Organization and Variation Analysis of 5S rDNA in Different Ploidy-level Hybrids of Red Crucian Carp × Topmouth Culter

Through distant crossing, diploid, triploid and tetraploid hybrids of red crucian carp (Carassius auratus red var., RCC♀, Cyprininae, 2n = 100) × topmouth culter (Erythroculter ilishaeformis Bleeker, TC♂, Cultrinae, 2n = 48) were successfully produced. Diploid hybrids possessed 74 chromosomes with one set from RCC and one set from TC; triploid hybrids harbored 124 chromosomes with two sets from RCC and one set from TC; tetraploid hybrids had 148 chromosomes with two sets from RCC and two sets from TC. The 5S rDNA of the three different ploidy-level hybrids and their parents were sequenced and analyzed. There were three monomeric 5S rDNA classes (designated class I: 203 bp; class II: 340 bp; and class III: 477 bp) in RCC and two monomeric 5S rDNA classes (designated class IV: 188 bp, and class V: 286 bp) in TC. In the hybrid offspring, diploid hybrids inherited three 5S rDNA classes from their female parent (RCC) and only class IV from their male parent (TC). Triploid hybrids inherited class II and class III from their female parent (RCC) and class IV from their male parent (TC). Tetraploid hybrids gained class II and class III from their female parent (RCC), and generated a new 5S rDNA sequence (designated class I–N). The specific paternal 5S rDNA sequence of class V was not found in the hybrid offspring. Sequence analysis of 5S rDNA revealed the influence of hybridization and polyploidization on the organization and variation of 5S rDNA in fish. This is the first report on the coexistence in vertebrates of viable diploid, triploid and tetraploid hybrids produced by crossing parents with different chromosome numbers, and these new hybrids are novel specimens for studying the genomic variation in the first generation of interspecific hybrids, which has significance for evolution and fish genetics.


Introduction
Polyploidization, the addition of an extra set (or sets) of chromosomes to the genome, is a predominant mechanism for speciation in plants and animals [1][2]. Polyploidization can occur via somatic doubling, the fusion of unreduced gametes, and by means of a triploid bridge or polyspermy [1][2][3]. The current prevailing opinion is that hybridization plays an important role in triggering polyploidization [1,[4][5][6]. Polyploid species are particularly frequent in the plant kingdom [2,7], and 40-70% of all plant species are polyploids [3,8]. However, while it is generally relatively rare in animals [9][10], polyploidy has occurred extensively, independently, and is often repeated in many groups of fish, from the sharks to the higher teleosts [11][12][13]. Furthermore, artificially induced polyploidy has been used in aquaculture to produce sterility and to improve production [14].
Polyploidy in fish represents a useful model system with which to test theories about the origin and consequences of polyploidy that have been derived from work on plants [15]. Using RCC (Carassius auratus red var.) and TC (Erythroculter ilishaeformis Bleeker), three new types of different ploidy-level hybrid fish were successfully obtained. This is the first report on the formation of these viable diploid, triploid and tetraploid hybrids by crossing different parents with a different chromosome number in vertebrates.
To further reveal the influence of hybridization and polyploidization on the 5S rDNA organization, we performed a study of nucleotide sequences and molecular organization of 5S rDNA in red crucian carp (RCC), topmouth culter (TC) and their hybrid offspring.

Ethics Statement
Administration of Affairs Concerning Animal Experimentation guidelines state that approval from the Science and Technology Bureau of China and the department of wildlife administration is not necessary when the fish in question are not rare or near extinction (first-class or second-class state protection level). Therefore, approval was not required for the experiments conducted in this paper.

Animals and Crossing Procedure
Specimens of RCC and TC were obtained from the Engineering Research Center of Polyploid Fish Breeding and Reproduction of State Education Ministry at Hunan Normal University. During the reproductive seasons (from June to July) in 2009, 2010 and 2011, 10 mature females and 10 mature males of both RCC and TC were chosen as the maternal and paternal parents, respectively. The crossings were performed by two groups. In the first group, RCC was used as the maternal, and TC as the paternal, While in the second group this was reversed. Mature eggs were fertilized and the embryos developed in the culture dishes at the water temperature of 20uC-22uC. In each crossing, 2000 embryos were chosen at random for the examination of the fertilization rate (number of embryos at the stage of gastrula/ number of eggs6100%), the hatching rate (number of hatched fry/number of eggs6100%) and the adulthood rate (number of adulthood/number of eggs6100%). By using the same methods, the same-species mating of RCC and TC were used as controls. The hatched fry were then transferred to a special pond for further culture.
The cross between RCCR6TC= resulted in three ploidy-level hybrid offspring: diploid, triploid and tetraploid hybrids. In the reverse cross between TCR6RCC= no living progeny were produced. Hereinafter, the diploid hybrids of RCCR6TC= are abbreviated as 2nRT hybrids, the triploid hybrids of RCCR6TC= as 3nRT hybrids, and the tetraploid hybrids of RCCR6TC= as 4nRT hybrids.

Preparation of Chromosome Spreads and Measurement of DNA Content
To determine ploidy, chromosome counts were carried out on kidney tissue from 10 RCC, 10 TC, 10 2nRT, 10 3nRT and 10 4nRT hybrids at 1 year of age. Preparations were made according to the method of Liu et.al (2007) [12] with minor modifications. After culture for 2-3 days at the water temperature of 20uC-22uC, the samples were injected with concanavalin one to three times at a dose of 6-10 mg/g body weight. The interval time of injection  was 12-24 hours. 2-3 hours prior to dissecting, each sample was injected with colchicines at a dose of 6-8 mg/g body weight. The kidney tissue was ground in 0.9% NaCl, followed by hypotonic treatment with 0.075 M KCl at 37uC for 40-60 min and then fixed in 3:1 methanol-acetic acid for three changes. Cells were dropped on cold, wet slides and stained with 4% Giemsa for 30-45 min. The shape and number of chromosome were analyzed under a light microscope. 200 metaphase spreads of chromosomes were analyzed for each type of fish (20 from each sample). Preparations were examined under an oil lens at a magnification of 6330 and good-quality metaphase spreads were photographed for analysis of karyotypes. Lengths of entire chromosomes (long and short arms) were measured. Chromosomes were classified on the basis of their long-arm to short-arm ratios according to the reported standards [41]. The DNA content of erythrocytes of RCC, TC and their hybrid offspring were measured using a flow cytometer (Cell Counter Analyzer, Partec, Germany). 1-2 ml of blood was gathered from the caudal vein using a syringe containing ,200-300 units of sodium heparin. Blood samples were then treated according to the method outlined in Liu et.al (2007) [12], and measured under the same condition. The DNA content of RCC and TC were used as the controls. To calculate the ratios of the DNA content of the hybrid offspring to the sum of that of RCC and TC, a x2 test was used, with Yate's correction applied to test for deviation from expected ratio values.

Genomic DNA Extraction, PCR and Sequencing
Total genomic DNA of RCC, TC and their hybrid offspring were extracted from the peripheral blood cells using a phenol/ chloroform extraction method as described in Sambrook et al. (1989) [42]. A set of primers based on those described by Qin et al. [40] (5S P1, 59-GCTATGCCCGATCTCGTCTGA-39: 5S P2R, 59-CAGGTTGGTATGGCCGTAAGC-39) were designed and synthesized to amplify the 5S rRNA genes and their nontranscribed spacer regions directly from genomic DNA. PCR reactions were carried out in a volume of 25 mL with approximately 20 ng of genomic DNA, 1.5 mM of MgCl 2 , 250 mM of each dNTP, 0.4 mM of each primer, and 1.25 U of Taq polymerase (TaKaRa, Dalian, China). The thermal program consisted of an initial denaturation step at 94uC for 5 min, followed by 25 cycles (94uC for 30 sec, 60uC for 30 sec, and 72uC for 1 min) and a final extension step at 72uC for 10 min. Amplification products were separated on a 1% agarose gel using TBE buffer. The DNA fragments were purified using a gel extraction kit (Sangon Biotech Co., Ltd., Shanghai, China) and ligated into the pMD18-T vector (TaKaRa, Dalian, China). The plasmids were transformed into E. coli DH5a and purified, and the inserted DNA fragments were sequenced using an automated DNA sequencer (ABI PRISM 3730, Applied Biosystems, Carlsbad, CA). To determine sequence homology and variation among the fragments amplified from RCC, TC, 2nRT, 3nRT and 4nRT hybrids, sequences were aligned using BioEdit [43] and Clustal W [44].

Fluorescence in situ Hybridization and Microscopy
Purified PCR products of 5S rDNA labeled with Dig-11-dUTP (Roche, Germany) were used as probes, and hybridization was performed according to the method described by Yi et al. (2003) [45] with minor modifications. After treatment with 30 mg/ml RNase A in 26SSC for 30 min at 37uC, the slides with chromosome metaphase spreads were denatured in 70% deionized formamide/26SSC for 2 min at 70uC, dehydrated in a 70%, 90% and 100% ethanol series for 5 min each (16SSC is 0.15 M NaCl/ 0.015 M sodium citrate, pH 7.6), and then air-dried. 4 ml of the hybridization mixture (approximately 100 ng of labeled probes, 50% formamide, 10 mg dextran sulfate/ml and 26SSC) was denatured for 10 min in boiling water, applied to the air-dried slides carrying denatured metaphase chromosomes under a 22622 mm coverslip, and sealed with rubber cement. The slides were then put in a moist chamber and allowed to incubate overnight at 37uC.
Following overnight incubation, the coverslips were removed and the slides were rinsed at 43uC in: 26SSC with 50% formamide, twice, 15 min each; 26SSC, 5 min; 16SSC, 5 min, then air-dried. The spectrum signals were achieved by application of 8 ml of 5 mg/ml FITC-conjugated antidigoxigenin antibody from sheep (Roche, Germany) and a final incubation in the humidity chamber at 37uC. After a series of washes with TNT (containing 0.1 M Tris-HCl, 0.15 M NaCl, 0.05% Tween 20) at 43uC, the slides were mounted in antifade solution containing 2 mg/ml 4',6-diamidino-2-phenylindole (DAPI) for 5 min. Slides were viewed under a Leica inverted CW4000 microscope and a Leica LCS SP2 confocal image system (Leica, Germany). Captured images were colored and overlapped in Adobe Photoshop cs4. At least 30 metaphases from each sample were analyzed.

Fertilization Rate, Hatching Rate and Adulthood Rate
In the cross of RCCR6TC= we observed a high fertilization rate (83.6%) and hatching rate (74.6%) but relatively low adulthood rate (28.6%). While in the same-species mating, the fertilization rate, hatching rate and adulthood rate of RCC were 89.9%, 85.4%, 75.5%, respectively, and that of TC were 94.9%, 89.8%, 77.4%. There were no living progeny in the reverse crosses between TCR6RCC=. Approximately 3000 2nRT, 2500 3nRT and 20 4nRT hybrids were obtained each year.
At the age of 2 years, because of the few number of 4nRT hybrids, only an artificial propagation of 2nRT and 3nRT hybrids was conducted during the reproductive season of last year, and neither 2nRT nor 3nRT hybrids were able to mate. According to our previous studies [11][12], we supposed that 4nRT hybrids were fertile though the confirmation of fertility or sterility requires further study.

Chromosome Number and Karyotypes
The chromosome numbers were determined for the mitotic metaphases of all samples. Table 1 shows the distribution of chromosome number in RCC, TC, 2nRT, 3nRT and 4nRT hybrids. For diploid RCC ( Figure 1A), 96.5% of chromosomal metaphases had 100 chromosomes with the karyotype formula of 22 m +34 sm +22 st +22 t (Figure 2A, F). For diploid TC ( Figure 1B), 94% of chromosomal metaphases possessed 48 chromosomes with the karyotype formula of 16 m +26 sm +6 st ( Figure 2B, G). In the hybrid offspring of RCCR6TC= without barbels and with low body height ( Figure 1C), 84% of chromosomal metaphases had 74 chromosomes with the karyotype formula of 19 m +30 sm +14 st +11 t ( Figure 2C, H). In the hybrid offspring of RCCR6TC= without barbels but having high  body height ( Figure 1D), 77.5% of chromosomal metaphases had 124 chromosomes with the karyotype formula of 30 m +47 sm +25 st +22 t ( Figure 2D, I). In the hybrid offspring of RCCR6TC= with barbels ( Figure 1E), 93% of chromosomal metaphases had 148 chromosomes with the karyotype formula of 38 m +60 sm +28 st +22 t ( Figure 2E, J).

DNA Content
The distribution of DNA content of all samples is shown in Table 2 and Figures 3A-E. The mean DNA content of 2nRT hybrids was equal (P.0.05) to the sum of half of RCC and TC, demonstrating that 2nRT hybrids had one set of chromosomes from both RCC and TC. The mean DNA content of 3nRT hybrids was equal (P.0.05) to the sum of RCC and half of TC, suggesting that 3nRT hybrids had two sets of chromosomes from RCC and one set of chromosomes from TC. The mean DNA content of 4nRT hybrids was equal (P.0.05) to that of RCC and TC, showing that 4nRT hybrids had two sets of chromosomes from both RCC and TC.    (Table 3). Sequence analysis showed that 2nRT hybrids presented two similarly sized 5S PCR products of 179 and 203 bp that were not distinguishable on the agarose gel, where they were seen as a single band of about 200 bp.

Nucleotide Sequence Analysis of 5S rDNA
Using BLASTn, all fragments of RCC, TC and their filial generation were confirmed to be 5S rDNA repeat units, each comprising a 3' end of the 5S rRNA region (positions 1-21), a whole NTS region, and a large 5' flanking of the coding region of the adjacent unit (positions 22-120) ( Figure 5A). In RCC, three fragments of 5S rDNA (designated class I: 203 bp; class II: 340 bp; and class III: 477 bp) were characterized by different NTS types (designated NTS-I, II, and III for the 83, 220, and 357 bp monomers, respectively). In TC, there were two fragments of 5S rDNA classes with the same coding region (designated class IV: 188 bp, and class V: 286 bp) but distinct NTS types (designated NTS-IV and NTS-V, for the 68 and 166 bp monomers, respectively) ( Figure 5B). The sequence data demonstrated that RCC and TC showed high conservation in 5S coding regions and large variation in NTS regions. The mean G + C content of the coding regions in RCC and TC were 55% and 55.8%, respectively, both GC-rich. Comparative analysis of homology of the 5S rDNA fragments between hybrid offspring and their parents showed that the diploid hybrids (2nRT) had four 5S rDNA classes, with three (class I, II and III) deriving from their female parent (RCC) and one (class IV) from their male parent (TC). The triploid hybrids (3nRT) inherited three 5S rDNA classes, with two (class II and class III) from RCC and one (class IV) from TC. The tetraploid hybrids (4nRT) inherited two 5S rDNA classes (class II and class III) from RCC and presented a new 5S rDNA sequence (designated class I-N: 203 bp) with a novel NTS sequence (designated NTS-I-N: 83 bp). The specific paternal 5S rDNA sequence class V was not found in the hybrid offspring. In addition, the 406 bp DNA fragment from 4nRT hybrids was a dimeric 5S rDNA tandem array consisting of two class I-N sequences ( Figure 5C). All the sequences, except the three that were shorter than 200 bp, were deposited in GenBank with the accession numbers JQ317900 to JQ317911.
All the internal control regions (including A box, internal element and C box) were identified in the coding regions of parents and their hybrid offspring. Although 5S rRNA genes were highly conserved in RCC and TC, several species-specific nucleotide variations were observed in the A box, positioned at 60, 61 and 63. Furthermore, among the hybrid progeny, several nucleotide variations were found not only in the ICRs but also at other four positions ( Figure 6).  (Figure 7). The thymidine residues necessary for transcription termination were also detected ( Figures 6 and 7).

Chromosomal Location of 5S rDNA
Through in situ hybridization, the 5S rDNA fluorescent probes prepared from the cloned 5S rDNA repeated units of TC (class IV and class V) were hybridized to the mitotic metaphase chromosomes of RCC, TC and their hybrid offspring. As shown in Figure 8, the probe from class IV of TC hybridized with the metaphase chromosomes of TC, 2nRT and 3nRT hybrids, but not with RCC and 4nRT hybrids; the probe from class V of TC hybridized only with the metaphase chromosomes of TC. These results and the data in Table 3 indicated the different heredity characteristic of paternal-specific 5S rDNA.

Formation of Diploid, Triploid and Tetraploid Hybrids
The different ploidy levels of hybrid fish were determined by counting the chromosomes and examining DNA content using flow cytometry. The former is a direct and accurate method for determining ploidy of samples, and the latter is a rapid, simple and equally accurate method.
Both chromosome number and karyotype formula of RCC were similar to the results of previous studies [11][12]. 2nRT, 3nRT and 4nRT hybrids possessed 74, 124 and 148 chromosomes respectively (Figure 2), in contrast to RCC (100) and TC (48). Furthermore, it is apparent that a pair of the largest submetacentric chromosomes in diploid TC can be used as marker chromosomes for identifying TC from RCC. Possessing 74 chromosomes and a submetacentric largest chromosome, 2nRT hybrids were proved to have obtained one set of chromosomes from RCC and one set of chromosomes from TC ( Figure 2C, H). With 124 chromosomes and a submetacentric largest chromosome, 3nRT hybrids apparently contain two sets of chromosomes from RCC and one set of chromosomes from TC ( Figure 2D, I). 4nRT hybrids had 148 chromosomes and one pair of submetacentric chromosomes and are suggested to harbor two sets of chromosomes from RCC and two sets of chromosomes from TC ( Figure 2E, J). In addition, there was a significant relationship between the sum of the mean DNA content of the hybrid offspring and that of their parents (Figure 3, Table 2). These results indicate the hybridization origin of the progeny of RCCR6TC=, rather than gynogenesis or androgenesis. A-E (mitotic metaphase chromosomes of RCC, TC, 2nRT, 3nRT and 4nRT hybrids respectively) show the single-label FISH results hybridized with a probe from the cloned repeated fragments of class IV; F-J (mitotic metaphase chromosomes of RCC, TC, 2nRT, 3nRT and 4nRT hybrids respectively) demonstrate the hybridization results after single-label FISH with a probe from the cloned repeated fragments of class V. The white arrowheads indicate the fluorescent signals (green) of 5S rDNA. Scale bar in A-J, 3 mm. doi:10.1371/journal.pone.0038976.g008 Polyploidization is an important and frequent event in lower vertebrate evolution, especially that of fish [46][47]. In our previous studies, we successfully obtained two kinds of bisexual fertile tetraploid hybrids (4 nAT and 4 nRB) [11][12]. In the case of 4 nAT, there were only diploid hybrids with 100 chromosomes in the F 1 -F 2 generation, while the tetraploid hybrids with 200 chromosomes were found in F 3 and the later generations, being formed via the fusion of unreduced gametes. For 4 nRB, only 3 nRB and 4 nRB hybrids were formed, and no diploid hybrids were found to survive past the first generation (although diploid embryos were observed). 3 nRB hybrids resulted from the retention of the second polar body, whereas 4 nRB hybrids resulted from somatic doubling. Recently, the coexistence of diploid, triploid and tetraploid crucian carp (Carassius auratus) in natural waters has been reported [48]. In the present study, viable 2nRT, 3nRT and 4nRT hybrids were created for the first time by crossing different parents (RCC and TC) belonging to distinct subfamilies with different chromosome numbers. We hypothesize that the different ploidy numbers of the hybrid offspring arose by non-disjunctive or incomplete cell division during the first doubling.

Molecular Organization of the 5S rDNA
Numerous studies have reported the retention of more than one 5S rDNA repeat class and the elimination of parental-specific 5S rDNA following polyploidization in plants and fishes [40,[49][50][51][52][53][54][55]. As previously reported by Qin et al. (2010), RCC was found to have three types of 5S rDNA sequences owing to a probably ancient polyploidization. 3 nRB and 4 nRB hybrids possessed four and three types of 5S rDNA sequences, respectively, following allopolyploidization, and a paternal-specific loss of 5S rDNA was also observed in 4 nRB hybrids. A similar result was found in the present study, with three types of 5S rDNA sequences observed in RCC having high similarity (100%, 99.7% and 98.7%, respectively) to sequences in GenBank (GQ485555, GQ485556 and GQ485557) which were uploaded by Qin et al. (2009). Moreover, only two types of 5S rDNA sequences were detected in TC, favoring the hypothesis that the presence of two classes of 5S rDNA is a general trend for the 5S rRNA gene organization in the fish genome [38]. However, 2nRT, 3nRT and 4nRT hybrids had four, three and three types of 5S rDNA sequences, respectively, partially inheriting 5S rDNA classes from their parents. FISH results (Figure 8) also confirmed the loss of paternal-specific 5S rDNA classes. Moreover, the internal control regions (A box, IE and C box), TATA-like control element and thymidine residues which are necessary for the correct gene expression, were observed in all the sequences of the parents and their hybrid offspring, indicating that all sequences analyzed here were functional genes.
Hybridization can result in genomic changes including alterations of gene expression, chromosomal structure, and genome size [56]. In the present work, we detected several nucleotide mutations in the coding regions of the hybrid progeny, even in the ICRs (Figure 6). We also observed nucleotide polymorphisms including base constitution and insertion-deletion in the NTS sequences of the hybrid offspring (Figure 7).
It has been reported that the NTS sequence could be employed as a molecular marker for species identification and phylogenetic studies [17,35,40,[57][58][59][60]. Here, three NTS types (NTS-I, NTS-II and NTS-III) with different length and base constitutions were detected in RCC. TC had two types of distinct NTS (NTS-IV and NTS-V) (Figure 7). Four types of NTS (NTS-I, NTS-II, NTS-III and NTS-IV) were detected in 2nRT hybrids, with types III and IV resulting from an insertion in the NTS-III sequence and a deletion in the NTS-IV sequence, respectively. The 3nRT hybrids had three NTS types (NTS-II, NTS-III and NTS-IV), and the same insertion was also detected in NTS-III. There were three types of NTS (NTS-I-N, NTS-II and NTS-III) in 4nRT hybrids, with NTS-I-N having several base mutations. These variations in NTS sequences indicate that 5S rDNA can serve as a suitable genetic marker for evolutionary studies and for the genetic identification of hybrid fish and their parent species.
In previous studies, a minimum length (59 bp) of NTS sequence was found to maintain the array and the expression/regulation dynamic of 5S rRNA genes in the fish genome [38,[61][62]. 2nRT hybrids inherited nearly all the 5S rDNA classes from their parents (with the exception of the class V 5S rDNA of TC). Furthermore, almost all the 5S rDNA classes preserved the structural organization characteristic of their parents (with the exception of a deletion of a T at position 259 and a sequence of GAATACCT at positions 230 to 237 in the NTS-IV sequences of TC) (Figure 7). It's the first time to describe this finding in F 1 hybrids of fish and provides further evidence that a 59 bp length of NTS sequence is the minimum necessary to guarantee the array and the dynamics of the 5S rRNA genes.
In this study, 4nRT hybrids lost all the paternal-specific 5S rDNA classes (IV and V) and presented a newly formed allopolyploid-special 5S rDNA sequence (class I-N). This novel 5S rDNA sequence showed a low similarity to that of their parents but a high homology to that of common carp (Cyprinus carpio, CC, Accession number: AB015590), even though this species belongs to a different genus (Figure 7, Table 4). It's the first report of this phenomenon in the cross of RCCR6TC=. Similar findings have previously been reported [12,40,63]. For example, the barbels and the Sox gene, which were special traits for CC, presented in 4 nRB hybrids but were absent in 3 nRB hybrids and their parents (RCC and BSB). Importantly, there was also an allopolyploid-special 5S rDNA sequence (Accession number: GU329957) of 4 nRB hybrids with low similarity to that of their parents but high homology to that of CC (Accession number: AB015590). The formation of CCspecific DNA sequences and genes in allopolyploid hybrids imply that hybridization and allopolyploidization could result in rapid genomic changes in the hybrid offspring of RCCR6TC=. In addition, it has been reported that a drastic elimination and recombination of the parental-specific rDNA repeat units has some relationship to the fertility of newly formed hybrids or allopolyploids [55,[64][65]. Thus, to improve fertility, rapidly genetic recombination was required in 4nRT hybrids, which led to the absence of parental-specific 5S rDNA and the presence of a novel 5S rDNA. Consequently, we can assume that 4nRT hybrids were fertile, although confirmation of fertility or sterility requires further study. However, there was no data available to allow interpretation of the possible evolutionary relationship between CC and the allopolyploid hybrids.
In conclusion, this is the first report on the formation of these viable diploid, triploid and tetraploid hybrids by crossing different parents with a different chromosome number in vertebrates. Results observed here confirmed the influence of hybridization and allopolyploidization on the 5S rDNA organization, and provided new information on the genetic variation of the 5S rDNA multigene family in vertebrates. We will continue the comparative study of genomic and phenotypic patterns in the hybrid offspring in our future studies.

Author Contributions
Conceived and designed the experiments: WH QQ SL YL Performed the experiments: WH TL LX. Analyzed the data: WH QQ SL TL JW JX. Contributed reagents/materials/analysis tools: WH TL JW JX CZ. Wrote the manuscript: WH SL.