Molecular characterization and application of a novel cytoplasmic male sterility-associated mitochondrial sequence in rice

Development and genetic analysis of SCAR marker

PCR amplification was performed using total genomic DNA of 28 accessions (Table 1) with 264 random primers (10 nucleotides). A 2100-bp product, named U-18/2100, was
amplified when using the RAPD primer OPN U-18 (5?-GAGGTCCACA-3?) with DNA templates
from YTA, CG-41A, HL-2, and C-M23 containing HL-type male sterile cytoplasm (Figure 1A). U-18/2100 was recovered, cloned, and sequenced according to TA-cloning protocols.
Following U-18/2100 sequence analysis, a SCAR marker was developed and named as L-sp1
(specific primers, H1: 5’-GAGGTCCACATCCTTCAATC-3’; H2: 5’-AGGTCCACAAACCACTGAAG-3’).
The genetic nature of the L-sp1 fragment was determined by PCR using total genomic
DNA of plants from two different backcross populations: BC₁F₁s CG-41A//CG-41B/MY23 and YTA//YTB/9311. These two backgrounds display HL cytoplasmic
male sterility and specific nuclear backgrounds (Figure 1B), as did plants from six different types of CMS lines (BC₇F₁) with similar SJB nucleic backgrounds (Figure 1C). The L-sp1 fragment was consistently amplified in all plants possessing the HL
cytoplasmic male sterility background, and amplification remained constant with changes
in the nucleic genome. L-sp1 can be inherited cytoplasmically or maternally, and has
specificity for CMS. Based on characteristics of the cytoplasmic genes and the possible
relationship between L-sp1 and CMS, we further verified stability and reliability
of the SCAR marker using mitochondrial DNA from 18 out of 28 accessions (Table 1) with specific primers H1 and H2. The L-sp1 fragment could be amplified from mitochondrial
genomic DNA of YTA, CG-41A, HL-2, and C-M23 (Figure 1D), suggesting that U-18/2100 is related to CMS.

Table 1. The three CMS rice types and F₁offspring used in this study

Figure 1. Genetic analysis and verification of L-sp1. A: PCR products amplified from genomic DNA of the studied accessions by OPN U-18. B: Amplified L-sp1 present in test-cross plants of YTA//YTB/9311 and CG-41A//CG-41B/MY23,
and F₂ plants of HL-2. C: Amplified L-sp1 fragment present in plants of different types of CMS lines with
the same nucleic background. D: Mitochondrial genomic DNA amplified by L-sp1 primers. M: DNA marker DL2000. Numbers
above the lanes match to the corresponding plant material No. listed in Table 1.

Molecular characterization of mitochondrial SCAR marker

To assay HL-CMS specificity and L-sp1 copy number, mitochondrial genomic DNA of CG-41A,
YTA, HL-2, YAS, YTB, and of all accessions was digested with EcoRI or BamHI and hybridized with L-sp1 probe. A single fragment of 23 kb was found when using
L-sp1 (2176 bp specific primer, H1: 5’-GAGGTCCACATCCTTCAATC-3’; H2: 5’-AGGTCCACAAACCACTGAAG-3’)
and N-atp6 (S59890) probes (F: 5’-CAATCCTTGGTAGAGTG-3’; R: 5’-TAATGGCAGTGGGACTCC-3’) for all
accessions following digestion with BamHI. The band was the same size in CG-41A, YTA, HL-2, and YAS, but smaller in YTB (Figure 2A and B). Three bands were detected when using the L-sp1 probe following digestion
with EcoR1, there were two bands detected in CG-41A, YTA, HL-2, and YAS, and one band in YTB
(Figure 2C). One band in YTB, and two bands in CG-41A, YTA, HL-2, and YAS were detected when
using the N-atp6 probe following digestion with EcoR1 (Figure 2D). These results indicated that L-sp1 was single copy, and could be used as a characteristic
molecular marker in the mitochondrial genome.

Figure 2. Southern analysis. M: Lambda DNA/Hind III Marker. A, B: Mitochondrial genomic DNA of CG-41A, YTA, HL-2, and YTB, and YAS cut with BamHI. C, D: Mitochondrial genomic DNA of CG-41A, YTA, HL-2, and YTB, and YAS cut with EcoRI. A, C: Southern analysis with L-sp1 probes. B, D: Southern analysis with N-atp6 probes. L-sp1 probe was amplified using the sequence-specific primers H1 and H2,
these contain one EcoRI restriction site and no BamHI restriction site. N-atp6 probe was amplified using sequence-specific primers F and R, and also contain one
EcoRI restriction site and no BamHI restriction site. A part of L-sp1 probe is homologous to a part of the N-atp6 probe.

L-sp1 is a chimeric mitochondrial genomic DNA sequence of 2176 bp (HQ267715). When
compared with the mitochondrial genomic DNA sequence, L-sp1 was determined to contain
four Japonica rice (Nipponbare) DNA fragments (Figure 3A). Except for a 10-bp gap located between bp 1724 and 1725, the sequence from bp
1 to 1933 of L-sp1 showed 99% similarity to bp 224994 to 223054 of the NC_011033.1
clone. The L-sp1 sequence from bp 1684 to 1933 showed 99% similarity to bp 282437
to 282180 and 413358 to 413001 of the same clone. The L-sp1 sequence from bp 1934
to 2100 showed 99% similarity to bp 343650 to 343483 and 424737 to 424570 of the NC_011033.1
clone, and 100% similarity to the cDNA sequence located at 1062 to 1228 bp of the
mitochondrial ribosomal protein L5 gene. The remaining 56-bp DNA sequence (bp 2119
to 2174) of L-sp1 was similar to bp 181916 to 181862 of the NC_011033.1 clone.

Figure 3. Comparison of sequences among L-sp1, Nipponbare mtDNA, N-atp6, and atp6-orfH79. A: Sequence comparison between L-sp1 and Nipponbare mtDNA. B: Sequence comparison between L-sp1 and Indica WA-CMS mtDNA. C: Sequence comparison between L-sp1, N-atp6, atp6-orfH79, and WA352. Colored lines: homologous sequence. Black lines: non-homologous sequence.

Sequence comparison of sequences between L-sp1 and indica WA-CMS mitochondrial genomic
DNA (Figure 3B) was as follows: the L-sp1 sequence from bp 1 to 1722 showed 98% similarity to bp
349531 to 351252 of JF281154.1; bp 1682 to 1931 showed 97% similarity to bp 37787
to 37530 and 219740 to 219997 of JF281154.1; and bp 1931 to 2098 of L-sp1 showed 100%
similarity to bp 338549 to 338716 of JF281154.1. Other sequences shared no homology
to the indica WA-CMS mitochondrial genome.

BLAST analysis of L-sp1, WA352 (AGG40956), N-atp6, and atp6-orfH79 sequences 25] revealed that L-sp1 sequences from bp 1 to 392 were entirely homologous to the 3’
flanking sequence of the ORF of N-atp6. Other DNA sequences showed no homology between L-sp1 and N-atp6. Furthermore, L-sp1 was non-homologous to the total DNA sequences of both atp6-orfH79 and WA352 (Figure 3C).

Distribution of L-sp1 in the AA genome of wild rice

To determine the distribution of L-sp1 in the AA genome of wild rice, PCR amplification
was performed using L-sp1 sequence-specific primers H1 and H2. Bands of approximately
2176 bp, the same size as those in YTA, were found in 11 of 102 investigated wild
rice accessions (Figure 4A). These 11 accessions belonged to four species: three from O. rufipogon (103423, 105696, 105698), five from O. nivara (101978, 103415, 103835, 105712, 106153), two from O. glumaepatula (100968, 105661), and one from O. meridionalis (82042), these accessions came from Cambodia, India, Sri Lanka, Suriname, Brazil,
Bangladesh and Laos in Southeast Asia, West Africa, South America, and Oceania, respectively.
To analyze the distribution of orfH79 in the 11 wild rice accessions, PCR was performed using orfH79 sequence-specific primers (O1: 5?-ATGACAAATCTGCTCCGAT-3?; O2: 5?-TTACTTAGGAAAGACTACAC-3?).
This revealed that 9 of the 11 wild accessions amplified the same sized band, approximately
240 bp, as YTA, while the other two accessions (103423, 105698) failed to amplify
a PCR product (Figure 4B).

Figure 4. Distribution of L-sp1 in wild rice and determination of cytoplasmic background of
HL-type hybrid seeds. A: Distribution of L-sp1 in wild rice. B: Distribution of orfH79 (orf79) in wild rice. YTA and the 11 strains of wild rice correspond to those listed in
Additional file 1: Table S1. M: DNA marker DL2000.

Development of new CMS lines via backcrosses from accessions containing L-sp1

To validate whether L-sp1 was related to CMS at the molecular genetic level, an interspecies
cross was performed using two accessions (103423 and 105698, now named w1 and w2,
respectively) carrying L-sp1 as maternal parents with YTB. Fertility analysis revealed
the percentage of stainable pollen grains of the F₁ hybrids w1/YTB and w2/YTB were 10.3%?±?1.5% and 15.8%?±?2.3%, respectively; over
50% of abortive pollen grains were spherical (Figure 5). The seed-setting rates of bagged spikelets for the same crosses were 13.6%?±?1.5%
and 23.5%?±?2.5%, respectively (Table 2). To elucidate whether male sterility of the test-cross derived from potential incompatibility
between species or subspecies, the HL maintainer YTB was crossed as a female parent
with w1 and w2. Fertility assessment revealed that YTB/w1 and YTB/w2 were both fertile
(~80% pollen fertility; ~50% seed-setting fertility). This indicated that the fertility
of crosses between wild rice and YTB was mainly influenced by the cytoplasm genome
as opposed to the nuclear genome in wild rice. Next, fertility of populations derived
from BC₁F₁ backcrosses of w1/YTB//YTB and w2/YTB//YTB were examined; the spherical abortive
grain rate increased from ~25% to ~85%, and fertility of pollen and seed-setting was
clearly reduced (Figure 5). Therefore, two new CMS lines could be developed from low fertility and sterile
plants belonging to BC₁F₁ (w1/YTB//YTB and w2/YTB//YTB) or BCnF₁ (n, generation number of backcross) maternal parents, by successive backcrossing
with YTB.

Figure 5. Pollen microspores. Bars?=?50 ?m.

Table 2. Fertility analysis of hybrid F₁and BC₁F₁backcrossed lines

Genetic and transcript analysis of L-sp1 in new CMS lines

PCR amplifications were performed using mtDNA from YTA, YTB, w1/YTB, w2/YTB, YTB/w1,
YTB/w2, w1/YTB//YTB, and w2/YTB//YTB plants. When using L-sp1 specific primers H1
and H2, an L-sp1 amplicon was observed in plants with the same cytoplasmic background
as w1 and w2 (w1/YTB, w2/YTB, w1/YTB//YTB, and w2/YTB//YTB), while the L-sp1 amplicon
was not amplified from plants with a different cytoplasm background than w1 and w2
(YTB, YTB/w1, and YTB/w2). This revealed that L-sp1 showed cytoplasmic or maternal
inheritance. In addition, the HL-CMS gene orfH79 was not amplified from the above plants (with the exception of YTA) when using the
orfH79 primer set (O1 and O2) (Figure 6A).

Figure 6. Genetic and transcript analysis of L-sp1 in the newly developed CMS lines. A: Genetic analysis: A1, Genetic pattern of L-sp1; A2, Genetic pattern of orfH79. B: Transcript analysis. M: DNA markers: 250-bp DNA ladder/DL2000. w1: accession 103423;
w2: accession 105698.

To examine the expression manner of L-sp1 in the F₁ and backcrossed BC₁F₁ w1 and w2 backgrounds, RT-PCR was performed using total RNA from rice anthers with
L-sp1 as probe. L-sp1 was found to be expressed in the anthers of w1/YTB, w2/YTB,
w1/YTB//YTB, and w2/YTB//YTB in a similar manner to YTA (Figure 6B).