 |
|
|| Home || || Contact || Free Research Publications || Eshop || Privacy Statement ||
Evaluating Diversity Array Technology (DArT)
for the NSW Rice Breeding Program
mm
by Russell Reinke June 2006
RIRDC Publication No 06/055 RIRDC Project No DAN-204A
The challenge is using the information and new biotechnology techniques in a breeding program where the emphasis is on traits with complex inheritance such as yield and quality. This study aimed to examine one such new technique, Diversity Array Technology (DArT), with a view to establishing its advantages and disadvantages, and the extent to which it should be integrated into the existing rice improvement program Outcomes from this project are relevant to the RIRDC rice research committee and to the rice improvement program.
Background
The NSW rice industry is
of central importance to the Riverina Region of NSW, it is both geographically
compact, and small by world standards. There is limited funding for research,
and the amount of resources directed at genetic improvement must be commensurate
with the overall size of the industry. The genome of rice has been sequenced,
many genes have been located on the rice genome map, and molecular markers
for a wide range of traits have been developed. Although an important adjunct
to the NSW Rice Breeding Program, the need for marker tests for simply
inherited traits is relatively small. Micro-satellite markers for starch
structure have been integrated as a routine test into the breeding program
on F4 breeding lines, and a marker for fragrance is used on specific populations
at this stage. Although many of the genes for disease resistance are relatively
simply inherited and require simple molecular marker tests, the absence
of significant rice pests and diseases in NSW means that these tests are
a lower priority and the number of individual marker tests needed is relatively
low.
The NSW rice breeding program focuses mainly on improving grain yield and quality, both of which are polygenic or complex traits, conditioned by many interacting genes. Rice biotechnology research has focused on regions of the genome conditioning complex traits (quantitative trait loci or QTL), but these are not widely applicable, effective or affordable in breeding programs. Often QTL are only relevant to the population in which they have been developed, not to the broader germplasm base used in most breeding programs. Any association between a trait observed in the breeding population and the regions of the genome which vary (QTL) are highly dependent on the accuracy with which both measurements are taken.
Measurement of phenotype is problematic, as the environment in which the plants are grown can have a significant impact on the expression of the trait. If the measurement of phenotype is flawed, then the association of genotype and phenotype is similarly flawed and the associations will have little relevance or application to practical plant breeding programs.
Given the worldwide expansion in rice biotechnology, and the need for the NSW Rice Breeding Program to focus on more complex traits, there is a need to test and use relevant technology to assess its value and balance the efficiency and affordability.
Aims and objectives
The aims and objectives
of this project were to:
Methods used
There were three phases
to this research project. The first involved constructing a DArT reference
panel, which is the array of DNA fragments shown to vary among a broad
cross-section of rice varieties.
Secondly relationships between all current NSW rice varieties and breeding lines were examined. The final phase involved an endeavour to relate phenotype to genotype, using the quantitatively inherited trait seedling vigour. A series of populations derived from crosses between NSW varieties and a variety derived from an inter-specific cross between the native African species Oryza glaberrima and Oryza sativa, were characterised for seedling vigour. A sub-set of lines derived from one of these populations were carefully phenotyped for seedling vigour under controlled environment conditions. A series of high and low-vigour lines from this experiment were genotyped using DArT, and the DArT markers which varied among these lines were examined against the seedling vigour values.
The DArT reference panel
The varieties selected for
the central pool of DNA included approximately 50 commonly used in the
hybridisation program, representatives of the indica, japonica and tropical
japonica sub-species, and a range of other Oryza species which have the
potential to be used in the breeding program to contribute useful traits
in future. These included 21 Oryza rufipogon accessions from Nepal, China,
India, Laos and Malaysia, Oryza barthii, Oryza meridionalis, Oryza glaberrima
and a range of Oryza sativa including representatives from the japonica,
indica and tropical japonica sub-groups.
Briefly, DNA was extracted from all of these varieties and broken into fragments of varying sizes using specific enzymes. To reduce the number of fragments to a manageable size, a complexity reduction process was carried out. This involved selecting only those fragments which contained a specific small sequence (a miniature inverted transposable element or MITE) known to be widely and evenly distributed throughout the genome, and which is often associated with functional regions of DNA (ie. previously identified genes). The selected fragments had fluorescent dye attached and were arranged in an array on a glass slide. DNA extracted from each of the reference varieties in turn was fragmented, dyed, and washed over (hybridised to) the array to determine which of the fragments varied among the varieties constituting the broad sample of Oryza DNA. An automated system was used to measure variation in the array by examining the colour of each fragment on the array. Many of the fragments are common to all varieties, and are of no use in determining genetic differences. Hence, those fragments showing differences between varieties (polymorphism) were retained for the final reference array comprising 6,144 points. A total of 254 rice varieties were used in construction of the reference panel. Thereafter, by using different coloured dyes attached to fragments derived from different varieties, it is possible to carry out a pair-wise comparison showing where varieties have the same sequence and where they differ, for each of the fragments on the array. This is particularly useful when examining breeding lines from a bi-parental cross, as at each point where the parents are polymorphic, markers in the progeny show from which parent the DNA was derived at that position in the genome.
Assessing diversity of
varieties included in the reference panel
Results from the initial
DArT analysis were examined in two ways. First, from the pattern of DArT
markers for each variety the degree to which each line varies from the
others can be measured in terms of a dissimilarity value.
All the pair-wise comparisons result in a 254 × 254 dissimilarity matrix. A way of presenting these values is in a dendrogram which shows visually the grouping of varieties according to their measured genetic differences and similarities.
The grouping of varieties was also analysed by principal coordinate analysis (PCO). It is a variant of principal components analysis (PCA) which is a technique used to simplify a dataset. In essence PCA is a linear transformation that chooses a new coordinate system for the data set such that the greatest variance in a single dimension of the data set comes to lie on the first axis (then called the first principal component), and the second greatest variance on the second axis, etc. In contrast however, instead of the finding the coordinates maximising variance, PCO finds dimensions that maximise similarities among the data points. This provided an alternative method of visualising the grouping of varieties, and the information is highly relevant to selection of parents for future hybridisations.
Seedling vigour in a range
of populations
Five populations were developed
by hybridising four NSW cultivars with three inter-specific crosses between
the West African Oryza glaberrima × Oryza sativa (Table 1 and Table
2, see page 9). The resulting F2 seed was sown at Leeton Farm in the 2001/02
season as a space-planted population to minimize the competitive effect
between plants. Single panicles that flowered in less than 120 days and
less than 1m tall at maturity were selected. These F2-derived F3 (F2:3)
seeds were direct sown during 2002 at Redland Bay in Queensland. The seeds
were sown at two dates, 4 September and 17 October. These sowing dates
were selected because the temperatures at this time were similar to the
temperatures experienced during the establishment of rice in south eastern
Australia. Measurements on individual seedlings were made 35 days after
sowing for both sowing dates (9 October and 22 November, respectively).
There were 233 entries including parental lines and standard varieties. Lines were randomized within each population in each of three replications. Each plot was 1 m in length and comprised of 10 seedlings at 10 cm spacing. Due to establishment variability three uniform plants were selected for each plot for detailed measurements. The measurements included leaf number, height from the shoot base to the highest ligule (height 1), height from shoot base to the tip of the longest leaf (height 2) and total dry matter. Analysis was conducted using restricted maximum likelihood (REML) methods, and revealed significant spatial effects associated with columns and rows within each population.
Detailed measurement of
seedling vigour and DArT analysis
A controlled-environment
experiment was carried out to measure seedling vigour and its components.
The experiment was conducted twice, under slightly different environmental
conditions to provide some assessment of the interaction between genotype
and environment. A total of 69 semi-dwarf lines selected from an F3:5 population
of Quest × WAB-450-I-B-P-160-HB (WAB450). Quest is an Australian
semidwarf medium grain semi-dwarf variety adapted to the temperate Australian
climate. WAB450 is a tall West African inter-specific variety bred from
the tropical japonica WAB56-104 and an O. glaberrima accession from Ivory
Coast, CG14. WAB450 is estimated to retain about 8-10% of the O. glaberrima
genome (Ndjiondjop et al. 2003).
Trials were conducted with a randomised block design, with both phenotyping trials conducted on the same layout.
Both parents where included in the trial, as were 4 control varieties: 2 high-vigour varieties (HSC55 and the tall variety Hungarian No. 1) as well as the most popular Australian cultivar (Amaroo) and the most widely grown rice world-wide, IR64 (Peng et al. 2000).
Only 9 replications of 1 individual each were used for each parent and control variety, giving a total experiment size of 882 individual plants.
Unblemished seeds were selected from 2-3 panicles within each line and each seed weighed to four decimal places. The seedling trays were kept at full soil moisture capacity by thorough misting each day.
To approximate field conditions at sowing temperatures were set to a maximum of 22°C and minimum of 13°C. As seeds germinated, elongation measurements of the growing shoot were taken, from the tip of the emerging leaf to the soil level. For phenotyping trial 1 measurements were taken almost daily from 7 days after sowing (DAS) until at least 3 elongation measurements were obtained. These measurements were averaged to obtain a mean elongation rate. Seeds not having germinated by 19 DAS were treated as non-germinating. Individual dates of emergence were extrapolated from the data by subtracting the average elongation rate from the first shoot length measurement for the seedling.
Between 22-24 DAS, key measurements were obtained by destructive sampling. The widths and lengths of individual leaves were measured, and the plants cut off at 5mm above soil level and dried in a 30-38°C oven for 3 weeks. The remaining dry matter was then weighed to 4 decimal places to obtain an aboveground biomass measurement (dry weight).
Analysis of the components of seedling vigour indicated that leaf area and in turn, leaf width were the best and most repeatable representatives of vigour. Therefore, lines were ranked for their leaf area in both trials, and the widths of leaf 2 and 3 in both trials. By calculating the sum of these 4 rankings, an overall rank for each line was determined, as a representative index for vigour in the population. A total of 20 lines were selected to undergo DArT analysis, the two parents WAB450 and Quest and 9 lines from the high and low vigour extremes. The 9 lines with highest ranking (excluding parents and standard varieties) were selected as the 9 high-vigour lines, and the 9 lowest ranking lines were selected for the 9 low-vigour lines. These high and low-vigour lines were genotyped using DArT, and of the 61 DNA fragments or DArT markers which varied among these lines, four markers were significantly associated with improved seedling vigour.
Results/Key findings
DArT reference panel development
The reference panel constructed
during this project used DNA extracted from a total of 254 rice varieties,
including 50 varieties commonly used in the breeding program. The reference
panel was constructed using a broad range of varieties and species constituting
the “meta-genome”, and the scope and diversity of these varieties has resulted
in a reference panel that should remain relevant and useful to the breeding
program for a considerable time. The reference panel should only need to
be updated if the program has substantial change in the varieties used
for hybridisation, or focuses on a species which has little representation
on the panel. Significant expansion in the use of Oryza glaberrima per
se. (rather than the inter-specific cross involving O. glaberrima and O.
sativa) may require future changes to the reference panel to ensure that
such germplasm is well-represented.
DArT fingerprinting
The project has highlighted
the capability of DArT to provide detailed genetic fingerprints of varieties,
to distinguish closely related lines, and to establish relationships between
lines providing useful information for choosing parental combinations for
breeding.
Development of the reference panel demonstrated the capacity of the technique to differentiate all of the closely related lines within the suite of existing and prospective varieties from the NSW Rice Breeding Program.. Each DArT analysis can provide up to 600 markers widely distributed across the genome and thus in diverse crosses enables selection of lines with small DNA segments of the diverse parent inserted into a locally-adapted line.
Future association studies relating agronomic and grain quality data will determine if DArT markers can replace existing molecular marker tests such as for fragrance. DArT analysis will play an increasing role in the rice breeding program because the cost per marker is low, its highly automated system has high throughput capacity, and the data generated will have increasing value with continuing advances in bioinformatics.
DArT Markers and seedling
vigour
Among the 20 lines genotyped
(high and low vigour lines, and the parents) a total of 61 DArT markers
varied resulting in 1189 individual data points.
The number of DArT markers reflected about 1% polymorphism out of the 6,144 clone array. The hybridisation pattern for each line can be seen in Table 8 on page 28. A ‘1’ in the table indicates that the genomic DNA fragments from the individual line matched the DNA fragment on the array. A ‘0’ indicates that the DNA fragment from the line was different to that on the array and no hybridisation was detected. This way, alleles for each DArT marker in the parent varieties can be traced to the progeny lines to identify which alleles in the progeny were derived from which parent.
DArT markers significantly associated with the high and low vigour groups were indicated by the probability value for Fisher’s Exact Test on the right hand column of Table 8 (page 28), and these results indicate that only 5 of the 61 markers varied significantly between the high and low vigour groups. Three of the significant DArT markers for vigour had identical marker distributions and are probably linked, although they may represent independent loci that influence the expression of vigour.
It is interesting to note that for these 3 clones, the high vigour allele originated from Quest, the supposedly low vigour parent. The occurrence of high vigour alleles from the low vigour parents has been reported in other vigour studies (Redoña & Mackill, 1996b and Zhang et al.2005b). This further emphasises that vigour is a genetically complex character and underscores the importance of genetic context, where successful expression of this character requires the interaction of genetic components dispersed across the genome.
Principle coordinates analysis, or metric multidimensional scaling, of the 61 DArT markers was performed, and a scatter plot of the first two axes (modelling a cumulative 91.68% of the variation observed) is presented in Figure 13 on page 27. This figure showed WAB450 and Quest well separated, although the progeny lines appeared more similar to Quest than WAB450. Several lines clustered with Quest, while the WAB450 parent was well separated from other lines. High and low vigour lines were not distinctly separated, however all high vigour lines fell below zero for axis 2, and 6 of 9 low vigour lines above. The high and low vigour lines did not separate into different clusters, although some smaller cluster emerged. For example, the high vigour lines 18:55, 18:59, 18:122, 19:287 and 19:289 as well as the low vigour line 18:136 clustered relatively close together. Given that most of the 61 DArT markers were not significantly associated with seedling vigour it is not surprising that the clustering did not clearly separate the high and low vigour groups.
Developing meaningful marker-trait associations is not a simple exercise, even with the relatively large numbers of markers generated with microarray technology. With complex traits such as vigour there is often no clear-cut phenotype, and the vigour groups genotyped represent a collection of lines with generally high or generally low vigour. This, in turn, reduces the ability to search for clear-cut genotypic differences.
Each DArT marker is sequence ready, so the next step in the process of identifying markers related to vigour would be to sequence the clone and search the published rice genome sequence to find the position within the genome.
The gives the opportunity to relate previously identified genes, located close to the DArT marker, to the trait of interest, and may provide clues as to the underlying mechanism which influences or controls the trait. Further, populations from this specific cross and other related crosses can be grown under field conditions to confirm the marker-trait association.
Implications
The use of DArT analysis
in this project has shown that the technology can provide detailed analysis
of the suite of parental lines used in the Rice Breeding Program, showing
the degree of genetic differences or similarities between all varieties.
Such information is useful for future selection of parents, allowing the
use of measurable genetic diversity rather than an estimate of diversity
based on phenotypic differences.
DArT is able to distinguish between closely related varieties and breeding lines emanating from the NSW Rice Breeding Program, demonstrating its capability to provide genetic fingerprinting in the case of uncertainty in seed identification, and as a means of quality assurance in the production of pure seed of existing commercial varieties and new breeding lines approaching release.
In diverse and inter-specific crosses DArT analysis is able to provide a fast and accurate means of determining the extent of introgression of the genome of the diverse parent. Capturing useful variation from such crosses inevitably requires multiple backcrossing to end up with a suitable genetic background for the NSW rice growing environment. Hence there is a need to ensure that at each hybridisation to the recurrent parent is a real cross, and not an inadvertent self-pollination. DArT analysis provides this information as well as an indication of the extent of the genome transferred.
If DArT analysis can be modified to include points on the array of known sequence that can test for existing well-defined molecular markers, then some of the resources devoted to existing simple markers can be used for DArT analysis, increasing the efficiency of the molecular marker part of the breeding program. The feasibility of this has not yet been canvassed in detail with Diversity Arrays Pty Ltd.
Development of usable QTL’s for complex traits such as seedling vigour and cold tolerance rests on the capacity to phenotype accurately and repeatably. In future studies DArT markers associated with useful traits should be sequenced and located on the rice genome to provide useful information as to the possible mechanism and genetic control of such traits. However, the increased emphasis on DArT and other biotechnological systems should not be at the expense of phenotyping capacity within the NSW Rice Breeding Program, since both disciplines are critical to future success.
There is a need for the rice improvement program to continue to develop the use of biotechnology to achieve greater efficiency, enhanced market responsiveness and address present and looming sustainability issues. As molecular markers for specific traits are added, greater efficiency is needed in the marker testing and selection program. DArT has the potential to encompass these needs through a staged introduction, drawing on the experience of other programs, while maintaining and improving existing phenotyping capacity.
|