Molecular Marker-assisted Breeding in Rice Jian-Long Xu Institute of Crop Sciences, CAAS Email: [email protected]
May 11, 2015
Molecular Marker-assisted
Breeding in Rice
Jian-Long Xu
Institute of Crop Sciences, CAAS
Email: [email protected]
Expertise & experiencesMolecular rice breeding (including allele mining& marker-assisted breeding)
August 2003 ~ presentMolecular Rice Breeder in the Institute of Crop Sciences, CAAS
2008 ~ 2012One month per year for Consultant in PBGB Division, IRRI
2005 ~ 2007Three months per year for Collaboration Research in PBGB Division, IRRI
January 2002 ~ October 2003Postdoctoral Fellow in PBGB Division, IRRIPostdoctoral Fellow in PBGB Division, IRRI
March 1999 ~ October 2000PhD thesis research in PBGB Division, IRRI
August 1990 – July 2003Senior Rice Breeder in Zhejiang Academy of Agricultural Sciences
PhD 2001 Zhejiang University,
China
Genetics (minor in
Statistics)
MSc 1990 Zhejiang Agricultural
University, China
Plant Breeding and
Genetics
BSc 1977 Zhejiang Agricultural
University, China
Plant Breeding and
Genetics
Successful breeding depends on:
(1)Variation: Sufficient (novel) genetic variation for
target traits in breeding populations
(2) Selection efficiency: Effective selection approach
to identify desirable alleles or allelic combinations for
the target traits in breeding populations
Traditional breeding depends on phenotypic selections.
Efficiency of selection is largely influenced by environment,
gene interaction, and gene by environment interaction.
Genetic markers can improve efficiency of selection. Genetic
markers include morphological marker (plant height, leaf
color), cytological marker (chr structure and no mutant),
biochemical marker (isozyme), and molecular marker (SSR).
DNA
RNA
Direct
selection:
Indirect
selection
Phenotypic indirect selection
(based on correlation between
traits)
Based on phenotypic value
Protein
Phenotype
selectiontraits)
Genotypic indirect selection
(based on markers associated
with a gene or QTL)
Marker-assisted selection (MAS) is a method whereby
a phenotype is selected on the genotype of the linked marker.
Note: marker isn’t the target gene itself, there is just an
association between them.
Linkage of the target gene with the marker
Genotypes of the parents
Resistant donor Recipient
Genotypes of the parents
Genotypes of the F1
Three genotypes of the F2 population
Selection with 95% confidence based on
marker genotypes when recombination
rate (r) of 5%
The advantages of MAS:
(1) Time saving from the substitution of complex field trials (that need
to be conducted at particular times of year or at specific locations,
or are technically complicated) with molecular tests;
(2) Elimination of unreliable phenotypic evaluation associated with
field trials due to environmental effects;
(3) Selection of genotypes at seedling stage;
(4) Gene ‘pyramiding’ or combining multiple genes simultaneously;(4) Gene ‘pyramiding’ or combining multiple genes simultaneously;
(5) Avoid the transfer of undesirable or deleterious genes (‘linkage
drag’; this is of particular relevance when the introgression of
genes from wild species is involved);
(6) Selecting for traits with low heritability;
(7) Testing for specific traits where phenotypic evaluation is not
feasible (e.g. quarantine restrictions may prevent exotic pathogens
to be used for screening).
Population development
Gene or QTL mapping
Linkage map construction/ phenotypic
evaluation for traits/ QTL analysis
QTL validation
Procedure of MAS
Considering mapping and
breeding purposes
QTL validation
Confirmation of position and effect of QTL/
verification of QTL in different populations and
genetic backgrounds / fine-mapping
Marker validation
Testing of marker in important
breeding parents
Marker-assisted selection
Requirements for large-scale application of MAS
◆◆◆◆ Validation of QTL in breeding materials
Multiple markers in vicinity of QTL desirable.
◆◆◆◆ Simple, quick, inexpensive protocols for tissue sampling,
DNA extraction, genotyping and data collectionDNA extraction, genotyping and data collection
◆◆◆◆ Efficient data tracking, management and intergration
with phenotypic data
◆◆◆◆ Decision support tools for breeders
optimal design of selection strategies
accurate selection of genotypes
Strategies of MAS
1 Foreground selection
Selection against the target gene.
◆◆◆◆ Single marker selection
Reliability: depends on linkage between the marker and the
target gene. For example, marker locus (M/m) links with the target gene. For example, marker locus (M/m) links with the
target gene locus (S/s), if the recombination rate between the
two loci is r, the probability of selection of genotype S/S based
on marker genotype of M/M is
P=(1-r)2
So, reliability of MAS will sharply decrease with the increase of
recombination rate. To ensure reliability of MAS more than
90%, the r should be lower than 5%.
If the probability to select 1 target plant is P, the minimum
number of plants with marker genotype M/M will be
calculated as:
N=log(1-P)2/log(1-r)2
So, when the recombination rate (r) is high as 30%, So, when the recombination rate (r) is high as 30%,
selection of 7 plants with M/M genotype will ensure to
obtain 1 target plant with probability of 99%, whereas we
must select 16 plants if MAS isn’t applied (namely, there is
no linkage between the marker and the target gene).
MAS scheme for early generation selection in a typical breeding program for disease resistance. A susceptible (S) parent is crossed with a resistant (R) parent and the F1
plant is self-pollinated to produce a F2 population. In this diagram, a robust marker has been developed for a major QTL controlling disease resistance (indicated by the arrow). By using a marker to assist selection, plant breeders may substitute large field trials and eliminate many unwanted genotypes (indicated by crosses) and retain only those plants possessing the desirable genotypes (indicated by arrows). Note that 75% of plants may be eliminated after one cycle of MAS.
◆◆◆◆ Bilateral marker selection
Bilateral marker selection will greatly improve reliability of
MAS.
If marker loci M1 and M2 locate each side of the target gene
locus S, and the recombination is r1 and r2 respectively,
thus F genotype is M SM /m sm , F -derived Fthus F1 genotype is M1SM2/m1sm2, F1-derived F2
population has two genotypes, M1SM2 (harbor the target
gene) and M1sM2 (without the target allele). In view of
probability of double crossing over is very low, so selecting
genotypes at M1 and M2 loci to track the garget gene S is
high reliable.
Without interrupt, the probability to obtain genotype S/S
by selection of bilateral marker genotypes M1M2/M1M2 is:
P=(1-r1)2 (1-r2)
2/[(1-r1)2 (1-r2)
2 + r1r2]
◆◆◆◆ When r1=r2 (the target gene is located in the middle of
the two marker loci), P will be minimum.the two marker loci), P will be minimum.
◆◆◆◆ In fact, two single crossing over generally interrupt
each other, thus resulting in even small probability of
double crossing over, so reliability of bilateral marker
selection is higher than expected.
Comparison of target control between single
marker and bilateral marker
It is clearly indicated that control of the
target gene by a single marker isn’t so
satisfactory in most cases. The marker
must be as close as 1 cM to the target to
keep the risk of ‘losing’ the target below
5% after five BC generations. Even with
a single marker at 1 cM, the risk of losing
the target is close to 10% in BC10. For
greater distance of a single marker, the
risk becomes rapidly too high.
For the case of bilateral markers, even if
the two marker loci are far apart, for
example 10 cM, efficiency of keeping the
risk of losing the target is almost same as
that in the case of 1 cM under single
marker. Obviously, breaking linkage
between marker locus and the target
gene in bilateral markers more difficult
than in single marker.
2 Background selection
Besides selection of the target gene (foreground selection), background
selection will be implemented if to keep original characters of a variety.
◆◆◆◆ MAS method: use a set of markers, which are evenly selected from
the whole genome to identify the genotype of the recurrent parent.
Normally screening background will be focused on those plants with
target gene.
◆◆◆◆ Consecutive backcrossing: backcrossing progeny will soon recover its
recurrent parental genome after several rounds of backcrossing.
Breeding method BC1F1 BC2F1 BC3F1 BC6F1
Traditional backcrossing 75 87.7 93.3 99
MAS-based backcrossing 85.5 98 100
% of the recurrent parental genome
Young & Tanksley 1989
Traditional
BC breeding
Year
Comparison of MAS and traditional BC breeding for
recovery of genetic background of the recurrent parent
MAS BC
breeding
YearBlack bar represents donor
genome
Only two BC generations, the target segment can be narrowed
down into 2 cM by MAS and completely diminish linkage drag
from donor parent.
MAS application in qualitative traits
In most cases, it is unnecessary to apply MAS for
qualitative traits. However, MAS does improve efficiency
of selection of qualitative traits in following cases:
◆◆◆◆ Pyramiding different resistance genes;◆◆◆◆ Pyramiding different resistance genes;
◆◆◆◆Difficulty in or high cost of phenotyping;
◆◆◆◆ Hope to select in early growing stage but the traits
normally express in late developing stages
◆◆◆◆ Screening genetic background besides the target
traits
1 Pyramiding of multiple genes
Pyramid different genes dispersed in various varieties into
one variety by MAS.
Different genes for the same target trait: to improve Different genes for the same target trait: to improve
trait value.
Multiple genes underlying different traits into the
same variety: ensure new variety having more
favorable traits
Example of genes for pyramiding in cereals
Chr6 Chr11 Chr12
Three bBlast resistance genes used for pyramiding
Zheng et al. 1995
C101LAC x C101A51
Pi-1 Pi-2
C101LAC x C101PKT
Pi-1 Pi-4
F1 F1
F2 150 plants F2 150 plants
Scheme of thre blast resistance genes pyramiding
10 plants homozygous
at Pi-1 & Pi-2
10 plants homozygous
at Pi-1 & Pi-4
Bilateral marker selection
X
F1
X
F2 150 plants
MAS
Plants with 3 resistance genes
Resistance
geneChr.
Marker
name
Linkage
distance
(cM)
Primer sequenceAnnealing
temperature
Size of
amplified
fragment (bp)
Pi-GD-1(t) 8
RM6208 3.4TCGAGCAGTACGTGGATCTG
55 90CACACGTACATCTGCAAGGG
To pyramid different blast resistant genes in Zanhuangzhan2 (3
major genes and 1 QTL) and one brown planthopper resistant gene
(Bph18(t)) in IR65482 into 3 dominant restorer lines (Chen et al. 2012)
Information of resistant genes and their linked markers
Pi-GD-1(t)
-G18
R8M10 3.4ACCAAACAAGCCCTAGAATT
56 235TGAGAAAGATGGCAGGACGC
Pi-GD-2(t)
–G29 RM3855 3.2
AATTTCTTGGGGAGGAGAGG55 424
AGTATCCGGTGATCTTCCCC
Pi-GD-3(t)
–G312 RM179 4.8
CCCCATTAGTCCACTCCACCAC
C 61 190
CCAATCAGCCTCATGCCTCCCC
GLP8-6(t)
–G88 G8-6ID-1 2.8
ATCCGGCACTACCTTTCCC55 235
CTGCTCCCACCGCATCTGT
Bph18(t) 12 7312.T4A 1.3AACAGCAGAGGGTTTGGCTA
50 1078CAGACTTTTCTTGGGGGTCA
Minghui86, Shuhui527 and
Zhehui7954 (Recurrent parent, RP)
Sanhuangzhan 2、、、、IR65482
(Donor parent, DP)x
F1
RP
BC1F1Pyramiding F1
Pyramiding
BC2F1
BC3F1
F2
F3
MAS
MAS
MAS
MAS
RP
RP
MAS MAS
BC3F2
BC3F3
F4
F5
MAS
MAS
MAS
Evaluation on resistance and agronomic traits for restorer
lines and their derived hybrids
Test-crosses with II-32A and Huhan11A
Scheme of molecular improvement of blast and brown
planthopper resistance for restorer lines
Restorer lines
Strain ReactionResistance
frequency
(%)S
1
S
2
S
3
S
4
S
5
S
6
S
7
S
8
S
9
S
10
S
11
S
12
S
13
S
14
S
15
S
16
S
17
S
18
S
19
S
20S R
CO39 S S S S S S S S S S S S S R R S S S S S 18 2 10
Sanhuangzhan2 R S R R R R R R R S R R R R R R R R R R 2 18 90
Minghui86 R R R R R R R R R S R R R R R R S R R R 2 18 90
Shuhui527 R S R R R S R R R R R S R R R R R R R R 3 17 85
Zhehui7954 R S S S S S S S S S R S S R R R S S R R 13 7 35
Evaluation of resistance of newly bred restorer lines to Pyricularia grisea Sacc.
Minghui86-G2 R R R R R R R R R S R R R R R R R R R R 1 19 95
Minghui86-G1-G2 R R R R R R R R R S R R R R R R S R R R 2 18 90
Shuhui527-G2 R R R R R S R R R R R R R R R R R R R R 1 19 95
Shuhui527-G1-G2 R R R R R S R R R R R R R R R R R R R R 1 19 95
Zhehui7954-G1-G2 R S R R R R R R R S R R R R R S R R R R 3 17 85
Zhehui7954-G1-G2-G8 S R R R R R R R R S R R R R R R R R R R 2 18 90
Zhehui7954-G1 -G8-
Bph18(t)R S S S R S R S S R R R S R R R S S R R 9 11 55
Zheshu-G2-G8 R S R R R R R R R S R R R R R R R R R R 2 18 90
Mingzhe-G2-G8 R S S R R R R R R S R R R R R R R R R S 4 16 80
Mingzhe-G1-G2-G8 R R R R R R R R R S R R R R R R R R R R 1 19 95
Mingzhe-G1-G2-Bph18(t) S R R R R R S S R R R S R S S R R R R R 6 14 70
NameResistant
gene
Seedlings
inoculated
No. of
survival
Resistant
score
Minghui86 - 19 0 9
Shuhui527 - 19 0 9
Zhehui7954 - 20 0 9
TN1(CK) - 20 0 9
Performance of resistance-improved restorer lines to brown planthopper
TN1(CK) - 20 0 9
IR65482 Bph18(t) 20 20 1~3
Shuhui527-Bph18(t) Bph18(t) 20 15 3
Zhehui7954-G1-G8-Bph18(t) Bph18(t) 18 12 5
Mingzhe-G1-G2-Bph18(t) Bph18(t) 20 16 3
Restorer line or combination PL SF SNP TGW PH HD GY
(cm) (%) (g) (cm) (d) (g/plant)
Minghui86 24.9 83.5 180.1 29.9 103.5 106.0 17.6
Minghui86-G2 26.0 86.8 174.7 26.6 99.7 110.0 18.3
Minghui86-G1-G2 22.9 92.8 169.8 27.5 98.2 107.0 17.9
II-32A/ II-32A/Minghui86 25.0 96.7 198.0 28.0 99.0 98.0 23.8
II-32A/ II-32A/Minghui86-G1-G2 24.0 94.9 188.2 25.9 95.3 97.0 24.9
LSD0.05 0.8 5.4 24.5 1.8 9.2 1.2 3.1
LSD0.01 1.1 7.7 34.9 2.5 13.1 1.7 4.4
Shuhui527 26.4 92.0 182.8 33.0 101.4 111.0 17.5
Shuhui527-G2 25.9 88.5 153.7 31.3 85.8 112.0 16.9
Shuhui527-G1-G2 24.5 85.7 179 27.1 97.5 107.0 21.0
Shuhui527-Bph18(t) 25.9 87.2 178.8 32.1 95.1 111.5 24.4
Agronomic performance of newly bred restorer lines and their hybrids during
2011 winter season in Hainan
Shuhui527-Bph18(t) 25.9 87.2 178.8 32.1 95.1 111.5 24.4
II-32A/ II-32A/Shuhui527 23.4 79.3 190.8 26.1 89.2 97.0 18.9
II-32A/ II-32A/Shuhui527-G2 24.4 92.8 203.4 26.9 93.6 101.0 21.5
II-32A/ Shuhui 527-G1-G2 22.7 92.4 170.9 26.6 89.5 100.0 17.8
II-32A/Shuhui527-Bph18(t) 23.8 92.1 174.8 29 95.7 102.0 21.2
LSD0.05 1 4.5 23.8 0.8 4.1 1.9 3.5
LSD0.01 1.3 6.1 32.3 1.1 5.6 2.5 4.7
Zhehui7954 19.6 83.5 203.3 26.8 89.2 104.0 19.0
Zhehui7954-G1-G2 19.7 90.6 178.7 25.7 91.2 100.0 23
Zhehui7954-G1-G2-G8 21.9 84.0 207.5 27.5 92.1 106.5 24.9
Zhehui7954-G1-G8-Bph18(t) 24.1 94.1 156.8 27.8 93.5 107.0 25.7
II-32A/Zhehui7954 22.2 89.3 211.1 26.1 90.7 97.5 23.8
II-32A/Zhehui7954-G1-G2-G8 22.0 93.1 176.6 27.4 96.7 100.0 23.5
II-32A/Zhehui7954-G1-G8-Bph18(t) 23.7 94.3 195.3 26.7 96.9 98.5 28.7
LSD0.05 0.9 3.4 24.5 1.3 4.6 1.1 4.1
Some important issues about MAS improvement of
resistance for restorer lines
(1) Firstly, the resistance improvement of parental lines of hybrids is
much different from that of conventional varieties. In the backcross
progenies of restorer parental lines, selections were performed not
only for similarity to the recurrent parents (RP), but also for their
fertility restoring gene(s) and specific combining ability to the CMS
lines.
◆◆◆◆ background recovery of the RP◆◆◆◆ background recovery of the RP
◆◆◆◆ the qualitatively inherited fertility restoring gene(s) of the RP
◆◆◆◆ the quantitatively inherited specific combining ability. It is gradually
recovered through backcrossing in different individuals to a varying
extent.
It was indicated that a minimum of three backcrosses in conjunction
with stringent phenotypic selection for the RP in each BC progenies
and combining ability testing on a relatively large scale, guarantees
the recovery of recurrent parental characteristics even without MAS
against the background of the RP
(2) Secondly, the level of hybrid rice resistance is determined by the
restorer line when CMS is susceptible, whereas the resistance level of
F1 is controlled by the interaction between CMS and restorer line
when CMS is resistant. Expression of many resistance genes such as
Xa21, etc., are affected by genetic background. So resistance of
hybrids derived from the resistant restorer lines probably compromise
and show resistance inferior to our expected. So we should choose
highly resistance genes for resistance improvement of hybrid rice.
(3) Backcrossing is a very efficient strategy to improve single trait. (3) Backcrossing is a very efficient strategy to improve single trait.
However, the newly released lines are phenotypically identical to the
RP, i.e. there is no break through in traits of the new variety. So
composite intercrossing is recommended to pyramid multiple
resistance genes as well as to create new variety. In MAS breeding
programs, polymorphic markers are the key problem when multiple
parents are involved. So it is better to develop linked markers showing
polymorphism among all parents, otherwise efficiency of MAS will be
degraded.
MAS for quantitative genes
Most important agronomic traits are genetically quantitative
and controlled by polygenes. In the past decades, some major
QTLs have been implemented by MAS.
Procedures MAS for quantitative traits:
◆◆◆◆ QTL initial mapping◆◆◆◆ QTL initial mapping
◆◆◆◆ Fine-mapping of major QTL
◆◆◆◆ Verification of gene effect using NILs
◆◆◆◆ Validation of molecular markers
◆◆◆◆ MAS application
RM283
R844
S2139
RM23
0.0
27.4
28.4
40.0
RM283
R844
S2139
RM23
0.0
27.4
28.4
40.0
RM283
R844
S2139
RM23
0.0
27.4
28.4
40.0
AP3206
RM3412CP03970
RM8094
0.0
1.0
1.3
Short arm of chromosome 1
Progress of Saltot locus
• Saturated map of the
Chromosome 1
(Saltol segment) is
developed
60.6 RM23
RM140
RM113
S1715S13994RM9
RM5
C1456RM237RM246
40.0
64.9
66.2
71.2
75.3
77.2
91.998.299.1103.1
119.5123.5
129.9
A
C52903S
C1733S
R2374B
C52903S
C1733S
R2374B
RM23
RM140
RM113
S1715S13994RM9
RM5
C1456RM237RM246
40.0
75.3
77.2
91.998.299.1103.1
119.5123.5
129.9
RM23
RM140
RM113
S1715S13994RM9
RM5
C1456RM237RM246
40.0
75.3
77.2
91.998.299.1103.1
119.5123.5
129.9
RM8094RM493CP6224
RM140
1.8
1.9
1.2
1.3
• Closely linked
markers linked to
the saltol locus
identified
• MAS is being
validated in 3
breeding populations
60.6
(Source: Glenn B. Gregorio)
RM3412
CP010136
AP3206
Chromosome location of associated QTL of Salinity tolerance trait
LOD threshold
RM140
CP6224
RM493
RM8094
CP03970
0.0
b
a
2.5
12.11Mb 12.27Mb
12.0Mb 12.27 Mb
preprotein translocase, SecA subunit
Sec23/Sec24 trunk
WD40
Ser Thr Kc
Receptor like kinase
SAM synthetase
cold shockprotein
chloroplast membrane protein
secretory peroxidase
CBL-interacting protein kinase 19
Peroxidase, putative
S_Tkc;WD40
0.27 Mb
SALtol Region ( Major QTL K+/Na+)
(~40 genes)
11.9 Mb 12.13 Mb
12.25Mb 12.40Mb
11.10Mb 12.7Mb
60.6 60.9 62.5 64.9 65.4 66.2 67.6 67.9
cM
65.8
Chromosome 1 of Rice
B1135C02
OSJNBa0011P19
P0426D06
B1153f04
Salt tolerant rice varieties developed by IRRI and released
in Philippines
IRRI 112 - PSBRc48 (Hagonoy) IRRI 113 - PSBRc50 (Bicol) IRRI 124 - PSBRc84 (Sipocot) IRRI 125 - PSBRc86 (Matnog) IRRI 126 - PSBRc88 (Naga) IRRI 128 - NSICRc106 IRRI 128 - NSICRc106
Other salt-tolerant rice varieties
CSR10, CSR13, CSR23, CSR27, CSR30, CSR36 and Lunishree, Vytilla 1, Vytilla 2, Vytilla 3, Vytilla 4, Panvel 1, Panvel 2, Sumati, Usar dhan 1, 2 & 3 (India); BRRI dhan 40, BRRI dhan 41 (Bangladesh); OM2717, OM2517, OM3242 (Vietnam)
Progress of Sub1A locus
A major QTL on chrom. 9 for
submergence tolerance – Sub1 QTL
10
15
20
IR40931-26 PI543851
0 10 20 30 40
LOD score
50cM
OPN4
OPAB16
C1232
RZ698
OPS14
RG553
R1016RZ206
RZ422
Sub-1(t)
1200
850
900
OPH7950
OPQ1600
1 2 3 4 5 6 7 8 9
0
5
Submergence tolerance score
Segregation in an F3 population
100cM
150cM
C985
RG570
RG451
RZ404
Xu and Mackill (1996) Mol Breed 2: 219
Sub1 locus, there are three structurally related genes Sub1A,
Sub1B, and Sub1C present in the same QTL region, encoding
ethylene-responsive factor (ERF) genes.
Fukao, et al., Annals of Botany, 2009,103: 143–150
Development of the submergence-tolerant Swarna-Sub1 with details of markers
used for foreground, recombinant, and background selection.
Field plot test of submergence tolerance of Sub1 and non-Sub1 varieties. The SUB1 locus from
FR13A was introduced into the rice varieties IR64 and Samba Mahsuri by marker-assisted
backcrossing and into IR49830-7-1-2-2 through conventional breeding. A field trial performed
at IRRI in 2007 included Sub1 lines, the progenitors, and IR49830-7-1-2-2 (tolerant, used as
SUB1 donor) and IR42 (sensitive) as checks. Fourteen-day-old seedlings were transplanted
into a field with high levees, grown for 14 days and then completely submerged with about 1.25
m of water for 17 days. The field was drained, and the plants were allowed to recover under
non-stress conditions. The photograph shows the performance of the lines about 60 days after
de-submergence.
Swarna with Sub1
MAS of Minor-effect QTLs
At present, using limited number of markers and small
mapping populations, only few QTLs with relatively large
phenotypic-effect have been identified, which account for a
small portion of QTLs affecting the target traits. Moreover,
QTL epistasis has great effect on selection. So, it is difficult QTL epistasis has great effect on selection. So, it is difficult
to implement MAS for minor-QTLs.
Genome selection (GS) will provide a new strategy for
mionr-QTLs (introduced later).
Genome-wide selection
Training population: used for genotyping with high throughput
SNP marker and phenotyping in the target environment, setting
up genetic predict model to estimate all possible QTL effects
affecting a trait
Breeding population: used for genotyping and predicting breeding
values for selection
In a training population (both genotypic and phenotypic data available),
fit a large number of markers as random effects in a linear model to
estimate all genetic effects simultaneously for a quantitative trait. The
aim is to capture all of the additive genetic variance due to alleles with
both large and small effects on the trait.
In a breeding population (only genotypic data available), use estimates
of marker effects to predict breeding values and select individuals with
the best GEBVs (genomic estimated breeding values).
GS consists of three steps:
(1) Prediction model training and validation
A training population (TP) consisting of germplasm having both
phenotypic and genome-wide marker data is used to estimate
marker effects.
(2) Breeding value prediction of single-crosses
The combination of all marker effect estimates and the marker data of
the single crosses is used to calculate genomic estimated breeding the single crosses is used to calculate genomic estimated breeding
values (GEBVs).
(3) selection based on these predictions
Selection is then imposed on the single crosses using GEBVs as
selection criterion. Thus, GS attempts to capture the total additive
genetic variance with genome-wide marker coverage and effect
estimates, contrasting with MARS strategies that utilize a small
number of significant markers for prediction and selection.
Advantages of GS:
◆◆◆◆ It is especially important for quantitative traits conferred by a
large number of genes each with a small effect.
◆◆◆◆ GS includes all markers in the model so that effect estimates are
unbiased and small effect QTL can be accounted for.
◆◆◆◆ Reduce the frequency of phenotyping because selection is based on
genotypic data rather than phenotypic data.
◆◆◆◆ Reduce cycle time, thereby increasing annual gains from selection.
Disadvantages of GS:
◆◆◆◆ Traits with lower heritability require larger TPs to maintain high
accuracies.
◆◆◆◆ When single crosses are unrelated to the training population (TP),
even if sufficient markers and training records are available, marker
effects could be inconsistent because of the presence of different
alleles, allele frequencies, and genetic background effects, i.e.
epistasis. So genetic model isn’t universal in different populations.
Most agronomic important traits are quantitatively inherited. A wide
range of segregating populations derived from bi-parental crosses,
including RILs, DHs, F2 and its derived populations, and BC or testcross
populations, have been used for QTL mapping. And many major
important QTLs have been cloned in rice. Oppositely, slow progresses
have been made so far in MAS-based breeding for complex traits, mainly
due to the following two aspects.
Summary of MAS for quantitative traits
due to the following two aspects.
(1) Segregation populations derived from bi-parents can’t identify
favorable alleles for the target traits. So we don’t have information about
favorable alleles for the target trait which will be best used in molecular
breeding.
(2) QTL mapping is separate from breeding program. Owing to QTL
mapping results are seriously dependent on genetic background. So QTL
information from mapping populations can’t be directly applied in MAS-
breeding.
So, integration of QTL mapping with MAS-based
breeding in the same genetic background has been
strongly recommended for complex quantitative traits by
Tanksley and Nelson (1996). So far, AB-QTL method has
been widely used in QTL identification from germplasm.
However, there are still some defects:
(1) Relative high expenses resulting from phenotyping and
genotyping for a large mapping population.
(2) Favorable alleles can not be mined using populations
derived from bi-parents.
With the development of sequencing technologies and the sharp
decreased sequencing cost, genome wide association (GWS) has
been recently used for QTL mapping and allele mining from
germplasm resources and made good progresses. However, there
are still some problems with this method.
(1) Wide variations in plant height and heading date of a natural
population seriously affect growth and development for some population seriously affect growth and development for some
early and dwarf entries, thus resulting in inaccurate phenotyping
for those parts of entries.
(2) There is population structure effect on QTL association
mapping.
(3) GWS and MAS-based breeding is still separate.
Germplasm holds a large of genetic variation for improving agricultural
crops. However, in the past favorable genes from germplasm have not
been efficiently used in plant breeding due to linkage drag. Although
backcross is effective to simple qualitative traits, it has not been
successful to improve quantitative traits by backcross breeding
procedure.
Here we demonstrate a new breeding strategy of backcross combined
molecular marker technology to efficiently identify QTL and improve
multiple complex traits based on designed QTL pyramiding (DQP).multiple complex traits based on designed QTL pyramiding (DQP).
RP x donors (many) F1s x RP BC1F1s x RP
~25 BC2F1s/donor x RP
x
BC2F3-5 bulk populations
BC3F1s x RP
1, 2, 3, 4, 5, 6, ……
BC3F2-3 bulk populations
Self and bulk
harvest
Selection for target traits
and backcrossing
BC4F1s
BC4F2s
x
x
Self and bulk
harvest
1, 2, 3, 4, 5, 6, ……
Screening for target traits such as tolerances to drought, salinity,
high temperature, anaerobic germ., P & Zn def., BPH, etc.
Strategy of integration of QTL mining with QTL-designed pyramiding using backcross introgression lines in elite background
Confirmation of the selected traits by replicated phenotyping
then genotyping of trait-specific lines (ILs)
Crosses made between sister ILs
having unlinked desirable
QTLs for target ecosystem
DQP & MAS for pyramiding desirable
QTLs and against undesirable donor
segments for target ecosystem
Develop multiple stress tolerant lines for different ecosystems and release
NILs for individual genes/QTLs for functional genomic studies
high temperature, anaerobic germ., P & Zn def., BPH, etc.
QTL identification and allele mining
ST-ILs selected from four
introgression populations in
Minghui86 background at the
overall growth stage
Minghui86/Shennong265 (40)
Minghui86/Zaoxian14 (33)
Minghui86/Y134 (40)
Minghui86/Gayabyeo (37)
Salt tolerant introgression lines (ILs) and QTL mapping
Principle of using selected ILs and molecular
markers to identify QTLs
QTL detection
Taken allele frequency of the random population as an expected value, a
significant deviation (excess or deficiency) of donor allele frequency at
single loci in the selected IL population from the expected level implies a
positive selection favoring the donor allele (in excess), or negative positive selection favoring the donor allele (in excess), or negative
selection against the donor allele (in deficiency). Significant deviation
loci are considered as QTLs affecting the selected traits.
Gene action at putative QTLs
●●●● Excess of the donor homozygote additive gene action
●●●● Excess of the heterozygote overdominance gene action
●●●● Excess of both the donor homozygote and heterozygote partial
or complete dominance gene action
0.33 0.31 0.64 <.000123.4 2.44Bin3,13RM231
–0.290.34 0.05 0.0001 17.9 31.06Bin2,62RM240
0.57 0.22 0.78 0.0000 68.5 17.80 Bin2,32LT62
0.25 0.03 0.28 <.0001102.60.56 0.25 0.81 <.000168.4 10.07Bin2,22RM29
0.38 0.63 1.00 <.000124.043.24 Bin1,81LT44
0.46 0.08 0.54 0.0000 104.8 34.64 Bin1,61LT35
Diff.Random
pop.ST-ILs
P X2
Frequency of
introgressionP X2
Minghui8686/Shennong265 (15)Minghui86/Gayabyeo (13)
Physical Physical Physical Physical position position position position
/Mb
BinChr.Chr.Chr.Chr.MarkerMarkerMarkerMarker
0.65 0.00 0.65 0.0002 16.90.49 0.20 0.69 0.0000 56.6 2.57 Bin1,11LT3
QTLs for ST detected in Minghui86/Gayabyeo and Minghui86/Shennong265 ILs
Frequency of
introgression
ST-ILs Random
pop.Diff.
0.39 0.13 0.51 <.000151.69.93 Bin12,212LT365
0.50 0.09 0.59 <.0001100.8 17.31Bin11,311RM209
0.30 0.16 0.46 <.000128.40.75 Bin11,111LT326
0.33 0.14 0.46 <.000138.817.68 Bin10,310LT319
0.54 0.24 0.78 0.0000 58.7 3.53 Bin10,110LT305
0.29 0.25 0.54 0.0006 15.0 5.46Bin9,19RM444
0.34 0.27 0.61 0.0001 19.2 18.40 Bin8,38LT268 0.77 0.00 0.77 <.000124.0
0.48 0.33 0.80 <.000141.14.49 Bin8,18LT253
–0.340.42 0.08 0.0001 18.6 20.50 Bin6,46LT207
0.53 0.31 0.84 <.000154.1 0.20 0.15 0.35 0.0006 15.0 0.63 Bin6,16LT186
0.52 0.28 0.80 <.000161.7 26.37Bin5,65RM26
0.35 0.29 0.64 <.0001187.5 6.99Bin5,25RM169
0.31 0.23 0.54 <.000121.5–0.380.49 0.11 0.0000 21.2 31.49 Bin4,64LT150
0.45 0.50 0.95 <.000132.421.14 Bin4,54LT140
0.28 0.20 0.49 0.0000 21.2 16.70 Bin3,33LT97
–0.310.42 0.11 0.0001 19.6 9.81Bin3,23RM7
0.33 0.31 0.64 <.000123.4 2.44Bin3,13RM231
0.31 0.10 0.41 <.000147.1 36.06Bin3,53
–0.310.34 0.03 0.0002 16.8 34.94Bin2,62
0.39 0.07 0.46 <.0001146.6 10.07Bin2,22
0.29 0.23 0.53 0.0001 17.7 27.11Bin1,51
1.31 0.15 1.47 <.000164.7 17.89 Bin1,41
2.74 Bin1,11
RM85
RM266
RM29
RM246
Mo18
Mo3
0.02 0.51 0.53 <.000145.50
0.16 0.03 0.19 <.000121.7
QTLs for ST detected in Minghui86/Zaoxian14 and Minghui86/Y134 ILs
Diff.Random
pop.ST-ILs
P X2
Frequency of
introgressionP X2
Minghui8686/Y134 (10)Minghui86/Zaoxian14 (9)
Physical Physical Physical Physical position position position position
/Mb
BinChr.Chr.Chr.Chr.MarkerMarkerMarkerMarkerFrequency of
introgression
ST-ILs Random
pop.Diff.
19.71Bin12,412
20.52Bin10,310
18.63Bin9,39
0.59Bin9,19
29.26Bin7,77
12.32Bin7,37
–0.801.08 0.28 0.0006 14.76 3.40 Bin6,16
0.51 0.03 0.54 <.0001110.28 26.91 Bin5,65
1.31 0.61 1.92 0.0017 12.70 15.55 Bin5,45
6.99Bin5,25
0.29 0.28 0.56 <.000120.5 2.02Bin4,14
0.31 0.10 0.41 <.000147.1 36.06Bin3,53
RM519
RM147
RM189
RM296
RM248
Mo233
Mo192
Mo185
Mo173
RM169
RM518
RM85
–0.460.67 0.21 <.000136.42
–0.020.36 0.34 <.000127.06
0.15 0.03 0.18 <.000128.90
0.22 0.11 0.33 0.0004 15.47
–0.360.51 0.15 <.000120.19
0.150.030.18<.000121.67
0.20 0.03 0.22 <.000133.3
0.18 0.04 0.21 <.000124.18
ST-QTLs detected in at least the two different
ST-IL populations
Gayabyeo Shennong265 Zaoxian14 Y134
Bin2.2 √ √ √ √
Bin1.1 √ √ √
Bin6.1 √ √ √
Bin2.6 √ √
Bin4.6 √ √
Bin5.2 √ √Bin5.2 √ √
Bin5.4 √ √
Bin5.6 √ √
Bin8.3 √ √
Bin9.1 √ √
Bin10.3 √ √
Based on phenotypic value and QTL allele distribution, we can easily
select ideal ILs to pyramid different alleles from different donors to
improve the target traits.
A case study of high yield (HY), drought
and salinity tolerance (DT, ST)
MAS-based pyramiding of QTLs
and salinity tolerance (DT, ST)
using the selected ILs
Development of HY-, DT- and
ST-ILs for QTL mapping
SN89366 Bg94-1 GH122 YJ7 JXSM
Feng-Ai-Zhan 1 (FAZ1) Backcross & selfing
with HY selection
Pyramiding of QTLs
for HY, DT and ST
IL1 IL2×××× IL3 IL4×××× IL5 IL6×××× IL7 IL8××××
For DT For ST
F1 F1 F1 F1
F2 populations
Pop. 1 Pop. 2 Pop. 3 Pop. 4 Pop. 5BC3F5
HY & DT ILs HY & ST ILs
DT screening ST screening
HY &
DT ILs
FAZ1/SN89366 (IL1)
FAZ1/Bg94-1 (IL2)
FAZ1/GH122 (IL3)
FAZ1/YJ7 (IL4)
FAZ1/SN89366 (IL5)
FAZ1/Bg94-1 (IL6)
FAZ1/JXSM (IL7)
FAZ1/BG94-1 (IL8)
HY &
ST ILs
60 random
plants
~30 HY
plants
~30 DT
plants
~30 ST
plants
Confirmed or cross-testing of
selected ILs for QTL mapping
New breeding lines with HY, DT and/or ST
Promising lines for RYT
QTL mapping QTL mapping
QTLs affecting high yield (HY), drought tolerance (DT) and salinity tolerance (ST)
detected in two pyramiding populations by frequency distortion of genotypes
Pop. Locus Ch. Posi. HY DT ST
X2
P Gene
action
X2
P Gene
action
X2
P Gene
action
IL3/IL4
(DTP2)
F2
RM486 1 153.5 18.75 0 OD 27.34 0 OD 25.87 0 OD
OSR14 2 6.9 7.76 0.0206 PD
RM471 4 53.8 13.46 0.0011 OD
RM584 6 26.2 7.74 0.0208 OD
RM3 6 74.3 7.67 0.0216 AD 13.66 0.001 OD
RM2 7 8.08 0.0175 OD
RM547 8 58.1 19.97 0 OD 27.89 0 OD 30.97 0 ODRM547 8 58.1 19.97 0 OD 27.89 0 OD 30.97 0 OD
RM21 11 85.7 10.78 0.0045 AD
RM4A 12 5.2 11.93 0.0025 OD
IL5/IL6
(STP1)
F2
RM297 1 155.9 10.45 0.0053 AD 6.49 0.0389 AD 9.93 0.0069 AD
RM324 2 66 6.31 0.0426 PD
RM55 3 168.2 6.51 0.0385 PD
RM3 6 74.3 13.44 0.0012 AD 9.48 0.0087 AD 7.7 0.0212 AD
RM444 9 3.3 56.43 0 PD
RM434 9 57.7 30.82 0 AD
RM4A 12 5.2 6.29 0.043 OD
RM519 12 62.6 8.19 0.0166 OD
RM235 12 91.3 12.67 0.0017 PD
RM582 RM57266.4RM31271.6RM2478.4
RM594.9RM488101.4
RM246115.2
RM302147.8RM212148.7RM486153.5RM297155.9
Chr1
RM764.0
RM25179.1
RM411127.9
RM55 RM186168.2
RM227182.1
Chr3
RM33521.5
RM47153.8
Chr4
RM1220.0
RM31118.8
RM87129.2
Chr5
1
1 11
4
2 2 2
3 3 3
2 1 1
1
1 1 3
1 1 1
4
2RM6
OSR14 RM1106.9
RM52158.4RM324 RM42466.0RM29068.0RM26270.2RM34182.7
RM47592.5
RM6154.7
Chr2
1 1
4
3
21 3 4
1 2 3 4
QTLs for HY identified in pyramiding populations
QTLs for DT identified in pyramiding populations
RM213186.4
RM4692.2RM190 RM5887.4RM58710.7RM51020.8RM225 RM584RM225
26.2
RM27640.3
RM374.3
Chr6RM236.0RM43243.5
RM1890.4
RM248116.6
Chr7RM408RM5060.0RM4075.7
RM54758.1
RM22380.5
RM21090.3
RM80103.7
RM447124.6
Chr8RM2960.0RM4443.3
RM56647.7
RM43457.7RM25766.1RM10873.3RM55376.7
Chr9RM2860.0
RM2185.7
RM206102.9
Chr11RM4A5.2
RM51962.6RM31365.5
RM23591.3
RM12 RM17109.1
Chr12
2
1
2 2 3 3 3
4 4
2
1 2 2 2
1
3 4
3
14
1
2
2 3
34
3
1 2 3 4
1 2 3 4
QTLs for DT identified in pyramiding populations
QTLs for ST identified in pyramiding populations
Distributions of QTLs
affecting HY, DT and ST
Selected pop. Intercross or
repeated screening
trait
No. of selected
lines
Line # Yield of introgression line (g) Salt tolerance of introgression line at the seedling stage
Trait value
Check of
higher value parent
±±±±% comp. with check
No. of survival days Score of salt toxicity of leaves
Trait value
Check of higher parent
±±±±% comp check
Trait value
Check of higher parent
±±±±% comp check
DT selected (30)
HY 1 QP49 43.5 30.1 44.8 10 8.8 13.6 4.5 5.5 18.2
ST 10
QP47 31.8 30.1 5.5 11 8.8 20.6 4.5 5.5 18.2
QP48 29.8 30.1 -0.9 11 8.8 22.9 4.5 5.5 18.2
QP63 24.3 30.1 -19.3 12 8.8 36.4 4.5 5.5 18.2
QP60 26.3 30.1 -12.6 12 8.8 31.8 4 5.5 27.3
QP61 28.8 30.1 -4.3 11 8.8 30.3 4 5.5 27.3
Promising pyramiding lines selected from intercross or repeated
screening for HY and ST from IL1x IL2 population
QP36 28 30.1 -7 11 8.8 29.5 4 5.5 27.3
QP37 28.2 30.1 -6.3 11 8.8 29.7 5 5.5 9.1
HY selected (30)
HY 2QP163 38.6 30.1 28.4 9.6 8.8 9.1 5 5.5 9.1
QP167 36.6 30.1 21.8 11.4 8.8 29.5 4 5.5 27.3
ST 7
QP171 35.8 30.1 18.9 10 8.8 17.1 4.5 5.5 18.2
QP169 32.1 30.1 6.7 12 8.8 33 4.5 5.5 18.2
QP168 25.4 30.1 -15.6 13 8.8 51.1 4 5.5 27.3
QP166 28.3 30.1 -6 11 8.8 29.1 4 5.5 27.3
QP164 23 30.1 -23.4 11 8.8 25.7 4 5.5 27.3
QP170 17.4 30.1 -42.2 11 8.8 25.1 4.5 5.5 18.2
QP165 24.5 30.1 -18.7 11 8.8 20.6 4 5.5 27.3
ST selected (33) HY 2QP327 36.6 30.1 21.6 NA NA NA NA NA NA
QP337 34.9 30.1 15.9 NA NA NA NA NA NA
Based on phenotypic and QTL information of trait-specific ILs, a new line with
HY, DT and ST was developed by pyramiding of different target QTLs
Zhong-Guang-Lv 1((((HY, DT & ST))))
RYT in Yunnan province in 2011
Zhong-Guang-You 2
RYT in Guangxi province in 2010-11
Molecular recurrent selection systems for improving
multiple complex traits based on trait-specific
ILs and dominant male sterile (DMS) line
Developments of MAS-based improvement strategies required for
multiple traits should include understanding the correlation between
different traits
◆◆◆◆ Interaction between components of a very complex trait such as
drought tolerance
◆◆◆◆ Genetic dissection of the developmental correlation
◆◆◆◆ Understanding of genetic networks
◆◆◆◆ Construction of selection indices across multiple traits.
Selection for multiple traits
◆◆◆◆ Construction of selection indices across multiple traits.
The methods for pyramiding genes affecting a specific trait can be used
to accumulate QTL alleles controlling different traits. A distinct
difference in concept is that alleles at different trait loci to be
accumulated may have different favorable directions, i.e. negative alleles
are favorable for some traits but positive alleles are favorable for others.
Therefore, we may need to combine the positive QTL alleles of some
traits with the negative alleles of others to meet breeding objectives.
Jiafuzhan (Rr, sterile)
Jiafuzhan (rr, fertile)
Development of a DMS line in HHZ background
Spontaneous mutation
x Jiafuzhan (rr, fertile)
Jiafuzhan (1Rr sterile : 1rr fertile)
x HHZ (rr)
F1 (1Rr sterile : 1rr fertile)
x HHZ (rr), backcross 4-5 times
HHZ (1Rr sterile : 1rr fertile)Anthers with different fertility
A: full sterile anther
B: full fertile anther
C,D: partial fertile anther
Composition of the molecular RS (MRS) populations:
30-50 ILs/PLs carrying favorable QTL alleles from different
donors plus the DMS line in the same genetic backgrounds (HHZ)
Ovals or boxes of different colors represent different ILs carrying genes/QTLs
MRS population in HHZ GB
Bulk harvest
seeds from
fertile plants
to be screened carrying genes/QTLs for different target traits
Development of RS population is still under the way
to be screened
for target traits
Bulk harvest
seeds from
sterile plants
for next round
of RS
HHZ MS
line
Each fertile individual has even chance to pollinate with DMS plants,
ensuring all possible recombination produced inside the RS population
50% fertile plants
RS populations based on trait-specific ILs and a DMS line in the same GB
Irrigated(YP)
Abiotic stresses
Biotic stresses
Trait-improved
New ILs/PLs
Continued introgression breeding/DQP
50% DMS plants
Combine DMS line-based RS system with whole genome selection
RILs
GS model
Trait screening
New MRS
population for
next round
New lines with multiple traits by pyramiding
Trait-improved lines
RYT and NCT under different
target Es
Farmers in dif. target Es
Continuation
of MRS
GS model
GS
GS
Precise and high-throughput phenotyping
High-throughput and precision phenotyping is critical for genetic
analysis of traits using molecular markers, and for time- and cost-
effective implementation of MAS in breeding. To match up with
the capacity and costefficiency of currently available genotyping
systems, a precision phenotyping system needs high-throughput
data generation, collection, processing, analysis, and delivery.
High Resolution Plant Phenomics The Plant Accelerator
The High Resolution Plant Phenomics Centre (HRPPC)
Phenomics technology in the field
Phenomobile
Designed: to straddle a plot and collect measurements of canopy temperature, crop stress indices, crop chemometrics, canopy volume, biomass and crop ground cover
Phenotower
From 16 meters above the crop canopy. Phenotower collects infra-red thermography and colour imagery of field plots.This data is used for spatial comparison of canopy temperature, leaf greenness and groundcover between genotypes at a single point in time.
Measurements include:
◆◆◆◆ Leaf size
◆◆◆◆ Number of leaves
◆◆◆◆ Shape
◆◆◆◆ Topology (study of constant properties)
◆◆◆◆ Surface orientation
◆◆◆◆ Leaf color
◆◆◆◆ Plant area and volume
Plant scan
Tethered blimp
The blimp will carry both infrared
and digital color cameras operating
in a height range of 10 m to 80 m
above the field.
It will identify the relative
differences in canopy temperature
indicating plant water use.
Remote Sensing techniques
A flowchart for whole-genome strategies in marker-assisted plant breeding. The system starts with natural and artificial crop populations to develop novel germplasm through four key platforms, genotyping, phenotyping, e-typing (environmental assay), and breeding informatics, which need decision support system in various steps towards product development.
Discussion
Thank You for
Your Attention!