A genome-wide microsatellite polymorphism database for the indica and japonica rice.

Microsatellite (MS) polymorphism is an important source of genetic diversity, providing support for map-based cloning and molecular breeding. We have developed a new database that contains 52 845 polymorphic MS loci between indica and japonica, composed of ample Class II MS markers, and integrated 18 828 MS loci from IRGSP and genetic markers from RGP. Based on genetic marker positions on the rice genome (http://rise.genomics.org.cn/rice2/index.jsp ), we determined the approximate genetic distances of these MS loci and validated 100 randomly selected markers experimentally with 90% success rate. In addition, we recorded polymorphic MS positions in indica cv. 9311 that is the most important paternal parent of the two-line hybrid rice in China. Our database will undoubtedly facilitate the application of MS markers in genetic researches and marker-assisted breeding. The data set is freely available from www.wigs.zju.edu.cn/achievment/polySSR.

Introduction

Microsatellites (MSs), or simple sequence repeats (SSRs), are short consecutive repetitive DNA sequence motifs with 1–6 bp in length. The frequency and distribution of MSs have been studied in detail in genomes of bacteria,[1] animals,[2] and plants.[3] MSs have become a valuable source of molecular markers, widely used in genetic diversity analysis[4] and positional cloning in rice[5,6] because of their ubiquity and polymorphism.[7]

Starting with screening genomic DNA libraries with MS probes,[8,9] many investigators have been involved in generating new MS markers[10-12] and integrating them into the rice genetic map.[13] However, the desire to have more markers is still strong, since the amount and their density over rice genomes are not sufficient for satisfying the need of geneticists and breeders.

It is fortunate that Beijing Genomics Institute[14] (BGI) and the International Rice Genome Sequencing Project[15] have released their rice genomic data that provide comparative basis for a thorough analysis on MS between indica and japonica subspecies. Although such a survey has been done recently for chromosome 4 without eliminating MSs with multiple copies,[16] the report focused on MSs in japonica rice cv. Nipponbare, mainly composed of DNRs, TNRs, and TTRs. In this study, we emphasize our newly discovered MS makers through a comparative analysis on MSs between the indica and japonica genome sequences, and have developed a genome-wide MS polymorphism database that have markers indexed in both rice genomes to facilitate molecular studies and marker-assisted breeding practice.

Materials and methods

Rice genome sequences and MS screening

The genome sequences for indica (Oryza sativa ssp. indica cv. 9311) and japonica (O. sativa ssp. japonica cv. Nipponbare) (Version build4) were downloaded from http://rise.genomics.org.cn/rice2/index.jsp[17] and http://rgp.dna.affrc.go.jp[15] respectively. The 5′-UTR, coding, intronic, 3′-UTR, and intergenic regions were determined based on their original annotations. An MS motif scanning tool, SSRIT[11] written in Perl was downloaded from Cornell University (http://www.gramene.org/) and modified to search for perfect MS repeats in the two rice genome sequences, including mono-nucleotide repeats (MNRs), di-nucleotide repeats (DNRs), tri-nucleotide repeats (TNRs), tetra-nucleotide repeats (TTRs), penta-nucleotide repeats (PNRs), and hexa-nucleotide repeats (HNRs) with numbers of repeat units of 10, 7, 6, 5, 4 and 4, respectively. Some of these MS sequences were combined into one repeat complex if the lengths between two adjacent units were less than 20 bp.

MS Polymorphism and primer design

An MS locus is defined as a tract of genomic sequences that include a perfect repeat motif and two unique 160-bp flanking sequences on each sides of the repeat. These MS candidate sequences were aligned to their source genome sequences with BLASTN (E-value cutoff of 1e-40); those with unique hit were selected as candidate MS loci together with their specific flanking sequences. The selected MS sequences were then used as queries against another genome without filtering low complexity sequences. We wrote a Perl script (Supplementary script: polymorphism.pl) to determine the presence of polymorphisms among MS motifs between indica and japonica. Primers pairs were automatically designed on the basis of unique flanking sequences aided by Primer3[18] with its default parameters. Candidate primer sequences were subsequently filtered through following parameters: length range from 18 to 24 bp, devoid of consecutive tracts of single nucleotide, GC content around 50% with a theoretical Tm approximately 60°C, and perfect ending with G- or C-rich at the 3′ end.

Experimental validation

One hundred primer pairs covering all rice chromosomes and covering all MS types were used for our validation experiment. The DNA templates used in this experiment are O. sativa ssp. indica cv. 9311 and O. sativa ssp. japonica cv. Nipponbare grown in a green house. Genomic DNA was extracted from 1-week-old seedlings, using the CTAB method.[19] The quality of DNA was evaluated electrophoretically with 0.8% agarose gel.

Oligonucleotide primers were made by Shanghai DNA Biotechnologies Co., Ltd. PCR was performed in 25 µl reactions containing 3.0 µl buffer, 2.5 µl MgCl2 (25 mM), 2.0 µl dNTP (2.5 mM), 0.5 µl Taq polymerase (5 U/µl), 3.0 µl template DNA (10 ng/µl), 13 µl ddH2O, 0.1 µg primers. A PTC-100TM thermal cycler (MJ Research Inc. Incline Village, NV, USA) was used along with the following PCR profile settings: 2 min at 96°C, 35 cycles of 40 s at 95°C, 30 s at 61°C, 1 min 30 s at 72°C, and a final extension at 72°C for 10 min. PCR products were electrophoresed on 1.0% agarose gel containing ethidium bromide. Montage TM PCR96 Cleanup Kits were used to purify PCR products before being sequenced with an ABI 377 DNA sequencer.

Results and discussion

MS identification and distribution in the indica and japonica genomes

We identified 102 706 and 102 423 candidate MSs in the indica and japonica genomes, respectively, and the average distances between the repeat units are shorter in the indica genome (3.6 kb) than in the japonica (3.7 kb) genome. These candidate MSs reside largely within intergenic regions (58 094 or 56.6% for the indica and 58 812 or 57.4% for the japonica), as compared to those in intronic regions (20 868 or 20.3% in the indica and 19 631 or 19.2% for the japonica) (Table 1). However, the average MS lengths in the indica and japonica genomes were found similar, from 17.4 bp to 17.8 bp. The fractions of MSs with unique flanking sequences in the indica and japonica are 67.1% and 72.5%, respectively. We also found that one-third of the total MSs are polymorphic and that the number of MS with unique flanking sequences was slightly lower in the indica (68 895) than in the japonica (74 227), but the number of MS sequences with polymorphisms and their average lengths were nearly the same between the two subspecies (Table 1). It implies that the number of MS with polymorphisms and the average length of polymorphic SSRs are independent parameters in the two rice subspecies. MS loci most likely to be useful as genetic markers are those that are single copy in both genomes and polymorphic when the genomes are compared.

Table 1

Frequency and polymorphism of MSs in different genomic regions of indica and japonica genomes

Genome region	Alla				Wsfsb			Wpmc			Rate of pmd (%)
	Count	Interval (kbp)	Lengthe	GC (%)	Count	Interval (kbp)	Perf (%)	Count	Interval (kbp)	Perg (%)
Indica
5′-UTR	12 411	2.1	17.4	40.1	9324	2.8	75.1	4613	5.6	37.2	49.5
CDS	5406	8.1	20.8	81.4	4217	10.4	78	1162	37.6	21.5	27.6
Intron	20 868	2.9	15.7	21.5	15 670	3.8	75.1	8551	7	41	54.6
3′-UTR	5927	3.2	16.6	20.1	4338	4.4	73.2	2312	8.3	39	53.3
Intergenic	58 094	3.9	17.8	19.9	35 346	6.4	60.8	19 053	11.8	32.8	53.9
Total	102 706	3.6	17.4	26.5	68 895	5.4	67.1	35 691	10.5	34.8	51.8
Japonica
5′-UTR	12 569	2	17.6	42.0	10 462	2.4	83.2	4871	5.1	38.8	46.6
CDS	5685	7.7	20.6	81.4	4854	9	85.4	1249	35	22	25.7
Intron	19 631	2.9	15.7	22.2	16 316	3.5	83.1	8281	6.9	42.2	50.8
3′-UTR	5726	3.2	16.7	20.8	4647	3.9	81.2	2321	7.9	40.5	49.9
Intergenic	58 812	4	18.5	21.1	37 948	6.3	64.5	19 057	12.5	32.4	50.2
Total	102 423	3.7	17.8	27.7	74 227	5.1	72.5	35 779	10.7	34.9	48.2

aAll MSs in the genomes.

bMSs with unique flanking sequences.

cMSs with polymorphisms.

dRate of MS Wpm in MS Wsfs.

eAverage length of MSs expressed in base pairs (bp).

fPercent of MSs with unique flanking sequences against all.

gPercent of MSs with polymorphisms against all.

MS polymorphisms in different genomic regions

We have noticed that the average MS length and distance between candidate MS loci within coding regions of the two subspecies were slightly different. The indica genome has an average MS length of 20.8 bp and an average distance between candidate MS loci of 8.1 kb; the corresponding numbers in the japonica genome are 20.6 bp and 7.7 kb, respectively. In the regions other than the coding sequences, the average MS length and the average distance between candidate MS loci become 2.1–3.9 bp and 2.0–4.0 kb, respectively, although MS sequences are most abundant in 5′-UTR, followed in an order of intronic, 3′-UTR, intergenic, and coding regions. The trend was found consistent with previous reports in Arabidopsis and rice.[3,20,21] In addition, GC contents of MS in other regions (19.9–40.1 and 20.8–42.0) are far lower than that of coding regions that are 81.4% and 81.4 % for the indica and japonica, respectively.

We also looked into details of MSs and their unique flanking sequences among different regions in the indica genome and found that there are higher numbers of MSs with unique flanking sequences in intergenic, intronic, and 5′-UTR regions, and the trend is similar in japonica. MSs with both polymorphism and unique flanking sequences vary in different genomic regions, there are as low as 21.5 and 22% in coding regions in the indica and japonica genomes, respectively, whereas there is up to 40% observed in the intronic region (Table 1). We concluded that MSs within intronic regions are most polymorphic, and followed by UTR, intergenic, and coding regions.

Polymorphisms and distribution of different MS repeat types

In our collection, there are 62.5% MNRs, 17.2% DNRs, and 10.6% TNRs, which form the rice genome sequences; these repeat types are collectively over 90% of the total MS sequences. The distributions of these MS types are rather similar between indica and japonica in general. TNRs were predominately found in coding regions (87.5% for indica and 87.6% for japonica, respectively). MNRs and DNRs are highly dispersed in intronic, 3′-UTR, intergenic, and 5′-UTR regions (Fig. 1A). The distribution of polymorphic MS loci is similar among all types, except the percentage of TNRs in various genomic regions varies significantly (Fig. 1B). With respect to the polymorphism rate (the ratio between MSs with polymorphism and the total) and those with unique flanking sequences, we found that MNRs and DNRs are better than TNRs in all genomic regions (Fig. 1C). The polymorphism rate of TTRs, PNRs, and HNRs varies from 32 to 40% (data not shown), and they may provide additional sources for molecular makers when called upon.

Figure 1

Percentages and polymorphic rates of MNR, DNR, and TNR in different regions of the two rice genomes. (A) Percentages of MNR, DNR, and TNR in the overall MS. (B) Percentages of MNR, DNR, and TNR with polymorphisms. (C) Polymorphic rates of MNR, DNR, and TNR.

3.4. Polymorphisms of the candidate MS loci with different repeat units

The number of A/T repeats is 8.5 and 6.8 times of all repeat types in the indica and japonica genomes, respectively; it is 7.5 and 6.6 times more polymorphic than C/G repeats. The average MS length (13.8 bp in the indica and 13.6 bp in the japonica) and the average length of MSs with polymorphisms (14.4 bp in indica and 14.1 bp in japonica) in C/G repeats are slightly longer than those of A/T repeats (10.8–11.2 bp). The polymorphism rate of C/G MNR class of SSR is slightly higher (68.6% for indica and 61.3% for japonica) than that of A/T MNRs (53.5% for indica and 49.6% for japonica). Therefore, when targeting MNRs as the basis for developing genetic markers, the C/G repeat class is likely to be more useful, at least when comparing indica and japonica genotypes. Among DNRs, both AG/CT (59.8–56.4%) and AT/TA (>50%) repeats are more polymorphic than CG/GC repeats (15.6%). AT/TA repeats in the indica and japonica assemblies have a longer average length (>30 bp) than AG/CT repeats (22.6–22.5 bp; Table 2). The polymorphism rate of GC-rich TNRs is relatively lower than that of GC-poor TNRs, despite the fact that CCG/CGG repeats are the most abundant, contributing more than 40% to the total TNRs. The GC-poor repeats, such as AAT/ATT, AAG/CTT, and ACT/AGT, have polymorphic rates greater than 40% (Table 2). Another example is AGG/CCT that contributes to 15% of the total TNRs but the polymorphic rate of AGG/CCT repeats is relatively low, 35.5% for indica and 31.7% for japonica. The numbers of polymorphic ATC/GCT, AAC/GTT, and ACT/AGT repeats are scarce. Among TTRs, AGAT/ATCT, AAAT/ATTT, AAAG/CTTT, ACAT/ATGT, AGCT/AGCT, and AATT/AATT are quite dominant, and the two kinds of TTR (ACAT/ATGT and AGAT/ATCT) have longer average lengths and higher polymorphism rates. Among PNRs, AAAAG/CTTTT, AAAAT/ATTTT, AGAGG/CCTCT, and AGGGG/CCCCT are rather dominant, AGAGG/CCTCT is more polymorphic, 24.8% in indica and 31.7% in japonica; all other kinds of PNR have much lower polymorphic rates (data not shown).

Table 2

Polymorphism distribution of MS motifs in indica and japonica genomes

Motifs	Indica				Rate of pm	Japonica				Rate of pm
	All		Wpm			All		Wpm
	Count	Length	Count	Length		Count	Length	Count	Length
A/T	57 467	10.9	21 393	11.2	53.5	53 940	10.8	20 637	11.1	49.6
C/G	6710	13.8	2849	14.4	68.6	7923	13.6	3104	14.1	61.3
AC/GT	1405	18.2	541	19.6	54.4	1318	18.1	560	19.4	52.5
AG/CT	8371	20.1	3220	22.6	59.8	8664	19.7	3351	22.5	56.4
AT/AT	7271	34.5	2147	36.8	56.6	8041	35.9	2287	38.1	54.4
CG/CG	668	14.8	59	15.7	15.6	755	14.7	64	15.5	15.6
AAC/GTT	186	24.5	45	30.7	37.8	185	22.7	57	25.2	41.6
AAG/CTT	892	25.1	304	29.0	46.5	863	26.0	305	30.1	41.8
AAT/ATT	683	45.0	248	51.7	55.6	671	49.5	253	59.5	49.6
ACC/GGT	603	20.1	111	21.3	23.8	632	20.1	120	21.7	22.2
ACG/CGT	832	19.7	117	21.6	20.8	942	19.5	122	20.6	19.5
ACT/AGT	151	22.2	44	24.8	40.7	139	22.2	57	24.9	44.5
AGC/GCT	958	19.7	216	20.9	31.3	999	19.9	264	20.7	33.2
AGG/CCT	1540	20.6	424	22.1	35.5	1557	20.5	415	21.8	31.7
ATC/GAT	310	22.2	70	22.7	37.8	303	21.6	74	22.4	37.0
CCG/CGG	4734	19.7	999	21.2	28.0	5009	19.6	1087	20.8	26.6

Labels are the same as in Table 1.

We detected 35 691 polymorphic MS loci with unique flanking sequences in the indica genome and 35 779 in the japonica genome (Table 3). If we took away 18 016 loci common for both, there were 53 454 subspecific MS loci left, and we were able to design primer pairs for 52 845 MSs (Table 3), which gave rise to one polymorphic MS for every 7.1 and 7.2 kb in lengths in indica and japonica, respectively. Of these MSs, there were 36 374 MNRs, 8298 DNRs, 3878 TNRs, 802 TTRs, 683 PNRs, 487 HNRs, and 2323 complex repeats that comprised two or more repeat units (Table 3). We also analyzed the length polymorphism of MSs in indica and japonica sequences (Fig. 2), and found that approximately 38% of MSs have a length difference of 1 bp. Only about 8% of MSs have a length polymorphism of more than 10 bp, and the number of MSs with polymorphism length of 5 and 7 bp is relatively small (Fig. 2), since the length polymorphism is related to the length of the repeat units or types.

Table 3

Summary of polymorphic MS loci

		MNR	DNR	TNR	TTR	PNR	HNR	Complex	Total
(1)	Indica MS	24 242	5967	2578	477	394	281	1752	35 691
(2)	Japonica MS	23 741	6262	2754	562	451	275	1734	35 779
(3)	Common MS	11 378	3658	1413	231	156	53	1127	18 016
(4)	Theoretical estimates of polymorphic MS	36 605	8571	3919	808	689	503	2359	53 454
(5)	Observed values of polymorphic MS	36 555	8546	3942	803	688	503	2417	53 454
(6)	MS with primers	36 374	8298	3878	802	683	487	2323	52 845

(1) The identified polymorphic MS loci in indica; (2) The identified polymorphic MS loci in japonica; (3) The common polymorphic MS loci that are found to be the same in both indica and japonica; (4) The theoretical estimates of all polymorphic MS loci in rice that can be estimated according to the following formula polymorphic MS = (1) + (2) − (3); (5) Observed values of all polymorphic MS loci in rice. The data discrepancies between (4) and (5) are due in part to the fact that some MS in indica belong to MNR or DNR, also some MS in japonica can be changed into the Complex kinds; (6) MS with primers: polymorphic MS loci with primers that were designed successfully with Primer3 (http://www.genome.wi.mit.edu/cgi-bin/primer/primer3).

Figure 2

Length distribution of polymorphic MSs in indica and japonica rice genomes.

3.5. Development of the polymorphic MS database for rice subspecies

A total of 18 828 SSR markers have been recently reported based on sequence analysis of the japonica cv. Nipponbare genome,[15] and 2240 of the total were experimentally confirmed.[13] Since some of the markers between the two genomes may not be polymorphic and their physical positions may not be unique, we may not be able to include all of them in our database. Although some MS polymorphisms in chromosome 4 between the indica and japonica have been reported,[16] it was limited to a single chromosome without eliminating multiple-hitting copies. In addition, previous works have mainly focused on Class I MSs, such as DNRs, TNRs, and TTRs, whereas Class II MSs, including MNRs, PNRs, and HNRs have not drawn enough attention for marker development.

By comparing MS sequences between the sequenced indica and japonica genomes, we developed a new database for 52 845 (Table 3) polymorphic MS loci with specific flanking sequences, including polymorphic ample Class II MSs. To avoid confusion in MS naming convention used by the rice community, we integrated 18 828 MSs published by IRGSP[15] into our collection, and the same loci are cross-referenced (6294 MSs from IRGSP). However, 6641 MSs were excluded because they have multiple copies in the rice genome. Some 3988 MSs were found lack of polymorphism between the indica and japonica genomes, and 1905 MSs were not mapped due to the absence of reciprocal polymorphic sequences between the two genomes. Only 4509 Class I polymorphic MS loci were detected in the indica genome (Table 4), 1775 (6284 − 4509) loci were missed because the length of their repeat motif was less than 20 bp and do not satisfy the criteria of Class I in indica, although they were Class I loci in japonica. These 1775 shorter MS alleles in indica may not represent any technical or biological anomalies but simply happened by chance since MSs are in general short in length. In addition, we also mapped the genetic markers consisting of mostly RFLPs from RGP (http://rgp.dna.affrc.go.jp) on the indica (from http://rise.genomics.org.cn), IRGSP, and TIGR pseudomolecules in silico. Although there is a misunderstanding about the BGI's sequence assemblies, the indica pseudomolecules are believed as useful as the BAC-to-BAC assembly when evaluated carefully according to scientific rationales.[22]

Table 4

Comparative analysis of MSs identified in indica with different cutoffs

	Class I						Class II
	All count	Wsfs		Wpm		Rate of pm	All count	Wsfs		Wpm		Rate of pm
		Count	Perf	Count	Perg			Count	Perf	Count	Perg
MNRs	789	641	81.2	417	52.9	65.1	61 074	45 996	75.3	23 324	38.2	50.7
DNRs	6753	4071	60.3	2747	40.7	67.5	16 947	9654	57.0	3825	22.6	39.6
TNRs	3763	2770	73.6	1345	35.7	48.6	54 824	33 480	61.1	2638	4.8	7.9

Class I: MNRs with 20, DNRs with 10, and TNRs with seven repeat units; Class II: MNRs with 10 repeat units, DNRs with seven, and TNRs with six repeat units. The rest of the labels are the same as in Table 1.

Classes of complex MS, such as MNRs, PNRs, and HNRs, could be used as genetic markers in the future

At present, only some of the MSs, such as DNRs, TNRs, and TTRs, are used as genetic markers for rice, and other types have been largely ignored, perhaps due to the lack of adequate experimental methods, such as in the case of MNRs, although they are accounted for more than 60% of all MSs and approximately half of them are believed polymorphic. In this report, we validated a few MNRs (such as P005 and P010 in Fig. 3), proving their usefulness, and a recent report in typing method[23] also suggested a new opportunity for the MNR application as genetic markers for rice. Although PNRs and HNRs have been previously used as molecular markers in sorghum and cotton,[24,25] they have not been developed for rice. We found many polymorphic PNRs and HNRs with polymorphisms in the rice genome, and verified some of them experimentally, including eight PNRs, five HNRs, and four complex MSs. We believe that they should be used as genetic markers in rice in the future. It is important to note that polymorphisms found between 9311 and Nipponbare may not be generalized to other indica and japonica genomes, and additional sequence and genotyping information from diverse cultivars will broaden our understanding of SSR polymorphism in O. sativa.

Figure 3

Experimental validation (PCR) of polymorphic MSs. Left lanes show PCR products from indica cv. 9311, and right lanes are samples from japonica cv. Nipponbare. MS involved MNR (P005, P010), DNR (P041, P031, P050, P043, P009, P039, P026, P040, P025, P024, P027), TNR (P055, P058, P062), TTR (P077), and other complex MS (P008, P095, P097, P100). Four PCR products for 9311 (P025, P026, P058, P062) and two for Nipponbare (P025, P058) showed complex band patterns and seven products (P008, P010, P025, P050, P055, P058, P062) appeared polymorphic.

Comparison of MSs identified in indica were made with different cut-off values. MS polymorphisms have two basic parameters, types of the repeat units and numbers of units. By adjusting cut-off values for the number of repeat units toward a lower boundary, we discovered more polymorphic MS loci that are both highly polymorphic and can be uniquely identified in the genomes. We did try to define Class 1 MS by using a standard protocol, but only detected 417 MNRs, 2747 DNRs, and 1345 TNRs in the indica genome assembly. In this report, we found that when the MNR length reaches 10 bp, the polymorphism numbers increases up to 23 324 for MNR with a length of 10 bp, 3825 for DNR with a length of 7 bp, and 2638 for TNR with a length of 6 bp (Table 3). It was evident that polymorphism counts of most Class II MNRs, DNRs, and TNRs are far higher than those of many Class I MS; the result suggests that different numbers of repeat units should be used for screening different MS types as genetic markers, such as 10 for MNRs, 7 for DNRs, 7 for TNRs, 6 for TTRs, 5 for PNRs, and 5 for HNRs.

Most MS repeats have unique flanking sequences (62.1% for indica and 73.1% for japonica in our study), and MSs may be preferentially associated with non-repetitive regions in rice genome as reported by Morgante et al.[3] except for AT/TA and AAT/ATT repeats.[3,10,11] Furthermore, the fraction of MSs with unique flanking sequences in the indica is about 10% lower than that in the japonica, suggesting that there may be more repetitive sequences assembled in the indica genome when compared with the japonica genome. It was reported previously that MS polymorphism is positively correlated with their length,[4,12] and the distribution of GC-rich DNRs and TNRs have a bias toward exons where polymorphisms occur scarcely.[3,11,17,21] We also found in our current study that the degree of the polymorphisms in MSs depends on their genomic locations and the repeat types. For instance, higher polymorphism rates were found within intronic and UTR regions, whereas the lowest within coding regions. The relationship between GC content and MS polymorphism is unclear. For instance, A/T-containing MNRs have a slightly lower polymorphism rate than the G/C containing repeats. The opposite seems true in DNRs, where AG/CG repeats have higher polymorphism rates of 52–60%, than those in CG/GC repeats (15–16%). Similarly, CCG/CGG repeats contributing almost half of the total TNRs have a polymorphism rate of less than 30%, whereas AAT/ATT AAG/CTT and ACT/AGT have polymorphism rates of greater than 40%. Our results are in agreement to previous observations.[4,12] The suggested reason was that GC-rich regions are relatively stable and resulted in less replication slippage.[26]

Experimental validation of 100 polymorphic MS loci

To validate our results, we chose 100 polymorphic MSs randomly for validation, using genomic DNA from indica cv. 9311 and japonica cv. Nipponbare as PCR templates. The PCR products were sequenced, and the result is summarized in Table 5 (other detailed information was summarized in Supplementary Table 1). We have 98 polymorphic MSs PCR-amplified in the two cultivars, and four of them worked for 9311 (P025, P026, P058, and P062 in Fig. 3) and two for Nipponbare (P025 and P058), showing complex banding patterns that were perhaps due to lack of sequence specificity in the genomic DNA sample. Sequencing results confirmed 91 PCR products to be polymorphic, and seven (P008, P010, P025, P050, P055, P058, and P062) of the remaining PCR products that failed in sequencing appeared polymorphic based on the existence of multiple DNA bands on agarose gels. Only one (P055) showed unexpected products with a length less than 150 bp. In addition, the sequences amplified with six primer pairs (P060, P016, P037, P087, P020, and P030 in Supplementary Table 1) appeared inconsistent with the expected genomic sequences. This could be explained by the fact that the low-complexity sequences were difficult to be sequenced and relatively error-prone. We concluded that our new MS collection contains at least 90% valid polymorphic MS markers.

Table 5

Summary of PCR amplification and sequencing in indica and japonica genomes

MS type	Total No.	9311		Nipponbare		No. both sequenced
		No. of amplified	No. of sequenced	No. of amplified	No. of sequenced
MNRs	12	12	10	12	11	10
DNRs	37	35 (1)	34	36 (1)	36	34
TNRs	23	23 (2)	22	23 (1)	19	19
TTRs	7	7	7	7	7	7
PNRs	8	8	8	8	8	8
HNRs	5	5	5	5	5	5
Complex	8	8	8	8	8	8
Total	100	98	94	99	94	91

Numbers in parentheses are the number of products with complex band pattern.