Lanaud C., Risterucci A.M., N'goran J.K.A.*, Kebe I.*, Pieretti I.
CIRAD/BIOTROP BP 5035 34032 Montpellier Cedex, France
*IDEFOR-DCC 01 BP 1827 Abidjan 01 Cote d'Ivoire

A high density linkage map of Theobroma cacao L. has been established from a population of 182 individuals derived from a cross of two heterozygous plants: a Trinitario and a Forastero. The map comprises 419 loci linked in 10 groups which correspond to the ten chromosomes of T. cacao. These loci correspond to 5 isozymes, 178 RFLP, 30 RAPD and 191 AFLP markers and 15 microsatellites. The length of the map is 868 cM with a 2.1 cM average distance between 2 markers.

This map is now used to study the basic components of characters and to localise loci (QTLs) affecting traits of interests such as resistance to Phytophthora palmivora, factors of yield and quality traits.

The first results concern the resistance to P. palmivora. Black pod disease due to P. palmivora is widespread all over the world and could be responsible of important losses of yield. A population of 144 individuals located at Bingerville (Ivory Coast) was analysed using AFLP and microsatellites. Several QTL were observed in the genome: two regions of the genome, similar in the two parents, explained 36% of the variability of the % of infected pods.

Two QTLs explaining 23% of yield have been also revealed by this study.

Molecular markers that are linked to QTLs could be used for an early "Marker Assisted Selection" (M.A.S.). The interest of such a M.A.S. is discussed in the case of breeding for quality traits.

The development of saturated linkage maps, with the markers spaced at small intervals throughout the genome, is particularly helpful to resolve complex quantitative traits into single Mendelian components (Paterson et al. 1988; Lander and Botstein, 1989) and to improve breeding strategies in a tree perennial crop such as cocoa with a long generation time.

The construction of a saturated linkage map requires the production of a large number of polymorphic DNA markers. The construction of a genetic linkage map in humans by use of restriction fragment length polymorphism (RFLP) was proposed by Botstein et al. 1980, this technique has been used to construct genetic maps for a wide variety of plant species such as tomato (Tanksley et al. 1992), potato (Bonierbale et al. 1988; Gebhart et al. 1989,1991) or Brassica (Landry et al. 1991; Chyi et al. 1992).

Other markers have been used later to construct genetic maps: Random amplified polymorphic DNA (RAPD) markers (Williams et al. 1990) have been used in map construction in species such as Arabidopsis thaliana (Reiter et al. 1992), slashpine (Nelson et al. 1993) or sorghum (Chittenden et al.. 1994). Microsatellites, also called simple sequence repeats (SSRs) (Tautz and Rentz. 1984) are small repeats of two tandemly arranged nucleotides, they are highly polymorphic and can be amplified by polymerase chain reaction (PCR) with unique flanking primers (Beckman and Soller, 1990). They have been used in mapping of several mamalian genomes (Weissenbach et al. 1992; Serikawa et al. 1992) and plant such as soybean (Akkaya et al. 1992), rice (Zhao et al. 1993) and maize (Senior and Heun, 1993). Amplified fragment length polymorphism (AFLP) markers (Vos et al. 1995) based on the detection of genomic selective restriction fragments by PCR amplification (Zabeau and Vos, 1992) have been also used more recently for mapping plant genomes ( Thomas et al. 1995; Meksem et al. 1995; Cho et al. 1996).

Most of important agronomic traits are under the control of several genes and their genetic control are generally not known. By establishing saturated maps with markers placed regularly along each chromosome of the species, it is possible to localise the several regions of the genome involved in traits of interest (Quantitative Trait Loci or QTL). This gives precise information on the genetic basis of the traits: the number of genes involved and the importance of each one on the variation of the character could be determined. The stability of their expression, according to different environments or genetic contexts could also be tested and constitute good guides to elaborate breeding strategies.

Using such maps, QTLs analyses were performed on a large number of species and were related to various kind of characters that could be resistances to fungus (Young, 1996; Lefebvre and Palloix, 1996), insects (Nienhuis et al. 1987), bacterial (Yoshimura et al. 1995) or nematodes (Pineda et al. 1993), physiological traits as tolerance to drought, dormancy (van den Berg et al. 1996), and flowering (Austin et al. 1996 ), factors of qualities (Osborn et al. 1987; Brummer et al. 1995).

Molecular markers that are linked to these QTLs will co-segregate with the genes involved in desirable traits and could be used efficiently to follow introgressions and accumulation of favourable traits during recombination cycles.

On cocoa, a first linkage map was established by Lanaud et al (1995) ; other map was established by Crouzillat et al (1996). More recently, markers were added to saturate the CIRAD map (Risterucci et al. 1996). In this work, we will present our results on the production of a high density map of the cocoa genome and on the first results obtained on QTLs analyses. Interest of a Marker Assisted Selection (MAS) will be discussed for improvement of cocoa.

The parents of the progeny analysed resulted from a cross between a Forastero (UPA 402) and a Trinitario (UF 676) (hybrid between Forastero and Criollo).

UPA 402 is an upper Amazon Forastero clone obtained from a sib matting involving IMC 60 and Na 34, two Forastero genotypes collected in Ecuador.

UF 676 is an Trinitario selection made by the United Company in Costa Rica.

UPA 402 has a higher level of homozygosity than UF 676, this is based on informations provided by pedigree and previous molecular analyses (Lanaud et al. 1995).

A mapping population of 182 progenies was used for the linkage analysis. This progeny is located in Cote d’Ivoire (IDEFOR/DCC).

Isozyme analysis was performed using five enzyme systems according to protocols described by Lanaud (1986, 1987). These included acid phosphatase (ACP), isocitrate dehydrogenase (IDH), malate dehydrogenase (MDH), phosphoglucose isomerase (PGI), phosphoglucomutase (PGM).

cDNA clones, genomic Pst I clones from librarires constructed at CIRAD were screened, ten genomic Pst I clones from Nestlé (Crouzillat et al. 1996) were also screened. A few clones of telomeres library obtained according to Kilian and Kleinhofs (1992) were screened on southern blots of parental DNA restricted with Eco RI, Eco R5, Bgl II, Hind III and Xba I. In addition 3 cloned genes were used. Histone (H4Cl4), isolated from maize (Philipps et al. 1986), alcohol dehydrogenase (Adh 1) isolated from maize (Gerlach et al. 1982), and rRNA gene (pTA 71) isolated from wheat (Gerlach et al. 1979).

Three fragments obtained from RAPD were also used as labelled probes. DNA fragments were extracted from low melting agarose gel and hybridised, without cloning, onto the restricted DNA.

Five micrograms of total DNA were restricted overnight with 3 u/µg of each of the five restriction endonucleases. Samples were loaded onto 0.8% agarose gel in TAE buffer separated by electrophoresis and blotted onto a Hybond N+ membrane (Amersham).

The DNA probes were labelled by using 32p dCTP by random priming (kit mega prime Amersham).

AFLP were performed with the Gibco BRL AFLP analysis system I as recommended by the supplier.

400 ng of genomic DNA were restricted with a EcoRI Mse I mix, the enzymes were inactived at 70 °C for 15 mn, the fragments were ligated with EcoRI and Mse I adapters.

5 µg of a 10 fold diluted ligation were amplified for a preselective amplification using EcoRI+1 and Mse I+1 primers.

EcoRI+3 primer was 33p labelled using a T4 polynucleotide kinase, this primer was mixed with Mse I+3 primer, PCR buffer and polymerase and 15µl of the mix were added to 5µl of 50 fold diluted preamplified DNA.

One cycle 94 °C for 30 s; 65 °C for 30 s and 72 °C for 60 s was performed, followed by 12 cycles with a lower annealing temperature of 0.7°c for each cycle and then 23 cycles of 94° C for 30 s; 56 °C for 30 s and 72 °C for 60 s.

After adding 20µl of loading buffer (98% formamide, 10mM EDTA, bromophenol blue, xylene cyanol), the mix were denatured at 92°c for 3 mn and 3 µl of each sample were loaded in a 5% polyacrylamide gel with 7.5 M urea and electrophoresed in 0.5% TBE buffer at 55 w for 1h40mn.

The gel was dried for 30 mn at 80 °C and exposed overnight to Xray film (Fuji RX).

A genomic library enriched in simple sequence repeats was constructed for microsatellites analysis with a modified version of the protocol of Karagyozov et al. 1993.

After screening on the parents, 15 microsatellite markers were analysed on the map population.The primers were end labelled with 33p ATP, the amplification was performed and the samples were analyzed in the same AFLP electrophoresis and revelation conditions.

The spacer 5S was mapped, by amplification with consensus primers (D’Hont et al. submitted) followed with a Hpa II restriction and analysed in 2 % TBE agarose gel.

A modified version of the protocol of Williams et al. (1990) was applied by use of primers provided by Operon Technologies. Amplification products were analysed in 2 % TBE agarose gel (Lanaud et al. 1995).

RFLP probes were named cTcCIR, gTcCIR, and rTcCIR ; The first letter c, g or r corresponding respectivly to cDNA, genomic and isolated RAPD genomic fragments.

Tc to Theobroma cacao and CIR to CIRAD.

Microsatellite markers were named mTcCIR.

TEL correspond to telomeric markers, N correspond to Nestle genomic probes.

RAPDs loci were named rOPX#, OP correspond to OPERON Technologies, X is the primer kit letter and # the approximate molecular weight of the band.

AFLP loci were named AFLP X/Y, x is the number of the primers combination and Y the number of the polymorphic band.

Segregation was studied on 100 individuals for RFLPs, RAPDs and isozymes and on 182 individuals for AFLP and microsatellites. Linkage analyses were performed using the program JOIN MAP version 1.4 (Stam, 1993). As both parents of the progeny are heterozygous, the markers, segregated according to the three possible Mendelian models and were integrated into the analyse: AxH, HxA or HxH where A is a homozygous locus and H is a heterozygous locus for the parent. The segregation of 420 markers was studied using a LOD score of 5.0 and 4,0 to identify the linkage groups. The kosambi mapping function was used to convert recombination frequencies in map distances (Kosambi, 1944).

A population of 144 individuals located in Bingerville (Cote d'Ivoire) obtained from the cross between UPA 402 and UF 676 was studied. The population was split between three different plots at Bingerville (Ivory Coast):

B7: 54 individuals planted in 1987.

C2-1: 39 individuals planted in 1988.

BL9: 51 individuals planted in 1989.

These 144 individuals were genetically marked using AFLP markers and microsatellites.

Two genetic maps were drawn up from these markers: one for the UPA 402 parent and the other for the UF 676 parent.

For each map, the markers that segregated according to a back-cross model were selected: heterozygote for the parent to be mapped, homozygote for the other. These models were used to analyse the QTLs provided by each parent independently.

The map of the UPA 402 parent comprised 43 linked markers and that of the UF 676 parent 104.

For each tree, the percentage of infected pods was evaluated in relation to the total harvest, i.e. over six years for the B7 trees and four years for the BL9 and C2-1 trees.

Only trees with more than 17 healthy or infected pods over the whole period were taken into account.

Harvest distribution was measured as the proportion of pods produced during the main harvest (i.e. September to December) in relation to total annual production.

Production precocity was evaluated over the first four harvesting years for all the trees, and is expressed as the ratio of yields for the first two harvesting years to the total for the first four years.

A yield index was calculated for each tree, as the ratio of its cumulated yields to the mean yield per tree in the plot concerned.

Mapmaker QTL (version 1.9) was used for QTL analysis. All the regions of the genome with an LOD score of over 1 were recorded. For the UPA 402 parent, the theoretical significance threshold for the LOD score (Lander and Botstein, 1989) was 2.2, compared to 2.5 for the UF 676 parent.

The polymorphism was evaluated for all the markers by testing the parents and some individuals of the progeny. The results are given in table 1. A total of 587 RFLP probes was screened using five restriction enzymes. Polymorphism was revealed in the progeny by 28 % of the cDNA probes and 27 % of the genomic probes. 21 microsatellites were screened and 15 of those were polymorphic (71%). 40 AFLP primer combinations ( resulted from 5 EcoRI+3 primers and 8 Mse I+3 primers) were screened. 30 of those were used, AFLP patterns revealed 1045 fragments, 191 were polymorphic and segregated in the progeny (19,4%). Only clearly scorable fragments were analysed, 5 to 15 fragments in a size range from 50 to 500 bases were used by primer combination. As RAPD, AFLP dominant markers were selected to give a band present in one parent and absent in the other parent and which segregated in the progeny. Only 23 AFLP fragments present on the both parents segregated.