabsen t in uninfected cells,thu sconfirmin g earlier observations by Verma& Bal (1976) and Robertson et al.(1984) .

DISCUSSION

Ourstud yo f nodulin geneexpressio n in anAgrobacteriwn transconjugant aimsa t identifying the Rhizobium genes exclusively involved in inducing nodulin gene expression (Moerman et al., 1987).I n the nodules formed by LBA2712o n Viciasativa roots noL bgene swer efoun d tob eexpresse d (Moermane t al., 1987),despit e thereleas e ofbacteri afro m theinfectio n threadan ddifferentiatio n intoinfecte d anduninfecte d cells, show thatdifferen t plant speciesallo w noduledevelopmen t toprocee d todifferen t stages of infection, suggesting that plant species mayvar y in their "tolerance" towards deviant bacteria. In apreviou s study wehav e shown that the sameAgrobacteriwn transconjugant LBA2712, harbouring a complete R. leguminosarum sym plasmid instead of its Ti plasmid, forms nodules on the rootso f peaplant s that are totally devoid of intracellular bacteria, although infection thread-like structures areobserve d (Goverse t al., 1986).Th e reason for thismarke d difference betweenpe aan dVicia noduledevelopmen t isno tclear . In alfalfa, Wong et al. (1983) and Truchet et al. (1984) have shown that an Agrobacterium transconjugant carrying theR. meliloti sy mplasmi d givesris et onodule -

41 "V

r*.i~! •S'-' 'T""

't ' • v '

^•;:'.^i'»W.^

Fig. 2. Toluidine blue stained section of a 15-day-old pea nodule. Capitals indicate the parts of the characteristic developmental zones from the indeterminate nodule that are shown in Fig. 3.A ) apical meristem.B )earl y symbioticzone .C - E )progressivel yolde rpart so f thelat esymbioti czone .Ba r= 10 0 p.m.

Fig.3 .Localizatio n of leghemoglobin ina 15-day-oldpe a noduleb yindirec t immunofluorescence. The left panel shows the fluorescence micrographs and the right panel the corresponding phase contrast micrographs of details of the developmental zones indicated by capitals in Fig. 2. In (B) an infection threadi sindicate db ya narrowhead . IC= infecte d cell,U C= uninfecte d cell.Ba r= 1 0 \im.

42 ï*.\3töB

tVf';.-'--' * -"

43 like structures thatar emor edisturbe d indevelopmen t than thenodule sforme d onpe ab y LBA2712, whereas in clover Hooykaas et al. (1981) have reported infected cells in nodules induced by a transconjugant carrying aR. trifolii sym plasmid. These results showtha tdifferen t plantspecie sallo wnodul edevelopmen tt oprocee d todifferen t stages of infection, suggesting thatplan t species mayvar y in their "tolerance"toward s deviant bacteria. The absence of late nodulin geneexpressio n in the nodules formed byLBA271 2 suggests that Rhizobium genes outside the sym plasmid must be responsible for the induction of latenoduli n geneexpression .However , the strategy of introducing the sym plasmidi n anAgrobacterium chromosoma l backgrounddemand stha tth eAgrobacterium chromosome itself does not interfere with the expression of nodulin genes.O n the one hand theAgrobacterium chromosome shouldno tharbou r genesinvolve d inth einductio n of nodulin gene expression, and on the other hand the Agrobacterium chromosome should notresul t in gene products that preclude nodulin gene expression at a particular stageo f development. Thefirs t condition ismet ,becaus e nonodulin s weredetectabl ei n tumoursforme d on Viciastem sb ywild-typ eA. tumefaciens (Moerman et al., 1987).Bu t the histological data presented in this paper suggest that the second condition is not fulfilled. Structural analyses on the electron microscopical level of thenodule s formed on theroot so fV. sativa byLBA271 2indicat etha tdevelopmen to f thesenodule si s similart o wild-type nodule development up to and including bacterial release from the infection thread anddifferentiatio n intoinfecte d anduninfecte d cells.However , soon after release from theinfectio n thread,furthe r development isseverel ydisturbe d andbot hbacteria lan d plant cells degenerate rapidly. Apparently, thereleas e of the bacteria from the infection threadsi scorrelate d with arecognitio n step,i nwhic hi sdecide dt oallo wo rcounterac tth e infection process. Robertsone tal .(1985 )hav epointe dou ttha tbacteri ai nth einfectio n threadca nb e considered as being extracellular, while after release they can be regarded as being intracellular, although still separated from the plant cytoplasm by the peribacteroid membrane.Bradle ye tal .(1986 )demonstrate d theassociatio n between thebacteroi doute r membrane and the peribacteroid membrane in pea root nodules. Bacterial lipopolysaccharides mayals ob edeposite di nth eperibacteroi d membrane (Bradleye t al., 1986).I ti sconceivabl etha tth eAgrobacterium transconjugan t LBA2712fail s in this first interaction between the membranes of plant and bacterium. As aconsequence , the host plant recognizes the transconjugant as a pathogen. A pathogen that is subject to host defence. The importance of the Rhizobium outer cell surface in bypassing the host defence mechanism is furthermore demonstrated by the disturbed infection process

44 caused by mutations inRhizobium genes that affect the outer surface of the bacterium. Thereactio n of theplan t in thisdisturbe d infection resembles ahypersensitiv e response (Rolfe et al., 1986). Toobtai n additional information about the absence of Lb gene expression in the nodules formed by LBA2712, the correlation between Lb gene expression and nodule development was investigated by an immunocytochemical method. The immunocytologicalobservation s showtha ti nwild-typ epe aroo tnodul edevelopmen t the complete morphological differentiation into infected and uninfected cells precedes the accumulation of Lb. Consequendy, this differentiation process seems independent from the expression of the Lb genes.Assumin g that Lb gene expression in pea and Vicia is regulated in a similar way, the Vicia Lb genes willb e first expressed when thecell s are completely filled with bacteria. As this developmental stagei s obviously notreache d in the nodules formed by theAgrobacterium transconjugant LBA2712, there will be no induction of Lbgen eexpressio n in thesenodules .Th e absence of Lbgen eexpressio n as found previously (Moermane t al., 1987)i stherefor e inperfec t agreementwit h theresult s ofth eimmunocytologica l localization studiespresente d in thispaper . Sinceth eAgrobacterium transconjugant isdegrade ddirectl y uponreleas efro m the infection thread, the nodule does not reach the correct stage of development for the induction of the expression of Lb genes.Apparently , the second condition stated above for the use of an engineered Agrobacterium in the identification of Rhizobium genes involved ininducin g nodulin geneexpressio n isno t fulfilled, because acounteractio n of the host defence response precludes nodulin gene expression. As a consequence, the Agrobacteriumtransconjugan t mighthav eal lth egeneti cinformatio n for theinductio no f the expression of the Lb genes. Therefore, definitive conclusions with respect to Rhizobium genes involved in inducing nodulin gene expression are not allowed for nodulin genes, which are not expressed in the nodules formed by such an engineered Agrobacterium. Chromosomal genes of Rhizobium may be involved in inducing late nodulin geneexpressio n ina n indirect manner byestablishin g a surface of the bacterium thatprevent sa ninteractio n withth eplan tdefenc e mechanism.

45 MATERIALS AND METHODS

Cultivationof plants andbacteria

Pisum sativum L. cv "Rondo" was cultured in gravel trays and inoculated as described by Bisseling et al. (1978). Vicia sativa subsp. nigra (L.)Ehrh. was germinated, cultured and inoculated as described by Moerman et al. (1987). The bacterial strains Agrobacterium tumefaciens LBA2712 and Rhizobium leguminosarum PRE (wild-type) wereculture d asdescribe d (Bhuvaneswari etal. , 1980).

Electron microscopy

Nodules from 16-day-old Vicia plants were fixed with 2.5% glutaraldehyde in 0.1 M sodiumphosphate buffer pH=7.2 for 1 hr at room temperature (RT). After washing in 0.1 M sodiumphosphate buffer pH=7.2 (1 hr, RT), the nodules were treated with 1% osmiumtetroxide in the same buffer for 1 hr (RT). After washing in 0.1 M sodiumphosphate buffer pH=7.2 (1 hr, RT), the nodules were dehydrated in a graded ethanol series and embedded in LR White resin according to the manufacturer's instructions (The London Resin Company Ltd., London). Sections were cut with glass knives on an LKB Ultrotome V. Semithin sections (0.5 - 2 (im) were stained with 1% toluidine blue 0 in 1% tetraborate in water. Ultrathin sections were stained at RT in an LKB Ultrostainer 2168 with uranyl acetate for 20 minutes and subsequently with lead citrate for 40 seconds. The sections were examined using a Philips EM301 transmission electron microscope operated at 60 kV.

Immunocytochemistry

Nodules from 15-day-old pea plants were fixed with 5% paraformaldehyde in 0.05 M sodiumphosphate buffer pH=7.2 for 1h r atRT , washed in 0.05 M sodiumphosphate buffer pH=7.2 (1 hr, RT), dehydrated in a graded ethanol series on ice, and embedded in LR White resin as above. Semithin sections (0.5 |im) were cut on an LKB Ultrotome V and attached topoly-L-lysin e coated slides. Sections were incubated for 1 hr at RT with an antiserum directed against pea leghemoglobin, diluted 1:500 in buffer A (Buffer A = 0.1 M Tris-HCl pH=7.4, 0.1 M NaCl, 1% Bovine Serum Albumine, 0.1%gelatin , 0.1% Tween-20, 0.1% Triton X-100, 0.05% sodium azide). In these experiments a recently made, monospecific, polyclonal anti-Lb serum was used that had a significantly higher titer than the antiserum used previously (by e.g. Govers et al., 1985). After washing for 1 hr in three changes of buffer A, the sections were incubated (1 hr, RT) with FITC-coupled goat-anti-rabbit (GAR) serum (Nordic, Tilburg),

46 diluted 1:25 inbuffe r A.Afte r washingfo r 1 hri n threechange so fbuffe r A,th e sectionswer emounte d with 20% (w/v) Mowiol 4-88 (Hoechst, Frankfurt am Main) in Citifluor (Citifluor Ltd., London)an d examined undera Leit zOrtholu x microscope with bothphas econtras t andepifluorescenc e (bya Xeno n 150lamp )illumination .I norde rt ocompar eth efluorescenc e intensitiesfro m different partso fa nodule , detailswer ephotographe dwit hth esam eexposur etime .Th eappropriat eexposur etim ewa sdetermine do n themos tintensel y fluorescing parto f thenodule .T oobtai n lowmagnificatio n micrographs,consecutiv e sectionsfro m thesam enodul ewer estaine dwit h 1% toluidineblu e0 i n 1% tetraboratei nwater .Negativ e controls included the incubation of sections without adding thefirst antiserum , and the incubation of sectionsfro m uninfected roottissue .

REFERENCES

Bergersen,F.J. , 1974,Formatio n and function of bacteroids, in: TheBiolog y of Nitrogen Fixation, A. Quispel,ed. ,Nort h HollandPublishin g Company,Amsterdam . Bhuvaneswari,T.V. , Turgeon,B.G. ,an d Bauer,W.D. , 1980, PlantPhysiol., 66:1027 . Bisseling,T. ,Va nde nBos ,R.C. ,an dVa nKammen ,A. , 1978,Biochim .Biophys .Acta ,539 :1 . Bradley, D.J.,Butcher , G.W.,Galfre , G.,Wood ,E.A. ,an d BrewingJ. , 1986,J. CellSei., 85: 47. Gloudemans, T., De Vries, S., Bussink, H.-J , Malik, N.S.A., Franssen, HJ., Louwerse, J., and BisselingX, 1987,Plant Mol. Biol., 8:395 . Govers, F., Gloudemans, T., Moerman, M., Van Kammen, A., and Bisseling, T., 1985,EMBO J.,4 : 861. Govers, F., Moerman, M., Downie, J.A., Hooykaas,P. , Franssen, H.J., Louwerse, J., Van Kammen, A.,an dBisseling.T. , 1986,Nature, 323: 564. Hooykaas, P.J.J., Van Brussel, A.A.N., Den Dulk-Ras, H., Van Slogteren, G.M.S., and Schilperoort, R.A., 1981,Nature, 291: 351. Hooykaas,P.J.J. ,Snijdewint , F.G.M., and Schilperoort, R.A., 1982,Plasmid, 8: 73. Hooykaas,PJJ. , Van Brussel,A.A.N. ,Va n Veen,R.J.M. ,an dWijffelman , CA., 1984,Th eexpressio n of Symplasmid san dT iplasmid si nrhizobi a andagrobacteria ,in: Advancesi nNitroge nFixatio n Research,C .Veege ran dW.E . Newton,eds. ,Nijhoff/Junk , TheHague . Kijne,J. , 1975,Physiol. Plant Pathol., 7:17 . Lang-Unnasch,N. ,an d Ausubel,F.M. , 1985,Plant Physiol., 77: 833. Legocki,R.P. ,an d Verma, D.P.S., 1980, Cell, 20:153 . Moerman,M. ,Nap ,J.P. ,Govers ,F. ,Schilperoort , R., Van Kammen,A. ,an dBisseling ,T. , 1987, Plant Mol.Biol., 9:171 .

47 Newcomb, W., 1981,Int. Rev. Cytol. Suppl., 13: 247. Newcomb,W. , Syono,K. ,an d TorreyJ.G., 1977,Can. J.Bot., 55:1891 . Robertson, J.G., Wells, B.,Bisseling , T., Farnden, K.J.F.,an d Johnston, A.W.B., 1984, Nature, 311: 254. Robertson,J.G. , Wells,B. ,Brewin , N.J.,Wood , E.,Knight ,CD. ,an d Downie,J.A. , 1985,J. Cell Sei. Suppl., 2:317 . Rolfe, B.G., Redmond, J.W., Batley, M., Chen, H., Djordjevic, S.P., Ridge, R.W., Bassam, B.J., Sargent, C.L., Dazzo, F.B., and Djordjevic, M.A., 1986, Intercellular communication and recognitioni nth eRhizobium-ltgume symbiosis,in: Recognitioni nMicrobe-Plan t Symbiotican d PathogenicInteractions , B.Lugtenberg ,ed. ,Springer-Verlag ,Berlin . Truchet, G., Rosenberg, C, Vasse, J., Julliot, J.-S., Camut, S., and Dénarié, J., 1984,J. Bacterioi, 157: 134. Vance,C.P. ,Johnson ,L.E.B. ,an d Hardarson,G. , 1980, Physiol. PlantPathol., 17: 167. Verma,D.P.S. ,an d Bal,A.K. , 1976,Proc. Natl. Acad. Sei.USA, 73: 3843. Werner,D. ,Mörschel ,E. ,Stripf ,R. ,an dWinchenbach ,B. , 1980,Planta, 147: 320. Wong,C.H. ,Pankhurst , E., Kondorosi, A.,an d Broughton,WJ. , 1983,/ . Cell. Biol., 97:787 .

48 CHAPTER 4

The relationship between nodulin gene expression and the Rhizobium nod genes in Vicia saliva root nodule development.

Jan-Peter Nap,Clemen sva n deWiel ,Herma n P.Spaink , Marja Moerman, Marcelva nde nHeuvel »Michae lA .Djordjevic , AndréA .M .va nLammeren , Albertva n Kammen,an dTo nBisselin g SUMMARY

The role of the Rhizobium nod genes in the induction of nodulin gene expression was examined byanalyzin gnodule sforme d onvetc hroot s bybacteria l strainscontainin gonl y thenod region .Introductio n of an 11 kbclone d nodregio n of theR. legwninosarum sym plasmidpRLlJ I into symplasmid-cure drhizobi a conferred upon therecipien t strainsth e ability to induce nodules in which all nodulin genes were expressed. This proves that from the sym plasmid only the nodregio n is involved in the induction of nodulin gene expression. An Agrobacterium transconjugant carrying the same nod region induces nodules in which only early nodulin geneexpressio n isdetected . Thus,th e nodregio n is essential for theinductio n ofearl y nodulin geneexpression . In thiscase ,nodul ecytolog y may indicate that a defense response of the plant interferes with the induction of late nodulin gene expression. Indirect evidence ispresente d that indeed theRhizobium nod genes are also in some way involved in the induction of the expression of late nodulin genes.Th ecombinatio n between histological data andpatter n of nodulin geneexpressio n furthermore revealsa correlatio n between nodulestructur e and nodulin geneexpression . Thiscorrelatio n mayai di n speculations about thefunction s ofnodulins .

INTRODUCTION

Leguminous plants are distinguished from otherplan t families by their ability to form nitrogen fixing root nodules in close cooperation with bacteria from the genera Rhizobiumo rBradyrhizobium. In severalleguminou s species,approximatel y thirty plant genes, so-called nodulin genes (Van Kammen, 1984), have been identified that are exclusively expressed in theroo t nodule (Legocki and Verma, 1980,Gover s et al, 1985, Verma and Brisson, 1987).I n recent years,i tha s been found that root nodule formation is attended by adifferentia l expression of nodulin genes.Thi s has resulted in adivisio n ofnoduli n genesint otw odistinc t classes,earl y andlat enoduli n genes. Early nodulin genes are expressed well before the onset of nitrogen fixation, and early nodulins are most likely involved in theformatio n of the nodule structure and the infection process (Govers et al, 1986, Gloudemans et al, 1987). The best studied example, so far, is the soybean early nodulin Ngm-75, which probably is a cell wall constituent (Franssen et al, 1987).Lat e nodulin gene expression starts around the onset

51 of nitrogenfixation, an dlat e nodulinsprobabl y function inestablishin g and maintaining the proper conditions within the nodule that allows nitrogen fixation and ammonia assimilation to occur. Type members of the class of late nodulin genes are the leghemoglobin genes. By analyzing the in vitro translation products of nodule and root RNA on two- dimensional (2-D) gels, we have previously identified one early, Nvs-40, and 15 late nodulin genes (Moerman etal , 1987)i nvetc h (Viciasativa subsp.nigra). Vetch wasuse d as test plant, because this small leguminous plant responds rapidly to inoculation with engineered bacteria (van Brussel et al, 1982). Late nodulin genes included the leghemoglobins, and Nvs-65. A second early nodulin, VsENOD2, was identified by Northern blot analysis using the soybean early nodulin cDNA clone pGmENOD2 (Franssen et al, 1987) as a probe. Insight in the mechanisms of regulation of these nodulin genes may be gained by studying the involvement of Rhizobium genes in the induction of nodulin gene expression. Investigations using overlapping cosmid clones allowed todra w theconclusio n that 10k b nodregio n of the symplasmi d is sufficient for the induction of early and late nodulin gene expression, if present in a rhizobial chromosomal background. Because in nodules induced by an Agrobacterium strain carrying the complete sym plasmid, the early but not the late nodulin genes were expressed (Moerman et al, 1987),w etentativel y concluded that the 10k b nodregio n on the symplasmi d carries at least the information for the induction of early nodulin gene expression. In thepresen t study,w e have analyzed nodulin geneexpressio n in nodules induced by engineered Rhizobium andAgrobacterium strains carrying exclusively the nodregio n from eitherth eR. leguminosarum or theR. trifoliisy mplasmid . We present further evidence that the Rhizobium nod genes are involved in the induction of the expression of early nodulin genes. Our results indicate that the nod genes are in some way involved in the induction of the expression of late nodulin genes. Analysis of nodulin gene expression in nodules increasingly disturbed in development, shows that the process of root nodule development can be dissected into successive steps, each characterized byth estar to fexpressio n ofdefine d nodulingenes .

52 RESULTS

Nodgenes are the only sym plasmid DNA required for nodulingene induction

Wehav euse d strainR. leguminosarum 248c(pMP104) which contains theclone d 11k b nod region of the sym plasmid pRLUI of R. leguminosarum. A physical and genetic map of the 11k b nod region carrying the nodE, F, D,A , B, C, I, and J genes, inserted into a low-copy-number Inc P vector to yield plasmid pMP104 (Spaink et al, 1987), is given in fig 1. Strain 248c(pMP104) has the ability to form nodules on both vetch and pea. After introduction ofpMP10 4 in the symplasmid-cure dR. trifoliistrai n ANU845, the resulting strain ANU845(pMP104) also obtained the ability to induce nodules on vetch and pea. The histology of the nodules induced on vetch by ANU845(pMP104) (fig. 2A) and by 248c(pMP104) is similar to that of the nodules induced by wild-type R. leguminosarum (Newcomb, 1981). These nodules have an apical meristem, peripherically located vascular bundles and acentra l tissue containing infected cells fully packed with bacteroids and uninfected cells.Th e bacteroids develop intopleiomorphi c forms (fig. 2D),bu tfai l tofi x nitrogen due toth eabsenc eo f nifand fix genes.

1kb

nod E F D A B C IJ E B BE E B pMPKM

nod E F D A B C IJ H -s— H l Tn5 i pRt032

Fig. 1. Simplified physical and genetic mapso f thenod region derived from the sym plasmid pRLUI of R. leguminosarum, aspresen t inpMP104 ,an dth e nod region derived from the sym plasmid pANU843 of R. trifolii, aspresen t in pRt032. With theTn 5 innodE at position Kll, thelatte r plasmid becomes pRt032(no<ΠKll::TnS). The maps are aligned tostres s their similarities. Inadditio n toth e nodgene s indicated, pRt032 contains thenod gene snodN, M and L,upstrea m from nodE, andpMP10 4 contains genesdownstrea m ofnodi]. H,Hi n dill;E ,Ec oRI , B,Ba m HI.

53 •-s- v 7;- */>• -

Fig. 2. Light micrograph montages (A - C) of nodules induced on vetch by (A) R. trifolii ANU845(pMP104), (B)R.trifolii ANU843(nodE Kll::Tn5), and (C)R.trifolii ANU845(pRt032)(nodE Kll::Tn5) respectively,an delectro n micrographs (D- F )o fbacteroid s of thesestrains .A )Three-week - old nodule induced byR.trifolii ANU845(pMP104) . D) Bacteroid ofR.trifolii ANU845(pMP104).B ) Four-week-old nodule induced by R.trifolii ANU843(nodE Kll::Tn5). E) Bacteroid of R.trifolii ANU843(nodE Kll::Tn5). C) Three-week-old nodule induced by R.trifolii ANU845(pRt032)(nodE Kll::Tn5). F) Bacteroid of R.trifolii ANU843(pRt032)(no

54 MVb

I00K_ Lf #M.' 61 - â ., 4t - % [31

30 - - ' ' * • C .^ 1*3- -J- • '0 pN (I 1 2 3 - , ! * «' "" *l * » m é * m 1 A' • irr . 1

i 1 **• *- T 1

Fig. 3. Expression of three major nodulin genes induced in vetch by R. trifolii ANU845(pMP104), R.trifolii ANU843(nodE Kll::Tn5), R.lrifolii ANU845(pRt032) (nodEKll::Tn5) , A. tumefacièns LBA4301(pMP104) and R. leguminosarum PRE,respectively . The upper parto f thefigure show sth e fluorograph of a2- Dge lo f invitr otranslatio n products obtained with totalRN Atha twa sisolate d from nitrogen fixing vetch root nodules, ISday safte r sowing and inoculation withR. leguminosarum PRE. Themajo r nodulinspot sNvs-65 ,Nvs-4 0an dvetc hleghemoglobi n(VsLb )ar eindicate db yarrowheads . In thelowe r parto f thefigure , fluorographs areshow n of in vitrotranslatio n products obtained from total RNA isolated from vetch root tips (panel 1),an d from nodules induced onvetc h bypane l 2:R. trifolii ANU845(pMP104), panel 3: R.trifolii ANU843(norfE Kll::Tn5), panel 4: R.trifolii ANU845(pRt032)(ntf<Œ Kll::Tn5), panel 5:A.lumefaciens LBA4301(pMP104) and panel 6:R. leguminosarum PRE.Onl y thepart so f the2- Dgel swithi n thesquare si nth e upperpar tar e shown,a s thesecontai nth emajo r nodulini nvitr otranslatio n products.Th ecompariso n ismad ei n(A )fo rNvs-65 , in (B) for Nvs-40,an d in (C)fo r VsLb.Th egel s used toobtai n (1) through (4) wereru n ina different seriestha nth egel suse d toobtai n (5)an d(6) ,whic hexplain sth emino rdifference s thatca nb eobserve d inth epatter no ftranslatio nproduct sbetwee n thesefluorographs .

Irrespective of the Rhizobium chromosomal background, both early and late nodulin genes are expressed in the nodules induced on vetch by each of the strains containing pMP104. In fig. 3 the expression of three major nodulin genes, Nvs-40, Nvs- 65 and vetch Lb, is shown for the ANU845(pMP104)-induced nodules in panels 2A, 2B, and 2C respectively. All other identified late nodulin mRNAs are also present (data not shown). The presence of the early nodulin VsENOD2 is demonstrated by Northern blot analysis using pGmENOD2 (Franssen et al, 1987) as probe (fig. 4). These results prove that the nod region is the only part of the sym plasmid that is essential for the induction of early and late nodulin gene expression.

55 Therole of the Rhizobium chromosome innodulin gene induction

To examine any role of theRhizobium chromosome in nodulin gene induction, weconstructe dAgrobacterium tumefaciens LBA4301(pMP104). This Tiplasmid-cure d Agrobacterium, containing plasmid pMP104 with the 11 kb nod region, efficiently induces noduleso nvetch .Thes e nodules have an apical meristem, and vascular bundles at the periphery (fig. 5A). Thus, such nodules are organized like wild-type nodules. In the early symbiotic zone, bacteria are released from the infection threads into the cytoplasm of the host cells, and become surrounded by a peribacteroid membrane (fig. 5D).Th ecytologica l data further indicate that after release from theinfectio n threadsth e development of the infected cells is severely disturbed. Unlike wild-type nodule development, some bacteria are observed within the central vacuole (fig. 5D). Bacteria never develop into Y-shaped structures and their cytoplasm becomes condensed (fig. 5E),whic h suggeststha tth ebacteri a are subject todegradation . In addition,th e nucleus andcytoplas m of theinfecte d plantcell shav ebecom eelectro n dense andcel l organelles havecompletel y disappeared (comparefigs. 5B, E withfigs. 2DJEJ 7).Thes e phenomena suggest agenera lcollaps e of theinfecte d plantcells .Thi scollaps ema yb edu et oth elac k ofa rhizobia l signal,bu ti ti sals oreminiscen t of aplan tdefens e response.I ncontrast , the uninfected cellshav e aprominen t centralvacuol e and they contain cell organelles, like plastids with prominent starch granules (fig. 5D),jus t as uninfected cells in nodules induced bywild-typ eRhizobium.

•EN0D2

Fig.4 .Expressio no fth eVsENOD 2 genei n vetch root nodules. Autoradiograph of a Northern blot containingRN Aisolate dfro mfour-week-ol dnodule s induced on vetch by A. tumefaciens LBA4301(pMP104),indicate da sAgr+pMP104 ,an d in three-week-old nodules induced by R. trifolii ANU845(pMP104), indicated as Rhiz+pMP104. The blot was hybridized with nick translated x pPsEN0D2 pGmENOD2.

56 •i- aéf&k J±. • . ff'

Fig. 5. Light micrograph montage (A) and electron micrographs (B - E) of a four-week-old nodule induced on vetchb y theAgrobacterium transconjugant LBA4301(pMP104). A) Section showing the apical meristem (M),vascula r bundles (VB) and endodermis (E).Bar = 100p.m .B ) Detail of thelat e symbioticzone .Th econtent so fth einfecte d cell(IC )hav edeteriorated .N oorganelle sca nb ediscerned , except for aver y dark staining nucleus (N)-Th euninfecte d cell (UC)appear s normal.Ba r= 5um . C) Detail of an uninfected cell in thelat esymbioti c zone.Th econtent s of theuninfecte d cell (UC)appea r unaffected, incontras t toth econtent so f theinfecte d cells(IC) .Th euninfecte d cell showsa prominen t central vacuole (V),a nucleus (N),an d plastids with starch granules (P).Ba r =3 um. D) Detail ofa n infected celli nth eearl ysymbioti czone .I nadditio nt obacteri asurrounde db ya peribacteroi dmembran e (arrowhead)i nth ecytoplasm ,severa lbacteri a(arrows )ar efoun dwithi nth ecentra lvacuol e(V) . Bar= 0, 4 Hm. E)Detai lo f aninfecte d cell showingbacteroid si nth elat esymbioti czone .Th ebacteroid sexhibi ta condensedcytoplasm ,an dth eplan tcytoplas mi sstainin gdark ,whic hi sa sig no fsever edegradation . Bar = 0,4 |im.

Analysis of the RNA isolated from these LBA4301(pMP104)-induced nodules shows that the two early nodulin mRNAs, VsENOD2 (fig. 4) and Nvs-40 (fig. 3, panel 5B) respectively, are present, but late nodulin mRNAs are not detectable (fig. 3, panels 5A and 5C). This result shows that the nod region is the only Rhizobium DNA that in combination with theAgrobacterium chromosome is necessary for the induction of early nodulin gene expression.

57 Nodgenes relate tothe induction of late nodulingene expression

Analogous to the 11k b nodregio n of R. leguminosarum, a 14k b nodregio n of R. trifoliicontain s all essential functions for the induction of a nodule on clover roots (Schofield et al, 1984). A physical and genetic map of the 14 kb region in plasmid pRt032, containing the nod genes «oûfN,M,L,F,E,D,A,B»C,I, and J, is shown in fig. 1. It hasbee n found that aTn 5 insertion in nodE ofR. trifolii extends thehos trang eo f the recipient mutant strain toth epea/vetc hcross-inoculatio n group (Djordjevic et al. 1985). After introduction ofplasmi d pRt032(nodEK ll::Tn5) , containing aTn 5insertio n inth e nodE genea tpositio n Kl1 ,i n thesy mplasmid-cure dstrai n ANU845,th eresultin g strain ANU845(pRt032)(ModEK ll::Tn5 ) differs from ANU845(pMP104) essentially in the R. trifolii origin of its nodregion . Thus, the capacities of two different nodregion s in the induction of nodulin geneexpressio n can beanalyze d in nodulesinduce d byeac ho f the strainso n the same hostplan t species.Strai nANU845(pRt032)(nod EK ll::Tn5 ) induces nodules on vetch with a frequency of only about one nodule per five plants. The few nodules formed develop without delay likewild-typ e nodules (fig. 2C)u p toth e stagei n which infected cells become fully packed with rhizobia. Bacteria develop into pleomorphic forms (fig. 2F), but, in contrast with wild-type nodule development, senescence occurs soon afterwards. In a three-week-old nodule, only a few layers of fully packedcell san d alarg e zoneo f senescence areobserve d (fig. 2C). Analysis of RNA isolated from these nodules show that both the Nvs-40 (fig. 3, panel4B )an dth eVsENOD 2(fig . 4)gen ear eexpressed .Als oth elat enoduli n geneNvs - 65 is expressed (fig. 3, panel 4A), but the expression of other late nodulin genes, including the leghemoglobin genes (fig. 3,pane l 4C), is not detectable in the nodules induced by ANU845(pRt032)(«odEK l1 ::Tn5) .Thes eobservation s dono texclud e that late nodulin genes are still expressed in the few fully infected cells observed in the ANU845(pRt032)(«odE Kll::Tn5)-induce d nodules (fig. 2C),becaus e their expression mightno t bedetectabl e ina total RNApreparation .Therefore , weexamine d thepresenc e ofleghemoglobi n by meanso f immunocytochemistry. Sectionsincubate d with antiserum directed against pea leghemoglobin, followed by immunogold silver labeling of the bound antibodies, showed only a low background labeling in the infected cells of the ANU845(pRt032)(wodEKll::Tn5)-induce d nodules (figs. 6Aan d 6B).I n contrast, high levelso f immunogold silver labeling arefoun d in theplan tcytoplas m surrounding wild- type R. leguminosarum bacteroids (figs. 6Can d 6D).Thes e analyses of individual cells show that the leghemoglobin genes are not expressed in the fully infected cells of the ANU845(pRt032)(«odE Kll::Tn5)-induce d nodules at levels found in infected cellso f nitrogenfixing nodules .

58 térs*1^'

Fig. 6. Localization of leghemoglobin by immunogold silver labeling in a three-week-old nodule induced on vetch by R.trifolii ANU845(pRt032)(7iodE Kll::Tn5) (A, B) and in a three-week-old nodule induced on vetch by wild-type R. leguminosarum PRE (C, D). A and C are the bright field micrographs of the epipolarization micrographs shown in B and D, respectively. The strong signal observed in the infected cells in the nodule induced by wild-type R. leguminosarum PRE (D) ispresen t as a dark silver label in the bright field micrograph C. No signal above background is observed in the cells of the nodule induced by ANU845(pRt032)(norfE Kl l::Tn5) (B).IC , infected cell,UC , uninfected cell,N , nucleus. Bar = 10 um.

Strain ANU845(pRt032)(nodE Kll::Tn5) induced the expression of early nodulin genes and a single late nodulin gene, Nvs-65. As discussed in the previous section, strain ANU845(pMP104) induced nodules in which all early and late nodulin genes examined, including the leghemoglobin genes, are expressed. Since the only difference between thetw ostrain si sth econstruc t carryingth enod region ,th enod regio n ofRhizobium leguminosarum present in pMP104appear s in somewa y tob e involvedi n theinductio n of theexpressio n of latenoduli ngenes .

R. trifoliican induceearly and late nodulingenes invetch nodules

The absence of most late nodulin gene transcripts in thevetc h nodules formed by ANU845(pRt032)(no

59 nod region is deficient in inducing the expression of late nodulin genes in vetch. To investigate further the genetic potentials of theR. trifolii sym plasmid, we used strain ANU843(no

cur

CI-

Fig.7 . Autoradiograph of a Western blot containing bacteroidprotein s from wild-typeR.trifolii bacteroids isolated from clover nodules (A) andANU843(nod E Kll::Tn5)bacteroid sisolate dfro m vetchnodule s (B). Theblo t wasincubate d with antiserum raisedagains t purifiedR. leguminosarum nitrogenasecomponent sC I andCI Ian d^^I-Iabele dprotei nA t odetec tth eimmun e A B complexes.

60 Fig.8 .Localizatio n of nitrogenaseb yimmunogol dsilve rlabelin g ina four-week-ol d noduleinduce do n vetch byR.trifolii ANU845(no

The nodules induced on vetch by ANU843(norfE Kl l::Tn5) do not fix nitrogen. Because all early and late nodulin genes, as far as identified, are expressed in these vetch nodules, it appears unlikely that the ineffective nature of the nodules is due to the absence of certain nodulins. For a better understanding of the Fix- phenotype of these nodules, we examined whether the enzyme nitrogenase was produced. Western blots of total protein isolated from wild-type R. trifolii bacteroids, from clover nodules, and from ANU843(«odE Kl l::Tn5) bacteroids, from vetch nodules, were incubated with antisera against the components CI and CII of the R. legwninosarum nitrogenase complex. The bound antibodies were visualized with 125I- protein A (fig. 7). The fluorograph shows that no detectable levels of CI and CII were present in the ANU843(«odE Kll::Tn5) bacteroids, whereas the R. trifolii nitrogenase was easily detected in clover nodules by the antisera used. The lack of nitrogenase was confirmed in immunocytological studies.

61 Noimmunogol d silver labeling above background was detectable in theANU843(nod E Kl1 ::Tn5 )bacteroid s after incubation of nodule sections with CIantiseru m (figs. 8Aan d B). In contrast, after the same treatment a high level of immunogold silver labeling is observed inwild-typ eR. trifolii bacteroids inclove rnodul e sections (figs. 8Can dD )an d wild-type R. leguminosarum bacteroids in vetch nodule sections (not shown). These observations indicate that despite the expression of all early and late nodulin genes, nitrogenaseprotei n isno tpresen t inANU843(noû K Kl1 ::Tn5)-induce d nodules.

DISCUSSION

Strains ofRhizobium carrying only thenod regio n from aR. leguminosarum sym plasmid areabl e toinduc enodule so n theroot so f vetch.I n suchnodules ,bot h early and late nodulin genes are expressed. Furthermore, we have shown that if the nod region from aR. leguminosarumsy m plasmid ispresen t in anAgrobacterium transconjugant, this nodregio n confers upon theAgrobacterium the ability tofor m anodul eo nvetch .I n these nodules only early and no late nodulin gene expression is found. These results confirm and extend our previous findings (Moerman et al, 1987). Similar results have recently beenfoun d innodule s induced on alfalfa (Medicagosativa) by anAgrobacterium transconjugant containing theclone d nodregio n of the^? . meliloti symplasmi d (Dickstein et al, 1988). Eight nod genes, nodE, F, D, A, B, C, I, and J, have been identified in the nod region ofR. leguminosarum present in pMP104. Analyzing thephenotype s of nodgen e Tn5 mutants, it has been found that mutations in the nodD, A, B, and C genes abolish nodulation. These four genes are apparently indispensable for nodulation. Mutations in the nodE,F ,I , and J genes dono t cause acomplet e inability toinduc e a nitrogen fixing root nodule,bu tresul t in adelaye d nodulation anda smaller numbero f nodules (Downie et al, 1985).Th e nodDgen e encodes aregulator y protein required for the expression of allothe r nodgene s (Rossen etal , 1987).Th e nodA, B,an d Cgene s areessentia l for root haircurling ,formatio n of theinfectio n thread and theinductio n of cortical cell divisions (Downie et al, 1985; Debellé et al, 1986; Dudley et al, 1987). Therefore, the gene products of the nodA, B, and C genes are likely to beresponsibl e for the generation of oneo rmor e signals thatresul t in these threephenomena , followed byinductio n of early nodulin gene expression and formation of a nodule. It is unclear whether the nodA, B, and C gene products accomplish these effects directly or by initiating a cascade of

62 reactions. Transfer of afragmen t carryingexclusivel y the«oi2D,A,B,C,E ,an dF regio n into a sym plasmid-curedRhizobium does not confer upon the recipient strain the ability to induce nodules, although such a strain still causes root hair curling and infection thread formation (Knight et al, 1985,va n Brussel et al, 1988).Thes e observations suggest that the nodD,A,B, and C genes are not sufficient by themselves for the induction of early nodulin gene expression and the formation of root nodules, but additional information encoded by the nodregio n isrequired . Indeed, mutations in the nodE,F , I, and J genes located on plasmid pMP104, if present in a sym plasmid-cured Rhizobium, result in a Nod" phenotype (van Brussel et al, 1988). The gene products of the nodE, F, I, and J genesthu sappea rindispensabl e for nodulation.O nth eothe rhand ,i fthes egene sar epar t of the complete sym plasmid, mutations in the nodE, F, J, and I genes result only in a delayed nodulation and a reduction of the number of nodules (Downie et al, 1985). Possibly the function of the mutated nodE,F ,J , or I gene is complemented by another symplasmi d gene,lik enodN, M,o rL , which are notpresen ti n pMP104.Thi squestio n needs further clarification. The results of the phenotypical effects of mutations do not allow the decision whether the nodE, F, J, and I gene products have any role in the induction ofearl y nodulin geneexpression . Asreporte d previously,earl ynoduli n geneexpressio n isno tdetectabl ei ntumors , andth eAgrobacteriwn chromosome doesno tappea r tocontribut e signalsinvolve d inth e induction ofearl y nodulin genes (Moerman et al, 1987).Recently , it was shown that the nod genes from R. meliloti are inducible to levels comparable with the level found in wild-type R. meliloti, if these genes are present in A. tumefaciens, but not if they are present in other Gram-negative bacteria such as Escherichia coli or Pseudomonas savastanoi(Yelto n et al, 1987).I f theRhizobium andAgrobacterium chromosome have common characteristics that allow the induction of the nod genes, the common chromosomal genes will be essential for root nodule formation. It seems unlikely that thesecommo n geneshav ea rol ei ngeneratin g signals towards theplan t for theinductio n of early nodulin geneexpression .Thes echromosoma l genes willrathe r support thebasi c physiology of the bacterium, which in turn will beimportan t for creating the conditions allowingth einteraction s between bacterium andhos tplant . Whereas theevidenc e of thenod regio n being sufficient for theinductio n of early nodulin geneexpressio n seems unequivocal, iti sno tclea r whether thenod regio n alone, or in cooperation with non-sym plasmid genes, also regulates the induction of late nodulin geneexpression .Non eo f thelat e nodulin genesi sexpresse d in nodules induced byth eAgrobacterium transconjugan t LBA4301(pMP104)carryin gth enod region ,whic h seems to imply that the Agrobacterium chromosome lacks one or more genes that are

63 present on the Rhizobium chromosome and that are involved in the induction of late nodulin gene expression. However, our data indicate that the agrobacteria start to degenerate after they are released from the infection threads. The cytological data are suggestive of a plant defense response (see also van de Wiel et al, 1988). The outer membrane of theAgrobacterium transconjugant is likely to differ from the Rhizobium outer membrane, andbacteria l membrane components become part of the peribacteroid membrane (Bradley et al, 1986).Upo n release of bacteria from theinfectio n threads,th e plant may thus detect an aberrant bacterial surface and react with a defense response. Such an active role of the plant inrejectin g the invading bacteria remains to be proven; morepassiv ephenomen ama y alsocaus eth eobserve ddegeneratio n ofbacteri aan dplan t tissue. However, irrespective of the precise mechanism of the interference of nodule development, late nodulin gene expression may have been prevented or aborted. Therefore, the lack of late nodulin gene expression in the nodules induced by LBA4301(pMP104)neithe r excludes, norproves ,tha t additional genesbeside s thenod region arerequire d for theinductio n of latenoduli n geneexpression . Indicationstha tth e nodgene s areindee dinvolve d inth einductio n of latenoduli n gene expression come from our studies of nodules induced by Rhizobium strains containing the nodregion s from R. trifoliian dR. leguminosarum,respectively . Strains ANU845(pMP104) and ANU845(pRt032) (nodE Kll::Tn5) differ only in the nod region construct they contain, andbot h strains are ablet o nodulate vetch. Inth e nodules induced onvetc h by ANU845(pMP104), all early and late nodulin genes are expressed, whereas in the nodules induced by ANU845(pRt032)(nodE Kll::Tn5) the majority of thelat enoduli n genesi sno texpressed .Thi sdifferenc e in thepatter n of latenoduli n gene expression should be attributed toth e only difference between the two strains,i.e .t oth e nodregio n construct. Hence, the nodgene s have arol e in the induction of late nodulin gene expression. This conclusion is supported by the observation of Schmidt et al. (1986) that the nodA and nodC genes are expressed inR. meliloti bacteroids, thus at a relatively advanced stage of nodule development. Although formal proof for the involvement of theRhizobium nodgene si n theinductio n oflat enoduli n geneexpressio n cannotb eobtained , these nodgene sma yver y wellb eth eonl yRhizobium genesessentia l for theinductio n ofth eexpressio n ofal lnoduli n genes. The reason for the absence of the transcripts from most late nodulin genes in nodules induced by the strain with the R.trifolii nod region is unclear. There are differences in the extent of nod genes between the two constructs pMP104 and pRtl032(rtodE Kll::Tn5), but the differences upstream nodF (i.e. nodNML) and downstream nodU appear not to be essential for nodule formation, in view of the phenotypical analysiso f mutationso f thesegene swhe npar to fth etota l symplasmid .I n

64 addition, iti shar d toimagin e howth e addition of thenod gene sNM Lwoul d disturb the action of theothe r nodgenes .Th e mutation innodE seem sno tresponsibl efo r the failure to induce the expression of late nodulin genes, becauseR. trifoliistrai n ANU843(nodE Kll::Tn5), carrying the nodE Kll mutation in the complete sym plasmid, induces noduleso nvetc h inwhic h alllat e nodulin genes areexpressed . Itha s been found thatth e amount of nodgen eproduct s appearscritica l for theprope r development of nodules.I fa high-copy numberplasmi d carrying thenodA,B,C genestranscribe d constitutively from a vector promoter, wasintroduce d in aR. leguminosarum with acomplet e symplasmid , nodulation ability was abolished completely (Knight et al, 1986). A two fold enhancement of nod gene expression was sufficient toresul t in a strain that induced on vetch only twenty procent of the number of nodules compared to wild-type strains and these nodules were ineffective (Hong et al, 1987). A slight difference in copy number between pMP104 and pRt032(wodE Kll::Tn5) may therefore explain the observed differences inth enoduli n geneexpressio n pattern. Aremarkabl e finding isth eineffectiv e natureo f thenodule s induced byR. trifolii ANU843(«odE Kll::Tn5) due to the lack of nitrogenase. The nifandfix genes were present inthi s strain,an dbot hearl yan dlat e nodulin geneswer eexpresse d in thenodule s formed, soal lprerequisite s for nitrogen fixation on vetch seemfulfilled . The samestrai n induces nitrogen fixing nodules on subterranean clover (Trifolium subterraneum; Djordjevic et al, 1985),showin g thatANU843(nod EK ll::Tn5 ) has allgeneti c potentials for nitrogen fixation. Wed o not know whether the nitrogenase protein has been broken down,o r nifgen eexpressio n isdisturbed . In the latter case,vetc h nodulespossibl y lack afacto r thati spresen t in subterranean clovernodule s andtha ti sinvolve d in theinductio n ofnitrogenas e geneexpression .Thi spresumabl y plant-specific factor maythu sb ea host - specific regulating factor in the induction of bacterial nitrogenase gene expression, in addition to therecentl y suggested role of low oxygen concentrations (Ditta et al, 1987, Fischer andHennecke , 1987). TheRhizobium andAgrobacterium strains used in this study induce nodules in which development is increasingly disturbed. Combining the histological data of the various noduletype s with thepatter n of nodulin geneexpressio n in these nodules (Table 2), a correlation is found between nodule structure and nodulin gene expression. The nodules formed on vetch by strain ANU845(pRt032)(«odEKll::Tn5 ) contain only 2-4 layers of fully packed infected cells.Th e absence of theexpressio n of most latenoduli n genes in these nodules suggests that these late nodulin genes are not expressed in the youngest cellstha tar efull y packed withbacteroids .Thi sconclusio n isi nagreemen t with our immunocytological localization studies of leghemoglobin in wild-type pea nodules (vand e Wiele t al,1988) .

65 Table 1.Patter n of nodulin gene expression in nodules induced by the strains indicated

Early Late Bacterialstrai n ENOD2 Nvs-40 Nvs-65 Lb

PRE + + + + c 248 (pMP104) + + + + ANU843(/wdEK l l::Tn5) + + + + ANU845(pMP104) + + + + ANU845(pRt032)(nodE Kll::Tn5) + + + - LBA4301(pMP104) + +

In nodules induced by LBA4301(pMP104), bacteria were released from the infection threads, but late nodulin gene expression was not detectable. Apparently, releasefro m theinfectio n threadsi sno t sufficient toinduc eth eexpressio n oflat enoduli n genes. Comparison of the histology of thenodule s induced byANU845(pRt032)(rt

66 geneo n theon ehand ,an dnodul edevelopmen t upt oa certai n stageo nth eother , mayb e of use in determining the cell type in which a particular nodulin gene is expressed and mustb eborn ei nmin di n speculations about thefunctio n ofnodulins .

MATERIALS AND METHODS

Plantsand bacteria

Vetch seeds were sterilized, germinated, inoculated and cultured as described (Moerman et al, 1987). Bacterial strains and their relevant characteristics are listed in Table 2. Bacterial crosses were performed asdescribe d (Spainke t al, 1987) using pRK2013(Ditt a etal , 1980)a s helperplasmid . Bacteria were grown in YEM medium as described (Gloudemans et al. 1987) with 2.5 mg/L tetracycline for pMP104 selection and 75 mg/Lkanamycin e for Tn5 selection. Nodules were excised from theroot s with a scalpel. Root tips from uninfected plants were isolated 8 days after sowing. All tissues were immediately frozen in liquid nitrogen and stored at -80"C until use.

Microscopy and immunocytocnemistry

Nodules were fixed with 2.5 % glutaraldehyde and 1 % osmiumtetroxide and embedded in LR White resin as described previously (Van de Wiel et al, 1988). Sections were cut with glass knives on an LKB Ultrotome V. Semithin sections (0.5-2.0 urn) were stained with 1% toluidine blue 0. Ultrathin sections were stained at room temperature in an LKB Ultrostainer 2168 with uranyl acetate for 20 min. and then with lead citrate for 40 sec. Sections were examined using a Philips EM 301 transmission electron microscope operated at 60 kV. For immunocytochemistry, nodules were fixed in 4 % paraformaldehyde, embedded in LR White resin and attached to slides as described (Van de Wiel et al, 1988). Semithin sections (0.5-2.0 |im) were incubated with antiserum, followed by incubation with 10 nm gold particles coupled to protein A (Janssen Pharmaceutica) as secondary label, and the signal was silver enhanced using the IntenSE™11 silver enhancement kit (Janssen Pharmaceutica) according to the manufacturer's manual. After treatment, sections were stained.wit h 0.1 % toluidine blue 0 for 1minute , mounted in Euparal (Chroma) and examined under a Nikon microscope equipped with epipolarization optics (Philips 100 Watt halogene lamp).

67 Table 2.Bacteria l strains and theirrelevan t characteristics

Strain Relevant characteristics* Reference or source

Rhaobium leguminosarum

PRE (wild type) pSym+ nod+fix+ Lie et al. 1979

248 (wild type) pSym+ nod+fix+ Josey et al. 1979

248° (= cured 248) nod-fix- Priem & Wijffelman 1984

248c (pMP104) pRlnod+ nod+fix- Tc* This study

Rkizobium trifolii

ANU843 (wUd type) pSym+ nod-fix- Schofield et al. 1983

ANU843 (nodE Kl 1::Tn5 ) pSym+ nod+fix- KmR Djordjevic et al 1985

ANU845 (=curedANU843 ) nod-fix- Schofield et al. 1983

ANU845 (pMP104) pRlnod+ nod+fix- TcR This study

ANU845(pRt032)(nodEKll::Tn5) pRtnod+ nod+fix- KmRCbR Djordjevic et al. 1985

Agrobacterium tumefaciens

LBA4301 (= curedAch5 ) nod-fix- Hooykaas et al. 1982

LBA4301 (pMP104) pRlnod+ nod+fix- TcR This study

*pSym, sym plasmid pRLIJI (R.leguminosarum) or pANU843 (R.trifolii); pRlnod, cloned nod region from the R.leguminosarum sym plasmid pRLIJI; pRtnod, cloned nod region from the R.trifolii sym plasmid pANU 843; nod, ability to nodulate vetch; fix, in planta nitrogen fixation on vetch; Tc, tetracycline; Km, kanamycin; Cb, carbenicillin.

RNA isolation, in vitro translation and 2-Dgel electrophoresis

Total RNA from plant tissue was isolated according to Govers et al. (1985). Approximately 2 Hgtota lRN A was translated in vitro in arabbi t reticulocyte lysate in a 6 (il reaction mixture as described (Moerman et al, 1987). Translation products were separated by 2-D gel electrophoresis, followed by fluorography of thedrie d gel topreflashe d Kodak XARSfil m (Govers etal , 1985).

Northern blot analysis

Total RNA was denatured in dimethyl sulfoxide/glyoxal, electrophoresed in 0.8% agarose gels (Maniatis et al, 1982), and transferred to GeneScreen (New England Nuclear) membranes as described (Govers et al, 1985). The membranes were hybridized with 32P-labeled probes (Maniatis et al, 1982)

68 underth econdition spreviousl y described (Franssene tal , 1987).

Proteinisolation and Western blotanalysis

Bacteroid proteins were isolated and separated by SDS/polyacrylamide gel electrophoresis as previously described (Bisseling etal , 1983).Protein s were transferred to nitrocellulose by electroblotting (Zabel et al, 1982) and after incubation with antiserum visualized with 125I-protein A according to Bisseling eta l (1983). Preparation of theantiser aagains t leghemoglobin andth ecomponent s CIan dCI I of theR. leguminosarum nitrogenase hasbee ndescribe dbefor e (Bisseling etal , 1980, Vand e Wiel etal , 1988).

REFERENCES

Bisseling, T., Moen, L.L., Van den Bos, R.C., and Van Kammen, A., 1980, J. Gen. Microbiol., 118: 377. Bisseling, T., Been, C, Klugkist,J. , Van Kammen, A., andNadler , K., 1983,EMBO J., 2:961 . Bradley, D.J., Butcher, G.W., Galfre, G., Wood, E.A., and BrewinJSlJ., 1986,/ . Cell Sei.,85 :47 . Debellé, F., Sharma, S.B., Rosenberg, C, Vasse, J., Maillet, F., Truchet, G., and Dénarié, J., 1986, Respective roles of common and specific Rhizobiummeliloti nod genes in thecontro l of lucerne infection, in: Recognition in Microbe-Plant Symbiotic and Pathogenic Interactions, B. Lugtenberg,ed. , Springer-Verlag,Berlin . Dickstein,R. , Bisseling, T., Reinhold, V.N., andAusubel , F.M., 1988, Genes andDevelopment 2: 677. Ditta, G., Stanfield, S., Corbin, D., and Helinski, D.R., 1980,Proc. Natl. Acad. Sei. USA,77 : 7347. Ditta, G., Virts, E., Palomares, A., and Kim, C.H., 1987,/ . Bacteriol., 169: 3217. Djordjevic, M.A., P.R. Schofield, and B.G. Rolfe. 1985. Tn5 mutagenesis of Rhizobium trifolii host- specific nodulation genes result in mutants with altered host-range ability. Mol. Gen. Genet 200: 463-471. Downie, J.A., Knight, CD., Johnston, A.W.B., and Rossen, L., 1985, Mol. Gen. Genet., 189: 255. Dudley, M.E.,Jacobs ,T.W. , andLong , S.R.,1987,Planta 171: 289. Fischer, H.M., and Hennecke, H., 1987,Mol. Gen. Genet., 209:621 . Franssen, HJ., Nap, J.-P., Gloudemans, T., Stiekema, W., Van Dam, H., Govers, F., Louwerse, J., Van Kammen, A., and Bisseling, T., 1987,Proc. Natl.Acad. Sä. USA,84 :4495 . Gloudemans, T., De Vries, S., Bussink, H.-J , Malik, N.S.A., Franssen, HJ., Louwerse, J., and

69 Bisseling.T., 1987,Plant Mol. Biol., 8:395 . Govers,F. ,Gloudemans ,T. , Moerman, M., Van Kammen, A., andBisseling , T., 1985, EMBO J., 4: 861. Govers, F., Moerman, M., Downie, J.A., Hooykaas, P., Franssen, HJ., Louwerse, J., Van Kammen,A. ,an dBisseling.T. , 1986,Nature, 323:564 . Hong, G.-F., Burn, J.L., and Johnston, A.W.B., 1987, Analysis of the mechanism of nod gene regulation inRhizobium leguminosarum, in: Plant Molecular Biology,D .vo nWettstei n andN. - H.Chua ,eds. ,Plenu m Press,Ne wYork . Hooykaas,P.J.J. , Snijdewint, F.G.M., and Schilperoort, R.A., 1982,Plasmid, 8: 73. Josey, D.P.,Beynon ,J.L. ,Johnston , A.W.B.,an d Beringer,J. , 1979,/ . Appl. Microbiol., 46: 343. Knight,CD. ,Rossen ,L. ,Robertson ,J.G. ,Wells ,B „an d Downie,J.A. , 1986, /. Bacteriol., 166:552 . Legocki,R.P. ,an d Verma,D.P.S. , 1980,Cell, 20:153 . Lie, T.A., Soe-Agnie,I.E. , Muller, G.J.L., andGökdan , D., 1979,Environmenta l control of symbiotic nitrogen fixation: limitation toan d flexibility of the\eg\ime-Rhizobium system,in: Proc .Symp . Soil Microbiol. Plants Nutrition 1976, W.J. Broughton, CK. John, J.C Rajaro and B. Lim, eds.,Universit y ofMalaya ,Kual aLumpur . Maniatis, T., Fritsch, E.F., and Sambrook, J., 1982, Molecular Cloning, ALaborator y Manual, Cold SpringHarbor ,Ne wYork . Moerman, M.,Nap ,J.P. , Govers,F. , Schilperoort, R., Van Kammen, A., and Bisseling, T., 1987, Plant Mol. Biol., 9:171 . Newcomb, W., 1981,Int. Rev. Cytol. Suppl., 13:247 . Priem, WJ.E., andWijffelman , CA., 1984,FEMS Microbiol. Lett.,35 :247 . Rossen,L. ,Davis ,E.O. ,an dJohnston , A.W.B., 1987,Trends Biochem. Sei.,12 : 430. Schmidt, J., John, M., Wieneke, U., Krössman, H.-D., and Schell, J., 1986, Proc.Natl. Acad. Sei. USA, 83: 9581. Schofield, P.R., Djordjevic, M.A.,Rolfe , B.G., Shine,J. , and Watson, J.M., 1983,Mol. Gen. Genet., 192: 459. Schofield, P.R., Ridge,R.W. ,Rolfe , B.G., Shine,J , andWatson ,J.M. , 1984,Plant Mol. Biol., 3: 3. Spaink, H.P., Okker, R.J.H., Wijffelman, CA., Pees, E., and Lugtenberg, B.J.J., 1987,Plant Mol. Biol., 9:27 . Van Brussel, A.A.N., Pees, E., Spaink, H.P., Tak, T., Wijffelman, CA., Okker, R.J.H., Truchet, G., and Lugtenberg, B.J.J., 1988, Correlation between Rhizobium leguminosarum nodgene s and nodulation phenotypes on Vicia, in:Nitroge n Fixation: Hundred Years After, H. Bothe,FJ . de Bruijn andW.E .Newton ,eds. ,Gusta vFische rVerlag ,Stuttgart . VanBrussel ,A.A.N. ,Tak ,T. ,Wetselaar , A.,Pees ,E. ,an dWijffelman , 1982,Plant Sei. Lett., 21: 317. Vand eWiel ,C , Nap,J.P. ,Va nLammeren ,A. ,an d Bisseling,T. , 1988,J. Plant Physiol., 132: 446. Van Kammen, A., 1984,Plant Mol. Biol. Rep.,2 : 43.

70 Verma,D.P.S. ,an d Brisson,N. ,eds. , 1987,Molecula r Genetics of Plant-Microbe Interactions,variou s reports,Nijhoff , Dordrecht. Yelton, M.M.,Mulligan ,J.T. , andLong ,S.R. , 1987,J. Bacterid.,169 :3094 . Zabel, P., Moerman, M., Van Straaten, F., Goldbach, R„ and Van Kammen, A., 1982,J. Virol., 41: 1083.

71 CHAPTER 5

The early nodulin transcript ENOD2 is located in the nodule parenchyma (inner cortex) of pea and soybean root nodules.

Clemensva nd eWiel ,Be n Scheres,Hen kFranssen ,Marie-Jos éva nLierop , Andréva nLammeren ,Alber tva nKamme n andTo nBisseling - SUMMARY

Ape a cDNAclon ehomologou s toth e soybean early nodulin clonepGmENOD 2 that most likely encodes a cell wall protein was isolated. The derived amino acid sequence of the pea ENOD2 protein shows that it contains the same repeating pentapeptides, ProProHisGluLys and ProProGluTyrGln, as the soybean ENOD2 protein. Byin situ hybridizatio n theexpressio n of theENOD 2gen e wasshow n tooccu r only in the inner cortex of the indeterminate pea nodule. The transcription of the pea ENOD2gen e starts when the inner cortical cells develop from the nodule meristem.I n thedeterminat e soybean noduleth eENOD 2gen ei sexpresse di n theinne rcorte x aswel l asi ncell ssurroundin g thevascula r bundletha tconnect sth enodul e with theroo t central cylinder. Thenam e "noduleinne rcortex "i smisleading , asther ei sn odirec t homology with the root inner cortex. Therefore, we propose to consider this tissue as nodule parenchyma. Apossibl e roleo f ENOD2i n amajo r function of the nodule parenchyma, namely creating an oxygen barrier for the central tissue with theRhizobium containing cells, is discussed.

INTRODUCTION

Root nodules formed on the roots of leguminous plants are unique organs for symbiotic nitrogen fixation by Rhizobium bacteria. Root nodules are organized structures which develop from meristems newly formed in the cortex of the root as a result of the interaction with rhizobia. The mature root nodule is made up of a central tissuecontainin g infected anduninfecte d cells,surrounde d by acortex .Th enodul eha sa common endodermis which divides the cortex into an outer and an inner cortex. The inner cortex is traversed by vascular strands,eac h surrounded by abundl eendodermis . The strands areconnecte d to the central cylinder of theroo t (for review seeNewcomb , 1981; Bergersen, 1982). By their morphology two main categories of leguminous nodules can be recognized, determinate and indeterminate nodules (for discussion see Sprent, 1980). Legumessuc ha sPisum (pea),Trifolium (clover) andMedicago (alfalfa) species develop indeterminate nodules, whereas determinate nodules are formed on the roots of for example Glycine (soybean) and Phaseolus (bean) species. Indeterminate root nodules

75 have apersisten t meristem at the apex from which cells are continuously added toth e cortical and central tissues. Consequently all tissues of these nodules are of graded age from the meristem to theroo t attachment point. The meristem of adeterminat e nodule ceasest odivid etw ot othre eweek s after inoculation andi tdifferentiate s completely into nodular tissue (Newcomb, 1981). The formation of root nodules involves the differential expression of a serieso f nodule-specific plant genes,th e nodulin genes (Van Kammen, 1984).Thes e geneshav e been divided into early and late nodulin genes. The early nodulin genes are already expressed atearl y stageso froo t noduledevelopment , well before theonse t of nitrogen fixation. Thelat enoduli n genes arefirs t expressed aroundth eonse to f nitrogen fixation, after a complete nodule structure has been formed. Several late nodulins e.g. leghemoglobin, n-uricase and nodulins present in the peribacteroid membrane, have been located in thecentra l tissueo f the nodule (Robertson etal., 1984;Va n den Bosch and Newcomb, 1986, 1988; Verma et al., 1986). Involvement of nodulins in the function of thecortica l tissuesi nth enodul eha sno tbee ndefine d for so far. Recently we have characterized the product of the early nodulin gene ENOD2 from soybean asa proline-ric hprotei n builtu po ftw orepeatin gpentapeptide s (Franssen etal., 1987) .I n thispape r werepor t the aminoaci d sequenceo f ahomologou sENOD 2 nodulin from pea. Moreover, we demonstrate that the ENOD2 gene is specifically expressed in the inner cortex of the determinate soybean nodule aswel l asi n the inner cortex of the indeterminate pea nodule. In the discussion we suggest that the ENOD2 nodulin hasa rol ei nth echaracteristi c morphology of theinne rcorte x andth efunctio n of thistissu ea sbarrie r for oxygendiffusio n intoth eroo tnodules .

RESULTS

Sequenceof the pea ENOD2 earlynodulin

From acDN Alibrar yprepare d againstpolyA(+ )RN A from 21-day-oldpe aroo tnodule s severalclone swer e selected thatspecificall y hybridized withth einser tfro m thesoybea n cDNAclon epGmENOD 2(Fransse n etal., 1987) .Th eclon e with the largest insert was named pPsENOD2. The insert of pPsENOD2, 558 bp in length, was sequenced and a partial amino acid sequence of the pea ENOD2 nodulin deduced from the cDNA sequence is shown in Figure 1. The sequence contains 336 nucleotides of an open

76 ProProHisGlu Ly sProProH isGluAs nThrP roP roGlu TyrGlnProProHis Glu CCCCCTCATGAGAAACCACCTCATGAAAATACACCACCAGAATACCAACCTCCTCATGAG 10 20 30 40 50 60

LysProProHisGluHisProProProGluTyrGlnProProHi sGluLysProProHis AAACCACCACATGAACATCCACCTCCAGAGTACCAACCTCCTCATGAGAAACCTCCTCAT 70 80 90 100 110 120

GluLysProSmrProLysTyrGlnProProHisGluHis SarProProGlu TyrGlnPro GAAAAGCCCTCACCAAAGTATCAACCACCACATGAACATTCGCCGCCAGAGTACCAACCT 130 140 150 160 170 180

ProHi sGluLysProProHisGluAsnProProProValTyrLysProProTyrGluAsn CCGCACGAGAAACCACCACATGAGAATCCACCACCAGTGTACAAACCGCCTTATGAGAAC 190 200 210 220 230 240

SerProProProHisValTyrHisArgProLeuPheGlnAlaProProProValLysPro TCACCCCCACCACATGTGTACCATCGTCCACTCTTTCAGGCACCTCCTCCTGTGAAGCCA 250 260 270 280 290 300

SerAzgP roPheGlyP roPheP roAlaPheLy sAsn * * * TCCCGACCTTTTGGCCCATTTCCAGCCTTTAAAAACTAATAATAACCACCACTGAAGAAT 310 320 330 340 350 360

CTGCACATTTAACTTGGTAAAGTAAAATTCAGAGTGGTTGTTTGTTATGCCTTTTATATC 370 380 390 400 410 420

AAGTGTTTATGTTCTTGTTTTCATTTGTTTTCCTTTTCTGTTTTAAAAGCTCTTTTAAGA 430 440 450 460 470 480

TGTAAAGCACAATGTGCCCTTTCTGCATGCAAATAAAGGCTCTATATATATTGCCTCTGT 490 500 510 520 530 540

AAAAAAAAAAAAAAAAAAAAAA 550 560

Fig. 1. cDNA and predicted amino acid sequence of the pPsENOD2 insert. Nucleotides 1-562 are determinedfrom th epPsENOD 2insert .Th esequenc eo fnucleotide s34-56 2i sconfirme d byanalysi so f anindependentl y obtainedENOD 2cDN Aclone .Th eamin oaci dsequenc eo f the onlylon gope nreadin g frame isdisplaye d overth enucleotid e sequence.Th eamin oaci dtriplet scharacteristi c for the different types of pentapeptide repeats described in the text are overlined with unbroken and dashed bars, respectively.Th ethre eterminatio ncodon sendin gth e readingfram e aremarke db yasterixes .Nucleotide s 511-517encompas sth epolyadenylatio n signal.

77 reading frame (ORF) encoding 112amin o acids of the C-terminal end of theENOD 2 protein. TheOR Fend swit h three successive termination codonsan di sfollowe d bya 3 ' non-translated region of about 235 nucleotides in which a potential poly(A) addition signali spresen t anda shortpar to f apoly(A ) tail.Th eamin oaci d sequencereveal s that the pea ENOD2protei n is very proline-rich and is mainly composed of two repeating pentapeptides, ProProHisGluLys and ProProGluTyrGln, respectively. Two ProProHisGluLys repeats alternate with oneProProGluTyrGl n element. Southern blots containing pea genomic DNA digested with EcoRI or SphI and a dilution series of pPsENOD2 were hybridized with the insert of pPsENOD2. A 7.2 kb EcoRI fragment and a 4.6 kb SphI fragment hybridized to pPsENOD2. Moreover, comparison of the levelso f hybridization ofth epPsENOD 2dilutio n series and thepe a genomic fragments, respectively,indicate d thatonl yon eENOD 2gen ei spresen t inth epe a genome (datano t shown).

Localizationof the ENOD2 transcriptin indeterminatepea nodules

Weexamine d with the insitu hybridizatio n techniquei nwhic h nodular tissueth e pea ENOD2 gene is expressed. Longitudinal sections of pea nodules from 20-day-old plants were hybridized with "s-labeled sensean d antisense RNA transcribed from the insert of pPsENOD2. After autoradiography the anti-sense RNA probe appeared to hybridize withRN Apresen ti n thesection s whereas the senseRN A probedi dno t (result not shown).Th e antisense RNA probe only hybridized with RNA present in the inner cortex of the nodule, suggesting that thepe a ENOD2 gene is exclusively expressed in this nodular tissue (Figure2A,B) .Th eENOD 2gen ei sexpresse d throughout thewhol e inner cortex; from the youngest cells directly adjacent to the meristem up to the oldest cells near the root attachment point. The vascular tissue traversing the nodule inner cortexdoe sno tcontai ndetectabl elevel so fth eENOD 2transcript . The presence of the ENOD2 transcript in the inner cortical cells close to the nodulemeriste m indicated thatexpressio n of theENOD 2gen ei sinduce d ata relativel y early stage of development. To test this we also hybridized serial sections of nodule primordia of seven-, eight-, and ten-day-old roots to antisense RNA from pPsENOD2. Thepe anodul e primordia areinitiate d in theinne rcel llayer so f theroo t cortex.A tda y seven noENOD 2messenge r wasdetectabl e in noduleprimordi a (data not shown).Th e ENOD2transcrip t is first detectable in nodule primordia of an eight-day-old pea plant (Figure 3A,B).A t this stage theinfectio n thread, which transports therhizobi a from an infected root hair to the nodule primordium, has already reached the primordium and

78 branchedof f intodifferen t cellso f thecentra lpar to f thepimordium .Moreove r the first differentiation ofprocambia l strands (not showni nFigur e 3A,bu tvisibl ei n consecutive sections of the same primordium) and the formation of an apical meristem have taken place (Figure 3A).Th eENOD 2messenge r ispresen t in a few inner cortical cells atth e baseo f the noduleprimordiu m (Figure 3B).I n noduleso f ten-day-old peaplants ,whic h isthre eday sbefor e theonse to f nitrogenfixation , infected cellsfille d withbacteroid sca n be seen at the base of the nodule (Fig. 3C). The ENOD2 transcript is now present throughout the inner cortex as in the 20-day-old nodule (compare Figure 3C,D with Figure 2A,B).

Localizationof the ENOD2 transcriptin determinate soybean nodules

Sections of soybean nodules of 21-day-old plants were hybridized with "S- labeled antisense RNA made from the insert of pGmENOD2. Figures 2C and D show that as in pea nodules the soybean ENOD2 messenger is located in the nodule inner cortex and in the tissue surrounding the vascular bundle connecting the nodule to the centralcylinde ro f theroot . Sincether ei sn opersisten t meristem in thistyp eo f nodule, the inner cortical tissue completely surrounds the central tissue of the mature nodule (compareFigur e 2CJDwit hFigur e 2A,B of pea).Th edistributio n of the silver grainsi n the different nodule tissues is better shown in Figures 2E and F, which represent magnifications of a section through a 21-day-old soybean nodule hybridized with ENOD2antisens eRNA . In order to obtain a good impression of the various tissues, a similar part of a section of a soybean nodule from a 21-day-old plant embedded in glycolmethacrylate resin, is shown in Figure 2G. Here the tissue morphology is better preserved than in paraffin. The inner cortical cells have fewer and smaller intercelullar spaces than the outer cortical cells. The endodermis that separates the inner and outer cortex mainly consists of large sclerenchymatic cells at this stage (Figure 2G).Figur e 2E and F show that thevas tmajorit y ofENOD 2transcrip t isfoun d in theinne rcortex , butlo wlevel so f thismessenge rar eals opresen ti nth eendodermi s andth eoute rcortica l celllaye rdirectl y adjacent to it. The boundary cell layers of uninfected cells between the cortex and the central tissue, like the central tissue itself, appear to contain no ENOD2 transcript (Figures 2E andF) . Also in soybean we studied the appearance of ENOD2 transcript during nodule development. The earliest stage that we investigated was six days after sowing and inoculation.A t thisstag e smallbump sbecom ejus t visibleo nth emai nroo tindicatin gth e

79 presence of nodule primordia. The primordia of the determinate nodule type originate in the outer cell layers of the root cortex. At six days cell divisions have also been induced in the inner cell layers of the root cortex and the central part of these dividing cells is developing into vascular tissue that connects the root nodule with the central cylinder of theroo t (Figure 4A and C).A t six days the soybean ENOD2 messenger is detectable in the newly formed tissue surrounding the procambial strand between the primordium and the root central cylinder and in inner cortical cells at the proximal and lateral sides of the noduleprimordiu m (Figures 4B and D). In a ten-day-old plant the globular meristem has further developed into a central and a cortical tissue (Figure 4E). The ENOD2 gene is expressed in the nodule inner cortex as well as in the tissue surrounding the vascular strand that connects the nodule with the central cylinder (Figure 4E, F). At this stage the inner cortex at the distal part of the nodule already contains the ENOD2 messenger, albeit still at a lower level than in the proximal part of the nodule (Figure 4E, F). In nodules from 21-day-old plants similar amounts of the ENOD2 transcripts are present in all parts of the nodule inner cortex (cf. Figure 2C, D).

Fig.2 .Localizatio n of ENOD2 transcripts by insitu hybridizatio n in pea (A,B) and soybean (C-G) nodules.A )Brigh t field micrograph of a longitudinal section through anodul e from a20-day-ol dpe a plant. In the nodule from the top to thebase , the apical meristem (M),an d early (ES) and late (LS) symbiotic growth stages of thecentra l tissue can be discerned. Over the nodule inner cortex (IC)a n autoradiographicsigna lo fblac ksilve rgrain si spresent . Nosigna lca nb eobserve dove rth enodul eoute r cortex (OC)no r thevascula rbundl e (VB).Th enodul eendodermi sca n notb eeasil yrecognize d inthi s section,sinc ei npe ath eendodermi s does notsclerif y likei n maturing soybean nodules (Fig.2(C )an d (E)). Atth ebas eo f thenodule ,par t of theroo t is visible in transversal section.Her eth ecorte x(RC) , anda grou po fphloe mfiber s(F )an da xyler npol e(X )o fth ecentra lcylinde rar eindicated . Barrepresent s 200um .B )Dar kfiel d micrograph ofth esam esectio na si n(A )showin gth eautoradiographi c signala s whitegrains .C ) Bright field micrograph of a longitudinal section through a nodulefro m a21-day-ol d soybeanplant .Th ecentra l tissue (CT)i scompletel y surrounded bya ninne rcorte x (IC)ove r whicha n autoradiographic signal of black silver grains can be observed. This signal continues over the tissue surrounding thevascula rbundl e (CVB) thatconnect s the nodulet oth ecentra l cylinder of theroot .E , endodermis,othe rabbreviation sa si n(A) .Ba rrepresent s20 0um .D )Dar kfield micrograp ho fth esam e sectiona si n (C)showin gth eautoradiographi c signala swhit egrains .E )Brigh tfield micrograp h of a detail of a section through the same nodule as in (C).Fro m top tobotto m the outer cortex (OC),th e sclerified endodermis(E) ,th einne rcorte x(IC )wit ha vascula rbundl e(VB) ,th eboundar ylaye r(BL) ,an d theinfecte d (In)an duninfecte d (Un)cell so fth ecentra ltissu eca nb ediscerned .Ba rrepresent s5 0um . F) Thesam edetai l as shown in (E), photographed with acombinatio n of brightfield an d epipolarization illumination. Astron gautoradiographi c signal of whitegrain si svisibl e over theinne rcortex .A lower signal is present over the endodermis and the adjacent layer of the outer cortex. G) Detail of a glycolmethacrylate section througha 21-day-ol d soybean noduleshowin g thesam etissue sa tth esam e magnification asi n(E )an d(F) .Abbreviation sa si n(E) . Thearrow sindicat eintercellula r spacesan dth e arrowheadscalciu moxalat ecrystal si nth eoute rcorte x (OC).

80 81 DISCUSSION

In thispape r wehav epresente d evidence that theearl y nodulin geneENOD 2i s specifically expressed during the formation of the tissue in determinate as well as in indeterminateroo tnodule stha tha s sofa r beendescribe d asth einne rcortex .Moreover , theoccurrenc e of homologous ENOD2 genes encoding polypeptides with a conserved structure in different legume species (Franssen et al., 1987; Dickstein et al., 1988) strongly suggests that theENOD 2protei n has arol e in the function of thisroo t nodule tissue (Govers etai, 1989). Inearlie r studiesw ehav edemonstrate d thati nbot h soybean andpe ath eENOD 2 gene isexpresse d during early stages of nodule morphogenesis (Franssen etal., 1987 ; Govers et al., 1986).Besides , it has been shown that in soybean and alfalfa this early nodulin gene is expressed in so-called empty nodules that contain neither infection threads nor intracellular bacteria (Franssen et al., 1987;Dickstei n etal., 1988). Such empty nodules are elicited on legume roots by certain Rhizobiwn andBradyrhizobium strains and mutants (Finan et al., 1985;Fransse n et al., 1987) and by Agrobacterium strains carrying theRhizobiwn meliloti nodgene s (Hirsch etal., 1985;Truche t et al., 1985).Th eexpressio n of theENOD 2gen e in theseempt y nodules strongly suggested a rolefo r theENOD 2earl ynoduli ni nth eformatio n ofth enodul e structurean dno ti nth e infection process. This conclusion has now been consolidated by our finding that the ENOD2gen ei s specifically expressed upon differentiation of the nodulemeriste m into innercortica lcells . Rootnodule sar eorgan swit h ahistologica lorganizatio n thati smarkedl y different from roots.Nevertheles s since theon e originates from theothe rthes e twoorgan s might share homologous tissues. Thus, the names nodule inner cortex and root inner cortex suggest thatthes e twotissue s areclosel y related.However , bydefinition , theroo t cortex isinwardl y delimited from thecentra l cylinderb y theendodermis .I n nodulesonl y what has hitherto been called the outer cortex has a similar position asth eroo t cortex andi s alsoconnecte d withi ta tth ebas eo f thenodule .I ncontrast ,wha tha sbee n hithertocalle d theinne rcorte xha sn opositiona lrelationshi p withth eroo tcortex :i ti slocate dinsid eth e nodule endodermis and around the vascular strands. In other plant parts, notably the stem,th ecorte x isalso ,b ydefinition , alwayslocate doutsid e thevascula r system and,t o our knowledge, never surrounding individual vascular strands. In addition, the morphology of thenodul e innercortica l cells distinguishes thistissu efro m root cortical tissues.Th enodul e inner cortical cells have fewer and smaller intercellular spaces than mostothe rcortica l cells (figure 2G,se eals oTjepkem a andYocum , 1974an dWitt yet

82 Fig. 3.Localizatio n of ENOD2 transcripts by in situ hybridization duringnodul edevelopmen ti npea .I nth e darkfiel dmicrograph s(B )an d(D) ,whic hcorrespon dt o thebrigh t field micrographs (A)an d (C),respectively , the autoradiographic signal is visible as whitegrains . A) Detail of a transection through an 8-day-old root showinga nodul eprimordiu m witha n apicalmeriste m (M). Thearro wpoint st oth epar to f theinfectio n thread that has grown through the root cortex to the primordium. A few inner cortical cells containing an autoradiographic signal of black silver grains are indicatedb yth elarg earrowhead .CC ,centra lcylinde ro f theroot .Ba rrepresent s5 0|im .B )Th eautoradiographi c signal over the inner cortical cells is indicated by the arrowhead.Q Transectionthroug ha roo twit ha 10-day - oldnodule .Th eautoradiographi c signalo f blacksilve r grains isvisibl eove r theinne rcorte x (IC).CC ,centra l cylinder of the root; RC, root cortex; M, apical meristem of the nodule; ES, early symbiotic growth zone of the nodule central tissue; OC, nodule outer cortex.Ba rrepresent s25 0um .

ai, 1986). Also at the molecular level the nodule inner cortex is different from the root cortex as we showed that at least one nodulin gene is specifically expressed during the formation of the nodule inner cortex. So both from an anatomical and a molecular point of view the name nodule inner cortex is misleading. Therefore we propose to consider this tissue as nodule parenchyma, while the nodule outer cortex can properly be described as nodule cortex. In determinate nodules the tissue that surrounds the vascular bundle connecting the nodule and the root central cylinder is morphologically very similar to the nodule parenchyma (see below). In addition, the ENOD2 gene is expressed in both tissues. Therefore we propose to consider also the tissue surrounding the connecting vascular bundle as nodule parenchyma. The determination of the nucleotide sequence of the cloned pea ENOD2 cDNA, and the amino acid sequence derived from it, allow a comparison with the structures of

83 thesoybea n andalfalf a ENOD2protein stha t havebee ndetermine dpreviousl y (Franssen ai, 1986).Als o atth emolecula r level thenodul e inner cortexi sdifferen t from theroo t cortex as we showed that at least one nodulin gene is specifically expressed during the formation of thenodul e innercortex . Sobot h from an anatomical and amolecula r point of view the name nodule inner cortex is misleading. Therefore we propose to consider this tissue as nodule parenchyma, while the nodule outer cortex can properly be described asnodul ecortex .I ndeterminat e nodules thetissu etha t surroundsth e vascular

Fig. 4. Localization of ENOD2 transcripts by in situ hybridization during nodule development insoybean .I n the dark field micrographs (B),(D )an d (F),whic h correspondt o thebrigh tfiel d micrographs(A) ,(C )an d(E) ,respectively ,th e autoradiographic signal is visible as white grains. A) ;•*$*#«:#>• '••^•Â- Transection through a six-day-old root with a nodule primordium (NP); CC, central vascular cylinder; RC, root cortex. Bar represents 250 urn.C )Detai l of the root in (A) showing the nodule primordium (NP)an dth e procambial strand (PC)betwee n theprimordiu m and thecentra l vascular cylindero fth eroot .Th earrowhea dindicate sa ninfectio nthrea d in thebasa l parto fa roo t haircell . Barrepresent s 50um . D) shows the autoradiographic signal over the newly developed tissue surrounding the procambial strand and over the developing innercortica l cellsi n thelatera l andbasa lpart so f the nodule primordium. E)Detai l ofa transectio n througha 10-day-oldroo t showinga nodul ewit h theprocambia l strand (PC)connectin g the nodule toth ecentra l vascular cylindera t oneo fth e xylem poles. An autoradiographic signal ofblac k silvergrain si svisibl eove rth einne rcorte x (IC)an dth etissu e surrounding theprocambia l strand. CT, central tissue ofth e nodule;OC ,nodul e outercortex ;F ,a grou po fphloe m fibers in the central vascular cylinder ofth e root; RC,roo t cortex. Barrepresent s 100um .

84 bundle connecting the nodule and the root central cylinder is morphologically very similar to the nodule parenchyma (see below). In addition, the ENOD2 gene is expressed inbot h tissues.Therefor e wepropos e toconside r alsoth e tissue surrounding theconnectin gvascula rbundl ea snodul eparenchyma . Thedeterminatio n of the nucleotide sequence of theclone d peaENOD 2cDNA , and the amino acid sequence derived from it, allow acompariso n with the structureso f thesoybea n andalfalf a ENOD2protein stha thav ebee ndetermine dpreviousl y (Franssen etal., 1987 ;Dickstei n etal., 1988 ,respectively) .Th epe aENOD 2protei n appearst ob e composed of the same two repeating pentapeptides as the soybean ENOD2protei n or variantso f these sequences withon e amino acidreplacement . However, whereasi n the soybean ENOD2protei n the repeating elements occur alternately, in the pea ENOD2 protein twoProProHisGluLy s repeats arealternate d withth eProProGluTyrGl n element. Thelatte rorganizatio n alsooccur si n thealfalf a ENOD2polypeptide ,i nwhic h thesam e pentapeptides are present. This difference in structure between the soybean and pea/alfalfa ENOD2proteins ,respectively , suggeststha t theamin oaci dcompositio n of thepentapeptide smigh tb eth emai nrequiremen t for thefunctio n of theENOD 2protein . A specific organization of the repeating elements seems less essential. The different distribution of the two pentapeptides in soybean and pea/alfalfa, respectively, might indicate thatindependen t duplication eventsinvolvin gdifferen t basicpolypeptid e units gave rise to the different ENOD2 genes during the evolution of these legumes. However, more sequence data from a wider variety of legumes will be needed to substantiate thishypothesis . The amino acid sequence of both the pea and soybean ENOD2protei n strongly resembles therecentl y described soybean protein 1A10tha t occurs in cell walls of the axis tissue of germinating soybean seeds (Averyhart-Fullard et al., 1988). This glycoprotein consistso f atleas t 40repeatin g ProProValTyrLys units and about 50%o f theproline sar e hydroxylated tohydroxyproline . Becauseo f thissimilarit y in structurei t is very likely that also ENOD2 is a (hydroxy)proline-rich cell wall protein. Together with thecarro t P33protein , thelAlO andENOD 2protein s seemt ofor m ane w classo f cellwal lprotein s that arecompose d of pentapeptides containing twoprolines .Thei rlo w Serconten t forms amajo r difference with another important group of hydroxyproline- rich cellwa lproteins ,th eextensins ,whic h arecharacterize d by (Hyp)4Ser-pentapeptide repeats (Cassab and Varner, 1988). Sequence analysis of two soybean ENOD2 genes revealingtha ta putativ e signalpeptid ei spresen ta tth eN-terminu so fth eENOD 2protei n lends further support to the hypothesis that ENOD2 represents a cell wall protein (Franssen etal., 1988). Thenodul eparenchym a ("innercortex" ) appears to bea n important tissuei nth e

85 Rhizobiwn-lcgumç symbiosis.Th efre e oxygen concentration in anodul e showsa sharp decline acrossth enodul eparenchym a toa ver y lowvalu ei nth ecentra ltissue, whic h isa necessity toprotec t the extremely oxygen-sensitive nitrogen-fixing enzyme nitrogenase (Tjepkema andYocum , 1974;Witt yet al., 1986) .I twa sshow n thatthi sdeclin emus tb e due to a high consumption rate of oxygen by the rhizobia in the infected cells of the central tissue combined with a diffusion barrier residing in the nodule parenchyma (Witty etal., 1986).A s oxygen diffusion through air is approximately 10^time s faster than through water, iti sver y likely that in nodulesoxyge n diffusion occursthroug h the intercellular spaces.A smentione d above,th enodul eparenchym acontain srelativel y few and small intercellular spaces. In contrast, in both ("outer") cortex and central tissue relatively wide intercellular spaces occur. By this specific morphology the nodule parenchyma will be able to form the oxygen diffusion barrier (Tjepkema and Yocum, 1974; Witty et ai, 1986). Since the differentiation of the cell wall will be a factor in determining tissuemorphology , wepropos etha tth eputativ ecel l wallprotei n ENOD2i s contributing tothi sspecia lmorpholog y of thenodul eparenchyma . In soybean the ENOD2 gene appears also to be expressed in the cells that surround thevascula r strandconnectin g the nodulewit h thecentra lcylinde ro f theroot . In pea such a long connecting vascular bundle is lacking, since here the nodule originates more closely to the central cylinder. The cells surrounding the connecting vascular strandar emorphologicall y similar toth enodul eparenchym a of the nodule,i.e . they have relatively few and small intercellular spaces.Thi s isconsisten t with the idea that theENOD 2 gene product can contribute tocel l morphology. There are, however, no experimental data indicating that this tissue has a function similar to the nodule parenchyma that surrounds thecentra ltissue .

MATERIALS AND METHODS

Growth conditionsfor plants

Soybean plants (Glycine max (L.) Men.cv .Williams ) and peaplant s (Pisum sativum (L)cv . Rondo) wereculture d asdescribe d before (Franssen etal., 1987;Bisselin get al., 1978) .A tth etim eo f sowingth esoybea nseed swer einoculate dwit hBradyrhizobium japonicum USDA11 0an dth epe aseed s werewit hRhizobium leguminosarum biovar.viciae PRE .

86 Isolation and sequencing ofpPsENOD2

A Xgtl1 cDN A library against RNA from root nodules of Pisum sativum (L.) cv. Sparkle was kindly provided by dr. G. Corruzi (Tigney et al., 1987).Nitrocellulos e replicas from plates containing 2000 plaques were made using standard procedures (Maniatis et al, 1982). The plaques were screened with nick translated (Maniatis elal., 1982) insert from the soybean cDNA clonepGmENOD 2 (Franssen et al., 1987).Phag e DNA purification, insert isolation and cloning in pUC18 was according to standard procedures (Maniatis et al., 1982). Both strands of the pPsENOD2 insert were sequenced using the chemical degradation method (Maxam andGilbert , 1980).

In situ hybridization

The in situ hybridizations were performed essentially as described by Cox & Goldberg (1988). Nodules were fixed with 3%paraformaldehyd e and 0.25% glutaraldehyde in 0.1 M sodium phosphate buffer pH 7.2 atroo m temperature (RT),dehydrate d ingrade d ethanol and xyleneserie san d embedded in paraplast. Sections, 7 \tm thick, were attached to poly-L-lysine-coated slides. Sections were deparaffinized with xylene and rehydrated through a graded ethanol series. They were subsequently pretreated with 1u.g/m l proteinase K in 200 mM TrisHCl pH 7.5, 2 mM CaCl2 at 37°C for 30 min and with 0.25% acetic anhydride in 0.1 M triethanolamine pH 8.0 at RT for 10 min, dehydrated in a graded ethanol seriesan d air-dried. Sections were hybridized with (anti)sense RNA probes, which were madeb y transcribing pT7 clones (a kind gift of dr. S. Tabor) containing the inserts of pPsENOD2 and pGmENOD2. The antisense RNA probes were radioactively labeled with 35S-UTP (1000-1500 Ci/mmole, NEN). The probes were partially degraded to a length of 150 nucleotides by heating at 65°C in 0.2 M Na2CO3/0.2 M NaHC03. Sections were hybridized with RNA probes in 50% formamide, 0.3 M NaCl, 10m M TrisHCl pH 7.5, 1m M EDTA, 10%dextransulfate , 1x Denhardt's, 70 mM DTT at 42°C for 16 hours. After washing three times in 4 x SSC, 5 mM DTT at RT slides were treated with 20 Hg/ml RNase A in 0.5 M NaCl, 10m M Tris/HCl pH 7.5, 5 mM EDTA at 37°C for 30 min and washed in the same buffer with 5 mM DTT at 37°C for 30 min. The final wash consisted of two times 2 x SSC, ImM DTTa t RT. Südes were dehydrated in graded ethanols (each with 300 mMammoniumacetate ) and 100%ethanol .Afte r air- drying,slide s werecoate d with KodakNTB 2nuclea remulsio n 1:1 diluted with 600 mM ammoniumacetate and exposed for one to three weeks at 4°C. They were developed in Kodak D19 developer for three minutes and fixed in Kodak Fix. Sections were stained with 0.025% toluidine blue 0 for 5 min and mounted with DPX. For embedding in glycolmethacrylateresi n nodules were fixed with 2.5%glutaraldehyd e in 0.1 M sodiumphosphate buffer, pH 7.2 for three hours. After dehydration in a graded ethanol series the

87 nodules were embedded in Technovit resin according to the manufacturer's instructions (Kulzer, Friedrichsdorf,FRG) .Sectio n of 4 pm thickness werestaine d with 1%toluidin e0 blu efo r 1mi nan d mountedwit hEuparal . Sections were photographed with a Nikon microscope equipped with dark field and epipolarizationoptics .

REFERENCES

Averyhart-Fullard,V. ,Datta ,K. ,an dMarcus ,A. , 1988,Proc. Natl. Acad. Sci.USA, 85:1082 . Bergersen,F J. , 1982,Roo tNodule so fLegumes :Structur ean dFunction .Researc h StudiesPress ,Ne w York. Bisseling,T. ,Va nde n Bos, R.C.an dVa n Kammen,A. , 1978,BiochimMophys. Acta,539 :1 . Cassab,G.J. , andVarner ,J.E. , 1988,Ann. Rev. Plant Physiol. Plant Mol. Biol., 39: 321. Cox,K.H. ,an dGoldberg ,R.B. , 1988,Analysi s ofplan tgen eexpression , in:Plan t Molecular Biology: Apractica lapproach ,C.H .Shaw ,ed. ,IR LPress , Oxford. Dickstein,R. , Bisseling,T. ,Reinhold ,V.N . andAusubel ,M. , 1988,Genes & Development, 2: 677. Finan, J.M., Hirsch, A.M., Leigh, J.A., Johansen, E., Kuldau, G.A., Deegan, S., Walker, G.C., and Signer,E. , 1985,Cell, 40:869 . Franssen, H.J., Nap,J.-P. , Gloudemans, T., Stiekema, W., Van Dam, H., Govers, F., Louwerse, J., VanKammen ,A. ,an dBisseling ,T. , 1987, Proc. Natl. Acad. Sei.USA, 84: 4495. Franssen, H.J., Scheres, B., Van de Wiel, C, and Bisseling, T., 1988,Characterizatio n of soybean (hydroxy)proline-rich early nodulins, in: Molecular Genetics of Plant-Microbe Interactions, R. Palaciosan d DP.S. Verma,eds. ,AP SPress ,St .Paul . Govers,F. ,Franssen , HJ., Pieterse,C , Wilmer, J., and Bisseling, T., 1990,Functio n and regulation of the early nodulin gene ENOD2, in: Genetic Engineering of Crop Plants, P. Grierson and G. Lycett,eds. ,Butterworth sScientific , Guildford. Govers,F. ,Moerman , M., Downie,J.A. , Hooykaas,P. , Franssen, HJ., Louwerse,J. , VanKammen , A.,an dBisseling ,T. , 1986,Nature, 323:564 . Hirsch,A.M. ,Drake ,D. ,Jacobs ,J.W. ,an dLong ,S.R. , 1985,J. Bacteriol.,16\: 223. Maniatis, T., Fritsch, E.F., and Sambrook, J., 1982, Molecular cloning. Alaborator y manual, Cold SpringHarbo rLaborator y Press,Col dSprin gHarbor ,NY . Maxam,A.M. ,an dGilbert ,W. , 1980,Methods Enzymoi, 65: 499. Newcomb, W., 1981,Int. Rev. Cytol. Suppl., 13:247 . Robertson,J.G. , Wells,B. ,Bisseling ,T. , Farnden, K.J.F., and Johnston, A.W.B., 1984,Nature, 311:

88 254. Sprent,J.I. , 1980,Plant Cell Env., 3: 35. Tigney, S.V., Walker,E.L . and Coruzzi,G.M. , 1987,EMBO J.,6 :1 . Tjepkema,J.D. ,an dYocum ,CS. , 1974, Planta, 119: 351. Truchet,G. , Debellé,F. , Vasse,J. , Terzaghi, B.,Gamerone , A.-M., Rosenberg.C, Batut,J. , Maillet, F.,an dDénarié ,J. , 1985,/ . Bacterioi,164 :1200 . Vande nBosch ,K.A. ,an dNewcomb ,E.H. , 1986, Planta, 167: 425. Vande nBosch ,K.A. ,an dNewcomb ,E.H. , 1988,Planta, 175:442. Van Kammen,A. , 1984,Plant Mol. Biol. Rep., 2: 43. Verma, D.P.S., Fortin, M.G., Stanley, J., Mauro, V.P., Purohit, S., and Morisson, N., 1986,Plant Mol.Biol., 7 : 51. Witty, J.F., Minchin, F.R., Sk0t, L., and Sheehy, J.E., 1986, Oxford Surv.Plant Mol. Cell Biol, 3: 275.

89 CHAPTER 6

Nodulin gene expression and ENOD2 localization in effective, nitrogen- fixing and ineffective, bacteria-free nodules of alfalfa (Medicago sativa).

Clemensva n deWiel ,Joann aNoms ,Birgi tBochenek ,Rebecc a Dickstein, Ton Bisseling,an dAn nHirsch . SUMMARY

Alfalfa plants form bacteria-free root nodules inrespons e toa numbe r of agents, includingRhizobium meliloti exo mutants ,Agrobacterium tumefaciens transconjugants carrying cloned R. meliloti nodulation genes, and compounds that function as auxin transport inhibitors, e.g. N-(l-naphthyl)phtalamic acid (NPA) or 2,3,5-triiodobenzoic acid (TIBA). These bacteria-free nodules contain transcripts for the early nodulin MsENOD2. In situ hybridization studies demonstrate that ENOD2 transcripts are localized inparenchym a cellsa tth ebas ean dalon gth eperipher y ofwild-typ eR. meliloti- induced, nitrogen-fixing alfalfa nodules. The ENOD2 gene is also expressed in a comparabletissue-specific manne ri nnodule selicite d byNP A andTIBA .I n bacteria-free nodulesinduce d byR. melilotiexo mutant s andA. tumefaciens transconjugants carrying either one or both R. meliloti symbiotic plasmids, ENOD2 transcripts are usually restricted inthei r localization toparenchym a cells atth ebas eo fth enodule .O nth ebasi s of the spatial pattern of ENOD2 gene expression, we conclude that the developmental pathway of bacteria-free nodules, whether bacterially or chemically induced, is comparable to that of wild-type R. meliloti-mducednodules , and furthermore, that the auxin transport inhibitors in their action mimic some factor(s) that trigger nodule development

INTRODUCTION

Legume root nodule development represents an excellent model system for investigating thequestio n of how differentiated plantcells ,i n thiscas eroo tcortica lcells , can beinduce d tochang e theirdevelopmenta l destiny toinitiat e ane w structure, in this caseth enodule .Roo t noduledevelopmen t in alfalfa (Medicago sativa)begin s following specific recognition between the host plant and Rhizobium meliloti. Cells of the inner cortical layers of the root divide anticlinally, thus initiating a nodule primordium. This nodule primordium is soon invaded by aninfectio n thread that originated in an infected root hair. Subsequently, rhizobia arerelease d from branches of theinfectio n thread into cellso f the central part of the noduleprimordium . Shortly thereafter, peripheral cellso f the nodule primordium start to differentiate into the peripheral tissues of the nodule, including thevascula r strands connecting the nodule toth eroo t vascular system.A t the

93 same time, the persistent nodule meristem, consisting of small, densely cytoplasmic, actively dividing cells,i s organized atth e apical (distal) endo f the nodule primordium. Thismeriste mprovide sth enodul ewit hne wcell stha tdifferentiat e inth e afore-mentioned noduletissues ,thu senablin g thenodul et ofollo w anindeterminat e growthpattern . Alfalfa plants form bacteria-free root nodules in response to auxin transport inhibitors (ATI's),suc h asN-(l-naphtyl)phthalami c acid (NPA) and 2,3,5-triiodobenzoic acid (TTBA) (Hirsch et al., 1989). These nodules show a histological organization resembling thato f nodules induced byRhizobium meliloti. In theATI-induce d nodulesa peripheral tissue and a central tissue, the latter consisting of cells rich in plastids with prominent starch grains, can be distinguished. A narrow zone of meristematic cells is located atth e distal end of thenodule .A t theproxima l end,vascula r tissueconnect s the nodule to the root stele. The ATI-induced nodules contain transcripts for the nodulins Nms-30an dMsENOD2 .However , transcriptsfo r latenodulins , such as leghemoglobin, areno tfoun d in theATI-induce d nodules (Hirschet ai, 1989).Thus ,i n severalrespects , ATI-induced nodulesresembl e the bacteria-free ("empty") noduleselicite d on alfalfa by certain bacterial strains, such as R. meliloti mutants defective in exopolysaccharide synthesis {exo) (Finan et al., 1985; Leigh et al., 1987; Dickstein et al., Noms et dl., 1988),o rA. tumefaciens transconjugants carrying clonedR. melilotinod gene s (Wonget al., 1983; Hirsch etai, 1984;Truche t etal, 1984;Hirsch.e f ai, 1985;Dickstei n etal., 1988). Because (early) nodulin genes areexpresse d inATI-induce d structures aswel l as in "empty" nodules formed in response to R. meliloti exo mutants or A. tumefaciens transconjugants carryingR. melilotinodulatio n sequences,i twa sconclude d thatth eATI - induced nodules truly represent nodules comparable to those induced byR. meliloti.I t wasals o suggested that theATI' s mimic the activity of therhizobia l nodgenes ,perhap s byperturbin g theroot' sendogenou s auxin/cytokinin balance (Hirsch etai, 1989). Recently, in situ hybridization studies have shown that the ENOD2 gene is exclusively expressed in the noduleparenchym a (the inner part of theperiphera l tissue, formerly called "innercortex" ) of soybean aswel l aspe a (Pisum sativum)nodule s (Van de Wiel et al., 1990). The nodule parenchyma is located between the central tissue of infected and uninfected cells, and the nodule cortex (the outer part of the peripheral tissue, formerly called "outer cortex"). The nodule parenchyma consists of highly vacuolated cells that are more densely packed, with fewer intercellular space, than the nodulecortex . In thisrepor t we have determined whether in ATI-induced nodules theENOD 2 gene is also expressed in a tissue-specific manner comparable to that in wild-type R. meliloti-induced nodules. Such a comparison is of significance, because the growth

94 pattern of theATI-induce d structures deviatesfro m the "normal" growthpatter n of root nodules in several ways. Their meristematic activity is frequently spread out over a relatively largepar t of the distal end of the nodule instead of being focused toa limited area of the apex as in wild-type nodules. Also, vascular traces are confined to the proximal part of the nodule and do not separate into distinct strands extending distally into the peripheral tissue of the nodule (Hirsch et al., 1989). However, even among bacterial-induced "empty"nodule sther ei s somevariatio n in growthpatter n (cf. Wonge t al., 1983;Hirsc h et al., 1984,1985; Finan et al., 1985).Therefore , weinclude d "empty" noduleselicite d bydifferen t bacterial strains inthi s study.Ou r goalwa st odetermin e the spatialpatter no fENOD 2gen eexpressio n inATI-induce d nodulesa swel la si nbacterial - induced "empty" nodules to test the proposition that the ATI-induced nodules are comparable to "true"nodules . In situ hybridization analysis, utilizing an alfalfa ENOD2 probe, was used to determine whereENOD 2transcrip t is localized inATI-induce d nodules.W e compared thislocalizatio n withtha ti nnodule sforme d following infection with wildtypeR. meliloti strains as well as several R. meliloti exo mutants and A. tumefaciens transconjugants carryingR. melilotimegaplasmids . Because some of the bacterial-induced nodules had not been analysed before, weals ochecke d nodulin geneexpressio n on two-dimensional gelso f invitro translated noduleRN A andexamine d RNAtransfe r blotshybridize d toa n alfalfa ENOD2insert .

RESULTS

Noduledevelopment and nodulin gene expression

Thepresen t study compares ATI-induced nodules with "empty" nodules formed after inoculation with R. meliloti exoA, exoB or exoY mutants or A. tumefaciens transconjugants carryingR. meliloti megaplasmids.A sth enodule selicite d bymos to f the strains usedi n thepresen t study have not been described before, they will be compared herewit hth e "empty"nodule salread y published. Previous analyses of in vitro translated RNA on two-dimensional .gels have shown that about2 0differen t nodulin translation productsca n beidentifie d in wild-type R. meliloti-induced alfalfa nodules (Dickstein et al., 1988;Nom se t al., 1988;cf . Figures 1A and B).I n nodules formed in response toR. melilotiexo mutants orA. tumefaciens

95 PT "— 69-I& •y^j** 1« S- 30- •«•

4»

0 14- %(S )

10 pH 3 10

96 transconjugants, only the nodulin translation product Nms30 can be detected (Dickstein et al., 1988; Norris et al., 1988; the latter describe our exoB mutant). In addition, a nodulin of variable isoforms, Nms25, that appears to be unique to exo mutant-induced nodules, has been described by Leigh et al. (1987). Figures IC and F illustrate that the "empty" nodules used in the present study conform to this pattern of gene expression. Nms30 as well as Nms25 translation products are detectable in nodules elicited by exoA and exoF mutants. On the other hand, nodules formed in response to A128 inoculation appear to contain only the Nms30 transcript (Figure ID). Thus, the pattern of nodulin gene expression is remarkably similar in all "empty" nodules, except for the nodulin Nms25, which apparently occurs only in R. meliloti exo mutant-induced nodules.

Localization ofENOD2 transcripts in nodules

Sections of the various alfalfa nodules were hybridized to antisense RNA transcribed from pA2ENOD2, which contains a 0.3 kb insert of alfalfa ENOD2 cDNA. The 4-week-old wild type Rm2011-induced nodule shows ahybridizatio n pattern like the one described for pea (cf. Van de Wiel et al., 1990). The ENOD2 mRNA is present in parenchyma cells, constituting the nodule parenchyma, surrounding the central tissue of infected and uninfected cells. There is no ENOD2 transcript detectable at the top of the nodule, at the site of the persistent meristem. No transcript is found within the vascular strands traversing the nodule parenchyma (Figures 2A,B). Four-week-old NPA- and TIBA-induced nodules are more or less similar to each other. Unlike a nodule elicited by wild type R. meliloti, the meristematic region commonly extends along the major part of the periphery of the nodule. A tissue

Fig. 1. Two-dimensional Polyacrylamide gel analysis of in vitrotranslation s of total RNA isolated from alfalfa tissue.A ) From root tissue.Th eope n arrow (in all panels)point s toglutamin esynthetas e (see Norris et al., 1988, for identification of this translation product as GS). The circles indicate prominent translation products that are evident following in vitro translations of RNA isolated from infected rootnodules ,bu twhic har eabsen t inroo tRNA ,namel yth eleghemoglobins .B )Fro mnitrogen - fixing nodules induced by wild typeR. meliloti strain Rml021. The boxed-in translation product isa nodule-specific/enhanced glutamine synthetase.Th edar k arrowhead points toNms-30 ,an d thearrow s indicate leghemoglobins. C) From bacteria-free nodules induced by Rm7055(«oF::Tn5). Nms-30 is present(arrowhead )a swel la sNms-2 5(tande marrows) ,bu tleghemoglobi n translationproduct s(circles ) areabsen t D)Fro mbacteria-fre e nodulesinduce db yA12 8 (Agrobacterium tumefacienscarryin gRml02 1 pSyma).Nms-3 0 is present (arrowhead), but leghemoglobin translation products (circles) are not. E) From bacteria-free nodules induced by Rm7061(e;coA::Tn5). The open arrow points to glutamine synthetase. Nms-30 is present (arrowhead) as well as Nms-25 (tandem arrows), but leghemoglobin translationproduct s(circles )ar eno t

97 98 consistingo fvacuolate d cellscontainin g amyloplastsca n bedelimite d inth ecente ro f the nodule. Around this central tissue a relatively narrow zone consisting of highly vacuolated, densely packed cells, reminiscent of the nodule parenchyma, contains transcripts that hybridize toth eENOD 2antisens e RNA.Outsid e this zone ales sdensel y packed ("outer")corte x wheren oENOD 2transcrip t can bedetected , ispresen t (Figures 2C,D). Figure2 E showsa nexampl eo f afour-week-ol d exoBmutant-induce d nodule.A s in wild-typeR. meliloti-mducednodules ,i ti spossibl e todistinguis h anapicall y localized persistent meristem, a peripheral zone and a central tissue, which, however, is free of intracellularbacteria .ENOD 2transcrip t isfoun d inth edensel y packed,highl y vacuolated parenchymatous tissue thatconstitute s theinne rpar t of theperiphera l tissue (the nodule paremchyma) (Figure 2E).Thi s parenchymatous tissue is traversed by avascula r strand thati sfre e ofENOD 2transcrip t (datano tshown) .Th eoute rpar to fth eperiphera ltissue , which ismor eloosel ypacked , isals ofre e ofENOD 2transcrip t (Figures2E) . Themajorit y of thenodule s induced by theexoA, exoB andexoF mutant s areno t of theelongate , wild type-like sorto f nodule.The y aresmal l and broad and meristematic activity, which issprea d outalon g arelativel y broadregio no f the subtendingroot , leads to the deposition of a limited amount of tissue in aplan eperpendicula r to theroo t axis. Consequently,littl eperiphera l andcentra l tissue isforme d distally inth enodul e (Figures 3A,B). ENOD2 transcripts are present in a zone of densely packed, highly vacuolated cellsextendin g along the baseo f the nodule (Figures 3A,B).Als o located basally in the nodule,severa lvascula r traces,connecte d toth eroo t stele,ar epresen t thatd ono tcontai n

Fig. 2. Detection of MsENOD2 transcript in alfalfa nodules by in situ hybridization. Bar = 100 um. A) Dark field photograph of a nitrogen-fixing nodule. MsENOD2 transcripts as visualized by the brightly reflecting silver grains (see Methods) are localized inparenchym a cells along theperipher y of the nodule surrounding the vascular bundle (vb) and also along the base of the nodule. The persistent meristem (m) is devoid of MsENOD2transcript . Median longitudinal section. B) Bright field photograph of a nitrogen- fixing nodule cut obliquely. The persistent meristem is out of the plane of section. MsENOD2 transcripts were localized on paraffin sections using the Boehringer-Mannheim RNA Labeling (Genius) kit (see Methods). A dark stain is present in parenchyma cells along the edge of the central tissue, including at the base of the nodule.Th e stain that appears tob e in the root (r) is actually at the base of a second lobe of a two-lobed nodule. The second lobe is not observed in this section. C) Dark field photograph and D) bright field photograph of a cluster of nodules elicited two weeks after addition of NPA. MsENOD2transcript s (brightly reflecting silver grains in C)ar e detected along theperipher y of the nodule, but not in the root (r). Slightly oblique section. E) Dark field photograph of a five-week-old, bacteria-free nodule elicited by Rm5078(exoB::Tn5). The meristem (m) is devoid of MsENOD2 mRNA as is the outer part of the peripheral tissue. MsENOD2 transcripts extend along the periphery of the nodule and are also localized in parenchyma cells at the base of the nodule. The vascular bundle of the root stele (vb, observed in transversal section) is also devoid of MsENOD2 transcript. Non- median longitudinal section. F) Bright field photograph of a. bacteria-free nodule induced by Rm7061(eJcoA::Tn5). The nodule has been sectioned transversely (see Fig. 3A for approximate location of thesection) .Severa l vascularbundle s (vb) areevident ; these aresurrounde d by parenchyma cells which express MsENOD2.Th e outerpar t of the peripheral tissue is devoid of MsENOD2 mRNA.

99 ;$•*• ne" *'::.^}$i&*h . tfWÊTS,

*-%»

* ,..• v.'-.vi *A. ..*'' <\&

Fig. 3. Detection of MsEN0D2 transcript in alfalfa nodules by in situ hybridization. Bar = 100 urn. (A)Brigh t field photograph and (B) darkfield photograp h of a three-week-old,bacteria-fre e nodule induced by Rm5078(e;coB::Tn5). The presumed meristematic region (m) is very broad and lacks MsENOD2 mRNA. MsENOD2 transcripts aredetecte d basally in thenodule , in parenchyma tissue that surrounds the vascular bundles (vb).Th e arrows denote the approximate location of the transverse section illustrated in Fig. 2F. Near median longitudinal section; nc: outer peripheral tissue of nodule. (C) Dark field photograph and (D) bright field photograph of a bacteria-free nodule induced by A135. MsENOD2 transcripts are localized inparenchym a cellsaroun d a vascular bundle (vb)o f the nodule. Anodul e outer peripheral tissue (nc) is observed, but cells typical of a meristematic region are not identifiable in this nodule. (E) Bright field photograph and (F) dark field photograph of a bacteria-free nodule induced by A128. The nodule is sectioned obliquely. MsENOD2 mRNA is detected basally in the nodule surrounding the vascular bundles (vb).Dividin g cells typical of a meristematic region are not observed in this nodule,bu t the loosely packed outerperiphera l tissue (nc) is present.

100 ENOD2 transcript. A more distally located tissue of the nodule, consisting of more cytoplasmic-richcell scontainin g amyloplasts,i sals ofre e ofENOD 2transcrip t (Figures 3A.B). The noduleselicite d by theA .tumefaciens transconjugants A128 (Figures 3E,F) and A135 (Figures 3C,D) are quite similar to the small and broad exo mutant-induced nodulesdescribe d above.Ther ei sa limite d amounto fmeristemati c activityove r abroa d zoneo f thesubtendin groot .Proximally , severalvascula r tracesar econnecte d toth eroo t stele (Figure 3E,F). ENOD2 transcripts are clearly detectable in a zone of densely packed, vacuolated parenchymatous tissue, containing amyloplasts, located basally in these nodules and/or around the vascular traces. The distally located tissue of these nodules, which is highly vacuolated and free of amyloplasts, contains no ENOD2 transcript (Figures 3CJDan dE,F) .

DISCUSSION

Our in situ hybridization experiments clearly show that the ENOD2 gene is expressed in the nodule parenchyma of nitrogen-fixing alfalfa nodules, following a pattern in agreement with the recent observations of pea nodules (Van de Wiel et al., 1990).Moreover , wefind tha t theENOD 2gen ei sexpresse d in atissue-specifi c manner innodule selicite d byth eATIs ,NP A andTIBA .ENOD 2transcript s arepresen t in azon e ofdensel ypacked , highly vacuolated parenchyma cells between acentra l tissueo fcells , rich in amyloplasts, anda noute rlaye ro fmor eloosel ypacke d cortical tissue.N oENOD 2 mRNAwa sdetecte d inth edista len do f thenodule ,th esit eo f meristematic activity.Th e position andmorpholog y of theENOD 2gene-expressin g zonecorrespon d tothos eo fth e nodule parenchyma of wild-type R. meliloti-induced nodules. Hence, by using the criterion of tissue-specific ENOD2gen eexpression , onema yconclud e that the "empty" ATI-elicited nodulesar eindee dcomparabl e tonodule sinduce d bywil dtyp eR. meliloti. The R. meliloti exo mutants and A. tumefaciens transconjugants induce a heterogeneous population of nodules (cf. Wong et al., 1983; Hirsch et al., 1984, 1985; Finane t al., 1985).Relativel y few areo f theelongat e typetha t mostresemble s thewild - typeR. meliloti-inducednodule . In such nodules, the presence of ENOD2 transcripts wasrestricte d toa tissu e that, bothpositionall y andmorphologically , iscomparabl e toth e nodule parenchyma of wild-type nodules (cf. Figures 2E and Figures 2A,B). The vast majority ofnodules ,however , areo f asmal l and broad typei n which littleperiphera lan d

101 central tissue is deposited perpendicularly to the long axis of the root. Nevertheless, ENOD2gen eexpressio n isrestricte d toa zoneo fdensel y packed parenchymatous tissue extending along thebas eo f the nodule and around thevasula r traces (cf. Figure 3).Th e locationo f thiszon ei sth e samea swher eENOD 2gene-expressin g noduleparenchym a is present in wild-typeR. meliloti-induced nodules.Thus ,i n spiteo f the anomalies related to the diffuse and limited meristematic activity of these "empty" nodules, thepatter n of ENOD2 gene expression does not differ fundamentally from that observed in wild type R. me/j/ori-elicited nodules. From the above-described examplesi t becomesapparen t thatlocalizatio n studies of mRNAs that mark specific nodule tissues may provide insight into the nature of aberrant root nodules. Although distinct tissues, like nodule parenchyma, should be recognizableb yth ecombinatio n ofpositiona l andcytologica lcriteria , suchdistinction si n tissue types are relatively difficult to assess by microscopical studies alone in "empty" nodules of the small and broad type described above. In the cases where a clearly differentiated nodule endodermis (cf. Truchet et al.,1984; Finan et al., 1985) is detectable,i ti srelativel y easy todelimi t noduleparenchym acell sfro m themor eloosel y packed ("outer") cortex.However , using the amount of intercellular space as a marker, delimitation from abacteria-fre e centraltissu ei sfrequentl y moredifficult , becausethes e possible differences in the extent of intercellular space between the two tissues are not easy todiscern .Furthermore , additional morphological criteria are hard toevaluate ; the noduleparenchym a cellsar eofte n rich in amyloplasts,lik eth ecentra l tissueitself .Also , the extent of vacuolation in thecentra l tissuei svariabl e in the nodules of the small and broad type. In these cases, localization studies of ENOD2 messenger are helpful. Such analysisma y alsob eimportan tfo r understanding thedevelopmenta l organization of even morepeculia r types of nodules, such asthos eelicite d onclove rb yA. tumefaciens or R. leguminosarum bv. trifolii transconjugants carrying R. meliloti nodulation sequences (Hirsch et al., 1985;Truche t et al., 1985),o r the oneselicite do n alfalfa by aR. meliloti with a Tn5 insertion 3 kb downstream from nodC (Dudley et al., 1987). Unlike other "empty" nodules, these pseudonodules more closely resemble lateral roots in that they have centrally located vascular tissue. In contrast to normal lateral roots, however, one such type of nodule has a nodule endodermis-like layer in its "cortex" (Dudley et al., 1987).Localizatio n ofENOD 2transcript s mightindicat e whether atissue comparabl e to the nodule parenchyma ispresen t in thesepseudonodule s and,i f so,wher ei t is located. Itthe n may bededuce d howfa r thesepseudonodule s aredevelopmentall y comparable to bonafide rootnodules . Thus,base d on theobservatio n that theATI sinduc e anorma l pattern ofENOD 2 gene expression, weconclud e that in their action theATI s mimic somebacteria l factors

102 responsible for nodule development. The ATIs are known to cause alterations in the plant's endogenous hormone balance.Cel l divisions, followed by noduleformatio n with concomitant tissue-specific expression of the ENOD2 gene,thu s may be related to the hormonal statuso f theroot .Recently ,LeRoug ee t al.(1990 )hav eidentifie d aR. meliloti- produced molecule, NodRm-1, which elicits alfalfa root hair deformation -one of the earliest stepsi n noduleformation- , asa ß-sulfate d tetraglucosamine with acetyl andacy l substitutions. The ATIs, which are often substituted benzoic acids, are structurally different from NodRm-1,a noligosaccharide . Thus,i ti sunlikel y thatbot hac ta tth esam e site.I tis ,however ,conceivabl e thatth eATI sinterac t with aconsecutiv e site somewhere in the cascade of events normally triggered by NodRm-1, resulting in nodule development.

METHODS

Plantand bacterial material

Alfalfa (Medicago sativa L. cv. Iroquois) plants were grown as described by Noms et al. (1988). At the time of sowing, individual bins containing alfalfa seeds were left uninoculated or inoculated with either Rhizobium meliloti wild type strains Rm2011 or Rml021, R. meliloti mutant in exoh (Rm5078), exoA (Rm7061) or exoF (Rm7055), or Agrobacterium tumefaciens transconjugants, either strain A128, which harbors the nodlnif symbiotic megaplasmid (pSyma), or strain A135, which carries pSyma and the megaplasmid bearing exo genes (pSymb) (Leigh et al., 1985;Fina n et al., 1986). Plants were treated with the ATIsa s described previously (Hirsch et al., 1989).

RNA analysis

Tissue was harvested, frozen in liquid nitrogen, and stored at -80oC until use. Total RNA was isolated from nodule or root tissue,in vitro translations performed, and the translation products separated by two-dimensional electrophoresis as described byNorri se t al.(1988) .

103 Insitu hybridization

Tissueswer efixed eithe ra sdescribe db yVa nd eWie le tal .(1990 )o ri n4 %glutaraldehyde , 1,5 % paraformaldehyde in 0.1 Mphosphat e buffer, pH 7.2. Following fixation, the tissues were rinsed twicei nbuffe r anddehydrate di na grade dethano lserie s untilS O %ethanol . Thetissu ewa stransferre d to the first tertiary butyl alcohol step, dehydrated in the tertiary butyl alcohol series (Sass, 1953),an d embeddedi nParaplast .Sections ,7 o r8 |i mthick ,wer eaffixe d topoly-L-lysine-coate dslides .Th ein situ hybridizationsusin g(35S)UTP-labele dantisens eRN Aprobe swer ebase do na protoco lpublishe db yCo x andGoldber g(1988 )an dwer eperforme d essentiallya sdescribe db yVa nd eWie le tal .(1990) . Following development of the emulsion, sections were stained either with 1% aqueou s safranin or with 0.05% toluidineblue . Alternatively, the RNA Labeling (Genius) kit using digoxigenin-labeled UTP (Boehringer Mannheim)wa sutilized . Sensean dantisens eprobe swer emad eaccordin gt oth emanufacturer' sdirection s andadde dt odeparaffinize d sections,whic hha dbee npretreate da sdescribe dfo rradioactiv eprobes . Some noduleswer efixed overnigh ti nformaldehyde-aceti c acid-alcohol(FAA )(Sass ,1953) ,rinse dovernigh ti n 50 %ethanol , slowly rehydrated, and then frozen on a block in liquid nitrogen. The nodules were sectioned at 16p m using a cryostat and the sections were placed on poly-L-lysine-coated slides.Th e sectionswer egraduall ywarme dt oroo mtemperatur ean dfurthe r treateda s forradioactiv eprobes .10-10 0 ngo fprobe/m lhybridizatio nbuffer/slid e wasadde dt oth ecryostat-sectione d material. The manufacturer's protocol for pre-hybridization, hybridization and post-hybridization was followed for both sensean d antisense probes; sections were hybridized at 37oC overnight. MsENOD2transcript s were detected immunologicallyaccordin gt oth emanufacturer' s directions. Forparaffin-embedde d material,12-7 2hour s wererequired fo r maximal colordevelopment , whilefo r cryostat-sectioned material,colo rdevelopmen t generallyoccurre dwithi n6 0minutes . Thesection swer eno tcounterstained . Following dehydration through an ethanol seriers, the sections werte mounted with DPX or Eukitt,an dwer ephotographe d eitherwit ha Niko n orZeis sAxiopho t microscopeequippe dwit hbrigh t anddar kfield optics .

REFERENCES

Cox,K.H. ,an dGoldberg ,R.8. , 1988,Analysi so f plantgen eexpression , in:Plan t Molecular Biology: Apractica lapproach ,C.H .Shaw ,ed. ,IR LPress ,Oxford . Dickstein, R., Bisseling, T., Reinhold, V.N., and Ausubel, F.M., 1988,Genes and Development, 2: 677.

104 Dudley,M.E. ,Jacobs ,T.W. , andLong ,S.R. , 1987, Planta, 171:289 . Finan, T.M., Hirsch, A.M., Leigh, J.A., Johansen, E., Kuldau, G.A., Deegan, S., Walker, G.C., and Signer,E.R . (1985)Symbioti c mutantso fRhizobium meliloti that uncouple plant from bacterial differentiation. Cell40,869-877 . Finan, T.M., Hirsch, A.M., Leigh, J.A., Johansen, E., Kuldau, G.A., Deegan, S., Walker, G.C., and Signer,E.R. , 1985,Cell 40,869-877 . Franssen, HJ., Nap,J.-P. ,Gloudemans , T., Stiekema,W. , Van Dam, H.,Govers ,F. ,Louwerse ,J. ,Va n Kammen,A. ,an d Bisseling,T. , 1987,Proc. Natl. Acad. Sei.USA, 84: 4495. Hirsch,A.M. ,Bhuvaneswari ,T.V. , Torrey,J.G. , and Bisseling, T., 1989,Proc. Natl. Acad. Sei.USA 86: 1244. Hirsch,A.M. ,Drake ,D. ,Jacobs ,T.W. , andLong ,S.R. , 1985,J. Bacterid., 161 : 223. Hirsch, A.M., Wilson, K.J., Jones, J.D.G., Bang, M., Walker, V.V., and Ausubel, F.M., 1984, J. Bacteriol,. 158:1133. Leigh,J.A. ,Reed ,J.W. , Hanks,J.F. ,Hirsch , A.M.,an d Walker,G.C. , 1987,Cell, 51: 579. Leigh,J.A. , Signer, E.R., and Walker,G.C. , 1985,Proc. Natl. Acad. Sei.USA, 82: 6231. Lerouge,P. ,Roche ,P. ,Faucher ,C , Maillet,F. ,Truchet ,G. ,Promé ,J.C. ,an d Dénarié,J. , 1990, Nature, 344: 781. Norris,J.H. ,Macol ,L.A. , and Hirsch,A.M. , 1988,Plant Physiol. 88, 321-328. Sass,J£. , 1953,Botanica l Microtechnique,3r ded. ,Iow aStat eUniversit y Press,Ames , IA. Truchet, G., Debelle,F. , Vasse,J. , Terzaghi, B.,Garnerone , A.-M.,Rosenberg , C, Batut, J., Maillet, F.,an dDénarié ,J. , 1985,J. Bacteriol., 164:1200 . Truchet, G., Rosenberg, C, Vasse, J., Julliot, J.-S., Camut, S„ and Dénarié, J., 1984,J. Bacteriol., 157: 134. Van de Wiel,C , Scheres, B.,Franssen , H., Van Lierop, M.-J., Van Lammeren, A., Van Kammen, A., andBisseling ,T. , 1990b, EMBO J.,9 :1 . Witty, J.F., Minchin, F.R., Sk0t, L., and Sheehy, J.E., 1986, Oxford Surv.Plant Mol. CellBiol., 3: 275. Wong,C.H. , Pankhurst, CE., Kondorosi, A.,an d Broughton, W.J., 1983,J. CellBiol., 97, 787.

105 CHAPTER 7

The EN0D12 gene product is involved in the infection process during the pea-Rhizobium interaction.

BenScheres ,Clemen sVa nD eWiel ,Andre iZalensky >Beatri xHorvath , Herman Spaink,Herma n VanEck ,Frie dZwartkruis ,Anne-Mari eWolters , TonGloudemans ,A bVa nKamme nan dTo nBisseling - SUMMARY

The pea cDNA clone pPsENOD12 represents a gene involved in the infection process during thePisum sativum L . -Rhizobium leguminosarum bv.viciae symbiosis . TheENOD1 2protei n is composed of pentapeptides containing two (hydroxy)prolines. The expression of the ENOD12 gene is induced in cells through which the infection thread ismigrating , but alsoi n cells that dono t yet contain an infection thread. Soluble compounds from Rhizobium are involved in eliciting ENOD12 gene expression. Rhizobiumcommo n and host-specific nodulation genes areessentia l for theproductio n of thesecompounds .Tw oENOD1 2gene sar eexpresse di nnodule s andi n stem tissueo f uninoculated plants. The gene represented by the cloned ENOD12 mRNA is also expressedi nflowers , but adifferen t transcription startmigh t beused .

INTRODUCTION

The symbiosis betweenRhizobium bacteria and legumesresult si n the formation of highlyorganize d structures,namel y theinfectio n threads and theroo t nodules.Thes e nodules,i n which nitrogen fixation takesplace ,consis t of different tissues,organize d in a specific way (Newcomb,1980). Theinfectio n process startswit h thedeformatio n andcurlin g ofroo thair s (Bauer, 1981). Curling is thought to achieve enclosure of attached bacteria, permitting them to locally modify and hydrolyse the cell wall of the root hairs (Callaham &Torrey , 1981; Turgeon & Bauer, 1985).A t this site cell wall material is deposited by the plant and it forms tubular structures, the infection threads. Infection threads, containing dividing bacteria, growint o theroo tcortex . In pearoot sth einfectio n thread proceeds toward the inner cortical cell layers, where the following events occur: prior to penetration of the infection thread into root cortex cells, the cells change remarkably, as microtubules rearrange,a nadditiona lcel lwal llaye ri sdeposite d anda cytoplasmi c bridgei sforme d by which the infection thread willmigrate .Th einfectio n threadpasse sthroug h theprepare d cells bya cel lwal ldegradation/depositio nmechanis mjus t like inroo thair sBakhuize n et al, 1988a, 1988b). Concomitantly with infection thread formation rhizobia induce the formation of a premeristem, the nodule primordium, in the inner cortical cell layers. Infection threads grow intoth enodul eprimordium , wherebacteri a arerelease d from the

109 infection threadtips (Libbeng a andBogers , 1974).Simultaneousl y atth eapica l siteo fth e nodule primordium the meristem is formed. The meristem cells are smaller in size and have smallervacuole stha n theprimordiu m cells.Th edirectio n of infection threadgrowt h is then reversed as it now follows the apical meristem that grows out of the root by adding cells which differentiate into the various nodule tissues. In this way there is a continuous infection process inth e so-called invasion zone,immediatel y adjacent toth e meristem. Upon release of bacteria in theplan t cells,th e bacteria areencapsulate d by a membraneo fplan torigin , anddifferentiat e intoN2-fixin g bacteroids (Newcomb, 1976). It has been shown that both bacterial and plant genes are involved in nodule formation. For example the bacterial common and host specific nodulation (nod)gene s areinvolve di nroo t hairdeformation , infection thread formation, andinductio n ofcortica l cell division (for review see Long, 1989). They are also essential for the induction of expression of nodule-specific plant genes, the nodulin genes (Van Kammen, 1984; Govers et al, 1986).Noduli n genes can, according to the the timing of their expression during nodule development, be divided into early and late nodulin genes (Govers et al, 1987). Early nodulins are involved in root hair deformation, infection or nodule morphogenesis. The best studied early nodulin isENOD2 . It is a (hydroxy)prolinerich protein which is mostlikel y acel l wall component (Franssen et al, 1987),tha t is formed in noduleparenchym a ('innercortex' ) cells (VanD e Wiele t al, 1990).Lat e nodulins are detectable after thenodul eha sdevelope d andbacteria lreleas eha stake nplace . Therefore, they are neither involved in infection nor in nodule morphogenesis. Well characterized late nodulins are the leghemoglobins (Brisson and Verma, 1982) and a nodule specific uricase (Bergmann et al, 1983),involve d in oxygen transport and nitrogen metabolism, respectively. Several late nodulins are located in the peribacteroid membrane, but their function is yet unknown (Fortin et al, 1985, 1987; Jacobs et al, 1987; Sandal et al, 1987). The early nodulins characterized so far, GmENOD2, GmENOD13, and GmENOD55, are related to nodule morphogenesis and not to the infection process (Franssen et al, 1987, 1988). Our aim was to isolate a cDNA clone encoding a nodulin involved in the infection process, in order to investigate the role of plant genes in this process and the regulation of their expression by Rhizobium. The infection process occurs abundantly inth einvasio n zone of young pea nodules,wher eman y cells derived from themeriste m arepenetrate d by aninfectio n thread. Therefore wedecide d toisolat e early nodulin cDNA clones from a pea nodule cDNA library. The infection process occursi nroo t hairs aswel la si nnodules ,s ow e selected putativeinfection-relate d clones bytestin g whether thecorrespondin g genesar eexpresse d bothi nroo t hairso f inoculated plants and in nodules. With this approach we obtained a cDNA clone representing an

110 early nodulin gene, involved in the infection process. In the following we report on the characterization of the ENOD12 cDNA clone, the regulation of expression of the corresponding gene by Rhizobium and its possible function in the infection process. Furthermore we discuss the evolutionary origin of this nodulin gene, since transcripts homologous to this cDNA clone were detected in stem and flower tissues.

RESULTS

Isolation of an infection-related early nodulin cDNA clone

A pea nodule cDNA library was differentially screened with cDNA probes made of RNA from nodules from 10da y old plants and uninfected roots from 8 day old plants, respectively. One of the isolated nodulin clones, pPsENOD12, appeared to encode an early nodulin potentially involved in the infection process. RNA transfer blot analysis revealed that the ENOD12 mRNA has a transient pattern of appearance during nodule development (figure 1A). It is already detectable in root segments of 8 day old infected pea plants, which do not possess macroscopically visible nodules. The mRNA reaches its maximum concentration from day 10t o day 13 and decreases in concentration thereafter. In contrast to ENOD12 mRNA the transcript of the pea early nodulin ENOD2 is first detectable on similar blots at day 10 and reaches a maximum concentration between day

Fig. 1. RNA transfer blot analysis of RNA from roots,nodule s androo t hairs. A. RNA transfer blots contain 10u g oftota lRN Afrom uninoculate d roots pPsEN0012 of 8 day old plants (R), and nodules (N), 8, 10, 13, and 17 days after sowing and inoculation. Blots were probed with pPsENOD12, pPsENOD2 and pPsLb inserts, pPsEN0D2 respectively. B. RNA transfer blot contains 20 ug of total RNA from root hairs (RH),an d nodules 13day s after inoculation (N).Plant s were inoculated 3 days after sowing and root hairs were harvested 48 hrs after inoculation (+); after 48 hrs without pPsLb inoculation (-);4 8hr safte r inoculation withR.leguminosarum bv. viciaeA1 0 (/io

Ill 13an d 17,wherea s theconcentratio n remainsconstan t thereafter. ThemRN Ao f thelat e nodulin leghemoglobin isfirst detectabl e atda y 13 andincrease si nconcentratio n during thefollowin g days (figure 1A) . Next,w edetermine d whetherth eENOD1 2gen ewa sexpresse di nroo thair s from inoculatedplant s usingRN Atransfe r blotanalysis .Thi s studyreveale dtha tth eENOD1 2 transcript ispresen t inroo thair s2 4hour s after inoculation,bu tno ti nth eroo thair s from uninoculated peas (figure IB, RH+ an d- lanes) .Th epresenc eo f theENOD1 2transcrip t in root hairs of plants shortly after inoculation indicates that ENOD12 gene expression maycorrelat ewit hth eoccurrenc e of theinfectio n process.

LocalizationofENODll mRNA inroot cortex and nodule cells

The correlation between ENOD12 gene expression and the occurrence of infection threads was further examined by in situ hybridization. Roots of pea plants, inoculated with bacteria three days after sowing, were harvested 2 and 3 days after inoculation. They were embedded in paraffin and serial sections containing nodule primordia and infection threads were selected. These sectionswer e hybridized to"s-labele d antisense ENOD12RNA . Two days after inoculation ENOD12 transcript appeared to be located in cortex cellscontainin g theinfectio n thread (figure 2A,figure 2B ,smal l arrowhead +arrow) .I n root hairs where infection threads are present ENOD12 mRNA is also detected (figure 2A,figure 2B ,arrow) .B yanalysi so f serial sectionsw edetermine d thelocatio no f theti p of theinfectio n thread (figure 2A,smal l arrowhead) andth eENOD1 2transcrip t appeared not to beconfine d to thecell scontainin g an infection thread. Atrac k of cortical cellsi n front of the thread towards the central cylinder also contains ENOD12 mRNA (figure 2B).Fro mcytologica l studiesi ti sknow n thatthes ecell sunderg o several morphological changes and become 'prepared' for infection thread passage (Bakhuizen et al, 1988a, 1988b). Moreover, ENOD12 transcript is also present in cells in the root inner cortex (figure 2A, figure 2B, large arrowhead), which will form the nodule primordium (Libbenga and Bogers, 1974). Three days after inoculation a centre of mitotic activity, the nodule primordium (NP), is clearly visible in the inner cortical cell layers, containing dividing cells which possess enlarged nuclei and a higher content of cytoplasm then the root cortex cells (figure 2C). The infection thread (arrow) has not yet reached the primordium. At this stageENOD1 2mRN A canb edetecte d in thecorte xcell scontainin g theinfectio n thread, inth ecell spreparin gfo r infection thread passage,an d alsoi nth ecell stha tfor m thene w

112 i%s&'&-^1l

fy y-ci > [ 'rrii&ï- ' •*•

'.*=*'&*-*•.

Fig. 2. Localization of ENOD12 transcripts in pea root segments at different stages of nodule developmentb y in situ hybridization Therigh tpane l showsdar kfiel d micrographs correspondingwit h the bright field micrographs in the left panel. In the dark field micrographs, silver grains representing hybridization signal are visible aswhit e spots.A/B ,C/D ,E/F ,an d G/Hrepresen t successive stageso f noduledevelopment . A:Transectio no fa five-day ol droot ,tw oday safte rinoculation .A ninfectio n thread which isclearl y visible at higher magnification can beobserve d ina roo t hair (arrow),th eti po fwhic h hasreache dth esecon dcortica lcel llaye ra sdeduce db yanalysi so fa complet ese to fseria lsection s(smal l arrowhead).I nth einne rcorte x thesit eo f thefutur e noduleprimordiu m ismarke d(larg earrowhead) .C : Transectiono fa si xda yol droot ,thre eday safte r inoculation.Infectio n thread(arrow) ,infectio n threadti p locatedb yanalyzin ga complet ese to fseria lsection s(arrowhead) ,an dth enodul eprimordiu m (NP)i nth e root innercorte x are indicated. E:Transectio n of aseve n dayol droot , four daysafte r inoculation.Th e infection thread(arrows )ha sreache dth enodul eprimordiu m (NP),an dbranche sof f intosevera lthinne r threads(smal larrowheads) ,whic hgro wint ocell sa tth ebas eo f theprimordium .A fe w celllayer sa tth e periphery of the primordium, which will most likely develop into the cortex and, at the top, into the apicalmeriste mo fth enodule ,d ono tcontai n hybridization signal(larg earrowhead) .G :Transectio n ofa tenda yol dpe aroot ,seve nday safte r inoculation.Th eorigina lsit eo fentranc eo fth einfectio n threadint o the nodule primordium isindicate d with an arrow.Th eapisa l meristem (M),invasio n zone (IZ),earl y symbiotic zone (ES),an dnodul ecorte x (NC)ar eindicated . Sectionswer e hybridized with -"S-labeled antisenseENOD1 2RNA . Usingsens eRN Aa sa prob ew edi dno tobserv ehybridizatio n signals(dat ano t shown). Bar= 10 0urn .Furthe rabbreviations :X = xyle mpole ,F = phloe mfiber so fcentra lcylinder .

113 centre of mitotic activity, the nodule primordium (figure 2C, figure 2D). We also localized ENOD12 transcript in pea nodules at later stages of development. Sections from pea nodules of 7,10 and 20 day old plants were hybridized with 35s labeled antisense ENOD12 mRNA. In 7 day old plants infection threads are penetrating the nodule primordium (NP) (figure 2E). At this stage ENOD12 mRNA is located in all cells of the part of the nodule primordium where the infection thread branches (figure 2E, small arrowheads). Only a few small cells at the periphery of the primordium do not contain ENOD12 transcript (figure 2E, figure 2F, large arrowhead). These cells will form the apical meristem and the nodule cortex, while the other cells of the primordium are destined to become the first cells of the infected and the uninfected cell type (Ubbenga & Bogers, 1974, C.v.d.W., unpubl. res.). In root cortex cells where the oldest part of the infection thread resides ENOD12 transcript is no longer detectable (figure 2E, figure 2F, arrows). Nodules from 10 day old plants posses an apical meristem (M), containing small, actively dividing cells which are rich in cytoplasm (figure 2G). The different tissues are

Figure 3. Localization of ENOD12 transcripts in nodules from 20 day old pea plants by in situ hybridization.A. Bright field micrograph of a longitudinalsectio no fa 2 0da yol dpe anodule . The arrow points to theremnan t of the infection thread thatoriginall ypenetrate dth enodul eprimordiu m (cf. figure 2E,G). Late symbiotic zone (LS), vascular bundles (VB), and nodule cortex (NC), which developed from the apical meristem, are indicated. Part of the root to which the nodule isattache d is visiblei na transversal section.Her eth eroo tcorte x (RC),a xyle mpol e(X )an dphloe mfibers (F )o fth e central cylinder are indicated. B: Dark field micrograph of thesectio n shown in A.Silve rgrain s representinghybridizatio nsigna lar evisibl ea swhit e spots. C: Magnification of the bright field micrograph in A, as outlined in figure A. Silver grainsrepresentin ghybridizatio n signalar evisbl ea s black spots.Th e hybridization signal is strongove r the invasion zone (IZ), and gradually diminishes over the early symbiotic zone (ES).Nodul e cortex (NC)an dmeriste m (M)ar eindicated .Section swer e hybridized with 35s-iabeled antisense ENOD12 RNA. Bar in A: 300 urn. Bari nC :5 0 urn.

114 graded in age from the apical meristem to the base of the nodule (Newcomb, 1976). Immediately adjacent to the meristem, inth einvasio n zone (IZ),cell s arepenetrate d by infection threads,growin g inreverse d direction asthe y nowfollo w the meristem. About half of the cells in the invasion zone are infected by bacteria released from infection threads. In the early symbiotic zone (ES) these cells differentiate into the infected cell type.Th eothe rcell swhic h areno t infected byrhizobi a become the uninfected cell type. ENOD12transcrip t ispresen t in the invasion zone,adjacen t to the meristem, but is not detectable in the meristematic cells (figure 2G, figure 2H). In the early symbiotic zone, wherecell sar eelongating ,th econcentratio n ofENOD1 2transcrip ti sdecreasing . In nodules from 20 day old plants ENOD12 mRNA is detectable at the nodule apex (figure 3A, figure 3B). A magnification of figure 3A shows that the transcript is located in the zone where infection threads are growing (figure 3C). No transcript is detectable in the meristem, and theconcentratio n of thetranscrip tdecrease s in theearl y symbiotic zone.W e conclude that, in 7da y old aswel l asi n 10an d 20da y old nodules, ENOD12mRN A is restricted to the region of the nodule where active infection thread growth occurs,an dtherefor e itmark sth einvasio n zone.

ENOD12 isa (hydroxy)proline rich protein

Further information on the nature of theENOD1 2earl y nodulin wasobtaine d by determining thenucleotid e sequenceo f theinser to fpPsENOD1 2(figur e 4).Th einser t is 553 bp in length, excluding a short poly A stretch at the 3' end of the sequence. The mRNA measures approx. 600bases ,a sdetermine d ona nRN A transfer blot, andprime r extension analysis showed that less than 20 bases from the 5' end of the mRNA are missing in pPsENOD12(figur e 10).Th esequenc e of the 5'en do f themRN A missingi n the cDNA clone was determined by direct RNA sequencing (figure 4, small typeface). ThecDN A sequence contains only one long open reading frame, starting with an ATG codon at position 7 which is the first and only methionine codon in the reading frame. From thederive d amino acid sequence amolecula r weighto f 12.5k Dwa scalculate d for theENOD1 2protein .A putativ esigna lpeptid econtainin ga hydrophobi c coresequenc ei s present atth eN-termina lpar t andth epossibl ecleavag e siteo f the signalpeptide ,marke d with an arrow infigur e 4,wa sdetermine d by therule so f Von Heijne (1983).Th e major part of thefollowin g protein sequence consists of twopentapepnd erepeatin g units.On e of these units, Pro-Pro-Gln-Lys-Glu, indicated with solid lines, is well conserved throughout theprotei n sequence.Th eothe runi ti spresen t asPro-Pro-Val-Asn-Gl ya tth e amino-terminal part and gradually,ever y amino acid except the prolines is permuted to

115 give a Pro-Pro-His-Lys-Lys unit at the carboxy-terminal part of the polypeptide chain (dashed lines).A t twoposition s aprolin e codon is changed into athreonin e codon by a single base substitution. Further downstream toth ecarbox y terminus theprolin e repeat units are absent. Invitr otranslatio n of hybrid-selected ENOD12mRN Afro m nodulesyielde d one radioactivepolypeptid e of 12.5k Dwhe n 35s-methionine wasuse da sradioactiv e amino acid. When the same selected mRNA was translated in the presence of ^H-leucine two radioactive polypeptides of 12.5 and 14k D were formed (figure 5). Since the smaller polypeptide is theon e labeled with 35s-methioninethi s cannot be abreakdow n product of the larger polypeptide. Hence there appear to be two different ENOD12mRNA s in nodules.Th eobservatio n thatther ei sn omethionin e encoded in theope nreadin g frame

MetAlaSerPhePheLeuSerSer nnnnnnnaaaaatcactaCTTAAAATGGCTTCCTTTTTCTTGTCCTCA -10 0 10 20 30 4. LeuValLeuPheLeuAlaAlaLeuIleLeuValProGlnGlyLeuAlaGlnTyrHisLeu CTAGTGTTGTTCCTTGCTGCTCTTATCCTTGTTCCTCAAGGACTTGCTCAATATCACCTT 40 50 60 70 80 90

AsnProValTyrGluProProVa1AsnGlyProProva1AsnLysProProGlnLys Glu AATCCTGTTTATGAACCACCAGTGAATGGGCCACCGGTGAATAAGCCACCACAGAAAGAG 100 110 120 130 140 150

TbrProValHisLysProProGlnlysGluThrProValHlsLysProProGlnLysGlu ACACCGGTTCATAAGCCACCACAGAAAGAGACACCGGTTCATAAGCCACCACAAAAAGAG 160 170 180 190 200 210

ProProArgHlsLysProProGlnLysGluProPvoArgHlsLysProProHlsLysLys CCACCGAGGCATAAGCCACCACAAAAAGAGCCACCGAGGCATAAACCACCACACAAGAAG 220 230 240 250 260 270

SerHisLeuHisValThrLysProSerTyrGlyLysHisProThrGluGluHlsAsnlle TCACATTTGCACGTGACAAAACCATCTTATGGTAAACATCCTACAGAAGAACATAACATC 280 290 300 310 320 330 HisPhe * CATTTCTAAAGCATTCTAGTACCAATGTTTCATTTGATATGTACCTTTTGTAACATGTGT 340 350 360 370 380 390 GGCCTTGTTGTTTTTCCATTTATGCATGGTTAAGTTATGTTTTTCTCTTATGTATGGCCA 400 410 420 430 440 450 AGTAAAGAGTAGCATATATTTGTTGCTTTTTGTTTAAAGGTACTTCCTGCTAGTGCAGTG 460 470 480 490 500 510 ATTTGTAGTCTTGATTGTGATGGTAAGCAGTTGTTTTGGTATTAAAAAAAAAAAA 520 530 540 550 560

Fig. 4. cDNA and predicted amino acid sequence of thepPsENOD1 2 insert. Nucleotides 1-565 are determined from thepPsENOD1 2 insert. Nucleotides -18 (the 5'en d ofth e mRNA) to0 ar e determined usingdirec tRN A sequencing (seematerial san d methods).Th eamin oaci d sequenceo f theonl y longope n readingfram e isdepicte dove rth enucleotid esequence .Th eputativ e signalpeptid ecleavag e sitei smarke d with an arrow.Proline s inth erepea t region are inbol d typeface. Theamin oaci d triplets characteristic of both types of pentapeptide repeats described in thetex t areoverline d with unbroken anddashe d bars, respectively.Th eterminatio n codonendin gth ereadin gfram e ismarke db ya nasterix .

116 Fig. 5. In vitro translation products of Met Leu hybrid-selected ENOD12mRNA . ENOD12 f 11 1 mRNAwa s selectedfor m totalnodul eRN A andtranslate di nvitr oi n thepresenc eo f ^S- - + ~ + methionine (Met) or 'n-leucine (Leu) as radioactiveamin oacid .Filter suse dfo rhybri d selection contained pBR322 (-) or pPsENOD12 insert (+).Th erigh ttw o lanes are exposed 6-fold longer than the neighbouring twolane st oth eleft . The size (kD)o f the in vitro translation productsi s indicatedt oth erigh tside .

f.. ••^Wa H i. ; 12.5

of pPsENOD12, except at the start, shows that the cDNA clone corresponds to the mRNA encoding the 14k Dpolypeptid e that can only bedetecte d with ^H-leucine.Th e observeddiscrepanc y between thecalculate d andth eapparen t mol.wt .i srathe rcommo n for proline-richprotein s (e.g.Fransse n et al, 1987).

ENOD12 gene expression requiresRhizobium nod genes and is induced byexcreted bacterialcompounds

To determine which bacterial genes are essential for the induction of ENOD12 gene expression weanalyze d the expression of thisgen e after inoculation with different Rhizobium mutants. The R.leguminosarum bv. viciae common nodulation genes nod ABC arerequire d for initiation of cortical cell division, root hair curling, and infection thread formation (Wijffelman etal , 1985).I nth eR.meliloti -alfalf a interaction thehost - specific nodEFgene s are also important for infection thread formation (Horvath et al, 1986),an drecentl y theinvolvemen t of nodEFi ninfectio n threadformatio n hasals obee n demonstrated for theR.leguminosaru m bv.vicia e- Vici ahirsut ainteractio n (Van Brussel et al, 1988).Therefor e weexamine d whether both common and host-specific nodgene s are essential for eliciting ENOD12 gene expression in root hairs. Pea plants were inoculated with various mutant R.leguminosarum bv. viciae strains.A s a control wild- type R. leguminosarum bv. viciae 248 was used. To obtain maximum sensitivity in detectingENOD1 2gen e expression inroo t hairs we amplified cDNA specifically made from total root hair RNA by thepolymeras e chain reaction (PCR, see Saiki et al, 1985;

117 Mullis and Faloona, 1987). The amplified cDNA was visualized by DNA transfer blotting using pPsENOD12 cDNA insert as a probe. In this way the presence of ENOD12 transcript in 1 u.g total root hair RNA, inoculated with wild-type R. leguminosarum bv. viciae could be visualized within several hours after exposure of a hybridized DNA transfer blot (figure 6, WT). Upon longer exposure a weak signal could also be observed in uninoculated root hair RNA (not shown). We do not know whether this signal is caused by low levels of ENOD12 mRNA or by residual chromosomal DNA present in the RNA preparations. The R. leguminosarum bv. viciae A10 strain carries a Tn5 mutation in nodA which blocks the formation of the nodA, nodB, and nodC products (Wijffelman et al, 1985). In our first experiments we demonstrated by RNA transfer blot analysis that ENOD12 transcript is found in root hairs from plants inoculated with wild type R.leguminosarum bv. viciae, but not in root hairs from plants inoculated with R. leguminosarum bv. viciae A10 (figure IB). Using the more sensitive PCR method it was confirmed that a mutation in nodA abolished the ability of bacteria to induce ENOD12 gene expression in root hairs (figure 6, lanes 2 and 3). The Sym-plasmid cured R. leguminosarum bv. viciae 248c (pMP104) strain, carrying the «odEFDABCIJ genes on the low-copy plasmid pMP104, is capable of nodulating Vicia (Spaink et al, 1987) and pea (H.P. Spaink, personal communitcation). This strain was able to induce ENOD12 gene expression (figure 6, lane 4). On the other hand the R. leguminosarum bv. viciae 248c (pMP104 nodE::Tn5) strain, carrying a mutation in nodE, which forms no infection threads on Vicia (Van Brussel et al, 1988), showed no induction of ENOD12 gene expression (figure 6, lane 5). We concluded that both common and host-specific nod genes of Rhizobium are essential for eliciting

Fig. 6. PCR analysis of ENOD12 gene expression in root hair RNA. ENOD12 sequences from 1|i g total root hair RNA were amplified using 12 PCR cycles, and after electrophoresis a DNA transfer blot was probed with 32p_jaj,eie(j pPsENOD12 insert. The different lanes contain amplified root hair ENOD12sequence sfro m uninoculated plants (1 and 6); from plants inoculated withR.leguminosarum bv . viciae A10 (2), wild-type R.leguminosarum bv . viciae 248 (3 and8) ,R.leguminosarum bv. viciae 248c (pMP104) (4);R.leguminosarum bv. viciae 248c (pMP104 nodE::Tn5) (5); and from plants inoculated with the cell-free supernatant of wild-type R.leguminosarum bv.viciae 24 8 induced with naringenin(7) .

118 EN0D12 geneexpressio n inpe aplants . It has been shown that, upon induction of the nod genes with a flavonoid, rhizobia excrete compounds that cause deformation of root hairs (Zaat et al, 1987). Therefore we studied whether the bacteria also excrete compounds capable of inducing the ENOD12 gene. Rhizobia were grown in the presence or absence of naringenin, a flavonoid inducing the nod genes. After removal of the bacteria (see experimental procedures) the Rhizobium-free culture medium wasapplie d to 3da yol d pea seedlings. Two days after inoculation root hairs were harvested and the presence of ENOD12 mRNAi nroo thai rRN Awa sstudied .Th eENOD1 2transcrip t wasdetectabl e inroo thai r RNA from plants treated with the medium of bacteria cultured in the presence of naringenin (figure 6,lan e7) .O nth eothe rhan d thecultur emedi ao fbacteri agrow ni nth e absenceo fnaringeni n couldno testablis h anincreas ei n theamoun to f transcript (datano t shown). We concluded that excreted compounds, formed after induction of the Rhizobium nodgenes , areabl et oelici tENOD1 2gen eexpressio n inroo thairs .

ENOD12 geneexpression isnot a defense reaction

Theprolin e repeat units in the ENOD12protei n are quite similar to those in the amino acid sequence of hydroxyproline-rich glycoproteins (HRGPs), accumulating in plant tissueafte r wounding orupo n interactions withpathogen s (Chen andVarner , 1985; Corbin et al, 1987). Furthermore, infection thread formation has been viewed as a modified plantdefens e response (Vance, 1983;Djordjevi c et al, 1987).Thi sprompte d us toinvestigat e whetherth eENOD1 2gen eo rsimila rgene sar einduce d aspar to f adefens e response in pea. For these experiments we used the pathogenic fungus Fusarium oxysporum f. sp. pisi. While accumulation of HRGP transcripts was observed upon RNA transfer blot analysis of total RNA from pea roots inoculated with the fungus, no ENOD12 mRNA was detectable in these RNA preparations (data not shown).W ecan , therefore, conclude that the expression of the ENOD12 gene(s) during the infection process cannot be attributed to a general defense response following Rhizobium infection.

ENOD12 geneexpression instem and flower

We studied whether the ENOD12 gene or genes resembling ENOD12 are expressed inothe rpart so f theplan t sincethi smigh t give someclue so nth eevolutionar y

119 origin of theENOD1 2earl y nodulin. By RNA transfer blot analysis we were unable to detect ENOD12 transcripts in root tip, root elongation zone, mature root, hypocotyl, epicotyl,plumule ,an dlea f (figure 7). On theothe r hand,hybridizin g RNA similari n size to ENOD12 transcript, but less abundant, was found in stem and flower RNA. (figure 7). In both tissuesthi s RNA wasals odetectabl e with anENOD1 2probe ,pPsENOD12 - 3', containing only the 3' region downstream from the Bgll site at position 239 in the cDNA sequence (figure 4) (datano tshown) . Using the in situ hybridization technique welocalize d ENOD12mRN A in stem internode sections. The transcript appeared to be located in a zone of cortical cells surrounding the central ring of vascular bundles and the interfascicular cambium cells (figure 8). In flowers we were not able to localize the ENOD12 transcript unambiguously. In vitro translation of hybrid-selected ENOD12 nodule mRNA resulted in two distinct polypeptides, as shown in figure 5. Therefore the existence of two or more ENOD12 genes in thepe a genome seemed plausible. In order toobtai n information on the number of ENOD12 genes we performed Southern blot analyses of restricted pea genomicDN A using thecDN A clonespPsENOD1 2 andpPsENOD12-3 'a sprobes .Tw o EcoRI fragments of 4.5 and 5.5 kb hybridized to pPsENOD12. Both fragments also hybridized to the 3'regio n probe. Restriction withHindi n again yielded two fragments, 1.8 and 7 kb in size, hybridizing as well to pPsENOD12 as to the 3'regio n probe (data not shown). Sinceth etw oHindü l and the twoEcoR I fragments hybridized to the same level with bothpPsENOD1 2an dpPsENOD12-3 'i tappear sver y likely thatther ear etw o ENOD12gene si nth epe agenome . Thepresenc e of moretha non eENOD1 2gen eraise sth equestio n of which genes aretranscribe d inth edifferen t tissues.T oanswe r thisquestio n wecoul dno tus ei n vitro translation of hybrid-selected ENOD12mRN A from steman d flower sincei nou r hands the sensitivity of hybrid-released translation wasinsufficien t with low abundant mRNA

Fig.7 .RN Atransfe r blotanalysi s ^. ^. ^. ^. ofvariou suninoculate dpe atissues . \ tV O O v> •?> RNA transfer blots contain 20 u.g ^^^V^v'ó^^ <: ^ of total RNAfro m the following tissues;fro m Sda yol dpe a plants: rootti p (RT), rootelongatio n zone (REL),hypocoty l (H), epicotyl (E), plumule(P) , leaf (L); from 12 day oldpe aplants :ste m (S),roo t (R); nodules 13day s after inoculation (N), and fromplant so fvaryin g age: flower (F).Blot swer eprobe d withpPsENOD1 2 insert;i nlane sP t oN of the left panela fain tban d migratingwit hlowe rmobilit ytha nENOD1 2mRN A can be observedwhic hrepresent saspecifi cbindin g ofprob e to the smallribosomal RNA .

120 and ^H-leucinea sth e labeled amino acid. Specific probesfo r different transcripts could neither be used since all isolated ENOD12 cDNA clones corresponded to the same ENOD12 mRNA.Therefor e we adapted the RNase mapping assay (Melton et al, 1984) to discriminate between different mRNAs by virtue of their complete or incomplete protection to RNase digestion. As aprob e for RNase mapping we used the 3' region of thepPsENOD1 2 inserta sth e5 'regio ncontain s sequenceduplication s which willpreven t accurate mapping. 32p_iabeled antisense ENOD12 mRNA was transcribed from the 3' region of the insert of pPsENOD12 as indicated in figure 9C1,2. This antisense RNA was hybridized to total RNA from roots and nodules,o r toroo t RNA which was mixed with 1 ng unlabeled ENOD12 sense RNA transcribed from the 3'regio n of theinser to f pPsENOD12,clone d in theopposit eorientatio n towards theT 7promoter , asindicate di n figure 9C3. After hybridization single-stranded RNA was digested using increasing amounts of RNase Tl and subsequently the RNA was separated by Polyacrylamide gel electrophoresis. The RNase mapping experiment showed that a21 6 bp sense-antisense ENOD12RN A hybrid remained fully protected using increasing amounts of RNase Tl

TTMS*e,

Fig.8 . Localization of ENOD12transcript s in stem tissue by in situ hybridization. A. Brightfiel dmicrograp ho fa transectio no fth e fourth internode of a 24-day-old pea plant, showinga centra lring o fvascula rbundle s(VB ) withtw ofibrovascula r bundles (FVB)an dtw o corticalfiber bundle s(CF )traversin gth ecorte x (C). P: pith; arrow: fascicular cambium; arrowhead: interfascicular cambium.B . Dark field micrographo f thesam esectio na si n A. Silvergrain srepresentin ghybridizatio nsigna l are visible as white spots. The section was hybridizedwit h35 S-labeledENOD1 2antisens e RNA. Bar= 30 0urn .

121 (figure 9A, R+S lanes). This 216 bp band was absent when root RNA was hybridized to antisense EN0D12 RNA (figure 9A, R lanes). Hybridization of the antisense probe to nodule RNA and subsequent digestion resulted in a fully protected hybrid of 202 bp (fig. 9A, N lanes 1/5T and IT) which was the expected size, as indicated in figure 9C. Using low concentrations of RNase Tl the trimming of the fully protected hybrid was still incomplete, resulting in products ranging in size between 202 and 220 bp (fig.9A, N lane 1/25T). In addition to the fully protected 202 bp hybrid, partially protected hybrids were formed (figure 9A, N lanes). Since these hybrids were not formed when the ENOD12 sense RNA was hybridized to the antisense probe (figure 9A, R+S lanes), we conclude that they originate from the second ENOD12 mRNA that occurs in nodules. Using increasing amounts of RNase many of these heterologous hybrids were degraded to smaller molecules, which cannot be analyzed on sequence gel as their size corresponds to the size of the fragments generated by digestion of the excess antisense RNA probes. However, the 74 and 48 bp fragments (figure 9A, N lanes) are examples of fragments which are not further degraded by increasing RNase concentrations. In conclusion, at fixed RNase concentration identical digestion patterns of the heterologous ENOD12 hybrid were obtained in different, independent experiments using nodule RNA from two different pea cultivars (figure 9A and 9B, N lanes). This indicates that with the RNase mapping method partially protected hybrids formed by hybridization of the antisense probe to a non-homologous mRNA are reproducibly detected as specific fragments on a

Fig.9 .RNas e mapping of ENOD12transcript s in nodule,stem ,roo t hair and flower tissues.A .2 0 u.g total RNA from 8da y old roots (R),nodule s 13day s after inoculation (N),an d roots mixed with 1 ng 'sense'ENOD1 2RN Atranscribe dfro m aT 7RN Apolymeras evecto r(fi g9C,3 )(R+S) , washybridize dt o a 227 nt 'antisense' ENOD12 RNA probe (fig 9C,2), followed by digestion with varying amountso f RNase Tl (1/25 T: 228 U/ml, 1/5 T: 1140 U/ml, 1T : 5760 U/ml). Protected RNA molecules were separatedb yelectrophoresi so na 6 %polyacrylamide/ure age lan dsize swer ecompare dt opBR32 2x Hinf l size markers (M) and the input ENOD12RN A probe (P). Vertical bars over the figure represent the borders of thedifferen t lanes. B.2 0u gtota lRN A form 8da y oldroot s(R) ,flower s (F),4t h internode stemsection s from 35da y oldplant s (S),roo t hairs4 8hr safte r inoculation (RH),nodule s 13 days after inoculation (N),an d root + 1n g 'sense'ENOD1 2RN A (R+S),wa s hybridized to 'antisense'ENOD1 2 RNAprob e(P )a si nA. ,followe d bydigestio n with 5760U/m lRNas e Tl and electrophoresis ona 6 % polyacrylamide/urea gel. Root, flower and nodule RNA was taken from two different cultivars, cv. 'Rondo' and cv. 'Sparkle'. A small portion of root RNA immediately after hybridization with probe without RNase digestion (C), and pBR322 x Hinfl size markers (M), were also subjected to electrophoresis. Vertical bars over the figure represent borders of the different lanes. C. 1:Schemati c representation of theclone dENOD1 2mRNA .Th epositio n of therestrictio n sitesi n thecorrespondin g cDNA clone that were used to subclone the fragment used for RNase mapping is indicated; 2 and3 : Sequences present in theENOD1 2'antisense ' (2),an d 'sense' (3) T7 RNA transcripts. 202 nucleotides betweenth eBg lI sit ea tpositio n24 6an dth eBa lI sit ea tpositio n44 8i nth epPsENOD1 2cDN A(figur e 9C1)( V//A ); 11 ntT 7RN Apolymeras epromote rregio n (I I ); 14n tpT 7polylinke r ( i^H ). Complementary sequences areindicate d by identically shaded inversed arrows.Dashe d vertical lines depict the size of the hybridizing fragments upon hybridization of the antisense RNA probe to the homologous ENOD12 mRNA (202 nt.) and to the sense RNA transcript (216 nt.), respectively. Restriction sitesi nth ecorrespondin g DNAfragments : Bal= Ball ,Bg l= Bgll ,E = EcoRI ,H =HindIII , S =SmaI .

122 B. rondo sparkle

ute u. IA te z ecJ.ec u. z K o. I l i 1 1 i l l l l l i l 1 i i i i i i i i

-227 -227 -216 -216 -202 -202

s* r- • m ' «k » -74 i

-74

-48 !• *

48 • é -

Bgt<246) Bal(448) Sn^/. ï.vi'yyyyyyyy/,, yyyy yyyyéyyyyy

202 i 14 Bal Bgl m H i S E

I 216 ! Bgt Bal 3. E S H

123 S R N N F R Fig. 10. Primer extension analysis of ENOD12 mRNA from nodule, stem and flower. An end-labeled oligonucleotide complementary to nt 70-90 of thecDN A sequencewa sanneale d totota lRN Afrom : — 124 4thinternod este msection sfro m 35 day old plants (S), 8da y old roots (R),nodule s1 3 daysafte r inoculation (N),an dflower s(F) , IMP extended with reverse transcriptase and m subjected to electrophoresis on 6% polyacrylamide/ureagels .2 0u g totalRN A — irj7 wasuse d for each extension,excep t inth e extensionsshow n inth eright pane lwher e8 0 u.gtota l RNA from flowers androot swa s used. The size of the largest reproducible - 1 t extension fragment in each of the lanesi s indicatedt oth eright sid e(nt )t o indicate the size differences between the twogroup s of extensionproducts .

Polyacrylamide gel.W e concluded that RNase Tl mapping is useful for distinguishing betweendifferen t mRNAs. Using this RNase mapping method we analyzed RNA from stem and flower to determine which ENOD12 mRNAs occur in these tissues. We also analysed root hair RNAt oinvestigat ewhethe rbot hENOD1 2mRNA sdetecte d in nodulear ealread ypresen t atth eearl y stageso f infection thread formation. The 202b pfull-siz e protected hybrid as well asth e7 5an d4 8 bpbands ,specifi c for thepartiall y degraded hybrid, arepresen t in the RNase protection pattern of stem and root hair RNA as shown in figure 9B (S and RH lanes), indicating that in these parts of the plant the same ENOD12 mRNAs are present as in nodules. Using flower RNA we could visualize the full-size protected hybrid,indicatin gtha tmRN Acorrespondin g toth eisolate dpPsENOD1 2clon e ispresen t in flower tissue. Wecoul d not demonstrate in thisexperiment , nori n experiments using moreflowe r RNA andprolonge d exposures, thepresenc eo f bothth e7 5an d4 8b pband s originatingfro m theprotecte dheterologou sENOD1 2mRNA . Longexposure s like infigur e 9B alwaysreveale d thepresenc e of21 6an d21 2b p fragments in all lanes including theroo t lane.Th econcentratio n of thesefragment s was dependent on the amount of probeinpu t and noto n theamoun to f hybridizingmRN Ai n the total RNA preparations. Therefore they are not due to the formation of a hybrid betweenprob ean dENOD1 2mRNA .

124 Since we could not detect differences between the ENOD12mRNA s present in nodule, stem and flower using RNase mapping with the 3' region of pPsENOD12, we investigated whether differences weredetectabl e atth e 5'en d of theENOD1 2mRN Ai n the different tissues. Therefore we extended a synthetic primer complementary to nucleotides 70-90 of the pPsENOD12 insert. After hybridizing with nodule, stem and flower RNA we compared the size of the extension products, as shown in figure 10 (lanes S and N). The sizes of the extension products with stem and nodule RNA appeared tob e identical.Henc e nodifferenc e in sizeo f the 5'end s can be detected with thisprimer . In contrast, theextensio n products with flower RNA differed in size (figure 10, lane F). The largest extension product with flower RNA measured 16 extra nucleotides compared toth e largest extension product with nodule RNA. This suggests that different transcriptional start sites are used on the gene corresponding to pPsENOD12i n flower and nodule, or alternatively the RNA transcript is differentially spliced atth e5 'en di n thesetissues .

DISCUSSION

TheENOD12 geneproduct is involved in the infectionprocess

The datapresente d in thispape r demonstrate that the cDNA clonepPsENOD1 2 represents a gene encoding a (hydroxy-)proline-rich early nodulin, involved in the infection processwhic h ispar to f thepea-Rhizobium interaction. pPsENOD12i sth e first clone that represents a nodulin gene involved in a process occurring in root hairs. The expression of the ENOD12 gene in root hairs requires the presence of functional nod genesi n Rhizobium. Therefore pPsENOD12ca n be animportan t helpfo r analyzing the mechanismb ywhic h thebacteria l nodgene sinitiat eth einfectio n process. Previously three cloned soybean early nodulins, GmENOD2, GmENOD13 and GmENOD55,hav e been shown tob einvolve d in stepsi nnodul e morphogenesis butno t in the infection process (Franssen et al, 1987, 1988). Likewise the alfalfa and pea ENOD2earl ynodulin s areinvolve d in noduleformatio n andno ti n theinfectio n process (Dickstein et al, 1988; Van De Wiel et al, 1990). These early nodulins from soybean, alfalfa, and pea are all (hydroxy)proline-rich proteins. Strikingly, the infection related ENOD12 early nodulin is also proline-rich. As in the ENOD2 early nodulins from different speciesan d inGmENOD13 ,th emajo r parto f theENOD1 2protei n iscompose d

125 of repeating elements containing three amino acids interspersed with two or three prolines. The structure of these three early nodulins is very similar to the soybean cell wall protein 1A10 (Averyhart-Fullard et al, 1988). Because of the homology between 1A10an dENOD1 2w eassum e that ENOD12i s also acel l wall protein, involved in the infection process.Th e occurrence of aputativ e signal peptide, which might function in excreting theprotein ,i sconsisten twit h thishypothesis . Allroo t and nodulecell si n which wehav efoun d ENOD12transcrip t are siteso f new cell wall synthesis and therefore possible sites of incorporation of the ENOD12 protein.I nth eroo tcorte xcell scontainin gth einfectio n thread,th einfectio n threadwal li s formed. The root cortical cells preparing for infection thread passage, which contain ENOD12 mRNA, form an additional cell wall layer (Bakhuizen et al, 1988b). The dividing cells in thenodul eprimordiu m alsofor m newcel lwalls .I nth einvasio n zoneo f thedevelopin g nodule again theinfectio n threadtip sar e siteso f cellwal l synthesis.Ou r presentknowledg e aboutENOD1 2doe sno t allowprediction so n abiochemica l function of the protein in cell walls yet, but the absence of ENOD12 transcripts in pea roots infected with F. oxysporum indicates that the protein is not functional in a defense response.

Rhizobium nodgenes are essential for theinduction ofENOD12 geneexpression.

Ourobservatio n that solublecompound si n aRhizobium- free culturemediu mca n induce ENOD12 gene expression, shows that physical contact between plant and bacterium is not a necessary prerequisite for ENOD12 gene expression. Therefore the direct role of bacterial genes inproducin g compounds involved inENOD1 2expressio n canb estudied .Th epresenc eo fENOD1 2transcrip ti nroo thair so fplant sinoculate d with the R. leguminosarum bv. viciae 248c (pMP104) strain, carrying only cloned nodEFDABCIJ genes, shows that these Sym-plasmid genes are sufficient to induce ENOD12 gene expression. The absence of ENOD12 RNA in root hairs from plants inoculated withR.leguminosarum bv.viciae carryin g aTn 5 mutation in nodA indicates that ENOD12 gene expression requires expression of at least one of the common nod genes.Thi s is consistent with the fact that these genes are essential for theinductio n of the infection process. Furthermore the host-specific nodE gene, and/or the nodF gene present onth e sameoperon ,i sals oessential .Henc e both common andhost-specifi c nod genes appears to be involved in producing the factor(s) that elicit ENOD12 gene expression. The R. melilotinod Aan d nodB genes have been shown to be involved in generating small solublecompound s that stimulate mitosiso f plant protoplasts (Schmidt

126 et al, 1988). Faucher et al (1989) reported that R. meliloti common nod genes are essential for the production of root hair deformation factor, and that the nodH gene is involved indeterminin g thehos t specificity of thisfactor . Theseauthor s hypothesize that the common nod genes produce a compound that can be modified to different factors, e.g.t oroo t hairdeformatio n factor byth enodR geneproduct .Whethe r thecompound(s ) inducing ENOD12 gene expression is also the result of a modification of a nodABC dependentfacto r byth e nodE product, orwhethe r nodABC and nodEenabl e production ofdifferen t factors which arebot h necessary for induction of ENOD12gen e expression, cannot yetb edecided . Which molecular mechanisms leadt oENOD1 2gen eexpression ?W ehav eshow n that soluble compounds from Rhizobiumwhic h are excreted upon induction of thenod genesar eth etrigge ri ninducin gexpressio n ofth eearl y nodulingene .Inductio n occursi n front of the growing infection thread and in the nodule primordium. This induction at significant distance from the bacteria indicates the involvement of factors which are capablet odiffus e throughsevera lcel llayers .I nth einvasio nzon eo fth enodul eENOD1 2 mRNA ispresen t in infected aswel l asi n uninfected cells,a sfa r asca n bejudge d from ourin situ hybridization s using 35s-labeledprobes .Thi sobservatio n iscompatibl e with thenotio no f diffusible inducing compounds.Whethe r these areth ebacteria l compounds made under influence of the nod genes, or plant substances influenced by these compounds, is presently unknown. Clues to the mechanism involved in ENOD12 gene expression come from the observation that ENOD12 genes are expressed in the cells preparing for infection thread growth aswel l asi n themitoticall y reactivated cellso f the initiating nodule primordium. A plant compound from the root vascular tissue, most likely present in the xylem,ha s been found toac t in concert with plant hormones for the induction of primordia in the root inner cortex, similar to the nodule primordium (Libbenga et al, 1973).Th eroo tcorte x cellspreparin g for infection threadpassag e show many structural analogiest ocell si nth enodul eprimordiu m andi tha sbee npostulate d that a similar compound from the xylem and phytohormones are also involved in the preparation of these cells (Bakhuizen et al, 1988b).Th e analogy between these cells is supported byou robservatio n thati n bothcel ltype sENOD1 2gene s areexpressed .Henc e wetak eint oaccoun t thatth epostulate d xylemfacto r andplan thormone s areinvolve d in the induction of ENOD12 gene expression. The necessary changes in phytohormone balance might be induced by the excreted bacterial compounds we have shown to be involved in induction of ENOD12 gene expression. The involvement of axyle m factor can explain the distribution of ENOD12 mRNA from the infection thread toward the noduleprimordiu mnea ra xyle mpole ,a tearl y stageso fth einfectio n process.

127 EN0D12 geneexpression instem and flower tissue

An important question concerning theevolutionar y origin of theENOD1 2gene s active in the Rhizobium infection process is whether theENOD1 2mRNA s in stem and flower aretranscribe d from the samegenes .Fro m Southern analysisan d hybrid-released translation experiments we conclude that two genes are present in thepe a genome, and they arebot h transcribed in nodules. Sinceth eoccurrenc eo f thesetw omRNA si n stem andflowe r tissue could not be analyzed by standard meansw e successfully modified an RNase mapping procedure to distinguish between different ENOD12 transcripts. In general this method might be a useful tool to analyse differential transcription of gene families, since extensive cDNA cloning is not required. In summary, the conclusions from ourRNas e mapping andprime rextensio n experiments aretha tbot hENOD1 2gene s areexpresse d in nodule and stem tissue,wherea si n flower tissue theexpressio n of only one gene, corresponding to pPsENOD12, can be detected. The 5' end of this mRNA differs from that ofit shomologou scounterpar t innodule .W eassum etha tth e difference in nodule and flower is due to adifferen t start of transcription on the same gene, or by alternative splicingo f anintro n nearth e5 'end . Nodulin genes are by definition genes exclusively expressed during root nodule formation and noti n any otherpar t of theplan t (Van Kammen, 1984).Ou rfindin g that ENOD12 genes are expressed at a low level in flower and stem tissue shows that the ENOD12 genes are not true nodulin genes. However, in most other studies on nodulin genes the analyses have been restricted to root and nodule tissues. One can therefore expect that several genes considered to represent true nodulins are also used in other developmental programs in the plant. Recently this was demonstrated for the nodule specific glutamine synthetase gene. More detailed analyses showed that this gene is expressed at low levels in e.g. the stemo f Phaseolus plants (Bennett et al, 1989).Als o theexpressio n of a globin genei n both roots andnodule s from Parasponia suggests that leghemoglobin, the 'archetype' of the nodulins, might be expressed in non-symbiotic tissues (Bogusz et al, 1988). In conclusion, nodule formation involves not only genes that are specifically evolved for the benefit of the symbiosis, but also genes that are normallyuse di nothe rpart so fth eplant , asexamplifie d byth eENOD1 2gene si npe aan d the "nodule specific" glutamine synthetasegen ei n bean.Thes egene sar eno texpresse di n uninoculated roots,an d therefore their expression must bedirectl y or indirectly induced byRhizobium factors. It becomes anintriguin gquestio n whetherRhizobium i sexploitin g the regulatory mechanisms used in other parts of the plant, or whether new symbiotic regulatory mechanisms haveevolved .

128 MATERIALS AND METHODS

Plantmaterials

Pea (Pisum sativumL . cv. rondo or sparkle) plants were cultured and inoculated with R. leguminosarum bv.viciae 248a sdescribe d by Bisseling eta l (1978).Nodule s wereexcise d from root tissue,excep ti nth ecas eo f pea plants 8day safte r inoculation,wher e2. 5c msection so f themai nroot , wherenodule swoul dappear ,wer eharvested .Uninfecte d peaplant swer eculture di nth esam eway , and pieces of themai n root werecollecte d 8day safte r sowing.Pe aroo t hairs werebrushe d from themai n root of seedlings, 48 hrs after inoculation of 3da y old seedlings, as described by Gloudemans et al. (1989). Fusarium oxysporum myceliumwa sinoculate di nCzapek-do xmediu m and grownfo r2 day sa t 30°C.Pe aplant s were inoculated with this suspension threeday safte r sowing,an d cultured asabove . Roottissu ewa sharveste dafte r variousincubatio n times.Al lplan t tissueswer efroze n inliqui dnitroge n immediatelyafte r harvestingan dstore da t-70°C .

Preparationof'Rhizobium-free culturemedium

Bacterial free culture medium for the inoculation of plants was prepared as follows: R.leguminosarum bv.viciae 248wa sgrow ni n YMB mediumt olat elo gphase ,dilute d toODÖO O= 0.0 1 in minimal medium and grown to late log phase,an d diluted again 1:100 in plant medium containing 2(iM naringenin. Bacteria were then grown to ODÖOO = 0-3- The culture was centrifiuged, and the supernatant was treatedwit hchlorofor m andinoculate d on 3 day oldpe aseedlings .

cDNA cloning

AXgtl l cDNA library, prepared from Pisum sativum cv. sparkle noduleRN A of 21 dayol d plants,wa skindl y provided bydr . G.Coruzz i (Tigney et al, 1987). Nitrocellulosereplica s weremade , containing phageDN Ao fapprox .300 0plaques ,usin gstandar dprocedure s (Maniatise tal , 1982).32p . labeledcDN Aprobe swer eprepare dfro m poly(A)+RNAo fnodule sfro m 10 day old plants,an do f 8da y old, uninoculated roots.Replic a filters werehybridize d toeithe rroo t or nodule cDNA asdescribe db y Franssen eta l(1987) . Plaques,givin ga nodule-specific signal,wer epurified , phageDN Awa sisolated , andcDN Ainsert swer esubclone dint o pUC vectorsusin gstandar dprocedure s(Maniati se tal ,1982) .

129 RNA expression analyses

Total RNA from nodules and other tissues was isolated by phenol extraction and LiCl precipitation according to De Vries et al. (1982). Total RNA concentrations were measured spectrophotometrically. Equal amounts of totalRNA , as indicated in thefigur e legends, were subjected to gel electrophoresis. RNA transfer blotting was performed as described by Franssen et al (1987), using GeneScreen membranesa s support Blots were hybridized tonick-translate d cDNA inserts.

Genomic DNA isolation and blotting

Genomic DNA from pea leaves was isolated using the CTAB method, described by Rogers and Bendich (1988).Restrictio n enzymedigestion s were performed under standard conditions.Digeste d DNA was electrophoresed on a 1% agarose gel and transferred to GeneScreen plus membranes (NEN) using ammonium acetate transfer (Rigaud et al, 1987).Th e blot was hybridized to nick-translated cDNA insert in IM NaCl, 1%SDS, 10% dextran sulphate and 100 ng/ml denatured salmon sperm DNA at 65°C during 24 hr. Subsequently blots were washed, 2x10' in 2xSSC and 2x20' in 2xSSC/l% SDS at 65°C.

Hybrid-releasedtranslation

Selection of ENOD12 mRNA was done with the insert of PsENOD12 bound on DPT paper (BioRad) as described by Maniatis (1982). 50n g denatured DNA was spotted on DPT discs of 0.5 cm^. Hybridization to 1m g total nodule RNA from 12da y old plants was done in 300 \ll containing 50%(v/v) formamide; 0,1%SDS; 0.6 M NaCl; 4mM EDTA; 80 mM Tris-HCl (pH 7.8). Hybridization was initiated at 50°C and temperature was decreased to 37°C over a 6h period. Hybrid selected mRNA was translated in a wheat germ extract (BRL) using 35s_met or 3jj.[eu (NEN) as radioactive amino acid. Proteins were separated on SDS containing 25% acrylamide gels. Gels were fluorographed to Kodak XAR-5 films.

DNA sequencing

pPsENOD12inser twa s sequenced usingth edoubl e stranded dideoxy chain termination procedure (Korneluk et al, 1985). Both strands were sequenced entirely. Additionally, the insert of a second, independently obtained ENOD12 clone was sequenced, spanning nt 16-565 of the sequence in fig.3. Sequence data were stored and analyzed using programs written by R.Staden on VAX/VMS and

130 microVAX/VM S computers.

Insitu hybridization

In situ hybridization wasperforme d essentially as described by Cox &Goldber g (1988).Th e inserto fpPsENOD1 2wa sclone d inpT7- 6(kindl yprovide d bydr .S.Tabor )an dantisens eRN Aprobe s were transcribed using T7RN A polymerase (New England Biolabs) and 35S-UTP (NEN, 1000-1500 Ci/mmol)a sth eradioactiv enucleotide ,unlabele dUT Pwa sno tadded .Th eprobe swer epartiall ydegrade d to an average length of 150 nucleotides by heating for 90 min. at 60°C in 0.2 MNa2CO3/0. 2 M NaHCC>3. Plant tissues were fixed with 3%paraformaldehyde , 0.25% glutaraldehyde in 0.1 Msodiu m phosphatebuffe r pH7.4 .Dehydratio n wasperforme d ingrade dethano lan dxylo lserie san dtissue swer e embedded in paraplast (Brunswick). 7 |xm thick sections wereattache d to poly-L-lysine-coatedslides . Sections were deparaffinized with xylol and rehydrated through a graded ethanol series. They were pretreated with 1 jig/ml proteinase Ki n 100m MTris/HCl ,p H 7.5,5 0m MEDT Aa t37° C for 30 min. and 0.25% acetic anhydride in 0.1 M triethanolamine pH 8.0 at room temperature for 10 min. Subsequentlythe ywer edehydrate di na grade dethano lserie san dai rdried .Section swer ehybridize dwit h RNA probes (106 cpm/ml) in 50%formamide , 0.3 MNaCl , 10m MTris/HC lp H 7.5, 1m MEDTA , 10%dextra n sulphate, lx Denhardt's,50 0ng/m l poly-A, 150ng/m l yeast tRNA and 70m MDT Tfo r 16hr s at42°C . Washing wasperforme d in 4 x SSC, 5 mMDT Ta t room temperature. Next sections were treated with 20 ng/ml RNase Ai n 0.5 M NaCl, 5m MEDTA , 10m MTris/HC lp H 7.5 at 37°C for 30 min. and washed in the samebuffe r with 5 mMDT T at 37°C for 30 min.Th e final wash was twice2 x SSC ,1 mM DTTa troo mtemperature .Section swer edehydrate di ngrade dethano lan dai rdried . Slides were coated with Kodak NTB2 nuclear emulsion diluted with an equal volume 600 mM ammoniumacetatean dexpose dfo r 1-4 weeksa t4°C .The ywer edevelope di nKoda kD1 9develope rfo r3 min. andfixe d inKoda kfix. Section swer estaine dwit h0.025 %toluidin eblu eO fo r 5min . andmounte d with DPX.

Polymerasechain reactions (PCR)

1 ng of the synthetic oligomer S'-CGTGCAAATGTGACTTCTTG-S', complementary to nt. 263-283o f theENOD1 2cDN Asequence ,an d 1 figroo thai rtota lRN Awer eanneale db yheatin g 3 min. at 85°C in 10 \i\ annealing buffer (250 mM KCl, 1 mM EDTA and 10 mM Tris/HCl, pH 8.3), incubating for 30min . at 52°C,an dgraduall y coolingt o42° Cfo r 30min . 15 nicDN Abuffe r (10 mM MgCl2,8 mMDTT ,0. 4 mMo f all four dNTPs,an d2 5 mMTris/HCl ,p H8.3 ) and 5U AM Vrevers e

131 transcriptase(Lif eScience )wer eadded ,an dENOD12-specifi ccDN Awa ssynthesize da t42° Cfo r6 0min . Then 55u lTa qpolymeras ebuffe r (30m M(NH4)2S04 ,9 m MMgCl2,1 0 mMb-mercaptoethano lan d 100m MTris/HCl ,p H8.8) ,2 0u l 5m MdNTPs , 10u l DMSO, 1.25 ul 2M KCl , 1 (lgo f the synthetic oligomer 5'-CTTGTCCTCACTAGTGTT-3'(n L21-4 1o f theENOD1 2cDN Asequence) ,an d2 U Ta q polymerase(Cetus )wer eadded . Themixtur ewa sheate dfo r 3 min. at92°C ,anneale dfo r2 min . at52°C , and 12-16o f thefollowin g amplification cycleswer eperformed : 5min .a t70°C ; 1 min.a t92°C ;an d1 min. at 52°C. Amplified cDNA was ethanol precipitated, the pellet was dissolved in TE buffer and nucleicacid swer eseparate do na 2 %agaros egel . UponDN Atransfe rblottin gth eamplifie d fragment was visualizedb yhybridizatio nt o^P-labele dENOD1 2inser t Asestablishe db ythre einitia lexperiment sth e concentration differences between amplified ENOD12 cDNA from the different root hair RNA preparationsreflecte d differences inth einitia lmRN Aconcentrations : 1) Differences inENOD1 2 mRNA concentrations asreveale db y thePC R method matched the differences revealed by RNA transfer blot analysiso f 20u g root hair RNA (cf figure 6an d IB, WTan d nodA"lanes) ;2 )PC R experiments ona dilution series of total nodule RNA revealed that, with 12-16 amplification cycles, initial mRNA concentrationdifference s werereflecte d inth edifference s ofth eamount so famplifie dENOD1 2cDNA ; 3) amplification ratesi ndifferen t totalroo thai rRN Apreparation s werecompare db ytakin g samples after different numberso fcycles .I tappeare dtha tth eENOD1 2cDN Aconcentratio nindee dincrease dwit hth e samerat ei ndifferen t roothai rRN Apreparations .

Primerextension analysis and RNA sequencing

Thesyntheti coligome r5'-AGGTGATATTGAGCAAGTCC-3' ,complementar yt onucleotid e70 - 90o f thepPsENOD1 2sequence ,wa s'^P-labele d using T4 polynucleotide kinase (Pharmacia). 1.10° cpmo f thisprime rwa scoprecipitate d with2 0|x gtota lRNA .Nuclei cacid swer eresuspende d in6.2 5 |xl annealingbuffe r (50m MTris/HCl ,p H 8.2, 60 mMNaCl , 10m MDTT) ,pu t at 68°C, and allowed to cooldow nt o 35°C.2.2 5 ulR Tbuffe r (250m MTris/HCl ,p H 8.2, 30m MMgC12 ,50 0m MNaCl ,5 0 mMDTT )2. 5 ul dNTPmixtur e (2mM )an d0. 5 ul AMVrevers etranscriptas e (Life Science,2 5U/nl ) wereadde d andprime r extension wasperforme d at45° C for 20min .Subsequently , 1 uiRNas eA wa s was added and incubation was prolonged for 15 min. The mixture was extracted once with phenol/chloroform (1:1)an dethano lprecipitate d using2 ug/m ltRN Aa sa carrier .Upo nresuspensio ni n 1.5 ui H2O loading buffer was added and after denaturation samples were analyzed on a 6% polyacrylamide/8M ure asequencin ggel . For RNA sequencing 5.10^cp m primer was coprecipitated with 80 jig total RNA. The precipitatewa sresuspende di n 12.5 ulannealin gbuffe r andanneale da sdescribe dabove .4. 5u lR Tbuffer , 5 ul dNTP mixture (2 mM) and 25 U AMV RT were added. 4 ul of this solution was added to four separatetubes ,containin g 1 ulo fon eo f thefou rdideoxyNTP' s (800uM) .Extensio n wasperforme d for

132 20min .a t45°C . Subsequently 1u io f thedNT Pmixtur e wasadde dfo r a chasereactio n for 15min .a t 45°C.Sample swer eextracted ,precipitated ,an dsubjecte d toge lanalysi sa sdescribe dabove .

RNase mapping

Theregio n ofpPsENOD1 2fro m theBgl lsit ea tpositio n 246u pt oth eBal lsit ea tpositio n44 8 containing 202nucleotide s from the 3' end of thecDN A wasclone d into pT7-6.Antisens e RNA was transcribed from this plasmid, after linearization immediately behind the insert, using T7 RNA polymerase (NewEnglan d Biolabs)an d32 -P UTP(NEN )a slabele d nucleotide.5 0u Munlabele dUT P wasadde dt oensur e95-100 %ful l sizetranscription .Afte r synthesisth ereactio nwa sstoppe dwit hDNase l (Boehringer)extracte donc ewit hphenol/chlorofor m (1:1)an donc ewit hchloroform ,an dunincorporate d nucleotideswer eremove db yspin-colum nchromatograph y(Maniati se ta l1982) . ForRNas emappin g 1.10^ cpmo fprob ewa scoprecipitate d with2 0|i gtota lRNA . Pelletswer e resuspended in 30|x lhybridizatio n buffer, and following denaturation at 85°C for 5mi n themi xwa s incubated 16h ra t45° C (Melton et al, 1984).Digestio n with64 0t o576 0U/m lRNas eT l (BRL)wa s performed at 45°C for 60 min, RNases were removed by an additional incubation for 15mi n with proteinaseK an dSD Sa t 37°C,al la sdescribe d byMelto ne ta l(1984) .Th emixtur ewa sextracte dwit h phenol/chloroform (1:1)an dprecipitate d with carrier tRNAan dethanol .Th epelle t wasresuspende di n H2O,loadin gbuffe r wasadde dan dupo ndenaturatio nsample swer eanalyze do n6 %Polyacrylamide /8 M ureasequencin ggels .

REFERENCES

Averyhart-Fullard,V. ,Datta ,K. ,an d Marcus,A .(1988) .A hydroxyprolin erich protei n in thesoybea n cellwall . Proc. Natl.Acad .Sei .US A85 ,1082-1085 . Bakhuizen, R., van Spronsen,P.C. ,Diaz ,C.L. , and Kijne, J.W. (1988a).Rearrangemen t of cytoplasm and endoplasmic microtubules in cortical cells of Pisum sativum L.root s prior toinfectio n by Rhizobium leguminosarum biovar viciae. In: The plant cytoskeleton in theRhizobium-legum e symbiosis,Ph. DThesi sLeide nUniversity ,Th eNetherlands ,pp .57-81 . Bakhuizen,R. , vanSpronsen ,P.C. ,an dKijne ,J.W .(1988b) .Cel l walldegradatio nan dinfectio n thread initiation in pea rootcortica l cellsdurin g infection by Rhizobium leguminosarum biovarviciae . In:Th eplan t cytoskeleton in theRhizobium-legum e symbiosis,Ph. D ThesisLeide n University, TheNetherlands ,pp . 83-107.

133 Bauer.W.D.(1981) .Infectio n of legumesb yRhizobia .Ann . Rev.Plan tPhysiol .32,407-449 . Bennett,M.J. ,Lightfoot , D.A.,an dCullimore ,J.V .(1989) .cDN Asequenc ean ddifferentia l expression of thegen eencodin g the glutamine synthetase polypeptideo f Phaseolus vulgarisL .Plan tMol . Biol. 12, 553-565. Bergmann,H. ,Preddie ,E. ,an dVerma ,D.P.S .(1983) .Nodulin-35 :a subuni to f specific uncase(uncas e II)induce dan dlocalize d inth euninfecte d cellso fsoybea nnodules . EMBOJ . 2,2333-2339. Bisseling,T. ,Va nde n Bos,R.C. ,an d VanKammen ,A .(1978) .Th eeffec t of ammonium nitrateo nth e synthesis of nitrogenase and theconcentratio n of leghemoglobin inpe aroo t nodules inducedb y Rhizobium leguminosarum.Biochim .Biophys .Act a539 , 1-11. Bogusz, D., Appleby, CA., Landsmann, J., Dennis, E.S., Trinick, M., and Peacock, W.J. (1988). Functioning hacmolglobini nnon-nodulatin gplants .Natur e331,178-180 . Brisson. N., and Verma, D.P.S. (1982). Soybean leghemoglobin gene family: normal, pseudo, and truncatedgenes .Proc .Natl .Acad .Sei . USA79,4055-4059 . Callaham, D.A., andTorrey , J.G. (1981).Th e structural basis for infection of root hairs of Trifolium repensb yRhizobium . Can.J . Bot 59,1947-1964 . Chen,J. ,an dVarner ,J.E .(1985) .Isolatio nan dcharacterizatio no fcDN Aclone sfo rcarro textensi nan da proline-rich33-kD aprotein . Proc.Nad .Acad .Sei . USA82,4399-4403 . Corbin, D.R., Sauer, N. and Lamb, C.J. (1987). Differential regulation of a hydroxyproline-rich glycoproteingen efamil y inwounde dan dinfecte d plants.Mol . andCell .Biol . 7,4337-4344. Cox,K.H. ,DeLeon ,D.V. ,Angerer ,L.M. ,an dAngerer ,R.C .(1984) .Detectio n of mRNAsi nse aurchi n embryosb yi nsit u hybridization usingasymmetri cRN Aprobes .Dev .Biol .101,485-502 . Darvill,A.G. ,an d Albersheim,P .(1984) .Phytoalexin san d their elicitors -a defense against microbial infection inplants .Ann .Rev .Physiol .35 , 243-275. DeVries , S.C., Springer,J. , and Wessels,J.H.G . (1982).Diversit y of abundant mRNA sequences and patternso fprotei nsynthesi si netiolate dan dgreene dpe aseedlings .Plant a 156,129-135. Dickstein, R., Bisseling,T. , Reinhold, V.N.,an d Ausubel,F.M . (1988).Expressio n of nodule-specific genesi nalfalf a rootnodule sblocke da ta nearl ystag eo fdevelopment .Gene san dDevelopmen t2 , 677-687. Djordjevic, M.A., Gabriel,D.W. , andRolfe , B.G.(1987) .Rhizobiu m -th erefine d parasiteo flegumes . Ann.Rev .Phytopathol . 25,145-168 . Faucher, C, Camut,S. , Dénarié,J. , and Truchet, G. (1989).Th e nodHan dnodQ host rangegene so f Rhizobiummeliloti behave as avirulence genes in Rhizobium leguminosarum bv. Viciae and determine changes in theproductio n of plant-specific extracellular signals.Mol . Plant-Microbe Interac. 2,291-300 Fortin,M.G. ,Zelechowska , M.,an d Verma,D.P.S .(1985) .Specifi c targeting of membrane nodulinst o thebacteroid-enclosin gcompartmen ti nsoybea nnodules .EMB OJ .4,3041-3046 . Fortin, M.G., Morrison, N.A., and Verma, D.P.S. (1987). Nodulin-26, a peribacteroid membrane

134 nodulin isexpresse d independently of the development of theperibacteroi d compartment.Nucl . AcidsRes . 15,813-824. Franssen, HJ., Nap,J.P. ,Gloudemans ,T. , Stiekema,W. , Van Dam,H. ,Govers ,F. ,Louwerse ,J. ,Va n Kammen,A. , and Bisseling, T. (1987).Characterizatio n of cDNA for nodulin-75o f soybean:a geneproduc tinvolve d inearl ystage so froo tnodul edevelopment .Proc .Natl .Acad .Sei .US A 84, 4495-4499. Franssen, HJ., Scheres, B.,Va n De Wiel,C , and Bisseling, T. (1988).Characterizatio n of soybean (hydroxy)proline-rich early nodulins. In: Molecular Genetics of Plant-Microbe interactions, RPalacios andD.P.Verma ,eds .(St .Paul :AP SPress) ,p p321-326 . Gloudemans,T. , Bhuvaneswari,T.V. ,Moerman , M.,Va nBrussel ,T. , VanKammen ,A. ,an dBisseling , T.(1989) .Involvemen to fRhizobiu m leguminosarum nodulation genesi ngen eexpressio n inpe a roothairs .Plan t Mol.Biol . 12,157-167 . Govers,F. ,Gloudemans ,T. ,Moerman ,M. ,Va nKammen ,A. ,an dBisseling ,T .(1986) .Rhizobiu m nod genesar einvolve di ninducin ga nearl ynoduli ngene . Nature323,564-566 . Govers, F., Nap, J.P., Van Kammen, A, and Bisseling, T. (1987) Nodulins in the developing root nodule. PlantPhysiol .Biochem .25 , 309-322. Horvath, B.,Kondorosi ,E. , John, M., Schmidt,J. , Török,J. , Gyórgypal,Z. , Barabas,I. ,Wieneke ,U. , Schell,J. , andKondorosi ,A .(1986) .Organization , structure and function of Rhizobium meliloti nodulationgene sdeterminin ghos tspecificit y inalfalfa . Cell46,335-343 . Jacobs, F.A., Zhang, M.,Fortin , M.G., and Verma,D.P.S .(1987) . Several nodulins of soybean share structuraldomain sbu tdiffe r inthei rsubcellula rlocations . Nucl. AcidsRes . 15,1271-1280. Korneluk,R.G. ,Quan ,F. ,an d Gravel,R.A . (1985).Rapi d and reliable dideoxy sequencing ofdouble - strandedDNA . Gene40,317-323 . Libbenga, KU., Van Iren, F., Bogers, R.J., Schraag-Lamers, M.F. (1973). The role of hormones and gradients inth einitiatio n ofcorte xproliferatio n andnodul eformatio n inPisu msativu mL . Planta 114, 29-39. Libbenga, K.R., and Bogers, RJ. (1974). Root nodule morphogenesis. In The Biology of Nitrogen Fixation,A .Quispel ,ed . (Amsterdam,Nort h HollandPublishin gCompany) ,pp . 430-472. Long,S J* .(1989) .Rhizobium-legum enodulation :lif e togetheri nth eunderground .Cel l56,203-214 . Maniatis, T., Fritsch, E.F., and Sambrook,J . (1982). Molecular cloning: a laboratory manual. (Cold SpringHarbor ,NY ,Col dSprin gHarbo rLaboratory ) Melton, D.A., Krieg,P.A. ,Rebabliati , M.R., Maniatis,T. ,Zinn ,K. ,an d Green, M.R. (1984). Efficient 'in vitro' synthesis of biologically active RNA and RNA hybridization probes from plasmids containinga bacteriophag eSP 6promoter .Nuc .Acid sRes . 12, 7035-7056. Mullis, K.B.,an dFaloona ,F.A . (1987).Specifi c synthesiso f DNAi n vitro viaa polymerase-catalyzed chainreaction .Meth .Enzymol . 155, 335-350. Newcomb, W. (1976). A correlated light and electron microscopic study of symbiotic growth and

135 differentiation inPisu msativu mroo tnodules .Can .J .Bot .54,2163-2186 . Newcomb,W . (1980).Contro l of morphogenesis and differentiation of pea root nodules. InNitroge n Fixation,vo lÜ , W.E.Newto nan dW.H .Orme-Johnson , eds.(Baltimore :Universit y ParkPress) , pp87-102 . Rigaud, G., Grange, T., and Pictet, R. (1987). The use of NaOH as transfer solution onto nylon membranedecrease sth ehybridizatio nefficiency . Nuc.Acid sRes . IS,857 . Rogers, S.O., and Bendich, AJ. (1988). Extraction of DNA from plant tissues. In Plant Molecular Biology Manual, S.B. Gelvin, and R.A. Schilperoort, eds. (Dordrecht, Kluwer Academic publishers),A6 , pp. 1-10. Saiki,S. , Scharf,F. ,Faloona ,K. , Mullis,G. , Horn,G. ,Erlich ,H. ,an d Arnheim,N .(1985) .Enzymati c amplification of b-globin genomic sequences andrestrictio n siteanalysi s for diagnosiso f sickle cellanemia .Scienc e230,1350-1354 . Sandal, N.N., Bojsen, K., and Marcker, K.A. (1987). A small family of nodule specific genes from soybean.Nucl .Acid sRes . 15,1507-1519 . Schmidt, J., Wingender, R., John, M., Wieneke, U., and Schell, J. (1988).Rhizobium meliloti nodA andnodB genesar einvolve d ingeneratin gcompound stha tstimulat e mitosiso f plantcells . Proc. Nad.Acad .Sei . USA85 ,8578-8582 . Showalter, A.M., Bell, J.N., Cramer, C.L., Bailey, J.A., Varner, J.E., and Lamb, CJ. (1985). Accumulation of hydroxyproline-rich glycoprotein mRNAs in response to fungal elicitor and infection. Proc. Natl.Acad .Sei .US A82,6551-655 5 Spaink, H.P.,Okker , R.J.H., Wijffelman, CA., Pees,E. , and Lugtenberg, B.J.J.(1987) .Promoter s in thenodulatio nregio no f theRhizobiu m leguminosarum symplasmi dpRLUI .Plan tMol .Biol .9 , 29-37. Tigney,S.V. ,Walker ,E.L. ,an dCoruzzi ,G.M .(1987) .Glutamin esyntetas egene so fpe aencod edistinc t polypeptideswhic har edifferentiall y expressedi nleaves ,root san dnodules .EMB OJ .6,1-9 . Turgeon, B.G., and Bauer, W.D. (1985). Ultrastructure of infection-thread development during the infection ofsoybea nb yRhizobiu mjaponicum . Planta 163,328-349. Van Brussel, A.A.N., Pees, E., Spaink, HP., Tak, T., Wijffelman, CA., Okker, R.J.H., Truchet, G., and Lugtenberg, B.JJ. (1988). Correlation between Rhizobium leguminosarum nod genes and nodulation phenotypes on vicia. In Nitrogen fixation, hundred years after, H. Bothe, F.J. de Bruijn,an dW.E .Newton ,eds .(Stuttgart-Ne w York: GustavFisher) , p.483 . Vance,C.P .(1983) .Rhizobiu m infection andnodulation :a beneficia lplan tdisease ? An. Rev. Microbiol. 37,399-424 . VanD eWiel ,C , Scheres,B. ,Franssen ,HJ. , vanLierop ,M.J. , van Lammeren,A. ,Va n Kammen,A. , andBisseling ,T .(1990) .Th eearl ynoduli ntranscrip tENOD 2 islocate di nth enodul eparenchym a (innercortex )o fpe aan dsoybea nroo tnodules .EMB OJ. ,i nth epress . VanKammen ,A .(1984) .Suggeste dnomenclatur efo rplan tgene sinvolve di nnodulatio nan dsymbiosis .

136 Plant Mol.Biol .Rep .2 , 43-45. Von Heijne, G. (1983).Pattern s of amino acids near signal-sequence cleavage sites.Eur .J . Biochem. 133, 17-21. Wijffelman, CA., Pees, E., Van Brussel, A.A.N., Okker, RJ.H., and Lugtenberg, B.J.J. (1985). Genetic and functional analysis of the nodulation region of Rhizobium leguminosarum sym PlasmidpRLUI .Arch .Microbiol . 143, 225-232. Zaat, S.A.J.,Va n Brussel,A.A.N. ,Tak ,T. ,Pees ,E. , andLugtenberg , BJ.J . (1987).Flavonoid s induce Rhizobium leguminosarum toproduc enodDAB Cgene-relate dfactor s thatcaus ethick ,shor troot s androo thai rresponse so ncommo n vetch.J .Bacteriol . 169,3388-3391.

137 CHAPTER 8

A molecular marker for the development of the uninfected cells in the central tissue of pea (Pisum sativum) root nodules.

Clemensva nd eWiel ,Marj a Moerman,Jok eva n Beckum,Marce lva nde nHeuvel , HenkKieft , Andréva nLammeren ,Alber tva n Kammen,an dTo n Bisseling SUMMARY

Thepe a early nodulin Nps-40' that has been identified as an invitro translation product of nodule mRNA,i s shown tocomigrat ewit h a4 0k Dprotei n isolated from pea nodules. Like the Nps-40' in vitro translation product, the putative Nps-40' in vivo protein appearsrelativel y earlydurin gnodul edevelopment , well aheado f thelat enoduli n leghemoglobin. An antiserumraise d against theputativ eNps-40 ' invivo protei n reacted with both the putative Nps-40' in vivo protein and the Nps-40' in vitro translation product,affirmin g thatthes emus t beidentica lproteins .Th eantiseru m alsoreacte dwit h some other 40 kD proteins that were identified, using an antiserum against glutamine synthetase (GS)o f bean,a svariou s forms of GS.Th ecros sreactio n of the anti-Nps-40'- serumwit hth eothe r4 0k Dprotein scoul d bediminishe d bytitratio n withG Sfro m bean. Using this titrated antiserum, the early nodulin was shown by immunolabeling to be exclusively located in the cytosol of the uninfected cells in the early and late symbiotic zoneo fpe anodules .Thi simplie stha tth euninfecte d cellsi nth eindeterminat epe anodul e have a nodule-specific role in the symbiosis with Rhizobium bacteria,jus t like in the noduleso f thedeterminat e growth type,suc h asthos eo f soybean. Furthermore, theanti - Nps-40'-serum cross-reacted with the nodulinsNvs-4 0an dNms-3 0o f the indeterminate noduleso f common vetch andalfalfa , respectively. Thepossibl efunctio n of Nps-40'and the regulation of the expression of the gene for this early nodulin are discussed in the lighto fearlie rstudie so nit soccurrenc evariou sindeterminat e nodulesblocke da t different stageso fdevelopmen t duet oth eus eo fengineere d bacterial strains.

INTRODUCTION

Root nodules of legumes formed upon infection with bacteria from the genera Rhizobium orBradyrhizobium contai n a specialized tissue, the central tissue, in which nitrogen fixation occurs. Two cell types can be recognized in this central tissue: the infected cells, which harbour the nitrogen-fixing bacteroids, and the uninfected cells, whichoccu rintermingle d withth einfecte d cells.Th euninfecte d cellsar echaracterize db y thepresenc eo f alarg ecentra lvacuol e andplastids ,whic h often containprominen t starch grains. Legume root nodules can be divided into two types: the determinate and the

141 indeterminate growth type (for review:Sprent , 1980).Durin g theformatio n of both types of nodules,nodule-specifi c proteins -nodulins- aremade .Mos t of thenodulin s localized sofar , are found in the infected cell type of determinate and/or indeterminate nodules (Nap &Bisseling , 1990; Scheres et al., 1990). Inth edeterminat e noduleso f soybean (Glycine max)a singl e nodulin was shown to occur specifically in the uninfected cells (Bergmann et al. 1983). This nodulin is a nodule-specific form of uricase(n-uricase )occurrin g inth eperoxisome so f the uninfected cells in soybean (Van den Bosch &Newcomb , 1986;Kanek o &Newcomb , 1987),an d also in cowpea (Vigna unguiculata) nodules (Webb & Newcomb, 1987). Uricase catalyzes one of the final steps in the ammonium assimilation pathway, leading to the production of the ureides, allantoin and allantoic acid. These compounds are in these speciesth emai nfor m in which thefixe d nitrogen istransporte d toothe rpart so f theplan t (cf. Schubert, 1986).I n thedeterminat e soybean nodule,th e uninfected cells have been shown toconstitut e a coherent network throughout the whole central tissue, which will allow for an efficient transportation of the ureides to the peripherally located vascular bundleso f thenodul e (Selker, 1988). Theindeterminat e pea (Pisumsativum) nodulesexpor tfixe d nitrogen in the form of aminoacid slik e glutamine and asparagine (Schubert, 1986).I n thesenodules ,uricas e isno tforme d andnodulin s specific for theuninfecte d cellso f thistyp eo fnodul ehav eno t yet been identified. Consequently, it is unclear whether the uninfected cells in the indeterminate nodules alsohav ea specifi c rolei n theassimilatio n of thefixe d nitrogeno r someothe rfunctio n related toth esymbiosis . In the indeterminate nodule the cells are of graded age from the persistent distal meristem toth eroo tattachmen t point,whic h enablesth eexaminatio n of thedevelopmen t of different cell types in a single longitudinal section of a nodule (Newcomb, 1981).I n that way, it was found that the development of the central tissue in pea nodules is attended byth e sequentialexpressio n of severalnoduli n genes (Scherese tal. , 1990).Th e firstnoduli n genes expressed during development are the early nodulin genesENOD1 2 andENOD5 .Transcript s of these genes aredetectabl e inth einvasio n zone,i.e .th e zone of cellsimmediatel y adjacent to thedista l meristem of the nodule.I n this zone infection threads penetrate cells just added by the meristem, and release rhizobia into them. Whereas ENOD12 messenger is present in all cells of the invasion zone, the ENOD5 genei sonl y expressed incell stha t have beenpenetrate d by aninfectio n thread (Scheres et al., 1990). So, as soon asa cytologica l recognition of infected cellsi s possible, the infected cells are marked by the presence of ENOD5 transcripts. At this stage, it is difficult to determine with certainty whether a cell is developing into the uninfected cell type. All

142 cellsar edevelopin ga centra lvacuol ean di ti shar dt oexclud etha ta bacteria-fre e cellwil l after all becomeinfected . Asmentione d above,w ear eno t awareo f anyrepor t indicating that the uninfected cells of indeterminate nodules are also marked by the expression of specific genes.A s a consequence, there are at present no molecular markers indicating whether a cell is determined to become an uninfected one, and neither it is feasible to define at which stageo fdevelopmen t acel lbecome scommitte d todevelopmen t intoth e uninfected type. This hampers the identification of the mechanism underlying development of thiscel ltype .Whil e thedifferentiatio n intoa ninfecte d cellmigh tjus t be relatedt oth ereleas eo f bacteriafro m aninfectio n thread,i ti sunclea rwhic heven ttrigger s thedevelopmen t intoa nuninfecte d cell. In this report we present evidence that the indeterminate pea nodule contains a nodule-specific protein that is found exclusively in the uninfected cells and probably corresponds to Nps-40', formerly identified as a pea early nodulin by Govers et al. (1985). With antibodies against Nps-40' we studied the time course of its appearance during pea nodule development and also its occurrence in other legume species. The resultso fou rpresen t studytogethe r withthos eo f earlier studieso nNps-40 ' genewil lb e used to discuss how the differentiation into uninfected cells might be induced by Rhizobium.

RESULTS

Raisingof an antiserum against the Nps-40' protein

The early nodulin Nps-40' was originally identified as an in vitro translation product of mRNA from pea root nodules by Govers et al. (1985). By comparing the proteins inextract so fpe anodule s androots ,afte r separation on two-dimensional gels,a nodule-specific protein was detected, with a molecular weight and iso-electric point comparable to that of the Nps-40' in vitro translation product. This nodule-specific protein already occurred in extracts from root nodules at day 10 after sowing and inoculation; this is well ahead of the appearance of the leghemoglobins at day 13,i n conformity toth etime cours eo f appearance of Nps-40' asdetecte d byin vitro translatio n of nodule mRNA.W econfirme d that the4 0k Dnodule-specifi c protein comigrated with the Nps-40' in vitro translation product, by electrophoresis of a mixture of protein extracted from nodules and in vitro translation products of nodule RNA (results not

143 •sativa alfalfa

i * r 4» 1 « * . * ~\Wfè -gr.

i • *•

Nvs-40 N-40' » t N30

Fig. 1.Two-dimensiona l gel electrophoresis patternso f thein vitro translation products of nodule mRNA (uppe rpanel )an do fth ein vitro translationproduct so fnodul emRN Aimmunoprecipitate db ya n anti-Nps-40'-senim(lowe rpanel ) frompea ,commo nvetc h (V. sativa),an dalfalfa , asindicate dabov eth e panels. Inpe aan dvetch , theNps-40 '(N-40' )an dNvs-4 0in vitro translationproducts , respectively,ar e precipitated by the anti-Nps-40'-serum, together with some other 40 kD products; in alfalfa predominantlyth eNms-3 0in vitro translatio nproduct .

shown). In order to raise an antiserum against Nps-40', we tried to purify the putative Nps-40' in vivo protein by ion exchange chromatography. However, for unknown reasons,w e did not succeed inrecoverin g the protein after chromatography on aDEA E column. Therefore, we separated proteins from ten-day-old nodules by preparative SDS/PAGE and used the mixture of proteins eluted from the 40 kD band to raise an antiserum. The specificity of this antiserum was tested by immuno-precipitation of in vitro translation products of nodule RNA. As shown in Fig. 1, the antiserum predominantly precipitated theNps-40 'protein .Hardl y anyprecipitatio n ofprotei n could bedetecte d upon additiono f theantiseru m toth eproduct sobtaine d byin vifro-translation ofroo t RNA (result not shown). On a Western blot of proteins from pea nodules, separated by two-dimensional SDS/PAGE,th e antiserum bound to several 40k Dproteins ,on eo f them beingNps-40' . Mostadditiona l proteins,bu tno tth eputativ e Nps-40' protein,reacte d with an antiserum

144 against glutamine synthetase from bean (Phaseolusvulgaris), suggesting that they are different isoforms of glutamine synthetase.Therefore , wedecide d toremov e antibodies thatmigh tb edirecte d againstglutamin e synthetase,b ytitratio n ofth eanti-Nps-40'-seru m with purified glutamine synthetase from bean. This resulted indeed in a more specific antiserum, as on Western blots of nodule proteins, the purified antiserum then predominantly bound toth e Nps-40'protein .A slight crossreactio n with some other 40 kD proteins of different pi was still visible (results not shown). These proteins might represent posttranslationally modified forms of Nps-40',bu t aremainin g cross-reaction with some forms of glutamine synthetase, or even other 40 kD proteins can not be excluded.Th etitrate d antiserum was used to localize theNps-40 'protei n on sectionso f pea nodules. Besides we examined whether a protein similar to Nps-40' occurred in other leguminous species.Fo r that purpose, weperforme d with the crude anti-Nps-40'-serum an immuno-precipitation experiment on proteins produced by in vitro translation of noduleRN Afro m vetch (Viciasativa) andalfalf a (Medicagosativa). As showni nFig . 1, theanti-Nps-40'-seru m precipitated translationproduct spreviousl y identified asth eNvs - 40 (Moerman et al., 1987) and Nms-30 nodulins (Lang-Unnasch & Ausubel, 1985), respectively. Thissuggest s thatearl y nodulinso f theNps-40 'typ ear elikewis e important in other legumes of the indeterminate growth type. No proteins were observed to precipitate from invitro translatio n products ofmRN A from the determinate noduleso f soybean withth e anti-Nps-40'-serum.

Localizationof Nps-40' in pea nodules

Sections of pea nodules were incubated with the titrated antiserum against Nps- 40' andbindin go f theseru mwa svisualize d bysilver-enhance dprotei nA-gol dlabelling . As shown inFig . 2A,th e titrated anti-Nps-40'-serum bindspredominantl y to uninfected cells,includin g the uninfected cells that form the boundary layer delimiting the central tissue from theperiphera l tissueo f the nodule.Moreover , areactio n with the anti-Nps - 40'-serumi salread yclearl ydetectabl ei nth eearl y symbiotic zoneo fth enodule ,whic hi s consistentwit hth eearl y appearanceo f Nps-40'durin gnodul edevelopment , asdescribe d above. The amount of anti-Nps-40'-serum binding diminishes in older parts of the late symbiotic zone(result sno tshown) . Since thepossibilit y of cross-reaction of the anti-Nps-40'-serum with glutamine synthetaseremained , sectionso fpe anodule swer eals oincubate d withth euntitrate d anti- Nps-40'-serum (Fig. 2B) and with a serum against glutamine synthetase of bean (Fig.

145 2C),respectively . By far the largest amounts of glutamine synthetase were detected in infected cells of the late symbiotic zone (Fig. 2C). This preponderance of glutamine synthetase in infected cells had already been shown for soybean (Verma et al., 1986). Contrary to the titrated anti-Nps-40'-serum, the untitrated antiserum bound not only to the uninfected cells, but also to the infected cells (Fig. 2B). These observations taken together imply that the binding of the anti-Nps-40'-antiserum to uninfected cellsca n not beattribute dt oa possibl ereactio nwit hglutamin e synthetase. The subcellular location of thereactio n with the anti-Nps-40'-serum was studied with the aido f electron microscopy, in which antibody binding wasvisualize d by 10 nm gold particles coupled to protein A. In this way, it was shown that anti-Nps-40'-serum binding waspredominantl y to the cytoplasm of theuninfecte d cells (Fig. 3).Thus ,ver y little antiserum binding was detected in the nucleus (Fig. 3A). Hardly any antiserum binding could be detected in mitochondria (Fig. 3B),plastid s or the vacuole. Thus, the änti-Nps-40'-serumbindin g appearst ob erestricte d toth ecytosol .A possibl e binding to proteins in theendoplasmi creticulu m can notb erule d out, since thelatte r cell organelle wasno tpreserve d wellenoug h inou rmateria l torecogniz ei t unequivocally.

Fig. 2.Immunogold-silve rlabelin go f semithincryosection sfro m atwo-week-ol dpe anodul e elicited byR. leguminosarum bv. viciae. Theepipolarizatio n micrographssho wdetail so fth e centraltissu e in the late symbiotic zone, incubated with anti-Nps-40'-serum titrated with GS from bean (A),wit h untitrated anti-Nps-40'-serum (B),an dwit hanti-bea nGS-seru m (C),respectively .I nA ,a prominent labelingsigna li nth efor mo fwhit e silvergrain si svisibl eove rth eparietall ylocalize dcytoplas mo fth e uninfected cells (UC); in B,silve rgrain sar evisibl eove rbot hth einfecte d (IC) andth euninfecte dcells ; andi nC ,silve rgrain sar epredominantl ypresen tove rth einfecte dcells .Ba r= 5 0urn .

146 DISCUSSION

Theresult s show thespecifi c occurrence of aprotei ni nth euninfecte d cellso fth e central tissue of pea nodules. This protein most likely corresponds to the early nodulin Nps-40',previousl y identified bytwo-dimensiona l SDS/PAGEo f invitro translate d pea nodule mRNA byGover se t al.(1985) . ForNps-40 'th ecorrespondin g cDNAi s stilllacking .I n ase to fpe aearl y nodulin cDNA clones (Scheres et al., 1990), none could be identified as coding for Nps-40'. Therefore, in order toobtai n aprob efo r thelocalizatio n of theexpressio n of theNps-40 ' gene,a nantiseru m wasraised ,eve n though wedi d not succeed in purifying from nodule extractsth eprotei n thatexactl ycomigrate swit hth ein vitro translated Nps-40'fro m other 40 kD proteins. These other 40 kD proteins most probably were isoforms of GS, as indicated by their reaction with an antiserum against purified GSfro m bean. Therefore, the specificity of our antiserum could beimprove d bytitration wit h GS from bean.Th e antiserumreacte d with thein vitro translate d Nps-40'i nimmunoprecipitatio nexperiment s and with the putative in vivo protein on Western blots, affirming that these must be identical proteins.Th etitrated antiseru m still gavea wea kreactio n onWester n blotswit h 40 kDprotein s other than Nps-40'.Anyhow , the titrated antiserum appeared to be very useful in demonstrating by immunolabeling the specific occurrence of Nps-40' in the uninfected cells,an d gavevirtuall y noreactio n with proteins in the infected cells. Since GS was demonstrated to be prominent in the infected cells and relatively scarce in the uninfected cells, the weak cross reaction of the titrated antiserum with other 40 kD proteins on Western blots may be attributable topost-translationall y modified forms of Nps-40', rather than to residual antibodies against GS.Also , there remains the remote possibility thatth ereactio no f thetitrate d antiserum withuninfecte d cellsi sa tleas tpartl y duet oth epresenc e ofprotein(s )othe rtha n GSo rNps-40' ,a sth eidentit yo f theothe r4 0 kDprotein sha sno tye tbee nconfirme d byindependen tmethods . In accordance with Nps-40' being an early nodulin, the protein specific for the uninfected cells is detectable histologically at a relatively early stage of nodule tissue differentiation, i.e.i n theearl y symbiotic zone.Th eprotei n can,therefore , beregarde d as amarke rfo r thedifferentiatio n ofth euninfecte d celltype . Homologous early nodulins arepresen t in vetch (Nvs-40,Moerma n et al., 1987) and alfalfa (Nms-30, Lang-Unnasch & Ausubel, 1985), as indicated by their cross reaction with the anti-Nps-40'-serum. This suggests that N-40'migh t serve arol e in the uninfected cells in several legumes of the indeterminate nodule growth type. Moreover, theexclusiv epresenc eo f anodule-specifi c protein inth euninfecte d cellso f bothth eearl y

147 148 and late symbiotic zones indicates that, like in the determinate nodules, this cell type mightpla ya specifi c rolei nth esymbiosi si nindeterminat enodule sa swell . It is not apparent from our results whether the Nps-40' protein has a role in assimilatingfixed nitroge n intoa for m suitablefo r transportation toth eres t of theplant , like e.g. uncase in soybean (Schubert, 1986). Such arol e for Nps-40' is not very likely for severalreasons : 1)Nps-40 'doe s not seemt o bea for m of glutamine synthetase,a si t does not bind anti-glutamine synthetase-serum; 2) in pea, the fixed nitrogen is mainly exportedou to fth enodul ei nth efor m ofglutamin e andasparagine ;thes eamin oacid sar e formed inth e infected cells (Schubert, 1986).Therefore , there isn oa priori needfo r an extra specific cell type for theproductio n of these aminoacid sca n bereasoned , like for the ureides in soybean; 3) the relatively early appearance during nodule development argues against arol efo r N-40' innitroge n assimilation. With regard toth eregulatio n of Nps-40' geneexpressio n and thepossibl erol eo f Nps-40'i n thesymbiosis ,th efollowin g observationsi nnodule sa tsom epoin tblocke di n their development are relevant. It was found that an Agrobacterium tumefaciens transconjugant, harbouring a Sym plasmid from Rhizobium leguminosarum bv. viciae, elicits nodules on pea that contain infection threads but no intracellular bacteria in the centraltissue. I ncontrast ,th e sametransconjugan t inducesnodule so nvetc h inwhic hth e release of bacteria from the infection thread and the differentiation of infected and uninfected cells take place. In the vetch nodules, Nvs-40 could be clearly detected (Moerman et al., 1987; cf. Nap et al, 1989),wherea s in the pea nodules Nps-40' was notfoun d (Goverse t al., 1986) (seeTabl e 1).However , in alfalfa, theNms-3 0 transcript is always detected, even in nodules lacking not only intracellular bacteria but also infection threads in the central tissue (Dickstein et al., 1988;Norri s et al., 1988) (Table 1).Thes e so-called "empty"nodule s areelicite d notonl y bya variet yo f bacteria,amon g whichA. tumefacienstransconjugant s carryingnodulatio n sequencesfro m theR. meliloti Sym plasmid, but also by treatment with auxin transport inhibitors, such as TIBA or NPA (Hirsch et al., 1989).

Fig. 3.Immunogol dlabelin go fultrathi nresi nsection sfro m atwo-week-ol dpe anodul eelicite db y R. leguminosarum bv.viciae. Theelectro nmicrograph ssho wdetail so fcell sfro mth ecentra ltissu ei nth e latesymbioti czone , incubatedwit h anti-Nps-40'-serum.A ) Detailo fa n uninfected celli nth eboundar y layer,showin glabelin go fblac kgol dparticle sove rth ecytoplasm ;hardl yan ylabe li spresen tove rth e nucleus(N) . B) Detailo ftw ocontiguou s uninfectedcell s inth eboundar ylayer ; againgol dparticle sar e present overth ecytoplasm , butno tove r themitochondrio n (Mt).C W =cel l wall.C )Detai lo fa n infected cell contiguous witha nuninfecte d cell;gol dparticle sar epresen tove rth ecytoplas m ofth e uninfectedcel l(right) ,bu tlac kove rth einfecte dcel l(left) ,i nwhic hbacteroid s(B )ca nb eobserved .Ba r= 0.3urn .

149 Table 1. N-40' gene expression in nodules induced byA. tumefaciens transconjugants carryingrhizobia l nodgene s

Hostplan t Bacterium ITi nC T BaR N-40'gen eexpressio n Reference

Pea Atum(pSym) -(Nps-40 ' Goverse tal , 1986 Vetch Atum(pSym) +(Nvs-4 0 Moermane tal. ,198 7 Alfalfa Atum(nod) +(Nms-30 ) Dicksteine tal. , 1988

Abbreviations: Atum(pSym) = A. tumefaciens (LBA2712) carrying the symbiotic plasmid of R. leguminosarum bv.viciae; Atum(nod) = A. tumefaciens (1038/pMH36) carrying nod geneso fR. meliloti; IT= infectio n threads;C T= centraltissu eo fnodule ;Ba R= bacterialreleas efro m infection threadi ncentra ltissu eo fnodule .

Thepresenc e of Nvs-40i nth eA. tumefaciens transconjugant-elicited noduleso n vetch,togethe rwit hth eobservatio n thati nthes enodule sinfecte d anduninfecte d cells become differentiated, are consistent with the observation that the homologousNps-40 ' genei sexclusivel y expressedi nth euninfecte d cellso f wildtyp eR. leguminosarumbv . viriae-induced noduleso npea .N oclea r differentiation between infected and uninfected cellsi srecognizabl e in theA.tumefaciens transconjugant-induced noduleso npea ,despit e thepresenc e of infection threads in the central tissue.Thus ,th e lack of detectable Nps- 40' might indicate that indeed nodifferentiatio n of uninfected cells hasoccurred , which might,i nturn ,b eattribute dt oth eabsenc eo fbacteria lreleas ei nth ecentra ltissu eo f these pea nodules. However, in the alfalfa "empty" nodules elicited by A. tumefaciens transconjugants there is at present no possible way of distinguishing infected and uninfected cells in the central tissue, not by using the presence of infection threads as criterion either. Yet, in such "empty" alfalfa nodules, Nms-30 has clearly been demonstrated. This suggests that, at least in alfalfa, induction of the expression of the Nps-40'-homologous gene happens - and that perhaps even uninfected cells can differentiate -, without prior infection of the central tissue. However, it is unknown whether Nms-30indee d is amolecula r markerfo r theuninfecte d cellsi n alfalfa. For the interpretation of the central tissue of "empty" nodules and the modeo f induction of the expression of theNms-3 0 gene,i t will thusb ever y important to beabl e tolocaliz e the Nms-30gen eproduc t in alfalfa. Inconclusion , ase to fearl y nodulinso fdifferen t legumespecie s(Nps-40 'o f pea, Nvs-40o fvetch ,an dNms-3 0o f alfalfa) havebee n shownt ob eimmunologicall y related. In order to determine to what extent they are comparable in the regulation of their production andthei r rolei n symbiosis, amor edirec t andreliabl eprob e than the above-

150 described antiserum has to be obtained for detecting the expression of the Nps-40' gene and its homologues in vetch and alfalfa. For that purpose, cDNA clones representing the Nps-40' early nodulin are now being isolated.

MATERIALS AND METHODS

Cultivationof plants andbacteria

Pisum sativum ssp. sativum L. cv "Rondo" and Vicia sativa ssp nigra L. (Ehrh.) were cultured in gravel trays and inoculated as described by Bisseling et al. (1978). Medicago sativa was cultured on agar slants and inoculated as described by Meade et al. (1982). The bacterial strains Rhizobium leguminosarum PRE (wild-type) and Rhizobium meliloti 7023 (nodA~) were cultured as described (Bhuvaneswari et al., 1980,an d Leigh et al., 1985,respectively) .

Protein analysisand raising of theantiserum

In vitro translation products (made according to Gloudemans et al., 1987) and in vivo proteins (isolated according to Studer et al., 1987) of pea nodules were compared by separating a mixture of 5 ul of in vitro translation products from RNA of 17-day-old nodules, and 40 u.g of in vivo proteins isolated from 20-day-old nodules,b y two-dimensional gel electrophoresis according toD e Vries et al. (1982); the resulting gel wassilver-stained ,drie d and fluorographed topre-flashe d Kodak XAR-5 film. For purification of theputativ e Nps-40' protein,anion-exchang e chromatography was performed with an FPLC type Mono Q HR 5/5 (Pharmacia) column, volume 1.8 ml.: 5 mg of proteins in 500 ul from 17-day-old pea nodules were centrifuged for 30 min. at 30.000g, filtrated through 0 0.22 um filters.and loaded. The column was washed with 5 ml of 10 mM KAc/50 mM Tris-HCL pH 7.5, and eluted with, successively, gradients of 50 to 200 mM KAc (10 mM/ml) and 200 to 500 mM KAc (35 mM/ml),an d 1 M KAc (10 ml),al l in 50 mM Tris-HCl pH 7.5.Fraction s of 0.5 ml were sampled every minute and 30 ul of each fraction was analyzed by 12.5 % PAGE with silver-staining. Fractions containing 40 kD proteins (3 times concentrated with Sephadex G-25 according to Saul & Don, 1984) were analyzed by two-dimensional electrophoresis. The fraction most abundant in 40 kD proteins was labeled with 125I according toZabe l et al. (1982) and used asa marker for preparative gel electrophoresis.

For preparative gel electrophoresis, 5 mg in vivo protein from 15-day-old nodules mixed with the 125I-labeled fraction rich in 40 kDprotein s from ion exchange chromatography, were separated on 10

151 %P Agel s for6 hrs.a t ISOV , driedan dfluorographe d toKoda k XAR-5 films. Slices withlabele d4 0 kD proteins were cut from the gels and washed 3 times for 20 hrs. at 4°C in 10 mM Tris-HCl pH 5; the eluentswer epooled ,freeze-drie dan dsave da t-80°C . Antiserum againstth eelectrophoreticall y purified 40 kDprotein s was preparedb y immunizing New Zealand white rabbits,essentiall y asdescribe db y Bisseling etal . (1979). Immunoprecipitation was performed according toGover se tal . (1985).Th eantiseru mwa stitrate dtwic eove ra colum n of glutamine synthetase (from bean, a generous gift of Dr. J. Cullimore; cf. Cullimore & Miflin, 1984) coupled to activated CH-Sepharose 4B (Pharmacia) according to the manufacturer's instructions. Western blotting wasperforme d asdescribe db yStude re tal .(1987) .

Immunocytochemistry

Forligh t microscopy, 15day-oldnodule s werefixed i n4 %formaldehyd e (freshly preparedfro m parafomaldehyde) in SO mMsodiu m phosphate buffer (pH 7.2) for 2 hrs.a troo m temperature (RT),an d subsequently infiltrated with0. 1 Msucros e inbuffe r for2 hrs.,wit h 1M sucros eovernight ,an dwit h 2.3 M sucrose for 4 hrs., before being frozen in liquid propane. Semithin sections (1 \im) were cut on a ReichertUltracu tE equipped with anFC 4cryoki t at-80°C , andtransferre d in drops of 2.3 M sucroset o polylysine-coated slides.Labelin g withanti-bea nGS-seru m (agenerou s gift from Dr.J.V . Cullimore; cf. Cullimore & Miflin, 1984) anti-Nps-40'-antiserum and visualization of the signal with silver-enhanced proteinA-goldparticle swer ea sdescribe d(Na pe tal. , 1989). For electron microscopy, lS-day-old nodules were fixed in 3% formaldehyde/0.25 % glutaraldehyde, embedded in LR White resin and sectioned as described (Van de Wiel et al., 1988). Sections werelabelle d withanti-Nps-40'-seru m andth esigna l visualizedwit hprotein Aattache dt o 10 nm gold particles as for light microscopy, omitting the silver enhancement treatment. Sections were post- stainedan dobserve dunde ra Philip sEM30 1a sdescribe d (Nape tal. , 1989).

REFERENCES

Bergmann, H., Freddie, E., and Verma, DP.S., 1983,Nodulin-35 : a subunit of specific uricase (uricase II) inducedan dlocalize di nth euninfecte d cellso f soybean nodules,EMBO J., 2:2333 . Bisseling, T., Van de Bos, R.C., Weststrate, M.W., Hakkaart, M.JJ., and Van Kammen, A., 1979, Biochim. Biophys. Acta, S62: SIS. Cullimore, J.V., and Miflin, BJ. , 1984,J. Exp.Bot., 35: 581.

152 DeVries ,S.C. , Springer,J. , andWessels ,J.G.H. , 1982,Planta, 156:129 . Dickstein, R., Bisseling, T., Reinhold, V.N.,an d Ausübet,F.M . (1988) Expression of nodule-specific genesi nalfalf a rootnodule sblocke da ta nearl ystag eo fdevelopment .Genes andDevelopment 2 , 677-687. Cloudemans, T., De Vries, S.C., Bussink, H.-J., Malik, N.S.A., Franssen, HJ., Louwerse, J., and Bisseling, T., 1987,Plant Mol. Biol., 8: 395. Govers,F. , Gloudemans, T., Moerman, M., Van Kammen, A., and Bisseling, T., 1985, EMBO J.,4 : 861. Govers, F., Moerman, M., Downie,J.A. , Hooykaas, P., Franssen, HJ., Louwerse, J., Van Kammen, A.,an dBisseling ,T. , 1986,Nature, 323:564 . Hirsch, A.M.,Bhuvaneswari ,T.V. , Torrey, J.G., and Bisseling, T., 1989,Proc. Natl. Acad. Sei. USA 86: 1244. Kaneko,Y. ,an dNewcomb ,E.H. ,1987 ,Protoplasma, 140: 1 Lang-Unnasch,N. ,an d Ausubel,F.M. , 1985,Plant Physiol., 77: 833. Leigh,J.A. , Signer,E.R. , and Walker, G.C., 1985,Proc. Natl. Acad. Sei.USA, 82: 6231. Meade, H.M., Long, S.R., Ruvkun, G.B., Brown, S.E., and Ausubel, F.M., 1982,J. Bacteriol., 149: 114. Moerman, M., Nap, J.P., Govers, F., Schilperoort, R., Van Kammen, A., and Bisseling, T., 1987, Plant Mol. Biol., 9:171 . Nap, J.-P., and Bisseling, T., 1990, Nodulin function and nodulin gene regulation in root nodule development, in: Molecularbiolog yo f symbiotic nitrogen fixation, P.M.Gresshoff , ed.,CR C Press, BocaRaton . Nap, J.P., Van de Wiel, C, Spaink, H.P., Moerman, M., Van den Heuvel, M., Djordjevic, M.A., Van Lammeren,A.A.M. ,Va nKammen ,A. ,an dBisseling ,T. , 1989, Mol. Plant-MicrobeInterac, 2: 53. Newcomb, W., 1981,Int. Rev. Cytol. Suppl., 13: 247. Norris,J.H. ,Macol ,L.A. , and Hirsch,A.M. , 1988,Plant Physiol. 88, 321-328. Saul,A. ,an d Don,M. , 1984,Anal. Biochem., 138: 451. Scheres, B.,Va n Engelen, F.,Va n der Knaap,E. , Van deWiel , C, Van Kammen,A. , and Bisseling, T., 1990b, Plant Cell, 2:687 . Schubert, K.R., 1986,Ann. Rev. Plant Physiol., 37:539 . Selker, J.M1., 1988,Protoplasma, 147:178 . Sprent,J.I. , 1980, PlantCell Env., 3: 35. Studer, D.,Gloudemans , T., Franssen, HJ., Fischer, H.-M., Bisseling, T., and Hennecke, H., 1987, Eur. J. CellBiol., 45: 177. Vande nBosch ,K.A. ,an dNewcomb ,E.H. , 1986,Planta, 167: 425. Van deWiel ,G , Nap,J P. , VanLammeren ,A.A.M. ,an dBisseling ,T. , 1988,J. Plant Physiol., 132: 446. Verma, D.P.S.,Fortin , M.G., Stanley, J., Mauro, V.P., Purohit, S., and Morrison, M., 1986,Plant Mol.

153 Biol, 7: 51. Webb,M.A. ,an dNewcomb ,E.H. , 1987, Planta, 172: 162. Zabel,P. , Moerman, M., Van Straaten, F., Goldbach,R. , and Van Kammen, A., 1982, J. Virol., 41: 1083.

154 CHAPTER 9

General discussion INTRODUCTION

One of the major goals of the studies described in this thesis has been the integration ofmolecula rdat ao n nodulin geneexpressio n withmicroscopica l observations inorde r toobtai nmor einsigh t intolegum eroo t noduledevelopment . In thefirs t parto f the thesis (chapters 3= Va n deWie l et al., 1988an d 4 =Na pe t al., 1989) nodulin gene expression in pea (Pisum sativum) and vetch (Vicia sativa) was determined by two- dimensional gel electrophoresis of in vitro translation products and by probing RNA transfer blots with nodulin cDNA both of nitrogen-fixing nodules and of nodules that wereblocke d in several stageso f development byth eus eo fengineere d bacterial strains. Todetermin e at which stage thedevelopmen t of these nodules was blocked, they were studied with the aid of light and electron microscopy. All these nodules exhibited the basic structureo f theleguminou s rootnodule :a centra l tissuesurrounde d by aperiphera l tissue, the latter divided by a nodule endodermis into an inner and an outer part. The inner part was traversed by the vascular bundles that link up with the root vascular system. Disturbances in the nodules, blocked in development, were apparent at several stagesi nth edifferentiatio n ofth einfecte d cellsi nth ecentra l tissue.Thus ,th eexpressio n of individual nodulin genes could be correlated with specific stages of central tissue development. In summary, the results indicated that for the expression of the Nps- 40'/Nvs-40 gene to occur the bacteria had to berelease d from the infection thread, but further development of the infected cells was not necessary; in nodules showing expression of the Nps-40'/Nvs-40 genes, the uninfected cells developed an apparently normal morphology. The expression of the Nvs-65 gene could only be detected when undisturbed developmento f theinfecte d cellst obecom ecompletel yfilled wit hbacteroid s was observed, whereas the induction of the expression of the leghemoglobin genes needed even further development of theinfecte d cells.O n the other hand, transcriptso f the ENOD2gen e could bedetecte d in all the nodules studied, even when noreleas eo f bacteria wasobserved . Theseresult s havebee nreviewe d byNa p &Bisselin g (1990). Moreover, in some cases, e.g. if the Rhizobium meliloti exo mutants or Agrobacterium tumefaciens transconjugants harbouringR. meliloti nodgene s areuse da s inducing agents, bacteria-free pseudonodules are formed on alfalfa (Medicagosativa) that, although formed under thedirectio n of the nodgenes ,diffe r in morphologic details from normal nodules (cf. Finan et al., 1985; Hirsch et al., 1985). Yet, these "empty" nodules express twonoduli n genes:ENOD 2an d Nms-30,respectivel y (Dickstein et al., 1988; Norris et al., 1988).Th e Nms-30 genei s most probably the alfalfa counterpart of

157 theNps-40 'gen e (chapter8) . The approach followed in the initial stage of this study provides only indirect information aboutth erespectiv eprocesse so f noduledevelopmen ti n which the nodulins are involved. More precise information could be obtained by utilizing a more direct approach, asfo r example insitu localization of the nodulin geneproducts .Th e datao n the appearance of leghemoglobin during development could indeed be confirmed and extendedb y sucha n approach, namelyb y immunolocalization of theprotei n in sections of pea andvetc hnodules ,i nchapter s 3an d4 ,respectively . In the second part of this thesis the localization of nodulin gene products was further pursued, by in situ hybridization for the detection of nodulin transcripts in the chapters 5 (= Van de Wiel et al., 1990b), 6 (= Van de Wiel et al., 1990a) and 7 (=Scherese t al., 1990a),an db yimmunolabelin g for thedetectio n of anoduli nprotei ni n chapter 8. By these methods more information was obtained, not only about the production of theNps-40 ' protein and theENOD 2transcript , but also about a set of pea nodulingene sfo rwhic hcDN Aclone sha dbecom eavailabl ei nth emeantime . Relevant data of the nodulins involved have been summarized in Table 1. Localization data are provided by the Figures 1, 2 and 3. One gene, ENOD2, is expressed in the inner part of the peripheral tissue of both growth types of nodule, i.e. theindeterminat e nodules (=nodule swit hth ecapabilit y ofbasicall yindefinit e growthb y thepossessio n of apersisten t meristem atth e tipo f the nodule)of pea and alfalfa aswel l as the determinate nodules (= nodules lacking in persistent meristematic activity) of soybean (Fig. 1,se eals ochapter s 5an d 6).Th emajorit y of nodulin genes,however , are expressedi n thedevelopin g centraltissu eo f theindeterminat e noduleo f pea (Fig.2 ,se e alsochapter s 7 and 8an d Scheres et al., 1990b):th ePsENOD1 2 gene in all cells of the invasion zone;th ePsENOD 5gen ei nth einfecte d cellso f bothth einvasio n zonean dth e early symbioticzone ;th ePsENOD 3an dPsENOD1 4gene si nth einfecte d cellso f thelat e parto fth eearl y symbiotic zonean dth eearl ypar to fth elat e symbiotic zone;an dth eNps - 40' gene to allprobabilit y in the uninfected cells of the early and late symbiotic zones. ThePsENOD 5an dPsENOD1 2gene s arealread y expressed ata relativel y early stageo f noduledevelopmen t (Fig.3 ,se eals o chapter7 an d Scherese tal. , 1990b):th ePsENOD 5 genei n all cells with actively growing infection thread tips,th ePsENOD1 2gen e in the cells forming the nodule primordium and the cells through which the infection thread (will) grow. Nocomparabl e cDNA clones were available for the study of central tissue development of thedeterminat e soybean nodule.I n thefollowin g willb ediscusse d what the approach of localization insitu has contributed to our understanding of the role of nodulinsi nroo tnodul e development.

158 Table 1.Earl ynodulin sfo rwhic ha prob ei s availablefo rin situ localization

Early nodulin Characteristics References

GmENOD2 HRGP-like Franssen et al., 1987 (extracellular protein?) Franssen et al., 1989

MsENOD2 homologous to GmENOD2 Dickstein et al., 1988

PsENOD2 homologous to GmENOD2 Van deWie le tal. , 1990b (chapter5 )

PsENOD3 clustero f4 cysteines Scheres et al., 1990b (metal-binding protein?)

PsENOD5 rich in Pro, Ala, Ser, Gly Scherese t al., 1990b (arabinogalactan protein?)

PsENOD12 HRGP-like Scheres et al., 1990a (extracellular protein?) (chapter 7)

PsENOD14 homologous to PsENOD3 Scheres et al., 1990b

Nps-40' sequenceunknown , antiserum raised Govers et al., 1985, chapter 8 GmENOD cloneswer eisolate dfro m asoybea n(Glycine max) nodulecDN A library,MsENO D clones wereisolate dfro m analfalf a (Medicago saliva) nodulecDN Alibrary , andPsENO D clonesfro m ape a (Pisum sativum) nodule cDNA library. For Nps-40' (nodulin 40' of pea) no clone is available. Abbreviation:HRGP ,hydroxiproline-ric hglycoprotein .

NODULINS AS MARKERS OF SPECIFIC NODULE TISSUES AND/OR OF SPECIFIC STAGES IN THE DEVELOPMENT OF THESE TISSUES

The expression of each of the nodulin genes has been shown to be restricted to specific cell types of the nodule. The expression of a specific nodulin gene could therefore beuse d asa nmarke r for thepossibl epresenc e of therespectiv e celltypes .Thi s might beo f helpi n theinterpretatio n of nodulestha tdiffe r inth eusua lorganizatio n ofth e relevant tissues, such assom eo f thealread ymentione d "empty" noduleso n alfalfa. This will be more fully discussed in relation to the expression pattern of the ENOD2 gene below. Moreover, several nodulin genes appear to be expressed only transiently during the development of their respective tissues. This opens the possibility to use them as

159 CO PsENOD 12 PsENOD 5

PsENOD 3 Nps-40' PsENOD1 4

Fig. 2.Schemati crepresentation so fth eti po fth e indeterminatenodul edepicte di nFig . 1, withth e in situ locations of transcripts of thepe a early nodulinsPsENOD12 ,PsENOD 3an d PsENOD14, and PsENOD5,respectively ,an dth emos tprobabl ein situ locatio no fth epe aearl ynoduli nNps-40 ' protein. Forth ename so fth edepicte dlayers ,se eth ecorrespondin gpar to fFig . 1.

markers for specific developmental stages of therelevan t tissue. This will be illustrated with thenodulin spresen t inth ecentra l tissueo f thepe anodule . In the indeterminate pea nodule, the sequential stages in the developmental process can be observed in a single longitudinal section, since the distal meristem continuously provides the nodule with new cells,whic h subsequently differentiate into thevariou snodula rcel ltypes .Th erelevan tnoduli n geneseac h areexpresse da t different developmental zonesi nth ecentra ltissu e(Fig .2 ,se eals ochapter s7 an d 8an dSchere se t al., 1990b), as defined by Newcomb (1981). Transcripts of the PsENOD12 gene are detected in allcell so f theinvasio n zone.Transcript so f thePsENOD 5gen ear efoun d in theinfecte d cellso f theinvasio n zonean dth eearl y symbioticzone .Th eexpressio n ofth e PsENOD5gen ei sstronge ri nth einfecte d cellso fth eearl y symbioticzon etha ni nthos e

161 Fig.3 .Localizatio n ofENOD 5 andENOD1 2transcript s inroo tsegment so f 6an d8 da yol dinoculate d pea plants.Th e B,D , F, and Hpanel s show dark field micrographs corresponding with brightfield micrographs in the A, C, E, and G panels, respectively. In the dark field micrographs silver grains representing hybridization signal are visible as white spots. Aan d C:Roo t transections showing two infection events at similar developmental stages on 6 day old plants, 3 days after inoculation. The infection thread(arrow )ha sreache dth ethir dcortica lcel llayer ,a sdetermine db yseria lsectioning .I nth e innercorte xth enodul eprimordiu m (NP)i sindicated . B: ENOD5 localization; D:ENOD1 2localization . Ean dG : Similarstage so fnodul edevelopmen t inroot so f8 da yol dplants ,5 daysafte r inoculation. The infection thread (arrow)ha sreache dth enodul eprimordiu m (NP),an dbranche sof f intosevera lthinne r threads(arrowheads) ,growin g into thecell so f theprimordiu m whereth efirst bacteri aar ereleased .F : ENOD5 localization; H: ENOD12localization . Sections were hybridized with 35S-labeled antisense ENOD5an dENOD1 2RNA . Bar= 5 0um .X = xyle mpol eo fcentra lcylinder .

of the invasion zone. The PsENOD3 and PsENOD14 genes are expressed in the infected cells of the late part of the early symbiotic zone and the early part of the late symbiotic zone. The Nps-40' protein seems restricted to the uninfected cells of the early and the late symbiotic zone. Thus, some of the early nodulin transcripts are specifically present in zones

162 coinciding with the ones defined by Newcomb (1981), while others are not. The different developmental zones according toNewcom b (1981) have been defined on the basis of morphological/cytological criteria, mainly bearing upon growth stages of the bacteria. With the isolation of clones representing the above-mentioned early nodulin genes and the generation of antisera against early nodulin proteins, a set of molecular markers is available that enablest orefin e thedescriptio n of thedevelopmen t of thetw o cell types in thecentra l tissue. In particular, the transient character of the expression of some early nodulin genes improves the distinction of successive stages in the development of the infected cell type. In the description of the development of the infected cell type based oncharacter s liketh efilling o f thecell s with bacteroids and the morphologic development of these bacteroids during the filling of the cells, the distinction of successive stages is more arbitrary, since these areprocesse s following a more gradual course. An improved distinction of developmental stages, in turn, will facilitate the unravelling of the successive steps in nodule development and the identification of signals, both from theplan tan dth ebacteria , involved ineffectin g these steps. In addition, the availibility of markers for the early development of infected cells (PsENODS) as well as uninfected cells (Nps-40'), together with a marker for both cell types (PsENOD12), may open up possibilities to study the regulatory mechanisms underlying the differentiation of cells into the infected and the uninfected type, respectively. Thedescriptio n of developmental stages will undoubtedly gain from knowledge of the function of the early nodulin gene products. This can be illustrated by recent studieso n thelocatio n of thetranscript s of abacteria l genewit h aknow n function inth e nitrogen fixation process, namely the nifH gene, which encodes component II of nitrogenase. It was shown thati n thefirst tw ocel llayer so f the late symbiotic zone nifH transcripts are not yet detectable, while in the subsequent cell layer nifHtranscript s are present at amaxima l level (Yange t al., submitted). So,i n thefirs t few cell layerso f the late symbiotic zone,a sdefine d byth ecytologica l criteriao f Newcomb (1981),nitroge n fixation willno t bepossible .Thes eresult scorrelat e wellwit h thosei nalfalf a ofVass ee t al.(1990),wh oextende d thedescriptio n ofth ecentra ltissu ezonatio ni nth eindeterminat e alfalfa nodule by the cytological distinction of five successive growth stages in the development of thebacteroids .The y showed inth efirs t layerso f thelat e symbioticzon e the presence of bacteroids of a growth stage that was correlated with the lack still of nitrogen fixation. Thelatte r correlation wasconfirme d by studying bacteroids of strains defective in nitrogen fixation. Therefore, Vasse et af. (1990) suggested to name these firstlayer so fth elat e symbiotic zone"interzon eII-III "(i nthei rterminolog y II= invasio n zone +earl y symbiotic zone,an d III =lat e symbiotic zone), and to name the parto f the

163 late symbiotic zonewit h actively nitrogen-fixing bacteroids"nitrogen-fixin g zone".Th e host cells themselves of this interzone II-III were characterized by an accumulation of starch. Such a starch accumulation is also observable in the infected cells at a corresponding position in the late symbiotic zone in pea root nodules where leghemoglobin is notye tdetecte d (chapter 3).Th e absence of leghemoglobin in theso - called"interzon eU-UI "als ocorrelate swel lwit hth eabsenc eo fnitroge nfixation, i nvie w ofit sfunctio n inregulatin g theoxyge n supply toprovid e the bacteria with oxygen while keeping theconcentratio n low soa st oprotec t nitrogenase.Thes eresult s taken together illustrate wellth evirtu eo f thecombinatio n ofcytologica l andmolecula r approacheswit h theapplicatio n ofgeneticall yengineere d bacterial strains (seefurthe r the sectiono nearl y nodulin geneexpressio n in"empty "nodule sbelow) . Thepotentia l ofusin gmolecula r markersfo r thedefinin g ofdevelopmenta l stages in thedevelopin gbacteroid sha sals obee nindicate d with aslightl ydifferen t approach by Sharma & Signer (1990). These authors monitored the expression of R. meliloti symbioticgene si nalfalf a noduledevelopmen t byusin gbacteri a thatcontaine d fusions of thepromoter s of the symbiotic geneswit h theEscherichia coligusA reporter gene.Th e induction ofth e symbiotic genepromoter s in thedifferen t symbiotic zoneso f the alfalfa nodule could thus be followed by the histochemical localization of the gusA gene product,ß-glucuronidase . When thefunction s ofth enodulin s areunknown ,the y mightb einferre d from the temporal and spatialpatter n of theirproduction , combined with sequencedata .Thi sca n beventured , if the properties of the nodulin deduced from the amino acid sequence are consistent with arol e in the specific function(s) of the tissue in which the nodulin is exclusively produced. The most clear-cut example of this is provided by the ENOD2 gene,o fwhic h theexpressio n isrestricte dt oal lcell so f apar to f theperiphera ltissue tha t has been shown to have a specialized function in regulating the oxygen influx into the nodule (chapter 5; cf. Witty et al.,1986). This is discussed more extensively in the sectiono nENOD 2gen eexpressio n below. When inferring functions along these lines is more difficult, nodulin gene expression might still provide clues as to how the cytologically distinguished stageso f noduledevelopmen t relatet oeac hothe r andho wthe y areregulate d byth einteractio no f bacterial- andhost-derive d signals.Fo rexample ,th ePsENOD1 2gene s areexpresse d in both themeristemati c cellsconstitutin g the noduleprimordiu m inth einne r layerso f the rootcorte x and,a ta late r stage,i n thenon-meristemati c cellso f the invasion zoneo f the nodule(Figs .2 an d 3,se eals ochapte r7) .Thes eobservation sma ypoin t toa n interesting functional similarity between these different tissues. Moreover, the PsENOD12 gene expression pattern may berelate d to the same host and/orbacterial-derive d regulatory

164 factors operating in both tissues. This will be discussed more fully in the following sectionso nth eexpressio n pattern of thePsENOD1 2an dPsENOD 5gene sdurin g nodule development

ThePsENOD12 gene

In the development of indeterminate nodules, two successive stages in meristematic activityca nb edistinguished :first, th emeristemati c activity atth ever yonse t of nodule formation in the inner cell layers of the root cortex that constitutes a nodule primordium, and secondly, the meristematic activity organized in apersisten t meristem thatestablishe s itself ina subsequent stagea tth edista l end of theprimordium , enabling thenodul et osustai nindeterminat e growth (Libbenga& Harkes , 1973). Themeristemati c cells in these two successive stages characteristically differ in the expression of the PsENOD12 genes (Figs. 2 and 3, see also chapter 7). No expression of these genes is detected in the persistent meristem of the second stage. On the other hand, the meristematiccell so fth efirs t staged oexpres sth ePsENOD1 2genes .A tth esecon dstag e ofmeristemati cactivity ,th eexpressio n ofth ePsENOD1 2gene sbecome srestricte d toth e invasion zone,i.e .th ezon eimmediatel y proximal toth edista lmeriste m where infection threadspenetrat ecell sproduce d byth emeristem . When theroo t cortical cells thatwil lfor m thenodul eprimordiu m areinduce d to express theirPsENOD1 2genes ,th einfectio n thread,containin g thebacteria ,i sstil li nth e outercel l layerso f theroo tcortex .Hence ,th enodul eprimordia l cellsmus tb e amenable toreac t toa rhizobia l signal carried over several cell layers.I n the invasion zoneo f the nodule all the cells contain PsENOD12 transcripts, whereas only a part of the cells is penetrated by a branch of the infection thread (chapter 7; Scheres et al., 1990b). Therefore, it seems that also in the nodule tip, a rhizobial signal is operating which, although itmigh t be adifferen t onefro m thatproduce d in theroo tcorte x at this stageo f noduledevelopment ,nevertheles si seffectiv e outsideth ecell sactuall ycontainin gbacteri a and should therefore be able to reach the persistent distal meristem as well. Still, the PsENOD12gene s are notexpresse d in thepersisten t distal meristem, whereas the level of PsENOD12 transcripts is already maximal in the invasion zone cell layer directly adjacent to it, indicating that the persistent distal meristem is not susceptible to such a signal. These observations suggest that the persistent distal meristem cells must be different from themeristemati c noduleprimordia l cells,no tonl y in thepatter n in which theydeposi t new cells,bu tals o atth emolecula r level,a sshow n byth eexpressio n of the PsENOD12 genes.Nevertheless ,i tca n not beexclude d that atthi sdevelopmenta l stage

165 the bacteria give off signals differing from those in the root cortex so that none of the meristematiccell si sresponsiv e tothem .O nth eothe r hand,i tma yb ehypothesized , that the activated cellso f the inner cortex not simply "dedifferentiate" tofor m new cells for thefutur e nodule,bu tals oalread y specializefo r aspecifi c function innodul e formation, asindicate d by thespecifi c expression of thePsENOD1 2gene si nthes ecells . Besidesi n thealread ymentione d sites,namel y thenodul eprimordia lcell san dth e nodule invasion zone,PsENOD1 2 transcripts are alsodetecte d in theroo t cortical cells through which the infection thread grows on its way to the nodule primordium (Fig.3 , chapter7) .Interestingly ,bot hth ecell so fth eoute rroo tcortica llayer s andth ecell so f the innerroo tcortica l layerstha tfor m thenodul eprimordiu m show a similarreactio n toth e stimulation by the rhizobia. In both cell types cytoplasmic strands start to intersect the large central vacuole of the original root cortical cells. In the inner cortical cells these cytoplasmic strandsserv ethei r usualrol e inpositionin g thenucleu s in thecentr eo f the cellprio r tocel l division; in theoute rroo t cortical cells,however , these strands attain a new function, namelyprovidin g a sorto f bridgefo r guiding theinfectio n thread through the space normally occupied by the central vacuole of the cell. In this connection, Bakhuizen et al. (1988) suggested that the root outer cortical cells through which the infection thread will migrate starta cel l division cycle inwhic h they become arrested at the G2 phase. In the nodule invasion zone, also a vacuolation process is taking place, although reverse to that in the development of the nodule primordium: the cells of the invasion zone,whic hjus t havebee n deposited by thepersisten t meristem with scattered vacuoles, developa singlecentra lvacuol e notintersecte d bycytoplasmi c strands.Fro m these observations,i t is still hard to tell if PsENOD12migh t have arol e in vacuolation phenomena (seeals obelow) . Despite the apparent differences between thenodul e primordium andth e nodule invasion zone,i ti stemptin g tohypothesiz e that theybot h expressth ePsENOD1 2gene s due to a comparable position in adevelopmenta l gradient of interacting signals.A t the primordium stage,suc h a gradient mightb ebrough t about bya ninteractio n between the rhizobia, coming in theinfectio n thread from theoutside , and (a) factor(s) derived from the protoxylem poles of the root. The nodule primordia mostly arise near these protoxylem poles and evidence for such a factor from the xylem comes from the experiments of Libbenga et al. (1973).I n addition, PsENOD12 transcript could only be detected in infected root hairs, if a primordium could also be observed (Van de Wiel, unpublished preliminary observations),indicatin g thatpenetratio n of aninfectio n thread per se isno t sufficient for inducing PsENOD12gen eexpression , buttha t an interaction with the initiation of nodule primordia is necessary for PsENOD12 gene expression to occuri ninfecte d roothairs .A ta late r stage,th einvasio n zonemigh t behypothesize d to

166 be a zone where an interaction occurs between, again, the rhizobia in the incoming infection threadsan da factor , nowcomin gfro m thepersisten t meristem. Sucha situatio n is conceivable, since during nodule development the growing infection thread is first directed towards the nodule primordium, but after establishment of the persistent meristem the growing infection thread isredirecte d over 180degree s toinfec t the cells deposited by the persistent meristem. It would be very interesting to compare the postulated factors from the root xylem and the nodule persistent meristem, since the afore-mentioned observations thatbot h should beinvolve d ininducin gPsENOD1 2gen e expression, suggest thatthe y mayb e similar. Also,whethe r thebacteri autiliz e thesam e factor(s) in both situations,wil lhav et ob edetermined . Further studies have shown that PsENOD12 transcripts can also be detected in other parts of the plant, namely in the flower and in older parts of the stem, though in smaller amounts than in the root nodule. Strictly speaking, PsENOD1 2 therefore no longer fulfils therequirement s defined for atru e nodulin. In the flower, notissue s have yet been found to express specifically the PsENOD1 2gene(s) . In the stem, however, PsENOD12 mRNA is only found in the inner cell layers of the cortex immediately adjacent to the ring of vascular bundles (chapter 7). These parenchymatic cells have relatively largecentra l vacuolesan d showneithe r signso fincipien t meristematic activity nor the appearance of cytoplasmic strandsthroug h theirvacuoles .Thus ,n o relationship withvacuolatio n phenomenacoul db eimplicate dhere . These data provide no obvious clues as to the function of the PsENOD12 nodulin. Itsamin o acid sequence, asdeduce d from the cDNA sequence,indicate s thati t might be acel l wall protein (cf. Table 1),whic h is supported by its homology with the soybean 1A10protei n (Averyhart-Fullard et al., 1988).Thi s isconsisten t with arol ei n noduleprimordia l meristematic activity aswel l asi ninfectio n thread growth, both being processes in which new cell walls are produced and with which PsENOD12 gene expression appears to be associated, as can be deduced from the PsENOD12 mRNA localization datadiscusse d above.Also ,th ecell so f theoute rroo tcortica l layersthroug h which theinfectio n thread makesit swa y toth enodul eprimordium ,deposi t anadditiona l wall layer. Invie wo f this,i twoul db eworthwhil e tochec kwhethe r thecell so fth einne r layerso f theste mcorte x thatexpres sth ePsENO D1 2 genesals odeposi t anadditiona lcel l wall layer and, if so, to compare the structure and the composition of this layer with those of the wall layers deposited in nodule development. Some cell wall-depositing activity mightb esuspecte di nth e stemcortica lcells ,a scel lstretchin g will.benecessitate d herea sa resul to fa nincreas ei nthicknes so fth eunderlyin g stele. Because the PsENOD1 2gene s areexpresse d in other parts of the plant than the nodule, it can be assumed that PsENOD1 2 has a role in normal plant developmental

167 processes not directly related to nodule formation, which it probably already had even before the development of the symbiosis in the course of evolution. Afterwards, PsENOD12als oprove d useful for specific functions in nodule development during the evolution ofth esymbiosi s(chapte r7) . If no specific clues toth e function of PsENOD12 in the nodule havecom e from thestudie so fit spresenc ei nothe rpart so fth eplan ttha n thenodule ,th e specific function of PsENOD12 in the nodule appears most likely to berelate d to the infection process. There is no consistent relationship of PsENOD12 gene expression with meristematic activity in the nodule, asi s evident from its absence in thedista l menstem . Neither are PsENOD12 transcripts detectable in root and shoot meristems. On the other hand, all sitesi nnodul edevelopmen t wherePsENOD1 2gen eexpressio n isfound , soonero r later become involved in the infection process. Yet the function of PsENOD12 needs not necessarily be in formation of the infection thread, since not all cells of the nodule primordium expressing the PsENOD12 genes will eventually be penetrated by an infection thread. Rather, we would hypothesize that ENOD12, instead of having a function in infection thread growth itself, functions in preparing the walls of cells that might receive aninfectio n thread, to enable these cells to sustain a contraled infection process. In the outer cells of the root cortex this preparation involves depositing an additionalcel lwal llayer ,i nth einne rcell so f theroo tcorte xi ti sexecute d concomitantly withth eformatio n of newcells ,an di nth ecell so f thenodul einvasio n zonei t is attained during the differentiation of cells produced by the cell division activity of the nodule distalmeristem .

ThePsENOD5 gene

Iti sinterestin g atthi spoin t tocompar eth epatter n ofPsENOD 5gen e expression with that of PsENOD12. The PsENOD5 gene is only expressed in cells with growing infection thread tips, including the outer cells of theroo t cortex in which no release of bacteria takes place. After the release of the bacteria from the infection threads, the amount of transcript increases concomitantly with the proliferation of bacteroids in the infected cells (Figs.2 and 3,Schere se tal. , 1990b).Thus ,contrar y toth e situation with the PsENOD12 gene,PsENOD 5 gene expression is apparently induced by a rhizobial factor that only operates at a distance of not more than a single cell. Moreover, there appears tob e aninductio n of PsENOD5 geneexpressio n proportional toth enumbe ro f intracellularbacteri au ptil lth epoin twhereupo nth ebacteri aceas et oproliferate . Regarding the possible function of the PsENOD5 protein, its amino acid

168 composition, as deduced from the cDNA sequence is reminiscent of that of arabinogalactan proteins. Nodule-specific forms of arabinogalactan proteins have been shown to accumulate in developing soybean nodules by Cassab (1986). In addition, therear esom ehydrophobi cdomain si nth eamin oaci d sequenceo fPsENOD 5indicatin g thatth ePsENOD 5protei nmigh tb einserte di na membrane .Arabinogalacta n proteinsca n be inserted in theplasmamembran e (Knox et al., 1989).Therefore , it might be inferred that the PsENOD5 protein is part of both the infection thread membrane and the peribacteroid membrane,wher e itcoul d serve arol e inth einteractio n between thehos t plant and the bacterium. In this connection, Bradley et al. (1986) showed an intimate physical association between theplan t peribacteroid membrane and the bacteroid outer membrane by immunolabeling with monoclonal antibodies against a plant membrane glycoprotein andbacteroi d lipopolysaccharide,respectively .

TheENOD2 gene

Oneearl y nodulin gene,ENOD2 ,i sexclusivel y expressed inth einne rpar to f the peripheral tissue, usually called "inner cortex", in both the determinate nodules of soybean {Glycinemax) (chapte r 5)an d theindeterminat e noduleso f pea (chapter 5)an d alfalfa (Medicago sativa) (chapter6 )(Fig . 1).Sinc eth eENOD 2gen eapparentl y encodes a nodule-specific protein,th e specific presence of theENOD 2transcrip t in theinne rpar t of the peripheral tissue indicates that not only the central tissue, but also a part of the peripheral tissueha sa nodule-specifi c character. The peripheral tissue has usually been called the nodule cortex (cf. Dart, 1977;Newcomb, 1981;Bergersen , 1982). The arrangement of tissues in this "cortex", however,i sdifferen t from thato f thecorte xo f thesubtendin groo t (cf. Fig. 1). Anodul e endodermis forms the boundary between a tissue (the "outer cortex") that morphologically resembles the root cortex, and a different tissue (the "inner cortex"). Both the nodule "outer cortex" and the root cortex parenchyma join at the base of the nodule, separated from the rest of the nodule and the root stele by the nodule endodermis.Th eothe rpar to f theperiphera l tissue,locate d insideth enodul e endodermis (the "innercortex") ,consist so f highlyvacuolated ,parenchymatou s cellstha t are smaller and more densely packed, i.e. with fewer and smaller intercellular spaces (Witty et al., 1986), and that have walls staining more densely (Goodchild, 1977) than those of the peripheral tissue outside theendoderma l layer (chapter 5). Moreover, it is traversed by vascular bundles each bounded by a second endodermal layer, named bundle endodermis,an di n which noENOD 2transcrip t wasdetected .Thus ,th eorganizatio n of

169 tissues inside of thenodul eendodermi si suniqu efo r nodules,i fcompare d toothe rplan t organs.Th euniqu e character of theinne r parto f the peripheral tissue, based both on its position and on its morphology, is emphasized by the specific presence of the early nodulinENOD 2transcrip t inthi stissu ei nsoybean ,pe a andalfalf a nodules. Theseobservation s taken together prompted us toreconside r theterminolog y of the peripheral tissue. Indeed, one might simply distinguish acorte x and amedull a (the central tissue) in nodules. However, according to strict appliance of plant-anatomical terminology, the term "inner cortex" in the nodule is potentially misleading. By definition, theroo t endodermis delimits theroo t cortex from the stele with the vascular tissue (cf. Esau, 1977). The nodule endodermis, on the contrary, delimits only the nodule "outer cortex" from the rest of the nodule, including the "inner cortex". Thus, only this "outercortex " has apositio n comparable totha to f theroo t cortex parenchyma and might becalle d acorte x proper, in agreement with itsmorpholog y that is similar to that of the root cortex. Inside the root endodermis are the stelar tissues, which are organized ina singlecentra l cylinder. In thenodule ,thes e stelartissues ar eorganize d in several bundles that traverse the "inner cortex", bounded from the latter by the bundle endodermis.Hence ,on emus tconclud e thati ti sa tissue fo r which nocounterpar t canb e found in the root. For that reason, we proposed to rename this tissue "nodule parenchyma", to avoid the potentially misleading term "inner cortex", which might be taken toimpl y arelationshi p withth eroo tcorte xtha tactuall ydoe sno texis t(Va nd eWie l et al., 1990). This term, however, is still unsatisfactory, since it does not describe the specific positiono f thetissu ei n thenodule . Next to the terminology of "inner" and "outer" cortex, there is also the terminology of Bond (1948), who named only the inner part of the peripheral tissue "nodule cortex".Thi s is basedo n herconclusio n from adevelopmenta l study of thepe a nodule, that the outer part of the peripheral tissue must be regarded as a few layers of root cortical cells that must have undergone some cell divisions and a considerable amounto f stretching toaccomodat e the growing nodule.I n thisview ,onl y theinne rpar t of theperiphera l tissuei sdeposite d byth epersisten t nodulemeristem ,wit h theinclusio n of thenodul eendodermis ,whic h thusbound sth enodul eprope rfro m rootcortica ltissue. Thisvie w is consistent with ourobservatio n that transcripts for thenoduli n ENOD2ar e only present in the inner part of the peripheral tissue and that no nodulins have been observed in theoute rpar to f theperiphera l tissue.O n theothe rhand , Newcomb statesi n his review of nodule morphogenesis and differentiation (1981), that all of the nodule peripheral tissuei sderive d from thedista l face of thepersisten t meristem.Evidently ,th e questiono f theprecis eontogen y of thevariou spart so fth eperiphera l tissuemus tfirst b e resolved,befor e asatisfactor y terminology canb eestablished .

170 The inner part of the peripheral tissue has been shown to function as a barrier towards the penetration of free oxygen into the central tissue (Witty et al., 1986). The flow of free oxygen intoth ecentra l tissue needs to beregulate d toprotec t the extremely oxygen-sensitive nitrogen-fixing enzymenitrogenase .Th erestricte dintercellula r air-filled space, the location through which oxygen can diffuse most easily, might enable the noduleparenchym a toperfor m afunctio n asoxyge n barrier.Fro m thecDN A nucleotide sequence it has been deduced that ENOD2 might be acel l wall protein (Table 1) and, therefore, we have suggested that ENOD2 contributes to a specific cell wall structure with limitedintercellula r space.Contrar yt oth etransien t appearance ofmos to fth eothe r early nodulins, the ENOD2transcrip t remains present during the nitrogen-fixing stage. This is alsoconsisten t with a role in theregulatio n of the oxygen supply to the central tissue.Tha tis ,th eoxyge n barrier hasbee n shownt ob ea variabl eon ei nconnectio n with changes in the environment, such as nitrate stress (Witty et al., 1986). The ENOD2 protein could alsocontribut e to,fo r instance,a flexibl e cell wall morphology, making it possible to change the extent of the intercellular space and thus the diffusion rate of oxygen. Studying the location of ENOD2 gene expression proved also to be helpful in interpreting aberrant typeso f nodules,a swil l bediscusse d inth enex t section.

Earlynodulin gene expression in "empty" nodules

In alfalfa, certainengineere d bacterial strains (cf. Finane t al., 1985;Hirsc he t al., 1985) as well as auxin transport inhibitors (ATIs; cf. Hirsch et al., 1989) elicit the formation of nodules which, in addition to being free of intracellular bacteria in their central tissue,ma y alsodiffe r in other detailso f their histology. Still,i tcoul d be shown that these nodules specifically express the MsENOD2 gene in cells that have a morphology and occupy apositio n in these nodules comparable to the cells expressing theMsENOD 2gen ei nnodule sforme d inrespons e towil d typeR. meliloti(chapte rVI) . Theseobservation s indicatetha t thesenodule shav ea histologica l structure homologous to thato f normal nitrogen-fixing nodules.Thi s isparticularl y intriguing, because, in the case of the nodules induced by ATIs, it strongly suggests that the whole sequence of stepsleadin g toth eformatio n ofa nodule ,includin g theprope rtissue-specific expressio n of theENOD 2gene ,ca n betriggere d bya singl eexogenou s signal.Suc ha propositio n is consistent withth erecen t observation thatth eNodRm- 1molecul eexcrete d byR. meliloti and identified by Lerouge et al. (1990), induces, in a host-specific way, nodule

171 formation on alfalfa (Roche et al., 1991).I n this connection, it iso f interest that certain vegetatively propagatedclone so f alfalfa form nodulesi nth eabsenc eo fbacteri a (Truchet et al. 1989). This implies that alfalfa has at its disposal the complete developmental program to form root nodules and only a trigger from the rhizobia is needed to set this program in motion. In the "spontaneously" nodulating alfalfa clones the need for this trigger apparently has been abolished by one or more mutations (Truchet et al., 1989). NodRm-1i sclearl ydifferen t from theATIs ,bu ti ti sconceivabl e thatbot htype so f signal moleculesinterac ta tdifferen t levelso fth esam ecascad eo fevent scontaine di nth enodul e developmental program.Onl yi f so,th e action ofplan t hormones will beinvolve d in the formation of rootnodules . In all the "empty" nodules discussed here, another early nodulin gene is expressed, namely the one encoding Nms-30, which is immunologically similar to the Nps-40' early nodulin of pea (chapter VIII). If Nms-30, like most probably Nps-40', turns out to be a marker of the uninfected cells, the question is raised whether indeed uninfected cellsca ndifferentiat e without thepresenc e of bacteria in thecentra l tissueo r that this lack of bacteria in the central tissue leads to disturbances in the normal specialization of cells in the developing central tissue. Further insight into the way uninfected cellsdifferentiat e or cell-specific gene expression isregulate d in the central tissue may be gained by a combination of in situ localization studies and the use of engineered bacteria by which nodule development is blocked at specific stages. The application ofengineere d bacteriai nelucidatin g thenatur eo f bacterial signalsinvolve di n theinductio n of nodulin geneexpressio n hasalread y been shown tob eusefu l in the first part of this thesis, although with one draw-back: If the expression of an individual nodulin gene fails, this needs not be simply due to the lack of a bacterial signal. Unfortunately, the possibility of the presence of bacterial signals eliciting a negative response, e.g. a defense response, in the host plant has to be considered, especially when transconjugants of theplan t pathogenAgrobacterium tumefaciens areapplie d (see chapters IIIan d IV,an d also Nap &Bisseling , 1990).Excep t for thisrestriction , strains like the ones used in chapter III and IV, together with mutants, like for example bar abacterial release) mutants (De Maagd et al., 1989),wil l beusefu l in afurthe r analysis of thedifferentiatio n of thecentra l tissue,i f combined with theabove-describe d markers ofdivers estage si ncentra l tissuedevelopment .

172 CONCLUDING REMARKS

During root nodule development, a range of tissues differentiates that can be distinguished by acombinatio n of cytologicalan d anatomical criteria.Th e development of several of these tissues is marked by the expression of nodule-specific genes. The majority ofthes egene sar eexpresse d inth edevelopin gcentra l tissue,whic h comesa sn o surprise since this tissue will harbour thenitrogen-fixin g bacteroids.However , also the innerpar to f theperiphera l tissueturn sou tt oexpress ,i na persisten t manner, atleas ton e early nodulin gene, i.e. ENOD2 and thishold s true for both the indeterminate pea and alfalfa nodulesan dth edeterminat e soybeannodule .I tha sbee n shown thatthi spatter no f gene expression can be related to functions of all these tissues that are specific for the symbiosis. Moreover, in pea nodules, the localization data of several nodulin and rhizobial transcripts have showna transien t expression patterni n thecentra ltissue ,whic h makesthes enodulin s useful inrefinin g thecharacterizatio n of successive developmental stages in the formation of the central tissue. This will be important in determining the type of signals by which the host plant and the bacteria interact to establish a properly functioning symbiosis. In addition, localization of tissue-specific nodulin gene expression has appeared helpful indeterminin g howfa r thepatter n of development of the "empty'roo tnodule si n alfalfa is comparable to that of "normal", nitrogen-fixing root nodules. It would be interesting totes twhethe rthi sals ocontribute s toth einterpretatio n of thedevelopmen to f even more aberrant types of nodules, such as those appearing to be rather lateral roots than "true"leguminou s rootnodule s (cf. Dudleye t al., 1987). Futureresearc h should bedirecte d toth elocalizatio n andfurthe r characterization of the corresponding nodulin proteins, in an attempt to understand their specific functions. For that purpose, antibodies will have to be obtained against these proteins. The study of thenoduli n proteins might beo f help inth efurthe r delimitation of specific nodule tissues and, furthermore, in the understanding of the specific functions of the various noduletissue s andperhap seve n theirorganizatio n intoa coherentl y functioning symbiotic nodule.A combinatio n ofth elocalizatio n of nodulin geneexpressio n with the use of nodules, blocked at different stages of development, will helpi n the analysis of theregulatio n of geneexpressio n effecting proper noduledevelopment .

173 REFERENCES

Averyhart-Fullard,V. ,Datta ,K .an dMarcus ,A .(1988 )Proc. Natl. Acad. Sci.USA, 85,1082-1085. Bakhuizen,R. , Van Spronsen,P.C. ,Dfaz , CL., and Kijne, J.W., 1988,Th eplan t cytoskeleton in the /Jtózoöium-legumesymbiosis ,Ph.D .thesis ,Leide nUniversity ,Th eNetherlands . Bergersen,F.J. , 1982,Roo tnodule so flegumes :structur ean dfunctions , Research studiesPres sWiley , Chichester. Bond,L. , 1948,Bot. Gaz., 109: 411. Bradley,DJ. , Butcher,G.W. ,Wood ,E.A. ,an d Brewin,N.J. , 1986,J. Cell Sei., 85: 47. Cassab,G.I. , 1986,Planta, 168: 441. Dart,P. , 1977,Infectio n anddevelopmen to f leguminous nodules,in: A treatis eo ndinitroge n fixation, vol.m , R.W.F.Hard yan dW.S .Silver ,eds. ,Wiley ,Ne wYork . De Maagd, R.A., Rao, A.S., Mulders, I.H.M., Goossen-De Roo, L., Van Loosdrecht, M.C.M., Wijffelman, CA.,and eLugtenberg ,B.J.J. , 1989,J. Bacterial, 171:1143. Dickstein,R. , Bisseling,T. ,Reinhold , V.N.,an d Ausübet,F.M.,1988 ,Genes andDevelopment 2 , 677- 687. Dudley,M.E. ,Jacobs ,T.W. ,an dLong ,S.R. , 1987,Planta 171,289-301 . Esau,K. , 1977,Anatom yo fsee dplants ,Wiley ,Ne wYork . Finan, T.M., Hirsch, A.M., Leigh, J.A., Johansen, E., Kuldau, G.A., Deegan, S., Walker, G.C., and Signer,E.R. , 1985,Cell 40,869-877 . Franssen, H.J., Nap, J.-P., Gloudemans, T., Stiekema, W., Van Dam, H., Govers, F., Louwerse, J., VanKammen ,A. ,an dBisseling ,T. , 1987,Proc. Natl. Acad. Sei. USA, 84: 4495. Franssen,H.J. , Thompson, D.V., Idler, K., Kormelink,R. , Van Kammen,A. , and Bisseling, T., 1989, Plant Mol. Biol.,14 : 103. Govers,F. ,Gloudemans , T., Moerman, M., Van Kammen, A., and Bisseling, T., 1985,EMBO J.,4 : 861. Hirsch, A.M., Bhuvaneswari, T.V., Torrey, J.G., and Bisseling, T., 1989,Proc. Natl. Acad. Sei. USA 86: 1244. Hirsch,A.M. ,Drake ,D. ,Jacobs ,T.W. ,an dLong ,S.R. , 1985,/ . Bacteriol., 161: 223. Knox,J.P. ,Day , S.,an d Roberts,K. , 1989,Development, 106: 47. Lerouge, P., Roche, P., Faucher, C, Maillet, F., Truchet, G., Promé, J.C, and Dénarié, J., 1990, Nature, 344: 781. Libbenga,K.R. , andHarkes ,P.A.A. , 1973,Planta, 114:17. Libbenga,K.R. , VanIren ,F. ,Bogers ,R.J. ,an d Schraag-Lamers,M.F. , 1973,Planta, 114: 29. Nap, J.-P., and Bisseling, T., 1990, Nodulin function and nodulin gene regulation in root nodule development, in:Molecula r biology of symbiotic nitrogen fixation, P.M.Gresshoff , ed., CRC

174 Press,Boc aRaton . Nap, J.P., Van deWiel ,C , Spaink, H.P.,Moerman , M.,Va n den Heuvel, M.,Djordjevic , M.A.,Va n Lammeren,A.A.M. ,Va n Kammen,A. ,an dBisseling ,T. , 1989,Mol. Plant-MicrobeInterac., 2: 53. Newcomb, W., 1981,Int. Rev. Cytol. Suppl., 13: 247. Norris,J.H. ,Macol ,L.A. , and Hirsch,A.M. , 1988,Plant Physiol. 88, 321-328. Roche, P., Lerouge, P., Prome, J.C., Faucher, C, Vasse, J., Maillet, F., Camut, S., De Billy, F., Dénarié, J., and Truchet, G., 1991, NodRm-1, a sulphated lipo-oligosaccharide signal of Rhizobium meliloti elicits hair deformation, cortical cell division and nodule organogenesis on alfalfa roots, in: Advances in molecular genetics of plant-microbe interactions, Vol. 1, H. Henneckean dD.P.S .Verma , eds.,Kluwe rAcademi cPublishers ,Dordrech t Scheres, B., Van de Wiel, C, Zalensky, A., Horvath, B., Spaink, H., Van Eck, H., Zwartkruis, F., Wolters,A.-M. ,Gloudemans ,T. , VanKammen ,A. ,an dBisseling ,T. , 1990a,Cell, 60,281. Scheres,B. ,Va nEngelen ,F. ,Va nde rKnaap ,E. ,Va nd eWiel ,C , VanKammen , A.,an dBisseling ,T. , 1990b,Plant Cell, 2: 687. Sharma, S.B.,an dSigner ,E.R. , 1990,Genes andDevelopment, 4:344 . Truchet,G. , Barker,D.G. , Camut, S., DeBilly ,F. ,Vasse ,J. , andHuguet , T., 1989,Alfalf a nodulation inth eabsenc eo fRhizobium, Mol. Gen. Genet., 219: 65. Van deWiel ,C , Nap,J.P. , Van Lammeren, A.A.M.,an d Bisseling, T., 1988, J.Plant Physiol., 132: 446. Van de Wiel, C, Norris,J.H. , Bochenek, B.,Dickstein , R., Bisseling, T., and Hirsch, A.M., 1990a, Plant Cell, 2:1009 . Vand eWiel ,C , Scheres,B. ,Franssen ,H. , VanLierop ,M.-J. ,Va nLammeren , A.,Va n Kammen,A. , andBisseling ,T. , 1990b, EMBO J.,9 :1 . Vasse,J. , DeBilly ,F. ,Camut ,S. ,an d Truchet,G. , 1990,/ . Bacteriol,172 : 4295. Witty, J.F., Minchin, F.R., Sk0t, L., and Sheehy, J.E., 1986, Oxford Surv.Plant Mol. Cell Biol., 3: 275. Yang,W.C. ,Horvath ,B. ,Hontelez ,J. , VanKammen , A.,an dBisseling ,T. , 1991,submitted.

175 Chapter 10

Samenvatting Vlinderbloemige planten, zoals bijvoorbeeld erwt, wikke, luzerneklaver en soja, kunnen een samenwerkingsverband aangaan met bodembacteriën van de geslachten Rhizobium enBradyrhizobium. Dit samenwerkingsverband heeft over het algemeen een uitermate specifiek karakter, d.w.z. de meeste bacteriesoorten kunnen slechts met een beperkt aantal gastheersoorten eeneffectiev e interactie aangaan.Z oi sRhizobium meliloti alleen compatibel met soorten uit de geslachten Medicago,Melilotus e nTrigonella. I n interactie met de bacterie vormt de plant een gespecialiseerd orgaan, de wortelknol, waarin de bacterie wordt opgenomen. Als tegenprestatie voor deze gastvrijheid van de plant legt de bacterie stikstof uit de lucht vast in de vorm van ammonia en levert dit vervolgens aand eplant . Gedurende de interactie komen zowel in de bacterie als in de plant specifieke genento texpressie .Wa tbetref t deplan tzij n eenaanta l genengeïdentificeer d dieallee ni n de wortelknol tot expressie komen. De producten van deze genen worden nodulines genoemd. Veel van deze genen -de late nodulinegenen- worden afgelezen kort voore n tijdens hetplaatsvinde n van de stikstoffixatie en zullen vermoedelijk ook voornamelijk een rol bij dit proces vervullen. Als voorbeeld hiervan kunnen de leghemoglobines gelden, die een functie in het zuurstoftransport naar de bacterie vervullen. Een andere groep van genen -de vroege nodulinegenen- komen aanmerkelijk eerder tot expressie, d.w.z. tijdens de infectie en de vorming van de wortelknol. Deze vroege nodulines vervullen een rol ind eOrganogenes e van deknol .He t verkrijgen van inzicht in dezero l eni nd eregulati eva n deexpressi eva ndez egene nvorme n hetthem ava ndi t proefschrift. Daartoe is vooral gekozen voor een aanpak waarbij de gegevens verkregen met moleculaire technieken voor het aantonen van genexpressie worden geïntegreerd met microscopische technieken omee n goedbeel dt ekrijge n van waare nwannee rgedurend e deknolontwikkelin g denodulinegene n totexpressi ekomen . Als achtergrond voor deze studies wordt in hoofdstuk 2 de ontwikkeling van de wortelknol beschreven. De bacteriën zetten de wortelharen aan zich te krommen en worden vervolgens ingevangen in deze kromming. Waar de bacteriën zich hebben gehecht aan de wortelhaar wordt de plantecelwand plaatselijk opgelost, waarop de wortelhaarcel nieuwe celwand afzet tegen de lesie in de oude celwand, in de vorm van eenbui sd ece l in.D ebacterië nkome n indez ebuis ,d ezogenaamd e infectiedraad, terecht en worden via de topgroei van deze infectiedraad de wortel ingeleid. Tegelijkertijd worden locaal in de wortelcortex cellen geactiveerd om te gaan delen, wat leidt tot de vormingva n een knolprimordium. De infectiedraad groeit naar het knolprimordium toe, waarhi j zichvertak t encelle n in het centrale gedeeltevan het primordium binnendringt. Uit zijtakken van deinfectiedraa d komen debacterië n vrij ind eplantecellen . Ze blijven daarbij omgeven dooree n membraan vanplantenoorsprong , de peribacteroidmembraan.

179 De bacteriën vullen de geïnfecteerde cellen en ontwikkelen zich tot stikstoffixerende bacteroiden. Daarnaast ontstaan ook ongeïnfecteerde cellen, die tesamen met de geïnfecteerde cellenhe tcentral eweefse l vand ekno lvormen .O mee nbuitenst e grenslaag vandi tcentral eweefsel , bestaandeui tuitsluiten d ongeïnfecteerde cellen,hee n wordtee n perifeer weefsel aangelegd. Ditperifer e weefsel op zijn beurt wordt in een binnenst en een buitenst gedeelte opgedeeld door de vorming van een endodermisachtige laag, de knolendodermis genaamd. Het binnenste gedeelte van het perifere weefsel wordt doorsneden door vaatbundels dievi a de centrale cylinderva n de wortel dekno l aan het transportsysteem vand eplan tkoppelen .Dez eknolvaatbundel s wordenva nbinne n naar buitenomgeve ndoo ree npericyke le nee n bundelendodermisme tbandje s vanCaspari . De knollen van leguminosen kunnen worden verdeeld in twee typen, uitgaande van het patroon van meristematische activiteit gedurende de knolontwikkeling. Bij het gedetermineerde typedifferentiëre n allecelle nva n hetknolprimordium ,da tontstaa t ind e buitenstecellage n van dewortelcortex , in gespecialiseerd knolweefsel. Daarbij gaat alle delingsactiviteit verloren ind euitgegroeide , stikstoffixerende knol.Voorbeelde n vandi t type zijn de knollen van soja en boon. Bij het ongedetermineerde type wordt een persistent meristeem gevormd in de apex van het primordium, dat in de binnenste cellagen van de wortelcortex ontstaat. Dit persistente (apicale,distale ) meristeem blijft gedurendeee n groot deelva n hetbestaa n vand ekno lnaa rbinne n toecelle n afzetten. De centraal afgezette cellenkunne n vervolgens weervanui the tcentral eweefse l geïnfecteerd worden. De infectiedraad, die eerst van buitenaf het knolprimordium ingroeide, draait daartoe zijn groeirichting 180°o mteneind e het persistente meristeem tekunne n volgen. In dit knoltype kunnen dus in één longitudinale doorsneded eopeenvolgend e stadia van knolontwikkeling bekeken worden. Voorbeelden van dit knoltype zijn de knollen van erwt, wikke, luzerneklaver en klaver. Tot slot worden nog enige varianten op de beschreven knolontwikkeling binnen devlinderbloemigenfamili e besproken. In de hoofdstukken 3 en 4 zijn knollen van smalbladige wikke bestudeerd die opgewekt zijn metverschillend e genetisch gemodificeerde bacteriestammen. Dergelijke knollen bleken in diverse ontwikkelingsstadia gestoord te zijn, zoals kon worden vastgesteld m.b.v. licht- en electronenemicroscopische waarnemingen. Tevens kwamen denodulinegene n inverschillend e mate totexpressie ,i n samenhang metd emat ewaari n knolontwikkeling gestoord was.D e gebruikte bacteriestammen waren transconjuganten vanRhizobium leguminosarwnbv . viciae (specifieke symbiont van erwt en wikke) en bv.trifolü (specifieke symbiontva n klaver),e nAgrobacterium tumefaciens(pathogee n op een breed scala van dicotylen), waarin de normale plasmiden (in het geval van A. tumefaciens het Ti-plasmide dat verantwoordelijk is voor de tumorinductie op waardplanten) vervangen waren door een symplasmide of een deel daarvan met de

180 zogenaamde nodgene nva nRhizobium erop .Dez enod gene n zijn verantwoordelijk voor deknolvormin g opd egastheer .Doo rhe tgebrui kva nverschillend e transconjuganten kon worden nagegaani nwelk emat ed enod gene nee nro l spelen inhe tcorrec t latenverlope n van de opeenvolgende ontwikkelingsstadia van de knol en in het tot expressie brengen van de verschillende nodulinegenen. Daarbij werd echter op één probleem gestuit: het uitblijven van de expressie van nodulinegenen kon niet zonder meer worden toegeschreven aan het ontbreken van genetische informatie in de transconjugant. Er werden aanwijzingen verkregen,da td ebacterië n eenafweerreacti e bijd eplan t opriepen. Dit zou kunnen betekenen dat bijv. de A. fM/ne/acie/iï-transconjugant wel over de benodigde genetische informatie beschikt voor het induceren van nodulinegenexpressie, maar datdez e informatie gemaskeerd wordtdoo r informatie van de transconjugant zelf metee nnegatie f effect opd egastheer . De waarnemingen aan de knollen die verstoord waren in hun ontwikkeling maakten hetmogelij k denodulinegenexpressi e incorrelati et e brengen methe t bereiken van een bepaald ontwikkelingsstadium. Zo bleek het vroege noduline Nvs-40 al detecteerbaar als slechts het vrijkomen van de bacteriën uit de infectiedraad was opgetreden,tesame n metd edifferentiati e vanongeüifecteerd ecellen ;he tnodulin eNvs-6 5 daarentegen kon slechtsgedetecteer d worden, nadatd e geïnfecteerde cellen geheel met bacteroiden gevuld waren.Transcripte n voor het vroegenodulin eENOD 2ware n in alle gevallen detecteerbaar. Het optreden van leghemoglobine vereiste een nog verdere ontwikkelingva nd egeïnfecteerd e cellen.Dez elaatst ewaarnemin gko nbevestig d worden door de directe detectie van leghemoglobine m.b.v. immunolocalisatie in longitudinale doorsnedes van erwteknollen in hoofdstuk 3. Hieruit bleek dat leghemoglobine in de eerste cellagen van de laat-symbiontische zone, waarin de geïnfecteerde cellen al wel compleetme tbacteroide n gevuld zijn, nognie tdetecteerbaa rwas . De directe benadering van in-situ-localisati e werd, nu uitgebreid met localisatie van transcripten middelsin-situ -hybridisatie,voortgeze t ind evolgend e hoofdstukken 5, 6, 7 en 8. In hoofdstuk 5 wordt de /«-jiru-localisatie van transcripten van het vroege noduline-genENOD 2i nd e gedetermineerde knollen van soja end e ongedetermineerde knollen van erwt beschreven. In beide knoltypen bleef de aanwezigheid van ENOD2 transcriptenbeperk t tothe tbinnenst egedeelt eva nhe tperifer e weefsel, meestal aangeduid als de binnenste cortex. Dit benadrukt nog eens het knolspecifieke karakter van dit weefsel, hetgeen ook al gebleken was uit zijn specifieke morfologie en zijn specifieke positie, namelijk binnen een endodermis.Di t gaf ons aanleiding een alternatieve naam voort e stellenvoo rd e "binnenstecortex" ,namelij k knolparenchym, teneinde verwarring tevermijde n metd ebinnenst ecorte xva nbijv .d ewortel ,di eduidelij k niet vergelijkbaar isme t dit zogenaamde knolparenchym. Deva n de cDNA afgeleide aminozuurvolgorde

181 laatee n herhaling van tweeverschillend epentapeptide-elementen ,beid e beginnend met tweeprolines , zien, dievoorafgegaa n wordtdoo r een signaalpeptide. Dit wijst erop,da t ENOD2ee nextracellulai reiwi tis .Daaro mword tgespeculeer do fENOD 2bijdraag t totd e specifieke celwandordening van het binnenste gedeelte van het perifere weefsel met weinig intercellulaire ruimte,di eo p zijn beurt verantwoordelijk geacht wordtvoo r een specifieke functie van dit weefsel, namelijk het reguleren van de instroom van vrije zuurstof uitd elucht . In hoofdstuk 6 zijn ENOD2-transcripten gelocaliseerd in knollen van luzerne- klaver.I ndi tgeva lwerde nnie tallee nnormal estikstoffixerend e knollenbestudeerd ,maa r ook knollen die geïnduceerd waren door R. meliloti exo-mutanten of A. tumefaciens transconjuganten met één of twee symplasmides van R. meliloti, en zelfs knollen die ontstonden inreacti eo pbehandelin gva nd ewortel sme tauxinetransportremmers .A ldez e knollen werden gekenmerkt door het ontbreken van bacteriën in het centrale weefsel, maar weken ook in andere details af van normale knollen. Niettemin werden in deze afwijkende knollenENOD2-transcripte nvastgestel di nee nweefse l datqu amorfologi e en positie overeenkwam met het binnenste gedeelte van het perifere weefsel van normale knollen.Di tleid tto td eopvattin gda tdez eknolle n nietprincipiee lafwijke n van normale knollen. In het gevalva n deinducti e door auxinetransportremmers impliceert dit,da td e basale wortelknolstructuur opgewekt kan worden door één betrekkelijk simpel signaalmolecuul. Hoofdstuk 7bespreek the tvroeg enodulin eENOD1 2va nd eerwt . Boodschapper- RNAvoo rENOD1 2wer d aangetroffen ind evroeg ecelle nva nhe tknolprimordiu m end e wortelcellen waardoord einfectiedraa d naarhe tknolprimordiu m groeit;i n delater e fasen van de knolvorming raakt het voorkomen van ENOD12-transcripten beperkt tot de invasiezone, waar een deel van de cellen die afgezet zijn door het persistente knolmeristeem gepenetreerd worden doorinfectiedraden . Netal sbi jENOD 2he tgeva lis , impliceert de uit de cDNA-sequentie afgeleide aminozuurvolgorde een eiwit met repeterendepentapeptide-elemente n dietelken sstarte nme ttwe eprolines .He trepeterend e gedeelte wordt voorafgegaan door een signaalpeptide, hetgeen wederom op een extracellulair eiwit wijst. Besproken wordt in hoeverre ENOD12betrokke n zou kunnen zijn bij hetinfectieproce s gedurende deknolontwikkeling . Verder wordt aangetoond dat door de bacteriën uitgescheiden factoren ENOD12-genexpressie induceren in wortelharen. De bacteriële «od-genen blijken hierbij een belangrijke rol te spelen. Tenslotteblijke n ENOD12-transcriptenoo kdetecteerbaa ri nd ebinnenst ecortexlage nva n de oudere stengelgedeelten en in bloemweefsel. Klaarblijkelijk is het ENOD12- genproduct ook betrokken bij andere processen in de plant dan alleen de knolontwikkeling.

182 In hoofdstuk 8 wordt m.b.v. immunolocalisatie aannemelijk gemaakt dat het vroege noduline Nps-40' specifiek isvoo rd ecytoso l van de ongeïnfecteerde cellen van de erwt. Dit laat voor de eerste keer, voor zover ons bekend, zien, dat in de ongedetermineerde knold eongeïnfecteerd e cellenee n specifieke rolmoete nvervulle n in de symbiose,zoal sdat reed si saangetoon d bijd e gedetermineerdeknollen ,bijv .di eva n soja. Immunoprecipitatie-experimenten geven aanda tvergelijkbar e eiwitten voorkomen in smalbladige wikke en luzemeklaver, en wel de nodulines Nvs-40, respectievelijk Nms-30. De regulatie van de genexpressie voor deze nodulines wordt besproken in het licht van vroegere waarnemingen aan knollen met verschillende ontwikkelingsstoornissen in het centrale weefsel. Tot slot wordt in hoofdstuk 9 besproken, wat de localisatiestudies hebben bijgedragen aan inzicht in de rol van de nodulines in de ontwikkeling van de diverse weefsels van de wortelknol. Tevens wordt gespeculeerd over de regulatie van hun aanmaak gedurende knolontwikkeling.

183 ACCOUNT

The contents of the preceding chapters have been based on the following publications:

Nap, J.P., Van de Wiel, C, Spaink, H.P., Moerman, M., Van den Heuvel, M., Djordjevic, M.A., Van Lammeren, A.A.M., Van Kammen, A., and Bisseling, T., 1989,Mol. Plant-Microbe Interne, 2: 53.

Scheres, B., Van de Wiel, C, Zalensky, A., Horvath, B., Spaink, H., Van Eck, H., Zwartkruis, F., Wolters, A.-M., Gloudemans, T., Van Kammen, A., and Bisseling, T., 1990a, Cell, 60, 281.

Scheres, B.,Va n Engelen, F., Van der Knaap, E., Van de Wiel, G, Van Kammen, A., and Bisseling,T. , 1990b,Plant Cell, 2:687 .

Van de Wiel, G, and Bisseling, T., 1990,Tissue-specifi c expression of early nodulin genes, in: Plant molecular biology, R.G. Herrmann and B. Larkins, eds., Plenum, London, inpress .

Van de Wiel, G, Nap,J.P. , Van Lammeren, A.A.M., and Bisseling, T., 1988,J. Plant Physiol., 132:446 .

Van de Wiel, G, Norris, J.H., Bochenek, B.,Dickstein , R., Bisseling, T., and Hirsch, A.M., 1990a,Plant Cell, 2: 1009.

Van de Wiel, G, Scheres, B., Franssen, H., Van Lierop, M.-J., Van Lammeren, A., Van Kammen, A., and Bisseling, T., 1990b,EMBO J., 9: 1.

185 NAWOORD

Het in dit boekje gepresenteerde onderzoek is tot stand gekomen binnen een samenwerkingsverband van twee Wageningse vakgroepen, te weten Moleculaire Biologie en Plantencytologie en -morfologie. Aan de inhoud van dit boekje hebben zodoende velen vanuit soms heel verschillende invalshoeken een belangrijke bijdrage geleverd.O pdez eplaat swi li ke rgraa gee naanta lme tnam evermelden . Ab van Kammen, mijn promotor, je stimulerende belangstelling voor de resultatenva nhe tonderzoe k enj ekritisch ekij k opd eerui tvoortvloeiend etekste nhebbe n belangrijk bijgedragen aandi t proefschrift Ton Bisseling, leider van de "knollengroep", het is al vaak gememoreerd, je gedrevenheid,j e kritische gevoel voor presentatie, zoweli n teksten alsi n illustratiese n lichtbeelden, zijn eenhee lbelangrijk e stimulansvoo rmi jgewees ti ndi tonderzoek . André van Lammeren, mijn co-promotor van de zijde van Plantencytologie en - morfologie, veel dank voor adviezen en de accuratesse waarmee je alle stukken controleerde. Jan-PeterNa p enBe n Scheres,va njulli ebeid ezitte n belangrijke stukken werki n ditproefschrift ; Jan-Peter,jou w vaardigheid inhe t schrijven van samenhangende teksten enjouw onvolprezen archief zijn voor mij van grote waarde geweest; Ben,jou w brede belangstelling binnen en buiten de wetenschap maaktej e voor mij tot een inspirerende collega. Deoverig e (ex-)ledenva nd eknollengroep ,Hen kFranssen ,di e zelfs een carnaval opofferde ommi j goedeprobe st ebezorgen , TonGloudemans ,Francin e Govers,Jeanin e Louwerse,i nhe tbijzonde r Marja Moerman,mij n "moleculaire"paranimf ,vee ldan kvoo r allehul pe nd elevendig e sfeer binnen degroep . Joke van Beckum, Herman van Eek, Fred van Engelen, Jeanne Jacobs, Marie- José van Lierop, Esther van der Knaap, Irma Vijn, Anne-Marie Wolters, Fried Zwartkruis, en alle niet bij namegenoemd e studenten,dan k voorjulli ebijdrag e aan het werkbinne nd e"knollengroep" . Speciaal wili kMarce lva nde nHeuve lnoemen ,d eenig e student met wie ik persoonlijk heb samengewerkt en wiens bijdrage in vele opzichten meteen ookee nwaardevoll ewas . Vand ekan tva n "molbi"wi li kverde r nogbedanken :Pete rva n Druten,Reinder t deFluiter ,Pie t Madern enJo bTielen svoo rfoto - en tekenwerk, waarvooroo k dankaa n de bemanning van de Dreijen-fotolocatie; GréHeitköni g en Marie-Joséva n Iersel voor alle secretariële activiteiten;Pie td eKa mvoo rhe tverzorge nva n devel eplante nvoo rhe t onderzoek; alle overige medewerkers voor adviezen en sfeerbijdragen, daarbij niet te

187 vergeten Rob Goldbach tevens voor het levendig houden van mijn ornithologische belangstelling. Dan van de kant van PCM wil ik de volgende personen niet ongenoemd laten: Siep Massalt, bedankt voor hetvel efotowerk ; AllexHaasdij k enPau lva n Snippenburg voor het tekenwerk envormgevingsadviezen ; Joke Cobben en Truus van derHoe f voor alle secretariëlewerkzaamheden ; HenkDuijt s voorall ebijdrage n diezic hnie t beperkten tot het labmaa r zichoo k uitstrekten over muziek en fietsen; Paul Fransz, kamergenoot, Ingrid Hoek, Juliette Janson, Christianne Marcelis, KeesTheuni s en Ton Timmers,oo k kamergenoot,julli everkeerde n weli n andere sectorenva nd eplantencytologie ,maa rhe t collegialecontac tme tjulli ehe bi kaltij d bijzonder opprij s gesteld;Hen kKieft , dankvoo r devel egoed eadvieze n endaadwerkelijk e hulpi nd eimmunolocalisatie ,Ha n Magendans enRo bde nOute rvoo r alletechnisch ee nplant-anatomisch eadviezen ;Adriaa nva nAels t enJa n Schelvoo rvelerle i adviezen;Prof .Willems e voor hetkritisc h bestuderen van de teksten; en "last but not least" Carmen Reinders, mijn PCM-paranimf en ook al kamergenoot, je bent een echte steun voor mij geweest gedurende de hele promotieperiode. Alle overige leden van de vakgroep en andere bewoners van het botanisch labdan kvoo rd ebelangstellin ge ngebode nhulp . Wat betreft de buitenwacht, dank aan BobBakhuize n voor het gastvrije onthaal inLeiden ,waa rBe ne ni kkwame nvoo r adviezen opknollengebied . Voorwa tbetref t het laatste ben ik ook dank verschuldigd aan vele andere leden van de Leidse groep. Ad Geurts van Kessel, bedankt voor de ontvangst in Rotterdam, waar ik me op de hoogte kwam stellen van dein-situ-hybridisatie . Jan van Lent (Virologie,LUW) ,dan kvoo r de immunocytologische adviezen. Bhuvana, Ann Hirsch, and Takis Katinakis, thanks for the cooperation and stimulating discussions during your stays at the Wageningen lab. Beatrice Horvâth, thank you for the support in thela ban dfo r guiding usthroug h Hungary. Tot slotdan k aanall ehuisgenote n voorhe tveraangename n van hetverblij f inhe t Wageningse, wat zeker heeft bijgedragen tot het tot een goed einde brengen van de promotie.

188 CURRICULUM VITAE

Clemens Caspar Maria van de Wiel werd op 9 augustus 1959 geboren in Amsterdam. Aldaar bezocht hijhe t St.-Nicolaaslyceum,waa r hij heteindexame n VWO behaaldei n 1977. Inda t zelfdejaa rvin ghi jaa nme td estudi eBiologi eaa nd eUniversitei t van Amsterdam. In 1985 behaalde hij daar het doctoraalexamen met als hoofdvak Bijzondere Plantkunde en als bijvakken Plantenfysiologie en Microbiologie. Vanaf februari 1986werkt ehi jbi jd evakgroepe n Moleculaire biologiee nPlantencytologi e en- morfologie van de Landbouwuniversiteit Wageningen aan het onderzoek dat in dit proefschrift beschreven staat.Va njul i 1990to te nme tjanuar i 1991wa shi j werkzaam bij devakgroe pMoleculair eCelbiologi eva nd eUniversitei tva nAmsterda m aanee n project dat doord e Stichtingvoo r Biologisch Onderzoek (BION) end e Stichtingvoo r Ruimte- Onderzoek (SRON) gezamenlijk gefinancierd werd. Op 1februar i 1991i shi j in dienst getreden van het Centrum voor Rassenonderzoek en Zaadtechnologie (CRZ-DLO) te Wageningen.

189