Identification of QTLs for root characteristics in maize grown in ...

13 downloads 0 Views 190KB Size Report
Doyle, 1978; Wen and Schnable, 1994; Neuffer et al.,. 1997; Hochholdinger and .... mary root (R1D) at 1 cm from the kernel; dry weight of the primary root (R1W) ...
Plant Molecular Biology 48: 697–712, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.

697

Identification of QTLs for root characteristics in maize grown in hydroponics and analysis of their overlap with QTLs for grain yield in the field at two water regimes Roberto Tuberosa∗ , Maria Corinna Sanguineti, Pierangelo Landi, Marcella Michela Giuliani, Silvio Salvi and Sergio Conti Department of Agroenvironmental Science and Technology, University of Bologna, Via Filippo Re 6-8, 40126 Bologna, Italy (∗ author for correspondence; e-mail [email protected]) Received 23 February 2001; accepted in revised form 1 August 2001

Key words: drought tolerance, hydroponics, molecular markers, quantitative trait locus (QTL), roots, Zea mays

Abstract We investigated the overlap among quantitative trait loci (QTLs) in maize for seminal root traits measured in hydroponics with QTLs for grain yield under well-watered (GY-WW) and water-stressed (GY-WS) field conditions as well as for a drought tolerance index (DTI) computed as GY-WS/GY-WW. In hydroponics, 11, 7, 9, and 10 QTLs were identified for primary root length (R1L), primary root diameter (R1D), primary root weight (R1W), and for the weight of the adventitious seminal roots (R2W), respectively. In the field, 7, 8, and 9 QTLs were identified for GY-WW, GY-WS, and DTI, respectively. Despite the weak correlation of root traits in hydroponics with GYWW, GY-WS, and DTI, a noticeable overlap between the corresponding QTLs was observed. QTLs for R2W most frequently and consistently overlapped with QTLs for GY-WW, GY-WS, and/or DTI. At four QTL regions, an increase in R2W was positively associated with GY-WW, GY-WS, and/or DTI. A 10 cM interval on chromosome 1 between PGAMCTA205 and php20644 showed the strongest effect on R1L, R1D, R2W, GY-WW, GY-WS, and DTI. These results indicate the feasibility of using hydroponics in maize to identify QTL regions controlling root traits at an early growth stage and also influencing GY in the field. A comparative analysis of the QTL regions herein identified with those described in previous studies investigating root traits in different maize populations revealed a number of QTLs in common.

Introduction Drought tolerance is a complex trait whose underlying components remain elusive or, at best, poorly understood (Ludlow and Muchow, 1990; Quarrie, 1996). Selection for a number of morpho-physiological traits has been pursued as a means of indirectly improving yield under conditions of limited water availability; until now, these efforts have largely neglected root traits. Because root characteristics influence water uptake and, consequently, the water balance of the plant, selection for root traits could lead to important benefits for improving and stabilizing yield under drought conditions, particularly in species with a limited capacity to adjust osmotically. Additionally, roots play

an important role in nutrient uptake and for anchoring the plant to the soil. The information available on the genetic control of root traits in the field and their relationships with yield is limited, mainly due to the difficulty of measuring root characteristics in a large number of plants. Furthermore, field studies on roots often require destructive approaches and are complicated by heterogeneity in soil profile, structure, and composition. As an alternative to root studies conducted in the field, hydroponics should allow for a rapid and lowcost screening of root characteristics in a large number of plants at an early growth stage. An additional advantage of hydroponics relates to the possibility of carrying out subsequent and non-destructive analyses

698 of root traits in the same set of plants. However, it remains to be ascertained to what extent, if any, the genetic variation for root traits in hydroponics at an early growth stage is associated with variation in root traits at a later growth stage under field conditions. This aspect is particularly relevant in maize (Zea mays L.) and other herbaceous monocots in which the importance of the seminal (embryogenic) root system in supplying water and nutrients declines as a result of the prevailing role in the adult plant of shoot-born roots, commonly named ‘adventitious’ nodal roots (Kiesselbach, 1949). In maize, a relationship between seminal root traits in hydroponics and root lodging in the field has been reported (Landi et al., 1998; Sanguineti et al., 1998). Furthermore, a signifcant, albeit weak, association has been found between seminal root traits in hydroponics and root pulling resistance in the field (Landi et al., 2001). More generally, a positive correlation between root traits of maize seedlings and those of mature plants has been reported by Nass and Zuber (1971). Other studies investigating the genetic basis of natural variation for root traits in maize have shown its quantitative nature and complexity (Rahman et al., 1988, 1994); however, no information on the number and location of genes affecting root traits and, most importantly, their effects on grain yield was reported. In addition, a number of mutants for root traits have been described in maize (Jenkins et al., 1930; Doyle, 1978; Wen and Schnable, 1994; Neuffer et al., 1997; Hochholdinger and Feix, 1998; Hochholdinger et al., 1998). Interestingly, one of these mutants (rtcs) reaches maturity and sets seed even if the root system is solely based on the seminal primary root (Hetz et al., 1996). The introduction of DNA-based molecular markers, allows for unprecedented opportunities to identify the genetic factors (quantitative trait loci, QTLs) underpinning the variation of quantitative traits (Tanksley, 1993; Lee, 1995; Ribaut et al., 1996, 1997; Agrama and Moussa, 1996; Quarrie, 1996; Nguyen et al., 1997; Quarrie et al., 1997; Tuberosa et al., 1998; Frova et al., 1999; Sari-Gorla et al., 1999) and to investigate to what extent linkage and/or pleiotropy may influence traits’ association (Lebreton et al., 1995; Simko et al., 1997; Sanguineti et al., 1999). Although QTLs for root characteristics have been extensively analyzed in rice (Champoux et al., 1995; Ray et al., 1996; Price and Tomos, 1997; Yadav et al., 1997; Price and Curtois, 1999; Price et al., 2000), limited information is available for other cereal species.

In maize, Lebreton et al. (1995) identified QTL regions influencing root pulling resistance at flowering and other morpho-physiological traits in an F2 population grown under greenhouse conditions, while Guingo et al. (1998) reported co-location of QTLs for root architecture and above-ground biomass production in a population of recombinant inbred lines tested in the field. More recently, QTLs for root pulling resistance at flowering have also been described in field experiments (Giuliani et al., 2000). These maize studies were carried out with one water regime only, thus precluding the distinction between the constitutive (per se) and adaptive nature of gene expression underlying QTL effects; such a distinction is made possible through an evaluation of the same mapping population at different water regimes (Frova et al., 1999; Sari-Gorla et al., 1999). This information would be valuable for the breeders in order to better define their ideotypes and would allow them to more effectively apply marker-assisted selection in relation to the expected water availability in the targeted environments. The objectives of this study in maize were (1) to identify QTLs for root characteristics in hydroponics at an early growth stage, (2) to identify QTLs for grain yield in the field under well-watered and waterstressed conditions as well as QTLs for a drought tolerance index, (3) to analyze the relationships among QTLs for root traits in hydroponics and QTLs for the traits investigated in the field, and (4) to analyse the overlap of the root QTLs herein described with those identified in two different maize populations in which root characteristics were investigated (Lebreton et al., 1995; Guingo et al., 1998).

Materials and methods Plant material The mapping population included 171 F3 families derived from a corresponding number of randomly chosen F2 plants of the cross between Lo964 and Lo1016, two maize inbred lines which were known to differ for root traits (Sanguineti et al., 1998). The F3 families, kindly provided by the Experimental Institute for Cereal Crops (Bergamo, Italy), were reproduced in our field nursery. For each F3 family, 20 randomly selected plants were sib-mated and used only once, either as a female or as a male parent. An equal amount of seed was then taken from each ear and pooled.

699 Experiment in hydroponics The study was conducted under greenhouse conditions on all the 171 F3 families and the two parental inbreds. Growing conditions were similar to those described by Landi et al. (1998) and are only summarized here. Seeds were sterilized and germinated in petri dishes. Seedlings with both the embryogenic primary root and coleoptile (ca. 1–2 cm long) were laid on wire netting supports placed on plastic tanks (37.5 cm long, 44.5 cm wide, 17.0 cm high). Each tank contained 28 litres of nutrient solution (Hoagland and Armon No. 1 modified solution) which was renewed every 4 days. The solution was continuously aerated through rubber tubes connected to an air compressor. The experiment was carried out according to a randomized complete block design with four replications (sowing dates). Each replication included 18 tanks, each one with 12 rows. The first and the last row of each tank were considered as border rows, while the other 10 rows were used to grow the investigated families (one row per family). In one tank for each replication, only three families were grown and seven border rows were added. Each experimental unit (row) included 11 seedlings, grown at 25 ± 1 ◦ C during the day (16 h light with 500 µmol m−2 s−1 photosynthetic photon flux density at the seedling level) and at 15 ± 1 ◦ C at night. The seven central plants of each row were harvested after ca. 3 weeks of growth when the collar of the second leaf was completely visible. Data were collected for the following traits of the seminal roots: length of the primary root (R1L); diameter of the primary root (R1D) at 1 cm from the kernel; dry weight of the primary root (R1W) including its lateral roots; dry weight of the adventitious roots (R2W) formed at the scutellar nodal region. R1L and R1D were measured on a single-plant basis while R1W and R2W were measured on a row basis. Field experiment A randomly chosen sample of 118 F3 families and the two parental inbreds were tested in 1999 at Cadriano (44◦33 N, 11◦ 24 E; Po valley, northern Italy) on a Udic Ustochrept (fine silty, mixed, mesic) soil. All entries were tested in two trials differing only for irrigation volumes: one irrigation volume corresponded to 120% (well-watered conditions, WW) and the other to 40% (water-stressed conditions, WS) of the evapotranspiration starting from planting and after accounting for rainfall. The procedure followed to calculate the crop evapotranspiration was described by

Landi et al. (1995). The irrigation volume of the WS treatment was adopted to achieve a level of moderate drought stress, a condition which is quite common for maize grown in the Po valley under limited availability of irrigation water. The two trials were adjacent and separated by four border rows. The field layout of both trials was a randomized complete block design with two replications. Each plot was a single row 4.05 m long and spaced 0.85 m between rows. Because of their different vigor, F3 families and parental inbreds were separated by two border rows. The two trials were sown on 13 May 1999 and 6 weeks later plots were thinned at 15 plants (4.4 plants/m2). Fertilizers were applied before sowing at rates of 120 kg N and 44 kg P per hectare; soon after thinning, N was further applied at 130 kg/ha. Weeds were controlled mechanically. Total rainfall from sowing to harvest equalled 322 mm which were quite uniformly distributed throughout the growing season, with the exception of July, in which only 8.8 mm of rainfall were recorded. Irrigation was applied from 4 July (corresponding to the mid-end of stem elongation) to 7 August (onset of the phase of linear dry matter accumulation into the grain). This 5week period was chosen because it was characterized by dry weather and corresponded to a phase in which maize is rather susceptible to drought. In total, 75 mm and 25 mm water were supplied by sprinkler irrigation in the WW and the WS trial, respectively. Ears were hand-harvested on 7 October after excluding one plant on each end of the plot. Grain yield (adjusted to 155 g per kg moisture) was determined on a plot basis in both well-watered (GY-WW) and water-stressed (GYWS) conditions; then a drought tolerance index (DTI) was calculated as ratio between grain yield of each plot in the WS trial and the mean across the two replications of the corresponding family in the WW trial (GY-WS/GY-WW). The values of DTI are expressed as a percentage. DNA extraction and molecular marker analysis For each F3 family, one leaf was harvested from 18 plants which were grown for 7 weeks in the greenhouse; the leaves of each family were bulked, freezedried, and milled to a fine powder. The isolation of total DNA was performed following the procedures described by Saghai-Maroof et al. (1984). For RFLP analysis, DNA restriction digests included 6 µg of total DNA and 18 U of each of three restriction enzymes (BamHI, EcoRI and HindIII). DNA

700 fragments were separated by electrophoresis on 0.8% agarose and transferred overnight by capillarity onto charged nylon membranes by using a 0.4 M NaOH solution. Membranes were air-dried at room temperature and stored at 4 ◦ C. Pre-hybridization, hybridization, and washings were carried out following the procedures described in Sharp et al. (1988). A preliminary screening was carried out with 131 RFLP probes kindly provided by the University of Missouri (Columbia, MO) to identify those probes revealing useful polymorphisms. Subsequently, membranes with the DNA of the F3 families were hybridized with 66 polymorphic RFLP probes. Probes were labelled with 32 P-dCTP using the random-hexamer procedure. AFLP analysis was conducted following the protocol described by Vos et al. (1995), with slight modifications, using PstI and MseI restriction enzymes. Selective amplifications were performed with primers carrying two (at the PstI-fragment ends) or three (at the MseI-fragment ends) 3 -selective nucleotides, in equal molar amount. Ten primer combinations were considered. The amplification products were visualized with a silver staining procedure (Bassam et al., 1991). SSR analysis was conducted with publicly available primer pairs (MaizeDB at www.agron.missouri. edu). Amplifications were performed in a reaction volume of 20 µl containing 20 ng of DNA, 200 mM dNTP, 300 nM for each primer, 1.5 mM MgCl2 , 50 mM Tris pH 8.5, 20 mM KCl and 1 unit of Taq polymerase (Promega). Amplification conditions were as described by Taramino and Tingey (1996). The silver staining protocol described by Bassam et al. (1991) was used to visualize amplification products. In total, 16 SSR primer pairs were tested to search for polymorphisms between parental lines in order to fill gaps in the AFLP-RFLP-based map. Construction of the linkage map The software package JoinMap version 1.4 (Stam, 1993) was used for the linkage analysis among the data obtained with the 171 F3 families scored with 166 marker loci and for the construction of the genetic map. The critical LOD (logarithm of odd value) score for the test of independence of pairs of markers was set at 3.0. With Haldane’s mapping function, the recombination frequency between linked loci was transformed into centimorgan (cM) distances. The χ 2 test was applied to identify markers with a distorted (P2.5 in the ‘cov/−’ analysis and a peak was also present in the ‘cov/+’ analysis or when the significance level was reached only in the ‘cov/+’ analysis. QTL positions were assigned at the point

701 of maximum LOD score in the regions under consideration. Two peaks for the same trait on the same chromosome were considered as two different QTLs when they were separated by at least two markers and they were at a distance greater than 20 cM (Groh et al., 1998); this distance could be lower when the signs of the additive and/or dominance effects at the two QTLs were different. QTLs of two or more traits were considered as overlapping when their LOD peaks were within a 10 cM region. The proportion of phenotypic variance (σp2 ) explained by a single QTL was computed as the square of the partial correlation coefficient (R2 ). For each trait, a multiple regression model including all the corresponding putative QTLs was fitted in order to estimate the additive (ai ) and dominance (di ) effects of each ith QTL, the total proportion of σp2 and of genotypic variance explained by the fitted model (Schön et al., 1993). The value of the additive effect at each QTL was computed as half the difference between the mean phenotypic value of the F3 families which, based on the information of the flanking markers, were assumed to be homozygous for the Lo1016 allele and that of the F3 families homozygous for the Lo964 allele. The dominance ratio (DR = |di|/|ai|) defined by Stuber et al. (1987) was used to describe the gene action of the QTL (additive for DR < 0.2; partially dominant for 0.2 ≤ DR < 0.8; dominant for 0.8 ≤ DR < 1.2; over-dominant for DR ≥ 1.2). Results Analysis of the phenotypic values The ANOVA (not reported) revealed that for the root traits investigated in hydroponics the difference between the parental inbreds was highly significant (P5.0) of 10 QTLs and their sizeable R2 values indicate that a number of QTLs with a major effect on root traits have been identified. Indeed, the choice of the parental lines was based on large differences in root traits, a condition likely to increase the chances of finding QTLs with major effects. Because of the good coverage (ca. 90%) of the maize genome provided by the present study, it seems unlikely that many other QTLs with major effects went undetected because they map in portions of the genome not covered by our map. Notwithstanding this, it should be noted that based on simulation studies conducted with data obtained from large mapping populations, it has been postulated that not all QTLs are detected in any given experiment, particularly when the population includes fewer than 500 families (Beavis, 1994, 1998); this hypothesis has been validated in a number of large-scale field experiments conducted in maize (Openshaw and Frascaroli, 1997). Therefore, it is likely that in our case a number of QTLs, especially those with a small effect, went undetected; this is supported by the fact that a sizeable proportion of the genetic variability (from 33.1% for R1L to 57.6% for R1D) among F3 families remained unaccounted for after fitting a multiple regression including all detected QTLs. Additionally, the comparison of the R2 values of the multiple regression with those found at the single QTLs indicates the possible presence of epistatic effects. For R1W and R2W, this hypothesis is supported by the observation that the sum of the additive effects estimated at the corresponding QTLs was higher than half of the difference between the parental lines. However,

an overestimation of QTL effects occurs more commonly with a population of small size (Openshaw and Frascaroli, 1997). In consideration that a meaningful evaluation of epistatic interactions between QTLs requires a population size considerably larger than the one herein evaluated, we did not proceed with the necessary computations. For R1L and R2W, the balanced presence in the parent lines of plus and minus alleles is in keeping with the high level of transgressivity for these traits shown by the F3 families as compared to the values of the parent lines. The more limited transgressivity of F3 families over the high and low parental values for R1D and R1W is in accordance with the more unbalanced distribution of the corresponding plus and minus QTL alleles in the parental lines. When data are available, the comparative analysis of the results obtained with different mapping populations allows for the identification of important QTL regions (Phillips et al., 1992; Lin et al., 1995) and, indirectly, also for their validation. In maize, two studies have reported detailed QTL information for root traits in genetic backgrounds different from that herein considered (Lebreton et al., 1995; Guingo et al., 1998). Using the UMC map as a cross-reference we compared the position of the root QTLs described in these two populations with those identified in Lo964 × Lo1016. This comparative analysis suggested a number of overlaps among the root QTLs of these three populations. Of the 6 and 3 QTLs described by Lebreton et al. (1995) for root pulling force and the number of nodal roots at the base of the stem, respectively, at least 3 and 2 QTLs could tentatively be assigned to chromosome regions with root QTLs also in Lo964 × Lo1016. The most noteworthy overlap of QTLs described in our study with those revealed by Guingo et al. (1998) occurred in bins 2.03 and 7.02: in both cases, a QTL for R1D was present in Lo964 × Lo1016 and a QTL for the diameter of nodal roots was also evidenced by Guingo et al. (1998). Additionally, when the mapping population tested by Ribaut et al. (1996) was evaluated in hydroponics for the root traits herein considered (Tuberosa et al., 2000), QTLs in common with those herein described were identified in 6 chromosome bins (1.03, 1.06, 1.11, 2.03, 7.04, and 10.07). QTL studies of traits for which mapped mutants are available provide an interesting opportunity to analyse the co-location between the map position of the mutants and that of QTLs for the same trait. In this case, Robertson (1985) postulated that a mutant

709 phenotype at a particular locus may be caused by an allele whose effect is more drastic than that of the QTL alleles at the same locus. In maize, Robertson’s hypothesis has been validated for plant height (Beavis et al., 1991) and for plant architecture (Doebley et al., 1995) but not for the concentration of abscisic acid (ABA) in the leaf measured under field conditions (Tuberosa et al., 1998). Among the root mutants so far described in maize, only rtcs (rootless for crown and lateral seminal root) has been mapped with sufficient precision (Hochholdinger et al., 1998) to allow for a reasonable comparison of its map position with those of the QTL peaks herein reported. Notably, the estimated position on our map of rtcs (bin 1.02) is ca. 15–20 cM away from a QTL peak for R2W identified near asg45; this QTL showed the highest R2 value (26.6%) for R2W and an absolute additive effect of considerable magnitude (1.69 mg), second only to that of the QTL near csu2d (1.71 mg). Considering that growth of the adventitious seminal roots is totally suppressed in rtcs (Hetz et al., 1996), it is tempting to hypothesize that the effects of our QTL and rtcs might be due to alleles at the same locus with different effects on the phenotype. Furthermore, also a QTL for root pulling resistance and the number of nodal roots at the base of the stem reported by Lebreton et al. (1995) could be assigned to the same region. QTL analysis for grain yield and drought tolerance index QTL analysis accounted for a considerably lower portion of the genetic variance among F3 families for GY-WW as compared to GY-WS (54.3 vs. 81.5%); this result should be related, at least in part, to the lower heritability of GY-WW as compared to that of GY-WS (0.63 vs. 0.72). Most frequently, severe drought stress reduces heritability values as compared to well-watered conditions, as a consequence of a reduction of the genetic variance and an increase in error variance as summarized by Blum (1988). In our case, the moderate intensity of the drought stress in the WS treatment allowed for an increase in the range in GY mean values among F3 families as compared to GYWW, while keeping the experimental error at similar levels in both water regimes. This result could tentatively be related to differences, among the F3 families, in root traits which should reasonably influence to a greater extent GY-WS as compared to GY-WW. When the same set of 118 F3 families was tested under rainfed conditions for variation in root pulling resistance,

significant differences were detected among families (Giuliani et al., 2000). Interestingly, maize genotypes tested for variation in nitrogen use efficiency, a trait likely to be influenced by root characteristics, also showed a greater range in GY values under nitrogenlimited conditions than under conditions of normal nitrogen supply (Di Fonzo et al., 1982; Bertin and Gallais, 2000). A number of QTLs (2 out of 12) for GY were detected at both water regimes, a result which suggests a prevailingly constitutive expression of the genes underlying these QTLs; conversely, the identification of 6 QTLs expressed only at low water availability indicates the presence of genes playing a more important role in the adaptation to drought. A similar role could account for the action of the 4 QTLs detected only for DTI but not for GY-WW or GY-WS. Because our main objective was to investigate root traits, we did not thoroughly analyze the overlap of QTLs for GY and DTI in our population and those reported for the same traits in all previous studies published in maize. Notwithstanding this, a number of co-locations are worth mentioning with a previous study investigating QTLs for GY in a maize population (Os420 × IABO78) tested under drought-stressed conditions (Sanguineti et al., 1999). The same population was also investigated for QTLs controlling the leaf concentration of ABA (Tuberosa et al., 1998), a growth regulator which influences the adaptive response of plants to drought at both shoot and root levels (Sharp, 1996). Three regions (bins 1.07, 3.09, and 7.02) harbouring QTLs for DTI in Lo964 × Lo1016 were found to influence GY in Os420 × IABO78; furthermore, bins 1.07 and 7.02 also influenced leaf ABA concentration in Os420 × IABO78. Sanguineti et al. (1999) had speculated that QTLs for leaf ABA might in some cases be related to genetic variation in root traits at the same regions. We are presently investigating leaf ABA concentration in the F3 families herein evaluated to verify the presence of QTLs for this trait and their overlap with QTLs for root traits. Co-location of QTLs for root traits, grain yield, and drought tolerance index An interesting contribution of QTL studies relates to the possibility of elucidating the genetic basis of associated traits. Although only through circumstantial evidence, QTL analysis provides the opportunity to verify the role of pleiotropy or linkage in traits’ association (Lebreton et al., 1995; Simko et al., 1997;

710 Sanguineti et al., 1999). In our case, the root trait in hydroponics which showed a more consistent association with GY was R2W: at all four QTL regions concomitantly influencing R2W, GY-WW, GY-WS and/or DTI, the sign of the corresponding additive effects always matched, a result in keeping also with the positive, albeit low in value, correlation value between R2W and GY-WS (r = 0.20, P