MARKER-ASSISTED EVALUATION AND IMPROVEMENT OF MAIZE

Transcription

1 MARKER-ASSISTED EVALUATION AND IMPROVEMENT OF MAIZE Charles W. Stuber Department of Genetics North Carolina State University Raleigh, North Carolina Q INTRODUCTION Plant and animal breeders are faced with many challenges in their improvement programs, many of which might be met with the development and application of new marker technology. Some of these challenges that have stimulated research in the application of this technology in maize are as follows: (1) prediction of hybrid performance, (2) identification of useful genetic factors in divergent populations or lines (such as exotic accessions), (3) introgression of desired genetic factors into elite breeding lines, (4) improvement of recurrent selection programs based on phenotypic responses. In order to meet these challenges, a concerted effort has been placed on the development of new tools, such as DNA-based markers,with a major focus on the improvement of breeding precision and efficiency. HYBRID PREDICTIONS Heterosis (or hybrid vigor) is a major reason for the success of the commercial maize industry as well as for the success of breeding efforts in many other crop and horticultural plants. Development of inbred lines suitable for use in production of superior hybrids is very costly and requires many years in traditional plant breeding programs. Much of the developmental effort is devoted to field testing of newly created lines in various single-cross combinations to identify those lines with superior combining ability. The search for a reliable method for prech'ctinghybrid performance without generating and testing hundreds or thousands of single-cross combinations has been the goal of numerous marker studies, particularly in maize. During the 1970s and early 1980s, our laboratory and several others conducted investigations in which marker (isozyme or RFLP) diversity of inbred lines was correlated with performance (usually grain yield) in single-cross hybrids. A major objective of these studies was to evaluate the use of markers for prediction of hybrid performance from crosses among untested inbred lines. The number of markers used in these studies varied from fewer than 11 isozymes to 230 RFLPs. These investigations showed that genetic distances based on marker data agreed well with pedigree data for assigning lines to lieterotic groups. However, in those studies that included field evaluations, it was concluded that isozyme and RFLP genotypic data were of limited usefulness for predicting the heterotic performance between unrelated inbred maize lines. Several factors contributed to the limited predictive value of marker data. In those studies using only isozyme genotypic data, the small number of isozyme loci assayed had effectively marked only a 19

2 small fraction of the genome. Thus, only a limited proportion of the QTLs contributing to the hybrid response would be sampled. Also, it cannot be assumed that allelic differences at marker loci equate to allelic differences at linked QTLs, or vice versa. For a limited number of markers to be effective as predictors for hybrid performance, the effects (including types of gene action) of the linked QTL "alleles" must be ascertained. MARKER ALLELE FREQUENCY CHANGES An earlier approach in the development of markers for plant breeding purposes was based on the association ofisozyme allelic frequencies with phenotypic changes of targeted quantitative traits in long term recurrent selection experiments. Studies in this area were designed not only to search for the presence of marker associations with quantitative traits but also to ascertain the strength of these associations; mapping QTLs came later in development of breeding strategies based on markertechnology. Significant changes in auelicfrequencies at several isozyme marker loci were found when monitored over different cycles of long-term recurrent selection in several populations of maize. More importantly, these frequency changes were highly correlated with phenotypic changes in the selected trait, grain yield. Results from studies such as these led to the hypothesis that manipulation of allelic frequencies at selected isozyme marker loci should produce significant responses in the correlated trait. An investigation to test this hypothesis was conducted in an open-pollinated maize... population, and showed that selections based solely on manipulations of allelic frequencies at seven isozyme loci significantly increased grain yield. Although these marker-allele manipulation studies produced statistically significant results, the findings were not dramatic. The level of linkage disequilibrium between marker loci and QTLs was probably low in these studies because the target populations had been subjected to several generations of random mating prior to the investigation. This undoubtedly reduced the effectiveness of the marker loci for manipulating the associated quantitative traits. Although these studies could be considered to be only mildly successfid, the results were sufficiently positive to stimulate further investigations using larger numbers of markers in more structured types of populations. The impetus for further study and development of marker-technology related to the breeding of quantitative traits is probably the major contribution that can be ascribed to these earlier attempts. QUANTITATIVE TRAIT EVALUATION AND MANIPULATION IN F2 POPULATIONS In the marker-facilitated research conducted in the maize genetics program at Raleigh, North Carolina, QTLs have been identified and mapped in 15 F2populations derived from seven elite inbred lines and five inbred lines with a partial exotic (Latin American, expected to be 50%) component. Both isozyme and RFLP marker loci were used in these studies, although the earlier studies used only isoztmes. Measurements recorded on individual plants in the field experiments included dimensions, weights, and counts of numerous vegetative and reproductive plant parts as well as silking and pollen shedding dates. Findings from these F2 studies showed that QTLs affecting most of the quantitative traits evaluated were generally distributed throughout the 2O

3 genome, however, certain chromosomal regions appeared to contribute greater effects than others to trait expression. Data from these F2 populations provided a much stronger case for marker-based selection for manipulating quantitative traits than that from the earlier isozyme allelic frequency studies. To evaluate the efficacy of marker-facilitated manipulation, data fi'om two F2population studies were used. Selections were based solely on the genotypes of 15 isozyme marker loci for the F2plants evaluated in the mapping studies. Results from one cycle of genotypic selection showed that the mean yield of the selected populations was about 20,4 greater than the unselected check population. From this investigation, it was concluded that marker-facilitated selection (based on 15 isozyme marker loci which probably represented no more than 30 to 40% of the genome) was as effective as phenotypic selection which would be expected to involve the entire genome. Furthermore, the results imply that a significant increase in the relative effectiveness of marker-based selection could be reasonably expected if the entire genome were marked with uniformly distributed loci (e.g., every 10 to 20 cm). It should be noted that Edwards and Johnson (1994) reported significant gains in hybrid performance in two sweet corn populations in which four cycles of recurrent selection were conducted based only on marker genotypes. Their study involved a number of quality traits as well as yielding ability. ENHANCEMENT OF THE B73 X Mo17 HYBRID Earlier mapping studies in our program suggested that two elite inbred lines, Tx303 and Oh43, contained genetic factors that might contribute to the heterotic response of the B73 X Mo17 single-cross hybrid. Therefore, when a major study ofheterosis in the B73 X Mo17 hybrid was conducted, a companion study was made to identify and locate the putative genetic factors in Tx303 and Oh43. Planned mean comparisons among backcrosses and testcrosses in the two studies were used to identify six chromosomal segments in Tx303 that (if transferred into B73) would be expected to enhance the B73 X Mo17 hybrid response for grain yield. Likewise, another six segments were identified in Oh43 for transfer into Mo17 that would also be expected to enhance the B73 X Molt hybrid response. Three backcross generations were used for transfer of the identified segments into the target lines, B73 and MolT. BC2 families were analyzed for marker genotypes which were then used to select individuals for the third backcrosses. Individuals were selected if they had the desired marker genotype in the vicinity of the donor segment to be transferred and the recipient line's genotype in the remainder of the genome. At least one marker was assayed on each of the 20 chromosome arms, and usually two markers were genotyped in the vicinity of the donor segment being transferred. Two selling generations followed the third backcross. At each backcrossing and selling stage, marker genotyping of plants was conducted in the same manner as for the BC2 families. However, if a marker locus became fixed in a line, that marker was not evaluated in that line in succeeding generations and only segregating loci were analyzed. This reduced laboratory analyses considerably. ARer the second seifing generation, 141 BC3S2 modified B73 lines were identified for testcrossing to the original MolT. Likewise, 116 BC3S2 modified Molt lines were targeted for 21

4 testcrossing to the original B73. These 257 testeross hybrids were evaluated in replicated field plots in three locations in North Carolina. Forty-five (32%) of the modified B73 x original Mo 17 testerosses yielded more than the cheek hybrid (normal B73 x normal Mo 17) by at least one standard deviation. Only 15 (11%) yielded less than the cheek. Evaluations of modified Mo17 x original B73 testcrosses showed that 51 (44%) yielded more grain than the normal cheek hybrid and only 10 (<9%) yielded less than the cheek hybrid. The highest yielding hybrids exceeded the cheek by 8 to 11% (9 to 12 bushels per acre). Based on these initial evaluations, the better performing modified lines were selected for intererossing and were designated as "enhanced" lines. Fifteen "enhanced" B73 lines were chosen for crossing with 18 "enlaanced" Mo17 lines producing 93 hybrids that were evaluated in replicated field trials in North Carolina in At a planting density of about 29,000 plants per acre, only 15 of the test hybrids produced less grain than either of two cheek hybrids (normal B73 x normal Mo17 and a high yielding commercial hybrid, Pioneer 3165).. Six exceeded the cheeks by two standard deviations or more. The two highest yielding "enlaaneed" B73 x "enhanced" Mo 17 hybrids exceeded the cheeks by more than 15% (22 to 24 bushels per acre). These "enhanced" B73 x "enhanced" Mo17 hybrids were evaluated again in 1994, with some additional reciprocal crosses, which increased the number of test hybrids to 149. Twenty of these hybrids exceeded the cheek hybrids (same as in 1993) by two standard deviations or more, with the three best exceeding the cheeks by 12 to 15% (18 to 24 bushels per acre). Although the rankings changed slightly over the two years of testing, the average yields over both years for the best yielding hybrids exceeded the cheek hybrids by 8 to 10% (10 to 16 bushels per acre). More importantly, the parental lines that showed superior general combining ability in 1993 also showed similar performance in Evaluations of the highest yielding "enhanced" B73 x "enhanced" Mo 17 hybrids were again conducted in 1995 in North Carolina and results corroborated those of the previous two years. In addition, nine of these hybrids were evaluated at four locations in central Iowa by Pioneer Hi-Bred International. Three of the nine exceeded the grain yields of the best Pioneer commercial cheek hybrids by 6 to 10 bushels per acre. These results were surprising because all of the selections were based on evaluations in North Carolina. Field tests in North Carolina in 1997 corroborated earlier results. Results from the introgression of the targeted segments from Tx303 into B73 and from Oh43 into Mo 17 have demonstrated that marker-facilitated baekerossing can be successfully employed to manipulate and improve complex traits such as grain yield in maize. Not all of the six targeted segments have been successfully transferred into a single modified B73 or modified Mo 17 line. There appears to be some indication that there may be no advantage in transferring more than two to four segments. In fact, there is some indication that there could be a disadvantage. Increasing the number of transferred segments may be replacing the recipient genome with an excessive amount of linked donor chromosomal segments that could cause a deleterious effect. Also, epistatie interactions between a larger number of introgressed segments may result in a negative effect. In addition, favorable epistatie complexes in coupling phase (e.g., between recurrent parent alleles) could be disrupted. Further evaluations are necessary to determine the effects of larger numbers of transferred segments. 22

5 BREEDING SCHEME USING NEAR-ISOGENIC LINES (NILs) Although the results discussed above showed that the enhancement of lines B73 and Mo 17 was successful, the procedure for development of the "enhanced" lines (NILs) was very inefficient and would not be recommended for a practical breeding program. That procedure required that the targeted segments (containing the putative QTLs) be identified in the donor lines prior to transfer to the recipient lines. In our maize program at Raleigh, we have outlined, and have tested, a marker-based breeding scheme for systematically generating superior lines without any prior identification of QTLs in the donor source(s). The identification and mapping of QTLs in the donor is a bonus derived from the evaluation of the NILs generated. Choice of the donor, usually will be based on prior knowledge of its potential for providing favorable genetic factors, and, in maize, may involve appropriate heterotic relationships. The procedure involves the generating of a series of NILs by sequentially replacing segments of an elite line (the recipient genome) with corresponding segments from the donor genome. The objective is for the group of NILs generated to contain the complete genome of the donor source, with each NIL containing a different segment from the donor. Marker-facilitated backcrossing, followed by marker-facilitated selfing to fix the introgressed segments, is used to monitor the targeted segments from the donor and to recover the recipient genotype in the remainder of the genome. The number ofbackcrosses required will depend on the number of evaluations that can be made in the marker laboratory. In maize, the NILs are then crossed to an appropriate tester(s) to create hybrid testcross progeny that are evaluated in replicated field trials (with appropriate checks) for the desired traits. The superior performing testcrosses will be presumed to have received donor segments that contain favorable QTLs. Therefore, QTLs are mapped by function, which should be an excellent criteria for QTL detection. The breeding scheme not only creates "enhanced" lines that are essentially identical to the original elite line, but it also provides for the identification and mapping of QTLs as a bonus with no additional cost. Obviously, the scheme is based on having a reasonably good marker map with alternate alleles in the donor and recipient lines. CONCLUSIONS Molecular-marker technology has been demonstrated to be effective for identifying and mapping QTLs in maize, as well as in many other plant species, and for studying phenomena such as heterosis and genotype by environment interaction. The positive results from marker-facilitated selection and introgression studies should encourage the use of this technology by commercial breeders for transferring desired genes between breeding lines. Appropriate use of markers should increase the precision and efficiency of plant breeding, as well as expedite the acquisition of favorable genes from exotic populations or from wild species. 23

6 SELECTED REFERENCES Edwards, M., and L. Johnson RFLPs for rapid recurrent selection. Proc. of Symposium on Analysis of Molecular Marker Data, August, 1994, Corvallis, OR. Graham, G.I., D.W. Wolff, and C.W. Stuber Characterization of a yield quantitative trait locus (QTL) in maize by fine mapping. 37: Crop Sci. Stuber, C. W Biochemical and molecular markers in plant breeding. In: Plant Breeding Reviews. J. Janick (Ed.). Vol. 9 pp Stuber, C. W Breeding multigenie traits. In: DNA-BasedMarkers in Plants. g. Phillips and I. Vasil (Eds.). Kluwer Academic Publishers, pp Stuber, C. W Heterosis in Plant Breeding. ln: Plant Breeding Reviews. J. Janiek (Ed.) John Wiley & Sons. 12: Stuber, C.W Enhancement of grain yield in maize hybrids using marker-facilitated introgression of QTLs. Proc. of Symposium on Analysis of Molecular Marker Data. Am. Soc. Hort. Sci. and Crop Sei. Soe. Am., August 1994, Corvallis, OR. pp Stuber, C.W Success in the use of molecular markers for yield enhancement in com. Proc. 49th Annual Corn and Sorghum Industry Res. Conf., American Seed Trade Assoc. 49: Stuber, C.W Mapping and manipulating quantitative traits in maize. Trends in Genetics 11: Stuber, C.W Case history in crop improvement: Yield heterosis in maize, ln: Molecular Analysis of Complex Traits, A. H. Paterson (Ed.) CRC Press, Inc. pp Stuber, C. W., S. E. Lincoln, D. W. Wolff, T. Helentjaris, and E. S. Lander Identification of genetic factors contributing to heterosis in a hybrid from two elite maize inbred lines using molecular markers. Genetics 132:

7 Question and Answers: 1. How important is epistasis in corn -- if it is important, epistatic combinations may not be identified in the NIL design that you are proposing? (From Dr. Jack Dekkers) Answer: Although several attempts have been made to evaluate epistasis using traditional biometrical methods as well as evaluations among QTLs, most have shown that epistasis was of little significance. However, I personally feel that epistasis is probably quite significant and that we have just not been able to design experiments to successfully measure it. I agree with you in that certain epistatic combinations may not be identified in the NIL design that I proposed, however, I still believe that considerable progress can be made with the design. Chromosomal segments with significant effects can later be pyramided to provide combinations that make take advantage of epistatic interactions. 2. How did you go about choosing your donor lines (Tx303 and Oh43) in your gene introgression experiment? (From Dr. Harris Wright) Answer: We had information from earlier experiments involving F2 populations in which Tx303, Oh43, B73 and Mo17 had been used as parents in various combinations. This information indicated that there might be factors in Tx303 that would enhance the B73 x Mo 17 single cross hybrid if transferred to B73. Likewise, the information indicated that there might be factors in Oh43 that could be transferred to Mo17 to also enhance the B73 x Mo17 single cross. 3. Given what you know today about the QTLs you have tested, how much of the potential might you have tapped (or exposed) using the enhanced B73 x Mo17 hybrids? Answer: I do not have a good answer for that question. I feel, however, that we have only scratched the surface in the use of DNA markers to enhance plant breeding eiliciencies and potentials. 25