Dihydroxyacid dehydratase is important for gametophyte development and disruption causes increased susceptibility to salinity stress in Arabidopsis

Highlight Molecular characterization of dihydroxyacid dehydratase in Arabidopsis reveals its important roles in gametophyte and root development, as well as involvement in salinity stress resistance.


Introduction
Isoleucine (Ile), valine (Val), and leucine (Leu) are grouped as branched-chain amino acids (BCAAs), as they all contain short branched hydrocarbon residues. Research interest in the biosynthetic pathway of BCAAs in plants was motivated by two main reasons. First, like animals, human are unable to synthesize BCAAs de novo, and have to obtain them through diet or other sources Joshi, 2009, 2010). Second, some enzymes involved in BCAA biosynthesis are targets of herbicide applications (Tan et al., 2006).
The enzymes participating in BCAA biosynthesis are widely conserved in bacteria, fungi, and plants. Threonine deaminase (TD) catalyses the first and committed step towards Ile biosynthesis by converting threonine into 2-oxobutanoate (Mourad and King, 1995;Joshi et al., 2006). When condensed with one molecule of pyruvate and following four more enzymatic reactions, 2-oxobutanoate is transformed into Ile (Yu et al., 2013). A unique feature of BCAA biosynthesis is that a single set of four enzymes catalysing Ile production are also utilized in Val biosynthesis. These enzymes are acetohydroxyacid synthase (AHAS; EC 4.1.3.18), ketolacid reductoisomerase (KARI; EC 1.1.1.86), dihydroxyacid dehydratase (DHAD: EC 4.2.1.9) and branched-chain aminotransferase (BCAT; EC 2.6.1.42). The last intermediate of Val biosynthesis, 3-methy-2-oxobutanoate, also serves as the initial substrate in the pathway toward Leu biosynthesis. After undergoing four enzymatic steps, including condensation, isomerization, decarboxylation and transamination, the intermediate is ultimately converted to Leu (de Kraker et al., 2007;Knill et al., 2008;He et al., 2009;He et al., 2010;He et al., 2011). To date, virtually all enzymatic activities required for BCAA biosynthesis have been detected in plants, including the model plant Arabidopsis. However, not all genes encoding these enzymes have been characterized in detail (Binder, 2010). For example, very little is known about KARI and DHAD, although the corresponding genes in Arabidopsis have been annotated for years based on sequence similarity to microbial homologues (Binder, 2010).
Proline accumulation has long been recognized to play an important role in osmotic regulation under a wide range of abiotic stresses (Liu and Zhu, 1997;Verbruggen and Hermans, 2008;Mattioli et al., 2009;Hayat et al., 2012). Partial deficiency of proline in mutants impaired in proline biosynthesis, such as p5cs1, caused plants to become hypersensitive to salt stress (Székely et al., 2008). However, it has been recognized that proline was not the only overproducing amino acid under osmotic stresses in some studies. In fact, the levels of other amino acids, particular three types of BCAAs, could be elevated to levels comparable to proline or even greater Joshi et al., 2010). This suggests that BCAA biosynthesis might also be involved in plant stress tolerance, but no direct evidence has been obtained to support this hypothesis.
To investigate the important roles of DHAD in regulating plant development and stress tolerance, we characterized the physiological alterations in Arabidopsis DHAD knockout and knockdown mutants. The lethality of the knockout mutants was identified and is partially caused by impairment in male and female gamete development. In the knockdown mutants, the levels of all three types of BCAA were reduced in roots, leading to a shorter root phenotype. More interestingly, the knockdown mutants exhibited higher sensitivity to salt stress, providing evidence for the first time that BCAA homeostasis plays a role in plant salt tolerance.

Plant materials
Seeds of Arabidopsis thaliana ecotype Columbia (Col-0, CS3879) and SALK mutant lines dhad-1 (SALK_062347), dhad-2 (SALK_075098/SALK_130404), dhad-3 (WiscDsLoxHs135_03D), and dhad-4 (WiscDsLoxHs184_11A) were obtained from the Arabidopsis Biological Resource Center (ABRC). Seeds of the lib mutant were kindly offered by Dong Liu (Yu et al., 2013). The seeds were surface sterilized using 70% ethanol for 1 min, 50% bleach for 10 min, and washing four times with sterilized water for 1 min each time. Seeds were germinated on half-strength Murashige & Skoog (half MS) agar medium containing 1% sucrose, and cultivated in a growth chamber with a photoperiod of 16 h light at 22°C and 8 h dark at 20°C, and 70% humidity, for 7 days. The seedlings were transferred to soil and grown under the same conditions. For salt and nickel stress experiments, Arabidopsis seedlings were germinated and grown on the same half MS medium containing different concentrations of Na + and Ni 2+ by adding NaCl and NiSO 4 , respectively. The root length was measured with a ruler and the number of lateral roots was recorded using dissecting light microscopy.

DNA extraction and genotyping
A rapid genomic DNA extraction protocol was utilized in the study. Briefly, pieces of leaf tissue collected from 3-week-old plants were grounded in extraction buffer (200 mM Tris/HCl pH 8.0, 250 mM NaCl, 25 mM EDTA, and 0.5% SDS) using a small blue pestle. After centrifugation at 14,000 rpm for 5 min at 4°C, the supernatant was transferred into a new tube and precipitated by adding an equal amount of isopropanol. After centrifugation at 14,000 rpm for 5 min at 4°C, DNA pellets were rinsed with 70% ethanol, air dried for 10 min, and finally resuspended in 1XTE buffer (10mM Tris/HCl pH 8.0 and 0.1mM EDTA).
RNA extraction and quantitative real-time PCR RNA was extracted from leaves and roots using an RNeasy plant mini kit (Qiagen, http://www.qiagen.com/). cDNA was synthesized from 5 μg of total RNA using the ProtoScript® First Strand cDNA Synthesis Kit (New England Biolabs) following the manufacturer's instructions. Quantitative real-time PCR was conducted on the ABI 7500 real-time PCR system (Applied Biosystems) according to the manufacturer's instructions. PCR reactions were performed in a total volume of 20 μl containing 10 μl iTaq Universal SYBR Green Supermix (Bio-Rad), 0.5 μl 10ng μl -1 forward and reverse primers (DHAD-F and DHAD-R), and 2 μl of diluted cDNA. The PCR programme was as follows: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 34s. The levels of gene expression were calculated using the comparative CT method and the expression of ACTIN2 was used as an internal control for data normalization. Data shown were the averages of three independent experiments and the primer sequences are listed in Supplementary Table 1.

Promoter-GUS fusion, Arabidopsis transformation, and GUS assay
The Arabidopsis DHAD promoter (from −965 to −1 bp) was amplified from genomic DNA using primers DHAD-1L and DHAD-1R, and the PCR product was cloned into pGEM-T vector (Promega). Once accuracy was verified by sequencing, the promoter sequence was subcloned into pCAMBIA1305 using SacI and NcoI restriction sites to produce the Promoter DHAD ::GUS construct (GUS, β-glucuronidase).
The construct was transformed into Agrobacterium tumefaciens strain GV3101 and subsequently transformed into plants through the floral dipping method (Clough and Bent, 1998). Transgenic plants were selected using 50 ng µl -1 hygromycin.
Histochemical GUS assays were performed on homozygous T3 transgenic plants as described previously (Jefferson et al., 1987).

Amino acid analysis
Free amino acids were extracted using a methanol and chloroform method from leaves (100 mg) and roots (50 mg) (Colebatch et al., 2004), and profiled using an HPLC-based pre-column derivatization protocol (Schuster et al., 2006).

AtDHAD is encoded by a single gene that is constitutively expressed
Based on high sequence similarity to orthologues in other organisms, the gene encoding DHAD in Arabidopsis has been assigned to AT3G23940 in the Plant Metabolic Pathway Database (www.PlantCyc.org) and in other reviews Binder, 2010). To evaluate the evolutionary relationships of DHAD across different plant organisms, a phylogenetic tree based on amino acid sequences was constructed. The result showed that DHAD genes from monocots and dicots formed two distinct clades, indicating that although the sequences were fundamentally conserved, DHAD genes underwent subsequent divergence accompanied by the evolution of these groupss (Fig. 1A). It is also worth noting that DHAD seems to occur as a single-copy gene in all plant species investigated, implying that there is selection pressure to preserve DHAD genes as singletons following genome-wide or small-scale duplication events during evolution.
To investigate the tissue-specific expression of DHAD, histochemical assay of GUS activity was performed using transgenic Arabidopsis containing a DHAD promoter-GUS construct. Of 20 independent transgenic lines investigated, 19 displayed similar patterns of GUS expression. Strong GUS staining was already observed in cotyledons and roots at 3 days after germination (Fig. 1B). At 10 days, intense staining was found in developing leaves and roots (Fig. 1C). In the flower, GUS staining was observed in nearly all the organ parts, including sepals, petals, anthers, and pistils ( Fig. 1D, E).

Identification of T-DNA mutants of AtDHAD
The AtDHAD gene contains 14 exons and 13 introns, and encodes a putative 608 amino acid protein ( Fig. 2A). To investigate the biological importance of AtDHAD in plant development and stress tolerance, we obtained five T-DNA insertion lines from the SALK and WiscDsLox collections (Alonso et al., 2003;Woody et al., 2007), named hereinafter dhad-1 (SALK_062347), dhad-2 (SALK_075098/ SALK_130404), dhad-3 (WiscDsLoxHs135_03D), and dhad-4 (WiscDsLoxHs184_11A). The insertions sites of all these mutant lines were verified by sequencing the junctions of the gene/T-DNA. It should be noted that the sequencing results showed that SALK_075098 and SALK_130404 harbour T-DNA at the same location, so only SALK_130404 was chosen for further study. The insertions of dhad-1 and dhad-2 reside in the putative promoter or 5'UTR regions with 273 bp and 88 bp upstream of the start codon, respectively. The insertions of dhad-3 and dhad-4 locate in exon 1 and intron 3, respectively ( Fig. 2A).

AtDHAD is an essential gene in Arabidopsis
The heterozygous plants of dhad-3 and dhad-4 did not show any apparent phenotypes under normal growth conditions, however we failed to find any homozygous plants from a large population of T2 progeny (Table 1), suggesting that loss of function of DHAD caused lethality. Meanwhile, segregation distortions were observed in both dhad-3 and dhad-4 alleles, with the ratios of wild-type (WT) to heterozygous plants far more different from the expected 1:2 (indicative of embryo lethality) or 1:1 (indicative of single gamete lethality), suggesting that abortion might occur in the development of both male and female gametophytes (Table 1).
To ascertain the origin of lethality from male, female, or both gametes, the reciprocal cross-pollination study was performed and the transmission efficiency (TE) through both male and female was calculated. As shown in Table 1, when heterozygous dhad-3 or dhad-4 plants were used as pollen donors, the TEs of dhad were 46.8% and 44.3% for dhad-3 or dhad-4, respectively, indicating that a proportion of dhad male gametes was aborted. Likewise, the female TEs were 43.9% and 41.3% for dhad-3 or dhad-4, respectively, indicating that the female gametophyte was affected to a similar extent to the male gamete (Table 1).
To clarify the effects of disruption of DHAD on male gamete development, Alexander staining was used to observe pollen viability and morphology (Alexander, 1969). For the WT, all pollens grains were uniformly shaped and sized (Fig. 2B). In comparison, although largely red stained, ~40% of pollen grains from DHAD-3/dhad-3 (40.6%; n = 1225) and DHAD-4/dhad-4 (38.3%; n = 1335) were smaller in size ( Fig. 2C-E). Given that the percentage of pollen grains displaying morphological alterations correlated well with the observed frequency of male gametophytes with the mutant genotype (Table 1), we inferred that the genotype of the small pollen grains was dhad, and that such abnormality may cause pollen grains that are less-or non-viable compared to WT pollen (Fig. 2C, D, E).
According to the TE data, female gametophyte development was also impaired in the dhad mutants (Table 1). This was further supported by examination showing that small, white, unfertilized ovules were substantially observed in the developing siliques of DHAD-3/dhad-3 and DHAD-4/ dhad-4 ( Fig. 2F-H). The frequencies of aborted ovules were 45.6% (n = 542) and 42.1% (n = 503) for DHAD-3/dhad-3 and DHAD-4/dhad-4, respectively (Fig. 2I). This is consistent with the values calculated from TE analysis (Table 1). In addition, it should be noted that although both male and female gamete development resulted in mostly sterile gametes, a small proportion were still viable; this is opposed to the genetic analysis in which none of the homozygous mutant plants could be recovered from F 2 progeny, implying that loss of function of DHAD may also cause embryo lethality.

Decreased expression of AtDHAD led to retarded root growth
In contrast to the lethality of dhad-3 and dhadh-4 knockouts, homozygous plants were successfully obtained from self-pollinating progenies of heterozygous dhad-1 and dhad-2, the alleles containing the insertions in the promoter or 5'UTR regions of DHAD (Fig. 1A). Real-time quantitative PCR analysis showed that the expression levels of DHAD in homozygous dhad-1 and dhad-2 were decreased to 36% and 25% compared to WT sibling plants (Supplementary Figure  S1A), indicating that T-DNA insertions at the promoter or 5'UTR regions downregulated the expression of DHAD.
When grown in soil, homozygous plants did not show any obvious phenotypes, such as morphological appearance, growth rate, and bolting time compared to WT (Supplementary Figure S1B). However, when seeds were germinated on half MS medium and grown vertically for 10 days, the average root length in both dhad-1 and dhad-2 was significantly shorter than WT (Fig. 3A), and the lateral root density was also decreased (Supplementary Figure S2), suggesting that both primary and lateral root development were affected. To clarify whether the shorter root phenotype was caused by the deficiency of free BCAAs, plants were grown in medium exogenously supplied with each individual type or a variety of combinations of two or all three BCAAs. Interestingly, it was observed that when supplied with Ile (Fig. 3B), not Leu ( Fig. 3C) or Val (Fig. 3D), the root length in mutants was restored to a level similar to WT, suggesting that the decreased root growth might be caused by the reduced level of Ile in mutants. It is further shown that any combinations containing Ile could rescue the root-growth phenotype (Fig. 3E, F, and H). It is worth mentioning that exogenous application of Leu caused cellular toxicity (Fig. 3C), and this toxicity was further exaggerated when Leu and Val were added together (Fig. 3G). All the above-described root-growth assays were independently repeated three times, and root growth rates are summarized and quantified in Fig. 3I.
A range of concentrations of 2-oxobutanote were exogenously applied in the growth medium. This was to test the possibility that the potential buildup of 2-oxobutanote, the Ile biosynthetic intermediate, will harm root growth in dhad knockdown mutants. As shown in Supplementary Figure  S3, up to 0.5 mM 2-oxobutanoate did not yield any distinct responses in dhad mutants compared to the WT, indicating that either the stunted root growth was not related to  the cellular concentration of 2-oxobutanoate, or alternatively that the toxicity of 2-oxobutanoate on plant growth is extremely minimal.
To corroborate that the shorter root phenotype correlated with the reduced level of BCAAs, free amino acids were quantified in root and leaf tissues. It is noted that since both alleles, dhad-1 and dhad-2, exhibited identical phenotypes in root-length assays, only dhad-2 was used in amino acid analysis. As shown in Table 2, in roots, the abundances of all three BCAAs in dhad-2 showed a statistically significant reduction compared to the WT, suggesting that the decreased gene dosage of DHAD could limit the rate of metabolic flux to form the end products, BCAAs. In addition, the contents of two precursor amino acids, aspartate and threonine, were moderately increased, suggesting that the partial blockage of BCAA biosynthesis could lead to the ectopic buildup of the precursor intermediates. By comparison, in leaves, no amino acids showed significant changes (Table 2), consistent with the indistinguishable phenotypes of the mutants and WT (Supplementary Figure S1B). Taken together, these results indicate that the decreased expression of DHAD resulted in partial deficiency in BCAA biosynthesis, and subsequent impairment of root development.

Decreased expression of DHAD resulted in increased sensitivity to salt stress
Previous studies have shown that the accumulation of BCAAs was greatly elevated under abiotic stresses Joshi et al., 2010), suggesting that BCAA homeostasis may contribute to the multifaceted stress tolerance systems in plants. To test the possibility that reduction in BCAA abundance may lead to hypersensitivity to abiotic stress, the effects of salt stress on root growth were investigated. As shown in Fig. 4A, with a low concentration of Na + (50 mM), the root length of dhad knockdown mutants was 62-64% of that in the WT (Fig. 4G), which is lower than the 68-71% without salt treatment (Fig. 3A and 3I). When Na + concentration increased to 75 mM, the ratio of the root length between the mutant and WT became 53-55% ( Fig. 4B and 4G), and further decreased to 39-41% after treatment with 100 mM Na + . This indicates that dhad knockdown mutants exhibited greater sensitivity compared to the WT towards increasing concentration of Na + . More importantly, the sensitivity to Na + treatment under different concentrations could be restored to similar levels as in the WT when 0.1 mM Ile was supplied exogenously. These results suggest that the high sensitivity of dhad to Na + stress can be attributed to the reduced level of Ile (Fig. 4D-G).
The hypersensitive response to salinity stress may be directly derived from salt ionic toxicity, or alternatively represent a universal reaction to osmotic stress. To ascertain these two possibilities, dhad knockdown mutants were  -test. b Values significantly lower that the WT (P < 0.05), Student's t-test. c Amino acid contents are given in pmol mg FW -1 (FW, fresh weight). Data are the mean ± SD of three replicates. Fig. 4. dhad-1 and dhad-2 are hypersensitive to salt stress. (A-F) 10-day-old WT, dhad-1, and dhad-2 seedlings grown on half MS medium supplemented with different levels of NaCl (A-C) and 0.1mM Ile (A-F). Bar, 10 mm. (G) Root length in the mutants was measured and shown as a percentage relative to that in the WT grown in the same conditions. Error bars indicate standard error. This figure is available in colour at JXB online. treated with increased concentrations of mannitol. As shown in Fig. 5, with a low concentration of mannitol, the root length of dhad mutants was 72-77% of that in the WT, which is comparable to the 68-71% without any treatment ( Fig. 3A  and 3I). However, with increased concentrations of mannitol, the ratio of root length between the mutant and WT was increased to 84-89%, which is clearly different from the salt response, indicating that there is not an exaggerated effect under osmotic stress. Therefore, we conclude that hypersensitivity in dhad to salinity stress can be attributed to the salt ionic toxicity.
The inhibition of root growth of dhad, by salt stress, might also reflect a common response to various abiotic stresses. To test this possibility, the root-growth assays were conducted under different concentrations of Ni 2+ treatment, which represents a type of heavy metal stress. As shown in Fig. 6, the mutants did not exhibit the signs of high sensitivity to the increased concentrations of Ni 2+ . In contrast, the ratios of root length between mutants and WT became slightly increased under elevated concentrations of Ni 2+ . This suggests that the high sensitivity to salt stress in dhad is quite specific.

The lib mutant shows the same pattern toward increased sensitivity to salt stress
To further establish the functional role of Ile biosynthesis in tolerance to salinity stress, the root-growth assays were conducted with lib (Yu et al., 2013). As shown in Fig. 7, similar to dhad alleles, lib shows higher sensitivity toward increased salt concentration, and this hypersensitivity could be rescued by exogenous application of Ile, unambiguously indicating that the Ile biosynthetic pathway plays a critical role in salt tolerance.  Bar, 10mm. (G) Root length in mutants was measured and is shown as a percentage relative to that in the WT grown in the same conditions. Error bars indicate standard error. This figure is available in colour at JXB online.

Discussion
As the essential components of protein synthesis and also as important signalling molecules, BCAAs have been thought to regulate numerous aspects of plant growth required for plant survival. However, little information is available about specific roles of BCAAs in plant growth and development as well as interaction with the environment. In recent years, extensive studies in Arabidopsis have led to the identification of most of the genes involved in BCAA biosynthesis. To date, only two genes, KARI and DHAD, remain functionally uncharacterized. In this study, we identified that null mutations of DHAD caused partial abortion in both male and female gametogenesis. This is reminiscent of the loss of function of LeuD3 and IPMDH2/IPMDH3, involved in Leu biosynthesis, where the lethality was detected in female or male gamete development, respectively (He et al., 2009;He et al., 2010). The underlying causes of lethality in distinct organs of different mutants include (i) different organs having exceptional requirements for the de novo biosynthesis of certain types of BCAAs; (ii) different biosynthetic intermediates having dissimilar abilities or requirements to be transported from sporophytic into reproductive tissues; and (iii) the potential cellular toxicities resulting from the abnormal accumulation of biosynthetic intermediates having different effects on plant development.
The decreased expression of DHAD resulted in a simultaneous reduction in the levels of the three BCAAs in roots, but not in leaves (Table 2), indicating that DHAD may exert rate-limiting control in regulating BCAA biosynthesis in a tissue-specific manner. However, the possibility of the existence of an alternative pathway for BCAA biosynthesis in leaves cannot be completely ruled out in this study. In addition, the retarded root growth could only be rescued by the exogenous application of Ile, not Leu or Val, suggesting that the impaired root growth was most probably caused by the deficiency in Ile accumulation. This result is consistent with results from a recent study in which a mutation in threonine deaminase/dehydratase catalysing the first and also committed step of Ile biosynthesis resulted in partial Ile deficiency, and consequently retarded root development (Yu et al., 2013). Therefore, it is clear that Ile homeostasis plays an important role in regulating root development in Arabidopsis. Moreover, the perturbation of DHAD may lead to a buildup of metabolic intermediates. For instance, it has previously been shown that the inhibition of AHAS can cause toxic accumulation of ketobutyrate in Salmonella (LaRossa, 1987). However, in this study, no hint of the cellular toxicity caused by 2-oxobutanoate was suggested, since the exogenous application of substantial amounts of 2-ketoburate didn't produce any distinct response in dhad relative to the WT. Furthermore, we observed that the concentrations of several other amino acids were affected, including increased levels of asparagine, aspartate, glutamate, and threonine, as well as a reduced level of lysine. As the precursor amino acid for Ile, the accumulation of threonine could simply be explained by potential build-up due to the blockage of downstream metabolic flux. Similarly, all the BCAAs are aspartate-derived amino acids, therefore the reduction in BCAAs could lead to a build-up of aspartate, subsequently further recruited into asparagine. However, the way in which glutamine increased and lysine decreased is unclear, suggesting that the biosynthesis of different amino acids is under the regulation of complex crosstalk, and the precise change leading to the imbalance of different amino acids deserves further investigation. In the lib mutant, in addition to the reduced level of Ile, only two other amino acids showed statistically significant changes in abundance; these were an increase in phenylalanine and a decrease in lysine. The consistent reduction in Ile content in both lib and dhad indicates that both genes are involved in Ile synthesis, but the distinct response of other amino acids towards the impairment in different steps along the same biosynthetic pathway suggests that the metabolic network associated with each step was slightly dissimilar. Taken together, although the possibility of the stunted root growth being a consequence of intermediate accumulation could not be completely ruled out, we speculate that the root phenotype is mostly likely caused by Ile deficiency.
Under abiotic stresses, the levels of some amino acids were significantly increased. Proline biosynthesis has been well studied and shown to play important roles in plant tolerance toward various stresses Liu and Zhu, 1997;Verbruggen and Hermans, 2008;Mattioli et al., 2009;Hayat et al., 2012). Disruption of proline biosynthesis renders plants hypersensitive to salinity stress (Liu and Zhu, 1997;Mattioli et al., 2009). Interestingly, it has long been recognized that the accumulation of the three BCAAs exhibited more significant induction than proline in some cases of osmotic stresses, implying that BCAAs may also play a role in plant stress responses Joshi et al., 2010). However, no functional evidence was obtained to draw such a conclusion. In this study, we found that the decreased expression of DHAD led to a reduction in BCAA levels and rendered plants hypersensitive to salinity stress ( Table 2 and Fig. 4), providing for the first time strong evidence that BCAA homeostasis has an important impact on plant salinity tolerance. In addition, such hypersensitivity seems to be rather specific for salinity stress, since the parallel assay to test the response to osmotic stress and heavy metal stress showed an opposite result. How does the deficiency in BCAAs change plant sensitivity to salinity stress? We speculate that this is related to both aspects of fundamental functions of BCAAs, in which they serve either as substrates for stress-induced protein biosynthesis or as signalling molecules for regulating stress-responsive gene expression.
To conclude, in this work, molecular and physiological responses to altered expression of DHAD were characterized in Arabidopsis. Our results demonstrate that DHAD encodes an essential gene and plays important roles in both male and female gametogenesis, as well as root development. More importantly, we have observed that partial deficiency of BCAAs renders plants hypersensitive to salt stress. The result provides novel evidence concerning the long-standing hypothesis that BCAA homeostasis contributes to the adaptation of plants to salinity stress.

Supplementary material
Supplementary data can be found at JXB online.
Supplementary Table S1. Primer sequences used in the study.

Funding
This work was supported by funding from the National Programme on Key Basic Research Project in China (973Programme: 2014CB147301) to YH, and partly by the US National Science Foundation CAREER Award 0845162 to SC.