Parallel domestication of the Shattering1 genes in cereals.

Abstract

A key step during crop domestication is the loss of seed shattering. Here, we show that seed shattering in sorghum is controlled by a single gene, Shattering1 (Sh1), which encodes a YABBY transcription factor. Domesticated sorghums harbor three different mutations at the Sh1 locus. Variants at regulatory sites in the promoter and intronic regions lead to a low level of expression, a 2.2-kb deletion causes a truncated transcript that lacks exons 2 and 3, and a GT-to-GG splice-site variant in the intron 4 results in removal of the exon 4. The distributions of these non-shattering haplotypes among sorghum landraces suggest three independent origins. The function of the rice ortholog (OsSh1) was subsequently validated with a shattering-resistant mutant, and two maize orthologs (ZmSh1-1 and ZmSh1-5.1+ZmSh1-5.2) were verified with a large mapping population. Our results indicate that Sh1 genes for seed shattering were under parallel selection during sorghum, rice and maize domestication.

Full text

Page 1

720?
VOLUME 44 | NUMBER 6 | JUNE 2012 Nature GeNetics
A?key?step?during?crop?domestication?is?the?loss?of?seed?
shattering.?Here,?we?show?that?seed?shattering?in?sorghum?
is?controlled?by?a?single?gene,?Shattering1 (Sh1),?which?
encodes?a?YABBY?transcription?factor.?Domesticated?sorghums?
harbor?three?different?mutations?at?the?Sh1?locus.?Variants?at?
regulatory?sites?in?the?promoter?and?intronic?regions?lead?to?a?
low?level?of?expression,?a?2.2-kb?deletion?causes?a?truncated?
transcript?that?lacks?exons?2?and?3,?and?a?GT-to-GG?splice-
site?variant?in?the?intron?4?results?in?removal?of?the?exon?4.?
The?distributions?of?these?non-shattering?haplotypes?among?
sorghum?landraces?suggest?three?independent?origins.?The?
function?of?the?rice?ortholog?(OsSh1)?was?subsequently?
validated?with?a?shattering-resistant?mutant,?and?two?maize?
orthologs?(ZmSh1-1?and?ZmSh1-5.1+ZmSh1-5.2)?were?verified?
with?a?large?mapping?population.?Our?results?indicate?that??
Sh1?genes?for?seed?shattering?were?under?parallel?selection?
during?sorghum,?rice?and?maize?domestication.
Cereal crops, the primary calorie source for humans, were domesti­
cated thousands of years ago1. During domestication, many morpho­
logical and physiological characteristics of the wild progenitors of
modern crops were reshaped to meet the needs of humans by artificial
selection. When ancient humans started to cultivate wild crops, one of
the most notable obstacles would have been the seed shattering habit.
Seeds on wild grasses shed naturally at maturity, ensuring their natural
propagation. Seed shattering, however, would have caused inefficient
harvesting and large losses in grain yield for ancient humans. Hence,
the non­shattering trait is likely to been placed under strong selection
early in domestication. Because the trait can be achieved by changes
in one or two major genetic loci with large effects2–4, non­shattering
variants could have appeared in the population at discernable fre­
quencies, leading to the fixation of the non­shattering variants in
ancient domesticated crop populations. Selection for non­shattering
crop plants would have greatly facilitated harvesting and improved
production, and propagation of cereal crops would have become
increasingly dependent on humans, a feature that distinguishes mod­
ern crops from their wild progenitors. Although several genes have
been identified as being responsible for seed shattering in rice and
wheat5–8, whether other cereals share the same molecular genetic
basis for shattering had not been determined.
Sorghum is the world’s fifth major crop and a new model plant with
applications in bioenergy and stress management9. Previous genetic
studies have shown that seed shattering in sorghum is governed by
a single locus10,11. To identify the molecular basis underlying seed
shattering in sorghum, we constructed an F2 population from a cross
between a wild sorghum with complete seed shattering, Sorghum
virgatum (SV), and a non­shattering domesticated sorghum line,
Tx430 (Fig. 1). The F1 plants showed the same complete shattering
as SV. The F2 segregation ratio suggested that a single gene with a
complete dominance effect explained this trait, and this gene was
designated Shattering1 (Sh1).
An initial scan with 94 F2 individuals mapped the Sh1 gene onto
sorghum chromosome 1 (Fig. 2a and Online Methods). This mapping
Parallel domestication of the Shattering1 genes in cereals
Zhongwei Lin1, Xianran Li1, Laura M Shannon2, Cheng-Ting Yeh3,4, Ming L Wang5, Guihua Bai1,6, Zhao Peng7,
Jiarui Li7, Harold N Trick7, Thomas E Clemente8, John Doebley2, Patrick S Schnable3,4, Mitchell R Tuinstra9,
Tesfaye T Tesso1, Frank White7 & Jianming Yu1
1Department of Agronomy, Kansas State University, Manhattan, Kansas, USA. 2Department of Genetics, University of Wisconsin, Madison, Wisconsin, USA.
3Center for Plant Genomics, Iowa State University, Ames, Iowa, USA. 4Department of Agronomy, Iowa State University, Ames, Iowa, USA. 5US Department of
Agriculture–Agricultural Research Service (USDA-ARS), Griffin, Georgia, USA. 6USDA-ARS, Manhattan, Kansas, USA. 7Department of Plant Pathology, Kansas State
University, Manhattan, Kansas, USA. 8Center for Plant Science Innovation, University of Nebraska, Lincoln, Nebraska, USA. 9Department of Agronomy, Purdue
University, West Lafayette, Indiana, USA. Correspondence should be addressed to J.Y. (jyu@ksu.edu).
Received 18 January; accepted 19 April; published online 13 May 2012; doi:10.1038/ng.2281
Figure 1 Seed shattering phenotype in sorghum. (a,b) Seeds were
scattered everywhere from the top of the wild sorghum SV plant (a),
whereas seeds were firmly retained on the head of the domesticated
sorghum Tx430 plant (b, shown only from a panicle branch) at maturity
after vigorous shaking. (c,d) Larger views of spikelets in a and b are
shown for SV (c) and Tx430 (d) plants after shaking. (e,f) Abscission
layers (of curved shape) were present at the junction between the hull and
pedicel on SV plants (e), whereas no abscission layer was observed on
Tx430 plants (f). AL, abscission layer; scale bars, 50 µm.
ab
SVTx430
cd
AL
Hull
Hull
Pedicel
Pedicel
fe
letters
npg
© 2012 Nature America, Inc. All rights reserved.

Page 2

Nature GeNetics? VOLUME 44 | NUMBER 6 | JUNE 2012
721
result was subsequently verified by linkage analysis with a simple
sequence repeat (SSR) marker (Xtxp302) across 286 F2 plants
(Fig. 2b). On the basis of the sorghum genome sequence9, four new
SSR markers (P1, P2, P3 and P4) were developed, and Sh1 was localized
within a 2.5­cM region (Fig. 2b). With three additional SSR makers
(P5, P6 and P7) and one SNP1, we fine­mapped Sh1 to a region
between P6 and SNP1, using 15,000 F2 plants
(Fig. 2c). A BAC clone (25K18) from a wild
sorghum with the shattering habit, Sorghum
propinquum (Propinquum), was found to
cover the Sh1 candidate region. Sequence
analysis of this BAC clone eventually placed
Sh1 within a fragment of approximately 17 kb
in size between P6 and SNP1 on sorghum
chromosome 1 (Fig. 2d). Sequence anno­
tation of this fragment revealed only two
37
a
ATG
Phenotype
SH
SV
–1,619
to –1,615 –1,194 –1,185152 3,985 4,881 5,076 6,251 5,449 6,608 8,122 to 8,825
P890003
SC265
SC35
Tx430
Tx623
Segaolane
Macia
SC283
28
26
14
12
10
b
c
Ajabsido
SC557
SC1345
Propinquum
SH
SH
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
Accession
25
Number
of samples
2.2-Kb indel
TAA
37
47
–log10 (P value)
8
6
4
2
SV
Tx430
SC265
1
1
1
1
61
61
61
55
60
60
60
55
120
120
120
55
180
180
125
55
121
121
121
55
181
181
125
55
191
191
125
55
Tx430
SC265
Tx430
SC265
Tx430
SC265
Tx623
Tx623
Tx623
Tx623
SV
Sv
Sv
Sh1
PropinquumSV Tx430SC265Tx623
d
Actin
Figure 3 Variant alleles and association
mapping at Sh1. (a) Gene structure and
haplotype analysis of Sh1. —SV-like, SC265-
like, Tx430-like and Tx623-like haplotypes were
identified on the basis of ten variants of the
Sh1 gene. The positions of these ten variants
are shown using the Sh1 gene sequence of SV
as reference, with the start codon designated as
position 0. The SV-like haplotype is highlighted
in yellow. Splicing mutation from GT to GG
at 6,608 in the SC265-like haplotype (blue),
four specific variants at −1,194, −1,185,
4,881 and 5,076 in the Tx430-like haplotype
(red) and a 2.2-kb deletion from 3,985 to
6,251 in the Tx623-like haplotype (green) are
indicated. Arrow bar, promoter region; thick bar,
downstream region after the stop codon; blank
box, exon; thin line, intron; ATG, start codon;
TAA, stop codon; adjacent unfilled boxes on a
dashed line, 2.2-kb deletion; SH, shattering;
NS, non-shattering. (b) Association testing
at sites of ten variants and two fine-mapping
markers P6 and SNP1. Black dots, ten variants
of Sh1; blue dots, P6 (left) and SNP1 (right);
red dot, supposed synthetic association site;
red dashed line, 5% significance threshold.
(c) Amino-acid sequence alignment for different
haplotypes. Two domains, zinc finger (blue) and
YABBY (red), are indicated. (d) There was strong
transcription of Sh1 in Propinquum and SV,
whereas there was weak transcription in
Tx430 and truncated transcripts for Tx623
and SC265.
6
4
2
–log10 (P value)
0
1
a
2
3
4
5
Chromosome
6
7
8
9
10
Xtxp302 P1P2P3 P4
P5
25K18
d
c
b
125F01
15G07
F2 plant
Phenotype of F3 plants
Segregated
SNP1
P6
P7
23
5 kb
351
Segregated
1
Sh1
1 15
R =
4.1 cM3.9 cM
2.5 cM
15.0 cM
Chr. 1
Chr. 1
236
Figure 2 Map-based cloning of Sh1 in sorghum. (a) DNA chip
screening across 94 F2 plants mapped the Sh1 gene onto sorghum
chromosome 1. Significant SNPs are marked as red dots; the red
dashed line represents the 5% significance threshold with
Bonferroni correction for 90 tests. (b) Genetic mapping of Sh1
with 286 F2 plants. Genetic distance between flanking pairwise
molecular markers is shown. (c) High-resolution mapping of Sh1 with 15,000 F2 plants. The genotypes of two F2 recombinant plants are shown; the F3
progenies of both F2 recombinant plants are segregated by phenotype. Gray bar, heterozygous region of Tx430; white bar, homozygous region of Tx430;
R, recombinant plant. (d) The Propinquum BAC clone 25K18 was identified by the two flanking markers of Sh1, and two genes were predicted within
the candidate region between these markers. Gray arrow, a hypothetical gene specifically expressed in pollen; red arrow, YABBY-like gene.
letters
npg
© 2012 Nature America, Inc. All rights reserved.

Page 3

722?
VOLUME 44 | NUMBER 6 | JUNE 2012 Nature GeNetics
letters
predicted genes: a hypothetical gene and a transcription factor gene
belonging to the YABBY family. The sequence of the hypothetical gene
from a shattering F2 recombinant plant, 15G07, was identical to that
of the non­shattering Tx430 parent (Fig. 2d), indicating that recom­
bination occurred between these two genes; therefore, the YABBY­like
gene was regarded as the functional candidate of Sh1 in sorghum.
We compared the 7,758­nt YABBY­like gene region from the start
codon to the stop codon in SV and Tx430 plants (Fig. 3a). No nucleo­
tide differences were detected in 6 exons, whereas 26 SNPs were
present in 5 introns. We then sequenced the YABBY­like gene from
the 2,761­nt upstream promoter region to the 1,315­nt downstream
fragment after the stop codon across 13 sorghum accessions. This set
of accessions comprised three wild shattering accessions and ten non­
shattering domesticated lines (Fig. 3a and Online Methods). Four
haplotypes emerged from the 10 representative nucleotide variants
across these 13 sorghum accessions. Three shattering wild accessions
had an SV­like haplotype; two non­shattering accessions had a SC265­
like haplotype harboring a GT­to­GG splice­site variant at nucleotide
position 6,608; four non­shattering accessions shared a Tx430­like
haplotype with two promoter variants at positions −1,194 and −1,185
and two intronic variants at positions 4,881 and 5,076; and four non­
shattering accessions had a Tx623­like haplotype with a 2.2­kb dele­
tion from 3,985 to 6,251 at the location of exons 2 and 3 (Fig. 3a,
green). We also sequenced 146 sorghum accessions from around the
world (Online Methods and Supplementary Tables 1 and 2). Of note,
the SV­like haplotype was conserved across all 25 shattering acces­
sions, and the three other haplotypes were retained in almost all non­
shattering accessions, except for in two that had rare recombination
events (Fig. 3a and Supplementary Tables 1 and 2).
We next performed an association test across 25 shattering and 121
non­shattering accessions. Significant associations at all ten repre­
sentative sites (P values = 6.0 × 10−4 to 1.6 × 10−11) were obtained.
Strong signals were observed at positions −1,619, 152, 5,449 and 8,122,
whereas medium­strength signals were detected at the locations of
the multiple causal variants, including the four specific mutations of the
Tx430­like haplotype, the 2.2­kb insertion and deletion (indel) and the
splice­site variant (Fig. 3b). The distributions of the common variants
at positions −1,619, 152, 5,449 and 8,122 (with an allele frequency of
83/146 for positions –1,619 and 8,122, or 84/146 for positions 152 and
5,449) were correlated with those of the presumptive causal variants
in the Tx430­like (37/146) and Tx623­like (47/146) haplotypes, which
occurred with lower frequencies. Because the common variants were
in linkage disequilibrium (LD) with the presumptive causal variants
but had higher frequencies (Supplementary Fig. 1), their synthetic
association signals were stronger than those of the causal variants
(Fig. 3b and Supplementary Table 2). The association signals at posi­
tions 4,881, 5,076 and 5,449 were strong because these sites contained
causal alleles in the Tx430­like and Tx623­like haplotypes. When
we tested three domesticated haplotypes as a group against the wild
haplotype, the YABBY­like gene was completely associated with the
shattering trait (P value = 1.1 × 10−28) (Fig. 3b, red dot).
The 576­bp coding sequence of the Sh1 gene encodes a YABBY
protein consisting of 191 amino acids. The zinc finger domain is
located from amino acids 11 to 52, and the YABBY domain is located
from amino acids 111 to 165 (Fig. 3c).
To identify the causal polymorphisms, we conducted expres­
sion analysis via RT­PCR, focusing on the entire coding transcript.
Compared with the wild shattering lines (SV and Propinquum), the
Tx430 line showed a low level of Sh1 transcription, whereas truncated
transcripts were found in Tx623 and SC265 (Fig. 3d). Sequence ana­
lysis of the transcripts of SV, Tx430, Tx623 and SC265 further showed
that Tx430 shared the same 576­bp coding region as SV, whereas
Tx623 encoded a 317­bp transcript without exons 2 and 3 and SC265
encoded a 527­bp transcript missing exon 4 (Supplementary Fig. 1).
Both truncated transcripts contained frameshifts that resulted in
the introduction of premature stop codons. The Tx623­like haplo­
type therefore encodes a protein lacking the zinc finger and YABBY
domains, whereas the SC265­like haplotype encodes a protein lacking
only the YABBY domain (Fig. 3c).
Microscopic examination revealed that abscission layers began to
form in the joint connecting the seed hull and pedicel in SV plants
during flowering, whereas no abscission layer was formed in Tx430
plants (Fig. 1e,f). These results indicate that the two promoter and two
intronic variants in the Tx430­like haplotype repressed the expression
of the Sh1 gene, whereas the 2.2­kb deletion in the Tx623­like hap­
lotype and the splice­site variation from GT to GG in the SC265­like
haplotype altered the Sh1­encoded protein, eliminating formation
of the abscission layer in the joint between the hull and pedicel and
thereby resulting in a loss of shattering.
Sorghum is known for its genetic diversity, and multiple domes­
tication events have been suggested on the basis of morphology and
0
a
b
4
8
12
16
S. bicolor chr. 1
20
24
28
32
83-bp insertion
22.279 kb
34.684 kb
1 kb
22.8 kb
7.758 kb
9.196 kb
7.514 kb
QTL interval and
peak
ExonIntron 19.3-kb intron
Sh1 gene
OsSh1 flanking markers from two different
QTL mapping studies
Unknown gene fragment
>4-kb insertion
SvSh1
ZmSh1-5.2
ZmSh1-1
Z. mays
chr. 1
Z. mays
chr. 5
O. sativa
chr. 3
S. italica
chr. IX
ZmSh1-5.1
OsSh1
SiSh1
300 287 274261
248
235
2220
13263939 2613
00
13
26
Figure 4 Genomic regions of Sh1 in cereals. (a) Genomic regions
corresponding to Sh1 were conserved in sorghum (Sorghum bicolor),
maize (Zea mays), rice (Oryza sativa) and foxtail millet (Setaria italica).
The genomic collinearity map was plotted on the basis of the BLASTP
result of pairwise genome analysis from CoGe; dot plot alignment
indicates the collinearity of genomic regions. (b) Sh1 gene structure
comparison. Sh1 gene structure is conserved, except for one extremely
large intron (19.3 kb) that was present only in the Sh1 ortholog on
maize chromosome 1 (ZmSh1-1) and a gene fusion that occurred in
one of two Sh1 orthologs on maize chromosome 5 (ZmSh1-5.1).
npg
© 2012 Nature America, Inc. All rights reserved.

Page 4

Nature GeNetics? VOLUME 44 | NUMBER 6 | JUNE 2012
723
letters
molecular marker analyses12,13. In this study, we found that the shatter­
ing haplotype remains conserved within 22 wild shattering accessions
and 3 shattering Sorghum bicolor strains from different regions of the
world (the shattering habit may have been introgressed into these
3 bicolor sorghum strains from wild sorghum), but 3 distinct haplo­
types were present among 121 non­shattering domesticated sorghums.
Seed shattering, vital for the propagation of wild crop progenitors in
nature, should be under very strong, direct natural selection, which
could explain the conserved shattering haplotype. Three ancient
human groups may have cultivated wild sorghum progenitors success­
fully by independently selecting and fixing three different sh1 muta­
tions that resulted in non­shattering in different original populations,
overpowering natural selection. Indeed, among 80 non­shattering
landraces, the Tx430­like haplotype group is dominated by acces­
sions from caudatum (Supplementary Table 2), a race thought to be
domesticated later than other races. The Tx623­like haplotype group
primarily contains accessions from the kafir and biocolor races from
South and East Africa. All accessions from durra, most from guinea
and almost half of those from bicolor have the SC265­like haplotype.
The lack of a dominant sh1 haplotype among accessions from bicolor
agrees with the wide distribution of this race.
Cereal crops such as rice, maize, wheat, barley and sorghum were
domesticated thousands of years ago. Although these crops were
domesticated from different wild progenitors by different ancient
human groups in different geographical zones, they all underwent
systemic and parallel changes during the domestication process14.
Whether these parallel changes in domestication share the same
genetic basis is still vigorously debated. Although the two major genes
for shattering in rice (Qsh1 and Sh4) and the Q gene for shattering in
wheat have not been found to be under selection in other crops, the
newly identified Sh1 gene in sorghum provides another entry point
to test this hypothesis. A shattering quantitative trait locus (QTL)
with minor genetic effect has been repeatedly mapped to a syntenic
block corresponding to Sh1 on rice chromosome 3 (refs. 15–17), two
disarticulation QTLs were identified in the syntenic blocks on maize
chromosomes 1 and 5 (refs. 11,18), and one of the two major QTLs
for shattering was also mapped within the same block on foxtail millet
chromosome 9 (ref. 19) (Fig. 4a).
The rice QTL for shattering in the Sh1 syntenic region has
its peak 80 kb away from the gene orthologous to Sh1 (OsSh1,
LOC_Os03g44710) (Fig. 4a, blue arrow, and Supplementary Note)
and a non­shattering mutant that has reduced cell numbers at the
abscission layer was available for further anlaysis15–17,20. Using the
non­shattering mutant (SR­5) and the wild­type rice breeding line
(Nanjing 11), we conducted genomic DNA amplification and expres­
sion analysis and Southern blotting (Supplementary Figs. 2–4). An
insertion of a >4­kb fragment was identified in intron 3 of OsSh1,
leading to reduced levels of transcription and the shattering­resistant
phenotype. Furthermore, in two recent whole­genome sequenc­
ing studies, OsSh1 was in a list of genes shown to be under strong
artificial selection21,22.
On maize chromosome 1, the Sh1 orthologous gene in the reference
genome of B73 (ZmSh1-1)23 contains an extremely large (19.3­kb)
intron 1 (Fig. 4b and Supplementary Note), similar to the key domestica­
tion gene for fruit size (fasciated) in tomato24. On maize chromosome 5,
the B73 genome contains two copies of the Sh1 orthologous gene
(ZmSh1-5.1+ZmSh1-5.2) within the syntenic block (Fig. 4b). The struc­
tural change in ZmSh1-5.1+ZmSh1-5.2 is present in the majority of maize
inbreds but absent in all teosinte inbreds, and the insertion in ZmSh1-1 is
present in almost all maize inbreds and a few teosinte inbreds, including
three with incomplete shattering (Fig. 5 and Supplementary Table 3).
For two maize inbreds that did not contain the B73­type insertion in
ZmSh1-1, sequence alignment revealed another 83­bp insertion in exon 3
of ZmSh1-1 on chromosome 1 causing a frameshift.
To further investigate ZmSh1 family genes, we conducted a whole­
genome linkage scan with a maize­teosinte population. A major QTL
for shattering with a very narrow confidence interval (genetic distance
of 1.4 cM, physical distance of 0.6 Mb) was identified on chromosome 5,
and another QTL for shattering was identified on chromosome 1
(2.1 cM, 2.4 Mb) (Fig. 5). These two QTL intervals correspond to the
genomic regions harboring the ZmSh1-1 and ZmSh1-5.1+ZmSh1-5.2
genes. The YABBY­like ZmSh1-1 gene is 1 of the 2 genes encod­
ing transcription factors among the 59 annotated genes within the
chromosome 1 interval, and the ZmSh1-5.1+ZmSh1-5.2 locus is the
only transcription factor gene among the 12 annotated genes within
the chromosome 5 interval (Supplementary Tables 4 and 5).
In summary, the identification of the Sh1 gene in sorghum, the con­
served collinearity of genomic regions containing the Sh1 orthologs
across several cereals, the identification of the rice ortholog OsSh1 and
the structural variation and QTL analyses of the two maize orthologs
(ZmSh1-1 and ZmSh1-5.1+ZmSh1-5.2) suggest that the Sh1 genes for
seed shattering have undergone parallel selection during domestica­
tion in multiple cereals.
URLs. CodonCode Aligner, http://www.codoncode.com/; GRIN, http://
www.ars­grin.gov/; CoGe, http://genomevolution.org/CoGe/; Maize
HapMap V2, http://www.panzea.org/.
10
a
b
50
Chr. 1
259.2 261.6 Mb
22.279 kb
Maize
Teosinte
ZmSh1-1ZmSh1-5.1ZmSh1-5.2
LOD
LOD
30
10
Chr. 5
15.8 16.4 Mb
Number
of lines
14
6
2
1
1
1
34.684 kb
22.8 kb
9
7
Figure 5 Maize Sh1 orthologs are located at seed shattering QTLs.
(a) Two QTLs explaining 3.5% and 23.1% of the phenotypic variation of
shattering were detected on maize chromosomes 1 and 5, respectively,
using a large mapping population. Physical positions of the QTL
confidence intervals and the maize Sh1 orthologs are indicated below
the chromosome axes. LOD, logarithm of odds. (b) Assembly of structural
variation for the Sh1 orthologs on maize chromosomes 1 and 5. ZmSh1-5.1
consists of a gene fusion that retains only the first three exons of the
YABBY-like gene and two exons from an unknown gene. The replacement
of exons 4–6 in ZmSh1-5.1 causes loss of the YABBY domain. A large
(23-kb) insertion is present between ZmSh1-5.1 and ZmSh1-5.2. Solid/
dashed red triangle, presence/absence of large insertion; solid/dashed red
arrow, presence/absence of 83-bp insertion in exon 3 of ZmSh1-1.
npg
© 2012 Nature America, Inc. All rights reserved.

Page 5

724?
VOLUME 44 | NUMBER 6 | JUNE 2012 Nature GeNetics
MeTHODs
Methods and any associated references are available in the online
version of the paper.
Note: Supplementary information is available in the online version of the paper.
ACkNoWLEDGMENTS
This work was supported by an Agriculture and Food Research Initiative
Competitive Grant (2011­03587 to J.Y. and T.T.T.) from the US Department of
Agriculture (USDA) National Institute of Food and Agriculture, the Plant Genome
Research Program of the National Science Foundation (DBI­0820610 to J.Y.,
DBI­0820619 to J.D. and DBI­1027527 to P.S.S.), the Plant Feedstock Genomics
Program of the US Department of Energy (DE­SC0002259 to J.Y.), the USDA­ARS
(M.L.W. and G.B.), the Targeted Excellence Program of Kansas State University and
the Kansas State University Center for Sorghum Improvement.
AUTHoR CoNTRIBUTIoNS
J.Y., F.W., T.T.T. and M.R.T. designed research. Z.L., X.L., L.M.S., C.­T.Y., Z.P.
and J.L. carried out the research. M.L.W., G.B., H.N.T., T.E.C., J.D. and P.S.S.
contributed new reagents. Z.L., X.L., L.M.S. and C.­T.Y. analyzed data. Z.L. and J.Y.
wrote the manuscript with input from all other coauthors.
CoMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Published online at http://www.nature.com/doifinder/10.1038/ng.2281.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html.
1. Doebley, J.F., Gaut, B.S. & Smith, B.D. The molecular genetics of crop domestication.
Cell 127, 1309–1321 (2006).
2. Li, C., Zhou, A. & Sang, T. Genetic analysis of rice domestication syndrome with
the wild annual species, Oryza nivara. New Phytol. 170, 185–193 (2006).
3. Matsui, K. et al. Identification of AFLP makers linked to non-seed shattering locus
(sht1) in buckwheat and conversion to STS markers for marker-assisted selection.
Genome 47, 469–474 (2004).
4. Nalam, V.J., Vales, M.I., Watson, C.J., Kianian, S.F. & Riera-Lizarazu, O. Map-based
analysis of genes affecting the brittle rachis character in tetraploid wheat
(Triticum turgidum L.). Theor. Appl. Genet. 112, 373–381 (2006).
5. Li, C., Zhou, A. & Sang, T. Rice domestication by reducing shattering. Science
311, 1936–1939 (2006).
6. Lin, Z. et al. Origin of seed shattering in rice (Oryza sativa L.). Planta 226, 11–20
(2007).
7. Konishi, S. et al. An SNP caused loss of seed shattering during rice domestication.
Science 312, 1392–1396 (2006).
8. Simons, K.J. et al. Molecular characterization of the major wheat domestication
gene Q. Genetics 172, 547–555 (2006).
9. Paterson, A.H. et al. The Sorghum bicolor genome and the diversification of grasses.
Nature 457, 551–556 (2009).
10. Harlan, J.R., de Wet, J.M.J. & Price, E.G. Comparative evolution of cereals.
Evolution 27, 311–325 (1973).
11. Paterson, A.H. et al. Convergent domestication of cereal crops by independent
mutations at corresponding genetic
(1995).
12. de Wet, J.M.J. & Huckabay, J.P. The origin of Sorghum bicolor. II. Distribution and
domestication. Evolution 21, 787–802 (1967).
13. Cui, Y.X., Xu, G.W., Magill, C.W., Schertz, K.F. & Hart, G.E. RFLP-based assay of
Sorghum bicolor (L.) Moench genetic diversity. Theor. Appl. Genet. 90, 787–796
(1995).
14. Harlan, J.R. Crop and Man. (American Society of Agronomy, Madison, Wisconsin,
1992).
15. Gu, X.Y., Kianian, S.F., Hareland, G.A., Hoffer, B.L. & Foley, M.E. Genetic analysis
of adaptive syndromes interrelated with seed dormancy in weedy rice (Oryza sativa).
Theor. Appl. Genet. 110, 1108–1118 (2005).
16. Onishi, K., Takagi, K., Kontani, M., Tanaka, T. & Sano, Y. Different patterns of
genealogical relationships found in the two major QTLs causing reduction of seed
shattering during rice domestication. Genome 50, 757–766 (2007).
17. Onishi, K. et al. A QTL cluster for plant architecture and its ecological significance
in Asian wild rice. Breed. Sci. 57, 7–16 (2007).
18. Doebley, J. & Stec, A. Genetic analysis of the morphological differences between
maize and teosinte. Genetics 129, 285–295 (1991).
19. Devos, K.M. & Gale, M.D. Genome relationships: the grass model in current
research. Plant Cell 12, 637–646 (2000).
20. Fukuta, Y. & Yagi, T. Mapping of a shattering resistance gene in a mutant line
SR-5 induced from an indica rice variety, Nan-jing 11. Breed. Sci. 48, 345–348
(1998).
21. He, Z. et al. Two evolutionary histories in the genome of rice: the roles of
domestication genes. PLoS Genet. 7, e1002100 (2011).
22. Xu, X. et al. Resequencing 50 accessions of cultivated and wild rice yields markers
for identifying agronomically important genes. Nat Biotechnol. 30, 105–111
(2011).
23. Schnable, P.S. et al. The B73 maize genome: complexity, diversity, and dynamics.
Science 326, 1112–1115 (2009).
24. Cong, B., Barrero, L.S. & Tanksley, S.D. Regulatory change in YABBY-like
transcription factor led to evolution of extreme fruit size during tomato domestication.
Nat. Genet. 40, 800–804 (2008).
loci. Science
269, 1714–1718
letters
npg
© 2012 Nature America, Inc. All rights reserved.

Page 6

Nature GeNetics
doi:10.1038/ng.2281
ONLINe?MeTHODs
Map-based cloning. A large F2 population with 15,286 plants was generated
from a cross between a wild sorghum virgatum (Sorghum bicolor (L.) Moench
ssp. verticilliflorum (Steud.) race virgatum) and a standard sorghum line Tx430
(Sorghum bicolor (L.) Moench ssp. bicolor). All these plants were planted in
the Kansas State University greenhouse.
First, 286 F2 plants were planted for genetic mapping of seed shattering. The
segregation between 217 shattering and 69 non­shattering F2 plants fit well with
a 3:1 ratio (χ2 = 0.12, P = 0.73), indicating a single gene with a complete domi­
nance effect for seed shattering. Next, 94 out of 286 random F2 plants were geno­
typed with a 384­SNP DNA chip25. SNP­trait relationship was tested for 90 SNPs
with polymorphisms distributed across the genome by Fisher’s exact test because
the shattering trait was scored as a binary trait. The significance threshold was
corrected for multiple testing by Bonferroni correction, α′ ≈ α/n = 0.05/90 =
5.6 × 10−4, where α is the nominal significance threshold and n is the number of
SNPs. A linkage map was constructed with MAPMAKER/EXP 3.0b26.
Subsequently, 15,000 additional F2 plants were grown to the 4­ or 5­leaf stage
in a planting tray with 14 × 7 wells. Rapid genotypic screening was performed
with two SSR markers, P1 and P2. Of the 587 recombinant F2 plants trans­
planted for phenotyping, 38 with recombination events between SSR markers P5
and P7 were self­pollinated to produce F3 families. From each of the 38 families,
20 random plants were used to confirm the corresponding F2 phenotype.
Association mapping. The 11.8­kb DNA fragment containing the Sh1 gene
was amplified from 146 sorghum accessions by PCR with seven pairs of
primers (Supplementary Table 6) using TaKaRa LA Taq (RR002A). All PCR
products were cleaned with the QIAquick PCR Purification kit (Qiagen) and
sequenced by ABI 3730 (Applied Biosystems). The sequences were assembled
with the CodonCode Aligner.
To confirm their phenotype, 25 shattering sorghum accessions obtained
from the Germplasm Resources Information Network (GRIN) were planted
in the greenhouse. These 22 shattering sorghum accessions mainly belong to
wild sorghum from Africa (Supplementary Table 1), whereas the other three
shattering sorghum accessions belong to domesticated sorghum. The shatter­
ing habit of wild sorghum may have introgressed into these three domesticated
sorghum accessions.
Including 41 breeding lines from a S. bicolor panel27 and 80 landraces with
diverse origins, 121 non­shattering domesticated sorghum accessions can be
divided into five domesticated races: bicolor, caudatum, durra, guinea and kafir
(Supplementary Table 2). To represent the diversity of the S. bicolor panel, we
selected eight domesticated sorghum accessions, Tx430, SC35, SC265, Ajabsido,
Macia, SC1103, Segaolane and SC1345, from the parents of sorghum nested asso­
ciation mapping (NAM) populations for sequencing of the whole Sh1 gene.
We used ten representative common variants with allele frequency of >10%
and two most closely flanking markers of the Sh1 gene for association testing
on 146 sorghum accessions, including 25 shattering and 121 non­shattering
accessions, with Fisher’s exact test, as Fisher’s exact test is powerful for qualita­
tive traits like shattering28. LD between two loci, A and B, was calculated as
rprr
p
)
pp
)(
p
i j ij
)
j
n
i
m
=∑
ij
ij
(
ij
i
22
11
2
2
==
−
=
∑
(
( (
A ( (
)
A
(
p
) (
B
))
p
(
A B and
A B
1
−
AB
1 )) (( ))
−
p
ijj
B
where A has m alleles, B has n alleles, p(Ai) is the frequency for Ai, p(Bj) is the
frequency for Bj and p(AiBj) is the frequency for AiBj. A triangle LD matrix
was constructed with PowerMarker29.
Expression analysis. Total RNA was extracted by Qiagen Plant Easy RNA kit
from the junction (1–2 cm) between the hull and pedicel where abscission layers
were located. First­strand cDNA synthesis was performed with SuperScript II
Reverse Transcriptase (18064­022, Invitrogen). Sh1 transcript was amplified, and
Actin transcript was amplified as an internal control (Supplementary Table 6).
Microscopy. Longitudinal sections cut by hand were stained with­ acridine
orange and imaged with a Zeiss Axioplan 2 microscope. Fluorescence images
were acquired with a long­pass 650­nm emission filter (red fluorescence)
under excitation at 550 nm.
Comparative genome analysis. BLASTP of SynMap on CoGe was used to
conduct pairwise genome comparisons. Genome sequence data sources were
selected for BLASTP analysis for the following cereals: sorghum (Tx623,
id331), maize (B73, id333), rice (Nipponbare, id3) and foxtail millet (Yugu1,
id32546). A genomic collinearity map was plotted in R on the basis of the
BLASTP results.
Rice mutant validation. The rice non­shattering mutant SR­5 was induced by
gamma­ray irradiation from rice breeding line Nanjing 11, which belongs to
the indica subspecies with an easy shattering habit. Scanning electron micro­
scope (SEM) analysis of the glume­pedicel junction revealed that the surface of
the glume­pedicel junction was very rough in SR­5 but smooth in Nanjing 11
when seeds were removed (Supplementary Fig. 2). The difference in the sur­
face of the glume­pedicel junction between SR­5 and Nanjing 11 suggested
development of the abscission layer was affected in the SR­5 mutant.
PCR for the YABBY­like Sh1 orthologous gene (LOC_Os03g44710) in the rice
SR­5 mutant and the wild­type Nanjing 11 line resulted in successful amplifica­
tion, except for a region located in intron 3 in SR­5 (Supplementary Fig. 3 and
Supplementary Table 7). Further analysis with Southern blotting showed that a
>4­kb insertion was present in intron 3 of the rice ortholog of Sh1 (OsSh1) in the
SR­5 mutant (Supplementary Fig. 4). This >4­kb insertion greatly decreased
transcription of the rice Sh1 ortholog, which affected the abscission layer, with
seeds being retained on the heads of SR­5 plants (Supplementary Fig. 2a,b).
These results suggested the Sh1 genes were functionally conserved between two
model cereal crops, sorghum and rice.
Maize gene structure analysis and QTL validation. To determine the extent
of structural variations across maize germplasm, we assembled the sequences
of the ZmSh1-1 (22 kb) and ZmSh1-5.1+ZmSh1-5.2 (35 kb) genes across a
diverse set of 27 maize and 17 teosinte inbred lines (Supplementary Table 3),
using short reads from maize HapMap V2 and RNA­seq data (P.S.S., unpub­
lished data) (Supplementary Table 3). Structural variation was scored on
the basis of the coverage and abundance of the uniquely aligned reads in the
corresponding region.
We then conducted a QTL mapping experiment using a large maize­
by­teosinte population with 866 recombinant inbred lines (RILs). A maize
inbred, W22, was crossed to a teosinte (Zea mays ssp. parviglumis, accession
CIMMYT 8759), and the resulting F1 generation was backcrossed to W22
for two generations; RILs were derived by selfing the BC2F1 for three gene­
rations. The 866 RILs were grown in two blocks during the summer of 2009
and in an additional block in the summer of 2010 at the West Madison
Agricultural Research Center in Madison, Wisconsin. Randomized complete­
block design was used with ten plants per plot. Shattering was scored
quantitatively as the number of segments into which the mature ears frac­
tured at harvest. Least squared means were determined for each RIL. QTL
mapping was conducted with 19,838 genotyping­by­sequencing (GBS)
markers of known physical positions in the maize B73 reference genome.
The multiple­interval mapping method implemented in R/qtl30 was used. The
significance threshold (5.87) was determined with 10,000 permutations.
Two major QTLs were identified at the genomic regions harboring ZmSh1-1
and ZmSh1-5.1+ZmSh1-5.2, and two other QTLs with smaller effects were
also identified (Supplementary Tables 6 and 8).
25. Murray, S.C., Rooney, W.L., Hamblin, M.T., Mitchell, S.E. & Kresovich, S. Sweet
sorghum genetic diversity and association mapping for brix and height. The Plant
Genome 2, 48–62 (2009).
26. Lander, E.S. et al. MAPMAKER: an interactive computer package for constructing
primary genetic linkage maps of experimental and natural populations. Genomics
1, 174–181 (1987).
27. Casa, A.M. et al. Community resources and strategies for association mapping in
sorghum. Crop Sci. 48, 30–40 (2008).
28. Balding, D.J. A tutorial on statistical methods for population association studies.
Nat. Rev. Genet. 7, 781–791 (2006).
29. Liu, K. & Muse, S.V. PowerMarker: an integrated analysis environment for genetic
marker analysis. Bioinformatics 21, 2128–2129 (2005).
30. Broman, K.W., Wu, H., Sen, S. & Churchill, G.A. R/qtl: QTL mapping in experimental
crosses. Bioinformatics 19, 889–890 (2003).
npg
© 2012 Nature America, Inc. All rights reserved.